Hand a container's cgroup to the user it runs as - #865
Closed
MayCXC wants to merge 4 commits into
Closed
Conversation
The reaper that runs at a concurrency of one deletes everything in the test directory except the unpacked rootfs it means to preserve. It held that rootfs path as FileManager reports it, under /var, and compared it against the entries of contentsOfDirectory, which reports them under /private/var, so the preserved path never matched any entry and the rootfs was deleted along with the per-test files. The unpack coordinator still held it as unpacked, so the next test opened a rootfs that was no longer there and failed with a missing file error. Resolve both sides with resolvingSymlinksInPathWithPrivate, which exists for this difference between the two views of the same directory.
Tests guarded for Linux are compiled out on macOS, so `make test` reports success without having run them, and nothing says so. The target that does run them is not mentioned anywhere outside the makefile.
The default kernel the tests fetch had drifted from the one users run. container installs kata 3.28.0 and its 6.18.15 kernel, while these tests fetched 3.17.0, so the suite exercised a guest with a different feature set to the one it is meant to represent. Nested runtimes are the visible case: 3.17.0 was built without nf_tables, so a docker daemon inside a container fails there and works on what ships. Kata moved from xz to zstd between those releases, so the archive is no longer named for its compression and tar is left to recognise it rather than being told, which also holds if the format changes again.
A container that runs as somebody other than root cannot use the cgroup it was placed in: the kernel keeps it for whoever owns the directory, so anything nested inside that wants to hold its own work to a limit finds the tree closed. On a machine the service manager solves this by delegating a subtree to a user session. Nothing here does, because the runtime is what creates the cgroup, and so is already the one privileged to give it away. Grant the user the directory, cgroup.procs, cgroup.threads and cgroup.subtree_control, which is what the kernel documents delegation to mean. The resource files stay where they are: they distribute the parent's resources rather than this cgroup's own, so the limits a container was given still bind everything it puts underneath. A container asks for this through its configuration, and a pod's containers ask the same way.
MayCXC
force-pushed
the
cgroup-delegation
branch
from
August 27, 2026 20:47
5e61a05 to
5c0d8dc
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
A container that runs as somebody other than root cannot use the cgroup it was placed in: the kernel keeps it for whoever owns the directory, so anything nested inside that wants to hold its own work to a limit finds the tree closed. On a machine the service manager solves this by delegating a subtree to a user session. Nothing here does, because the runtime is what creates the cgroup, and so is already the one privileged to give it away.
Grant the user the directory,
cgroup.procs,cgroup.threadsandcgroup.subtree_control, which is what the kernel documents delegation to mean. The resource files stay where they are: they distribute the parent's resources rather than this cgroup's own, so the limits a container was given still bind everything it puts underneath.A container asks for this through its configuration, and a pod's containers ask the same way.
Dependency and Merge Order
This targets stock
apple/containerization:mainand depends on #863. Please merge #863 first.The head carries #863's three commits as prerequisites, because both PRs target stock
main. After #863 merges this branch can take currentmain, leaving only the delegation change and its test.Motivation and Context
This is what a nested runtime looks for. Rootless runc asks a service manager to delegate a cgroup, and without one it ends up with no cgroup at all, so a container runtime running inside a container cannot hold its own workloads to limits. The delegation set is the one the kernel's cgroup-v2 documentation names, not a guess:
https://github.com/opencontainers/runc/blob/main/docs/cgroup-v2.md
It is opt-in through the configuration, so nothing changes for containers that do not ask.
Testing
swift buildandmake checkclean.swift test: 603 tests in 83 suites passed.container cgroup delegationruns a container as a non-root user and asserts it can place its own work under a limit, which fails without the delegation.Type of Change