Merge pull request #51 from heartlock/support-privileged

Support privileged

Godeps/Godeps.json

@@ -283,6 +283,16 @@
 			"Comment": "v1.0.0-rc4-51-ga431dbc",
 			"Rev": "a431dbcf6a74fca2e0e040b819a836dbe3fb23ca"
 		},
+		{
+			"ImportPath": "github.com/opencontainers/runc/libcontainer/configs",
+			"Comment": "v1.0.0-rc3-21-g50401b5",
+			"Rev": "50401b5b4c2e01e4f1372b73a021742deeaf4e2d"
+		},
+		{
+			"ImportPath": "github.com/opencontainers/runc/libcontainer/devices",
+			"Comment": "v1.0.0-rc3-21-g50401b5",
+			"Rev": "50401b5b4c2e01e4f1372b73a021742deeaf4e2d"
+		},
 		{
 			"ImportPath": "github.com/opencontainers/runc/libcontainer/system",
 			"Comment": "v1.0.0-rc3-21-g50401b5",

pkg/server/container_start.go

@@ -31,6 +31,7 @@ import (
 	prototypes "github.com/gogo/protobuf/types"
 	"github.com/golang/glog"
 	imagespec "github.com/opencontainers/image-spec/specs-go/v1"
+	"github.com/opencontainers/runc/libcontainer/devices"
 	runtimespec "github.com/opencontainers/runtime-spec/specs-go"
 	"github.com/opencontainers/runtime-tools/generate"
 	"golang.org/x/net/context"
@@ -250,11 +251,17 @@ func (c *criContainerdService) generateContainerSpec(id string, sandboxPid uint3
 		g.AddProcessEnv(e.GetKey(), e.GetValue())
 	}
 
-	// Add extra mounts first so that CRI specified mounts can override.
-	addOCIBindMounts(&g, append(extraMounts, config.GetMounts()...))
+	// TODO: add setOCIPrivileged group all privileged logic together
+	securityContext := config.GetLinux().GetSecurityContext()
 
-	// TODO(random-liu): [P1] Set device mapping.
-	// Ref https://github.com/moby/moby/blob/master/oci/devices_linux.go.
+	// Add extra mounts first so that CRI specified mounts can override.
+	addOCIBindMounts(&g, append(extraMounts, config.GetMounts()...), securityContext.GetPrivileged())
+
+	g.SetRootReadonly(securityContext.GetReadonlyRootfs())
+
+	if err := addOCIDevices(&g, config.GetDevices(), securityContext.GetPrivileged()); err != nil {
+		return nil, fmt.Errorf("failed to set devices mapping %+v: %v", config.GetDevices(), err)
+	}
 
 	// TODO(random-liu): [P1] Handle container logging, decorate and redirect to file.
 
@@ -267,15 +274,11 @@ func (c *criContainerdService) generateContainerSpec(id string, sandboxPid uint3
 
 	g.SetProcessTerminal(config.GetTty())
 
-	securityContext := config.GetLinux().GetSecurityContext()
-
-	if err := setOCICapabilities(&g, securityContext.GetCapabilities()); err != nil {
+	if err := setOCICapabilities(&g, securityContext.GetCapabilities(), securityContext.GetPrivileged()); err != nil {
 		return nil, fmt.Errorf("failed to set capabilities %+v: %v",
 			securityContext.GetCapabilities(), err)
 	}
 
-	// TODO(random-liu): [P0] Handle privileged.
-
 	// Set namespaces, share namespace with sandbox container.
 	setOCINamespaces(&g, securityContext.GetNamespaceOptions(), sandboxPid)
 
@@ -288,8 +291,6 @@ func (c *criContainerdService) generateContainerSpec(id string, sandboxPid uint3
 		g.AddProcessAdditionalGid(uint32(group))
 	}
 
-	g.SetRootReadonly(securityContext.GetReadonlyRootfs())
-
 	// TODO(random-liu): [P2] Add apparmor and seccomp.
 
 	return g.Spec(), nil
@@ -352,8 +353,70 @@ func addImageEnvs(g *generate.Generator, imageEnvs []string) error {
 	return nil
 }
 
+func clearReadOnly(m *runtimespec.Mount) {
+	var opt []string
+	for _, o := range m.Options {
+		if o != "ro" {
+			opt = append(opt, o)
+		}
+	}
+	m.Options = opt
+}
+
+// addDevices set device mapping.
+func addOCIDevices(g *generate.Generator, devs []*runtime.Device, privileged bool) error {
+	spec := g.Spec()
+	if privileged {
+		hostDevices, err := devices.HostDevices()
+		if err != nil {
+			return err
+		}
+		for _, hostDevice := range hostDevices {
+			rd := runtimespec.LinuxDevice{
+				Path:  hostDevice.Path,
+				Type:  string(hostDevice.Type),
+				Major: hostDevice.Major,
+				Minor: hostDevice.Minor,
+				UID:   &hostDevice.Uid,
+				GID:   &hostDevice.Gid,
+			}
+			g.AddDevice(rd)
+		}
+		spec.Linux.Resources.Devices = []runtimespec.LinuxDeviceCgroup{
+			{
+				Allow:  true,
+				Access: "rwm",
+			},
+		}
+		return nil
+	}
+	for _, device := range devs {
+		dev, err := devices.DeviceFromPath(device.HostPath, device.Permissions)
+		if err != nil {
+			return err
+		}
+		rd := runtimespec.LinuxDevice{
+			Path:  device.ContainerPath,
+			Type:  string(dev.Type),
+			Major: dev.Major,
+			Minor: dev.Minor,
+			UID:   &dev.Uid,
+			GID:   &dev.Gid,
+		}
+		g.AddDevice(rd)
+		spec.Linux.Resources.Devices = append(spec.Linux.Resources.Devices, runtimespec.LinuxDeviceCgroup{
+			Allow:  true,
+			Type:   string(dev.Type),
+			Major:  &dev.Major,
+			Minor:  &dev.Minor,
+			Access: dev.Permissions,
+		})
+	}
+	return nil
+}
+
 // addOCIBindMounts adds bind mounts.
-func addOCIBindMounts(g *generate.Generator, mounts []*runtime.Mount) {
+func addOCIBindMounts(g *generate.Generator, mounts []*runtime.Mount, privileged bool) {
 	for _, mount := range mounts {
 		dst := mount.GetContainerPath()
 		src := mount.GetHostPath()
@@ -364,6 +427,21 @@ func addOCIBindMounts(g *generate.Generator, mounts []*runtime.Mount) {
 		// TODO(random-liu): [P1] Apply selinux label
 		g.AddBindMount(src, dst, options)
 	}
+	if !privileged {
+		return
+	}
+	spec := g.Spec()
+	// clear readonly for /sys and cgroup
+	for i, m := range spec.Mounts {
+		if spec.Mounts[i].Destination == "/sys" && !spec.Root.Readonly {
+			clearReadOnly(&spec.Mounts[i])
+		}
+		if m.Type == "cgroup" {
+			clearReadOnly(&spec.Mounts[i])
+		}
+	}
+	spec.Linux.ReadonlyPaths = nil
+	spec.Linux.MaskedPaths = nil
 }
 
 // setOCILinuxResource set container resource limit.
@@ -379,7 +457,12 @@ func setOCILinuxResource(g *generate.Generator, resources *runtime.LinuxContaine
 }
 
 // setOCICapabilities adds/drops process capabilities.
-func setOCICapabilities(g *generate.Generator, capabilities *runtime.Capability) error {
+func setOCICapabilities(g *generate.Generator, capabilities *runtime.Capability, privileged bool) error {
+	if privileged {
+		// Add all capabilities in privileged mode.
+		g.SetupPrivileged(true)
+		return nil
+	}
 	if capabilities == nil {
 		return nil
 	}
@@ -395,7 +478,6 @@ func setOCICapabilities(g *generate.Generator, capabilities *runtime.Capability)
 			return err
 		}
 	}
-
 	return nil
 }
 

vendor/github.com/opencontainers/runc/libcontainer/configs/blkio_device.go

@@ -0,0 +1,61 @@
+package configs
+
+import "fmt"
+
+// blockIODevice holds major:minor format supported in blkio cgroup
+type blockIODevice struct {
+	// Major is the device's major number
+	Major int64 `json:"major"`
+	// Minor is the device's minor number
+	Minor int64 `json:"minor"`
+}
+
+// WeightDevice struct holds a `major:minor weight`|`major:minor leaf_weight` pair
+type WeightDevice struct {
+	blockIODevice
+	// Weight is the bandwidth rate for the device, range is from 10 to 1000
+	Weight uint16 `json:"weight"`
+	// LeafWeight is the bandwidth rate for the device while competing with the cgroup's child cgroups, range is from 10 to 1000, cfq scheduler only
+	LeafWeight uint16 `json:"leafWeight"`
+}
+
+// NewWeightDevice returns a configured WeightDevice pointer
+func NewWeightDevice(major, minor int64, weight, leafWeight uint16) *WeightDevice {
+	wd := &WeightDevice{}
+	wd.Major = major
+	wd.Minor = minor
+	wd.Weight = weight
+	wd.LeafWeight = leafWeight
+	return wd
+}
+
+// WeightString formats the struct to be writable to the cgroup specific file
+func (wd *WeightDevice) WeightString() string {
+	return fmt.Sprintf("%d:%d %d", wd.Major, wd.Minor, wd.Weight)
+}
+
+// LeafWeightString formats the struct to be writable to the cgroup specific file
+func (wd *WeightDevice) LeafWeightString() string {
+	return fmt.Sprintf("%d:%d %d", wd.Major, wd.Minor, wd.LeafWeight)
+}
+
+// ThrottleDevice struct holds a `major:minor rate_per_second` pair
+type ThrottleDevice struct {
+	blockIODevice
+	// Rate is the IO rate limit per cgroup per device
+	Rate uint64 `json:"rate"`
+}
+
+// NewThrottleDevice returns a configured ThrottleDevice pointer
+func NewThrottleDevice(major, minor int64, rate uint64) *ThrottleDevice {
+	td := &ThrottleDevice{}
+	td.Major = major
+	td.Minor = minor
+	td.Rate = rate
+	return td
+}
+
+// String formats the struct to be writable to the cgroup specific file
+func (td *ThrottleDevice) String() string {
+	return fmt.Sprintf("%d:%d %d", td.Major, td.Minor, td.Rate)
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/cgroup_unix.go

@@ -0,0 +1,124 @@
+// +build linux freebsd
+
+package configs
+
+type FreezerState string
+
+const (
+	Undefined FreezerState = ""
+	Frozen    FreezerState = "FROZEN"
+	Thawed    FreezerState = "THAWED"
+)
+
+type Cgroup struct {
+	// Deprecated, use Path instead
+	Name string `json:"name,omitempty"`
+
+	// name of parent of cgroup or slice
+	// Deprecated, use Path instead
+	Parent string `json:"parent,omitempty"`
+
+	// Path specifies the path to cgroups that are created and/or joined by the container.
+	// The path is assumed to be relative to the host system cgroup mountpoint.
+	Path string `json:"path"`
+
+	// ScopePrefix describes prefix for the scope name
+	ScopePrefix string `json:"scope_prefix"`
+
+	// Paths represent the absolute cgroups paths to join.
+	// This takes precedence over Path.
+	Paths map[string]string
+
+	// Resources contains various cgroups settings to apply
+	*Resources
+}
+
+type Resources struct {
+	// If this is true allow access to any kind of device within the container.  If false, allow access only to devices explicitly listed in the allowed_devices list.
+	// Deprecated
+	AllowAllDevices *bool `json:"allow_all_devices,omitempty"`
+	// Deprecated
+	AllowedDevices []*Device `json:"allowed_devices,omitempty"`
+	// Deprecated
+	DeniedDevices []*Device `json:"denied_devices,omitempty"`
+
+	Devices []*Device `json:"devices"`
+
+	// Memory limit (in bytes)
+	Memory uint64 `json:"memory"`
+
+	// Memory reservation or soft_limit (in bytes)
+	MemoryReservation uint64 `json:"memory_reservation"`
+
+	// Total memory usage (memory + swap); set `-1` to enable unlimited swap
+	MemorySwap uint64 `json:"memory_swap"`
+
+	// Kernel memory limit (in bytes)
+	KernelMemory uint64 `json:"kernel_memory"`
+
+	// Kernel memory limit for TCP use (in bytes)
+	KernelMemoryTCP uint64 `json:"kernel_memory_tcp"`
+
+	// CPU shares (relative weight vs. other containers)
+	CpuShares uint64 `json:"cpu_shares"`
+
+	// CPU hardcap limit (in usecs). Allowed cpu time in a given period.
+	CpuQuota int64 `json:"cpu_quota"`
+
+	// CPU period to be used for hardcapping (in usecs). 0 to use system default.
+	CpuPeriod uint64 `json:"cpu_period"`
+
+	// How many time CPU will use in realtime scheduling (in usecs).
+	CpuRtRuntime int64 `json:"cpu_rt_quota"`
+
+	// CPU period to be used for realtime scheduling (in usecs).
+	CpuRtPeriod uint64 `json:"cpu_rt_period"`
+
+	// CPU to use
+	CpusetCpus string `json:"cpuset_cpus"`
+
+	// MEM to use
+	CpusetMems string `json:"cpuset_mems"`
+
+	// Process limit; set <= `0' to disable limit.
+	PidsLimit int64 `json:"pids_limit"`
+
+	// Specifies per cgroup weight, range is from 10 to 1000.
+	BlkioWeight uint16 `json:"blkio_weight"`
+
+	// Specifies tasks' weight in the given cgroup while competing with the cgroup's child cgroups, range is from 10 to 1000, cfq scheduler only
+	BlkioLeafWeight uint16 `json:"blkio_leaf_weight"`
+
+	// Weight per cgroup per device, can override BlkioWeight.
+	BlkioWeightDevice []*WeightDevice `json:"blkio_weight_device"`
+
+	// IO read rate limit per cgroup per device, bytes per second.
+	BlkioThrottleReadBpsDevice []*ThrottleDevice `json:"blkio_throttle_read_bps_device"`
+
+	// IO write rate limit per cgroup per device, bytes per second.
+	BlkioThrottleWriteBpsDevice []*ThrottleDevice `json:"blkio_throttle_write_bps_device"`
+
+	// IO read rate limit per cgroup per device, IO per second.
+	BlkioThrottleReadIOPSDevice []*ThrottleDevice `json:"blkio_throttle_read_iops_device"`
+
+	// IO write rate limit per cgroup per device, IO per second.
+	BlkioThrottleWriteIOPSDevice []*ThrottleDevice `json:"blkio_throttle_write_iops_device"`
+
+	// set the freeze value for the process
+	Freezer FreezerState `json:"freezer"`
+
+	// Hugetlb limit (in bytes)
+	HugetlbLimit []*HugepageLimit `json:"hugetlb_limit"`
+
+	// Whether to disable OOM Killer
+	OomKillDisable bool `json:"oom_kill_disable"`
+
+	// Tuning swappiness behaviour per cgroup
+	MemorySwappiness *uint64 `json:"memory_swappiness"`
+
+	// Set priority of network traffic for container
+	NetPrioIfpriomap []*IfPrioMap `json:"net_prio_ifpriomap"`
+
+	// Set class identifier for container's network packets
+	NetClsClassid uint32 `json:"net_cls_classid_u"`
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/cgroup_unsupported.go

@@ -0,0 +1,6 @@
+// +build !windows,!linux,!freebsd
+
+package configs
+
+type Cgroup struct {
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/cgroup_windows.go

@@ -0,0 +1,6 @@
+package configs
+
+// TODO Windows: This can ultimately be entirely factored out on Windows as
+// cgroups are a Unix-specific construct.
+type Cgroup struct {
+}

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

@@ -0,0 +1,343 @@
+package configs
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"time"
+
+	"github.com/Sirupsen/logrus"
+	"github.com/opencontainers/runtime-spec/specs-go"
+)
+
+type Rlimit struct {
+	Type int    `json:"type"`
+	Hard uint64 `json:"hard"`
+	Soft uint64 `json:"soft"`
+}
+
+// IDMap represents UID/GID Mappings for User Namespaces.
+type IDMap struct {
+	ContainerID int `json:"container_id"`
+	HostID      int `json:"host_id"`
+	Size        int `json:"size"`
+}
+
+// Seccomp represents syscall restrictions
+// By default, only the native architecture of the kernel is allowed to be used
+// for syscalls. Additional architectures can be added by specifying them in
+// Architectures.
+type Seccomp struct {
+	DefaultAction Action     `json:"default_action"`
+	Architectures []string   `json:"architectures"`
+	Syscalls      []*Syscall `json:"syscalls"`
+}
+
+// Action is taken upon rule match in Seccomp
+type Action int
+
+const (
+	Kill Action = iota + 1
+	Errno
+	Trap
+	Allow
+	Trace
+)
+
+// Operator is a comparison operator to be used when matching syscall arguments in Seccomp
+type Operator int
+
+const (
+	EqualTo Operator = iota + 1
+	NotEqualTo
+	GreaterThan
+	GreaterThanOrEqualTo
+	LessThan
+	LessThanOrEqualTo
+	MaskEqualTo
+)
+
+// Arg is a rule to match a specific syscall argument in Seccomp
+type Arg struct {
+	Index    uint     `json:"index"`
+	Value    uint64   `json:"value"`
+	ValueTwo uint64   `json:"value_two"`
+	Op       Operator `json:"op"`
+}
+
+// Syscall is a rule to match a syscall in Seccomp
+type Syscall struct {
+	Name   string `json:"name"`
+	Action Action `json:"action"`
+	Args   []*Arg `json:"args"`
+}
+
+// TODO Windows. Many of these fields should be factored out into those parts
+// which are common across platforms, and those which are platform specific.
+
+// Config defines configuration options for executing a process inside a contained environment.
+type Config struct {
+	// NoPivotRoot will use MS_MOVE and a chroot to jail the process into the container's rootfs
+	// This is a common option when the container is running in ramdisk
+	NoPivotRoot bool `json:"no_pivot_root"`
+
+	// ParentDeathSignal specifies the signal that is sent to the container's process in the case
+	// that the parent process dies.
+	ParentDeathSignal int `json:"parent_death_signal"`
+
+	// Path to a directory containing the container's root filesystem.
+	Rootfs string `json:"rootfs"`
+
+	// Readonlyfs will remount the container's rootfs as readonly where only externally mounted
+	// bind mounts are writtable.
+	Readonlyfs bool `json:"readonlyfs"`
+
+	// Specifies the mount propagation flags to be applied to /.
+	RootPropagation int `json:"rootPropagation"`
+
+	// Mounts specify additional source and destination paths that will be mounted inside the container's
+	// rootfs and mount namespace if specified
+	Mounts []*Mount `json:"mounts"`
+
+	// The device nodes that should be automatically created within the container upon container start.  Note, make sure that the node is marked as allowed in the cgroup as well!
+	Devices []*Device `json:"devices"`
+
+	MountLabel string `json:"mount_label"`
+
+	// Hostname optionally sets the container's hostname if provided
+	Hostname string `json:"hostname"`
+
+	// Namespaces specifies the container's namespaces that it should setup when cloning the init process
+	// If a namespace is not provided that namespace is shared from the container's parent process
+	Namespaces Namespaces `json:"namespaces"`
+
+	// Capabilities specify the capabilities to keep when executing the process inside the container
+	// All capabilities not specified will be dropped from the processes capability mask
+	Capabilities *Capabilities `json:"capabilities"`
+
+	// Networks specifies the container's network setup to be created
+	Networks []*Network `json:"networks"`
+
+	// Routes can be specified to create entries in the route table as the container is started
+	Routes []*Route `json:"routes"`
+
+	// Cgroups specifies specific cgroup settings for the various subsystems that the container is
+	// placed into to limit the resources the container has available
+	Cgroups *Cgroup `json:"cgroups"`
+
+	// AppArmorProfile specifies the profile to apply to the process running in the container and is
+	// change at the time the process is execed
+	AppArmorProfile string `json:"apparmor_profile,omitempty"`
+
+	// ProcessLabel specifies the label to apply to the process running in the container.  It is
+	// commonly used by selinux
+	ProcessLabel string `json:"process_label,omitempty"`
+
+	// Rlimits specifies the resource limits, such as max open files, to set in the container
+	// If Rlimits are not set, the container will inherit rlimits from the parent process
+	Rlimits []Rlimit `json:"rlimits,omitempty"`
+
+	// OomScoreAdj specifies the adjustment to be made by the kernel when calculating oom scores
+	// for a process. Valid values are between the range [-1000, '1000'], where processes with
+	// higher scores are preferred for being killed.
+	// More information about kernel oom score calculation here: https://lwn.net/Articles/317814/
+	OomScoreAdj int `json:"oom_score_adj"`
+
+	// UidMappings is an array of User ID mappings for User Namespaces
+	UidMappings []IDMap `json:"uid_mappings"`
+
+	// GidMappings is an array of Group ID mappings for User Namespaces
+	GidMappings []IDMap `json:"gid_mappings"`
+
+	// MaskPaths specifies paths within the container's rootfs to mask over with a bind
+	// mount pointing to /dev/null as to prevent reads of the file.
+	MaskPaths []string `json:"mask_paths"`
+
+	// ReadonlyPaths specifies paths within the container's rootfs to remount as read-only
+	// so that these files prevent any writes.
+	ReadonlyPaths []string `json:"readonly_paths"`
+
+	// Sysctl is a map of properties and their values. It is the equivalent of using
+	// sysctl -w my.property.name value in Linux.
+	Sysctl map[string]string `json:"sysctl"`
+
+	// Seccomp allows actions to be taken whenever a syscall is made within the container.
+	// A number of rules are given, each having an action to be taken if a syscall matches it.
+	// A default action to be taken if no rules match is also given.
+	Seccomp *Seccomp `json:"seccomp"`
+
+	// NoNewPrivileges controls whether processes in the container can gain additional privileges.
+	NoNewPrivileges bool `json:"no_new_privileges,omitempty"`
+
+	// Hooks are a collection of actions to perform at various container lifecycle events.
+	// CommandHooks are serialized to JSON, but other hooks are not.
+	Hooks *Hooks
+
+	// Version is the version of opencontainer specification that is supported.
+	Version string `json:"version"`
+
+	// Labels are user defined metadata that is stored in the config and populated on the state
+	Labels []string `json:"labels"`
+
+	// NoNewKeyring will not allocated a new session keyring for the container.  It will use the
+	// callers keyring in this case.
+	NoNewKeyring bool `json:"no_new_keyring"`
+
+	// Rootless specifies whether the container is a rootless container.
+	Rootless bool `json:"rootless"`
+}
+
+type Hooks struct {
+	// Prestart commands are executed after the container namespaces are created,
+	// but before the user supplied command is executed from init.
+	Prestart []Hook
+
+	// Poststart commands are executed after the container init process starts.
+	Poststart []Hook
+
+	// Poststop commands are executed after the container init process exits.
+	Poststop []Hook
+}
+
+type Capabilities struct {
+	// Bounding is the set of capabilities checked by the kernel.
+	Bounding []string
+	// Effective is the set of capabilities checked by the kernel.
+	Effective []string
+	// Inheritable is the capabilities preserved across execve.
+	Inheritable []string
+	// Permitted is the limiting superset for effective capabilities.
+	Permitted []string
+	// Ambient is the ambient set of capabilities that are kept.
+	Ambient []string
+}
+
+func (hooks *Hooks) UnmarshalJSON(b []byte) error {
+	var state struct {
+		Prestart  []CommandHook
+		Poststart []CommandHook
+		Poststop  []CommandHook
+	}
+
+	if err := json.Unmarshal(b, &state); err != nil {
+		return err
+	}
+
+	deserialize := func(shooks []CommandHook) (hooks []Hook) {
+		for _, shook := range shooks {
+			hooks = append(hooks, shook)
+		}
+
+		return hooks
+	}
+
+	hooks.Prestart = deserialize(state.Prestart)
+	hooks.Poststart = deserialize(state.Poststart)
+	hooks.Poststop = deserialize(state.Poststop)
+	return nil
+}
+
+func (hooks Hooks) MarshalJSON() ([]byte, error) {
+	serialize := func(hooks []Hook) (serializableHooks []CommandHook) {
+		for _, hook := range hooks {
+			switch chook := hook.(type) {
+			case CommandHook:
+				serializableHooks = append(serializableHooks, chook)
+			default:
+				logrus.Warnf("cannot serialize hook of type %T, skipping", hook)
+			}
+		}
+
+		return serializableHooks
+	}
+
+	return json.Marshal(map[string]interface{}{
+		"prestart":  serialize(hooks.Prestart),
+		"poststart": serialize(hooks.Poststart),
+		"poststop":  serialize(hooks.Poststop),
+	})
+}
+
+// HookState is the payload provided to a hook on execution.
+type HookState specs.State
+
+type Hook interface {
+	// Run executes the hook with the provided state.
+	Run(HookState) error
+}
+
+// NewFunctionHook will call the provided function when the hook is run.
+func NewFunctionHook(f func(HookState) error) FuncHook {
+	return FuncHook{
+		run: f,
+	}
+}
+
+type FuncHook struct {
+	run func(HookState) error
+}
+
+func (f FuncHook) Run(s HookState) error {
+	return f.run(s)
+}
+
+type Command struct {
+	Path    string         `json:"path"`
+	Args    []string       `json:"args"`
+	Env     []string       `json:"env"`
+	Dir     string         `json:"dir"`
+	Timeout *time.Duration `json:"timeout"`
+}
+
+// NewCommandHook will execute the provided command when the hook is run.
+func NewCommandHook(cmd Command) CommandHook {
+	return CommandHook{
+		Command: cmd,
+	}
+}
+
+type CommandHook struct {
+	Command
+}
+
+func (c Command) Run(s HookState) error {
+	b, err := json.Marshal(s)
+	if err != nil {
+		return err
+	}
+	var stdout, stderr bytes.Buffer
+	cmd := exec.Cmd{
+		Path:   c.Path,
+		Args:   c.Args,
+		Env:    c.Env,
+		Stdin:  bytes.NewReader(b),
+		Stdout: &stdout,
+		Stderr: &stderr,
+	}
+	if err := cmd.Start(); err != nil {
+		return err
+	}
+	errC := make(chan error, 1)
+	go func() {
+		err := cmd.Wait()
+		if err != nil {
+			err = fmt.Errorf("error running hook: %v, stdout: %s, stderr: %s", err, stdout.String(), stderr.String())
+		}
+		errC <- err
+	}()
+	var timerCh <-chan time.Time
+	if c.Timeout != nil {
+		timer := time.NewTimer(*c.Timeout)
+		defer timer.Stop()
+		timerCh = timer.C
+	}
+	select {
+	case err := <-errC:
+		return err
+	case <-timerCh:
+		cmd.Process.Kill()
+		cmd.Wait()
+		return fmt.Errorf("hook ran past specified timeout of %.1fs", c.Timeout.Seconds())
+	}
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/config_unix.go

@@ -0,0 +1,63 @@
+// +build freebsd linux
+
+package configs
+
+import "fmt"
+
+// HostUID gets the translated uid for the process on host which could be
+// different when user namespaces are enabled.
+func (c Config) HostUID(containerId int) (int, error) {
+	if c.Namespaces.Contains(NEWUSER) {
+		if c.UidMappings == nil {
+			return -1, fmt.Errorf("User namespaces enabled, but no uid mappings found.")
+		}
+		id, found := c.hostIDFromMapping(containerId, c.UidMappings)
+		if !found {
+			return -1, fmt.Errorf("User namespaces enabled, but no user mapping found.")
+		}
+		return id, nil
+	}
+	// Return unchanged id.
+	return containerId, nil
+}
+
+// HostRootUID gets the root uid for the process on host which could be non-zero
+// when user namespaces are enabled.
+func (c Config) HostRootUID() (int, error) {
+	return c.HostUID(0)
+}
+
+// HostGID gets the translated gid for the process on host which could be
+// different when user namespaces are enabled.
+func (c Config) HostGID(containerId int) (int, error) {
+	if c.Namespaces.Contains(NEWUSER) {
+		if c.GidMappings == nil {
+			return -1, fmt.Errorf("User namespaces enabled, but no gid mappings found.")
+		}
+		id, found := c.hostIDFromMapping(containerId, c.GidMappings)
+		if !found {
+			return -1, fmt.Errorf("User namespaces enabled, but no group mapping found.")
+		}
+		return id, nil
+	}
+	// Return unchanged id.
+	return containerId, nil
+}
+
+// HostRootGID gets the root gid for the process on host which could be non-zero
+// when user namespaces are enabled.
+func (c Config) HostRootGID() (int, error) {
+	return c.HostGID(0)
+}
+
+// Utility function that gets a host ID for a container ID from user namespace map
+// if that ID is present in the map.
+func (c Config) hostIDFromMapping(containerID int, uMap []IDMap) (int, bool) {
+	for _, m := range uMap {
+		if (containerID >= m.ContainerID) && (containerID <= (m.ContainerID + m.Size - 1)) {
+			hostID := m.HostID + (containerID - m.ContainerID)
+			return hostID, true
+		}
+	}
+	return -1, false
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/device.go

@@ -0,0 +1,57 @@
+package configs
+
+import (
+	"fmt"
+	"os"
+)
+
+const (
+	Wildcard = -1
+)
+
+// TODO Windows: This can be factored out in the future
+
+type Device struct {
+	// Device type, block, char, etc.
+	Type rune `json:"type"`
+
+	// Path to the device.
+	Path string `json:"path"`
+
+	// Major is the device's major number.
+	Major int64 `json:"major"`
+
+	// Minor is the device's minor number.
+	Minor int64 `json:"minor"`
+
+	// Cgroup permissions format, rwm.
+	Permissions string `json:"permissions"`
+
+	// FileMode permission bits for the device.
+	FileMode os.FileMode `json:"file_mode"`
+
+	// Uid of the device.
+	Uid uint32 `json:"uid"`
+
+	// Gid of the device.
+	Gid uint32 `json:"gid"`
+
+	// Write the file to the allowed list
+	Allow bool `json:"allow"`
+}
+
+func (d *Device) CgroupString() string {
+	return fmt.Sprintf("%c %s:%s %s", d.Type, deviceNumberString(d.Major), deviceNumberString(d.Minor), d.Permissions)
+}
+
+func (d *Device) Mkdev() int {
+	return int((d.Major << 8) | (d.Minor & 0xff) | ((d.Minor & 0xfff00) << 12))
+}
+
+// deviceNumberString converts the device number to a string return result.
+func deviceNumberString(number int64) string {
+	if number == Wildcard {
+		return "*"
+	}
+	return fmt.Sprint(number)
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/device_defaults.go

@@ -0,0 +1,111 @@
+// +build linux freebsd
+
+package configs
+
+var (
+	// DefaultSimpleDevices are devices that are to be both allowed and created.
+	DefaultSimpleDevices = []*Device{
+		// /dev/null and zero
+		{
+			Path:        "/dev/null",
+			Type:        'c',
+			Major:       1,
+			Minor:       3,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+		{
+			Path:        "/dev/zero",
+			Type:        'c',
+			Major:       1,
+			Minor:       5,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		{
+			Path:        "/dev/full",
+			Type:        'c',
+			Major:       1,
+			Minor:       7,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		// consoles and ttys
+		{
+			Path:        "/dev/tty",
+			Type:        'c',
+			Major:       5,
+			Minor:       0,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		// /dev/urandom,/dev/random
+		{
+			Path:        "/dev/urandom",
+			Type:        'c',
+			Major:       1,
+			Minor:       9,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+		{
+			Path:        "/dev/random",
+			Type:        'c',
+			Major:       1,
+			Minor:       8,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+	}
+	DefaultAllowedDevices = append([]*Device{
+		// allow mknod for any device
+		{
+			Type:        'c',
+			Major:       Wildcard,
+			Minor:       Wildcard,
+			Permissions: "m",
+		},
+		{
+			Type:        'b',
+			Major:       Wildcard,
+			Minor:       Wildcard,
+			Permissions: "m",
+		},
+
+		{
+			Path:        "/dev/console",
+			Type:        'c',
+			Major:       5,
+			Minor:       1,
+			Permissions: "rwm",
+		},
+		// /dev/pts/ - pts namespaces are "coming soon"
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       136,
+			Minor:       Wildcard,
+			Permissions: "rwm",
+		},
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       5,
+			Minor:       2,
+			Permissions: "rwm",
+		},
+
+		// tuntap
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       10,
+			Minor:       200,
+			Permissions: "rwm",
+		},
+	}, DefaultSimpleDevices...)
+	DefaultAutoCreatedDevices = append([]*Device{}, DefaultSimpleDevices...)
+)

vendor/github.com/opencontainers/runc/libcontainer/configs/hugepage_limit.go

@@ -0,0 +1,9 @@
+package configs
+
+type HugepageLimit struct {
+	// which type of hugepage to limit.
+	Pagesize string `json:"page_size"`
+
+	// usage limit for hugepage.
+	Limit uint64 `json:"limit"`
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/interface_priority_map.go

@@ -0,0 +1,14 @@
+package configs
+
+import (
+	"fmt"
+)
+
+type IfPrioMap struct {
+	Interface string `json:"interface"`
+	Priority  int64  `json:"priority"`
+}
+
+func (i *IfPrioMap) CgroupString() string {
+	return fmt.Sprintf("%s %d", i.Interface, i.Priority)
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/mount.go

@@ -0,0 +1,39 @@
+package configs
+
+const (
+	// EXT_COPYUP is a directive to copy up the contents of a directory when
+	// a tmpfs is mounted over it.
+	EXT_COPYUP = 1 << iota
+)
+
+type Mount struct {
+	// Source path for the mount.
+	Source string `json:"source"`
+
+	// Destination path for the mount inside the container.
+	Destination string `json:"destination"`
+
+	// Device the mount is for.
+	Device string `json:"device"`
+
+	// Mount flags.
+	Flags int `json:"flags"`
+
+	// Propagation Flags
+	PropagationFlags []int `json:"propagation_flags"`
+
+	// Mount data applied to the mount.
+	Data string `json:"data"`
+
+	// Relabel source if set, "z" indicates shared, "Z" indicates unshared.
+	Relabel string `json:"relabel"`
+
+	// Extensions are additional flags that are specific to runc.
+	Extensions int `json:"extensions"`
+
+	// Optional Command to be run before Source is mounted.
+	PremountCmds []Command `json:"premount_cmds"`
+
+	// Optional Command to be run after Source is mounted.
+	PostmountCmds []Command `json:"postmount_cmds"`
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/namespaces.go

@@ -0,0 +1,5 @@
+package configs
+
+type NamespaceType string
+
+type Namespaces []Namespace

vendor/github.com/opencontainers/runc/libcontainer/configs/namespaces_syscall.go

@@ -0,0 +1,31 @@
+// +build linux
+
+package configs
+
+import "syscall"
+
+func (n *Namespace) Syscall() int {
+	return namespaceInfo[n.Type]
+}
+
+var namespaceInfo = map[NamespaceType]int{
+	NEWNET:  syscall.CLONE_NEWNET,
+	NEWNS:   syscall.CLONE_NEWNS,
+	NEWUSER: syscall.CLONE_NEWUSER,
+	NEWIPC:  syscall.CLONE_NEWIPC,
+	NEWUTS:  syscall.CLONE_NEWUTS,
+	NEWPID:  syscall.CLONE_NEWPID,
+}
+
+// CloneFlags parses the container's Namespaces options to set the correct
+// flags on clone, unshare. This function returns flags only for new namespaces.
+func (n *Namespaces) CloneFlags() uintptr {
+	var flag int
+	for _, v := range *n {
+		if v.Path != "" {
+			continue
+		}
+		flag |= namespaceInfo[v.Type]
+	}
+	return uintptr(flag)
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/namespaces_syscall_unsupported.go

@@ -0,0 +1,13 @@
+// +build !linux,!windows
+
+package configs
+
+func (n *Namespace) Syscall() int {
+	panic("No namespace syscall support")
+}
+
+// CloneFlags parses the container's Namespaces options to set the correct
+// flags on clone, unshare. This function returns flags only for new namespaces.
+func (n *Namespaces) CloneFlags() uintptr {
+	panic("No namespace syscall support")
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/namespaces_unix.go

@@ -0,0 +1,127 @@
+// +build linux freebsd
+
+package configs
+
+import (
+	"fmt"
+	"os"
+	"sync"
+)
+
+const (
+	NEWNET  NamespaceType = "NEWNET"
+	NEWPID  NamespaceType = "NEWPID"
+	NEWNS   NamespaceType = "NEWNS"
+	NEWUTS  NamespaceType = "NEWUTS"
+	NEWIPC  NamespaceType = "NEWIPC"
+	NEWUSER NamespaceType = "NEWUSER"
+)
+
+var (
+	nsLock              sync.Mutex
+	supportedNamespaces = make(map[NamespaceType]bool)
+)
+
+// NsName converts the namespace type to its filename
+func NsName(ns NamespaceType) string {
+	switch ns {
+	case NEWNET:
+		return "net"
+	case NEWNS:
+		return "mnt"
+	case NEWPID:
+		return "pid"
+	case NEWIPC:
+		return "ipc"
+	case NEWUSER:
+		return "user"
+	case NEWUTS:
+		return "uts"
+	}
+	return ""
+}
+
+// IsNamespaceSupported returns whether a namespace is available or
+// not
+func IsNamespaceSupported(ns NamespaceType) bool {
+	nsLock.Lock()
+	defer nsLock.Unlock()
+	supported, ok := supportedNamespaces[ns]
+	if ok {
+		return supported
+	}
+	nsFile := NsName(ns)
+	// if the namespace type is unknown, just return false
+	if nsFile == "" {
+		return false
+	}
+	_, err := os.Stat(fmt.Sprintf("/proc/self/ns/%s", nsFile))
+	// a namespace is supported if it exists and we have permissions to read it
+	supported = err == nil
+	supportedNamespaces[ns] = supported
+	return supported
+}
+
+func NamespaceTypes() []NamespaceType {
+	return []NamespaceType{
+		NEWNET,
+		NEWPID,
+		NEWNS,
+		NEWUTS,
+		NEWIPC,
+		NEWUSER,
+	}
+}
+
+// Namespace defines configuration for each namespace.  It specifies an
+// alternate path that is able to be joined via setns.
+type Namespace struct {
+	Type NamespaceType `json:"type"`
+	Path string        `json:"path"`
+}
+
+func (n *Namespace) GetPath(pid int) string {
+	if n.Path != "" {
+		return n.Path
+	}
+	return fmt.Sprintf("/proc/%d/ns/%s", pid, NsName(n.Type))
+}
+
+func (n *Namespaces) Remove(t NamespaceType) bool {
+	i := n.index(t)
+	if i == -1 {
+		return false
+	}
+	*n = append((*n)[:i], (*n)[i+1:]...)
+	return true
+}
+
+func (n *Namespaces) Add(t NamespaceType, path string) {
+	i := n.index(t)
+	if i == -1 {
+		*n = append(*n, Namespace{Type: t, Path: path})
+		return
+	}
+	(*n)[i].Path = path
+}
+
+func (n *Namespaces) index(t NamespaceType) int {
+	for i, ns := range *n {
+		if ns.Type == t {
+			return i
+		}
+	}
+	return -1
+}
+
+func (n *Namespaces) Contains(t NamespaceType) bool {
+	return n.index(t) != -1
+}
+
+func (n *Namespaces) PathOf(t NamespaceType) string {
+	i := n.index(t)
+	if i == -1 {
+		return ""
+	}
+	return (*n)[i].Path
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/namespaces_unsupported.go

@@ -0,0 +1,8 @@
+// +build !linux,!freebsd
+
+package configs
+
+// Namespace defines configuration for each namespace.  It specifies an
+// alternate path that is able to be joined via setns.
+type Namespace struct {
+}

vendor/github.com/opencontainers/runc/libcontainer/configs/network.go

@@ -0,0 +1,72 @@
+package configs
+
+// Network defines configuration for a container's networking stack
+//
+// The network configuration can be omitted from a container causing the
+// container to be setup with the host's networking stack
+type Network struct {
+	// Type sets the networks type, commonly veth and loopback
+	Type string `json:"type"`
+
+	// Name of the network interface
+	Name string `json:"name"`
+
+	// The bridge to use.
+	Bridge string `json:"bridge"`
+
+	// MacAddress contains the MAC address to set on the network interface
+	MacAddress string `json:"mac_address"`
+
+	// Address contains the IPv4 and mask to set on the network interface
+	Address string `json:"address"`
+
+	// Gateway sets the gateway address that is used as the default for the interface
+	Gateway string `json:"gateway"`
+
+	// IPv6Address contains the IPv6 and mask to set on the network interface
+	IPv6Address string `json:"ipv6_address"`
+
+	// IPv6Gateway sets the ipv6 gateway address that is used as the default for the interface
+	IPv6Gateway string `json:"ipv6_gateway"`
+
+	// Mtu sets the mtu value for the interface and will be mirrored on both the host and
+	// container's interfaces if a pair is created, specifically in the case of type veth
+	// Note: This does not apply to loopback interfaces.
+	Mtu int `json:"mtu"`
+
+	// TxQueueLen sets the tx_queuelen value for the interface and will be mirrored on both the host and
+	// container's interfaces if a pair is created, specifically in the case of type veth
+	// Note: This does not apply to loopback interfaces.
+	TxQueueLen int `json:"txqueuelen"`
+
+	// HostInterfaceName is a unique name of a veth pair that resides on in the host interface of the
+	// container.
+	HostInterfaceName string `json:"host_interface_name"`
+
+	// HairpinMode specifies if hairpin NAT should be enabled on the virtual interface
+	// bridge port in the case of type veth
+	// Note: This is unsupported on some systems.
+	// Note: This does not apply to loopback interfaces.
+	HairpinMode bool `json:"hairpin_mode"`
+}
+
+// Routes can be specified to create entries in the route table as the container is started
+//
+// All of destination, source, and gateway should be either IPv4 or IPv6.
+// One of the three options must be present, and omitted entries will use their
+// IP family default for the route table.  For IPv4 for example, setting the
+// gateway to 1.2.3.4 and the interface to eth0 will set up a standard
+// destination of 0.0.0.0(or *) when viewed in the route table.
+type Route struct {
+	// Sets the destination and mask, should be a CIDR.  Accepts IPv4 and IPv6
+	Destination string `json:"destination"`
+
+	// Sets the source and mask, should be a CIDR.  Accepts IPv4 and IPv6
+	Source string `json:"source"`
+
+	// Sets the gateway.  Accepts IPv4 and IPv6
+	Gateway string `json:"gateway"`
+
+	// The device to set this route up for, for example: eth0
+	InterfaceName string `json:"interface_name"`
+}

vendor/github.com/opencontainers/runc/libcontainer/devices/devices_unix.go

@@ -0,0 +1,106 @@
+// +build linux freebsd
+
+package devices
+
+import (
+	"errors"
+	"fmt"
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"syscall"
+
+	"github.com/opencontainers/runc/libcontainer/configs"
+)
+
+var (
+	ErrNotADevice = errors.New("not a device node")
+)
+
+// Testing dependencies
+var (
+	osLstat       = os.Lstat
+	ioutilReadDir = ioutil.ReadDir
+)
+
+// Given the path to a device and its cgroup_permissions(which cannot be easily queried) look up the information about a linux device and return that information as a Device struct.
+func DeviceFromPath(path, permissions string) (*configs.Device, error) {
+	fileInfo, err := osLstat(path)
+	if err != nil {
+		return nil, err
+	}
+	var (
+		devType                rune
+		mode                   = fileInfo.Mode()
+		fileModePermissionBits = os.FileMode.Perm(mode)
+	)
+	switch {
+	case mode&os.ModeDevice == 0:
+		return nil, ErrNotADevice
+	case mode&os.ModeCharDevice != 0:
+		fileModePermissionBits |= syscall.S_IFCHR
+		devType = 'c'
+	default:
+		fileModePermissionBits |= syscall.S_IFBLK
+		devType = 'b'
+	}
+	stat_t, ok := fileInfo.Sys().(*syscall.Stat_t)
+	if !ok {
+		return nil, fmt.Errorf("cannot determine the device number for device %s", path)
+	}
+	devNumber := int(stat_t.Rdev)
+	return &configs.Device{
+		Type:        devType,
+		Path:        path,
+		Major:       Major(devNumber),
+		Minor:       Minor(devNumber),
+		Permissions: permissions,
+		FileMode:    fileModePermissionBits,
+		Uid:         stat_t.Uid,
+		Gid:         stat_t.Gid,
+	}, nil
+}
+
+func HostDevices() ([]*configs.Device, error) {
+	return getDevices("/dev")
+}
+
+func getDevices(path string) ([]*configs.Device, error) {
+	files, err := ioutilReadDir(path)
+	if err != nil {
+		return nil, err
+	}
+	out := []*configs.Device{}
+	for _, f := range files {
+		switch {
+		case f.IsDir():
+			switch f.Name() {
+			// ".lxc" & ".lxd-mounts" added to address https://github.com/lxc/lxd/issues/2825
+			case "pts", "shm", "fd", "mqueue", ".lxc", ".lxd-mounts":
+				continue
+			default:
+				sub, err := getDevices(filepath.Join(path, f.Name()))
+				if err != nil {
+					return nil, err
+				}
+
+				out = append(out, sub...)
+				continue
+			}
+		case f.Name() == "console":
+			continue
+		}
+		device, err := DeviceFromPath(filepath.Join(path, f.Name()), "rwm")
+		if err != nil {
+			if err == ErrNotADevice {
+				continue
+			}
+			if os.IsNotExist(err) {
+				continue
+			}
+			return nil, err
+		}
+		out = append(out, device)
+	}
+	return out, nil
+}

vendor/github.com/opencontainers/runc/libcontainer/devices/devices_unsupported.go

@@ -0,0 +1,3 @@
+// +build windows
+
+package devices

vendor/github.com/opencontainers/runc/libcontainer/devices/number.go

@@ -0,0 +1,24 @@
+// +build linux freebsd
+
+package devices
+
+/*
+
+This code provides support for manipulating linux device numbers.  It should be replaced by normal syscall functions once http://code.google.com/p/go/issues/detail?id=8106 is solved.
+
+You can read what they are here:
+
+ - http://www.makelinux.net/ldd3/chp-3-sect-2
+ - http://www.linux-tutorial.info/modules.php?name=MContent&pageid=94
+
+Note! These are NOT the same as the MAJOR(dev_t device);, MINOR(dev_t device); and MKDEV(int major, int minor); functions as defined in <linux/kdev_t.h> as the representation of device numbers used by go is different than the one used internally to the kernel! - https://github.com/torvalds/linux/blob/master/include/linux/kdev_t.h#L9
+
+*/
+
+func Major(devNumber int) int64 {
+	return int64((devNumber >> 8) & 0xfff)
+}
+
+func Minor(devNumber int) int64 {
+	return int64((devNumber & 0xff) | ((devNumber >> 12) & 0xfff00))
+}