update vendor

Signed-off-by: Jess Frazelle <acidburn@microsoft.com>

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

@@ -27,9 +27,9 @@ type Manager interface {
 	// Destroys the cgroup set
 	Destroy() error
 
-	// NewCgroupManager() and LoadCgroupManager() require following attributes:
+	// The option func SystemdCgroups() and Cgroupfs() require following attributes:
 	// 	Paths   map[string]string
-	// 	Cgroups *cgroups.Cgroup
+	// 	Cgroups *configs.Cgroup
 	// Paths maps cgroup subsystem to path at which it is mounted.
 	// Cgroups specifies specific cgroup settings for the various subsystems
 
@@ -37,7 +37,7 @@ type Manager interface {
 	// restore the object later.
 	GetPaths() map[string]string
 
-	// Set the cgroup as configured.
+	// Sets the cgroup as configured.
 	Set(container *configs.Config) error
 }
 

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

@@ -0,0 +1,18 @@
+// +build linux
+
+package cgroups
+
+import (
+	"testing"
+)
+
+func TestParseCgroups(t *testing.T) {
+	cgroups, err := ParseCgroupFile("/proc/self/cgroup")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if _, ok := cgroups["cpu"]; !ok {
+		t.Fail()
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/apply_raw.go

@@ -9,10 +9,8 @@ import (
 	"io/ioutil"
 	"os"
 	"path/filepath"
-	"strconv"
 	"sync"
 
-	"github.com/Sirupsen/logrus"
 	"github.com/opencontainers/runc/libcontainer/cgroups"
 	"github.com/opencontainers/runc/libcontainer/configs"
 	libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
@@ -34,7 +32,6 @@ var (
 		&FreezerGroup{},
 		&NameGroup{GroupName: "name=systemd", Join: true},
 	}
-	CgroupProcesses  = "cgroup.procs"
 	HugePageSizes, _ = cgroups.GetHugePageSize()
 )
 
@@ -107,6 +104,8 @@ func (m *Manager) Apply(pid int) (err error) {
 	if m.Cgroups == nil {
 		return nil
 	}
+	m.mu.Lock()
+	defer m.mu.Unlock()
 
 	var c = m.Cgroups
 
@@ -115,8 +114,8 @@ func (m *Manager) Apply(pid int) (err error) {
 		return err
 	}
 
+	m.Paths = make(map[string]string)
 	if c.Paths != nil {
-		paths := make(map[string]string)
 		for name, path := range c.Paths {
 			_, err := d.path(name)
 			if err != nil {
@@ -125,37 +124,44 @@ func (m *Manager) Apply(pid int) (err error) {
 				}
 				return err
 			}
-			paths[name] = path
+			m.Paths[name] = path
 		}
-		m.Paths = paths
 		return cgroups.EnterPid(m.Paths, pid)
 	}
 
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	paths := make(map[string]string)
 	for _, sys := range subsystems {
-		if err := sys.Apply(d); err != nil {
-			return err
-		}
 		// TODO: Apply should, ideally, be reentrant or be broken up into a separate
 		// create and join phase so that the cgroup hierarchy for a container can be
 		// created then join consists of writing the process pids to cgroup.procs
 		p, err := d.path(sys.Name())
 		if err != nil {
-			if cgroups.IsNotFound(err) {
+			// The non-presence of the devices subsystem is
+			// considered fatal for security reasons.
+			if cgroups.IsNotFound(err) && sys.Name() != "devices" {
 				continue
 			}
 			return err
 		}
-		paths[sys.Name()] = p
+		m.Paths[sys.Name()] = p
+
+		if err := sys.Apply(d); err != nil {
+			if os.IsPermission(err) && m.Cgroups.Path == "" {
+				// If we didn't set a cgroup path, then let's defer the error here
+				// until we know whether we have set limits or not.
+				// If we hadn't set limits, then it's ok that we couldn't join this cgroup, because
+				// it will have the same limits as its parent.
+				delete(m.Paths, sys.Name())
+				continue
+			}
+			return err
+		}
+
 	}
-	m.Paths = paths
 	return nil
 }
 
 func (m *Manager) Destroy() error {
-	if m.Cgroups.Paths != nil {
+	if m.Cgroups == nil || m.Cgroups.Paths != nil {
 		return nil
 	}
 	m.mu.Lock()
@@ -191,19 +197,20 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
 }
 
 func (m *Manager) Set(container *configs.Config) error {
-	for _, sys := range subsystems {
-		// Generate fake cgroup data.
-		d, err := getCgroupData(container.Cgroups, -1)
-		if err != nil {
-			return err
-		}
-		// Get the path, but don't error out if the cgroup wasn't found.
-		path, err := d.path(sys.Name())
-		if err != nil && !cgroups.IsNotFound(err) {
-			return err
-		}
+	// If Paths are set, then we are just joining cgroups paths
+	// and there is no need to set any values.
+	if m.Cgroups.Paths != nil {
+		return nil
+	}
 
+	paths := m.GetPaths()
+	for _, sys := range subsystems {
+		path := paths[sys.Name()]
 		if err := sys.Set(path, container.Cgroups); err != nil {
+			if path == "" {
+				// cgroup never applied
+				return fmt.Errorf("cannot set limits on the %s cgroup, as the container has not joined it", sys.Name())
+			}
 			return err
 		}
 	}
@@ -219,14 +226,8 @@ func (m *Manager) Set(container *configs.Config) error {
 // Freeze toggles the container's freezer cgroup depending on the state
 // provided
 func (m *Manager) Freeze(state configs.FreezerState) error {
-	d, err := getCgroupData(m.Cgroups, 0)
-	if err != nil {
-		return err
-	}
-	dir, err := d.path("freezer")
-	if err != nil {
-		return err
-	}
+	paths := m.GetPaths()
+	dir := paths["freezer"]
 	prevState := m.Cgroups.Resources.Freezer
 	m.Cgroups.Resources.Freezer = state
 	freezer, err := subsystems.Get("freezer")
@@ -242,28 +243,13 @@ func (m *Manager) Freeze(state configs.FreezerState) error {
 }
 
 func (m *Manager) GetPids() ([]int, error) {
-	dir, err := getCgroupPath(m.Cgroups)
-	if err != nil {
-		return nil, err
-	}
-	return cgroups.GetPids(dir)
+	paths := m.GetPaths()
+	return cgroups.GetPids(paths["devices"])
 }
 
 func (m *Manager) GetAllPids() ([]int, error) {
-	dir, err := getCgroupPath(m.Cgroups)
-	if err != nil {
-		return nil, err
-	}
-	return cgroups.GetAllPids(dir)
-}
-
-func getCgroupPath(c *configs.Cgroup) (string, error) {
-	d, err := getCgroupData(c, 0)
-	if err != nil {
-		return "", err
-	}
-
-	return d.path("devices")
+	paths := m.GetPaths()
+	return cgroups.GetAllPids(paths["devices"])
 }
 
 func getCgroupData(c *configs.Cgroup, pid int) (*cgroupData, error) {
@@ -294,25 +280,8 @@ func getCgroupData(c *configs.Cgroup, pid int) (*cgroupData, error) {
 	}, nil
 }
 
-func (raw *cgroupData) parentPath(subsystem, mountpoint, root string) (string, error) {
-	// Use GetThisCgroupDir instead of GetInitCgroupDir, because the creating
-	// process could in container and shared pid namespace with host, and
-	// /proc/1/cgroup could point to whole other world of cgroups.
-	initPath, err := cgroups.GetThisCgroupDir(subsystem)
-	if err != nil {
-		return "", err
-	}
-	// This is needed for nested containers, because in /proc/self/cgroup we
-	// see pathes from host, which don't exist in container.
-	relDir, err := filepath.Rel(root, initPath)
-	if err != nil {
-		return "", err
-	}
-	return filepath.Join(mountpoint, relDir), nil
-}
-
 func (raw *cgroupData) path(subsystem string) (string, error) {
-	mnt, root, err := cgroups.FindCgroupMountpointAndRoot(subsystem)
+	mnt, err := cgroups.FindCgroupMountpoint(subsystem)
 	// If we didn't mount the subsystem, there is no point we make the path.
 	if err != nil {
 		return "", err
@@ -320,11 +289,14 @@ func (raw *cgroupData) path(subsystem string) (string, error) {
 
 	// If the cgroup name/path is absolute do not look relative to the cgroup of the init process.
 	if filepath.IsAbs(raw.innerPath) {
-		// Sometimes subsystems can be mounted togethger as 'cpu,cpuacct'.
+		// Sometimes subsystems can be mounted together as 'cpu,cpuacct'.
 		return filepath.Join(raw.root, filepath.Base(mnt), raw.innerPath), nil
 	}
 
-	parentPath, err := raw.parentPath(subsystem, mnt, root)
+	// Use GetOwnCgroupPath instead of GetInitCgroupPath, because the creating
+	// process could in container and shared pid namespace with host, and
+	// /proc/1/cgroup could point to whole other world of cgroups.
+	parentPath, err := cgroups.GetOwnCgroupPath(subsystem)
 	if err != nil {
 		return "", err
 	}
@@ -340,7 +312,7 @@ func (raw *cgroupData) join(subsystem string) (string, error) {
 	if err := os.MkdirAll(path, 0755); err != nil {
 		return "", err
 	}
-	if err := writeFile(path, CgroupProcesses, strconv.Itoa(raw.pid)); err != nil {
+	if err := cgroups.WriteCgroupProc(path, raw.pid); err != nil {
 		return "", err
 	}
 	return path, nil
@@ -353,8 +325,7 @@ func writeFile(dir, file, data string) error {
 		return fmt.Errorf("no such directory for %s", file)
 	}
 	if err := ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700); err != nil {
-		//return fmt.Errorf("failed to write %v to %v: %v", data, file, err)
-		logrus.Debugf("failed to write %v to %v: %v", data, file, err)
+		return fmt.Errorf("failed to write %v to %v: %v", data, file, err)
 	}
 	return nil
 }
@@ -374,8 +345,8 @@ func removePath(p string, err error) error {
 	return nil
 }
 
-func CheckCpushares(path string, c int64) error {
-	var cpuShares int64
+func CheckCpushares(path string, c uint64) error {
+	var cpuShares uint64
 
 	if c == 0 {
 		return nil

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/apply_raw_test.go

@@ -0,0 +1,272 @@
+// +build linux
+
+package fs
+
+import (
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+func TestInvalidCgroupPath(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Path: "../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+func TestInvalidAbsoluteCgroupPath(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Path: "/../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidCgroupParent(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "../../../../../../../../../../some/path",
+		Name:   "name",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidAbsoluteCgroupParent(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "/../../../../../../../../../../some/path",
+		Name:   "name",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidCgroupName(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "parent",
+		Name:   "../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidAbsoluteCgroupName(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "parent",
+		Name:   "/../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidCgroupNameAndParent(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "../../../../../../../../../../some/path",
+		Name:   "../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}
+
+// XXX: Remove me after we get rid of configs.Cgroup.Name and configs.Cgroup.Parent.
+func TestInvalidAbsoluteCgroupNameAndParent(t *testing.T) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		t.Errorf("couldn't get cgroup root: %v", err)
+	}
+
+	config := &configs.Cgroup{
+		Parent: "/../../../../../../../../../../some/path",
+		Name:   "/../../../../../../../../../../some/path",
+	}
+
+	data, err := getCgroupData(config, 0)
+	if err != nil {
+		t.Errorf("couldn't get cgroup data: %v", err)
+	}
+
+	// Make sure the final innerPath doesn't go outside the cgroup mountpoint.
+	if strings.HasPrefix(data.innerPath, "..") {
+		t.Errorf("SECURITY: cgroup innerPath is outside cgroup mountpoint!")
+	}
+
+	// Double-check, using an actual cgroup.
+	deviceRoot := filepath.Join(root, "devices")
+	devicePath, err := data.path("devices")
+	if err != nil {
+		t.Errorf("couldn't get cgroup path: %v", err)
+	}
+	if !strings.HasPrefix(devicePath, deviceRoot) {
+		t.Errorf("SECURITY: cgroup path() is outside cgroup mountpoint!")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/blkio_test.go

@@ -0,0 +1,636 @@
+// +build linux
+
+package fs
+
+import (
+	"strconv"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+const (
+	sectorsRecursiveContents      = `8:0 1024`
+	serviceBytesRecursiveContents = `8:0 Read 100
+8:0 Write 200
+8:0 Sync 300
+8:0 Async 500
+8:0 Total 500
+Total 500`
+	servicedRecursiveContents = `8:0 Read 10
+8:0 Write 40
+8:0 Sync 20
+8:0 Async 30
+8:0 Total 50
+Total 50`
+	queuedRecursiveContents = `8:0 Read 1
+8:0 Write 4
+8:0 Sync 2
+8:0 Async 3
+8:0 Total 5
+Total 5`
+	serviceTimeRecursiveContents = `8:0 Read 173959
+8:0 Write 0
+8:0 Sync 0
+8:0 Async 173959
+8:0 Total 17395
+Total 17395`
+	waitTimeRecursiveContents = `8:0 Read 15571
+8:0 Write 0
+8:0 Sync 0
+8:0 Async 15571
+8:0 Total 15571`
+	mergedRecursiveContents = `8:0 Read 5
+8:0 Write 10
+8:0 Sync 0
+8:0 Async 0
+8:0 Total 15
+Total 15`
+	timeRecursiveContents = `8:0 8`
+	throttleServiceBytes  = `8:0 Read 11030528
+8:0 Write 23
+8:0 Sync 42
+8:0 Async 11030528
+8:0 Total 11030528
+252:0 Read 11030528
+252:0 Write 23
+252:0 Sync 42
+252:0 Async 11030528
+252:0 Total 11030528
+Total 22061056`
+	throttleServiced = `8:0 Read 164
+8:0 Write 23
+8:0 Sync 42
+8:0 Async 164
+8:0 Total 164
+252:0 Read 164
+252:0 Write 23
+252:0 Sync 42
+252:0 Async 164
+252:0 Total 164
+Total 328`
+)
+
+func appendBlkioStatEntry(blkioStatEntries *[]cgroups.BlkioStatEntry, major, minor, value uint64, op string) {
+	*blkioStatEntries = append(*blkioStatEntries, cgroups.BlkioStatEntry{Major: major, Minor: minor, Value: value, Op: op})
+}
+
+func TestBlkioSetWeight(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		weightBefore = 100
+		weightAfter  = 200
+	)
+
+	helper.writeFileContents(map[string]string{
+		"blkio.weight": strconv.Itoa(weightBefore),
+	})
+
+	helper.CgroupData.config.Resources.BlkioWeight = weightAfter
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "blkio.weight")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.weight - %s", err)
+	}
+
+	if value != weightAfter {
+		t.Fatal("Got the wrong value, set blkio.weight failed.")
+	}
+}
+
+func TestBlkioSetWeightDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		weightDeviceBefore = "8:0 400"
+	)
+
+	wd := configs.NewWeightDevice(8, 0, 500, 0)
+	weightDeviceAfter := wd.WeightString()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.weight_device": weightDeviceBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioWeightDevice = []*configs.WeightDevice{wd}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.weight_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.weight_device - %s", err)
+	}
+
+	if value != weightDeviceAfter {
+		t.Fatal("Got the wrong value, set blkio.weight_device failed.")
+	}
+}
+
+// regression #274
+func TestBlkioSetMultipleWeightDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		weightDeviceBefore = "8:0 400"
+	)
+
+	wd1 := configs.NewWeightDevice(8, 0, 500, 0)
+	wd2 := configs.NewWeightDevice(8, 16, 500, 0)
+	// we cannot actually set and check both because normal ioutil.WriteFile
+	// when writing to cgroup file will overwrite the whole file content instead
+	// of updating it as the kernel is doing. Just check the second device
+	// is present will suffice for the test to ensure multiple writes are done.
+	weightDeviceAfter := wd2.WeightString()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.weight_device": weightDeviceBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioWeightDevice = []*configs.WeightDevice{wd1, wd2}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.weight_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.weight_device - %s", err)
+	}
+
+	if value != weightDeviceAfter {
+		t.Fatal("Got the wrong value, set blkio.weight_device failed.")
+	}
+}
+
+func TestBlkioStats(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Verify expected stats.
+	expectedStats := cgroups.BlkioStats{}
+	appendBlkioStatEntry(&expectedStats.SectorsRecursive, 8, 0, 1024, "")
+
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 100, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 200, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 300, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 500, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 500, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 10, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 40, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 20, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 30, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 50, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoQueuedRecursive, 8, 0, 1, "Read")
+	appendBlkioStatEntry(&expectedStats.IoQueuedRecursive, 8, 0, 4, "Write")
+	appendBlkioStatEntry(&expectedStats.IoQueuedRecursive, 8, 0, 2, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoQueuedRecursive, 8, 0, 3, "Async")
+	appendBlkioStatEntry(&expectedStats.IoQueuedRecursive, 8, 0, 5, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoServiceTimeRecursive, 8, 0, 173959, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServiceTimeRecursive, 8, 0, 0, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServiceTimeRecursive, 8, 0, 0, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServiceTimeRecursive, 8, 0, 173959, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServiceTimeRecursive, 8, 0, 17395, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoWaitTimeRecursive, 8, 0, 15571, "Read")
+	appendBlkioStatEntry(&expectedStats.IoWaitTimeRecursive, 8, 0, 0, "Write")
+	appendBlkioStatEntry(&expectedStats.IoWaitTimeRecursive, 8, 0, 0, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoWaitTimeRecursive, 8, 0, 15571, "Async")
+	appendBlkioStatEntry(&expectedStats.IoWaitTimeRecursive, 8, 0, 15571, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoMergedRecursive, 8, 0, 5, "Read")
+	appendBlkioStatEntry(&expectedStats.IoMergedRecursive, 8, 0, 10, "Write")
+	appendBlkioStatEntry(&expectedStats.IoMergedRecursive, 8, 0, 0, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoMergedRecursive, 8, 0, 0, "Async")
+	appendBlkioStatEntry(&expectedStats.IoMergedRecursive, 8, 0, 15, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoTimeRecursive, 8, 0, 8, "")
+
+	expectBlkioStatsEquals(t, expectedStats, actualStats.BlkioStats)
+}
+
+func TestBlkioStatsNoSectorsFile(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoServiceBytesFile(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_serviced_recursive":     servicedRecursiveContents,
+		"blkio.io_queued_recursive":       queuedRecursiveContents,
+		"blkio.sectors_recursive":         sectorsRecursiveContents,
+		"blkio.io_service_time_recursive": serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":    waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":       mergedRecursiveContents,
+		"blkio.time_recursive":            timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoServicedFile(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoQueuedFile(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoServiceTimeFile(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping test in short mode.")
+	}
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoWaitTimeFile(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping test in short mode.")
+	}
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoMergedFile(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping test in short mode.")
+	}
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsNoTimeFile(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping test in short mode.")
+	}
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": serviceBytesRecursiveContents,
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatalf("Failed unexpectedly: %s", err)
+	}
+}
+
+func TestBlkioStatsUnexpectedNumberOfFields(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": "8:0 Read 100 100",
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected to fail, but did not")
+	}
+}
+
+func TestBlkioStatsUnexpectedFieldType(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": "8:0 Read Write",
+		"blkio.io_serviced_recursive":      servicedRecursiveContents,
+		"blkio.io_queued_recursive":        queuedRecursiveContents,
+		"blkio.sectors_recursive":          sectorsRecursiveContents,
+		"blkio.io_service_time_recursive":  serviceTimeRecursiveContents,
+		"blkio.io_wait_time_recursive":     waitTimeRecursiveContents,
+		"blkio.io_merged_recursive":        mergedRecursiveContents,
+		"blkio.time_recursive":             timeRecursiveContents,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected to fail, but did not")
+	}
+}
+
+func TestNonCFQBlkioStats(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"blkio.io_service_bytes_recursive": "",
+		"blkio.io_serviced_recursive":      "",
+		"blkio.io_queued_recursive":        "",
+		"blkio.sectors_recursive":          "",
+		"blkio.io_service_time_recursive":  "",
+		"blkio.io_wait_time_recursive":     "",
+		"blkio.io_merged_recursive":        "",
+		"blkio.time_recursive":             "",
+		"blkio.throttle.io_service_bytes":  throttleServiceBytes,
+		"blkio.throttle.io_serviced":       throttleServiced,
+	})
+
+	blkio := &BlkioGroup{}
+	actualStats := *cgroups.NewStats()
+	err := blkio.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Verify expected stats.
+	expectedStats := cgroups.BlkioStats{}
+
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 11030528, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 23, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 42, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 11030528, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 8, 0, 11030528, "Total")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 252, 0, 11030528, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 252, 0, 23, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 252, 0, 42, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 252, 0, 11030528, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServiceBytesRecursive, 252, 0, 11030528, "Total")
+
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 164, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 23, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 42, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 164, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 8, 0, 164, "Total")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 252, 0, 164, "Read")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 252, 0, 23, "Write")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 252, 0, 42, "Sync")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 252, 0, 164, "Async")
+	appendBlkioStatEntry(&expectedStats.IoServicedRecursive, 252, 0, 164, "Total")
+
+	expectBlkioStatsEquals(t, expectedStats, actualStats.BlkioStats)
+}
+
+func TestBlkioSetThrottleReadBpsDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		throttleBefore = `8:0 1024`
+	)
+
+	td := configs.NewThrottleDevice(8, 0, 2048)
+	throttleAfter := td.String()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.throttle.read_bps_device": throttleBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioThrottleReadBpsDevice = []*configs.ThrottleDevice{td}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.throttle.read_bps_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.throttle.read_bps_device - %s", err)
+	}
+
+	if value != throttleAfter {
+		t.Fatal("Got the wrong value, set blkio.throttle.read_bps_device failed.")
+	}
+}
+func TestBlkioSetThrottleWriteBpsDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		throttleBefore = `8:0 1024`
+	)
+
+	td := configs.NewThrottleDevice(8, 0, 2048)
+	throttleAfter := td.String()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.throttle.write_bps_device": throttleBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioThrottleWriteBpsDevice = []*configs.ThrottleDevice{td}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.throttle.write_bps_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.throttle.write_bps_device - %s", err)
+	}
+
+	if value != throttleAfter {
+		t.Fatal("Got the wrong value, set blkio.throttle.write_bps_device failed.")
+	}
+}
+func TestBlkioSetThrottleReadIOpsDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		throttleBefore = `8:0 1024`
+	)
+
+	td := configs.NewThrottleDevice(8, 0, 2048)
+	throttleAfter := td.String()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.throttle.read_iops_device": throttleBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioThrottleReadIOPSDevice = []*configs.ThrottleDevice{td}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.throttle.read_iops_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.throttle.read_iops_device - %s", err)
+	}
+
+	if value != throttleAfter {
+		t.Fatal("Got the wrong value, set blkio.throttle.read_iops_device failed.")
+	}
+}
+func TestBlkioSetThrottleWriteIOpsDevice(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		throttleBefore = `8:0 1024`
+	)
+
+	td := configs.NewThrottleDevice(8, 0, 2048)
+	throttleAfter := td.String()
+
+	helper.writeFileContents(map[string]string{
+		"blkio.throttle.write_iops_device": throttleBefore,
+	})
+
+	helper.CgroupData.config.Resources.BlkioThrottleWriteIOPSDevice = []*configs.ThrottleDevice{td}
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "blkio.throttle.write_iops_device")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.throttle.write_iops_device - %s", err)
+	}
+
+	if value != throttleAfter {
+		t.Fatal("Got the wrong value, set blkio.throttle.write_iops_device failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/cpu.go

@@ -22,21 +22,59 @@ func (s *CpuGroup) Name() string {
 func (s *CpuGroup) Apply(d *cgroupData) error {
 	// We always want to join the cpu group, to allow fair cpu scheduling
 	// on a container basis
-	_, err := d.join("cpu")
+	path, err := d.path("cpu")
 	if err != nil && !cgroups.IsNotFound(err) {
 		return err
 	}
+	return s.ApplyDir(path, d.config, d.pid)
+}
+
+func (s *CpuGroup) ApplyDir(path string, cgroup *configs.Cgroup, pid int) error {
+	// This might happen if we have no cpu cgroup mounted.
+	// Just do nothing and don't fail.
+	if path == "" {
+		return nil
+	}
+	if err := os.MkdirAll(path, 0755); err != nil {
+		return err
+	}
+	// We should set the real-Time group scheduling settings before moving
+	// in the process because if the process is already in SCHED_RR mode
+	// and no RT bandwidth is set, adding it will fail.
+	if err := s.SetRtSched(path, cgroup); err != nil {
+		return err
+	}
+	// because we are not using d.join we need to place the pid into the procs file
+	// unlike the other subsystems
+	if err := cgroups.WriteCgroupProc(path, pid); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
+	if cgroup.Resources.CpuRtPeriod != 0 {
+		if err := writeFile(path, "cpu.rt_period_us", strconv.FormatUint(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
+			return err
+		}
+	}
+	if cgroup.Resources.CpuRtRuntime != 0 {
+		if err := writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil {
+			return err
+		}
+	}
 	return nil
 }
 
 func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
 	if cgroup.Resources.CpuShares != 0 {
-		if err := writeFile(path, "cpu.shares", strconv.FormatInt(cgroup.Resources.CpuShares, 10)); err != nil {
+		if err := writeFile(path, "cpu.shares", strconv.FormatUint(cgroup.Resources.CpuShares, 10)); err != nil {
 			return err
 		}
 	}
 	if cgroup.Resources.CpuPeriod != 0 {
-		if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatInt(cgroup.Resources.CpuPeriod, 10)); err != nil {
+		if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatUint(cgroup.Resources.CpuPeriod, 10)); err != nil {
 			return err
 		}
 	}
@@ -45,15 +83,8 @@ func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
 			return err
 		}
 	}
-	if cgroup.Resources.CpuRtPeriod != 0 {
-		if err := writeFile(path, "cpu.rt_period_us", strconv.FormatInt(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
-			return err
-		}
-	}
-	if cgroup.Resources.CpuRtRuntime != 0 {
-		if err := writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil {
-			return err
-		}
+	if err := s.SetRtSched(path, cgroup); err != nil {
+		return err
 	}
 
 	return nil

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/cpu_test.go

@@ -0,0 +1,209 @@
+// +build linux
+
+package fs
+
+import (
+	"fmt"
+	"strconv"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+)
+
+func TestCpuSetShares(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		sharesBefore = 1024
+		sharesAfter  = 512
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpu.shares": strconv.Itoa(sharesBefore),
+	})
+
+	helper.CgroupData.config.Resources.CpuShares = sharesAfter
+	cpu := &CpuGroup{}
+	if err := cpu.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "cpu.shares")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.shares - %s", err)
+	}
+
+	if value != sharesAfter {
+		t.Fatal("Got the wrong value, set cpu.shares failed.")
+	}
+}
+
+func TestCpuSetBandWidth(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		quotaBefore     = 8000
+		quotaAfter      = 5000
+		periodBefore    = 10000
+		periodAfter     = 7000
+		rtRuntimeBefore = 8000
+		rtRuntimeAfter  = 5000
+		rtPeriodBefore  = 10000
+		rtPeriodAfter   = 7000
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpu.cfs_quota_us":  strconv.Itoa(quotaBefore),
+		"cpu.cfs_period_us": strconv.Itoa(periodBefore),
+		"cpu.rt_runtime_us": strconv.Itoa(rtRuntimeBefore),
+		"cpu.rt_period_us":  strconv.Itoa(rtPeriodBefore),
+	})
+
+	helper.CgroupData.config.Resources.CpuQuota = quotaAfter
+	helper.CgroupData.config.Resources.CpuPeriod = periodAfter
+	helper.CgroupData.config.Resources.CpuRtRuntime = rtRuntimeAfter
+	helper.CgroupData.config.Resources.CpuRtPeriod = rtPeriodAfter
+	cpu := &CpuGroup{}
+	if err := cpu.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	quota, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_quota_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.cfs_quota_us - %s", err)
+	}
+	if quota != quotaAfter {
+		t.Fatal("Got the wrong value, set cpu.cfs_quota_us failed.")
+	}
+
+	period, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_period_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.cfs_period_us - %s", err)
+	}
+	if period != periodAfter {
+		t.Fatal("Got the wrong value, set cpu.cfs_period_us failed.")
+	}
+	rtRuntime, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_runtime_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.rt_runtime_us - %s", err)
+	}
+	if rtRuntime != rtRuntimeAfter {
+		t.Fatal("Got the wrong value, set cpu.rt_runtime_us failed.")
+	}
+	rtPeriod, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_period_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.rt_period_us - %s", err)
+	}
+	if rtPeriod != rtPeriodAfter {
+		t.Fatal("Got the wrong value, set cpu.rt_period_us failed.")
+	}
+}
+
+func TestCpuStats(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		nrPeriods     = 2000
+		nrThrottled   = 200
+		throttledTime = uint64(18446744073709551615)
+	)
+
+	cpuStatContent := fmt.Sprintf("nr_periods %d\n nr_throttled %d\n throttled_time %d\n",
+		nrPeriods, nrThrottled, throttledTime)
+	helper.writeFileContents(map[string]string{
+		"cpu.stat": cpuStatContent,
+	})
+
+	cpu := &CpuGroup{}
+	actualStats := *cgroups.NewStats()
+	err := cpu.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	expectedStats := cgroups.ThrottlingData{
+		Periods:          nrPeriods,
+		ThrottledPeriods: nrThrottled,
+		ThrottledTime:    throttledTime}
+
+	expectThrottlingDataEquals(t, expectedStats, actualStats.CpuStats.ThrottlingData)
+}
+
+func TestNoCpuStatFile(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	cpu := &CpuGroup{}
+	actualStats := *cgroups.NewStats()
+	err := cpu.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal("Expected not to fail, but did")
+	}
+}
+
+func TestInvalidCpuStat(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+	cpuStatContent := `nr_periods 2000
+	nr_throttled 200
+	throttled_time fortytwo`
+	helper.writeFileContents(map[string]string{
+		"cpu.stat": cpuStatContent,
+	})
+
+	cpu := &CpuGroup{}
+	actualStats := *cgroups.NewStats()
+	err := cpu.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failed stat parsing.")
+	}
+}
+
+func TestCpuSetRtSchedAtApply(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		rtRuntimeBefore = 0
+		rtRuntimeAfter  = 5000
+		rtPeriodBefore  = 0
+		rtPeriodAfter   = 7000
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpu.rt_runtime_us": strconv.Itoa(rtRuntimeBefore),
+		"cpu.rt_period_us":  strconv.Itoa(rtPeriodBefore),
+	})
+
+	helper.CgroupData.config.Resources.CpuRtRuntime = rtRuntimeAfter
+	helper.CgroupData.config.Resources.CpuRtPeriod = rtPeriodAfter
+	cpu := &CpuGroup{}
+	if err := cpu.ApplyDir(helper.CgroupPath, helper.CgroupData.config, 1234); err != nil {
+		t.Fatal(err)
+	}
+
+	rtRuntime, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_runtime_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.rt_runtime_us - %s", err)
+	}
+	if rtRuntime != rtRuntimeAfter {
+		t.Fatal("Got the wrong value, set cpu.rt_runtime_us failed.")
+	}
+	rtPeriod, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_period_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.rt_period_us - %s", err)
+	}
+	if rtPeriod != rtPeriodAfter {
+		t.Fatal("Got the wrong value, set cpu.rt_period_us failed.")
+	}
+	pid, err := getCgroupParamUint(helper.CgroupPath, "cgroup.procs")
+	if err != nil {
+		t.Fatalf("Failed to parse cgroup.procs - %s", err)
+	}
+	if pid != 1234 {
+		t.Fatal("Got the wrong value, set cgroup.procs failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/cpuset.go

@@ -8,7 +8,6 @@ import (
 	"io/ioutil"
 	"os"
 	"path/filepath"
-	"strconv"
 
 	"github.com/opencontainers/runc/libcontainer/cgroups"
 	"github.com/opencontainers/runc/libcontainer/configs"
@@ -58,16 +57,34 @@ func (s *CpusetGroup) ApplyDir(dir string, cgroup *configs.Cgroup, pid int) erro
 	if dir == "" {
 		return nil
 	}
-	root, err := getCgroupRoot()
+	mountInfo, err := ioutil.ReadFile("/proc/self/mountinfo")
 	if err != nil {
 		return err
 	}
-	if err := s.ensureParent(dir, root); err != nil {
+	root := filepath.Dir(cgroups.GetClosestMountpointAncestor(dir, string(mountInfo)))
+	// 'ensureParent' start with parent because we don't want to
+	// explicitly inherit from parent, it could conflict with
+	// 'cpuset.cpu_exclusive'.
+	if err := s.ensureParent(filepath.Dir(dir), root); err != nil {
 		return err
 	}
+	if err := os.MkdirAll(dir, 0755); err != nil {
+		return err
+	}
+	// We didn't inherit cpuset configs from parent, but we have
+	// to ensure cpuset configs are set before moving task into the
+	// cgroup.
+	// The logic is, if user specified cpuset configs, use these
+	// specified configs, otherwise, inherit from parent. This makes
+	// cpuset configs work correctly with 'cpuset.cpu_exclusive', and
+	// keep backward compatbility.
+	if err := s.ensureCpusAndMems(dir, cgroup); err != nil {
+		return err
+	}
+
 	// because we are not using d.join we need to place the pid into the procs file
 	// unlike the other subsystems
-	if err := writeFile(dir, "cgroup.procs", strconv.Itoa(pid)); err != nil {
+	if err := cgroups.WriteCgroupProc(dir, pid); err != nil {
 		return err
 	}
 
@@ -137,3 +154,10 @@ func (s *CpusetGroup) copyIfNeeded(current, parent string) error {
 func (s *CpusetGroup) isEmpty(b []byte) bool {
 	return len(bytes.Trim(b, "\n")) == 0
 }
+
+func (s *CpusetGroup) ensureCpusAndMems(path string, cgroup *configs.Cgroup) error {
+	if err := s.Set(path, cgroup); err != nil {
+		return err
+	}
+	return s.copyIfNeeded(path, filepath.Dir(path))
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/cpuset_test.go

@@ -0,0 +1,65 @@
+// +build linux
+
+package fs
+
+import (
+	"testing"
+)
+
+func TestCpusetSetCpus(t *testing.T) {
+	helper := NewCgroupTestUtil("cpuset", t)
+	defer helper.cleanup()
+
+	const (
+		cpusBefore = "0"
+		cpusAfter  = "1-3"
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpuset.cpus": cpusBefore,
+	})
+
+	helper.CgroupData.config.Resources.CpusetCpus = cpusAfter
+	cpuset := &CpusetGroup{}
+	if err := cpuset.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "cpuset.cpus")
+	if err != nil {
+		t.Fatalf("Failed to parse cpuset.cpus - %s", err)
+	}
+
+	if value != cpusAfter {
+		t.Fatal("Got the wrong value, set cpuset.cpus failed.")
+	}
+}
+
+func TestCpusetSetMems(t *testing.T) {
+	helper := NewCgroupTestUtil("cpuset", t)
+	defer helper.cleanup()
+
+	const (
+		memsBefore = "0"
+		memsAfter  = "1"
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpuset.mems": memsBefore,
+	})
+
+	helper.CgroupData.config.Resources.CpusetMems = memsAfter
+	cpuset := &CpusetGroup{}
+	if err := cpuset.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "cpuset.mems")
+	if err != nil {
+		t.Fatalf("Failed to parse cpuset.mems - %s", err)
+	}
+
+	if value != memsAfter {
+		t.Fatal("Got the wrong value, set cpuset.mems failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/devices.go

@@ -43,21 +43,23 @@ func (s *DevicesGroup) Set(path string, cgroup *configs.Cgroup) error {
 		}
 		return nil
 	}
-	if !cgroup.Resources.AllowAllDevices {
-		if err := writeFile(path, "devices.deny", "a"); err != nil {
-			return err
-		}
-
-		for _, dev := range cgroup.Resources.AllowedDevices {
-			if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
+	if cgroup.Resources.AllowAllDevices != nil {
+		if *cgroup.Resources.AllowAllDevices == false {
+			if err := writeFile(path, "devices.deny", "a"); err != nil {
 				return err
 			}
-		}
-		return nil
-	}
 
-	if err := writeFile(path, "devices.allow", "a"); err != nil {
-		return err
+			for _, dev := range cgroup.Resources.AllowedDevices {
+				if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+
+		if err := writeFile(path, "devices.allow", "a"); err != nil {
+			return err
+		}
 	}
 
 	for _, dev := range cgroup.Resources.DeniedDevices {

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/devices_test.go

@@ -0,0 +1,98 @@
+// +build linux
+
+package fs
+
+import (
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var (
+	allowedDevices = []*configs.Device{
+		{
+			Path:        "/dev/zero",
+			Type:        'c',
+			Major:       1,
+			Minor:       5,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+	}
+	allowedList   = "c 1:5 rwm"
+	deniedDevices = []*configs.Device{
+		{
+			Path:        "/dev/null",
+			Type:        'c',
+			Major:       1,
+			Minor:       3,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+	}
+	deniedList = "c 1:3 rwm"
+)
+
+func TestDevicesSetAllow(t *testing.T) {
+	helper := NewCgroupTestUtil("devices", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"devices.deny": "a",
+	})
+	allowAllDevices := false
+	helper.CgroupData.config.Resources.AllowAllDevices = &allowAllDevices
+	helper.CgroupData.config.Resources.AllowedDevices = allowedDevices
+	devices := &DevicesGroup{}
+	if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "devices.allow")
+	if err != nil {
+		t.Fatalf("Failed to parse devices.allow - %s", err)
+	}
+
+	if value != allowedList {
+		t.Fatal("Got the wrong value, set devices.allow failed.")
+	}
+
+	// When AllowAllDevices is nil, devices.allow file should not be modified.
+	helper.CgroupData.config.Resources.AllowAllDevices = nil
+	if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+	value, err = getCgroupParamString(helper.CgroupPath, "devices.allow")
+	if err != nil {
+		t.Fatalf("Failed to parse devices.allow - %s", err)
+	}
+	if value != allowedList {
+		t.Fatal("devices policy shouldn't have changed on AllowedAllDevices=nil.")
+	}
+}
+
+func TestDevicesSetDeny(t *testing.T) {
+	helper := NewCgroupTestUtil("devices", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"devices.allow": "a",
+	})
+
+	allowAllDevices := true
+	helper.CgroupData.config.Resources.AllowAllDevices = &allowAllDevices
+	helper.CgroupData.config.Resources.DeniedDevices = deniedDevices
+	devices := &DevicesGroup{}
+	if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "devices.deny")
+	if err != nil {
+		t.Fatalf("Failed to parse devices.deny - %s", err)
+	}
+
+	if value != deniedList {
+		t.Fatal("Got the wrong value, set devices.deny failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/freezer.go

@@ -29,11 +29,15 @@ func (s *FreezerGroup) Apply(d *cgroupData) error {
 func (s *FreezerGroup) Set(path string, cgroup *configs.Cgroup) error {
 	switch cgroup.Resources.Freezer {
 	case configs.Frozen, configs.Thawed:
-		if err := writeFile(path, "freezer.state", string(cgroup.Resources.Freezer)); err != nil {
-			return err
-		}
-
 		for {
+			// In case this loop does not exit because it doesn't get the expected
+			// state, let's write again this state, hoping it's going to be properly
+			// set this time. Otherwise, this loop could run infinitely, waiting for
+			// a state change that would never happen.
+			if err := writeFile(path, "freezer.state", string(cgroup.Resources.Freezer)); err != nil {
+				return err
+			}
+
 			state, err := readFile(path, "freezer.state")
 			if err != nil {
 				return err
@@ -41,6 +45,7 @@ func (s *FreezerGroup) Set(path string, cgroup *configs.Cgroup) error {
 			if strings.TrimSpace(state) == string(cgroup.Resources.Freezer) {
 				break
 			}
+
 			time.Sleep(1 * time.Millisecond)
 		}
 	case configs.Undefined:

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/freezer_test.go

@@ -0,0 +1,47 @@
+// +build linux
+
+package fs
+
+import (
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+func TestFreezerSetState(t *testing.T) {
+	helper := NewCgroupTestUtil("freezer", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"freezer.state": string(configs.Frozen),
+	})
+
+	helper.CgroupData.config.Resources.Freezer = configs.Thawed
+	freezer := &FreezerGroup{}
+	if err := freezer.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "freezer.state")
+	if err != nil {
+		t.Fatalf("Failed to parse freezer.state - %s", err)
+	}
+	if value != string(configs.Thawed) {
+		t.Fatal("Got the wrong value, set freezer.state failed.")
+	}
+}
+
+func TestFreezerSetInvalidState(t *testing.T) {
+	helper := NewCgroupTestUtil("freezer", t)
+	defer helper.cleanup()
+
+	const (
+		invalidArg configs.FreezerState = "Invalid"
+	)
+
+	helper.CgroupData.config.Resources.Freezer = invalidArg
+	freezer := &FreezerGroup{}
+	if err := freezer.Set(helper.CgroupPath, helper.CgroupData.config); err == nil {
+		t.Fatal("Failed to return invalid argument error")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/hugetlb_test.go

@@ -0,0 +1,154 @@
+// +build linux
+
+package fs
+
+import (
+	"fmt"
+	"strconv"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+const (
+	hugetlbUsageContents    = "128\n"
+	hugetlbMaxUsageContents = "256\n"
+	hugetlbFailcnt          = "100\n"
+)
+
+var (
+	usage    = "hugetlb.%s.usage_in_bytes"
+	limit    = "hugetlb.%s.limit_in_bytes"
+	maxUsage = "hugetlb.%s.max_usage_in_bytes"
+	failcnt  = "hugetlb.%s.failcnt"
+)
+
+func TestHugetlbSetHugetlb(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+
+	const (
+		hugetlbBefore = 256
+		hugetlbAfter  = 512
+	)
+
+	for _, pageSize := range HugePageSizes {
+		helper.writeFileContents(map[string]string{
+			fmt.Sprintf(limit, pageSize): strconv.Itoa(hugetlbBefore),
+		})
+	}
+
+	for _, pageSize := range HugePageSizes {
+		helper.CgroupData.config.Resources.HugetlbLimit = []*configs.HugepageLimit{
+			{
+				Pagesize: pageSize,
+				Limit:    hugetlbAfter,
+			},
+		}
+		hugetlb := &HugetlbGroup{}
+		if err := hugetlb.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	for _, pageSize := range HugePageSizes {
+		limit := fmt.Sprintf(limit, pageSize)
+		value, err := getCgroupParamUint(helper.CgroupPath, limit)
+		if err != nil {
+			t.Fatalf("Failed to parse %s - %s", limit, err)
+		}
+		if value != hugetlbAfter {
+			t.Fatalf("Set hugetlb.limit_in_bytes failed. Expected: %v, Got: %v", hugetlbAfter, value)
+		}
+	}
+}
+
+func TestHugetlbStats(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+	for _, pageSize := range HugePageSizes {
+		helper.writeFileContents(map[string]string{
+			fmt.Sprintf(usage, pageSize):    hugetlbUsageContents,
+			fmt.Sprintf(maxUsage, pageSize): hugetlbMaxUsageContents,
+			fmt.Sprintf(failcnt, pageSize):  hugetlbFailcnt,
+		})
+	}
+
+	hugetlb := &HugetlbGroup{}
+	actualStats := *cgroups.NewStats()
+	err := hugetlb.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedStats := cgroups.HugetlbStats{Usage: 128, MaxUsage: 256, Failcnt: 100}
+	for _, pageSize := range HugePageSizes {
+		expectHugetlbStatEquals(t, expectedStats, actualStats.HugetlbStats[pageSize])
+	}
+}
+
+func TestHugetlbStatsNoUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		maxUsage: hugetlbMaxUsageContents,
+	})
+
+	hugetlb := &HugetlbGroup{}
+	actualStats := *cgroups.NewStats()
+	err := hugetlb.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestHugetlbStatsNoMaxUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+	for _, pageSize := range HugePageSizes {
+		helper.writeFileContents(map[string]string{
+			fmt.Sprintf(usage, pageSize): hugetlbUsageContents,
+		})
+	}
+
+	hugetlb := &HugetlbGroup{}
+	actualStats := *cgroups.NewStats()
+	err := hugetlb.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestHugetlbStatsBadUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+	for _, pageSize := range HugePageSizes {
+		helper.writeFileContents(map[string]string{
+			fmt.Sprintf(usage, pageSize): "bad",
+			maxUsage:                     hugetlbMaxUsageContents,
+		})
+	}
+
+	hugetlb := &HugetlbGroup{}
+	actualStats := *cgroups.NewStats()
+	err := hugetlb.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestHugetlbStatsBadMaxUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("hugetlb", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		usage:    hugetlbUsageContents,
+		maxUsage: "bad",
+	})
+
+	hugetlb := &HugetlbGroup{}
+	actualStats := *cgroups.NewStats()
+	err := hugetlb.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/memory.go

@@ -5,13 +5,23 @@ package fs
 import (
 	"bufio"
 	"fmt"
+	"io/ioutil"
 	"os"
 	"path/filepath"
 	"strconv"
 	"strings"
+	"syscall" // only for Errno
 
 	"github.com/opencontainers/runc/libcontainer/cgroups"
 	"github.com/opencontainers/runc/libcontainer/configs"
+
+	"golang.org/x/sys/unix"
+)
+
+const (
+	cgroupKernelMemoryLimit = "memory.kmem.limit_in_bytes"
+	cgroupMemorySwapLimit   = "memory.memsw.limit_in_bytes"
+	cgroupMemoryLimit       = "memory.limit_in_bytes"
 )
 
 type MemoryGroup struct {
@@ -25,20 +35,23 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
 	path, err := d.path("memory")
 	if err != nil && !cgroups.IsNotFound(err) {
 		return err
+	} else if path == "" {
+		return nil
 	}
 	if memoryAssigned(d.config) {
-		if path != "" {
+		if _, err := os.Stat(path); os.IsNotExist(err) {
 			if err := os.MkdirAll(path, 0755); err != nil {
 				return err
 			}
-		}
-		// We have to set kernel memory here, as we can't change it once
-		// processes have been attached.
-		if err := s.SetKernelMemory(path, d.config); err != nil {
-			return err
+			// Only enable kernel memory accouting when this cgroup
+			// is created by libcontainer, otherwise we might get
+			// error when people use `cgroupsPath` to join an existed
+			// cgroup whose kernel memory is not initialized.
+			if err := EnableKernelMemoryAccounting(path); err != nil {
+				return err
+			}
 		}
 	}
-
 	defer func() {
 		if err != nil {
 			os.RemoveAll(path)
@@ -54,21 +67,57 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
 	return nil
 }
 
-func (s *MemoryGroup) SetKernelMemory(path string, cgroup *configs.Cgroup) error {
-	// This has to be done separately because it has special constraints (it
-	// can't be done after there are processes attached to the cgroup).
-	if cgroup.Resources.KernelMemory > 0 {
-		if err := writeFile(path, "memory.kmem.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemory, 10)); err != nil {
+func EnableKernelMemoryAccounting(path string) error {
+	// Check if kernel memory is enabled
+	// We have to limit the kernel memory here as it won't be accounted at all
+	// until a limit is set on the cgroup and limit cannot be set once the
+	// cgroup has children, or if there are already tasks in the cgroup.
+	for _, i := range []int64{1, -1} {
+		if err := setKernelMemory(path, i); err != nil {
 			return err
 		}
 	}
 	return nil
 }
 
+func setKernelMemory(path string, kernelMemoryLimit int64) error {
+	if path == "" {
+		return fmt.Errorf("no such directory for %s", cgroupKernelMemoryLimit)
+	}
+	if !cgroups.PathExists(filepath.Join(path, cgroupKernelMemoryLimit)) {
+		// kernel memory is not enabled on the system so we should do nothing
+		return nil
+	}
+	if err := ioutil.WriteFile(filepath.Join(path, cgroupKernelMemoryLimit), []byte(strconv.FormatInt(kernelMemoryLimit, 10)), 0700); err != nil {
+		// Check if the error number returned by the syscall is "EBUSY"
+		// The EBUSY signal is returned on attempts to write to the
+		// memory.kmem.limit_in_bytes file if the cgroup has children or
+		// once tasks have been attached to the cgroup
+		if pathErr, ok := err.(*os.PathError); ok {
+			if errNo, ok := pathErr.Err.(syscall.Errno); ok {
+				if errNo == unix.EBUSY {
+					return fmt.Errorf("failed to set %s, because either tasks have already joined this cgroup or it has children", cgroupKernelMemoryLimit)
+				}
+			}
+		}
+		return fmt.Errorf("failed to write %v to %v: %v", kernelMemoryLimit, cgroupKernelMemoryLimit, err)
+	}
+	return nil
+}
+
 func setMemoryAndSwap(path string, cgroup *configs.Cgroup) error {
+	// If the memory update is set to -1 we should also
+	// set swap to -1, it means unlimited memory.
+	if cgroup.Resources.Memory == -1 {
+		// Only set swap if it's enabled in kernel
+		if cgroups.PathExists(filepath.Join(path, cgroupMemorySwapLimit)) {
+			cgroup.Resources.MemorySwap = -1
+		}
+	}
+
 	// When memory and swap memory are both set, we need to handle the cases
 	// for updating container.
-	if cgroup.Resources.Memory != 0 && cgroup.Resources.MemorySwap > 0 {
+	if cgroup.Resources.Memory != 0 && cgroup.Resources.MemorySwap != 0 {
 		memoryUsage, err := getMemoryData(path, "")
 		if err != nil {
 			return err
@@ -77,29 +126,29 @@ func setMemoryAndSwap(path string, cgroup *configs.Cgroup) error {
 		// When update memory limit, we should adapt the write sequence
 		// for memory and swap memory, so it won't fail because the new
 		// value and the old value don't fit kernel's validation.
-		if memoryUsage.Limit < uint64(cgroup.Resources.MemorySwap) {
-			if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
+		if cgroup.Resources.MemorySwap == -1 || memoryUsage.Limit < uint64(cgroup.Resources.MemorySwap) {
+			if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
 				return err
 			}
-			if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
+			if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
 				return err
 			}
 		} else {
-			if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
+			if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
 				return err
 			}
-			if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
+			if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
 				return err
 			}
 		}
 	} else {
 		if cgroup.Resources.Memory != 0 {
-			if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
+			if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
 				return err
 			}
 		}
-		if cgroup.Resources.MemorySwap > 0 {
-			if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
+		if cgroup.Resources.MemorySwap != 0 {
+			if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
 				return err
 			}
 		}
@@ -113,11 +162,18 @@ func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
 		return err
 	}
 
+	if cgroup.Resources.KernelMemory != 0 {
+		if err := setKernelMemory(path, cgroup.Resources.KernelMemory); err != nil {
+			return err
+		}
+	}
+
 	if cgroup.Resources.MemoryReservation != 0 {
 		if err := writeFile(path, "memory.soft_limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemoryReservation, 10)); err != nil {
 			return err
 		}
 	}
+
 	if cgroup.Resources.KernelMemoryTCP != 0 {
 		if err := writeFile(path, "memory.kmem.tcp.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemoryTCP, 10)); err != nil {
 			return err
@@ -130,12 +186,12 @@ func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
 	}
 	if cgroup.Resources.MemorySwappiness == nil || int64(*cgroup.Resources.MemorySwappiness) == -1 {
 		return nil
-	} else if int64(*cgroup.Resources.MemorySwappiness) >= 0 && int64(*cgroup.Resources.MemorySwappiness) <= 100 {
-		if err := writeFile(path, "memory.swappiness", strconv.FormatInt(*cgroup.Resources.MemorySwappiness, 10)); err != nil {
+	} else if *cgroup.Resources.MemorySwappiness <= 100 {
+		if err := writeFile(path, "memory.swappiness", strconv.FormatUint(*cgroup.Resources.MemorySwappiness, 10)); err != nil {
 			return err
 		}
 	} else {
-		return fmt.Errorf("invalid value:%d. valid memory swappiness range is 0-100", int64(*cgroup.Resources.MemorySwappiness))
+		return fmt.Errorf("invalid value:%d. valid memory swappiness range is 0-100", *cgroup.Resources.MemorySwappiness)
 	}
 
 	return nil
@@ -187,6 +243,14 @@ func (s *MemoryGroup) GetStats(path string, stats *cgroups.Stats) error {
 	}
 	stats.MemoryStats.KernelTCPUsage = kernelTCPUsage
 
+	useHierarchy := strings.Join([]string{"memory", "use_hierarchy"}, ".")
+	value, err := getCgroupParamUint(path, useHierarchy)
+	if err != nil {
+		return err
+	}
+	if value == 1 {
+		stats.MemoryStats.UseHierarchy = true
+	}
 	return nil
 }
 
@@ -197,7 +261,7 @@ func memoryAssigned(cgroup *configs.Cgroup) bool {
 		cgroup.Resources.KernelMemory > 0 ||
 		cgroup.Resources.KernelMemoryTCP > 0 ||
 		cgroup.Resources.OomKillDisable ||
-		(cgroup.Resources.MemorySwappiness != nil && *cgroup.Resources.MemorySwappiness != -1)
+		(cgroup.Resources.MemorySwappiness != nil && int64(*cgroup.Resources.MemorySwappiness) != -1)
 }
 
 func getMemoryData(path, name string) (cgroups.MemoryData, error) {

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/memory_test.go

@@ -0,0 +1,455 @@
+// +build linux
+
+package fs
+
+import (
+	"strconv"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+)
+
+const (
+	memoryStatContents = `cache 512
+rss 1024`
+	memoryUsageContents        = "2048\n"
+	memoryMaxUsageContents     = "4096\n"
+	memoryFailcnt              = "100\n"
+	memoryLimitContents        = "8192\n"
+	memoryUseHierarchyContents = "1\n"
+)
+
+func TestMemorySetMemory(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		memoryBefore      = 314572800 // 300M
+		memoryAfter       = 524288000 // 500M
+		reservationBefore = 209715200 // 200M
+		reservationAfter  = 314572800 // 300M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.limit_in_bytes":      strconv.Itoa(memoryBefore),
+		"memory.soft_limit_in_bytes": strconv.Itoa(reservationBefore),
+	})
+
+	helper.CgroupData.config.Resources.Memory = memoryAfter
+	helper.CgroupData.config.Resources.MemoryReservation = reservationAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
+	}
+	if value != memoryAfter {
+		t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
+	}
+
+	value, err = getCgroupParamUint(helper.CgroupPath, "memory.soft_limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.soft_limit_in_bytes - %s", err)
+	}
+	if value != reservationAfter {
+		t.Fatal("Got the wrong value, set memory.soft_limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetMemoryswap(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		memoryswapBefore = 314572800 // 300M
+		memoryswapAfter  = 524288000 // 500M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.memsw.limit_in_bytes": strconv.Itoa(memoryswapBefore),
+	})
+
+	helper.CgroupData.config.Resources.MemorySwap = memoryswapAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.memsw.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.memsw.limit_in_bytes - %s", err)
+	}
+	if value != memoryswapAfter {
+		t.Fatal("Got the wrong value, set memory.memsw.limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetMemoryLargerThanSwap(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		memoryBefore     = 314572800 // 300M
+		memoryswapBefore = 524288000 // 500M
+		memoryAfter      = 629145600 // 600M
+		memoryswapAfter  = 838860800 // 800M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.limit_in_bytes":       strconv.Itoa(memoryBefore),
+		"memory.memsw.limit_in_bytes": strconv.Itoa(memoryswapBefore),
+		// Set will call getMemoryData when memory and swap memory are
+		// both set, fake these fields so we don't get error.
+		"memory.usage_in_bytes":     "0",
+		"memory.max_usage_in_bytes": "0",
+		"memory.failcnt":            "0",
+	})
+
+	helper.CgroupData.config.Resources.Memory = memoryAfter
+	helper.CgroupData.config.Resources.MemorySwap = memoryswapAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
+	}
+	if value != memoryAfter {
+		t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
+	}
+	value, err = getCgroupParamUint(helper.CgroupPath, "memory.memsw.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.memsw.limit_in_bytes - %s", err)
+	}
+	if value != memoryswapAfter {
+		t.Fatal("Got the wrong value, set memory.memsw.limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetSwapSmallerThanMemory(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		memoryBefore     = 629145600 // 600M
+		memoryswapBefore = 838860800 // 800M
+		memoryAfter      = 314572800 // 300M
+		memoryswapAfter  = 524288000 // 500M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.limit_in_bytes":       strconv.Itoa(memoryBefore),
+		"memory.memsw.limit_in_bytes": strconv.Itoa(memoryswapBefore),
+		// Set will call getMemoryData when memory and swap memory are
+		// both set, fake these fields so we don't get error.
+		"memory.usage_in_bytes":     "0",
+		"memory.max_usage_in_bytes": "0",
+		"memory.failcnt":            "0",
+	})
+
+	helper.CgroupData.config.Resources.Memory = memoryAfter
+	helper.CgroupData.config.Resources.MemorySwap = memoryswapAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
+	}
+	if value != memoryAfter {
+		t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
+	}
+	value, err = getCgroupParamUint(helper.CgroupPath, "memory.memsw.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.memsw.limit_in_bytes - %s", err)
+	}
+	if value != memoryswapAfter {
+		t.Fatal("Got the wrong value, set memory.memsw.limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetKernelMemory(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		kernelMemoryBefore = 314572800 // 300M
+		kernelMemoryAfter  = 524288000 // 500M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.kmem.limit_in_bytes": strconv.Itoa(kernelMemoryBefore),
+	})
+
+	helper.CgroupData.config.Resources.KernelMemory = kernelMemoryAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.kmem.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.kmem.limit_in_bytes - %s", err)
+	}
+	if value != kernelMemoryAfter {
+		t.Fatal("Got the wrong value, set memory.kmem.limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetKernelMemoryTCP(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		kernelMemoryTCPBefore = 314572800 // 300M
+		kernelMemoryTCPAfter  = 524288000 // 500M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.kmem.tcp.limit_in_bytes": strconv.Itoa(kernelMemoryTCPBefore),
+	})
+
+	helper.CgroupData.config.Resources.KernelMemoryTCP = kernelMemoryTCPAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.kmem.tcp.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.kmem.tcp.limit_in_bytes - %s", err)
+	}
+	if value != kernelMemoryTCPAfter {
+		t.Fatal("Got the wrong value, set memory.kmem.tcp.limit_in_bytes failed.")
+	}
+}
+
+func TestMemorySetMemorySwappinessDefault(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	swappinessBefore := 60 //default is 60
+	swappinessAfter := uint64(0)
+
+	helper.writeFileContents(map[string]string{
+		"memory.swappiness": strconv.Itoa(swappinessBefore),
+	})
+
+	helper.CgroupData.config.Resources.MemorySwappiness = &swappinessAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.swappiness")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.swappiness - %s", err)
+	}
+	if value != swappinessAfter {
+		t.Fatalf("Got the wrong value (%d), set memory.swappiness = %d failed.", value, swappinessAfter)
+	}
+}
+
+func TestMemoryStats(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":                     memoryStatContents,
+		"memory.usage_in_bytes":           memoryUsageContents,
+		"memory.limit_in_bytes":           memoryLimitContents,
+		"memory.max_usage_in_bytes":       memoryMaxUsageContents,
+		"memory.failcnt":                  memoryFailcnt,
+		"memory.memsw.usage_in_bytes":     memoryUsageContents,
+		"memory.memsw.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.memsw.failcnt":            memoryFailcnt,
+		"memory.memsw.limit_in_bytes":     memoryLimitContents,
+		"memory.kmem.usage_in_bytes":      memoryUsageContents,
+		"memory.kmem.max_usage_in_bytes":  memoryMaxUsageContents,
+		"memory.kmem.failcnt":             memoryFailcnt,
+		"memory.kmem.limit_in_bytes":      memoryLimitContents,
+		"memory.use_hierarchy":            memoryUseHierarchyContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedStats := cgroups.MemoryStats{Cache: 512, Usage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, SwapUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, KernelUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, Stats: map[string]uint64{"cache": 512, "rss": 1024}, UseHierarchy: true}
+	expectMemoryStatEquals(t, expectedStats, actualStats.MemoryStats)
+}
+
+func TestMemoryStatsNoStatFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.usage_in_bytes":     memoryUsageContents,
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.limit_in_bytes":     memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err != nil {
+		t.Fatal(err)
+	}
+}
+
+func TestMemoryStatsNoUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               memoryStatContents,
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.limit_in_bytes":     memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsNoMaxUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":           memoryStatContents,
+		"memory.usage_in_bytes": memoryUsageContents,
+		"memory.limit_in_bytes": memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsNoLimitInBytesFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               memoryStatContents,
+		"memory.usage_in_bytes":     memoryUsageContents,
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsBadStatFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               "rss rss",
+		"memory.usage_in_bytes":     memoryUsageContents,
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.limit_in_bytes":     memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsBadUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               memoryStatContents,
+		"memory.usage_in_bytes":     "bad",
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.limit_in_bytes":     memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsBadMaxUsageFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               memoryStatContents,
+		"memory.usage_in_bytes":     memoryUsageContents,
+		"memory.max_usage_in_bytes": "bad",
+		"memory.limit_in_bytes":     memoryLimitContents,
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemoryStatsBadLimitInBytesFile(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+	helper.writeFileContents(map[string]string{
+		"memory.stat":               memoryStatContents,
+		"memory.usage_in_bytes":     memoryUsageContents,
+		"memory.max_usage_in_bytes": memoryMaxUsageContents,
+		"memory.limit_in_bytes":     "bad",
+	})
+
+	memory := &MemoryGroup{}
+	actualStats := *cgroups.NewStats()
+	err := memory.GetStats(helper.CgroupPath, &actualStats)
+	if err == nil {
+		t.Fatal("Expected failure")
+	}
+}
+
+func TestMemorySetOomControl(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		oomKillDisable = 1 // disable oom killer, default is 0
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.oom_control": strconv.Itoa(oomKillDisable),
+	})
+
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.oom_control")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.oom_control - %s", err)
+	}
+
+	if value != oomKillDisable {
+		t.Fatalf("Got the wrong value, set memory.oom_control failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/net_cls.go

@@ -3,6 +3,8 @@
 package fs
 
 import (
+	"strconv"
+
 	"github.com/opencontainers/runc/libcontainer/cgroups"
 	"github.com/opencontainers/runc/libcontainer/configs"
 )
@@ -23,8 +25,8 @@ func (s *NetClsGroup) Apply(d *cgroupData) error {
 }
 
 func (s *NetClsGroup) Set(path string, cgroup *configs.Cgroup) error {
-	if cgroup.Resources.NetClsClassid != "" {
-		if err := writeFile(path, "net_cls.classid", cgroup.Resources.NetClsClassid); err != nil {
+	if cgroup.Resources.NetClsClassid != 0 {
+		if err := writeFile(path, "net_cls.classid", strconv.FormatUint(uint64(cgroup.Resources.NetClsClassid), 10)); err != nil {
 			return err
 		}
 	}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/net_cls_test.go

@@ -0,0 +1,39 @@
+// +build linux
+
+package fs
+
+import (
+	"strconv"
+	"testing"
+)
+
+const (
+	classidBefore = 0x100002
+	classidAfter  = 0x100001
+)
+
+func TestNetClsSetClassid(t *testing.T) {
+	helper := NewCgroupTestUtil("net_cls", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"net_cls.classid": strconv.FormatUint(classidBefore, 10),
+	})
+
+	helper.CgroupData.config.Resources.NetClsClassid = classidAfter
+	netcls := &NetClsGroup{}
+	if err := netcls.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	// As we are in mock environment, we can't get correct value of classid from
+	// net_cls.classid.
+	// So. we just judge if we successfully write classid into file
+	value, err := getCgroupParamUint(helper.CgroupPath, "net_cls.classid")
+	if err != nil {
+		t.Fatalf("Failed to parse net_cls.classid - %s", err)
+	}
+	if value != classidAfter {
+		t.Fatal("Got the wrong value, set net_cls.classid failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/net_prio_test.go

@@ -0,0 +1,38 @@
+// +build linux
+
+package fs
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var (
+	prioMap = []*configs.IfPrioMap{
+		{
+			Interface: "test",
+			Priority:  5,
+		},
+	}
+)
+
+func TestNetPrioSetIfPrio(t *testing.T) {
+	helper := NewCgroupTestUtil("net_prio", t)
+	defer helper.cleanup()
+
+	helper.CgroupData.config.Resources.NetPrioIfpriomap = prioMap
+	netPrio := &NetPrioGroup{}
+	if err := netPrio.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "net_prio.ifpriomap")
+	if err != nil {
+		t.Fatalf("Failed to parse net_prio.ifpriomap - %s", err)
+	}
+	if !strings.Contains(value, "test 5") {
+		t.Fatal("Got the wrong value, set net_prio.ifpriomap failed.")
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/pids_test.go

@@ -0,0 +1,111 @@
+// +build linux
+
+package fs
+
+import (
+	"strconv"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+)
+
+const (
+	maxUnlimited = -1
+	maxLimited   = 1024
+)
+
+func TestPidsSetMax(t *testing.T) {
+	helper := NewCgroupTestUtil("pids", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"pids.max": "max",
+	})
+
+	helper.CgroupData.config.Resources.PidsLimit = maxLimited
+	pids := &PidsGroup{}
+	if err := pids.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "pids.max")
+	if err != nil {
+		t.Fatalf("Failed to parse pids.max - %s", err)
+	}
+
+	if value != maxLimited {
+		t.Fatalf("Expected %d, got %d for setting pids.max - limited", maxLimited, value)
+	}
+}
+
+func TestPidsSetUnlimited(t *testing.T) {
+	helper := NewCgroupTestUtil("pids", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"pids.max": strconv.Itoa(maxLimited),
+	})
+
+	helper.CgroupData.config.Resources.PidsLimit = maxUnlimited
+	pids := &PidsGroup{}
+	if err := pids.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "pids.max")
+	if err != nil {
+		t.Fatalf("Failed to parse pids.max - %s", err)
+	}
+
+	if value != "max" {
+		t.Fatalf("Expected %s, got %s for setting pids.max - unlimited", "max", value)
+	}
+}
+
+func TestPidsStats(t *testing.T) {
+	helper := NewCgroupTestUtil("pids", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"pids.current": strconv.Itoa(1337),
+		"pids.max":     strconv.Itoa(maxLimited),
+	})
+
+	pids := &PidsGroup{}
+	stats := *cgroups.NewStats()
+	if err := pids.GetStats(helper.CgroupPath, &stats); err != nil {
+		t.Fatal(err)
+	}
+
+	if stats.PidsStats.Current != 1337 {
+		t.Fatalf("Expected %d, got %d for pids.current", 1337, stats.PidsStats.Current)
+	}
+
+	if stats.PidsStats.Limit != maxLimited {
+		t.Fatalf("Expected %d, got %d for pids.max", maxLimited, stats.PidsStats.Limit)
+	}
+}
+
+func TestPidsStatsUnlimited(t *testing.T) {
+	helper := NewCgroupTestUtil("pids", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"pids.current": strconv.Itoa(4096),
+		"pids.max":     "max",
+	})
+
+	pids := &PidsGroup{}
+	stats := *cgroups.NewStats()
+	if err := pids.GetStats(helper.CgroupPath, &stats); err != nil {
+		t.Fatal(err)
+	}
+
+	if stats.PidsStats.Current != 4096 {
+		t.Fatalf("Expected %d, got %d for pids.current", 4096, stats.PidsStats.Current)
+	}
+
+	if stats.PidsStats.Limit != 0 {
+		t.Fatalf("Expected %d, got %d for pids.max", 0, stats.PidsStats.Limit)
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/stats_util_test.go

@@ -0,0 +1,123 @@
+// +build linux
+
+package fs
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/cgroups"
+
+	"github.com/sirupsen/logrus"
+)
+
+func blkioStatEntryEquals(expected, actual []cgroups.BlkioStatEntry) error {
+	if len(expected) != len(actual) {
+		return fmt.Errorf("blkioStatEntries length do not match")
+	}
+	for i, expValue := range expected {
+		actValue := actual[i]
+		if expValue != actValue {
+			return fmt.Errorf("Expected blkio stat entry %v but found %v", expValue, actValue)
+		}
+	}
+	return nil
+}
+
+func expectBlkioStatsEquals(t *testing.T, expected, actual cgroups.BlkioStats) {
+	if err := blkioStatEntryEquals(expected.IoServiceBytesRecursive, actual.IoServiceBytesRecursive); err != nil {
+		logrus.Printf("blkio IoServiceBytesRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoServicedRecursive, actual.IoServicedRecursive); err != nil {
+		logrus.Printf("blkio IoServicedRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoQueuedRecursive, actual.IoQueuedRecursive); err != nil {
+		logrus.Printf("blkio IoQueuedRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.SectorsRecursive, actual.SectorsRecursive); err != nil {
+		logrus.Printf("blkio SectorsRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoServiceTimeRecursive, actual.IoServiceTimeRecursive); err != nil {
+		logrus.Printf("blkio IoServiceTimeRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoWaitTimeRecursive, actual.IoWaitTimeRecursive); err != nil {
+		logrus.Printf("blkio IoWaitTimeRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoMergedRecursive, actual.IoMergedRecursive); err != nil {
+		logrus.Printf("blkio IoMergedRecursive do not match - %v vs %v\n", expected.IoMergedRecursive, actual.IoMergedRecursive)
+		t.Fail()
+	}
+
+	if err := blkioStatEntryEquals(expected.IoTimeRecursive, actual.IoTimeRecursive); err != nil {
+		logrus.Printf("blkio IoTimeRecursive do not match - %s\n", err)
+		t.Fail()
+	}
+}
+
+func expectThrottlingDataEquals(t *testing.T, expected, actual cgroups.ThrottlingData) {
+	if expected != actual {
+		logrus.Printf("Expected throttling data %v but found %v\n", expected, actual)
+		t.Fail()
+	}
+}
+
+func expectHugetlbStatEquals(t *testing.T, expected, actual cgroups.HugetlbStats) {
+	if expected != actual {
+		logrus.Printf("Expected hugetlb stats %v but found %v\n", expected, actual)
+		t.Fail()
+	}
+}
+
+func expectMemoryStatEquals(t *testing.T, expected, actual cgroups.MemoryStats) {
+	expectMemoryDataEquals(t, expected.Usage, actual.Usage)
+	expectMemoryDataEquals(t, expected.SwapUsage, actual.SwapUsage)
+	expectMemoryDataEquals(t, expected.KernelUsage, actual.KernelUsage)
+
+	if expected.UseHierarchy != actual.UseHierarchy {
+		logrus.Printf("Expected memory use hiearchy %v, but found %v\n", expected.UseHierarchy, actual.UseHierarchy)
+		t.Fail()
+	}
+
+	for key, expValue := range expected.Stats {
+		actValue, ok := actual.Stats[key]
+		if !ok {
+			logrus.Printf("Expected memory stat key %s not found\n", key)
+			t.Fail()
+		}
+		if expValue != actValue {
+			logrus.Printf("Expected memory stat value %d but found %d\n", expValue, actValue)
+			t.Fail()
+		}
+	}
+}
+
+func expectMemoryDataEquals(t *testing.T, expected, actual cgroups.MemoryData) {
+	if expected.Usage != actual.Usage {
+		logrus.Printf("Expected memory usage %d but found %d\n", expected.Usage, actual.Usage)
+		t.Fail()
+	}
+	if expected.MaxUsage != actual.MaxUsage {
+		logrus.Printf("Expected memory max usage %d but found %d\n", expected.MaxUsage, actual.MaxUsage)
+		t.Fail()
+	}
+	if expected.Failcnt != actual.Failcnt {
+		logrus.Printf("Expected memory failcnt %d but found %d\n", expected.Failcnt, actual.Failcnt)
+		t.Fail()
+	}
+	if expected.Limit != actual.Limit {
+		logrus.Printf("Expected memory limit %d but found %d\n", expected.Limit, actual.Limit)
+		t.Fail()
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/util_test.go

@@ -0,0 +1,67 @@
+// +build linux
+
+/*
+Utility for testing cgroup operations.
+
+Creates a mock of the cgroup filesystem for the duration of the test.
+*/
+package fs
+
+import (
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+type cgroupTestUtil struct {
+	// cgroup data to use in tests.
+	CgroupData *cgroupData
+
+	// Path to the mock cgroup directory.
+	CgroupPath string
+
+	// Temporary directory to store mock cgroup filesystem.
+	tempDir string
+	t       *testing.T
+}
+
+// Creates a new test util for the specified subsystem
+func NewCgroupTestUtil(subsystem string, t *testing.T) *cgroupTestUtil {
+	d := &cgroupData{
+		config: &configs.Cgroup{},
+	}
+	d.config.Resources = &configs.Resources{}
+	tempDir, err := ioutil.TempDir("", "cgroup_test")
+	if err != nil {
+		t.Fatal(err)
+	}
+	d.root = tempDir
+	testCgroupPath := filepath.Join(d.root, subsystem)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Ensure the full mock cgroup path exists.
+	err = os.MkdirAll(testCgroupPath, 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return &cgroupTestUtil{CgroupData: d, CgroupPath: testCgroupPath, tempDir: tempDir, t: t}
+}
+
+func (c *cgroupTestUtil) cleanup() {
+	os.RemoveAll(c.tempDir)
+}
+
+// Write the specified contents on the mock of the specified cgroup files.
+func (c *cgroupTestUtil) writeFileContents(fileContents map[string]string) {
+	for file, contents := range fileContents {
+		err := writeFile(c.CgroupPath, file, contents)
+		if err != nil {
+			c.t.Fatal(err)
+		}
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/utils.go

@@ -12,7 +12,6 @@ import (
 )
 
 var (
-	ErrNotSupportStat = errors.New("stats are not supported for subsystem")
 	ErrNotValidFormat = errors.New("line is not a valid key value format")
 )
 

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/utils_test.go

@@ -0,0 +1,97 @@
+// +build linux
+
+package fs
+
+import (
+	"io/ioutil"
+	"math"
+	"os"
+	"path/filepath"
+	"strconv"
+	"testing"
+)
+
+const (
+	cgroupFile  = "cgroup.file"
+	floatValue  = 2048.0
+	floatString = "2048"
+)
+
+func TestGetCgroupParamsInt(t *testing.T) {
+	// Setup tempdir.
+	tempDir, err := ioutil.TempDir("", "cgroup_utils_test")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(tempDir)
+	tempFile := filepath.Join(tempDir, cgroupFile)
+
+	// Success.
+	err = ioutil.WriteFile(tempFile, []byte(floatString), 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	value, err := getCgroupParamUint(tempDir, cgroupFile)
+	if err != nil {
+		t.Fatal(err)
+	} else if value != floatValue {
+		t.Fatalf("Expected %d to equal %f", value, floatValue)
+	}
+
+	// Success with new line.
+	err = ioutil.WriteFile(tempFile, []byte(floatString+"\n"), 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	value, err = getCgroupParamUint(tempDir, cgroupFile)
+	if err != nil {
+		t.Fatal(err)
+	} else if value != floatValue {
+		t.Fatalf("Expected %d to equal %f", value, floatValue)
+	}
+
+	// Success with negative values
+	err = ioutil.WriteFile(tempFile, []byte("-12345"), 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	value, err = getCgroupParamUint(tempDir, cgroupFile)
+	if err != nil {
+		t.Fatal(err)
+	} else if value != 0 {
+		t.Fatalf("Expected %d to equal %d", value, 0)
+	}
+
+	// Success with negative values lesser than min int64
+	s := strconv.FormatFloat(math.MinInt64, 'f', -1, 64)
+	err = ioutil.WriteFile(tempFile, []byte(s), 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	value, err = getCgroupParamUint(tempDir, cgroupFile)
+	if err != nil {
+		t.Fatal(err)
+	} else if value != 0 {
+		t.Fatalf("Expected %d to equal %d", value, 0)
+	}
+
+	// Not a float.
+	err = ioutil.WriteFile(tempFile, []byte("not-a-float"), 0755)
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = getCgroupParamUint(tempDir, cgroupFile)
+	if err == nil {
+		t.Fatal("Expecting error, got none")
+	}
+
+	// Unknown file.
+	err = os.Remove(tempFile)
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = getCgroupParamUint(tempDir, cgroupFile)
+	if err == nil {
+		t.Fatal("Expecting error, got none")
+	}
+}

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

@@ -51,6 +51,8 @@ type MemoryStats struct {
 	KernelUsage MemoryData `json:"kernel_usage,omitempty"`
 	// usage of kernel TCP memory
 	KernelTCPUsage MemoryData `json:"kernel_tcp_usage,omitempty"`
+	// if true, memory usage is accounted for throughout a hierarchy of cgroups.
+	UseHierarchy bool `json:"use_hierarchy"`
 
 	Stats map[string]uint64 `json:"stats,omitempty"`
 }

vendor/github.com/opencontainers/runc/libcontainer/cgroups/systemd/apply_nosystemd.go

@@ -1,4 +1,4 @@
-// +build !linux
+// +build !linux static_build
 
 package systemd
 
@@ -43,7 +43,7 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
 }
 
 func (m *Manager) Set(container *configs.Config) error {
-	return nil, fmt.Errorf("Systemd not supported")
+	return fmt.Errorf("Systemd not supported")
 }
 
 func (m *Manager) Freeze(state configs.FreezerState) error {

vendor/github.com/opencontainers/runc/libcontainer/cgroups/systemd/apply_systemd.go

@@ -1,14 +1,12 @@
-// +build linux
+// +build linux,!static_build
 
 package systemd
 
 import (
 	"errors"
 	"fmt"
-	"io/ioutil"
 	"os"
 	"path/filepath"
-	"strconv"
 	"strings"
 	"sync"
 	"time"
@@ -19,6 +17,7 @@ import (
 	"github.com/opencontainers/runc/libcontainer/cgroups"
 	"github.com/opencontainers/runc/libcontainer/cgroups/fs"
 	"github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/sirupsen/logrus"
 )
 
 type Manager struct {
@@ -67,13 +66,16 @@ var subsystems = subsystemSet{
 
 const (
 	testScopeWait = 4
+	testSliceWait = 4
 )
 
 var (
 	connLock                        sync.Mutex
 	theConn                         *systemdDbus.Conn
 	hasStartTransientUnit           bool
+	hasStartTransientSliceUnit      bool
 	hasTransientDefaultDependencies bool
+	hasDelegate                     bool
 )
 
 func newProp(name string, units interface{}) systemdDbus.Property {
@@ -146,6 +148,48 @@ func UseSystemd() bool {
 
 		// Not critical because of the stop unit logic above.
 		theConn.StopUnit(scope, "replace", nil)
+
+		// Assume StartTransientUnit on a scope allows Delegate
+		hasDelegate = true
+		dl := newProp("Delegate", true)
+		if _, err := theConn.StartTransientUnit(scope, "replace", []systemdDbus.Property{dl}, nil); err != nil {
+			if dbusError, ok := err.(dbus.Error); ok {
+				if strings.Contains(dbusError.Name, "org.freedesktop.DBus.Error.PropertyReadOnly") {
+					hasDelegate = false
+				}
+			}
+		}
+
+		// Assume we have the ability to start a transient unit as a slice
+		// This was broken until systemd v229, but has been back-ported on RHEL environments >= 219
+		// For details, see: https://bugzilla.redhat.com/show_bug.cgi?id=1370299
+		hasStartTransientSliceUnit = true
+
+		// To ensure simple clean-up, we create a slice off the root with no hierarchy
+		slice := fmt.Sprintf("libcontainer_%d_systemd_test_default.slice", os.Getpid())
+		if _, err := theConn.StartTransientUnit(slice, "replace", nil, nil); err != nil {
+			if _, ok := err.(dbus.Error); ok {
+				hasStartTransientSliceUnit = false
+			}
+		}
+
+		for i := 0; i <= testSliceWait; i++ {
+			if _, err := theConn.StopUnit(slice, "replace", nil); err != nil {
+				if dbusError, ok := err.(dbus.Error); ok {
+					if strings.Contains(dbusError.Name, "org.freedesktop.systemd1.NoSuchUnit") {
+						hasStartTransientSliceUnit = false
+						break
+					}
+				}
+			} else {
+				break
+			}
+			time.Sleep(time.Millisecond)
+		}
+
+		// Not critical because of the stop unit logic above.
+		theConn.StopUnit(scope, "replace", nil)
+		theConn.StopUnit(slice, "replace", nil)
 	}
 	return hasStartTransientUnit
 }
@@ -179,13 +223,29 @@ func (m *Manager) Apply(pid int) error {
 		slice = c.Parent
 	}
 
-	properties = append(properties,
-		systemdDbus.PropSlice(slice),
-		systemdDbus.PropDescription("docker container "+c.Name),
-		newProp("PIDs", []uint32{uint32(pid)}),
+	properties = append(properties, systemdDbus.PropDescription("libcontainer container "+c.Name))
+
+	// if we create a slice, the parent is defined via a Wants=
+	if strings.HasSuffix(unitName, ".slice") {
+		// This was broken until systemd v229, but has been back-ported on RHEL environments >= 219
+		if !hasStartTransientSliceUnit {
+			return fmt.Errorf("systemd version does not support ability to start a slice as transient unit")
+		}
+		properties = append(properties, systemdDbus.PropWants(slice))
+	} else {
+		// otherwise, we use Slice=
+		properties = append(properties, systemdDbus.PropSlice(slice))
+	}
+
+	// only add pid if its valid, -1 is used w/ general slice creation.
+	if pid != -1 {
+		properties = append(properties, newProp("PIDs", []uint32{uint32(pid)}))
+	}
+
+	if hasDelegate {
 		// This is only supported on systemd versions 218 and above.
-		newProp("Delegate", true),
-	)
+		properties = append(properties, newProp("Delegate", true))
+	}
 
 	// Always enable accounting, this gets us the same behaviour as the fs implementation,
 	// plus the kernel has some problems with joining the memory cgroup at a later time.
@@ -206,7 +266,21 @@ func (m *Manager) Apply(pid int) error {
 
 	if c.Resources.CpuShares != 0 {
 		properties = append(properties,
-			newProp("CPUShares", uint64(c.Resources.CpuShares)))
+			newProp("CPUShares", c.Resources.CpuShares))
+	}
+
+	// cpu.cfs_quota_us and cpu.cfs_period_us are controlled by systemd.
+	if c.Resources.CpuQuota != 0 && c.Resources.CpuPeriod != 0 {
+		cpuQuotaPerSecUSec := uint64(c.Resources.CpuQuota*1000000) / c.Resources.CpuPeriod
+		// systemd converts CPUQuotaPerSecUSec (microseconds per CPU second) to CPUQuota
+		// (integer percentage of CPU) internally.  This means that if a fractional percent of
+		// CPU is indicated by Resources.CpuQuota, we need to round up to the nearest
+		// 10ms (1% of a second) such that child cgroups can set the cpu.cfs_quota_us they expect.
+		if cpuQuotaPerSecUSec%10000 != 0 {
+			cpuQuotaPerSecUSec = ((cpuQuotaPerSecUSec / 10000) + 1) * 10000
+		}
+		properties = append(properties,
+			newProp("CPUQuotaPerSecUSec", cpuQuotaPerSecUSec))
 	}
 
 	if c.Resources.BlkioWeight != 0 {
@@ -214,20 +288,25 @@ func (m *Manager) Apply(pid int) error {
 			newProp("BlockIOWeight", uint64(c.Resources.BlkioWeight)))
 	}
 
-	// We need to set kernel memory before processes join cgroup because
-	// kmem.limit_in_bytes can only be set when the cgroup is empty.
-	// And swap memory limit needs to be set after memory limit, only
-	// memory limit is handled by systemd, so it's kind of ugly here.
-	if c.Resources.KernelMemory > 0 {
+	// We have to set kernel memory here, as we can't change it once
+	// processes have been attached to the cgroup.
+	if c.Resources.KernelMemory != 0 {
 		if err := setKernelMemory(c); err != nil {
 			return err
 		}
 	}
 
-	if _, err := theConn.StartTransientUnit(unitName, "replace", properties, nil); err != nil {
+	statusChan := make(chan string)
+	if _, err := theConn.StartTransientUnit(unitName, "replace", properties, statusChan); err != nil && !isUnitExists(err) {
 		return err
 	}
 
+	select {
+	case <-statusChan:
+	case <-time.After(time.Second):
+		logrus.Warnf("Timed out while waiting for StartTransientUnit completion signal from dbus. Continuing...")
+	}
+
 	if err := joinCgroups(c, pid); err != nil {
 		return err
 	}
@@ -269,15 +348,6 @@ func (m *Manager) GetPaths() map[string]string {
 	return paths
 }
 
-func writeFile(dir, file, data string) error {
-	// Normally dir should not be empty, one case is that cgroup subsystem
-	// is not mounted, we will get empty dir, and we want it fail here.
-	if dir == "" {
-		return fmt.Errorf("no such directory for %s", file)
-	}
-	return ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700)
-}
-
 func join(c *configs.Cgroup, subsystem string, pid int) (string, error) {
 	path, err := getSubsystemPath(c, subsystem)
 	if err != nil {
@@ -286,10 +356,9 @@ func join(c *configs.Cgroup, subsystem string, pid int) (string, error) {
 	if err := os.MkdirAll(path, 0755); err != nil {
 		return "", err
 	}
-	if err := writeFile(path, "cgroup.procs", strconv.Itoa(pid)); err != nil {
+	if err := cgroups.WriteCgroupProc(path, pid); err != nil {
 		return "", err
 	}
-
 	return path, nil
 }
 
@@ -299,7 +368,6 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
 		switch name {
 		case "name=systemd":
 			// let systemd handle this
-			break
 		case "cpuset":
 			path, err := getSubsystemPath(c, name)
 			if err != nil && !cgroups.IsNotFound(err) {
@@ -309,7 +377,6 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
 			if err := s.ApplyDir(path, c, pid); err != nil {
 				return err
 			}
-			break
 		default:
 			_, err := join(c, name, pid)
 			if err != nil {
@@ -331,10 +398,10 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
 	return nil
 }
 
-// systemd represents slice heirarchy using `-`, so we need to follow suit when
+// systemd represents slice hierarchy using `-`, so we need to follow suit when
 // generating the path of slice. Essentially, test-a-b.slice becomes
-// test.slice/test-a.slice/test-a-b.slice.
-func expandSlice(slice string) (string, error) {
+// /test.slice/test-a.slice/test-a-b.slice.
+func ExpandSlice(slice string) (string, error) {
 	suffix := ".slice"
 	// Name has to end with ".slice", but can't be just ".slice".
 	if len(slice) < len(suffix) || !strings.HasSuffix(slice, suffix) {
@@ -348,6 +415,10 @@ func expandSlice(slice string) (string, error) {
 
 	var path, prefix string
 	sliceName := strings.TrimSuffix(slice, suffix)
+	// if input was -.slice, we should just return root now
+	if sliceName == "-" {
+		return "/", nil
+	}
 	for _, component := range strings.Split(sliceName, "-") {
 		// test--a.slice isn't permitted, nor is -test.slice.
 		if component == "" {
@@ -355,10 +426,9 @@ func expandSlice(slice string) (string, error) {
 		}
 
 		// Append the component to the path and to the prefix.
-		path += prefix + component + suffix + "/"
+		path += "/" + prefix + component + suffix
 		prefix += component + "-"
 	}
-
 	return path, nil
 }
 
@@ -368,17 +438,19 @@ func getSubsystemPath(c *configs.Cgroup, subsystem string) (string, error) {
 		return "", err
 	}
 
-	initPath, err := cgroups.GetInitCgroupDir(subsystem)
+	initPath, err := cgroups.GetInitCgroup(subsystem)
 	if err != nil {
 		return "", err
 	}
+	// if pid 1 is systemd 226 or later, it will be in init.scope, not the root
+	initPath = strings.TrimSuffix(filepath.Clean(initPath), "init.scope")
 
 	slice := "system.slice"
 	if c.Parent != "" {
 		slice = c.Parent
 	}
 
-	slice, err = expandSlice(slice)
+	slice, err = ExpandSlice(slice)
 	if err != nil {
 		return "", err
 	}
@@ -439,6 +511,11 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
 }
 
 func (m *Manager) Set(container *configs.Config) error {
+	// If Paths are set, then we are just joining cgroups paths
+	// and there is no need to set any values.
+	if m.Cgroups.Paths != nil {
+		return nil
+	}
 	for _, sys := range subsystems {
 		// Get the subsystem path, but don't error out for not found cgroups.
 		path, err := getSubsystemPath(container.Cgroups, sys.Name())
@@ -460,7 +537,11 @@ func (m *Manager) Set(container *configs.Config) error {
 }
 
 func getUnitName(c *configs.Cgroup) string {
-	return fmt.Sprintf("%s-%s.scope", c.ScopePrefix, c.Name)
+	// by default, we create a scope unless the user explicitly asks for a slice.
+	if !strings.HasSuffix(c.Name, ".slice") {
+		return fmt.Sprintf("%s-%s.scope", c.ScopePrefix, c.Name)
+	}
+	return c.Name
 }
 
 func setKernelMemory(c *configs.Cgroup) error {
@@ -472,8 +553,15 @@ func setKernelMemory(c *configs.Cgroup) error {
 	if err := os.MkdirAll(path, 0755); err != nil {
 		return err
 	}
-
-	// This doesn't get called by manager.Set, so we need to do it here.
-	s := &fs.MemoryGroup{}
-	return s.SetKernelMemory(path, c)
+	return fs.EnableKernelMemoryAccounting(path)
+}
+
+// isUnitExists returns true if the error is that a systemd unit already exists.
+func isUnitExists(err error) bool {
+	if err != nil {
+		if dbusError, ok := err.(dbus.Error); ok {
+			return strings.Contains(dbusError.Name, "org.freedesktop.systemd1.UnitExists")
+		}
+	}
+	return false
 }

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

@@ -16,37 +16,24 @@ import (
 	"github.com/docker/go-units"
 )
 
-const cgroupNamePrefix = "name="
+const (
+	cgroupNamePrefix = "name="
+	CgroupProcesses  = "cgroup.procs"
+)
 
-// https://www.kernel.org/doc/Documentation/cgroups/cgroups.txt
+// https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt
 func FindCgroupMountpoint(subsystem string) (string, error) {
-	// We are not using mount.GetMounts() because it's super-inefficient,
-	// parsing it directly sped up x10 times because of not using Sscanf.
-	// It was one of two major performance drawbacks in container start.
-	f, err := os.Open("/proc/self/mountinfo")
-	if err != nil {
-		return "", err
-	}
-	defer f.Close()
-
-	scanner := bufio.NewScanner(f)
-	for scanner.Scan() {
-		txt := scanner.Text()
-		fields := strings.Split(txt, " ")
-		for _, opt := range strings.Split(fields[len(fields)-1], ",") {
-			if opt == subsystem {
-				return fields[4], nil
-			}
-		}
-	}
-	if err := scanner.Err(); err != nil {
-		return "", err
-	}
-
-	return "", NewNotFoundError(subsystem)
+	mnt, _, err := FindCgroupMountpointAndRoot(subsystem)
+	return mnt, err
 }
 
 func FindCgroupMountpointAndRoot(subsystem string) (string, string, error) {
+	// We are not using mount.GetMounts() because it's super-inefficient,
+	// parsing it directly sped up x10 times because of not using Sscanf.
+	// It was one of two major performance drawbacks in container start.
+	if !isSubsystemAvailable(subsystem) {
+		return "", "", NewNotFoundError(subsystem)
+	}
 	f, err := os.Open("/proc/self/mountinfo")
 	if err != nil {
 		return "", "", err
@@ -70,6 +57,30 @@ func FindCgroupMountpointAndRoot(subsystem string) (string, string, error) {
 	return "", "", NewNotFoundError(subsystem)
 }
 
+func isSubsystemAvailable(subsystem string) bool {
+	cgroups, err := ParseCgroupFile("/proc/self/cgroup")
+	if err != nil {
+		return false
+	}
+	_, avail := cgroups[subsystem]
+	return avail
+}
+
+func GetClosestMountpointAncestor(dir, mountinfo string) string {
+	deepestMountPoint := ""
+	for _, mountInfoEntry := range strings.Split(mountinfo, "\n") {
+		mountInfoParts := strings.Fields(mountInfoEntry)
+		if len(mountInfoParts) < 5 {
+			continue
+		}
+		mountPoint := mountInfoParts[4]
+		if strings.HasPrefix(mountPoint, deepestMountPoint) && strings.HasPrefix(dir, mountPoint) {
+			deepestMountPoint = mountPoint
+		}
+	}
+	return deepestMountPoint
+}
+
 func FindCgroupMountpointDir() (string, error) {
 	f, err := os.Open("/proc/self/mountinfo")
 	if err != nil {
@@ -113,7 +124,7 @@ type Mount struct {
 	Subsystems []string
 }
 
-func (m Mount) GetThisCgroupDir(cgroups map[string]string) (string, error) {
+func (m Mount) GetOwnCgroup(cgroups map[string]string) (string, error) {
 	if len(m.Subsystems) == 0 {
 		return "", fmt.Errorf("no subsystem for mount")
 	}
@@ -121,16 +132,17 @@ func (m Mount) GetThisCgroupDir(cgroups map[string]string) (string, error) {
 	return getControllerPath(m.Subsystems[0], cgroups)
 }
 
-func getCgroupMountsHelper(ss map[string]bool, mi io.Reader) ([]Mount, error) {
+func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount, error) {
 	res := make([]Mount, 0, len(ss))
 	scanner := bufio.NewScanner(mi)
-	for scanner.Scan() {
+	numFound := 0
+	for scanner.Scan() && numFound < len(ss) {
 		txt := scanner.Text()
 		sepIdx := strings.Index(txt, " - ")
 		if sepIdx == -1 {
 			return nil, fmt.Errorf("invalid mountinfo format")
 		}
-		if txt[sepIdx+3:sepIdx+9] != "cgroup" {
+		if txt[sepIdx+3:sepIdx+10] == "cgroup2" || txt[sepIdx+3:sepIdx+9] != "cgroup" {
 			continue
 		}
 		fields := strings.Split(txt, " ")
@@ -139,12 +151,17 @@ func getCgroupMountsHelper(ss map[string]bool, mi io.Reader) ([]Mount, error) {
 			Root:       fields[3],
 		}
 		for _, opt := range strings.Split(fields[len(fields)-1], ",") {
+			if !ss[opt] {
+				continue
+			}
 			if strings.HasPrefix(opt, cgroupNamePrefix) {
 				m.Subsystems = append(m.Subsystems, opt[len(cgroupNamePrefix):])
-			}
-			if ss[opt] {
+			} else {
 				m.Subsystems = append(m.Subsystems, opt)
 			}
+			if !all {
+				numFound++
+			}
 		}
 		res = append(res, m)
 	}
@@ -154,23 +171,25 @@ func getCgroupMountsHelper(ss map[string]bool, mi io.Reader) ([]Mount, error) {
 	return res, nil
 }
 
-func GetCgroupMounts() ([]Mount, error) {
+// GetCgroupMounts returns the mounts for the cgroup subsystems.
+// all indicates whether to return just the first instance or all the mounts.
+func GetCgroupMounts(all bool) ([]Mount, error) {
 	f, err := os.Open("/proc/self/mountinfo")
 	if err != nil {
 		return nil, err
 	}
 	defer f.Close()
 
-	all, err := GetAllSubsystems()
+	allSubsystems, err := ParseCgroupFile("/proc/self/cgroup")
 	if err != nil {
 		return nil, err
 	}
 
 	allMap := make(map[string]bool)
-	for _, s := range all {
+	for s := range allSubsystems {
 		allMap[s] = true
 	}
-	return getCgroupMountsHelper(allMap, f)
+	return getCgroupMountsHelper(allMap, f, all)
 }
 
 // GetAllSubsystems returns all the cgroup subsystems supported by the kernel
@@ -185,9 +204,6 @@ func GetAllSubsystems() ([]string, error) {
 
 	s := bufio.NewScanner(f)
 	for s.Scan() {
-		if err := s.Err(); err != nil {
-			return nil, err
-		}
 		text := s.Text()
 		if text[0] != '#' {
 			parts := strings.Fields(text)
@@ -196,11 +212,14 @@ func GetAllSubsystems() ([]string, error) {
 			}
 		}
 	}
+	if err := s.Err(); err != nil {
+		return nil, err
+	}
 	return subsystems, nil
 }
 
-// GetThisCgroupDir returns the relative path to the cgroup docker is running in.
-func GetThisCgroupDir(subsystem string) (string, error) {
+// GetOwnCgroup returns the relative path to the cgroup docker is running in.
+func GetOwnCgroup(subsystem string) (string, error) {
 	cgroups, err := ParseCgroupFile("/proc/self/cgroup")
 	if err != nil {
 		return "", err
@@ -209,8 +228,16 @@ func GetThisCgroupDir(subsystem string) (string, error) {
 	return getControllerPath(subsystem, cgroups)
 }
 
-func GetInitCgroupDir(subsystem string) (string, error) {
+func GetOwnCgroupPath(subsystem string) (string, error) {
+	cgroup, err := GetOwnCgroup(subsystem)
+	if err != nil {
+		return "", err
+	}
 
+	return getCgroupPathHelper(subsystem, cgroup)
+}
+
+func GetInitCgroup(subsystem string) (string, error) {
 	cgroups, err := ParseCgroupFile("/proc/1/cgroup")
 	if err != nil {
 		return "", err
@@ -219,8 +246,33 @@ func GetInitCgroupDir(subsystem string) (string, error) {
 	return getControllerPath(subsystem, cgroups)
 }
 
+func GetInitCgroupPath(subsystem string) (string, error) {
+	cgroup, err := GetInitCgroup(subsystem)
+	if err != nil {
+		return "", err
+	}
+
+	return getCgroupPathHelper(subsystem, cgroup)
+}
+
+func getCgroupPathHelper(subsystem, cgroup string) (string, error) {
+	mnt, root, err := FindCgroupMountpointAndRoot(subsystem)
+	if err != nil {
+		return "", err
+	}
+
+	// This is needed for nested containers, because in /proc/self/cgroup we
+	// see pathes from host, which don't exist in container.
+	relCgroup, err := filepath.Rel(root, cgroup)
+	if err != nil {
+		return "", err
+	}
+
+	return filepath.Join(mnt, relCgroup), nil
+}
+
 func readProcsFile(dir string) ([]int, error) {
-	f, err := os.Open(filepath.Join(dir, "cgroup.procs"))
+	f, err := os.Open(filepath.Join(dir, CgroupProcesses))
 	if err != nil {
 		return nil, err
 	}
@@ -243,6 +295,8 @@ func readProcsFile(dir string) ([]int, error) {
 	return out, nil
 }
 
+// ParseCgroupFile parses the given cgroup file, typically from
+// /proc/<pid>/cgroup, into a map of subgroups to cgroup names.
 func ParseCgroupFile(path string) (map[string]string, error) {
 	f, err := os.Open(path)
 	if err != nil {
@@ -250,21 +304,35 @@ func ParseCgroupFile(path string) (map[string]string, error) {
 	}
 	defer f.Close()
 
-	s := bufio.NewScanner(f)
+	return parseCgroupFromReader(f)
+}
+
+// helper function for ParseCgroupFile to make testing easier
+func parseCgroupFromReader(r io.Reader) (map[string]string, error) {
+	s := bufio.NewScanner(r)
 	cgroups := make(map[string]string)
 
 	for s.Scan() {
-		if err := s.Err(); err != nil {
-			return nil, err
-		}
-
 		text := s.Text()
-		parts := strings.Split(text, ":")
+		// from cgroups(7):
+		// /proc/[pid]/cgroup
+		// ...
+		// For each cgroup hierarchy ... there is one entry
+		// containing three colon-separated fields of the form:
+		//     hierarchy-ID:subsystem-list:cgroup-path
+		parts := strings.SplitN(text, ":", 3)
+		if len(parts) < 3 {
+			return nil, fmt.Errorf("invalid cgroup entry: must contain at least two colons: %v", text)
+		}
 
 		for _, subs := range strings.Split(parts[1], ",") {
 			cgroups[subs] = parts[2]
 		}
 	}
+	if err := s.Err(); err != nil {
+		return nil, err
+	}
+
 	return cgroups, nil
 }
 
@@ -291,8 +359,7 @@ func PathExists(path string) bool {
 func EnterPid(cgroupPaths map[string]string, pid int) error {
 	for _, path := range cgroupPaths {
 		if PathExists(path) {
-			if err := ioutil.WriteFile(filepath.Join(path, "cgroup.procs"),
-				[]byte(strconv.Itoa(pid)), 0700); err != nil {
+			if err := WriteCgroupProc(path, pid); err != nil {
 				return err
 			}
 		}
@@ -361,7 +428,7 @@ func GetAllPids(path string) ([]int, error) {
 	// collect pids from all sub-cgroups
 	err := filepath.Walk(path, func(p string, info os.FileInfo, iErr error) error {
 		dir, file := filepath.Split(p)
-		if file != "cgroup.procs" {
+		if file != CgroupProcesses {
 			return nil
 		}
 		if iErr != nil {
@@ -376,3 +443,20 @@ func GetAllPids(path string) ([]int, error) {
 	})
 	return pids, err
 }
+
+// WriteCgroupProc writes the specified pid into the cgroup's cgroup.procs file
+func WriteCgroupProc(dir string, pid int) error {
+	// Normally dir should not be empty, one case is that cgroup subsystem
+	// is not mounted, we will get empty dir, and we want it fail here.
+	if dir == "" {
+		return fmt.Errorf("no such directory for %s", CgroupProcesses)
+	}
+
+	// Dont attach any pid to the cgroup if -1 is specified as a pid
+	if pid != -1 {
+		if err := ioutil.WriteFile(filepath.Join(dir, CgroupProcesses), []byte(strconv.Itoa(pid)), 0700); err != nil {
+			return fmt.Errorf("failed to write %v to %v: %v", pid, CgroupProcesses, err)
+		}
+	}
+	return nil
+}

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

@@ -0,0 +1,333 @@
+// +build linux
+
+package cgroups
+
+import (
+	"bytes"
+	"fmt"
+	"reflect"
+	"strings"
+	"testing"
+)
+
+const fedoraMountinfo = `15 35 0:3 / /proc rw,nosuid,nodev,noexec,relatime shared:5 - proc proc rw
+16 35 0:14 / /sys rw,nosuid,nodev,noexec,relatime shared:6 - sysfs sysfs rw,seclabel
+17 35 0:5 / /dev rw,nosuid shared:2 - devtmpfs devtmpfs rw,seclabel,size=8056484k,nr_inodes=2014121,mode=755
+18 16 0:15 / /sys/kernel/security rw,nosuid,nodev,noexec,relatime shared:7 - securityfs securityfs rw
+19 16 0:13 / /sys/fs/selinux rw,relatime shared:8 - selinuxfs selinuxfs rw
+20 17 0:16 / /dev/shm rw,nosuid,nodev shared:3 - tmpfs tmpfs rw,seclabel
+21 17 0:10 / /dev/pts rw,nosuid,noexec,relatime shared:4 - devpts devpts rw,seclabel,gid=5,mode=620,ptmxmode=000
+22 35 0:17 / /run rw,nosuid,nodev shared:21 - tmpfs tmpfs rw,seclabel,mode=755
+23 16 0:18 / /sys/fs/cgroup rw,nosuid,nodev,noexec shared:9 - tmpfs tmpfs rw,seclabel,mode=755
+24 23 0:19 / /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime shared:10 - cgroup cgroup rw,xattr,release_agent=/usr/lib/systemd/systemd-cgroups-agent,name=systemd
+25 16 0:20 / /sys/fs/pstore rw,nosuid,nodev,noexec,relatime shared:20 - pstore pstore rw
+26 23 0:21 / /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime shared:11 - cgroup cgroup rw,cpuset,clone_children
+27 23 0:22 / /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime shared:12 - cgroup cgroup rw,cpuacct,cpu,clone_children
+28 23 0:23 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:13 - cgroup cgroup rw,memory,clone_children
+29 23 0:24 / /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime shared:14 - cgroup cgroup rw,devices,clone_children
+30 23 0:25 / /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime shared:15 - cgroup cgroup rw,freezer,clone_children
+31 23 0:26 / /sys/fs/cgroup/net_cls rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,net_cls,clone_children
+32 23 0:27 / /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime shared:17 - cgroup cgroup rw,blkio,clone_children
+33 23 0:28 / /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime shared:18 - cgroup cgroup rw,perf_event,clone_children
+34 23 0:29 / /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime shared:19 - cgroup cgroup rw,hugetlb,clone_children
+35 1 253:2 / / rw,relatime shared:1 - ext4 /dev/mapper/ssd-root--f20 rw,seclabel,data=ordered
+36 15 0:30 / /proc/sys/fs/binfmt_misc rw,relatime shared:22 - autofs systemd-1 rw,fd=38,pgrp=1,timeout=300,minproto=5,maxproto=5,direct
+37 17 0:12 / /dev/mqueue rw,relatime shared:23 - mqueue mqueue rw,seclabel
+38 35 0:31 / /tmp rw shared:24 - tmpfs tmpfs rw,seclabel
+39 17 0:32 / /dev/hugepages rw,relatime shared:25 - hugetlbfs hugetlbfs rw,seclabel
+40 16 0:7 / /sys/kernel/debug rw,relatime shared:26 - debugfs debugfs rw
+41 16 0:33 / /sys/kernel/config rw,relatime shared:27 - configfs configfs rw
+42 35 0:34 / /var/lib/nfs/rpc_pipefs rw,relatime shared:28 - rpc_pipefs sunrpc rw
+43 15 0:35 / /proc/fs/nfsd rw,relatime shared:29 - nfsd sunrpc rw
+45 35 8:17 / /boot rw,relatime shared:30 - ext4 /dev/sdb1 rw,seclabel,data=ordered
+46 35 253:4 / /home rw,relatime shared:31 - ext4 /dev/mapper/ssd-home rw,seclabel,data=ordered
+47 35 253:5 / /var/lib/libvirt/images rw,noatime,nodiratime shared:32 - ext4 /dev/mapper/ssd-virt rw,seclabel,discard,data=ordered
+48 35 253:12 / /mnt/old rw,relatime shared:33 - ext4 /dev/mapper/HelpDeskRHEL6-FedoraRoot rw,seclabel,data=ordered
+121 22 0:36 / /run/user/1000/gvfs rw,nosuid,nodev,relatime shared:104 - fuse.gvfsd-fuse gvfsd-fuse rw,user_id=1000,group_id=1000
+124 16 0:37 / /sys/fs/fuse/connections rw,relatime shared:107 - fusectl fusectl rw
+165 38 253:3 / /tmp/mnt rw,relatime shared:147 - ext4 /dev/mapper/ssd-root rw,seclabel,data=ordered
+167 35 253:15 / /var/lib/docker/devicemapper/mnt/aae4076022f0e2b80a2afbf8fc6df450c52080191fcef7fb679a73e6f073e5c2 rw,relatime shared:149 - ext4 /dev/mapper/docker-253:2-425882-aae4076022f0e2b80a2afbf8fc6df450c52080191fcef7fb679a73e6f073e5c2 rw,seclabel,discard,stripe=16,data=ordered
+171 35 253:16 / /var/lib/docker/devicemapper/mnt/c71be651f114db95180e472f7871b74fa597ee70a58ccc35cb87139ddea15373 rw,relatime shared:153 - ext4 /dev/mapper/docker-253:2-425882-c71be651f114db95180e472f7871b74fa597ee70a58ccc35cb87139ddea15373 rw,seclabel,discard,stripe=16,data=ordered
+175 35 253:17 / /var/lib/docker/devicemapper/mnt/1bac6ab72862d2d5626560df6197cf12036b82e258c53d981fa29adce6f06c3c rw,relatime shared:157 - ext4 /dev/mapper/docker-253:2-425882-1bac6ab72862d2d5626560df6197cf12036b82e258c53d981fa29adce6f06c3c rw,seclabel,discard,stripe=16,data=ordered
+179 35 253:18 / /var/lib/docker/devicemapper/mnt/d710a357d77158e80d5b2c55710ae07c94e76d34d21ee7bae65ce5418f739b09 rw,relatime shared:161 - ext4 /dev/mapper/docker-253:2-425882-d710a357d77158e80d5b2c55710ae07c94e76d34d21ee7bae65ce5418f739b09 rw,seclabel,discard,stripe=16,data=ordered
+183 35 253:19 / /var/lib/docker/devicemapper/mnt/6479f52366114d5f518db6837254baab48fab39f2ac38d5099250e9a6ceae6c7 rw,relatime shared:165 - ext4 /dev/mapper/docker-253:2-425882-6479f52366114d5f518db6837254baab48fab39f2ac38d5099250e9a6ceae6c7 rw,seclabel,discard,stripe=16,data=ordered
+187 35 253:20 / /var/lib/docker/devicemapper/mnt/8d9df91c4cca5aef49eeb2725292aab324646f723a7feab56be34c2ad08268e1 rw,relatime shared:169 - ext4 /dev/mapper/docker-253:2-425882-8d9df91c4cca5aef49eeb2725292aab324646f723a7feab56be34c2ad08268e1 rw,seclabel,discard,stripe=16,data=ordered
+191 35 253:21 / /var/lib/docker/devicemapper/mnt/c8240b768603d32e920d365dc9d1dc2a6af46cd23e7ae819947f969e1b4ec661 rw,relatime shared:173 - ext4 /dev/mapper/docker-253:2-425882-c8240b768603d32e920d365dc9d1dc2a6af46cd23e7ae819947f969e1b4ec661 rw,seclabel,discard,stripe=16,data=ordered
+195 35 253:22 / /var/lib/docker/devicemapper/mnt/2eb3a01278380bbf3ed12d86ac629eaa70a4351301ee307a5cabe7b5f3b1615f rw,relatime shared:177 - ext4 /dev/mapper/docker-253:2-425882-2eb3a01278380bbf3ed12d86ac629eaa70a4351301ee307a5cabe7b5f3b1615f rw,seclabel,discard,stripe=16,data=ordered
+199 35 253:23 / /var/lib/docker/devicemapper/mnt/37a17fb7c9d9b80821235d5f2662879bd3483915f245f9b49cdaa0e38779b70b rw,relatime shared:181 - ext4 /dev/mapper/docker-253:2-425882-37a17fb7c9d9b80821235d5f2662879bd3483915f245f9b49cdaa0e38779b70b rw,seclabel,discard,stripe=16,data=ordered
+203 35 253:24 / /var/lib/docker/devicemapper/mnt/aea459ae930bf1de913e2f29428fd80ee678a1e962d4080019d9f9774331ee2b rw,relatime shared:185 - ext4 /dev/mapper/docker-253:2-425882-aea459ae930bf1de913e2f29428fd80ee678a1e962d4080019d9f9774331ee2b rw,seclabel,discard,stripe=16,data=ordered
+207 35 253:25 / /var/lib/docker/devicemapper/mnt/928ead0bc06c454bd9f269e8585aeae0a6bd697f46dc8754c2a91309bc810882 rw,relatime shared:189 - ext4 /dev/mapper/docker-253:2-425882-928ead0bc06c454bd9f269e8585aeae0a6bd697f46dc8754c2a91309bc810882 rw,seclabel,discard,stripe=16,data=ordered
+211 35 253:26 / /var/lib/docker/devicemapper/mnt/0f284d18481d671644706e7a7244cbcf63d590d634cc882cb8721821929d0420 rw,relatime shared:193 - ext4 /dev/mapper/docker-253:2-425882-0f284d18481d671644706e7a7244cbcf63d590d634cc882cb8721821929d0420 rw,seclabel,discard,stripe=16,data=ordered
+215 35 253:27 / /var/lib/docker/devicemapper/mnt/d9dd16722ab34c38db2733e23f69e8f4803ce59658250dd63e98adff95d04919 rw,relatime shared:197 - ext4 /dev/mapper/docker-253:2-425882-d9dd16722ab34c38db2733e23f69e8f4803ce59658250dd63e98adff95d04919 rw,seclabel,discard,stripe=16,data=ordered
+219 35 253:28 / /var/lib/docker/devicemapper/mnt/bc4500479f18c2c08c21ad5282e5f826a016a386177d9874c2764751c031d634 rw,relatime shared:201 - ext4 /dev/mapper/docker-253:2-425882-bc4500479f18c2c08c21ad5282e5f826a016a386177d9874c2764751c031d634 rw,seclabel,discard,stripe=16,data=ordered
+223 35 253:29 / /var/lib/docker/devicemapper/mnt/7770c8b24eb3d5cc159a065910076938910d307ab2f5d94e1dc3b24c06ee2c8a rw,relatime shared:205 - ext4 /dev/mapper/docker-253:2-425882-7770c8b24eb3d5cc159a065910076938910d307ab2f5d94e1dc3b24c06ee2c8a rw,seclabel,discard,stripe=16,data=ordered
+227 35 253:30 / /var/lib/docker/devicemapper/mnt/c280cd3d0bf0aa36b478b292279671624cceafc1a67eaa920fa1082601297adf rw,relatime shared:209 - ext4 /dev/mapper/docker-253:2-425882-c280cd3d0bf0aa36b478b292279671624cceafc1a67eaa920fa1082601297adf rw,seclabel,discard,stripe=16,data=ordered
+231 35 253:31 / /var/lib/docker/devicemapper/mnt/8b59a7d9340279f09fea67fd6ad89ddef711e9e7050eb647984f8b5ef006335f rw,relatime shared:213 - ext4 /dev/mapper/docker-253:2-425882-8b59a7d9340279f09fea67fd6ad89ddef711e9e7050eb647984f8b5ef006335f rw,seclabel,discard,stripe=16,data=ordered
+235 35 253:32 / /var/lib/docker/devicemapper/mnt/1a28059f29eda821578b1bb27a60cc71f76f846a551abefabce6efd0146dce9f rw,relatime shared:217 - ext4 /dev/mapper/docker-253:2-425882-1a28059f29eda821578b1bb27a60cc71f76f846a551abefabce6efd0146dce9f rw,seclabel,discard,stripe=16,data=ordered
+239 35 253:33 / /var/lib/docker/devicemapper/mnt/e9aa60c60128cad1 rw,relatime shared:221 - ext4 /dev/mapper/docker-253:2-425882-e9aa60c60128cad1 rw,seclabel,discard,stripe=16,data=ordered
+243 35 253:34 / /var/lib/docker/devicemapper/mnt/5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d-init rw,relatime shared:225 - ext4 /dev/mapper/docker-253:2-425882-5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d-init rw,seclabel,discard,stripe=16,data=ordered
+247 35 253:35 / /var/lib/docker/devicemapper/mnt/5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d rw,relatime shared:229 - ext4 /dev/mapper/docker-253:2-425882-5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d rw,seclabel,discard,stripe=16,data=ordered
+31 21 0:23 / /DATA/foo_bla_bla rw,relatime - cifs //foo/BLA\040BLA\040BLA/ rw,sec=ntlm,cache=loose,unc=\\foo\BLA BLA BLA,username=my_login,domain=mydomain.com,uid=12345678,forceuid,gid=12345678,forcegid,addr=10.1.30.10,file_mode=0755,dir_mode=0755,nounix,rsize=61440,wsize=65536,actimeo=1`
+
+const systemdMountinfo = `115 83 0:32 / / rw,relatime - aufs none rw,si=c0bd3d3,dio,dirperm1
+116 115 0:35 / /proc rw,nosuid,nodev,noexec,relatime - proc proc rw
+117 115 0:36 / /dev rw,nosuid - tmpfs tmpfs rw,mode=755
+118 117 0:37 / /dev/pts rw,nosuid,noexec,relatime - devpts devpts rw,gid=5,mode=620,ptmxmode=666
+119 115 0:38 / /sys rw,nosuid,nodev,noexec,relatime - sysfs sysfs rw
+120 119 0:39 / /sys/fs/cgroup rw,nosuid,nodev,noexec,relatime - tmpfs tmpfs rw,mode=755
+121 120 0:19 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,xattr,release_agent=/lib/systemd/systemd-cgroups-agent,name=systemd
+122 120 0:20 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,devices
+123 120 0:21 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,freezer
+124 120 0:22 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,memory
+125 120 0:23 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/net_cls,net_prio rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,net_cls,net_prio
+126 120 0:24 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,blkio
+127 120 0:25 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,cpuset,clone_children
+128 120 0:26 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,cpu,cpuacct
+129 120 0:27 /system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,perf_event,release_agent=/run/cgmanager/agents/cgm-release-agent.perf_event
+130 115 43:0 /var/lib/docker/volumes/a44a712176377f57c094397330ee04387284c478364eb25f4c3d25f775f25c26/_data /var/lib/docker rw,relatime - ext4 /dev/nbd0 rw,data=ordered
+131 115 43:0 /var/lib/docker/containers/dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e/resolv.conf /etc/resolv.conf rw,relatime - ext4 /dev/nbd0 rw,data=ordered
+132 115 43:0 /var/lib/docker/containers/dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e/hostname /etc/hostname rw,relatime - ext4 /dev/nbd0 rw,data=ordered
+133 115 43:0 /var/lib/docker/containers/dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e/hosts /etc/hosts rw,relatime - ext4 /dev/nbd0 rw,data=ordered
+134 117 0:33 / /dev/shm rw,nosuid,nodev,noexec,relatime - tmpfs shm rw,size=65536k
+135 117 0:13 / /dev/mqueue rw,nosuid,nodev,noexec,relatime - mqueue mqueue rw
+136 117 0:12 /1 /dev/console rw,nosuid,noexec,relatime - devpts none rw,gid=5,mode=620,ptmxmode=000
+84 115 0:40 / /tmp rw,relatime - tmpfs none rw`
+
+const cgroup2Mountinfo = `18 64 0:18 / /sys rw,nosuid,nodev,noexec,relatime shared:6 - sysfs sysfs rw,seclabel
+19 64 0:4 / /proc rw,nosuid,nodev,noexec,relatime shared:5 - proc proc rw
+20 64 0:6 / /dev rw,nosuid shared:2 - devtmpfs devtmpfs rw,seclabel,size=8171204k,nr_inodes=2042801,mode=755
+21 18 0:19 / /sys/kernel/security rw,nosuid,nodev,noexec,relatime shared:7 - securityfs securityfs rw
+22 20 0:20 / /dev/shm rw,nosuid,nodev shared:3 - tmpfs tmpfs rw,seclabel
+23 20 0:21 / /dev/pts rw,nosuid,noexec,relatime shared:4 - devpts devpts rw,seclabel,gid=5,mode=620,ptmxmode=000
+24 64 0:22 / /run rw,nosuid,nodev shared:24 - tmpfs tmpfs rw,seclabel,mode=755
+25 18 0:23 / /sys/fs/cgroup ro,nosuid,nodev,noexec shared:8 - tmpfs tmpfs ro,seclabel,mode=755
+26 25 0:24 / /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime shared:9 - cgroup2 cgroup rw
+27 18 0:25 / /sys/fs/pstore rw,nosuid,nodev,noexec,relatime shared:20 - pstore pstore rw,seclabel
+28 18 0:26 / /sys/firmware/efi/efivars rw,nosuid,nodev,noexec,relatime shared:21 - efivarfs efivarfs rw
+29 25 0:27 / /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime shared:10 - cgroup cgroup rw,cpu,cpuacct
+30 25 0:28 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:11 - cgroup cgroup rw,memory
+31 25 0:29 / /sys/fs/cgroup/net_cls,net_prio rw,nosuid,nodev,noexec,relatime shared:12 - cgroup cgroup rw,net_cls,net_prio
+32 25 0:30 / /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime shared:13 - cgroup cgroup rw,blkio
+33 25 0:31 / /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime shared:14 - cgroup cgroup rw,perf_event
+34 25 0:32 / /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime shared:15 - cgroup cgroup rw,hugetlb
+35 25 0:33 / /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,freezer
+36 25 0:34 / /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime shared:17 - cgroup cgroup rw,cpuset
+37 25 0:35 / /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime shared:18 - cgroup cgroup rw,devices
+38 25 0:36 / /sys/fs/cgroup/pids rw,nosuid,nodev,noexec,relatime shared:19 - cgroup cgroup rw,pids
+61 18 0:37 / /sys/kernel/config rw,relatime shared:22 - configfs configfs rw
+64 0 253:0 / / rw,relatime shared:1 - ext4 /dev/mapper/fedora_dhcp--16--129-root rw,seclabel,data=ordered
+39 18 0:17 / /sys/fs/selinux rw,relatime shared:23 - selinuxfs selinuxfs rw
+40 20 0:16 / /dev/mqueue rw,relatime shared:25 - mqueue mqueue rw,seclabel
+41 20 0:39 / /dev/hugepages rw,relatime shared:26 - hugetlbfs hugetlbfs rw,seclabel
+`
+
+func TestGetCgroupMounts(t *testing.T) {
+	type testData struct {
+		mountInfo  string
+		root       string
+		subsystems map[string]bool
+	}
+	testTable := []testData{
+		{
+			mountInfo: fedoraMountinfo,
+			root:      "/",
+			subsystems: map[string]bool{
+				"cpuset":     true,
+				"cpu":        true,
+				"cpuacct":    true,
+				"memory":     true,
+				"devices":    true,
+				"freezer":    true,
+				"net_cls":    true,
+				"blkio":      true,
+				"perf_event": true,
+				"hugetlb":    true,
+			},
+		},
+		{
+			mountInfo: systemdMountinfo,
+			root:      "/system.slice/docker-dc4eaa1a34ec4d593bc0125d31eea823a1d76ae483aeb1409cca80304e34da2e.scope",
+			subsystems: map[string]bool{
+				"cpuset":     true,
+				"cpu":        true,
+				"cpuacct":    true,
+				"memory":     true,
+				"devices":    true,
+				"freezer":    true,
+				"net_cls":    true,
+				"blkio":      true,
+				"perf_event": true,
+			},
+		},
+	}
+	for _, td := range testTable {
+		mi := bytes.NewBufferString(td.mountInfo)
+		cgMounts, err := getCgroupMountsHelper(td.subsystems, mi, false)
+		if err != nil {
+			t.Fatal(err)
+		}
+		cgMap := make(map[string]Mount)
+		for _, m := range cgMounts {
+			for _, ss := range m.Subsystems {
+				cgMap[ss] = m
+			}
+		}
+		for ss := range td.subsystems {
+			m, ok := cgMap[ss]
+			if !ok {
+				t.Fatalf("%s not found", ss)
+			}
+			if m.Root != td.root {
+				t.Fatalf("unexpected root for %s: %s", ss, m.Root)
+			}
+			if !strings.HasPrefix(m.Mountpoint, "/sys/fs/cgroup/") && !strings.Contains(m.Mountpoint, ss) {
+				t.Fatalf("unexpected mountpoint for %s: %s", ss, m.Mountpoint)
+			}
+			var ssFound bool
+			for _, mss := range m.Subsystems {
+				if mss == ss {
+					ssFound = true
+					break
+				}
+			}
+			if !ssFound {
+				t.Fatalf("subsystem %s not found in Subsystems field %v", ss, m.Subsystems)
+			}
+		}
+	}
+}
+
+func BenchmarkGetCgroupMounts(b *testing.B) {
+	subsystems := map[string]bool{
+		"cpuset":     true,
+		"cpu":        true,
+		"cpuacct":    true,
+		"memory":     true,
+		"devices":    true,
+		"freezer":    true,
+		"net_cls":    true,
+		"blkio":      true,
+		"perf_event": true,
+		"hugetlb":    true,
+	}
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		b.StopTimer()
+		mi := bytes.NewBufferString(fedoraMountinfo)
+		b.StartTimer()
+		if _, err := getCgroupMountsHelper(subsystems, mi, false); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+func TestParseCgroupString(t *testing.T) {
+	testCases := []struct {
+		input          string
+		expectedError  error
+		expectedOutput map[string]string
+	}{
+		{
+			// Taken from a CoreOS instance running systemd 225 with CPU/Mem
+			// accounting enabled in systemd
+			input: `9:blkio:/
+8:freezer:/
+7:perf_event:/
+6:devices:/system.slice/system-sshd.slice
+5:cpuset:/
+4:cpu,cpuacct:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service
+3:net_cls,net_prio:/
+2:memory:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service
+1:name=systemd:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service`,
+			expectedOutput: map[string]string{
+				"name=systemd": "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
+				"blkio":        "/",
+				"freezer":      "/",
+				"perf_event":   "/",
+				"devices":      "/system.slice/system-sshd.slice",
+				"cpuset":       "/",
+				"cpu":          "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
+				"cpuacct":      "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
+				"net_cls":      "/",
+				"net_prio":     "/",
+				"memory":       "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
+			},
+		},
+		{
+			input:         `malformed input`,
+			expectedError: fmt.Errorf(`invalid cgroup entry: must contain at least two colons: malformed input`),
+		},
+	}
+
+	for ndx, testCase := range testCases {
+		out, err := parseCgroupFromReader(strings.NewReader(testCase.input))
+		if err != nil {
+			if testCase.expectedError == nil || testCase.expectedError.Error() != err.Error() {
+				t.Errorf("%v: expected error %v, got error %v", ndx, testCase.expectedError, err)
+			}
+		} else {
+			if !reflect.DeepEqual(testCase.expectedOutput, out) {
+				t.Errorf("%v: expected output %v, got error %v", ndx, testCase.expectedOutput, out)
+			}
+		}
+	}
+
+}
+
+func TestIgnoreCgroup2Mount(t *testing.T) {
+	subsystems := map[string]bool{
+		"cpuset":       true,
+		"cpu":          true,
+		"cpuacct":      true,
+		"memory":       true,
+		"devices":      true,
+		"freezer":      true,
+		"net_cls":      true,
+		"blkio":        true,
+		"perf_event":   true,
+		"pids":         true,
+		"name=systemd": true,
+	}
+
+	mi := bytes.NewBufferString(cgroup2Mountinfo)
+	cgMounts, err := getCgroupMountsHelper(subsystems, mi, false)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, m := range cgMounts {
+		if m.Mountpoint == "/sys/fs/cgroup/systemd" {
+			t.Errorf("parsed a cgroup2 mount at /sys/fs/cgroup/systemd instead of ignoring it")
+		}
+	}
+}
+
+const fakeMountInfo = ` 18 24 0:17 / /sys rw,nosuid,nodev,noexec,relatime - sysfs sysfs rw
+100 99 1:31 / /foo/bar rw,relatime - fake fake rw,fake
+100 99 1:31 / /foo/bar/baz2 rw,relatime - fake fake rw,fake
+100 99 1:31 / /foo/bar/baz rw,relatime - fake fake rw,fake
+100 99 1:31 / /foo/bar/bazza rw,relatime - fake fake rw,fake
+100 99 1:31 / /foo/bar/baz3 rw,relatime - fake fake rw,fake
+100 99 1:31 / /foo rw,relatime - fake fake rw,fake
+100 99 1:31 / /unrelated rw,relatime - fake fake rw,fake
+100 99 1:31 / / rw,relatime - fake fake rw,fake
+`
+
+func TestGetClosestMountpointAncestor(t *testing.T) {
+	testCases := []struct {
+		input      string
+		mountinfos string
+		output     string
+	}{
+		{input: "/foo/bar/baz/a/b/c", mountinfos: fakeMountInfo, output: "/foo/bar/baz"},
+		{input: "/foo/bar/baz", mountinfos: fakeMountInfo, output: "/foo/bar/baz"},
+		{input: "/foo/bar/bazza", mountinfos: fakeMountInfo, output: "/foo/bar/bazza"},
+		{input: "/a/b/c/d", mountinfos: fakeMountInfo, output: "/"},
+	}
+
+	for _, c := range testCases {
+		mountpoint := GetClosestMountpointAncestor(c.input, c.mountinfos)
+		if mountpoint != c.output {
+			t.Errorf("expected %s, got %s", c.output, mountpoint)
+		}
+	}
+}