Files
Michael Crosby b438e97b93 Add cloud-hypervisor VMM backend for Linux hosts (#782)
apple/containerization currently runs containers in per-container VMs on
macOS hosts via Virtualization.framework. This adds a second VMM backend
so the same Swift orchestration layer (LinuxContainer / LinuxPod /
Vminitd gRPC contract) runs on Linux hosts via cloud-hypervisor + KVM.

**CloudHypervisor Swift package** (`Sources/CloudHypervisor/`) — a thin
client for cloud-hypervisor's REST-over-UDS API, layered on
AsyncHTTPClient. Endpoints cover VMM / VM lifecycle / hotplug (disk, fs,
net, vsock, remove-device). Cross-platform (compiles on macOS for unit
tests; consumed at runtime only by the Linux side of Containerization).

**CH backend in Containerization** — one cloud-hypervisor subprocess per
VM, gated behind `#if os(Linux)`. CHVirtualMachineManager /
CHVirtualMachineInstance mirror the VZ shape behind the existing
VirtualMachineManager / VirtualMachineInstance protocol. CHProcess and
VirtiofsdProcess manage the binaries; CHHotplugProvider handles
virtio-blk and virtio-fs runtime hotplug (with one virtiofsd per unique
source-hash tag, refcounted across containers).

**Linux host networking** — BridgeManager brings up a Linux bridge with
an IPv4 subnet and (opt-in via `--enable-nat`) iptables MASQUERADE +
scoped FORWARD rules. LinuxBridgedNetwork enslaves a fresh TAP per
container to the bridge. State is recorded under `/run/containerization`
so `cctl bridge delete` reverses exactly what create did. Bridge
teardown verifies the link kind via sysfs to refuse deleting non-bridge
interfaces.

**cctl run / bridge** — end-to-end Linux container run path (image pull,
ext4 rootfs assembly, VM boot, container exec) plus `cctl bridge
create|delete` for the host network plumbing.

**Build & dist** — `make linux-build` / `make linux-integration` build
and exercise the host side inside an apple/container `--virtualization`
dev container. `make dist-x86_64` produces a deployment tarball (cctl +
cloud-hypervisor + virtiofsd + initfs + kernel) cross-compiled from the
aarch64 dev container; pipeline documented in `docs/x86_64-build.md`.
Static-musl C deps and the Zig cross compiler are pinned by SHA256.

The host orchestrator runs as root. Per-VM runtime state lives under
`/run/containerization/ch/<UUID>` with mode 0700; UDS sockets inside are
bound with mode 0600. Vminitd's gRPC channel inherits that trust
boundary — socket-file perms are the auth.

Sandbox flags are upstream-secure by default. Two per-component opt-outs
exist for the apple/container dev-container case (where the host seccomp
profile SIGSYS-kills CH and virtiofsd):
- `CONTAINERIZATION_NO_CH_SECCOMP=1` — `cloud-hypervisor --seccomp
false`.
- `CONTAINERIZATION_NO_VIRTIOFSD_SANDBOX=1` — `virtiofsd --sandbox
none`. Each emits a one-shot `logger.warning` at process start. Legacy
alias `CONTAINERIZATION_RELAXED_SANDBOX=1` flips both. cctl spawns both
binaries with `setsid` and a minimal env allowlist (PATH / HOME /
RUST_LOG / RUST_BACKTRACE) so the parent's secrets don't leak to
children.

`make linux-integration` runs the cross-platform integration suite
against a real cloud-hypervisor VM inside the dev container. Linux runs
the cross-platform subset (`process true`/`false`/`echo hi`, virtiofs
round-trip, hotplug); the macOS suite is unchanged.

Signed-off-by: michael_crosby <michael_crosby@apple.com>
2026-07-02 11:20:22 -04:00

188 lines
6.8 KiB
Swift

//===----------------------------------------------------------------------===//
// Copyright © 2026 Apple Inc. and the Containerization project authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//===----------------------------------------------------------------------===//
#if os(Linux)
import ContainerizationError
import ContainerizationExtras
import ContainerizationOS
import Foundation
import Logging
import Synchronization
#if canImport(Musl)
import Musl
#elseif canImport(Glibc)
import Glibc
#endif
/// A managed `virtiofsd` subprocess serving a single shared directory.
///
/// One `VirtiofsdProcess` per virtio-fs share. Cloud Hypervisor connects to
/// the published UDS via its `FsConfig.socket` field. Lifecycle mirrors
/// `CHProcess`: spawn + wait-for-socket on `start()`, SIGTERM/SIGKILL on
/// `terminate()`.
final class VirtiofsdProcess: Sendable {
struct Config: Sendable {
let binary: URL
let socketPath: URL
let sharedDir: URL
let readonly: Bool
}
private struct State {
var command: Command?
var exitTask: Task<Void, Never>?
}
private let config: Config
private let logger: Logger?
private let state: Mutex<State>
init(config: Config, logger: Logger?) {
self.config = config
self.logger = logger
self.state = Mutex(State(command: nil, exitTask: nil))
}
/// Spawn virtiofsd and wait for its UDS to accept connections.
func start() async throws {
var arguments = [
"--socket-path", config.socketPath.path,
"--shared-dir", config.sharedDir.path,
]
if SandboxOverrides.virtiofsdSandboxDisabled {
// virtiofsd defaults to `--sandbox namespace`, which sets up a
// userns + pivot_root + seccomp filter. Inside apple/container's
// --virtualization dev container the default seccomp profile
// SIGSYS-kills processes that hit unfiltered syscalls (same
// reason CH runs with `--seccomp false`). `--sandbox none`
// skips both userns setup and seccomp; safe inside the per-VM
// dev container only. Opt-in via
// CONTAINERIZATION_NO_VIRTIOFSD_SANDBOX=1.
logger?.warning(
"virtiofsd launching with --sandbox none (CONTAINERIZATION_NO_VIRTIOFSD_SANDBOX=1) — userns/pivot_root/seccomp setup disabled"
)
arguments.append(contentsOf: ["--sandbox", "none"])
}
if config.readonly {
arguments.append("--readonly")
}
var command = Command(
config.binary.path,
arguments: arguments,
environment: ChildEnvironment.minimal()
)
// Inherit stderr so virtiofsd's startup logs surface in the host's
// log stream rather than vanishing into /dev/null (Command's default).
command.stderr = FileHandle.standardError
// Same rationale as CHProcess: keep virtiofsd out of the parent's
// controlling-TTY signal group so Ctrl-C doesn't kill it before our
// own terminate() ladder runs.
command.attrs.setsid = true
do {
try command.start()
} catch {
throw error
}
let exitTask = Task<Void, Never>.detached { [command, logger] in
do {
_ = try command.wait()
} catch {
logger?.error("virtiofsd wait failed: \(error)")
}
}
state.withLock {
$0.command = command
$0.exitTask = exitTask
}
try await waitForSocket()
}
/// SIGTERM → grace window → SIGKILL. Returns once virtiofsd is reaped.
func terminate(graceSeconds: UInt32) async {
guard let command = state.withLock({ $0.command }) else { return }
_ = command.kill(SIGTERM)
do {
try await Timeout.run(for: .seconds(Int(graceSeconds))) {
await self.waitForExit()
}
} catch {
logger?.warning("virtiofsd did not exit within \(graceSeconds)s, sending SIGKILL")
_ = command.kill(SIGKILL)
await waitForExit()
}
}
// MARK: - Private helpers
private static let socketDeadline: Duration = .seconds(10)
private static let socketPollInterval: Duration = .milliseconds(50)
private func waitForExit() async {
guard let task = state.withLock({ $0.exitTask }) else { return }
await task.value
}
private func waitForSocket() async throws {
let clock = ContinuousClock()
let started = clock.now
let deadline = started.advanced(by: Self.socketDeadline)
while clock.now < deadline {
if Self.isSocketReady(at: config.socketPath) {
let elapsed = clock.now - started
logger?.debug("virtiofsd socket bound in \(elapsed) at \(config.socketPath.path)")
return
}
try? await Task.sleep(for: Self.socketPollInterval)
}
// Capture diagnostic state before terminating.
let fm = FileManager.default
let socketExists = fm.fileExists(atPath: config.socketPath.path)
let parentExists = fm.fileExists(atPath: config.socketPath.deletingLastPathComponent().path)
let sharedExists = fm.fileExists(atPath: config.sharedDir.path)
let detail = "socketExists=\(socketExists) parentDirExists=\(parentExists) sharedDirExists=\(sharedExists)"
await terminate(graceSeconds: 5)
throw ContainerizationError(
.timeout,
message: "virtiofsd socket not connectable at \(config.socketPath.path) within \(Self.socketDeadline) [\(detail)]"
)
}
private static func isSocketReady(at url: URL) -> Bool {
// Only check that the socket file exists. Do NOT connect — virtiofsd
// runs in vhost-user mode where the first incoming connection is
// treated as the VMM (cloud-hypervisor); when that connection closes,
// virtiofsd exits. A connect-then-close readiness probe therefore
// kills virtiofsd before CH ever gets to it, leaving CH's vm.boot
// failing with "vhost-user: can't connect to peer: No such file
// or directory".
var st = stat()
guard stat(url.path, &st) == 0 else { return false }
return (st.st_mode & S_IFMT) == S_IFSOCK
}
}
#endif