Files
runc/libcontainer/utils/utils.go
Aleksa Sarai 8e8b136c49 tree-wide: use /proc/thread-self for thread-local state
With the idmap work, we will have a tainted Go thread in our
thread-group that has a different mount namespace to the other threads.
It seems that (due to some bad luck) the Go scheduler tends to make this
thread the thread-group leader in our tests, which results in very
baffling failures where /proc/self/mountinfo produces gibberish results.

In order to avoid this, switch to using /proc/thread-self for everything
that is thread-local. This primarily includes switching all file
descriptor paths (CLONE_FS), all of the places that check the current
cgroup (technically we never will run a single runc thread in a separate
cgroup, but better to be safe than sorry), and the aforementioned
mountinfo code. We don't need to do anything for the following because
the results we need aren't thread-local:

 * Checks that certain namespaces are supported by stat(2)ing
   /proc/self/ns/...

 * /proc/self/exe and /proc/self/cmdline are not thread-local.

 * While threads can be in different cgroups, we do not do this for the
   runc binary (or libcontainer) and thus we do not need to switch to
   the thread-local version of /proc/self/cgroups.

 * All of the CLONE_NEWUSER files are not thread-local because you
   cannot set the usernamespace of a single thread (setns(CLONE_NEWUSER)
   is blocked for multi-threaded programs).

Note that we have to use runtime.LockOSThread when we have an open
handle to a tid-specific procfs file that we are operating on multiple
times. Go can reschedule us such that we are running on a different
thread and then kill the original thread (causing -ENOENT or similarly
confusing errors). This is not strictly necessary for most usages of
/proc/thread-self (such as using /proc/thread-self/fd/$n directly) since
only operating on the actual inodes associated with the tid requires
this locking, but because of the pre-3.17 fallback for CentOS, we have
to do this in most cases.

In addition, CentOS's kernel is too old for /proc/thread-self, which
requires us to emulate it -- however in rootfs_linux.go, we are in the
container pid namespace but /proc is the host's procfs. This leads to
the incredibly frustrating situation where there is no way (on pre-4.1
Linux) to figure out which /proc/self/task/... entry refers to the
current tid. We can just use /proc/self in this case.

Yes this is all pretty ugly. I also wish it wasn't necessary.

Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
2023-12-14 11:36:41 +11:00

132 lines
3.8 KiB
Go

package utils
import (
"encoding/binary"
"encoding/json"
"io"
"os"
"path/filepath"
"strings"
"unsafe"
"golang.org/x/sys/unix"
)
const (
exitSignalOffset = 128
)
// NativeEndian is the native byte order of the host system.
var NativeEndian binary.ByteOrder
func init() {
// Copied from <golang.org/x/net/internal/socket/sys.go>.
i := uint32(1)
b := (*[4]byte)(unsafe.Pointer(&i))
if b[0] == 1 {
NativeEndian = binary.LittleEndian
} else {
NativeEndian = binary.BigEndian
}
}
// ExitStatus returns the correct exit status for a process based on if it
// was signaled or exited cleanly
func ExitStatus(status unix.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
// without a trailing newline. This is used instead of json.Encoder because
// there might be a problem in json decoder in some cases, see:
// https://github.com/docker/docker/issues/14203#issuecomment-174177790
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 {
// Deal with empty strings nicely.
if path == "" {
return ""
}
// 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)
}
// stripRoot returns the passed path, stripping the root path if it was
// (lexicially) inside it. Note that both passed paths will always be treated
// as absolute, and the returned path will also always be absolute. In
// addition, the paths are cleaned before stripping the root.
func stripRoot(root, path string) string {
// Make the paths clean and absolute.
root, path = CleanPath("/"+root), CleanPath("/"+path)
switch {
case path == root:
path = "/"
case root == "/":
// do nothing
case strings.HasPrefix(path, root+"/"):
path = strings.TrimPrefix(path, root+"/")
}
return CleanPath("/" + path)
}
// SearchLabels searches through a list of key=value pairs for a given key,
// returning its value, and the binary flag telling whether the key exist.
func SearchLabels(labels []string, key string) (string, bool) {
key += "="
for _, s := range labels {
if strings.HasPrefix(s, key) {
return s[len(key):], true
}
}
return "", false
}
// Annotations returns the bundle path and user defined annotations from the
// libcontainer state. We need to remove the bundle because that is a label
// added by libcontainer.
func Annotations(labels []string) (bundle string, userAnnotations map[string]string) {
userAnnotations = make(map[string]string)
for _, l := range labels {
parts := strings.SplitN(l, "=", 2)
if len(parts) < 2 {
continue
}
if parts[0] == "bundle" {
bundle = parts[1]
} else {
userAnnotations[parts[0]] = parts[1]
}
}
return
}