mirror of
https://github.com/opencontainers/runc.git
synced 2025-11-02 20:04:01 +08:00
With open_tree(OPEN_TREE_CLONE), it is possible to implement both the id-mapped mounts and bind-mount source file descriptor logic entirely in Go without requiring any complicated handling from nsexec. However, implementing it the naive way (do the OPEN_TREE_CLONE in the host namespace before the rootfs is set up -- which is what the existing implementation did) exposes issues in how mount ordering (in particular when handling mount sources from inside the container rootfs, but also in relation to mount propagation) was handled for idmapped mounts and bind-mount sources. In order to solve this problem completely, it is necessary to spawn a thread which joins the container mount namespace and provides mountfds when requested by the rootfs setup code (ensuring that the mount order and mount propagation of the source of the bind-mount are handled correctly). While the need to join the mount namespace leads to other complicated (such as with the usage of /proc/self -- fixed in a later patch) the resulting code is still reasonable and is the only real way to solve the issue. This allows us to reduce the amount of C code we have in nsexec, as well as simplifying a whole host of places that were made more complicated with the addition of id-mapped mounts and the bind sourcefd logic. Because we join the container namespace, we can continue to use regular O_PATH file descriptors for non-id-mapped bind-mount sources (which means we don't have to raise the kernel requirement for that case). In addition, we can easily add support for id-mappings that don't match the container's user namespace. The approach taken here is to use Go's officially supported mechanism for spawning a process in a user namespace, but (ab)use PTRACE_TRACEME to avoid actually having to exec a different process. The most efficient way to implement this would be to do clone() in cgo directly to run a function that just does kill(getpid(), SIGSTOP) -- we can always switch to that if it turns out this approach is too slow. It should be noted that the included micro-benchmark seems to indicate this is Fast Enough(TM): goos: linux goarch: amd64 pkg: github.com/opencontainers/runc/libcontainer/userns cpu: Intel(R) Core(TM) i5-10210U CPU @ 1.60GHz BenchmarkSpawnProc BenchmarkSpawnProc-8 1670 770065 ns/op Fixes:fda12ab101("Support idmap mounts on volumes") Fixes:9c444070ec("Open bind mount sources from the host userns") Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
217 lines
6.0 KiB
Go
217 lines
6.0 KiB
Go
package libcontainer
|
|
|
|
import (
|
|
"encoding/json"
|
|
"errors"
|
|
"fmt"
|
|
"os"
|
|
|
|
securejoin "github.com/cyphar/filepath-securejoin"
|
|
"golang.org/x/sys/unix"
|
|
|
|
//nolint:revive // Enable cgroup manager to manage devices
|
|
_ "github.com/opencontainers/runc/libcontainer/cgroups/devices"
|
|
"github.com/opencontainers/runc/libcontainer/cgroups/manager"
|
|
"github.com/opencontainers/runc/libcontainer/configs"
|
|
"github.com/opencontainers/runc/libcontainer/configs/validate"
|
|
"github.com/opencontainers/runc/libcontainer/intelrdt"
|
|
"github.com/opencontainers/runc/libcontainer/utils"
|
|
)
|
|
|
|
const (
|
|
stateFilename = "state.json"
|
|
execFifoFilename = "exec.fifo"
|
|
)
|
|
|
|
// Create creates a new container with the given id inside a given state
|
|
// directory (root), and returns a Container object.
|
|
//
|
|
// The root is a state directory which many containers can share. It can be
|
|
// used later to get the list of containers, or to get information about a
|
|
// particular container (see Load).
|
|
//
|
|
// The id must not be empty and consist of only the following characters:
|
|
// ASCII letters, digits, underscore, plus, minus, period. The id must be
|
|
// unique and non-existent for the given root path.
|
|
func Create(root, id string, config *configs.Config) (*Container, error) {
|
|
if root == "" {
|
|
return nil, errors.New("root not set")
|
|
}
|
|
if err := validateID(id); err != nil {
|
|
return nil, err
|
|
}
|
|
if err := validate.Validate(config); err != nil {
|
|
return nil, err
|
|
}
|
|
if err := os.MkdirAll(root, 0o700); err != nil {
|
|
return nil, err
|
|
}
|
|
stateDir, err := securejoin.SecureJoin(root, id)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
if _, err := os.Stat(stateDir); err == nil {
|
|
return nil, ErrExist
|
|
} else if !os.IsNotExist(err) {
|
|
return nil, err
|
|
}
|
|
|
|
cm, err := manager.New(config.Cgroups)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
// Check that cgroup does not exist or empty (no processes).
|
|
// Note for cgroup v1 this check is not thorough, as there are multiple
|
|
// separate hierarchies, while both Exists() and GetAllPids() only use
|
|
// one for "devices" controller (assuming others are the same, which is
|
|
// probably true in almost all scenarios). Checking all the hierarchies
|
|
// would be too expensive.
|
|
if cm.Exists() {
|
|
pids, err := cm.GetAllPids()
|
|
// Reading PIDs can race with cgroups removal, so ignore ENOENT and ENODEV.
|
|
if err != nil && !errors.Is(err, os.ErrNotExist) && !errors.Is(err, unix.ENODEV) {
|
|
return nil, fmt.Errorf("unable to get cgroup PIDs: %w", err)
|
|
}
|
|
if len(pids) != 0 {
|
|
return nil, fmt.Errorf("container's cgroup is not empty: %d process(es) found", len(pids))
|
|
}
|
|
}
|
|
|
|
// Check that cgroup is not frozen. Do not use Exists() here
|
|
// since in cgroup v1 it only checks "devices" controller.
|
|
st, err := cm.GetFreezerState()
|
|
if err != nil {
|
|
return nil, fmt.Errorf("unable to get cgroup freezer state: %w", err)
|
|
}
|
|
if st == configs.Frozen {
|
|
return nil, errors.New("container's cgroup unexpectedly frozen")
|
|
}
|
|
|
|
// Parent directory is already created above, so Mkdir is enough.
|
|
if err := os.Mkdir(stateDir, 0o711); err != nil {
|
|
return nil, err
|
|
}
|
|
c := &Container{
|
|
id: id,
|
|
stateDir: stateDir,
|
|
config: config,
|
|
cgroupManager: cm,
|
|
intelRdtManager: intelrdt.NewManager(config, id, ""),
|
|
}
|
|
c.state = &stoppedState{c: c}
|
|
return c, nil
|
|
}
|
|
|
|
// Load takes a path to the state directory (root) and an id of an existing
|
|
// container, and returns a Container object reconstructed from the saved
|
|
// state. This presents a read only view of the container.
|
|
func Load(root, id string) (*Container, error) {
|
|
if root == "" {
|
|
return nil, errors.New("root not set")
|
|
}
|
|
// when load, we need to check id is valid or not.
|
|
if err := validateID(id); err != nil {
|
|
return nil, err
|
|
}
|
|
stateDir, err := securejoin.SecureJoin(root, id)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
state, err := loadState(stateDir)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
r := &nonChildProcess{
|
|
processPid: state.InitProcessPid,
|
|
processStartTime: state.InitProcessStartTime,
|
|
fds: state.ExternalDescriptors,
|
|
}
|
|
cm, err := manager.NewWithPaths(state.Config.Cgroups, state.CgroupPaths)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
c := &Container{
|
|
initProcess: r,
|
|
initProcessStartTime: state.InitProcessStartTime,
|
|
id: id,
|
|
config: &state.Config,
|
|
cgroupManager: cm,
|
|
intelRdtManager: intelrdt.NewManager(&state.Config, id, state.IntelRdtPath),
|
|
stateDir: stateDir,
|
|
created: state.Created,
|
|
}
|
|
c.state = &loadedState{c: c}
|
|
if err := c.refreshState(); err != nil {
|
|
return nil, err
|
|
}
|
|
return c, nil
|
|
}
|
|
|
|
func loadState(root string) (*State, error) {
|
|
stateFilePath, err := securejoin.SecureJoin(root, stateFilename)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
f, err := os.Open(stateFilePath)
|
|
if err != nil {
|
|
if os.IsNotExist(err) {
|
|
return nil, ErrNotExist
|
|
}
|
|
return nil, err
|
|
}
|
|
defer f.Close()
|
|
var state *State
|
|
if err := json.NewDecoder(f).Decode(&state); err != nil {
|
|
return nil, err
|
|
}
|
|
return state, nil
|
|
}
|
|
|
|
// validateID checks if the supplied container ID is valid, returning
|
|
// the ErrInvalidID in case it is not.
|
|
//
|
|
// The format of valid ID was never formally defined, instead the code
|
|
// was modified to allow or disallow specific characters.
|
|
//
|
|
// Currently, a valid ID is a non-empty string consisting only of
|
|
// the following characters:
|
|
// - uppercase (A-Z) and lowercase (a-z) Latin letters;
|
|
// - digits (0-9);
|
|
// - underscore (_);
|
|
// - plus sign (+);
|
|
// - minus sign (-);
|
|
// - period (.).
|
|
//
|
|
// In addition, IDs that can't be used to represent a file name
|
|
// (such as . or ..) are rejected.
|
|
|
|
func validateID(id string) error {
|
|
if len(id) < 1 {
|
|
return ErrInvalidID
|
|
}
|
|
|
|
// Allowed characters: 0-9 A-Z a-z _ + - .
|
|
for i := 0; i < len(id); i++ {
|
|
c := id[i]
|
|
switch {
|
|
case c >= 'a' && c <= 'z':
|
|
case c >= 'A' && c <= 'Z':
|
|
case c >= '0' && c <= '9':
|
|
case c == '_':
|
|
case c == '+':
|
|
case c == '-':
|
|
case c == '.':
|
|
default:
|
|
return ErrInvalidID
|
|
}
|
|
|
|
}
|
|
|
|
if string(os.PathSeparator)+id != utils.CleanPath(string(os.PathSeparator)+id) {
|
|
return ErrInvalidID
|
|
}
|
|
|
|
return nil
|
|
}
|