specify cgroup ownership semantics

cgroups v2 supports secure delegation of cgroups.  Accordingly,
control over a cgroup (that is, creation of new child cgroups and
movement of processes and threads among the cgroup subtree exposed
to a container) can be safely delegated to a container.  Adjusting
the ownership enables real-world use cases like systemd-based
containers fully isolated in user namespaces.

To encourage adoption of this feature, and secure implementation,
define the semantics of cgroup ownership.  Changing/setting the
cgroup ownership should only be performed when:

- using cgroups v2, and
- container will have a new cgroup namespace, and
- cgroupfs will be mounted read/write.

The specific files whose ownership should be changed are listed.

In terms of current practice, this is already the behaviour of crun
(which also chown's the memory.oom.group file), and there is a pull
request for runc: opencontainers/runc#3057.

Signed-off-by: Fraser Tweedale <ftweedal@redhat.com>

config-linux.md

@@ -196,6 +196,44 @@ For example, to run a new process in an existing container without updating limi
 
 Runtimes MAY attach the container process to additional cgroup controllers beyond those necessary to fulfill the `resources` settings.
 
+### Cgroup ownership
+
+Runtimes MAY change (or cause to be changed) the owner of the
+container's cgroup to the host uid that maps to uid 0 in the
+container's user namespace, according to the following rules.
+
+Runtimes SHOULD NOT change the ownership of container cgroups when
+cgroups v1 is in use. Cgroup delegation is not secure in cgroups
+v1.
+
+A runtime SHOULD NOT change the ownership of a container cgroup
+unless it will also create a new cgroup namespace for the container.
+Typically this occurs when the `linux.namespaces` array contains an
+object with `type` equal to `"cgroup"` and `path` unset.
+
+Runtimes SHOULD change the cgroup ownership if and only if the
+cgroup filesystem is to be mounted read/write; that is, when the
+configuration's `mounts` array contains an object where:
+
+- The `source` field is equal to `"cgroup"`
+- The `destination` field is equal to `"/sys/fs/cgroup"`
+- The `options` field does not contain the value `"ro"`
+
+If the configuration does not specify such a mount, the runtime
+SHOULD NOT change the cgroup ownership.
+
+A runtime that changes the cgroup ownership SHOULD only change the
+ownership of the container's cgroup directory and the following
+files within that directory:
+
+- `cgroup.procs`
+- `cgroup.subtree_control`
+- `cgroup.threads`
+
+The runtime SHOULD NOT change the ownership of any other files.
+Changing other files may allow the container to elevate its own
+resource limits or perform other unwanted behaviour.
+
 ### Example
 
 ```json