Remove link/packetsocket

This change removes NetworkDispatcher.DeliverOutboundPacket.

Since all packet writes go through the NIC (the only NetworkDispatcher),
we can deliver outgoing packets to interested packet endpoints before
writing the packet to the link endpoint as the stack expects that all
packets that get delivered to a link endpoint are transmitted on the
wire. That is, link endpoints no longer need to let the stack know when
it writes a packet as the stack already knows about the packet it writes
through a link endpoint.

PiperOrigin-RevId: 395761629

2 files changed, 51 insertions, 41 deletions

diff --git a/pkg/tcpip/stack/nic.go b/pkg/tcpip/stack/nic.go
index ddc1ddab6..bb56ca9cb 100644
--- a/pkg/tcpip/stack/nic.go
+++ b/pkg/tcpip/stack/nic.go
@@ -41,6 +41,17 @@ func (l *linkResolver) confirmReachable(addr tcpip.Address) {
 
 var _ NetworkInterface = (*nic)(nil)
 
+// TODO(https://gvisor.dev/issue/6558): Use an anonymous struct in nic for this
+// once copylocks supports anonymous structs.
+type packetEPs struct {
+	mu sync.RWMutex
+
+	// eps is protected by the mutex, but the values contained in it are not.
+	//
+	// +checklocks:mu
+	eps map[tcpip.NetworkProtocolNumber]*packetEndpointList
+}
+
 // nic represents a "network interface card" to which the networking stack is
 // attached.
 type nic struct {
@@ -72,10 +83,9 @@ type nic struct {
 		sync.RWMutex
 		spoofing    bool
 		promiscuous bool
-		// packetEPs is protected by mu, but the contained packetEndpointList are
-		// not.
-		packetEPs map[tcpip.NetworkProtocolNumber]*packetEndpointList
 	}
+
+	packetEPs packetEPs
 }
 
 // makeNICStats initializes the NIC statistics and associates them to the global
@@ -143,17 +153,21 @@ func newNIC(stack *Stack, id tcpip.NICID, name string, ep LinkEndpoint, ctx NICC
 		duplicateAddressDetectors: make(map[tcpip.NetworkProtocolNumber]DuplicateAddressDetector),
 	}
 	nic.linkResQueue.init(nic)
-	nic.mu.packetEPs = make(map[tcpip.NetworkProtocolNumber]*packetEndpointList)
+
+	nic.packetEPs.mu.Lock()
+	defer nic.packetEPs.mu.Unlock()
+
+	nic.packetEPs.eps = make(map[tcpip.NetworkProtocolNumber]*packetEndpointList)
 
 	resolutionRequired := ep.Capabilities()&CapabilityResolutionRequired != 0
 
 	// Register supported packet and network endpoint protocols.
 	for _, netProto := range header.Ethertypes {
-		nic.mu.packetEPs[netProto] = new(packetEndpointList)
+		nic.packetEPs.eps[netProto] = new(packetEndpointList)
 	}
 	for _, netProto := range stack.networkProtocols {
 		netNum := netProto.Number()
-		nic.mu.packetEPs[netNum] = new(packetEndpointList)
+		nic.packetEPs.eps[netNum] = new(packetEndpointList)
 
 		netEP := netProto.NewEndpoint(nic, nic)
 		nic.networkEndpoints[netNum] = netEP
@@ -365,6 +379,8 @@ func (n *nic) writePacket(r RouteInfo, protocol tcpip.NetworkProtocolNumber, pkt
 
 	pkt.EgressRoute = r
 	pkt.NetworkProtocolNumber = protocol
+	n.deliverOutboundPacket(r.RemoteLinkAddress, pkt)
+
 	if err := n.LinkEndpoint.WritePacket(r, protocol, pkt); err != nil {
 		return err
 	}
@@ -383,6 +399,7 @@ func (n *nic) writePackets(r RouteInfo, protocol tcpip.NetworkProtocolNumber, pk
 	for pkt := pkts.Front(); pkt != nil; pkt = pkt.Next() {
 		pkt.EgressRoute = r
 		pkt.NetworkProtocolNumber = protocol
+		n.deliverOutboundPacket(r.RemoteLinkAddress, pkt)
 	}
 
 	writtenPackets, err := n.LinkEndpoint.WritePackets(r, pkts, protocol)
@@ -699,12 +716,9 @@ func (n *nic) isInGroup(addr tcpip.Address) bool {
 // This rule applies only to the slice itself, not to the items of the slice;
 // the ownership of the items is not retained by the caller.
 func (n *nic) DeliverNetworkPacket(remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt *PacketBuffer) {
-	n.mu.RLock()
 	enabled := n.Enabled()
 	// If the NIC is not yet enabled, don't receive any packets.
 	if !enabled {
-		n.mu.RUnlock()
-
 		n.stats.disabledRx.packets.Increment()
 		n.stats.disabledRx.bytes.IncrementBy(uint64(pkt.Data().Size()))
 		return
@@ -715,7 +729,6 @@ func (n *nic) DeliverNetworkPacket(remote, local tcpip.LinkAddress, protocol tcp
 
 	networkEndpoint, ok := n.networkEndpoints[protocol]
 	if !ok {
-		n.mu.RUnlock()
 		n.stats.unknownL3ProtocolRcvdPackets.Increment()
 		return
 	}
@@ -727,12 +740,6 @@ func (n *nic) DeliverNetworkPacket(remote, local tcpip.LinkAddress, protocol tcp
 	}
 	pkt.RXTransportChecksumValidated = n.LinkEndpoint.Capabilities()&CapabilityRXChecksumOffload != 0
 
-	// Are any packet type sockets listening for this network protocol?
-	protoEPs := n.mu.packetEPs[protocol]
-	// Other packet type sockets that are listening for all protocols.
-	anyEPs := n.mu.packetEPs[header.EthernetProtocolAll]
-	n.mu.RUnlock()
-
 	// Deliver to interested packet endpoints without holding NIC lock.
 	var packetEPPkt *PacketBuffer
 	deliverPacketEPs := func(ep PacketEndpoint) {
@@ -758,24 +765,37 @@ func (n *nic) DeliverNetworkPacket(remote, local tcpip.LinkAddress, protocol tcp
 
 		ep.HandlePacket(n.id, local, protocol, packetEPPkt.Clone())
 	}
-	if protoEPs != nil {
+
+	n.packetEPs.mu.Lock()
+	// Are any packet type sockets listening for this network protocol?
+	protoEPs, protoEPsOK := n.packetEPs.eps[protocol]
+	// Other packet type sockets that are listening for all protocols.
+	anyEPs, anyEPsOK := n.packetEPs.eps[header.EthernetProtocolAll]
+	n.packetEPs.mu.Unlock()
+
+	if protoEPsOK {
 		protoEPs.forEach(deliverPacketEPs)
 	}
-	if anyEPs != nil {
+	if anyEPsOK {
 		anyEPs.forEach(deliverPacketEPs)
 	}
 
 	networkEndpoint.HandlePacket(pkt)
 }
 
-// DeliverOutboundPacket implements NetworkDispatcher.DeliverOutboundPacket.
-func (n *nic) DeliverOutboundPacket(remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt *PacketBuffer) {
-	n.mu.RLock()
+// deliverOutboundPacket delivers outgoing packets to interested endpoints.
+func (n *nic) deliverOutboundPacket(remote tcpip.LinkAddress, pkt *PacketBuffer) {
+	n.packetEPs.mu.RLock()
+	defer n.packetEPs.mu.RUnlock()
 	// We do not deliver to protocol specific packet endpoints as on Linux
 	// only ETH_P_ALL endpoints get outbound packets.
 	// Add any other packet sockets that maybe listening for all protocols.
-	eps := n.mu.packetEPs[header.EthernetProtocolAll]
-	n.mu.RUnlock()
+	eps, ok := n.packetEPs.eps[header.EthernetProtocolAll]
+	if !ok {
+		return
+	}
+
+	local := n.LinkAddress()
 
 	var packetEPPkt *PacketBuffer
 	eps.forEach(func(ep PacketEndpoint) {
@@ -796,11 +816,11 @@ func (n *nic) DeliverOutboundPacket(remote, local tcpip.LinkAddress, protocol tc
 			// Add the link layer header as outgoing packets are intercepted before
 			// the link layer header is created and packet endpoints are interested
 			// in the link header.
-			n.LinkEndpoint.AddHeader(local, remote, protocol, packetEPPkt)
+			n.LinkEndpoint.AddHeader(local, remote, pkt.NetworkProtocolNumber, packetEPPkt)
 			packetEPPkt.PktType = tcpip.PacketOutgoing
 		}
 
-		ep.HandlePacket(n.id, local, protocol, packetEPPkt.Clone())
+		ep.HandlePacket(n.id, local, pkt.NetworkProtocolNumber, packetEPPkt.Clone())
 	})
 }
 
@@ -953,10 +973,10 @@ func (n *nic) setNUDConfigs(protocol tcpip.NetworkProtocolNumber, c NUDConfigura
 }
 
 func (n *nic) registerPacketEndpoint(netProto tcpip.NetworkProtocolNumber, ep PacketEndpoint) tcpip.Error {
-	n.mu.Lock()
-	defer n.mu.Unlock()
+	n.packetEPs.mu.Lock()
+	defer n.packetEPs.mu.Unlock()
 
-	eps, ok := n.mu.packetEPs[netProto]
+	eps, ok := n.packetEPs.eps[netProto]
 	if !ok {
 		return &tcpip.ErrNotSupported{}
 	}
@@ -966,10 +986,10 @@ func (n *nic) registerPacketEndpoint(netProto tcpip.NetworkProtocolNumber, ep Pa
 }
 
 func (n *nic) unregisterPacketEndpoint(netProto tcpip.NetworkProtocolNumber, ep PacketEndpoint) {
-	n.mu.Lock()
-	defer n.mu.Unlock()
+	n.packetEPs.mu.Lock()
+	defer n.packetEPs.mu.Unlock()
 
-	eps, ok := n.mu.packetEPs[netProto]
+	eps, ok := n.packetEPs.eps[netProto]
 	if !ok {
 		return
 	}
diff --git a/pkg/tcpip/stack/registration.go b/pkg/tcpip/stack/registration.go
index 57b3348b2..113baaaae 100644
--- a/pkg/tcpip/stack/registration.go
+++ b/pkg/tcpip/stack/registration.go
@@ -733,16 +733,6 @@ type NetworkDispatcher interface {
 	//
 	// DeliverNetworkPacket takes ownership of pkt.
 	DeliverNetworkPacket(remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt *PacketBuffer)
-
-	// DeliverOutboundPacket is called by link layer when a packet is being
-	// sent out.
-	//
-	// pkt.LinkHeader may or may not be set before calling
-	// DeliverOutboundPacket. Some packets do not have link headers (e.g.
-	// packets sent via loopback), and won't have the field set.
-	//
-	// DeliverOutboundPacket takes ownership of pkt.
-	DeliverOutboundPacket(remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt *PacketBuffer)
 }
 
 // LinkEndpointCapabilities is the type associated with the capabilities