Vendor in files for runc/libcontainer

vndr was previously removing .go files in runc/libcontainer since they weren't used anywhere
in cri-o, but kpod stats will use them

Signed-off-by: Ryan Cole <rcyoalne@gmail.com>

vendor/github.com/opencontainers/runc/libcontainer/capabilities_linux.go

@@ -0,0 +1,114 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"strings"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/syndtr/gocapability/capability"
+)
+
+const allCapabilityTypes = capability.CAPS | capability.BOUNDS | capability.AMBS
+
+var capabilityMap map[string]capability.Cap
+
+func init() {
+	capabilityMap = make(map[string]capability.Cap)
+	last := capability.CAP_LAST_CAP
+	// workaround for RHEL6 which has no /proc/sys/kernel/cap_last_cap
+	if last == capability.Cap(63) {
+		last = capability.CAP_BLOCK_SUSPEND
+	}
+	for _, cap := range capability.List() {
+		if cap > last {
+			continue
+		}
+		capKey := fmt.Sprintf("CAP_%s", strings.ToUpper(cap.String()))
+		capabilityMap[capKey] = cap
+	}
+}
+
+func newContainerCapList(capConfig *configs.Capabilities) (*containerCapabilities, error) {
+	bounding := []capability.Cap{}
+	for _, c := range capConfig.Bounding {
+		v, ok := capabilityMap[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		bounding = append(bounding, v)
+	}
+	effective := []capability.Cap{}
+	for _, c := range capConfig.Effective {
+		v, ok := capabilityMap[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		effective = append(effective, v)
+	}
+	inheritable := []capability.Cap{}
+	for _, c := range capConfig.Inheritable {
+		v, ok := capabilityMap[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		inheritable = append(inheritable, v)
+	}
+	permitted := []capability.Cap{}
+	for _, c := range capConfig.Permitted {
+		v, ok := capabilityMap[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		permitted = append(permitted, v)
+	}
+	ambient := []capability.Cap{}
+	for _, c := range capConfig.Ambient {
+		v, ok := capabilityMap[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		ambient = append(ambient, v)
+	}
+	pid, err := capability.NewPid(os.Getpid())
+	if err != nil {
+		return nil, err
+	}
+	return &containerCapabilities{
+		bounding:    bounding,
+		effective:   effective,
+		inheritable: inheritable,
+		permitted:   permitted,
+		ambient:     ambient,
+		pid:         pid,
+	}, nil
+}
+
+type containerCapabilities struct {
+	pid         capability.Capabilities
+	bounding    []capability.Cap
+	effective   []capability.Cap
+	inheritable []capability.Cap
+	permitted   []capability.Cap
+	ambient     []capability.Cap
+}
+
+// ApplyBoundingSet sets the capability bounding set to those specified in the whitelist.
+func (c *containerCapabilities) ApplyBoundingSet() error {
+	c.pid.Clear(capability.BOUNDS)
+	c.pid.Set(capability.BOUNDS, c.bounding...)
+	return c.pid.Apply(capability.BOUNDS)
+}
+
+// Apply sets all the capabilities for the current process in the config.
+func (c *containerCapabilities) ApplyCaps() error {
+	c.pid.Clear(allCapabilityTypes)
+	c.pid.Set(capability.BOUNDS, c.bounding...)
+	c.pid.Set(capability.PERMITTED, c.permitted...)
+	c.pid.Set(capability.INHERITABLE, c.inheritable...)
+	c.pid.Set(capability.EFFECTIVE, c.effective...)
+	c.pid.Set(capability.AMBIENT, c.ambient...)
+	return c.pid.Apply(allCapabilityTypes)
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/rootless/rootless.go

@@ -0,0 +1,128 @@
+// +build linux
+
+package rootless
+
+import (
+	"fmt"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/cgroups/fs"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/configs/validate"
+)
+
+// TODO: This is copied from libcontainer/cgroups/fs, which duplicates this code
+//       needlessly. We should probably export this list.
+
+var subsystems = []subsystem{
+	&fs.CpusetGroup{},
+	&fs.DevicesGroup{},
+	&fs.MemoryGroup{},
+	&fs.CpuGroup{},
+	&fs.CpuacctGroup{},
+	&fs.PidsGroup{},
+	&fs.BlkioGroup{},
+	&fs.HugetlbGroup{},
+	&fs.NetClsGroup{},
+	&fs.NetPrioGroup{},
+	&fs.PerfEventGroup{},
+	&fs.FreezerGroup{},
+	&fs.NameGroup{GroupName: "name=systemd"},
+}
+
+type subsystem interface {
+	// Name returns the name of the subsystem.
+	Name() string
+
+	// Returns the stats, as 'stats', corresponding to the cgroup under 'path'.
+	GetStats(path string, stats *cgroups.Stats) error
+}
+
+// The noop cgroup manager is used for rootless containers, because we currently
+// cannot manage cgroups if we are in a rootless setup. This manager is chosen
+// by factory if we are in rootless mode. We error out if any cgroup options are
+// set in the config -- this may change in the future with upcoming kernel features
+// like the cgroup namespace.
+
+type Manager struct {
+	Cgroups *configs.Cgroup
+	Paths   map[string]string
+}
+
+func (m *Manager) Apply(pid int) error {
+	// If there are no cgroup settings, there's nothing to do.
+	if m.Cgroups == nil {
+		return nil
+	}
+
+	// We can't set paths.
+	// TODO(cyphar): Implement the case where the runner of a rootless container
+	//               owns their own cgroup, which would allow us to set up a
+	//               cgroup for each path.
+	if m.Cgroups.Paths != nil {
+		return fmt.Errorf("cannot change cgroup path in rootless container")
+	}
+
+	// We load the paths into the manager.
+	paths := make(map[string]string)
+	for _, sys := range subsystems {
+		name := sys.Name()
+
+		path, err := cgroups.GetOwnCgroupPath(name)
+		if err != nil {
+			// Ignore paths we couldn't resolve.
+			continue
+		}
+
+		paths[name] = path
+	}
+
+	m.Paths = paths
+	return nil
+}
+
+func (m *Manager) GetPaths() map[string]string {
+	return m.Paths
+}
+
+func (m *Manager) Set(container *configs.Config) error {
+	// We have to re-do the validation here, since someone might decide to
+	// update a rootless container.
+	return validate.New().Validate(container)
+}
+
+func (m *Manager) GetPids() ([]int, error) {
+	dir, err := cgroups.GetOwnCgroupPath("devices")
+	if err != nil {
+		return nil, err
+	}
+	return cgroups.GetPids(dir)
+}
+
+func (m *Manager) GetAllPids() ([]int, error) {
+	dir, err := cgroups.GetOwnCgroupPath("devices")
+	if err != nil {
+		return nil, err
+	}
+	return cgroups.GetAllPids(dir)
+}
+
+func (m *Manager) GetStats() (*cgroups.Stats, error) {
+	// TODO(cyphar): We can make this work if we figure out a way to allow usage
+	//               of cgroups with a rootless container. While this doesn't
+	//               actually require write access to a cgroup directory, the
+	//               statistics are not useful if they can be affected by
+	//               non-container processes.
+	return nil, fmt.Errorf("cannot get cgroup stats in rootless container")
+}
+
+func (m *Manager) Freeze(state configs.FreezerState) error {
+	// TODO(cyphar): We can make this work if we figure out a way to allow usage
+	//               of cgroups with a rootless container.
+	return fmt.Errorf("cannot use freezer cgroup in rootless container")
+}
+
+func (m *Manager) Destroy() error {
+	// We don't have to do anything here because we didn't do any setup.
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/compat_1.5_linux.go

@@ -0,0 +1,10 @@
+// +build linux,!go1.5
+
+package libcontainer
+
+import "syscall"
+
+// GidMappingsEnableSetgroups was added in Go 1.5, so do nothing when building
+// with earlier versions
+func enableSetgroups(sys *syscall.SysProcAttr) {
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/validate/rootless.go

@@ -0,0 +1,117 @@
+package validate
+
+import (
+	"fmt"
+	"os"
+	"reflect"
+	"strings"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var (
+	geteuid = os.Geteuid
+	getegid = os.Getegid
+)
+
+func (v *ConfigValidator) rootless(config *configs.Config) error {
+	if err := rootlessMappings(config); err != nil {
+		return err
+	}
+	if err := rootlessMount(config); err != nil {
+		return err
+	}
+	// Currently, cgroups cannot effectively be used in rootless containers.
+	// The new cgroup namespace doesn't really help us either because it doesn't
+	// have nice interactions with the user namespace (we're working with upstream
+	// to fix this).
+	if err := rootlessCgroup(config); err != nil {
+		return err
+	}
+
+	// XXX: We currently can't verify the user config at all, because
+	//      configs.Config doesn't store the user-related configs. So this
+	//      has to be verified by setupUser() in init_linux.go.
+
+	return nil
+}
+
+func rootlessMappings(config *configs.Config) error {
+	rootuid, err := config.HostRootUID()
+	if err != nil {
+		return fmt.Errorf("failed to get root uid from uidMappings: %v", err)
+	}
+	if euid := geteuid(); euid != 0 {
+		if !config.Namespaces.Contains(configs.NEWUSER) {
+			return fmt.Errorf("rootless containers require user namespaces")
+		}
+		if rootuid != euid {
+			return fmt.Errorf("rootless containers cannot map container root to a different host user")
+		}
+	}
+
+	rootgid, err := config.HostRootGID()
+	if err != nil {
+		return fmt.Errorf("failed to get root gid from gidMappings: %v", err)
+	}
+
+	// Similar to the above test, we need to make sure that we aren't trying to
+	// map to a group ID that we don't have the right to be.
+	if rootgid != getegid() {
+		return fmt.Errorf("rootless containers cannot map container root to a different host group")
+	}
+
+	// We can only map one user and group inside a container (our own).
+	if len(config.UidMappings) != 1 || config.UidMappings[0].Size != 1 {
+		return fmt.Errorf("rootless containers cannot map more than one user")
+	}
+	if len(config.GidMappings) != 1 || config.GidMappings[0].Size != 1 {
+		return fmt.Errorf("rootless containers cannot map more than one group")
+	}
+
+	return nil
+}
+
+// cgroup verifies that the user isn't trying to set any cgroup limits or paths.
+func rootlessCgroup(config *configs.Config) error {
+	// Nothing set at all.
+	if config.Cgroups == nil || config.Cgroups.Resources == nil {
+		return nil
+	}
+
+	// Used for comparing to the zero value.
+	left := reflect.ValueOf(*config.Cgroups.Resources)
+	right := reflect.Zero(left.Type())
+
+	// This is all we need to do, since specconv won't add cgroup options in
+	// rootless mode.
+	if !reflect.DeepEqual(left.Interface(), right.Interface()) {
+		return fmt.Errorf("cannot specify resource limits in rootless container")
+	}
+
+	return nil
+}
+
+// mount verifies that the user isn't trying to set up any mounts they don't have
+// the rights to do. In addition, it makes sure that no mount has a `uid=` or
+// `gid=` option that doesn't resolve to root.
+func rootlessMount(config *configs.Config) error {
+	// XXX: We could whitelist allowed devices at this point, but I'm not
+	//      convinced that's a good idea. The kernel is the best arbiter of
+	//      access control.
+
+	for _, mount := range config.Mounts {
+		// Check that the options list doesn't contain any uid= or gid= entries
+		// that don't resolve to root.
+		for _, opt := range strings.Split(mount.Data, ",") {
+			if strings.HasPrefix(opt, "uid=") && opt != "uid=0" {
+				return fmt.Errorf("cannot specify uid= mount options in rootless containers where argument isn't 0")
+			}
+			if strings.HasPrefix(opt, "gid=") && opt != "gid=0" {
+				return fmt.Errorf("cannot specify gid= mount options in rootless containers where argument isn't 0")
+			}
+		}
+	}
+
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/validate/validator.go

@@ -0,0 +1,195 @@
+package validate
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+	selinux "github.com/opencontainers/selinux/go-selinux"
+)
+
+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
+	}
+	if err := v.sysctl(config); err != nil {
+		return err
+	}
+	if config.Rootless {
+		if err := v.rootless(config); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// rootfs validates if 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 {
+	if _, err := os.Stat(config.Rootfs); err != nil {
+		if os.IsNotExist(err) {
+			return fmt.Errorf("rootfs (%s) does not exist", config.Rootfs)
+		}
+		return err
+	}
+	cleaned, err := filepath.Abs(config.Rootfs)
+	if err != nil {
+		return err
+	}
+	if cleaned, err = filepath.EvalSymlinks(cleaned); err != nil {
+		return err
+	}
+	if filepath.Clean(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")
+	}
+	if config.ProcessLabel != "" && !selinux.GetEnabled() {
+		return fmt.Errorf("selinux label is specified in config, but selinux is disabled or not supported")
+	}
+
+	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
+}
+
+// sysctl validates that the specified sysctl keys are valid or not.
+// /proc/sys isn't completely namespaced and depending on which namespaces
+// are specified, a subset of sysctls are permitted.
+func (v *ConfigValidator) sysctl(config *configs.Config) error {
+	validSysctlMap := map[string]bool{
+		"kernel.msgmax":          true,
+		"kernel.msgmnb":          true,
+		"kernel.msgmni":          true,
+		"kernel.sem":             true,
+		"kernel.shmall":          true,
+		"kernel.shmmax":          true,
+		"kernel.shmmni":          true,
+		"kernel.shm_rmid_forced": true,
+	}
+
+	for s := range config.Sysctl {
+		if validSysctlMap[s] || strings.HasPrefix(s, "fs.mqueue.") {
+			if config.Namespaces.Contains(configs.NEWIPC) {
+				continue
+			} else {
+				return fmt.Errorf("sysctl %q is not allowed in the hosts ipc namespace", s)
+			}
+		}
+		if strings.HasPrefix(s, "net.") {
+			if config.Namespaces.Contains(configs.NEWNET) {
+				if path := config.Namespaces.PathOf(configs.NEWNET); path != "" {
+					if err := checkHostNs(s, path); err != nil {
+						return err
+					}
+				}
+				continue
+			} else {
+				return fmt.Errorf("sysctl %q is not allowed in the hosts network namespace", s)
+			}
+		}
+		return fmt.Errorf("sysctl %q is not in a separate kernel namespace", s)
+	}
+
+	return nil
+}
+
+func isSymbolicLink(path string) (bool, error) {
+	fi, err := os.Lstat(path)
+	if err != nil {
+		return false, err
+	}
+
+	return fi.Mode()&os.ModeSymlink == os.ModeSymlink, nil
+}
+
+// checkHostNs checks whether network sysctl is used in host namespace.
+func checkHostNs(sysctlConfig string, path string) error {
+	var currentProcessNetns = "/proc/self/ns/net"
+	// readlink on the current processes network namespace
+	destOfCurrentProcess, err := os.Readlink(currentProcessNetns)
+	if err != nil {
+		return fmt.Errorf("read soft link %q error", currentProcessNetns)
+	}
+
+	// First check if the provided path is a symbolic link
+	symLink, err := isSymbolicLink(path)
+	if err != nil {
+		return fmt.Errorf("could not check that %q is a symlink: %v", path, err)
+	}
+
+	if symLink == false {
+		// The provided namespace is not a symbolic link,
+		// it is not the host namespace.
+		return nil
+	}
+
+	// readlink on the path provided in the struct
+	destOfContainer, err := os.Readlink(path)
+	if err != nil {
+		return fmt.Errorf("read soft link %q error", path)
+	}
+	if destOfContainer == destOfCurrentProcess {
+		return fmt.Errorf("sysctl %q is not allowed in the hosts network namespace", sysctlConfig)
+	}
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/console.go

@@ -0,0 +1,17 @@
+package libcontainer
+
+import (
+	"io"
+	"os"
+)
+
+// Console represents a pseudo TTY.
+type Console interface {
+	io.ReadWriteCloser
+
+	// Path returns the filesystem path to the slave side of the pty.
+	Path() string
+
+	// Fd returns the fd for the master of the pty.
+	File() *os.File
+}

vendor/github.com/opencontainers/runc/libcontainer/console_freebsd.go

@@ -0,0 +1,13 @@
+// +build freebsd
+
+package libcontainer
+
+import (
+	"errors"
+)
+
+// newConsole returns an initialized 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() (Console, error) {
+	return nil, errors.New("libcontainer console is not supported on FreeBSD")
+}

vendor/github.com/opencontainers/runc/libcontainer/console_linux.go

@@ -0,0 +1,152 @@
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"unsafe"
+
+	"golang.org/x/sys/unix"
+)
+
+func ConsoleFromFile(f *os.File) Console {
+	return &linuxConsole{
+		master: f,
+	}
+}
+
+// newConsole returns an initialized 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() (Console, error) {
+	master, err := os.OpenFile("/dev/ptmx", unix.O_RDWR|unix.O_NOCTTY|unix.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 &linuxConsole{
+		slavePath: console,
+		master:    master,
+	}, nil
+}
+
+// linuxConsole is a linux pseudo TTY for use within a container.
+type linuxConsole struct {
+	master    *os.File
+	slavePath string
+}
+
+func (c *linuxConsole) File() *os.File {
+	return c.master
+}
+
+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() error {
+	oldMask := unix.Umask(0000)
+	defer unix.Umask(oldMask)
+	f, err := os.Create("/dev/console")
+	if err != nil && !os.IsExist(err) {
+		return err
+	}
+	if f != nil {
+		f.Close()
+	}
+	return unix.Mount(c.slavePath, "/dev/console", "bind", unix.MS_BIND, "")
+}
+
+// dupStdio opens the slavePath for the console and dups the fds to the current
+// processes stdio, fd 0,1,2.
+func (c *linuxConsole) dupStdio() error {
+	slave, err := c.open(unix.O_RDWR)
+	if err != nil {
+		return err
+	}
+	fd := int(slave.Fd())
+	for _, i := range []int{0, 1, 2} {
+		if err := unix.Dup3(fd, i, 0); 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 := unix.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 := unix.Syscall(unix.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(), unix.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) {
+	n, err := unix.IoctlGetInt(int(f.Fd()), unix.TIOCGPTN)
+	if err != nil {
+		return "", err
+	}
+	return fmt.Sprintf("/dev/pts/%d", n), nil
+}
+
+// SaneTerminal sets the necessary tty_ioctl(4)s to ensure that a pty pair
+// created by us acts normally. In particular, a not-very-well-known default of
+// Linux unix98 ptys is that they have +onlcr by default. While this isn't a
+// problem for terminal emulators, because we relay data from the terminal we
+// also relay that funky line discipline.
+func SaneTerminal(terminal *os.File) error {
+	termios, err := unix.IoctlGetTermios(int(terminal.Fd()), unix.TCGETS)
+	if err != nil {
+		return fmt.Errorf("ioctl(tty, tcgets): %s", err.Error())
+	}
+
+	// Set -onlcr so we don't have to deal with \r.
+	termios.Oflag &^= unix.ONLCR
+
+	if err := unix.IoctlSetTermios(int(terminal.Fd()), unix.TCSETS, termios); err != nil {
+		return fmt.Errorf("ioctl(tty, tcsets): %s", err.Error())
+	}
+
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/console_solaris.go

@@ -0,0 +1,11 @@
+package libcontainer
+
+import (
+	"errors"
+)
+
+// newConsole returns an initialized 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() (Console, error) {
+	return nil, errors.New("libcontainer console is not supported on Solaris")
+}

vendor/github.com/opencontainers/runc/libcontainer/console_windows.go

@@ -0,0 +1,30 @@
+package libcontainer
+
+// newConsole returns an initialized console that can be used within a container
+func newConsole() (Console, error) {
+	return &windowsConsole{}, nil
+}
+
+// windowsConsole is a Windows pseudo TTY for use within a container.
+type windowsConsole struct {
+}
+
+func (c *windowsConsole) Fd() uintptr {
+	return 0
+}
+
+func (c *windowsConsole) Path() string {
+	return ""
+}
+
+func (c *windowsConsole) Read(b []byte) (int, error) {
+	return 0, nil
+}
+
+func (c *windowsConsole) Write(b []byte) (int, error) {
+	return 0, nil
+}
+
+func (c *windowsConsole) Close() error {
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/container.go

@@ -0,0 +1,166 @@
+// Package 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 (
+	"os"
+	"time"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+// Status is the status of a container.
+type Status int
+
+const (
+	// Created is the status that denotes the container exists but has not been run yet.
+	Created Status = iota
+	// Running is the status that denotes the container exists and is running.
+	Running
+	// Pausing is the status that denotes the container exists, it is in the process of being paused.
+	Pausing
+	// Paused is the status that denotes the container exists, but all its processes are paused.
+	Paused
+	// Stopped is the status that denotes the container does not have a created or running process.
+	Stopped
+)
+
+func (s Status) String() string {
+	switch s {
+	case Created:
+		return "created"
+	case Running:
+		return "running"
+	case Pausing:
+		return "pausing"
+	case Paused:
+		return "paused"
+	case Stopped:
+		return "stopped"
+	default:
+		return "unknown"
+	}
+}
+
+// BaseState represents the platform agnostic pieces relating to a
+// running container's state
+type BaseState 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 in clock cycles since boot time.
+	InitProcessStartTime uint64 `json:"init_process_start"`
+
+	// Created is the unix timestamp for the creation time of the container in UTC
+	Created time.Time `json:"created"`
+
+	// Config is the container's configuration.
+	Config configs.Config `json:"config"`
+}
+
+// BaseContainer is a libcontainer container object.
+//
+// Each container is thread-safe within the same process. Since a container can
+// be destroyed by a separate process, any function may return that the container
+// was not found. BaseContainer includes methods that are platform agnostic.
+type BaseContainer interface {
+	// Returns the ID of the container
+	ID() string
+
+	// Returns the current status of the container.
+	//
+	// errors:
+	// ContainerNotExists - Container no longer exists,
+	// Systemerror - System error.
+	Status() (Status, error)
+
+	// State returns the current container's state information.
+	//
+	// errors:
+	// SystemError - System error.
+	State() (*State, error)
+
+	// Returns the current config of the container.
+	Config() configs.Config
+
+	// Returns the PIDs inside this container. The PIDs are in the namespace of the calling process.
+	//
+	// errors:
+	// ContainerNotExists - 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:
+	// ContainerNotExists - Container no longer exists,
+	// Systemerror - System error.
+	Stats() (*Stats, error)
+
+	// Set resources of container as configured
+	//
+	// We can use this to change resources when containers are running.
+	//
+	// errors:
+	// SystemError - System error.
+	Set(config configs.Config) error
+
+	// Start a process inside the container. Returns error if process fails to
+	// start. You can track process lifecycle with passed Process structure.
+	//
+	// errors:
+	// ContainerNotExists - Container no longer exists,
+	// ConfigInvalid - config is invalid,
+	// ContainerPaused - Container is paused,
+	// SystemError - System error.
+	Start(process *Process) (err error)
+
+	// Run immediately starts the process inside the container.  Returns error if process
+	// fails to start.  It does not block waiting for the exec fifo  after start returns but
+	// opens the fifo after start returns.
+	//
+	// errors:
+	// ContainerNotExists - Container no longer exists,
+	// ConfigInvalid - config is invalid,
+	// ContainerPaused - Container is paused,
+	// SystemError - System error.
+	Run(process *Process) (err error)
+
+	// Destroys the container, if its in a valid state, after killing any
+	// remaining running processes.
+	//
+	// Any event registrations are removed before the container is destroyed.
+	// No error is returned if the container is already destroyed.
+	//
+	// Running containers must first be stopped using Signal(..).
+	// Paused containers must first be resumed using Resume(..).
+	//
+	// errors:
+	// ContainerNotStopped - Container is still running,
+	// ContainerPaused - Container is paused,
+	// SystemError - System error.
+	Destroy() error
+
+	// Signal sends the provided signal code to the container's initial process.
+	//
+	// If all is specified the signal is sent to all processes in the container
+	// including the initial process.
+	//
+	// errors:
+	// SystemError - System error.
+	Signal(s os.Signal, all bool) error
+
+	// Exec signals the container to exec the users process at the end of the init.
+	//
+	// errors:
+	// SystemError - System error.
+	Exec() error
+}

vendor/github.com/opencontainers/runc/libcontainer/container_solaris.go

@@ -0,0 +1,20 @@
+package libcontainer
+
+// State represents a running container's state
+type State struct {
+	BaseState
+
+	// Platform specific fields below here
+}
+
+// A libcontainer container object.
+//
+// Each container is thread-safe within the same process. Since a container can
+// be destroyed by a separate process, any function may return that the container
+// was not found.
+type Container interface {
+	BaseContainer
+
+	// Methods below here are platform specific
+
+}

vendor/github.com/opencontainers/runc/libcontainer/container_windows.go

@@ -0,0 +1,20 @@
+package libcontainer
+
+// State represents a running container's state
+type State struct {
+	BaseState
+
+	// Platform specific fields below here
+}
+
+// A libcontainer container object.
+//
+// Each container is thread-safe within the same process. Since a container can
+// be destroyed by a separate process, any function may return that the container
+// was not found.
+type Container interface {
+	BaseContainer
+
+	// Methods below here are platform specific
+
+}

vendor/github.com/opencontainers/runc/libcontainer/criu_opts_linux.go

@@ -0,0 +1,37 @@
+package libcontainer
+
+// cgroup restoring strategy provided by criu
+type cgMode uint32
+
+const (
+	CRIU_CG_MODE_SOFT    cgMode = 3 + iota // restore cgroup properties if only dir created by criu
+	CRIU_CG_MODE_FULL                      // always restore all cgroups and their properties
+	CRIU_CG_MODE_STRICT                    // restore all, requiring them to not present in the system
+	CRIU_CG_MODE_DEFAULT                   // the same as CRIU_CG_MODE_SOFT
+)
+
+type CriuPageServerInfo struct {
+	Address string // IP address of CRIU page server
+	Port    int32  // port number of CRIU page server
+}
+
+type VethPairName struct {
+	ContainerInterfaceName string
+	HostInterfaceName      string
+}
+
+type CriuOpts struct {
+	ImagesDirectory         string             // directory for storing image files
+	WorkDirectory           string             // directory to cd and write logs/pidfiles/stats to
+	ParentImage             string             // direcotry for storing parent image files in pre-dump and dump
+	LeaveRunning            bool               // leave container in running state after checkpoint
+	TcpEstablished          bool               // checkpoint/restore established TCP connections
+	ExternalUnixConnections bool               // allow external unix connections
+	ShellJob                bool               // allow to dump and restore shell jobs
+	FileLocks               bool               // handle file locks, for safety
+	PreDump                 bool               // call criu predump to perform iterative checkpoint
+	PageServer              CriuPageServerInfo // allow to dump to criu page server
+	VethPairs               []VethPairName     // pass the veth to criu when restore
+	ManageCgroupsMode       cgMode             // dump or restore cgroup mode
+	EmptyNs                 uint32             // don't c/r properties for namespace from this mask
+}

vendor/github.com/opencontainers/runc/libcontainer/criu_opts_windows.go

@@ -0,0 +1,6 @@
+package libcontainer
+
+// TODO Windows: This can ultimately be entirely factored out as criu is
+// a Unix concept not relevant on Windows.
+type CriuOpts struct {
+}

vendor/github.com/opencontainers/runc/libcontainer/criurpc/criurpc.proto

@@ -0,0 +1,195 @@
+syntax = "proto2";
+
+message criu_page_server_info {
+	optional string		address	= 1;
+	optional int32		port	= 2;
+	optional int32		pid	= 3;
+	optional int32		fd	= 4;
+}
+
+message criu_veth_pair {
+	required string		if_in	= 1;
+	required string		if_out	= 2;
+};
+
+message ext_mount_map {
+	required string		key	= 1;
+	required string		val	= 2;
+};
+
+message join_namespace {
+	required string		ns		= 1;
+	required string		ns_file		= 2;
+	optional string		extra_opt	= 3;
+}
+
+message inherit_fd {
+	required string		key	= 1;
+	required int32		fd	= 2;
+};
+
+message cgroup_root {
+	optional string		ctrl	= 1;
+	required string		path	= 2;
+};
+
+message unix_sk {
+	required uint32		inode 	= 1;
+};
+
+enum criu_cg_mode {
+	IGNORE	= 0;
+	CG_NONE	= 1;
+	PROPS	= 2;
+	SOFT	= 3;
+	FULL	= 4;
+	STRICT	= 5;
+	DEFAULT = 6;
+};
+
+message criu_opts {
+	required int32			images_dir_fd	= 1;
+	optional int32			pid		= 2; /* if not set on dump, will dump requesting process */
+
+	optional bool			leave_running	= 3;
+	optional bool			ext_unix_sk	= 4;
+	optional bool			tcp_established	= 5;
+	optional bool			evasive_devices	= 6;
+	optional bool			shell_job	= 7;
+	optional bool			file_locks	= 8;
+	optional int32			log_level	= 9 [default = 2];
+	optional string			log_file	= 10; /* No subdirs are allowed. Consider using work-dir */
+
+	optional criu_page_server_info	ps		= 11;
+
+	optional bool			notify_scripts	= 12;
+
+	optional string			root		= 13;
+	optional string			parent_img	= 14;
+	optional bool			track_mem	= 15;
+	optional bool			auto_dedup	= 16;
+
+	optional int32			work_dir_fd	= 17;
+	optional bool			link_remap	= 18;
+	repeated criu_veth_pair		veths		= 19; /* DEPRECATED, use external instead */
+
+	optional uint32			cpu_cap		= 20 [default = 0xffffffff];
+	optional bool			force_irmap	= 21;
+	repeated string			exec_cmd	= 22;
+
+	repeated ext_mount_map		ext_mnt		= 23; /* DEPRECATED, use external instead */
+	optional bool			manage_cgroups	= 24; /* backward compatibility */
+	repeated cgroup_root		cg_root		= 25;
+
+	optional bool			rst_sibling	= 26; /* swrk only */
+	repeated inherit_fd		inherit_fd	= 27; /* swrk only */
+
+	optional bool			auto_ext_mnt	= 28;
+	optional bool			ext_sharing 	= 29;
+	optional bool			ext_masters	= 30;
+
+	repeated string			skip_mnt	= 31;
+	repeated string			enable_fs	= 32;
+
+	repeated unix_sk                unix_sk_ino     = 33; /* DEPRECATED, use external instead */
+
+	optional criu_cg_mode		manage_cgroups_mode = 34;
+	optional uint32			ghost_limit	= 35 [default = 0x100000];
+	repeated string			irmap_scan_paths = 36;
+	repeated string			external	= 37;
+	optional uint32			empty_ns	= 38;
+	repeated join_namespace		join_ns		= 39;
+
+	optional string			cgroup_props		= 41;
+	optional string			cgroup_props_file	= 42;
+	repeated string			cgroup_dump_controller	= 43;
+
+	optional string			freeze_cgroup		= 44;
+	optional uint32			timeout			= 45;
+	optional bool			tcp_skip_in_flight	= 46;
+	optional bool			weak_sysctls		= 47;
+	optional bool			lazy_pages		= 48;
+	optional int32			status_fd		= 49;
+	optional bool			orphan_pts_master	= 50;
+}
+
+message criu_dump_resp {
+	optional bool restored		= 1;
+}
+
+message criu_restore_resp {
+	required int32 pid		= 1;
+}
+
+message criu_notify {
+	optional string script		= 1;
+	optional int32	pid		= 2;
+}
+
+enum criu_req_type {
+	EMPTY		= 0;
+	DUMP		= 1;
+	RESTORE		= 2;
+	CHECK		= 3;
+	PRE_DUMP	= 4;
+	PAGE_SERVER	= 5;
+
+	NOTIFY		= 6;
+
+	CPUINFO_DUMP	= 7;
+	CPUINFO_CHECK	= 8;
+
+	FEATURE_CHECK	= 9;
+}
+
+/*
+ * List of features which can queried via
+ * CRIU_REQ_TYPE__FEATURE_CHECK
+ */
+message criu_features {
+	optional bool			mem_track	= 1;
+}
+
+/*
+ * Request -- each type corresponds to must-be-there
+ * request arguments of respective type
+ */
+
+message criu_req {
+	required criu_req_type		type		= 1;
+
+	optional criu_opts		opts		= 2;
+	optional bool			notify_success	= 3;
+
+	/*
+	 * When set service won't close the connection but
+	 * will wait for more req-s to appear. Works not
+	 * for all request types.
+	 */
+	optional bool			keep_open	= 4;
+	/*
+	 * 'features' can be used to query which features
+	 * are supported by the installed criu/kernel
+	 * via RPC.
+	 */
+	optional criu_features		features	= 5;
+}
+
+/*
+ * Response -- it states whether the request was served
+ * and additional request-specific information
+ */
+
+message criu_resp {
+	required criu_req_type		type		= 1;
+	required bool			success		= 2;
+
+	optional criu_dump_resp		dump		= 3;
+	optional criu_restore_resp	restore		= 4;
+	optional criu_notify		notify		= 5;
+	optional criu_page_server_info	ps		= 6;
+
+	optional int32			cr_errno	= 7;
+	optional criu_features		features	= 8;
+	optional string			cr_errmsg	= 9;
+}

vendor/github.com/opencontainers/runc/libcontainer/error.go

@@ -0,0 +1,70 @@
+package libcontainer
+
+import "io"
+
+// ErrorCode is the API error code type.
+type ErrorCode int
+
+// API error codes.
+const (
+	// Factory errors
+	IdInUse ErrorCode = iota
+	InvalidIdFormat
+
+	// Container errors
+	ContainerNotExists
+	ContainerPaused
+	ContainerNotStopped
+	ContainerNotRunning
+	ContainerNotPaused
+
+	// Process errors
+	NoProcessOps
+
+	// Common errors
+	ConfigInvalid
+	ConsoleExists
+	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"
+	case ConsoleExists:
+		return "Console exists for process"
+	case ContainerNotPaused:
+		return "Container is not paused"
+	case NoProcessOps:
+		return "No process operations"
+	default:
+		return "Unknown error"
+	}
+}
+
+// Error is the API error type.
+type Error interface {
+	error
+
+	// Returns an error if it failed to write the detail of the Error to w.
+	// The detail of the Error may include the error message and a
+	// representation of the stack trace.
+	Detail(w io.Writer) error
+
+	// Returns the error code for this error.
+	Code() ErrorCode
+}

vendor/github.com/opencontainers/runc/libcontainer/factory.go

@@ -0,0 +1,44 @@
+package libcontainer
+
+import (
+	"github.com/opencontainers/runc/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.
+	//
+	// The id must not already be in use by an existing container. Containers created using
+	// a factory with the same path (and filesystem) must have distinct ids.
+	//
+	// Returns the new container with a running process.
+	//
+	// errors:
+	// IdInUse - id is already in use by a container
+	// InvalidIdFormat - id has incorrect format
+	// ConfigInvalid - config is invalid
+	// Systemerror - System error
+	//
+	// On error, any partially created container parts are cleaned up (the operation is atomic).
+	Create(id string, config *configs.Config) (Container, error)
+
+	// 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:
+	// Path does not exist
+	// System error
+	Load(id string) (Container, error)
+
+	// StartInitialization is an internal API to libcontainer used during the reexec of the
+	// container.
+	//
+	// Errors:
+	// Pipe connection error
+	// System error
+	StartInitialization() error
+
+	// Type returns info string about factory type (e.g. lxc, libcontainer...)
+	Type() string
+}

vendor/github.com/opencontainers/runc/libcontainer/factory_linux.go

@@ -0,0 +1,325 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"regexp"
+	"runtime/debug"
+	"strconv"
+
+	"github.com/docker/docker/pkg/mount"
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/cgroups/fs"
+	"github.com/opencontainers/runc/libcontainer/cgroups/rootless"
+	"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/configs/validate"
+	"github.com/opencontainers/runc/libcontainer/utils"
+
+	"golang.org/x/sys/unix"
+)
+
+const (
+	stateFilename    = "state.json"
+	execFifoFilename = "exec.fifo"
+)
+
+var idRegex = regexp.MustCompile(`^[\w+-\.]+$`)
+
+// InitArgs returns an options func to configure a LinuxFactory with the
+// provided init binary path and arguments.
+func InitArgs(args ...string) func(*LinuxFactory) error {
+	return func(l *LinuxFactory) (err error) {
+		if len(args) > 0 {
+			// Resolve relative paths to ensure that its available
+			// after directory changes.
+			if args[0], err = filepath.Abs(args[0]); err != nil {
+				return newGenericError(err, ConfigInvalid)
+			}
+		}
+
+		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
+}
+
+// RootlessCgroups is an options func to configure a LinuxFactory to
+// return containers that use the "rootless" cgroup manager, which will
+// fail to do any operations not possible to do with an unprivileged user.
+// It should only be used in conjunction with rootless containers.
+func RootlessCgroups(l *LinuxFactory) error {
+	l.NewCgroupsManager = func(config *configs.Cgroup, paths map[string]string) cgroups.Manager {
+		return &rootless.Manager{
+			Cgroups: config,
+			Paths:   paths,
+		}
+	}
+	return nil
+}
+
+// TmpfsRoot is an option func to mount LinuxFactory.Root to tmpfs.
+func TmpfsRoot(l *LinuxFactory) error {
+	mounted, err := mount.Mounted(l.Root)
+	if err != nil {
+		return err
+	}
+	if !mounted {
+		if err := unix.Mount("tmpfs", l.Root, "tmpfs", 0, ""); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// CriuPath returns an option func to configure a LinuxFactory with the
+// provided criupath
+func CriuPath(criupath string) func(*LinuxFactory) error {
+	return func(l *LinuxFactory) error {
+		l.CriuPath = criupath
+		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,
+		InitArgs:  []string{"/proc/self/exe", "init"},
+		Validator: validate.New(),
+		CriuPath:  "criu",
+	}
+	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
+
+	// InitArgs are arguments for calling the init responsibilities for spawning
+	// a container.
+	InitArgs []string
+
+	// CriuPath is the path to the criu binary used for checkpoint and restore of
+	// containers.
+	CriuPath 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, 0711); err != nil {
+		return nil, newGenericError(err, SystemError)
+	}
+	if err := os.Chown(containerRoot, unix.Geteuid(), unix.Getegid()); err != nil {
+		return nil, newGenericError(err, SystemError)
+	}
+	if config.Rootless {
+		RootlessCgroups(l)
+	}
+	c := &linuxContainer{
+		id:            id,
+		root:          containerRoot,
+		config:        config,
+		initArgs:      l.InitArgs,
+		criuPath:      l.CriuPath,
+		cgroupManager: l.NewCgroupsManager(config.Cgroups, nil),
+	}
+	c.state = &stoppedState{c: c}
+	return c, 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, id)
+	if err != nil {
+		return nil, err
+	}
+	r := &nonChildProcess{
+		processPid:       state.InitProcessPid,
+		processStartTime: state.InitProcessStartTime,
+		fds:              state.ExternalDescriptors,
+	}
+	// We have to use the RootlessManager.
+	if state.Rootless {
+		RootlessCgroups(l)
+	}
+	c := &linuxContainer{
+		initProcess:          r,
+		initProcessStartTime: state.InitProcessStartTime,
+		id:                   id,
+		config:               &state.Config,
+		initArgs:             l.InitArgs,
+		criuPath:             l.CriuPath,
+		cgroupManager:        l.NewCgroupsManager(state.Config.Cgroups, state.CgroupPaths),
+		root:                 containerRoot,
+		created:              state.Created,
+	}
+	c.state = &loadedState{c: c}
+	if err := c.refreshState(); err != nil {
+		return nil, err
+	}
+	return c, nil
+}
+
+func (l *LinuxFactory) Type() string {
+	return "libcontainer"
+}
+
+// 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() (err error) {
+	var (
+		pipefd, rootfd int
+		consoleSocket  *os.File
+		envInitPipe    = os.Getenv("_LIBCONTAINER_INITPIPE")
+		envStateDir    = os.Getenv("_LIBCONTAINER_STATEDIR")
+		envConsole     = os.Getenv("_LIBCONTAINER_CONSOLE")
+	)
+
+	// Get the INITPIPE.
+	pipefd, err = strconv.Atoi(envInitPipe)
+	if err != nil {
+		return fmt.Errorf("unable to convert _LIBCONTAINER_INITPIPE=%s to int: %s", envInitPipe, err)
+	}
+
+	var (
+		pipe = os.NewFile(uintptr(pipefd), "pipe")
+		it   = initType(os.Getenv("_LIBCONTAINER_INITTYPE"))
+	)
+	defer pipe.Close()
+
+	// Only init processes have STATEDIR.
+	rootfd = -1
+	if it == initStandard {
+		if rootfd, err = strconv.Atoi(envStateDir); err != nil {
+			return fmt.Errorf("unable to convert _LIBCONTAINER_STATEDIR=%s to int: %s", envStateDir, err)
+		}
+	}
+
+	if envConsole != "" {
+		console, err := strconv.Atoi(envConsole)
+		if err != nil {
+			return fmt.Errorf("unable to convert _LIBCONTAINER_CONSOLE=%s to int: %s", envConsole, err)
+		}
+		consoleSocket = os.NewFile(uintptr(console), "console-socket")
+		defer consoleSocket.Close()
+	}
+
+	// clear the current process's environment to clean any libcontainer
+	// specific env vars.
+	os.Clearenv()
+
+	defer func() {
+		// We have an error during the initialization of the container's init,
+		// send it back to the parent process in the form of an initError.
+		if werr := utils.WriteJSON(pipe, syncT{procError}); werr != nil {
+			fmt.Fprintln(os.Stderr, err)
+			return
+		}
+		if werr := utils.WriteJSON(pipe, newSystemError(err)); werr != nil {
+			fmt.Fprintln(os.Stderr, err)
+			return
+		}
+	}()
+	defer func() {
+		if e := recover(); e != nil {
+			err = fmt.Errorf("panic from initialization: %v, %v", e, string(debug.Stack()))
+		}
+	}()
+
+	i, err := newContainerInit(it, pipe, consoleSocket, rootfd)
+	if err != nil {
+		return err
+	}
+
+	// If Init succeeds, syscall.Exec will not return, hence none of the defers will be called.
+	return i.Init()
+}
+
+func (l *LinuxFactory) loadState(root, id string) (*State, error) {
+	f, err := os.Open(filepath.Join(root, stateFilename))
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil, newGenericError(fmt.Errorf("container %q does not exist", id), 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)
+	}
+
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/generic_error.go

@@ -0,0 +1,92 @@
+package libcontainer
+
+import (
+	"fmt"
+	"io"
+	"text/template"
+	"time"
+
+	"github.com/opencontainers/runc/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 {
+	return createSystemError(err, "")
+}
+
+func newSystemErrorWithCausef(err error, cause string, v ...interface{}) Error {
+	return createSystemError(err, fmt.Sprintf(cause, v...))
+}
+
+func newSystemErrorWithCause(err error, cause string) Error {
+	return createSystemError(err, cause)
+}
+
+// createSystemError creates the specified error with the correct number of
+// stack frames skipped. This is only to be called by the other functions for
+// formatting the error.
+func createSystemError(err error, cause string) Error {
+	gerr := &genericError{
+		Timestamp: time.Now(),
+		Err:       err,
+		ECode:     SystemError,
+		Cause:     cause,
+		Stack:     stacktrace.Capture(2),
+	}
+	if err != nil {
+		gerr.Message = err.Error()
+	}
+	return gerr
+}
+
+type genericError struct {
+	Timestamp time.Time
+	ECode     ErrorCode
+	Err       error `json:"-"`
+	Cause     string
+	Message   string
+	Stack     stacktrace.Stacktrace
+}
+
+func (e *genericError) Error() string {
+	if e.Cause == "" {
+		return e.Message
+	}
+	frame := e.Stack.Frames[0]
+	return fmt.Sprintf("%s:%d: %s caused %q", frame.File, frame.Line, e.Cause, e.Message)
+}
+
+func (e *genericError) Code() ErrorCode {
+	return e.ECode
+}
+
+func (e *genericError) Detail(w io.Writer) error {
+	return errorTemplate.Execute(w, e)
+}

vendor/github.com/opencontainers/runc/libcontainer/init_linux.go

@@ -0,0 +1,502 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"io"
+	"net"
+	"os"
+	"strings"
+	"syscall" // only for Errno
+	"unsafe"
+
+	"github.com/Sirupsen/logrus"
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/system"
+	"github.com/opencontainers/runc/libcontainer/user"
+	"github.com/opencontainers/runc/libcontainer/utils"
+	"github.com/vishvananda/netlink"
+
+	"golang.org/x/sys/unix"
+)
+
+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 temporary 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"`
+	Capabilities     *configs.Capabilities `json:"capabilities"`
+	ProcessLabel     string                `json:"process_label"`
+	AppArmorProfile  string                `json:"apparmor_profile"`
+	NoNewPrivileges  bool                  `json:"no_new_privileges"`
+	User             string                `json:"user"`
+	AdditionalGroups []string              `json:"additional_groups"`
+	Config           *configs.Config       `json:"config"`
+	Networks         []*network            `json:"network"`
+	PassedFilesCount int                   `json:"passed_files_count"`
+	ContainerId      string                `json:"containerid"`
+	Rlimits          []configs.Rlimit      `json:"rlimits"`
+	CreateConsole    bool                  `json:"create_console"`
+	Rootless         bool                  `json:"rootless"`
+}
+
+type initer interface {
+	Init() error
+}
+
+func newContainerInit(t initType, pipe *os.File, consoleSocket *os.File, stateDirFD int) (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{
+			pipe:          pipe,
+			consoleSocket: consoleSocket,
+			config:        config,
+		}, nil
+	case initStandard:
+		return &linuxStandardInit{
+			pipe:          pipe,
+			consoleSocket: consoleSocket,
+			parentPid:     unix.Getppid(),
+			config:        config,
+			stateDirFD:    stateDirFD,
+		}, 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 executing the command inside the namespace
+func finalizeNamespace(config *initConfig) error {
+	// Ensure that all unwanted fds we may have accidentally
+	// inherited are marked close-on-exec so they stay out of the
+	// container
+	if err := utils.CloseExecFrom(config.PassedFilesCount + 3); err != nil {
+		return err
+	}
+
+	capabilities := &configs.Capabilities{}
+	if config.Capabilities != nil {
+		capabilities = config.Capabilities
+	} else if config.Config.Capabilities != nil {
+		capabilities = config.Config.Capabilities
+	}
+	w, err := newContainerCapList(capabilities)
+	if err != nil {
+		return err
+	}
+	// drop capabilities in bounding set before changing user
+	if err := w.ApplyBoundingSet(); 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
+	}
+	if err := w.ApplyCaps(); err != nil {
+		return err
+	}
+	if config.Cwd != "" {
+		if err := unix.Chdir(config.Cwd); err != nil {
+			return fmt.Errorf("chdir to cwd (%q) set in config.json failed: %v", config.Cwd, err)
+		}
+	}
+	return nil
+}
+
+// setupConsole sets up the console from inside the container, and sends the
+// master pty fd to the config.Pipe (using cmsg). This is done to ensure that
+// consoles are scoped to a container properly (see runc#814 and the many
+// issues related to that). This has to be run *after* we've pivoted to the new
+// rootfs (and the users' configuration is entirely set up).
+func setupConsole(socket *os.File, config *initConfig, mount bool) error {
+	defer socket.Close()
+	// At this point, /dev/ptmx points to something that we would expect. We
+	// used to change the owner of the slave path, but since the /dev/pts mount
+	// can have gid=X set (at the users' option). So touching the owner of the
+	// slave PTY is not necessary, as the kernel will handle that for us. Note
+	// however, that setupUser (specifically fixStdioPermissions) *will* change
+	// the UID owner of the console to be the user the process will run as (so
+	// they can actually control their console).
+	console, err := newConsole()
+	if err != nil {
+		return err
+	}
+	// After we return from here, we don't need the console anymore.
+	defer console.Close()
+
+	linuxConsole, ok := console.(*linuxConsole)
+	if !ok {
+		return fmt.Errorf("failed to cast console to *linuxConsole")
+	}
+	// Mount the console inside our rootfs.
+	if mount {
+		if err := linuxConsole.mount(); err != nil {
+			return err
+		}
+	}
+	// While we can access console.master, using the API is a good idea.
+	if err := utils.SendFd(socket, linuxConsole.File()); err != nil {
+		return err
+	}
+	// Now, dup over all the things.
+	return linuxConsole.dupStdio()
+}
+
+// syncParentReady sends to the given pipe a JSON payload which indicates that
+// the init is ready to Exec the child process. It then waits for the parent to
+// indicate that it is cleared to Exec.
+func syncParentReady(pipe io.ReadWriter) error {
+	// Tell parent.
+	if err := writeSync(pipe, procReady); err != nil {
+		return err
+	}
+
+	// Wait for parent to give the all-clear.
+	if err := readSync(pipe, procRun); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// syncParentHooks sends to the given pipe a JSON payload which indicates that
+// the parent should execute pre-start hooks. It then waits for the parent to
+// indicate that it is cleared to resume.
+func syncParentHooks(pipe io.ReadWriter) error {
+	// Tell parent.
+	if err := writeSync(pipe, procHooks); err != nil {
+		return err
+	}
+
+	// Wait for parent to give the all-clear.
+	if err := readSync(pipe, procResume); 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:  0,
+		Gid:  0,
+		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
+	}
+
+	var addGroups []int
+	if len(config.AdditionalGroups) > 0 {
+		addGroups, err = user.GetAdditionalGroupsPath(config.AdditionalGroups, groupPath)
+		if err != nil {
+			return err
+		}
+	}
+
+	if config.Rootless {
+		if execUser.Uid != 0 {
+			return fmt.Errorf("cannot run as a non-root user in a rootless container")
+		}
+
+		if execUser.Gid != 0 {
+			return fmt.Errorf("cannot run as a non-root group in a rootless container")
+		}
+
+		// We cannot set any additional groups in a rootless container and thus we
+		// bail if the user asked us to do so. TODO: We currently can't do this
+		// earlier, but if libcontainer.Process.User was typesafe this might work.
+		if len(addGroups) > 0 {
+			return fmt.Errorf("cannot set any additional groups in a rootless container")
+		}
+	}
+
+	// before we change to the container's user make sure that the processes STDIO
+	// is correctly owned by the user that we are switching to.
+	if err := fixStdioPermissions(config, execUser); err != nil {
+		return err
+	}
+
+	// This isn't allowed in an unprivileged user namespace since Linux 3.19.
+	// There's nothing we can do about /etc/group entries, so we silently
+	// ignore setting groups here (since the user didn't explicitly ask us to
+	// set the group).
+	if !config.Rootless {
+		suppGroups := append(execUser.Sgids, addGroups...)
+		if err := unix.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
+}
+
+// fixStdioPermissions fixes the permissions of PID 1's STDIO within the container to the specified user.
+// The ownership needs to match because it is created outside of the container and needs to be
+// localized.
+func fixStdioPermissions(config *initConfig, u *user.ExecUser) error {
+	var null unix.Stat_t
+	if err := unix.Stat("/dev/null", &null); err != nil {
+		return err
+	}
+	for _, fd := range []uintptr{
+		os.Stdin.Fd(),
+		os.Stderr.Fd(),
+		os.Stdout.Fd(),
+	} {
+		var s unix.Stat_t
+		if err := unix.Fstat(int(fd), &s); err != nil {
+			return err
+		}
+
+		// Skip chown of /dev/null if it was used as one of the STDIO fds.
+		if s.Rdev == null.Rdev {
+			continue
+		}
+
+		// Skip chown if s.Gid is actually an unmapped gid in the host. While
+		// this is a bit dodgy if it just so happens that the console _is_
+		// owned by overflow_gid, there's no way for us to disambiguate this as
+		// a userspace program.
+		if _, err := config.Config.HostGID(int(s.Gid)); err != nil {
+			continue
+		}
+
+		// We only change the uid owner (as it is possible for the mount to
+		// prefer a different gid, and there's no reason for us to change it).
+		// The reason why we don't just leave the default uid=X mount setup is
+		// that users expect to be able to actually use their console. Without
+		// this code, you couldn't effectively run as a non-root user inside a
+		// container and also have a console set up.
+		if err := unix.Fchown(int(fd), u.Uid, int(s.Gid)); 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 {
+		_, dst, err := net.ParseCIDR(config.Destination)
+		if err != nil {
+			return err
+		}
+		src := net.ParseIP(config.Source)
+		if src == nil {
+			return fmt.Errorf("Invalid source for route: %s", config.Source)
+		}
+		gw := net.ParseIP(config.Gateway)
+		if gw == nil {
+			return fmt.Errorf("Invalid gateway for route: %s", config.Gateway)
+		}
+		l, err := netlink.LinkByName(config.InterfaceName)
+		if err != nil {
+			return err
+		}
+		route := &netlink.Route{
+			Scope:     netlink.SCOPE_UNIVERSE,
+			Dst:       dst,
+			Src:       src,
+			Gw:        gw,
+			LinkIndex: l.Attrs().Index,
+		}
+		if err := netlink.RouteAdd(route); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func setupRlimits(limits []configs.Rlimit, pid int) error {
+	for _, rlimit := range limits {
+		if err := system.Prlimit(pid, rlimit.Type, unix.Rlimit{Max: rlimit.Hard, Cur: rlimit.Soft}); err != nil {
+			return fmt.Errorf("error setting rlimit type %v: %v", rlimit.Type, err)
+		}
+	}
+	return nil
+}
+
+const _P_PID = 1
+
+type siginfo struct {
+	si_signo int32
+	si_errno int32
+	si_code  int32
+	// below here is a union; si_pid is the only field we use
+	si_pid int32
+	// Pad to 128 bytes as detailed in blockUntilWaitable
+	pad [96]byte
+}
+
+// isWaitable returns true if the process has exited false otherwise.
+// Its based off blockUntilWaitable in src/os/wait_waitid.go
+func isWaitable(pid int) (bool, error) {
+	si := &siginfo{}
+	_, _, e := unix.Syscall6(unix.SYS_WAITID, _P_PID, uintptr(pid), uintptr(unsafe.Pointer(si)), unix.WEXITED|unix.WNOWAIT|unix.WNOHANG, 0, 0)
+	if e != 0 {
+		return false, os.NewSyscallError("waitid", e)
+	}
+
+	return si.si_pid != 0, nil
+}
+
+// isNoChildren returns true if err represents a unix.ECHILD (formerly syscall.ECHILD) false otherwise
+func isNoChildren(err error) bool {
+	switch err := err.(type) {
+	case syscall.Errno:
+		if err == unix.ECHILD {
+			return true
+		}
+	case *os.SyscallError:
+		if err.Err == unix.ECHILD {
+			return true
+		}
+	}
+	return false
+}
+
+// signalAllProcesses freezes then iterates over all the processes inside the
+// manager's cgroups sending the signal s to them.
+// If s is SIGKILL then it will wait for each process to exit.
+// For all other signals it will check if the process is ready to report its
+// exit status and only if it is will a wait be performed.
+func signalAllProcesses(m cgroups.Manager, s os.Signal) error {
+	var procs []*os.Process
+	if err := m.Freeze(configs.Frozen); err != nil {
+		logrus.Warn(err)
+	}
+	pids, err := m.GetAllPids()
+	if err != nil {
+		m.Freeze(configs.Thawed)
+		return err
+	}
+	for _, pid := range pids {
+		p, err := os.FindProcess(pid)
+		if err != nil {
+			logrus.Warn(err)
+			continue
+		}
+		procs = append(procs, p)
+		if err := p.Signal(s); err != nil {
+			logrus.Warn(err)
+		}
+	}
+	if err := m.Freeze(configs.Thawed); err != nil {
+		logrus.Warn(err)
+	}
+
+	for _, p := range procs {
+		if s != unix.SIGKILL {
+			if ok, err := isWaitable(p.Pid); err != nil {
+				if !isNoChildren(err) {
+					logrus.Warn("signalAllProcesses: ", p.Pid, err)
+				}
+				continue
+			} else if !ok {
+				// Not ready to report so don't wait
+				continue
+			}
+		}
+
+		if _, err := p.Wait(); err != nil {
+			if !isNoChildren(err) {
+				logrus.Warn("wait: ", err)
+			}
+		}
+	}
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/keys/keyctl.go

@@ -0,0 +1,50 @@
+// +build linux
+
+package keys
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	"golang.org/x/sys/unix"
+)
+
+type KeySerial uint32
+
+func JoinSessionKeyring(name string) (KeySerial, error) {
+	sessKeyId, err := unix.KeyctlJoinSessionKeyring(name)
+	if err != nil {
+		return 0, fmt.Errorf("could not create session key: %v", err)
+	}
+	return KeySerial(sessKeyId), nil
+}
+
+// ModKeyringPerm modifies permissions on a keyring by reading the current permissions,
+// anding the bits with the given mask (clearing permissions) and setting
+// additional permission bits
+func ModKeyringPerm(ringId KeySerial, mask, setbits uint32) error {
+	dest, err := unix.KeyctlString(unix.KEYCTL_DESCRIBE, int(ringId))
+	if err != nil {
+		return err
+	}
+
+	res := strings.Split(string(dest), ";")
+	if len(res) < 5 {
+		return fmt.Errorf("Destination buffer for key description is too small")
+	}
+
+	// parse permissions
+	perm64, err := strconv.ParseUint(res[3], 16, 32)
+	if err != nil {
+		return err
+	}
+
+	perm := (uint32(perm64) & mask) | setbits
+
+	if err := unix.KeyctlSetperm(int(ringId), perm); err != nil {
+		return err
+	}
+
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/message_linux.go

@@ -0,0 +1,87 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"github.com/vishvananda/netlink/nl"
+	"golang.org/x/sys/unix"
+)
+
+// list of known message types we want to send to bootstrap program
+// The number is randomly chosen to not conflict with known netlink types
+const (
+	InitMsg         uint16 = 62000
+	CloneFlagsAttr  uint16 = 27281
+	NsPathsAttr     uint16 = 27282
+	UidmapAttr      uint16 = 27283
+	GidmapAttr      uint16 = 27284
+	SetgroupAttr    uint16 = 27285
+	OomScoreAdjAttr uint16 = 27286
+	RootlessAttr    uint16 = 27287
+)
+
+type Int32msg struct {
+	Type  uint16
+	Value uint32
+}
+
+// Serialize serializes the message.
+// Int32msg has the following representation
+// | nlattr len | nlattr type |
+// | uint32 value             |
+func (msg *Int32msg) Serialize() []byte {
+	buf := make([]byte, msg.Len())
+	native := nl.NativeEndian()
+	native.PutUint16(buf[0:2], uint16(msg.Len()))
+	native.PutUint16(buf[2:4], msg.Type)
+	native.PutUint32(buf[4:8], msg.Value)
+	return buf
+}
+
+func (msg *Int32msg) Len() int {
+	return unix.NLA_HDRLEN + 4
+}
+
+// Bytemsg has the following representation
+// | nlattr len | nlattr type |
+// | value              | pad |
+type Bytemsg struct {
+	Type  uint16
+	Value []byte
+}
+
+func (msg *Bytemsg) Serialize() []byte {
+	l := msg.Len()
+	buf := make([]byte, (l+unix.NLA_ALIGNTO-1) & ^(unix.NLA_ALIGNTO-1))
+	native := nl.NativeEndian()
+	native.PutUint16(buf[0:2], uint16(l))
+	native.PutUint16(buf[2:4], msg.Type)
+	copy(buf[4:], msg.Value)
+	return buf
+}
+
+func (msg *Bytemsg) Len() int {
+	return unix.NLA_HDRLEN + len(msg.Value) + 1 // null-terminated
+}
+
+type Boolmsg struct {
+	Type  uint16
+	Value bool
+}
+
+func (msg *Boolmsg) Serialize() []byte {
+	buf := make([]byte, msg.Len())
+	native := nl.NativeEndian()
+	native.PutUint16(buf[0:2], uint16(msg.Len()))
+	native.PutUint16(buf[2:4], msg.Type)
+	if msg.Value {
+		buf[4] = 1
+	} else {
+		buf[4] = 0
+	}
+	return buf
+}
+
+func (msg *Boolmsg) Len() int {
+	return unix.NLA_HDRLEN + 1
+}

vendor/github.com/opencontainers/runc/libcontainer/network_linux.go

@@ -0,0 +1,259 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"io/ioutil"
+	"net"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/utils"
+	"github.com/vishvananda/netlink"
+)
+
+var strategies = map[string]networkStrategy{
+	"veth":     &veth{},
+	"loopback": &loopback{},
+}
+
+// networkStrategy represents a specific network configuration for
+// a container's networking stack
+type networkStrategy interface {
+	create(*network, int) error
+	initialize(*network) error
+	detach(*configs.Network) error
+	attach(*configs.Network) error
+}
+
+// getStrategy returns the specific network strategy for the
+// provided type.
+func getStrategy(tpe string) (networkStrategy, error) {
+	s, exists := strategies[tpe]
+	if !exists {
+		return nil, fmt.Errorf("unknown strategy type %q", tpe)
+	}
+	return s, nil
+}
+
+// Returns the network statistics for the network interfaces represented by the NetworkRuntimeInfo.
+func getNetworkInterfaceStats(interfaceName string) (*NetworkInterface, error) {
+	out := &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)
+}
+
+// loopback is a network strategy that provides a basic loopback device
+type loopback struct {
+}
+
+func (l *loopback) create(n *network, nspid int) error {
+	return nil
+}
+
+func (l *loopback) initialize(config *network) error {
+	return netlink.LinkSetUp(&netlink.Device{LinkAttrs: netlink.LinkAttrs{Name: "lo"}})
+}
+
+func (l *loopback) attach(n *configs.Network) (err error) {
+	return nil
+}
+
+func (l *loopback) detach(n *configs.Network) (err error) {
+	return nil
+}
+
+// veth is a network strategy that uses a bridge and creates
+// a veth pair, one that is attached to the bridge on the host and the other
+// is placed inside the container's namespace
+type veth struct {
+}
+
+func (v *veth) detach(n *configs.Network) (err error) {
+	return netlink.LinkSetMaster(&netlink.Device{LinkAttrs: netlink.LinkAttrs{Name: n.HostInterfaceName}}, nil)
+}
+
+// attach a container network interface to an external network
+func (v *veth) attach(n *configs.Network) (err error) {
+	brl, err := netlink.LinkByName(n.Bridge)
+	if err != nil {
+		return err
+	}
+	br, ok := brl.(*netlink.Bridge)
+	if !ok {
+		return fmt.Errorf("Wrong device type %T", brl)
+	}
+	host, err := netlink.LinkByName(n.HostInterfaceName)
+	if err != nil {
+		return err
+	}
+
+	if err := netlink.LinkSetMaster(host, br); err != nil {
+		return err
+	}
+	if err := netlink.LinkSetMTU(host, n.Mtu); err != nil {
+		return err
+	}
+	if n.HairpinMode {
+		if err := netlink.LinkSetHairpin(host, true); err != nil {
+			return err
+		}
+	}
+	if err := netlink.LinkSetUp(host); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (v *veth) create(n *network, nspid int) (err error) {
+	tmpName, err := v.generateTempPeerName()
+	if err != nil {
+		return err
+	}
+	n.TempVethPeerName = tmpName
+	if n.Bridge == "" {
+		return fmt.Errorf("bridge is not specified")
+	}
+	veth := &netlink.Veth{
+		LinkAttrs: netlink.LinkAttrs{
+			Name:   n.HostInterfaceName,
+			TxQLen: n.TxQueueLen,
+		},
+		PeerName: n.TempVethPeerName,
+	}
+	if err := netlink.LinkAdd(veth); err != nil {
+		return err
+	}
+	defer func() {
+		if err != nil {
+			netlink.LinkDel(veth)
+		}
+	}()
+	if err := v.attach(&n.Network); err != nil {
+		return err
+	}
+	child, err := netlink.LinkByName(n.TempVethPeerName)
+	if err != nil {
+		return err
+	}
+	return netlink.LinkSetNsPid(child, nspid)
+}
+
+func (v *veth) generateTempPeerName() (string, error) {
+	return utils.GenerateRandomName("veth", 7)
+}
+
+func (v *veth) initialize(config *network) error {
+	peer := config.TempVethPeerName
+	if peer == "" {
+		return fmt.Errorf("peer is not specified")
+	}
+	child, err := netlink.LinkByName(peer)
+	if err != nil {
+		return err
+	}
+	if err := netlink.LinkSetDown(child); err != nil {
+		return err
+	}
+	if err := netlink.LinkSetName(child, config.Name); err != nil {
+		return err
+	}
+	// get the interface again after we changed the name as the index also changes.
+	if child, err = netlink.LinkByName(config.Name); err != nil {
+		return err
+	}
+	if config.MacAddress != "" {
+		mac, err := net.ParseMAC(config.MacAddress)
+		if err != nil {
+			return err
+		}
+		if err := netlink.LinkSetHardwareAddr(child, mac); err != nil {
+			return err
+		}
+	}
+	ip, err := netlink.ParseAddr(config.Address)
+	if err != nil {
+		return err
+	}
+	if err := netlink.AddrAdd(child, ip); err != nil {
+		return err
+	}
+	if config.IPv6Address != "" {
+		ip6, err := netlink.ParseAddr(config.IPv6Address)
+		if err != nil {
+			return err
+		}
+		if err := netlink.AddrAdd(child, ip6); err != nil {
+			return err
+		}
+	}
+	if err := netlink.LinkSetMTU(child, config.Mtu); err != nil {
+		return err
+	}
+	if err := netlink.LinkSetUp(child); err != nil {
+		return err
+	}
+	if config.Gateway != "" {
+		gw := net.ParseIP(config.Gateway)
+		if err := netlink.RouteAdd(&netlink.Route{
+			Scope:     netlink.SCOPE_UNIVERSE,
+			LinkIndex: child.Attrs().Index,
+			Gw:        gw,
+		}); err != nil {
+			return err
+		}
+	}
+	if config.IPv6Gateway != "" {
+		gw := net.ParseIP(config.IPv6Gateway)
+		if err := netlink.RouteAdd(&netlink.Route{
+			Scope:     netlink.SCOPE_UNIVERSE,
+			LinkIndex: child.Attrs().Index,
+			Gw:        gw,
+		}); err != nil {
+			return err
+		}
+	}
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/notify_linux.go

@@ -0,0 +1,90 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"io/ioutil"
+	"os"
+	"path/filepath"
+
+	"golang.org/x/sys/unix"
+)
+
+const oomCgroupName = "memory"
+
+type PressureLevel uint
+
+const (
+	LowPressure PressureLevel = iota
+	MediumPressure
+	CriticalPressure
+)
+
+func registerMemoryEvent(cgDir string, evName string, arg string) (<-chan struct{}, error) {
+	evFile, err := os.Open(filepath.Join(cgDir, evName))
+	if err != nil {
+		return nil, err
+	}
+	fd, err := unix.Eventfd(0, unix.EFD_CLOEXEC)
+	if err != nil {
+		evFile.Close()
+		return nil, err
+	}
+
+	eventfd := os.NewFile(uintptr(fd), "eventfd")
+
+	eventControlPath := filepath.Join(cgDir, "cgroup.event_control")
+	data := fmt.Sprintf("%d %d %s", eventfd.Fd(), evFile.Fd(), arg)
+	if err := ioutil.WriteFile(eventControlPath, []byte(data), 0700); err != nil {
+		eventfd.Close()
+		evFile.Close()
+		return nil, err
+	}
+	ch := make(chan struct{})
+	go func() {
+		defer func() {
+			close(ch)
+			eventfd.Close()
+			evFile.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
+}
+
+// notifyOnOOM returns channel on which you can expect event about OOM,
+// if process died without OOM this channel will be closed.
+func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
+	dir := paths[oomCgroupName]
+	if dir == "" {
+		return nil, fmt.Errorf("path %q missing", oomCgroupName)
+	}
+
+	return registerMemoryEvent(dir, "memory.oom_control", "")
+}
+
+func notifyMemoryPressure(paths map[string]string, level PressureLevel) (<-chan struct{}, error) {
+	dir := paths[oomCgroupName]
+	if dir == "" {
+		return nil, fmt.Errorf("path %q missing", oomCgroupName)
+	}
+
+	if level > CriticalPressure {
+		return nil, fmt.Errorf("invalid pressure level %d", level)
+	}
+
+	levelStr := []string{"low", "medium", "critical"}[level]
+	return registerMemoryEvent(dir, "memory.pressure_level", levelStr)
+}

vendor/github.com/opencontainers/runc/libcontainer/process.go

@@ -0,0 +1,106 @@
+package libcontainer
+
+import (
+	"fmt"
+	"io"
+	"math"
+	"os"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+type processOperations interface {
+	wait() (*os.ProcessState, error)
+	signal(sig os.Signal) error
+	pid() int
+}
+
+// Process specifies the configuration and IO for a process inside
+// a container.
+type Process struct {
+	// The command to be run followed by any arguments.
+	Args []string
+
+	// Env specifies the environment variables for the process.
+	Env []string
+
+	// User will set the uid and gid of the executing process running inside the container
+	// local to the container's user and group configuration.
+	User string
+
+	// AdditionalGroups specifies the gids that should be added to supplementary groups
+	// in addition to those that the user belongs to.
+	AdditionalGroups []string
+
+	// Cwd will change the processes current working directory inside the container's rootfs.
+	Cwd string
+
+	// Stdin is a pointer to a reader which provides the standard input stream.
+	Stdin io.Reader
+
+	// Stdout is a pointer to a writer which receives the standard output stream.
+	Stdout io.Writer
+
+	// Stderr is a pointer to a writer which receives the standard error stream.
+	Stderr io.Writer
+
+	// ExtraFiles specifies additional open files to be inherited by the container
+	ExtraFiles []*os.File
+
+	// Capabilities specify the capabilities to keep when executing the process inside the container
+	// All capabilities not specified will be dropped from the processes capability mask
+	Capabilities *configs.Capabilities
+
+	// AppArmorProfile specifies the profile to apply to the process and is
+	// changed at the time the process is execed
+	AppArmorProfile string
+
+	// Label specifies the label to apply to the process.  It is commonly used by selinux
+	Label string
+
+	// NoNewPrivileges controls whether processes can gain additional privileges.
+	NoNewPrivileges *bool
+
+	// 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 []configs.Rlimit
+
+	// ConsoleSocket provides the masterfd console.
+	ConsoleSocket *os.File
+
+	ops processOperations
+}
+
+// Wait waits for the process to exit.
+// Wait releases any resources associated with the Process
+func (p Process) Wait() (*os.ProcessState, error) {
+	if p.ops == nil {
+		return nil, newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
+	}
+	return p.ops.wait()
+}
+
+// Pid returns the process ID
+func (p Process) Pid() (int, error) {
+	// math.MinInt32 is returned here, because it's invalid value
+	// for the kill() system call.
+	if p.ops == nil {
+		return math.MinInt32, newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
+	}
+	return p.ops.pid(), nil
+}
+
+// Signal sends a signal to the Process.
+func (p Process) Signal(sig os.Signal) error {
+	if p.ops == nil {
+		return newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
+	}
+	return p.ops.signal(sig)
+}
+
+// IO holds the process's STDIO
+type IO struct {
+	Stdin  io.WriteCloser
+	Stdout io.ReadCloser
+	Stderr io.ReadCloser
+}

vendor/github.com/opencontainers/runc/libcontainer/process_linux.go

@@ -0,0 +1,493 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strconv"
+	"syscall" // only for Signal
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/system"
+	"github.com/opencontainers/runc/libcontainer/utils"
+
+	"golang.org/x/sys/unix"
+)
+
+type parentProcess interface {
+	// pid returns the pid for the running process.
+	pid() int
+
+	// start starts the process execution.
+	start() error
+
+	// send a SIGKILL to the process and wait for the exit.
+	terminate() error
+
+	// wait waits on the process returning the process state.
+	wait() (*os.ProcessState, error)
+
+	// startTime returns the process start time.
+	startTime() (uint64, error)
+
+	signal(os.Signal) error
+
+	externalDescriptors() []string
+
+	setExternalDescriptors(fds []string)
+}
+
+type setnsProcess struct {
+	cmd           *exec.Cmd
+	parentPipe    *os.File
+	childPipe     *os.File
+	cgroupPaths   map[string]string
+	config        *initConfig
+	fds           []string
+	process       *Process
+	bootstrapData io.Reader
+}
+
+func (p *setnsProcess) startTime() (uint64, error) {
+	stat, err := system.Stat(p.pid())
+	return stat.StartTime, err
+}
+
+func (p *setnsProcess) signal(sig os.Signal) error {
+	s, ok := sig.(syscall.Signal)
+	if !ok {
+		return errors.New("os: unsupported signal type")
+	}
+	return unix.Kill(p.pid(), s)
+}
+
+func (p *setnsProcess) start() (err error) {
+	defer p.parentPipe.Close()
+	err = p.cmd.Start()
+	p.childPipe.Close()
+	if err != nil {
+		return newSystemErrorWithCause(err, "starting setns process")
+	}
+	if p.bootstrapData != nil {
+		if _, err := io.Copy(p.parentPipe, p.bootstrapData); err != nil {
+			return newSystemErrorWithCause(err, "copying bootstrap data to pipe")
+		}
+	}
+	if err = p.execSetns(); err != nil {
+		return newSystemErrorWithCause(err, "executing setns process")
+	}
+	// We can't join cgroups if we're in a rootless container.
+	if !p.config.Rootless && len(p.cgroupPaths) > 0 {
+		if err := cgroups.EnterPid(p.cgroupPaths, p.pid()); err != nil {
+			return newSystemErrorWithCausef(err, "adding pid %d to cgroups", p.pid())
+		}
+	}
+	// set rlimits, this has to be done here because we lose permissions
+	// to raise the limits once we enter a user-namespace
+	if err := setupRlimits(p.config.Rlimits, p.pid()); err != nil {
+		return newSystemErrorWithCause(err, "setting rlimits for process")
+	}
+	if err := utils.WriteJSON(p.parentPipe, p.config); err != nil {
+		return newSystemErrorWithCause(err, "writing config to pipe")
+	}
+
+	ierr := parseSync(p.parentPipe, func(sync *syncT) error {
+		switch sync.Type {
+		case procReady:
+			// This shouldn't happen.
+			panic("unexpected procReady in setns")
+		case procHooks:
+			// This shouldn't happen.
+			panic("unexpected procHooks in setns")
+		default:
+			return newSystemError(fmt.Errorf("invalid JSON payload from child"))
+		}
+	})
+
+	if err := unix.Shutdown(int(p.parentPipe.Fd()), unix.SHUT_WR); err != nil {
+		return newSystemErrorWithCause(err, "calling shutdown on init pipe")
+	}
+	// Must be done after Shutdown so the child will exit and we can wait for it.
+	if ierr != nil {
+		p.wait()
+		return ierr
+	}
+	return nil
+}
+
+// execSetns runs the process that executes C code to perform the setns calls
+// because setns support requires the C process to fork off a child and perform the setns
+// before the go runtime boots, we wait on the process to die and receive the child's pid
+// over the provided pipe.
+func (p *setnsProcess) execSetns() error {
+	status, err := p.cmd.Process.Wait()
+	if err != nil {
+		p.cmd.Wait()
+		return newSystemErrorWithCause(err, "waiting on setns process to finish")
+	}
+	if !status.Success() {
+		p.cmd.Wait()
+		return newSystemError(&exec.ExitError{ProcessState: status})
+	}
+	var pid *pid
+	if err := json.NewDecoder(p.parentPipe).Decode(&pid); err != nil {
+		p.cmd.Wait()
+		return newSystemErrorWithCause(err, "reading pid from init pipe")
+	}
+	process, err := os.FindProcess(pid.Pid)
+	if err != nil {
+		return err
+	}
+	p.cmd.Process = process
+	p.process.ops = p
+	return nil
+}
+
+// terminate sends a SIGKILL to the forked process for the setns routine then waits to
+// avoid the process becoming a zombie.
+func (p *setnsProcess) terminate() error {
+	if p.cmd.Process == nil {
+		return nil
+	}
+	err := p.cmd.Process.Kill()
+	if _, werr := p.wait(); err == nil {
+		err = werr
+	}
+	return err
+}
+
+func (p *setnsProcess) wait() (*os.ProcessState, error) {
+	err := p.cmd.Wait()
+
+	// Return actual ProcessState even on Wait error
+	return p.cmd.ProcessState, err
+}
+
+func (p *setnsProcess) pid() int {
+	return p.cmd.Process.Pid
+}
+
+func (p *setnsProcess) externalDescriptors() []string {
+	return p.fds
+}
+
+func (p *setnsProcess) setExternalDescriptors(newFds []string) {
+	p.fds = newFds
+}
+
+type initProcess struct {
+	cmd           *exec.Cmd
+	parentPipe    *os.File
+	childPipe     *os.File
+	config        *initConfig
+	manager       cgroups.Manager
+	container     *linuxContainer
+	fds           []string
+	process       *Process
+	bootstrapData io.Reader
+	sharePidns    bool
+	rootDir       *os.File
+}
+
+func (p *initProcess) pid() int {
+	return p.cmd.Process.Pid
+}
+
+func (p *initProcess) externalDescriptors() []string {
+	return p.fds
+}
+
+// execSetns runs the process that executes C code to perform the setns calls
+// because setns support requires the C process to fork off a child and perform the setns
+// before the go runtime boots, we wait on the process to die and receive the child's pid
+// over the provided pipe.
+// This is called by initProcess.start function
+func (p *initProcess) execSetns() error {
+	status, err := p.cmd.Process.Wait()
+	if err != nil {
+		p.cmd.Wait()
+		return err
+	}
+	if !status.Success() {
+		p.cmd.Wait()
+		return &exec.ExitError{ProcessState: status}
+	}
+	var pid *pid
+	if err := json.NewDecoder(p.parentPipe).Decode(&pid); err != nil {
+		p.cmd.Wait()
+		return err
+	}
+	process, err := os.FindProcess(pid.Pid)
+	if err != nil {
+		return err
+	}
+	p.cmd.Process = process
+	p.process.ops = p
+	return nil
+}
+
+func (p *initProcess) start() error {
+	defer p.parentPipe.Close()
+	err := p.cmd.Start()
+	p.process.ops = p
+	p.childPipe.Close()
+	p.rootDir.Close()
+	if err != nil {
+		p.process.ops = nil
+		return newSystemErrorWithCause(err, "starting init process command")
+	}
+	if _, err := io.Copy(p.parentPipe, p.bootstrapData); err != nil {
+		return newSystemErrorWithCause(err, "copying bootstrap data to pipe")
+	}
+	if err := p.execSetns(); err != nil {
+		return newSystemErrorWithCause(err, "running exec setns process for init")
+	}
+	// Save the standard descriptor names before the container process
+	// can potentially move them (e.g., via dup2()).  If we don't do this now,
+	// we won't know at checkpoint time which file descriptor to look up.
+	fds, err := getPipeFds(p.pid())
+	if err != nil {
+		return newSystemErrorWithCausef(err, "getting pipe fds for pid %d", p.pid())
+	}
+	p.setExternalDescriptors(fds)
+	// Do this before syncing with child so that no children can escape the
+	// cgroup. We don't need to worry about not doing this and not being root
+	// because we'd be using the rootless cgroup manager in that case.
+	if err := p.manager.Apply(p.pid()); err != nil {
+		return newSystemErrorWithCause(err, "applying cgroup configuration for process")
+	}
+	defer func() {
+		if err != nil {
+			// TODO: should not be the responsibility to call here
+			p.manager.Destroy()
+		}
+	}()
+	if err := p.createNetworkInterfaces(); err != nil {
+		return newSystemErrorWithCause(err, "creating network interfaces")
+	}
+	if err := p.sendConfig(); err != nil {
+		return newSystemErrorWithCause(err, "sending config to init process")
+	}
+	var (
+		sentRun    bool
+		sentResume bool
+	)
+
+	ierr := parseSync(p.parentPipe, func(sync *syncT) error {
+		switch sync.Type {
+		case procReady:
+			// set rlimits, this has to be done here because we lose permissions
+			// to raise the limits once we enter a user-namespace
+			if err := setupRlimits(p.config.Rlimits, p.pid()); err != nil {
+				return newSystemErrorWithCause(err, "setting rlimits for ready process")
+			}
+			// call prestart hooks
+			if !p.config.Config.Namespaces.Contains(configs.NEWNS) {
+				// Setup cgroup before prestart hook, so that the prestart hook could apply cgroup permissions.
+				if err := p.manager.Set(p.config.Config); err != nil {
+					return newSystemErrorWithCause(err, "setting cgroup config for ready process")
+				}
+
+				if p.config.Config.Hooks != nil {
+					s := configs.HookState{
+						Version: p.container.config.Version,
+						ID:      p.container.id,
+						Pid:     p.pid(),
+						Bundle:  utils.SearchLabels(p.config.Config.Labels, "bundle"),
+					}
+					for i, hook := range p.config.Config.Hooks.Prestart {
+						if err := hook.Run(s); err != nil {
+							return newSystemErrorWithCausef(err, "running prestart hook %d", i)
+						}
+					}
+				}
+			}
+			// Sync with child.
+			if err := writeSync(p.parentPipe, procRun); err != nil {
+				return newSystemErrorWithCause(err, "writing syncT 'run'")
+			}
+			sentRun = true
+		case procHooks:
+			// Setup cgroup before prestart hook, so that the prestart hook could apply cgroup permissions.
+			if err := p.manager.Set(p.config.Config); err != nil {
+				return newSystemErrorWithCause(err, "setting cgroup config for procHooks process")
+			}
+			if p.config.Config.Hooks != nil {
+				s := configs.HookState{
+					Version: p.container.config.Version,
+					ID:      p.container.id,
+					Pid:     p.pid(),
+					Bundle:  utils.SearchLabels(p.config.Config.Labels, "bundle"),
+				}
+				for i, hook := range p.config.Config.Hooks.Prestart {
+					if err := hook.Run(s); err != nil {
+						return newSystemErrorWithCausef(err, "running prestart hook %d", i)
+					}
+				}
+			}
+			// Sync with child.
+			if err := writeSync(p.parentPipe, procResume); err != nil {
+				return newSystemErrorWithCause(err, "writing syncT 'resume'")
+			}
+			sentResume = true
+		default:
+			return newSystemError(fmt.Errorf("invalid JSON payload from child"))
+		}
+
+		return nil
+	})
+
+	if !sentRun {
+		return newSystemErrorWithCause(ierr, "container init")
+	}
+	if p.config.Config.Namespaces.Contains(configs.NEWNS) && !sentResume {
+		return newSystemError(fmt.Errorf("could not synchronise after executing prestart hooks with container process"))
+	}
+	if err := unix.Shutdown(int(p.parentPipe.Fd()), unix.SHUT_WR); err != nil {
+		return newSystemErrorWithCause(err, "shutting down init pipe")
+	}
+
+	// Must be done after Shutdown so the child will exit and we can wait for it.
+	if ierr != nil {
+		p.wait()
+		return ierr
+	}
+	return nil
+}
+
+func (p *initProcess) wait() (*os.ProcessState, error) {
+	err := p.cmd.Wait()
+	if err != nil {
+		return p.cmd.ProcessState, err
+	}
+	// we should kill all processes in cgroup when init is died if we use host PID namespace
+	if p.sharePidns {
+		signalAllProcesses(p.manager, unix.SIGKILL)
+	}
+	return p.cmd.ProcessState, nil
+}
+
+func (p *initProcess) terminate() error {
+	if p.cmd.Process == nil {
+		return nil
+	}
+	err := p.cmd.Process.Kill()
+	if _, werr := p.wait(); err == nil {
+		err = werr
+	}
+	return err
+}
+
+func (p *initProcess) startTime() (uint64, error) {
+	stat, err := system.Stat(p.pid())
+	return stat.StartTime, err
+}
+
+func (p *initProcess) sendConfig() error {
+	// send the config to the container's init process, we don't use JSON Encode
+	// here because there might be a problem in JSON decoder in some cases, see:
+	// https://github.com/docker/docker/issues/14203#issuecomment-174177790
+	return utils.WriteJSON(p.parentPipe, p.config)
+}
+
+func (p *initProcess) createNetworkInterfaces() error {
+	for _, config := range p.config.Config.Networks {
+		strategy, err := getStrategy(config.Type)
+		if err != nil {
+			return err
+		}
+		n := &network{
+			Network: *config,
+		}
+		if err := strategy.create(n, p.pid()); err != nil {
+			return err
+		}
+		p.config.Networks = append(p.config.Networks, n)
+	}
+	return nil
+}
+
+func (p *initProcess) signal(sig os.Signal) error {
+	s, ok := sig.(syscall.Signal)
+	if !ok {
+		return errors.New("os: unsupported signal type")
+	}
+	return unix.Kill(p.pid(), s)
+}
+
+func (p *initProcess) setExternalDescriptors(newFds []string) {
+	p.fds = newFds
+}
+
+func getPipeFds(pid int) ([]string, error) {
+	fds := make([]string, 3)
+
+	dirPath := filepath.Join("/proc", strconv.Itoa(pid), "/fd")
+	for i := 0; i < 3; i++ {
+		// XXX: This breaks if the path is not a valid symlink (which can
+		//      happen in certain particularly unlucky mount namespace setups).
+		f := filepath.Join(dirPath, strconv.Itoa(i))
+		target, err := os.Readlink(f)
+		if err != nil {
+			// Ignore permission errors, for rootless containers and other
+			// non-dumpable processes. if we can't get the fd for a particular
+			// file, there's not much we can do.
+			if os.IsPermission(err) {
+				continue
+			}
+			return fds, err
+		}
+		fds[i] = target
+	}
+	return fds, nil
+}
+
+// InitializeIO creates pipes for use with the process's stdio and returns the
+// opposite side for each. Do not use this if you want to have a pseudoterminal
+// set up for you by libcontainer (TODO: fix that too).
+// TODO: This is mostly unnecessary, and should be handled by clients.
+func (p *Process) InitializeIO(rootuid, rootgid int) (i *IO, err error) {
+	var fds []uintptr
+	i = &IO{}
+	// cleanup in case of an error
+	defer func() {
+		if err != nil {
+			for _, fd := range fds {
+				unix.Close(int(fd))
+			}
+		}
+	}()
+	// STDIN
+	r, w, err := os.Pipe()
+	if err != nil {
+		return nil, err
+	}
+	fds = append(fds, r.Fd(), w.Fd())
+	p.Stdin, i.Stdin = r, w
+	// STDOUT
+	if r, w, err = os.Pipe(); err != nil {
+		return nil, err
+	}
+	fds = append(fds, r.Fd(), w.Fd())
+	p.Stdout, i.Stdout = w, r
+	// STDERR
+	if r, w, err = os.Pipe(); err != nil {
+		return nil, err
+	}
+	fds = append(fds, r.Fd(), w.Fd())
+	p.Stderr, i.Stderr = w, r
+	// change ownership of the pipes incase we are in a user namespace
+	for _, fd := range fds {
+		if err := unix.Fchown(int(fd), rootuid, rootgid); err != nil {
+			return nil, err
+		}
+	}
+	return i, nil
+}

vendor/github.com/opencontainers/runc/libcontainer/restored_process.go

@@ -0,0 +1,122 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/opencontainers/runc/libcontainer/system"
+)
+
+func newRestoredProcess(pid int, fds []string) (*restoredProcess, error) {
+	var (
+		err error
+	)
+	proc, err := os.FindProcess(pid)
+	if err != nil {
+		return nil, err
+	}
+	stat, err := system.Stat(pid)
+	if err != nil {
+		return nil, err
+	}
+	return &restoredProcess{
+		proc:             proc,
+		processStartTime: stat.StartTime,
+		fds:              fds,
+	}, nil
+}
+
+type restoredProcess struct {
+	proc             *os.Process
+	processStartTime uint64
+	fds              []string
+}
+
+func (p *restoredProcess) start() error {
+	return newGenericError(fmt.Errorf("restored process cannot be started"), SystemError)
+}
+
+func (p *restoredProcess) pid() int {
+	return p.proc.Pid
+}
+
+func (p *restoredProcess) terminate() error {
+	err := p.proc.Kill()
+	if _, werr := p.wait(); err == nil {
+		err = werr
+	}
+	return err
+}
+
+func (p *restoredProcess) wait() (*os.ProcessState, error) {
+	// TODO: how do we wait on the actual process?
+	// maybe use --exec-cmd in criu
+	st, err := p.proc.Wait()
+	if err != nil {
+		return nil, err
+	}
+	return st, nil
+}
+
+func (p *restoredProcess) startTime() (uint64, error) {
+	return p.processStartTime, nil
+}
+
+func (p *restoredProcess) signal(s os.Signal) error {
+	return p.proc.Signal(s)
+}
+
+func (p *restoredProcess) externalDescriptors() []string {
+	return p.fds
+}
+
+func (p *restoredProcess) setExternalDescriptors(newFds []string) {
+	p.fds = newFds
+}
+
+// nonChildProcess represents a process where the calling process is not
+// the parent process.  This process is created when a factory loads a container from
+// a persisted state.
+type nonChildProcess struct {
+	processPid       int
+	processStartTime uint64
+	fds              []string
+}
+
+func (p *nonChildProcess) start() error {
+	return newGenericError(fmt.Errorf("restored process cannot be started"), SystemError)
+}
+
+func (p *nonChildProcess) pid() int {
+	return p.processPid
+}
+
+func (p *nonChildProcess) terminate() error {
+	return newGenericError(fmt.Errorf("restored process cannot be terminated"), SystemError)
+}
+
+func (p *nonChildProcess) wait() (*os.ProcessState, error) {
+	return nil, newGenericError(fmt.Errorf("restored process cannot be waited on"), SystemError)
+}
+
+func (p *nonChildProcess) startTime() (uint64, error) {
+	return p.processStartTime, nil
+}
+
+func (p *nonChildProcess) signal(s os.Signal) error {
+	proc, err := os.FindProcess(p.processPid)
+	if err != nil {
+		return err
+	}
+	return proc.Signal(s)
+}
+
+func (p *nonChildProcess) externalDescriptors() []string {
+	return p.fds
+}
+
+func (p *nonChildProcess) setExternalDescriptors(newFds []string) {
+	p.fds = newFds
+}

vendor/github.com/opencontainers/runc/libcontainer/rootfs_linux.go

@@ -0,0 +1,812 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"io"
+	"io/ioutil"
+	"os"
+	"os/exec"
+	"path"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/docker/docker/pkg/mount"
+	"github.com/docker/docker/pkg/symlink"
+	"github.com/mrunalp/fileutils"
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/system"
+	libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
+	"github.com/opencontainers/selinux/go-selinux/label"
+
+	"golang.org/x/sys/unix"
+)
+
+const defaultMountFlags = unix.MS_NOEXEC | unix.MS_NOSUID | unix.MS_NODEV
+
+// needsSetupDev returns true if /dev needs to be set up.
+func needsSetupDev(config *configs.Config) bool {
+	for _, m := range config.Mounts {
+		if m.Device == "bind" && libcontainerUtils.CleanPath(m.Destination) == "/dev" {
+			return false
+		}
+	}
+	return true
+}
+
+// prepareRootfs sets up the devices, mount points, and filesystems for use
+// inside a new mount namespace. It doesn't set anything as ro. You must call
+// finalizeRootfs after this function to finish setting up the rootfs.
+func prepareRootfs(pipe io.ReadWriter, config *configs.Config) (err error) {
+	if err := prepareRoot(config); err != nil {
+		return newSystemErrorWithCause(err, "preparing rootfs")
+	}
+
+	setupDev := needsSetupDev(config)
+	for _, m := range config.Mounts {
+		for _, precmd := range m.PremountCmds {
+			if err := mountCmd(precmd); err != nil {
+				return newSystemErrorWithCause(err, "running premount command")
+			}
+		}
+
+		if err := mountToRootfs(m, config.Rootfs, config.MountLabel); err != nil {
+			return newSystemErrorWithCausef(err, "mounting %q to rootfs %q at %q", m.Source, config.Rootfs, m.Destination)
+		}
+
+		for _, postcmd := range m.PostmountCmds {
+			if err := mountCmd(postcmd); err != nil {
+				return newSystemErrorWithCause(err, "running postmount command")
+			}
+		}
+	}
+
+	if setupDev {
+		if err := createDevices(config); err != nil {
+			return newSystemErrorWithCause(err, "creating device nodes")
+		}
+		if err := setupPtmx(config); err != nil {
+			return newSystemErrorWithCause(err, "setting up ptmx")
+		}
+		if err := setupDevSymlinks(config.Rootfs); err != nil {
+			return newSystemErrorWithCause(err, "setting up /dev symlinks")
+		}
+	}
+
+	// Signal the parent to run the pre-start hooks.
+	// The hooks are run after the mounts are setup, but before we switch to the new
+	// root, so that the old root is still available in the hooks for any mount
+	// manipulations.
+	if err := syncParentHooks(pipe); err != nil {
+		return err
+	}
+
+	// The reason these operations are done here rather than in finalizeRootfs
+	// is because the console-handling code gets quite sticky if we have to set
+	// up the console before doing the pivot_root(2). This is because the
+	// Console API has to also work with the ExecIn case, which means that the
+	// API must be able to deal with being inside as well as outside the
+	// container. It's just cleaner to do this here (at the expense of the
+	// operation not being perfectly split).
+
+	if err := unix.Chdir(config.Rootfs); err != nil {
+		return newSystemErrorWithCausef(err, "changing dir to %q", config.Rootfs)
+	}
+
+	if config.NoPivotRoot {
+		err = msMoveRoot(config.Rootfs)
+	} else {
+		err = pivotRoot(config.Rootfs)
+	}
+	if err != nil {
+		return newSystemErrorWithCause(err, "jailing process inside rootfs")
+	}
+
+	if setupDev {
+		if err := reOpenDevNull(); err != nil {
+			return newSystemErrorWithCause(err, "reopening /dev/null inside container")
+		}
+	}
+
+	return nil
+}
+
+// finalizeRootfs sets anything to ro if necessary. You must call
+// prepareRootfs first.
+func finalizeRootfs(config *configs.Config) (err error) {
+	// remount dev as ro if specified
+	for _, m := range config.Mounts {
+		if libcontainerUtils.CleanPath(m.Destination) == "/dev" {
+			if m.Flags&unix.MS_RDONLY == unix.MS_RDONLY {
+				if err := remountReadonly(m); err != nil {
+					return newSystemErrorWithCausef(err, "remounting %q as readonly", m.Destination)
+				}
+			}
+			break
+		}
+	}
+
+	// set rootfs ( / ) as readonly
+	if config.Readonlyfs {
+		if err := setReadonly(); err != nil {
+			return newSystemErrorWithCause(err, "setting rootfs as readonly")
+		}
+	}
+
+	unix.Umask(0022)
+	return nil
+}
+
+func mountCmd(cmd configs.Command) error {
+	command := exec.Command(cmd.Path, cmd.Args[:]...)
+	command.Env = cmd.Env
+	command.Dir = cmd.Dir
+	if out, err := command.CombinedOutput(); err != nil {
+		return fmt.Errorf("%#v failed: %s: %v", cmd, string(out), err)
+	}
+	return nil
+}
+
+func mountToRootfs(m *configs.Mount, rootfs, mountLabel string) error {
+	var (
+		dest = m.Destination
+	)
+	if !strings.HasPrefix(dest, rootfs) {
+		dest = filepath.Join(rootfs, dest)
+	}
+
+	switch m.Device {
+	case "proc", "sysfs":
+		if err := os.MkdirAll(dest, 0755); err != nil {
+			return err
+		}
+		// Selinux kernels do not support labeling of /proc or /sys
+		return mountPropagate(m, rootfs, "")
+	case "mqueue":
+		if err := os.MkdirAll(dest, 0755); err != nil {
+			return err
+		}
+		if err := mountPropagate(m, rootfs, mountLabel); err != nil {
+			// older kernels do not support labeling of /dev/mqueue
+			if err := mountPropagate(m, rootfs, ""); err != nil {
+				return err
+			}
+			return label.SetFileLabel(dest, mountLabel)
+		}
+		return nil
+	case "tmpfs":
+		copyUp := m.Extensions&configs.EXT_COPYUP == configs.EXT_COPYUP
+		tmpDir := ""
+		stat, err := os.Stat(dest)
+		if err != nil {
+			if err := os.MkdirAll(dest, 0755); err != nil {
+				return err
+			}
+		}
+		if copyUp {
+			tmpDir, err = ioutil.TempDir("/tmp", "runctmpdir")
+			if err != nil {
+				return newSystemErrorWithCause(err, "tmpcopyup: failed to create tmpdir")
+			}
+			defer os.RemoveAll(tmpDir)
+			m.Destination = tmpDir
+		}
+		if err := mountPropagate(m, rootfs, mountLabel); err != nil {
+			return err
+		}
+		if copyUp {
+			if err := fileutils.CopyDirectory(dest, tmpDir); err != nil {
+				errMsg := fmt.Errorf("tmpcopyup: failed to copy %s to %s: %v", dest, tmpDir, err)
+				if err1 := unix.Unmount(tmpDir, unix.MNT_DETACH); err1 != nil {
+					return newSystemErrorWithCausef(err1, "tmpcopyup: %v: failed to unmount", errMsg)
+				}
+				return errMsg
+			}
+			if err := unix.Mount(tmpDir, dest, "", unix.MS_MOVE, ""); err != nil {
+				errMsg := fmt.Errorf("tmpcopyup: failed to move mount %s to %s: %v", tmpDir, dest, err)
+				if err1 := unix.Unmount(tmpDir, unix.MNT_DETACH); err1 != nil {
+					return newSystemErrorWithCausef(err1, "tmpcopyup: %v: failed to unmount", errMsg)
+				}
+				return errMsg
+			}
+		}
+		if stat != nil {
+			if err = os.Chmod(dest, stat.Mode()); err != nil {
+				return err
+			}
+		}
+		return nil
+	case "bind":
+		stat, err := os.Stat(m.Source)
+		if err != nil {
+			// error out if the source of a bind mount does not exist as we will be
+			// unable to bind anything to it.
+			return err
+		}
+		// ensure that the destination of the bind mount is resolved of symlinks at mount time because
+		// any previous mounts can invalidate the next mount's destination.
+		// this can happen when a user specifies mounts within other mounts to cause breakouts or other
+		// evil stuff to try to escape the container's rootfs.
+		if dest, err = symlink.FollowSymlinkInScope(dest, rootfs); err != nil {
+			return err
+		}
+		if err := checkMountDestination(rootfs, dest); err != nil {
+			return err
+		}
+		// update the mount with the correct dest after symlinks are resolved.
+		m.Destination = dest
+		if err := createIfNotExists(dest, stat.IsDir()); err != nil {
+			return err
+		}
+		if err := mountPropagate(m, rootfs, mountLabel); err != nil {
+			return err
+		}
+		// bind mount won't change mount options, we need remount to make mount options effective.
+		// first check that we have non-default options required before attempting a remount
+		if m.Flags&^(unix.MS_REC|unix.MS_REMOUNT|unix.MS_BIND) != 0 {
+			// only remount if unique mount options are set
+			if err := remount(m, rootfs); err != nil {
+				return err
+			}
+		}
+
+		if m.Relabel != "" {
+			if err := label.Validate(m.Relabel); err != nil {
+				return err
+			}
+			shared := label.IsShared(m.Relabel)
+			if err := label.Relabel(m.Source, mountLabel, shared); err != nil {
+				return err
+			}
+		}
+	case "cgroup":
+		binds, err := getCgroupMounts(m)
+		if err != nil {
+			return err
+		}
+		var merged []string
+		for _, b := range binds {
+			ss := filepath.Base(b.Destination)
+			if strings.Contains(ss, ",") {
+				merged = append(merged, ss)
+			}
+		}
+		tmpfs := &configs.Mount{
+			Source:           "tmpfs",
+			Device:           "tmpfs",
+			Destination:      m.Destination,
+			Flags:            defaultMountFlags,
+			Data:             "mode=755",
+			PropagationFlags: m.PropagationFlags,
+		}
+		if err := mountToRootfs(tmpfs, rootfs, mountLabel); err != nil {
+			return err
+		}
+		for _, b := range binds {
+			if err := mountToRootfs(b, rootfs, mountLabel); err != nil {
+				return err
+			}
+		}
+		for _, mc := range merged {
+			for _, ss := range strings.Split(mc, ",") {
+				// symlink(2) is very dumb, it will just shove the path into
+				// the link and doesn't do any checks or relative path
+				// conversion. Also, don't error out if the cgroup already exists.
+				if err := os.Symlink(mc, filepath.Join(rootfs, m.Destination, ss)); err != nil && !os.IsExist(err) {
+					return err
+				}
+			}
+		}
+		if m.Flags&unix.MS_RDONLY != 0 {
+			// remount cgroup root as readonly
+			mcgrouproot := &configs.Mount{
+				Source:      m.Destination,
+				Device:      "bind",
+				Destination: m.Destination,
+				Flags:       defaultMountFlags | unix.MS_RDONLY | unix.MS_BIND,
+			}
+			if err := remount(mcgrouproot, rootfs); err != nil {
+				return err
+			}
+		}
+	default:
+		// ensure that the destination of the mount is resolved of symlinks at mount time because
+		// any previous mounts can invalidate the next mount's destination.
+		// this can happen when a user specifies mounts within other mounts to cause breakouts or other
+		// evil stuff to try to escape the container's rootfs.
+		var err error
+		if dest, err = symlink.FollowSymlinkInScope(dest, rootfs); err != nil {
+			return err
+		}
+		if err := checkMountDestination(rootfs, dest); err != nil {
+			return err
+		}
+		// update the mount with the correct dest after symlinks are resolved.
+		m.Destination = dest
+		if err := os.MkdirAll(dest, 0755); err != nil {
+			return err
+		}
+		return mountPropagate(m, rootfs, mountLabel)
+	}
+	return nil
+}
+
+func getCgroupMounts(m *configs.Mount) ([]*configs.Mount, error) {
+	mounts, err := cgroups.GetCgroupMounts(false)
+	if err != nil {
+		return nil, err
+	}
+
+	cgroupPaths, err := cgroups.ParseCgroupFile("/proc/self/cgroup")
+	if err != nil {
+		return nil, err
+	}
+
+	var binds []*configs.Mount
+
+	for _, mm := range mounts {
+		dir, err := mm.GetOwnCgroup(cgroupPaths)
+		if err != nil {
+			return nil, err
+		}
+		relDir, err := filepath.Rel(mm.Root, dir)
+		if err != nil {
+			return nil, err
+		}
+		binds = append(binds, &configs.Mount{
+			Device:           "bind",
+			Source:           filepath.Join(mm.Mountpoint, relDir),
+			Destination:      filepath.Join(m.Destination, filepath.Base(mm.Mountpoint)),
+			Flags:            unix.MS_BIND | unix.MS_REC | m.Flags,
+			PropagationFlags: m.PropagationFlags,
+		})
+	}
+
+	return binds, nil
+}
+
+// checkMountDestination checks to ensure that the mount destination is not over the top of /proc.
+// dest is required to be an abs path and have any symlinks resolved before calling this function.
+func checkMountDestination(rootfs, dest string) error {
+	invalidDestinations := []string{
+		"/proc",
+	}
+	// White list, it should be sub directories of invalid destinations
+	validDestinations := []string{
+		// These entries can be bind mounted by files emulated by fuse,
+		// so commands like top, free displays stats in container.
+		"/proc/cpuinfo",
+		"/proc/diskstats",
+		"/proc/meminfo",
+		"/proc/stat",
+		"/proc/swaps",
+		"/proc/uptime",
+		"/proc/net/dev",
+	}
+	for _, valid := range validDestinations {
+		path, err := filepath.Rel(filepath.Join(rootfs, valid), dest)
+		if err != nil {
+			return err
+		}
+		if path == "." {
+			return nil
+		}
+	}
+	for _, invalid := range invalidDestinations {
+		path, err := filepath.Rel(filepath.Join(rootfs, invalid), dest)
+		if err != nil {
+			return err
+		}
+		if path == "." || !strings.HasPrefix(path, "..") {
+			return fmt.Errorf("%q cannot be mounted because it is located inside %q", dest, invalid)
+		}
+	}
+	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/core"})
+	}
+	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
+}
+
+// If stdin, stdout, and/or stderr are pointing to `/dev/null` in the parent's rootfs
+// this method will make them point to `/dev/null` in this container's rootfs.  This
+// needs to be called after we chroot/pivot into the container's rootfs so that any
+// symlinks are resolved locally.
+func reOpenDevNull() error {
+	var stat, devNullStat unix.Stat_t
+	file, err := os.OpenFile("/dev/null", os.O_RDWR, 0)
+	if err != nil {
+		return fmt.Errorf("Failed to open /dev/null - %s", err)
+	}
+	defer file.Close()
+	if err := unix.Fstat(int(file.Fd()), &devNullStat); err != nil {
+		return err
+	}
+	for fd := 0; fd < 3; fd++ {
+		if err := unix.Fstat(fd, &stat); err != nil {
+			return err
+		}
+		if stat.Rdev == devNullStat.Rdev {
+			// Close and re-open the fd.
+			if err := unix.Dup3(int(file.Fd()), fd, 0); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+// Create the device nodes in the container.
+func createDevices(config *configs.Config) error {
+	useBindMount := system.RunningInUserNS() || config.Namespaces.Contains(configs.NEWUSER)
+	oldMask := unix.Umask(0000)
+	for _, node := range config.Devices {
+		// containers running in a user namespace are not allowed to mknod
+		// devices so we can just bind mount it from the host.
+		if err := createDeviceNode(config.Rootfs, node, useBindMount); err != nil {
+			unix.Umask(oldMask)
+			return err
+		}
+	}
+	unix.Umask(oldMask)
+	return nil
+}
+
+func bindMountDeviceNode(dest string, node *configs.Device) error {
+	f, err := os.Create(dest)
+	if err != nil && !os.IsExist(err) {
+		return err
+	}
+	if f != nil {
+		f.Close()
+	}
+	return unix.Mount(node.Path, dest, "bind", unix.MS_BIND, "")
+}
+
+// Creates the device node in the rootfs of the container.
+func createDeviceNode(rootfs string, node *configs.Device, bind bool) error {
+	dest := filepath.Join(rootfs, node.Path)
+	if err := os.MkdirAll(filepath.Dir(dest), 0755); err != nil {
+		return err
+	}
+
+	if bind {
+		return bindMountDeviceNode(dest, node)
+	}
+	if err := mknodDevice(dest, node); err != nil {
+		if os.IsExist(err) {
+			return nil
+		} else if os.IsPermission(err) {
+			return bindMountDeviceNode(dest, node)
+		}
+		return err
+	}
+	return nil
+}
+
+func mknodDevice(dest string, node *configs.Device) error {
+	fileMode := node.FileMode
+	switch node.Type {
+	case 'c', 'u':
+		fileMode |= unix.S_IFCHR
+	case 'b':
+		fileMode |= unix.S_IFBLK
+	case 'p':
+		fileMode |= unix.S_IFIFO
+	default:
+		return fmt.Errorf("%c is not a valid device type for device %s", node.Type, node.Path)
+	}
+	if err := unix.Mknod(dest, uint32(fileMode), node.Mkdev()); err != nil {
+		return err
+	}
+	return unix.Chown(dest, int(node.Uid), int(node.Gid))
+}
+
+func getMountInfo(mountinfo []*mount.Info, dir string) *mount.Info {
+	for _, m := range mountinfo {
+		if m.Mountpoint == dir {
+			return m
+		}
+	}
+	return nil
+}
+
+// Get the parent mount point of directory passed in as argument. Also return
+// optional fields.
+func getParentMount(rootfs string) (string, string, error) {
+	var path string
+
+	mountinfos, err := mount.GetMounts()
+	if err != nil {
+		return "", "", err
+	}
+
+	mountinfo := getMountInfo(mountinfos, rootfs)
+	if mountinfo != nil {
+		return rootfs, mountinfo.Optional, nil
+	}
+
+	path = rootfs
+	for {
+		path = filepath.Dir(path)
+
+		mountinfo = getMountInfo(mountinfos, path)
+		if mountinfo != nil {
+			return path, mountinfo.Optional, nil
+		}
+
+		if path == "/" {
+			break
+		}
+	}
+
+	// If we are here, we did not find parent mount. Something is wrong.
+	return "", "", fmt.Errorf("Could not find parent mount of %s", rootfs)
+}
+
+// Make parent mount private if it was shared
+func rootfsParentMountPrivate(rootfs string) error {
+	sharedMount := false
+
+	parentMount, optionalOpts, err := getParentMount(rootfs)
+	if err != nil {
+		return err
+	}
+
+	optsSplit := strings.Split(optionalOpts, " ")
+	for _, opt := range optsSplit {
+		if strings.HasPrefix(opt, "shared:") {
+			sharedMount = true
+			break
+		}
+	}
+
+	// Make parent mount PRIVATE if it was shared. It is needed for two
+	// reasons. First of all pivot_root() will fail if parent mount is
+	// shared. Secondly when we bind mount rootfs it will propagate to
+	// parent namespace and we don't want that to happen.
+	if sharedMount {
+		return unix.Mount("", parentMount, "", unix.MS_PRIVATE, "")
+	}
+
+	return nil
+}
+
+func prepareRoot(config *configs.Config) error {
+	flag := unix.MS_SLAVE | unix.MS_REC
+	if config.RootPropagation != 0 {
+		flag = config.RootPropagation
+	}
+	if err := unix.Mount("", "/", "", uintptr(flag), ""); err != nil {
+		return err
+	}
+
+	// Make parent mount private to make sure following bind mount does
+	// not propagate in other namespaces. Also it will help with kernel
+	// check pass in pivot_root. (IS_SHARED(new_mnt->mnt_parent))
+	if err := rootfsParentMountPrivate(config.Rootfs); err != nil {
+		return err
+	}
+
+	return unix.Mount(config.Rootfs, config.Rootfs, "bind", unix.MS_BIND|unix.MS_REC, "")
+}
+
+func setReadonly() error {
+	return unix.Mount("/", "/", "bind", unix.MS_BIND|unix.MS_REMOUNT|unix.MS_RDONLY|unix.MS_REC, "")
+}
+
+func setupPtmx(config *configs.Config) error {
+	ptmx := filepath.Join(config.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)
+	}
+	return nil
+}
+
+// pivotRoot will call pivot_root such that rootfs becomes the new root
+// filesystem, and everything else is cleaned up.
+func pivotRoot(rootfs string) error {
+	// While the documentation may claim otherwise, pivot_root(".", ".") is
+	// actually valid. What this results in is / being the new root but
+	// /proc/self/cwd being the old root. Since we can play around with the cwd
+	// with pivot_root this allows us to pivot without creating directories in
+	// the rootfs. Shout-outs to the LXC developers for giving us this idea.
+
+	oldroot, err := unix.Open("/", unix.O_DIRECTORY|unix.O_RDONLY, 0)
+	if err != nil {
+		return err
+	}
+	defer unix.Close(oldroot)
+
+	newroot, err := unix.Open(rootfs, unix.O_DIRECTORY|unix.O_RDONLY, 0)
+	if err != nil {
+		return err
+	}
+	defer unix.Close(newroot)
+
+	// Change to the new root so that the pivot_root actually acts on it.
+	if err := unix.Fchdir(newroot); err != nil {
+		return err
+	}
+
+	if err := unix.PivotRoot(".", "."); err != nil {
+		return fmt.Errorf("pivot_root %s", err)
+	}
+
+	// Currently our "." is oldroot (according to the current kernel code).
+	// However, purely for safety, we will fchdir(oldroot) since there isn't
+	// really any guarantee from the kernel what /proc/self/cwd will be after a
+	// pivot_root(2).
+
+	if err := unix.Fchdir(oldroot); err != nil {
+		return err
+	}
+
+	// Make oldroot rprivate to make sure our unmounts don't propagate to the
+	// host (and thus bork the machine).
+	if err := unix.Mount("", ".", "", unix.MS_PRIVATE|unix.MS_REC, ""); err != nil {
+		return err
+	}
+	// Preform the unmount. MNT_DETACH allows us to unmount /proc/self/cwd.
+	if err := unix.Unmount(".", unix.MNT_DETACH); err != nil {
+		return err
+	}
+
+	// Switch back to our shiny new root.
+	if err := unix.Chdir("/"); err != nil {
+		return fmt.Errorf("chdir / %s", err)
+	}
+	return nil
+}
+
+func msMoveRoot(rootfs string) error {
+	if err := unix.Mount(rootfs, "/", "", unix.MS_MOVE, ""); err != nil {
+		return err
+	}
+	if err := unix.Chroot("."); err != nil {
+		return err
+	}
+	return unix.Chdir("/")
+}
+
+// createIfNotExists creates a file or a directory only if it does not already exist.
+func createIfNotExists(path string, isDir bool) error {
+	if _, err := os.Stat(path); err != nil {
+		if os.IsNotExist(err) {
+			if isDir {
+				return os.MkdirAll(path, 0755)
+			}
+			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
+}
+
+// readonlyPath will make a path read only.
+func readonlyPath(path string) error {
+	if err := unix.Mount(path, path, "", unix.MS_BIND|unix.MS_REC, ""); err != nil {
+		if os.IsNotExist(err) {
+			return nil
+		}
+		return err
+	}
+	return unix.Mount(path, path, "", unix.MS_BIND|unix.MS_REMOUNT|unix.MS_RDONLY|unix.MS_REC, "")
+}
+
+// remountReadonly will remount an existing mount point and ensure that it is read-only.
+func remountReadonly(m *configs.Mount) error {
+	var (
+		dest  = m.Destination
+		flags = m.Flags
+	)
+	for i := 0; i < 5; i++ {
+		if err := unix.Mount("", dest, "", uintptr(flags|unix.MS_REMOUNT|unix.MS_RDONLY), ""); err != nil {
+			switch err {
+			case unix.EBUSY:
+				time.Sleep(100 * time.Millisecond)
+				continue
+			default:
+				return err
+			}
+		}
+		return nil
+	}
+	return fmt.Errorf("unable to mount %s as readonly max retries reached", dest)
+}
+
+// maskPath masks the top of the specified path inside a container to avoid
+// security issues from processes reading information from non-namespace aware
+// mounts ( proc/kcore ).
+// For files, maskPath bind mounts /dev/null over the top of the specified path.
+// For directories, maskPath mounts read-only tmpfs over the top of the specified path.
+func maskPath(path string) error {
+	if err := unix.Mount("/dev/null", path, "", unix.MS_BIND, ""); err != nil && !os.IsNotExist(err) {
+		if err == unix.ENOTDIR {
+			return unix.Mount("tmpfs", path, "tmpfs", unix.MS_RDONLY, "")
+		}
+		return err
+	}
+	return nil
+}
+
+// writeSystemProperty writes the value to a path under /proc/sys as determined from the key.
+// For e.g. net.ipv4.ip_forward translated to /proc/sys/net/ipv4/ip_forward.
+func writeSystemProperty(key, value string) error {
+	keyPath := strings.Replace(key, ".", "/", -1)
+	return ioutil.WriteFile(path.Join("/proc/sys", keyPath), []byte(value), 0644)
+}
+
+func remount(m *configs.Mount, rootfs string) error {
+	var (
+		dest = m.Destination
+	)
+	if !strings.HasPrefix(dest, rootfs) {
+		dest = filepath.Join(rootfs, dest)
+	}
+	if err := unix.Mount(m.Source, dest, m.Device, uintptr(m.Flags|unix.MS_REMOUNT), ""); err != nil {
+		return err
+	}
+	return nil
+}
+
+// Do the mount operation followed by additional mounts required to take care
+// of propagation flags.
+func mountPropagate(m *configs.Mount, rootfs string, mountLabel string) error {
+	var (
+		dest  = m.Destination
+		data  = label.FormatMountLabel(m.Data, mountLabel)
+		flags = m.Flags
+	)
+	if libcontainerUtils.CleanPath(dest) == "/dev" {
+		flags &= ^unix.MS_RDONLY
+	}
+
+	copyUp := m.Extensions&configs.EXT_COPYUP == configs.EXT_COPYUP
+	if !(copyUp || strings.HasPrefix(dest, rootfs)) {
+		dest = filepath.Join(rootfs, dest)
+	}
+
+	if err := unix.Mount(m.Source, dest, m.Device, uintptr(flags), data); err != nil {
+		return err
+	}
+
+	for _, pflag := range m.PropagationFlags {
+		if err := unix.Mount("", dest, "", uintptr(pflag), ""); err != nil {
+			return err
+		}
+	}
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/seccomp/config.go

@@ -0,0 +1,76 @@
+package seccomp
+
+import (
+	"fmt"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var operators = map[string]configs.Operator{
+	"SCMP_CMP_NE":        configs.NotEqualTo,
+	"SCMP_CMP_LT":        configs.LessThan,
+	"SCMP_CMP_LE":        configs.LessThanOrEqualTo,
+	"SCMP_CMP_EQ":        configs.EqualTo,
+	"SCMP_CMP_GE":        configs.GreaterThanOrEqualTo,
+	"SCMP_CMP_GT":        configs.GreaterThan,
+	"SCMP_CMP_MASKED_EQ": configs.MaskEqualTo,
+}
+
+var actions = map[string]configs.Action{
+	"SCMP_ACT_KILL":  configs.Kill,
+	"SCMP_ACT_ERRNO": configs.Errno,
+	"SCMP_ACT_TRAP":  configs.Trap,
+	"SCMP_ACT_ALLOW": configs.Allow,
+	"SCMP_ACT_TRACE": configs.Trace,
+}
+
+var archs = map[string]string{
+	"SCMP_ARCH_X86":         "x86",
+	"SCMP_ARCH_X86_64":      "amd64",
+	"SCMP_ARCH_X32":         "x32",
+	"SCMP_ARCH_ARM":         "arm",
+	"SCMP_ARCH_AARCH64":     "arm64",
+	"SCMP_ARCH_MIPS":        "mips",
+	"SCMP_ARCH_MIPS64":      "mips64",
+	"SCMP_ARCH_MIPS64N32":   "mips64n32",
+	"SCMP_ARCH_MIPSEL":      "mipsel",
+	"SCMP_ARCH_MIPSEL64":    "mipsel64",
+	"SCMP_ARCH_MIPSEL64N32": "mipsel64n32",
+	"SCMP_ARCH_PPC":         "ppc",
+	"SCMP_ARCH_PPC64":       "ppc64",
+	"SCMP_ARCH_PPC64LE":     "ppc64le",
+	"SCMP_ARCH_S390":        "s390",
+	"SCMP_ARCH_S390X":       "s390x",
+}
+
+// ConvertStringToOperator converts a string into a Seccomp comparison operator.
+// Comparison operators use the names they are assigned by Libseccomp's header.
+// Attempting to convert a string that is not a valid operator results in an
+// error.
+func ConvertStringToOperator(in string) (configs.Operator, error) {
+	if op, ok := operators[in]; ok == true {
+		return op, nil
+	}
+	return 0, fmt.Errorf("string %s is not a valid operator for seccomp", in)
+}
+
+// ConvertStringToAction converts a string into a Seccomp rule match action.
+// Actions use the names they are assigned in Libseccomp's header, though some
+// (notable, SCMP_ACT_TRACE) are not available in this implementation and will
+// return errors.
+// Attempting to convert a string that is not a valid action results in an
+// error.
+func ConvertStringToAction(in string) (configs.Action, error) {
+	if act, ok := actions[in]; ok == true {
+		return act, nil
+	}
+	return 0, fmt.Errorf("string %s is not a valid action for seccomp", in)
+}
+
+// ConvertStringToArch converts a string into a Seccomp comparison arch.
+func ConvertStringToArch(in string) (string, error) {
+	if arch, ok := archs[in]; ok == true {
+		return arch, nil
+	}
+	return "", fmt.Errorf("string %s is not a valid arch for seccomp", in)
+}

vendor/github.com/opencontainers/runc/libcontainer/seccomp/seccomp_linux.go

@@ -0,0 +1,227 @@
+// +build linux,cgo,seccomp
+
+package seccomp
+
+import (
+	"bufio"
+	"fmt"
+	"os"
+	"strings"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+	libseccomp "github.com/seccomp/libseccomp-golang"
+
+	"golang.org/x/sys/unix"
+)
+
+var (
+	actAllow = libseccomp.ActAllow
+	actTrap  = libseccomp.ActTrap
+	actKill  = libseccomp.ActKill
+	actTrace = libseccomp.ActTrace.SetReturnCode(int16(unix.EPERM))
+	actErrno = libseccomp.ActErrno.SetReturnCode(int16(unix.EPERM))
+)
+
+// Filters given syscalls in a container, preventing them from being used
+// Started in the container init process, and carried over to all child processes
+// Setns calls, however, require a separate invocation, as they are not children
+// of the init until they join the namespace
+func InitSeccomp(config *configs.Seccomp) error {
+	if config == nil {
+		return fmt.Errorf("cannot initialize Seccomp - nil config passed")
+	}
+
+	defaultAction, err := getAction(config.DefaultAction)
+	if err != nil {
+		return fmt.Errorf("error initializing seccomp - invalid default action")
+	}
+
+	filter, err := libseccomp.NewFilter(defaultAction)
+	if err != nil {
+		return fmt.Errorf("error creating filter: %s", err)
+	}
+
+	// Add extra architectures
+	for _, arch := range config.Architectures {
+		scmpArch, err := libseccomp.GetArchFromString(arch)
+		if err != nil {
+			return err
+		}
+
+		if err := filter.AddArch(scmpArch); err != nil {
+			return err
+		}
+	}
+
+	// Unset no new privs bit
+	if err := filter.SetNoNewPrivsBit(false); err != nil {
+		return fmt.Errorf("error setting no new privileges: %s", err)
+	}
+
+	// Add a rule for each syscall
+	for _, call := range config.Syscalls {
+		if call == nil {
+			return fmt.Errorf("encountered nil syscall while initializing Seccomp")
+		}
+
+		if err = matchCall(filter, call); err != nil {
+			return err
+		}
+	}
+
+	if err = filter.Load(); err != nil {
+		return fmt.Errorf("error loading seccomp filter into kernel: %s", err)
+	}
+
+	return nil
+}
+
+// IsEnabled returns if the kernel has been configured to support seccomp.
+func IsEnabled() bool {
+	// Try to read from /proc/self/status for kernels > 3.8
+	s, err := parseStatusFile("/proc/self/status")
+	if err != nil {
+		// Check if Seccomp is supported, via CONFIG_SECCOMP.
+		if err := unix.Prctl(unix.PR_GET_SECCOMP, 0, 0, 0, 0); err != unix.EINVAL {
+			// Make sure the kernel has CONFIG_SECCOMP_FILTER.
+			if err := unix.Prctl(unix.PR_SET_SECCOMP, unix.SECCOMP_MODE_FILTER, 0, 0, 0); err != unix.EINVAL {
+				return true
+			}
+		}
+		return false
+	}
+	_, ok := s["Seccomp"]
+	return ok
+}
+
+// Convert Libcontainer Action to Libseccomp ScmpAction
+func getAction(act configs.Action) (libseccomp.ScmpAction, error) {
+	switch act {
+	case configs.Kill:
+		return actKill, nil
+	case configs.Errno:
+		return actErrno, nil
+	case configs.Trap:
+		return actTrap, nil
+	case configs.Allow:
+		return actAllow, nil
+	case configs.Trace:
+		return actTrace, nil
+	default:
+		return libseccomp.ActInvalid, fmt.Errorf("invalid action, cannot use in rule")
+	}
+}
+
+// Convert Libcontainer Operator to Libseccomp ScmpCompareOp
+func getOperator(op configs.Operator) (libseccomp.ScmpCompareOp, error) {
+	switch op {
+	case configs.EqualTo:
+		return libseccomp.CompareEqual, nil
+	case configs.NotEqualTo:
+		return libseccomp.CompareNotEqual, nil
+	case configs.GreaterThan:
+		return libseccomp.CompareGreater, nil
+	case configs.GreaterThanOrEqualTo:
+		return libseccomp.CompareGreaterEqual, nil
+	case configs.LessThan:
+		return libseccomp.CompareLess, nil
+	case configs.LessThanOrEqualTo:
+		return libseccomp.CompareLessOrEqual, nil
+	case configs.MaskEqualTo:
+		return libseccomp.CompareMaskedEqual, nil
+	default:
+		return libseccomp.CompareInvalid, fmt.Errorf("invalid operator, cannot use in rule")
+	}
+}
+
+// Convert Libcontainer Arg to Libseccomp ScmpCondition
+func getCondition(arg *configs.Arg) (libseccomp.ScmpCondition, error) {
+	cond := libseccomp.ScmpCondition{}
+
+	if arg == nil {
+		return cond, fmt.Errorf("cannot convert nil to syscall condition")
+	}
+
+	op, err := getOperator(arg.Op)
+	if err != nil {
+		return cond, err
+	}
+
+	return libseccomp.MakeCondition(arg.Index, op, arg.Value, arg.ValueTwo)
+}
+
+// Add a rule to match a single syscall
+func matchCall(filter *libseccomp.ScmpFilter, call *configs.Syscall) error {
+	if call == nil || filter == nil {
+		return fmt.Errorf("cannot use nil as syscall to block")
+	}
+
+	if len(call.Name) == 0 {
+		return fmt.Errorf("empty string is not a valid syscall")
+	}
+
+	// If we can't resolve the syscall, assume it's not supported on this kernel
+	// Ignore it, don't error out
+	callNum, err := libseccomp.GetSyscallFromName(call.Name)
+	if err != nil {
+		return nil
+	}
+
+	// Convert the call's action to the libseccomp equivalent
+	callAct, err := getAction(call.Action)
+	if err != nil {
+		return err
+	}
+
+	// Unconditional match - just add the rule
+	if len(call.Args) == 0 {
+		if err = filter.AddRule(callNum, callAct); err != nil {
+			return err
+		}
+	} else {
+		// Conditional match - convert the per-arg rules into library format
+		conditions := []libseccomp.ScmpCondition{}
+
+		for _, cond := range call.Args {
+			newCond, err := getCondition(cond)
+			if err != nil {
+				return err
+			}
+
+			conditions = append(conditions, newCond)
+		}
+
+		if err = filter.AddRuleConditional(callNum, callAct, conditions); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func parseStatusFile(path string) (map[string]string, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return nil, err
+	}
+	defer f.Close()
+
+	s := bufio.NewScanner(f)
+	status := make(map[string]string)
+
+	for s.Scan() {
+		text := s.Text()
+		parts := strings.Split(text, ":")
+
+		if len(parts) <= 1 {
+			continue
+		}
+
+		status[parts[0]] = parts[1]
+	}
+	if err := s.Err(); err != nil {
+		return nil, err
+	}
+
+	return status, nil
+}

vendor/github.com/opencontainers/runc/libcontainer/seccomp/seccomp_unsupported.go

@@ -0,0 +1,24 @@
+// +build !linux !cgo !seccomp
+
+package seccomp
+
+import (
+	"errors"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var ErrSeccompNotEnabled = errors.New("seccomp: config provided but seccomp not supported")
+
+// InitSeccomp does nothing because seccomp is not supported.
+func InitSeccomp(config *configs.Seccomp) error {
+	if config != nil {
+		return ErrSeccompNotEnabled
+	}
+	return nil
+}
+
+// IsEnabled returns false, because it is not supported.
+func IsEnabled() bool {
+	return false
+}

vendor/github.com/opencontainers/runc/libcontainer/setgroups_linux.go

@@ -0,0 +1,11 @@
+// +build linux,go1.5
+
+package libcontainer
+
+import "syscall"
+
+// Set the GidMappingsEnableSetgroups member to true, so the process's
+// setgroups proc entry wont be set to 'deny' if GidMappings are set
+func enableSetgroups(sys *syscall.SysProcAttr) {
+	sys.GidMappingsEnableSetgroups = true
+}

vendor/github.com/opencontainers/runc/libcontainer/setns_init_linux.go

@@ -0,0 +1,65 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/opencontainers/runc/libcontainer/apparmor"
+	"github.com/opencontainers/runc/libcontainer/keys"
+	"github.com/opencontainers/runc/libcontainer/seccomp"
+	"github.com/opencontainers/runc/libcontainer/system"
+	"github.com/opencontainers/selinux/go-selinux/label"
+
+	"golang.org/x/sys/unix"
+)
+
+// linuxSetnsInit performs the container's initialization for running a new process
+// inside an existing container.
+type linuxSetnsInit struct {
+	pipe          *os.File
+	consoleSocket *os.File
+	config        *initConfig
+}
+
+func (l *linuxSetnsInit) getSessionRingName() string {
+	return fmt.Sprintf("_ses.%s", l.config.ContainerId)
+}
+
+func (l *linuxSetnsInit) Init() error {
+	if !l.config.Config.NoNewKeyring {
+		// do not inherit the parent's session keyring
+		if _, err := keys.JoinSessionKeyring(l.getSessionRingName()); err != nil {
+			return err
+		}
+	}
+	if l.config.CreateConsole {
+		if err := setupConsole(l.consoleSocket, l.config, false); err != nil {
+			return err
+		}
+		if err := system.Setctty(); err != nil {
+			return err
+		}
+	}
+	if l.config.NoNewPrivileges {
+		if err := unix.Prctl(unix.PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0); err != nil {
+			return err
+		}
+	}
+	if l.config.Config.Seccomp != nil {
+		if err := seccomp.InitSeccomp(l.config.Config.Seccomp); err != nil {
+			return err
+		}
+	}
+	if err := finalizeNamespace(l.config); err != nil {
+		return err
+	}
+	if err := apparmor.ApplyProfile(l.config.AppArmorProfile); err != nil {
+		return err
+	}
+	if err := label.SetProcessLabel(l.config.ProcessLabel); err != nil {
+		return err
+	}
+	return system.Execv(l.config.Args[0], l.config.Args[0:], os.Environ())
+}

vendor/github.com/opencontainers/runc/libcontainer/stacktrace/capture.go

@@ -0,0 +1,27 @@
+package stacktrace
+
+import "runtime"
+
+// Capture captures a stacktrace for the current calling go program
+//
+// skip is the number of frames to skip
+func Capture(userSkip int) Stacktrace {
+	var (
+		skip   = userSkip + 1 // add one for our own function
+		frames []Frame
+		prevPc uintptr
+	)
+	for i := skip; ; i++ {
+		pc, file, line, ok := runtime.Caller(i)
+		//detect if caller is repeated to avoid loop, gccgo
+		//currently runs  into a loop without this check
+		if !ok || pc == prevPc {
+			break
+		}
+		frames = append(frames, NewFrame(pc, file, line))
+		prevPc = pc
+	}
+	return Stacktrace{
+		Frames: frames,
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/stacktrace/frame.go

@@ -0,0 +1,38 @@
+package stacktrace
+
+import (
+	"path/filepath"
+	"runtime"
+	"strings"
+)
+
+// NewFrame returns a new stack frame for the provided information
+func NewFrame(pc uintptr, file string, line int) Frame {
+	fn := runtime.FuncForPC(pc)
+	if fn == nil {
+		return Frame{}
+	}
+	pack, name := parseFunctionName(fn.Name())
+	return Frame{
+		Line:     line,
+		File:     filepath.Base(file),
+		Package:  pack,
+		Function: name,
+	}
+}
+
+func parseFunctionName(name string) (string, string) {
+	i := strings.LastIndex(name, ".")
+	if i == -1 {
+		return "", name
+	}
+	return name[:i], name[i+1:]
+}
+
+// Frame contains all the information for a stack frame within a go program
+type Frame struct {
+	File     string
+	Function string
+	Package  string
+	Line     int
+}

vendor/github.com/opencontainers/runc/libcontainer/stacktrace/stacktrace.go

@@ -0,0 +1,5 @@
+package stacktrace
+
+type Stacktrace struct {
+	Frames []Frame
+}

vendor/github.com/opencontainers/runc/libcontainer/standard_init_linux.go

@@ -0,0 +1,188 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"os/exec"
+	"syscall" //only for Exec
+
+	"github.com/opencontainers/runc/libcontainer/apparmor"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/keys"
+	"github.com/opencontainers/runc/libcontainer/seccomp"
+	"github.com/opencontainers/runc/libcontainer/system"
+	"github.com/opencontainers/selinux/go-selinux/label"
+
+	"golang.org/x/sys/unix"
+)
+
+type linuxStandardInit struct {
+	pipe          *os.File
+	consoleSocket *os.File
+	parentPid     int
+	stateDirFD    int
+	config        *initConfig
+}
+
+func (l *linuxStandardInit) getSessionRingParams() (string, uint32, uint32) {
+	var newperms uint32
+
+	if l.config.Config.Namespaces.Contains(configs.NEWUSER) {
+		// with user ns we need 'other' search permissions
+		newperms = 0x8
+	} else {
+		// without user ns we need 'UID' search permissions
+		newperms = 0x80000
+	}
+
+	// create a unique per session container name that we can
+	// join in setns; however, other containers can also join it
+	return fmt.Sprintf("_ses.%s", l.config.ContainerId), 0xffffffff, newperms
+}
+
+func (l *linuxStandardInit) Init() error {
+	if !l.config.Config.NoNewKeyring {
+		ringname, keepperms, newperms := l.getSessionRingParams()
+
+		// do not inherit the parent's session keyring
+		sessKeyId, err := keys.JoinSessionKeyring(ringname)
+		if err != nil {
+			return err
+		}
+		// make session keyring searcheable
+		if err := keys.ModKeyringPerm(sessKeyId, keepperms, newperms); err != nil {
+			return err
+		}
+	}
+
+	if err := setupNetwork(l.config); err != nil {
+		return err
+	}
+	if err := setupRoute(l.config.Config); err != nil {
+		return err
+	}
+
+	label.Init()
+
+	// prepareRootfs() can be executed only for a new mount namespace.
+	if l.config.Config.Namespaces.Contains(configs.NEWNS) {
+		if err := prepareRootfs(l.pipe, l.config.Config); err != nil {
+			return err
+		}
+	}
+
+	// Set up the console. This has to be done *before* we finalize the rootfs,
+	// but *after* we've given the user the chance to set up all of the mounts
+	// they wanted.
+	if l.config.CreateConsole {
+		if err := setupConsole(l.consoleSocket, l.config, true); err != nil {
+			return err
+		}
+		if err := system.Setctty(); err != nil {
+			return err
+		}
+	}
+
+	// Finish the rootfs setup.
+	if l.config.Config.Namespaces.Contains(configs.NEWNS) {
+		if err := finalizeRootfs(l.config.Config); err != nil {
+			return err
+		}
+	}
+
+	if hostname := l.config.Config.Hostname; hostname != "" {
+		if err := unix.Sethostname([]byte(hostname)); err != nil {
+			return err
+		}
+	}
+	if err := apparmor.ApplyProfile(l.config.AppArmorProfile); err != nil {
+		return err
+	}
+	if err := label.SetProcessLabel(l.config.ProcessLabel); err != nil {
+		return err
+	}
+
+	for key, value := range l.config.Config.Sysctl {
+		if err := writeSystemProperty(key, value); err != nil {
+			return err
+		}
+	}
+	for _, path := range l.config.Config.ReadonlyPaths {
+		if err := readonlyPath(path); err != nil {
+			return err
+		}
+	}
+	for _, path := range l.config.Config.MaskPaths {
+		if err := maskPath(path); err != nil {
+			return err
+		}
+	}
+	pdeath, err := system.GetParentDeathSignal()
+	if err != nil {
+		return err
+	}
+	if l.config.NoNewPrivileges {
+		if err := unix.Prctl(unix.PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0); err != nil {
+			return err
+		}
+	}
+	// Tell our parent that we're ready to Execv. This must be done before the
+	// Seccomp rules have been applied, because we need to be able to read and
+	// write to a socket.
+	if err := syncParentReady(l.pipe); err != nil {
+		return err
+	}
+	// Without NoNewPrivileges seccomp is a privileged operation, so we need to
+	// do this before dropping capabilities; otherwise do it as late as possible
+	// just before execve so as few syscalls take place after it as possible.
+	if l.config.Config.Seccomp != nil && !l.config.NoNewPrivileges {
+		if err := seccomp.InitSeccomp(l.config.Config.Seccomp); err != nil {
+			return err
+		}
+	}
+	if err := finalizeNamespace(l.config); err != nil {
+		return err
+	}
+	// finalizeNamespace can change user/group which clears the parent death
+	// signal, so we restore it here.
+	if err := pdeath.Restore(); err != nil {
+		return err
+	}
+	// compare the parent from the initial start of the init process and make sure that it did not change.
+	// if the parent changes that means it died and we were reparented to something else so we should
+	// just kill ourself and not cause problems for someone else.
+	if unix.Getppid() != l.parentPid {
+		return unix.Kill(unix.Getpid(), unix.SIGKILL)
+	}
+	// check for the arg before waiting to make sure it exists and it is returned
+	// as a create time error.
+	name, err := exec.LookPath(l.config.Args[0])
+	if err != nil {
+		return err
+	}
+	// close the pipe to signal that we have completed our init.
+	l.pipe.Close()
+	// wait for the fifo to be opened on the other side before
+	// exec'ing the users process.
+	fd, err := unix.Openat(l.stateDirFD, execFifoFilename, os.O_WRONLY|unix.O_CLOEXEC, 0)
+	if err != nil {
+		return newSystemErrorWithCause(err, "openat exec fifo")
+	}
+	if _, err := unix.Write(fd, []byte("0")); err != nil {
+		return newSystemErrorWithCause(err, "write 0 exec fifo")
+	}
+	if l.config.Config.Seccomp != nil && l.config.NoNewPrivileges {
+		if err := seccomp.InitSeccomp(l.config.Config.Seccomp); err != nil {
+			return newSystemErrorWithCause(err, "init seccomp")
+		}
+	}
+	// close the statedir fd before exec because the kernel resets dumpable in the wrong order
+	// https://github.com/torvalds/linux/blob/v4.9/fs/exec.c#L1290-L1318
+	unix.Close(l.stateDirFD)
+	if err := syscall.Exec(name, l.config.Args[0:], os.Environ()); err != nil {
+		return newSystemErrorWithCause(err, "exec user process")
+	}
+	return nil
+}

vendor/github.com/opencontainers/runc/libcontainer/state_linux.go

@@ -0,0 +1,248 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"github.com/Sirupsen/logrus"
+	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/utils"
+
+	"golang.org/x/sys/unix"
+)
+
+func newStateTransitionError(from, to containerState) error {
+	return &stateTransitionError{
+		From: from.status().String(),
+		To:   to.status().String(),
+	}
+}
+
+// stateTransitionError is returned when an invalid state transition happens from one
+// state to another.
+type stateTransitionError struct {
+	From string
+	To   string
+}
+
+func (s *stateTransitionError) Error() string {
+	return fmt.Sprintf("invalid state transition from %s to %s", s.From, s.To)
+}
+
+type containerState interface {
+	transition(containerState) error
+	destroy() error
+	status() Status
+}
+
+func destroy(c *linuxContainer) error {
+	if !c.config.Namespaces.Contains(configs.NEWPID) {
+		if err := signalAllProcesses(c.cgroupManager, unix.SIGKILL); err != nil {
+			logrus.Warn(err)
+		}
+	}
+	err := c.cgroupManager.Destroy()
+	if rerr := os.RemoveAll(c.root); err == nil {
+		err = rerr
+	}
+	c.initProcess = nil
+	if herr := runPoststopHooks(c); err == nil {
+		err = herr
+	}
+	c.state = &stoppedState{c: c}
+	return err
+}
+
+func runPoststopHooks(c *linuxContainer) error {
+	if c.config.Hooks != nil {
+		s := configs.HookState{
+			Version: c.config.Version,
+			ID:      c.id,
+			Bundle:  utils.SearchLabels(c.config.Labels, "bundle"),
+		}
+		for _, hook := range c.config.Hooks.Poststop {
+			if err := hook.Run(s); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+// stoppedState represents a container is a stopped/destroyed state.
+type stoppedState struct {
+	c *linuxContainer
+}
+
+func (b *stoppedState) status() Status {
+	return Stopped
+}
+
+func (b *stoppedState) transition(s containerState) error {
+	switch s.(type) {
+	case *runningState, *restoredState:
+		b.c.state = s
+		return nil
+	case *stoppedState:
+		return nil
+	}
+	return newStateTransitionError(b, s)
+}
+
+func (b *stoppedState) destroy() error {
+	return destroy(b.c)
+}
+
+// runningState represents a container that is currently running.
+type runningState struct {
+	c *linuxContainer
+}
+
+func (r *runningState) status() Status {
+	return Running
+}
+
+func (r *runningState) transition(s containerState) error {
+	switch s.(type) {
+	case *stoppedState:
+		t, err := r.c.runType()
+		if err != nil {
+			return err
+		}
+		if t == Running {
+			return newGenericError(fmt.Errorf("container still running"), ContainerNotStopped)
+		}
+		r.c.state = s
+		return nil
+	case *pausedState:
+		r.c.state = s
+		return nil
+	case *runningState:
+		return nil
+	}
+	return newStateTransitionError(r, s)
+}
+
+func (r *runningState) destroy() error {
+	t, err := r.c.runType()
+	if err != nil {
+		return err
+	}
+	if t == Running {
+		return newGenericError(fmt.Errorf("container is not destroyed"), ContainerNotStopped)
+	}
+	return destroy(r.c)
+}
+
+type createdState struct {
+	c *linuxContainer
+}
+
+func (i *createdState) status() Status {
+	return Created
+}
+
+func (i *createdState) transition(s containerState) error {
+	switch s.(type) {
+	case *runningState, *pausedState, *stoppedState:
+		i.c.state = s
+		return nil
+	case *createdState:
+		return nil
+	}
+	return newStateTransitionError(i, s)
+}
+
+func (i *createdState) destroy() error {
+	i.c.initProcess.signal(unix.SIGKILL)
+	return destroy(i.c)
+}
+
+// pausedState represents a container that is currently pause.  It cannot be destroyed in a
+// paused state and must transition back to running first.
+type pausedState struct {
+	c *linuxContainer
+}
+
+func (p *pausedState) status() Status {
+	return Paused
+}
+
+func (p *pausedState) transition(s containerState) error {
+	switch s.(type) {
+	case *runningState, *stoppedState:
+		p.c.state = s
+		return nil
+	case *pausedState:
+		return nil
+	}
+	return newStateTransitionError(p, s)
+}
+
+func (p *pausedState) destroy() error {
+	t, err := p.c.runType()
+	if err != nil {
+		return err
+	}
+	if t != Running && t != Created {
+		if err := p.c.cgroupManager.Freeze(configs.Thawed); err != nil {
+			return err
+		}
+		return destroy(p.c)
+	}
+	return newGenericError(fmt.Errorf("container is paused"), ContainerPaused)
+}
+
+// restoredState is the same as the running state but also has associated checkpoint
+// information that maybe need destroyed when the container is stopped and destroy is called.
+type restoredState struct {
+	imageDir string
+	c        *linuxContainer
+}
+
+func (r *restoredState) status() Status {
+	return Running
+}
+
+func (r *restoredState) transition(s containerState) error {
+	switch s.(type) {
+	case *stoppedState, *runningState:
+		return nil
+	}
+	return newStateTransitionError(r, s)
+}
+
+func (r *restoredState) destroy() error {
+	if _, err := os.Stat(filepath.Join(r.c.root, "checkpoint")); err != nil {
+		if !os.IsNotExist(err) {
+			return err
+		}
+	}
+	return destroy(r.c)
+}
+
+// loadedState is used whenever a container is restored, loaded, or setting additional
+// processes inside and it should not be destroyed when it is exiting.
+type loadedState struct {
+	c *linuxContainer
+	s Status
+}
+
+func (n *loadedState) status() Status {
+	return n.s
+}
+
+func (n *loadedState) transition(s containerState) error {
+	n.c.state = s
+	return nil
+}
+
+func (n *loadedState) destroy() error {
+	if err := n.c.refreshState(); err != nil {
+		return err
+	}
+	return n.c.state.destroy()
+}

vendor/github.com/opencontainers/runc/libcontainer/stats.go

@@ -0,0 +1,15 @@
+package libcontainer
+
+type NetworkInterface struct {
+	// Name is the name of the network interface.
+	Name string
+
+	RxBytes   uint64
+	RxPackets uint64
+	RxErrors  uint64
+	RxDropped uint64
+	TxBytes   uint64
+	TxPackets uint64
+	TxErrors  uint64
+	TxDropped uint64
+}

vendor/github.com/opencontainers/runc/libcontainer/stats_freebsd.go

@@ -0,0 +1,5 @@
+package libcontainer
+
+type Stats struct {
+	Interfaces []*NetworkInterface
+}

vendor/github.com/opencontainers/runc/libcontainer/stats_linux.go

@@ -0,0 +1,8 @@
+package libcontainer
+
+import "github.com/opencontainers/runc/libcontainer/cgroups"
+
+type Stats struct {
+	Interfaces  []*NetworkInterface
+	CgroupStats *cgroups.Stats
+}

vendor/github.com/opencontainers/runc/libcontainer/stats_solaris.go

@@ -0,0 +1,7 @@
+package libcontainer
+
+// Solaris - TODO
+
+type Stats struct {
+	Interfaces []*NetworkInterface
+}

vendor/github.com/opencontainers/runc/libcontainer/stats_windows.go

@@ -0,0 +1,5 @@
+package libcontainer
+
+type Stats struct {
+	Interfaces []*NetworkInterface
+}

vendor/github.com/opencontainers/runc/libcontainer/sync.go

@@ -0,0 +1,107 @@
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"io"
+
+	"github.com/opencontainers/runc/libcontainer/utils"
+)
+
+type syncType string
+
+// Constants that are used for synchronisation between the parent and child
+// during container setup. They come in pairs (with procError being a generic
+// response which is followed by a &genericError).
+//
+// [  child  ] <-> [   parent   ]
+//
+// procHooks   --> [run hooks]
+//             <-- procResume
+//
+// procConsole -->
+//             <-- procConsoleReq
+//  [send(fd)] --> [recv(fd)]
+//             <-- procConsoleAck
+//
+// procReady   --> [final setup]
+//             <-- procRun
+const (
+	procError  syncType = "procError"
+	procReady  syncType = "procReady"
+	procRun    syncType = "procRun"
+	procHooks  syncType = "procHooks"
+	procResume syncType = "procResume"
+)
+
+type syncT struct {
+	Type syncType `json:"type"`
+}
+
+// writeSync is used to write to a synchronisation pipe. An error is returned
+// if there was a problem writing the payload.
+func writeSync(pipe io.Writer, sync syncType) error {
+	if err := utils.WriteJSON(pipe, syncT{sync}); err != nil {
+		return err
+	}
+	return nil
+}
+
+// readSync is used to read from a synchronisation pipe. An error is returned
+// if we got a genericError, the pipe was closed, or we got an unexpected flag.
+func readSync(pipe io.Reader, expected syncType) error {
+	var procSync syncT
+	if err := json.NewDecoder(pipe).Decode(&procSync); err != nil {
+		if err == io.EOF {
+			return fmt.Errorf("parent closed synchronisation channel")
+		}
+
+		if procSync.Type == procError {
+			var ierr genericError
+
+			if err := json.NewDecoder(pipe).Decode(&ierr); err != nil {
+				return fmt.Errorf("failed reading error from parent: %v", err)
+			}
+
+			return &ierr
+		}
+
+		if procSync.Type != expected {
+			return fmt.Errorf("invalid synchronisation flag from parent")
+		}
+	}
+	return nil
+}
+
+// parseSync runs the given callback function on each syncT received from the
+// child. It will return once io.EOF is returned from the given pipe.
+func parseSync(pipe io.Reader, fn func(*syncT) error) error {
+	dec := json.NewDecoder(pipe)
+	for {
+		var sync syncT
+		if err := dec.Decode(&sync); err != nil {
+			if err == io.EOF {
+				break
+			}
+			return err
+		}
+
+		// We handle this case outside fn for cleanliness reasons.
+		var ierr *genericError
+		if sync.Type == procError {
+			if err := dec.Decode(&ierr); err != nil && err != io.EOF {
+				return newSystemErrorWithCause(err, "decoding proc error from init")
+			}
+			if ierr != nil {
+				return ierr
+			}
+			// Programmer error.
+			panic("No error following JSON procError payload.")
+		}
+
+		if err := fn(&sync); err != nil {
+			return err
+		}
+	}
+	return nil
+}