device: reduce redundant per-packet overhead in RX path

Peer.RoutineSequentialReceiver() deals with packet vectors and does not
need to perform timer and endpoint operations for every packet in a
given vector. Changing these per-packet operations to per-vector
improves throughput by as much as 10% in some environments.

Signed-off-by: Jordan Whited <jordan@tailscale.com>

Author: jwhited

device/receive.go

@@ -445,7 +445,9 @@ func (peer *Peer) RoutineSequentialReceiver(maxBatchSize int) {
 			return
 		}
 		elemsContainer.Lock()
-		for _, elem := range elemsContainer.elems {
+		validTailPacket := -1
+		dataPacketReceived := false
+		for i, elem := range elemsContainer.elems {
 			if elem.packet == nil {
 				// decryption failed
 				continue
@@ -455,21 +457,19 @@ func (peer *Peer) RoutineSequentialReceiver(maxBatchSize int) {
 				continue
 			}
 
-			peer.SetEndpointFromPacket(elem.endpoint)
+			validTailPacket = i
 			if peer.ReceivedWithKeypair(elem.keypair) {
+				peer.SetEndpointFromPacket(elem.endpoint)
 				peer.timersHandshakeComplete()
 				peer.SendStagedPackets()
 			}
-			peer.keepKeyFreshReceiving()
-			peer.timersAnyAuthenticatedPacketTraversal()
-			peer.timersAnyAuthenticatedPacketReceived()
 			peer.rxBytes.Add(uint64(len(elem.packet) + MinMessageSize))
 
 			if len(elem.packet) == 0 {
 				device.log.Verbosef("%v - Receiving keepalive packet", peer)
 				continue
 			}
-			peer.timersDataReceived()
+			dataPacketReceived = true
 
 			switch elem.packet[0] >> 4 {
 			case 4:
@@ -512,6 +512,15 @@ func (peer *Peer) RoutineSequentialReceiver(maxBatchSize int) {
 
 			bufs = append(bufs, elem.buffer[:MessageTransportOffsetContent+len(elem.packet)])
 		}
+		if validTailPacket >= 0 {
+			peer.SetEndpointFromPacket(elemsContainer.elems[validTailPacket].endpoint)
+			peer.keepKeyFreshReceiving()
+			peer.timersAnyAuthenticatedPacketTraversal()
+			peer.timersAnyAuthenticatedPacketReceived()
+		}
+		if dataPacketReceived {
+			peer.timersDataReceived()
+		}
 		if len(bufs) > 0 {
 			_, err := device.tun.device.Write(bufs, MessageTransportOffsetContent)
 			if err != nil && !device.isClosed() {

device/send.go

@@ -436,7 +436,7 @@ func calculatePaddingSize(packetSize, mtu int) int {
 	return paddedSize - lastUnit
 }
 
-/* Encrypts the elems in the queue
+/* Encrypts the elements in the queue
  * and marks them for sequential consumption (by releasing the mutex)
  *
  * Obs. One instance per core
@@ -495,7 +495,7 @@ func (peer *Peer) RoutineSequentialSender(maxBatchSize int) {
 			return
 		}
 		if !peer.isRunning.Load() {
-			// peer has been stopped; return re-usable elemsContainer to the shared pool.
+			// peer has been stopped; return re-usable elems to the shared pool.
 			// This is an optimization only. It is possible for the peer to be stopped
 			// immediately after this check, in which case, elem will get processed.
 			// The timers and SendBuffers code are resilient to a few stragglers.