Merge branch 'main' into embedded-vnc

# Conflicts:
#	client/ui/frontend/src/app.tsx
#	client/ui/frontend/src/modules/main/MainConnectionStatusSwitch.tsx
#	client/ui/i18n/locales/uk/common.json
#	go.sum

client/internal/engine.go

@@ -14,12 +14,14 @@ import (
 	"sort"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/hashicorp/go-multierror"
 	"github.com/pion/ice/v4"
 	"github.com/pion/stun/v3"
 	log "github.com/sirupsen/logrus"
+	wgdevice "golang.zx2c4.com/wireguard/device"
 	"golang.zx2c4.com/wireguard/tun/netstack"
 	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
 
@@ -239,6 +241,12 @@ type Engine struct {
 
 	wgInterface WGIface
 
+	// wgDevice is a lock-free handle on the WireGuard device behind
+	// wgInterface. Reaching the device through wgInterface requires
+	// syncMsgMux, which handleSync holds while it adds and removes peers;
+	// SetPerformance must stay reachable exactly when that work is stuck.
+	wgDevice atomic.Pointer[wgdevice.Device]
+
 	udpMux *udpmux.UniversalUDPMuxDefault
 
 	// networkSerial is the latest CurrentSerial (state ID) of the network sent by the Management service
@@ -661,6 +669,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 		log.Errorf("failed to pull up wgInterface [%s]: %s", e.wgInterface.Name(), err.Error())
 		return fmt.Errorf("up wg interface: %w", err)
 	}
+	e.wgDevice.Store(e.wgInterface.GetWGDevice())
 
 	// Set up notrack rules immediately after proxy is listening to prevent
 	// conntrack entries from being created before the rules are in place
@@ -2164,6 +2173,10 @@ func (e *Engine) close() {
 	log.Debugf("removing Netbird interface %s", e.config.WgIfaceName)
 
 	if e.wgInterface != nil {
+		// Drop the handle before the close starts: a retune that loads it
+		// afterwards would touch a device on its way out and report success
+		// for an engine that is already gone.
+		e.wgDevice.Store(nil)
 		if err := e.wgInterface.Close(); err != nil {
 			log.Errorf("failed closing Netbird interface %s %v", e.config.WgIfaceName, err)
 		}
@@ -2323,15 +2336,16 @@ type Performance struct {
 }
 
 // SetPerformance applies the given tuning to this engine's live Device.
+//
+// It deliberately does not take syncMsgMux. Raising the buffer pool cap is the
+// recovery path for a device whose pool is exhausted, and an exhausted pool
+// blocks peer removal inside handleSync, which holds syncMsgMux for as long as
+// it stays blocked. Taking the lock here would make the retune unreachable in
+// the one situation that needs it.
 func (e *Engine) SetPerformance(t Performance) error {
-	e.syncMsgMux.Lock()
-	defer e.syncMsgMux.Unlock()
-	if e.wgInterface == nil {
-		return fmt.Errorf("wg interface not initialized")
-	}
-	dev := e.wgInterface.GetWGDevice()
+	dev := e.wgDevice.Load()
 	if dev == nil {
-		return fmt.Errorf("wg device not initialized")
+		return errors.New("wg device not initialized")
 	}
 	if t.PreallocatedBuffersPerPool != nil {
 		dev.SetPreallocatedBuffersPerPool(*t.PreallocatedBuffersPerPool)

client/internal/peer/conn.go

@@ -135,9 +135,10 @@ type Conn struct {
 	// used to store the remote Rosenpass key for Relayed connection in case of connection update from ice
 	rosenpassRemoteKey []byte
 
-	wgProxyICE   wgproxy.Proxy
-	wgProxyRelay wgproxy.Proxy
-	handshaker   *Handshaker
+	wgProxyICE     wgproxy.Proxy
+	wgProxyRelay   wgproxy.Proxy
+	relayedConnRef *relayClient.Conn
+	handshaker     *Handshaker
 
 	guard *guard.Guard
 	wg    sync.WaitGroup
@@ -560,7 +561,7 @@ func (conn *Conn) onRelayConnectionIsReady(rci RelayConnInfo) {
 	conn.mu.Lock()
 	defer conn.mu.Unlock()
 
-	if conn.ctx.Err() != nil {
+	if conn.ctx.Err() != nil || rci.relayedConn.Context().Err() != nil {
 		if err := rci.relayedConn.Close(); err != nil {
 			conn.Log.Warnf("failed to close unnecessary relayed connection: %v", err)
 		}
@@ -575,7 +576,9 @@ func (conn *Conn) onRelayConnectionIsReady(rci RelayConnInfo) {
 		conn.Log.Errorf("failed to add relayed net.Conn to local proxy: %v", err)
 		return
 	}
-	wgProxy.SetDisconnectListener(conn.onRelayDisconnected)
+	wgProxy.SetDisconnectListener(func() {
+		conn.onRelayDisconnected(rci.relayedConn)
+	})
 
 	conn.dumpState.NewLocalProxy()
 
@@ -583,7 +586,7 @@ func (conn *Conn) onRelayConnectionIsReady(rci RelayConnInfo) {
 
 	if conn.isICEActive() {
 		conn.Log.Debugf("do not switch to relay because current priority is: %s", conn.currentConnPriority.String())
-		conn.setRelayedProxy(wgProxy)
+		conn.setRelayedProxy(wgProxy, rci.relayedConn)
 		conn.statusRelay.SetConnected()
 		conn.updateRelayStatus(rci.relayedConn.RemoteAddr().String(), rci.rosenpassPubKey, time.Now())
 		return
@@ -614,15 +617,26 @@ func (conn *Conn) onRelayConnectionIsReady(rci RelayConnInfo) {
 	conn.rosenpassRemoteKey = rci.rosenpassPubKey
 	conn.currentConnPriority = conntype.Relay
 	conn.statusRelay.SetConnected()
-	conn.setRelayedProxy(wgProxy)
+	conn.setRelayedProxy(wgProxy, rci.relayedConn)
 	conn.updateRelayStatus(rci.relayedConn.RemoteAddr().String(), rci.rosenpassPubKey, updateTime)
 	conn.Log.Infof("start to communicate with peer via relay")
 	conn.doOnConnected(rci.rosenpassPubKey, rci.rosenpassAddr, updateTime)
 }
 
-func (conn *Conn) onRelayDisconnected() {
+// onRelayDisconnected reports the teardown of a relayed connection. relayedConn
+// names the connection the signal belongs to, so a signal that arrives after
+// its connection was replaced is ignored instead of tearing down its successor.
+// A nil relayedConn means the caller does not track generations and the current
+// connection is always torn down.
+func (conn *Conn) onRelayDisconnected(relayedConn *relayClient.Conn) {
 	conn.mu.Lock()
 	defer conn.mu.Unlock()
+
+	if relayedConn != nil && conn.relayedConnRef != relayedConn {
+		conn.Log.Debugf("ignoring relay disconnect of a superseded connection")
+		return
+	}
+
 	conn.handleRelayDisconnectedLocked()
 }
 
@@ -646,6 +660,7 @@ func (conn *Conn) handleRelayDisconnectedLocked() {
 		_ = conn.wgProxyRelay.CloseConn()
 		conn.wgProxyRelay = nil
 	}
+	conn.relayedConnRef = nil
 
 	changed := conn.statusRelay.Get() != worker.StatusDisconnected
 	if changed {
@@ -930,13 +945,14 @@ func (conn *Conn) logTraceConnState() {
 	}
 }
 
-func (conn *Conn) setRelayedProxy(proxy wgproxy.Proxy) {
+func (conn *Conn) setRelayedProxy(proxy wgproxy.Proxy, relayedConn *relayClient.Conn) {
 	if conn.wgProxyRelay != nil {
 		if err := conn.wgProxyRelay.CloseConn(); err != nil {
 			conn.Log.Warnf("failed to close deprecated wg proxy conn: %v", err)
 		}
 	}
 	conn.wgProxyRelay = proxy
+	conn.relayedConnRef = relayedConn
 }
 
 // onWGHandshakeSuccess is called when the first WireGuard handshake is detected

client/internal/peer/handshaker.go

@@ -116,7 +116,7 @@ func (h *Handshaker) Listen(ctx context.Context) {
 	for {
 		select {
 		case remoteOfferAnswer := <-h.remoteOffersCh:
-			h.log.Infof("received offer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials())
+			h.log.Infof("received offer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t, relay server: %s, relay IP: %s", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials(), remoteOfferAnswer.RelaySrvAddress, remoteOfferAnswer.RelaySrvIP)
 
 			// Record signaling received for reconnection attempts
 			if h.metricsStages != nil {
@@ -138,7 +138,7 @@ func (h *Handshaker) Listen(ctx context.Context) {
 				continue
 			}
 		case remoteOfferAnswer := <-h.remoteAnswerCh:
-			h.log.Infof("received answer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials())
+			h.log.Infof("received answer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t, relay server: %s, relay IP: %s", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials(), remoteOfferAnswer.RelaySrvAddress, remoteOfferAnswer.RelaySrvIP)
 
 			// Record signaling received for reconnection attempts
 			if h.metricsStages != nil {
@@ -209,14 +209,14 @@ func (h *Handshaker) sendOffer() error {
 	}
 
 	offer := h.buildOfferAnswer()
-	h.log.Debugf("sending offer with serial: %s", offer.SessionIDString())
+	h.log.Debugf("sending offer with serial: %s, relay server: %s, relay IP: %s", offer.SessionIDString(), offer.RelaySrvAddress, offer.RelaySrvIP)
 
 	return h.signaler.SignalOffer(offer, h.config.Key)
 }
 
 func (h *Handshaker) sendAnswer() error {
 	answer := h.buildOfferAnswer()
-	h.log.Debugf("sending answer with serial: %s", answer.SessionIDString())
+	h.log.Debugf("sending answer with serial: %s, relay server: %s, relay IP: %s", answer.SessionIDString(), answer.RelaySrvAddress, answer.RelaySrvIP)
 
 	return h.signaler.SignalAnswer(answer, h.config.Key)
 }

client/internal/peer/status.go

@@ -819,8 +819,8 @@ func (d *Status) SetSessionExpiresAt(deadline time.Time) {
 // "none" would blank the UI at the exact moment it should say the session
 // ended.
 func (d *Status) GetSessionExpiresAt() time.Time {
-	d.mux.Lock()
-	defer d.mux.Unlock()
+	d.mux.RLock()
+	defer d.mux.RUnlock()
 	return d.sessionExpiresAt
 }
 

client/internal/peer/worker_relay.go

@@ -3,7 +3,6 @@ package peer
 import (
 	"context"
 	"errors"
-	"net"
 	"net/netip"
 	"sync"
 	"sync/atomic"
@@ -14,7 +13,7 @@ import (
 )
 
 type RelayConnInfo struct {
-	relayedConn     net.Conn
+	relayedConn     *relayClient.Conn
 	rosenpassPubKey []byte
 	rosenpassAddr   string
 }
@@ -27,7 +26,7 @@ type WorkerRelay struct {
 	conn         *Conn
 	relayManager *relayClient.Manager
 
-	relayedConn net.Conn
+	relayedConn *relayClient.Conn
 	relayLock   sync.Mutex
 
 	relaySupportedOnRemotePeer atomic.Bool
@@ -80,12 +79,7 @@ func (w *WorkerRelay) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
 	w.relayedConn = relayedConn
 	w.relayLock.Unlock()
 
-	err = w.relayManager.AddCloseListener(srv, w.onRelayClientDisconnected)
-	if err != nil {
-		log.Errorf("failed to add close listener: %s", err)
-		_ = relayedConn.Close()
-		return
-	}
+	go w.watchRelayedConn(relayedConn)
 
 	w.log.Debugf("peer conn opened via Relay: %s", srv)
 	go w.conn.onRelayConnectionIsReady(RelayConnInfo{
@@ -109,12 +103,15 @@ func (w *WorkerRelay) RelayIsSupportedLocally() bool {
 
 func (w *WorkerRelay) CloseConn() {
 	w.relayLock.Lock()
-	defer w.relayLock.Unlock()
-	if w.relayedConn == nil {
+	conn := w.relayedConn
+	w.relayedConn = nil
+	w.relayLock.Unlock()
+
+	if conn == nil {
 		return
 	}
 
-	if err := w.relayedConn.Close(); err != nil {
+	if err := conn.Close(); err != nil {
 		w.log.Warnf("failed to close relay connection: %v", err)
 	}
 }
@@ -133,6 +130,8 @@ func (w *WorkerRelay) preferredRelayServer(myRelayAddress, remoteRelayAddress st
 	return remoteRelayAddress
 }
 
-func (w *WorkerRelay) onRelayClientDisconnected() {
-	go w.conn.onRelayDisconnected()
+func (w *WorkerRelay) watchRelayedConn(relayedConn *relayClient.Conn) {
+	<-relayedConn.Context().Done()
+
+	w.conn.onRelayDisconnected(relayedConn)
 }

client/internal/profilemanager/invoking_user.go

@@ -6,6 +6,7 @@ import (
 	"os/user"
 	"path/filepath"
 	"runtime"
+	"strconv"
 
 	log "github.com/sirupsen/logrus"
 )
@@ -13,17 +14,21 @@ import (
 const envSudoUser = "SUDO_USER"
 
 var (
-	geteuid    = os.Geteuid
-	lookupUser = user.Lookup
+	currentUser = user.Current
+	getegid     = os.Getegid
+	geteuid     = os.Geteuid
+	lookupUser  = user.Lookup
 )
 
 // InvokingUser returns the user a CLI invocation acts for. Under sudo that is
 // the user who ran sudo, not root: privileged flags force commands through
 // sudo, and resolving profiles as root would silently switch the daemon to
-// root's (default) profile instead of the invoking user's. Privilege decisions
-// are not made here — those stay on the kernel credentials of the daemon
-// connection, which SUDO_USER (a plain environment variable) can never
-// influence; a forged value only selects a profile root could select anyway.
+// root's (default) profile instead of the invoking user's. An unmapped positive
+// process UID uses its numeric kernel identity; root, sudo lookup failures, and
+// unavailable platform identities still fail closed. Privilege decisions stay
+// on the kernel credentials of the daemon connection, which SUDO_USER (a plain
+// environment variable) can never influence; a forged value only selects a
+// profile root could select anyway.
 func InvokingUser() (*user.User, error) {
 	if u, ok := sudoInvokingUser(); ok {
 		return u, nil
@@ -35,7 +40,23 @@ func InvokingUser() (*user.User, error) {
 	if sudoActive() {
 		return nil, fmt.Errorf("resolve sudo invoking user %q: refusing to fall back to root", os.Getenv(envSudoUser))
 	}
-	return user.Current()
+	u, err := currentUser()
+	if err == nil {
+		return u, nil
+	}
+
+	uid := geteuid()
+	if uid <= 0 {
+		return nil, err
+	}
+
+	log.Debugf("current user lookup for UID %d: %v; using numeric UID", uid, err)
+	uidString := strconv.Itoa(uid)
+	return &user.User{
+		Username: uidString,
+		Uid:      uidString,
+		Gid:      strconv.Itoa(getegid()),
+	}, nil
 }
 
 // IsPlainRoot reports that the process runs as root with no usable sudo

client/internal/profilemanager/invoking_user_test.go

@@ -2,6 +2,7 @@ package profilemanager
 
 import (
 	"errors"
+	"fmt"
 	"io/fs"
 	"os"
 	"os/user"
@@ -21,7 +22,51 @@ func TestInvokingUserFallsBackToProcessUser(t *testing.T) {
 
 	current, err := user.Current()
 	require.NoError(t, err)
-	assert.Equal(t, current.Username, got.Username)
+	assert.Equal(t, current.Username, got.Username, "invoking user should match the process user without sudo")
+}
+
+func TestInvokingUserFailsClosedWithoutPositiveUID(t *testing.T) {
+	for _, uid := range []int{0, -1} {
+		t.Run(fmt.Sprintf("UID%d", uid), func(t *testing.T) {
+			t.Setenv(envSudoUser, "")
+			lookupErr := errors.New("current user unavailable")
+			fakeUnmappedUser(t, uid, 0, lookupErr)
+
+			got, err := InvokingUser()
+			require.ErrorIs(t, err, lookupErr)
+			assert.Nil(t, got, "root or unavailable UID must not become a synthetic identity")
+		})
+	}
+}
+
+func TestProfileFilePathUsesNumericIdentityForUnmappedNonRoot(t *testing.T) {
+	t.Setenv(envSudoUser, "")
+	fakeUnmappedUser(t, 1001230000, 0, errors.New("user: unknown userid 1001230000"))
+
+	profilesRoot := t.TempDir()
+	origDir := DefaultConfigPathDir
+	origOverride := ConfigDirOverride
+	DefaultConfigPathDir = profilesRoot
+	ConfigDirOverride = ""
+	t.Cleanup(func() {
+		DefaultConfigPathDir = origDir
+		ConfigDirOverride = origOverride
+	})
+
+	profileID := ID("0123456789abcdef0123456789abcdef")
+	got, err := (&Profile{ID: profileID}).FilePath()
+	require.NoError(t, err)
+	assert.Equal(t,
+		filepath.Join(profilesRoot, "1001230000", profileID.String()+".json"),
+		got,
+		"profile path should use the numeric UID namespace",
+	)
+
+	entries, err := os.ReadDir(profilesRoot)
+	require.NoError(t, err)
+	require.Len(t, entries, 1, "only the numeric UID directory should be created")
+	assert.Equal(t, "1001230000", entries[0].Name(), "profile namespace should be numeric")
+	assert.True(t, entries[0].IsDir(), "profile namespace should be a directory")
 }
 
 func TestSudoInvokingUserInactiveWithoutSudoContext(t *testing.T) {
@@ -60,6 +105,13 @@ func TestInvokingUserFailsClosedWhenSudoLookupFails(t *testing.T) {
 	fakeSudo(t, filepath.Join("/home", "misha"))
 	lookupUser = func(string) (*user.User, error) { return nil, errors.New("nss unavailable") }
 
+	origCurrentUser := currentUser
+	currentUser = func() (*user.User, error) {
+		t.Fatal("currentUser must not be called after a sudo lookup failure")
+		return nil, errors.New("currentUser called unexpectedly")
+	}
+	t.Cleanup(func() { currentUser = origCurrentUser })
+
 	got, err := InvokingUser()
 	require.Error(t, err)
 	assert.Nil(t, got, "must not resolve to the root process user")
@@ -215,6 +267,22 @@ func fakeSudo(t *testing.T, home string) {
 	})
 }
 
+func fakeUnmappedUser(t *testing.T, uid, gid int, lookupErr error) {
+	t.Helper()
+
+	origCurrentUser := currentUser
+	origEuid := geteuid
+	origEgid := getegid
+	currentUser = func() (*user.User, error) { return nil, lookupErr }
+	geteuid = func() int { return uid }
+	getegid = func() int { return gid }
+	t.Cleanup(func() {
+		currentUser = origCurrentUser
+		geteuid = origEuid
+		getegid = origEgid
+	})
+}
+
 func assertNoEntries(t *testing.T, root string) {
 	t.Helper()
 	err := filepath.WalkDir(root, func(path string, _ fs.DirEntry, err error) error {