go:mod: runtime-spec v1.1.0-rc.1; add docs/spec-conformance.md

[AkihiroSuda]

runc now implements the OCI Runtime Spec v1.1.0-rc.1 for the linux platform.

The following features are not implemented yet:

Spec version	Feature	PR
v1.0.0	SCMP_ARCH_PARISC	Unplanned, due to lack of users
v1.0.0	SCMP_ARCH_PARISC64	Unplanned, due to lack of users
v1.0.2	.linux.personality	#3126
v1.1.0-rc.1	.linux.resources.cpu.burst	#3205
v1.1.0-rc.1	.domainname	#3600
v1.1.0-rc.1	.[]mounts.uidMappings	#3717
v1.1.0-rc.1	.[]mounts.gidMappings	#3717
v1.1.0-rc.1	SECCOMP_FILTER_FLAG_WAIT_KILLABLE_RECV	TODO

[AkihiroSuda]

Let's release runc v1.2.0-rc.1 after merging this
@opencontainers/runc-maintainers

[kolyshkin]

LGTM

[AkihiroSuda]

Rebased, as #3377 was merged

[thaJeztah]

LGTM

[AkihiroSuda]

Rebased

[AkihiroSuda]

Rerebased

[kolyshkin]

Failure in centos-stream-9 CI:

=== RUN   TestPodSkipDevicesUpdate
    systemd_test.go:114: container stderr not empty: /bin/basename: missing operand
        Try '/bin/basename --help' for more information.
--- FAIL: TestPodSkipDevicesUpdate (0.14s)

Hmm, I have absolutely no idea where the "basename" is coming from in this test. Most probably it's from some bashrc on the testing host (in this case we run bash from the host inside the cgroup, see

runc/libcontainer/cgroups/devices/systemd_test.go

Line 162 in 5c2a1a0

cmd := exec.Command("bash", "-c", "read; echo > /dev/full; cat /dev/null; true")

). We can maybe add --norc to work around this, but in general this should not ever happen.

Update: ah, this is centos-stream-9, perhaps something was broken in there.

[AkihiroSuda]

Restarted the CI and it is now green

[rata]

Update: ah, this is centos-stream-9, perhaps something was broken in there.

I've seen this error yesterday too :-/ I completely forgot about it.

[AkihiroSuda]

@kolyshkin @thaJeztah Lost your LGTMs after rebasing, could you LGTM again 🙏

[thaJeztah]

LGTM

[kolyshkin]

still LGTM