Another strict size workaround

Reviewed-on: #14

README.md

@@ -62,3 +62,7 @@ sudo setcap 'CAP_NET_BIND_SERVICE=ep CAP_NET_RAW=ep' ~/go/bin/jrouter
 * `NET_RAW` is needed for EtherTalk
 
 TODO: instructions for non-Linux machines
+
+## Bug reports? Feature requests? Complaints? Praise? 
+
+You can contact me on the Fediverse at @DrJosh9000@cloudisland.nz, or email me at josh.deprez@gmail.com. 

aurp/routing_info.go

@@ -173,9 +173,12 @@ func (e EventTuples) WriteTo(w io.Writer) (int64, error) {
 }
 
 func parseEventTuples(p []byte) (EventTuples, error) {
-	// Each event tuple is at least 4 bytes, so we need to store at most
+	// Event tuples can be 1, 4, or 6 bytes long. But the only type of length 1
-	// len(p)/4 of them.
+	// is the Null event type sent to probe whether or not the data receiver is
-	e := make(EventTuples, 0, len(p)/4)
+	// still listening. If that's present there probably aren't any other
+	// tuples. Hence len(p)/4 (rounded up) is a reasonable estimate of max tuple
+	// count.
+	e := make(EventTuples, 0, (len(p)+3)/4)
 	for len(p) > 0 {
 		et, nextp, err := parseEventTuple(p)
 		if err != nil {
@@ -198,6 +201,10 @@ type EventTuple struct {
 func (et *EventTuple) WriteTo(w io.Writer) (int64, error) {
 	a := acc(w)
 	a.write8(uint8(et.EventCode))
+	if et.EventCode == EventCodeNull {
+		// null tuple
+		return a.ret()
+	}
 	a.write16(uint16(et.RangeStart))
 	if !et.Extended {
 		// non-extended tuple
@@ -211,12 +218,18 @@ func (et *EventTuple) WriteTo(w io.Writer) (int64, error) {
 }
 
 func parseEventTuple(p []byte) (EventTuple, []byte, error) {
-	if len(p) < 4 {
+	if len(p) < 1 {
-		return EventTuple{}, p, fmt.Errorf("insufficient input length %d for network event tuple", len(p))
+		return EventTuple{}, p, fmt.Errorf("insufficient input length %d for any network event tuple", len(p))
 	}
 
 	var et EventTuple
 	et.EventCode = EventCode(p[0])
+	if et.EventCode == EventCodeNull {
+		return et, p[1:], nil
+	}
+	if len(p) < 4 {
+		return EventTuple{}, p, fmt.Errorf("insufficient input length %d for non-Null network event tuple", len(p))
+	}
 	et.RangeStart = ddp.Network(binary.BigEndian.Uint16(p[1:3]))
 	et.RangeEnd = et.RangeStart
 	et.Distance = p[3]

main.go

@@ -87,6 +87,11 @@ const peerTableTemplate = `
 		<th>Remote addr</th>
 		<th>Receiver state</th>
 		<th>Sender state</th>
+		<th>Last heard from</th>
+		<th>Last reconnect</th>
+		<th>Last update</th>
+		<th>Last send</th>
+		<th>Send retries</th>
 	</tr></thead>
 	<tbody>
 {{range $peer := . }}
@@ -95,6 +100,11 @@ const peerTableTemplate = `
 		<td>{{$peer.RemoteAddr}}</td>
 		<td>{{$peer.ReceiverState}}</td>
 		<td>{{$peer.SenderState}}</td>
+		<td>{{$peer.LastHeardFromAgo}}</td>
+		<td>{{$peer.LastReconnectAgo}}</td>
+		<td>{{$peer.LastUpdateAgo}}</td>
+		<td>{{$peer.LastSendAgo}}</td>
+		<td>{{$peer.SendRetries}}</td>		
 	</tr>
 {{end}}
 	</tbody>
@@ -287,18 +297,7 @@ func main() {
 			continue
 		}
 
-		peer := &router.AURPPeer{
+		peer := router.NewAURPPeer(routes, ln, peerStr, raddr, localDI, nil, nextConnID)
-			Transport: &aurp.Transport{
-				LocalDI:     localDI,
-				RemoteDI:    aurp.IPDomainIdentifier(raddr.IP),
-				LocalConnID: nextConnID,
-			},
-			UDPConn:        ln,
-			ConfiguredAddr: peerStr,
-			RemoteAddr:     raddr,
-			ReceiveCh:      make(chan aurp.Packet, 1024),
-			RouteTable:     routes,
-		}
 		aurp.Inc(&nextConnID)
 		peersMu.Lock()
 		peers[udpAddrFromNet(raddr)] = peer
@@ -394,17 +393,7 @@ func main() {
 					continue
 				}
 				// New peer!
-				pr = &router.AURPPeer{
+				pr = router.NewAURPPeer(routes, ln, "", raddr, localDI, dh.SourceDI, nextConnID)
-					Transport: &aurp.Transport{
-						LocalDI:     localDI,
-						RemoteDI:    dh.SourceDI, // platinum rule
-						LocalConnID: nextConnID,
-					},
-					UDPConn:    ln,
-					RemoteAddr: raddr,
-					ReceiveCh:  make(chan aurp.Packet, 1024),
-					RouteTable: routes,
-				}
 				aurp.Inc(&nextConnID)
 				peers[ra] = pr
 				goPeerHandler(pr)

router/aarp.go

@@ -37,6 +37,8 @@ const (
 	maxAMTEntryAge        = 30 * time.Second
 	aarpRequestRetransmit = 1 * time.Second
 	aarpRequestTimeout    = 10 * time.Second
+
+	aarpBodyLength = 28 // bytes
 )
 
 const aarpStatusTemplate = `
@@ -181,8 +183,17 @@ func (a *AARPMachine) Run(ctx context.Context) error {
 				a.incomingCh = nil
 			}
 
+			// sfiera/multitalk will return an "excess data" error if the
+			// payload is too big. Most traffic I've seen locally does not have
+			// this problem, but I've seen one report with some junk trailing
+			// data on AARP packets.
+			payload := ethFrame.Payload
+			if len(payload) > aarpBodyLength {
+				payload = payload[:aarpBodyLength]
+			}
+
 			var aapkt aarp.Packet
-			if err := aarp.Unmarshal(ethFrame.Payload, &aapkt); err != nil {
+			if err := aarp.Unmarshal(payload, &aapkt); err != nil {
 				log.Printf("Couldn't unmarshal AARP packet: %v", err)
 				continue
 			}
@@ -378,10 +389,7 @@ func (e AMTEntry) Valid() bool {
 // LastUpdatedAgo is a friendly string reporting how long ago the entry was
 // updated/resolved.
 func (e AMTEntry) LastUpdatedAgo() string {
-	if e.LastUpdated.IsZero() {
+	return ago(e.LastUpdated)
-		return "never"
-	}
-	return fmt.Sprintf("%v ago", time.Since(e.LastUpdated).Truncate(time.Millisecond))
 }
 
 // addressMappingTable implements a concurrent-safe Address Mapping Table for

router/misc.go

@@ -16,6 +16,11 @@
 
 package router
 
+import (
+	"fmt"
+	"time"
+)
+
 // StringSet is a set of strings.
 // Yep, yet another string set implementation. Took me 2 minutes to write *shrug*
 type StringSet map[string]struct{}
@@ -50,3 +55,11 @@ func SetFromSlice(ss []string) StringSet {
 	set.Insert(ss...)
 	return set
 }
+
+// ago is a helper for formatting times.
+func ago(t time.Time) string {
+	if t.IsZero() {
+		return "never"
+	}
+	return fmt.Sprintf("%v ago", time.Since(t).Truncate(time.Millisecond))
+}

router/peer_aurp.go

@@ -114,9 +114,75 @@ type AURPPeer struct {
 	// Route table (the peer will add/remove/update routes and zones)
 	RouteTable *RouteTable
 
-	mu     sync.RWMutex
+	// Event tuples yet to be sent to this peer in an RI-Upd.
-	rstate ReceiverState
+	pendingEventsMu sync.Mutex
-	sstate SenderState
+	pendingEvents   aurp.EventTuples
+
+	// The internal states below are only set within the Handle loop, but can
+	// be read concurrently from outside.
+	mu            sync.RWMutex
+	rstate        ReceiverState
+	sstate        SenderState
+	lastReconnect time.Time
+	lastHeardFrom time.Time
+	lastSend      time.Time // TODO: clarify use of lastSend / sendRetries
+	lastUpdate    time.Time
+	sendRetries   int
+}
+
+func NewAURPPeer(routes *RouteTable, udpConn *net.UDPConn, peerAddr string, raddr *net.UDPAddr, localDI, remoteDI aurp.DomainIdentifier, connID uint16) *AURPPeer {
+	if remoteDI == nil {
+		remoteDI = aurp.IPDomainIdentifier(raddr.IP)
+	}
+	return &AURPPeer{
+		Transport: &aurp.Transport{
+			LocalDI:     localDI,
+			RemoteDI:    remoteDI,
+			LocalConnID: connID,
+		},
+		UDPConn:        udpConn,
+		ConfiguredAddr: peerAddr,
+		RemoteAddr:     raddr,
+		ReceiveCh:      make(chan aurp.Packet, 1024),
+		RouteTable:     routes,
+	}
+}
+
+func (p *AURPPeer) addPendingEvent(ec aurp.EventCode, route *Route) {
+	// Don't advertise routes to AURP peers to other AURP peers
+	if route.AURPPeer != nil {
+		return
+	}
+	et := aurp.EventTuple{
+		EventCode:  ec,
+		Extended:   route.Extended,
+		RangeStart: route.NetStart,
+		Distance:   route.Distance,
+		RangeEnd:   route.NetEnd,
+	}
+	switch ec {
+	case aurp.EventCodeND, aurp.EventCodeNRC:
+		et.Distance = 0 // "The distance field does not apply to ND or NRC event tuples and should be set to 0."
+	}
+	p.pendingEventsMu.Lock()
+	defer p.pendingEventsMu.Unlock()
+	p.pendingEvents = append(p.pendingEvents, et)
+}
+
+func (p *AURPPeer) RouteAdded(route *Route) {
+	p.addPendingEvent(aurp.EventCodeNA, route)
+}
+
+func (p *AURPPeer) RouteDeleted(route *Route) {
+	p.addPendingEvent(aurp.EventCodeND, route)
+}
+
+func (p *AURPPeer) RouteDistanceChanged(route *Route) {
+	p.addPendingEvent(aurp.EventCodeNDC, route)
+}
+
+func (p *AURPPeer) RouteForwarderChanged(route *Route) {
+	p.addPendingEvent(aurp.EventCodeNRC, route)
 }
 
 func (p *AURPPeer) Forward(ddpkt *ddp.ExtPacket) error {
@@ -124,7 +190,7 @@ func (p *AURPPeer) Forward(ddpkt *ddp.ExtPacket) error {
 	if err != nil {
 		return err
 	}
-	_, err = p.Send(p.Transport.NewAppleTalkPacket(outPkt))
+	_, err = p.send(p.Transport.NewAppleTalkPacket(outPkt))
 	return err
 }
 
@@ -140,6 +206,36 @@ func (p *AURPPeer) SenderState() SenderState {
 	return p.sstate
 }
 
+func (p *AURPPeer) LastReconnectAgo() string {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return ago(p.lastReconnect)
+}
+
+func (p *AURPPeer) LastHeardFromAgo() string {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return ago(p.lastHeardFrom)
+}
+
+func (p *AURPPeer) LastSendAgo() string {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return ago(p.lastSend)
+}
+
+func (p *AURPPeer) LastUpdateAgo() string {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return ago(p.lastUpdate)
+}
+
+func (p *AURPPeer) SendRetries() int {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return p.sendRetries
+}
+
 func (p *AURPPeer) setRState(rstate ReceiverState) {
 	p.mu.Lock()
 	defer p.mu.Unlock()
@@ -152,6 +248,42 @@ func (p *AURPPeer) setSState(sstate SenderState) {
 	p.sstate = sstate
 }
 
+func (p *AURPPeer) incSendRetries() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.sendRetries++
+}
+
+func (p *AURPPeer) resetSendRetries() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.sendRetries = 0
+}
+
+func (p *AURPPeer) bumpLastHeardFrom() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.lastHeardFrom = time.Now()
+}
+
+func (p *AURPPeer) bumpLastReconnect() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.lastReconnect = time.Now()
+}
+
+func (p *AURPPeer) bumpLastSend() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.lastSend = time.Now()
+}
+
+func (p *AURPPeer) bumpLastUpdate() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.lastUpdate = time.Now()
+}
+
 func (p *AURPPeer) disconnect() {
 	p.mu.Lock()
 	defer p.mu.Unlock()
@@ -159,8 +291,8 @@ func (p *AURPPeer) disconnect() {
 	p.sstate = SenderUnconnected
 }
 
-// Send encodes and sends pkt to the remote host.
+// send encodes and sends pkt to the remote host.
-func (p *AURPPeer) Send(pkt aurp.Packet) (int, error) {
+func (p *AURPPeer) send(pkt aurp.Packet) (int, error) {
 	var b bytes.Buffer
 	if _, err := pkt.WriteTo(&b); err != nil {
 		return 0, err
@@ -170,23 +302,28 @@ func (p *AURPPeer) Send(pkt aurp.Packet) (int, error) {
 }
 
 func (p *AURPPeer) Handle(ctx context.Context) error {
+	// Stop listening to events if the goroutine exits
+	defer p.RouteTable.RemoveObserver(p)
+
 	rticker := time.NewTicker(1 * time.Second)
 	defer rticker.Stop()
 	sticker := time.NewTicker(1 * time.Second)
 	defer sticker.Stop()
 
-	lastReconnect := time.Now()
+	p.mu.Lock()
-	lastHeardFrom := time.Now()
+	p.lastReconnect = time.Now()
-	lastSend := time.Now() // TODO: clarify use of lastSend / sendRetries
+	p.lastHeardFrom = time.Now()
-	lastUpdate := time.Now()
+	p.lastSend = time.Now() // TODO: clarify use of lastSend / sendRetries
-	sendRetries := 0
+	p.lastUpdate = time.Now()
+	p.sendRetries = 0
+	p.mu.Unlock()
 
 	var lastRISent aurp.Packet
 
 	p.disconnect()
 
 	// Write an Open-Req packet
-	if _, err := p.Send(p.Transport.NewOpenReqPacket(nil)); err != nil {
+	if _, err := p.send(p.Transport.NewOpenReqPacket(nil)); err != nil {
 		log.Printf("AURP Peer: Couldn't send Open-Req packet: %v", err)
 		return err
 	}
@@ -202,7 +339,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 			}
 			// Send a best-effort Router Down before returning
 			lastRISent = p.Transport.NewRDPacket(aurp.ErrCodeNormalClose)
-			if _, err := p.Send(lastRISent); err != nil {
+			if _, err := p.send(lastRISent); err != nil {
 				log.Printf("Couldn't send RD packet: %v", err)
 			}
 			return ctx.Err()
@@ -210,60 +347,60 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 		case <-rticker.C:
 			switch p.rstate {
 			case ReceiverWaitForOpenRsp:
-				if time.Since(lastSend) <= sendRetryTimer {
+				if time.Since(p.lastSend) <= sendRetryTimer {
 					break
 				}
-				if sendRetries >= sendRetryLimit {
+				if p.sendRetries >= sendRetryLimit {
 					log.Printf("AURP Peer: Send retry limit reached while waiting for Open-Rsp, closing connection")
 					p.setRState(ReceiverUnconnected)
 					break
 				}
 
 				// Send another Open-Req
-				sendRetries++
+				p.incSendRetries()
-				lastSend = time.Now()
+				p.bumpLastSend()
-				if _, err := p.Send(p.Transport.NewOpenReqPacket(nil)); err != nil {
+				if _, err := p.send(p.Transport.NewOpenReqPacket(nil)); err != nil {
 					log.Printf("AURP Peer: Couldn't send Open-Req packet: %v", err)
 					return err
 				}
 
 			case ReceiverConnected:
 				// Check LHFT, send tickle?
-				if time.Since(lastHeardFrom) <= lastHeardFromTimer {
+				if time.Since(p.lastHeardFrom) <= lastHeardFromTimer {
 					break
 				}
-				if _, err := p.Send(p.Transport.NewTicklePacket()); err != nil {
+				if _, err := p.send(p.Transport.NewTicklePacket()); err != nil {
 					log.Printf("AURP Peer: Couldn't send Tickle: %v", err)
 					return err
 				}
 				p.setRState(ReceiverWaitForTickleAck)
-				sendRetries = 0
+				p.resetSendRetries()
-				lastSend = time.Now()
+				p.bumpLastSend()
 
 			case ReceiverWaitForTickleAck:
-				if time.Since(lastSend) <= sendRetryTimer {
+				if time.Since(p.lastSend) <= sendRetryTimer {
 					break
 				}
-				if sendRetries >= tickleRetryLimit {
+				if p.sendRetries >= tickleRetryLimit {
 					log.Printf("AURP Peer: Send retry limit reached while waiting for Tickle-Ack, closing connection")
 					p.setRState(ReceiverUnconnected)
 					p.RouteTable.DeleteAURPPeer(p)
 					break
 				}
 
-				sendRetries++
+				p.incSendRetries()
-				lastSend = time.Now()
+				p.bumpLastSend()
-				if _, err := p.Send(p.Transport.NewTicklePacket()); err != nil {
+				if _, err := p.send(p.Transport.NewTicklePacket()); err != nil {
 					log.Printf("AURP Peer: Couldn't send Tickle: %v", err)
 					return err
 				}
 				// still in Wait For Tickle-Ack
 
 			case ReceiverWaitForRIRsp:
-				if time.Since(lastSend) <= sendRetryTimer {
+				if time.Since(p.lastSend) <= sendRetryTimer {
 					break
 				}
-				if sendRetries >= sendRetryLimit {
+				if p.sendRetries >= sendRetryLimit {
 					log.Printf("AURP Peer: Send retry limit reached while waiting for RI-Rsp, closing connection")
 					p.setRState(ReceiverUnconnected)
 					p.RouteTable.DeleteAURPPeer(p)
@@ -272,8 +409,9 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 
 				// RI-Req is stateless, so we don't need to cache the one we
 				// sent earlier just to send it again
-				sendRetries++
+				p.incSendRetries()
-				if _, err := p.Send(p.Transport.NewRIReqPacket()); err != nil {
+				p.bumpLastSend()
+				if _, err := p.send(p.Transport.NewRIReqPacket()); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Req packet: %v", err)
 					return err
 				}
@@ -282,18 +420,18 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 			case ReceiverUnconnected:
 				// Data receiver is unconnected. If data sender is connected,
 				// send a null RI-Upd to check if the sender is also unconnected
-				if p.sstate == SenderConnected && time.Since(lastSend) > sendRetryTimer {
+				if p.sstate == SenderConnected && time.Since(p.lastSend) > sendRetryTimer {
-					if sendRetries >= sendRetryLimit {
+					if p.sendRetries >= sendRetryLimit {
 						log.Printf("AURP Peer: Send retry limit reached while probing sender connect, closing connection")
 					}
-					sendRetries++
+					p.incSendRetries()
-					lastSend = time.Now()
+					p.bumpLastSend()
 					aurp.Inc(&p.Transport.LocalSeq)
 					events := aurp.EventTuples{{
 						EventCode: aurp.EventCodeNull,
 					}}
 					lastRISent = p.Transport.NewRIUpdPacket(events)
-					if _, err := p.Send(lastRISent); err != nil {
+					if _, err := p.send(lastRISent); err != nil {
 						log.Printf("AURP Peer: Couldn't send RI-Upd packet: %v", err)
 						return err
 					}
@@ -302,7 +440,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 
 				if p.ConfiguredAddr != "" {
 					// Periodically try to reconnect, if this peer is in the config file
-					if time.Since(lastReconnect) <= reconnectTimer {
+					if time.Since(p.lastReconnect) <= reconnectTimer {
 						break
 					}
 
@@ -315,10 +453,10 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 					log.Printf("AURP Peer: resolved %q to %v", p.ConfiguredAddr, raddr)
 					p.RemoteAddr = raddr
 
-					lastReconnect = time.Now()
+					p.bumpLastReconnect()
-					sendRetries = 0
+					p.resetSendRetries()
-					lastSend = time.Now()
+					p.bumpLastSend()
-					if _, err := p.Send(p.Transport.NewOpenReqPacket(nil)); err != nil {
+					if _, err := p.send(p.Transport.NewOpenReqPacket(nil)); err != nil {
 						log.Printf("AURP Peer: Couldn't send Open-Req packet: %v", err)
 						return err
 					}
@@ -332,40 +470,64 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				// Do nothing
 
 			case SenderConnected:
-				if time.Since(lastUpdate) <= updateTimer {
+				if time.Since(p.lastUpdate) <= updateTimer {
 					break
 				}
-				// TODO: is there a routing update to send?
+
+				// Are there routing updates to send?
+				p.pendingEventsMu.Lock()
+				if len(p.pendingEvents) == 0 {
+					p.pendingEventsMu.Unlock()
+					break
+				}
+				// Yes - swap the slices, release the mutex, then send them
+				pending := p.pendingEvents
+				p.pendingEvents = make(aurp.EventTuples, 0, cap(pending))
+				p.pendingEventsMu.Unlock()
+
+				// TODO: eliminate events that cancel out (e.g. NA then ND)
+				// TODO: split pending events to fit within a packet
+
+				p.bumpLastUpdate()
+				aurp.Inc(&p.Transport.LocalSeq)
+				lastRISent = p.Transport.NewRIUpdPacket(pending)
+				if _, err := p.send(lastRISent); err != nil {
+					log.Printf("AURP Peer: Couldn't send RI-Upd packet: %v", err)
+					return err
+				}
+				p.setSState(SenderWaitForRIUpdAck)
 
 			case SenderWaitForRIRspAck, SenderWaitForRIUpdAck:
-				if time.Since(lastSend) <= sendRetryTimer {
+				if time.Since(p.lastSend) <= sendRetryTimer {
 					break
 				}
 				if lastRISent == nil {
 					log.Print("AURP Peer: sender retry: lastRISent = nil?")
 					continue
 				}
-				if sendRetries >= sendRetryLimit {
+				if p.sendRetries >= sendRetryLimit {
 					log.Printf("AURP Peer: Send retry limit reached, closing connection")
 					p.setSState(SenderUnconnected)
+					p.RouteTable.RemoveObserver(p)
 					continue
 				}
-				sendRetries++
+				p.incSendRetries()
-				lastSend = time.Now()
+				p.bumpLastSend()
-				if _, err := p.Send(lastRISent); err != nil {
+				if _, err := p.send(lastRISent); err != nil {
 					log.Printf("AURP Peer: Couldn't re-send %T: %v", lastRISent, err)
 					return err
 				}
 
 			case SenderWaitForRDAck:
-				if time.Since(lastSend) <= sendRetryTimer {
+				if time.Since(p.lastSend) <= sendRetryTimer {
 					break
 				}
 				p.setSState(SenderUnconnected)
+				p.RouteTable.RemoveObserver(p)
 			}
 
 		case pkt := <-p.ReceiveCh:
-			lastHeardFrom = time.Now()
+			p.bumpLastHeardFrom()
 
 			switch pkt := pkt.(type) {
 			case *aurp.OpenReqPacket:
@@ -392,19 +554,21 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 					orsp = p.Transport.NewOpenRspPacket(0, 1, nil)
 				}
 
-				if _, err := p.Send(orsp); err != nil {
+				if _, err := p.send(orsp); err != nil {
 					log.Printf("AURP Peer: Couldn't send Open-Rsp: %v", err)
 					return err
 				}
 				if orsp.RateOrErrCode >= 0 {
+					// Data sender is successfully in connected state
 					p.setSState(SenderConnected)
+					p.RouteTable.AddObserver(p)
 				}
 
 				// If receiver is unconnected, commence connecting
 				if p.rstate == ReceiverUnconnected {
-					lastSend = time.Now()
+					p.resetSendRetries()
-					sendRetries = 0
+					p.bumpLastSend()
-					if _, err := p.Send(p.Transport.NewOpenReqPacket(nil)); err != nil {
+					if _, err := p.send(p.Transport.NewOpenReqPacket(nil)); err != nil {
 						log.Printf("AURP Peer: Couldn't send Open-Req packet: %v", err)
 						return err
 					}
@@ -425,8 +589,8 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				p.setRState(ReceiverConnected)
 
 				// Send an RI-Req
-				sendRetries = 0
+				p.resetSendRetries()
-				if _, err := p.Send(p.Transport.NewRIReqPacket()); err != nil {
+				if _, err := p.send(p.Transport.NewRIReqPacket()); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Req packet: %v", err)
 					return err
 				}
@@ -449,7 +613,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				p.Transport.LocalSeq = 1
 				// TODO: Split tuples across multiple packets as required
 				lastRISent = p.Transport.NewRIRspPacket(aurp.RoutingFlagLast, nets)
-				if _, err := p.Send(lastRISent); err != nil {
+				if _, err := p.send(lastRISent); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Rsp packet: %v", err)
 					return err
 				}
@@ -474,7 +638,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 
 				// TODO: track which networks we don't have zone info for, and
 				// only set SZI for those ?
-				if _, err := p.Send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, aurp.RoutingFlagSendZoneInfo)); err != nil {
+				if _, err := p.send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, aurp.RoutingFlagSendZoneInfo)); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Ack packet: %v", err)
 					return err
 				}
@@ -500,7 +664,8 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				}
 
 				p.setSState(SenderConnected)
-				sendRetries = 0
+				p.resetSendRetries()
+				p.RouteTable.AddObserver(p)
 
 				// If SZI flag is set, send ZI-Rsp (transaction)
 				if pkt.Flags&aurp.RoutingFlagSendZoneInfo != 0 {
@@ -525,7 +690,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 					}
 					zones := p.RouteTable.ZonesForNetworks(nets)
 					// TODO: split ZI-Rsp packets similarly to ZIP Replies
-					if _, err := p.Send(p.Transport.NewZIRspPacket(zones)); err != nil {
+					if _, err := p.send(p.Transport.NewZIRspPacket(zones)); err != nil {
 						log.Printf("AURP Peer: Couldn't send ZI-Rsp packet: %v", err)
 					}
 				}
@@ -536,9 +701,9 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 					// Receiver is unconnected, but their receiver sent us an
 					// RI-Ack for something
 					// Try to reconnect?
-					lastSend = time.Now()
+					p.resetSendRetries()
-					sendRetries = 0
+					p.bumpLastSend()
-					if _, err := p.Send(p.Transport.NewOpenReqPacket(nil)); err != nil {
+					if _, err := p.send(p.Transport.NewOpenReqPacket(nil)); err != nil {
 						log.Printf("AURP Peer: Couldn't send Open-Req packet: %v", err)
 						return err
 					}
@@ -546,7 +711,6 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				}
 
 			case *aurp.RIUpdPacket:
-
 				var ackFlag aurp.RoutingFlag
 
 				for _, et := range pkt.Events {
@@ -586,7 +750,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 					}
 				}
 
-				if _, err := p.Send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, ackFlag)); err != nil {
+				if _, err := p.send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, ackFlag)); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Ack: %v", err)
 					return err
 				}
@@ -600,7 +764,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				p.RouteTable.DeleteAURPPeer(p)
 
 				// Respond with RI-Ack
-				if _, err := p.Send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, 0)); err != nil {
+				if _, err := p.send(p.Transport.NewRIAckPacket(pkt.ConnectionID, pkt.Sequence, 0)); err != nil {
 					log.Printf("AURP Peer: Couldn't send RI-Ack: %v", err)
 					return err
 				}
@@ -610,7 +774,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 			case *aurp.ZIReqPacket:
 				// TODO: split ZI-Rsp packets similarly to ZIP Replies
 				zones := p.RouteTable.ZonesForNetworks(pkt.Networks)
-				if _, err := p.Send(p.Transport.NewZIRspPacket(zones)); err != nil {
+				if _, err := p.send(p.Transport.NewZIRspPacket(zones)); err != nil {
 					log.Printf("AURP Peer: Couldn't send ZI-Rsp packet: %v", err)
 					return err
 				}
@@ -622,7 +786,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				}
 
 			case *aurp.GDZLReqPacket:
-				if _, err := p.Send(p.Transport.NewGDZLRspPacket(-1, nil)); err != nil {
+				if _, err := p.send(p.Transport.NewGDZLRspPacket(-1, nil)); err != nil {
 					log.Printf("AURP Peer: Couldn't send GDZL-Rsp packet: %v", err)
 					return err
 				}
@@ -631,7 +795,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 				log.Printf("AURP Peer: Received a GDZL-Rsp, but I wouldn't have sent a GDZL-Req - that's weird")
 
 			case *aurp.GZNReqPacket:
-				if _, err := p.Send(p.Transport.NewGZNRspPacket(pkt.ZoneName, false, nil)); err != nil {
+				if _, err := p.send(p.Transport.NewGZNRspPacket(pkt.ZoneName, false, nil)); err != nil {
 					log.Printf("AURP Peer: Couldn't send GZN-Rsp packet: %v", err)
 					return err
 				}
@@ -641,7 +805,7 @@ func (p *AURPPeer) Handle(ctx context.Context) error {
 
 			case *aurp.TicklePacket:
 				// Immediately respond with Tickle-Ack
-				if _, err := p.Send(p.Transport.NewTickleAckPacket()); err != nil {
+				if _, err := p.send(p.Transport.NewTickleAckPacket()); err != nil {
 					log.Printf("AURP Peer: Couldn't send Tickle-Ack: %v", err)
 					return err
 				}

router/port.go

@@ -18,6 +18,7 @@ package router
 
 import (
 	"context"
+	"encoding/binary"
 	"errors"
 	"io"
 	"log"
@@ -79,8 +80,18 @@ func (port *EtherTalkPort) Serve(ctx context.Context) {
 
 		case ethertalk.AppleTalkProto:
 			// log.Print("Got an AppleTalk frame")
+
+			// Workaround for strict length checking in sfiera/multitalk
+			payload := ethFrame.Payload
+			if len(payload) < 2 {
+				log.Printf("Couldn't unmarshal DDP packet: too small (length = %d)", len(payload))
+			}
+			if size := binary.BigEndian.Uint16(payload[:2]) & 0x3ff; len(payload) > int(size) {
+				payload = payload[:size]
+			}
+
 			ddpkt := new(ddp.ExtPacket)
-			if err := ddp.ExtUnmarshal(ethFrame.Payload, ddpkt); err != nil {
+			if err := ddp.ExtUnmarshal(payload, ddpkt); err != nil {
 				log.Printf("Couldn't unmarshal DDP packet: %v", err)
 				continue
 			}

router/route.go

@@ -45,10 +45,7 @@ type Route struct {
 }
 
 func (r Route) LastSeenAgo() string {
-	if r.LastSeen.IsZero() {
+	return ago(r.LastSeen)
-		return "never"
-	}
-	return fmt.Sprintf("%v ago", time.Since(r.LastSeen).Truncate(time.Millisecond))
 }
 
 // Valid reports whether the route is valid.
@@ -58,17 +55,40 @@ func (r *Route) Valid() bool {
 	return len(r.ZoneNames) > 0 && (r.EtherTalkPeer == nil || time.Since(r.LastSeen) <= maxRouteAge)
 }
 
+type RouteTableObserver interface {
+	RouteAdded(*Route)
+	RouteDeleted(*Route)
+	RouteDistanceChanged(*Route)
+	RouteForwarderChanged(*Route)
+}
+
 type RouteTable struct {
-	mu     sync.Mutex
+	routesMu sync.RWMutex
-	routes map[*Route]struct{}
+	routes   map[*Route]struct{}
+
+	observersMu sync.RWMutex
+	observers   map[RouteTableObserver]struct{}
 }
 
 func NewRouteTable() *RouteTable {
 	return &RouteTable{
-		routes: make(map[*Route]struct{}),
+		routes:    make(map[*Route]struct{}),
+		observers: make(map[RouteTableObserver]struct{}),
 	}
 }
 
+func (rt *RouteTable) AddObserver(obs RouteTableObserver) {
+	rt.observersMu.Lock()
+	defer rt.observersMu.Unlock()
+	rt.observers[obs] = struct{}{}
+}
+
+func (rt *RouteTable) RemoveObserver(obs RouteTableObserver) {
+	rt.observersMu.Lock()
+	defer rt.observersMu.Unlock()
+	delete(rt.observers, obs)
+}
+
 func (rt *RouteTable) InsertEtherTalkDirect(port *EtherTalkPort) {
 	r := &Route{
 		Extended:        true,
@@ -80,14 +100,14 @@ func (rt *RouteTable) InsertEtherTalkDirect(port *EtherTalkPort) {
 		EtherTalkDirect: port,
 	}
 
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 	rt.routes[r] = struct{}{}
 }
 
 func (rt *RouteTable) Dump() []Route {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	table := make([]Route, 0, len(rt.routes))
 	for r := range rt.routes {
@@ -97,8 +117,8 @@ func (rt *RouteTable) Dump() []Route {
 }
 
 func (rt *RouteTable) LookupRoute(network ddp.Network) *Route {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	var bestRoute *Route
 	for r := range rt.routes {
@@ -120,8 +140,8 @@ func (rt *RouteTable) LookupRoute(network ddp.Network) *Route {
 }
 
 func (rt *RouteTable) DeleteAURPPeer(peer *AURPPeer) {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	for route := range rt.routes {
 		if route.AURPPeer == peer {
@@ -131,8 +151,8 @@ func (rt *RouteTable) DeleteAURPPeer(peer *AURPPeer) {
 }
 
 func (rt *RouteTable) DeleteAURPPeerNetwork(peer *AURPPeer, network ddp.Network) {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	for route := range rt.routes {
 		if route.AURPPeer == peer && route.NetStart == network {
@@ -142,8 +162,8 @@ func (rt *RouteTable) DeleteAURPPeerNetwork(peer *AURPPeer, network ddp.Network)
 }
 
 func (rt *RouteTable) UpdateAURPRouteDistance(peer *AURPPeer, network ddp.Network, distance uint8) {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	for route := range rt.routes {
 		if route.AURPPeer == peer && route.NetStart == network {
@@ -161,8 +181,8 @@ func (rt *RouteTable) UpsertEtherTalkRoute(peer *EtherTalkPeer, extended bool, n
 		return nil, fmt.Errorf("invalid network range [%d, %d] for nonextended network", netStart, netEnd)
 	}
 
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	// Update?
 	for r := range rt.routes {
@@ -213,16 +233,16 @@ func (rt *RouteTable) InsertAURPRoute(peer *AURPPeer, extended bool, netStart, n
 		AURPPeer: peer,
 	}
 
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 	rt.routes[r] = struct{}{}
 	return nil
 }
 
 // ValidRoutes returns all valid routes.
 func (rt *RouteTable) ValidRoutes() []*Route {
-	rt.mu.Lock()
+	rt.routesMu.RLock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.RUnlock()
 	valid := make([]*Route, 0, len(rt.routes))
 	for r := range rt.routes {
 		if r.Valid() {
@@ -234,8 +254,8 @@ func (rt *RouteTable) ValidRoutes() []*Route {
 
 // ValidNonAURPRoutes returns all valid routes that were not learned via AURP.
 func (rt *RouteTable) ValidNonAURPRoutes() []*Route {
-	rt.mu.Lock()
+	rt.routesMu.RLock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.RUnlock()
 	valid := make([]*Route, 0, len(rt.routes))
 	for r := range rt.routes {
 		if r.AURPPeer != nil {

router/zones.go

@@ -23,8 +23,8 @@ import (
 )
 
 func (rt *RouteTable) AddZonesToNetwork(n ddp.Network, zs ...string) {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 	for r := range rt.routes {
 		if n < r.NetStart || n > r.NetEnd {
 			continue
@@ -39,8 +39,8 @@ func (rt *RouteTable) AddZonesToNetwork(n ddp.Network, zs ...string) {
 func (rt *RouteTable) ZonesForNetworks(ns []ddp.Network) map[ddp.Network][]string {
 	zs := make(map[ddp.Network][]string)
 
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 	for r := range rt.routes {
 		if !r.Valid() {
 			continue
@@ -55,8 +55,8 @@ func (rt *RouteTable) ZonesForNetworks(ns []ddp.Network) map[ddp.Network][]strin
 }
 
 func (rt *RouteTable) RoutesForZone(zone string) []*Route {
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	var routes []*Route
 	for r := range rt.routes {
@@ -73,8 +73,8 @@ func (rt *RouteTable) RoutesForZone(zone string) []*Route {
 func (rt *RouteTable) AllZoneNames() (zones []string) {
 	defer slices.Sort(zones)
 
-	rt.mu.Lock()
+	rt.routesMu.Lock()
-	defer rt.mu.Unlock()
+	defer rt.routesMu.Unlock()
 
 	zs := make(StringSet)
 	for r := range rt.routes {