regression: entrypoint is broken on cgroup v2

[AkihiroSuda]

What happened:

fix_cgroup() change in https://github.com/kubernetes-sigs/kind/pull/1995/files broke cgroup v2

What you expected to happen:
It should work

How to reproduce it (as minimally and precisely as possible):

Minimal repro:

$ docker build -t kindest/base:latest ./images/base
$ docker run --rm --privileged kindest/base
stat: cannot read file system information for '/kind': No such file or directory
INFO: ensuring we can execute mount/umount even with userns-remap
INFO: remounting /sys read-only
INFO: making mounts shared
INFO: fix cgroup mounts for all subsystems
$ echo $?
1

Environment:

• kind version: (use kind version): 26637e2 (Jan 19, 2021)
• Ubuntu 20.10, cgroup v2
• moby/moby@0456e058d2, containerd/containerd@092f9e6, opencontainers/runc@1a90116

[BenTheElder]

perhaps we should check for v1/v2 and split the code-path here?
we don't have v2 in CI and this change fixed a number of v1 platforms.

[BenTheElder]

I'm also actually not sure if Kubernetes even has any cgroupsv2 CI at all yet, last I checked someone decided that meant we had to use fedora VMs ... which we didn't have a GCE VM image handy for. AFAIK SIG Node never sorted that out.

EDIT: which is not to say kind shouldn't be fixed, but rather that we probably also need to look at getting upstream CI as well still.

[AkihiroSuda]

PR: #2014