fleet-agent: on-host TPM-mTLS telemetry client (v1)

Build the on-host agent that pairs with fleet-hostapi and feeds the
fleet-manager control plane (TODO #14 "STILL TO BUILD (a)"). Nothing on the
workstation talked to the manager before this; Monitor + InfluxDB stayed empty.

Scope: telemetry only. The agent reuses the host TPM identity for mTLS (read-only
tpmkey.LoadSigner at 0x81000100 + cert from NV 0x01800100; the private key never
leaves the TPM), opens the long-lived FleetAgent.Session bidi stream, sends Hello
+ periodic HealthSnapshot (load/cpu/temp/mem/tmpfs/nvme/auth/selinux/chrony/kiosk,
probed read-only from /proc+/sys plus hostapi facts), and LOGS but does not act on
the DesiredState/Command it receives. Reconcile actuators (kiosk URL, reboot,
drain) are deferred until the telemetry half is bench-verified.

- files/agent/: new module (go-tpm + grpc), static CGO_ENABLED=0 like enrol/hostapi.
  Copies the enrol tpmkey (+ read-only LoadSigner) and the generated fleet.agent.v1
  proto so it stays self-contained without importing the initramfs enrol module.
- files/agent/simtest/: proves the TPM-signed cert completes a real mTLS client-auth
  handshake (RequireAndVerifyClientCert) against the go-tpm software simulator.
- files/etc/fleet-agent/agent.env: fleet-generic config (manager DNS name, fleet CA
  reused from radius-ca.pem, TPM handle/nvindex, interval). No per-host state.
- fleet-agent.service: hardened unit (tss group for /dev/tpmrm0, INET egress).
- CI builds fleet-agent out-of-VM + runs vet/test + the simulator handshake test.
- Packer: enable_agent flag (default on) + 2b-agent provisioner + fapolicyd trust.
  Post-switch-root only, so no dracut/initramfs changes.
- docs: TODO #14 + fleet-management-design as-built updated.

Requires the host already enrolled (broker approve births the inventory row) —
else the manager returns PermissionDenied, the correct "not enrolled yet" signal.

files/agent/identity.go

@@ -0,0 +1,96 @@
+// identity.go — the host's mTLS identity for the agent plane, reused from the TPM.
+//
+// The private key never leaves the TPM: we present the enrolment cert (DER read
+// from NV 0x01800100) as the leaf and hand crypto/tls a crypto.Signer backed by the
+// persistent signing key at 0x81000100 (tpmkey.LoadSigner). The manager verifies
+// this chain to the fleet CA and derives the cert SERIAL as the inventory key
+// (agent.go: RequireAndVerifyClientCert + peerSerial), so a host can only ever be
+// itself. The server cert is verified against the same fleet CA bundle.
+package main
+
+import (
+	"crypto/tls"
+	"crypto/x509"
+	"fmt"
+	"os"
+
+	"fleet-agent/tpmkey"
+
+	"github.com/google/go-tpm/tpm2"
+	"github.com/google/go-tpm/tpm2/transport"
+)
+
+// tpmIdentity holds the open TPM transport plus the loaded identity so the caller
+// can close the device when the agent exits.
+type tpmIdentity struct {
+	dev  *os.File
+	cert tls.Certificate
+}
+
+func (id *tpmIdentity) Close() error {
+	if id.dev != nil {
+		return id.dev.Close()
+	}
+	return nil
+}
+
+// loadTPMIdentity opens the TPM, loads the signing key at handle, reads the cert
+// from NV index, and assembles a tls.Certificate whose PrivateKey is the TPM
+// signer. The cert must parse and its public key must match the loaded signer
+// (a stale NV cert for a rotated key would fail the handshake anyway — we surface
+// it here with a clear error instead).
+func loadTPMIdentity(tpmPath string, handle, nvIndex tpm2.TPMHandle) (*tpmIdentity, error) {
+	// /dev/tpmrm0 (kernel resource manager) is a command/response char device;
+	// transport.FromReadWriter over the *os.File avoids the linuxtpm transport and
+	// its extra deps — the same lean approach fleet-enrol uses.
+	f, err := os.OpenFile(tpmPath, os.O_RDWR, 0)
+	if err != nil {
+		return nil, fmt.Errorf("open TPM %s: %w", tpmPath, err)
+	}
+	tpm := transport.FromReadWriter(f)
+
+	signer, err := tpmkey.LoadSigner(tpm, handle)
+	if err != nil {
+		f.Close()
+		return nil, fmt.Errorf("load signing key: %w (host enrolled?)", err)
+	}
+	der, err := tpmkey.ReadNVCert(tpm, nvIndex)
+	if err != nil {
+		f.Close()
+		return nil, fmt.Errorf("read identity cert from NV %#x: %w (host enrolled?)", uint32(nvIndex), err)
+	}
+	leaf, err := x509.ParseCertificate(der)
+	if err != nil {
+		f.Close()
+		return nil, fmt.Errorf("parse identity cert: %w", err)
+	}
+
+	return &tpmIdentity{
+		dev: f,
+		cert: tls.Certificate{
+			Certificate: [][]byte{der},
+			PrivateKey:  signer, // crypto.Signer — TLS signs the handshake in the TPM
+			Leaf:        leaf,
+		},
+	}, nil
+}
+
+// clientTLS builds the mTLS config: our TPM-backed client cert + the fleet CA as
+// the root that verifies the manager's server cert. serverName is the manager's
+// DNS name (SNI + hostname verification).
+func clientTLS(id *tpmIdentity, caFile, serverName string) (*tls.Config, error) {
+	pem, err := os.ReadFile(caFile)
+	if err != nil {
+		return nil, fmt.Errorf("read fleet CA %s: %w", caFile, err)
+	}
+	pool := x509.NewCertPool()
+	if !pool.AppendCertsFromPEM(pem) {
+		return nil, fmt.Errorf("no certificates parsed from fleet CA %s", caFile)
+	}
+	return &tls.Config{
+		MinVersion:   tls.VersionTLS12,
+		Certificates: []tls.Certificate{id.cert},
+		RootCAs:      pool,
+		ServerName:   serverName,
+	}, nil
+}