Merge remote-tracking branch 'origin/master' into multiport

* Use a list for relay IPs instead of a map

* linter

.github/ISSUE_TEMPLATE/config.yml

@@ -1,13 +1,21 @@
 blank_issues_enabled: true
 contact_links:
+  - name: 💨 Performance Issues
+    url: https://github.com/slackhq/nebula/discussions/new/choose
+    about: 'We ask that you create a discussion instead of an issue for performance-related questions. This allows us to have a more open conversation about the issue and helps us to better understand the problem.'
+
+  - name: 📄 Documentation Issues
+    url: https://github.com/definednet/nebula-docs
+    about: "If you've found an issue with the website documentation, please file it in the nebula-docs repository."
+
+  - name: 📱 Mobile Nebula Issues
+    url: https://github.com/definednet/mobile_nebula
+    about: "If you're using the mobile Nebula app and have found an issue, please file it in the mobile_nebula repository."
+
   - name: 📘 Documentation
     url: https://nebula.defined.net/docs/
-    about: Review documentation.
+    about: 'The documentation is the best place to start if you are new to Nebula.'
 
   - name: 💁 Support/Chat
-    url: https://join.slack.com/t/nebulaoss/shared_invite/enQtOTA5MDI4NDg3MTg4LTkwY2EwNTI4NzQyMzc0M2ZlODBjNWI3NTY1MzhiOThiMmZlZjVkMTI0NGY4YTMyNjUwMWEyNzNkZTJmYzQxOGU
-    about: 'This issue tracker is not for support questions. Join us on Slack for assistance!'
-
-  - name: 📱 Mobile Nebula
-    url: https://github.com/definednet/mobile_nebula
-    about: 'This issue tracker is not for mobile support. Try the Mobile Nebula repo instead!'
+    url: https://join.slack.com/t/nebulaoss/shared_invite/zt-39pk4xopc-CUKlGcb5Z39dQ0cK1v7ehA
+    about: 'For faster support, join us on Slack for assistance!'

.github/pull_request_template.md

@@ -0,0 +1,11 @@
+<!--
+Thank you for taking the time to submit a pull request!
+
+Please be sure to provide a clear description of what you're trying to achieve with the change.
+
+- If you're submitting a new feature, please explain how to use it and document any new config options in the example config.
+- If you're submitting a bugfix, please link the related issue or describe the circumstances surrounding the issue.
+- If you're changing a default, explain why you believe the new default is appropriate for most users.
+
+P.S. If you're only updating the README or other docs, please file a pull request here instead: https://github.com/DefinedNet/nebula-docs
+-->

.github/workflows/smoke.yml

@@ -26,7 +26,7 @@ jobs:
         check-latest: true
 
     - name: build
-      run: make bin-docker CGO_ENABLED=1 BUILD_ARGS="-race -tags=mutex_debug"
+      run: make bin-docker CGO_ENABLED=1 BUILD_ARGS=-race
 
     - name: setup docker image
       working-directory: ./.github/workflows/smoke
@@ -52,4 +52,12 @@ jobs:
       working-directory: ./.github/workflows/smoke
       run: NAME="smoke-p256" ./smoke.sh
 
+    - name: setup docker image for multiport
+      working-directory: ./.github/workflows/smoke
+      run: NAME="smoke-multiport" MULTIPORT_TX=true MULTIPORT_RX=true MULTIPORT_HANDSHAKE=true ./build.sh
+
+    - name: run smoke
+      working-directory: ./.github/workflows/smoke
+      run: NAME="smoke-multiport" ./smoke.sh
+
     timeout-minutes: 10

.github/workflows/smoke/genconfig.sh

@@ -48,6 +48,10 @@ listen:
 
 tun:
   dev: ${TUN_DEV:-tun0}
+  multiport:
+    tx_enabled: ${MULTIPORT_TX:-false}
+    rx_enabled: ${MULTIPORT_RX:-false}
+    tx_handshake: ${MULTIPORT_HANDSHAKE:-false}
 
 firewall:
   inbound_action: reject

.github/workflows/test.yml

@@ -32,15 +32,15 @@ jobs:
       run: make vet
 
     - name: golangci-lint
-      uses: golangci/golangci-lint-action@v7
+      uses: golangci/golangci-lint-action@v8
       with:
-        version: v2.0
+        version: v2.1
 
     - name: Test
       run: make test
 
     - name: End 2 end
-      run: make e2e-mutex-debug TEST_LOGS=1 TEST_FLAGS=-v
+      run: make e2evv
 
     - name: Build test mobile
       run: make build-test-mobile
@@ -115,9 +115,9 @@ jobs:
       run: make vet
 
     - name: golangci-lint
-      uses: golangci/golangci-lint-action@v7
+      uses: golangci/golangci-lint-action@v8
       with:
-        version: v2.0
+        version: v2.1
 
     - name: Test
       run: make test

Makefile

@@ -63,9 +63,6 @@ ALL = $(ALL_LINUX) \
 e2e:
 	$(TEST_ENV) go test -tags=e2e_testing -count=1 $(TEST_FLAGS) ./e2e
 
-e2e-mutex-debug:
-	$(TEST_ENV) go test -tags=mutex_debug,e2e_testing -count=1 $(TEST_FLAGS) ./e2e
-
 e2ev: TEST_FLAGS += -v
 e2ev: e2e
 
@@ -218,7 +215,6 @@ ifeq ($(words $(MAKECMDGOALS)),1)
 	@$(MAKE) service ${.DEFAULT_GOAL} --no-print-directory
 endif
 
-bin-docker: BUILD_ARGS = -tags=mutex_debug
 bin-docker: bin build/linux-amd64/nebula build/linux-amd64/nebula-cert
 
 smoke-docker: bin-docker
@@ -231,6 +227,10 @@ smoke-relay-docker: bin-docker
 	cd .github/workflows/smoke/ && ./build-relay.sh
 	cd .github/workflows/smoke/ && ./smoke-relay.sh
 
+smoke-multiport-docker: bin-docker
+	cd .github/workflows/smoke/ && NAME="smoke-multiport" MULTIPORT_TX=true MULTIPORT_RX=true MULTIPORT_HANDSHAKE=true ./build.sh
+	cd .github/workflows/smoke/ && NAME="smoke-multiport" ./smoke.sh
+
 smoke-docker-race: BUILD_ARGS = -race
 smoke-docker-race: CGO_ENABLED = 1
 smoke-docker-race: smoke-docker

README.md

@@ -4,7 +4,7 @@ It lets you seamlessly connect computers anywhere in the world. Nebula is portab
 It can be used to connect a small number of computers, but is also able to connect tens of thousands of computers.
 
 Nebula incorporates a number of existing concepts like encryption, security groups, certificates,
-and tunneling, and each of those individual pieces existed before Nebula in various forms.
+and tunneling.
 What makes Nebula different to existing offerings is that it brings all of these ideas together,
 resulting in a sum that is greater than its individual parts.
 
@@ -12,7 +12,7 @@ Further documentation can be found [here](https://nebula.defined.net/docs/).
 
 You can read more about Nebula [here](https://medium.com/p/884110a5579).
 
-You can also join the NebulaOSS Slack group [here](https://join.slack.com/t/nebulaoss/shared_invite/zt-2xqe6e7vn-k_KGi8s13nsr7cvHVvHvuQ).
+You can also join the NebulaOSS Slack group [here](https://join.slack.com/t/nebulaoss/shared_invite/zt-39pk4xopc-CUKlGcb5Z39dQ0cK1v7ehA).
 
 ## Supported Platforms
 
@@ -28,33 +28,33 @@ Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for
 #### Distribution Packages
 
 - [Arch Linux](https://archlinux.org/packages/extra/x86_64/nebula/)
-    ```
-    $ sudo pacman -S nebula
+    ```sh
+    sudo pacman -S nebula
     ```
 
 - [Fedora Linux](https://src.fedoraproject.org/rpms/nebula)
-    ```
-    $ sudo dnf install nebula
+    ```sh
+    sudo dnf install nebula
     ```
 
 - [Debian Linux](https://packages.debian.org/source/stable/nebula)
-    ```
-    $ sudo apt install nebula
+    ```sh
+    sudo apt install nebula
     ```
 
 - [Alpine Linux](https://pkgs.alpinelinux.org/packages?name=nebula)
-    ```
-    $ sudo apk add nebula
+    ```sh
+    sudo apk add nebula
     ```
 
 - [macOS Homebrew](https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/n/nebula.rb)
-    ```
-    $ brew install nebula
+    ```sh
+    brew install nebula
     ```
 
 - [Docker](https://hub.docker.com/r/nebulaoss/nebula)
-    ```
-    $ docker pull nebulaoss/nebula
+    ```sh
+    docker pull nebulaoss/nebula
     ```
 
 #### Mobile
@@ -64,10 +64,10 @@ Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for
 
 ## Technical Overview
 
-Nebula is a mutually authenticated peer-to-peer software defined network based on the [Noise Protocol Framework](https://noiseprotocol.org/).
+Nebula is a mutually authenticated peer-to-peer software-defined network based on the [Noise Protocol Framework](https://noiseprotocol.org/).
 Nebula uses certificates to assert a node's IP address, name, and membership within user-defined groups.
 Nebula's user-defined groups allow for provider agnostic traffic filtering between nodes.
-Discovery nodes allow individual peers to find each other and optionally use UDP hole punching to establish connections from behind most firewalls or NATs.
+Discovery nodes (aka lighthouses) allow individual peers to find each other and optionally use UDP hole punching to establish connections from behind most firewalls or NATs.
 Users can move data between nodes in any number of cloud service providers, datacenters, and endpoints, without needing to maintain a particular addressing scheme.
 
 Nebula uses Elliptic-curve Diffie-Hellman (`ECDH`) key exchange and `AES-256-GCM` in its default configuration.
@@ -82,28 +82,34 @@ To set up a Nebula network, you'll need:
 
 #### 2. (Optional, but you really should..) At least one discovery node with a routable IP address, which we call a lighthouse.
 
-Nebula lighthouses allow nodes to find each other, anywhere in the world. A lighthouse is the only node in a Nebula network whose IP should not change. Running a lighthouse requires very few compute resources, and you can easily use the least expensive option from a cloud hosting provider. If you're not sure which provider to use, a number of us have used $5/mo [DigitalOcean](https://digitalocean.com) droplets as lighthouses.
-
-  Once you have launched an instance, ensure that Nebula udp traffic (default port udp/4242) can reach it over the internet.
+Nebula lighthouses allow nodes to find each other, anywhere in the world. A lighthouse is the only node in a Nebula network whose IP should not change. Running a lighthouse requires very few compute resources, and you can easily use the least expensive option from a cloud hosting provider. If you're not sure which provider to use, a number of us have used $6/mo [DigitalOcean](https://digitalocean.com) droplets as lighthouses.
 
+Once you have launched an instance, ensure that Nebula udp traffic (default port udp/4242) can reach it over the internet.
 
 #### 3. A Nebula certificate authority, which will be the root of trust for a particular Nebula network.
 
-  ```
-  ./nebula-cert ca -name "Myorganization, Inc"
-  ```
-  This will create files named `ca.key` and `ca.cert` in the current directory. The `ca.key` file is the most sensitive file you'll create, because it is the key used to sign the certificates for individual nebula nodes/hosts. Please store this file somewhere safe, preferably with strong encryption.
+```sh
+./nebula-cert ca -name "Myorganization, Inc"
+```
+
+This will create files named `ca.key` and `ca.cert` in the current directory. The `ca.key` file is the most sensitive file you'll create, because it is the key used to sign the certificates for individual nebula nodes/hosts. Please store this file somewhere safe, preferably with strong encryption.
+
+**Be aware!** By default, certificate authorities have a 1-year lifetime before expiration. See [this guide](https://nebula.defined.net/docs/guides/rotating-certificate-authority/) for details on rotating a CA.
 
 #### 4. Nebula host keys and certificates generated from that certificate authority
+
 This assumes you have four nodes, named lighthouse1, laptop, server1, host3. You can name the nodes any way you'd like, including FQDN. You'll also need to choose IP addresses and the associated subnet. In this example, we are creating a nebula network that will use 192.168.100.x/24 as its network range. This example also demonstrates nebula groups, which can later be used to define traffic rules in a nebula network.
-```
+```sh
 ./nebula-cert sign -name "lighthouse1" -ip "192.168.100.1/24"
 ./nebula-cert sign -name "laptop" -ip "192.168.100.2/24" -groups "laptop,home,ssh"
 ./nebula-cert sign -name "server1" -ip "192.168.100.9/24" -groups "servers"
 ./nebula-cert sign -name "host3" -ip "192.168.100.10/24"
 ```
 
+By default, host certificates will expire 1 second before the CA expires. Use the `-duration` flag to specify a shorter lifetime.
+
 #### 5. Configuration files for each host
+
 Download a copy of the nebula [example configuration](https://github.com/slackhq/nebula/blob/master/examples/config.yml).
 
 * On the lighthouse node, you'll need to ensure `am_lighthouse: true` is set.
@@ -118,10 +124,13 @@ For each host, copy the nebula binary to the host, along with `config.yml` from
 **DO NOT COPY `ca.key` TO INDIVIDUAL NODES.**
 
 #### 7. Run nebula on each host
-```
+
+```sh
 ./nebula -config /path/to/config.yml
 ```
 
+For more detailed instructions, [find the full documentation here](https://nebula.defined.net/docs/).
+
 ## Building Nebula from source
 
 Make sure you have [go](https://go.dev/doc/install) installed and clone this repo. Change to the nebula directory.
@@ -140,8 +149,10 @@ The default curve used for cryptographic handshakes and signatures is Curve25519
 
 In addition, Nebula can be built using the [BoringCrypto GOEXPERIMENT](https://github.com/golang/go/blob/go1.20/src/crypto/internal/boring/README.md) by running either of the following make targets:
 
-    make bin-boringcrypto
-    make release-boringcrypto
+```sh
+make bin-boringcrypto
+make release-boringcrypto
+```
 
 This is not the recommended default deployment, but may be useful based on your compliance requirements.
 
@@ -149,5 +160,3 @@ This is not the recommended default deployment, but may be useful based on your
 
 Nebula was created at Slack Technologies, Inc by Nate Brown and Ryan Huber, with contributions from Oliver Fross, Alan Lam, Wade Simmons, and Lining Wang.
 
-
-

cert/crypto_test.go

@@ -26,21 +26,21 @@ func TestNewArgon2Parameters(t *testing.T) {
 }
 
 func TestDecryptAndUnmarshalSigningPrivateKey(t *testing.T) {
-	passphrase := []byte("DO NOT USE THIS KEY")
+	passphrase := []byte("DO NOT USE")
 	privKey := []byte(`# A good key
 -----BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
-CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCognnjujd67Vsv99p22wfAjQaDT
-oCMW1mdjkU3gACKNW4MSXOWR9Sts4C81yk1RUku2gvGKs3TB9LYoklLsIizSYOLl
-+Vs//O1T0I1Xbml2XBAROsb/VSoDln/6LMqR4B6fn6B3GOsLBBqRI8daDl9lRMPB
-qrlJ69wer3ZUHFXA
+CjsKC0FFUy0yNTYtR0NNEiwIExCAgAQYAyAEKiCPoDfGQiosxNPTbPn5EsMlc2MI
+c0Bt4oz6gTrFQhX3aBJcimhHKeAuhyTGvllD0Z19fe+DFPcLH3h5VrdjVfIAajg0
+KrbV3n9UHif/Au5skWmquNJzoW1E4MTdRbvpti6o+WdQ49DxjBFhx0YH8LBqrbPU
+0BGkUHmIO7daP24=
 -----END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
 `)
 	shortKey := []byte(`# A key which, once decrypted, is too short
 -----BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
-CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCoga5h8owMEBWRSMMJKzuUvWce7
-k0qlBkQmCxiuLh80MuASW70YcKt8jeEIS2axo2V6zAKA9TSMcCsJW1kDDXEtL/xe
-GLF5T7sDl5COp4LU3pGxpV+KoeQ/S3gQCAAcnaOtnJQX+aSDnbO3jCHyP7U9CHbs
-rQr3bdH3Oy/WiYU=
+CjsKC0FFUy0yNTYtR0NNEiwIExCAgAQYAyAEKiAVJwdfl3r+eqi/vF6S7OMdpjfo
+hAzmTCRnr58Su4AqmBJbCv3zleYCEKYJP6UI3S8ekLMGISsgO4hm5leukCCyqT0Z
+cQ76yrberpzkJKoPLGisX8f+xdy4aXSZl7oEYWQte1+vqbtl/eY9PGZhxUQdcyq7
+hqzIyrRqfUgVuA==
 -----END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
 `)
 	invalidBanner := []byte(`# Invalid banner (not encrypted)

config/config.go

@@ -243,7 +243,7 @@ func (c *C) GetInt(k string, d int) int {
 // GetUint32 will get the uint32 for k or return the default d if not found or invalid
 func (c *C) GetUint32(k string, d uint32) uint32 {
 	r := c.GetInt(k, int(d))
-	if uint64(r) > uint64(math.MaxUint32) {
+	if r < 0 || uint64(r) > uint64(math.MaxUint32) {
 		return d
 	}
 	return uint32(r)

connection_manager.go

@@ -4,12 +4,16 @@ import (
 	"bytes"
 	"context"
 	"encoding/binary"
+	"fmt"
 	"net/netip"
+	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
 )
 
@@ -26,130 +30,124 @@ const (
 )
 
 type connectionManager struct {
-	in     map[uint32]struct{}
-	inLock syncRWMutex
-
-	out     map[uint32]struct{}
-	outLock syncRWMutex
-
 	// relayUsed holds which relay localIndexs are in use
 	relayUsed     map[uint32]struct{}
-	relayUsedLock syncRWMutex
+	relayUsedLock *sync.RWMutex
 
-	hostMap                 *HostMap
-	trafficTimer            *LockingTimerWheel[uint32]
-	intf                    *Interface
-	pendingDeletion         map[uint32]struct{}
-	punchy                  *Punchy
+	hostMap      *HostMap
+	trafficTimer *LockingTimerWheel[uint32]
+	intf         *Interface
+	punchy       *Punchy
+
+	// Configuration settings
 	checkInterval           time.Duration
 	pendingDeletionInterval time.Duration
-	metricsTxPunchy         metrics.Counter
+	inactivityTimeout       atomic.Int64
+	dropInactive            atomic.Bool
+
+	metricsTxPunchy metrics.Counter
 
 	l *logrus.Logger
 }
 
-func newConnectionManager(ctx context.Context, l *logrus.Logger, intf *Interface, checkInterval, pendingDeletionInterval time.Duration, punchy *Punchy) *connectionManager {
-	var max time.Duration
-	if checkInterval < pendingDeletionInterval {
-		max = pendingDeletionInterval
-	} else {
-		max = checkInterval
+func newConnectionManagerFromConfig(l *logrus.Logger, c *config.C, hm *HostMap, p *Punchy) *connectionManager {
+	cm := &connectionManager{
+		hostMap:         hm,
+		l:               l,
+		punchy:          p,
+		relayUsed:       make(map[uint32]struct{}),
+		relayUsedLock:   &sync.RWMutex{},
+		metricsTxPunchy: metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
 	}
 
-	nc := &connectionManager{
-		hostMap:                 intf.hostMap,
-		in:                      make(map[uint32]struct{}),
-		inLock:                  newSyncRWMutex("connection-manager-in"),
-		out:                     make(map[uint32]struct{}),
-		outLock:                 newSyncRWMutex("connection-manager-out"),
-		relayUsed:               make(map[uint32]struct{}),
-		relayUsedLock:           newSyncRWMutex("connection-manager-relay-used"),
-		trafficTimer:            NewLockingTimerWheel[uint32]("connection-manager-timer", time.Millisecond*500, max),
-		intf:                    intf,
-		pendingDeletion:         make(map[uint32]struct{}),
-		checkInterval:           checkInterval,
-		pendingDeletionInterval: pendingDeletionInterval,
-		punchy:                  punchy,
-		metricsTxPunchy:         metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
-		l:                       l,
-	}
+	cm.reload(c, true)
+	c.RegisterReloadCallback(func(c *config.C) {
+		cm.reload(c, false)
+	})
 
-	nc.Start(ctx)
-	return nc
+	return cm
 }
 
-func (n *connectionManager) In(localIndex uint32) {
-	n.inLock.RLock()
-	// If this already exists, return
-	if _, ok := n.in[localIndex]; ok {
-		n.inLock.RUnlock()
-		return
+func (cm *connectionManager) reload(c *config.C, initial bool) {
+	if initial {
+		cm.checkInterval = time.Duration(c.GetInt("timers.connection_alive_interval", 5)) * time.Second
+		cm.pendingDeletionInterval = time.Duration(c.GetInt("timers.pending_deletion_interval", 10)) * time.Second
+
+		// We want at least a minimum resolution of 500ms per tick so that we can hit these intervals
+		// pretty close to their configured duration.
+		// The inactivity duration is checked each time a hostinfo ticks through so we don't need the wheel to contain it.
+		minDuration := min(time.Millisecond*500, cm.checkInterval, cm.pendingDeletionInterval)
+		maxDuration := max(cm.checkInterval, cm.pendingDeletionInterval)
+		cm.trafficTimer = NewLockingTimerWheel[uint32](minDuration, maxDuration)
+	}
+
+	if initial || c.HasChanged("tunnels.inactivity_timeout") {
+		old := cm.getInactivityTimeout()
+		cm.inactivityTimeout.Store((int64)(c.GetDuration("tunnels.inactivity_timeout", 10*time.Minute)))
+		if !initial {
+			cm.l.WithField("oldDuration", old).
+				WithField("newDuration", cm.getInactivityTimeout()).
+				Info("Inactivity timeout has changed")
+		}
+	}
+
+	if initial || c.HasChanged("tunnels.drop_inactive") {
+		old := cm.dropInactive.Load()
+		cm.dropInactive.Store(c.GetBool("tunnels.drop_inactive", false))
+		if !initial {
+			cm.l.WithField("oldBool", old).
+				WithField("newBool", cm.dropInactive.Load()).
+				Info("Drop inactive setting has changed")
+		}
 	}
-	n.inLock.RUnlock()
-	n.inLock.Lock()
-	n.in[localIndex] = struct{}{}
-	n.inLock.Unlock()
 }
 
-func (n *connectionManager) Out(localIndex uint32) {
-	n.outLock.RLock()
-	// If this already exists, return
-	if _, ok := n.out[localIndex]; ok {
-		n.outLock.RUnlock()
-		return
-	}
-	n.outLock.RUnlock()
-	n.outLock.Lock()
-	n.out[localIndex] = struct{}{}
-	n.outLock.Unlock()
+func (cm *connectionManager) getInactivityTimeout() time.Duration {
+	return (time.Duration)(cm.inactivityTimeout.Load())
 }
 
-func (n *connectionManager) RelayUsed(localIndex uint32) {
-	n.relayUsedLock.RLock()
+func (cm *connectionManager) In(h *HostInfo) {
+	h.in.Store(true)
+}
+
+func (cm *connectionManager) Out(h *HostInfo) {
+	h.out.Store(true)
+}
+
+func (cm *connectionManager) RelayUsed(localIndex uint32) {
+	cm.relayUsedLock.RLock()
 	// If this already exists, return
-	if _, ok := n.relayUsed[localIndex]; ok {
-		n.relayUsedLock.RUnlock()
+	if _, ok := cm.relayUsed[localIndex]; ok {
+		cm.relayUsedLock.RUnlock()
 		return
 	}
-	n.relayUsedLock.RUnlock()
-	n.relayUsedLock.Lock()
-	n.relayUsed[localIndex] = struct{}{}
-	n.relayUsedLock.Unlock()
+	cm.relayUsedLock.RUnlock()
+	cm.relayUsedLock.Lock()
+	cm.relayUsed[localIndex] = struct{}{}
+	cm.relayUsedLock.Unlock()
 }
 
 // getAndResetTrafficCheck returns if there was any inbound or outbound traffic within the last tick and
 // resets the state for this local index
-func (n *connectionManager) getAndResetTrafficCheck(localIndex uint32) (bool, bool) {
-	n.inLock.Lock()
-	n.outLock.Lock()
-	_, in := n.in[localIndex]
-	_, out := n.out[localIndex]
-	delete(n.in, localIndex)
-	delete(n.out, localIndex)
-	n.inLock.Unlock()
-	n.outLock.Unlock()
+func (cm *connectionManager) getAndResetTrafficCheck(h *HostInfo, now time.Time) (bool, bool) {
+	in := h.in.Swap(false)
+	out := h.out.Swap(false)
+	if in || out {
+		h.lastUsed = now
+	}
 	return in, out
 }
 
-func (n *connectionManager) AddTrafficWatch(localIndex uint32) {
-	// Use a write lock directly because it should be incredibly rare that we are ever already tracking this index
-	n.outLock.Lock()
-	if _, ok := n.out[localIndex]; ok {
-		n.outLock.Unlock()
-		return
+// AddTrafficWatch must be called for every new HostInfo.
+// We will continue to monitor the HostInfo until the tunnel is dropped.
+func (cm *connectionManager) AddTrafficWatch(h *HostInfo) {
+	if h.out.Swap(true) == false {
+		cm.trafficTimer.Add(h.localIndexId, cm.checkInterval)
 	}
-	n.out[localIndex] = struct{}{}
-	n.trafficTimer.Add(localIndex, n.checkInterval)
-	n.outLock.Unlock()
 }
 
-func (n *connectionManager) Start(ctx context.Context) {
-	go n.Run(ctx)
-}
-
-func (n *connectionManager) Run(ctx context.Context) {
-	//TODO: this tick should be based on the min wheel tick? Check firewall
-	clockSource := time.NewTicker(500 * time.Millisecond)
+func (cm *connectionManager) Start(ctx context.Context) {
+	clockSource := time.NewTicker(cm.trafficTimer.t.tickDuration)
 	defer clockSource.Stop()
 
 	p := []byte("")
@@ -162,61 +160,61 @@ func (n *connectionManager) Run(ctx context.Context) {
 			return
 
 		case now := <-clockSource.C:
-			n.trafficTimer.Advance(now)
+			cm.trafficTimer.Advance(now)
 			for {
-				localIndex, has := n.trafficTimer.Purge()
+				localIndex, has := cm.trafficTimer.Purge()
 				if !has {
 					break
 				}
 
-				n.doTrafficCheck(localIndex, p, nb, out, now)
+				cm.doTrafficCheck(localIndex, p, nb, out, now)
 			}
 		}
 	}
 }
 
-func (n *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
-	decision, hostinfo, primary := n.makeTrafficDecision(localIndex, now)
+func (cm *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
+	decision, hostinfo, primary := cm.makeTrafficDecision(localIndex, now)
 
 	switch decision {
 	case deleteTunnel:
-		if n.hostMap.DeleteHostInfo(hostinfo) {
+		if cm.hostMap.DeleteHostInfo(hostinfo) {
 			// Only clearing the lighthouse cache if this is the last hostinfo for this vpn ip in the hostmap
-			n.intf.lightHouse.DeleteVpnAddrs(hostinfo.vpnAddrs)
+			cm.intf.lightHouse.DeleteVpnAddrs(hostinfo.vpnAddrs)
 		}
 
 	case closeTunnel:
-		n.intf.sendCloseTunnel(hostinfo)
-		n.intf.closeTunnel(hostinfo)
+		cm.intf.sendCloseTunnel(hostinfo)
+		cm.intf.closeTunnel(hostinfo)
 
 	case swapPrimary:
-		n.swapPrimary(hostinfo, primary)
+		cm.swapPrimary(hostinfo, primary)
 
 	case migrateRelays:
-		n.migrateRelayUsed(hostinfo, primary)
+		cm.migrateRelayUsed(hostinfo, primary)
 
 	case tryRehandshake:
-		n.tryRehandshake(hostinfo)
+		cm.tryRehandshake(hostinfo)
 
 	case sendTestPacket:
-		n.intf.SendMessageToHostInfo(header.Test, header.TestRequest, hostinfo, p, nb, out)
+		cm.intf.SendMessageToHostInfo(header.Test, header.TestRequest, hostinfo, p, nb, out)
 	}
 
-	n.resetRelayTrafficCheck(hostinfo)
+	cm.resetRelayTrafficCheck(hostinfo)
 }
 
-func (n *connectionManager) resetRelayTrafficCheck(hostinfo *HostInfo) {
+func (cm *connectionManager) resetRelayTrafficCheck(hostinfo *HostInfo) {
 	if hostinfo != nil {
-		n.relayUsedLock.Lock()
-		defer n.relayUsedLock.Unlock()
+		cm.relayUsedLock.Lock()
+		defer cm.relayUsedLock.Unlock()
 		// No need to migrate any relays, delete usage info now.
 		for _, idx := range hostinfo.relayState.CopyRelayForIdxs() {
-			delete(n.relayUsed, idx)
+			delete(cm.relayUsed, idx)
 		}
 	}
 }
 
-func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo) {
+func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo) {
 	relayFor := oldhostinfo.relayState.CopyAllRelayFor()
 
 	for _, r := range relayFor {
@@ -226,46 +224,51 @@ func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo)
 		var relayFrom netip.Addr
 		var relayTo netip.Addr
 		switch {
-		case ok && existing.State == Established:
-			// This relay already exists in newhostinfo, then do nothing.
-			continue
-		case ok && existing.State == Requested:
-			// The relay exists in a Requested state; re-send the request
-			index = existing.LocalIndex
-			switch r.Type {
-			case TerminalType:
-				relayFrom = n.intf.myVpnAddrs[0]
-				relayTo = existing.PeerAddr
-			case ForwardingType:
-				relayFrom = existing.PeerAddr
-				relayTo = newhostinfo.vpnAddrs[0]
-			default:
-				// should never happen
+		case ok:
+			switch existing.State {
+			case Established, PeerRequested, Disestablished:
+				// This relay already exists in newhostinfo, then do nothing.
+				continue
+			case Requested:
+				// The relay exists in a Requested state; re-send the request
+				index = existing.LocalIndex
+				switch r.Type {
+				case TerminalType:
+					relayFrom = cm.intf.myVpnAddrs[0]
+					relayTo = existing.PeerAddr
+				case ForwardingType:
+					relayFrom = existing.PeerAddr
+					relayTo = newhostinfo.vpnAddrs[0]
+				default:
+					// should never happen
+					panic(fmt.Sprintf("Migrating unknown relay type: %v", r.Type))
+				}
 			}
 		case !ok:
-			n.relayUsedLock.RLock()
-			if _, relayUsed := n.relayUsed[r.LocalIndex]; !relayUsed {
+			cm.relayUsedLock.RLock()
+			if _, relayUsed := cm.relayUsed[r.LocalIndex]; !relayUsed {
 				// The relay hasn't been used; don't migrate it.
-				n.relayUsedLock.RUnlock()
+				cm.relayUsedLock.RUnlock()
 				continue
 			}
-			n.relayUsedLock.RUnlock()
+			cm.relayUsedLock.RUnlock()
 			// The relay doesn't exist at all; create some relay state and send the request.
 			var err error
-			index, err = AddRelay(n.l, newhostinfo, n.hostMap, r.PeerAddr, nil, r.Type, Requested)
+			index, err = AddRelay(cm.l, newhostinfo, cm.hostMap, r.PeerAddr, nil, r.Type, Requested)
 			if err != nil {
-				n.l.WithError(err).Error("failed to migrate relay to new hostinfo")
+				cm.l.WithError(err).Error("failed to migrate relay to new hostinfo")
 				continue
 			}
 			switch r.Type {
 			case TerminalType:
-				relayFrom = n.intf.myVpnAddrs[0]
+				relayFrom = cm.intf.myVpnAddrs[0]
 				relayTo = r.PeerAddr
 			case ForwardingType:
 				relayFrom = r.PeerAddr
 				relayTo = newhostinfo.vpnAddrs[0]
 			default:
 				// should never happen
+				panic(fmt.Sprintf("Migrating unknown relay type: %v", r.Type))
 			}
 		}
 
@@ -278,12 +281,12 @@ func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo)
 		switch newhostinfo.GetCert().Certificate.Version() {
 		case cert.Version1:
 			if !relayFrom.Is4() {
-				n.l.Error("can not migrate v1 relay with a v6 network because the relay is not running a current nebula version")
+				cm.l.Error("can not migrate v1 relay with a v6 network because the relay is not running a current nebula version")
 				continue
 			}
 
 			if !relayTo.Is4() {
-				n.l.Error("can not migrate v1 relay with a v6 remote network because the relay is not running a current nebula version")
+				cm.l.Error("can not migrate v1 relay with a v6 remote network because the relay is not running a current nebula version")
 				continue
 			}
 
@@ -295,16 +298,16 @@ func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo)
 			req.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 			req.RelayToAddr = netAddrToProtoAddr(relayTo)
 		default:
-			newhostinfo.logger(n.l).Error("Unknown certificate version found while attempting to migrate relay")
+			newhostinfo.logger(cm.l).Error("Unknown certificate version found while attempting to migrate relay")
 			continue
 		}
 
 		msg, err := req.Marshal()
 		if err != nil {
-			n.l.WithError(err).Error("failed to marshal Control message to migrate relay")
+			cm.l.WithError(err).Error("failed to marshal Control message to migrate relay")
 		} else {
-			n.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
-			n.l.WithFields(logrus.Fields{
+			cm.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
+			cm.l.WithFields(logrus.Fields{
 				"relayFrom":           req.RelayFromAddr,
 				"relayTo":             req.RelayToAddr,
 				"initiatorRelayIndex": req.InitiatorRelayIndex,
@@ -315,46 +318,45 @@ func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo)
 	}
 }
 
-func (n *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
-	n.hostMap.RLock()
-	defer n.hostMap.RUnlock()
+func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
+	// Read lock the main hostmap to order decisions based on tunnels being the primary tunnel
+	cm.hostMap.RLock()
+	defer cm.hostMap.RUnlock()
 
-	hostinfo := n.hostMap.Indexes[localIndex]
+	hostinfo := cm.hostMap.Indexes[localIndex]
 	if hostinfo == nil {
-		n.l.WithField("localIndex", localIndex).Debugf("Not found in hostmap")
-		delete(n.pendingDeletion, localIndex)
+		cm.l.WithField("localIndex", localIndex).Debugln("Not found in hostmap")
 		return doNothing, nil, nil
 	}
 
-	if n.isInvalidCertificate(now, hostinfo) {
-		delete(n.pendingDeletion, hostinfo.localIndexId)
+	if cm.isInvalidCertificate(now, hostinfo) {
 		return closeTunnel, hostinfo, nil
 	}
 
-	primary := n.hostMap.Hosts[hostinfo.vpnAddrs[0]]
+	primary := cm.hostMap.Hosts[hostinfo.vpnAddrs[0]]
 	mainHostInfo := true
 	if primary != nil && primary != hostinfo {
 		mainHostInfo = false
 	}
 
 	// Check for traffic on this hostinfo
-	inTraffic, outTraffic := n.getAndResetTrafficCheck(localIndex)
+	inTraffic, outTraffic := cm.getAndResetTrafficCheck(hostinfo, now)
 
 	// A hostinfo is determined alive if there is incoming traffic
 	if inTraffic {
 		decision := doNothing
-		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(n.l).
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.l).
 				WithField("tunnelCheck", m{"state": "alive", "method": "passive"}).
 				Debug("Tunnel status")
 		}
-		delete(n.pendingDeletion, hostinfo.localIndexId)
+		hostinfo.pendingDeletion.Store(false)
 
 		if mainHostInfo {
 			decision = tryRehandshake
 
 		} else {
-			if n.shouldSwapPrimary(hostinfo, primary) {
+			if cm.shouldSwapPrimary(hostinfo, primary) {
 				decision = swapPrimary
 			} else {
 				// migrate the relays to the primary, if in use.
@@ -362,46 +364,55 @@ func (n *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time
 			}
 		}
 
-		n.trafficTimer.Add(hostinfo.localIndexId, n.checkInterval)
+		cm.trafficTimer.Add(hostinfo.localIndexId, cm.checkInterval)
 
 		if !outTraffic {
 			// Send a punch packet to keep the NAT state alive
-			n.sendPunch(hostinfo)
+			cm.sendPunch(hostinfo)
 		}
 
 		return decision, hostinfo, primary
 	}
 
-	if _, ok := n.pendingDeletion[hostinfo.localIndexId]; ok {
+	if hostinfo.pendingDeletion.Load() {
 		// We have already sent a test packet and nothing was returned, this hostinfo is dead
-		hostinfo.logger(n.l).
+		hostinfo.logger(cm.l).
 			WithField("tunnelCheck", m{"state": "dead", "method": "active"}).
 			Info("Tunnel status")
 
-		delete(n.pendingDeletion, hostinfo.localIndexId)
 		return deleteTunnel, hostinfo, nil
 	}
 
 	decision := doNothing
 	if hostinfo != nil && hostinfo.ConnectionState != nil && mainHostInfo {
 		if !outTraffic {
+			inactiveFor, isInactive := cm.isInactive(hostinfo, now)
+			if isInactive {
+				// Tunnel is inactive, tear it down
+				hostinfo.logger(cm.l).
+					WithField("inactiveDuration", inactiveFor).
+					WithField("primary", mainHostInfo).
+					Info("Dropping tunnel due to inactivity")
+
+				return closeTunnel, hostinfo, primary
+			}
+
 			// If we aren't sending or receiving traffic then its an unused tunnel and we don't to test the tunnel.
 			// Just maintain NAT state if configured to do so.
-			n.sendPunch(hostinfo)
-			n.trafficTimer.Add(hostinfo.localIndexId, n.checkInterval)
+			cm.sendPunch(hostinfo)
+			cm.trafficTimer.Add(hostinfo.localIndexId, cm.checkInterval)
 			return doNothing, nil, nil
-
 		}
 
-		if n.punchy.GetTargetEverything() {
+		if cm.punchy.GetTargetEverything() {
 			// This is similar to the old punchy behavior with a slight optimization.
 			// We aren't receiving traffic but we are sending it, punch on all known
 			// ips in case we need to re-prime NAT state
-			n.sendPunch(hostinfo)
+			cm.sendPunch(hostinfo)
 		}
 
-		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(n.l).
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.l).
 				WithField("tunnelCheck", m{"state": "testing", "method": "active"}).
 				Debug("Tunnel status")
 		}
@@ -410,17 +421,33 @@ func (n *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time
 		decision = sendTestPacket
 
 	} else {
-		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(n.l).Debugf("Hostinfo sadness")
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.l).Debugf("Hostinfo sadness")
 		}
 	}
 
-	n.pendingDeletion[hostinfo.localIndexId] = struct{}{}
-	n.trafficTimer.Add(hostinfo.localIndexId, n.pendingDeletionInterval)
+	hostinfo.pendingDeletion.Store(true)
+	cm.trafficTimer.Add(hostinfo.localIndexId, cm.pendingDeletionInterval)
 	return decision, hostinfo, nil
 }
 
-func (n *connectionManager) shouldSwapPrimary(current, primary *HostInfo) bool {
+func (cm *connectionManager) isInactive(hostinfo *HostInfo, now time.Time) (time.Duration, bool) {
+	if cm.dropInactive.Load() == false {
+		// We aren't configured to drop inactive tunnels
+		return 0, false
+	}
+
+	inactiveDuration := now.Sub(hostinfo.lastUsed)
+	if inactiveDuration < cm.getInactivityTimeout() {
+		// It's not considered inactive
+		return inactiveDuration, false
+	}
+
+	// The tunnel is inactive
+	return inactiveDuration, true
+}
+
+func (cm *connectionManager) shouldSwapPrimary(current, primary *HostInfo) bool {
 	// The primary tunnel is the most recent handshake to complete locally and should work entirely fine.
 	// If we are here then we have multiple tunnels for a host pair and neither side believes the same tunnel is primary.
 	// Let's sort this out.
@@ -428,83 +455,90 @@ func (n *connectionManager) shouldSwapPrimary(current, primary *HostInfo) bool {
 	// Only one side should swap because if both swap then we may never resolve to a single tunnel.
 	// vpn addr is static across all tunnels for this host pair so lets
 	// use that to determine if we should consider swapping.
-	if current.vpnAddrs[0].Compare(n.intf.myVpnAddrs[0]) < 0 {
+	if current.vpnAddrs[0].Compare(cm.intf.myVpnAddrs[0]) < 0 {
 		// Their primary vpn addr is less than mine. Do not swap.
 		return false
 	}
 
-	crt := n.intf.pki.getCertState().getCertificate(current.ConnectionState.myCert.Version())
+	crt := cm.intf.pki.getCertState().getCertificate(current.ConnectionState.myCert.Version())
 	// If this tunnel is using the latest certificate then we should swap it to primary for a bit and see if things
 	// settle down.
 	return bytes.Equal(current.ConnectionState.myCert.Signature(), crt.Signature())
 }
 
-func (n *connectionManager) swapPrimary(current, primary *HostInfo) {
-	n.hostMap.Lock()
+func (cm *connectionManager) swapPrimary(current, primary *HostInfo) {
+	cm.hostMap.Lock()
 	// Make sure the primary is still the same after the write lock. This avoids a race with a rehandshake.
-	if n.hostMap.Hosts[current.vpnAddrs[0]] == primary {
-		n.hostMap.unlockedMakePrimary(current)
+	if cm.hostMap.Hosts[current.vpnAddrs[0]] == primary {
+		cm.hostMap.unlockedMakePrimary(current)
 	}
-	n.hostMap.Unlock()
+	cm.hostMap.Unlock()
 }
 
 // isInvalidCertificate will check if we should destroy a tunnel if pki.disconnect_invalid is true and
 // the certificate is no longer valid. Block listed certificates will skip the pki.disconnect_invalid
 // check and return true.
-func (n *connectionManager) isInvalidCertificate(now time.Time, hostinfo *HostInfo) bool {
+func (cm *connectionManager) isInvalidCertificate(now time.Time, hostinfo *HostInfo) bool {
 	remoteCert := hostinfo.GetCert()
 	if remoteCert == nil {
 		return false
 	}
 
-	caPool := n.intf.pki.GetCAPool()
+	caPool := cm.intf.pki.GetCAPool()
 	err := caPool.VerifyCachedCertificate(now, remoteCert)
 	if err == nil {
 		return false
 	}
 
-	if !n.intf.disconnectInvalid.Load() && err != cert.ErrBlockListed {
+	if !cm.intf.disconnectInvalid.Load() && err != cert.ErrBlockListed {
 		// Block listed certificates should always be disconnected
 		return false
 	}
 
-	hostinfo.logger(n.l).WithError(err).
+	hostinfo.logger(cm.l).WithError(err).
 		WithField("fingerprint", remoteCert.Fingerprint).
 		Info("Remote certificate is no longer valid, tearing down the tunnel")
 
 	return true
 }
 
-func (n *connectionManager) sendPunch(hostinfo *HostInfo) {
-	if !n.punchy.GetPunch() {
+func (cm *connectionManager) sendPunch(hostinfo *HostInfo) {
+	if !cm.punchy.GetPunch() {
 		// Punching is disabled
 		return
 	}
 
-	if n.punchy.GetTargetEverything() {
-		hostinfo.remotes.ForEach(n.hostMap.GetPreferredRanges(), func(addr netip.AddrPort, preferred bool) {
-			n.metricsTxPunchy.Inc(1)
-			n.intf.outside.WriteTo([]byte{1}, addr)
+	if cm.intf.lightHouse.IsAnyLighthouseAddr(hostinfo.vpnAddrs) {
+		// Do not punch to lighthouses, we assume our lighthouse update interval is good enough.
+		// In the event the update interval is not sufficient to maintain NAT state then a publicly available lighthouse
+		// would lose the ability to notify us and punchy.respond would become unreliable.
+		return
+	}
+
+	if cm.punchy.GetTargetEverything() {
+		hostinfo.remotes.ForEach(cm.hostMap.GetPreferredRanges(), func(addr netip.AddrPort, preferred bool) {
+			cm.metricsTxPunchy.Inc(1)
+			cm.intf.outside.WriteTo([]byte{1}, addr)
 		})
 
 	} else if hostinfo.remote.IsValid() {
-		n.metricsTxPunchy.Inc(1)
-		n.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
+		cm.metricsTxPunchy.Inc(1)
+		cm.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
 	}
 }
 
-func (n *connectionManager) tryRehandshake(hostinfo *HostInfo) {
-	cs := n.intf.pki.getCertState()
+func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
+	cs := cm.intf.pki.getCertState()
 	curCrt := hostinfo.ConnectionState.myCert
 	myCrt := cs.getCertificate(curCrt.Version())
-	if curCrt.Version() >= cs.defaultVersion && bytes.Equal(curCrt.Signature(), myCrt.Signature()) == true {
+	if curCrt.Version() >= cs.initiatingVersion && bytes.Equal(curCrt.Signature(), myCrt.Signature()) == true {
 		// The current tunnel is using the latest certificate and version, no need to rehandshake.
 		return
 	}
 
-	n.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+	cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
 		WithField("reason", "local certificate is not current").
 		Info("Re-handshaking with remote")
 
-	n.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
+	cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 }

connection_manager_test.go

@@ -1,7 +1,6 @@
 package nebula
 
 import (
-	"context"
 	"crypto/ed25519"
 	"crypto/rand"
 	"net/netip"
@@ -44,10 +43,10 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	hostMap.preferredRanges.Store(&preferredRanges)
 
 	cs := &CertState{
-		defaultVersion:   cert.Version1,
-		privateKey:       []byte{},
-		v1Cert:           &dummyCert{version: cert.Version1},
-		v1HandshakeBytes: []byte{},
+		initiatingVersion: cert.Version1,
+		privateKey:        []byte{},
+		v1Cert:            &dummyCert{version: cert.Version1},
+		v1HandshakeBytes:  []byte{},
 	}
 
 	lh := newTestLighthouse()
@@ -64,10 +63,10 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	nc.intf = ifce
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
@@ -85,32 +84,33 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
 	// We saw traffic out to vpnIp
-	nc.Out(hostinfo.localIndexId)
-	nc.In(hostinfo.localIndexId)
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	nc.Out(hostinfo)
+	nc.In(hostinfo)
+	assert.False(t, hostinfo.pendingDeletion.Load())
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
-	assert.Contains(t, nc.out, hostinfo.localIndexId)
+	assert.True(t, hostinfo.out.Load())
+	assert.True(t, hostinfo.in.Load())
 
 	// Do a traffic check tick, should not be pending deletion but should not have any in/out packets recorded
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.NotContains(t, nc.in, hostinfo.localIndexId)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
 
 	// Do another traffic check tick, this host should be pending deletion now
-	nc.Out(hostinfo.localIndexId)
+	nc.Out(hostinfo)
+	assert.True(t, hostinfo.out.Load())
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.NotContains(t, nc.in, hostinfo.localIndexId)
+	assert.True(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 
 	// Do a final traffic check tick, the host should now be removed
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
+	assert.NotContains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs)
 	assert.NotContains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 }
 
@@ -126,10 +126,10 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	hostMap.preferredRanges.Store(&preferredRanges)
 
 	cs := &CertState{
-		defaultVersion:   cert.Version1,
-		privateKey:       []byte{},
-		v1Cert:           &dummyCert{version: cert.Version1},
-		v1HandshakeBytes: []byte{},
+		initiatingVersion: cert.Version1,
+		privateKey:        []byte{},
+		v1Cert:            &dummyCert{version: cert.Version1},
+		v1HandshakeBytes:  []byte{},
 	}
 
 	lh := newTestLighthouse()
@@ -146,10 +146,10 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	nc.intf = ifce
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
@@ -167,33 +167,129 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
 	// We saw traffic out to vpnIp
-	nc.Out(hostinfo.localIndexId)
-	nc.In(hostinfo.localIndexId)
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.vpnAddrs[0])
+	nc.Out(hostinfo)
+	nc.In(hostinfo)
+	assert.True(t, hostinfo.in.Load())
+	assert.True(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.pendingDeletion.Load())
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 
 	// Do a traffic check tick, should not be pending deletion but should not have any in/out packets recorded
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.NotContains(t, nc.in, hostinfo.localIndexId)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
 
 	// Do another traffic check tick, this host should be pending deletion now
-	nc.Out(hostinfo.localIndexId)
+	nc.Out(hostinfo)
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.NotContains(t, nc.in, hostinfo.localIndexId)
+	assert.True(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 
 	// We saw traffic, should no longer be pending deletion
-	nc.In(hostinfo.localIndexId)
+	nc.In(hostinfo)
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.NotContains(t, nc.in, hostinfo.localIndexId)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
+	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
+}
+
+func Test_NewConnectionManager_DisconnectInactive(t *testing.T) {
+	l := test.NewLogger()
+	localrange := netip.MustParsePrefix("10.1.1.1/24")
+	vpnAddrs := []netip.Addr{netip.MustParseAddr("172.1.1.2")}
+	preferredRanges := []netip.Prefix{localrange}
+
+	// Very incomplete mock objects
+	hostMap := newHostMap(l)
+	hostMap.preferredRanges.Store(&preferredRanges)
+
+	cs := &CertState{
+		initiatingVersion: cert.Version1,
+		privateKey:        []byte{},
+		v1Cert:            &dummyCert{version: cert.Version1},
+		v1HandshakeBytes:  []byte{},
+	}
+
+	lh := newTestLighthouse()
+	ifce := &Interface{
+		hostMap:          hostMap,
+		inside:           &test.NoopTun{},
+		outside:          &udp.NoopConn{},
+		firewall:         &Firewall{},
+		lightHouse:       lh,
+		pki:              &PKI{},
+		handshakeManager: NewHandshakeManager(l, hostMap, lh, &udp.NoopConn{}, defaultHandshakeConfig),
+		l:                l,
+	}
+	ifce.pki.cs.Store(cs)
+
+	// Create manager
+	conf := config.NewC(l)
+	conf.Settings["tunnels"] = map[string]any{
+		"drop_inactive": true,
+	}
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	assert.True(t, nc.dropInactive.Load())
+	nc.intf = ifce
+
+	// Add an ip we have established a connection w/ to hostmap
+	hostinfo := &HostInfo{
+		vpnAddrs:      vpnAddrs,
+		localIndexId:  1099,
+		remoteIndexId: 9901,
+	}
+	hostinfo.ConnectionState = &ConnectionState{
+		myCert: &dummyCert{version: cert.Version1},
+		H:      &noise.HandshakeState{},
+	}
+	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
+
+	// Do a traffic check tick, in and out should be cleared but should not be pending deletion
+	nc.Out(hostinfo)
+	nc.In(hostinfo)
+	assert.True(t, hostinfo.out.Load())
+	assert.True(t, hostinfo.in.Load())
+
+	now := time.Now()
+	decision, _, _ := nc.makeTrafficDecision(hostinfo.localIndexId, now)
+	assert.Equal(t, tryRehandshake, decision)
+	assert.Equal(t, now, hostinfo.lastUsed)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
+
+	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Second*5))
+	assert.Equal(t, doNothing, decision)
+	assert.Equal(t, now, hostinfo.lastUsed)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
+
+	// Do another traffic check tick, should still not be pending deletion
+	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Second*10))
+	assert.Equal(t, doNothing, decision)
+	assert.Equal(t, now, hostinfo.lastUsed)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
+	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
+
+	// Finally advance beyond the inactivity timeout
+	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Minute*10))
+	assert.Equal(t, closeTunnel, decision)
+	assert.Equal(t, now, hostinfo.lastUsed)
+	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.False(t, hostinfo.out.Load())
+	assert.False(t, hostinfo.in.Load())
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 }
@@ -264,10 +360,10 @@ func Test_NewConnectionManagerTest_DisconnectInvalid(t *testing.T) {
 	ifce.disconnectInvalid.Store(true)
 
 	// Create manager
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	nc.intf = ifce
 	ifce.connectionManager = nc
 
 	hostinfo := &HostInfo{

connection_state.go

@@ -4,6 +4,7 @@ import (
 	"crypto/rand"
 	"encoding/json"
 	"fmt"
+	"sync"
 	"sync/atomic"
 
 	"github.com/flynn/noise"
@@ -23,7 +24,7 @@ type ConnectionState struct {
 	initiator      bool
 	messageCounter atomic.Uint64
 	window         *Bits
-	writeLock      syncMutex
+	writeLock      sync.Mutex
 }
 
 func NewConnectionState(l *logrus.Logger, cs *CertState, crt cert.Certificate, initiator bool, pattern noise.HandshakePattern) (*ConnectionState, error) {
@@ -75,8 +76,6 @@ func NewConnectionState(l *logrus.Logger, cs *CertState, crt cert.Certificate, i
 		initiator: initiator,
 		window:    b,
 		myCert:    crt,
-
-		writeLock: newSyncMutex("connection-state-write"),
 	}
 	// always start the counter from 2, as packet 1 and packet 2 are handshake packets.
 	ci.messageCounter.Add(2)

control.go

@@ -26,14 +26,15 @@ type controlHostLister interface {
 }
 
 type Control struct {
-	f               *Interface
-	l               *logrus.Logger
-	ctx             context.Context
-	cancel          context.CancelFunc
-	sshStart        func()
-	statsStart      func()
-	dnsStart        func()
-	lighthouseStart func()
+	f                      *Interface
+	l                      *logrus.Logger
+	ctx                    context.Context
+	cancel                 context.CancelFunc
+	sshStart               func()
+	statsStart             func()
+	dnsStart               func()
+	lighthouseStart        func()
+	connectionManagerStart func(context.Context)
 }
 
 type ControlHostInfo struct {
@@ -63,6 +64,9 @@ func (c *Control) Start() {
 	if c.dnsStart != nil {
 		go c.dnsStart()
 	}
+	if c.connectionManagerStart != nil {
+		go c.connectionManagerStart(c.ctx)
+	}
 	if c.lighthouseStart != nil {
 		c.lighthouseStart()
 	}
@@ -131,8 +135,7 @@ func (c *Control) ListHostmapIndexes(pendingMap bool) []ControlHostInfo {
 
 // GetCertByVpnIp returns the authenticated certificate of the given vpn IP, or nil if not found
 func (c *Control) GetCertByVpnIp(vpnIp netip.Addr) cert.Certificate {
-	_, found := c.f.myVpnAddrsTable.Lookup(vpnIp)
-	if found {
+	if c.f.myVpnAddrsTable.Contains(vpnIp) {
 		// Only returning the default certificate since its impossible
 		// for any other host but ourselves to have more than 1
 		return c.f.pki.getCertState().GetDefaultCertificate().Copy()

control_test.go

@@ -53,7 +53,7 @@ func TestControl_GetHostInfoByVpnIp(t *testing.T) {
 		localIndexId:  201,
 		vpnAddrs:      []netip.Addr{vpnIp},
 		relayState: RelayState{
-			relays:         map[netip.Addr]struct{}{},
+			relays:         nil,
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
@@ -72,7 +72,7 @@ func TestControl_GetHostInfoByVpnIp(t *testing.T) {
 		localIndexId:  201,
 		vpnAddrs:      []netip.Addr{vpnIp2},
 		relayState: RelayState{
-			relays:         map[netip.Addr]struct{}{},
+			relays:         nil,
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},

dns_server.go

@@ -6,6 +6,7 @@ import (
 	"net/netip"
 	"strconv"
 	"strings"
+	"sync"
 
 	"github.com/gaissmai/bart"
 	"github.com/miekg/dns"
@@ -20,17 +21,16 @@ var dnsServer *dns.Server
 var dnsAddr string
 
 type dnsRecords struct {
-	syncRWMutex
+	sync.RWMutex
 	l               *logrus.Logger
 	dnsMap4         map[string]netip.Addr
 	dnsMap6         map[string]netip.Addr
 	hostMap         *HostMap
-	myVpnAddrsTable *bart.Table[struct{}]
+	myVpnAddrsTable *bart.Lite
 }
 
 func newDnsRecords(l *logrus.Logger, cs *CertState, hostMap *HostMap) *dnsRecords {
 	return &dnsRecords{
-		syncRWMutex:     newSyncRWMutex("dns-records"),
 		l:               l,
 		dnsMap4:         make(map[string]netip.Addr),
 		dnsMap6:         make(map[string]netip.Addr),
@@ -112,8 +112,8 @@ func (d *dnsRecords) isSelfNebulaOrLocalhost(addr string) bool {
 		return true
 	}
 
-	_, found := d.myVpnAddrsTable.Lookup(b)
-	return found //if we found it in this table, it's good
+	//if we found it in this table, it's good
+	return d.myVpnAddrsTable.Contains(b)
 }
 
 func (d *dnsRecords) parseQuery(m *dns.Msg, w dns.ResponseWriter) {

e2e/handshakes_test.go

@@ -506,7 +506,7 @@ func TestReestablishRelays(t *testing.T) {
 	curIndexes := len(myControl.GetHostmap().Indexes)
 	for curIndexes >= start {
 		curIndexes = len(myControl.GetHostmap().Indexes)
-		r.Logf("Wait for the dead index to go away:start=%v indexes, currnet=%v indexes", start, curIndexes)
+		r.Logf("Wait for the dead index to go away:start=%v indexes, current=%v indexes", start, curIndexes)
 		myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi from me should fail"))
 
 		r.RouteForAllExitFunc(func(p *udp.Packet, c *nebula.Control) router.ExitType {
@@ -1052,6 +1052,9 @@ func TestRehandshakingLoser(t *testing.T) {
 	t.Log("Stand up a tunnel between me and them")
 	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
 
+	myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
+	theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
+
 	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
 
 	r.Log("Renew their certificate and spin until mine sees it")

e2e/router/router.go

@@ -700,6 +700,7 @@ func (r *R) FlushAll() {
 			r.Unlock()
 			panic("Can't FlushAll for host: " + p.To.String())
 		}
+		receiver.InjectUDPPacket(p)
 		r.Unlock()
 	}
 }

e2e/tunnels_test.go

@@ -0,0 +1,57 @@
+//go:build e2e_testing
+// +build e2e_testing
+
+package e2e
+
+import (
+	"testing"
+	"time"
+
+	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/cert_test"
+	"github.com/slackhq/nebula/e2e/router"
+)
+
+func TestDropInactiveTunnels(t *testing.T) {
+	// The goal of this test is to ensure the shortest inactivity timeout will close the tunnel on both sides
+	// under ideal conditions
+	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "5s"}})
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "10m"}})
+
+	// Share our underlay information
+	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	r := router.NewR(t, myControl, theirControl)
+
+	r.Log("Assert the tunnel between me and them works")
+	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
+
+	r.Log("Go inactive and wait for the tunnels to get dropped")
+	waitStart := time.Now()
+	for {
+		myIndexes := len(myControl.GetHostmap().Indexes)
+		theirIndexes := len(theirControl.GetHostmap().Indexes)
+		if myIndexes == 0 && theirIndexes == 0 {
+			break
+		}
+
+		since := time.Since(waitStart)
+		r.Logf("my tunnels: %v; their tunnels: %v; duration: %v", myIndexes, theirIndexes, since)
+		if since > time.Second*30 {
+			t.Fatal("Tunnel should have been declared inactive after 5 seconds and before 30 seconds")
+		}
+
+		time.Sleep(1 * time.Second)
+		r.FlushAll()
+	}
+
+	r.Logf("Inactive tunnels were dropped within %v", time.Since(waitStart))
+	myControl.Stop()
+	theirControl.Stop()
+}

examples/config.yml

@@ -13,11 +13,11 @@ pki:
   # disconnect_invalid is a toggle to force a client to be disconnected if the certificate is expired or invalid.
   #disconnect_invalid: true
 
-  # default_version controls which certificate version is used in handshakes.
+  # initiating_version controls which certificate version is used when initiating handshakes.
   # This setting only applies if both a v1 and a v2 certificate are configured, in which case it will default to `1`.
   # Once all hosts in the mesh are configured with both a v1 and v2 certificate then this should be changed to `2`.
   # After all hosts in the mesh are using a v2 certificate then v1 certificates are no longer needed.
-  # default_version: 1
+  # initiating_version: 1
 
 # The static host map defines a set of hosts with fixed IP addresses on the internet (or any network).
 # A host can have multiple fixed IP addresses defined here, and nebula will try each when establishing a tunnel.
@@ -275,6 +275,51 @@ tun:
   # On linux only, set to true to manage unsafe routes directly on the system route table with gateway routes instead of
   # in nebula configuration files. Default false, not reloadable.
   #use_system_route_table: false
+  # Buffer size for reading routes updates. 0 means default system buffer size. (/proc/sys/net/core/rmem_default).
+  # If using massive routes updates, for example BGP, you may need to increase this value to avoid packet loss.
+  # SO_RCVBUFFORCE is used to avoid having to raise the system wide max
+  #use_system_route_table_buffer_size: 0
+
+  # EXPERIMENTAL: This option may change or disappear in the future.
+  # Multiport spreads outgoing UDP packets across multiple UDP send ports,
+  # which allows nebula to work around any issues on the underlay network.
+  # Some example issues this could work around:
+  #  - UDP rate limits on a per flow basis.
+  #  - Partial underlay network failure in which some flows work and some don't
+  # Agreement is done during the handshake to decide if multiport mode will
+  # be used for a given tunnel (one side must have tx_enabled set, the other
+  # side must have rx_enabled set)
+  #
+  # NOTE: you cannot use multiport on a host if you are relying on UDP hole
+  # punching to get through a NAT or firewall.
+  #
+  # NOTE: Linux only (uses raw sockets to send). Also currently only works
+  # with IPv4 underlay network remotes.
+  #
+  # The default values are listed below:
+  #multiport:
+    # This host support sending via multiple UDP ports.
+    #tx_enabled: false
+    #
+    # This host supports receiving packets sent from multiple UDP ports.
+    #rx_enabled: false
+    #
+    # How many UDP ports to use when sending. The lowest source port will be
+    # listen.port and go up to (but not including) listen.port + tx_ports.
+    #tx_ports: 100
+    #
+    # NOTE: All of your hosts must be running a version of Nebula that supports
+    # multiport if you want to enable this feature. Older versions of Nebula
+    # will be confused by these multiport handshakes.
+    #
+    # If handshakes are not getting a response, attempt to transmit handshakes
+    # using random UDP source ports (to get around partial underlay network
+    # failures).
+    #tx_handshake: false
+    #
+    # How many unresponded handshakes we should send before we attempt to
+    # send multiport handshakes.
+    #tx_handshake_delay: 2
 
 # Configure logging level
 logging:
@@ -334,6 +379,18 @@ logging:
   # after receiving the response for lighthouse queries
   #trigger_buffer: 64
 
+# Tunnel manager settings
+#tunnels:
+  # drop_inactive controls whether inactive tunnels are maintained or dropped after the inactive_timeout period has
+  # elapsed.
+  # In general, it is a good idea to enable this setting. It will be enabled by default in a future release.
+  # This setting is reloadable
+  #drop_inactive: false
+
+  # inactivity_timeout controls how long a tunnel MUST NOT see any inbound or outbound traffic before being considered
+  # inactive and eligible to be dropped.
+  # This setting is reloadable
+  #inactivity_timeout: 10m
 
 # Nebula security group configuration
 firewall:

examples/go_service/main.go

@@ -5,8 +5,12 @@ import (
 	"fmt"
 	"log"
 	"net"
+	"os"
 
+	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/overlay"
 	"github.com/slackhq/nebula/service"
 )
 
@@ -59,7 +63,16 @@ pki:
 	if err := cfg.LoadString(configStr); err != nil {
 		return err
 	}
-	svc, err := service.New(&cfg)
+
+	logger := logrus.New()
+	logger.Out = os.Stdout
+
+	ctrl, err := nebula.Main(&cfg, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
+	if err != nil {
+		return err
+	}
+
+	svc, err := service.New(ctrl)
 	if err != nil {
 		return err
 	}

firewall.go

@@ -10,6 +10,7 @@ import (
 	"reflect"
 	"strconv"
 	"strings"
+	"sync"
 	"time"
 
 	"github.com/gaissmai/bart"
@@ -52,7 +53,7 @@ type Firewall struct {
 
 	// routableNetworks describes the vpn addresses as well as any unsafe networks issued to us in the certificate.
 	// The vpn addresses are a full bit match while the unsafe networks only match the prefix
-	routableNetworks *bart.Table[struct{}]
+	routableNetworks *bart.Lite
 
 	// assignedNetworks is a list of vpn networks assigned to us in the certificate.
 	assignedNetworks  []netip.Prefix
@@ -75,7 +76,7 @@ type firewallMetrics struct {
 }
 
 type FirewallConntrack struct {
-	syncMutex
+	sync.Mutex
 
 	Conns      map[firewall.Packet]*conn
 	TimerWheel *TimerWheel[firewall.Packet]
@@ -124,7 +125,7 @@ type firewallPort map[int32]*FirewallCA
 
 type firewallLocalCIDR struct {
 	Any       bool
-	LocalCIDR *bart.Table[struct{}]
+	LocalCIDR *bart.Lite
 }
 
 // NewFirewall creates a new Firewall object. A TimerWheel is created for you from the provided timeouts.
@@ -147,23 +148,22 @@ func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.D
 		tmax = defaultTimeout
 	}
 
-	routableNetworks := new(bart.Table[struct{}])
+	routableNetworks := new(bart.Lite)
 	var assignedNetworks []netip.Prefix
 	for _, network := range c.Networks() {
 		nprefix := netip.PrefixFrom(network.Addr(), network.Addr().BitLen())
-		routableNetworks.Insert(nprefix, struct{}{})
+		routableNetworks.Insert(nprefix)
 		assignedNetworks = append(assignedNetworks, network)
 	}
 
 	hasUnsafeNetworks := false
 	for _, n := range c.UnsafeNetworks() {
-		routableNetworks.Insert(n, struct{}{})
+		routableNetworks.Insert(n)
 		hasUnsafeNetworks = true
 	}
 
 	return &Firewall{
 		Conntrack: &FirewallConntrack{
-			syncMutex:  newSyncMutex("firewall-conntrack"),
 			Conns:      make(map[firewall.Packet]*conn),
 			TimerWheel: NewTimerWheel[firewall.Packet](tmin, tmax),
 		},
@@ -431,8 +431,7 @@ func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *
 
 	// Make sure remote address matches nebula certificate
 	if h.networks != nil {
-		_, ok := h.networks.Lookup(fp.RemoteAddr)
-		if !ok {
+		if !h.networks.Contains(fp.RemoteAddr) {
 			f.metrics(incoming).droppedRemoteAddr.Inc(1)
 			return ErrInvalidRemoteIP
 		}
@@ -445,8 +444,7 @@ func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *
 	}
 
 	// Make sure we are supposed to be handling this local ip address
-	_, ok := f.routableNetworks.Lookup(fp.LocalAddr)
-	if !ok {
+	if !f.routableNetworks.Contains(fp.LocalAddr) {
 		f.metrics(incoming).droppedLocalAddr.Inc(1)
 		return ErrInvalidLocalIP
 	}
@@ -752,7 +750,7 @@ func (fc *FirewallCA) match(p firewall.Packet, c *cert.CachedCertificate, caPool
 func (fr *FirewallRule) addRule(f *Firewall, groups []string, host string, ip, localCIDR netip.Prefix) error {
 	flc := func() *firewallLocalCIDR {
 		return &firewallLocalCIDR{
-			LocalCIDR: new(bart.Table[struct{}]),
+			LocalCIDR: new(bart.Lite),
 		}
 	}
 
@@ -879,7 +877,7 @@ func (flc *firewallLocalCIDR) addRule(f *Firewall, localIp netip.Prefix) error {
 		}
 
 		for _, network := range f.assignedNetworks {
-			flc.LocalCIDR.Insert(network, struct{}{})
+			flc.LocalCIDR.Insert(network)
 		}
 		return nil
 
@@ -888,7 +886,7 @@ func (flc *firewallLocalCIDR) addRule(f *Firewall, localIp netip.Prefix) error {
 		return nil
 	}
 
-	flc.LocalCIDR.Insert(localIp, struct{}{})
+	flc.LocalCIDR.Insert(localIp)
 	return nil
 }
 
@@ -901,8 +899,7 @@ func (flc *firewallLocalCIDR) match(p firewall.Packet, c *cert.CachedCertificate
 		return true
 	}
 
-	_, ok := flc.LocalCIDR.Lookup(p.LocalAddr)
-	return ok
+	return flc.LocalCIDR.Contains(p.LocalAddr)
 }
 
 type rule struct {

firewall/packet.go

@@ -3,6 +3,7 @@ package firewall
 import (
 	"encoding/json"
 	"fmt"
+	mathrand "math/rand"
 	"net/netip"
 )
 
@@ -60,3 +61,30 @@ func (fp Packet) MarshalJSON() ([]byte, error) {
 		"Fragment":   fp.Fragment,
 	})
 }
+
+// UDPSendPort calculates the UDP port to send from when using multiport mode.
+// The result will be from [0, numBuckets)
+func (fp Packet) UDPSendPort(numBuckets int) uint16 {
+	if numBuckets <= 1 {
+		return 0
+	}
+
+	// If there is no port (like an ICMP packet), pick a random UDP send port
+	if fp.LocalPort == 0 {
+		return uint16(mathrand.Intn(numBuckets))
+	}
+
+	// A decent enough 32bit hash function
+	// Prospecting for Hash Functions
+	// - https://nullprogram.com/blog/2018/07/31/
+	// - https://github.com/skeeto/hash-prospector
+	//   [16 21f0aaad 15 d35a2d97 15] = 0.10760229515479501
+	x := (uint32(fp.LocalPort) << 16) | uint32(fp.RemotePort)
+	x ^= x >> 16
+	x *= 0x21f0aaad
+	x ^= x >> 15
+	x *= 0xd35a2d97
+	x ^= x >> 15
+
+	return uint16(x) % uint16(numBuckets)
+}

go.mod

@@ -1,37 +1,35 @@
 module github.com/slackhq/nebula
 
-go 1.23.6
+go 1.23.0
 
 toolchain go1.24.1
 
 require (
-	dario.cat/mergo v1.0.1
+	dario.cat/mergo v1.0.2
 	github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be
 	github.com/armon/go-radix v1.0.0
-	github.com/clarkmcc/go-dag v0.0.0-20220908000337-9c3ba5b365fc
 	github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432
 	github.com/flynn/noise v1.1.0
-	github.com/gaissmai/bart v0.20.1
+	github.com/gaissmai/bart v0.20.4
 	github.com/gogo/protobuf v1.3.2
 	github.com/google/gopacket v1.1.19
 	github.com/kardianos/service v1.2.2
-	github.com/miekg/dns v1.1.64
+	github.com/miekg/dns v1.1.65
 	github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b
 	github.com/nbrownus/go-metrics-prometheus v0.0.0-20210712211119-974a6260965f
-	github.com/prometheus/client_golang v1.21.1
+	github.com/prometheus/client_golang v1.22.0
 	github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475
 	github.com/sirupsen/logrus v1.9.3
 	github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e
 	github.com/stefanberger/go-pkcs11uri v0.0.0-20230803200340-78284954bff6
 	github.com/stretchr/testify v1.10.0
-	github.com/timandy/routine v1.1.5
-	github.com/vishvananda/netlink v1.3.0
-	golang.org/x/crypto v0.36.0
+	github.com/vishvananda/netlink v1.3.1
+	golang.org/x/crypto v0.37.0
 	golang.org/x/exp v0.0.0-20230725093048-515e97ebf090
-	golang.org/x/net v0.38.0
-	golang.org/x/sync v0.12.0
-	golang.org/x/sys v0.31.0
-	golang.org/x/term v0.30.0
+	golang.org/x/net v0.39.0
+	golang.org/x/sync v0.13.0
+	golang.org/x/sys v0.32.0
+	golang.org/x/term v0.31.0
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2
 	golang.zx2c4.com/wireguard v0.0.0-20230325221338-052af4a8072b
 	golang.zx2c4.com/wireguard/windows v0.5.3
@@ -45,13 +43,12 @@ require (
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/google/btree v1.1.2 // indirect
-	github.com/klauspost/compress v1.17.11 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/prometheus/client_model v0.6.1 // indirect
 	github.com/prometheus/common v0.62.0 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect
-	github.com/vishvananda/netns v0.0.4 // indirect
+	github.com/vishvananda/netns v0.0.5 // indirect
 	golang.org/x/mod v0.23.0 // indirect
 	golang.org/x/time v0.5.0 // indirect
 	golang.org/x/tools v0.30.0 // indirect

go.sum

@@ -1,6 +1,6 @@
 cloud.google.com/go v0.34.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
-dario.cat/mergo v1.0.1 h1:Ra4+bf83h2ztPIQYNP99R6m+Y7KfnARDfID+a+vLl4s=
-dario.cat/mergo v1.0.1/go.mod h1:uNxQE+84aUszobStD9th8a29P2fMDhsBdgRYvZOxGmk=
+dario.cat/mergo v1.0.2 h1:85+piFYR1tMbRrLcDwR18y4UKJ3aH1Tbzi24VRW1TK8=
+dario.cat/mergo v1.0.2/go.mod h1:E/hbnu0NxMFBjpMIE34DRGLWqDy0g5FuKDhCb31ngxA=
 github.com/alecthomas/template v0.0.0-20160405071501-a0175ee3bccc/go.mod h1:LOuyumcjzFXgccqObfd/Ljyb9UuFJ6TxHnclSeseNhc=
 github.com/alecthomas/template v0.0.0-20190718012654-fb15b899a751/go.mod h1:LOuyumcjzFXgccqObfd/Ljyb9UuFJ6TxHnclSeseNhc=
 github.com/alecthomas/units v0.0.0-20151022065526-2efee857e7cf/go.mod h1:ybxpYRFXyAe+OPACYpWeL0wqObRcbAqCMya13uyzqw0=
@@ -17,8 +17,6 @@ github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6r
 github.com/cespare/xxhash/v2 v2.1.1/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
 github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
 github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
-github.com/clarkmcc/go-dag v0.0.0-20220908000337-9c3ba5b365fc h1:6e91sWiDE69Jl0WUsY/LvTCBPRBe6b2j8H7W96JGJ4s=
-github.com/clarkmcc/go-dag v0.0.0-20220908000337-9c3ba5b365fc/go.mod h1:RGIcF96ORCYAsdz60Ou9mPBNa4+DjoQFS8nelPniFoY=
 github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432 h1:M5QgkYacWj0Xs8MhpIK/5uwU02icXpEoSo9sM2aRCps=
 github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432/go.mod h1:xwIwAxMvYnVrGJPe2FKx5prTrnAjGOD8zvDOnxnrrkM=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
@@ -26,8 +24,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/flynn/noise v1.1.0 h1:KjPQoQCEFdZDiP03phOvGi11+SVVhBG2wOWAorLsstg=
 github.com/flynn/noise v1.1.0/go.mod h1:xbMo+0i6+IGbYdJhF31t2eR1BIU0CYc12+BNAKwUTag=
-github.com/gaissmai/bart v0.20.1 h1:igNss0zDsSY8e+ophKgD9KJVPKBOo7uSVjyKCL7nIzo=
-github.com/gaissmai/bart v0.20.1/go.mod h1:JJzMAhNF5Rjo4SF4jWBrANuJfqY+FvsFhW7t1UZJ+XY=
+github.com/gaissmai/bart v0.20.4 h1:Ik47r1fy3jRVU+1eYzKSW3ho2UgBVTVnUS8O993584U=
+github.com/gaissmai/bart v0.20.4/go.mod h1:cEed+ge8dalcbpi8wtS9x9m2hn/fNJH5suhdGQOHnYk=
 github.com/go-kit/kit v0.8.0/go.mod h1:xBxKIO96dXMWWy0MnWVtmwkA9/13aqxPnvrjFYMA2as=
 github.com/go-kit/kit v0.9.0/go.mod h1:xBxKIO96dXMWWy0MnWVtmwkA9/13aqxPnvrjFYMA2as=
 github.com/go-kit/log v0.1.0/go.mod h1:zbhenjAZHb184qTLMA9ZjW7ThYL0H2mk7Q6pNt4vbaY=
@@ -55,8 +53,8 @@ github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMyw
 github.com/google/go-cmp v0.4.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.4/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
-github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/gopacket v1.1.19 h1:ves8RnFZPGiFnTS0uPQStjwru6uO6h+nlr9j6fL7kF8=
 github.com/google/gopacket v1.1.19/go.mod h1:iJ8V8n6KS+z2U1A8pUwu8bW5SyEMkXJB8Yo/Vo+TKTo=
@@ -70,8 +68,8 @@ github.com/kardianos/service v1.2.2 h1:ZvePhAHfvo0A7Mftk/tEzqEZ7Q4lgnR8sGz4xu1YX
 github.com/kardianos/service v1.2.2/go.mod h1:CIMRFEJVL+0DS1a3Nx06NaMn4Dz63Ng6O7dl0qH0zVM=
 github.com/kisielk/errcheck v1.5.0/go.mod h1:pFxgyoBC7bSaBwPgfKdkLd5X25qrDl4LWUI2bnpBCr8=
 github.com/kisielk/gotool v1.0.0/go.mod h1:XhKaO+MFFWcvkIS/tQcRk01m1F5IRFswLeQ+oQHNcck=
-github.com/klauspost/compress v1.17.11 h1:In6xLpyWOi1+C7tXUUWv2ot1QvBjxevKAaI6IXrJmUc=
-github.com/klauspost/compress v1.17.11/go.mod h1:pMDklpSncoRMuLFrf1W9Ss9KT+0rH90U12bZKk7uwG0=
+github.com/klauspost/compress v1.18.0 h1:c/Cqfb0r+Yi+JtIEq73FWXVkRonBlf0CRNYc8Zttxdo=
+github.com/klauspost/compress v1.18.0/go.mod h1:2Pp+KzxcywXVXMr50+X0Q/Lsb43OQHYWRCY2AiWywWQ=
 github.com/konsorten/go-windows-terminal-sequences v1.0.1/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/konsorten/go-windows-terminal-sequences v1.0.3/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/kr/logfmt v0.0.0-20140226030751-b84e30acd515/go.mod h1:+0opPa2QZZtGFBFZlji/RkVcI2GknAs/DXo4wKdlNEc=
@@ -85,8 +83,8 @@ github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0SNc=
 github.com/kylelemons/godebug v1.1.0/go.mod h1:9/0rRGxNHcop5bhtWyNeEfOS8JIWk580+fNqagV/RAw=
 github.com/matttproud/golang_protobuf_extensions v1.0.1/go.mod h1:D8He9yQNgCq6Z5Ld7szi9bcBfOoFv/3dc6xSMkL2PC0=
-github.com/miekg/dns v1.1.64 h1:wuZgD9wwCE6XMT05UU/mlSko71eRSXEAm2EbjQXLKnQ=
-github.com/miekg/dns v1.1.64/go.mod h1:Dzw9769uoKVaLuODMDZz9M6ynFU6Em65csPuoi8G0ck=
+github.com/miekg/dns v1.1.65 h1:0+tIPHzUW0GCge7IiK3guGP57VAw7hoPDfApjkMD1Fc=
+github.com/miekg/dns v1.1.65/go.mod h1:Dzw9769uoKVaLuODMDZz9M6ynFU6Em65csPuoi8G0ck=
 github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b h1:J/AzCvg5z0Hn1rqZUJjpbzALUmkKX0Zwbc/i4fw7Sfk=
 github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b/go.mod h1:XsNlhZGX73bx86s2hdc/FuaLm2CPZJemRLMA+WTFxgs=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
@@ -108,8 +106,8 @@ github.com/prometheus/client_golang v0.9.1/go.mod h1:7SWBe2y4D6OKWSNQJUaRYU/AaXP
 github.com/prometheus/client_golang v1.0.0/go.mod h1:db9x61etRT2tGnBNRi70OPL5FsnadC4Ky3P0J6CfImo=
 github.com/prometheus/client_golang v1.7.1/go.mod h1:PY5Wy2awLA44sXw4AOSfFBetzPP4j5+D6mVACh+pe2M=
 github.com/prometheus/client_golang v1.11.0/go.mod h1:Z6t4BnS23TR94PD6BsDNk8yVqroYurpAkEiz0P2BEV0=
-github.com/prometheus/client_golang v1.21.1 h1:DOvXXTqVzvkIewV/CDPFdejpMCGeMcbGCQ8YOmu+Ibk=
-github.com/prometheus/client_golang v1.21.1/go.mod h1:U9NM32ykUErtVBxdvD3zfi+EuFkkaBvMb09mIfe0Zgg=
+github.com/prometheus/client_golang v1.22.0 h1:rb93p9lokFEsctTys46VnV1kLCDpVZ0a/Y92Vm0Zc6Q=
+github.com/prometheus/client_golang v1.22.0/go.mod h1:R7ljNsLXhuQXYZYtw6GAE9AZg8Y7vEW5scdCXrWRXC0=
 github.com/prometheus/client_model v0.0.0-20180712105110-5c3871d89910/go.mod h1:MbSGuTsp3dbXC40dX6PRTWyKYBIrTGTE9sqQNg2J8bo=
 github.com/prometheus/client_model v0.0.0-20190129233127-fd36f4220a90/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
 github.com/prometheus/client_model v0.2.0/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
@@ -147,14 +145,10 @@ github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81P
 github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.10.0 h1:Xv5erBjTwe/5IxqUQTdXv5kgmIvbHo3QQyRwhJsOfJA=
 github.com/stretchr/testify v1.10.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
-github.com/timandy/routine v1.1.1 h1:6/Z7qLFZj3GrzuRksBFzIG8YGUh8CLhjnnMePBQTrEI=
-github.com/timandy/routine v1.1.1/go.mod h1:OZHPOKSvqL/ZvqXFkNZyit0xIVelERptYXdAHH00adQ=
-github.com/timandy/routine v1.1.5 h1:LSpm7Iijwb9imIPlucl4krpr2EeCeAUvifiQ9Uf5X+M=
-github.com/timandy/routine v1.1.5/go.mod h1:kXslgIosdY8LW0byTyPnenDgn4/azt2euufAq9rK51w=
-github.com/vishvananda/netlink v1.3.0 h1:X7l42GfcV4S6E4vHTsw48qbrV+9PVojNfIhZcwQdrZk=
-github.com/vishvananda/netlink v1.3.0/go.mod h1:i6NetklAujEcC6fK0JPjT8qSwWyO0HLn4UKG+hGqeJs=
-github.com/vishvananda/netns v0.0.4 h1:Oeaw1EM2JMxD51g9uhtC0D7erkIjgmj8+JZc26m1YX8=
-github.com/vishvananda/netns v0.0.4/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+github.com/vishvananda/netlink v1.3.1 h1:3AEMt62VKqz90r0tmNhog0r/PpWKmrEShJU0wJW6bV0=
+github.com/vishvananda/netlink v1.3.1/go.mod h1:ARtKouGSTGchR8aMwmkzC0qiNPrrWO5JS/XMVl45+b4=
+github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zdEY=
+github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 golang.org/x/crypto v0.0.0-20180904163835-0709b304e793/go.mod h1:6SG95UA2DQfeDnfUPMdvaQW0Q7yPrPDi9nlGo2tz2b4=
@@ -162,8 +156,8 @@ golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACk
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20210322153248-0c34fe9e7dc2/go.mod h1:T9bdIzuCu7OtxOm1hfPfRQxPLYneinmdGuTeoZ9dtd4=
-golang.org/x/crypto v0.36.0 h1:AnAEvhDddvBdpY+uR+MyHmuZzzNqXSe/GvuDeob5L34=
-golang.org/x/crypto v0.36.0/go.mod h1:Y4J0ReaxCR1IMaabaSMugxJES1EpwhBHhv2bDHklZvc=
+golang.org/x/crypto v0.37.0 h1:kJNSjF/Xp7kU0iB2Z+9viTPMW4EqqsrywMXLJOOsXSE=
+golang.org/x/crypto v0.37.0/go.mod h1:vg+k43peMZ0pUMhYmVAWysMK35e6ioLh3wB8ZCAfbVc=
 golang.org/x/exp v0.0.0-20230725093048-515e97ebf090 h1:Di6/M8l0O2lCLc6VVRWhgCiApHV8MnQurBnFSHsQtNY=
 golang.org/x/exp v0.0.0-20230725093048-515e97ebf090/go.mod h1:FXUEEKJgO7OQYeo8N01OfiKP8RXMtf6e8aTskBGqWdc=
 golang.org/x/lint v0.0.0-20200302205851-738671d3881b/go.mod h1:3xt1FjdF8hUf6vQPIChWIBhFzV8gjjsPE/fR3IyQdNY=
@@ -182,8 +176,8 @@ golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLL
 golang.org/x/net v0.0.0-20200625001655-4c5254603344/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.38.0 h1:vRMAPTMaeGqVhG5QyLJHqNDwecKTomGeqbnfZyKlBI8=
-golang.org/x/net v0.38.0/go.mod h1:ivrbrMbzFq5J41QOQh0siUuly180yBYtLp+CKbEaFx8=
+golang.org/x/net v0.39.0 h1:ZCu7HMWDxpXpaiKdhzIfaltL9Lp31x/3fCP11bc6/fY=
+golang.org/x/net v0.39.0/go.mod h1:X7NRbYVEA+ewNkCNyJ513WmMdQ3BineSwVtN2zD/d+E=
 golang.org/x/oauth2 v0.0.0-20190226205417-e64efc72b421/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw=
 golang.org/x/sync v0.0.0-20181108010431-42b317875d0f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
@@ -191,8 +185,8 @@ golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201207232520-09787c993a3a/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.12.0 h1:MHc5BpPuC30uJk597Ri8TV3CNZcTLu6B6z4lJy+g6Jw=
-golang.org/x/sync v0.12.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
+golang.org/x/sync v0.13.0 h1:AauUjRAJ9OSnvULf/ARrrVywoJDy0YS2AwQ98I37610=
+golang.org/x/sync v0.13.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
 golang.org/x/sys v0.0.0-20180905080454-ebe1bf3edb33/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20181116152217-5ac8a444bdc5/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@@ -210,11 +204,11 @@ golang.org/x/sys v0.0.0-20210603081109-ebe580a85c40/go.mod h1:oPkhp1MJrh7nUepCBc
 golang.org/x/sys v0.0.0-20220715151400-c0bba94af5f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.31.0 h1:ioabZlmFYtWhL+TRYpcnNlLwhyxaM9kWTDEmfnprqik=
-golang.org/x/sys v0.31.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
+golang.org/x/sys v0.32.0 h1:s77OFDvIQeibCmezSnk/q6iAfkdiQaJi4VzroCFrN20=
+golang.org/x/sys v0.32.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
-golang.org/x/term v0.30.0 h1:PQ39fJZ+mfadBm0y5WlL4vlM7Sx1Hgf13sMIY2+QS9Y=
-golang.org/x/term v0.30.0/go.mod h1:NYYFdzHoI5wRh/h5tDMdMqCqPJZEuNqVR5xJLd/n67g=
+golang.org/x/term v0.31.0 h1:erwDkOK1Msy6offm1mOgvspSkslFnIGsFnxOKoufg3o=
+golang.org/x/term v0.31.0/go.mod h1:R4BeIy7D95HzImkxGkTW1UQTtP54tio2RyHz7PwK0aw=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=

handshake_ix.go

@@ -9,6 +9,7 @@ import (
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/udp"
 )
 
 // NOISE IX Handshakes
@@ -25,7 +26,7 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 
 	// If we're connecting to a v6 address we must use a v2 cert
 	cs := f.pki.getCertState()
-	v := cs.defaultVersion
+	v := cs.initiatingVersion
 	for _, a := range hh.hostinfo.vpnAddrs {
 		if a.Is6() {
 			v = cert.Version2
@@ -69,6 +70,15 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 		},
 	}
 
+	if f.multiPort.Tx || f.multiPort.Rx {
+		hs.Details.InitiatorMultiPort = &MultiPortDetails{
+			RxSupported: f.multiPort.Rx,
+			TxSupported: f.multiPort.Tx,
+			BasePort:    uint32(f.multiPort.TxBasePort),
+			TotalPorts:  uint32(f.multiPort.TxPorts),
+		}
+	}
+
 	hsBytes, err := hs.Marshal()
 	if err != nil {
 		f.l.WithError(err).WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
@@ -101,7 +111,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 	if crt == nil {
 		f.l.WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).
-			WithField("certVersion", cs.defaultVersion).
+			WithField("certVersion", cs.initiatingVersion).
 			Error("Unable to handshake with host because no certificate is available")
 	}
 
@@ -192,8 +202,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 
 	for _, network := range remoteCert.Certificate.Networks() {
 		vpnAddr := network.Addr()
-		_, found := f.myVpnAddrsTable.Lookup(vpnAddr)
-		if found {
+		if f.myVpnAddrsTable.Contains(vpnAddr) {
 			f.l.WithField("vpnAddr", vpnAddr).WithField("udpAddr", addr).
 				WithField("certName", certName).
 				WithField("certVersion", certVersion).
@@ -204,7 +213,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 		}
 
 		// vpnAddrs outside our vpn networks are of no use to us, filter them out
-		if _, ok := f.myVpnNetworksTable.Lookup(vpnAddr); !ok {
+		if !f.myVpnNetworksTable.Contains(vpnAddr) {
 			continue
 		}
 
@@ -242,17 +251,37 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 		return
 	}
 
+	var multiportTx, multiportRx bool
+	if f.multiPort.Rx || f.multiPort.Tx {
+		if hs.Details.InitiatorMultiPort != nil {
+			multiportTx = hs.Details.InitiatorMultiPort.RxSupported && f.multiPort.Tx
+			multiportRx = hs.Details.InitiatorMultiPort.TxSupported && f.multiPort.Rx
+		}
+
+		hs.Details.ResponderMultiPort = &MultiPortDetails{
+			TxSupported: f.multiPort.Tx,
+			RxSupported: f.multiPort.Rx,
+			BasePort:    uint32(f.multiPort.TxBasePort),
+			TotalPorts:  uint32(f.multiPort.TxPorts),
+		}
+	}
+	if hs.Details.InitiatorMultiPort != nil && hs.Details.InitiatorMultiPort.BasePort != uint32(addr.Port()) {
+		// The other side sent us a handshake from a different port, make sure
+		// we send responses back to the BasePort
+		addr = netip.AddrPortFrom(addr.Addr(), uint16(hs.Details.InitiatorMultiPort.BasePort))
+	}
+
 	hostinfo := &HostInfo{
-		syncRWMutex:       newSyncRWMutex("hostinfo"),
 		ConnectionState:   ci,
 		localIndexId:      myIndex,
 		remoteIndexId:     hs.Details.InitiatorIndex,
 		vpnAddrs:          vpnAddrs,
 		HandshakePacket:   make(map[uint8][]byte, 0),
 		lastHandshakeTime: hs.Details.Time,
+		multiportTx:       multiportTx,
+		multiportRx:       multiportRx,
 		relayState: RelayState{
-			syncRWMutex:    newSyncRWMutex("relay-state"),
-			relays:         map[netip.Addr]struct{}{},
+			relays:         nil,
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
@@ -265,6 +294,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 		WithField("issuer", issuer).
 		WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
 		WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+		WithField("multiportTx", multiportTx).WithField("multiportRx", multiportRx).
 		Info("Handshake message received")
 
 	hs.Details.ResponderIndex = myIndex
@@ -341,6 +371,10 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 	if err != nil {
 		switch err {
 		case ErrAlreadySeen:
+			if hostinfo.multiportRx {
+				// The other host is sending to us with multiport, so only grab the IP
+				addr = netip.AddrPortFrom(addr.Addr(), hostinfo.remote.Port())
+			}
 			// Update remote if preferred
 			if existing.SetRemoteIfPreferred(f.hostMap, addr) {
 				// Send a test packet to ensure the other side has also switched to
@@ -351,7 +385,14 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 			msg = existing.HandshakePacket[2]
 			f.messageMetrics.Tx(header.Handshake, header.MessageSubType(msg[1]), 1)
 			if addr.IsValid() {
-				err := f.outside.WriteTo(msg, addr)
+				if multiportTx {
+					// TODO remove alloc here
+					raw := make([]byte, len(msg)+udp.RawOverhead)
+					copy(raw[udp.RawOverhead:], msg)
+					err = f.udpRaw.WriteTo(raw, udp.RandomSendPort.UDPSendPort(f.multiPort.TxPorts), addr)
+				} else {
+					err = f.outside.WriteTo(msg, addr)
+				}
 				if err != nil {
 					f.l.WithField("vpnAddrs", existing.vpnAddrs).WithField("udpAddr", addr).
 						WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
@@ -420,7 +461,14 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 	// Do the send
 	f.messageMetrics.Tx(header.Handshake, header.MessageSubType(msg[1]), 1)
 	if addr.IsValid() {
-		err = f.outside.WriteTo(msg, addr)
+		if multiportTx {
+			// TODO remove alloc here
+			raw := make([]byte, len(msg)+udp.RawOverhead)
+			copy(raw[udp.RawOverhead:], msg)
+			err = f.udpRaw.WriteTo(raw, udp.RandomSendPort.UDPSendPort(f.multiPort.TxPorts), addr)
+		} else {
+			err = f.outside.WriteTo(msg, addr)
+		}
 		if err != nil {
 			f.l.WithField("vpnAddrs", vpnAddrs).WithField("udpAddr", addr).
 				WithField("certName", certName).
@@ -460,7 +508,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 			Info("Handshake message sent")
 	}
 
-	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
+	f.connectionManager.AddTrafficWatch(hostinfo)
 
 	hostinfo.remotes.ResetBlockedRemotes()
 
@@ -516,6 +564,20 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 		return true
 	}
 
+	if (f.multiPort.Tx || f.multiPort.Rx) && hs.Details.ResponderMultiPort != nil {
+		hostinfo.multiportTx = hs.Details.ResponderMultiPort.RxSupported && f.multiPort.Tx
+		hostinfo.multiportRx = hs.Details.ResponderMultiPort.TxSupported && f.multiPort.Rx
+	}
+
+	if hs.Details.ResponderMultiPort != nil && hs.Details.ResponderMultiPort.BasePort != uint32(addr.Port()) {
+		// The other side sent us a handshake from a different port, make sure
+		// we send responses back to the BasePort
+		addr = netip.AddrPortFrom(
+			addr.Addr(),
+			uint16(hs.Details.ResponderMultiPort.BasePort),
+		)
+	}
+
 	rc, err := cert.Recombine(cert.Version(hs.Details.CertVersion), hs.Details.Cert, ci.H.PeerStatic(), ci.Curve())
 	if err != nil {
 		f.l.WithError(err).WithField("udpAddr", addr).
@@ -581,7 +643,7 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 	for _, network := range vpnNetworks {
 		// vpnAddrs outside our vpn networks are of no use to us, filter them out
 		vpnAddr := network.Addr()
-		if _, ok := f.myVpnNetworksTable.Lookup(vpnAddr); !ok {
+		if !f.myVpnNetworksTable.Contains(vpnAddr) {
 			continue
 		}
 
@@ -647,6 +709,7 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 		WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 		WithField("durationNs", duration).
 		WithField("sentCachedPackets", len(hh.packetStore)).
+		WithField("multiportTx", hostinfo.multiportTx).WithField("multiportRx", hostinfo.multiportRx).
 		Info("Handshake message received")
 
 	// Build up the radix for the firewall if we have subnets in the cert
@@ -655,7 +718,7 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 
 	// Complete our handshake and update metrics, this will replace any existing tunnels for the vpnAddrs here
 	f.handshakeManager.Complete(hostinfo, f)
-	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
+	f.connectionManager.AddTrafficWatch(hostinfo)
 
 	if f.l.Level >= logrus.DebugLevel {
 		hostinfo.logger(f.l).Debugf("Sending %d stored packets", len(hh.packetStore))

handshake_manager.go

@@ -8,6 +8,7 @@ import (
 	"errors"
 	"net/netip"
 	"slices"
+	"sync"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
@@ -44,7 +45,7 @@ type HandshakeConfig struct {
 
 type HandshakeManager struct {
 	// Mutex for interacting with the vpnIps and indexes maps
-	syncRWMutex
+	sync.RWMutex
 
 	vpnIps  map[netip.Addr]*HandshakeHostInfo
 	indexes map[uint32]*HandshakeHostInfo
@@ -60,12 +61,15 @@ type HandshakeManager struct {
 	f                      *Interface
 	l                      *logrus.Logger
 
+	multiPort MultiPortConfig
+	udpRaw    *udp.RawConn
+
 	// can be used to trigger outbound handshake for the given vpnIp
 	trigger chan netip.Addr
 }
 
 type HandshakeHostInfo struct {
-	syncMutex
+	sync.Mutex
 
 	startTime   time.Time        // Time that we first started trying with this handshake
 	ready       bool             // Is the handshake ready
@@ -103,7 +107,6 @@ func (hh *HandshakeHostInfo) cachePacket(l *logrus.Logger, t header.MessageType,
 
 func NewHandshakeManager(l *logrus.Logger, mainHostMap *HostMap, lightHouse *LightHouse, outside udp.Conn, config HandshakeConfig) *HandshakeManager {
 	return &HandshakeManager{
-		syncRWMutex:            newSyncRWMutex("handshake-manager"),
 		vpnIps:                 map[netip.Addr]*HandshakeHostInfo{},
 		indexes:                map[uint32]*HandshakeHostInfo{},
 		mainHostMap:            mainHostMap,
@@ -111,7 +114,7 @@ func NewHandshakeManager(l *logrus.Logger, mainHostMap *HostMap, lightHouse *Lig
 		outside:                outside,
 		config:                 config,
 		trigger:                make(chan netip.Addr, config.triggerBuffer),
-		OutboundHandshakeTimer: NewLockingTimerWheel[netip.Addr]("handshake-manager-timer", config.tryInterval, hsTimeout(config.retries, config.tryInterval)),
+		OutboundHandshakeTimer: NewLockingTimerWheel[netip.Addr](config.tryInterval, hsTimeout(config.retries, config.tryInterval)),
 		messageMetrics:         config.messageMetrics,
 		metricInitiated:        metrics.GetOrRegisterCounter("handshake_manager.initiated", nil),
 		metricTimedOut:         metrics.GetOrRegisterCounter("handshake_manager.timed_out", nil),
@@ -236,6 +239,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 
 	// Send the handshake to all known ips, stage 2 takes care of assigning the hostinfo.remote based on the first to reply
 	var sentTo []netip.AddrPort
+	var sentMultiport bool
 	hostinfo.remotes.ForEach(hm.mainHostMap.GetPreferredRanges(), func(addr netip.AddrPort, _ bool) {
 		hm.messageMetrics.Tx(header.Handshake, header.MessageSubType(hostinfo.HandshakePacket[0][1]), 1)
 		err := hm.outside.WriteTo(hostinfo.HandshakePacket[0], addr)
@@ -248,6 +252,27 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 		} else {
 			sentTo = append(sentTo, addr)
 		}
+
+		// Attempt a multiport handshake if we are past the TxHandshakeDelay attempts
+		if hm.multiPort.TxHandshake && hm.udpRaw != nil && hh.counter >= hm.multiPort.TxHandshakeDelay {
+			sentMultiport = true
+			// We need to re-allocate with 8 bytes at the start of SOCK_RAW
+			raw := hostinfo.HandshakePacket[0x80]
+			if raw == nil {
+				raw = make([]byte, len(hostinfo.HandshakePacket[0])+udp.RawOverhead)
+				copy(raw[udp.RawOverhead:], hostinfo.HandshakePacket[0])
+				hostinfo.HandshakePacket[0x80] = raw
+			}
+
+			hm.messageMetrics.Tx(header.Handshake, header.MessageSubType(hostinfo.HandshakePacket[0][1]), 1)
+			err = hm.udpRaw.WriteTo(raw, udp.RandomSendPort.UDPSendPort(hm.multiPort.TxPorts), addr)
+			if err != nil {
+				hostinfo.logger(hm.l).WithField("udpAddr", addr).
+					WithField("initiatorIndex", hostinfo.localIndexId).
+					WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+					WithError(err).Error("Failed to send handshake message")
+			}
+		}
 	})
 
 	// Don't be too noisy or confusing if we fail to send a handshake - if we don't get through we'll eventually log a timeout,
@@ -256,6 +281,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 		hostinfo.logger(hm.l).WithField("udpAddrs", sentTo).
 			WithField("initiatorIndex", hostinfo.localIndexId).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			WithField("multiportHandshake", sentMultiport).
 			Info("Handshake message sent")
 	} else if hm.l.Level >= logrus.DebugLevel {
 		hostinfo.logger(hm.l).WithField("udpAddrs", sentTo).
@@ -274,8 +300,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 			}
 
 			// Don't relay through the host I'm trying to connect to
-			_, found := hm.f.myVpnAddrsTable.Lookup(relay)
-			if found {
+			if hm.f.myVpnAddrsTable.Contains(relay) {
 				continue
 			}
 
@@ -448,19 +473,16 @@ func (hm *HandshakeManager) StartHandshake(vpnAddr netip.Addr, cacheCb func(*Han
 	}
 
 	hostinfo := &HostInfo{
-		syncRWMutex:     newSyncRWMutex("hostinfo"),
 		vpnAddrs:        []netip.Addr{vpnAddr},
 		HandshakePacket: make(map[uint8][]byte, 0),
 		relayState: RelayState{
-			syncRWMutex:    newSyncRWMutex("relay-state"),
-			relays:         map[netip.Addr]struct{}{},
+			relays:         nil,
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
 	}
 
 	hh := &HandshakeHostInfo{
-		syncMutex: newSyncMutex("handshake-hostinfo"),
 		hostinfo:  hostinfo,
 		startTime: time.Now(),
 	}

handshake_manager_test.go

@@ -24,10 +24,10 @@ func Test_NewHandshakeManagerVpnIp(t *testing.T) {
 	lh := newTestLighthouse()
 
 	cs := &CertState{
-		defaultVersion:   cert.Version1,
-		privateKey:       []byte{},
-		v1Cert:           &dummyCert{version: cert.Version1},
-		v1HandshakeBytes: []byte{},
+		initiatingVersion: cert.Version1,
+		privateKey:        []byte{},
+		v1Cert:            &dummyCert{version: cert.Version1},
+		v1HandshakeBytes:  []byte{},
 	}
 
 	blah := NewHandshakeManager(l, mainHM, lh, &udp.NoopConn{}, defaultHandshakeConfig)
@@ -98,5 +98,5 @@ func (mw *mockEncWriter) GetHostInfo(_ netip.Addr) *HostInfo {
 }
 
 func (mw *mockEncWriter) GetCertState() *CertState {
-	return &CertState{defaultVersion: cert.Version2}
+	return &CertState{initiatingVersion: cert.Version2}
 }

hostmap.go

@@ -4,6 +4,8 @@ import (
 	"errors"
 	"net"
 	"net/netip"
+	"slices"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -52,7 +54,7 @@ type Relay struct {
 }
 
 type HostMap struct {
-	syncRWMutex     //Because we concurrently read and write to our maps
+	sync.RWMutex    //Because we concurrently read and write to our maps
 	Indexes         map[uint32]*HostInfo
 	Relays          map[uint32]*HostInfo // Maps a Relay IDX to a Relay HostInfo object
 	RemoteIndexes   map[uint32]*HostInfo
@@ -65,9 +67,9 @@ type HostMap struct {
 // struct, make a copy of an existing value, edit the fileds in the copy, and
 // then store a pointer to the new copy in both realyForBy* maps.
 type RelayState struct {
-	syncRWMutex
+	sync.RWMutex
 
-	relays map[netip.Addr]struct{} // Set of vpnAddr's of Hosts to use as relays to access this peer
+	relays []netip.Addr // Ordered set of VpnAddrs of Hosts to use as relays to access this peer
 	// For data race avoidance, the contents of a *Relay are treated immutably. To update a *Relay, copy the existing data,
 	// modify what needs to be updated, and store the new modified copy in the relayForByIp and relayForByIdx maps (with
 	// the RelayState Lock held)
@@ -78,7 +80,12 @@ type RelayState struct {
 func (rs *RelayState) DeleteRelay(ip netip.Addr) {
 	rs.Lock()
 	defer rs.Unlock()
-	delete(rs.relays, ip)
+	for idx, val := range rs.relays {
+		if val == ip {
+			rs.relays = append(rs.relays[:idx], rs.relays[idx+1:]...)
+			return
+		}
+	}
 }
 
 func (rs *RelayState) UpdateRelayForByIpState(vpnIp netip.Addr, state int) {
@@ -123,16 +130,16 @@ func (rs *RelayState) GetRelayForByAddr(addr netip.Addr) (*Relay, bool) {
 func (rs *RelayState) InsertRelayTo(ip netip.Addr) {
 	rs.Lock()
 	defer rs.Unlock()
-	rs.relays[ip] = struct{}{}
+	if !slices.Contains(rs.relays, ip) {
+		rs.relays = append(rs.relays, ip)
+	}
 }
 
 func (rs *RelayState) CopyRelayIps() []netip.Addr {
+	ret := make([]netip.Addr, len(rs.relays))
 	rs.RLock()
 	defer rs.RUnlock()
-	ret := make([]netip.Addr, 0, len(rs.relays))
-	for ip := range rs.relays {
-		ret = append(ret, ip)
-	}
+	copy(ret, rs.relays)
 	return ret
 }
 
@@ -208,7 +215,6 @@ func (rs *RelayState) InsertRelay(ip netip.Addr, idx uint32, r *Relay) {
 }
 
 type HostInfo struct {
-	syncRWMutex
 	remote          netip.AddrPort
 	remotes         *RemoteList
 	promoteCounter  atomic.Uint32
@@ -223,9 +229,15 @@ type HostInfo struct {
 	recvError atomic.Uint32
 
 	// networks are both all vpn and unsafe networks assigned to this host
-	networks   *bart.Table[struct{}]
+	networks   *bart.Lite
 	relayState RelayState
 
+	// If true, we should send to this remote using multiport
+	multiportTx bool
+
+	// If true, we will receive from this remote using multiport
+	multiportRx bool
+
 	// HandshakePacket records the packets used to create this hostinfo
 	// We need these to avoid replayed handshake packets creating new hostinfos which causes churn
 	HandshakePacket map[uint8][]byte
@@ -250,6 +262,14 @@ type HostInfo struct {
 	// Used to track other hostinfos for this vpn ip since only 1 can be primary
 	// Synchronised via hostmap lock and not the hostinfo lock.
 	next, prev *HostInfo
+
+	//TODO: in, out, and others might benefit from being an atomic.Int32. We could collapse connectionManager pendingDeletion, relayUsed, and in/out into this 1 thing
+	in, out, pendingDeletion atomic.Bool
+
+	// lastUsed tracks the last time ConnectionManager checked the tunnel and it was in use.
+	// This value will be behind against actual tunnel utilization in the hot path.
+	// This should only be used by the ConnectionManagers ticker routine.
+	lastUsed time.Time
 }
 
 type ViaSender struct {
@@ -288,7 +308,6 @@ func NewHostMapFromConfig(l *logrus.Logger, c *config.C) *HostMap {
 
 func newHostMap(l *logrus.Logger) *HostMap {
 	return &HostMap{
-		syncRWMutex:   newSyncRWMutex("hostmap"),
 		Indexes:       map[uint32]*HostInfo{},
 		Relays:        map[uint32]*HostInfo{},
 		RemoteIndexes: map[uint32]*HostInfo{},
@@ -733,13 +752,13 @@ func (i *HostInfo) buildNetworks(networks, unsafeNetworks []netip.Prefix) {
 		return
 	}
 
-	i.networks = new(bart.Table[struct{}])
+	i.networks = new(bart.Lite)
 	for _, network := range networks {
-		i.networks.Insert(network, struct{}{})
+		i.networks.Insert(network)
 	}
 
 	for _, network := range unsafeNetworks {
-		i.networks.Insert(network, struct{}{})
+		i.networks.Insert(network)
 	}
 }
 

hostmap_test.go

@@ -7,6 +7,7 @@ import (
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestHostMap_MakePrimary(t *testing.T) {
@@ -215,3 +216,31 @@ func TestHostMap_reload(t *testing.T) {
 	c.ReloadConfigString("preferred_ranges: [1.1.1.1/32]")
 	assert.Equal(t, []string{"1.1.1.1/32"}, toS(hm.GetPreferredRanges()))
 }
+
+func TestHostMap_RelayState(t *testing.T) {
+	h1 := &HostInfo{vpnAddrs: []netip.Addr{netip.MustParseAddr("0.0.0.1")}, localIndexId: 1}
+	a1 := netip.MustParseAddr("::1")
+	a2 := netip.MustParseAddr("2001::1")
+
+	h1.relayState.InsertRelayTo(a1)
+	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
+	h1.relayState.InsertRelayTo(a2)
+	assert.Equal(t, []netip.Addr{a1, a2}, h1.relayState.relays)
+	// Ensure that the first relay added is the first one returned in the copy
+	currentRelays := h1.relayState.CopyRelayIps()
+	require.Len(t, currentRelays, 2)
+	assert.Equal(t, a1, currentRelays[0])
+
+	// Deleting the last one in the list works ok
+	h1.relayState.DeleteRelay(a2)
+	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
+
+	// Deleting an element not in the list works ok
+	h1.relayState.DeleteRelay(a2)
+	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
+
+	// Deleting the only element in the list works ok
+	h1.relayState.DeleteRelay(a1)
+	assert.Equal(t, []netip.Addr{}, h1.relayState.relays)
+
+}

inside.go

@@ -9,6 +9,7 @@ import (
 	"github.com/slackhq/nebula/iputil"
 	"github.com/slackhq/nebula/noiseutil"
 	"github.com/slackhq/nebula/routing"
+	"github.com/slackhq/nebula/udp"
 )
 
 func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet, nb, out []byte, q int, localCache firewall.ConntrackCache) {
@@ -22,14 +23,12 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 
 	// Ignore local broadcast packets
 	if f.dropLocalBroadcast {
-		_, found := f.myBroadcastAddrsTable.Lookup(fwPacket.RemoteAddr)
-		if found {
+		if f.myBroadcastAddrsTable.Contains(fwPacket.RemoteAddr) {
 			return
 		}
 	}
 
-	_, found := f.myVpnAddrsTable.Lookup(fwPacket.RemoteAddr)
-	if found {
+	if f.myVpnAddrsTable.Contains(fwPacket.RemoteAddr) {
 		// Immediately forward packets from self to self.
 		// This should only happen on Darwin-based and FreeBSD hosts, which
 		// routes packets from the Nebula addr to the Nebula addr through the Nebula
@@ -70,7 +69,7 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 
 	dropReason := f.firewall.Drop(*fwPacket, false, hostinfo, f.pki.GetCAPool(), localCache)
 	if dropReason == nil {
-		f.sendNoMetrics(header.Message, 0, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, packet, nb, out, q)
+		f.sendNoMetrics(header.Message, 0, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, packet, nb, out, q, fwPacket)
 
 	} else {
 		f.rejectInside(packet, out, q)
@@ -119,7 +118,7 @@ func (f *Interface) rejectOutside(packet []byte, ci *ConnectionState, hostinfo *
 		return
 	}
 
-	f.sendNoMetrics(header.Message, 0, ci, hostinfo, netip.AddrPort{}, out, nb, packet, q)
+	f.sendNoMetrics(header.Message, 0, ci, hostinfo, netip.AddrPort{}, out, nb, packet, q, nil)
 }
 
 // Handshake will attempt to initiate a tunnel with the provided vpn address if it is within our vpn networks. This is a no-op if the tunnel is already established or being established
@@ -130,8 +129,7 @@ func (f *Interface) Handshake(vpnAddr netip.Addr) {
 // getOrHandshakeNoRouting returns nil if the vpnAddr is not routable.
 // If the 2nd return var is false then the hostinfo is not ready to be used in a tunnel
 func (f *Interface) getOrHandshakeNoRouting(vpnAddr netip.Addr, cacheCallback func(*HandshakeHostInfo)) (*HostInfo, bool) {
-	_, found := f.myVpnNetworksTable.Lookup(vpnAddr)
-	if found {
+	if f.myVpnNetworksTable.Contains(vpnAddr) {
 		return f.handshakeManager.GetOrHandshake(vpnAddr, cacheCallback)
 	}
 
@@ -231,7 +229,7 @@ func (f *Interface) sendMessageNow(t header.MessageType, st header.MessageSubTyp
 		return
 	}
 
-	f.sendNoMetrics(header.Message, st, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, p, nb, out, 0)
+	f.sendNoMetrics(header.Message, st, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, p, nb, out, 0, nil)
 }
 
 // SendMessageToVpnAddr handles real addr:port lookup and sends to the current best known address for vpnAddr
@@ -261,12 +259,12 @@ func (f *Interface) SendMessageToHostInfo(t header.MessageType, st header.Messag
 
 func (f *Interface) send(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, p, nb, out []byte) {
 	f.messageMetrics.Tx(t, st, 1)
-	f.sendNoMetrics(t, st, ci, hostinfo, netip.AddrPort{}, p, nb, out, 0)
+	f.sendNoMetrics(t, st, ci, hostinfo, netip.AddrPort{}, p, nb, out, 0, nil)
 }
 
 func (f *Interface) sendTo(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote netip.AddrPort, p, nb, out []byte) {
 	f.messageMetrics.Tx(t, st, 1)
-	f.sendNoMetrics(t, st, ci, hostinfo, remote, p, nb, out, 0)
+	f.sendNoMetrics(t, st, ci, hostinfo, remote, p, nb, out, 0, nil)
 }
 
 // SendVia sends a payload through a Relay tunnel. No authentication or encryption is done
@@ -291,7 +289,7 @@ func (f *Interface) SendVia(via *HostInfo,
 	c := via.ConnectionState.messageCounter.Add(1)
 
 	out = header.Encode(out, header.Version, header.Message, header.MessageRelay, relay.RemoteIndex, c)
-	f.connectionManager.Out(via.localIndexId)
+	f.connectionManager.Out(via)
 
 	// Authenticate the header and payload, but do not encrypt for this message type.
 	// The payload consists of the inner, unencrypted Nebula header, as well as the end-to-end encrypted payload.
@@ -334,10 +332,27 @@ func (f *Interface) SendVia(via *HostInfo,
 	f.connectionManager.RelayUsed(relay.LocalIndex)
 }
 
-func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote netip.AddrPort, p, nb, out []byte, q int) {
+func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote netip.AddrPort, p, nb, out []byte, q int, udpPortGetter udp.SendPortGetter) {
 	if ci.eKey == nil {
 		return
 	}
+
+	multiport := f.multiPort.Tx && hostinfo.multiportTx
+	rawOut := out
+	if multiport {
+		if len(out) < udp.RawOverhead {
+			// NOTE: This is because some spots in the code send us `out[:0]`, so
+			// we need to expand the slice back out to get our 8 bytes back.
+			out = out[:udp.RawOverhead]
+		}
+		// Preserve bytes needed for the raw socket
+		out = out[udp.RawOverhead:]
+
+		if udpPortGetter == nil {
+			udpPortGetter = udp.RandomSendPort
+		}
+	}
+
 	useRelay := !remote.IsValid() && !hostinfo.remote.IsValid()
 	fullOut := out
 
@@ -359,7 +374,7 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 
 	//l.WithField("trace", string(debug.Stack())).Error("out Header ", &Header{Version, t, st, 0, hostinfo.remoteIndexId, c}, p)
 	out = header.Encode(out, header.Version, t, st, hostinfo.remoteIndexId, c)
-	f.connectionManager.Out(hostinfo.localIndexId)
+	f.connectionManager.Out(hostinfo)
 
 	// Query our LH if we haven't since the last time we've been rebound, this will cause the remote to punch against
 	// all our addrs and enable a faster roaming.
@@ -387,13 +402,25 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 	}
 
 	if remote.IsValid() {
-		err = f.writers[q].WriteTo(out, remote)
+		if multiport {
+			rawOut = rawOut[:len(out)+udp.RawOverhead]
+			port := udpPortGetter.UDPSendPort(f.multiPort.TxPorts)
+			err = f.udpRaw.WriteTo(rawOut, port, remote)
+		} else {
+			err = f.writers[q].WriteTo(out, remote)
+		}
 		if err != nil {
 			hostinfo.logger(f.l).WithError(err).
 				WithField("udpAddr", remote).Error("Failed to write outgoing packet")
 		}
 	} else if hostinfo.remote.IsValid() {
-		err = f.writers[q].WriteTo(out, hostinfo.remote)
+		if multiport {
+			rawOut = rawOut[:len(out)+udp.RawOverhead]
+			port := udpPortGetter.UDPSendPort(f.multiPort.TxPorts)
+			err = f.udpRaw.WriteTo(rawOut, port, hostinfo.remote)
+		} else {
+			err = f.writers[q].WriteTo(out, hostinfo.remote)
+		}
 		if err != nil {
 			hostinfo.logger(f.l).WithError(err).
 				WithField("udpAddr", remote).Error("Failed to write outgoing packet")

interface.go

@@ -24,23 +24,23 @@ import (
 const mtu = 9001
 
 type InterfaceConfig struct {
-	HostMap                 *HostMap
-	Outside                 udp.Conn
-	Inside                  overlay.Device
-	pki                     *PKI
-	Firewall                *Firewall
-	ServeDns                bool
-	HandshakeManager        *HandshakeManager
-	lightHouse              *LightHouse
-	checkInterval           time.Duration
-	pendingDeletionInterval time.Duration
-	DropLocalBroadcast      bool
-	DropMulticast           bool
-	routines                int
-	MessageMetrics          *MessageMetrics
-	version                 string
-	relayManager            *relayManager
-	punchy                  *Punchy
+	HostMap            *HostMap
+	Outside            udp.Conn
+	Inside             overlay.Device
+	pki                *PKI
+	Cipher             string
+	Firewall           *Firewall
+	ServeDns           bool
+	HandshakeManager   *HandshakeManager
+	lightHouse         *LightHouse
+	connectionManager  *connectionManager
+	DropLocalBroadcast bool
+	DropMulticast      bool
+	routines           int
+	MessageMetrics     *MessageMetrics
+	version            string
+	relayManager       *relayManager
+	punchy             *Punchy
 
 	tryPromoteEvery uint32
 	reQueryEvery    uint32
@@ -61,11 +61,11 @@ type Interface struct {
 	serveDns              bool
 	createTime            time.Time
 	lightHouse            *LightHouse
-	myBroadcastAddrsTable *bart.Table[struct{}]
-	myVpnAddrs            []netip.Addr          // A list of addresses assigned to us via our certificate
-	myVpnAddrsTable       *bart.Table[struct{}] // A table of addresses assigned to us via our certificate
-	myVpnNetworks         []netip.Prefix        // A list of networks assigned to us via our certificate
-	myVpnNetworksTable    *bart.Table[struct{}] // A table of networks assigned to us via our certificate
+	myBroadcastAddrsTable *bart.Lite
+	myVpnAddrs            []netip.Addr // A list of addresses assigned to us via our certificate
+	myVpnAddrsTable       *bart.Lite
+	myVpnNetworks         []netip.Prefix // A list of networks assigned to us via our certificate
+	myVpnNetworksTable    *bart.Lite
 	dropLocalBroadcast    bool
 	dropMulticast         bool
 	routines              int
@@ -87,6 +87,9 @@ type Interface struct {
 
 	writers []udp.Conn
 	readers []io.ReadWriteCloser
+	udpRaw  *udp.RawConn
+
+	multiPort MultiPortConfig
 
 	metricHandshakes    metrics.Histogram
 	messageMetrics      *MessageMetrics
@@ -95,6 +98,15 @@ type Interface struct {
 	l *logrus.Logger
 }
 
+type MultiPortConfig struct {
+	Tx               bool
+	Rx               bool
+	TxBasePort       uint16
+	TxPorts          int
+	TxHandshake      bool
+	TxHandshakeDelay int64
+}
+
 type EncWriter interface {
 	SendVia(via *HostInfo,
 		relay *Relay,
@@ -157,6 +169,9 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 	if c.Firewall == nil {
 		return nil, errors.New("no firewall rules")
 	}
+	if c.connectionManager == nil {
+		return nil, errors.New("no connection manager")
+	}
 
 	cs := c.pki.getCertState()
 	ifce := &Interface{
@@ -181,7 +196,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 		myVpnAddrsTable:       cs.myVpnAddrsTable,
 		myBroadcastAddrsTable: cs.myVpnBroadcastAddrsTable,
 		relayManager:          c.relayManager,
-
+		connectionManager:     c.connectionManager,
 		conntrackCacheTimeout: c.ConntrackCacheTimeout,
 
 		metricHandshakes: metrics.GetOrRegisterHistogram("handshakes", nil, metrics.NewExpDecaySample(1028, 0.015)),
@@ -198,7 +213,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 	ifce.reQueryEvery.Store(c.reQueryEvery)
 	ifce.reQueryWait.Store(int64(c.reQueryWait))
 
-	ifce.connectionManager = newConnectionManager(ctx, c.l, ifce, c.checkInterval, c.pendingDeletionInterval, c.punchy)
+	ifce.connectionManager.intf = ifce
 
 	return ifce, nil
 }
@@ -221,6 +236,8 @@ func (f *Interface) activate() {
 
 	metrics.GetOrRegisterGauge("routines", nil).Update(int64(f.routines))
 
+	metrics.GetOrRegisterGauge("multiport.tx_ports", nil).Update(int64(f.multiPort.TxPorts))
+
 	// Prepare n tun queues
 	var reader io.ReadWriteCloser = f.inside
 	for i := 0; i < f.routines; i++ {
@@ -409,8 +426,10 @@ func (f *Interface) emitStats(ctx context.Context, i time.Duration) {
 
 	udpStats := udp.NewUDPStatsEmitter(f.writers)
 
+	var rawStats func()
+
 	certExpirationGauge := metrics.GetOrRegisterGauge("certificate.ttl_seconds", nil)
-	certDefaultVersion := metrics.GetOrRegisterGauge("certificate.default_version", nil)
+	certInitiatingVersion := metrics.GetOrRegisterGauge("certificate.initiating_version", nil)
 	certMaxVersion := metrics.GetOrRegisterGauge("certificate.max_version", nil)
 
 	for {
@@ -425,7 +444,14 @@ func (f *Interface) emitStats(ctx context.Context, i time.Duration) {
 			certState := f.pki.getCertState()
 			defaultCrt := certState.GetDefaultCertificate()
 			certExpirationGauge.Update(int64(defaultCrt.NotAfter().Sub(time.Now()) / time.Second))
-			certDefaultVersion.Update(int64(defaultCrt.Version()))
+			certInitiatingVersion.Update(int64(defaultCrt.Version()))
+
+			if f.udpRaw != nil {
+				if rawStats == nil {
+					rawStats = udp.NewRawStatsEmitter(f.udpRaw)
+				}
+				rawStats()
+			}
 
 			// Report the max certificate version we are capable of using
 			if certState.v2Cert != nil {

lighthouse.go

@@ -9,6 +9,7 @@ import (
 	"net/netip"
 	"slices"
 	"strconv"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -26,12 +27,12 @@ var ErrHostNotKnown = errors.New("host not known")
 
 type LightHouse struct {
 	//TODO: We need a timer wheel to kick out vpnAddrs that haven't reported in a long time
-	syncRWMutex  //Because we concurrently read and write to our maps
+	sync.RWMutex //Because we concurrently read and write to our maps
 	ctx          context.Context
 	amLighthouse bool
 
 	myVpnNetworks      []netip.Prefix
-	myVpnNetworksTable *bart.Table[struct{}]
+	myVpnNetworksTable *bart.Lite
 	punchConn          udp.Conn
 	punchy             *Punchy
 
@@ -95,7 +96,6 @@ func NewLightHouseFromConfig(ctx context.Context, l *logrus.Logger, c *config.C,
 	}
 
 	h := LightHouse{
-		syncRWMutex:        newSyncRWMutex("lighthouse"),
 		ctx:                ctx,
 		amLighthouse:       amLighthouse,
 		myVpnNetworks:      cs.myVpnNetworks,
@@ -201,8 +201,7 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 
 			//TODO: we could technically insert all returned addrs instead of just the first one if a dns lookup was used
 			addr := addrs[0].Unmap()
-			_, found := lh.myVpnNetworksTable.Lookup(addr)
-			if found {
+			if lh.myVpnNetworksTable.Contains(addr) {
 				lh.l.WithField("addr", rawAddr).WithField("entry", i+1).
 					Warn("Ignoring lighthouse.advertise_addrs report because it is within the nebula network range")
 				continue
@@ -359,8 +358,7 @@ func (lh *LightHouse) parseLighthouses(c *config.C, lhMap map[netip.Addr]struct{
 			return util.NewContextualError("Unable to parse lighthouse host entry", m{"host": host, "entry": i + 1}, err)
 		}
 
-		_, found := lh.myVpnNetworksTable.Lookup(addr)
-		if !found {
+		if !lh.myVpnNetworksTable.Contains(addr) {
 			return util.NewContextualError("lighthouse host is not in our networks, invalid", m{"vpnAddr": addr, "networks": lh.myVpnNetworks}, nil)
 		}
 		lhMap[addr] = struct{}{}
@@ -431,8 +429,7 @@ func (lh *LightHouse) loadStaticMap(c *config.C, staticList map[netip.Addr]struc
 			return util.NewContextualError("Unable to parse static_host_map entry", m{"host": k, "entry": i + 1}, err)
 		}
 
-		_, found := lh.myVpnNetworksTable.Lookup(vpnAddr)
-		if !found {
+		if !lh.myVpnNetworksTable.Contains(vpnAddr) {
 			return util.NewContextualError("static_host_map key is not in our network, invalid", m{"vpnAddr": vpnAddr, "networks": lh.myVpnNetworks, "entry": i + 1}, nil)
 		}
 
@@ -475,7 +472,6 @@ func (lh *LightHouse) QueryServer(vpnAddr netip.Addr) {
 		return
 	}
 
-	chanDebugSend("lighthouse-query-chan")
 	lh.queryChan <- vpnAddr
 }
 
@@ -654,8 +650,7 @@ func (lh *LightHouse) shouldAdd(vpnAddr netip.Addr, to netip.Addr) bool {
 		return false
 	}
 
-	_, found := lh.myVpnNetworksTable.Lookup(to)
-	if found {
+	if lh.myVpnNetworksTable.Contains(to) {
 		return false
 	}
 
@@ -675,8 +670,7 @@ func (lh *LightHouse) unlockedShouldAddV4(vpnAddr netip.Addr, to *V4AddrPort) bo
 		return false
 	}
 
-	_, found := lh.myVpnNetworksTable.Lookup(udpAddr.Addr())
-	if found {
+	if lh.myVpnNetworksTable.Contains(udpAddr.Addr()) {
 		return false
 	}
 
@@ -696,8 +690,7 @@ func (lh *LightHouse) unlockedShouldAddV6(vpnAddr netip.Addr, to *V6AddrPort) bo
 		return false
 	}
 
-	_, found := lh.myVpnNetworksTable.Lookup(udpAddr.Addr())
-	if found {
+	if lh.myVpnNetworksTable.Contains(udpAddr.Addr()) {
 		return false
 	}
 
@@ -732,8 +725,6 @@ func (lh *LightHouse) startQueryWorker() {
 		nb := make([]byte, 12, 12)
 		out := make([]byte, mtu)
 
-		chanDebugRecv("lighthouse-query-chan")
-
 		for {
 			select {
 			case <-lh.ctx.Done():
@@ -766,7 +757,7 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 		if hi != nil {
 			v = hi.ConnectionState.myCert.Version()
 		} else {
-			v = lh.ifce.GetCertState().defaultVersion
+			v = lh.ifce.GetCertState().initiatingVersion
 		}
 
 		if v == cert.Version1 {
@@ -859,8 +850,7 @@ func (lh *LightHouse) SendUpdate() {
 
 	lal := lh.GetLocalAllowList()
 	for _, e := range localAddrs(lh.l, lal) {
-		_, found := lh.myVpnNetworksTable.Lookup(e)
-		if found {
+		if lh.myVpnNetworksTable.Contains(e) {
 			continue
 		}
 
@@ -886,7 +876,7 @@ func (lh *LightHouse) SendUpdate() {
 		if hi != nil {
 			v = hi.ConnectionState.myCert.Version()
 		} else {
-			v = lh.ifce.GetCertState().defaultVersion
+			v = lh.ifce.GetCertState().initiatingVersion
 		}
 		if v == cert.Version1 {
 			if v1Update == nil {
@@ -1117,7 +1107,7 @@ func (lhh *LightHouseHandler) sendHostPunchNotification(n *NebulaMeta, fromVpnAd
 		targetHI := lhh.lh.ifce.GetHostInfo(punchNotifDest)
 		var useVersion cert.Version
 		if targetHI == nil {
-			useVersion = lhh.lh.ifce.GetCertState().defaultVersion
+			useVersion = lhh.lh.ifce.GetCertState().initiatingVersion
 		} else {
 			crt := targetHI.GetCert().Certificate
 			useVersion = crt.Version()

lighthouse_test.go

@@ -31,8 +31,8 @@ func TestOldIPv4Only(t *testing.T) {
 func Test_lhStaticMapping(t *testing.T) {
 	l := test.NewLogger()
 	myVpnNet := netip.MustParsePrefix("10.128.0.1/16")
-	nt := new(bart.Table[struct{}])
-	nt.Insert(myVpnNet, struct{}{})
+	nt := new(bart.Lite)
+	nt.Insert(myVpnNet)
 	cs := &CertState{
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
@@ -56,8 +56,8 @@ func Test_lhStaticMapping(t *testing.T) {
 func TestReloadLighthouseInterval(t *testing.T) {
 	l := test.NewLogger()
 	myVpnNet := netip.MustParsePrefix("10.128.0.1/16")
-	nt := new(bart.Table[struct{}])
-	nt.Insert(myVpnNet, struct{}{})
+	nt := new(bart.Lite)
+	nt.Insert(myVpnNet)
 	cs := &CertState{
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
@@ -91,8 +91,8 @@ func TestReloadLighthouseInterval(t *testing.T) {
 func BenchmarkLighthouseHandleRequest(b *testing.B) {
 	l := test.NewLogger()
 	myVpnNet := netip.MustParsePrefix("10.128.0.1/0")
-	nt := new(bart.Table[struct{}])
-	nt.Insert(myVpnNet, struct{}{})
+	nt := new(bart.Lite)
+	nt.Insert(myVpnNet)
 	cs := &CertState{
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
@@ -196,8 +196,8 @@ func TestLighthouse_Memory(t *testing.T) {
 	c.Settings["listen"] = map[string]any{"port": 4242}
 
 	myVpnNet := netip.MustParsePrefix("10.128.0.1/24")
-	nt := new(bart.Table[struct{}])
-	nt.Insert(myVpnNet, struct{}{})
+	nt := new(bart.Lite)
+	nt.Insert(myVpnNet)
 	cs := &CertState{
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
@@ -281,8 +281,8 @@ func TestLighthouse_reload(t *testing.T) {
 	c.Settings["listen"] = map[string]any{"port": 4242}
 
 	myVpnNet := netip.MustParsePrefix("10.128.0.1/24")
-	nt := new(bart.Table[struct{}])
-	nt.Insert(myVpnNet, struct{}{})
+	nt := new(bart.Lite)
+	nt.Insert(myVpnNet)
 	cs := &CertState{
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
@@ -417,7 +417,7 @@ func (tw *testEncWriter) GetHostInfo(vpnIp netip.Addr) *HostInfo {
 }
 
 func (tw *testEncWriter) GetCertState() *CertState {
-	return &CertState{defaultVersion: tw.protocolVersion}
+	return &CertState{initiatingVersion: tw.protocolVersion}
 }
 
 // assertIp4InArray asserts every address in want is at the same position in have and that the lengths match

main.go

@@ -185,6 +185,7 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 
 	hostMap := NewHostMapFromConfig(l, c)
 	punchy := NewPunchyFromConfig(l, c)
+	connManager := newConnectionManagerFromConfig(l, c, hostMap, punchy)
 	lightHouse, err := NewLightHouseFromConfig(ctx, l, c, pki.getCertState(), udpConns[0], punchy)
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Failed to initialize lighthouse handler", err)
@@ -220,31 +221,26 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		}
 	}
 
-	checkInterval := c.GetInt("timers.connection_alive_interval", 5)
-	pendingDeletionInterval := c.GetInt("timers.pending_deletion_interval", 10)
-
 	ifConfig := &InterfaceConfig{
-		HostMap:                 hostMap,
-		Inside:                  tun,
-		Outside:                 udpConns[0],
-		pki:                     pki,
-		Firewall:                fw,
-		ServeDns:                serveDns,
-		HandshakeManager:        handshakeManager,
-		lightHouse:              lightHouse,
-		checkInterval:           time.Second * time.Duration(checkInterval),
-		pendingDeletionInterval: time.Second * time.Duration(pendingDeletionInterval),
-		tryPromoteEvery:         c.GetUint32("counters.try_promote", defaultPromoteEvery),
-		reQueryEvery:            c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
-		reQueryWait:             c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
-		DropLocalBroadcast:      c.GetBool("tun.drop_local_broadcast", false),
-		DropMulticast:           c.GetBool("tun.drop_multicast", false),
-		routines:                routines,
-		MessageMetrics:          messageMetrics,
-		version:                 buildVersion,
-		relayManager:            NewRelayManager(ctx, l, hostMap, c),
-		punchy:                  punchy,
-
+		HostMap:               hostMap,
+		Inside:                tun,
+		Outside:               udpConns[0],
+		pki:                   pki,
+		Firewall:              fw,
+		ServeDns:              serveDns,
+		HandshakeManager:      handshakeManager,
+		connectionManager:     connManager,
+		lightHouse:            lightHouse,
+		tryPromoteEvery:       c.GetUint32("counters.try_promote", defaultPromoteEvery),
+		reQueryEvery:          c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
+		reQueryWait:           c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
+		DropLocalBroadcast:    c.GetBool("tun.drop_local_broadcast", false),
+		DropMulticast:         c.GetBool("tun.drop_multicast", false),
+		routines:              routines,
+		MessageMetrics:        messageMetrics,
+		version:               buildVersion,
+		relayManager:          NewRelayManager(ctx, l, hostMap, c),
+		punchy:                punchy,
 		ConntrackCacheTimeout: conntrackCacheTimeout,
 		l:                     l,
 	}
@@ -259,6 +255,39 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		ifce.writers = udpConns
 		lightHouse.ifce = ifce
 
+		loadMultiPortConfig := func(c *config.C) {
+			ifce.multiPort.Rx = c.GetBool("tun.multiport.rx_enabled", false)
+
+			tx := c.GetBool("tun.multiport.tx_enabled", false)
+
+			if tx && ifce.udpRaw == nil {
+				ifce.udpRaw, err = udp.NewRawConn(l, c.GetString("listen.host", "0.0.0.0"), port, uint16(port))
+				if err != nil {
+					l.WithError(err).Error("Failed to get raw socket for tun.multiport.tx_enabled")
+					ifce.udpRaw = nil
+					tx = false
+				}
+			}
+
+			if tx {
+				ifce.multiPort.TxBasePort = uint16(port)
+				ifce.multiPort.TxPorts = c.GetInt("tun.multiport.tx_ports", 100)
+				ifce.multiPort.TxHandshake = c.GetBool("tun.multiport.tx_handshake", false)
+				ifce.multiPort.TxHandshakeDelay = int64(c.GetInt("tun.multiport.tx_handshake_delay", 2))
+				ifce.udpRaw.ReloadConfig(c)
+			}
+			ifce.multiPort.Tx = tx
+
+			// TODO: if we upstream this, make this cleaner
+			handshakeManager.udpRaw = ifce.udpRaw
+			handshakeManager.multiPort = ifce.multiPort
+
+			l.WithField("multiPort", ifce.multiPort).Info("Multiport configured")
+		}
+
+		loadMultiPortConfig(c)
+		c.RegisterReloadCallback(loadMultiPortConfig)
+
 		ifce.RegisterConfigChangeCallbacks(c)
 		ifce.reloadDisconnectInvalid(c)
 		ifce.reloadSendRecvError(c)
@@ -296,5 +325,6 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		statsStart,
 		dnsStart,
 		lightHouse.StartUpdateWorker,
+		connManager.Start,
 	}, nil
 }

mutex_debug.go

@@ -1,195 +0,0 @@
-//go:build mutex_debug
-// +build mutex_debug
-
-package nebula
-
-import (
-	"fmt"
-	"runtime"
-	"sync"
-
-	"github.com/clarkmcc/go-dag"
-	"github.com/timandy/routine"
-)
-
-type mutexKey = string
-
-// For each Key in this map, the Value is a list of lock types you can already have
-// when you want to grab that Key. This ensures that locks are always fetched
-// in the same order, to prevent deadlocks.
-var allowedConcurrentLocks = map[mutexKey][]mutexKey{
-	"connection-manager-in":         {"hostmap"},
-	"connection-manager-out":        {"connection-manager-in", "handshake-hostinfo", "handshake-manager"},
-	"connection-manager-relay-used": {"handshake-hostinfo"},
-	"connection-manager-timer":      {"connection-manager-out"},
-	// "connection-state-write":        {"hostmap"},
-	"firewall-conntrack":      {"handshake-hostinfo"},
-	"handshake-manager":       {"handshake-hostinfo", "hostmap"},
-	"handshake-manager-timer": {"handshake-manager"},
-	"hostmap":                 {"lighthouse-query-chan", "handshake-hostinfo"},
-	"lighthouse":              {"handshake-hostinfo"},
-	"relay-state":             {"connection-manager-relay-used", "hostmap"},
-	"remote-list":             {"lighthouse", "handshake-manager"},
-	"lighthouse-query-chan":   {"handshake-hostinfo"},
-}
-
-type mutexValue struct {
-	file string
-	line int
-}
-
-func (m mutexValue) String() string {
-	return fmt.Sprintf("%s:%d", m.file, m.line)
-}
-
-var threadLocal = routine.NewThreadLocalWithInitial[map[mutexKey]mutexValue](func() map[mutexKey]mutexValue { return map[mutexKey]mutexValue{} })
-
-var allowedDAG dag.AcyclicGraph
-
-// We build a directed acyclic graph to assert that the locks can only be
-// acquired in a determined order, If there are cycles in the DAG, then we
-// know that the locking order is not guaranteed.
-func init() {
-	for k, v := range allowedConcurrentLocks {
-		allowedDAG.Add(k)
-		for _, t := range v {
-			allowedDAG.Add(t)
-		}
-	}
-	for k, v := range allowedConcurrentLocks {
-		for _, t := range v {
-			allowedDAG.Connect(dag.BasicEdge(k, t))
-		}
-	}
-
-	if cycles := allowedDAG.Cycles(); len(cycles) > 0 {
-		panic(fmt.Errorf("Cycles found in allowedConcurrentLocks: %v", cycles))
-	}
-
-	// Rebuild allowedConcurrentLocks as a flattened list of all possibilities
-	for k := range allowedConcurrentLocks {
-		ancestors, err := allowedDAG.Ancestors(k)
-		if err != nil {
-			panic(err)
-		}
-
-		var allowed []mutexKey
-		for t := range ancestors {
-			allowed = append(allowed, t.(mutexKey))
-		}
-		allowedConcurrentLocks[k] = allowed
-	}
-}
-
-type syncRWMutex struct {
-	sync.RWMutex
-	mutexKey
-}
-
-type syncMutex struct {
-	sync.Mutex
-	mutexKey
-}
-
-func newSyncRWMutex(key mutexKey) syncRWMutex {
-	return syncRWMutex{
-		mutexKey: key,
-	}
-}
-
-func newSyncMutex(key mutexKey) syncMutex {
-	return syncMutex{
-		mutexKey: key,
-	}
-}
-
-func alertMutex(err error) {
-	panic(err)
-	// NOTE: you could switch to this log Line and remove the panic if you want
-	// to log all failures instead of panicking on the first one
-	//log.Print(err, string(debug.Stack()))
-}
-
-func checkMutex(state map[mutexKey]mutexValue, add mutexKey) {
-	if add == "" {
-		alertMutex(fmt.Errorf("mutex not initialized with mutexKey"))
-	}
-
-	allowedConcurrent := allowedConcurrentLocks[add]
-
-	for k, v := range state {
-		if add == k {
-			alertMutex(fmt.Errorf("re-entrant lock: %s. previous allocation: %s", add, v))
-		}
-
-		// TODO use slices.Contains, but requires go1.21
-		var found bool
-		for _, a := range allowedConcurrent {
-			if a == k {
-				found = true
-				break
-			}
-		}
-		if !found {
-			alertMutex(fmt.Errorf("grabbing %s lock and already have these locks: %s", add, state))
-		}
-	}
-}
-
-func chanDebugRecv(key mutexKey) {
-	m := threadLocal.Get()
-	checkMutex(m, key)
-	v := mutexValue{}
-	_, v.file, v.line, _ = runtime.Caller(1)
-	m[key] = v
-}
-
-func chanDebugSend(key mutexKey) {
-	m := threadLocal.Get()
-	checkMutex(m, key)
-}
-
-func (s *syncRWMutex) Lock() {
-	m := threadLocal.Get()
-	checkMutex(m, s.mutexKey)
-	v := mutexValue{}
-	_, v.file, v.line, _ = runtime.Caller(1)
-	m[s.mutexKey] = v
-	s.RWMutex.Lock()
-}
-
-func (s *syncRWMutex) Unlock() {
-	m := threadLocal.Get()
-	delete(m, s.mutexKey)
-	s.RWMutex.Unlock()
-}
-
-func (s *syncRWMutex) RLock() {
-	m := threadLocal.Get()
-	checkMutex(m, s.mutexKey)
-	v := mutexValue{}
-	_, v.file, v.line, _ = runtime.Caller(1)
-	m[s.mutexKey] = v
-	s.RWMutex.RLock()
-}
-
-func (s *syncRWMutex) RUnlock() {
-	m := threadLocal.Get()
-	delete(m, s.mutexKey)
-	s.RWMutex.RUnlock()
-}
-
-func (s *syncMutex) Lock() {
-	m := threadLocal.Get()
-	checkMutex(m, s.mutexKey)
-	v := mutexValue{}
-	_, v.file, v.line, _ = runtime.Caller(1)
-	m[s.mutexKey] = v
-	s.Mutex.Lock()
-}
-
-func (s *syncMutex) Unlock() {
-	m := threadLocal.Get()
-	delete(m, s.mutexKey)
-	s.Mutex.Unlock()
-}

mutex_nodebug.go

@@ -1,23 +0,0 @@
-//go:build !mutex_debug
-// +build !mutex_debug
-
-package nebula
-
-import (
-	"sync"
-)
-
-type mutexKey = string
-type syncRWMutex = sync.RWMutex
-type syncMutex = sync.Mutex
-
-func newSyncRWMutex(mutexKey) syncRWMutex {
-	return sync.RWMutex{}
-}
-
-func newSyncMutex(mutexKey) syncMutex {
-	return sync.Mutex{}
-}
-
-func chanDebugRecv(key mutexKey) {}
-func chanDebugSend(key mutexKey) {}

nebula.pb.go

@@ -124,7 +124,7 @@ func (x NebulaControl_MessageType) String() string {
 }
 
 func (NebulaControl_MessageType) EnumDescriptor() ([]byte, []int) {
-	return fileDescriptor_2d65afa7693df5ef, []int{8, 0}
+	return fileDescriptor_2d65afa7693df5ef, []int{9, 0}
 }
 
 type NebulaMeta struct {
@@ -541,20 +541,90 @@ func (m *NebulaHandshake) GetHmac() []byte {
 	return nil
 }
 
+type MultiPortDetails struct {
+	RxSupported bool   `protobuf:"varint,1,opt,name=RxSupported,proto3" json:"RxSupported,omitempty"`
+	TxSupported bool   `protobuf:"varint,2,opt,name=TxSupported,proto3" json:"TxSupported,omitempty"`
+	BasePort    uint32 `protobuf:"varint,3,opt,name=BasePort,proto3" json:"BasePort,omitempty"`
+	TotalPorts  uint32 `protobuf:"varint,4,opt,name=TotalPorts,proto3" json:"TotalPorts,omitempty"`
+}
+
+func (m *MultiPortDetails) Reset()         { *m = MultiPortDetails{} }
+func (m *MultiPortDetails) String() string { return proto.CompactTextString(m) }
+func (*MultiPortDetails) ProtoMessage()    {}
+func (*MultiPortDetails) Descriptor() ([]byte, []int) {
+	return fileDescriptor_2d65afa7693df5ef, []int{7}
+}
+func (m *MultiPortDetails) XXX_Unmarshal(b []byte) error {
+	return m.Unmarshal(b)
+}
+func (m *MultiPortDetails) XXX_Marshal(b []byte, deterministic bool) ([]byte, error) {
+	if deterministic {
+		return xxx_messageInfo_MultiPortDetails.Marshal(b, m, deterministic)
+	} else {
+		b = b[:cap(b)]
+		n, err := m.MarshalToSizedBuffer(b)
+		if err != nil {
+			return nil, err
+		}
+		return b[:n], nil
+	}
+}
+func (m *MultiPortDetails) XXX_Merge(src proto.Message) {
+	xxx_messageInfo_MultiPortDetails.Merge(m, src)
+}
+func (m *MultiPortDetails) XXX_Size() int {
+	return m.Size()
+}
+func (m *MultiPortDetails) XXX_DiscardUnknown() {
+	xxx_messageInfo_MultiPortDetails.DiscardUnknown(m)
+}
+
+var xxx_messageInfo_MultiPortDetails proto.InternalMessageInfo
+
+func (m *MultiPortDetails) GetRxSupported() bool {
+	if m != nil {
+		return m.RxSupported
+	}
+	return false
+}
+
+func (m *MultiPortDetails) GetTxSupported() bool {
+	if m != nil {
+		return m.TxSupported
+	}
+	return false
+}
+
+func (m *MultiPortDetails) GetBasePort() uint32 {
+	if m != nil {
+		return m.BasePort
+	}
+	return 0
+}
+
+func (m *MultiPortDetails) GetTotalPorts() uint32 {
+	if m != nil {
+		return m.TotalPorts
+	}
+	return 0
+}
+
 type NebulaHandshakeDetails struct {
-	Cert           []byte `protobuf:"bytes,1,opt,name=Cert,proto3" json:"Cert,omitempty"`
-	InitiatorIndex uint32 `protobuf:"varint,2,opt,name=InitiatorIndex,proto3" json:"InitiatorIndex,omitempty"`
-	ResponderIndex uint32 `protobuf:"varint,3,opt,name=ResponderIndex,proto3" json:"ResponderIndex,omitempty"`
-	Cookie         uint64 `protobuf:"varint,4,opt,name=Cookie,proto3" json:"Cookie,omitempty"`
-	Time           uint64 `protobuf:"varint,5,opt,name=Time,proto3" json:"Time,omitempty"`
-	CertVersion    uint32 `protobuf:"varint,8,opt,name=CertVersion,proto3" json:"CertVersion,omitempty"`
+	Cert               []byte            `protobuf:"bytes,1,opt,name=Cert,proto3" json:"Cert,omitempty"`
+	InitiatorIndex     uint32            `protobuf:"varint,2,opt,name=InitiatorIndex,proto3" json:"InitiatorIndex,omitempty"`
+	ResponderIndex     uint32            `protobuf:"varint,3,opt,name=ResponderIndex,proto3" json:"ResponderIndex,omitempty"`
+	Cookie             uint64            `protobuf:"varint,4,opt,name=Cookie,proto3" json:"Cookie,omitempty"`
+	Time               uint64            `protobuf:"varint,5,opt,name=Time,proto3" json:"Time,omitempty"`
+	CertVersion        uint32            `protobuf:"varint,8,opt,name=CertVersion,proto3" json:"CertVersion,omitempty"`
+	InitiatorMultiPort *MultiPortDetails `protobuf:"bytes,6,opt,name=InitiatorMultiPort,proto3" json:"InitiatorMultiPort,omitempty"`
+	ResponderMultiPort *MultiPortDetails `protobuf:"bytes,7,opt,name=ResponderMultiPort,proto3" json:"ResponderMultiPort,omitempty"`
 }
 
 func (m *NebulaHandshakeDetails) Reset()         { *m = NebulaHandshakeDetails{} }
 func (m *NebulaHandshakeDetails) String() string { return proto.CompactTextString(m) }
 func (*NebulaHandshakeDetails) ProtoMessage()    {}
 func (*NebulaHandshakeDetails) Descriptor() ([]byte, []int) {
-	return fileDescriptor_2d65afa7693df5ef, []int{7}
+	return fileDescriptor_2d65afa7693df5ef, []int{8}
 }
 func (m *NebulaHandshakeDetails) XXX_Unmarshal(b []byte) error {
 	return m.Unmarshal(b)
@@ -625,6 +695,20 @@ func (m *NebulaHandshakeDetails) GetCertVersion() uint32 {
 	return 0
 }
 
+func (m *NebulaHandshakeDetails) GetInitiatorMultiPort() *MultiPortDetails {
+	if m != nil {
+		return m.InitiatorMultiPort
+	}
+	return nil
+}
+
+func (m *NebulaHandshakeDetails) GetResponderMultiPort() *MultiPortDetails {
+	if m != nil {
+		return m.ResponderMultiPort
+	}
+	return nil
+}
+
 type NebulaControl struct {
 	Type                NebulaControl_MessageType `protobuf:"varint,1,opt,name=Type,proto3,enum=nebula.NebulaControl_MessageType" json:"Type,omitempty"`
 	InitiatorRelayIndex uint32                    `protobuf:"varint,2,opt,name=InitiatorRelayIndex,proto3" json:"InitiatorRelayIndex,omitempty"`
@@ -639,7 +723,7 @@ func (m *NebulaControl) Reset()         { *m = NebulaControl{} }
 func (m *NebulaControl) String() string { return proto.CompactTextString(m) }
 func (*NebulaControl) ProtoMessage()    {}
 func (*NebulaControl) Descriptor() ([]byte, []int) {
-	return fileDescriptor_2d65afa7693df5ef, []int{8}
+	return fileDescriptor_2d65afa7693df5ef, []int{9}
 }
 func (m *NebulaControl) XXX_Unmarshal(b []byte) error {
 	return m.Unmarshal(b)
@@ -730,6 +814,7 @@ func init() {
 	proto.RegisterType((*V6AddrPort)(nil), "nebula.V6AddrPort")
 	proto.RegisterType((*NebulaPing)(nil), "nebula.NebulaPing")
 	proto.RegisterType((*NebulaHandshake)(nil), "nebula.NebulaHandshake")
+	proto.RegisterType((*MultiPortDetails)(nil), "nebula.MultiPortDetails")
 	proto.RegisterType((*NebulaHandshakeDetails)(nil), "nebula.NebulaHandshakeDetails")
 	proto.RegisterType((*NebulaControl)(nil), "nebula.NebulaControl")
 }
@@ -737,57 +822,61 @@ func init() {
 func init() { proto.RegisterFile("nebula.proto", fileDescriptor_2d65afa7693df5ef) }
 
 var fileDescriptor_2d65afa7693df5ef = []byte{
-	// 785 bytes of a gzipped FileDescriptorProto
-	0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x84, 0x55, 0xcd, 0x6e, 0xeb, 0x44,
-	0x14, 0x8e, 0x1d, 0x27, 0x4e, 0x4f, 0x7e, 0xae, 0x39, 0x15, 0xc1, 0x41, 0x22, 0x0a, 0x5e, 0x54,
-	0x57, 0x2c, 0x72, 0x51, 0x5a, 0xae, 0x58, 0x72, 0x1b, 0x84, 0xd2, 0xaa, 0x3f, 0x61, 0x54, 0x8a,
-	0xc4, 0x06, 0xb9, 0xf6, 0xd0, 0x58, 0x71, 0x3c, 0xa9, 0x3d, 0x41, 0xcd, 0x5b, 0xf0, 0x30, 0x3c,
-	0x04, 0xec, 0xba, 0x42, 0x2c, 0x51, 0xbb, 0x64, 0xc9, 0x0b, 0xa0, 0x19, 0xff, 0x27, 0x86, 0xbb,
-	0x9b, 0x73, 0xbe, 0xef, 0x3b, 0x73, 0xe6, 0xf3, 0x9c, 0x31, 0x74, 0x02, 0x7a, 0xb7, 0xf1, 0xed,
-	0xf1, 0x3a, 0x64, 0x9c, 0x61, 0x33, 0x8e, 0xac, 0xbf, 0x55, 0x80, 0x2b, 0xb9, 0xbc, 0xa4, 0xdc,
-	0xc6, 0x09, 0x68, 0x37, 0xdb, 0x35, 0x35, 0x95, 0x91, 0xf2, 0xba, 0x37, 0x19, 0x8e, 0x13, 0x4d,
-	0xce, 0x18, 0x5f, 0xd2, 0x28, 0xb2, 0xef, 0xa9, 0x60, 0x11, 0xc9, 0xc5, 0x63, 0xd0, 0xbf, 0xa6,
-	0xdc, 0xf6, 0xfc, 0xc8, 0x54, 0x47, 0xca, 0xeb, 0xf6, 0x64, 0xb0, 0x2f, 0x4b, 0x08, 0x24, 0x65,
-	0x5a, 0xff, 0x28, 0xd0, 0x2e, 0x94, 0xc2, 0x16, 0x68, 0x57, 0x2c, 0xa0, 0x46, 0x0d, 0xbb, 0x70,
-	0x30, 0x63, 0x11, 0xff, 0x76, 0x43, 0xc3, 0xad, 0xa1, 0x20, 0x42, 0x2f, 0x0b, 0x09, 0x5d, 0xfb,
-	0x5b, 0x43, 0xc5, 0x8f, 0xa1, 0x2f, 0x72, 0xdf, 0xad, 0x5d, 0x9b, 0xd3, 0x2b, 0xc6, 0xbd, 0x9f,
-	0x3c, 0xc7, 0xe6, 0x1e, 0x0b, 0x8c, 0x3a, 0x0e, 0xe0, 0x43, 0x81, 0x5d, 0xb2, 0x9f, 0xa9, 0x5b,
-	0x82, 0xb4, 0x14, 0x9a, 0x6f, 0x02, 0x67, 0x51, 0x82, 0x1a, 0xd8, 0x03, 0x10, 0xd0, 0xf7, 0x0b,
-	0x66, 0xaf, 0x3c, 0xa3, 0x89, 0x87, 0xf0, 0x2a, 0x8f, 0xe3, 0x6d, 0x75, 0xd1, 0xd9, 0xdc, 0xe6,
-	0x8b, 0xe9, 0x82, 0x3a, 0x4b, 0xa3, 0x25, 0x3a, 0xcb, 0xc2, 0x98, 0x72, 0x80, 0x9f, 0xc0, 0xa0,
-	0xba, 0xb3, 0x77, 0xce, 0xd2, 0x00, 0xeb, 0x77, 0x15, 0x3e, 0xd8, 0x33, 0x05, 0x2d, 0x80, 0x6b,
-	0xdf, 0xbd, 0x5d, 0x07, 0xef, 0x5c, 0x37, 0x94, 0xd6, 0x77, 0x4f, 0x55, 0x53, 0x21, 0x85, 0x2c,
-	0x1e, 0x81, 0x9e, 0x12, 0x9a, 0xd2, 0xe4, 0x4e, 0x6a, 0xb2, 0xc8, 0x91, 0x14, 0xc4, 0x31, 0x18,
-	0xd7, 0xbe, 0x4b, 0xa8, 0x6f, 0x6f, 0x93, 0x54, 0x64, 0x36, 0x46, 0xf5, 0xa4, 0xe2, 0x1e, 0x86,
-	0x13, 0xe8, 0x96, 0xc9, 0xfa, 0xa8, 0xbe, 0x57, 0xbd, 0x4c, 0xc1, 0x13, 0x68, 0xdf, 0x9e, 0x88,
-	0xe5, 0x9c, 0x85, 0x5c, 0x7c, 0x74, 0xa1, 0xc0, 0x54, 0x91, 0x43, 0xa4, 0x48, 0x93, 0xaa, 0xb7,
-	0xb9, 0x4a, 0xdb, 0x51, 0xbd, 0x2d, 0xa8, 0x72, 0x1a, 0x9a, 0xa0, 0x3b, 0x6c, 0x13, 0x70, 0x1a,
-	0x9a, 0x75, 0x61, 0x0c, 0x49, 0x43, 0xeb, 0x08, 0x34, 0x79, 0xe2, 0x1e, 0xa8, 0x33, 0x4f, 0xba,
-	0xa6, 0x11, 0x75, 0xe6, 0x89, 0xf8, 0x82, 0xc9, 0x9b, 0xa8, 0x11, 0xf5, 0x82, 0x59, 0x27, 0x00,
-	0x79, 0x1b, 0x88, 0xb1, 0x2a, 0x76, 0x99, 0xc4, 0x15, 0x10, 0x34, 0x81, 0x49, 0x4d, 0x97, 0xc8,
-	0xb5, 0xf5, 0x15, 0x40, 0xde, 0xc6, 0xfb, 0xf6, 0xc8, 0x2a, 0xd4, 0x0b, 0x15, 0x1e, 0xd3, 0xc1,
-	0x9a, 0x7b, 0xc1, 0xfd, 0xff, 0x0f, 0x96, 0x60, 0x54, 0x0c, 0x16, 0x82, 0x76, 0xe3, 0xad, 0x68,
-	0xb2, 0x8f, 0x5c, 0x5b, 0xd6, 0xde, 0xd8, 0x08, 0xb1, 0x51, 0xc3, 0x03, 0x68, 0xc4, 0x97, 0x50,
-	0xb1, 0x7e, 0x84, 0x57, 0x71, 0xdd, 0x99, 0x1d, 0xb8, 0xd1, 0xc2, 0x5e, 0x52, 0xfc, 0x32, 0x9f,
-	0x51, 0x45, 0x5e, 0x9f, 0x9d, 0x0e, 0x32, 0xe6, 0xee, 0xa0, 0x8a, 0x26, 0x66, 0x2b, 0xdb, 0x91,
-	0x4d, 0x74, 0x88, 0x5c, 0x5b, 0x7f, 0x28, 0xd0, 0xaf, 0xd6, 0x09, 0xfa, 0x94, 0x86, 0x5c, 0xee,
-	0xd2, 0x21, 0x72, 0x8d, 0x47, 0xd0, 0x3b, 0x0b, 0x3c, 0xee, 0xd9, 0x9c, 0x85, 0x67, 0x81, 0x4b,
-	0x1f, 0x13, 0xa7, 0x77, 0xb2, 0x82, 0x47, 0x68, 0xb4, 0x66, 0x81, 0x4b, 0x13, 0x5e, 0xec, 0xe7,
-	0x4e, 0x16, 0xfb, 0xd0, 0x9c, 0x32, 0xb6, 0xf4, 0xa8, 0xa9, 0x49, 0x67, 0x92, 0x28, 0xf3, 0xab,
-	0x91, 0xfb, 0x85, 0x23, 0x68, 0x8b, 0x1e, 0x6e, 0x69, 0x18, 0x79, 0x2c, 0x30, 0x5b, 0xb2, 0x60,
-	0x31, 0x75, 0xae, 0xb5, 0x9a, 0x86, 0x7e, 0xae, 0xb5, 0x74, 0xa3, 0x65, 0xfd, 0x5a, 0x87, 0x6e,
-	0x7c, 0xb0, 0x29, 0x0b, 0x78, 0xc8, 0x7c, 0xfc, 0xa2, 0xf4, 0xdd, 0x3e, 0x2d, 0xbb, 0x96, 0x90,
-	0x2a, 0x3e, 0xdd, 0xe7, 0x70, 0x98, 0x1d, 0x4e, 0x0e, 0x4f, 0xf1, 0xdc, 0x55, 0x90, 0x50, 0x64,
-	0xc7, 0x2c, 0x28, 0x62, 0x07, 0xaa, 0x20, 0xfc, 0x0c, 0x7a, 0xe9, 0x38, 0xdf, 0x30, 0x79, 0xa9,
-	0xb5, 0xec, 0xe9, 0xd8, 0x41, 0x8a, 0xcf, 0xc2, 0x37, 0x21, 0x5b, 0x49, 0x76, 0x23, 0x63, 0xef,
-	0x61, 0x38, 0x86, 0x76, 0xb1, 0x70, 0xd5, 0x93, 0x53, 0x24, 0x64, 0xcf, 0x48, 0x56, 0x5c, 0xaf,
-	0x50, 0x94, 0x29, 0xd6, 0xec, 0xbf, 0xfe, 0x00, 0x7d, 0xc0, 0x69, 0x48, 0x6d, 0x4e, 0x25, 0x9f,
-	0xd0, 0x87, 0x0d, 0x8d, 0xb8, 0xa1, 0xe0, 0x47, 0x70, 0x58, 0xca, 0x0b, 0x4b, 0x22, 0x6a, 0xa8,
-	0xa7, 0xc7, 0xbf, 0x3d, 0x0f, 0x95, 0xa7, 0xe7, 0xa1, 0xf2, 0xd7, 0xf3, 0x50, 0xf9, 0xe5, 0x65,
-	0x58, 0x7b, 0x7a, 0x19, 0xd6, 0xfe, 0x7c, 0x19, 0xd6, 0x7e, 0x18, 0xdc, 0x7b, 0x7c, 0xb1, 0xb9,
-	0x1b, 0x3b, 0x6c, 0xf5, 0x26, 0xf2, 0x6d, 0x67, 0xb9, 0x78, 0x78, 0x13, 0xb7, 0x74, 0xd7, 0x94,
-	0x3f, 0xc2, 0xe3, 0x7f, 0x03, 0x00, 0x00, 0xff, 0xff, 0xea, 0x6f, 0xbc, 0x50, 0x18, 0x07, 0x00,
-	0x00,
+	// 864 bytes of a gzipped FileDescriptorProto
+	0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x84, 0x56, 0x4f, 0x6f, 0xe3, 0x44,
+	0x14, 0x8f, 0x1d, 0xe7, 0x4f, 0x5f, 0x9a, 0xac, 0x79, 0x15, 0x25, 0x5d, 0x89, 0x28, 0xf8, 0x50,
+	0xad, 0x38, 0x64, 0x51, 0x5b, 0x56, 0x1c, 0xd9, 0x06, 0xa1, 0xac, 0xb4, 0xed, 0x96, 0x21, 0x14,
+	0x89, 0x0b, 0x9a, 0xc6, 0x43, 0x63, 0xc5, 0xf1, 0x78, 0xed, 0x31, 0x6a, 0xbe, 0x05, 0xe2, 0xb3,
+	0xf0, 0x21, 0xe0, 0xb6, 0x47, 0x4e, 0x08, 0xb5, 0x47, 0x8e, 0x7c, 0x01, 0x34, 0xe3, 0x7f, 0xe3,
+	0xc4, 0x6c, 0x6f, 0xf3, 0xde, 0xef, 0xf7, 0x7b, 0xfe, 0xcd, 0x9b, 0x79, 0x93, 0xc0, 0x7e, 0xc0,
+	0x6e, 0x12, 0x9f, 0x4e, 0xc2, 0x88, 0x0b, 0x8e, 0xed, 0x34, 0x72, 0xfe, 0x31, 0x01, 0x2e, 0xd5,
+	0xf2, 0x82, 0x09, 0x8a, 0x27, 0x60, 0xcd, 0x37, 0x21, 0x1b, 0x1a, 0x63, 0xe3, 0xd9, 0xe0, 0x64,
+	0x34, 0xc9, 0x34, 0x25, 0x63, 0x72, 0xc1, 0xe2, 0x98, 0xde, 0x32, 0xc9, 0x22, 0x8a, 0x8b, 0xa7,
+	0xd0, 0xf9, 0x8a, 0x09, 0xea, 0xf9, 0xf1, 0xd0, 0x1c, 0x1b, 0xcf, 0x7a, 0x27, 0x47, 0xbb, 0xb2,
+	0x8c, 0x40, 0x72, 0xa6, 0xf3, 0xaf, 0x01, 0x3d, 0xad, 0x14, 0x76, 0xc1, 0xba, 0xe4, 0x01, 0xb3,
+	0x1b, 0xd8, 0x87, 0xbd, 0x19, 0x8f, 0xc5, 0x37, 0x09, 0x8b, 0x36, 0xb6, 0x81, 0x08, 0x83, 0x22,
+	0x24, 0x2c, 0xf4, 0x37, 0xb6, 0x89, 0x4f, 0xe1, 0x50, 0xe6, 0xbe, 0x0b, 0x5d, 0x2a, 0xd8, 0x25,
+	0x17, 0xde, 0x4f, 0xde, 0x82, 0x0a, 0x8f, 0x07, 0x76, 0x13, 0x8f, 0xe0, 0x43, 0x89, 0x5d, 0xf0,
+	0x9f, 0x99, 0x5b, 0x81, 0xac, 0x1c, 0xba, 0x4a, 0x82, 0xc5, 0xb2, 0x02, 0xb5, 0x70, 0x00, 0x20,
+	0xa1, 0xef, 0x97, 0x9c, 0xae, 0x3d, 0xbb, 0x8d, 0x07, 0xf0, 0xa4, 0x8c, 0xd3, 0xcf, 0x76, 0xa4,
+	0xb3, 0x2b, 0x2a, 0x96, 0xd3, 0x25, 0x5b, 0xac, 0xec, 0xae, 0x74, 0x56, 0x84, 0x29, 0x65, 0x0f,
+	0x3f, 0x86, 0xa3, 0x7a, 0x67, 0x2f, 0x17, 0x2b, 0x1b, 0x9c, 0x3f, 0x4c, 0xf8, 0x60, 0xa7, 0x29,
+	0xe8, 0x00, 0xbc, 0xf1, 0xdd, 0xeb, 0x30, 0x78, 0xe9, 0xba, 0x91, 0x6a, 0x7d, 0xff, 0xdc, 0x1c,
+	0x1a, 0x44, 0xcb, 0xe2, 0x31, 0x74, 0x72, 0x42, 0x5b, 0x35, 0x79, 0x3f, 0x6f, 0xb2, 0xcc, 0x91,
+	0x1c, 0xc4, 0x09, 0xd8, 0x6f, 0x7c, 0x97, 0x30, 0x9f, 0x6e, 0xb2, 0x54, 0x3c, 0x6c, 0x8d, 0x9b,
+	0x59, 0xc5, 0x1d, 0x0c, 0x4f, 0xa0, 0x5f, 0x25, 0x77, 0xc6, 0xcd, 0x9d, 0xea, 0x55, 0x0a, 0x9e,
+	0x41, 0xef, 0xfa, 0x4c, 0x2e, 0xaf, 0x78, 0x24, 0xe4, 0xa1, 0x4b, 0x05, 0xe6, 0x8a, 0x12, 0x22,
+	0x3a, 0x4d, 0xa9, 0x5e, 0x94, 0x2a, 0x6b, 0x4b, 0xf5, 0x42, 0x53, 0x95, 0x34, 0x1c, 0x42, 0x67,
+	0xc1, 0x93, 0x40, 0xb0, 0x68, 0xd8, 0x94, 0x8d, 0x21, 0x79, 0xe8, 0x1c, 0x83, 0xa5, 0x76, 0x3c,
+	0x00, 0x73, 0xe6, 0xa9, 0xae, 0x59, 0xc4, 0x9c, 0x79, 0x32, 0x7e, 0xcd, 0xd5, 0x4d, 0xb4, 0x88,
+	0xf9, 0x9a, 0x3b, 0x67, 0x00, 0xa5, 0x0d, 0xc4, 0x54, 0x95, 0x76, 0x99, 0xa4, 0x15, 0x10, 0x2c,
+	0x89, 0x29, 0x4d, 0x9f, 0xa8, 0xb5, 0xf3, 0x25, 0x40, 0x69, 0xe3, 0xb1, 0x6f, 0x14, 0x15, 0x9a,
+	0x5a, 0x85, 0xbb, 0x7c, 0xb0, 0xae, 0xbc, 0xe0, 0xf6, 0xfd, 0x83, 0x25, 0x19, 0x35, 0x83, 0x85,
+	0x60, 0xcd, 0xbd, 0x35, 0xcb, 0xbe, 0xa3, 0xd6, 0x8e, 0xb3, 0x33, 0x36, 0x52, 0x6c, 0x37, 0x70,
+	0x0f, 0x5a, 0xe9, 0x25, 0x34, 0x9c, 0x1f, 0xe1, 0x49, 0x5a, 0x77, 0x46, 0x03, 0x37, 0x5e, 0xd2,
+	0x15, 0xc3, 0x2f, 0xca, 0x19, 0x35, 0xd4, 0xf5, 0xd9, 0x72, 0x50, 0x30, 0xb7, 0x07, 0x55, 0x9a,
+	0x98, 0xad, 0xe9, 0x42, 0x99, 0xd8, 0x27, 0x6a, 0xed, 0xfc, 0x6a, 0x80, 0x7d, 0x91, 0xf8, 0xc2,
+	0x93, 0x1b, 0xcd, 0x89, 0x63, 0xe8, 0x91, 0xbb, 0x6f, 0x93, 0x30, 0xe4, 0x91, 0x60, 0xae, 0xfa,
+	0x4c, 0x97, 0xe8, 0x29, 0xc9, 0x98, 0x6b, 0x0c, 0x33, 0x65, 0x68, 0x29, 0x7c, 0x0a, 0xdd, 0x73,
+	0x1a, 0x33, 0xad, 0x97, 0x45, 0x8c, 0x23, 0x80, 0x39, 0x17, 0xd4, 0xcf, 0xaf, 0x8f, 0x44, 0xb5,
+	0x8c, 0xf3, 0x97, 0x09, 0x87, 0xf5, 0x9b, 0x91, 0x7b, 0x98, 0xb2, 0x48, 0x28, 0x4f, 0xfb, 0x44,
+	0xad, 0xf1, 0x18, 0x06, 0xaf, 0x02, 0x4f, 0x78, 0x54, 0xf0, 0xe8, 0x55, 0xe0, 0xb2, 0xbb, 0xec,
+	0xf8, 0xb7, 0xb2, 0x92, 0x47, 0x58, 0x1c, 0xf2, 0xc0, 0x65, 0x19, 0x2f, 0x35, 0xb6, 0x95, 0xc5,
+	0x43, 0x68, 0x4f, 0x39, 0x5f, 0x79, 0x4c, 0x59, 0xb3, 0x48, 0x16, 0x15, 0x87, 0xd8, 0x2a, 0x0f,
+	0x51, 0x36, 0x42, 0x7a, 0xb8, 0x66, 0x51, 0xec, 0xf1, 0x60, 0xd8, 0x55, 0x05, 0xf5, 0x14, 0xce,
+	0x00, 0x0b, 0x1f, 0x45, 0xa7, 0xb3, 0xc9, 0x1f, 0xe6, 0x47, 0xb7, 0x7d, 0x04, 0xa4, 0x46, 0x23,
+	0x2b, 0x15, 0x4e, 0xcb, 0x4a, 0x9d, 0xc7, 0x2a, 0xed, 0x6a, 0x9c, 0xdf, 0x9a, 0xd0, 0x4f, 0x1b,
+	0x3c, 0xe5, 0x81, 0x88, 0xb8, 0x8f, 0x9f, 0x57, 0x2e, 0xf5, 0x27, 0xd5, 0x2b, 0x95, 0x91, 0x6a,
+	0xee, 0xf5, 0x67, 0x70, 0x50, 0x18, 0x55, 0x2f, 0x8b, 0xde, 0xff, 0x3a, 0x48, 0x2a, 0x0a, 0x43,
+	0x9a, 0x22, 0x3d, 0x89, 0x3a, 0x08, 0x3f, 0x85, 0x41, 0xfe, 0xd6, 0xcd, 0xb9, 0x9a, 0x78, 0xab,
+	0x78, 0x57, 0xb7, 0x10, 0xfd, 0xcd, 0xfc, 0x3a, 0xe2, 0x6b, 0xc5, 0x6e, 0x15, 0xec, 0x1d, 0x0c,
+	0x27, 0xd0, 0xd3, 0x0b, 0xd7, 0xbd, 0xc7, 0x3a, 0xa1, 0x78, 0x63, 0x8b, 0xe2, 0x9d, 0x1a, 0x45,
+	0x95, 0xe2, 0xcc, 0xfe, 0xef, 0xe7, 0xf1, 0x10, 0x70, 0x1a, 0x31, 0x2a, 0x98, 0xe2, 0x13, 0xf6,
+	0x36, 0x61, 0xb1, 0xb0, 0x0d, 0xfc, 0x08, 0x0e, 0x2a, 0x79, 0xd9, 0x92, 0x98, 0xd9, 0xe6, 0xf9,
+	0xe9, 0xef, 0xf7, 0x23, 0xe3, 0xdd, 0xfd, 0xc8, 0xf8, 0xfb, 0x7e, 0x64, 0xfc, 0xf2, 0x30, 0x6a,
+	0xbc, 0x7b, 0x18, 0x35, 0xfe, 0x7c, 0x18, 0x35, 0x7e, 0x38, 0xba, 0xf5, 0xc4, 0x32, 0xb9, 0x99,
+	0x2c, 0xf8, 0xfa, 0x79, 0xec, 0xd3, 0xc5, 0x6a, 0xf9, 0xf6, 0x79, 0x6a, 0xe9, 0xa6, 0xad, 0xfe,
+	0x25, 0x9c, 0xfe, 0x17, 0x00, 0x00, 0xff, 0xff, 0xd6, 0x71, 0x5a, 0xf8, 0x35, 0x08, 0x00, 0x00,
 }
 
 func (m *NebulaMeta) Marshal() (dAtA []byte, err error) {
@@ -1114,6 +1203,59 @@ func (m *NebulaHandshake) MarshalToSizedBuffer(dAtA []byte) (int, error) {
 	return len(dAtA) - i, nil
 }
 
+func (m *MultiPortDetails) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalToSizedBuffer(dAtA[:size])
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *MultiPortDetails) MarshalTo(dAtA []byte) (int, error) {
+	size := m.Size()
+	return m.MarshalToSizedBuffer(dAtA[:size])
+}
+
+func (m *MultiPortDetails) MarshalToSizedBuffer(dAtA []byte) (int, error) {
+	i := len(dAtA)
+	_ = i
+	var l int
+	_ = l
+	if m.TotalPorts != 0 {
+		i = encodeVarintNebula(dAtA, i, uint64(m.TotalPorts))
+		i--
+		dAtA[i] = 0x20
+	}
+	if m.BasePort != 0 {
+		i = encodeVarintNebula(dAtA, i, uint64(m.BasePort))
+		i--
+		dAtA[i] = 0x18
+	}
+	if m.TxSupported {
+		i--
+		if m.TxSupported {
+			dAtA[i] = 1
+		} else {
+			dAtA[i] = 0
+		}
+		i--
+		dAtA[i] = 0x10
+	}
+	if m.RxSupported {
+		i--
+		if m.RxSupported {
+			dAtA[i] = 1
+		} else {
+			dAtA[i] = 0
+		}
+		i--
+		dAtA[i] = 0x8
+	}
+	return len(dAtA) - i, nil
+}
+
 func (m *NebulaHandshakeDetails) Marshal() (dAtA []byte, err error) {
 	size := m.Size()
 	dAtA = make([]byte, size)
@@ -1139,6 +1281,30 @@ func (m *NebulaHandshakeDetails) MarshalToSizedBuffer(dAtA []byte) (int, error)
 		i--
 		dAtA[i] = 0x40
 	}
+	if m.ResponderMultiPort != nil {
+		{
+			size, err := m.ResponderMultiPort.MarshalToSizedBuffer(dAtA[:i])
+			if err != nil {
+				return 0, err
+			}
+			i -= size
+			i = encodeVarintNebula(dAtA, i, uint64(size))
+		}
+		i--
+		dAtA[i] = 0x3a
+	}
+	if m.InitiatorMultiPort != nil {
+		{
+			size, err := m.InitiatorMultiPort.MarshalToSizedBuffer(dAtA[:i])
+			if err != nil {
+				return 0, err
+			}
+			i -= size
+			i = encodeVarintNebula(dAtA, i, uint64(size))
+		}
+		i--
+		dAtA[i] = 0x32
+	}
 	if m.Time != 0 {
 		i = encodeVarintNebula(dAtA, i, uint64(m.Time))
 		i--
@@ -1392,6 +1558,27 @@ func (m *NebulaHandshake) Size() (n int) {
 	return n
 }
 
+func (m *MultiPortDetails) Size() (n int) {
+	if m == nil {
+		return 0
+	}
+	var l int
+	_ = l
+	if m.RxSupported {
+		n += 2
+	}
+	if m.TxSupported {
+		n += 2
+	}
+	if m.BasePort != 0 {
+		n += 1 + sovNebula(uint64(m.BasePort))
+	}
+	if m.TotalPorts != 0 {
+		n += 1 + sovNebula(uint64(m.TotalPorts))
+	}
+	return n
+}
+
 func (m *NebulaHandshakeDetails) Size() (n int) {
 	if m == nil {
 		return 0
@@ -1414,6 +1601,14 @@ func (m *NebulaHandshakeDetails) Size() (n int) {
 	if m.Time != 0 {
 		n += 1 + sovNebula(uint64(m.Time))
 	}
+	if m.InitiatorMultiPort != nil {
+		l = m.InitiatorMultiPort.Size()
+		n += 1 + l + sovNebula(uint64(l))
+	}
+	if m.ResponderMultiPort != nil {
+		l = m.ResponderMultiPort.Size()
+		n += 1 + l + sovNebula(uint64(l))
+	}
 	if m.CertVersion != 0 {
 		n += 1 + sovNebula(uint64(m.CertVersion))
 	}
@@ -2356,6 +2551,134 @@ func (m *NebulaHandshake) Unmarshal(dAtA []byte) error {
 	}
 	return nil
 }
+func (m *MultiPortDetails) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowNebula
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= uint64(b&0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: MultiPortDetails: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: MultiPortDetails: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field RxSupported", wireType)
+			}
+			var v int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				v |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			m.RxSupported = bool(v != 0)
+		case 2:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field TxSupported", wireType)
+			}
+			var v int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				v |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			m.TxSupported = bool(v != 0)
+		case 3:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field BasePort", wireType)
+			}
+			m.BasePort = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.BasePort |= uint32(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 4:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field TotalPorts", wireType)
+			}
+			m.TotalPorts = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.TotalPorts |= uint32(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		default:
+			iNdEx = preIndex
+			skippy, err := skipNebula(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if (skippy < 0) || (iNdEx+skippy) < 0 {
+				return ErrInvalidLengthNebula
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
 func (m *NebulaHandshakeDetails) Unmarshal(dAtA []byte) error {
 	l := len(dAtA)
 	iNdEx := 0
@@ -2495,6 +2818,78 @@ func (m *NebulaHandshakeDetails) Unmarshal(dAtA []byte) error {
 					break
 				}
 			}
+		case 6:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field InitiatorMultiPort", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return ErrInvalidLengthNebula
+			}
+			postIndex := iNdEx + msglen
+			if postIndex < 0 {
+				return ErrInvalidLengthNebula
+			}
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			if m.InitiatorMultiPort == nil {
+				m.InitiatorMultiPort = &MultiPortDetails{}
+			}
+			if err := m.InitiatorMultiPort.Unmarshal(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
+		case 7:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field ResponderMultiPort", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowNebula
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return ErrInvalidLengthNebula
+			}
+			postIndex := iNdEx + msglen
+			if postIndex < 0 {
+				return ErrInvalidLengthNebula
+			}
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			if m.ResponderMultiPort == nil {
+				m.ResponderMultiPort = &MultiPortDetails{}
+			}
+			if err := m.ResponderMultiPort.Unmarshal(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
 		case 8:
 			if wireType != 0 {
 				return fmt.Errorf("proto: wrong wireType = %d for field CertVersion", wireType)

nebula.proto

@@ -65,6 +65,13 @@ message NebulaHandshake {
   bytes Hmac = 2;
 }
 
+message MultiPortDetails {
+  bool RxSupported = 1;
+  bool TxSupported = 2;
+  uint32 BasePort = 3;
+  uint32 TotalPorts = 4;
+}
+
 message NebulaHandshakeDetails {
   bytes Cert = 1;
   uint32 InitiatorIndex = 2;
@@ -72,8 +79,9 @@ message NebulaHandshakeDetails {
   uint64 Cookie = 4;
   uint64 Time = 5;
   uint32 CertVersion = 8;
-  // reserved for WIP multiport
-  reserved 6, 7;
+
+  MultiPortDetails InitiatorMultiPort = 6;
+  MultiPortDetails ResponderMultiPort = 7;
 }
 
 message NebulaControl {

outside.go

@@ -31,8 +31,7 @@ func (f *Interface) readOutsidePackets(ip netip.AddrPort, via *ViaSender, out []
 
 	//l.Error("in packet ", header, packet[HeaderLen:])
 	if ip.IsValid() {
-		_, found := f.myVpnNetworksTable.Lookup(ip.Addr())
-		if found {
+		if f.myVpnNetworksTable.Contains(ip.Addr()) {
 			if f.l.Level >= logrus.DebugLevel {
 				f.l.WithField("udpAddr", ip).Debug("Refusing to process double encrypted packet")
 			}
@@ -82,7 +81,7 @@ func (f *Interface) readOutsidePackets(ip netip.AddrPort, via *ViaSender, out []
 			// Pull the Roaming parts up here, and return in all call paths.
 			f.handleHostRoaming(hostinfo, ip)
 			// Track usage of both the HostInfo and the Relay for the received & authenticated packet
-			f.connectionManager.In(hostinfo.localIndexId)
+			f.connectionManager.In(hostinfo)
 			f.connectionManager.RelayUsed(h.RemoteIndex)
 
 			relay, ok := hostinfo.relayState.QueryRelayForByIdx(h.RemoteIndex)
@@ -214,7 +213,7 @@ func (f *Interface) readOutsidePackets(ip netip.AddrPort, via *ViaSender, out []
 
 	f.handleHostRoaming(hostinfo, ip)
 
-	f.connectionManager.In(hostinfo.localIndexId)
+	f.connectionManager.In(hostinfo)
 }
 
 // closeTunnel closes a tunnel locally, it does not send a closeTunnel packet to the remote
@@ -233,6 +232,16 @@ func (f *Interface) sendCloseTunnel(h *HostInfo) {
 
 func (f *Interface) handleHostRoaming(hostinfo *HostInfo, udpAddr netip.AddrPort) {
 	if udpAddr.IsValid() && hostinfo.remote != udpAddr {
+		if hostinfo.multiportRx {
+			// If the remote is sending with multiport, we aren't roaming unless
+			// the IP has changed
+			if hostinfo.remote.Addr().Compare(udpAddr.Addr()) == 0 {
+				return
+			}
+			// Keep the port from the original hostinfo, because the remote is transmitting from multiport ports
+			udpAddr = netip.AddrPortFrom(udpAddr.Addr(), hostinfo.remote.Port())
+		}
+
 		if !f.lightHouse.GetRemoteAllowList().AllowAll(hostinfo.vpnAddrs, udpAddr.Addr()) {
 			hostinfo.logger(f.l).WithField("newAddr", udpAddr).Debug("lighthouse.remote_allow_list denied roaming")
 			return
@@ -313,12 +322,11 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 	offset := ipv6.HeaderLen // Start at the end of the ipv6 header
 	next := 0
 	for {
-		if dataLen < offset {
+		if protoAt >= dataLen {
 			break
 		}
-
 		proto := layers.IPProtocol(data[protoAt])
-		//fmt.Println(proto, protoAt)
+
 		switch proto {
 		case layers.IPProtocolICMPv6, layers.IPProtocolESP, layers.IPProtocolNoNextHeader:
 			fp.Protocol = uint8(proto)
@@ -366,7 +374,7 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 
 		case layers.IPProtocolAH:
 			// Auth headers, used by IPSec, have a different meaning for header length
-			if dataLen < offset+1 {
+			if dataLen <= offset+1 {
 				break
 			}
 
@@ -374,7 +382,7 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 
 		default:
 			// Normal ipv6 header length processing
-			if dataLen < offset+1 {
+			if dataLen <= offset+1 {
 				break
 			}
 
@@ -500,7 +508,7 @@ func (f *Interface) decryptToTun(hostinfo *HostInfo, messageCounter uint64, out
 		return false
 	}
 
-	f.connectionManager.In(hostinfo.localIndexId)
+	f.connectionManager.In(hostinfo)
 	_, err = f.readers[q].Write(out)
 	if err != nil {
 		f.l.WithError(err).Error("Failed to write to tun")

outside_test.go

@@ -117,6 +117,45 @@ func Test_newPacket_v6(t *testing.T) {
 	err = newPacket(buffer.Bytes(), true, p)
 	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
 
+	// A v6 packet with a hop-by-hop extension
+	// ICMPv6 Payload (Echo Request)
+	icmpLayer := layers.ICMPv6{
+		TypeCode: layers.ICMPv6TypeEchoRequest,
+	}
+	// Hop-by-Hop Extension Header
+	hopOption := layers.IPv6HopByHopOption{}
+	hopOption.OptionData = []byte{0, 0, 0, 0}
+	hopByHop := layers.IPv6HopByHop{}
+	hopByHop.Options = append(hopByHop.Options, &hopOption)
+
+	ip = layers.IPv6{
+		Version:    6,
+		HopLimit:   128,
+		NextHeader: layers.IPProtocolIPv6Destination,
+		SrcIP:      net.IPv6linklocalallrouters,
+		DstIP:      net.IPv6linklocalallnodes,
+	}
+
+	buffer.Clear()
+	err = gopacket.SerializeLayers(buffer, gopacket.SerializeOptions{
+		ComputeChecksums: false,
+		FixLengths:       true,
+	}, &ip, &hopByHop, &icmpLayer)
+	if err != nil {
+		panic(err)
+	}
+	// Ensure buffer length checks during parsing with the next 2 tests.
+
+	// A full IPv6 header and 1 byte in the first extension, but missing
+	// the length byte.
+	err = newPacket(buffer.Bytes()[:41], true, p)
+	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
+
+	// A full IPv6 header plus 1 full extension, but only 1 byte of the
+	// next layer, missing length byte
+	err = newPacket(buffer.Bytes()[:49], true, p)
+	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
+
 	// A good ICMP packet
 	ip = layers.IPv6{
 		Version:    6,
@@ -288,6 +327,10 @@ func Test_newPacket_v6(t *testing.T) {
 	assert.Equal(t, uint16(22), p.LocalPort)
 	assert.False(t, p.Fragment)
 
+	// Ensure buffer bounds checking during processing
+	err = newPacket(b[:41], true, p)
+	require.ErrorIs(t, err, ErrIPv6PacketTooShort)
+
 	// Invalid AH header
 	b = buffer.Bytes()
 	err = newPacket(b, true, p)

overlay/tun_linux.go

@@ -34,10 +34,11 @@ type tun struct {
 	deviceIndex int
 	ioctlFd     uintptr
 
-	Routes          atomic.Pointer[[]Route]
-	routeTree       atomic.Pointer[bart.Table[routing.Gateways]]
-	routeChan       chan struct{}
-	useSystemRoutes bool
+	Routes                    atomic.Pointer[[]Route]
+	routeTree                 atomic.Pointer[bart.Table[routing.Gateways]]
+	routeChan                 chan struct{}
+	useSystemRoutes           bool
+	useSystemRoutesBufferSize int
 
 	l *logrus.Logger
 }
@@ -124,12 +125,13 @@ func newTun(c *config.C, l *logrus.Logger, vpnNetworks []netip.Prefix, multiqueu
 
 func newTunGeneric(c *config.C, l *logrus.Logger, file *os.File, vpnNetworks []netip.Prefix) (*tun, error) {
 	t := &tun{
-		ReadWriteCloser: file,
-		fd:              int(file.Fd()),
-		vpnNetworks:     vpnNetworks,
-		TXQueueLen:      c.GetInt("tun.tx_queue", 500),
-		useSystemRoutes: c.GetBool("tun.use_system_route_table", false),
-		l:               l,
+		ReadWriteCloser:           file,
+		fd:                        int(file.Fd()),
+		vpnNetworks:               vpnNetworks,
+		TXQueueLen:                c.GetInt("tun.tx_queue", 500),
+		useSystemRoutes:           c.GetBool("tun.use_system_route_table", false),
+		useSystemRoutesBufferSize: c.GetInt("tun.use_system_route_table_buffer_size", 0),
+		l:                         l,
 	}
 
 	err := t.reload(c, true)
@@ -531,7 +533,13 @@ func (t *tun) watchRoutes() {
 	rch := make(chan netlink.RouteUpdate)
 	doneChan := make(chan struct{})
 
-	if err := netlink.RouteSubscribe(rch, doneChan); err != nil {
+	netlinkOptions := netlink.RouteSubscribeOptions{
+		ReceiveBufferSize:      t.useSystemRoutesBufferSize,
+		ReceiveBufferForceSize: t.useSystemRoutesBufferSize != 0,
+		ErrorCallback:          func(e error) { t.l.WithError(e).Errorf("netlink error") },
+	}
+
+	if err := netlink.RouteSubscribeWithOptions(rch, doneChan, netlinkOptions); err != nil {
 		t.l.WithError(err).Errorf("failed to subscribe to system route changes")
 		return
 	}
@@ -541,8 +549,14 @@ func (t *tun) watchRoutes() {
 	go func() {
 		for {
 			select {
-			case r := <-rch:
-				t.updateRoutes(r)
+			case r, ok := <-rch:
+				if ok {
+					t.updateRoutes(r)
+				} else {
+					// may be should do something here as
+					// netlink stops sending updates
+					return
+				}
 			case <-doneChan:
 				// netlink.RouteSubscriber will close the rch for us
 				return

pki.go

@@ -33,16 +33,16 @@ type CertState struct {
 	v2Cert           cert.Certificate
 	v2HandshakeBytes []byte
 
-	defaultVersion cert.Version
-	privateKey     []byte
-	pkcs11Backed   bool
-	cipher         string
+	initiatingVersion cert.Version
+	privateKey        []byte
+	pkcs11Backed      bool
+	cipher            string
 
 	myVpnNetworks            []netip.Prefix
-	myVpnNetworksTable       *bart.Table[struct{}]
+	myVpnNetworksTable       *bart.Lite
 	myVpnAddrs               []netip.Addr
-	myVpnAddrsTable          *bart.Table[struct{}]
-	myVpnBroadcastAddrsTable *bart.Table[struct{}]
+	myVpnAddrsTable          *bart.Lite
+	myVpnBroadcastAddrsTable *bart.Lite
 }
 
 func NewPKIFromConfig(l *logrus.Logger, c *config.C) (*PKI, error) {
@@ -194,7 +194,7 @@ func (p *PKI) reloadCAPool(c *config.C) *util.ContextualError {
 }
 
 func (cs *CertState) GetDefaultCertificate() cert.Certificate {
-	c := cs.getCertificate(cs.defaultVersion)
+	c := cs.getCertificate(cs.initiatingVersion)
 	if c == nil {
 		panic("No default certificate found")
 	}
@@ -317,37 +317,37 @@ func newCertStateFromConfig(c *config.C) (*CertState, error) {
 		return nil, errors.New("no certificates found in pki.cert")
 	}
 
-	useDefaultVersion := uint32(1)
+	useInitiatingVersion := uint32(1)
 	if v1 == nil {
 		// The only condition that requires v2 as the default is if only a v2 certificate is present
 		// We do this to avoid having to configure it specifically in the config file
-		useDefaultVersion = 2
+		useInitiatingVersion = 2
 	}
 
-	rawDefaultVersion := c.GetUint32("pki.default_version", useDefaultVersion)
-	var defaultVersion cert.Version
-	switch rawDefaultVersion {
+	rawInitiatingVersion := c.GetUint32("pki.initiating_version", useInitiatingVersion)
+	var initiatingVersion cert.Version
+	switch rawInitiatingVersion {
 	case 1:
 		if v1 == nil {
-			return nil, fmt.Errorf("can not use pki.default_version 1 without a v1 certificate in pki.cert")
+			return nil, fmt.Errorf("can not use pki.initiating_version 1 without a v1 certificate in pki.cert")
 		}
-		defaultVersion = cert.Version1
+		initiatingVersion = cert.Version1
 	case 2:
-		defaultVersion = cert.Version2
+		initiatingVersion = cert.Version2
 	default:
-		return nil, fmt.Errorf("unknown pki.default_version: %v", rawDefaultVersion)
+		return nil, fmt.Errorf("unknown pki.initiating_version: %v", rawInitiatingVersion)
 	}
 
-	return newCertState(defaultVersion, v1, v2, isPkcs11, curve, rawKey)
+	return newCertState(initiatingVersion, v1, v2, isPkcs11, curve, rawKey)
 }
 
 func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, privateKeyCurve cert.Curve, privateKey []byte) (*CertState, error) {
 	cs := CertState{
 		privateKey:               privateKey,
 		pkcs11Backed:             pkcs11backed,
-		myVpnNetworksTable:       new(bart.Table[struct{}]),
-		myVpnAddrsTable:          new(bart.Table[struct{}]),
-		myVpnBroadcastAddrsTable: new(bart.Table[struct{}]),
+		myVpnNetworksTable:       new(bart.Lite),
+		myVpnAddrsTable:          new(bart.Lite),
+		myVpnBroadcastAddrsTable: new(bart.Lite),
 	}
 
 	if v1 != nil && v2 != nil {
@@ -361,7 +361,7 @@ func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, p
 
 		//TODO: CERT-V2 make sure v2 has v1s address
 
-		cs.defaultVersion = dv
+		cs.initiatingVersion = dv
 	}
 
 	if v1 != nil {
@@ -380,8 +380,8 @@ func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, p
 		cs.v1Cert = v1
 		cs.v1HandshakeBytes = v1hs
 
-		if cs.defaultVersion == 0 {
-			cs.defaultVersion = cert.Version1
+		if cs.initiatingVersion == 0 {
+			cs.initiatingVersion = cert.Version1
 		}
 	}
 
@@ -401,8 +401,8 @@ func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, p
 		cs.v2Cert = v2
 		cs.v2HandshakeBytes = v2hs
 
-		if cs.defaultVersion == 0 {
-			cs.defaultVersion = cert.Version2
+		if cs.initiatingVersion == 0 {
+			cs.initiatingVersion = cert.Version2
 		}
 	}
 
@@ -415,16 +415,16 @@ func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, p
 
 	for _, network := range crt.Networks() {
 		cs.myVpnNetworks = append(cs.myVpnNetworks, network)
-		cs.myVpnNetworksTable.Insert(network, struct{}{})
+		cs.myVpnNetworksTable.Insert(network)
 
 		cs.myVpnAddrs = append(cs.myVpnAddrs, network.Addr())
-		cs.myVpnAddrsTable.Insert(netip.PrefixFrom(network.Addr(), network.Addr().BitLen()), struct{}{})
+		cs.myVpnAddrsTable.Insert(netip.PrefixFrom(network.Addr(), network.Addr().BitLen()))
 
 		if network.Addr().Is4() {
 			addr := network.Masked().Addr().As4()
 			mask := net.CIDRMask(network.Bits(), network.Addr().BitLen())
 			binary.BigEndian.PutUint32(addr[:], binary.BigEndian.Uint32(addr[:])|^binary.BigEndian.Uint32(mask))
-			cs.myVpnBroadcastAddrsTable.Insert(netip.PrefixFrom(netip.AddrFrom4(addr), network.Addr().BitLen()), struct{}{})
+			cs.myVpnBroadcastAddrsTable.Insert(netip.PrefixFrom(netip.AddrFrom4(addr), network.Addr().BitLen()))
 		}
 	}
 

relay_manager.go

@@ -241,15 +241,13 @@ func (rm *relayManager) handleCreateRelayRequest(v cert.Version, h *HostInfo, f
 	logMsg.Info("handleCreateRelayRequest")
 	// Is the source of the relay me? This should never happen, but did happen due to
 	// an issue migrating relays over to newly re-handshaked host info objects.
-	_, found := f.myVpnAddrsTable.Lookup(from)
-	if found {
+	if f.myVpnAddrsTable.Contains(from) {
 		logMsg.WithField("myIP", from).Error("Discarding relay request from myself")
 		return
 	}
 
 	// Is the target of the relay me?
-	_, found = f.myVpnAddrsTable.Lookup(target)
-	if found {
+	if f.myVpnAddrsTable.Contains(target) {
 		existingRelay, ok := h.relayState.QueryRelayForByIp(from)
 		if ok {
 			switch existingRelay.State {

remote_list.go

@@ -7,6 +7,7 @@ import (
 	"slices"
 	"sort"
 	"strconv"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -184,7 +185,7 @@ func (hr *hostnamesResults) GetAddrs() []netip.AddrPort {
 // It serves as a local cache of query replies, host update notifications, and locally learned addresses
 type RemoteList struct {
 	// Every interaction with internals requires a lock!
-	syncRWMutex
+	sync.RWMutex
 
 	// The full list of vpn addresses assigned to this host
 	vpnAddrs []netip.Addr
@@ -214,12 +215,11 @@ type RemoteList struct {
 // NewRemoteList creates a new empty RemoteList
 func NewRemoteList(vpnAddrs []netip.Addr, shouldAdd func(netip.Addr) bool) *RemoteList {
 	r := &RemoteList{
-		syncRWMutex: newSyncRWMutex("remote-list"),
-		vpnAddrs:    make([]netip.Addr, len(vpnAddrs)),
-		addrs:       make([]netip.AddrPort, 0),
-		relays:      make([]netip.Addr, 0),
-		cache:       make(map[netip.Addr]*cache),
-		shouldAdd:   shouldAdd,
+		vpnAddrs:  make([]netip.Addr, len(vpnAddrs)),
+		addrs:     make([]netip.AddrPort, 0),
+		relays:    make([]netip.Addr, 0),
+		cache:     make(map[netip.Addr]*cache),
+		shouldAdd: shouldAdd,
 	}
 	copy(r.vpnAddrs, vpnAddrs)
 	return r

service/service.go

@@ -9,13 +9,10 @@ import (
 	"math"
 	"net"
 	"net/netip"
-	"os"
 	"strings"
 	"sync"
 
-	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
-	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/overlay"
 	"golang.org/x/sync/errgroup"
 	"gvisor.dev/gvisor/pkg/buffer"
@@ -46,14 +43,7 @@ type Service struct {
 	}
 }
 
-func New(config *config.C) (*Service, error) {
-	logger := logrus.New()
-	logger.Out = os.Stdout
-
-	control, err := nebula.Main(config, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
-	if err != nil {
-		return nil, err
-	}
+func New(control *nebula.Control) (*Service, error) {
 	control.Start()
 
 	ctx := control.Context()

service/service_test.go

@@ -5,13 +5,17 @@ import (
 	"context"
 	"errors"
 	"net/netip"
+	"os"
 	"testing"
 	"time"
 
 	"dario.cat/mergo"
+	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/cert_test"
 	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/overlay"
 	"golang.org/x/sync/errgroup"
 	"gopkg.in/yaml.v3"
 )
@@ -71,7 +75,15 @@ func newSimpleService(caCrt cert.Certificate, caKey []byte, name string, udpIp n
 		panic(err)
 	}
 
-	s, err := New(&c)
+	logger := logrus.New()
+	logger.Out = os.Stdout
+
+	control, err := nebula.Main(&c, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
+	if err != nil {
+		panic(err)
+	}
+
+	s, err := New(control)
 	if err != nil {
 		panic(err)
 	}

timeout.go

@@ -1,6 +1,7 @@
 package nebula
 
 import (
+	"sync"
 	"time"
 )
 
@@ -33,7 +34,7 @@ type TimerWheel[T any] struct {
 }
 
 type LockingTimerWheel[T any] struct {
-	m syncMutex
+	m sync.Mutex
 	t *TimerWheel[T]
 }
 
@@ -80,9 +81,8 @@ func NewTimerWheel[T any](min, max time.Duration) *TimerWheel[T] {
 }
 
 // NewLockingTimerWheel is version of TimerWheel that is safe for concurrent use with a small performance penalty
-func NewLockingTimerWheel[T any](name string, min, max time.Duration) *LockingTimerWheel[T] {
+func NewLockingTimerWheel[T any](min, max time.Duration) *LockingTimerWheel[T] {
 	return &LockingTimerWheel[T]{
-		m: newSyncMutex(name),
 		t: NewTimerWheel[T](min, max),
 	}
 }

udp/errors.go

@@ -0,0 +1,5 @@
+package udp
+
+import "errors"
+
+var ErrInvalidIPv6RemoteForSocket = errors.New("listener is IPv4, but writing to IPv6 remote")

udp/udp_darwin.go

@@ -3,20 +3,62 @@
 
 package udp
 
-// Darwin support is primarily implemented in udp_generic, besides NewListenConfig
-
 import (
+	"context"
+	"encoding/binary"
+	"errors"
 	"fmt"
 	"net"
 	"net/netip"
 	"syscall"
+	"unsafe"
 
 	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/config"
 	"golang.org/x/sys/unix"
 )
 
+type StdConn struct {
+	*net.UDPConn
+	isV4  bool
+	sysFd uintptr
+	l     *logrus.Logger
+}
+
+var _ Conn = &StdConn{}
+
 func NewListener(l *logrus.Logger, ip netip.Addr, port int, multi bool, batch int) (Conn, error) {
-	return NewGenericListener(l, ip, port, multi, batch)
+	lc := NewListenConfig(multi)
+	pc, err := lc.ListenPacket(context.TODO(), "udp", net.JoinHostPort(ip.String(), fmt.Sprintf("%v", port)))
+	if err != nil {
+		return nil, err
+	}
+
+	if uc, ok := pc.(*net.UDPConn); ok {
+		c := &StdConn{UDPConn: uc, l: l}
+
+		rc, err := uc.SyscallConn()
+		if err != nil {
+			return nil, fmt.Errorf("failed to open udp socket: %w", err)
+		}
+
+		err = rc.Control(func(fd uintptr) {
+			c.sysFd = fd
+		})
+		if err != nil {
+			return nil, fmt.Errorf("failed to get udp fd: %w", err)
+		}
+
+		la, err := c.LocalAddr()
+		if err != nil {
+			return nil, err
+		}
+		c.isV4 = la.Addr().Is4()
+
+		return c, nil
+	}
+
+	return nil, fmt.Errorf("unexpected PacketConn: %T %#v", pc, pc)
 }
 
 func NewListenConfig(multi bool) net.ListenConfig {
@@ -43,16 +85,116 @@ func NewListenConfig(multi bool) net.ListenConfig {
 	}
 }
 
-func (u *GenericConn) Rebind() error {
-	rc, err := u.UDPConn.SyscallConn()
-	if err != nil {
-		return err
+//go:linkname sendto golang.org/x/sys/unix.sendto
+//go:noescape
+func sendto(s int, buf []byte, flags int, to unsafe.Pointer, addrlen int32) (err error)
+
+func (u *StdConn) WriteTo(b []byte, ap netip.AddrPort) error {
+	var sa unsafe.Pointer
+	var addrLen int32
+
+	if u.isV4 {
+		if ap.Addr().Is6() {
+			return ErrInvalidIPv6RemoteForSocket
+		}
+
+		var rsa unix.RawSockaddrInet6
+		rsa.Family = unix.AF_INET6
+		rsa.Addr = ap.Addr().As16()
+		binary.BigEndian.PutUint16((*[2]byte)(unsafe.Pointer(&rsa.Port))[:], ap.Port())
+		sa = unsafe.Pointer(&rsa)
+		addrLen = syscall.SizeofSockaddrInet4
+	} else {
+		var rsa unix.RawSockaddrInet6
+		rsa.Family = unix.AF_INET6
+		rsa.Addr = ap.Addr().As16()
+		binary.BigEndian.PutUint16((*[2]byte)(unsafe.Pointer(&rsa.Port))[:], ap.Port())
+		sa = unsafe.Pointer(&rsa)
+		addrLen = syscall.SizeofSockaddrInet6
 	}
 
-	return rc.Control(func(fd uintptr) {
-		err := syscall.SetsockoptInt(int(fd), unix.IPPROTO_IPV6, unix.IPV6_BOUND_IF, 0)
-		if err != nil {
-			u.l.WithError(err).Error("Failed to rebind udp socket")
+	// Golang stdlib doesn't handle EAGAIN correctly in some situations so we do writes ourselves
+	// See https://github.com/golang/go/issues/73919
+	for {
+		//_, _, err := unix.Syscall6(unix.SYS_SENDTO, u.sysFd, uintptr(unsafe.Pointer(&b[0])), uintptr(len(b)), 0, sa, addrLen)
+		err := sendto(int(u.sysFd), b, 0, sa, addrLen)
+		if err == nil {
+			// Written, get out before the error handling
+			return nil
 		}
-	})
+
+		if errors.Is(err, syscall.EINTR) {
+			// Write was interrupted, retry
+			continue
+		}
+
+		if errors.Is(err, syscall.EAGAIN) {
+			return &net.OpError{Op: "sendto", Err: unix.EWOULDBLOCK}
+		}
+
+		if errors.Is(err, syscall.EBADF) {
+			return net.ErrClosed
+		}
+
+		return &net.OpError{Op: "sendto", Err: err}
+	}
+}
+
+func (u *StdConn) LocalAddr() (netip.AddrPort, error) {
+	a := u.UDPConn.LocalAddr()
+
+	switch v := a.(type) {
+	case *net.UDPAddr:
+		addr, ok := netip.AddrFromSlice(v.IP)
+		if !ok {
+			return netip.AddrPort{}, fmt.Errorf("LocalAddr returned invalid IP address: %s", v.IP)
+		}
+		return netip.AddrPortFrom(addr, uint16(v.Port)), nil
+
+	default:
+		return netip.AddrPort{}, fmt.Errorf("LocalAddr returned: %#v", a)
+	}
+}
+
+func (u *StdConn) ReloadConfig(c *config.C) {
+	// TODO
+}
+
+func NewUDPStatsEmitter(udpConns []Conn) func() {
+	// No UDP stats for non-linux
+	return func() {}
+}
+
+func (u *StdConn) ListenOut(r EncReader) {
+	buffer := make([]byte, MTU)
+
+	for {
+		// Just read one packet at a time
+		n, rua, err := u.ReadFromUDPAddrPort(buffer)
+		if err != nil {
+			if errors.Is(err, net.ErrClosed) {
+				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
+				return
+			}
+
+			u.l.WithError(err).Error("unexpected udp socket receive error")
+		}
+
+		r(netip.AddrPortFrom(rua.Addr().Unmap(), rua.Port()), buffer[:n])
+	}
+}
+
+func (u *StdConn) Rebind() error {
+	var err error
+	if u.isV4 {
+		err = syscall.SetsockoptInt(int(u.sysFd), syscall.IPPROTO_IP, syscall.IP_BOUND_IF, 0)
+	} else {
+		err = syscall.SetsockoptInt(int(u.sysFd), syscall.IPPROTO_IPV6, syscall.IPV6_BOUND_IF, 0)
+	}
+
+	if err != nil {
+		u.l.WithError(err).Error("Failed to rebind udp socket")
+	}
+
+	return nil
 }

udp/udp_generic.go

@@ -1,6 +1,7 @@
-//go:build (!linux || android) && !e2e_testing
+//go:build (!linux || android) && !e2e_testing && !darwin
 // +build !linux android
 // +build !e2e_testing
+// +build !darwin
 
 // udp_generic implements the nebula UDP interface in pure Go stdlib. This
 // means it can be used on platforms like Darwin and Windows.

udp/udp_linux.go

@@ -221,7 +221,7 @@ func (u *StdConn) writeTo6(b []byte, ip netip.AddrPort) error {
 
 func (u *StdConn) writeTo4(b []byte, ip netip.AddrPort) error {
 	if !ip.Addr().Is4() {
-		return fmt.Errorf("Listener is IPv4, but writing to IPv6 remote")
+		return ErrInvalidIPv6RemoteForSocket
 	}
 
 	var rsa unix.RawSockaddrInet4

udp/udp_raw.go

@@ -0,0 +1,16 @@
+package udp
+
+import mathrand "math/rand"
+
+type SendPortGetter interface {
+	// UDPSendPort returns the port to use
+	UDPSendPort(maxPort int) uint16
+}
+
+type randomSendPort struct{}
+
+func (randomSendPort) UDPSendPort(maxPort int) uint16 {
+	return uint16(mathrand.Intn(maxPort))
+}
+
+var RandomSendPort = randomSendPort{}

udp/udp_raw_linux.go

@@ -0,0 +1,191 @@
+//go:build !android && !e2e_testing
+// +build !android,!e2e_testing
+
+package udp
+
+import (
+	"encoding/binary"
+	"fmt"
+	"net"
+	"net/netip"
+	"syscall"
+	"unsafe"
+
+	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/config"
+	"golang.org/x/net/ipv4"
+	"golang.org/x/sys/unix"
+)
+
+// RawOverhead is the number of bytes that need to be reserved at the start of
+// the raw bytes passed to (*RawConn).WriteTo. This is used by WriteTo to prefix
+// the IP and UDP headers.
+const RawOverhead = 28
+
+type RawConn struct {
+	sysFd    int
+	basePort uint16
+	l        *logrus.Logger
+}
+
+func NewRawConn(l *logrus.Logger, ip string, port int, basePort uint16) (*RawConn, error) {
+	syscall.ForkLock.RLock()
+	// With IPPROTO_UDP, the linux kernel tries to deliver every UDP packet
+	// received in the system to our socket. This constantly overflows our
+	// buffer and marks our socket as having dropped packets. This makes the
+	// stats on the socket useless.
+	//
+	// In contrast, IPPROTO_RAW is not delivered any packets and thus our read
+	// buffer will not fill up and mark as having dropped packets. The only
+	// difference is that we have to assemble the IP header as well, but this
+	// is fairly easy since Linux does the checksum for us.
+	//
+	// TODO: How to get this working with Inet6 correctly? I was having issues
+	// with the source address when testing before, probably need to `bind(2)`?
+	fd, err := unix.Socket(unix.AF_INET, unix.SOCK_RAW, unix.IPPROTO_RAW)
+	if err == nil {
+		unix.CloseOnExec(fd)
+	}
+	syscall.ForkLock.RUnlock()
+	if err != nil {
+		return nil, err
+	}
+
+	// We only want to send, not recv. This will hopefully help the kernel avoid
+	// wasting time on us
+	if err = unix.SetsockoptInt(fd, unix.SOL_SOCKET, unix.SO_RCVBUF, 0); err != nil {
+		return nil, fmt.Errorf("unable to set SO_RCVBUF: %s", err)
+	}
+
+	var lip [16]byte
+	copy(lip[:], net.ParseIP(ip))
+
+	// TODO do we need to `bind(2)` so that we send from the correct address/interface?
+	if err = unix.Bind(fd, &unix.SockaddrInet6{Addr: lip, Port: port}); err != nil {
+		return nil, fmt.Errorf("unable to bind to socket: %s", err)
+	}
+
+	return &RawConn{
+		sysFd:    fd,
+		basePort: basePort,
+		l:        l,
+	}, nil
+}
+
+// WriteTo must be called with raw leaving the first `udp.RawOverhead` bytes empty,
+// for the IP/UDP headers.
+func (u *RawConn) WriteTo(raw []byte, fromPort uint16, ip netip.AddrPort) error {
+	var rsa unix.RawSockaddrInet4
+	rsa.Family = unix.AF_INET
+	rsa.Addr = ip.Addr().As4()
+
+	totalLen := len(raw)
+	udpLen := totalLen - ipv4.HeaderLen
+
+	// IP header
+	raw[0] = byte(ipv4.Version<<4 | (ipv4.HeaderLen >> 2 & 0x0f))
+	raw[1] = 0 // tos
+	binary.BigEndian.PutUint16(raw[2:4], uint16(totalLen))
+	binary.BigEndian.PutUint16(raw[4:6], 0)   // id (linux does it for us)
+	binary.BigEndian.PutUint16(raw[6:8], 0)   // frag options
+	raw[8] = byte(64)                         // ttl
+	raw[9] = byte(17)                         // protocol
+	binary.BigEndian.PutUint16(raw[10:12], 0) // checksum (linux does it for us)
+	binary.BigEndian.PutUint32(raw[12:16], 0) // src (linux does it for us)
+	copy(raw[16:20], rsa.Addr[:])             // dst
+
+	// UDP header
+	fromPort = u.basePort + fromPort
+	binary.BigEndian.PutUint16(raw[20:22], uint16(fromPort))  // src port
+	binary.BigEndian.PutUint16(raw[22:24], uint16(ip.Port())) // dst port
+	binary.BigEndian.PutUint16(raw[24:26], uint16(udpLen))    // UDP length
+	binary.BigEndian.PutUint16(raw[26:28], 0)                 // checksum (optional)
+
+	for {
+		_, _, err := unix.Syscall6(
+			unix.SYS_SENDTO,
+			uintptr(u.sysFd),
+			uintptr(unsafe.Pointer(&raw[0])),
+			uintptr(len(raw)),
+			uintptr(0),
+			uintptr(unsafe.Pointer(&rsa)),
+			uintptr(unix.SizeofSockaddrInet4),
+		)
+
+		if err != 0 {
+			return &net.OpError{Op: "sendto", Err: err}
+		}
+
+		//TODO: handle incomplete writes
+
+		return nil
+	}
+}
+
+func (u *RawConn) ReloadConfig(c *config.C) {
+	b := c.GetInt("listen.write_buffer", 0)
+	if b <= 0 {
+		return
+	}
+
+	if err := u.SetSendBuffer(b); err != nil {
+		u.l.WithError(err).Error("Failed to set listen.write_buffer")
+		return
+	}
+
+	s, err := u.GetSendBuffer()
+	if err != nil {
+		u.l.WithError(err).Warn("Failed to get listen.write_buffer")
+		return
+	}
+
+	u.l.WithField("size", s).Info("listen.write_buffer was set")
+}
+
+func (u *RawConn) SetSendBuffer(n int) error {
+	return unix.SetsockoptInt(u.sysFd, unix.SOL_SOCKET, unix.SO_SNDBUFFORCE, n)
+}
+
+func (u *RawConn) GetSendBuffer() (int, error) {
+	return unix.GetsockoptInt(u.sysFd, unix.SOL_SOCKET, unix.SO_SNDBUF)
+}
+
+func (u *RawConn) getMemInfo(meminfo *[unix.SK_MEMINFO_VARS]uint32) error {
+	var vallen uint32 = 4 * unix.SK_MEMINFO_VARS
+	_, _, err := unix.Syscall6(unix.SYS_GETSOCKOPT, uintptr(u.sysFd), uintptr(unix.SOL_SOCKET), uintptr(unix.SO_MEMINFO), uintptr(unsafe.Pointer(meminfo)), uintptr(unsafe.Pointer(&vallen)), 0)
+	if err != 0 {
+		return err
+	}
+	return nil
+}
+
+func NewRawStatsEmitter(rawConn *RawConn) func() {
+	// Check if our kernel supports SO_MEMINFO before registering the gauges
+	var gauges [unix.SK_MEMINFO_VARS]metrics.Gauge
+	var meminfo [unix.SK_MEMINFO_VARS]uint32
+	if err := rawConn.getMemInfo(&meminfo); err == nil {
+		gauges = [unix.SK_MEMINFO_VARS]metrics.Gauge{
+			metrics.GetOrRegisterGauge("raw.rmem_alloc", nil),
+			metrics.GetOrRegisterGauge("raw.rcvbuf", nil),
+			metrics.GetOrRegisterGauge("raw.wmem_alloc", nil),
+			metrics.GetOrRegisterGauge("raw.sndbuf", nil),
+			metrics.GetOrRegisterGauge("raw.fwd_alloc", nil),
+			metrics.GetOrRegisterGauge("raw.wmem_queued", nil),
+			metrics.GetOrRegisterGauge("raw.optmem", nil),
+			metrics.GetOrRegisterGauge("raw.backlog", nil),
+			metrics.GetOrRegisterGauge("raw.drops", nil),
+		}
+	} else {
+		// return no-op because we don't support SO_MEMINFO
+		return func() {}
+	}
+
+	return func() {
+		if err := rawConn.getMemInfo(&meminfo); err == nil {
+			for j := 0; j < unix.SK_MEMINFO_VARS; j++ {
+				gauges[j].Update(int64(meminfo[j]))
+			}
+		}
+	}
+}

udp/udp_raw_unsupported.go

@@ -0,0 +1,29 @@
+//go:build !linux || android || e2e_testing
+// +build !linux android e2e_testing
+
+package udp
+
+import (
+	"fmt"
+	"net/netip"
+	"runtime"
+
+	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/config"
+)
+
+const RawOverhead = 0
+
+type RawConn struct{}
+
+func NewRawConn(l *logrus.Logger, ip string, port int, basePort uint16) (*RawConn, error) {
+	return nil, fmt.Errorf("multiport tx is not supported on %s", runtime.GOOS)
+}
+
+func (u *RawConn) WriteTo(raw []byte, fromPort uint16, addr netip.AddrPort) error {
+	return fmt.Errorf("multiport tx is not supported on %s", runtime.GOOS)
+}
+
+func (u *RawConn) ReloadConfig(c *config.C) {}
+
+func NewRawStatsEmitter(rawConn *RawConn) func() { return func() {} }

udp/udp_rio_windows.go

@@ -92,6 +92,25 @@ func (u *RIOConn) bind(sa windows.Sockaddr) error {
 	// Enable v4 for this socket
 	syscall.SetsockoptInt(syscall.Handle(u.sock), syscall.IPPROTO_IPV6, syscall.IPV6_V6ONLY, 0)
 
+	// Disable reporting of PORT_UNREACHABLE and NET_UNREACHABLE errors from the UDP socket receive call.
+	// These errors are returned on Windows during UDP receives based on the receipt of ICMP packets. Disable
+	// the UDP receive error returns with these ioctl calls.
+	ret := uint32(0)
+	flag := uint32(0)
+	size := uint32(unsafe.Sizeof(flag))
+	err = syscall.WSAIoctl(syscall.Handle(u.sock), syscall.SIO_UDP_CONNRESET, (*byte)(unsafe.Pointer(&flag)), size, nil, 0, &ret, nil, 0)
+	if err != nil {
+		return err
+	}
+	ret = 0
+	flag = 0
+	size = uint32(unsafe.Sizeof(flag))
+	SIO_UDP_NETRESET := uint32(syscall.IOC_IN | syscall.IOC_VENDOR | 15)
+	err = syscall.WSAIoctl(syscall.Handle(u.sock), SIO_UDP_NETRESET, (*byte)(unsafe.Pointer(&flag)), size, nil, 0, &ret, nil, 0)
+	if err != nil {
+		return err
+	}
+
 	err = u.rx.Open()
 	if err != nil {
 		return err
@@ -122,8 +141,12 @@ func (u *RIOConn) ListenOut(r EncReader) {
 		// Just read one packet at a time
 		n, rua, err := u.receive(buffer)
 		if err != nil {
-			u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-			return
+			if errors.Is(err, net.ErrClosed) {
+				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
+				return
+			}
+			u.l.WithError(err).Error("unexpected udp socket receive error")
+			continue
 		}
 
 		r(netip.AddrPortFrom(netip.AddrFrom16(rua.Addr).Unmap(), (rua.Port>>8)|((rua.Port&0xff)<<8)), buffer[:n])