update cadvisor, docker, and runc godeps

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

@@ -34,6 +34,7 @@ filegroup(
     srcs = [
         ":package-srcs",
         "//vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs:all-srcs",
+        "//vendor/github.com/opencontainers/runc/libcontainer/cgroups/rootless:all-srcs",
         "//vendor/github.com/opencontainers/runc/libcontainer/cgroups/systemd:all-srcs",
     ],
     tags = ["automanaged"],

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
 

vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs/BUILD

@@ -31,6 +31,7 @@ go_library(
             "//vendor/github.com/opencontainers/runc/libcontainer/configs:go_default_library",
             "//vendor/github.com/opencontainers/runc/libcontainer/system:go_default_library",
             "//vendor/github.com/opencontainers/runc/libcontainer/utils:go_default_library",
+            "//vendor/golang.org/x/sys/unix:go_default_library",
         ],
         "//conditions:default": [],
     }),

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

@@ -114,8 +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 {
@@ -124,17 +124,12 @@ 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)
 	}
 
-	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
@@ -147,9 +142,12 @@ func (m *Manager) Apply(pid int) (err error) {
 			}
 			return err
 		}
-		paths[sys.Name()] = p
+		m.Paths[sys.Name()] = p
+
+		if err := sys.Apply(d); err != nil {
+			return err
+		}
 	}
-	m.Paths = paths
 	return nil
 }
 
@@ -269,25 +267,8 @@ 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
@@ -299,7 +280,10 @@ func (raw *cgroupData) path(subsystem string) (string, error) {
 		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
 	}
@@ -348,8 +332,8 @@ 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/cpu.go

@@ -55,7 +55,7 @@ func (s *CpuGroup) ApplyDir(path string, cgroup *configs.Cgroup, pid int) error
 
 func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
 	if cgroup.Resources.CpuRtPeriod != 0 {
-		if err := writeFile(path, "cpu.rt_period_us", strconv.FormatInt(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
+		if err := writeFile(path, "cpu.rt_period_us", strconv.FormatUint(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
 			return err
 		}
 	}
@@ -69,12 +69,12 @@ func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
 
 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
 		}
 	}

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

@@ -57,13 +57,31 @@ 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 := cgroups.WriteCgroupProc(dir, pid); err != nil {
@@ -136,3 +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/memory.go

@@ -10,13 +10,19 @@ import (
 	"path/filepath"
 	"strconv"
 	"strings"
-	"syscall"
+	"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"
+const (
+	cgroupKernelMemoryLimit = "memory.kmem.limit_in_bytes"
+	cgroupMemorySwapLimit   = "memory.memsw.limit_in_bytes"
+	cgroupMemoryLimit       = "memory.limit_in_bytes"
+)
 
 type MemoryGroup struct {
 }
@@ -29,14 +35,18 @@ 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
 			}
-		}
-		if d.config.KernelMemory != 0 {
+			// 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
 			}
@@ -85,7 +95,7 @@ func setKernelMemory(path string, kernelMemoryLimit int64) error {
 		// 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 == syscall.EBUSY {
+				if errNo == unix.EBUSY {
 					return fmt.Errorf("failed to set %s, because either tasks have already joined this cgroup or it has children", cgroupKernelMemoryLimit)
 				}
 			}
@@ -96,9 +106,18 @@ func setKernelMemory(path string, kernelMemoryLimit int64) error {
 }
 
 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
@@ -107,29 +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
 			}
 		}
@@ -167,12 +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
@@ -224,6 +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
 }
 
@@ -234,7 +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/rootless/BUILD

@@ -0,0 +1,35 @@
+load("@io_bazel_rules_go//go:def.bzl", "go_library")
+
+go_library(
+    name = "go_default_library",
+    srcs = select({
+        "@io_bazel_rules_go//go/platform:linux_amd64": [
+            "rootless.go",
+        ],
+        "//conditions:default": [],
+    }),
+    visibility = ["//visibility:public"],
+    deps = select({
+        "@io_bazel_rules_go//go/platform:linux_amd64": [
+            "//vendor/github.com/opencontainers/runc/libcontainer/cgroups:go_default_library",
+            "//vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs:go_default_library",
+            "//vendor/github.com/opencontainers/runc/libcontainer/configs:go_default_library",
+            "//vendor/github.com/opencontainers/runc/libcontainer/configs/validate:go_default_library",
+        ],
+        "//conditions:default": [],
+    }),
+)
+
+filegroup(
+    name = "package-srcs",
+    srcs = glob(["**"]),
+    tags = ["automanaged"],
+    visibility = ["//visibility:private"],
+)
+
+filegroup(
+    name = "all-srcs",
+    srcs = [":package-srcs"],
+    tags = ["automanaged"],
+    visibility = ["//visibility:public"],
+)

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

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

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

@@ -5,7 +5,6 @@ package systemd
 import (
 	"errors"
 	"fmt"
-	"io/ioutil"
 	"os"
 	"path/filepath"
 	"strings"
@@ -266,7 +265,14 @@ 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
+		properties = append(properties,
+			newProp("CPUQuotaPerSecUSec", cpuQuotaPerSecUSec))
 	}
 
 	if c.Resources.BlkioWeight != 0 {
@@ -327,15 +333,6 @@ 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 {
@@ -356,7 +353,6 @@ 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) {
@@ -366,7 +362,6 @@ 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 {
@@ -429,7 +424,7 @@ 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
 	}

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

@@ -23,36 +23,14 @@ const (
 
 // 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.
-	if !isSubsystemAvailable(subsystem) {
-		return "", NewNotFoundError(subsystem)
-	}
-	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)
 	}
@@ -88,6 +66,21 @@ func isSubsystemAvailable(subsystem string) bool {
 	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 {
@@ -131,7 +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")
 	}
@@ -149,7 +142,7 @@ func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount,
 		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, " ")
@@ -211,9 +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)
@@ -222,11 +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
@@ -235,8 +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
@@ -245,6 +246,31 @@ 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, CgroupProcesses))
 	if err != nil {
@@ -287,10 +313,6 @@ func parseCgroupFromReader(r io.Reader) (map[string]string, error) {
 	cgroups := make(map[string]string)
 
 	for s.Scan() {
-		if err := s.Err(); err != nil {
-			return nil, err
-		}
-
 		text := s.Text()
 		// from cgroups(7):
 		// /proc/[pid]/cgroup
@@ -307,6 +329,10 @@ func parseCgroupFromReader(r io.Reader) (map[string]string, error) {
 			cgroups[subs] = parts[2]
 		}
 	}
+	if err := s.Err(); err != nil {
+		return nil, err
+	}
+
 	return cgroups, nil
 }