opensource/podman
1
0
Fork 0
mirror of https://github.com/containers/podman.git synced 2025-09-27 00:34:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

Vendor in new version of Buildah

Browse Source 
This also pulls in latest runc and containers/common

Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
This commit is contained in:
Daniel J Walsh
2020-07-08 16:05:12 -04:00
parent 5c6002bf9d
commit b020d1ad13
42 changed files with 985 additions and 752 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
vendor/github.com/opencontainers/runc/libcontainer/cgroups/cgroups.go generated vendored
View File

@ -3,8 +3,6 @@
package cgroups
import (
"fmt"
"github.com/opencontainers/runc/libcontainer/configs"
)
@ -27,48 +25,27 @@ type Manager interface {
// Destroys the cgroup set
Destroy() error
// The option func SystemdCgroups() and Cgroupfs() require following attributes:
// Paths map[string]string
// Cgroups *configs.Cgroup
// Paths maps cgroup subsystem to path at which it is mounted.
// Cgroups specifies specific cgroup settings for the various subsystems
// Returns cgroup paths to save in a state file and to be able to
// restore the object later.
GetPaths() map[string]string
// GetUnifiedPath returns the unified path when running in unified mode.
// The value corresponds to the all values of GetPaths() map.
//
// GetUnifiedPath returns error when running in hybrid mode as well as
// in legacy mode.
GetUnifiedPath() (string, error)
// Path returns a cgroup path to the specified controller/subsystem.
// For cgroupv2, the argument is unused and can be empty.
Path(string) string
// Sets the cgroup as configured.
Set(container *configs.Config) error
// Gets the cgroup as configured.
// GetPaths returns cgroup path(s) to save in a state file in order to restore later.
//
// For cgroup v1, a key is cgroup subsystem name, and the value is the path
// to the cgroup for this subsystem.
//
// For cgroup v2 unified hierarchy, a key is "", and the value is the unified path.
GetPaths() map[string]string
// GetCgroups returns the cgroup data as configured.
GetCgroups() (*configs.Cgroup, error)
}
type NotFoundError struct {
Subsystem string
}
// GetFreezerState retrieves the current FreezerState of the cgroup.
GetFreezerState() (configs.FreezerState, error)
func (e *NotFoundError) Error() string {
return fmt.Sprintf("mountpoint for %s not found", e.Subsystem)
}
func NewNotFoundError(sub string) error {
return &NotFoundError{
Subsystem: sub,
}
}
func IsNotFound(err error) bool {
if err == nil {
return false
}
_, ok := err.(*NotFoundError)
return ok
// Whether the cgroup path exists or not
Exists() bool
}

27
vendor/github.com/opencontainers/runc/libcontainer/cgroups/stats.go generated vendored
View File

@ -20,6 +20,12 @@ type CpuUsage struct {
// Total CPU time consumed per core.
// Units: nanoseconds.
PercpuUsage []uint64 `json:"percpu_usage,omitempty"`
// CPU time consumed per core in kernel mode
// Units: nanoseconds.
PercpuUsageInKernelmode []uint64 `json:"percpu_usage_in_kernelmode"`
// CPU time consumed per core in user mode
// Units: nanoseconds.
PercpuUsageInUsermode []uint64 `json:"percpu_usage_in_usermode"`
// Time spent by tasks of the cgroup in kernel mode.
// Units: nanoseconds.
UsageInKernelmode uint64 `json:"usage_in_kernelmode"`
@ -51,12 +57,33 @@ type MemoryStats struct {
KernelUsage MemoryData `json:"kernel_usage,omitempty"`
// usage of kernel TCP memory
KernelTCPUsage MemoryData `json:"kernel_tcp_usage,omitempty"`
// usage of memory pages by NUMA node
// see chapter 5.6 of memory controller documentation
PageUsageByNUMA PageUsageByNUMA `json:"page_usage_by_numa,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"`
}
type PageUsageByNUMA struct {
// Embedding is used as types can't be recursive.
PageUsageByNUMAInner
Hierarchical PageUsageByNUMAInner `json:"hierarchical,omitempty"`
}
type PageUsageByNUMAInner struct {
Total PageStats `json:"total,omitempty"`
File PageStats `json:"file,omitempty"`
Anon PageStats `json:"anon,omitempty"`
Unevictable PageStats `json:"unevictable,omitempty"`
}
type PageStats struct {
Total uint64 `json:"total,omitempty"`
Nodes map[uint8]uint64 `json:"nodes,omitempty"`
}
type PidsStats struct {
// number of pids in the cgroup
Current uint64 `json:"current,omitempty"`

357
vendor/github.com/opencontainers/runc/libcontainer/cgroups/utils.go generated vendored
View File

@ -4,6 +4,7 @@ package cgroups
import (
"bufio"
"errors"
"fmt"
"io"
"io/ioutil"
@ -12,7 +13,6 @@ import (
"strconv"
"strings"
"sync"
"syscall"
"time"
units "github.com/docker/go-units"
@ -20,7 +20,6 @@ import (
)
const (
CgroupNamePrefix = "name="
CgroupProcesses = "cgroup.procs"
unifiedMountpoint = "/sys/fs/cgroup"
)
@ -40,8 +39,8 @@ var HugePageSizeUnitList = []string{"B", "KB", "MB", "GB", "TB", "PB"}
// IsCgroup2UnifiedMode returns whether we are running in cgroup v2 unified mode.
func IsCgroup2UnifiedMode() bool {
isUnifiedOnce.Do(func() {
var st syscall.Statfs_t
if err := syscall.Statfs(unifiedMountpoint, &st); err != nil {
var st unix.Statfs_t
if err := unix.Statfs(unifiedMountpoint, &st); err != nil {
panic("cannot statfs cgroup root")
}
isUnified = st.Type == unix.CGROUP2_SUPER_MAGIC
@ -49,191 +48,19 @@ func IsCgroup2UnifiedMode() bool {
return isUnified
}
// https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt
func FindCgroupMountpoint(cgroupPath, subsystem string) (string, error) {
if IsCgroup2UnifiedMode() {
return unifiedMountpoint, nil
}
mnt, _, err := FindCgroupMountpointAndRoot(cgroupPath, subsystem)
return mnt, err
}
func FindCgroupMountpointAndRoot(cgroupPath, 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
}
defer f.Close()
if IsCgroup2UnifiedMode() {
subsystem = ""
}
return findCgroupMountpointAndRootFromReader(f, cgroupPath, subsystem)
}
func findCgroupMountpointAndRootFromReader(reader io.Reader, cgroupPath, subsystem string) (string, string, error) {
scanner := bufio.NewScanner(reader)
for scanner.Scan() {
txt := scanner.Text()
fields := strings.Fields(txt)
if len(fields) < 9 {
continue
}
if strings.HasPrefix(fields[4], cgroupPath) {
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
if (subsystem == "" && fields[9] == "cgroup2") || opt == subsystem {
return fields[4], fields[3], nil
}
}
}
}
if err := scanner.Err(); err != nil {
return "", "", err
}
return "", "", NewNotFoundError(subsystem)
}
func isSubsystemAvailable(subsystem string) bool {
if IsCgroup2UnifiedMode() {
controllers, err := GetAllSubsystems()
if err != nil {
return false
}
for _, c := range controllers {
if c == subsystem {
return true
}
}
return false
}
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 {
return "", err
}
defer f.Close()
scanner := bufio.NewScanner(f)
for scanner.Scan() {
text := scanner.Text()
fields := strings.Split(text, " ")
// Safe as mountinfo encodes mountpoints with spaces as \040.
index := strings.Index(text, " - ")
postSeparatorFields := strings.Fields(text[index+3:])
numPostFields := len(postSeparatorFields)
// This is an error as we can't detect if the mount is for "cgroup"
if numPostFields == 0 {
return "", fmt.Errorf("Found no fields post '-' in %q", text)
}
if postSeparatorFields[0] == "cgroup" || postSeparatorFields[0] == "cgroup2" {
// Check that the mount is properly formatted.
if numPostFields < 3 {
return "", fmt.Errorf("Error found less than 3 fields post '-' in %q", text)
}
return filepath.Dir(fields[4]), nil
}
}
if err := scanner.Err(); err != nil {
return "", err
}
return "", NewNotFoundError("cgroup")
}
type Mount struct {
Mountpoint string
Root string
Subsystems []string
}
func (m Mount) GetOwnCgroup(cgroups map[string]string) (string, error) {
if len(m.Subsystems) == 0 {
return "", fmt.Errorf("no subsystem for mount")
}
return getControllerPath(m.Subsystems[0], cgroups)
}
func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount, error) {
res := make([]Mount, 0, len(ss))
scanner := bufio.NewScanner(mi)
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+10] == "cgroup2" || txt[sepIdx+3:sepIdx+9] != "cgroup" {
continue
}
fields := strings.Split(txt, " ")
m := Mount{
Mountpoint: fields[4],
Root: fields[3],
}
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
seen, known := ss[opt]
if !known || (!all && seen) {
continue
}
ss[opt] = true
if strings.HasPrefix(opt, CgroupNamePrefix) {
opt = opt[len(CgroupNamePrefix):]
}
m.Subsystems = append(m.Subsystems, opt)
numFound++
}
if len(m.Subsystems) > 0 || all {
res = append(res, m)
}
}
if err := scanner.Err(); err != nil {
return nil, err
}
return res, nil
}
// GetCgroupMounts returns the mounts for the cgroup subsystems.
// all indicates whether to return just the first instance or all the mounts.
// This function should not be used from cgroupv2 code, as in this case
// all the controllers are available under the constant unifiedMountpoint.
func GetCgroupMounts(all bool) ([]Mount, error) {
if IsCgroup2UnifiedMode() {
// TODO: remove cgroupv2 case once all external users are converted
availableControllers, err := GetAllSubsystems()
if err != nil {
return nil, err
@ -246,22 +73,7 @@ func GetCgroupMounts(all bool) ([]Mount, error) {
return []Mount{m}, nil
}
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return nil, err
}
defer f.Close()
allSubsystems, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return nil, err
}
allMap := make(map[string]bool)
for s := range allSubsystems {
allMap[s] = false
}
return getCgroupMountsHelper(allMap, f, all)
return getCgroupMountsV1(all)
}
// GetAllSubsystems returns all the cgroup subsystems supported by the kernel
@ -305,61 +117,8 @@ func GetAllSubsystems() ([]string, error) {
return subsystems, nil
}
// 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
}
return getControllerPath(subsystem, cgroups)
}
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
}
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 paths 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))
func readProcsFile(file string) ([]int, error) {
f, err := os.Open(file)
if err != nil {
return nil, err
}
@ -379,11 +138,18 @@ func readProcsFile(dir string) ([]int, error) {
out = append(out, pid)
}
}
return out, nil
return out, s.Err()
}
// ParseCgroupFile parses the given cgroup file, typically from
// /proc/<pid>/cgroup, into a map of subgroups to cgroup names.
// ParseCgroupFile parses the given cgroup file, typically /proc/self/cgroup
// or /proc/<pid>/cgroup, into a map of subsystems to cgroup paths, e.g.
// "cpu": "/user.slice/user-1000.slice"
// "pids": "/user.slice/user-1000.slice"
// etc.
//
// Note that for cgroup v2 unified hierarchy, there are no per-controller
// cgroup paths, so the resulting map will have a single element where the key
// is empty string ("") and the value is the cgroup path the <pid> is in.
func ParseCgroupFile(path string) (map[string]string, error) {
f, err := os.Open(path)
if err != nil {
@ -423,22 +189,6 @@ func parseCgroupFromReader(r io.Reader) (map[string]string, error) {
return cgroups, nil
}
func getControllerPath(subsystem string, cgroups map[string]string) (string, error) {
if IsCgroup2UnifiedMode() {
return "/", nil
}
if p, ok := cgroups[subsystem]; ok {
return p, nil
}
if p, ok := cgroups[CgroupNamePrefix+subsystem]; ok {
return p, nil
}
return "", NewNotFoundError(subsystem)
}
func PathExists(path string) bool {
if _, err := os.Stat(path); err != nil {
return false
@ -514,8 +264,8 @@ func getHugePageSizeFromFilenames(fileNames []string) ([]string, error) {
}
// GetPids returns all pids, that were added to cgroup at path.
func GetPids(path string) ([]int, error) {
return readProcsFile(path)
func GetPids(dir string) ([]int, error) {
return readProcsFile(filepath.Join(dir, CgroupProcesses))
}
// GetAllPids returns all pids, that were added to cgroup at path and to all its
@ -524,14 +274,13 @@ func GetAllPids(path string) ([]int, error) {
var pids []int
// 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 != CgroupProcesses {
return nil
}
if iErr != nil {
return iErr
}
cPids, err := readProcsFile(dir)
if info.IsDir() || info.Name() != CgroupProcesses {
return nil
}
cPids, err := readProcsFile(p)
if err != nil {
return err
}
@ -568,7 +317,7 @@ func WriteCgroupProc(dir string, pid int) error {
// EINVAL might mean that the task being added to cgroup.procs is in state
// TASK_NEW. We should attempt to do so again.
if isEINVAL(err) {
if errors.Is(err, unix.EINVAL) {
time.Sleep(30 * time.Millisecond)
continue
}
@ -578,11 +327,53 @@ func WriteCgroupProc(dir string, pid int) error {
return err
}
func isEINVAL(err error) bool {
switch err := err.(type) {
case *os.PathError:
return err.Err == unix.EINVAL
default:
return false
// Since the OCI spec is designed for cgroup v1, in some cases
// there is need to convert from the cgroup v1 configuration to cgroup v2
// the formula for BlkIOWeight is y = (1 + (x - 10) * 9999 / 990)
// convert linearly from [10-1000] to [1-10000]
func ConvertBlkIOToCgroupV2Value(blkIoWeight uint16) uint64 {
if blkIoWeight == 0 {
return 0
}
return uint64(1 + (uint64(blkIoWeight)-10)*9999/990)
}
// Since the OCI spec is designed for cgroup v1, in some cases
// there is need to convert from the cgroup v1 configuration to cgroup v2
// the formula for cpuShares is y = (1 + ((x - 2) * 9999) / 262142)
// convert from [2-262144] to [1-10000]
// 262144 comes from Linux kernel definition "#define MAX_SHARES (1UL << 18)"
func ConvertCPUSharesToCgroupV2Value(cpuShares uint64) uint64 {
if cpuShares == 0 {
return 0
}
return (1 + ((cpuShares-2)*9999)/262142)
}
// ConvertMemorySwapToCgroupV2Value converts MemorySwap value from OCI spec
// for use by cgroup v2 drivers. A conversion is needed since Resources.MemorySwap
// is defined as memory+swap combined, while in cgroup v2 swap is a separate value.
func ConvertMemorySwapToCgroupV2Value(memorySwap, memory int64) (int64, error) {
// for compatibility with cgroup1 controller, set swap to unlimited in
// case the memory is set to unlimited, and swap is not explicitly set,
// treating the request as "set both memory and swap to unlimited".
if memory == -1 && memorySwap == 0 {
return -1, nil
}
if memorySwap == -1 || memorySwap == 0 {
// -1 is "max", 0 is "unset", so treat as is
return memorySwap, nil
}
// sanity checks
if memory == 0 || memory == -1 {
return 0, errors.New("unable to set swap limit without memory limit")
}
if memory < 0 {
return 0, fmt.Errorf("invalid memory value: %d", memory)
}
if memorySwap < memory {
return 0, errors.New("memory+swap limit should be >= memory limit")
}
return memorySwap - memory, nil
}

250
vendor/github.com/opencontainers/runc/libcontainer/cgroups/v1_utils.go generated vendored Normal file
View File

@ -0,0 +1,250 @@
package cgroups
import (
"bufio"
"errors"
"fmt"
"io"
"os"
"path/filepath"
"strings"
)
// Code in this source file are specific to cgroup v1,
// and must not be used from any cgroup v2 code.
const (
CgroupNamePrefix = "name="
)
var (
errUnified = errors.New("not implemented for cgroup v2 unified hierarchy")
)
type NotFoundError struct {
Subsystem string
}
func (e *NotFoundError) Error() string {
return fmt.Sprintf("mountpoint for %s not found", e.Subsystem)
}
func NewNotFoundError(sub string) error {
return &NotFoundError{
Subsystem: sub,
}
}
func IsNotFound(err error) bool {
if err == nil {
return false
}
_, ok := err.(*NotFoundError)
return ok
}
// https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt
func FindCgroupMountpoint(cgroupPath, subsystem string) (string, error) {
if IsCgroup2UnifiedMode() {
return "", errUnified
}
mnt, _, err := FindCgroupMountpointAndRoot(cgroupPath, subsystem)
return mnt, err
}
func FindCgroupMountpointAndRoot(cgroupPath, subsystem string) (string, string, error) {
if IsCgroup2UnifiedMode() {
return "", "", errUnified
}
// 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()
return findCgroupMountpointAndRootFromReader(f, cgroupPath, subsystem)
}
func findCgroupMountpointAndRootFromReader(reader io.Reader, cgroupPath, subsystem string) (string, string, error) {
scanner := bufio.NewScanner(reader)
for scanner.Scan() {
txt := scanner.Text()
fields := strings.Fields(txt)
if len(fields) < 9 {
continue
}
if strings.HasPrefix(fields[4], cgroupPath) {
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
if opt == subsystem {
return fields[4], fields[3], nil
}
}
}
}
if err := scanner.Err(); err != nil {
return "", "", err
}
return "", "", NewNotFoundError(subsystem)
}
func isSubsystemAvailable(subsystem string) bool {
if IsCgroup2UnifiedMode() {
panic("don't call isSubsystemAvailable from cgroupv2 code")
}
cgroups, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return false
}
_, avail := cgroups[subsystem]
return avail
}
func (m Mount) GetOwnCgroup(cgroups map[string]string) (string, error) {
if len(m.Subsystems) == 0 {
return "", fmt.Errorf("no subsystem for mount")
}
return getControllerPath(m.Subsystems[0], cgroups)
}
func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount, error) {
res := make([]Mount, 0, len(ss))
scanner := bufio.NewScanner(mi)
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+10] == "cgroup2" || txt[sepIdx+3:sepIdx+9] != "cgroup" {
continue
}
fields := strings.Split(txt, " ")
m := Mount{
Mountpoint: fields[4],
Root: fields[3],
}
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
seen, known := ss[opt]
if !known || (!all && seen) {
continue
}
ss[opt] = true
opt = strings.TrimPrefix(opt, CgroupNamePrefix)
m.Subsystems = append(m.Subsystems, opt)
numFound++
}
if len(m.Subsystems) > 0 || all {
res = append(res, m)
}
}
if err := scanner.Err(); err != nil {
return nil, err
}
return res, nil
}
func getCgroupMountsV1(all bool) ([]Mount, error) {
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return nil, err
}
defer f.Close()
allSubsystems, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return nil, err
}
allMap := make(map[string]bool)
for s := range allSubsystems {
allMap[s] = false
}
return getCgroupMountsHelper(allMap, f, all)
}
// GetOwnCgroup returns the relative path to the cgroup docker is running in.
func GetOwnCgroup(subsystem string) (string, error) {
if IsCgroup2UnifiedMode() {
return "", errUnified
}
cgroups, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return "", err
}
return getControllerPath(subsystem, cgroups)
}
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) {
if IsCgroup2UnifiedMode() {
return "", errUnified
}
cgroups, err := ParseCgroupFile("/proc/1/cgroup")
if err != nil {
return "", err
}
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 paths 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 getControllerPath(subsystem string, cgroups map[string]string) (string, error) {
if IsCgroup2UnifiedMode() {
return "", errUnified
}
if p, ok := cgroups[subsystem]; ok {
return p, nil
}
if p, ok := cgroups[CgroupNamePrefix+subsystem]; ok {
return p, nil
}
return "", NewNotFoundError(subsystem)
}

28
vendor/github.com/opencontainers/runc/libcontainer/configs/cgroup_linux.go generated vendored
View File

@ -1,5 +1,9 @@
package configs
import (
systemdDbus "github.com/coreos/go-systemd/v22/dbus"
)
type FreezerState string
const (
@ -29,18 +33,16 @@ type Cgroup struct {
// Resources contains various cgroups settings to apply
*Resources
// SystemdProps are any additional properties for systemd,
// derived from org.systemd.property.xxx annotations.
// Ignored unless systemd is used for managing cgroups.
SystemdProps []systemdDbus.Property `json:"-"`
}
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"`
// Devices is the set of access rules for devices in the container.
Devices []*DeviceRule `json:"devices"`
// Memory limit (in bytes)
Memory int64 `json:"memory"`
@ -125,6 +127,10 @@ type Resources struct {
// CpuWeight sets a proportional bandwidth limit.
CpuWeight uint64 `json:"cpu_weight"`
// CpuMax sets she maximum bandwidth limit (format: max period).
CpuMax string `json:"cpu_max"`
// SkipDevices allows to skip configuring device permissions.
// Used by e.g. kubelet while creating a parent cgroup (kubepods)
// common for many containers.
//
// NOTE it is impossible to start a container which has this flag set.
SkipDevices bool `json:"skip_devices"`
}

96
vendor/github.com/opencontainers/runc/libcontainer/configs/config.go generated vendored
View File

@ -8,7 +8,7 @@ import (
"time"
"github.com/opencontainers/runtime-spec/specs-go"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
)
@ -70,9 +70,10 @@ type Arg struct {
// 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"`
Name string `json:"name"`
Action Action `json:"action"`
ErrnoRet *uint `json:"errnoRet"`
Args []*Arg `json:"args"`
}
// TODO Windows. Many of these fields should be factored out into those parts
@ -175,7 +176,7 @@ type Config struct {
// 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
Hooks Hooks
// Version is the version of opencontainer specification that is supported.
Version string `json:"version"`
@ -202,17 +203,50 @@ type Config struct {
RootlessCgroups bool `json:"rootless_cgroups,omitempty"`
}
type Hooks struct {
type HookName string
type HookList []Hook
type Hooks map[HookName]HookList
const (
// Prestart commands are executed after the container namespaces are created,
// but before the user supplied command is executed from init.
Prestart []Hook
// Note: This hook is now deprecated
// Prestart commands are called in the Runtime namespace.
Prestart HookName = "prestart"
// CreateRuntime commands MUST be called as part of the create operation after
// the runtime environment has been created but before the pivot_root has been executed.
// CreateRuntime is called immediately after the deprecated Prestart hook.
// CreateRuntime commands are called in the Runtime Namespace.
CreateRuntime = "createRuntime"
// CreateContainer commands MUST be called as part of the create operation after
// the runtime environment has been created but before the pivot_root has been executed.
// CreateContainer commands are called in the Container namespace.
CreateContainer = "createContainer"
// StartContainer commands MUST be called as part of the start operation and before
// the container process is started.
// StartContainer commands are called in the Container namespace.
StartContainer = "startContainer"
// Poststart commands are executed after the container init process starts.
Poststart []Hook
// Poststart commands are called in the Runtime Namespace.
Poststart = "poststart"
// Poststop commands are executed after the container init process exits.
Poststop []Hook
}
// Poststop commands are called in the Runtime Namespace.
Poststop = "poststop"
)
// TODO move this to runtime-spec
// See: https://github.com/opencontainers/runtime-spec/pull/1046
const (
Creating = "creating"
Created = "created"
Running = "running"
Stopped = "stopped"
)
type Capabilities struct {
// Bounding is the set of capabilities checked by the kernel.
@ -227,32 +261,39 @@ type Capabilities struct {
Ambient []string
}
func (hooks *Hooks) UnmarshalJSON(b []byte) error {
var state struct {
Prestart []CommandHook
Poststart []CommandHook
Poststop []CommandHook
func (hooks HookList) RunHooks(state *specs.State) error {
for i, h := range hooks {
if err := h.Run(state); err != nil {
return errors.Wrapf(err, "Running hook #%d:", i)
}
}
return nil
}
func (hooks *Hooks) UnmarshalJSON(b []byte) error {
var state map[HookName][]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)
*hooks = Hooks{}
for n, commandHooks := range state {
if len(commandHooks) == 0 {
continue
}
return hooks
(*hooks)[n] = HookList{}
for _, h := range commandHooks {
(*hooks)[n] = append((*hooks)[n], h)
}
}
hooks.Prestart = deserialize(state.Prestart)
hooks.Poststart = deserialize(state.Poststart)
hooks.Poststop = deserialize(state.Poststop)
return nil
}
func (hooks Hooks) MarshalJSON() ([]byte, error) {
func (hooks *Hooks) MarshalJSON() ([]byte, error) {
serialize := func(hooks []Hook) (serializableHooks []CommandHook) {
for _, hook := range hooks {
switch chook := hook.(type) {
@ -267,9 +308,12 @@ func (hooks Hooks) MarshalJSON() ([]byte, error) {
}
return json.Marshal(map[string]interface{}{
"prestart": serialize(hooks.Prestart),
"poststart": serialize(hooks.Poststart),
"poststop": serialize(hooks.Poststop),
"prestart": serialize((*hooks)[Prestart]),
"createRuntime": serialize((*hooks)[CreateRuntime]),
"createContainer": serialize((*hooks)[CreateContainer]),
"startContainer": serialize((*hooks)[StartContainer]),
"poststart": serialize((*hooks)[Poststart]),
"poststop": serialize((*hooks)[Poststop]),
})
}

167
vendor/github.com/opencontainers/runc/libcontainer/configs/device.go generated vendored
View File

@ -3,30 +3,19 @@ package configs
import (
"fmt"
"os"
"strconv"
)
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"`
DeviceRule
// 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"`
@ -35,23 +24,147 @@ type Device struct {
// Gid of the device.
Gid uint32 `json:"gid"`
}
// Write the file to the allowed list
// DevicePermissions is a cgroupv1-style string to represent device access. It
// has to be a string for backward compatibility reasons, hence why it has
// methods to do set operations.
type DevicePermissions string
const (
deviceRead uint = (1 << iota)
deviceWrite
deviceMknod
)
func (p DevicePermissions) toSet() uint {
var set uint
for _, perm := range p {
switch perm {
case 'r':
set |= deviceRead
case 'w':
set |= deviceWrite
case 'm':
set |= deviceMknod
}
}
return set
}
func fromSet(set uint) DevicePermissions {
var perm string
if set&deviceRead == deviceRead {
perm += "r"
}
if set&deviceWrite == deviceWrite {
perm += "w"
}
if set&deviceMknod == deviceMknod {
perm += "m"
}
return DevicePermissions(perm)
}
// Union returns the union of the two sets of DevicePermissions.
func (p DevicePermissions) Union(o DevicePermissions) DevicePermissions {
lhs := p.toSet()
rhs := o.toSet()
return fromSet(lhs | rhs)
}
// Difference returns the set difference of the two sets of DevicePermissions.
// In set notation, A.Difference(B) gives you A\B.
func (p DevicePermissions) Difference(o DevicePermissions) DevicePermissions {
lhs := p.toSet()
rhs := o.toSet()
return fromSet(lhs &^ rhs)
}
// Intersection computes the intersection of the two sets of DevicePermissions.
func (p DevicePermissions) Intersection(o DevicePermissions) DevicePermissions {
lhs := p.toSet()
rhs := o.toSet()
return fromSet(lhs & rhs)
}
// IsEmpty returns whether the set of permissions in a DevicePermissions is
// empty.
func (p DevicePermissions) IsEmpty() bool {
return p == DevicePermissions("")
}
// IsValid returns whether the set of permissions is a subset of valid
// permissions (namely, {r,w,m}).
func (p DevicePermissions) IsValid() bool {
return p == fromSet(p.toSet())
}
type DeviceType rune
const (
WildcardDevice DeviceType = 'a'
BlockDevice DeviceType = 'b'
CharDevice DeviceType = 'c' // or 'u'
FifoDevice DeviceType = 'p'
)
func (t DeviceType) IsValid() bool {
switch t {
case WildcardDevice, BlockDevice, CharDevice, FifoDevice:
return true
default:
return false
}
}
func (t DeviceType) CanMknod() bool {
switch t {
case BlockDevice, CharDevice, FifoDevice:
return true
default:
return false
}
}
func (t DeviceType) CanCgroup() bool {
switch t {
case WildcardDevice, BlockDevice, CharDevice:
return true
default:
return false
}
}
type DeviceRule struct {
// Type of device ('c' for char, 'b' for block). If set to 'a', this rule
// acts as a wildcard and all fields other than Allow are ignored.
Type DeviceType `json:"type"`
// Major is the device's major number.
Major int64 `json:"major"`
// Minor is the device's minor number.
Minor int64 `json:"minor"`
// Permissions is the set of permissions that this rule applies to (in the
// cgroupv1 format -- any combination of "rwm").
Permissions DevicePermissions `json:"permissions"`
// Allow specifies whether this rule is allowed.
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 "*"
func (d *DeviceRule) CgroupString() string {
var (
major = strconv.FormatInt(d.Major, 10)
minor = strconv.FormatInt(d.Minor, 10)
)
if d.Major == Wildcard {
major = "*"
}
return fmt.Sprint(number)
if d.Minor == Wildcard {
minor = "*"
}
return fmt.Sprintf("%c %s:%s %s", d.Type, major, minor, d.Permissions)
}

111
vendor/github.com/opencontainers/runc/libcontainer/configs/device_defaults.go generated vendored
View File

@ -1,111 +0,0 @@
// +build linux
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...)
)

16
vendor/github.com/opencontainers/runc/libcontainer/configs/device_unix.go generated vendored Normal file
View File

@ -0,0 +1,16 @@
// +build !windows
package configs
import (
"errors"
"golang.org/x/sys/unix"
)
func (d *DeviceRule) Mkdev() (uint64, error) {
if d.Major == Wildcard || d.Minor == Wildcard {
return 0, errors.New("cannot mkdev() device with wildcards")
}
return unix.Mkdev(uint32(d.Major), uint32(d.Minor)), nil
}

5
vendor/github.com/opencontainers/runc/libcontainer/configs/device_windows.go generated vendored Normal file
View File

@ -0,0 +1,5 @@
package configs
func (d *DeviceRule) Mkdev() (uint64, error) {
return 0, nil
}

36
vendor/github.com/opencontainers/runc/libcontainer/devices/devices.go generated vendored
View File

@ -31,33 +31,33 @@ func DeviceFromPath(path, permissions string) (*configs.Device, error) {
}
var (
devType configs.DeviceType
mode = stat.Mode
devNumber = uint64(stat.Rdev)
major = unix.Major(devNumber)
minor = unix.Minor(devNumber)
)
if major == 0 {
return nil, ErrNotADevice
}
var (
devType rune
mode = stat.Mode
)
switch {
case mode&unix.S_IFBLK == unix.S_IFBLK:
devType = 'b'
devType = configs.BlockDevice
case mode&unix.S_IFCHR == unix.S_IFCHR:
devType = 'c'
devType = configs.CharDevice
case mode&unix.S_IFIFO == unix.S_IFIFO:
devType = configs.FifoDevice
default:
return nil, ErrNotADevice
}
return &configs.Device{
Type: devType,
Path: path,
Major: int64(major),
Minor: int64(minor),
Permissions: permissions,
FileMode: os.FileMode(mode),
Uid: stat.Uid,
Gid: stat.Gid,
DeviceRule: configs.DeviceRule{
Type: devType,
Major: int64(major),
Minor: int64(minor),
Permissions: configs.DevicePermissions(permissions),
},
Path: path,
FileMode: os.FileMode(mode),
Uid: stat.Uid,
Gid: stat.Gid,
}, nil
}

39
vendor/github.com/opencontainers/runc/libcontainer/system/linux.go generated vendored
View File

@ -5,26 +5,13 @@ package system
import (
"os"
"os/exec"
"syscall" // only for exec
"sync"
"unsafe"
"github.com/opencontainers/runc/libcontainer/user"
"golang.org/x/sys/unix"
)
// If arg2 is nonzero, set the "child subreaper" attribute of the
// calling process; if arg2 is zero, unset the attribute. When a
// process is marked as a child subreaper, all of the children
// that it creates, and their descendants, will be marked as
// having a subreaper. In effect, a subreaper fulfills the role
// of init(1) for its descendant processes. Upon termination of
// a process that is orphaned (i.e., its immediate parent has
// already terminated) and marked as having a subreaper, the
// nearest still living ancestor subreaper will receive a SIGCHLD
// signal and be able to wait(2) on the process to discover its
// termination status.
const PR_SET_CHILD_SUBREAPER = 36
type ParentDeathSignal int
func (p ParentDeathSignal) Restore() error {
@ -51,7 +38,7 @@ func Execv(cmd string, args []string, env []string) error {
return err
}
return syscall.Exec(name, args, env)
return unix.Exec(name, args, env)
}
func Prlimit(pid, resource int, limit unix.Rlimit) error {
@ -100,15 +87,23 @@ func Setctty() error {
return nil
}
var (
inUserNS bool
nsOnce sync.Once
)
// RunningInUserNS detects whether we are currently running in a user namespace.
// Originally copied from github.com/lxc/lxd/shared/util.go
func RunningInUserNS() bool {
uidmap, err := user.CurrentProcessUIDMap()
if err != nil {
// This kernel-provided file only exists if user namespaces are supported
return false
}
return UIDMapInUserNS(uidmap)
nsOnce.Do(func() {
uidmap, err := user.CurrentProcessUIDMap()
if err != nil {
// This kernel-provided file only exists if user namespaces are supported
return
}
inUserNS = UIDMapInUserNS(uidmap)
})
return inUserNS
}
func UIDMapInUserNS(uidmap []user.IDMap) bool {
@ -140,7 +135,7 @@ func GetParentNSeuid() int64 {
// SetSubreaper sets the value i as the subreaper setting for the calling process
func SetSubreaper(i int) error {
return unix.Prctl(PR_SET_CHILD_SUBREAPER, uintptr(i), 0, 0, 0)
return unix.Prctl(unix.PR_SET_CHILD_SUBREAPER, uintptr(i), 0, 0, 0)
}
// GetSubreaper returns the subreaper setting for the calling process

12
vendor/github.com/opencontainers/runc/libcontainer/system/sysconfig.go generated vendored
View File

@ -1,12 +0,0 @@
// +build cgo,linux
package system
/*
#include <unistd.h>
*/
import "C"
func GetClockTicks() int {
return int(C.sysconf(C._SC_CLK_TCK))
}

15
vendor/github.com/opencontainers/runc/libcontainer/system/sysconfig_notcgo.go generated vendored
View File

@ -1,15 +0,0 @@
// +build !cgo windows
package system
func GetClockTicks() int {
// TODO figure out a better alternative for platforms where we're missing cgo
//
// TODO Windows. This could be implemented using Win32 QueryPerformanceFrequency().
// https://msdn.microsoft.com/en-us/library/windows/desktop/ms644905(v=vs.85).aspx
//
// An example of its usage can be found here.
// https://msdn.microsoft.com/en-us/library/windows/desktop/dn553408(v=vs.85).aspx
return 100
}

30
vendor/github.com/opencontainers/runc/libcontainer/user/user.go generated vendored
View File

@ -60,7 +60,7 @@ type Group struct {
// groupFromOS converts an os/user.(*Group) to local Group
//
// (This does not include Pass, Shell or Gecos)
// (This does not include Pass or List)
func groupFromOS(g *user.Group) (Group, error) {
newGroup := Group{
Name: g.Name,
@ -162,10 +162,6 @@ func ParsePasswdFilter(r io.Reader, filter func(User) bool) ([]User, error) {
)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
line := strings.TrimSpace(s.Text())
if line == "" {
continue
@ -183,6 +179,9 @@ func ParsePasswdFilter(r io.Reader, filter func(User) bool) ([]User, error) {
out = append(out, p)
}
}
if err := s.Err(); err != nil {
return nil, err
}
return out, nil
}
@ -221,10 +220,6 @@ func ParseGroupFilter(r io.Reader, filter func(Group) bool) ([]Group, error) {
)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
text := s.Text()
if text == "" {
continue
@ -242,6 +237,9 @@ func ParseGroupFilter(r io.Reader, filter func(Group) bool) ([]Group, error) {
out = append(out, p)
}
}
if err := s.Err(); err != nil {
return nil, err
}
return out, nil
}
@ -532,10 +530,6 @@ func ParseSubIDFilter(r io.Reader, filter func(SubID) bool) ([]SubID, error) {
)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
line := strings.TrimSpace(s.Text())
if line == "" {
continue
@ -549,6 +543,9 @@ func ParseSubIDFilter(r io.Reader, filter func(SubID) bool) ([]SubID, error) {
out = append(out, p)
}
}
if err := s.Err(); err != nil {
return nil, err
}
return out, nil
}
@ -586,10 +583,6 @@ func ParseIDMapFilter(r io.Reader, filter func(IDMap) bool) ([]IDMap, error) {
)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
line := strings.TrimSpace(s.Text())
if line == "" {
continue
@ -603,6 +596,9 @@ func ParseIDMapFilter(r io.Reader, filter func(IDMap) bool) ([]IDMap, error) {
out = append(out, p)
}
}
if err := s.Err(); err != nil {
return nil, err
}
return out, nil
}