moby--moby/daemon/runtime_unix.go

175 lines
6.1 KiB
Go
Raw Normal View History

//go:build !windows
// +build !windows
package daemon
import (
"fmt"
"os"
"os/exec"
"path/filepath"
"strings"
v2runcoptions "github.com/containerd/containerd/runtime/v2/runc/options"
daemon: support other containerd runtimes (MVP) Contrary to popular belief, the OCI Runtime specification does not specify the command-line API for runtimes. Looking at containerd's architecture from the lens of the OCI Runtime spec, the _shim_ is the OCI Runtime and runC is "just" an implementation detail of the io.containerd.runc.v2 runtime. When one configures a non-default runtime in Docker, what they're really doing is instructing Docker to create containers using the io.containerd.runc.v2 runtime with a configuration option telling the runtime that the runC binary is at some non-default path. Consequently, only OCI runtimes which are compatible with the io.containerd.runc.v2 shim, such as crun, can be used in this manner. Other OCI runtimes, including kata-containers v2, come with their own containerd shim and are not compatible with io.containerd.runc.v2. As Docker has not historically provided a way to select a non-default runtime which requires its own shim, runtimes such as kata-containers v2 could not be used with Docker. Allow other containerd shims to be used with Docker; no daemon configuration required. If the daemon is instructed to create a container with a runtime name which does not match any of the configured or stock runtimes, it passes the name along to containerd verbatim. A user can start a container with the kata-containers runtime, for example, simply by calling docker run --runtime io.containerd.kata.v2 Runtime names which containerd would interpret as a path to an arbitrary binary are disallowed. While handy for development and testing it is not strictly necessary and would allow anyone with Engine API access to trivially execute any binary on the host as root, so we have decided it would be safest for our users if it was not allowed. It is not yet possible to set an alternative containerd shim as the default runtime; it can only be configured per-container. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-07-20 20:12:01 +00:00
"github.com/containerd/containerd/runtime/v2/shim"
"github.com/docker/docker/api/types"
"github.com/docker/docker/daemon/config"
"github.com/docker/docker/errdefs"
"github.com/docker/docker/pkg/ioutils"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
)
const (
defaultRuntimeName = "runc"
linuxShimV2 = "io.containerd.runc.v2"
)
func configureRuntimes(conf *config.Config) {
if conf.DefaultRuntime == "" {
conf.DefaultRuntime = config.StockRuntimeName
}
if conf.Runtimes == nil {
conf.Runtimes = make(map[string]types.Runtime)
}
conf.Runtimes[config.LinuxV2RuntimeName] = types.Runtime{Path: defaultRuntimeName, Shim: defaultV2ShimConfig(conf, defaultRuntimeName)}
conf.Runtimes[config.StockRuntimeName] = conf.Runtimes[config.LinuxV2RuntimeName]
}
func defaultV2ShimConfig(conf *config.Config, runtimePath string) *types.ShimConfig {
return &types.ShimConfig{
Binary: linuxShimV2,
Opts: &v2runcoptions.Options{
BinaryName: runtimePath,
Root: filepath.Join(conf.ExecRoot, "runtime-"+defaultRuntimeName),
SystemdCgroup: UsingSystemd(conf),
NoPivotRoot: os.Getenv("DOCKER_RAMDISK") != "",
},
}
}
func (daemon *Daemon) loadRuntimes() error {
return daemon.initRuntimes(daemon.configStore.Runtimes)
}
func (daemon *Daemon) initRuntimes(runtimes map[string]types.Runtime) (err error) {
runtimeDir := filepath.Join(daemon.configStore.Root, "runtimes")
// Remove old temp directory if any
os.RemoveAll(runtimeDir + "-old")
tmpDir, err := ioutils.TempDir(daemon.configStore.Root, "gen-runtimes")
if err != nil {
return errors.Wrap(err, "failed to get temp dir to generate runtime scripts")
}
defer func() {
if err != nil {
if err1 := os.RemoveAll(tmpDir); err1 != nil {
logrus.WithError(err1).WithField("dir", tmpDir).
Warn("failed to remove tmp dir")
}
return
}
if err = os.Rename(runtimeDir, runtimeDir+"-old"); err != nil {
return
}
if err = os.Rename(tmpDir, runtimeDir); err != nil {
err = errors.Wrap(err, "failed to setup runtimes dir, new containers may not start")
return
}
if err = os.RemoveAll(runtimeDir + "-old"); err != nil {
logrus.WithError(err).WithField("dir", tmpDir).
Warn("failed to remove old runtimes dir")
}
}()
for name, rt := range runtimes {
if len(rt.Args) > 0 {
script := filepath.Join(tmpDir, name)
content := fmt.Sprintf("#!/bin/sh\n%s %s $@\n", rt.Path, strings.Join(rt.Args, " "))
if err := os.WriteFile(script, []byte(content), 0700); err != nil {
return err
}
}
if rt.Shim == nil {
rt.Shim = defaultV2ShimConfig(daemon.configStore, rt.Path)
}
}
return nil
}
// rewriteRuntimePath is used for runtimes which have custom arguments supplied.
// This is needed because the containerd API only calls the OCI runtime binary, there is no options for extra arguments.
// To support this case, the daemon wraps the specified runtime in a script that passes through those arguments.
func (daemon *Daemon) rewriteRuntimePath(name, p string, args []string) (string, error) {
if len(args) == 0 {
return p, nil
}
// Check that the runtime path actually exists here so that we can return a well known error.
if _, err := exec.LookPath(p); err != nil {
return "", errors.Wrap(err, "error while looking up the specified runtime path")
}
return filepath.Join(daemon.configStore.Root, "runtimes", name), nil
}
func (daemon *Daemon) getRuntime(name string) (*types.Runtime, error) {
rt := daemon.configStore.GetRuntime(name)
if rt == nil {
daemon: support other containerd runtimes (MVP) Contrary to popular belief, the OCI Runtime specification does not specify the command-line API for runtimes. Looking at containerd's architecture from the lens of the OCI Runtime spec, the _shim_ is the OCI Runtime and runC is "just" an implementation detail of the io.containerd.runc.v2 runtime. When one configures a non-default runtime in Docker, what they're really doing is instructing Docker to create containers using the io.containerd.runc.v2 runtime with a configuration option telling the runtime that the runC binary is at some non-default path. Consequently, only OCI runtimes which are compatible with the io.containerd.runc.v2 shim, such as crun, can be used in this manner. Other OCI runtimes, including kata-containers v2, come with their own containerd shim and are not compatible with io.containerd.runc.v2. As Docker has not historically provided a way to select a non-default runtime which requires its own shim, runtimes such as kata-containers v2 could not be used with Docker. Allow other containerd shims to be used with Docker; no daemon configuration required. If the daemon is instructed to create a container with a runtime name which does not match any of the configured or stock runtimes, it passes the name along to containerd verbatim. A user can start a container with the kata-containers runtime, for example, simply by calling docker run --runtime io.containerd.kata.v2 Runtime names which containerd would interpret as a path to an arbitrary binary are disallowed. While handy for development and testing it is not strictly necessary and would allow anyone with Engine API access to trivially execute any binary on the host as root, so we have decided it would be safest for our users if it was not allowed. It is not yet possible to set an alternative containerd shim as the default runtime; it can only be configured per-container. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-07-20 20:12:01 +00:00
if !isPermissibleC8dRuntimeName(name) {
return nil, errdefs.InvalidParameter(errors.Errorf("unknown or invalid runtime name: %s", name))
}
return &types.Runtime{Shim: &types.ShimConfig{Binary: name}}, nil
}
if len(rt.Args) > 0 {
p, err := daemon.rewriteRuntimePath(name, rt.Path, rt.Args)
if err != nil {
return nil, err
}
rt.Path = p
rt.Args = nil
}
if rt.Shim == nil {
rt.Shim = defaultV2ShimConfig(daemon.configStore, rt.Path)
}
return rt, nil
}
daemon: support other containerd runtimes (MVP) Contrary to popular belief, the OCI Runtime specification does not specify the command-line API for runtimes. Looking at containerd's architecture from the lens of the OCI Runtime spec, the _shim_ is the OCI Runtime and runC is "just" an implementation detail of the io.containerd.runc.v2 runtime. When one configures a non-default runtime in Docker, what they're really doing is instructing Docker to create containers using the io.containerd.runc.v2 runtime with a configuration option telling the runtime that the runC binary is at some non-default path. Consequently, only OCI runtimes which are compatible with the io.containerd.runc.v2 shim, such as crun, can be used in this manner. Other OCI runtimes, including kata-containers v2, come with their own containerd shim and are not compatible with io.containerd.runc.v2. As Docker has not historically provided a way to select a non-default runtime which requires its own shim, runtimes such as kata-containers v2 could not be used with Docker. Allow other containerd shims to be used with Docker; no daemon configuration required. If the daemon is instructed to create a container with a runtime name which does not match any of the configured or stock runtimes, it passes the name along to containerd verbatim. A user can start a container with the kata-containers runtime, for example, simply by calling docker run --runtime io.containerd.kata.v2 Runtime names which containerd would interpret as a path to an arbitrary binary are disallowed. While handy for development and testing it is not strictly necessary and would allow anyone with Engine API access to trivially execute any binary on the host as root, so we have decided it would be safest for our users if it was not allowed. It is not yet possible to set an alternative containerd shim as the default runtime; it can only be configured per-container. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-07-20 20:12:01 +00:00
// isPermissibleC8dRuntimeName tests whether name is safe to pass into
// containerd as a runtime name, and whether the name is well-formed.
// It does not check if the runtime is installed.
//
// A runtime name containing slash characters is interpreted by containerd as
// the path to a runtime binary. If we allowed this, anyone with Engine API
// access could get containerd to execute an arbitrary binary as root. Although
// Engine API access is already equivalent to root on the host, the runtime name
// has not historically been a vector to run arbitrary code as root so users are
// not expecting it to become one.
//
// This restriction is not configurable. There are viable workarounds for
// legitimate use cases: administrators and runtime developers can make runtimes
// available for use with Docker by installing them onto PATH following the
// [binary naming convention] for containerd Runtime v2.
//
// [binary naming convention]: https://github.com/containerd/containerd/blob/main/runtime/v2/README.md#binary-naming
func isPermissibleC8dRuntimeName(name string) bool {
// containerd uses a rather permissive test to validate runtime names:
//
// - Any name for which filepath.IsAbs(name) is interpreted as the absolute
// path to a shim binary. We want to block this behaviour.
// - Any name which contains at least one '.' character and no '/' characters
// and does not begin with a '.' character is a valid runtime name. The shim
// binary name is derived from the final two components of the name and
// searched for on the PATH. The name "a.." is technically valid per
// containerd's implementation: it would resolve to a binary named
// "containerd-shim---".
//
// https://github.com/containerd/containerd/blob/11ded166c15f92450958078cd13c6d87131ec563/runtime/v2/manager.go#L297-L317
// https://github.com/containerd/containerd/blob/11ded166c15f92450958078cd13c6d87131ec563/runtime/v2/shim/util.go#L83-L93
return !filepath.IsAbs(name) && !strings.ContainsRune(name, '/') && shim.BinaryName(name) != ""
}