Use stack.Route exclusively for writing packets

* Remove stack.Route from incoming packet path.
There is no need to pass around a stack.Route during the incoming path
of a packet. Instead, pass around the packet's link/network layer
information in the packet buffer since all layers may need this
information.

* Support address bound and outgoing packet NIC in routes.
When forwarding is enabled, the source address of a packet may be bound
to a different interface than the outgoing interface. This change
updates stack.Route to hold both NICs so that one can be used to write
packets while the other is used to check if the route's bound address
is valid. Note, we need to hold the address's interface so we can check
if the address is a spoofed address.

* Introduce the concept of a local route.
Local routes are routes where the packet never needs to leave the stack;
the destination is stack-local. We can now route between interfaces
within a stack if the packet never needs to leave the stack, even when
forwarding is disabled.

* Always obtain a route from the stack before sending a packet.
If a packet needs to be sent in response to an incoming packet, a route
must be obtained from the stack to ensure the stack is configured to
send packets to the packet's source from the packet's destination.

* Enable spoofing if a stack may send packets from unowned addresses.
This change required changes to some netgophers since previously,
promiscuous mode was enough to let the netstack respond to all
incoming packets regardless of the packet's destination address. Now
that a stack.Route is not held for each incoming packet, finding a route
may fail with local addresses we don't own but accepted packets for
while in promiscuous mode. Since we also want to be able to send from
any address (in response the received promiscuous mode packets), we need
to enable spoofing.

* Skip transport layer checksum checks for locally generated packets.
If a packet is locally generated, the stack can safely assume that no
errors were introduced while being locally routed since the packet is
never sent out the wire.

Some bugs fixed:
- transport layer checksum was never calculated after NAT.
- handleLocal didn't handle routing across interfaces.
- stack didn't support forwarding across interfaces.
- always consult the routing table before creating an endpoint.

Updates #4688
Fixes #3906

PiperOrigin-RevId: 340943442

3 files changed, 33 insertions, 25 deletions

diff --git a/pkg/tcpip/transport/udp/endpoint.go b/pkg/tcpip/transport/udp/endpoint.go
index cdb5127ab..56bdf6c34 100644
--- a/pkg/tcpip/transport/udp/endpoint.go
+++ b/pkg/tcpip/transport/udp/endpoint.go
@@ -1012,7 +1012,7 @@ func sendUDP(r *stack.Route, data buffer.VectorisedView, localPort, remotePort u
 	// On IPv4, UDP checksum is optional, and a zero value indicates the
 	// transmitter skipped the checksum generation (RFC768).
 	// On IPv6, UDP checksum is not optional (RFC2460 Section 8.1).
-	if r.Capabilities()&stack.CapabilityTXChecksumOffload == 0 &&
+	if r.RequiresTXTransportChecksum() &&
 		(!noChecksum || r.NetProto == header.IPv6ProtocolNumber) {
 		xsum := r.PseudoHeaderChecksum(ProtocolNumber, length)
 		for _, v := range data.Views() {
@@ -1382,10 +1382,11 @@ func (e *endpoint) Readiness(mask waiter.EventMask) waiter.EventMask {
 // On IPv4, UDP checksum is optional, and a zero value means the transmitter
 // omitted the checksum generation (RFC768).
 // On IPv6, UDP checksum is not optional (RFC2460 Section 8.1).
-func verifyChecksum(r *stack.Route, hdr header.UDP, pkt *stack.PacketBuffer) bool {
-	if r.Capabilities()&stack.CapabilityRXChecksumOffload == 0 &&
-		(hdr.Checksum() != 0 || r.NetProto == header.IPv6ProtocolNumber) {
-		xsum := r.PseudoHeaderChecksum(ProtocolNumber, hdr.Length())
+func verifyChecksum(hdr header.UDP, pkt *stack.PacketBuffer) bool {
+	if !pkt.RXTransportChecksumValidated &&
+		(hdr.Checksum() != 0 || pkt.NetworkProtocolNumber == header.IPv6ProtocolNumber) {
+		netHdr := pkt.Network()
+		xsum := header.PseudoHeaderChecksum(ProtocolNumber, netHdr.DestinationAddress(), netHdr.SourceAddress(), hdr.Length())
 		for _, v := range pkt.Data.Views() {
 			xsum = header.Checksum(v, xsum)
 		}
@@ -1396,7 +1397,7 @@ func verifyChecksum(r *stack.Route, hdr header.UDP, pkt *stack.PacketBuffer) boo
 
 // HandlePacket is called by the stack when new packets arrive to this transport
 // endpoint.
-func (e *endpoint) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pkt *stack.PacketBuffer) {
+func (e *endpoint) HandlePacket(id stack.TransportEndpointID, pkt *stack.PacketBuffer) {
 	// Get the header then trim it from the view.
 	hdr := header.UDP(pkt.TransportHeader().View())
 	if int(hdr.Length()) > pkt.Data.Size()+header.UDPMinimumSize {
@@ -1406,7 +1407,7 @@ func (e *endpoint) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pk
 		return
 	}
 
-	if !verifyChecksum(r, hdr, pkt) {
+	if !verifyChecksum(hdr, pkt) {
 		// Checksum Error.
 		e.stack.Stats().UDP.ChecksumErrors.Increment()
 		e.stats.ReceiveErrors.ChecksumErrors.Increment()
@@ -1437,7 +1438,7 @@ func (e *endpoint) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pk
 	// Push new packet into receive list and increment the buffer size.
 	packet := &udpPacket{
 		senderAddress: tcpip.FullAddress{
-			NIC:  r.NICID(),
+			NIC:  pkt.NICID,
 			Addr: id.RemoteAddress,
 			Port: header.UDP(hdr).SourcePort(),
 		},
@@ -1447,7 +1448,7 @@ func (e *endpoint) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pk
 	e.rcvBufSize += pkt.Data.Size()
 
 	// Save any useful information from the network header to the packet.
-	switch r.NetProto {
+	switch pkt.NetworkProtocolNumber {
 	case header.IPv4ProtocolNumber:
 		packet.tos, _ = header.IPv4(pkt.NetworkHeader().View()).TOS()
 	case header.IPv6ProtocolNumber:
@@ -1457,9 +1458,10 @@ func (e *endpoint) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pk
 	// TODO(gvisor.dev/issue/3556): r.LocalAddress may be a multicast or broadcast
 	// address. packetInfo.LocalAddr should hold a unicast address that can be
 	// used to respond to the incoming packet.
-	packet.packetInfo.LocalAddr = r.LocalAddress
-	packet.packetInfo.DestinationAddr = r.LocalAddress
-	packet.packetInfo.NIC = r.NICID()
+	localAddr := pkt.Network().DestinationAddress()
+	packet.packetInfo.LocalAddr = localAddr
+	packet.packetInfo.DestinationAddr = localAddr
+	packet.packetInfo.NIC = pkt.NICID
 	packet.timestamp = e.stack.Clock().NowNanoseconds()
 
 	e.rcvMu.Unlock()
diff --git a/pkg/tcpip/transport/udp/forwarder.go b/pkg/tcpip/transport/udp/forwarder.go
index 3ae6cc221..14e4648cd 100644
--- a/pkg/tcpip/transport/udp/forwarder.go
+++ b/pkg/tcpip/transport/udp/forwarder.go
@@ -43,10 +43,9 @@ func NewForwarder(s *stack.Stack, handler func(*ForwarderRequest)) *Forwarder {
 //
 // This function is expected to be passed as an argument to the
 // stack.SetTransportProtocolHandler function.
-func (f *Forwarder) HandlePacket(r *stack.Route, id stack.TransportEndpointID, pkt *stack.PacketBuffer) bool {
+func (f *Forwarder) HandlePacket(id stack.TransportEndpointID, pkt *stack.PacketBuffer) bool {
 	f.handler(&ForwarderRequest{
 		stack: f.stack,
-		route: r,
 		id:    id,
 		pkt:   pkt,
 	})
@@ -59,7 +58,6 @@ func (f *Forwarder) HandlePacket(r *stack.Route, id stack.TransportEndpointID, p
 // it via CreateEndpoint.
 type ForwarderRequest struct {
 	stack *stack.Stack
-	route *stack.Route
 	id    stack.TransportEndpointID
 	pkt   *stack.PacketBuffer
 }
@@ -72,17 +70,25 @@ func (r *ForwarderRequest) ID() stack.TransportEndpointID {
 
 // CreateEndpoint creates a connected UDP endpoint for the session request.
 func (r *ForwarderRequest) CreateEndpoint(queue *waiter.Queue) (tcpip.Endpoint, *tcpip.Error) {
-	ep := newEndpoint(r.stack, r.route.NetProto, queue)
-	if err := r.stack.RegisterTransportEndpoint(r.route.NICID(), []tcpip.NetworkProtocolNumber{r.route.NetProto}, ProtocolNumber, r.id, ep, ep.portFlags, ep.bindToDevice); err != nil {
+	netHdr := r.pkt.Network()
+	route, err := r.stack.FindRoute(r.pkt.NICID, netHdr.DestinationAddress(), netHdr.SourceAddress(), r.pkt.NetworkProtocolNumber, false /* multicastLoop */)
+	if err != nil {
+		return nil, err
+	}
+	route.ResolveWith(r.pkt.SourceLinkAddress())
+
+	ep := newEndpoint(r.stack, r.pkt.NetworkProtocolNumber, queue)
+	if err := r.stack.RegisterTransportEndpoint(r.pkt.NICID, []tcpip.NetworkProtocolNumber{r.pkt.NetworkProtocolNumber}, ProtocolNumber, r.id, ep, ep.portFlags, ep.bindToDevice); err != nil {
 		ep.Close()
+		route.Release()
 		return nil, err
 	}
 
 	ep.ID = r.id
-	ep.route = r.route.Clone()
+	ep.route = route
 	ep.dstPort = r.id.RemotePort
-	ep.effectiveNetProtos = []tcpip.NetworkProtocolNumber{r.route.NetProto}
-	ep.RegisterNICID = r.route.NICID()
+	ep.effectiveNetProtos = []tcpip.NetworkProtocolNumber{r.pkt.NetworkProtocolNumber}
+	ep.RegisterNICID = r.pkt.NICID
 	ep.boundPortFlags = ep.portFlags
 
 	ep.state = StateConnected
@@ -91,7 +97,7 @@ func (r *ForwarderRequest) CreateEndpoint(queue *waiter.Queue) (tcpip.Endpoint,
 	ep.rcvReady = true
 	ep.rcvMu.Unlock()
 
-	ep.HandlePacket(r.route, r.id, r.pkt)
+	ep.HandlePacket(r.id, r.pkt)
 
 	return ep, nil
 }
diff --git a/pkg/tcpip/transport/udp/protocol.go b/pkg/tcpip/transport/udp/protocol.go
index da5b1deb2..91420edd3 100644
--- a/pkg/tcpip/transport/udp/protocol.go
+++ b/pkg/tcpip/transport/udp/protocol.go
@@ -78,15 +78,15 @@ func (*protocol) ParsePorts(v buffer.View) (src, dst uint16, err *tcpip.Error) {
 
 // HandleUnknownDestinationPacket handles packets that are targeted at this
 // protocol but don't match any existing endpoint.
-func (p *protocol) HandleUnknownDestinationPacket(r *stack.Route, id stack.TransportEndpointID, pkt *stack.PacketBuffer) stack.UnknownDestinationPacketDisposition {
+func (p *protocol) HandleUnknownDestinationPacket(id stack.TransportEndpointID, pkt *stack.PacketBuffer) stack.UnknownDestinationPacketDisposition {
 	hdr := header.UDP(pkt.TransportHeader().View())
 	if int(hdr.Length()) > pkt.Data.Size()+header.UDPMinimumSize {
-		r.Stack().Stats().UDP.MalformedPacketsReceived.Increment()
+		p.stack.Stats().UDP.MalformedPacketsReceived.Increment()
 		return stack.UnknownDestinationPacketMalformed
 	}
 
-	if !verifyChecksum(r, hdr, pkt) {
-		r.Stack().Stats().UDP.ChecksumErrors.Increment()
+	if !verifyChecksum(hdr, pkt) {
+		p.stack.Stats().UDP.ChecksumErrors.Increment()
 		return stack.UnknownDestinationPacketMalformed
 	}