Use PacketBuffers, rather than VectorisedViews, in netstack.

PacketBuffers are analogous to Linux's sk_buff. They hold all information about
a packet, headers, and payload. This is important for:

* iptables to access various headers of packets
* Preventing the clutter of passing different net and link headers along with
  VectorisedViews to packet handling functions.

This change only affects the incoming packet path, and a future change will
change the outgoing path.

Benchmark               Regular         PacketBufferPtr  PacketBufferConcrete
--------------------------------------------------------------------------------
BM_Recvmsg             400.715MB/s      373.676MB/s      396.276MB/s
BM_Sendmsg             361.832MB/s      333.003MB/s      335.571MB/s
BM_Recvfrom            453.336MB/s      393.321MB/s      381.650MB/s
BM_Sendto              378.052MB/s      372.134MB/s      341.342MB/s
BM_SendmsgTCP/0/1k     353.711MB/s      316.216MB/s      322.747MB/s
BM_SendmsgTCP/0/2k     600.681MB/s      588.776MB/s      565.050MB/s
BM_SendmsgTCP/0/4k     995.301MB/s      888.808MB/s      941.888MB/s
BM_SendmsgTCP/0/8k     1.517GB/s        1.274GB/s        1.345GB/s
BM_SendmsgTCP/0/16k    1.872GB/s        1.586GB/s        1.698GB/s
BM_SendmsgTCP/0/32k    1.017GB/s        1.020GB/s        1.133GB/s
BM_SendmsgTCP/0/64k    475.626MB/s      584.587MB/s      627.027MB/s
BM_SendmsgTCP/0/128k   416.371MB/s      503.434MB/s      409.850MB/s
BM_SendmsgTCP/0/256k   323.449MB/s      449.599MB/s      388.852MB/s
BM_SendmsgTCP/0/512k   243.992MB/s      267.676MB/s      314.474MB/s
BM_SendmsgTCP/0/1M     95.138MB/s       95.874MB/s       95.417MB/s
BM_SendmsgTCP/0/2M     96.261MB/s       94.977MB/s       96.005MB/s
BM_SendmsgTCP/0/4M     96.512MB/s       95.978MB/s       95.370MB/s
BM_SendmsgTCP/0/8M     95.603MB/s       95.541MB/s       94.935MB/s
BM_SendmsgTCP/0/16M    94.598MB/s       94.696MB/s       94.521MB/s
BM_SendmsgTCP/0/32M    94.006MB/s       94.671MB/s       94.768MB/s
BM_SendmsgTCP/0/64M    94.133MB/s       94.333MB/s       94.746MB/s
BM_SendmsgTCP/0/128M   93.615MB/s       93.497MB/s       93.573MB/s
BM_SendmsgTCP/0/256M   93.241MB/s       95.100MB/s       93.272MB/s
BM_SendmsgTCP/1/1k     303.644MB/s      316.074MB/s      308.430MB/s
BM_SendmsgTCP/1/2k     537.093MB/s      584.962MB/s      529.020MB/s
BM_SendmsgTCP/1/4k     882.362MB/s      939.087MB/s      892.285MB/s
BM_SendmsgTCP/1/8k     1.272GB/s        1.394GB/s        1.296GB/s
BM_SendmsgTCP/1/16k    1.802GB/s        2.019GB/s        1.830GB/s
BM_SendmsgTCP/1/32k    2.084GB/s        2.173GB/s        2.156GB/s
BM_SendmsgTCP/1/64k    2.515GB/s        2.463GB/s        2.473GB/s
BM_SendmsgTCP/1/128k   2.811GB/s        3.004GB/s        2.946GB/s
BM_SendmsgTCP/1/256k   3.008GB/s        3.159GB/s        3.171GB/s
BM_SendmsgTCP/1/512k   2.980GB/s        3.150GB/s        3.126GB/s
BM_SendmsgTCP/1/1M     2.165GB/s        2.233GB/s        2.163GB/s
BM_SendmsgTCP/1/2M     2.370GB/s        2.219GB/s        2.453GB/s
BM_SendmsgTCP/1/4M     2.005GB/s        2.091GB/s        2.214GB/s
BM_SendmsgTCP/1/8M     2.111GB/s        2.013GB/s        2.109GB/s
BM_SendmsgTCP/1/16M    1.902GB/s        1.868GB/s        1.897GB/s
BM_SendmsgTCP/1/32M    1.655GB/s        1.665GB/s        1.635GB/s
BM_SendmsgTCP/1/64M    1.575GB/s        1.547GB/s        1.575GB/s
BM_SendmsgTCP/1/128M   1.524GB/s        1.584GB/s        1.580GB/s
BM_SendmsgTCP/1/256M   1.579GB/s        1.607GB/s        1.593GB/s

PiperOrigin-RevId: 278940079

3 files changed, 48 insertions, 33 deletions

diff --git a/pkg/tcpip/network/ipv4/icmp.go b/pkg/tcpip/network/ipv4/icmp.go
index 50b363dc4..ce771631c 100644
--- a/pkg/tcpip/network/ipv4/icmp.go
+++ b/pkg/tcpip/network/ipv4/icmp.go
@@ -15,6 +15,7 @@
 package ipv4
 
 import (
+	"gvisor.dev/gvisor/pkg/tcpip"
 	"gvisor.dev/gvisor/pkg/tcpip/buffer"
 	"gvisor.dev/gvisor/pkg/tcpip/header"
 	"gvisor.dev/gvisor/pkg/tcpip/stack"
@@ -24,8 +25,8 @@ import (
 // the original packet that caused the ICMP one to be sent. This information is
 // used to find out which transport endpoint must be notified about the ICMP
 // packet.
-func (e *endpoint) handleControl(typ stack.ControlType, extra uint32, vv buffer.VectorisedView) {
-	h := header.IPv4(vv.First())
+func (e *endpoint) handleControl(typ stack.ControlType, extra uint32, pkt tcpip.PacketBuffer) {
+	h := header.IPv4(pkt.Data.First())
 
 	// We don't use IsValid() here because ICMP only requires that the IP
 	// header plus 8 bytes of the transport header be included. So it's
@@ -39,7 +40,7 @@ func (e *endpoint) handleControl(typ stack.ControlType, extra uint32, vv buffer.
 	}
 
 	hlen := int(h.HeaderLength())
-	if vv.Size() < hlen || h.FragmentOffset() != 0 {
+	if pkt.Data.Size() < hlen || h.FragmentOffset() != 0 {
 		// We won't be able to handle this if it doesn't contain the
 		// full IPv4 header, or if it's a fragment not at offset 0
 		// (because it won't have the transport header).
@@ -47,15 +48,15 @@ func (e *endpoint) handleControl(typ stack.ControlType, extra uint32, vv buffer.
 	}
 
 	// Skip the ip header, then deliver control message.
-	vv.TrimFront(hlen)
+	pkt.Data.TrimFront(hlen)
 	p := h.TransportProtocol()
-	e.dispatcher.DeliverTransportControlPacket(e.id.LocalAddress, h.DestinationAddress(), ProtocolNumber, p, typ, extra, vv)
+	e.dispatcher.DeliverTransportControlPacket(e.id.LocalAddress, h.DestinationAddress(), ProtocolNumber, p, typ, extra, pkt)
 }
 
-func (e *endpoint) handleICMP(r *stack.Route, netHeader buffer.View, vv buffer.VectorisedView) {
+func (e *endpoint) handleICMP(r *stack.Route, pkt tcpip.PacketBuffer) {
 	stats := r.Stats()
 	received := stats.ICMP.V4PacketsReceived
-	v := vv.First()
+	v := pkt.Data.First()
 	if len(v) < header.ICMPv4MinimumSize {
 		received.Invalid.Increment()
 		return
@@ -73,20 +74,23 @@ func (e *endpoint) handleICMP(r *stack.Route, netHeader buffer.View, vv buffer.V
 		// checksum. We'll have to reset this before we hand the packet
 		// off.
 		h.SetChecksum(0)
-		gotChecksum := ^header.ChecksumVV(vv, 0 /* initial */)
+		gotChecksum := ^header.ChecksumVV(pkt.Data, 0 /* initial */)
 		if gotChecksum != wantChecksum {
 			// It's possible that a raw socket expects to receive this.
 			h.SetChecksum(wantChecksum)
-			e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, netHeader, vv)
+			e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, pkt)
 			received.Invalid.Increment()
 			return
 		}
 
 		// It's possible that a raw socket expects to receive this.
 		h.SetChecksum(wantChecksum)
-		e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, netHeader, vv)
+		e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, tcpip.PacketBuffer{
+			Data:          pkt.Data.Clone(nil),
+			NetworkHeader: append(buffer.View(nil), pkt.NetworkHeader...),
+		})
 
-		vv := vv.Clone(nil)
+		vv := pkt.Data.Clone(nil)
 		vv.TrimFront(header.ICMPv4MinimumSize)
 		hdr := buffer.NewPrependable(int(r.MaxHeaderLength()) + header.ICMPv4MinimumSize)
 		pkt := header.ICMPv4(hdr.Prepend(header.ICMPv4MinimumSize))
@@ -104,19 +108,19 @@ func (e *endpoint) handleICMP(r *stack.Route, netHeader buffer.View, vv buffer.V
 	case header.ICMPv4EchoReply:
 		received.EchoReply.Increment()
 
-		e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, netHeader, vv)
+		e.dispatcher.DeliverTransportPacket(r, header.ICMPv4ProtocolNumber, pkt)
 
 	case header.ICMPv4DstUnreachable:
 		received.DstUnreachable.Increment()
 
-		vv.TrimFront(header.ICMPv4MinimumSize)
+		pkt.Data.TrimFront(header.ICMPv4MinimumSize)
 		switch h.Code() {
 		case header.ICMPv4PortUnreachable:
-			e.handleControl(stack.ControlPortUnreachable, 0, vv)
+			e.handleControl(stack.ControlPortUnreachable, 0, pkt)
 
 		case header.ICMPv4FragmentationNeeded:
 			mtu := uint32(h.MTU())
-			e.handleControl(stack.ControlPacketTooBig, calculateMTU(mtu), vv)
+			e.handleControl(stack.ControlPacketTooBig, calculateMTU(mtu), pkt)
 		}
 
 	case header.ICMPv4SrcQuench:
diff --git a/pkg/tcpip/network/ipv4/ipv4.go b/pkg/tcpip/network/ipv4/ipv4.go
index 1339f8474..26f1402ed 100644
--- a/pkg/tcpip/network/ipv4/ipv4.go
+++ b/pkg/tcpip/network/ipv4/ipv4.go
@@ -198,7 +198,7 @@ func (e *endpoint) writePacketFragments(r *stack.Route, gso *stack.GSO, hdr buff
 	return nil
 }
 
-func (e *endpoint) addIPHeader(r *stack.Route, hdr *buffer.Prependable, payloadSize int, params stack.NetworkHeaderParams) {
+func (e *endpoint) addIPHeader(r *stack.Route, hdr *buffer.Prependable, payloadSize int, params stack.NetworkHeaderParams) header.IPv4 {
 	ip := header.IPv4(hdr.Prepend(header.IPv4MinimumSize))
 	length := uint16(hdr.UsedLength() + payloadSize)
 	id := uint32(0)
@@ -218,19 +218,24 @@ func (e *endpoint) addIPHeader(r *stack.Route, hdr *buffer.Prependable, payloadS
 		DstAddr:     r.RemoteAddress,
 	})
 	ip.SetChecksum(^ip.CalculateChecksum())
+	return ip
 }
 
 // WritePacket writes a packet to the given destination address and protocol.
 func (e *endpoint) WritePacket(r *stack.Route, gso *stack.GSO, hdr buffer.Prependable, payload buffer.VectorisedView, params stack.NetworkHeaderParams, loop stack.PacketLooping) *tcpip.Error {
-	e.addIPHeader(r, &hdr, payload.Size(), params)
+	ip := e.addIPHeader(r, &hdr, payload.Size(), params)
 
 	if loop&stack.PacketLoop != 0 {
 		views := make([]buffer.View, 1, 1+len(payload.Views()))
 		views[0] = hdr.View()
 		views = append(views, payload.Views()...)
-		vv := buffer.NewVectorisedView(len(views[0])+payload.Size(), views)
 		loopedR := r.MakeLoopedRoute()
-		e.HandlePacket(&loopedR, vv)
+
+		e.HandlePacket(&loopedR, tcpip.PacketBuffer{
+			Data:          buffer.NewVectorisedView(len(views[0])+payload.Size(), views),
+			NetworkHeader: buffer.View(ip),
+		})
+
 		loopedR.Release()
 	}
 	if loop&stack.PacketOut == 0 {
@@ -301,7 +306,10 @@ func (e *endpoint) WriteHeaderIncludedPacket(r *stack.Route, payload buffer.Vect
 	ip.SetChecksum(^ip.CalculateChecksum())
 
 	if loop&stack.PacketLoop != 0 {
-		e.HandlePacket(r, payload)
+		e.HandlePacket(r, tcpip.PacketBuffer{
+			Data:          payload,
+			NetworkHeader: buffer.View(ip),
+		})
 	}
 	if loop&stack.PacketOut == 0 {
 		return nil
@@ -314,22 +322,23 @@ func (e *endpoint) WriteHeaderIncludedPacket(r *stack.Route, payload buffer.Vect
 
 // HandlePacket is called by the link layer when new ipv4 packets arrive for
 // this endpoint.
-func (e *endpoint) HandlePacket(r *stack.Route, vv buffer.VectorisedView) {
-	headerView := vv.First()
+func (e *endpoint) HandlePacket(r *stack.Route, pkt tcpip.PacketBuffer) {
+	headerView := pkt.Data.First()
 	h := header.IPv4(headerView)
-	if !h.IsValid(vv.Size()) {
+	if !h.IsValid(pkt.Data.Size()) {
 		r.Stats().IP.MalformedPacketsReceived.Increment()
 		return
 	}
+	pkt.NetworkHeader = headerView[:h.HeaderLength()]
 
 	hlen := int(h.HeaderLength())
 	tlen := int(h.TotalLength())
-	vv.TrimFront(hlen)
-	vv.CapLength(tlen - hlen)
+	pkt.Data.TrimFront(hlen)
+	pkt.Data.CapLength(tlen - hlen)
 
 	more := (h.Flags() & header.IPv4FlagMoreFragments) != 0
 	if more || h.FragmentOffset() != 0 {
-		if vv.Size() == 0 {
+		if pkt.Data.Size() == 0 {
 			// Drop the packet as it's marked as a fragment but has
 			// no payload.
 			r.Stats().IP.MalformedPacketsReceived.Increment()
@@ -337,10 +346,10 @@ func (e *endpoint) HandlePacket(r *stack.Route, vv buffer.VectorisedView) {
 			return
 		}
 		// The packet is a fragment, let's try to reassemble it.
-		last := h.FragmentOffset() + uint16(vv.Size()) - 1
+		last := h.FragmentOffset() + uint16(pkt.Data.Size()) - 1
 		// Drop the packet if the fragmentOffset is incorrect. i.e the
-		// combination of fragmentOffset and vv.size() causes a wrap
-		// around resulting in last being less than the offset.
+		// combination of fragmentOffset and pkt.Data.size() causes a
+		// wrap around resulting in last being less than the offset.
 		if last < h.FragmentOffset() {
 			r.Stats().IP.MalformedPacketsReceived.Increment()
 			r.Stats().IP.MalformedFragmentsReceived.Increment()
@@ -348,7 +357,7 @@ func (e *endpoint) HandlePacket(r *stack.Route, vv buffer.VectorisedView) {
 		}
 		var ready bool
 		var err error
-		vv, ready, err = e.fragmentation.Process(hash.IPv4FragmentHash(h), h.FragmentOffset(), last, more, vv)
+		pkt.Data, ready, err = e.fragmentation.Process(hash.IPv4FragmentHash(h), h.FragmentOffset(), last, more, pkt.Data)
 		if err != nil {
 			r.Stats().IP.MalformedPacketsReceived.Increment()
 			r.Stats().IP.MalformedFragmentsReceived.Increment()
@@ -361,11 +370,11 @@ func (e *endpoint) HandlePacket(r *stack.Route, vv buffer.VectorisedView) {
 	p := h.TransportProtocol()
 	if p == header.ICMPv4ProtocolNumber {
 		headerView.CapLength(hlen)
-		e.handleICMP(r, headerView, vv)
+		e.handleICMP(r, pkt)
 		return
 	}
 	r.Stats().IP.PacketsDelivered.Increment()
-	e.dispatcher.DeliverTransportPacket(r, p, headerView, vv)
+	e.dispatcher.DeliverTransportPacket(r, p, pkt)
 }
 
 // Close cleans up resources associated with the endpoint.
diff --git a/pkg/tcpip/network/ipv4/ipv4_test.go b/pkg/tcpip/network/ipv4/ipv4_test.go
index 99f84acd7..f100d84ee 100644
--- a/pkg/tcpip/network/ipv4/ipv4_test.go
+++ b/pkg/tcpip/network/ipv4/ipv4_test.go
@@ -464,7 +464,9 @@ func TestInvalidFragments(t *testing.T) {
 			s.CreateNIC(nicid, sniffer.New(ep))
 
 			for _, pkt := range tc.packets {
-				ep.InjectLinkAddr(header.IPv4ProtocolNumber, remoteLinkAddr, buffer.NewVectorisedView(len(pkt), []buffer.View{pkt}))
+				ep.InjectLinkAddr(header.IPv4ProtocolNumber, remoteLinkAddr, tcpip.PacketBuffer{
+					Data: buffer.NewVectorisedView(len(pkt), []buffer.View{pkt}),
+				})
 			}
 
 			if got, want := s.Stats().IP.MalformedPacketsReceived.Value(), tc.wantMalformedIPPackets; got != want {