Add all pids to state output

Also update libcontainer dep

Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
This commit is contained in:
Michael Crosby 2016-02-11 14:07:34 -08:00
parent d6bd304c92
commit 532697f32f
123 changed files with 11277 additions and 547 deletions

View file

@ -10,80 +10,165 @@ host system and which is (optionally) isolated from other containers in the syst
#### Using libcontainer
To create a container you first have to initialize an instance of a factory
that will handle the creation and initialization for a container.
Because containers are spawned in a two step process you will need to provide
arguments to a binary that will be executed as the init process for the container.
To use the current binary that is spawning the containers and acting as the parent
you can use `os.Args[0]` and we have a command called `init` setup.
Because containers are spawned in a two step process you will need a binary that
will be executed as the init process for the container. In libcontainer, we use
the current binary (/proc/self/exe) to be executed as the init process, and use
arg "init", we call the first step process "bootstrap", so you always need a "init"
function as the entry of "bootstrap".
```go
root, err := libcontainer.New("/var/lib/container", libcontainer.InitArgs(os.Args[0], "init"))
func init() {
if len(os.Args) > 1 && os.Args[1] == "init" {
runtime.GOMAXPROCS(1)
runtime.LockOSThread()
factory, _ := libcontainer.New("")
if err := factory.StartInitialization(); err != nil {
logrus.Fatal(err)
}
panic("--this line should have never been executed, congratulations--")
}
}
```
Then to create a container you first have to initialize an instance of a factory
that will handle the creation and initialization for a container.
```go
factory, err := libcontainer.New("/var/lib/container", libcontainer.Cgroupfs, libcontainer.InitArgs(os.Args[0], "init"))
if err != nil {
log.Fatal(err)
logrus.Fatal(err)
return
}
```
Once you have an instance of the factory created we can create a configuration
struct describing how the container is to be created. A sample would look similar to this:
struct describing how the container is to be created. A sample would look similar to this:
```go
defaultMountFlags := syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
config := &configs.Config{
Rootfs: rootfs,
Capabilities: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Namespaces: configs.Namespaces([]configs.Namespace{
{Type: configs.NEWNS},
{Type: configs.NEWUTS},
{Type: configs.NEWIPC},
{Type: configs.NEWPID},
{Type: configs.NEWNET},
}),
Cgroups: &configs.Cgroup{
Name: "test-container",
Parent: "system",
AllowAllDevices: false,
AllowedDevices: configs.DefaultAllowedDevices,
},
Devices: configs.DefaultAutoCreatedDevices,
Hostname: "testing",
Networks: []*configs.Network{
{
Type: "loopback",
Address: "127.0.0.1/0",
Gateway: "localhost",
},
},
Rlimits: []configs.Rlimit{
{
Type: syscall.RLIMIT_NOFILE,
Hard: uint64(1024),
Soft: uint64(1024),
},
},
Rootfs: "/your/path/to/rootfs",
Capabilities: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Namespaces: configs.Namespaces([]configs.Namespace{
{Type: configs.NEWNS},
{Type: configs.NEWUTS},
{Type: configs.NEWIPC},
{Type: configs.NEWPID},
{Type: configs.NEWUSER},
{Type: configs.NEWNET},
}),
Cgroups: &configs.Cgroup{
Name: "test-container",
Parent: "system",
Resources: &configs.Resources{
MemorySwappiness: -1,
AllowAllDevices: false,
AllowedDevices: configs.DefaultAllowedDevices,
},
},
MaskPaths: []string{
"/proc/kcore",
},
ReadonlyPaths: []string{
"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
},
Devices: configs.DefaultAutoCreatedDevices,
Hostname: "testing",
Mounts: []*configs.Mount{
{
Source: "proc",
Destination: "/proc",
Device: "proc",
Flags: defaultMountFlags,
},
{
Source: "tmpfs",
Destination: "/dev",
Device: "tmpfs",
Flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME,
Data: "mode=755",
},
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
Flags: syscall.MS_NOSUID | syscall.MS_NOEXEC,
Data: "newinstance,ptmxmode=0666,mode=0620,gid=5",
},
{
Device: "tmpfs",
Source: "shm",
Destination: "/dev/shm",
Data: "mode=1777,size=65536k",
Flags: defaultMountFlags,
},
{
Source: "mqueue",
Destination: "/dev/mqueue",
Device: "mqueue",
Flags: defaultMountFlags,
},
{
Source: "sysfs",
Destination: "/sys",
Device: "sysfs",
Flags: defaultMountFlags | syscall.MS_RDONLY,
},
},
UidMappings: []configs.IDMap{
{
ContainerID: 0,
Host: 1000,
size: 65536,
},
},
GidMappings: []configs.IDMap{
{
ContainerID: 0,
Host: 1000,
size: 65536,
},
},
Networks: []*configs.Network{
{
Type: "loopback",
Address: "127.0.0.1/0",
Gateway: "localhost",
},
},
Rlimits: []configs.Rlimit{
{
Type: syscall.RLIMIT_NOFILE,
Hard: uint64(1025),
Soft: uint64(1025),
},
},
}
```
Once you have the configuration populated you can create a container:
```go
container, err := root.Create("container-id", config)
container, err := factory.Create("container-id", config)
if err != nil {
logrus.Fatal(err)
return
}
```
To spawn bash as the initial process inside the container and have the
@ -91,23 +176,25 @@ processes pid returned in order to wait, signal, or kill the process:
```go
process := &libcontainer.Process{
Args: []string{"/bin/bash"},
Env: []string{"PATH=/bin"},
User: "daemon",
Stdin: os.Stdin,
Stdout: os.Stdout,
Stderr: os.Stderr,
Args: []string{"/bin/bash"},
Env: []string{"PATH=/bin"},
User: "daemon",
Stdin: os.Stdin,
Stdout: os.Stdout,
Stderr: os.Stderr,
}
err := container.Start(process)
if err != nil {
log.Fatal(err)
logrus.Fatal(err)
container.Destroy()
return
}
// wait for the process to finish.
status, err := process.Wait()
_, err := process.Wait()
if err != nil {
log.Fatal(err)
logrus.Fatal(err)
}
// destroy the container.
@ -124,7 +211,6 @@ processes, err := container.Processes()
// it's processes.
stats, err := container.Stats()
// pause all processes inside the container.
container.Pause()

View file

@ -142,6 +142,7 @@ system resources like cpu, memory, and device access.
| perf_event | 1 |
| freezer | 1 |
| hugetlb | 1 |
| pids | 1 |
All cgroup subsystem are joined so that statistics can be collected from
@ -199,7 +200,7 @@ provide a good default for security and flexibility for the applications.
| CAP_SYS_BOOT | 0 |
| CAP_LEASE | 0 |
| CAP_WAKE_ALARM | 0 |
| CAP_BLOCK_SUSPE | 0 |
| CAP_BLOCK_SUSPEND | 0 |
Additional security layers like [apparmor](https://wiki.ubuntu.com/AppArmor)

View file

@ -15,6 +15,9 @@ type Manager interface {
// Returns the PIDs inside the cgroup set
GetPids() ([]int, error)
// Returns the PIDs inside the cgroup set & all sub-cgroups
GetAllPids() ([]int, error)
// Returns statistics for the cgroup set
GetStats() (*Stats, error)

View file

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

View file

@ -23,6 +23,7 @@ var (
&MemoryGroup{},
&CpuGroup{},
&CpuacctGroup{},
&PidsGroup{},
&BlkioGroup{},
&HugetlbGroup{},
&NetClsGroup{},
@ -93,11 +94,10 @@ func getCgroupRoot() (string, error) {
}
type cgroupData struct {
root string
parent string
name string
config *configs.Cgroup
pid int
root string
innerPath string
config *configs.Cgroup
pid int
}
func (m *Manager) Apply(pid int) (err error) {
@ -112,6 +112,22 @@ func (m *Manager) Apply(pid int) (err error) {
return err
}
if c.Paths != nil {
paths := make(map[string]string)
for name, path := range c.Paths {
_, err := d.path(name)
if err != nil {
if cgroups.IsNotFound(err) {
continue
}
return err
}
paths[name] = path
}
m.Paths = paths
return cgroups.EnterPid(m.Paths, pid)
}
paths := make(map[string]string)
defer func() {
if err != nil {
@ -135,17 +151,13 @@ func (m *Manager) Apply(pid int) (err error) {
paths[sys.Name()] = p
}
m.Paths = paths
if paths["cpu"] != "" {
if err := CheckCpushares(paths["cpu"], c.Resources.CpuShares); err != nil {
return err
}
}
return nil
}
func (m *Manager) Destroy() error {
if m.Cgroups.Paths != nil {
return nil
}
m.mu.Lock()
defer m.mu.Unlock()
if err := cgroups.RemovePaths(m.Paths); err != nil {
@ -179,15 +191,28 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
}
func (m *Manager) Set(container *configs.Config) error {
for name, path := range m.Paths {
sys, err := subsystems.Get(name)
if err == errSubsystemDoesNotExist || !cgroups.PathExists(path) {
continue
for _, sys := range subsystems {
// Generate fake cgroup data.
d, err := getCgroupData(container.Cgroups, -1)
if err != nil {
return err
}
// Get the path, but don't error out if the cgroup wasn't found.
path, err := d.path(sys.Name())
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := sys.Set(path, container.Cgroups); err != nil {
return err
}
}
if m.Paths["cpu"] != "" {
if err := CheckCpushares(m.Paths["cpu"], container.Cgroups.Resources.CpuShares); err != nil {
return err
}
}
return nil
}
@ -217,31 +242,50 @@ func (m *Manager) Freeze(state configs.FreezerState) error {
}
func (m *Manager) GetPids() ([]int, error) {
d, err := getCgroupData(m.Cgroups, 0)
dir, err := getCgroupPath(m.Cgroups)
if err != nil {
return nil, err
}
dir, err := d.path("devices")
if err != nil {
return nil, err
}
return cgroups.GetPids(dir)
}
func (m *Manager) GetAllPids() ([]int, error) {
dir, err := getCgroupPath(m.Cgroups)
if err != nil {
return nil, err
}
return cgroups.GetAllPids(dir)
}
func getCgroupPath(c *configs.Cgroup) (string, error) {
d, err := getCgroupData(c, 0)
if err != nil {
return "", err
}
return d.path("devices")
}
func getCgroupData(c *configs.Cgroup, pid int) (*cgroupData, error) {
root, err := getCgroupRoot()
if err != nil {
return nil, err
}
if (c.Name != "" || c.Parent != "") && c.Path != "" {
return nil, fmt.Errorf("cgroup: either Path or Name and Parent should be used")
}
innerPath := c.Path
if innerPath == "" {
innerPath = filepath.Join(c.Parent, c.Name)
}
return &cgroupData{
root: root,
parent: c.Parent,
name: c.Name,
config: c,
pid: pid,
root: root,
innerPath: c.Path,
config: c,
pid: pid,
}, nil
}
@ -269,11 +313,10 @@ func (raw *cgroupData) path(subsystem string) (string, error) {
return "", err
}
cgPath := filepath.Join(raw.parent, raw.name)
// If the cgroup name/path is absolute do not look relative to the cgroup of the init process.
if filepath.IsAbs(cgPath) {
if filepath.IsAbs(raw.innerPath) {
// Sometimes subsystems can be mounted togethger as 'cpu,cpuacct'.
return filepath.Join(raw.root, filepath.Base(mnt), cgPath), nil
return filepath.Join(raw.root, filepath.Base(mnt), raw.innerPath), nil
}
parentPath, err := raw.parentPath(subsystem, mnt, root)
@ -281,7 +324,7 @@ func (raw *cgroupData) path(subsystem string) (string, error) {
return "", err
}
return filepath.Join(parentPath, cgPath), nil
return filepath.Join(parentPath, raw.innerPath), nil
}
func (raw *cgroupData) join(subsystem string) (string, error) {

View file

@ -22,15 +22,10 @@ func (s *BlkioGroup) Name() string {
}
func (s *BlkioGroup) Apply(d *cgroupData) error {
dir, err := d.join("blkio")
_, err := d.join("blkio")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

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

View file

@ -22,15 +22,10 @@ func (s *CpuGroup) Name() string {
func (s *CpuGroup) Apply(d *cgroupData) error {
// We always want to join the cpu group, to allow fair cpu scheduling
// on a container basis
dir, err := d.join("cpu")
_, err := d.join("cpu")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

@ -0,0 +1,163 @@
// +build linux
package fs
import (
"fmt"
"strconv"
"testing"
"github.com/opencontainers/runc/libcontainer/cgroups"
)
func TestCpuSetShares(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
const (
sharesBefore = 1024
sharesAfter = 512
)
helper.writeFileContents(map[string]string{
"cpu.shares": strconv.Itoa(sharesBefore),
})
helper.CgroupData.config.Resources.CpuShares = sharesAfter
cpu := &CpuGroup{}
if err := cpu.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "cpu.shares")
if err != nil {
t.Fatalf("Failed to parse cpu.shares - %s", err)
}
if value != sharesAfter {
t.Fatal("Got the wrong value, set cpu.shares failed.")
}
}
func TestCpuSetBandWidth(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
const (
quotaBefore = 8000
quotaAfter = 5000
periodBefore = 10000
periodAfter = 7000
rtRuntimeBefore = 8000
rtRuntimeAfter = 5000
rtPeriodBefore = 10000
rtPeriodAfter = 7000
)
helper.writeFileContents(map[string]string{
"cpu.cfs_quota_us": strconv.Itoa(quotaBefore),
"cpu.cfs_period_us": strconv.Itoa(periodBefore),
"cpu.rt_runtime_us": strconv.Itoa(rtRuntimeBefore),
"cpu.rt_period_us": strconv.Itoa(rtPeriodBefore),
})
helper.CgroupData.config.Resources.CpuQuota = quotaAfter
helper.CgroupData.config.Resources.CpuPeriod = periodAfter
helper.CgroupData.config.Resources.CpuRtRuntime = rtRuntimeAfter
helper.CgroupData.config.Resources.CpuRtPeriod = rtPeriodAfter
cpu := &CpuGroup{}
if err := cpu.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
quota, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_quota_us")
if err != nil {
t.Fatalf("Failed to parse cpu.cfs_quota_us - %s", err)
}
if quota != quotaAfter {
t.Fatal("Got the wrong value, set cpu.cfs_quota_us failed.")
}
period, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_period_us")
if err != nil {
t.Fatalf("Failed to parse cpu.cfs_period_us - %s", err)
}
if period != periodAfter {
t.Fatal("Got the wrong value, set cpu.cfs_period_us failed.")
}
rtRuntime, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_runtime_us")
if err != nil {
t.Fatalf("Failed to parse cpu.rt_runtime_us - %s", err)
}
if rtRuntime != rtRuntimeAfter {
t.Fatal("Got the wrong value, set cpu.rt_runtime_us failed.")
}
rtPeriod, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_period_us")
if err != nil {
t.Fatalf("Failed to parse cpu.rt_period_us - %s", err)
}
if rtPeriod != rtPeriodAfter {
t.Fatal("Got the wrong value, set cpu.rt_period_us failed.")
}
}
func TestCpuStats(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
const (
kNrPeriods = 2000
kNrThrottled = 200
kThrottledTime = uint64(18446744073709551615)
)
cpuStatContent := fmt.Sprintf("nr_periods %d\n nr_throttled %d\n throttled_time %d\n",
kNrPeriods, kNrThrottled, kThrottledTime)
helper.writeFileContents(map[string]string{
"cpu.stat": cpuStatContent,
})
cpu := &CpuGroup{}
actualStats := *cgroups.NewStats()
err := cpu.GetStats(helper.CgroupPath, &actualStats)
if err != nil {
t.Fatal(err)
}
expectedStats := cgroups.ThrottlingData{
Periods: kNrPeriods,
ThrottledPeriods: kNrThrottled,
ThrottledTime: kThrottledTime}
expectThrottlingDataEquals(t, expectedStats, actualStats.CpuStats.ThrottlingData)
}
func TestNoCpuStatFile(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
cpu := &CpuGroup{}
actualStats := *cgroups.NewStats()
err := cpu.GetStats(helper.CgroupPath, &actualStats)
if err != nil {
t.Fatal("Expected not to fail, but did")
}
}
func TestInvalidCpuStat(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
cpuStatContent := `nr_periods 2000
nr_throttled 200
throttled_time fortytwo`
helper.writeFileContents(map[string]string{
"cpu.stat": cpuStatContent,
})
cpu := &CpuGroup{}
actualStats := *cgroups.NewStats()
err := cpu.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failed stat parsing.")
}
}

View file

@ -4,6 +4,7 @@ package fs
import (
"bytes"
"fmt"
"io/ioutil"
"os"
"path/filepath"
@ -11,6 +12,7 @@ import (
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
)
type CpusetGroup struct {
@ -63,11 +65,6 @@ func (s *CpusetGroup) ApplyDir(dir string, cgroup *configs.Cgroup, pid int) erro
if err := s.ensureParent(dir, root); err != nil {
return err
}
// the default values inherit from parent cgroup are already set in
// s.ensureParent, cover these if we have our own
if err := s.Set(dir, cgroup); err != nil {
return err
}
// because we are not using d.join we need to place the pid into the procs file
// unlike the other subsystems
if err := writeFile(dir, "cgroup.procs", strconv.Itoa(pid)); err != nil {
@ -92,9 +89,13 @@ func (s *CpusetGroup) getSubsystemSettings(parent string) (cpus []byte, mems []b
// it's parent.
func (s *CpusetGroup) ensureParent(current, root string) error {
parent := filepath.Dir(current)
if filepath.Clean(parent) == root {
if libcontainerUtils.CleanPath(parent) == root {
return nil
}
// Avoid infinite recursion.
if parent == current {
return fmt.Errorf("cpuset: cgroup parent path outside cgroup root")
}
if err := s.ensureParent(parent, root); err != nil {
return err
}

View file

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

View file

@ -15,21 +15,29 @@ func (s *DevicesGroup) Name() string {
}
func (s *DevicesGroup) Apply(d *cgroupData) error {
dir, err := d.join("devices")
_, err := d.join("devices")
if err != nil {
// We will return error even it's `not found` error, devices
// cgroup is hard requirement for container's security.
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}
func (s *DevicesGroup) Set(path string, cgroup *configs.Cgroup) error {
devices := cgroup.Resources.Devices
if len(devices) > 0 {
for _, dev := range devices {
file := "devices.deny"
if dev.Allow {
file = "devices.allow"
}
if err := writeFile(path, file, dev.CgroupString()); err != nil {
return err
}
}
return nil
}
if !cgroup.Resources.AllowAllDevices {
if err := writeFile(path, "devices.deny", "a"); err != nil {
return err

View file

@ -0,0 +1,84 @@
// +build linux
package fs
import (
"testing"
"github.com/opencontainers/runc/libcontainer/configs"
)
var (
allowedDevices = []*configs.Device{
{
Path: "/dev/zero",
Type: 'c',
Major: 1,
Minor: 5,
Permissions: "rwm",
FileMode: 0666,
},
}
allowedList = "c 1:5 rwm"
deniedDevices = []*configs.Device{
{
Path: "/dev/null",
Type: 'c',
Major: 1,
Minor: 3,
Permissions: "rwm",
FileMode: 0666,
},
}
deniedList = "c 1:3 rwm"
)
func TestDevicesSetAllow(t *testing.T) {
helper := NewCgroupTestUtil("devices", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"devices.deny": "a",
})
helper.CgroupData.config.Resources.AllowAllDevices = false
helper.CgroupData.config.Resources.AllowedDevices = allowedDevices
devices := &DevicesGroup{}
if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamString(helper.CgroupPath, "devices.allow")
if err != nil {
t.Fatalf("Failed to parse devices.allow - %s", err)
}
if value != allowedList {
t.Fatal("Got the wrong value, set devices.allow failed.")
}
}
func TestDevicesSetDeny(t *testing.T) {
helper := NewCgroupTestUtil("devices", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"devices.allow": "a",
})
helper.CgroupData.config.Resources.AllowAllDevices = true
helper.CgroupData.config.Resources.DeniedDevices = deniedDevices
devices := &DevicesGroup{}
if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamString(helper.CgroupPath, "devices.deny")
if err != nil {
t.Fatalf("Failed to parse devices.deny - %s", err)
}
if value != deniedList {
t.Fatal("Got the wrong value, set devices.deny failed.")
}
}

View file

@ -19,15 +19,10 @@ func (s *FreezerGroup) Name() string {
}
func (s *FreezerGroup) Apply(d *cgroupData) error {
dir, err := d.join("freezer")
_, err := d.join("freezer")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

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

View file

@ -19,15 +19,10 @@ func (s *HugetlbGroup) Name() string {
}
func (s *HugetlbGroup) Apply(d *cgroupData) error {
dir, err := d.join("hugetlb")
_, err := d.join("hugetlb")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

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

View file

@ -32,8 +32,9 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
return err
}
}
if err := s.Set(path, d.config); err != nil {
// We have to set kernel memory here, as we can't change it once
// processes have been attached.
if err := s.SetKernelMemory(path, d.config); err != nil {
return err
}
}
@ -50,7 +51,17 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
if err != nil && !cgroups.IsNotFound(err) {
return err
}
return nil
}
func (s *MemoryGroup) SetKernelMemory(path string, cgroup *configs.Cgroup) error {
// This has to be done separately because it has special constraints (it
// can't be done after there are processes attached to the cgroup).
if cgroup.Resources.KernelMemory > 0 {
if err := writeFile(path, "memory.kmem.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemory, 10)); err != nil {
return err
}
}
return nil
}
@ -70,12 +81,6 @@ func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
return err
}
}
if cgroup.Resources.KernelMemory > 0 {
if err := writeFile(path, "memory.kmem.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemory, 10)); err != nil {
return err
}
}
if cgroup.Resources.OomKillDisable {
if err := writeFile(path, "memory.oom_control", "1"); err != nil {
return err
@ -157,6 +162,7 @@ func getMemoryData(path, name string) (cgroups.MemoryData, error) {
usage := strings.Join([]string{moduleName, "usage_in_bytes"}, ".")
maxUsage := strings.Join([]string{moduleName, "max_usage_in_bytes"}, ".")
failcnt := strings.Join([]string{moduleName, "failcnt"}, ".")
limit := strings.Join([]string{moduleName, "limit_in_bytes"}, ".")
value, err := getCgroupParamUint(path, usage)
if err != nil {
@ -182,6 +188,14 @@ func getMemoryData(path, name string) (cgroups.MemoryData, error) {
return cgroups.MemoryData{}, fmt.Errorf("failed to parse %s - %v", failcnt, err)
}
memoryData.Failcnt = value
value, err = getCgroupParamUint(path, limit)
if err != nil {
if moduleName != "memory" && os.IsNotExist(err) {
return cgroups.MemoryData{}, nil
}
return cgroups.MemoryData{}, fmt.Errorf("failed to parse %s - %v", limit, err)
}
memoryData.Limit = value
return memoryData, nil
}

View file

@ -0,0 +1,339 @@
// +build linux
package fs
import (
"strconv"
"testing"
"github.com/opencontainers/runc/libcontainer/cgroups"
)
const (
memoryStatContents = `cache 512
rss 1024`
memoryUsageContents = "2048\n"
memoryMaxUsageContents = "4096\n"
memoryFailcnt = "100\n"
memoryLimitContents = "8192\n"
)
func TestMemorySetMemory(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
memoryBefore = 314572800 // 300M
memoryAfter = 524288000 // 500M
reservationBefore = 209715200 // 200M
reservationAfter = 314572800 // 300M
)
helper.writeFileContents(map[string]string{
"memory.limit_in_bytes": strconv.Itoa(memoryBefore),
"memory.soft_limit_in_bytes": strconv.Itoa(reservationBefore),
})
helper.CgroupData.config.Resources.Memory = memoryAfter
helper.CgroupData.config.Resources.MemoryReservation = reservationAfter
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
}
if value != memoryAfter {
t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
}
value, err = getCgroupParamUint(helper.CgroupPath, "memory.soft_limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.soft_limit_in_bytes - %s", err)
}
if value != reservationAfter {
t.Fatal("Got the wrong value, set memory.soft_limit_in_bytes failed.")
}
}
func TestMemorySetMemoryswap(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
memoryswapBefore = 314572800 // 300M
memoryswapAfter = 524288000 // 500M
)
helper.writeFileContents(map[string]string{
"memory.memsw.limit_in_bytes": strconv.Itoa(memoryswapBefore),
})
helper.CgroupData.config.Resources.MemorySwap = memoryswapAfter
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.memsw.limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.memsw.limit_in_bytes - %s", err)
}
if value != memoryswapAfter {
t.Fatal("Got the wrong value, set memory.memsw.limit_in_bytes failed.")
}
}
func TestMemorySetKernelMemory(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
kernelMemoryBefore = 314572800 // 300M
kernelMemoryAfter = 524288000 // 500M
)
helper.writeFileContents(map[string]string{
"memory.kmem.limit_in_bytes": strconv.Itoa(kernelMemoryBefore),
})
helper.CgroupData.config.Resources.KernelMemory = kernelMemoryAfter
memory := &MemoryGroup{}
if err := memory.SetKernelMemory(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.kmem.limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.kmem.limit_in_bytes - %s", err)
}
if value != kernelMemoryAfter {
t.Fatal("Got the wrong value, set memory.kmem.limit_in_bytes failed.")
}
}
func TestMemorySetMemorySwappinessDefault(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
swappinessBefore = 60 //deafult is 60
swappinessAfter = 0
)
helper.writeFileContents(map[string]string{
"memory.swappiness": strconv.Itoa(swappinessBefore),
})
helper.CgroupData.config.Resources.Memory = swappinessAfter
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.swappiness")
if err != nil {
t.Fatalf("Failed to parse memory.swappiness - %s", err)
}
if value != swappinessAfter {
t.Fatal("Got the wrong value, set memory.swappiness failed.")
}
}
func TestMemoryStats(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": memoryUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.failcnt": memoryFailcnt,
"memory.memsw.usage_in_bytes": memoryUsageContents,
"memory.memsw.max_usage_in_bytes": memoryMaxUsageContents,
"memory.memsw.failcnt": memoryFailcnt,
"memory.memsw.limit_in_bytes": memoryLimitContents,
"memory.kmem.usage_in_bytes": memoryUsageContents,
"memory.kmem.max_usage_in_bytes": memoryMaxUsageContents,
"memory.kmem.failcnt": memoryFailcnt,
"memory.kmem.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err != nil {
t.Fatal(err)
}
expectedStats := cgroups.MemoryStats{Cache: 512, Usage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, SwapUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, KernelUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, Stats: map[string]uint64{"cache": 512, "rss": 1024}}
expectMemoryStatEquals(t, expectedStats, actualStats.MemoryStats)
}
func TestMemoryStatsNoStatFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.usage_in_bytes": memoryUsageContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err != nil {
t.Fatal(err)
}
}
func TestMemoryStatsNoUsageFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsNoMaxUsageFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": memoryUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsNoLimitInBytesFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": memoryUsageContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsBadStatFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": "rss rss",
"memory.usage_in_bytes": memoryUsageContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsBadUsageFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": "bad",
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsBadMaxUsageFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": memoryUsageContents,
"memory.max_usage_in_bytes": "bad",
"memory.limit_in_bytes": memoryLimitContents,
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemoryStatsBadLimitInBytesFile(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"memory.stat": memoryStatContents,
"memory.usage_in_bytes": memoryUsageContents,
"memory.max_usage_in_bytes": memoryMaxUsageContents,
"memory.limit_in_bytes": "bad",
})
memory := &MemoryGroup{}
actualStats := *cgroups.NewStats()
err := memory.GetStats(helper.CgroupPath, &actualStats)
if err == nil {
t.Fatal("Expected failure")
}
}
func TestMemorySetOomControl(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
oom_kill_disable = 1 // disable oom killer, default is 0
)
helper.writeFileContents(map[string]string{
"memory.oom_control": strconv.Itoa(oom_kill_disable),
})
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.oom_control")
if err != nil {
t.Fatalf("Failed to parse memory.oom_control - %s", err)
}
if value != oom_kill_disable {
t.Fatalf("Got the wrong value, set memory.oom_control failed.")
}
}

View file

@ -15,15 +15,10 @@ func (s *NetClsGroup) Name() string {
}
func (s *NetClsGroup) Apply(d *cgroupData) error {
dir, err := d.join("net_cls")
_, err := d.join("net_cls")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

@ -0,0 +1,38 @@
// +build linux
package fs
import (
"testing"
)
const (
classidBefore = "0x100002"
classidAfter = "0x100001"
)
func TestNetClsSetClassid(t *testing.T) {
helper := NewCgroupTestUtil("net_cls", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"net_cls.classid": classidBefore,
})
helper.CgroupData.config.Resources.NetClsClassid = classidAfter
netcls := &NetClsGroup{}
if err := netcls.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
// As we are in mock environment, we can't get correct value of classid from
// net_cls.classid.
// So. we just judge if we successfully write classid into file
value, err := getCgroupParamString(helper.CgroupPath, "net_cls.classid")
if err != nil {
t.Fatalf("Failed to parse net_cls.classid - %s", err)
}
if value != classidAfter {
t.Fatal("Got the wrong value, set net_cls.classid failed.")
}
}

View file

@ -15,15 +15,10 @@ func (s *NetPrioGroup) Name() string {
}
func (s *NetPrioGroup) Apply(d *cgroupData) error {
dir, err := d.join("net_prio")
_, err := d.join("net_prio")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := s.Set(dir, d.config); err != nil {
return err
}
return nil
}

View file

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

View file

@ -0,0 +1,57 @@
// +build linux
package fs
import (
"fmt"
"strconv"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
)
type PidsGroup struct {
}
func (s *PidsGroup) Name() string {
return "pids"
}
func (s *PidsGroup) Apply(d *cgroupData) error {
_, err := d.join("pids")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
return nil
}
func (s *PidsGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.PidsLimit != 0 {
// "max" is the fallback value.
limit := "max"
if cgroup.Resources.PidsLimit > 0 {
limit = strconv.FormatInt(cgroup.Resources.PidsLimit, 10)
}
if err := writeFile(path, "pids.max", limit); err != nil {
return err
}
}
return nil
}
func (s *PidsGroup) Remove(d *cgroupData) error {
return removePath(d.path("pids"))
}
func (s *PidsGroup) GetStats(path string, stats *cgroups.Stats) error {
value, err := getCgroupParamUint(path, "pids.current")
if err != nil {
return fmt.Errorf("failed to parse pids.current - %s", err)
}
stats.PidsStats.Current = value
return nil
}

View file

@ -0,0 +1,83 @@
// +build linux
package fs
import (
"strconv"
"testing"
"github.com/opencontainers/runc/libcontainer/cgroups"
)
const (
maxUnlimited = -1
maxLimited = 1024
)
func TestPidsSetMax(t *testing.T) {
helper := NewCgroupTestUtil("pids", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"pids.max": "max",
})
helper.CgroupData.config.Resources.PidsLimit = maxLimited
pids := &PidsGroup{}
if err := pids.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "pids.max")
if err != nil {
t.Fatalf("Failed to parse pids.max - %s", err)
}
if value != maxLimited {
t.Fatalf("Expected %d, got %d for setting pids.max - limited", maxLimited, value)
}
}
func TestPidsSetUnlimited(t *testing.T) {
helper := NewCgroupTestUtil("pids", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"pids.max": strconv.Itoa(maxLimited),
})
helper.CgroupData.config.Resources.PidsLimit = maxUnlimited
pids := &PidsGroup{}
if err := pids.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamString(helper.CgroupPath, "pids.max")
if err != nil {
t.Fatalf("Failed to parse pids.max - %s", err)
}
if value != "max" {
t.Fatalf("Expected %s, got %s for setting pids.max - unlimited", "max", value)
}
}
func TestPidsStats(t *testing.T) {
helper := NewCgroupTestUtil("pids", t)
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"pids.current": strconv.Itoa(1337),
"pids.max": strconv.Itoa(maxLimited),
})
pids := &PidsGroup{}
stats := *cgroups.NewStats()
if err := pids.GetStats(helper.CgroupPath, &stats); err != nil {
t.Fatal(err)
}
if stats.PidsStats.Current != 1337 {
t.Fatalf("Expected %d, got %d for pids.current", 1337, stats.PidsStats.Current)
}
}

View file

@ -0,0 +1,117 @@
// +build linux
package fs
import (
"fmt"
"testing"
"github.com/Sirupsen/logrus"
"github.com/opencontainers/runc/libcontainer/cgroups"
)
func blkioStatEntryEquals(expected, actual []cgroups.BlkioStatEntry) error {
if len(expected) != len(actual) {
return fmt.Errorf("blkioStatEntries length do not match")
}
for i, expValue := range expected {
actValue := actual[i]
if expValue != actValue {
return fmt.Errorf("Expected blkio stat entry %v but found %v", expValue, actValue)
}
}
return nil
}
func expectBlkioStatsEquals(t *testing.T, expected, actual cgroups.BlkioStats) {
if err := blkioStatEntryEquals(expected.IoServiceBytesRecursive, actual.IoServiceBytesRecursive); err != nil {
logrus.Printf("blkio IoServiceBytesRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoServicedRecursive, actual.IoServicedRecursive); err != nil {
logrus.Printf("blkio IoServicedRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoQueuedRecursive, actual.IoQueuedRecursive); err != nil {
logrus.Printf("blkio IoQueuedRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.SectorsRecursive, actual.SectorsRecursive); err != nil {
logrus.Printf("blkio SectorsRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoServiceTimeRecursive, actual.IoServiceTimeRecursive); err != nil {
logrus.Printf("blkio IoServiceTimeRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoWaitTimeRecursive, actual.IoWaitTimeRecursive); err != nil {
logrus.Printf("blkio IoWaitTimeRecursive do not match - %s\n", err)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoMergedRecursive, actual.IoMergedRecursive); err != nil {
logrus.Printf("blkio IoMergedRecursive do not match - %v vs %v\n", expected.IoMergedRecursive, actual.IoMergedRecursive)
t.Fail()
}
if err := blkioStatEntryEquals(expected.IoTimeRecursive, actual.IoTimeRecursive); err != nil {
logrus.Printf("blkio IoTimeRecursive do not match - %s\n", err)
t.Fail()
}
}
func expectThrottlingDataEquals(t *testing.T, expected, actual cgroups.ThrottlingData) {
if expected != actual {
logrus.Printf("Expected throttling data %v but found %v\n", expected, actual)
t.Fail()
}
}
func expectHugetlbStatEquals(t *testing.T, expected, actual cgroups.HugetlbStats) {
if expected != actual {
logrus.Printf("Expected hugetlb stats %v but found %v\n", expected, actual)
t.Fail()
}
}
func expectMemoryStatEquals(t *testing.T, expected, actual cgroups.MemoryStats) {
expectMemoryDataEquals(t, expected.Usage, actual.Usage)
expectMemoryDataEquals(t, expected.SwapUsage, actual.SwapUsage)
expectMemoryDataEquals(t, expected.KernelUsage, actual.KernelUsage)
for key, expValue := range expected.Stats {
actValue, ok := actual.Stats[key]
if !ok {
logrus.Printf("Expected memory stat key %s not found\n", key)
t.Fail()
}
if expValue != actValue {
logrus.Printf("Expected memory stat value %d but found %d\n", expValue, actValue)
t.Fail()
}
}
}
func expectMemoryDataEquals(t *testing.T, expected, actual cgroups.MemoryData) {
if expected.Usage != actual.Usage {
logrus.Printf("Expected memory usage %d but found %d\n", expected.Usage, actual.Usage)
t.Fail()
}
if expected.MaxUsage != actual.MaxUsage {
logrus.Printf("Expected memory max usage %d but found %d\n", expected.MaxUsage, actual.MaxUsage)
t.Fail()
}
if expected.Failcnt != actual.Failcnt {
logrus.Printf("Expected memory failcnt %d but found %d\n", expected.Failcnt, actual.Failcnt)
t.Fail()
}
if expected.Limit != actual.Limit {
logrus.Printf("Expected memory limit %d but found %d\n", expected.Limit, actual.Limit)
t.Fail()
}
}

View file

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

View file

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

View file

@ -36,7 +36,9 @@ type MemoryData struct {
Usage uint64 `json:"usage,omitempty"`
MaxUsage uint64 `json:"max_usage,omitempty"`
Failcnt uint64 `json:"failcnt"`
Limit uint64 `json:"limit"`
}
type MemoryStats struct {
// memory used for cache
Cache uint64 `json:"cache,omitempty"`
@ -49,6 +51,11 @@ type MemoryStats struct {
Stats map[string]uint64 `json:"stats,omitempty"`
}
type PidsStats struct {
// number of pids in the cgroup
Current uint64 `json:"current,omitempty"`
}
type BlkioStatEntry struct {
Major uint64 `json:"major,omitempty"`
Minor uint64 `json:"minor,omitempty"`
@ -80,6 +87,7 @@ type HugetlbStats struct {
type Stats struct {
CpuStats CpuStats `json:"cpu_stats,omitempty"`
MemoryStats MemoryStats `json:"memory_stats,omitempty"`
PidsStats PidsStats `json:"pids_stats,omitempty"`
BlkioStats BlkioStats `json:"blkio_stats,omitempty"`
// the map is in the format "size of hugepage: stats of the hugepage"
HugetlbStats map[string]HugetlbStats `json:"hugetlb_stats,omitempty"`

View file

@ -26,6 +26,10 @@ func (m *Manager) GetPids() ([]int, error) {
return nil, fmt.Errorf("Systemd not supported")
}
func (m *Manager) GetAllPids() ([]int, error) {
return nil, fmt.Errorf("Systemd not supported")
}
func (m *Manager) Destroy() error {
return fmt.Errorf("Systemd not supported")
}

View file

@ -55,6 +55,7 @@ var subsystems = subsystemSet{
&fs.MemoryGroup{},
&fs.CpuGroup{},
&fs.CpuacctGroup{},
&fs.PidsGroup{},
&fs.BlkioGroup{},
&fs.HugetlbGroup{},
&fs.PerfEventGroup{},
@ -167,6 +168,23 @@ func (m *Manager) Apply(pid int) error {
properties []systemdDbus.Property
)
if c.Paths != nil {
paths := make(map[string]string)
for name, path := range c.Paths {
_, err := getSubsystemPath(m.Cgroups, name)
if err != nil {
// Don't fail if a cgroup hierarchy was not found, just skip this subsystem
if cgroups.IsNotFound(err) {
continue
}
return err
}
paths[name] = path
}
m.Paths = paths
return cgroups.EnterPid(m.Paths, pid)
}
if c.Parent != "" {
slice = c.Parent
}
@ -233,7 +251,7 @@ func (m *Manager) Apply(pid int) error {
return err
}
// we need to manually join the freezer, net_cls, net_prio and cpuset cgroup in systemd
// we need to manually join the freezer, net_cls, net_prio, pids and cpuset cgroup in systemd
// because it does not currently support it via the dbus api.
if err := joinFreezer(c, pid); err != nil {
return err
@ -246,6 +264,10 @@ func (m *Manager) Apply(pid int) error {
return err
}
if err := joinPids(c, pid); err != nil {
return err
}
if err := joinCpuset(c, pid); err != nil {
return err
}
@ -277,17 +299,13 @@ func (m *Manager) Apply(pid int) error {
paths[s.Name()] = subsystemPath
}
m.Paths = paths
if paths["cpu"] != "" {
if err := fs.CheckCpushares(paths["cpu"], c.Resources.CpuShares); err != nil {
return err
}
}
return nil
}
func (m *Manager) Destroy() error {
if m.Cgroups.Paths != nil {
return nil
}
m.mu.Lock()
defer m.mu.Unlock()
theConn.StopUnit(getUnitName(m.Cgroups), "replace", nil)
@ -330,68 +348,74 @@ func join(c *configs.Cgroup, subsystem string, pid int) (string, error) {
}
func joinCpu(c *configs.Cgroup, pid int) error {
path, err := getSubsystemPath(c, "cpu")
_, err := join(c, "cpu", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if c.Resources.CpuQuota != 0 {
if err = writeFile(path, "cpu.cfs_quota_us", strconv.FormatInt(c.Resources.CpuQuota, 10)); err != nil {
return err
}
}
if c.Resources.CpuPeriod != 0 {
if err = writeFile(path, "cpu.cfs_period_us", strconv.FormatInt(c.Resources.CpuPeriod, 10)); err != nil {
return err
}
}
if c.Resources.CpuRtPeriod != 0 {
if err = writeFile(path, "cpu.rt_period_us", strconv.FormatInt(c.Resources.CpuRtPeriod, 10)); err != nil {
return err
}
}
if c.Resources.CpuRtRuntime != 0 {
if err = writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(c.Resources.CpuRtRuntime, 10)); err != nil {
return err
}
}
return nil
}
func joinFreezer(c *configs.Cgroup, pid int) error {
path, err := join(c, "freezer", pid)
_, err := join(c, "freezer", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
freezer, err := subsystems.Get("freezer")
if err != nil {
return err
}
return freezer.Set(path, c)
return nil
}
func joinNetPrio(c *configs.Cgroup, pid int) error {
path, err := join(c, "net_prio", pid)
_, err := join(c, "net_prio", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
netPrio, err := subsystems.Get("net_prio")
if err != nil {
return err
}
return netPrio.Set(path, c)
return nil
}
func joinNetCls(c *configs.Cgroup, pid int) error {
path, err := join(c, "net_cls", pid)
_, err := join(c, "net_cls", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
netcls, err := subsystems.Get("net_cls")
if err != nil {
return nil
}
func joinPids(c *configs.Cgroup, pid int) error {
_, err := join(c, "pids", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
return netcls.Set(path, c)
return nil
}
// systemd represents slice heirarchy using `-`, so we need to follow suit when
// generating the path of slice. Essentially, test-a-b.slice becomes
// test.slice/test-a.slice/test-a-b.slice.
func expandSlice(slice string) (string, error) {
suffix := ".slice"
// Name has to end with ".slice", but can't be just ".slice".
if len(slice) < len(suffix) || !strings.HasSuffix(slice, suffix) {
return "", fmt.Errorf("invalid slice name: %s", slice)
}
// Path-separators are not allowed.
if strings.Contains(slice, "/") {
return "", fmt.Errorf("invalid slice name: %s", slice)
}
var path, prefix string
sliceName := strings.TrimSuffix(slice, suffix)
for _, component := range strings.Split(sliceName, "-") {
// test--a.slice isn't permitted, nor is -test.slice.
if component == "" {
return "", fmt.Errorf("invalid slice name: %s", slice)
}
// Append the component to the path and to the prefix.
path += prefix + component + suffix + "/"
prefix += component + "-"
}
return path, nil
}
func getSubsystemPath(c *configs.Cgroup, subsystem string) (string, error) {
@ -410,6 +434,11 @@ func getSubsystemPath(c *configs.Cgroup, subsystem string) (string, error) {
slice = c.Parent
}
slice, err = expandSlice(slice)
if err != nil {
return "", err
}
return filepath.Join(mountpoint, initPath, slice, getUnitName(c)), nil
}
@ -440,6 +469,14 @@ func (m *Manager) GetPids() ([]int, error) {
return cgroups.GetPids(path)
}
func (m *Manager) GetAllPids() ([]int, error) {
path, err := getSubsystemPath(m.Cgroups, "devices")
if err != nil {
return nil, err
}
return cgroups.GetAllPids(path)
}
func (m *Manager) GetStats() (*cgroups.Stats, error) {
m.mu.Lock()
defer m.mu.Unlock()
@ -458,16 +495,23 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
}
func (m *Manager) Set(container *configs.Config) error {
for name, path := range m.Paths {
sys, err := subsystems.Get(name)
if err == errSubsystemDoesNotExist || !cgroups.PathExists(path) {
continue
for _, sys := range subsystems {
// Get the subsystem path, but don't error out for not found cgroups.
path, err := getSubsystemPath(container.Cgroups, sys.Name())
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := sys.Set(path, container.Cgroups); err != nil {
return err
}
}
if m.Paths["cpu"] != "" {
if err := fs.CheckCpushares(m.Paths["cpu"], container.Cgroups.Resources.CpuShares); err != nil {
return err
}
}
return nil
}
@ -487,17 +531,13 @@ func getUnitName(c *configs.Cgroup) string {
// because systemd will re-write the device settings if it needs to re-apply the cgroup context.
// This happens at least for v208 when any sibling unit is started.
func joinDevices(c *configs.Cgroup, pid int) error {
path, err := join(c, "devices", pid)
_, err := join(c, "devices", pid)
// Even if it's `not found` error, we'll return err because devices cgroup
// is hard requirement for container security.
if err != nil {
return err
}
devices, err := subsystems.Get("devices")
if err != nil {
return err
}
return devices.Set(path, c)
return nil
}
func setKernelMemory(c *configs.Cgroup) error {
@ -510,52 +550,16 @@ func setKernelMemory(c *configs.Cgroup) error {
return err
}
if c.Resources.KernelMemory > 0 {
err = writeFile(path, "memory.kmem.limit_in_bytes", strconv.FormatInt(c.Resources.KernelMemory, 10))
if err != nil {
return err
}
}
return nil
// This doesn't get called by manager.Set, so we need to do it here.
s := &fs.MemoryGroup{}
return s.SetKernelMemory(path, c)
}
func joinMemory(c *configs.Cgroup, pid int) error {
path, err := getSubsystemPath(c, "memory")
_, err := join(c, "memory", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
// -1 disables memoryswap
if c.Resources.MemorySwap > 0 {
err = writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(c.Resources.MemorySwap, 10))
if err != nil {
return err
}
}
if c.Resources.MemoryReservation > 0 {
err = writeFile(path, "memory.soft_limit_in_bytes", strconv.FormatInt(c.Resources.MemoryReservation, 10))
if err != nil {
return err
}
}
if c.Resources.OomKillDisable {
if err := writeFile(path, "memory.oom_control", "1"); err != nil {
return err
}
}
if c.Resources.MemorySwappiness >= 0 && c.Resources.MemorySwappiness <= 100 {
err = writeFile(path, "memory.swappiness", strconv.FormatInt(c.Resources.MemorySwappiness, 10))
if err != nil {
return err
}
} else if c.Resources.MemorySwappiness == -1 {
return nil
} else {
return fmt.Errorf("invalid value:%d. valid memory swappiness range is 0-100", c.Resources.MemorySwappiness)
}
return nil
}
@ -577,68 +581,25 @@ func joinCpuset(c *configs.Cgroup, pid int) error {
// expects device path instead of major minor numbers, which is also confusing
// for users. So we use fs work around for now.
func joinBlkio(c *configs.Cgroup, pid int) error {
path, err := getSubsystemPath(c, "blkio")
_, err := join(c, "blkio", pid)
if err != nil {
return err
}
// systemd doesn't directly support this in the dbus properties
if c.Resources.BlkioLeafWeight != 0 {
if err := writeFile(path, "blkio.leaf_weight", strconv.FormatUint(uint64(c.Resources.BlkioLeafWeight), 10)); err != nil {
return err
}
}
for _, wd := range c.Resources.BlkioWeightDevice {
if err := writeFile(path, "blkio.weight_device", wd.WeightString()); err != nil {
return err
}
if err := writeFile(path, "blkio.leaf_weight_device", wd.LeafWeightString()); err != nil {
return err
}
}
for _, td := range c.Resources.BlkioThrottleReadBpsDevice {
if err := writeFile(path, "blkio.throttle.read_bps_device", td.String()); err != nil {
return err
}
}
for _, td := range c.Resources.BlkioThrottleWriteBpsDevice {
if err := writeFile(path, "blkio.throttle.write_bps_device", td.String()); err != nil {
return err
}
}
for _, td := range c.Resources.BlkioThrottleReadIOPSDevice {
if err := writeFile(path, "blkio.throttle.read_iops_device", td.String()); err != nil {
return err
}
}
for _, td := range c.Resources.BlkioThrottleWriteIOPSDevice {
if err := writeFile(path, "blkio.throttle.write_iops_device", td.String()); err != nil {
return err
}
}
return nil
}
func joinHugetlb(c *configs.Cgroup, pid int) error {
path, err := join(c, "hugetlb", pid)
_, err := join(c, "hugetlb", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
hugetlb, err := subsystems.Get("hugetlb")
if err != nil {
return err
}
return hugetlb.Set(path, c)
return nil
}
func joinPerfEvent(c *configs.Cgroup, pid int) error {
path, err := join(c, "perf_event", pid)
_, err := join(c, "perf_event", pid)
if err != nil && !cgroups.IsNotFound(err) {
return err
}
perfEvent, err := subsystems.Get("perf_event")
if err != nil {
return err
}
return perfEvent.Set(path, c)
return nil
}

View file

@ -5,6 +5,7 @@ package cgroups
import (
"bufio"
"fmt"
"io"
"io/ioutil"
"os"
"path/filepath"
@ -12,7 +13,6 @@ import (
"strings"
"time"
"github.com/docker/docker/pkg/mount"
"github.com/docker/go-units"
)
@ -84,10 +84,19 @@ func FindCgroupMountpointDir() (string, error) {
// Safe as mountinfo encodes mountpoints with spaces as \040.
index := strings.Index(text, " - ")
postSeparatorFields := strings.Fields(text[index+3:])
if len(postSeparatorFields) < 3 {
return "", fmt.Errorf("Error found less than 3 fields post '-' in %q", text)
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" {
// Check that the mount is properly formated.
if numPostFields < 3 {
return "", fmt.Errorf("Error found less than 3 fields post '-' in %q", text)
}
return filepath.Dir(fields[4]), nil
}
}
@ -112,11 +121,45 @@ func (m Mount) GetThisCgroupDir(cgroups map[string]string) (string, error) {
return getControllerPath(m.Subsystems[0], cgroups)
}
func getCgroupMountsHelper(ss map[string]bool, mi io.Reader) ([]Mount, error) {
res := make([]Mount, 0, len(ss))
scanner := bufio.NewScanner(mi)
for scanner.Scan() {
txt := scanner.Text()
sepIdx := strings.IndexByte(txt, '-')
if sepIdx == -1 {
return nil, fmt.Errorf("invalid mountinfo format")
}
if txt[sepIdx+2:sepIdx+8] != "cgroup" {
continue
}
fields := strings.Split(txt, " ")
m := Mount{
Mountpoint: fields[4],
Root: fields[3],
}
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
if strings.HasPrefix(opt, cgroupNamePrefix) {
m.Subsystems = append(m.Subsystems, opt[len(cgroupNamePrefix):])
}
if ss[opt] {
m.Subsystems = append(m.Subsystems, opt)
}
}
res = append(res, m)
}
if err := scanner.Err(); err != nil {
return nil, err
}
return res, nil
}
func GetCgroupMounts() ([]Mount, error) {
mounts, err := mount.GetMounts()
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return nil, err
}
defer f.Close()
all, err := GetAllSubsystems()
if err != nil {
@ -127,24 +170,7 @@ func GetCgroupMounts() ([]Mount, error) {
for _, s := range all {
allMap[s] = true
}
res := []Mount{}
for _, mount := range mounts {
if mount.Fstype == "cgroup" {
m := Mount{Mountpoint: mount.Mountpoint, Root: mount.Root}
for _, opt := range strings.Split(mount.VfsOpts, ",") {
if strings.HasPrefix(opt, cgroupNamePrefix) {
m.Subsystems = append(m.Subsystems, opt[len(cgroupNamePrefix):])
}
if allMap[opt] {
m.Subsystems = append(m.Subsystems, opt)
}
}
res = append(res, m)
}
}
return res, nil
return getCgroupMountsHelper(allMap, f)
}
// Returns all the cgroup subsystems supported by the kernel
@ -323,9 +349,14 @@ func GetHugePageSize() ([]string, error) {
return pageSizes, nil
}
// GetPids returns all pids, that were added to cgroup at path and to all its
// subcgroups.
// GetPids returns all pids, that were added to cgroup at path.
func GetPids(path string) ([]int, error) {
return readProcsFile(path)
}
// GetAllPids returns all pids, that were added to cgroup at path and to all its
// subcgroups.
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 {

View file

@ -0,0 +1,138 @@
package cgroups
import (
"bytes"
"strings"
"testing"
)
const fedoraMountinfo = `15 35 0:3 / /proc rw,nosuid,nodev,noexec,relatime shared:5 - proc proc rw
16 35 0:14 / /sys rw,nosuid,nodev,noexec,relatime shared:6 - sysfs sysfs rw,seclabel
17 35 0:5 / /dev rw,nosuid shared:2 - devtmpfs devtmpfs rw,seclabel,size=8056484k,nr_inodes=2014121,mode=755
18 16 0:15 / /sys/kernel/security rw,nosuid,nodev,noexec,relatime shared:7 - securityfs securityfs rw
19 16 0:13 / /sys/fs/selinux rw,relatime shared:8 - selinuxfs selinuxfs rw
20 17 0:16 / /dev/shm rw,nosuid,nodev shared:3 - tmpfs tmpfs rw,seclabel
21 17 0:10 / /dev/pts rw,nosuid,noexec,relatime shared:4 - devpts devpts rw,seclabel,gid=5,mode=620,ptmxmode=000
22 35 0:17 / /run rw,nosuid,nodev shared:21 - tmpfs tmpfs rw,seclabel,mode=755
23 16 0:18 / /sys/fs/cgroup rw,nosuid,nodev,noexec shared:9 - tmpfs tmpfs rw,seclabel,mode=755
24 23 0:19 / /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime shared:10 - cgroup cgroup rw,xattr,release_agent=/usr/lib/systemd/systemd-cgroups-agent,name=systemd
25 16 0:20 / /sys/fs/pstore rw,nosuid,nodev,noexec,relatime shared:20 - pstore pstore rw
26 23 0:21 / /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime shared:11 - cgroup cgroup rw,cpuset,clone_children
27 23 0:22 / /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime shared:12 - cgroup cgroup rw,cpuacct,cpu,clone_children
28 23 0:23 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:13 - cgroup cgroup rw,memory,clone_children
29 23 0:24 / /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime shared:14 - cgroup cgroup rw,devices,clone_children
30 23 0:25 / /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime shared:15 - cgroup cgroup rw,freezer,clone_children
31 23 0:26 / /sys/fs/cgroup/net_cls rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,net_cls,clone_children
32 23 0:27 / /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime shared:17 - cgroup cgroup rw,blkio,clone_children
33 23 0:28 / /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime shared:18 - cgroup cgroup rw,perf_event,clone_children
34 23 0:29 / /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime shared:19 - cgroup cgroup rw,hugetlb,clone_children
35 1 253:2 / / rw,relatime shared:1 - ext4 /dev/mapper/ssd-root--f20 rw,seclabel,data=ordered
36 15 0:30 / /proc/sys/fs/binfmt_misc rw,relatime shared:22 - autofs systemd-1 rw,fd=38,pgrp=1,timeout=300,minproto=5,maxproto=5,direct
37 17 0:12 / /dev/mqueue rw,relatime shared:23 - mqueue mqueue rw,seclabel
38 35 0:31 / /tmp rw shared:24 - tmpfs tmpfs rw,seclabel
39 17 0:32 / /dev/hugepages rw,relatime shared:25 - hugetlbfs hugetlbfs rw,seclabel
40 16 0:7 / /sys/kernel/debug rw,relatime shared:26 - debugfs debugfs rw
41 16 0:33 / /sys/kernel/config rw,relatime shared:27 - configfs configfs rw
42 35 0:34 / /var/lib/nfs/rpc_pipefs rw,relatime shared:28 - rpc_pipefs sunrpc rw
43 15 0:35 / /proc/fs/nfsd rw,relatime shared:29 - nfsd sunrpc rw
45 35 8:17 / /boot rw,relatime shared:30 - ext4 /dev/sdb1 rw,seclabel,data=ordered
46 35 253:4 / /home rw,relatime shared:31 - ext4 /dev/mapper/ssd-home rw,seclabel,data=ordered
47 35 253:5 / /var/lib/libvirt/images rw,noatime,nodiratime shared:32 - ext4 /dev/mapper/ssd-virt rw,seclabel,discard,data=ordered
48 35 253:12 / /mnt/old rw,relatime shared:33 - ext4 /dev/mapper/HelpDeskRHEL6-FedoraRoot rw,seclabel,data=ordered
121 22 0:36 / /run/user/1000/gvfs rw,nosuid,nodev,relatime shared:104 - fuse.gvfsd-fuse gvfsd-fuse rw,user_id=1000,group_id=1000
124 16 0:37 / /sys/fs/fuse/connections rw,relatime shared:107 - fusectl fusectl rw
165 38 253:3 / /tmp/mnt rw,relatime shared:147 - ext4 /dev/mapper/ssd-root rw,seclabel,data=ordered
167 35 253:15 / /var/lib/docker/devicemapper/mnt/aae4076022f0e2b80a2afbf8fc6df450c52080191fcef7fb679a73e6f073e5c2 rw,relatime shared:149 - ext4 /dev/mapper/docker-253:2-425882-aae4076022f0e2b80a2afbf8fc6df450c52080191fcef7fb679a73e6f073e5c2 rw,seclabel,discard,stripe=16,data=ordered
171 35 253:16 / /var/lib/docker/devicemapper/mnt/c71be651f114db95180e472f7871b74fa597ee70a58ccc35cb87139ddea15373 rw,relatime shared:153 - ext4 /dev/mapper/docker-253:2-425882-c71be651f114db95180e472f7871b74fa597ee70a58ccc35cb87139ddea15373 rw,seclabel,discard,stripe=16,data=ordered
175 35 253:17 / /var/lib/docker/devicemapper/mnt/1bac6ab72862d2d5626560df6197cf12036b82e258c53d981fa29adce6f06c3c rw,relatime shared:157 - ext4 /dev/mapper/docker-253:2-425882-1bac6ab72862d2d5626560df6197cf12036b82e258c53d981fa29adce6f06c3c rw,seclabel,discard,stripe=16,data=ordered
179 35 253:18 / /var/lib/docker/devicemapper/mnt/d710a357d77158e80d5b2c55710ae07c94e76d34d21ee7bae65ce5418f739b09 rw,relatime shared:161 - ext4 /dev/mapper/docker-253:2-425882-d710a357d77158e80d5b2c55710ae07c94e76d34d21ee7bae65ce5418f739b09 rw,seclabel,discard,stripe=16,data=ordered
183 35 253:19 / /var/lib/docker/devicemapper/mnt/6479f52366114d5f518db6837254baab48fab39f2ac38d5099250e9a6ceae6c7 rw,relatime shared:165 - ext4 /dev/mapper/docker-253:2-425882-6479f52366114d5f518db6837254baab48fab39f2ac38d5099250e9a6ceae6c7 rw,seclabel,discard,stripe=16,data=ordered
187 35 253:20 / /var/lib/docker/devicemapper/mnt/8d9df91c4cca5aef49eeb2725292aab324646f723a7feab56be34c2ad08268e1 rw,relatime shared:169 - ext4 /dev/mapper/docker-253:2-425882-8d9df91c4cca5aef49eeb2725292aab324646f723a7feab56be34c2ad08268e1 rw,seclabel,discard,stripe=16,data=ordered
191 35 253:21 / /var/lib/docker/devicemapper/mnt/c8240b768603d32e920d365dc9d1dc2a6af46cd23e7ae819947f969e1b4ec661 rw,relatime shared:173 - ext4 /dev/mapper/docker-253:2-425882-c8240b768603d32e920d365dc9d1dc2a6af46cd23e7ae819947f969e1b4ec661 rw,seclabel,discard,stripe=16,data=ordered
195 35 253:22 / /var/lib/docker/devicemapper/mnt/2eb3a01278380bbf3ed12d86ac629eaa70a4351301ee307a5cabe7b5f3b1615f rw,relatime shared:177 - ext4 /dev/mapper/docker-253:2-425882-2eb3a01278380bbf3ed12d86ac629eaa70a4351301ee307a5cabe7b5f3b1615f rw,seclabel,discard,stripe=16,data=ordered
199 35 253:23 / /var/lib/docker/devicemapper/mnt/37a17fb7c9d9b80821235d5f2662879bd3483915f245f9b49cdaa0e38779b70b rw,relatime shared:181 - ext4 /dev/mapper/docker-253:2-425882-37a17fb7c9d9b80821235d5f2662879bd3483915f245f9b49cdaa0e38779b70b rw,seclabel,discard,stripe=16,data=ordered
203 35 253:24 / /var/lib/docker/devicemapper/mnt/aea459ae930bf1de913e2f29428fd80ee678a1e962d4080019d9f9774331ee2b rw,relatime shared:185 - ext4 /dev/mapper/docker-253:2-425882-aea459ae930bf1de913e2f29428fd80ee678a1e962d4080019d9f9774331ee2b rw,seclabel,discard,stripe=16,data=ordered
207 35 253:25 / /var/lib/docker/devicemapper/mnt/928ead0bc06c454bd9f269e8585aeae0a6bd697f46dc8754c2a91309bc810882 rw,relatime shared:189 - ext4 /dev/mapper/docker-253:2-425882-928ead0bc06c454bd9f269e8585aeae0a6bd697f46dc8754c2a91309bc810882 rw,seclabel,discard,stripe=16,data=ordered
211 35 253:26 / /var/lib/docker/devicemapper/mnt/0f284d18481d671644706e7a7244cbcf63d590d634cc882cb8721821929d0420 rw,relatime shared:193 - ext4 /dev/mapper/docker-253:2-425882-0f284d18481d671644706e7a7244cbcf63d590d634cc882cb8721821929d0420 rw,seclabel,discard,stripe=16,data=ordered
215 35 253:27 / /var/lib/docker/devicemapper/mnt/d9dd16722ab34c38db2733e23f69e8f4803ce59658250dd63e98adff95d04919 rw,relatime shared:197 - ext4 /dev/mapper/docker-253:2-425882-d9dd16722ab34c38db2733e23f69e8f4803ce59658250dd63e98adff95d04919 rw,seclabel,discard,stripe=16,data=ordered
219 35 253:28 / /var/lib/docker/devicemapper/mnt/bc4500479f18c2c08c21ad5282e5f826a016a386177d9874c2764751c031d634 rw,relatime shared:201 - ext4 /dev/mapper/docker-253:2-425882-bc4500479f18c2c08c21ad5282e5f826a016a386177d9874c2764751c031d634 rw,seclabel,discard,stripe=16,data=ordered
223 35 253:29 / /var/lib/docker/devicemapper/mnt/7770c8b24eb3d5cc159a065910076938910d307ab2f5d94e1dc3b24c06ee2c8a rw,relatime shared:205 - ext4 /dev/mapper/docker-253:2-425882-7770c8b24eb3d5cc159a065910076938910d307ab2f5d94e1dc3b24c06ee2c8a rw,seclabel,discard,stripe=16,data=ordered
227 35 253:30 / /var/lib/docker/devicemapper/mnt/c280cd3d0bf0aa36b478b292279671624cceafc1a67eaa920fa1082601297adf rw,relatime shared:209 - ext4 /dev/mapper/docker-253:2-425882-c280cd3d0bf0aa36b478b292279671624cceafc1a67eaa920fa1082601297adf rw,seclabel,discard,stripe=16,data=ordered
231 35 253:31 / /var/lib/docker/devicemapper/mnt/8b59a7d9340279f09fea67fd6ad89ddef711e9e7050eb647984f8b5ef006335f rw,relatime shared:213 - ext4 /dev/mapper/docker-253:2-425882-8b59a7d9340279f09fea67fd6ad89ddef711e9e7050eb647984f8b5ef006335f rw,seclabel,discard,stripe=16,data=ordered
235 35 253:32 / /var/lib/docker/devicemapper/mnt/1a28059f29eda821578b1bb27a60cc71f76f846a551abefabce6efd0146dce9f rw,relatime shared:217 - ext4 /dev/mapper/docker-253:2-425882-1a28059f29eda821578b1bb27a60cc71f76f846a551abefabce6efd0146dce9f rw,seclabel,discard,stripe=16,data=ordered
239 35 253:33 / /var/lib/docker/devicemapper/mnt/e9aa60c60128cad1 rw,relatime shared:221 - ext4 /dev/mapper/docker-253:2-425882-e9aa60c60128cad1 rw,seclabel,discard,stripe=16,data=ordered
243 35 253:34 / /var/lib/docker/devicemapper/mnt/5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d-init rw,relatime shared:225 - ext4 /dev/mapper/docker-253:2-425882-5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d-init rw,seclabel,discard,stripe=16,data=ordered
247 35 253:35 / /var/lib/docker/devicemapper/mnt/5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d rw,relatime shared:229 - ext4 /dev/mapper/docker-253:2-425882-5fec11304b6f4713fea7b6ccdcc1adc0a1966187f590fe25a8227428a8df275d rw,seclabel,discard,stripe=16,data=ordered
31 21 0:23 / /DATA/foo_bla_bla rw,relatime - cifs //foo/BLA\040BLA\040BLA/ rw,sec=ntlm,cache=loose,unc=\\foo\BLA BLA BLA,username=my_login,domain=mydomain.com,uid=12345678,forceuid,gid=12345678,forcegid,addr=10.1.30.10,file_mode=0755,dir_mode=0755,nounix,rsize=61440,wsize=65536,actimeo=1`
func TestGetCgroupMounts(t *testing.T) {
subsystems := map[string]bool{
"cpuset": true,
"cpu": true,
"cpuacct": true,
"memory": true,
"devices": true,
"freezer": true,
"net_cls": true,
"blkio": true,
"perf_event": true,
"hugetlb": true,
}
mi := bytes.NewBufferString(fedoraMountinfo)
cgMounts, err := getCgroupMountsHelper(subsystems, mi)
if err != nil {
t.Fatal(err)
}
cgMap := make(map[string]Mount)
for _, m := range cgMounts {
for _, ss := range m.Subsystems {
cgMap[ss] = m
}
}
for ss := range subsystems {
m, ok := cgMap[ss]
if !ok {
t.Fatalf("%s not found", ss)
}
if m.Root != "/" {
t.Fatalf("unexpected root for %s: %s", ss, m.Root)
}
if !strings.HasPrefix(m.Mountpoint, "/sys/fs/cgroup/") && !strings.Contains(m.Mountpoint, ss) {
t.Fatalf("unexpected mountpoint for %s: %s", ss, m.Mountpoint)
}
var ssFound bool
for _, mss := range m.Subsystems {
if mss == ss {
ssFound = true
break
}
}
if !ssFound {
t.Fatalf("subsystem %s not found in Subsystems field %v", ss, m.Subsystems)
}
}
}
func BenchmarkGetCgroupMounts(b *testing.B) {
subsystems := map[string]bool{
"cpuset": true,
"cpu": true,
"cpuacct": true,
"memory": true,
"devices": true,
"freezer": true,
"net_cls": true,
"blkio": true,
"perf_event": true,
"hugetlb": true,
}
b.ResetTimer()
for i := 0; i < b.N; i++ {
b.StopTimer()
mi := bytes.NewBufferString(fedoraMountinfo)
b.StartTimer()
if _, err := getCgroupMountsHelper(subsystems, mi); err != nil {
b.Fatal(err)
}
}
}

View file

@ -11,25 +11,38 @@ const (
)
type Cgroup struct {
Name string `json:"name"`
// Deprecated, use Path instead
Name string `json:"name,omitempty"`
// name of parent cgroup or slice
Parent string `json:"parent"`
// 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 decribes 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 *Resources `json:"resources"`
*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.
AllowAllDevices bool `json:"allow_all_devices"`
// Deprecated
AllowAllDevices bool `json:"allow_all_devices,omitempty"`
// Deprecated
AllowedDevices []*Device `json:"allowed_devices,omitempty"`
// Deprecated
DeniedDevices []*Device `json:"denied_devices,omitempty"`
AllowedDevices []*Device `json:"allowed_devices"`
DeniedDevices []*Device `json:"denied_devices"`
Devices []*Device `json:"devices"`
// Memory limit (in bytes)
Memory int64 `json:"memory"`
@ -37,7 +50,7 @@ type Resources struct {
// Memory reservation or soft_limit (in bytes)
MemoryReservation int64 `json:"memory_reservation"`
// Total memory usage (memory + swap); set `-1' to disable swap
// Total memory usage (memory + swap); set `-1` to enable unlimited swap
MemorySwap int64 `json:"memory_swap"`
// Kernel memory limit (in bytes)
@ -64,6 +77,9 @@ type Resources struct {
// 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"`

View file

@ -0,0 +1,156 @@
// +build linux freebsd
package configs
import (
"encoding/json"
"fmt"
"os"
"path/filepath"
"testing"
)
// Checks whether the expected capability is specified in the capabilities.
func contains(expected string, values []string) bool {
for _, v := range values {
if v == expected {
return true
}
}
return false
}
func containsDevice(expected *Device, values []*Device) bool {
for _, d := range values {
if d.Path == expected.Path &&
d.Permissions == expected.Permissions &&
d.FileMode == expected.FileMode &&
d.Major == expected.Major &&
d.Minor == expected.Minor &&
d.Type == expected.Type {
return true
}
}
return false
}
func loadConfig(name string) (*Config, error) {
f, err := os.Open(filepath.Join("../sample_configs", name))
if err != nil {
return nil, err
}
defer f.Close()
var container *Config
if err := json.NewDecoder(f).Decode(&container); err != nil {
return nil, err
}
// Check that a config doesn't contain extra fields
var configMap, abstractMap map[string]interface{}
if _, err := f.Seek(0, 0); err != nil {
return nil, err
}
if err := json.NewDecoder(f).Decode(&abstractMap); err != nil {
return nil, err
}
configData, err := json.Marshal(&container)
if err != nil {
return nil, err
}
if err := json.Unmarshal(configData, &configMap); err != nil {
return nil, err
}
for k := range configMap {
delete(abstractMap, k)
}
if len(abstractMap) != 0 {
return nil, fmt.Errorf("unknown fields: %s", abstractMap)
}
return container, nil
}
func TestRemoveNamespace(t *testing.T) {
ns := Namespaces{
{Type: NEWNET},
}
if !ns.Remove(NEWNET) {
t.Fatal("NEWNET was not removed")
}
if len(ns) != 0 {
t.Fatalf("namespaces should have 0 items but reports %d", len(ns))
}
}
func TestHostUIDNoUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{},
}
uid, err := config.HostUID()
if err != nil {
t.Fatal(err)
}
if uid != 0 {
t.Fatalf("expected uid 0 with no USERNS but received %d", uid)
}
}
func TestHostUIDWithUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{{Type: NEWUSER}},
UidMappings: []IDMap{
{
ContainerID: 0,
HostID: 1000,
Size: 1,
},
},
}
uid, err := config.HostUID()
if err != nil {
t.Fatal(err)
}
if uid != 1000 {
t.Fatalf("expected uid 1000 with no USERNS but received %d", uid)
}
}
func TestHostGIDNoUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{},
}
uid, err := config.HostGID()
if err != nil {
t.Fatal(err)
}
if uid != 0 {
t.Fatalf("expected gid 0 with no USERNS but received %d", uid)
}
}
func TestHostGIDWithUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{{Type: NEWUSER}},
GidMappings: []IDMap{
{
ContainerID: 0,
HostID: 1000,
Size: 1,
},
},
}
uid, err := config.HostGID()
if err != nil {
t.Fatal(err)
}
if uid != 1000 {
t.Fatalf("expected gid 1000 with no USERNS but received %d", uid)
}
}

View file

@ -0,0 +1,3 @@
package configs
// All current tests are for Unix-specific functionality

View file

@ -35,6 +35,9 @@ type Device struct {
// Gid of the device.
Gid uint32 `json:"gid"`
// Write the file to the allowed list
Allow bool `json:"allow"`
}
func (d *Device) CgroupString() string {

View file

@ -82,20 +82,6 @@ var (
Minor: 1,
Permissions: "rwm",
},
{
Path: "/dev/tty0",
Type: 'c',
Major: 4,
Minor: 0,
Permissions: "rwm",
},
{
Path: "/dev/tty1",
Type: 'c',
Major: 4,
Minor: 1,
Permissions: "rwm",
},
// /dev/pts/ - pts namespaces are "coming soon"
{
Path: "",

View file

@ -6,6 +6,7 @@ package libcontainer
import (
"os"
"time"
"github.com/opencontainers/runc/libcontainer/configs"
)
@ -14,8 +15,11 @@ import (
type Status int
const (
// The container exists but has not been run yet
Created Status = iota
// The container exists and is running.
Running Status = iota + 1
Running
// The container exists, it is in the process of being paused.
Pausing
@ -32,6 +36,8 @@ const (
func (s Status) String() string {
switch s {
case Created:
return "created"
case Running:
return "running"
case Pausing:
@ -43,7 +49,7 @@ func (s Status) String() string {
case Destroyed:
return "destroyed"
default:
return "undefined"
return "unknown"
}
}
@ -56,9 +62,12 @@ type BaseState struct {
// InitProcessPid is the init process id in the parent namespace.
InitProcessPid int `json:"init_process_pid"`
// InitProcessStartTime is the init process start time.
// InitProcessStartTime is the init process start time in clock cycles since boot time.
InitProcessStartTime string `json:"init_process_start"`
// Created is the unix timestamp for the creation time of the container in UTC
Created time.Time `json:"created"`
// Config is the container's configuration.
Config configs.Config `json:"config"`
}

View file

@ -15,12 +15,14 @@ import (
"strings"
"sync"
"syscall"
"time"
"github.com/Sirupsen/logrus"
"github.com/golang/protobuf/proto"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/criurpc"
"github.com/opencontainers/runc/libcontainer/utils"
"github.com/vishvananda/netlink/nl"
)
@ -38,6 +40,7 @@ type linuxContainer struct {
m sync.Mutex
criuVersion int
state containerState
created time.Time
}
// State represents a running container's state
@ -104,6 +107,12 @@ type Container interface {
// errors:
// Systemerror - System error.
NotifyOOM() (<-chan struct{}, error)
// NotifyMemoryPressure returns a read-only channel signaling when the container reaches a given pressure level
//
// errors:
// Systemerror - System error.
NotifyMemoryPressure(level PressureLevel) (<-chan struct{}, error)
}
// ID returns the container's unique ID
@ -129,7 +138,7 @@ func (c *linuxContainer) State() (*State, error) {
}
func (c *linuxContainer) Processes() ([]int, error) {
pids, err := c.cgroupManager.GetPids()
pids, err := c.cgroupManager.GetAllPids()
if err != nil {
return nil, newSystemError(err)
}
@ -183,29 +192,30 @@ func (c *linuxContainer) Start(process *Process) error {
}
return newSystemError(err)
}
// generate a timestamp indicating when the container was started
c.created = time.Now().UTC()
c.state = &runningState{
c: c,
}
if doInit {
if err := c.updateState(parent); err != nil {
return err
}
} else {
c.state.transition(&nullState{
c: c,
s: Running,
})
}
if c.config.Hooks != nil {
s := configs.HookState{
Version: c.config.Version,
ID: c.id,
Pid: parent.pid(),
Root: c.config.Rootfs,
}
for _, hook := range c.config.Hooks.Poststart {
if err := hook.Run(s); err != nil {
if err := parent.terminate(); err != nil {
logrus.Warn(err)
if c.config.Hooks != nil {
s := configs.HookState{
Version: c.config.Version,
ID: c.id,
Pid: parent.pid(),
Root: c.config.Rootfs,
}
for _, hook := range c.config.Hooks.Poststart {
if err := hook.Run(s); err != nil {
if err := parent.terminate(); err != nil {
logrus.Warn(err)
}
return newSystemError(err)
}
return newSystemError(err)
}
}
}
@ -258,7 +268,7 @@ func (c *linuxContainer) commandTemplate(p *Process, childPipe *os.File) (*exec.
}
func (c *linuxContainer) newInitProcess(p *Process, cmd *exec.Cmd, parentPipe, childPipe *os.File) (*initProcess, error) {
t := "_LIBCONTAINER_INITTYPE=standard"
t := "_LIBCONTAINER_INITTYPE=" + string(initStandard)
cloneFlags := c.config.Namespaces.CloneFlags()
if cloneFlags&syscall.CLONE_NEWUSER != 0 {
if err := c.addUidGidMappings(cmd.SysProcAttr); err != nil {
@ -285,7 +295,7 @@ func (c *linuxContainer) newInitProcess(p *Process, cmd *exec.Cmd, parentPipe, c
}
func (c *linuxContainer) newSetnsProcess(p *Process, cmd *exec.Cmd, parentPipe, childPipe *os.File) (*setnsProcess, error) {
cmd.Env = append(cmd.Env, "_LIBCONTAINER_INITTYPE=setns")
cmd.Env = append(cmd.Env, "_LIBCONTAINER_INITTYPE="+string(initSetns))
// for setns process, we dont have to set cloneflags as the process namespaces
// will only be set via setns syscall
data, err := c.bootstrapData(0, c.initProcess.pid(), p.consolePath)
@ -334,6 +344,13 @@ func (c *linuxContainer) Destroy() error {
func (c *linuxContainer) Pause() error {
c.m.Lock()
defer c.m.Unlock()
status, err := c.currentStatus()
if err != nil {
return err
}
if status != Running {
return newGenericError(fmt.Errorf("container not running"), ContainerNotRunning)
}
if err := c.cgroupManager.Freeze(configs.Frozen); err != nil {
return err
}
@ -345,6 +362,13 @@ func (c *linuxContainer) Pause() error {
func (c *linuxContainer) Resume() error {
c.m.Lock()
defer c.m.Unlock()
status, err := c.currentStatus()
if err != nil {
return err
}
if status != Paused {
return newGenericError(fmt.Errorf("container not paused"), ContainerNotPaused)
}
if err := c.cgroupManager.Freeze(configs.Thawed); err != nil {
return err
}
@ -357,6 +381,10 @@ func (c *linuxContainer) NotifyOOM() (<-chan struct{}, error) {
return notifyOnOOM(c.cgroupManager.GetPaths())
}
func (c *linuxContainer) NotifyMemoryPressure(level PressureLevel) (<-chan struct{}, error) {
return notifyMemoryPressure(c.cgroupManager.GetPaths(), level)
}
// XXX debug support, remove when debugging done.
func addArgsFromEnv(evar string, args *[]string) {
if e := os.Getenv(evar); e != "" {
@ -929,9 +957,6 @@ func (c *linuxContainer) criuNotifications(resp *criurpc.CriuResp, process *Proc
func (c *linuxContainer) updateState(process parentProcess) error {
c.initProcess = process
if err := c.refreshState(); err != nil {
return err
}
state, err := c.currentState()
if err != nil {
return err
@ -945,7 +970,7 @@ func (c *linuxContainer) saveState(s *State) error {
return err
}
defer f.Close()
return json.NewEncoder(f).Encode(s)
return utils.WriteJSON(f, s)
}
func (c *linuxContainer) deleteState() error {
@ -1007,35 +1032,37 @@ func (c *linuxContainer) isPaused() (bool, error) {
}
func (c *linuxContainer) currentState() (*State, error) {
status, err := c.currentStatus()
if err != nil {
return nil, err
}
if status == Destroyed {
return nil, newGenericError(fmt.Errorf("container destroyed"), ContainerNotExists)
}
startTime, err := c.initProcess.startTime()
if err != nil {
return nil, newSystemError(err)
var (
startTime string
externalDescriptors []string
pid = -1
)
if c.initProcess != nil {
pid = c.initProcess.pid()
startTime, _ = c.initProcess.startTime()
externalDescriptors = c.initProcess.externalDescriptors()
}
state := &State{
BaseState: BaseState{
ID: c.ID(),
Config: *c.config,
InitProcessPid: c.initProcess.pid(),
InitProcessPid: pid,
InitProcessStartTime: startTime,
Created: c.created,
},
CgroupPaths: c.cgroupManager.GetPaths(),
NamespacePaths: make(map[configs.NamespaceType]string),
ExternalDescriptors: c.initProcess.externalDescriptors(),
ExternalDescriptors: externalDescriptors,
}
for _, ns := range c.config.Namespaces {
state.NamespacePaths[ns.Type] = ns.GetPath(c.initProcess.pid())
}
for _, nsType := range configs.NamespaceTypes() {
if _, ok := state.NamespacePaths[nsType]; !ok {
ns := configs.Namespace{Type: nsType}
state.NamespacePaths[ns.Type] = ns.GetPath(c.initProcess.pid())
if pid > 0 {
for _, ns := range c.config.Namespaces {
state.NamespacePaths[ns.Type] = ns.GetPath(pid)
}
for _, nsType := range configs.NamespaceTypes() {
if _, ok := state.NamespacePaths[nsType]; !ok {
ns := configs.Namespace{Type: nsType}
state.NamespacePaths[ns.Type] = ns.GetPath(pid)
}
}
}
return state, nil

View file

@ -0,0 +1,218 @@
// +build linux
package libcontainer
import (
"fmt"
"os"
"testing"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
)
type mockCgroupManager struct {
pids []int
allPids []int
stats *cgroups.Stats
paths map[string]string
}
func (m *mockCgroupManager) GetPids() ([]int, error) {
return m.pids, nil
}
func (m *mockCgroupManager) GetAllPids() ([]int, error) {
return m.allPids, nil
}
func (m *mockCgroupManager) GetStats() (*cgroups.Stats, error) {
return m.stats, nil
}
func (m *mockCgroupManager) Apply(pid int) error {
return nil
}
func (m *mockCgroupManager) Set(container *configs.Config) error {
return nil
}
func (m *mockCgroupManager) Destroy() error {
return nil
}
func (m *mockCgroupManager) GetPaths() map[string]string {
return m.paths
}
func (m *mockCgroupManager) Freeze(state configs.FreezerState) error {
return nil
}
type mockProcess struct {
_pid int
started string
}
func (m *mockProcess) terminate() error {
return nil
}
func (m *mockProcess) pid() int {
return m._pid
}
func (m *mockProcess) startTime() (string, error) {
return m.started, nil
}
func (m *mockProcess) start() error {
return nil
}
func (m *mockProcess) wait() (*os.ProcessState, error) {
return nil, nil
}
func (m *mockProcess) signal(_ os.Signal) error {
return nil
}
func (p *mockProcess) externalDescriptors() []string {
return []string{}
}
func (p *mockProcess) setExternalDescriptors(newFds []string) {
}
func TestGetContainerPids(t *testing.T) {
container := &linuxContainer{
id: "myid",
config: &configs.Config{},
cgroupManager: &mockCgroupManager{allPids: []int{1, 2, 3}},
}
pids, err := container.Processes()
if err != nil {
t.Fatal(err)
}
for i, expected := range []int{1, 2, 3} {
if pids[i] != expected {
t.Fatalf("expected pid %d but received %d", expected, pids[i])
}
}
}
func TestGetContainerStats(t *testing.T) {
container := &linuxContainer{
id: "myid",
config: &configs.Config{},
cgroupManager: &mockCgroupManager{
pids: []int{1, 2, 3},
stats: &cgroups.Stats{
MemoryStats: cgroups.MemoryStats{
Usage: cgroups.MemoryData{
Usage: 1024,
},
},
},
},
}
stats, err := container.Stats()
if err != nil {
t.Fatal(err)
}
if stats.CgroupStats == nil {
t.Fatal("cgroup stats are nil")
}
if stats.CgroupStats.MemoryStats.Usage.Usage != 1024 {
t.Fatalf("expected memory usage 1024 but recevied %d", stats.CgroupStats.MemoryStats.Usage.Usage)
}
}
func TestGetContainerState(t *testing.T) {
var (
pid = os.Getpid()
expectedMemoryPath = "/sys/fs/cgroup/memory/myid"
expectedNetworkPath = "/networks/fd"
)
container := &linuxContainer{
id: "myid",
config: &configs.Config{
Namespaces: []configs.Namespace{
{Type: configs.NEWPID},
{Type: configs.NEWNS},
{Type: configs.NEWNET, Path: expectedNetworkPath},
{Type: configs.NEWUTS},
// emulate host for IPC
//{Type: configs.NEWIPC},
},
},
initProcess: &mockProcess{
_pid: pid,
started: "010",
},
cgroupManager: &mockCgroupManager{
pids: []int{1, 2, 3},
stats: &cgroups.Stats{
MemoryStats: cgroups.MemoryStats{
Usage: cgroups.MemoryData{
Usage: 1024,
},
},
},
paths: map[string]string{
"memory": expectedMemoryPath,
},
},
}
container.state = &createdState{c: container}
state, err := container.State()
if err != nil {
t.Fatal(err)
}
if state.InitProcessPid != pid {
t.Fatalf("expected pid %d but received %d", pid, state.InitProcessPid)
}
if state.InitProcessStartTime != "010" {
t.Fatalf("expected process start time 010 but received %s", state.InitProcessStartTime)
}
paths := state.CgroupPaths
if paths == nil {
t.Fatal("cgroup paths should not be nil")
}
if memPath := paths["memory"]; memPath != expectedMemoryPath {
t.Fatalf("expected memory path %q but received %q", expectedMemoryPath, memPath)
}
for _, ns := range container.config.Namespaces {
path := state.NamespacePaths[ns.Type]
if path == "" {
t.Fatalf("expected non nil namespace path for %s", ns.Type)
}
if ns.Type == configs.NEWNET {
if path != expectedNetworkPath {
t.Fatalf("expected path %q but received %q", expectedNetworkPath, path)
}
} else {
file := ""
switch ns.Type {
case configs.NEWNET:
file = "net"
case configs.NEWNS:
file = "mnt"
case configs.NEWPID:
file = "pid"
case configs.NEWIPC:
file = "ipc"
case configs.NEWUSER:
file = "user"
case configs.NEWUTS:
file = "uts"
}
expected := fmt.Sprintf("/proc/%d/ns/%s", pid, file)
if expected != path {
t.Fatalf("expected path %q but received %q", expected, path)
}
}
}
}

View file

@ -0,0 +1,63 @@
// +build linux freebsd
package devices
import (
"errors"
"os"
"testing"
)
func TestDeviceFromPathLstatFailure(t *testing.T) {
testError := errors.New("test error")
// Override os.Lstat to inject error.
osLstat = func(path string) (os.FileInfo, error) {
return nil, testError
}
_, err := DeviceFromPath("", "")
if err != testError {
t.Fatalf("Unexpected error %v, expected %v", err, testError)
}
}
func TestHostDevicesIoutilReadDirFailure(t *testing.T) {
testError := errors.New("test error")
// Override ioutil.ReadDir to inject error.
ioutilReadDir = func(dirname string) ([]os.FileInfo, error) {
return nil, testError
}
_, err := HostDevices()
if err != testError {
t.Fatalf("Unexpected error %v, expected %v", err, testError)
}
}
func TestHostDevicesIoutilReadDirDeepFailure(t *testing.T) {
testError := errors.New("test error")
called := false
// Override ioutil.ReadDir to inject error after the first call.
ioutilReadDir = func(dirname string) ([]os.FileInfo, error) {
if called {
return nil, testError
}
called = true
// Provoke a second call.
fi, err := os.Lstat("/tmp")
if err != nil {
t.Fatalf("Unexpected error %v", err)
}
return []os.FileInfo{fi}, nil
}
_, err := HostDevices()
if err != testError {
t.Fatalf("Unexpected error %v, expected %v", err, testError)
}
}

View file

@ -0,0 +1,102 @@
// +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 it's 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() {
case "pts", "shm", "fd", "mqueue":
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
}
return nil, err
}
out = append(out, device)
}
return out, nil
}

View file

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

View file

@ -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))
}

View file

@ -16,9 +16,10 @@ const (
ContainerPaused
ContainerNotStopped
ContainerNotRunning
ContainerNotPaused
// Process errors
ProcessNotExecuted
NoProcessOps
// Common errors
ConfigInvalid
@ -46,6 +47,10 @@ func (c ErrorCode) String() string {
return "Container is not running"
case ConsoleExists:
return "Console exists for process"
case ContainerNotPaused:
return "Container is not paused"
case NoProcessOps:
return "No process operations"
default:
return "Unknown error"
}

View file

@ -0,0 +1,20 @@
package libcontainer
import "testing"
func TestErrorCode(t *testing.T) {
codes := map[ErrorCode]string{
IdInUse: "Id already in use",
InvalidIdFormat: "Invalid format",
ContainerPaused: "Container paused",
ConfigInvalid: "Invalid configuration",
SystemError: "System error",
ContainerNotExists: "Container does not exist",
}
for code, expected := range codes {
if actual := code.String(); actual != expected {
t.Fatalf("expected string %q but received %q", expected, actual)
}
}
}

View file

@ -5,7 +5,6 @@ package libcontainer
import (
"encoding/json"
"fmt"
"io/ioutil"
"os"
"os/exec"
"path/filepath"
@ -19,6 +18,7 @@ import (
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/configs/validate"
"github.com/opencontainers/runc/libcontainer/utils"
)
const (
@ -202,8 +202,12 @@ func (l *LinuxFactory) Load(id string) (Container, error) {
criuPath: l.CriuPath,
cgroupManager: l.NewCgroupsManager(state.Config.Cgroups, state.CgroupPaths),
root: containerRoot,
created: state.Created,
}
c.state = &createdState{c: c, s: Created}
if err := c.refreshState(); err != nil {
return nil, err
}
c.state = &nullState{c: c}
return c, nil
}
@ -226,21 +230,29 @@ func (l *LinuxFactory) StartInitialization() (err error) {
// clear the current process's environment to clean any libcontainer
// specific env vars.
os.Clearenv()
var i initer
defer func() {
// if we have an error during the initialization of the container's init then send it back to the
// parent process in the form of an initError.
if err != nil {
// ensure that any data sent from the parent is consumed so it doesn't
// receive ECONNRESET when the child writes to the pipe.
ioutil.ReadAll(pipe)
if err := json.NewEncoder(pipe).Encode(newSystemError(err)); err != nil {
if _, ok := i.(*linuxStandardInit); ok {
// Synchronisation only necessary for standard init.
if err := utils.WriteJSON(pipe, syncT{procError}); err != nil {
panic(err)
}
}
if err := utils.WriteJSON(pipe, newSystemError(err)); err != nil {
panic(err)
}
} else {
if err := utils.WriteJSON(pipe, syncT{procStart}); err != nil {
panic(err)
}
}
// ensure that this pipe is always closed
pipe.Close()
}()
i, err := newContainerInit(it, pipe)
i, err = newContainerInit(it, pipe)
if err != nil {
return err
}

View file

@ -0,0 +1,183 @@
// +build linux
package libcontainer
import (
"io/ioutil"
"os"
"path/filepath"
"syscall"
"testing"
"github.com/docker/docker/pkg/mount"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/utils"
)
func newTestRoot() (string, error) {
dir, err := ioutil.TempDir("", "libcontainer")
if err != nil {
return "", err
}
return dir, nil
}
func TestFactoryNew(t *testing.T) {
root, rerr := newTestRoot()
if rerr != nil {
t.Fatal(rerr)
}
defer os.RemoveAll(root)
factory, err := New(root, Cgroupfs)
if err != nil {
t.Fatal(err)
}
if factory == nil {
t.Fatal("factory should not be nil")
}
lfactory, ok := factory.(*LinuxFactory)
if !ok {
t.Fatal("expected linux factory returned on linux based systems")
}
if lfactory.Root != root {
t.Fatalf("expected factory root to be %q but received %q", root, lfactory.Root)
}
if factory.Type() != "libcontainer" {
t.Fatalf("unexpected factory type: %q, expected %q", factory.Type(), "libcontainer")
}
}
func TestFactoryNewTmpfs(t *testing.T) {
root, rerr := newTestRoot()
if rerr != nil {
t.Fatal(rerr)
}
defer os.RemoveAll(root)
factory, err := New(root, Cgroupfs, TmpfsRoot)
if err != nil {
t.Fatal(err)
}
if factory == nil {
t.Fatal("factory should not be nil")
}
lfactory, ok := factory.(*LinuxFactory)
if !ok {
t.Fatal("expected linux factory returned on linux based systems")
}
if lfactory.Root != root {
t.Fatalf("expected factory root to be %q but received %q", root, lfactory.Root)
}
if factory.Type() != "libcontainer" {
t.Fatalf("unexpected factory type: %q, expected %q", factory.Type(), "libcontainer")
}
mounted, err := mount.Mounted(lfactory.Root)
if err != nil {
t.Fatal(err)
}
if !mounted {
t.Fatalf("Factory Root is not mounted")
}
mounts, err := mount.GetMounts()
if err != nil {
t.Fatal(err)
}
var found bool
for _, m := range mounts {
if m.Mountpoint == lfactory.Root {
if m.Fstype != "tmpfs" {
t.Fatalf("Fstype of root: %s, expected %s", m.Fstype, "tmpfs")
}
if m.Source != "tmpfs" {
t.Fatalf("Source of root: %s, expected %s", m.Source, "tmpfs")
}
found = true
}
}
if !found {
t.Fatalf("Factory Root is not listed in mounts list")
}
defer syscall.Unmount(root, syscall.MNT_DETACH)
}
func TestFactoryLoadNotExists(t *testing.T) {
root, rerr := newTestRoot()
if rerr != nil {
t.Fatal(rerr)
}
defer os.RemoveAll(root)
factory, err := New(root, Cgroupfs)
if err != nil {
t.Fatal(err)
}
_, err = factory.Load("nocontainer")
if err == nil {
t.Fatal("expected nil error loading non-existing container")
}
lerr, ok := err.(Error)
if !ok {
t.Fatal("expected libcontainer error type")
}
if lerr.Code() != ContainerNotExists {
t.Fatalf("expected error code %s but received %s", ContainerNotExists, lerr.Code())
}
}
func TestFactoryLoadContainer(t *testing.T) {
root, err := newTestRoot()
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(root)
// setup default container config and state for mocking
var (
id = "1"
expectedConfig = &configs.Config{
Rootfs: "/mycontainer/root",
}
expectedState = &State{
BaseState: BaseState{
InitProcessPid: 1024,
Config: *expectedConfig,
},
}
)
if err := os.Mkdir(filepath.Join(root, id), 0700); err != nil {
t.Fatal(err)
}
if err := marshal(filepath.Join(root, id, stateFilename), expectedState); err != nil {
t.Fatal(err)
}
factory, err := New(root, Cgroupfs)
if err != nil {
t.Fatal(err)
}
container, err := factory.Load(id)
if err != nil {
t.Fatal(err)
}
if container.ID() != id {
t.Fatalf("expected container id %q but received %q", id, container.ID())
}
config := container.Config()
if config.Rootfs != expectedConfig.Rootfs {
t.Fatalf("expected rootfs %q but received %q", expectedConfig.Rootfs, config.Rootfs)
}
lcontainer, ok := container.(*linuxContainer)
if !ok {
t.Fatal("expected linux container on linux based systems")
}
if lcontainer.initProcess.pid() != expectedState.InitProcessPid {
t.Fatalf("expected init pid %d but received %d", expectedState.InitProcessPid, lcontainer.initProcess.pid())
}
}
func marshal(path string, v interface{}) error {
f, err := os.Create(path)
if err != nil {
return err
}
defer f.Close()
return utils.WriteJSON(f, v)
}

View file

@ -9,6 +9,19 @@ import (
"github.com/opencontainers/runc/libcontainer/stacktrace"
)
type syncType uint8
const (
procReady syncType = iota
procError
procStart
procRun
)
type syncT struct {
Type syncType `json:"type"`
}
var errorTemplate = template.Must(template.New("error").Parse(`Timestamp: {{.Timestamp}}
Code: {{.ECode}}
{{if .Message }}

View file

@ -0,0 +1,14 @@
package libcontainer
import (
"fmt"
"io/ioutil"
"testing"
)
func TestErrorDetail(t *testing.T) {
err := newGenericError(fmt.Errorf("test error"), SystemError)
if derr := err.Detail(ioutil.Discard); derr != nil {
t.Fatal(derr)
}
}

View file

@ -5,6 +5,7 @@ package libcontainer
import (
"encoding/json"
"fmt"
"io"
"io/ioutil"
"net"
"os"
@ -73,6 +74,7 @@ func newContainerInit(t initType, pipe *os.File) (initer, error) {
}, nil
case initStandard:
return &linuxStandardInit{
pipe: pipe,
parentPid: syscall.Getppid(),
config: config,
}, nil
@ -140,6 +142,27 @@ func finalizeNamespace(config *initConfig) error {
return nil
}
// syncParentReady sends to the given pipe a JSON payload which indicates that
// the init is ready to Exec the child process. It then waits for the parent to
// indicate that it is cleared to Exec.
func syncParentReady(pipe io.ReadWriter) error {
// Tell parent.
if err := utils.WriteJSON(pipe, syncT{procReady}); err != nil {
return err
}
// Wait for parent to give the all-clear.
var procSync syncT
if err := json.NewDecoder(pipe).Decode(&procSync); err != nil {
if err == io.EOF {
return fmt.Errorf("parent closed synchronisation channel")
}
if procSync.Type != procRun {
return fmt.Errorf("invalid synchronisation flag from parent")
}
}
return nil
}
// joinExistingNamespaces gets all the namespace paths specified for the container and
// does a setns on the namespace fd so that the current process joins the namespace.
func joinExistingNamespaces(namespaces []configs.Namespace) error {
@ -309,7 +332,7 @@ func killCgroupProcesses(m cgroups.Manager) error {
if err := m.Freeze(configs.Frozen); err != nil {
logrus.Warn(err)
}
pids, err := m.GetPids()
pids, err := m.GetAllPids()
if err != nil {
m.Freeze(configs.Thawed)
return err

View file

@ -0,0 +1,204 @@
package integration
import (
"bufio"
"bytes"
"io/ioutil"
"os"
"path/filepath"
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
)
func showFile(t *testing.T, fname string) error {
t.Logf("=== %s ===\n", fname)
f, err := os.Open(fname)
if err != nil {
t.Log(err)
return err
}
defer f.Close()
scanner := bufio.NewScanner(f)
for scanner.Scan() {
t.Log(scanner.Text())
}
if err := scanner.Err(); err != nil {
return err
}
t.Logf("=== END ===\n")
return nil
}
func TestCheckpoint(t *testing.T) {
if testing.Short() {
return
}
root, err := newTestRoot()
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(root)
rootfs, err := newRootfs()
if err != nil {
t.Fatal(err)
}
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Mounts = append(config.Mounts, &configs.Mount{
Destination: "/sys/fs/cgroup",
Device: "cgroup",
Flags: defaultMountFlags | syscall.MS_RDONLY,
})
factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
if err != nil {
t.Fatal(err)
}
container, err := factory.Create("test", config)
if err != nil {
t.Fatal(err)
}
defer container.Destroy()
stdinR, stdinW, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
Stdout: &stdout,
}
err = container.Start(&pconfig)
stdinR.Close()
defer stdinW.Close()
if err != nil {
t.Fatal(err)
}
pid, err := pconfig.Pid()
if err != nil {
t.Fatal(err)
}
process, err := os.FindProcess(pid)
if err != nil {
t.Fatal(err)
}
imagesDir, err := ioutil.TempDir("", "criu")
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(imagesDir)
checkpointOpts := &libcontainer.CriuOpts{
ImagesDirectory: imagesDir,
WorkDirectory: imagesDir,
}
dumpLog := filepath.Join(checkpointOpts.WorkDirectory, "dump.log")
restoreLog := filepath.Join(checkpointOpts.WorkDirectory, "restore.log")
if err := container.Checkpoint(checkpointOpts); err != nil {
showFile(t, dumpLog)
t.Fatal(err)
}
state, err := container.Status()
if err != nil {
t.Fatal(err)
}
if state != libcontainer.Running {
t.Fatal("Unexpected state checkpoint: ", state)
}
stdinW.Close()
_, err = process.Wait()
if err != nil {
t.Fatal(err)
}
// reload the container
container, err = factory.Load("test")
if err != nil {
t.Fatal(err)
}
restoreStdinR, restoreStdinW, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
restoreProcessConfig := &libcontainer.Process{
Cwd: "/",
Stdin: restoreStdinR,
Stdout: &stdout,
}
err = container.Restore(restoreProcessConfig, checkpointOpts)
restoreStdinR.Close()
defer restoreStdinW.Close()
if err != nil {
showFile(t, restoreLog)
t.Fatal(err)
}
state, err = container.Status()
if err != nil {
t.Fatal(err)
}
if state != libcontainer.Running {
t.Fatal("Unexpected restore state: ", state)
}
pid, err = restoreProcessConfig.Pid()
if err != nil {
t.Fatal(err)
}
process, err = os.FindProcess(pid)
if err != nil {
t.Fatal(err)
}
_, err = restoreStdinW.WriteString("Hello!")
if err != nil {
t.Fatal(err)
}
restoreStdinW.Close()
s, err := process.Wait()
if err != nil {
t.Fatal(err)
}
if !s.Success() {
t.Fatal(s.String(), pid)
}
output := string(stdout.Bytes())
if !strings.Contains(output, "Hello!") {
t.Fatal("Did not restore the pipe correctly:", output)
}
}

View file

@ -0,0 +1,2 @@
// integration is used for integration testing of libcontainer
package integration

File diff suppressed because it is too large Load diff

View file

@ -0,0 +1,402 @@
package integration
import (
"bytes"
"io"
"os"
"strconv"
"strings"
"testing"
"time"
"github.com/opencontainers/runc/libcontainer"
)
func TestExecIn(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
buffers := newStdBuffers()
ps := &libcontainer.Process{
Cwd: "/",
Args: []string{"ps"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
stdinW.Close()
waitProcess(process, t)
out := buffers.Stdout.String()
if !strings.Contains(out, "cat") || !strings.Contains(out, "ps") {
t.Fatalf("unexpected running process, output %q", out)
}
}
func TestExecInRlimit(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
buffers := newStdBuffers()
ps := &libcontainer.Process{
Cwd: "/",
Args: []string{"/bin/sh", "-c", "ulimit -n"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
stdinW.Close()
waitProcess(process, t)
out := buffers.Stdout.String()
if limit := strings.TrimSpace(out); limit != "1025" {
t.Fatalf("expected rlimit to be 1025, got %s", limit)
}
}
func TestExecInError(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer func() {
stdinW.Close()
if _, err := process.Wait(); err != nil {
t.Log(err)
}
}()
ok(t, err)
for i := 0; i < 42; i++ {
var out bytes.Buffer
unexistent := &libcontainer.Process{
Cwd: "/",
Args: []string{"unexistent"},
Env: standardEnvironment,
Stdout: &out,
}
err = container.Start(unexistent)
if err == nil {
t.Fatal("Should be an error")
}
if !strings.Contains(err.Error(), "executable file not found") {
t.Fatalf("Should be error about not found executable, got %s", err)
}
if !bytes.Contains(out.Bytes(), []byte("executable file not found")) {
t.Fatalf("executable file not found error not delivered to stdio:\n%s", out.String())
}
}
}
func TestExecInTTY(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
var stdout bytes.Buffer
ps := &libcontainer.Process{
Cwd: "/",
Args: []string{"ps"},
Env: standardEnvironment,
}
console, err := ps.NewConsole(0)
copy := make(chan struct{})
go func() {
io.Copy(&stdout, console)
close(copy)
}()
ok(t, err)
err = container.Start(ps)
ok(t, err)
select {
case <-time.After(5 * time.Second):
t.Fatal("Waiting for copy timed out")
case <-copy:
}
waitProcess(ps, t)
stdinW.Close()
waitProcess(process, t)
out := stdout.String()
if !strings.Contains(out, "cat") || !strings.Contains(string(out), "ps") {
t.Fatalf("unexpected running process, output %q", out)
}
}
func TestExecInEnvironment(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
buffers := newStdBuffers()
process2 := &libcontainer.Process{
Cwd: "/",
Args: []string{"env"},
Env: []string{
"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
"DEBUG=true",
"DEBUG=false",
"ENV=test",
},
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(process2)
ok(t, err)
waitProcess(process2, t)
stdinW.Close()
waitProcess(process, t)
out := buffers.Stdout.String()
// check execin's process environment
if !strings.Contains(out, "DEBUG=false") ||
!strings.Contains(out, "ENV=test") ||
!strings.Contains(out, "HOME=/root") ||
!strings.Contains(out, "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin") ||
strings.Contains(out, "DEBUG=true") {
t.Fatalf("unexpected running process, output %q", out)
}
}
func TestExecinPassExtraFiles(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
if err != nil {
t.Fatal(err)
}
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
if err != nil {
t.Fatal(err)
}
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
if err != nil {
t.Fatal(err)
}
var stdout bytes.Buffer
pipeout1, pipein1, err := os.Pipe()
pipeout2, pipein2, err := os.Pipe()
inprocess := &libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "cd /proc/$$/fd; echo -n *; echo -n 1 >3; echo -n 2 >4"},
Env: []string{"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"},
ExtraFiles: []*os.File{pipein1, pipein2},
Stdin: nil,
Stdout: &stdout,
}
err = container.Start(inprocess)
if err != nil {
t.Fatal(err)
}
waitProcess(inprocess, t)
stdinW.Close()
waitProcess(process, t)
out := string(stdout.Bytes())
// fd 5 is the directory handle for /proc/$$/fd
if out != "0 1 2 3 4 5" {
t.Fatalf("expected to have the file descriptors '0 1 2 3 4 5' passed to exec, got '%s'", out)
}
var buf = []byte{0}
_, err = pipeout1.Read(buf)
if err != nil {
t.Fatal(err)
}
out1 := string(buf)
if out1 != "1" {
t.Fatalf("expected first pipe to receive '1', got '%s'", out1)
}
_, err = pipeout2.Read(buf)
if err != nil {
t.Fatal(err)
}
out2 := string(buf)
if out2 != "2" {
t.Fatalf("expected second pipe to receive '2', got '%s'", out2)
}
}
func TestExecInOomScoreAdj(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.OomScoreAdj = 200
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
buffers := newStdBuffers()
ps := &libcontainer.Process{
Cwd: "/",
Args: []string{"/bin/sh", "-c", "cat /proc/self/oom_score_adj"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
stdinW.Close()
waitProcess(process, t)
out := buffers.Stdout.String()
if oomScoreAdj := strings.TrimSpace(out); oomScoreAdj != strconv.Itoa(config.OomScoreAdj) {
t.Fatalf("expected oomScoreAdj to be %d, got %s", config.OomScoreAdj, oomScoreAdj)
}
}

View file

@ -0,0 +1,60 @@
package integration
import (
"os"
"runtime"
"testing"
"github.com/Sirupsen/logrus"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
_ "github.com/opencontainers/runc/libcontainer/nsenter"
)
// init runs the libcontainer initialization code because of the busybox style needs
// to work around the go runtime and the issues with forking
func init() {
if len(os.Args) < 2 || os.Args[1] != "init" {
return
}
runtime.GOMAXPROCS(1)
runtime.LockOSThread()
factory, err := libcontainer.New("")
if err != nil {
logrus.Fatalf("unable to initialize for container: %s", err)
}
if err := factory.StartInitialization(); err != nil {
logrus.Fatal(err)
}
}
var (
factory libcontainer.Factory
systemdFactory libcontainer.Factory
)
func TestMain(m *testing.M) {
var (
err error
ret int = 0
)
logrus.SetOutput(os.Stderr)
logrus.SetLevel(logrus.InfoLevel)
factory, err = libcontainer.New(".", libcontainer.Cgroupfs)
if err != nil {
logrus.Error(err)
os.Exit(1)
}
if systemd.UseSystemd() {
systemdFactory, err = libcontainer.New(".", libcontainer.SystemdCgroups)
if err != nil {
logrus.Error(err)
os.Exit(1)
}
}
ret = m.Run()
os.Exit(ret)
}

View file

@ -0,0 +1,219 @@
// +build linux,cgo,seccomp
package integration
import (
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
libseccomp "github.com/seccomp/libseccomp-golang"
)
func TestSeccompDenyGetcwd(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
if err != nil {
t.Fatal(err)
}
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Seccomp = &configs.Seccomp{
DefaultAction: configs.Allow,
Syscalls: []*configs.Syscall{
{
Name: "getcwd",
Action: configs.Errno,
},
},
}
container, err := newContainer(config)
if err != nil {
t.Fatal(err)
}
defer container.Destroy()
buffers := newStdBuffers()
pwd := &libcontainer.Process{
Cwd: "/",
Args: []string{"pwd"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(pwd)
if err != nil {
t.Fatal(err)
}
ps, err := pwd.Wait()
if err == nil {
t.Fatal("Expecting error (negative return code); instead exited cleanly!")
}
var exitCode int
status := ps.Sys().(syscall.WaitStatus)
if status.Exited() {
exitCode = status.ExitStatus()
} else if status.Signaled() {
exitCode = -int(status.Signal())
} else {
t.Fatalf("Unrecognized exit reason!")
}
if exitCode == 0 {
t.Fatalf("Getcwd should fail with negative exit code, instead got %d!", exitCode)
}
expected := "pwd: getcwd: Operation not permitted"
actual := strings.Trim(buffers.Stderr.String(), "\n")
if actual != expected {
t.Fatalf("Expected output %s but got %s\n", expected, actual)
}
}
func TestSeccompPermitWriteConditional(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
if err != nil {
t.Fatal(err)
}
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Seccomp = &configs.Seccomp{
DefaultAction: configs.Allow,
Syscalls: []*configs.Syscall{
{
Name: "write",
Action: configs.Errno,
Args: []*configs.Arg{
{
Index: 0,
Value: 1,
Op: configs.GreaterThan,
},
},
},
},
}
container, err := newContainer(config)
if err != nil {
t.Fatal(err)
}
defer container.Destroy()
buffers := newStdBuffers()
dmesg := &libcontainer.Process{
Cwd: "/",
Args: []string{"busybox", "ls", "/"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(dmesg)
if err != nil {
t.Fatal(err)
}
if _, err := dmesg.Wait(); err != nil {
t.Fatalf("%s: %s", err, buffers.Stderr)
}
}
func TestSeccompDenyWriteConditional(t *testing.T) {
if testing.Short() {
return
}
// Only test if library version is v2.2.1 or higher
// Conditional filtering will always error in v2.2.0 and lower
major, minor, micro := libseccomp.GetLibraryVersion()
if (major == 2 && minor < 2) || (major == 2 && minor == 2 && micro < 1) {
return
}
rootfs, err := newRootfs()
if err != nil {
t.Fatal(err)
}
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Seccomp = &configs.Seccomp{
DefaultAction: configs.Allow,
Syscalls: []*configs.Syscall{
{
Name: "write",
Action: configs.Errno,
Args: []*configs.Arg{
{
Index: 0,
Value: 1,
Op: configs.GreaterThan,
},
},
},
},
}
container, err := newContainer(config)
if err != nil {
t.Fatal(err)
}
defer container.Destroy()
buffers := newStdBuffers()
dmesg := &libcontainer.Process{
Cwd: "/",
Args: []string{"busybox", "ls", "does_not_exist"},
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(dmesg)
if err != nil {
t.Fatal(err)
}
ps, err := dmesg.Wait()
if err == nil {
t.Fatal("Expecting negative return, instead got 0!")
}
var exitCode int
status := ps.Sys().(syscall.WaitStatus)
if status.Exited() {
exitCode = status.ExitStatus()
} else if status.Signaled() {
exitCode = -int(status.Signal())
} else {
t.Fatalf("Unrecognized exit reason!")
}
if exitCode == 0 {
t.Fatalf("Busybox should fail with negative exit code, instead got %d!", exitCode)
}
// We're denying write to stderr, so we expect an empty buffer
expected := ""
actual := strings.Trim(buffers.Stderr.String(), "\n")
if actual != expected {
t.Fatalf("Expected output %s but got %s\n", expected, actual)
}
}

View file

@ -0,0 +1,120 @@
package integration
import (
"syscall"
"github.com/opencontainers/runc/libcontainer/configs"
)
var standardEnvironment = []string{
"HOME=/root",
"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
"HOSTNAME=integration",
"TERM=xterm",
}
const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
// newTemplateConfig returns a base template for running a container
//
// it uses a network strategy of just setting a loopback interface
// and the default setup for devices
func newTemplateConfig(rootfs string) *configs.Config {
return &configs.Config{
Rootfs: rootfs,
Capabilities: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Namespaces: configs.Namespaces([]configs.Namespace{
{Type: configs.NEWNS},
{Type: configs.NEWUTS},
{Type: configs.NEWIPC},
{Type: configs.NEWPID},
{Type: configs.NEWNET},
}),
Cgroups: &configs.Cgroup{
Path: "integration/test",
Resources: &configs.Resources{
MemorySwappiness: -1,
AllowAllDevices: false,
AllowedDevices: configs.DefaultAllowedDevices,
},
},
MaskPaths: []string{
"/proc/kcore",
},
ReadonlyPaths: []string{
"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
},
Devices: configs.DefaultAutoCreatedDevices,
Hostname: "integration",
Mounts: []*configs.Mount{
{
Source: "proc",
Destination: "/proc",
Device: "proc",
Flags: defaultMountFlags,
},
{
Source: "tmpfs",
Destination: "/dev",
Device: "tmpfs",
Flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME,
Data: "mode=755",
},
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
Flags: syscall.MS_NOSUID | syscall.MS_NOEXEC,
Data: "newinstance,ptmxmode=0666,mode=0620,gid=5",
},
{
Device: "tmpfs",
Source: "shm",
Destination: "/dev/shm",
Data: "mode=1777,size=65536k",
Flags: defaultMountFlags,
},
{
Source: "mqueue",
Destination: "/dev/mqueue",
Device: "mqueue",
Flags: defaultMountFlags,
},
{
Source: "sysfs",
Destination: "/sys",
Device: "sysfs",
Flags: defaultMountFlags | syscall.MS_RDONLY,
},
},
Networks: []*configs.Network{
{
Type: "loopback",
Address: "127.0.0.1/0",
Gateway: "localhost",
},
},
Rlimits: []configs.Rlimit{
{
Type: syscall.RLIMIT_NOFILE,
Hard: uint64(1025),
Soft: uint64(1025),
},
},
}
}

View file

@ -0,0 +1,141 @@
package integration
import (
"bytes"
"fmt"
"io/ioutil"
"os"
"os/exec"
"path/filepath"
"runtime"
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
)
func newStdBuffers() *stdBuffers {
return &stdBuffers{
Stdin: bytes.NewBuffer(nil),
Stdout: bytes.NewBuffer(nil),
Stderr: bytes.NewBuffer(nil),
}
}
type stdBuffers struct {
Stdin *bytes.Buffer
Stdout *bytes.Buffer
Stderr *bytes.Buffer
}
func (b *stdBuffers) String() string {
s := []string{}
if b.Stderr != nil {
s = append(s, b.Stderr.String())
}
if b.Stdout != nil {
s = append(s, b.Stdout.String())
}
return strings.Join(s, "|")
}
// ok fails the test if an err is not nil.
func ok(t testing.TB, err error) {
if err != nil {
_, file, line, _ := runtime.Caller(1)
t.Fatalf("%s:%d: unexpected error: %s\n\n", filepath.Base(file), line, err.Error())
}
}
func waitProcess(p *libcontainer.Process, t *testing.T) {
_, file, line, _ := runtime.Caller(1)
status, err := p.Wait()
if err != nil {
t.Fatalf("%s:%d: unexpected error: %s\n\n", filepath.Base(file), line, err.Error())
}
if !status.Success() {
t.Fatalf("%s:%d: unexpected status: %s\n\n", filepath.Base(file), line, status.String())
}
}
// newRootfs creates a new tmp directory and copies the busybox root filesystem
func newRootfs() (string, error) {
dir, err := ioutil.TempDir("", "")
if err != nil {
return "", err
}
if err := os.MkdirAll(dir, 0700); err != nil {
return "", err
}
if err := copyBusybox(dir); err != nil {
return "", err
}
return dir, nil
}
func remove(dir string) {
os.RemoveAll(dir)
}
// copyBusybox copies the rootfs for a busybox container created for the test image
// into the new directory for the specific test
func copyBusybox(dest string) error {
out, err := exec.Command("sh", "-c", fmt.Sprintf("cp -R /busybox/* %s/", dest)).CombinedOutput()
if err != nil {
return fmt.Errorf("copy error %q: %q", err, out)
}
return nil
}
func newContainer(config *configs.Config) (libcontainer.Container, error) {
f := factory
if config.Cgroups != nil && config.Cgroups.Parent == "system.slice" {
f = systemdFactory
}
return f.Create("testCT", config)
}
// runContainer runs the container with the specific config and arguments
//
// buffers are returned containing the STDOUT and STDERR output for the run
// along with the exit code and any go error
func runContainer(config *configs.Config, console string, args ...string) (buffers *stdBuffers, exitCode int, err error) {
container, err := newContainer(config)
if err != nil {
return nil, -1, err
}
defer container.Destroy()
buffers = newStdBuffers()
process := &libcontainer.Process{
Cwd: "/",
Args: args,
Env: standardEnvironment,
Stdin: buffers.Stdin,
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Start(process)
if err != nil {
return buffers, -1, err
}
ps, err := process.Wait()
if err != nil {
return buffers, -1, err
}
status := ps.Sys().(syscall.WaitStatus)
if status.Exited() {
exitCode = status.ExitStatus()
} else if status.Signaled() {
exitCode = -int(status.Signal())
} else {
return buffers, -1, err
}
return
}

View file

@ -0,0 +1,67 @@
// +build linux
package keyctl
import (
"fmt"
"syscall"
"strings"
"strconv"
"unsafe"
)
const KEYCTL_JOIN_SESSION_KEYRING = 1
const KEYCTL_SETPERM = 5
const KEYCTL_DESCRIBE = 6
type KeySerial uint32
func JoinSessionKeyring(name string) (KeySerial, error) {
var _name *byte = nil
var err error
if len(name) > 0 {
_name, err = syscall.BytePtrFromString(name)
if err != nil {
return KeySerial(0), err
}
}
sessKeyId, _, errn := syscall.Syscall(syscall.SYS_KEYCTL, KEYCTL_JOIN_SESSION_KEYRING, uintptr(unsafe.Pointer(_name)), 0)
if errn != 0 {
return 0, fmt.Errorf("could not create session key: %v", errn)
}
return KeySerial(sessKeyId), nil
}
// modify permissions on a keyring by reading the current permissions,
// anding the bits with the given mask (clearing permissions) and setting
// additional permission bits
func ModKeyringPerm(ringId KeySerial, mask, setbits uint32) error {
dest := make([]byte, 1024)
destBytes := unsafe.Pointer(&dest[0])
if _, _, err := syscall.Syscall6(syscall.SYS_KEYCTL, uintptr(KEYCTL_DESCRIBE), uintptr(ringId), uintptr(destBytes), uintptr(len(dest)), 0, 0); err != 0 {
return err
}
res := strings.Split(string(dest), ";")
if len(res) < 5 {
return fmt.Errorf("Destination buffer for key description is too small")
}
// parse permissions
perm64, err := strconv.ParseUint(res[3], 16, 32)
if err != nil {
return err
}
perm := (uint32(perm64) & mask) | setbits
if _, _, err := syscall.Syscall(syscall.SYS_KEYCTL, uintptr(KEYCTL_SETPERM), uintptr(ringId), uintptr(perm)); err != 0 {
return err
}
return nil
}

View file

@ -0,0 +1,144 @@
// +build selinux,linux
package label
import (
"os"
"strings"
"testing"
"github.com/opencontainers/runc/libcontainer/selinux"
)
func TestInit(t *testing.T) {
if selinux.SelinuxEnabled() {
var testNull []string
plabel, mlabel, err := InitLabels(testNull)
if err != nil {
t.Log("InitLabels Failed")
t.Fatal(err)
}
testDisabled := []string{"disable"}
plabel, mlabel, err = InitLabels(testDisabled)
if err != nil {
t.Log("InitLabels Disabled Failed")
t.Fatal(err)
}
if plabel != "" {
t.Log("InitLabels Disabled Failed")
t.Fatal()
}
testUser := []string{"user:user_u", "role:user_r", "type:user_t", "level:s0:c1,c15"}
plabel, mlabel, err = InitLabels(testUser)
if err != nil {
t.Log("InitLabels User Failed")
t.Fatal(err)
}
if plabel != "user_u:user_r:user_t:s0:c1,c15" || mlabel != "user_u:object_r:svirt_sandbox_file_t:s0:c1,c15" {
t.Log("InitLabels User Match Failed")
t.Log(plabel, mlabel)
t.Fatal(err)
}
testBadData := []string{"user", "role:user_r", "type:user_t", "level:s0:c1,c15"}
plabel, mlabel, err = InitLabels(testBadData)
if err == nil {
t.Log("InitLabels Bad Failed")
t.Fatal(err)
}
}
}
func TestDuplicateLabel(t *testing.T) {
secopt := DupSecOpt("system_u:system_r:svirt_lxc_net_t:s0:c1,c2")
t.Log(secopt)
for _, opt := range secopt {
con := strings.SplitN(opt, ":", 3)
if len(con) != 3 || con[0] != "label" {
t.Errorf("Invalid DupSecOpt return value")
continue
}
if con[1] == "user" {
if con[2] != "system_u" {
t.Errorf("DupSecOpt Failed user incorrect")
}
continue
}
if con[1] == "role" {
if con[2] != "system_r" {
t.Errorf("DupSecOpt Failed role incorrect")
}
continue
}
if con[1] == "type" {
if con[2] != "svirt_lxc_net_t" {
t.Errorf("DupSecOpt Failed type incorrect")
}
continue
}
if con[1] == "level" {
if con[2] != "s0:c1,c2" {
t.Errorf("DupSecOpt Failed level incorrect")
}
continue
}
t.Errorf("DupSecOpt Failed invalid field %q", con[1])
}
secopt = DisableSecOpt()
if secopt[0] != "label:disable" {
t.Errorf("DisableSecOpt Failed level incorrect")
}
}
func TestRelabel(t *testing.T) {
testdir := "/tmp/test"
if err := os.Mkdir(testdir, 0755); err != nil {
t.Fatal(err)
}
defer os.RemoveAll(testdir)
label := "system_u:system_r:svirt_sandbox_file_t:s0:c1,c2"
if err := Relabel(testdir, "", true); err != nil {
t.Fatal("Relabel with no label failed: %v", err)
}
if err := Relabel(testdir, label, true); err != nil {
t.Fatal("Relabel shared failed: %v", err)
}
if err := Relabel(testdir, label, false); err != nil {
t.Fatal("Relabel unshared failed: %v", err)
}
if err := Relabel("/etc", label, false); err == nil {
t.Fatal("Relabel /etc succeeded")
}
if err := Relabel("/", label, false); err == nil {
t.Fatal("Relabel / succeeded")
}
if err := Relabel("/usr", label, false); err == nil {
t.Fatal("Relabel /usr succeeded")
}
}
func TestValidate(t *testing.T) {
if err := Validate("zZ"); err != ErrIncompatibleLabel {
t.Fatalf("Expected incompatible error, got %v", err)
}
if err := Validate("Z"); err != nil {
t.Fatal(err)
}
if err := Validate("z"); err != nil {
t.Fatal(err)
}
if err := Validate(""); err != nil {
t.Fatal(err)
}
}
func TestIsShared(t *testing.T) {
if shared := IsShared("Z"); shared {
t.Fatal("Expected label `Z` to not be shared, got %v", shared)
}
if shared := IsShared("z"); !shared {
t.Fatal("Expected label `z` to be shared, got %v", shared)
}
if shared := IsShared("Zz"); !shared {
t.Fatal("Expected label `Zz` to be shared, got %v", shared)
}
}

View file

@ -12,31 +12,32 @@ import (
const oomCgroupName = "memory"
// notifyOnOOM returns channel on which you can expect event about OOM,
// if process died without OOM this channel will be closed.
// s is current *libcontainer.State for container.
func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
dir := paths[oomCgroupName]
if dir == "" {
return nil, fmt.Errorf("There is no path for %q in state", oomCgroupName)
}
oomControl, err := os.Open(filepath.Join(dir, "memory.oom_control"))
type PressureLevel uint
const (
LowPressure PressureLevel = iota
MediumPressure
CriticalPressure
)
func registerMemoryEvent(cgDir string, evName string, arg string) (<-chan struct{}, error) {
evFile, err := os.Open(filepath.Join(cgDir, evName))
if err != nil {
return nil, err
}
fd, _, syserr := syscall.RawSyscall(syscall.SYS_EVENTFD2, 0, syscall.FD_CLOEXEC, 0)
if syserr != 0 {
oomControl.Close()
evFile.Close()
return nil, syserr
}
eventfd := os.NewFile(fd, "eventfd")
eventControlPath := filepath.Join(dir, "cgroup.event_control")
data := fmt.Sprintf("%d %d", eventfd.Fd(), oomControl.Fd())
eventControlPath := filepath.Join(cgDir, "cgroup.event_control")
data := fmt.Sprintf("%d %d %s", eventfd.Fd(), evFile.Fd(), arg)
if err := ioutil.WriteFile(eventControlPath, []byte(data), 0700); err != nil {
eventfd.Close()
oomControl.Close()
evFile.Close()
return nil, err
}
ch := make(chan struct{})
@ -44,7 +45,7 @@ func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
defer func() {
close(ch)
eventfd.Close()
oomControl.Close()
evFile.Close()
}()
buf := make([]byte, 8)
for {
@ -61,3 +62,28 @@ func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
}()
return ch, nil
}
// notifyOnOOM returns channel on which you can expect event about OOM,
// if process died without OOM this channel will be closed.
func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
dir := paths[oomCgroupName]
if dir == "" {
return nil, fmt.Errorf("path %q missing", oomCgroupName)
}
return registerMemoryEvent(dir, "memory.oom_control", "")
}
func notifyMemoryPressure(paths map[string]string, level PressureLevel) (<-chan struct{}, error) {
dir := paths[oomCgroupName]
if dir == "" {
return nil, fmt.Errorf("path %q missing", oomCgroupName)
}
if level > CriticalPressure {
return nil, fmt.Errorf("invalid pressure level %d", level)
}
levelStr := []string{"low", "medium", "critical"}[level]
return registerMemoryEvent(dir, "memory.pressure_level", levelStr)
}

View file

@ -0,0 +1,128 @@
// +build linux
package libcontainer
import (
"encoding/binary"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"syscall"
"testing"
"time"
)
type notifyFunc func(paths map[string]string) (<-chan struct{}, error)
func testMemoryNotification(t *testing.T, evName string, notify notifyFunc, targ string) {
memoryPath, err := ioutil.TempDir("", "testmemnotification-"+evName)
if err != nil {
t.Fatal(err)
}
evFile := filepath.Join(memoryPath, evName)
eventPath := filepath.Join(memoryPath, "cgroup.event_control")
if err := ioutil.WriteFile(evFile, []byte{}, 0700); err != nil {
t.Fatal(err)
}
if err := ioutil.WriteFile(eventPath, []byte{}, 0700); err != nil {
t.Fatal(err)
}
paths := map[string]string{
"memory": memoryPath,
}
ch, err := notify(paths)
if err != nil {
t.Fatal("expected no error, got:", err)
}
data, err := ioutil.ReadFile(eventPath)
if err != nil {
t.Fatal("couldn't read event control file:", err)
}
var eventFd, evFd int
var arg string
if targ != "" {
_, err = fmt.Sscanf(string(data), "%d %d %s", &eventFd, &evFd, &arg)
} else {
_, err = fmt.Sscanf(string(data), "%d %d", &eventFd, &evFd)
}
if err != nil || arg != targ {
t.Fatalf("invalid control data %q: %s", data, err)
}
// re-open the eventfd
efd, err := syscall.Dup(eventFd)
if err != nil {
t.Fatal("unable to reopen eventfd:", err)
}
defer syscall.Close(efd)
if err != nil {
t.Fatal("unable to dup event fd:", err)
}
buf := make([]byte, 8)
binary.LittleEndian.PutUint64(buf, 1)
if _, err := syscall.Write(efd, buf); err != nil {
t.Fatal("unable to write to eventfd:", err)
}
select {
case <-ch:
case <-time.After(100 * time.Millisecond):
t.Fatal("no notification on channel after 100ms")
}
// simulate what happens when a cgroup is destroyed by cleaning up and then
// writing to the eventfd.
if err := os.RemoveAll(memoryPath); err != nil {
t.Fatal(err)
}
if _, err := syscall.Write(efd, buf); err != nil {
t.Fatal("unable to write to eventfd:", err)
}
// give things a moment to shut down
select {
case _, ok := <-ch:
if ok {
t.Fatal("expected no notification to be triggered")
}
case <-time.After(100 * time.Millisecond):
}
if _, _, err := syscall.Syscall(syscall.SYS_FCNTL, uintptr(evFd), syscall.F_GETFD, 0); err != syscall.EBADF {
t.Error("expected event control to be closed")
}
if _, _, err := syscall.Syscall(syscall.SYS_FCNTL, uintptr(eventFd), syscall.F_GETFD, 0); err != syscall.EBADF {
t.Error("expected event fd to be closed")
}
}
func TestNotifyOnOOM(t *testing.T) {
f := func(paths map[string]string) (<-chan struct{}, error) {
return notifyOnOOM(paths)
}
testMemoryNotification(t, "memory.oom_control", f, "")
}
func TestNotifyMemoryPressure(t *testing.T) {
tests := map[PressureLevel]string{
LowPressure: "low",
MediumPressure: "medium",
CriticalPressure: "critical",
}
for level, arg := range tests {
f := func(paths map[string]string) (<-chan struct{}, error) {
return notifyMemoryPressure(paths, level)
}
testMemoryNotification(t, "memory.pressure_level", f, arg)
}
}

View file

@ -0,0 +1,143 @@
package nsenter
import (
"bytes"
"encoding/json"
"io"
"os"
"os/exec"
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/vishvananda/netlink/nl"
)
type pid struct {
Pid int `json:"Pid"`
}
func TestNsenterAlivePid(t *testing.T) {
args := []string{"nsenter-exec"}
parent, child, err := newPipe()
if err != nil {
t.Fatalf("failed to create pipe %v", err)
}
cmd := &exec.Cmd{
Path: os.Args[0],
Args: args,
ExtraFiles: []*os.File{child},
Env: []string{"_LIBCONTAINER_INITTYPE=setns", "_LIBCONTAINER_INITPIPE=3"},
}
if err := cmd.Start(); err != nil {
t.Fatalf("nsenter failed to start %v", err)
}
r := nl.NewNetlinkRequest(int(libcontainer.InitMsg), 0)
r.AddData(&libcontainer.Int32msg{
Type: libcontainer.PidAttr,
Value: uint32(os.Getpid()),
})
if _, err := io.Copy(parent, bytes.NewReader(r.Serialize())); err != nil {
t.Fatal(err)
}
decoder := json.NewDecoder(parent)
var pid *pid
if err := decoder.Decode(&pid); err != nil {
t.Fatalf("%v", err)
}
if err := cmd.Wait(); err != nil {
t.Fatalf("nsenter exits with a non-zero exit status")
}
p, err := os.FindProcess(pid.Pid)
if err != nil {
t.Fatalf("%v", err)
}
p.Wait()
}
func TestNsenterInvalidPid(t *testing.T) {
args := []string{"nsenter-exec"}
parent, child, err := newPipe()
if err != nil {
t.Fatalf("failed to create pipe %v", err)
}
cmd := &exec.Cmd{
Path: os.Args[0],
Args: args,
ExtraFiles: []*os.File{child},
Env: []string{"_LIBCONTAINER_INITTYPE=setns", "_LIBCONTAINER_INITPIPE=3"},
}
if err := cmd.Start(); err != nil {
t.Fatal("nsenter exits with a zero exit status")
}
r := nl.NewNetlinkRequest(int(libcontainer.InitMsg), 0)
r.AddData(&libcontainer.Int32msg{
Type: libcontainer.PidAttr,
Value: 0,
})
if _, err := io.Copy(parent, bytes.NewReader(r.Serialize())); err != nil {
t.Fatal(err)
}
if err := cmd.Wait(); err == nil {
t.Fatal("nsenter exits with a zero exit status")
}
}
func TestNsenterDeadPid(t *testing.T) {
deadCmd := exec.Command("true")
if err := deadCmd.Run(); err != nil {
t.Fatal(err)
}
args := []string{"nsenter-exec"}
parent, child, err := newPipe()
if err != nil {
t.Fatalf("failed to create pipe %v", err)
}
cmd := &exec.Cmd{
Path: os.Args[0],
Args: args,
ExtraFiles: []*os.File{child},
Env: []string{"_LIBCONTAINER_INITTYPE=setns", "_LIBCONTAINER_INITPIPE=3"},
}
if err := cmd.Start(); err != nil {
t.Fatal("nsenter exits with a zero exit status")
}
r := nl.NewNetlinkRequest(int(libcontainer.InitMsg), 0)
r.AddData(&libcontainer.Int32msg{
Type: libcontainer.PidAttr,
Value: uint32(deadCmd.Process.Pid),
})
if _, err := io.Copy(parent, bytes.NewReader(r.Serialize())); err != nil {
t.Fatal(err)
}
if err := cmd.Wait(); err == nil {
t.Fatal("nsenter exits with a zero exit status")
}
}
func init() {
if strings.HasPrefix(os.Args[0], "nsenter-") {
os.Exit(0)
}
return
}
func newPipe() (parent *os.File, child *os.File, err error) {
fds, err := syscall.Socketpair(syscall.AF_LOCAL, syscall.SOCK_STREAM|syscall.SOCK_CLOEXEC, 0)
if err != nil {
return nil, nil, err
}
return os.NewFile(uintptr(fds[1]), "parent"), os.NewFile(uintptr(fds[0]), "child"), nil
}

View file

@ -17,6 +17,7 @@
#include <sched.h>
#include <signal.h>
#include <bits/sockaddr.h>
#include <linux/netlink.h>
#include <linux/types.h>
#include <stdint.h>

View file

@ -55,7 +55,7 @@ type Process struct {
// Wait releases any resources associated with the Process
func (p Process) Wait() (*os.ProcessState, error) {
if p.ops == nil {
return nil, newGenericError(fmt.Errorf("invalid process"), ProcessNotExecuted)
return nil, newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
}
return p.ops.wait()
}
@ -65,7 +65,7 @@ func (p Process) Pid() (int, error) {
// math.MinInt32 is returned here, because it's invalid value
// for the kill() system call.
if p.ops == nil {
return math.MinInt32, newGenericError(fmt.Errorf("invalid process"), ProcessNotExecuted)
return math.MinInt32, newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
}
return p.ops.pid(), nil
}
@ -73,7 +73,7 @@ func (p Process) Pid() (int, error) {
// Signal sends a signal to the Process.
func (p Process) Signal(sig os.Signal) error {
if p.ops == nil {
return newGenericError(fmt.Errorf("invalid process"), ProcessNotExecuted)
return newGenericError(fmt.Errorf("invalid process"), NoProcessOps)
}
return p.ops.signal(sig)
}

View file

@ -5,6 +5,7 @@ package libcontainer
import (
"encoding/json"
"errors"
"fmt"
"io"
"os"
"os/exec"
@ -15,6 +16,7 @@ import (
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/system"
"github.com/opencontainers/runc/libcontainer/utils"
)
type parentProcess interface {
@ -83,9 +85,10 @@ func (p *setnsProcess) start() (err error) {
return newSystemError(err)
}
}
if err := json.NewEncoder(p.parentPipe).Encode(p.config); err != nil {
if err := utils.WriteJSON(p.parentPipe, p.config); err != nil {
return newSystemError(err)
}
if err := syscall.Shutdown(int(p.parentPipe.Fd()), syscall.SHUT_WR); err != nil {
return newSystemError(err)
}
@ -95,6 +98,7 @@ func (p *setnsProcess) start() (err error) {
if err := json.NewDecoder(p.parentPipe).Decode(&ierr); err != nil && err != io.EOF {
return newSystemError(err)
}
// Must be done after Shutdown so the child will exit and we can wait for it.
if ierr != nil {
p.wait()
return newSystemError(ierr)
@ -198,7 +202,6 @@ func (p *initProcess) start() (err error) {
return newSystemError(err)
}
p.setExternalDescriptors(fds)
// Do this before syncing with child so that no children
// can escape the cgroup
if err := p.manager.Apply(p.pid()); err != nil {
@ -229,13 +232,56 @@ func (p *initProcess) start() (err error) {
if err := p.sendConfig(); err != nil {
return newSystemError(err)
}
// wait for the child process to fully complete and receive an error message
// if one was encoutered
var ierr *genericError
if err := json.NewDecoder(p.parentPipe).Decode(&ierr); err != nil && err != io.EOF {
var (
procSync syncT
sentRun bool
ierr *genericError
)
loop:
for {
if err := json.NewDecoder(p.parentPipe).Decode(&procSync); err != nil {
if err == io.EOF {
break loop
}
return newSystemError(err)
}
switch procSync.Type {
case procStart:
break loop
case procReady:
if err := p.manager.Set(p.config.Config); err != nil {
return newSystemError(err)
}
// Sync with child.
if err := utils.WriteJSON(p.parentPipe, syncT{procRun}); err != nil {
return newSystemError(err)
}
sentRun = true
case procError:
// wait for the child process to fully complete and receive an error message
// if one was encoutered
if err := json.NewDecoder(p.parentPipe).Decode(&ierr); err != nil && err != io.EOF {
return newSystemError(err)
}
if ierr != nil {
break loop
}
// Programmer error.
panic("No error following JSON procError payload.")
default:
return newSystemError(fmt.Errorf("invalid JSON synchronisation payload from child"))
}
}
if !sentRun {
return newSystemError(fmt.Errorf("could not synchronise with container process"))
}
if err := syscall.Shutdown(int(p.parentPipe.Fd()), syscall.SHUT_WR); err != nil {
return newSystemError(err)
}
// Must be done after Shutdown so the child will exit and we can wait for it.
if ierr != nil {
p.wait()
return newSystemError(ierr)
}
return nil
@ -269,12 +315,10 @@ func (p *initProcess) startTime() (string, error) {
}
func (p *initProcess) sendConfig() error {
// send the state to the container's init process then shutdown writes for the parent
if err := json.NewEncoder(p.parentPipe).Encode(p.config); err != nil {
return err
}
// shutdown writes for the parent side of the pipe
return syscall.Shutdown(int(p.parentPipe.Fd()), syscall.SHUT_WR)
// send the config to the container's init process, we don't use JSON Encode
// here because there might be a problem in JSON decoder in some cases, see:
// https://github.com/docker/docker/issues/14203#issuecomment-174177790
return utils.WriteJSON(p.parentPipe, p.config)
}
func (p *initProcess) createNetworkInterfaces() error {

View file

@ -18,6 +18,8 @@ import (
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/label"
"github.com/opencontainers/runc/libcontainer/system"
libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
)
const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
@ -293,12 +295,30 @@ func getCgroupMounts(m *configs.Mount) ([]*configs.Mount, error) {
// checkMountDestination checks to ensure that the mount destination is not over the top of /proc.
// dest is required to be an abs path and have any symlinks resolved before calling this function.
func checkMountDestination(rootfs, dest string) error {
if filepath.Clean(rootfs) == filepath.Clean(dest) {
if libcontainerUtils.CleanPath(rootfs) == libcontainerUtils.CleanPath(dest) {
return fmt.Errorf("mounting into / is prohibited")
}
invalidDestinations := []string{
"/proc",
}
// White list, it should be sub directories of invalid destinations
validDestinations := []string{
// These entries can be bind mounted by files emulated by fuse,
// so commands like top, free displays stats in container.
"/proc/cpuinfo",
"/proc/diskstats",
"/proc/meminfo",
"/proc/stats",
}
for _, valid := range validDestinations {
path, err := filepath.Rel(filepath.Join(rootfs, valid), dest)
if err != nil {
return err
}
if path == "." {
return nil
}
}
for _, invalid := range invalidDestinations {
path, err := filepath.Rel(filepath.Join(rootfs, invalid), dest)
if err != nil {
@ -365,11 +385,12 @@ func reOpenDevNull() error {
// Create the device nodes in the container.
func createDevices(config *configs.Config) error {
useBindMount := system.RunningInUserNS() || config.Namespaces.Contains(configs.NEWUSER)
oldMask := syscall.Umask(0000)
for _, node := range config.Devices {
// containers running in a user namespace are not allowed to mknod
// devices so we can just bind mount it from the host.
if err := createDeviceNode(config.Rootfs, node, config.Namespaces.Contains(configs.NEWUSER)); err != nil {
if err := createDeviceNode(config.Rootfs, node, useBindMount); err != nil {
syscall.Umask(oldMask)
return err
}

View file

@ -0,0 +1,37 @@
// +build linux
package libcontainer
import "testing"
func TestCheckMountDestOnProc(t *testing.T) {
dest := "/rootfs/proc/"
err := checkMountDestination("/rootfs", dest)
if err == nil {
t.Fatal("destination inside proc should return an error")
}
}
func TestCheckMountDestInSys(t *testing.T) {
dest := "/rootfs//sys/fs/cgroup"
err := checkMountDestination("/rootfs", dest)
if err != nil {
t.Fatal("destination inside /sys should not return an error")
}
}
func TestCheckMountDestFalsePositive(t *testing.T) {
dest := "/rootfs/sysfiles/fs/cgroup"
err := checkMountDestination("/rootfs", dest)
if err != nil {
t.Fatal(err)
}
}
func TestCheckMountRoot(t *testing.T) {
dest := "/rootfs"
err := checkMountDestination("/rootfs", dest)
if err == nil {
t.Fatal(err)
}
}

View file

@ -0,0 +1,47 @@
Name: cat
State: R (running)
Tgid: 19383
Ngid: 0
Pid: 19383
PPid: 19275
TracerPid: 0
Uid: 1000 1000 1000 1000
Gid: 1000 1000 1000 1000
FDSize: 256
Groups: 24 25 27 29 30 44 46 102 104 108 111 1000 1001
NStgid: 19383
NSpid: 19383
NSpgid: 19383
NSsid: 19275
VmPeak: 5944 kB
VmSize: 5944 kB
VmLck: 0 kB
VmPin: 0 kB
VmHWM: 744 kB
VmRSS: 744 kB
VmData: 324 kB
VmStk: 136 kB
VmExe: 48 kB
VmLib: 1776 kB
VmPTE: 32 kB
VmPMD: 12 kB
VmSwap: 0 kB
Threads: 1
SigQ: 0/30067
SigPnd: 0000000000000000
ShdPnd: 0000000000000000
SigBlk: 0000000000000000
SigIgn: 0000000000000080
SigCgt: 0000000000000000
CapInh: 0000000000000000
CapPrm: 0000000000000000
CapEff: 0000000000000000
CapBnd: 0000003fffffffff
CapAmb: 0000000000000000
Seccomp: 0
Cpus_allowed: f
Cpus_allowed_list: 0-3
Mems_allowed: 00000000,00000001
Mems_allowed_list: 0
voluntary_ctxt_switches: 0
nonvoluntary_ctxt_switches: 1

View file

@ -3,8 +3,11 @@
package seccomp
import (
"bufio"
"fmt"
"log"
"os"
"strings"
"syscall"
"github.com/opencontainers/runc/libcontainer/configs"
@ -17,6 +20,9 @@ var (
actKill = libseccomp.ActKill
actTrace = libseccomp.ActTrace.SetReturnCode(int16(syscall.EPERM))
actErrno = libseccomp.ActErrno.SetReturnCode(int16(syscall.EPERM))
// SeccompModeFilter refers to the syscall argument SECCOMP_MODE_FILTER.
SeccompModeFilter = uintptr(2)
)
// Filters given syscalls in a container, preventing them from being used
@ -73,6 +79,24 @@ func InitSeccomp(config *configs.Seccomp) error {
return nil
}
// IsEnabled returns if the kernel has been configured to support seccomp.
func IsEnabled() bool {
// Try to read from /proc/self/status for kernels > 3.8
s, err := parseStatusFile("/proc/self/status")
if err != nil {
// Check if Seccomp is supported, via CONFIG_SECCOMP.
if _, _, err := syscall.RawSyscall(syscall.SYS_PRCTL, syscall.PR_GET_SECCOMP, 0, 0); err != syscall.EINVAL {
// Make sure the kernel has CONFIG_SECCOMP_FILTER.
if _, _, err := syscall.RawSyscall(syscall.SYS_PRCTL, syscall.PR_SET_SECCOMP, SeccompModeFilter, 0); err != syscall.EINVAL {
return true
}
}
return false
}
_, ok := s["Seccomp"]
return ok
}
// Convert Libcontainer Action to Libseccomp ScmpAction
func getAction(act configs.Action) (libseccomp.ScmpAction, error) {
switch act {
@ -178,3 +202,30 @@ func matchCall(filter *libseccomp.ScmpFilter, call *configs.Syscall) error {
return nil
}
func parseStatusFile(path string) (map[string]string, error) {
f, err := os.Open(path)
if err != nil {
return nil, err
}
defer f.Close()
s := bufio.NewScanner(f)
status := make(map[string]string)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
text := s.Text()
parts := strings.Split(text, ":")
if len(parts) <= 1 {
continue
}
status[parts[0]] = parts[1]
}
return status, nil
}

View file

@ -0,0 +1,17 @@
// +build linux,cgo,seccomp
package seccomp
import "testing"
func TestParseStatusFile(t *testing.T) {
s, err := parseStatusFile("fixtures/proc_self_status")
if err != nil {
t.Fatal(err)
}
if _, ok := s["Seccomp"]; !ok {
t.Fatal("expected to find 'Seccomp' in the map but did not.")
}
}

View file

@ -17,3 +17,8 @@ func InitSeccomp(config *configs.Seccomp) error {
}
return nil
}
// IsEnabled returns false, because it is not supported.
func IsEnabled() bool {
return false
}

View file

@ -0,0 +1,477 @@
// +build linux
package selinux
import (
"bufio"
"crypto/rand"
"encoding/binary"
"fmt"
"io"
"os"
"path/filepath"
"regexp"
"strconv"
"strings"
"syscall"
"github.com/docker/docker/pkg/mount"
"github.com/opencontainers/runc/libcontainer/system"
)
const (
Enforcing = 1
Permissive = 0
Disabled = -1
selinuxDir = "/etc/selinux/"
selinuxConfig = selinuxDir + "config"
selinuxTypeTag = "SELINUXTYPE"
selinuxTag = "SELINUX"
selinuxPath = "/sys/fs/selinux"
xattrNameSelinux = "security.selinux"
stRdOnly = 0x01
)
var (
assignRegex = regexp.MustCompile(`^([^=]+)=(.*)$`)
mcsList = make(map[string]bool)
selinuxfs = "unknown"
selinuxEnabled = false // Stores whether selinux is currently enabled
selinuxEnabledChecked = false // Stores whether selinux enablement has been checked or established yet
)
type SELinuxContext map[string]string
// SetDisabled disables selinux support for the package
func SetDisabled() {
selinuxEnabled, selinuxEnabledChecked = false, true
}
// getSelinuxMountPoint returns the path to the mountpoint of an selinuxfs
// filesystem or an empty string if no mountpoint is found. Selinuxfs is
// a proc-like pseudo-filesystem that exposes the selinux policy API to
// processes. The existence of an selinuxfs mount is used to determine
// whether selinux is currently enabled or not.
func getSelinuxMountPoint() string {
if selinuxfs != "unknown" {
return selinuxfs
}
selinuxfs = ""
mounts, err := mount.GetMounts()
if err != nil {
return selinuxfs
}
for _, mount := range mounts {
if mount.Fstype == "selinuxfs" {
selinuxfs = mount.Mountpoint
break
}
}
if selinuxfs != "" {
var buf syscall.Statfs_t
syscall.Statfs(selinuxfs, &buf)
if (buf.Flags & stRdOnly) == 1 {
selinuxfs = ""
}
}
return selinuxfs
}
// SelinuxEnabled returns whether selinux is currently enabled.
func SelinuxEnabled() bool {
if selinuxEnabledChecked {
return selinuxEnabled
}
selinuxEnabledChecked = true
if fs := getSelinuxMountPoint(); fs != "" {
if con, _ := Getcon(); con != "kernel" {
selinuxEnabled = true
}
}
return selinuxEnabled
}
func readConfig(target string) (value string) {
var (
val, key string
bufin *bufio.Reader
)
in, err := os.Open(selinuxConfig)
if err != nil {
return ""
}
defer in.Close()
bufin = bufio.NewReader(in)
for done := false; !done; {
var line string
if line, err = bufin.ReadString('\n'); err != nil {
if err != io.EOF {
return ""
}
done = true
}
line = strings.TrimSpace(line)
if len(line) == 0 {
// Skip blank lines
continue
}
if line[0] == ';' || line[0] == '#' {
// Skip comments
continue
}
if groups := assignRegex.FindStringSubmatch(line); groups != nil {
key, val = strings.TrimSpace(groups[1]), strings.TrimSpace(groups[2])
if key == target {
return strings.Trim(val, "\"")
}
}
}
return ""
}
func getSELinuxPolicyRoot() string {
return selinuxDir + readConfig(selinuxTypeTag)
}
func readCon(name string) (string, error) {
var val string
in, err := os.Open(name)
if err != nil {
return "", err
}
defer in.Close()
_, err = fmt.Fscanf(in, "%s", &val)
return val, err
}
// Setfilecon sets the SELinux label for this path or returns an error.
func Setfilecon(path string, scon string) error {
return system.Lsetxattr(path, xattrNameSelinux, []byte(scon), 0)
}
// Getfilecon returns the SELinux label for this path or returns an error.
func Getfilecon(path string) (string, error) {
con, err := system.Lgetxattr(path, xattrNameSelinux)
// Trim the NUL byte at the end of the byte buffer, if present.
if con[len(con)-1] == '\x00' {
con = con[:len(con)-1]
}
return string(con), err
}
func Setfscreatecon(scon string) error {
return writeCon(fmt.Sprintf("/proc/self/task/%d/attr/fscreate", syscall.Gettid()), scon)
}
func Getfscreatecon() (string, error) {
return readCon(fmt.Sprintf("/proc/self/task/%d/attr/fscreate", syscall.Gettid()))
}
// Getcon returns the SELinux label of the current process thread, or an error.
func Getcon() (string, error) {
return readCon(fmt.Sprintf("/proc/self/task/%d/attr/current", syscall.Gettid()))
}
// Getpidcon returns the SELinux label of the given pid, or an error.
func Getpidcon(pid int) (string, error) {
return readCon(fmt.Sprintf("/proc/%d/attr/current", pid))
}
func Getexeccon() (string, error) {
return readCon(fmt.Sprintf("/proc/self/task/%d/attr/exec", syscall.Gettid()))
}
func writeCon(name string, val string) error {
out, err := os.OpenFile(name, os.O_WRONLY, 0)
if err != nil {
return err
}
defer out.Close()
if val != "" {
_, err = out.Write([]byte(val))
} else {
_, err = out.Write(nil)
}
return err
}
func Setexeccon(scon string) error {
return writeCon(fmt.Sprintf("/proc/self/task/%d/attr/exec", syscall.Gettid()), scon)
}
func (c SELinuxContext) Get() string {
return fmt.Sprintf("%s:%s:%s:%s", c["user"], c["role"], c["type"], c["level"])
}
func NewContext(scon string) SELinuxContext {
c := make(SELinuxContext)
if len(scon) != 0 {
con := strings.SplitN(scon, ":", 4)
c["user"] = con[0]
c["role"] = con[1]
c["type"] = con[2]
c["level"] = con[3]
}
return c
}
func ReserveLabel(scon string) {
if len(scon) != 0 {
con := strings.SplitN(scon, ":", 4)
mcsAdd(con[3])
}
}
func selinuxEnforcePath() string {
return fmt.Sprintf("%s/enforce", selinuxPath)
}
func SelinuxGetEnforce() int {
var enforce int
enforceS, err := readCon(selinuxEnforcePath())
if err != nil {
return -1
}
enforce, err = strconv.Atoi(string(enforceS))
if err != nil {
return -1
}
return enforce
}
func SelinuxSetEnforce(mode int) error {
return writeCon(selinuxEnforcePath(), fmt.Sprintf("%d", mode))
}
func SelinuxGetEnforceMode() int {
switch readConfig(selinuxTag) {
case "enforcing":
return Enforcing
case "permissive":
return Permissive
}
return Disabled
}
func mcsAdd(mcs string) error {
if mcsList[mcs] {
return fmt.Errorf("MCS Label already exists")
}
mcsList[mcs] = true
return nil
}
func mcsDelete(mcs string) {
mcsList[mcs] = false
}
func IntToMcs(id int, catRange uint32) string {
var (
SETSIZE = int(catRange)
TIER = SETSIZE
ORD = id
)
if id < 1 || id > 523776 {
return ""
}
for ORD > TIER {
ORD = ORD - TIER
TIER -= 1
}
TIER = SETSIZE - TIER
ORD = ORD + TIER
return fmt.Sprintf("s0:c%d,c%d", TIER, ORD)
}
func uniqMcs(catRange uint32) string {
var (
n uint32
c1, c2 uint32
mcs string
)
for {
binary.Read(rand.Reader, binary.LittleEndian, &n)
c1 = n % catRange
binary.Read(rand.Reader, binary.LittleEndian, &n)
c2 = n % catRange
if c1 == c2 {
continue
} else {
if c1 > c2 {
t := c1
c1 = c2
c2 = t
}
}
mcs = fmt.Sprintf("s0:c%d,c%d", c1, c2)
if err := mcsAdd(mcs); err != nil {
continue
}
break
}
return mcs
}
func FreeLxcContexts(scon string) {
if len(scon) != 0 {
con := strings.SplitN(scon, ":", 4)
mcsDelete(con[3])
}
}
func GetLxcContexts() (processLabel string, fileLabel string) {
var (
val, key string
bufin *bufio.Reader
)
if !SelinuxEnabled() {
return "", ""
}
lxcPath := fmt.Sprintf("%s/contexts/lxc_contexts", getSELinuxPolicyRoot())
in, err := os.Open(lxcPath)
if err != nil {
return "", ""
}
defer in.Close()
bufin = bufio.NewReader(in)
for done := false; !done; {
var line string
if line, err = bufin.ReadString('\n'); err != nil {
if err == io.EOF {
done = true
} else {
goto exit
}
}
line = strings.TrimSpace(line)
if len(line) == 0 {
// Skip blank lines
continue
}
if line[0] == ';' || line[0] == '#' {
// Skip comments
continue
}
if groups := assignRegex.FindStringSubmatch(line); groups != nil {
key, val = strings.TrimSpace(groups[1]), strings.TrimSpace(groups[2])
if key == "process" {
processLabel = strings.Trim(val, "\"")
}
if key == "file" {
fileLabel = strings.Trim(val, "\"")
}
}
}
if processLabel == "" || fileLabel == "" {
return "", ""
}
exit:
// mcs := IntToMcs(os.Getpid(), 1024)
mcs := uniqMcs(1024)
scon := NewContext(processLabel)
scon["level"] = mcs
processLabel = scon.Get()
scon = NewContext(fileLabel)
scon["level"] = mcs
fileLabel = scon.Get()
return processLabel, fileLabel
}
func SecurityCheckContext(val string) error {
return writeCon(fmt.Sprintf("%s.context", selinuxPath), val)
}
func CopyLevel(src, dest string) (string, error) {
if src == "" {
return "", nil
}
if err := SecurityCheckContext(src); err != nil {
return "", err
}
if err := SecurityCheckContext(dest); err != nil {
return "", err
}
scon := NewContext(src)
tcon := NewContext(dest)
mcsDelete(tcon["level"])
mcsAdd(scon["level"])
tcon["level"] = scon["level"]
return tcon.Get(), nil
}
// Prevent users from relabing system files
func badPrefix(fpath string) error {
var badprefixes = []string{"/usr"}
for _, prefix := range badprefixes {
if fpath == prefix || strings.HasPrefix(fpath, fmt.Sprintf("%s/", prefix)) {
return fmt.Errorf("Relabeling content in %s is not allowed.", prefix)
}
}
return nil
}
// Change the fpath file object to the SELinux label scon.
// If the fpath is a directory and recurse is true Chcon will walk the
// directory tree setting the label
func Chcon(fpath string, scon string, recurse bool) error {
if scon == "" {
return nil
}
if err := badPrefix(fpath); err != nil {
return err
}
callback := func(p string, info os.FileInfo, err error) error {
return Setfilecon(p, scon)
}
if recurse {
return filepath.Walk(fpath, callback)
}
return Setfilecon(fpath, scon)
}
// DupSecOpt takes an SELinux process label and returns security options that
// can will set the SELinux Type and Level for future container processes
func DupSecOpt(src string) []string {
if src == "" {
return nil
}
con := NewContext(src)
if con["user"] == "" ||
con["role"] == "" ||
con["type"] == "" ||
con["level"] == "" {
return nil
}
return []string{"label:user:" + con["user"],
"label:role:" + con["role"],
"label:type:" + con["type"],
"label:level:" + con["level"]}
}
// DisableSecOpt returns a security opt that can be used to disabling SELinux
// labeling support for future container processes
func DisableSecOpt() []string {
return []string{"label:disable"}
}

View file

@ -0,0 +1,75 @@
// +build linux,selinux
package selinux_test
import (
"os"
"testing"
"github.com/opencontainers/runc/libcontainer/selinux"
)
func TestSetfilecon(t *testing.T) {
if selinux.SelinuxEnabled() {
tmp := "selinux_test"
out, _ := os.OpenFile(tmp, os.O_WRONLY|os.O_CREATE, 0)
out.Close()
err := selinux.Setfilecon(tmp, "system_u:object_r:bin_t:s0")
if err != nil {
t.Log("Setfilecon failed")
t.Fatal(err)
}
os.Remove(tmp)
}
}
func TestSELinux(t *testing.T) {
var (
err error
plabel, flabel string
)
if selinux.SelinuxEnabled() {
t.Log("Enabled")
plabel, flabel = selinux.GetLxcContexts()
t.Log(plabel)
t.Log(flabel)
selinux.FreeLxcContexts(plabel)
plabel, flabel = selinux.GetLxcContexts()
t.Log(plabel)
t.Log(flabel)
selinux.FreeLxcContexts(plabel)
t.Log("getenforce ", selinux.SelinuxGetEnforce())
mode := selinux.SelinuxGetEnforceMode()
t.Log("getenforcemode ", mode)
defer selinux.SelinuxSetEnforce(mode)
if err := selinux.SelinuxSetEnforce(selinux.Enforcing); err != nil {
t.Fatalf("enforcing selinux failed: %v", err)
}
if err := selinux.SelinuxSetEnforce(selinux.Permissive); err != nil {
t.Fatalf("setting selinux mode to permissive failed: %v", err)
}
selinux.SelinuxSetEnforce(mode)
pid := os.Getpid()
t.Logf("PID:%d MCS:%s\n", pid, selinux.IntToMcs(pid, 1023))
err = selinux.Setfscreatecon("unconfined_u:unconfined_r:unconfined_t:s0")
if err == nil {
t.Log(selinux.Getfscreatecon())
} else {
t.Log("setfscreatecon failed", err)
t.Fatal(err)
}
err = selinux.Setfscreatecon("")
if err == nil {
t.Log(selinux.Getfscreatecon())
} else {
t.Log("setfscreatecon failed", err)
t.Fatal(err)
}
t.Log(selinux.Getpidcon(1))
} else {
t.Log("Disabled")
}
}

View file

@ -6,6 +6,7 @@ import (
"os"
"github.com/opencontainers/runc/libcontainer/apparmor"
"github.com/opencontainers/runc/libcontainer/keys"
"github.com/opencontainers/runc/libcontainer/label"
"github.com/opencontainers/runc/libcontainer/seccomp"
"github.com/opencontainers/runc/libcontainer/system"
@ -18,6 +19,10 @@ type linuxSetnsInit struct {
}
func (l *linuxSetnsInit) Init() error {
// do not inherit the parent's session keyring
if _, err := keyctl.JoinSessionKeyring("_ses"); err != nil {
return err
}
if err := setupRlimits(l.config.Config); err != nil {
return err
}

View file

@ -0,0 +1,27 @@
package stacktrace
import "testing"
func captureFunc() Stacktrace {
return Capture(0)
}
func TestCaptureTestFunc(t *testing.T) {
stack := captureFunc()
if len(stack.Frames) == 0 {
t.Fatal("expected stack frames to be returned")
}
// the first frame is the caller
frame := stack.Frames[0]
if expected := "captureFunc"; frame.Function != expected {
t.Fatalf("expteced function %q but recevied %q", expected, frame.Function)
}
if expected := "github.com/opencontainers/runc/libcontainer/stacktrace"; frame.Package != expected {
t.Fatalf("expected package %q but received %q", expected, frame.Package)
}
if expected := "capture_test.go"; frame.File != expected {
t.Fatalf("expected file %q but received %q", expected, frame.File)
}
}

View file

@ -0,0 +1,20 @@
package stacktrace
import "testing"
func TestParsePackageName(t *testing.T) {
var (
name = "github.com/opencontainers/runc/libcontainer/stacktrace.captureFunc"
expectedPackage = "github.com/opencontainers/runc/libcontainer/stacktrace"
expectedFunction = "captureFunc"
)
pack, funcName := parseFunctionName(name)
if pack != expectedPackage {
t.Fatalf("expected package %q but received %q", expectedPackage, pack)
}
if funcName != expectedFunction {
t.Fatalf("expected function %q but received %q", expectedFunction, funcName)
}
}

View file

@ -3,22 +3,37 @@
package libcontainer
import (
"io"
"os"
"syscall"
"github.com/opencontainers/runc/libcontainer/apparmor"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/keys"
"github.com/opencontainers/runc/libcontainer/label"
"github.com/opencontainers/runc/libcontainer/seccomp"
"github.com/opencontainers/runc/libcontainer/system"
)
type linuxStandardInit struct {
pipe io.ReadWriter
parentPid int
config *initConfig
}
func (l *linuxStandardInit) Init() error {
// do not inherit the parent's session keyring
sessKeyId, err := keyctl.JoinSessionKeyring("")
if err != nil {
return err
}
// make session keyring searcheable
// without user ns we need 'UID' search permissions
// with user ns we need 'other' search permissions
if err := keyctl.ModKeyringPerm(sessKeyId, 0xffffffff, 0x080008); err != nil {
return err
}
// join any namespaces via a path to the namespace fd if provided
if err := joinExistingNamespaces(l.config.Config.Namespaces); err != nil {
return err
@ -50,7 +65,6 @@ func (l *linuxStandardInit) Init() error {
if err := setOomScoreAdj(l.config.Config.OomScoreAdj); err != nil {
return err
}
label.Init()
// InitializeMountNamespace() can be executed only for a new mount namespace
if l.config.Config.Namespaces.Contains(configs.NEWNS) {
@ -75,7 +89,6 @@ func (l *linuxStandardInit) Init() error {
return err
}
}
for _, path := range l.config.Config.ReadonlyPaths {
if err := remountReadonly(path); err != nil {
return err
@ -90,6 +103,12 @@ func (l *linuxStandardInit) Init() error {
if err != nil {
return err
}
// Tell our parent that we're ready to Execv. This must be done before the
// Seccomp rules have been applied, because we need to be able to read and
// write to a socket.
if err := syncParentReady(l.pipe); err != nil {
return err
}
if l.config.Config.Seccomp != nil {
if err := seccomp.InitSeccomp(l.config.Config.Seccomp); err != nil {
return err

View file

@ -49,6 +49,7 @@ func destroy(c *linuxContainer) error {
if herr := runPoststopHooks(c); err == nil {
err = herr
}
c.state = &stoppedState{c: c}
return err
}
@ -119,7 +120,7 @@ func (r *runningState) transition(s containerState) error {
case *pausedState:
r.c.state = s
return nil
case *runningState, *nullState:
case *runningState:
return nil
}
return newStateTransitionError(r, s)
@ -148,7 +149,7 @@ func (p *pausedState) status() Status {
func (p *pausedState) transition(s containerState) error {
switch s.(type) {
case *runningState:
case *runningState, *stoppedState:
p.c.state = s
return nil
case *pausedState:
@ -158,6 +159,16 @@ func (p *pausedState) transition(s containerState) error {
}
func (p *pausedState) destroy() error {
isRunning, err := p.c.isRunning()
if err != nil {
return err
}
if !isRunning {
if err := p.c.cgroupManager.Freeze(configs.Thawed); err != nil {
return err
}
return destroy(p.c)
}
return newGenericError(fmt.Errorf("container is paused"), ContainerPaused)
}
@ -191,27 +202,25 @@ func (r *restoredState) destroy() error {
return destroy(r.c)
}
// nullState is used whenever a container is restored, loaded, or setting additional
// createdState is used whenever a container is restored, loaded, or setting additional
// processes inside and it should not be destroyed when it is exiting.
type nullState struct {
type createdState struct {
c *linuxContainer
s Status
}
func (n *nullState) status() Status {
func (n *createdState) status() Status {
return n.s
}
func (n *nullState) transition(s containerState) error {
switch s.(type) {
case *restoredState:
n.c.state = s
default:
// do nothing for null states
}
func (n *createdState) transition(s containerState) error {
n.c.state = s
return nil
}
func (n *nullState) destroy() error {
return nil
func (n *createdState) destroy() error {
if err := n.c.refreshState(); err != nil {
return err
}
return n.c.state.destroy()
}

View file

@ -0,0 +1,79 @@
// +build linux
package libcontainer
import "testing"
func TestStateStatus(t *testing.T) {
states := map[containerState]Status{
&stoppedState{}: Destroyed,
&runningState{}: Running,
&restoredState{}: Running,
&pausedState{}: Paused,
}
for s, status := range states {
if s.status() != status {
t.Fatalf("state returned %s but expected %s", s.status(), status)
}
}
}
func isStateTransitionError(err error) bool {
_, ok := err.(*stateTransitionError)
return ok
}
func TestStoppedStateTransition(t *testing.T) {
s := &stoppedState{c: &linuxContainer{}}
valid := []containerState{
&stoppedState{},
&runningState{},
&restoredState{},
}
for _, v := range valid {
if err := s.transition(v); err != nil {
t.Fatal(err)
}
}
err := s.transition(&pausedState{})
if err == nil {
t.Fatal("transition to paused state should fail")
}
if !isStateTransitionError(err) {
t.Fatal("expected stateTransitionError")
}
}
func TestPausedStateTransition(t *testing.T) {
s := &pausedState{c: &linuxContainer{}}
valid := []containerState{
&pausedState{},
&runningState{},
&stoppedState{},
}
for _, v := range valid {
if err := s.transition(v); err != nil {
t.Fatal(err)
}
}
}
func TestRestoredStateTransition(t *testing.T) {
s := &restoredState{c: &linuxContainer{}}
valid := []containerState{
&stoppedState{},
&runningState{},
}
for _, v := range valid {
if err := s.transition(v); err != nil {
t.Fatal(err)
}
}
err := s.transition(&createdState{})
if err == nil {
t.Fatal("transition to created state should fail")
}
if !isStateTransitionError(err) {
t.Fatal("expected stateTransitionError")
}
}

View file

@ -3,6 +3,9 @@
package system
import (
"bufio"
"fmt"
"os"
"os/exec"
"syscall"
"unsafe"
@ -75,3 +78,37 @@ func Setctty() error {
}
return nil
}
/*
* Detect whether we are currently running in a user namespace.
* Copied from github.com/lxc/lxd/shared/util.go
*/
func RunningInUserNS() bool {
file, err := os.Open("/proc/self/uid_map")
if err != nil {
/*
* This kernel-provided file only exists if user namespaces are
* supported
*/
return false
}
defer file.Close()
buf := bufio.NewReader(file)
l, _, err := buf.ReadLine()
if err != nil {
return false
}
line := string(l)
var a, b, c int64
fmt.Sscanf(line, "%d %d %d", &a, &b, &c)
/*
* We assume we are in the initial user namespace if we have a full
* range - 4294967295 uids starting at uid 0.
*/
if a == 0 && b == 0 && c == 4294967295 {
return false
}
return true
}

View file

@ -0,0 +1,472 @@
package user
import (
"io"
"reflect"
"sort"
"strconv"
"strings"
"testing"
)
func TestUserParseLine(t *testing.T) {
var (
a, b string
c []string
d int
)
parseLine("", &a, &b)
if a != "" || b != "" {
t.Fatalf("a and b should be empty ('%v', '%v')", a, b)
}
parseLine("a", &a, &b)
if a != "a" || b != "" {
t.Fatalf("a should be 'a' and b should be empty ('%v', '%v')", a, b)
}
parseLine("bad boys:corny cows", &a, &b)
if a != "bad boys" || b != "corny cows" {
t.Fatalf("a should be 'bad boys' and b should be 'corny cows' ('%v', '%v')", a, b)
}
parseLine("", &c)
if len(c) != 0 {
t.Fatalf("c should be empty (%#v)", c)
}
parseLine("d,e,f:g:h:i,j,k", &c, &a, &b, &c)
if a != "g" || b != "h" || len(c) != 3 || c[0] != "i" || c[1] != "j" || c[2] != "k" {
t.Fatalf("a should be 'g', b should be 'h', and c should be ['i','j','k'] ('%v', '%v', '%#v')", a, b, c)
}
parseLine("::::::::::", &a, &b, &c)
if a != "" || b != "" || len(c) != 0 {
t.Fatalf("a, b, and c should all be empty ('%v', '%v', '%#v')", a, b, c)
}
parseLine("not a number", &d)
if d != 0 {
t.Fatalf("d should be 0 (%v)", d)
}
parseLine("b:12:c", &a, &d, &b)
if a != "b" || b != "c" || d != 12 {
t.Fatalf("a should be 'b' and b should be 'c', and d should be 12 ('%v', '%v', %v)", a, b, d)
}
}
func TestUserParsePasswd(t *testing.T) {
users, err := ParsePasswdFilter(strings.NewReader(`
root:x:0:0:root:/root:/bin/bash
adm:x:3:4:adm:/var/adm:/bin/false
this is just some garbage data
`), nil)
if err != nil {
t.Fatalf("Unexpected error: %v", err)
}
if len(users) != 3 {
t.Fatalf("Expected 3 users, got %v", len(users))
}
if users[0].Uid != 0 || users[0].Name != "root" {
t.Fatalf("Expected users[0] to be 0 - root, got %v - %v", users[0].Uid, users[0].Name)
}
if users[1].Uid != 3 || users[1].Name != "adm" {
t.Fatalf("Expected users[1] to be 3 - adm, got %v - %v", users[1].Uid, users[1].Name)
}
}
func TestUserParseGroup(t *testing.T) {
groups, err := ParseGroupFilter(strings.NewReader(`
root:x:0:root
adm:x:4:root,adm,daemon
this is just some garbage data
`), nil)
if err != nil {
t.Fatalf("Unexpected error: %v", err)
}
if len(groups) != 3 {
t.Fatalf("Expected 3 groups, got %v", len(groups))
}
if groups[0].Gid != 0 || groups[0].Name != "root" || len(groups[0].List) != 1 {
t.Fatalf("Expected groups[0] to be 0 - root - 1 member, got %v - %v - %v", groups[0].Gid, groups[0].Name, len(groups[0].List))
}
if groups[1].Gid != 4 || groups[1].Name != "adm" || len(groups[1].List) != 3 {
t.Fatalf("Expected groups[1] to be 4 - adm - 3 members, got %v - %v - %v", groups[1].Gid, groups[1].Name, len(groups[1].List))
}
}
func TestValidGetExecUser(t *testing.T) {
const passwdContent = `
root:x:0:0:root user:/root:/bin/bash
adm:x:42:43:adm:/var/adm:/bin/false
this is just some garbage data
`
const groupContent = `
root:x:0:root
adm:x:43:
grp:x:1234:root,adm
this is just some garbage data
`
defaultExecUser := ExecUser{
Uid: 8888,
Gid: 8888,
Sgids: []int{8888},
Home: "/8888",
}
tests := []struct {
ref string
expected ExecUser
}{
{
ref: "root",
expected: ExecUser{
Uid: 0,
Gid: 0,
Sgids: []int{0, 1234},
Home: "/root",
},
},
{
ref: "adm",
expected: ExecUser{
Uid: 42,
Gid: 43,
Sgids: []int{1234},
Home: "/var/adm",
},
},
{
ref: "root:adm",
expected: ExecUser{
Uid: 0,
Gid: 43,
Sgids: defaultExecUser.Sgids,
Home: "/root",
},
},
{
ref: "adm:1234",
expected: ExecUser{
Uid: 42,
Gid: 1234,
Sgids: defaultExecUser.Sgids,
Home: "/var/adm",
},
},
{
ref: "42:1234",
expected: ExecUser{
Uid: 42,
Gid: 1234,
Sgids: defaultExecUser.Sgids,
Home: "/var/adm",
},
},
{
ref: "1337:1234",
expected: ExecUser{
Uid: 1337,
Gid: 1234,
Sgids: defaultExecUser.Sgids,
Home: defaultExecUser.Home,
},
},
{
ref: "1337",
expected: ExecUser{
Uid: 1337,
Gid: defaultExecUser.Gid,
Sgids: defaultExecUser.Sgids,
Home: defaultExecUser.Home,
},
},
{
ref: "",
expected: ExecUser{
Uid: defaultExecUser.Uid,
Gid: defaultExecUser.Gid,
Sgids: defaultExecUser.Sgids,
Home: defaultExecUser.Home,
},
},
}
for _, test := range tests {
passwd := strings.NewReader(passwdContent)
group := strings.NewReader(groupContent)
execUser, err := GetExecUser(test.ref, &defaultExecUser, passwd, group)
if err != nil {
t.Logf("got unexpected error when parsing '%s': %s", test.ref, err.Error())
t.Fail()
continue
}
if !reflect.DeepEqual(test.expected, *execUser) {
t.Logf("got: %#v", execUser)
t.Logf("expected: %#v", test.expected)
t.Fail()
continue
}
}
}
func TestInvalidGetExecUser(t *testing.T) {
const passwdContent = `
root:x:0:0:root user:/root:/bin/bash
adm:x:42:43:adm:/var/adm:/bin/false
this is just some garbage data
`
const groupContent = `
root:x:0:root
adm:x:43:
grp:x:1234:root,adm
this is just some garbage data
`
tests := []string{
// No such user/group.
"notuser",
"notuser:notgroup",
"root:notgroup",
"notuser:adm",
"8888:notgroup",
"notuser:8888",
// Invalid user/group values.
"-1:0",
"0:-3",
"-5:-2",
}
for _, test := range tests {
passwd := strings.NewReader(passwdContent)
group := strings.NewReader(groupContent)
execUser, err := GetExecUser(test, nil, passwd, group)
if err == nil {
t.Logf("got unexpected success when parsing '%s': %#v", test, execUser)
t.Fail()
continue
}
}
}
func TestGetExecUserNilSources(t *testing.T) {
const passwdContent = `
root:x:0:0:root user:/root:/bin/bash
adm:x:42:43:adm:/var/adm:/bin/false
this is just some garbage data
`
const groupContent = `
root:x:0:root
adm:x:43:
grp:x:1234:root,adm
this is just some garbage data
`
defaultExecUser := ExecUser{
Uid: 8888,
Gid: 8888,
Sgids: []int{8888},
Home: "/8888",
}
tests := []struct {
ref string
passwd, group bool
expected ExecUser
}{
{
ref: "",
passwd: false,
group: false,
expected: ExecUser{
Uid: 8888,
Gid: 8888,
Sgids: []int{8888},
Home: "/8888",
},
},
{
ref: "root",
passwd: true,
group: false,
expected: ExecUser{
Uid: 0,
Gid: 0,
Sgids: []int{8888},
Home: "/root",
},
},
{
ref: "0",
passwd: false,
group: false,
expected: ExecUser{
Uid: 0,
Gid: 8888,
Sgids: []int{8888},
Home: "/8888",
},
},
{
ref: "0:0",
passwd: false,
group: false,
expected: ExecUser{
Uid: 0,
Gid: 0,
Sgids: []int{8888},
Home: "/8888",
},
},
}
for _, test := range tests {
var passwd, group io.Reader
if test.passwd {
passwd = strings.NewReader(passwdContent)
}
if test.group {
group = strings.NewReader(groupContent)
}
execUser, err := GetExecUser(test.ref, &defaultExecUser, passwd, group)
if err != nil {
t.Logf("got unexpected error when parsing '%s': %s", test.ref, err.Error())
t.Fail()
continue
}
if !reflect.DeepEqual(test.expected, *execUser) {
t.Logf("got: %#v", execUser)
t.Logf("expected: %#v", test.expected)
t.Fail()
continue
}
}
}
func TestGetAdditionalGroups(t *testing.T) {
const groupContent = `
root:x:0:root
adm:x:43:
grp:x:1234:root,adm
adm:x:4343:root,adm-duplicate
this is just some garbage data
`
tests := []struct {
groups []string
expected []int
hasError bool
}{
{
// empty group
groups: []string{},
expected: []int{},
},
{
// single group
groups: []string{"adm"},
expected: []int{43},
},
{
// multiple groups
groups: []string{"adm", "grp"},
expected: []int{43, 1234},
},
{
// invalid group
groups: []string{"adm", "grp", "not-exist"},
expected: nil,
hasError: true,
},
{
// group with numeric id
groups: []string{"43"},
expected: []int{43},
},
{
// group with unknown numeric id
groups: []string{"adm", "10001"},
expected: []int{43, 10001},
},
{
// groups specified twice with numeric and name
groups: []string{"adm", "43"},
expected: []int{43},
},
{
// groups with too small id
groups: []string{"-1"},
expected: nil,
hasError: true,
},
{
// groups with too large id
groups: []string{strconv.Itoa(1 << 31)},
expected: nil,
hasError: true,
},
}
for _, test := range tests {
group := strings.NewReader(groupContent)
gids, err := GetAdditionalGroups(test.groups, group)
if test.hasError && err == nil {
t.Errorf("Parse(%#v) expects error but has none", test)
continue
}
if !test.hasError && err != nil {
t.Errorf("Parse(%#v) has error %v", test, err)
continue
}
sort.Sort(sort.IntSlice(gids))
if !reflect.DeepEqual(gids, test.expected) {
t.Errorf("Gids(%v), expect %v from groups %v", gids, test.expected, test.groups)
}
}
}
func TestGetAdditionalGroupsNumeric(t *testing.T) {
tests := []struct {
groups []string
expected []int
hasError bool
}{
{
// numeric groups only
groups: []string{"1234", "5678"},
expected: []int{1234, 5678},
},
{
// numeric and alphabetic
groups: []string{"1234", "fake"},
expected: nil,
hasError: true,
},
}
for _, test := range tests {
gids, err := GetAdditionalGroups(test.groups, nil)
if test.hasError && err == nil {
t.Errorf("Parse(%#v) expects error but has none", test)
continue
}
if !test.hasError && err != nil {
t.Errorf("Parse(%#v) has error %v", test, err)
continue
}
sort.Sort(sort.IntSlice(gids))
if !reflect.DeepEqual(gids, test.expected) {
t.Errorf("Gids(%v), expect %v from groups %v", gids, test.expected, test.groups)
}
}
}

View file

@ -3,7 +3,9 @@ package utils
import (
"crypto/rand"
"encoding/hex"
"encoding/json"
"io"
"os"
"path/filepath"
"syscall"
)
@ -36,10 +38,44 @@ func ResolveRootfs(uncleanRootfs string) (string, error) {
}
// ExitStatus returns the correct exit status for a process based on if it
// was signaled or exited cleanly.
// was signaled or exited cleanly
func ExitStatus(status syscall.WaitStatus) int {
if status.Signaled() {
return exitSignalOffset + int(status.Signal())
}
return status.ExitStatus()
}
// WriteJSON writes the provided struct v to w using standard json marshaling
func WriteJSON(w io.Writer, v interface{}) error {
data, err := json.Marshal(v)
if err != nil {
return err
}
_, err = w.Write(data)
return err
}
// CleanPath makes a path safe for use with filepath.Join. This is done by not
// only cleaning the path, but also (if the path is relative) adding a leading
// '/' and cleaning it (then removing the leading '/'). This ensures that a
// path resulting from prepending another path will always resolve to lexically
// be a subdirectory of the prefixed path. This is all done lexically, so paths
// that include symlinks won't be safe as a result of using CleanPath.
func CleanPath(path string) string {
// Ensure that all paths are cleaned (especially problematic ones like
// "/../../../../../" which can cause lots of issues).
path = filepath.Clean(path)
// If the path isn't absolute, we need to do more processing to fix paths
// such as "../../../../<etc>/some/path". We also shouldn't convert absolute
// paths to relative ones.
if !filepath.IsAbs(path) {
path = filepath.Clean(string(os.PathSeparator) + path)
// This can't fail, as (by definition) all paths are relative to root.
path, _ = filepath.Rel(string(os.PathSeparator), path)
}
// Clean the path again for good measure.
return filepath.Clean(path)
}

View file

@ -0,0 +1,25 @@
package utils
import "testing"
func TestGenerateName(t *testing.T) {
name, err := GenerateRandomName("veth", 5)
if err != nil {
t.Fatal(err)
}
expected := 5 + len("veth")
if len(name) != expected {
t.Fatalf("expected name to be %d chars but received %d", expected, len(name))
}
name, err = GenerateRandomName("veth", 65)
if err != nil {
t.Fatal(err)
}
expected = 64 + len("veth")
if len(name) != expected {
t.Fatalf("expected name to be %d chars but received %d", expected, len(name))
}
}

View file

@ -0,0 +1,8 @@
package xattr
import (
"fmt"
"runtime"
)
var ErrNotSupportedPlatform = fmt.Errorf("platform and architecture is not supported %s %s", runtime.GOOS, runtime.GOARCH)

View file

@ -0,0 +1,53 @@
// +build linux
package xattr
import (
"syscall"
"github.com/opencontainers/runc/libcontainer/system"
)
func XattrEnabled(path string) bool {
if Setxattr(path, "user.test", "") == syscall.ENOTSUP {
return false
}
return true
}
func stringsfromByte(buf []byte) (result []string) {
offset := 0
for index, b := range buf {
if b == 0 {
result = append(result, string(buf[offset:index]))
offset = index + 1
}
}
return
}
func Listxattr(path string) ([]string, error) {
size, err := system.Llistxattr(path, nil)
if err != nil {
return nil, err
}
buf := make([]byte, size)
read, err := system.Llistxattr(path, buf)
if err != nil {
return nil, err
}
names := stringsfromByte(buf[:read])
return names, nil
}
func Getxattr(path, attr string) (string, error) {
value, err := system.Lgetxattr(path, attr)
if err != nil {
return "", err
}
return string(value), nil
}
func Setxattr(path, xattr, value string) error {
return system.Lsetxattr(path, xattr, []byte(value), 0)
}

View file

@ -0,0 +1,78 @@
// +build linux
package xattr_test
import (
"os"
"testing"
"github.com/opencontainers/runc/libcontainer/xattr"
)
func TestXattr(t *testing.T) {
tmp := "xattr_test"
out, err := os.OpenFile(tmp, os.O_WRONLY|os.O_CREATE, 0)
if err != nil {
t.Fatal("failed")
}
defer os.Remove(tmp)
attr := "user.test"
out.Close()
if !xattr.XattrEnabled(tmp) {
t.Log("Disabled")
t.Fatal("failed")
}
t.Log("Success")
err = xattr.Setxattr(tmp, attr, "test")
if err != nil {
t.Fatal("failed")
}
var value string
value, err = xattr.Getxattr(tmp, attr)
if err != nil {
t.Fatal("failed")
}
if value != "test" {
t.Fatal("failed")
}
t.Log("Success")
var names []string
names, err = xattr.Listxattr(tmp)
if err != nil {
t.Fatal("failed")
}
var found int
for _, name := range names {
if name == attr {
found = 1
}
}
// Listxattr doesn't return trusted.* and system.* namespace
// attrs when run in unprevileged mode.
if found != 1 {
t.Fatal("failed")
}
t.Log("Success")
big := "0000000000000000000000000000000000000000000000000000000000000000000008c6419ad822dfe29283fb3ac98dcc5908810cb31f4cfe690040c42c144b7492eicompslf20dxmlpgz"
// Test for long xattrs larger than 128 bytes
err = xattr.Setxattr(tmp, attr, big)
if err != nil {
t.Fatal("failed to add long value")
}
value, err = xattr.Getxattr(tmp, attr)
if err != nil {
t.Fatal("failed to get long value")
}
t.Log("Success")
if value != big {
t.Fatal("failed, value doesn't match")
}
t.Log("Success")
}

View file

@ -0,0 +1,15 @@
// +build !linux
package xattr
func Listxattr(path string) ([]string, error) {
return nil, ErrNotSupportedPlatform
}
func Getxattr(path, attr string) (string, error) {
return "", ErrNotSupportedPlatform
}
func Setxattr(path, xattr, value string) error {
return ErrNotSupportedPlatform
}