add new files for compat layer

wgstack/conn/bind_std.go

@@ -0,0 +1,513 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"context"
+	"errors"
+	"net"
+	"net/netip"
+	"runtime"
+	"strconv"
+	"sync"
+	"syscall"
+
+	"golang.org/x/net/ipv4"
+	"golang.org/x/net/ipv6"
+	"golang.org/x/sys/unix"
+)
+
+var (
+	_ Bind = (*StdNetBind)(nil)
+)
+
+// StdNetBind implements Bind for all platforms. While Windows has its own Bind
+// (see bind_windows.go), it may fall back to StdNetBind.
+// TODO: Remove usage of ipv{4,6}.PacketConn when net.UDPConn has comparable
+// methods for sending and receiving multiple datagrams per-syscall. See the
+// proposal in https://github.com/golang/go/issues/45886#issuecomment-1218301564.
+type StdNetBind struct {
+	mu     sync.Mutex // protects all fields except as specified
+	ipv4   *net.UDPConn
+	ipv6   *net.UDPConn
+	ipv4PC *ipv4.PacketConn // will be nil on non-Linux
+	ipv6PC *ipv6.PacketConn // will be nil on non-Linux
+
+	// these three fields are not guarded by mu
+	udpAddrPool  sync.Pool
+	ipv4MsgsPool sync.Pool
+	ipv6MsgsPool sync.Pool
+
+	blackhole4 bool
+	blackhole6 bool
+
+	listenAddr4 string
+	listenAddr6 string
+	bindV4      bool
+	bindV6      bool
+	reusePort   bool
+}
+
+func newStdNetBind() *StdNetBind {
+	return &StdNetBind{
+		udpAddrPool: sync.Pool{
+			New: func() any {
+				return &net.UDPAddr{
+					IP: make([]byte, 16),
+				}
+			},
+		},
+
+		ipv4MsgsPool: sync.Pool{
+			New: func() any {
+				msgs := make([]ipv4.Message, IdealBatchSize)
+				for i := range msgs {
+					msgs[i].Buffers = make(net.Buffers, 1)
+					msgs[i].OOB = make([]byte, srcControlSize)
+				}
+				return &msgs
+			},
+		},
+
+		ipv6MsgsPool: sync.Pool{
+			New: func() any {
+				msgs := make([]ipv6.Message, IdealBatchSize)
+				for i := range msgs {
+					msgs[i].Buffers = make(net.Buffers, 1)
+					msgs[i].OOB = make([]byte, srcControlSize)
+				}
+				return &msgs
+			},
+		},
+		bindV4:    true,
+		bindV6:    true,
+		reusePort: false,
+	}
+}
+
+// NewStdNetBind creates a bind that listens on all interfaces.
+func NewStdNetBind() *StdNetBind {
+	return newStdNetBind()
+}
+
+// NewStdNetBindForAddr creates a bind that listens on a specific address.
+// If addr is IPv4, only the IPv4 socket will be created. For IPv6, only the
+// IPv6 socket will be created.
+func NewStdNetBindForAddr(addr netip.Addr, reusePort bool) *StdNetBind {
+	b := newStdNetBind()
+	if addr.IsValid() {
+		if addr.Is4() {
+			b.listenAddr4 = addr.Unmap().String()
+			b.bindV4 = true
+			b.bindV6 = false
+		} else {
+			b.listenAddr6 = addr.Unmap().String()
+			b.bindV6 = true
+			b.bindV4 = false
+		}
+	}
+	b.reusePort = reusePort
+	return b
+}
+
+type StdNetEndpoint struct {
+	// AddrPort is the endpoint destination.
+	netip.AddrPort
+	// src is the current sticky source address and interface index, if supported.
+	src struct {
+		netip.Addr
+		ifidx int32
+	}
+}
+
+var (
+	_ Bind     = (*StdNetBind)(nil)
+	_ Endpoint = &StdNetEndpoint{}
+)
+
+func (*StdNetBind) ParseEndpoint(s string) (Endpoint, error) {
+	e, err := netip.ParseAddrPort(s)
+	if err != nil {
+		return nil, err
+	}
+	return &StdNetEndpoint{
+		AddrPort: e,
+	}, nil
+}
+
+func (e *StdNetEndpoint) ClearSrc() {
+	e.src.ifidx = 0
+	e.src.Addr = netip.Addr{}
+}
+
+func (e *StdNetEndpoint) DstIP() netip.Addr {
+	return e.AddrPort.Addr()
+}
+
+func (e *StdNetEndpoint) SrcIP() netip.Addr {
+	return e.src.Addr
+}
+
+func (e *StdNetEndpoint) SrcIfidx() int32 {
+	return e.src.ifidx
+}
+
+func (e *StdNetEndpoint) DstToBytes() []byte {
+	b, _ := e.AddrPort.MarshalBinary()
+	return b
+}
+
+func (e *StdNetEndpoint) DstToString() string {
+	return e.AddrPort.String()
+}
+
+func (e *StdNetEndpoint) SrcToString() string {
+	return e.src.Addr.String()
+}
+
+func (s *StdNetBind) listenNet(network string, host string, port int) (*net.UDPConn, int, error) {
+	lc := listenConfig()
+	if s.reusePort {
+		base := lc.Control
+		lc.Control = func(network, address string, c syscall.RawConn) error {
+			if base != nil {
+				if err := base(network, address, c); err != nil {
+					return err
+				}
+			}
+			return c.Control(func(fd uintptr) {
+				_ = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_REUSEPORT, 1)
+			})
+		}
+	}
+
+	addr := ":" + strconv.Itoa(port)
+	if host != "" {
+		addr = net.JoinHostPort(host, strconv.Itoa(port))
+	}
+
+	conn, err := lc.ListenPacket(context.Background(), network, addr)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	// Retrieve port.
+	laddr := conn.LocalAddr()
+	uaddr, err := net.ResolveUDPAddr(
+		laddr.Network(),
+		laddr.String(),
+	)
+	if err != nil {
+		return nil, 0, err
+	}
+	return conn.(*net.UDPConn), uaddr.Port, nil
+}
+
+func (s *StdNetBind) openIPv4(port int) (*net.UDPConn, *ipv4.PacketConn, int, error) {
+	if !s.bindV4 {
+		return nil, nil, port, nil
+	}
+	host := s.listenAddr4
+	conn, actualPort, err := s.listenNet("udp4", host, port)
+	if err != nil {
+		if errors.Is(err, syscall.EAFNOSUPPORT) {
+			return nil, nil, port, nil
+		}
+		return nil, nil, port, err
+	}
+	if runtime.GOOS != "linux" {
+		return conn, nil, actualPort, nil
+	}
+	pc := ipv4.NewPacketConn(conn)
+	return conn, pc, actualPort, nil
+}
+
+func (s *StdNetBind) openIPv6(port int) (*net.UDPConn, *ipv6.PacketConn, int, error) {
+	if !s.bindV6 {
+		return nil, nil, port, nil
+	}
+	host := s.listenAddr6
+	conn, actualPort, err := s.listenNet("udp6", host, port)
+	if err != nil {
+		if errors.Is(err, syscall.EAFNOSUPPORT) {
+			return nil, nil, port, nil
+		}
+		return nil, nil, port, err
+	}
+	if runtime.GOOS != "linux" {
+		return conn, nil, actualPort, nil
+	}
+	pc := ipv6.NewPacketConn(conn)
+	return conn, pc, actualPort, nil
+}
+
+func (s *StdNetBind) Open(uport uint16) ([]ReceiveFunc, uint16, error) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	var err error
+	var tries int
+
+	if s.ipv4 != nil || s.ipv6 != nil {
+		return nil, 0, ErrBindAlreadyOpen
+	}
+
+	// Attempt to open ipv4 and ipv6 listeners on the same port.
+	// If uport is 0, we can retry on failure.
+again:
+	port := int(uport)
+	var v4conn *net.UDPConn
+	var v6conn *net.UDPConn
+	var v4pc *ipv4.PacketConn
+	var v6pc *ipv6.PacketConn
+
+	v4conn, v4pc, port, err = s.openIPv4(port)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	// Listen on the same port as we're using for ipv4.
+	v6conn, v6pc, port, err = s.openIPv6(port)
+	if uport == 0 && errors.Is(err, syscall.EADDRINUSE) && tries < 100 {
+		if v4conn != nil {
+			v4conn.Close()
+		}
+		tries++
+		goto again
+	}
+	if err != nil {
+		if v4conn != nil {
+			v4conn.Close()
+		}
+		return nil, 0, err
+	}
+
+	var fns []ReceiveFunc
+	if v4conn != nil {
+		s.ipv4 = v4conn
+		if v4pc != nil {
+			s.ipv4PC = v4pc
+		}
+		fns = append(fns, s.makeReceiveIPv4(v4pc, v4conn))
+	}
+	if v6conn != nil {
+		s.ipv6 = v6conn
+		if v6pc != nil {
+			s.ipv6PC = v6pc
+		}
+		fns = append(fns, s.makeReceiveIPv6(v6pc, v6conn))
+	}
+	if len(fns) == 0 {
+		return nil, 0, syscall.EAFNOSUPPORT
+	}
+
+	return fns, uint16(port), nil
+}
+
+func (s *StdNetBind) makeReceiveIPv4(pc *ipv4.PacketConn, conn *net.UDPConn) ReceiveFunc {
+	return func(bufs [][]byte, sizes []int, eps []Endpoint) (n int, err error) {
+		msgs := s.ipv4MsgsPool.Get().(*[]ipv4.Message)
+		defer s.ipv4MsgsPool.Put(msgs)
+		for i := range bufs {
+			(*msgs)[i].Buffers[0] = bufs[i]
+		}
+		var numMsgs int
+		if runtime.GOOS == "linux" && pc != nil {
+			numMsgs, err = pc.ReadBatch(*msgs, 0)
+			if err != nil {
+				return 0, err
+			}
+		} else {
+			msg := &(*msgs)[0]
+			msg.N, msg.NN, _, msg.Addr, err = conn.ReadMsgUDP(msg.Buffers[0], msg.OOB)
+			if err != nil {
+				return 0, err
+			}
+			numMsgs = 1
+		}
+		for i := 0; i < numMsgs; i++ {
+			msg := &(*msgs)[i]
+			sizes[i] = msg.N
+			addrPort := msg.Addr.(*net.UDPAddr).AddrPort()
+			ep := &StdNetEndpoint{AddrPort: addrPort} // TODO: remove allocation
+			getSrcFromControl(msg.OOB[:msg.NN], ep)
+			eps[i] = ep
+		}
+		return numMsgs, nil
+	}
+}
+
+func (s *StdNetBind) makeReceiveIPv6(pc *ipv6.PacketConn, conn *net.UDPConn) ReceiveFunc {
+	return func(bufs [][]byte, sizes []int, eps []Endpoint) (n int, err error) {
+		msgs := s.ipv6MsgsPool.Get().(*[]ipv6.Message)
+		defer s.ipv6MsgsPool.Put(msgs)
+		for i := range bufs {
+			(*msgs)[i].Buffers[0] = bufs[i]
+		}
+		var numMsgs int
+		if runtime.GOOS == "linux" && pc != nil {
+			numMsgs, err = pc.ReadBatch(*msgs, 0)
+			if err != nil {
+				return 0, err
+			}
+		} else {
+			msg := &(*msgs)[0]
+			msg.N, msg.NN, _, msg.Addr, err = conn.ReadMsgUDP(msg.Buffers[0], msg.OOB)
+			if err != nil {
+				return 0, err
+			}
+			numMsgs = 1
+		}
+		for i := 0; i < numMsgs; i++ {
+			msg := &(*msgs)[i]
+			sizes[i] = msg.N
+			addrPort := msg.Addr.(*net.UDPAddr).AddrPort()
+			ep := &StdNetEndpoint{AddrPort: addrPort} // TODO: remove allocation
+			getSrcFromControl(msg.OOB[:msg.NN], ep)
+			eps[i] = ep
+		}
+		return numMsgs, nil
+	}
+}
+
+// TODO: When all Binds handle IdealBatchSize, remove this dynamic function and
+// rename the IdealBatchSize constant to BatchSize.
+func (s *StdNetBind) BatchSize() int {
+	if runtime.GOOS == "linux" {
+		return IdealBatchSize
+	}
+	return 1
+}
+
+func (s *StdNetBind) Close() error {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	var err1, err2 error
+	if s.ipv4 != nil {
+		err1 = s.ipv4.Close()
+		s.ipv4 = nil
+		s.ipv4PC = nil
+	}
+	if s.ipv6 != nil {
+		err2 = s.ipv6.Close()
+		s.ipv6 = nil
+		s.ipv6PC = nil
+	}
+	s.blackhole4 = false
+	s.blackhole6 = false
+	if err1 != nil {
+		return err1
+	}
+	return err2
+}
+
+func (s *StdNetBind) Send(bufs [][]byte, endpoint Endpoint) error {
+	s.mu.Lock()
+	blackhole := s.blackhole4
+	conn := s.ipv4
+	var (
+		pc4 *ipv4.PacketConn
+		pc6 *ipv6.PacketConn
+	)
+	is6 := false
+	if endpoint.DstIP().Is6() {
+		blackhole = s.blackhole6
+		conn = s.ipv6
+		pc6 = s.ipv6PC
+		is6 = true
+	} else {
+		pc4 = s.ipv4PC
+	}
+	s.mu.Unlock()
+
+	if blackhole {
+		return nil
+	}
+	if conn == nil {
+		return syscall.EAFNOSUPPORT
+	}
+	if is6 {
+		return s.send6(conn, pc6, endpoint, bufs)
+	} else {
+		return s.send4(conn, pc4, endpoint, bufs)
+	}
+}
+
+func (s *StdNetBind) send4(conn *net.UDPConn, pc *ipv4.PacketConn, ep Endpoint, bufs [][]byte) error {
+	ua := s.udpAddrPool.Get().(*net.UDPAddr)
+	as4 := ep.DstIP().As4()
+	copy(ua.IP, as4[:])
+	ua.IP = ua.IP[:4]
+	ua.Port = int(ep.(*StdNetEndpoint).Port())
+	msgs := s.ipv4MsgsPool.Get().(*[]ipv4.Message)
+	for i, buf := range bufs {
+		(*msgs)[i].Buffers[0] = buf
+		(*msgs)[i].Addr = ua
+		setSrcControl(&(*msgs)[i].OOB, ep.(*StdNetEndpoint))
+	}
+	var (
+		n     int
+		err   error
+		start int
+	)
+	if runtime.GOOS == "linux" && pc != nil {
+		for {
+			n, err = pc.WriteBatch((*msgs)[start:len(bufs)], 0)
+			if err != nil || n == len((*msgs)[start:len(bufs)]) {
+				break
+			}
+			start += n
+		}
+	} else {
+		for i, buf := range bufs {
+			_, _, err = conn.WriteMsgUDP(buf, (*msgs)[i].OOB, ua)
+			if err != nil {
+				break
+			}
+		}
+	}
+	s.udpAddrPool.Put(ua)
+	s.ipv4MsgsPool.Put(msgs)
+	return err
+}
+
+func (s *StdNetBind) send6(conn *net.UDPConn, pc *ipv6.PacketConn, ep Endpoint, bufs [][]byte) error {
+	ua := s.udpAddrPool.Get().(*net.UDPAddr)
+	as16 := ep.DstIP().As16()
+	copy(ua.IP, as16[:])
+	ua.IP = ua.IP[:16]
+	ua.Port = int(ep.(*StdNetEndpoint).Port())
+	msgs := s.ipv6MsgsPool.Get().(*[]ipv6.Message)
+	for i, buf := range bufs {
+		(*msgs)[i].Buffers[0] = buf
+		(*msgs)[i].Addr = ua
+		setSrcControl(&(*msgs)[i].OOB, ep.(*StdNetEndpoint))
+	}
+	var (
+		n     int
+		err   error
+		start int
+	)
+	if runtime.GOOS == "linux" && pc != nil {
+		for {
+			n, err = pc.WriteBatch((*msgs)[start:len(bufs)], 0)
+			if err != nil || n == len((*msgs)[start:len(bufs)]) {
+				break
+			}
+			start += n
+		}
+	} else {
+		for i, buf := range bufs {
+			_, _, err = conn.WriteMsgUDP(buf, (*msgs)[i].OOB, ua)
+			if err != nil {
+				break
+			}
+		}
+	}
+	s.udpAddrPool.Put(ua)
+	s.ipv6MsgsPool.Put(msgs)
+	return err
+}

wgstack/conn/conn.go

@@ -0,0 +1,131 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"errors"
+	"fmt"
+	"net/netip"
+	"reflect"
+	"runtime"
+	"strings"
+)
+
+const (
+	IdealBatchSize = 128 // maximum number of packets handled per read and write
+)
+
+// A ReceiveFunc receives at least one packet from the network and writes them
+// into packets. On a successful read it returns the number of elements of
+// sizes, packets, and endpoints that should be evaluated. Some elements of
+// sizes may be zero, and callers should ignore them. Callers must pass a sizes
+// and eps slice with a length greater than or equal to the length of packets.
+// These lengths must not exceed the length of the associated Bind.BatchSize().
+type ReceiveFunc func(packets [][]byte, sizes []int, eps []Endpoint) (n int, err error)
+
+// A Bind listens on a port for both IPv6 and IPv4 UDP traffic.
+//
+// A Bind interface may also be a PeekLookAtSocketFd or BindSocketToInterface,
+// depending on the platform-specific implementation.
+type Bind interface {
+	// Open puts the Bind into a listening state on a given port and reports the actual
+	// port that it bound to. Passing zero results in a random selection.
+	// fns is the set of functions that will be called to receive packets.
+	Open(port uint16) (fns []ReceiveFunc, actualPort uint16, err error)
+
+	// Close closes the Bind listener.
+	// All fns returned by Open must return net.ErrClosed after a call to Close.
+	Close() error
+
+	// SetMark sets the mark for each packet sent through this Bind.
+	// This mark is passed to the kernel as the socket option SO_MARK.
+	SetMark(mark uint32) error
+
+	// Send writes one or more packets in bufs to address ep. The length of
+	// bufs must not exceed BatchSize().
+	Send(bufs [][]byte, ep Endpoint) error
+
+	// ParseEndpoint creates a new endpoint from a string.
+	ParseEndpoint(s string) (Endpoint, error)
+
+	// BatchSize is the number of buffers expected to be passed to
+	// the ReceiveFuncs, and the maximum expected to be passed to SendBatch.
+	BatchSize() int
+}
+
+// BindSocketToInterface is implemented by Bind objects that support being
+// tied to a single network interface. Used by wireguard-windows.
+type BindSocketToInterface interface {
+	BindSocketToInterface4(interfaceIndex uint32, blackhole bool) error
+	BindSocketToInterface6(interfaceIndex uint32, blackhole bool) error
+}
+
+// PeekLookAtSocketFd is implemented by Bind objects that support having their
+// file descriptor peeked at. Used by wireguard-android.
+type PeekLookAtSocketFd interface {
+	PeekLookAtSocketFd4() (fd int, err error)
+	PeekLookAtSocketFd6() (fd int, err error)
+}
+
+// An Endpoint maintains the source/destination caching for a peer.
+//
+//	dst: the remote address of a peer ("endpoint" in uapi terminology)
+//	src: the local address from which datagrams originate going to the peer
+type Endpoint interface {
+	ClearSrc()           // clears the source address
+	SrcToString() string // returns the local source address (ip:port)
+	DstToString() string // returns the destination address (ip:port)
+	DstToBytes() []byte  // used for mac2 cookie calculations
+	DstIP() netip.Addr
+	SrcIP() netip.Addr
+}
+
+var (
+	ErrBindAlreadyOpen   = errors.New("bind is already open")
+	ErrWrongEndpointType = errors.New("endpoint type does not correspond with bind type")
+)
+
+func (fn ReceiveFunc) PrettyName() string {
+	name := runtime.FuncForPC(reflect.ValueOf(fn).Pointer()).Name()
+	// 0. cheese/taco.beansIPv6.func12.func21218-fm
+	name = strings.TrimSuffix(name, "-fm")
+	// 1. cheese/taco.beansIPv6.func12.func21218
+	if idx := strings.LastIndexByte(name, '/'); idx != -1 {
+		name = name[idx+1:]
+		// 2. taco.beansIPv6.func12.func21218
+	}
+	for {
+		var idx int
+		for idx = len(name) - 1; idx >= 0; idx-- {
+			if name[idx] < '0' || name[idx] > '9' {
+				break
+			}
+		}
+		if idx == len(name)-1 {
+			break
+		}
+		const dotFunc = ".func"
+		if !strings.HasSuffix(name[:idx+1], dotFunc) {
+			break
+		}
+		name = name[:idx+1-len(dotFunc)]
+		// 3. taco.beansIPv6.func12
+		// 4. taco.beansIPv6
+	}
+	if idx := strings.LastIndexByte(name, '.'); idx != -1 {
+		name = name[idx+1:]
+		// 5. beansIPv6
+	}
+	if name == "" {
+		return fmt.Sprintf("%p", fn)
+	}
+	if strings.HasSuffix(name, "IPv4") {
+		return "v4"
+	}
+	if strings.HasSuffix(name, "IPv6") {
+		return "v6"
+	}
+	return name
+}

wgstack/conn/controlfns.go

@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"net"
+	"syscall"
+)
+
+// UDP socket read/write buffer size (7MB). The value of 7MB is chosen as it is
+// the max supported by a default configuration of macOS. Some platforms will
+// silently clamp the value to other maximums, such as linux clamping to
+// net.core.{r,w}mem_max (see _linux.go for additional implementation that works
+// around this limitation)
+const socketBufferSize = 7 << 20
+
+// controlFn is the callback function signature from net.ListenConfig.Control.
+// It is used to apply platform specific configuration to the socket prior to
+// bind.
+type controlFn func(network, address string, c syscall.RawConn) error
+
+// controlFns is a list of functions that are called from the listen config
+// that can apply socket options.
+var controlFns = []controlFn{}
+
+// listenConfig returns a net.ListenConfig that applies the controlFns to the
+// socket prior to bind. This is used to apply socket buffer sizing and packet
+// information OOB configuration for sticky sockets.
+func listenConfig() *net.ListenConfig {
+	return &net.ListenConfig{
+		Control: func(network, address string, c syscall.RawConn) error {
+			for _, fn := range controlFns {
+				if err := fn(network, address, c); err != nil {
+					return err
+				}
+			}
+			return nil
+		},
+	}
+}

wgstack/conn/controlfns_linux.go

@@ -0,0 +1,62 @@
+//go:build linux
+
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"fmt"
+	"runtime"
+	"syscall"
+
+	"golang.org/x/sys/unix"
+)
+
+func init() {
+	controlFns = append(controlFns,
+
+		// Attempt to set the socket buffer size beyond net.core.{r,w}mem_max by
+		// using SO_*BUFFORCE. This requires CAP_NET_ADMIN, and is allowed here to
+		// fail silently - the result of failure is lower performance on very fast
+		// links or high latency links.
+		func(network, address string, c syscall.RawConn) error {
+			return c.Control(func(fd uintptr) {
+				// Set up to *mem_max
+				_ = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_RCVBUF, socketBufferSize)
+				_ = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_SNDBUF, socketBufferSize)
+				// Set beyond *mem_max if CAP_NET_ADMIN
+				_ = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_RCVBUFFORCE, socketBufferSize)
+				_ = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_SNDBUFFORCE, socketBufferSize)
+			})
+		},
+
+		// Enable receiving of the packet information (IP_PKTINFO for IPv4,
+		// IPV6_PKTINFO for IPv6) that is used to implement sticky socket support.
+		func(network, address string, c syscall.RawConn) error {
+			var err error
+			switch network {
+			case "udp4":
+				if runtime.GOOS != "android" {
+					c.Control(func(fd uintptr) {
+						err = unix.SetsockoptInt(int(fd), unix.IPPROTO_IP, unix.IP_PKTINFO, 1)
+					})
+				}
+			case "udp6":
+				c.Control(func(fd uintptr) {
+					if runtime.GOOS != "android" {
+						err = unix.SetsockoptInt(int(fd), unix.IPPROTO_IPV6, unix.IPV6_RECVPKTINFO, 1)
+						if err != nil {
+							return
+						}
+					}
+					err = unix.SetsockoptInt(int(fd), unix.IPPROTO_IPV6, unix.IPV6_V6ONLY, 1)
+				})
+			default:
+				err = fmt.Errorf("unhandled network: %s: %w", network, unix.EINVAL)
+			}
+			return err
+		},
+	)
+}

wgstack/conn/default.go

@@ -0,0 +1,9 @@
+//go:build !windows
+
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+func NewDefaultBind() Bind { return NewStdNetBind() }

wgstack/conn/mark_unix.go

@@ -0,0 +1,64 @@
+//go:build linux || openbsd || freebsd
+
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"runtime"
+
+	"golang.org/x/sys/unix"
+)
+
+var fwmarkIoctl int
+
+func init() {
+	switch runtime.GOOS {
+	case "linux", "android":
+		fwmarkIoctl = 36 /* unix.SO_MARK */
+	case "freebsd":
+		fwmarkIoctl = 0x1015 /* unix.SO_USER_COOKIE */
+	case "openbsd":
+		fwmarkIoctl = 0x1021 /* unix.SO_RTABLE */
+	}
+}
+
+func (s *StdNetBind) SetMark(mark uint32) error {
+	var operr error
+	if fwmarkIoctl == 0 {
+		return nil
+	}
+	if s.ipv4 != nil {
+		fd, err := s.ipv4.SyscallConn()
+		if err != nil {
+			return err
+		}
+		err = fd.Control(func(fd uintptr) {
+			operr = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, fwmarkIoctl, int(mark))
+		})
+		if err == nil {
+			err = operr
+		}
+		if err != nil {
+			return err
+		}
+	}
+	if s.ipv6 != nil {
+		fd, err := s.ipv6.SyscallConn()
+		if err != nil {
+			return err
+		}
+		err = fd.Control(func(fd uintptr) {
+			operr = unix.SetsockoptInt(int(fd), unix.SOL_SOCKET, fwmarkIoctl, int(mark))
+		})
+		if err == nil {
+			err = operr
+		}
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}

wgstack/conn/sticky_linux.go

@@ -0,0 +1,116 @@
+//go:build linux && !android
+
+// SPDX-License-Identifier: MIT
+//
+// Copyright (C) 2017-2023 WireGuard LLC. All Rights Reserved.
+
+package conn
+
+import (
+	"net/netip"
+	"unsafe"
+
+	"golang.org/x/sys/unix"
+)
+
+// getSrcFromControl parses the control for PKTINFO and if found updates ep with
+// the source information found.
+func getSrcFromControl(control []byte, ep *StdNetEndpoint) {
+	ep.ClearSrc()
+
+	var (
+		hdr  unix.Cmsghdr
+		data []byte
+		rem  []byte = control
+		err  error
+	)
+
+	for len(rem) > unix.SizeofCmsghdr {
+		hdr, data, rem, err = unix.ParseOneSocketControlMessage(rem)
+		if err != nil {
+			return
+		}
+
+		if hdr.Level == unix.IPPROTO_IP &&
+			hdr.Type == unix.IP_PKTINFO {
+
+			info := pktInfoFromBuf[unix.Inet4Pktinfo](data)
+			ep.src.Addr = netip.AddrFrom4(info.Spec_dst)
+			ep.src.ifidx = info.Ifindex
+
+			return
+		}
+
+		if hdr.Level == unix.IPPROTO_IPV6 &&
+			hdr.Type == unix.IPV6_PKTINFO {
+
+			info := pktInfoFromBuf[unix.Inet6Pktinfo](data)
+			ep.src.Addr = netip.AddrFrom16(info.Addr)
+			ep.src.ifidx = int32(info.Ifindex)
+
+			return
+		}
+	}
+}
+
+// pktInfoFromBuf returns type T populated from the provided buf via copy(). It
+// panics if buf is of insufficient size.
+func pktInfoFromBuf[T unix.Inet4Pktinfo | unix.Inet6Pktinfo](buf []byte) (t T) {
+	size := int(unsafe.Sizeof(t))
+	if len(buf) < size {
+		panic("pktInfoFromBuf: buffer too small")
+	}
+	copy(unsafe.Slice((*byte)(unsafe.Pointer(&t)), size), buf)
+	return t
+}
+
+// setSrcControl sets an IP{V6}_PKTINFO in control based on the source address
+// and source ifindex found in ep. control's len will be set to 0 in the event
+// that ep is a default value.
+func setSrcControl(control *[]byte, ep *StdNetEndpoint) {
+	*control = (*control)[:cap(*control)]
+	if len(*control) < int(unsafe.Sizeof(unix.Cmsghdr{})) {
+		*control = (*control)[:0]
+		return
+	}
+
+	if ep.src.ifidx == 0 && !ep.SrcIP().IsValid() {
+		*control = (*control)[:0]
+		return
+	}
+
+	if len(*control) < srcControlSize {
+		*control = (*control)[:0]
+		return
+	}
+
+	hdr := (*unix.Cmsghdr)(unsafe.Pointer(&(*control)[0]))
+	if ep.SrcIP().Is4() {
+		hdr.Level = unix.IPPROTO_IP
+		hdr.Type = unix.IP_PKTINFO
+		hdr.SetLen(unix.CmsgLen(unix.SizeofInet4Pktinfo))
+
+		info := (*unix.Inet4Pktinfo)(unsafe.Pointer(&(*control)[unix.SizeofCmsghdr]))
+		info.Ifindex = ep.src.ifidx
+		if ep.SrcIP().IsValid() {
+			info.Spec_dst = ep.SrcIP().As4()
+		}
+		*control = (*control)[:unix.CmsgSpace(unix.SizeofInet4Pktinfo)]
+	} else {
+		hdr.Level = unix.IPPROTO_IPV6
+		hdr.Type = unix.IPV6_PKTINFO
+		hdr.SetLen(unix.CmsgLen(unix.SizeofInet6Pktinfo))
+
+		info := (*unix.Inet6Pktinfo)(unsafe.Pointer(&(*control)[unix.SizeofCmsghdr]))
+		info.Ifindex = uint32(ep.src.ifidx)
+		if ep.SrcIP().IsValid() {
+			info.Addr = ep.SrcIP().As16()
+		}
+		*control = (*control)[:unix.CmsgSpace(unix.SizeofInet6Pktinfo)]
+	}
+
+}
+
+var srcControlSize = unix.CmsgSpace(unix.SizeofInet6Pktinfo)
+
+const StdNetSupportsStickySockets = true