feat(tunnel): zero-touch tunnel enrollment — control-plane PKI, host agent, iOS, nginx isolation

Customers install one command / log in once; hardware-generated keys never leave the
device; CSRs return certs + subdomain; frpc + base-app run as durable services. No .p12,
no manual cert import. Implements the MVP fast-path of docs/PLAN_TUNNEL_AUTOMATION.md.

Control-plane / PKI (control-plane/):
- ca/x509-assembler.ts: single KMS-signed real X.509 issuance primitive (Ed25519 + P-256)
- ca/csr-ec.ts: P-256 PKCS#10 proof-of-possession (verifyCsrPoPEc) + CSR-key routing
- ca/frpclient-issue.ts, ca/device-issue.ts: P-256 frp-client + device leaf signers
- ca/rotate.ts + api/renew.ts: real-X.509 /renew + /device/:id/renew (mTLS current cert)
- registry/devices.ts: device registry + per-account cap/rate-limit
- auth/session.ts: device:enroll capability token mint/verify
- api/device-enroll.ts: POST /device/enroll (ownership-gated, deny-by-default)
- pairing/native-redeem.ts + shared gateAndConsumePairingCode; api/provision.ts native arm
- boot/native-ca.ts + main.ts: wire two P-256 CAs + issuers + routers (dev / KMS fail-fast)

Contracts: relay-contracts enroll right; relay-auth SPIFFE /device/ arm + spiffeIdFor(kind)

Host agent (agent/):
- transport/frpcToml.ts; provision/frpcBinary.ts + untar.ts (verify-download + traversal-safe extract)
- keys P-256 keygen/CSR/loadIdentity; service two-unit install + BIND_HOST loopback S-GATE
- net/loopbackLiteral.ts strict guard; health/probe.ts + transport/frpSupervise.ts; cli pair --install

iOS (ios/Packages/ClientTLS): SecureEnclaveKey + CertificateSigningRequest + DeviceEnrollmentClient
+ Keychain enroll refactor (SecKey/Security.framework end-to-end, avoids the -25300 trap)

Isolation (deploy/nginx): njs/getCertSub.js SAN parser + zone-anchored map -> 403

Verified: 758 tests green (control-plane 246, agent 267, relay-auth 133, relay-contracts 85,
iOS ClientTLS 27), all tsc clean; real nginx+njs docker 403/200/400; Swift CSR accepted by
the real control-plane verifier; frpc extract byte-identical to `tar -xO`. Cross-validation
caught + fixed 5 real defects (1 critical, 4 high). Remaining = infra (KMS, nginx deploy,
VPS frps, physical iPhone) per PROGRESS_LOG runbook.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Yaojia Wang
2026-07-10 16:11:13 +02:00
parent 31054450fc
commit e7f3bd05f0
79 changed files with 9920 additions and 385 deletions

View File

@@ -23,6 +23,7 @@ import { createSessionRegistry } from './registry/sessions.js'
import { createSubdomainAssigner } from './subdomain/assign.js'
import { createPairingIssuer } from './pairing/issue.js'
import { createPairingRedeemer } from './pairing/redeem.js'
import { createNativeHostEnroller } from './pairing/native-redeem.js'
import { createLeafSigner, type LeafSigner } from './ca/sign.js'
import { loadRealLeafSigner } from './ca/issue.js'
import { createRoutingTable } from './routing/table.js'
@@ -33,6 +34,17 @@ import { createAuthorizer, type CapabilityVerifier } from './api/authz.js'
import { buildRouter } from './api/provision.js'
import { buildCaSigner, inProcessCaSigner, type KmsResolver, type CaSigner } from './boot/ca-wiring.js'
import { configureCapabilityVerifyKey } from './boot/verifier.js'
import {
createDeviceRegistry,
createMemoryDeviceStore,
type DeviceRegistry,
} from './registry/devices.js'
import { createFrpClientLeafSigner } from './ca/frpclient-issue.js'
import { createDeviceLeafSigner, type DeviceLeafSigner } from './ca/device-issue.js'
import { createLeafRenewer } from './ca/rotate.js'
import { buildDeviceEnrollRouter, type SubdomainOwnershipResolver } from './api/device-enroll.js'
import { buildRenewRouter } from './api/renew.js'
import { buildNativeCas, DEFAULT_NATIVE_DNS_ZONE, type NativeCas, type NativeCaMaterial } from './boot/native-ca.js'
import type { RevocationBus } from 'relay-contracts'
export interface ControlPlaneOverrides {
@@ -41,6 +53,29 @@ export interface ControlPlaneOverrides {
readonly bus?: RevocationBus & Partial<TestableRevocationBus>
readonly kmsResolver?: KmsResolver
readonly caChainDer?: readonly Uint8Array[]
/** Production mode → native-CA + renew-anchor material is fail-closed (INV9). Defaults to NODE_ENV. */
readonly production?: boolean
/** Production-loaded native-tunnel CA material (per-CA KMS ref + public cert DER). */
readonly nativeCaMaterial?: NativeCaMaterial
/** DNS zone the native-tunnel leaves are stamped under (defaults to `terminal.yaojia.wang`). */
readonly nativeDnsZone?: string
}
/** Native-tunnel PKI handles exposed for wiring + tests (the enroll/renew issuers behind the routes). */
export interface NativeTunnelHandles {
readonly nativeCas: NativeCas
/** Host frp-client P-256 leaf signer (used at onboarding to mint the first leaf). */
readonly hostSigner: LeafSigner
/** Device P-256 leaf signer (shares the device store with the registry + renew path). */
readonly deviceSigner: DeviceLeafSigner
/** Device registry (ownership + cap/rate + expiry-renewal source of truth). */
readonly deviceRegistry: DeviceRegistry
}
export interface BuiltControlPlane {
readonly app: FastifyInstance
readonly stores: Stores
readonly nativeTunnel: NativeTunnelHandles
}
/**
@@ -97,7 +132,7 @@ function devKmsResolver(): KmsResolver {
export async function buildControlPlane(
env: ControlPlaneEnv,
overrides: ControlPlaneOverrides = {},
): Promise<{ app: FastifyInstance; stores: Stores }> {
): Promise<BuiltControlPlane> {
const stores = overrides.stores ?? createMemoryStores()
const bus: RevocationBus = overrides.bus ?? createInMemoryRevocationBus()
const caChainDer = overrides.caChainDer ?? []
@@ -142,7 +177,81 @@ export async function buildControlPlane(
expectedAud: env.baseDomain,
})
// ── Native-tunnel PKI: frp-client-CA (host) + device-CA (device), both P-256 (§1.3) ───────────────
// Production is fail-closed: `buildNativeCas` refuses to boot without real KMS-backed material, and
// the renew route MUST validate presented certs against non-empty anchors (renew.ts). DEV generates
// self-signed in-process P-256 CAs so leaves chain to a real, re-parseable CA.
const production = overrides.production ?? process.env.NODE_ENV === 'production'
const nativeCas = await buildNativeCas(env, {
production,
...(overrides.kmsResolver !== undefined ? { kmsResolver: overrides.kmsResolver } : {}),
...(overrides.nativeCaMaterial !== undefined ? { material: overrides.nativeCaMaterial } : {}),
})
const nativeDnsZone = overrides.nativeDnsZone ?? DEFAULT_NATIVE_DNS_ZONE
// ONE DeviceStore shared across the device registry, the device signer, and the renew path so the
// signer's gate + renew-expiry bump all see the device the enroll route just registered.
const deviceStore = createMemoryDeviceStore()
const deviceRegistry = createDeviceRegistry({ devices: deviceStore })
const hostSigner = createFrpClientLeafSigner({
hosts: stores.hosts,
signer: nativeCas.frpClientCa.signer,
issuerName: nativeCas.frpClientCa.issuerName,
caChainDer: nativeCas.frpClientCa.anchorsDer,
trustDomain: env.relayTrustDomain,
dnsZone: nativeDnsZone,
})
// Native (EC-P256) `/enroll` arm: pairing-gated host onboarding that mints the FIRST frp-client leaf
// (mirrors the Ed25519 relay redeemer, but issues a P-256 leaf under a SERVER-assigned subdomain and
// returns no E2E-relay content secret — L-host-hcs). Shares the frp-client `hostSigner` above so the
// enroll-issued leaf and the later /renew leaf are stamped by the SAME CA + subdomain.
const nativeEnroller = createNativeHostEnroller({
pairing: stores.pairing,
hosts,
subdomains,
hostSigner,
pairingMaxRedeemAttempts: env.pairingMaxRedeemAttempts,
audit,
})
const deviceSigner = createDeviceLeafSigner({
signer: nativeCas.deviceCa.signer,
issuer: nativeCas.deviceCa.issuerName,
caChainDer: nativeCas.deviceCa.anchorsDer,
sanBaseDomain: nativeDnsZone,
trustDomain: env.relayTrustDomain,
devices: deviceStore,
})
const renewer = createLeafRenewer({ hostSigner, deviceSigner, deviceExpiry: deviceRegistry })
// The device leaf's dNSName SAN is nginx :8470's single tenant boundary, so the enroll route must
// NOT trust the client-supplied subdomain: resolve who OWNS it against the host-onboarding registry
// (deny-by-default: unknown → null). Same source of truth as `HostStore.getBySubdomain(...)`.
const ownership: SubdomainOwnershipResolver = {
async ownerOfSubdomain(subdomain) {
return (await hosts.getHostBySubdomain(subdomain))?.accountId ?? null
},
}
// Fail-closed: the renew route validates presented certs against these anchors; in production they
// MUST be non-empty (renew.ts: "production wiring MUST supply them"). Refuse to boot otherwise.
if (production && (nativeCas.frpClientCa.anchorsDer.length === 0 || nativeCas.deviceCa.anchorsDer.length === 0)) {
throw new Error('native-tunnel renew anchors must be non-empty in production (fail-closed) — refusing to boot')
}
const app = Fastify({ logger: false })
await app.register(buildRouter({ authorizer, accounts, hosts, pairingIssuer, redeemer, deprovisioner }))
return { app, stores }
await app.register(buildRouter({ authorizer, accounts, hosts, pairingIssuer, redeemer, deprovisioner, nativeEnroller }))
// Device enrollment is bearer-gated by the SAME capability verifier seam the admin API uses.
await app.register(buildDeviceEnrollRouter({ verifier, devices: deviceRegistry, signer: deviceSigner, ownership }))
// Leaf renewal is mTLS-authenticated (current client cert) — anchors chain-validate the presented cert.
await app.register(
buildRenewRouter({
hosts,
devices: deviceRegistry,
renewer,
hostCaAnchorsDer: nativeCas.frpClientCa.anchorsDer,
deviceCaAnchorsDer: nativeCas.deviceCa.anchorsDer,
}),
)
return { app, stores, nativeTunnel: { nativeCas, hostSigner, deviceSigner, deviceRegistry } }
}