v1.7.2 (#887)

Update CHANGELOG for Nebula v1.7.2

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -0,0 +1,57 @@
+name: "\U0001F41B Bug Report"
+description: Report an issue or possible bug
+title: "\U0001F41B BUG:"
+labels: []
+assignees: []
+body:
+  - type: markdown
+    attributes:
+      value: |
+        ### Thank you for taking the time to file a bug report!
+        
+        Please fill out this form as completely as possible.
+
+  - type: input
+    id: version
+    attributes:
+      label: What version of `nebula` are you using?
+      placeholder: 0.0.0
+    validations:
+      required: true
+
+  - type: input
+    id: os
+    attributes:
+      label: What operating system are you using?
+      description: iOS and Android specific issues belong in the [mobile_nebula](https://github.com/DefinedNet/mobile_nebula) repo.
+      placeholder: Linux, Mac, Windows
+    validations:
+      required: true
+
+  - type: textarea
+    id: description
+    attributes:
+      label: Describe the Bug
+      description: A clear and concise description of what the bug is.
+    validations:
+      required: true
+
+  - type: textarea
+    id: logs
+    attributes:
+      label: Logs from affected hosts
+      description: |
+        Provide logs from all affected hosts during the time of the issue.
+        Improve formatting by using <code>```</code> at the beginning and end of each log block.
+    validations:
+      required: false
+
+  - type: textarea
+    id: configs
+    attributes:
+      label: Config files from affected hosts
+      description: |
+        Provide config files for all affected hosts.
+        Improve formatting by using <code>```</code> at the beginning and end of each config file.
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,13 @@
+blank_issues_enabled: true
+contact_links:
+  - name: 📘 Documentation
+    url: https://nebula.defined.net/docs/
+    about: Review documentation.
+
+  - 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!'

.github/workflows/gofmt.yml

@@ -14,31 +14,21 @@ jobs:
     runs-on: ubuntu-latest
     steps:
 
-    - name: Set up Go 1.16
-      uses: actions/setup-go@v1
-      with:
-        go-version: 1.16
-      id: go
+    - uses: actions/checkout@v3
 
-    - name: Check out code into the Go module directory
-      uses: actions/checkout@v1
-
-    - uses: actions/cache@v1
+    - uses: actions/setup-go@v4
       with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-gofmt1.16-${{ hashFiles('**/go.sum') }}
-        restore-keys: |
-          ${{ runner.os }}-gofmt1.16-
+        go-version-file: 'go.mod'
+        check-latest: true
 
     - name: Install goimports
       run: |
-        go get golang.org/x/tools/cmd/goimports
-        go build golang.org/x/tools/cmd/goimports
+        go install golang.org/x/tools/cmd/goimports@latest
 
     - name: gofmt
       run: |
-        if [ "$(find . -iname '*.go' | grep -v '\.pb\.go$' | xargs ./goimports -l)" ]
+        if [ "$(find . -iname '*.go' | grep -v '\.pb\.go$' | xargs goimports -l)" ]
         then
-          find . -iname '*.go' | grep -v '\.pb\.go$' | xargs ./goimports -d
+          find . -iname '*.go' | grep -v '\.pb\.go$' | xargs goimports -d
           exit 1
         fi

.github/workflows/release.yml

@@ -10,13 +10,12 @@ jobs:
     name: Build Linux All
     runs-on: ubuntu-latest
     steps:
-      - name: Set up Go 1.16
-        uses: actions/setup-go@v1
-        with:
-          go-version: 1.16
+      - uses: actions/checkout@v3
 
-      - name: Checkout code
-        uses: actions/checkout@v2
+      - uses: actions/setup-go@v4
+        with:
+          go-version-file: 'go.mod'
+          check-latest: true
 
       - name: Build
         run: |
@@ -25,84 +24,106 @@ jobs:
           mv build/*.tar.gz release
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v1
+        uses: actions/upload-artifact@v2
         with:
           name: linux-latest
           path: release
 
   build-windows:
-    name: Build Windows amd64
+    name: Build Windows
     runs-on: windows-latest
     steps:
-      - name: Set up Go 1.16
-        uses: actions/setup-go@v1
-        with:
-          go-version: 1.16
+      - uses: actions/checkout@v3
 
-      - name: Checkout code
-        uses: actions/checkout@v2
+      - uses: actions/setup-go@v4
+        with:
+          go-version-file: 'go.mod'
+          check-latest: true
 
       - name: Build
         run: |
           echo $Env:GITHUB_REF.Substring(11)
-          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\nebula.exe ./cmd/nebula-service
-          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\nebula-cert.exe ./cmd/nebula-cert
+          mkdir build\windows-amd64
+          $Env:GOARCH = "amd64"
+          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\windows-amd64\nebula.exe ./cmd/nebula-service
+          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\windows-amd64\nebula-cert.exe ./cmd/nebula-cert
+          mkdir build\windows-arm64
+          $Env:GOARCH = "arm64"
+          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\windows-arm64\nebula.exe ./cmd/nebula-service
+          go build -trimpath -ldflags "-X main.Build=$($Env:GITHUB_REF.Substring(11))" -o build\windows-arm64\nebula-cert.exe ./cmd/nebula-cert
+          mkdir build\dist\windows
+          mv dist\windows\wintun build\dist\windows\
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v1
+        uses: actions/upload-artifact@v2
         with:
           name: windows-latest
           path: build
 
   build-darwin:
-    name: Build Darwin amd64
-    runs-on: macOS-latest
+    name: Build Universal Darwin
+    env:
+      HAS_SIGNING_CREDS: ${{ secrets.AC_USERNAME != '' }}
+    runs-on: macos-11
     steps:
-      - name: Set up Go 1.16
-        uses: actions/setup-go@v1
+      - uses: actions/checkout@v3
+
+      - uses: actions/setup-go@v4
         with:
-          go-version: 1.16
+          go-version-file: 'go.mod'
+          check-latest: true
 
-      - name: Checkout code
-        uses: actions/checkout@v2
+      - name: Import certificates
+        if: env.HAS_SIGNING_CREDS == 'true'
+        uses: Apple-Actions/import-codesign-certs@v1
+        with:
+          p12-file-base64: ${{ secrets.APPLE_DEVELOPER_CERTIFICATE_P12_BASE64 }}
+          p12-password: ${{ secrets.APPLE_DEVELOPER_CERTIFICATE_PASSWORD }}
 
-      - name: Build
+      - name: Build, sign, and notarize
+        env:
+          AC_USERNAME: ${{ secrets.AC_USERNAME }}
+          AC_PASSWORD: ${{ secrets.AC_PASSWORD }}
         run: |
-          make BUILD_NUMBER="${GITHUB_REF#refs/tags/v}" service build/nebula-darwin-amd64.tar.gz
-          make BUILD_NUMBER="${GITHUB_REF#refs/tags/v}" service build/nebula-darwin-arm64.tar.gz
+          rm -rf release
           mkdir release
-          mv build/*.tar.gz release
+          make BUILD_NUMBER="${GITHUB_REF#refs/tags/v}" service build/darwin-amd64/nebula build/darwin-amd64/nebula-cert
+          make BUILD_NUMBER="${GITHUB_REF#refs/tags/v}" service build/darwin-arm64/nebula build/darwin-arm64/nebula-cert
+          lipo -create -output ./release/nebula ./build/darwin-amd64/nebula ./build/darwin-arm64/nebula
+          lipo -create -output ./release/nebula-cert ./build/darwin-amd64/nebula-cert ./build/darwin-arm64/nebula-cert
+
+          if [ -n "$AC_USERNAME" ]; then
+            codesign -s "10BC1FDDEB6CE753550156C0669109FAC49E4D1E" -f -v --timestamp --options=runtime -i "net.defined.nebula" ./release/nebula
+            codesign -s "10BC1FDDEB6CE753550156C0669109FAC49E4D1E" -f -v --timestamp --options=runtime -i "net.defined.nebula-cert" ./release/nebula-cert
+          fi
+
+          zip -j release/nebula-darwin.zip release/nebula-cert release/nebula
+
+          if [ -n "$AC_USERNAME" ]; then
+            xcrun notarytool submit ./release/nebula-darwin.zip --team-id "576H3XS7FP" --apple-id "$AC_USERNAME" --password "$AC_PASSWORD" --wait
+          fi
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v1
+        uses: actions/upload-artifact@v2
         with:
           name: darwin-latest
-          path: release
+          path: ./release/*
 
   release:
     name: Create and Upload Release
     needs: [build-linux, build-darwin, build-windows]
     runs-on: ubuntu-latest
     steps:
-      - name: Download Linux artifacts
-        uses: actions/download-artifact@v1
-        with:
-          name: linux-latest
-
-      - name: Download Darwin artifacts
-        uses: actions/download-artifact@v1
-        with:
-          name: darwin-latest
-
-      - name: Download Windows artifacts
-        uses: actions/download-artifact@v1
-        with:
-          name: windows-latest
+      - name: Download artifacts
+        uses: actions/download-artifact@v2
 
       - name: Zip Windows
         run: |
           cd windows-latest
-          zip nebula-windows-amd64.zip nebula.exe nebula-cert.exe
+          cp windows-amd64/* .
+          zip -r nebula-windows-amd64.zip nebula.exe nebula-cert.exe dist
+          cp windows-arm64/* .
+          zip -r nebula-windows-arm64.zip nebula.exe nebula-cert.exe dist
 
       - name: Create sha256sum
         run: |
@@ -112,15 +133,23 @@ jobs:
               cd $dir
               if [ "$dir" = windows-latest ]
               then
-                sha256sum <nebula.exe | sed 's=-$=nebula-windows-amd64.zip/nebula.exe='
-                sha256sum <nebula-cert.exe | sed 's=-$=nebula-windows-amd64.zip/nebula-cert.exe='
+                sha256sum <windows-amd64/nebula.exe | sed 's=-$=nebula-windows-amd64.zip/nebula.exe='
+                sha256sum <windows-amd64/nebula-cert.exe | sed 's=-$=nebula-windows-amd64.zip/nebula-cert.exe='
+                sha256sum <windows-arm64/nebula.exe | sed 's=-$=nebula-windows-arm64.zip/nebula.exe='
+                sha256sum <windows-arm64/nebula-cert.exe | sed 's=-$=nebula-windows-arm64.zip/nebula-cert.exe='
                 sha256sum nebula-windows-amd64.zip
+                sha256sum nebula-windows-arm64.zip
+              elif [ "$dir" = darwin-latest ]
+              then
+                sha256sum <nebula-darwin.zip | sed 's=-$=nebula-darwin.zip='
+                sha256sum <nebula | sed 's=-$=nebula-darwin.zip/nebula='
+                sha256sum <nebula-cert | sed 's=-$=nebula-darwin.zip/nebula-cert='
               else
-                  for v in *.tar.gz
-                  do
-                    sha256sum $v
-                    tar zxf $v --to-command='sh -c "sha256sum | sed s=-$='$v'/$TAR_FILENAME="'
-                  done
+                for v in *.tar.gz
+                do
+                  sha256sum $v
+                  tar zxf $v --to-command='sh -c "sha256sum | sed s=-$='$v'/$TAR_FILENAME="'
+                done
               fi
             )
           done | sort -k 2 >SHASUM256.txt
@@ -150,25 +179,15 @@ jobs:
           asset_name: SHASUM256.txt
           asset_content_type: text/plain
 
-      - name: Upload darwin-amd64
+      - name: Upload darwin zip
         uses: actions/upload-release-asset@v1.0.1
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         with:
           upload_url: ${{ steps.create_release.outputs.upload_url }}
-          asset_path: ./darwin-latest/nebula-darwin-amd64.tar.gz
-          asset_name: nebula-darwin-amd64.tar.gz
-          asset_content_type: application/gzip
-
-      - name: Upload darwin-arm64
-        uses: actions/upload-release-asset@v1.0.1
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          upload_url: ${{ steps.create_release.outputs.upload_url }}
-          asset_path: ./darwin-latest/nebula-darwin-arm64.tar.gz
-          asset_name: nebula-darwin-arm64.tar.gz
-          asset_content_type: application/gzip
+          asset_path: ./darwin-latest/nebula-darwin.zip
+          asset_name: nebula-darwin.zip
+          asset_content_type: application/zip
 
       - name: Upload windows-amd64
         uses: actions/upload-release-asset@v1.0.1
@@ -180,6 +199,16 @@ jobs:
           asset_name: nebula-windows-amd64.zip
           asset_content_type: application/zip
 
+      - name: Upload windows-arm64
+        uses: actions/upload-release-asset@v1.0.1
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          upload_url: ${{ steps.create_release.outputs.upload_url }}
+          asset_path: ./windows-latest/nebula-windows-arm64.zip
+          asset_name: nebula-windows-arm64.zip
+          asset_content_type: application/zip
+
       - name: Upload linux-amd64
         uses: actions/upload-release-asset@v1.0.1
         env:
@@ -300,6 +329,16 @@ jobs:
           asset_name: nebula-linux-mips-softfloat.tar.gz
           asset_content_type: application/gzip
 
+      - name: Upload linux-riscv64
+        uses: actions/upload-release-asset@v1.0.1
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          upload_url: ${{ steps.create_release.outputs.upload_url }}
+          asset_path: ./linux-latest/nebula-linux-riscv64.tar.gz
+          asset_name: nebula-linux-riscv64.tar.gz
+          asset_content_type: application/gzip
+
       - name: Upload freebsd-amd64
         uses: actions/upload-release-asset@v1.0.1
         env:

.github/workflows/smoke.yml

@@ -18,21 +18,12 @@ jobs:
     runs-on: ubuntu-latest
     steps:
 
-    - name: Set up Go 1.16
-      uses: actions/setup-go@v1
-      with:
-        go-version: 1.16
-      id: go
+    - uses: actions/checkout@v3
 
-    - name: Check out code into the Go module directory
-      uses: actions/checkout@v1
-
-    - uses: actions/cache@v1
+    - uses: actions/setup-go@v4
       with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-go1.16-${{ hashFiles('**/go.sum') }}
-        restore-keys: |
-          ${{ runner.os }}-go1.16-
+        go-version-file: 'go.mod'
+        check-latest: true
 
     - name: build
       run: make bin-docker
@@ -45,4 +36,20 @@ jobs:
       working-directory: ./.github/workflows/smoke
       run: ./smoke.sh
 
+    - name: setup relay docker image
+      working-directory: ./.github/workflows/smoke
+      run: ./build-relay.sh
+
+    - name: run smoke relay
+      working-directory: ./.github/workflows/smoke
+      run: ./smoke-relay.sh
+
+    - name: setup docker image for P256
+      working-directory: ./.github/workflows/smoke
+      run: NAME="smoke-p256" CURVE=P256 ./build.sh
+
+    - name: run smoke-p256
+      working-directory: ./.github/workflows/smoke
+      run: NAME="smoke-p256" ./smoke.sh
+
     timeout-minutes: 10

.github/workflows/smoke/Dockerfile

@@ -1,4 +1,6 @@
-FROM debian:buster
+FROM ubuntu:jammy
+
+RUN apt-get update && apt-get install -y iputils-ping ncat tcpdump
 
 ADD ./build /nebula
 

.github/workflows/smoke/build-relay.sh

@@ -0,0 +1,44 @@
+#!/bin/sh
+
+set -e -x
+
+rm -rf ./build
+mkdir ./build
+
+(
+    cd build
+
+    cp ../../../../build/linux-amd64/nebula .
+    cp ../../../../build/linux-amd64/nebula-cert .
+
+    HOST="lighthouse1" AM_LIGHTHOUSE=true ../genconfig.sh >lighthouse1.yml <<EOF
+relay:
+  am_relay: true
+EOF
+
+    export LIGHTHOUSES="192.168.100.1 172.17.0.2:4242"
+    export REMOTE_ALLOW_LIST='{"172.17.0.4/32": false, "172.17.0.5/32": false}'
+
+    HOST="host2" ../genconfig.sh >host2.yml <<EOF
+relay:
+  relays:
+    - 192.168.100.1
+EOF
+
+    export REMOTE_ALLOW_LIST='{"172.17.0.3/32": false}'
+
+    HOST="host3" ../genconfig.sh >host3.yml
+
+    HOST="host4" ../genconfig.sh >host4.yml <<EOF
+relay:
+  use_relays: false
+EOF
+
+    ../../../../nebula-cert ca -name "Smoke Test"
+    ../../../../nebula-cert sign -name "lighthouse1" -groups "lighthouse,lighthouse1" -ip "192.168.100.1/24"
+    ../../../../nebula-cert sign -name "host2" -groups "host,host2" -ip "192.168.100.2/24"
+    ../../../../nebula-cert sign -name "host3" -groups "host,host3" -ip "192.168.100.3/24"
+    ../../../../nebula-cert sign -name "host4" -groups "host,host4" -ip "192.168.100.4/24"
+)
+
+sudo docker build -t nebula:smoke-relay .

.github/workflows/smoke/build.sh

@@ -29,11 +29,11 @@ mkdir ./build
         OUTBOUND='[{"port": "any", "proto": "icmp", "group": "lighthouse"}]' \
         ../genconfig.sh >host4.yml
 
-    ../../../../nebula-cert ca -name "Smoke Test"
+    ../../../../nebula-cert ca -curve "${CURVE:-25519}" -name "Smoke Test"
     ../../../../nebula-cert sign -name "lighthouse1" -groups "lighthouse,lighthouse1" -ip "192.168.100.1/24"
     ../../../../nebula-cert sign -name "host2" -groups "host,host2" -ip "192.168.100.2/24"
     ../../../../nebula-cert sign -name "host3" -groups "host,host3" -ip "192.168.100.3/24"
     ../../../../nebula-cert sign -name "host4" -groups "host,host4" -ip "192.168.100.4/24"
 )
 
-sudo docker build -t nebula:smoke .
+sudo docker build -t "nebula:${NAME:-smoke}" .

.github/workflows/smoke/genconfig.sh

@@ -40,6 +40,7 @@ pki:
 lighthouse:
   am_lighthouse: ${AM_LIGHTHOUSE:-false}
   hosts: $(lighthouse_hosts)
+  remote_allow_list: ${REMOTE_ALLOW_LIST}
 
 listen:
   host: 0.0.0.0
@@ -49,6 +50,10 @@ tun:
   dev: ${TUN_DEV:-nebula1}
 
 firewall:
+  inbound_action: reject
+  outbound_action: reject
   outbound: ${OUTBOUND:-$FIREWALL_ALL}
   inbound: ${INBOUND:-$FIREWALL_ALL}
+
+$(test -t 0 || cat)
 EOF

.github/workflows/smoke/smoke-relay.sh

@@ -0,0 +1,85 @@
+#!/bin/bash
+
+set -e -x
+
+set -o pipefail
+
+mkdir -p logs
+
+cleanup() {
+    echo
+    echo " *** cleanup"
+    echo
+
+    set +e
+    if [ "$(jobs -r)" ]
+    then
+        sudo docker kill lighthouse1 host2 host3 host4
+    fi
+}
+
+trap cleanup EXIT
+
+sudo docker run --name lighthouse1 --rm nebula:smoke-relay -config lighthouse1.yml -test
+sudo docker run --name host2 --rm nebula:smoke-relay -config host2.yml -test
+sudo docker run --name host3 --rm nebula:smoke-relay -config host3.yml -test
+sudo docker run --name host4 --rm nebula:smoke-relay -config host4.yml -test
+
+sudo docker run --name lighthouse1 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
+sleep 1
+sudo docker run --name host2 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
+sleep 1
+sudo docker run --name host3 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config host3.yml 2>&1 | tee logs/host3 | sed -u 's/^/  [host3]  /' &
+sleep 1
+sudo docker run --name host4 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config host4.yml 2>&1 | tee logs/host4 | sed -u 's/^/  [host4]  /' &
+sleep 1
+
+set +x
+echo
+echo " *** Testing ping from lighthouse1"
+echo
+set -x
+sudo docker exec lighthouse1 ping -c1 192.168.100.2
+sudo docker exec lighthouse1 ping -c1 192.168.100.3
+sudo docker exec lighthouse1 ping -c1 192.168.100.4
+
+set +x
+echo
+echo " *** Testing ping from host2"
+echo
+set -x
+sudo docker exec host2 ping -c1 192.168.100.1
+# Should fail because no relay configured in this direction
+! sudo docker exec host2 ping -c1 192.168.100.3 -w5 || exit 1
+! sudo docker exec host2 ping -c1 192.168.100.4 -w5 || exit 1
+
+set +x
+echo
+echo " *** Testing ping from host3"
+echo
+set -x
+sudo docker exec host3 ping -c1 192.168.100.1
+sudo docker exec host3 ping -c1 192.168.100.2
+sudo docker exec host3 ping -c1 192.168.100.4
+
+set +x
+echo
+echo " *** Testing ping from host4"
+echo
+set -x
+sudo docker exec host4 ping -c1 192.168.100.1
+# Should fail because relays not allowed
+! sudo docker exec host4 ping -c1 192.168.100.2 -w5 || exit 1
+sudo docker exec host4 ping -c1 192.168.100.3
+
+sudo docker exec host4 sh -c 'kill 1'
+sudo docker exec host3 sh -c 'kill 1'
+sudo docker exec host2 sh -c 'kill 1'
+sudo docker exec lighthouse1 sh -c 'kill 1'
+sleep 1
+
+if [ "$(jobs -r)" ]
+then
+    echo "nebula still running after SIGTERM sent" >&2
+    exit 1
+fi

.github/workflows/smoke/smoke.sh

@@ -7,6 +7,10 @@ set -o pipefail
 mkdir -p logs
 
 cleanup() {
+    echo
+    echo " *** cleanup"
+    echo
+
     set +e
     if [ "$(jobs -r)" ]
     then
@@ -16,20 +20,37 @@ cleanup() {
 
 trap cleanup EXIT
 
-sudo docker run --name lighthouse1 --rm nebula:smoke -config lighthouse1.yml -test
-sudo docker run --name host2 --rm nebula:smoke -config host2.yml -test
-sudo docker run --name host3 --rm nebula:smoke -config host3.yml -test
-sudo docker run --name host4 --rm nebula:smoke -config host4.yml -test
+CONTAINER="nebula:${NAME:-smoke}"
 
-sudo docker run --name lighthouse1 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 &
+sudo docker run --name lighthouse1 --rm "$CONTAINER" -config lighthouse1.yml -test
+sudo docker run --name host2 --rm "$CONTAINER" -config host2.yml -test
+sudo docker run --name host3 --rm "$CONTAINER" -config host3.yml -test
+sudo docker run --name host4 --rm "$CONTAINER" -config host4.yml -test
+
+sudo docker run --name lighthouse1 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
 sleep 1
-sudo docker run --name host2 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke -config host2.yml 2>&1 | tee logs/host2 &
+sudo docker run --name host2 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
 sleep 1
-sudo docker run --name host3 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke -config host3.yml 2>&1 | tee logs/host3 &
+sudo docker run --name host3 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host3.yml 2>&1 | tee logs/host3 | sed -u 's/^/  [host3]  /' &
 sleep 1
-sudo docker run --name host4 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke -config host4.yml 2>&1 | tee logs/host4 &
+sudo docker run --name host4 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host4.yml 2>&1 | tee logs/host4 | sed -u 's/^/  [host4]  /' &
 sleep 1
 
+# grab tcpdump pcaps for debugging
+sudo docker exec lighthouse1 tcpdump -i nebula1 -q -w - -U 2>logs/lighthouse1.inside.log >logs/lighthouse1.inside.pcap &
+sudo docker exec lighthouse1 tcpdump -i eth0 -q -w - -U 2>logs/lighthouse1.outside.log >logs/lighthouse1.outside.pcap &
+sudo docker exec host2 tcpdump -i nebula1 -q -w - -U 2>logs/host2.inside.log >logs/host2.inside.pcap &
+sudo docker exec host2 tcpdump -i eth0 -q -w - -U 2>logs/host2.outside.log >logs/host2.outside.pcap &
+sudo docker exec host3 tcpdump -i nebula1 -q -w - -U 2>logs/host3.inside.log >logs/host3.inside.pcap &
+sudo docker exec host3 tcpdump -i eth0 -q -w - -U 2>logs/host3.outside.log >logs/host3.outside.pcap &
+sudo docker exec host4 tcpdump -i nebula1 -q -w - -U 2>logs/host4.inside.log >logs/host4.inside.pcap &
+sudo docker exec host4 tcpdump -i eth0 -q -w - -U 2>logs/host4.outside.log >logs/host4.outside.pcap &
+
+sudo docker exec host2 ncat -nklv 0.0.0.0 2000 &
+sudo docker exec host3 ncat -nklv 0.0.0.0 2000 &
+sudo docker exec host2 ncat -e '/usr/bin/echo host2' -nkluv 0.0.0.0 3000 &
+sudo docker exec host3 ncat -e '/usr/bin/echo host3' -nkluv 0.0.0.0 3000 &
+
 set +x
 echo
 echo " *** Testing ping from lighthouse1"
@@ -47,6 +68,15 @@ sudo docker exec host2 ping -c1 192.168.100.1
 # Should fail because not allowed by host3 inbound firewall
 ! sudo docker exec host2 ping -c1 192.168.100.3 -w5 || exit 1
 
+set +x
+echo
+echo " *** Testing ncat from host2"
+echo
+set -x
+# Should fail because not allowed by host3 inbound firewall
+! sudo docker exec host2 ncat -nzv -w5 192.168.100.3 2000 || exit 1
+! sudo docker exec host2 ncat -nzuv -w5 192.168.100.3 3000 | grep -q host3 || exit 1
+
 set +x
 echo
 echo " *** Testing ping from host3"
@@ -55,6 +85,14 @@ set -x
 sudo docker exec host3 ping -c1 192.168.100.1
 sudo docker exec host3 ping -c1 192.168.100.2
 
+set +x
+echo
+echo " *** Testing ncat from host3"
+echo
+set -x
+sudo docker exec host3 ncat -nzv -w5 192.168.100.2 2000
+sudo docker exec host3 ncat -nzuv -w5 192.168.100.2 3000 | grep -q host2
+
 set +x
 echo
 echo " *** Testing ping from host4"
@@ -65,6 +103,17 @@ sudo docker exec host4 ping -c1 192.168.100.1
 ! sudo docker exec host4 ping -c1 192.168.100.2 -w5 || exit 1
 ! sudo docker exec host4 ping -c1 192.168.100.3 -w5 || exit 1
 
+set +x
+echo
+echo " *** Testing ncat from host4"
+echo
+set -x
+# Should fail because not allowed by host4 outbound firewall
+! sudo docker exec host4 ncat -nzv -w5 192.168.100.2 2000 || exit 1
+! sudo docker exec host4 ncat -nzv -w5 192.168.100.3 2000 || exit 1
+! sudo docker exec host4 ncat -nzuv -w5 192.168.100.2 3000 | grep -q host2 || exit 1
+! sudo docker exec host4 ncat -nzuv -w5 192.168.100.3 3000 | grep -q host3 || exit 1
+
 set +x
 echo
 echo " *** Testing conntrack"
@@ -81,3 +130,9 @@ sudo docker exec host3 sh -c 'kill 1'
 sudo docker exec host2 sh -c 'kill 1'
 sudo docker exec lighthouse1 sh -c 'kill 1'
 sleep 1
+
+if [ "$(jobs -r)" ]
+then
+    echo "nebula still running after SIGTERM sent" >&2
+    exit 1
+fi

.github/workflows/test.yml

@@ -18,54 +18,66 @@ jobs:
     runs-on: ubuntu-latest
     steps:
 
-    - name: Set up Go 1.16
-      uses: actions/setup-go@v1
-      with:
-        go-version: 1.16
-      id: go
+    - uses: actions/checkout@v3
 
-    - name: Check out code into the Go module directory
-      uses: actions/checkout@v1
-
-    - uses: actions/cache@v1
+    - uses: actions/setup-go@v4
       with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-go1.16-${{ hashFiles('**/go.sum') }}
-        restore-keys: |
-          ${{ runner.os }}-go1.16-
+        go-version-file: 'go.mod'
+        check-latest: true
 
     - name: Build
       run: make all
 
+    - name: Vet
+      run: make vet
+
     - name: Test
       run: make test
 
     - name: End 2 end
       run: make e2evv
 
+    - uses: actions/upload-artifact@v3
+      with:
+        name: e2e packet flow
+        path: e2e/mermaid/
+        if-no-files-found: warn
+
+  test-linux-boringcrypto:
+    name: Build and test on linux with boringcrypto
+    runs-on: ubuntu-latest
+    steps:
+
+    - uses: actions/checkout@v3
+
+    - uses: actions/setup-go@v4
+      with:
+        go-version-file: 'go.mod'
+        check-latest: true
+
+    - name: Build
+      run: make bin-boringcrypto
+
+    - name: Test
+      run: make test-boringcrypto
+
+    - name: End 2 end
+      run: make e2evv GOEXPERIMENT=boringcrypto CGO_ENABLED=1
+
   test:
     name: Build and test on ${{ matrix.os }}
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [windows-latest, macOS-latest]
+        os: [windows-latest, macos-11]
     steps:
 
-    - name: Set up Go 1.16
-      uses: actions/setup-go@v1
-      with:
-        go-version: 1.16
-      id: go
+    - uses: actions/checkout@v3
 
-    - name: Check out code into the Go module directory
-      uses: actions/checkout@v1
-
-    - uses: actions/cache@v1
+    - uses: actions/setup-go@v4
       with:
-        path: ~/go/pkg/mod
-        key: ${{ runner.os }}-go1.16-${{ hashFiles('**/go.sum') }}
-        restore-keys: |
-          ${{ runner.os }}-go1.16-
+        go-version-file: 'go.mod'
+        check-latest: true
 
     - name: Build nebula
       run: go build ./cmd/nebula
@@ -73,8 +85,17 @@ jobs:
     - name: Build nebula-cert
       run: go build ./cmd/nebula-cert
 
+    - name: Vet
+      run: make vet
+
     - name: Test
-      run: go test -v ./...
+      run: make test
 
     - name: End 2 end
       run: make e2evv
+
+    - uses: actions/upload-artifact@v3
+      with:
+        name: e2e packet flow
+        path: e2e/mermaid/
+        if-no-files-found: warn

.gitignore

@@ -4,9 +4,14 @@
 /nebula-arm6
 /nebula-darwin
 /nebula.exe
-/cert/*.crt
-/cert/*.key
+/nebula-cert.exe
 /coverage.out
 /cpu.pprof
 /build
 /*.tar.gz
+/e2e/mermaid/
+**.crt
+**.key
+**.pem
+!/examples/quickstart-vagrant/ansible/roles/nebula/files/vagrant-test-ca.key
+!/examples/quickstart-vagrant/ansible/roles/nebula/files/vagrant-test-ca.crt

CHANGELOG.md

@@ -7,6 +7,249 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.7.2] - 2023-06-01
+
+### Fixed
+
+- Fix a freeze during config reload if the `static_host_map` config was changed. (#886)
+
+## [1.7.1] - 2023-05-18
+
+### Fixed
+
+- Fix IPv4 addresses returned by `static_host_map` DNS lookup queries being
+  treated as IPv6 addresses. (#877)
+
+## [1.7.0] - 2023-05-17
+
+### Added
+
+- `nebula-cert ca` now supports encrypting the CA's private key with a
+  passphrase. Pass `-encrypt` in order to be prompted for a passphrase.
+  Encryption is performed using AES-256-GCM and Argon2id for KDF. KDF
+  parameters default to RFC recommendations, but can be overridden via CLI
+  flags `-argon-memory`, `-argon-parallelism`, and `-argon-iterations`. (#386)
+
+- Support for curve P256 and BoringCrypto has been added. See README section
+  "Curve P256 and BoringCrypto" for more details. (#865, #861, #769, #856, #803)
+
+- New firewall rule `local_cidr`. This could be used to filter destinations
+  when using `unsafe_routes`. (#507)
+
+- Add `unsafe_route` option `install`. This controls whether the route is
+  installed in the systems routing table. (#831)
+
+- Add `tun.use_system_route_table` option. Set to true to manage unsafe routes
+  directly on the system route table with gateway routes instead of in Nebula
+  configuration files. This is only supported on Linux. (#839)
+
+- The metric `certificate.ttl_seconds` is now exposed via stats. (#782)
+
+- Add `punchy.respond_delay` option. This allows you to change the delay
+  before attempting punchy.respond. Default is 5 seconds. (#721)
+
+- Added SSH commands to allow the capture of a mutex profile. (#737)
+
+- You can now set `lighthouse.calculated_remotes` to make it possible to do
+  handshakes without a lighthouse in certain configurations. (#759)
+
+- The firewall can be configured to send REJECT replies instead of the default
+  DROP behavior. (#738)
+
+- For macOS, an example launchd configuration file is now provided. (#762)
+
+### Changed
+
+- Lighthouses and other `static_host_map` entries that use DNS names will now
+  be automatically refreshed to detect when the IP address changes. (#796)
+
+- Lighthouses send ACK replies back to clients so that they do not fall into
+  connection testing as often by clients. (#851, #408)
+
+- Allow the `listen.host` option to contain a hostname. (#825)
+
+- When Nebula switches to a new certificate (such as via SIGHUP), we now
+  rehandshake with all existing tunnels. This allows firewall groups to be
+  updated and `pki.disconnect_invalid` to know about the new certificate
+  expiration time. (#838, #857, #842, #840, #835, #828, #820, #807)
+
+### Fixed
+
+- Always disconnect blocklisted hosts, even if `pki.disconnect_invalid` is
+  not set. (#858)
+
+- Dependencies updated and go1.20 required. (#780, #824, #855, #854)
+
+- Fix possible race condition with relays. (#827)
+
+- FreeBSD: Fix connection to the localhost's own Nebula IP. (#808)
+
+- Normalize and document some common log field values. (#837, #811)
+
+- Fix crash if you set unlucky values for the firewall timeout configuration
+  options. (#802)
+
+- Make DNS queries case insensitive. (#793)
+
+- Update example systemd configurations to want `nss-lookup`. (#791)
+
+- Errors with SSH commands now go to the SSH tunnel instead of stderr. (#757)
+
+- Fix a hang when shutting down Android. (#772)
+
+## [1.6.1] - 2022-09-26
+
+### Fixed
+
+- Refuse to process underlay packets received from overlay IPs. This prevents
+  confusion on hosts that have unsafe routes configured. (#741)
+
+- The ssh `reload` command did not work on Windows, since it relied on sending
+  a SIGHUP signal internally. This has been fixed. (#725)
+
+- A regression in v1.5.2 that broke unsafe routes on Mobile clients has been
+  fixed. (#729)
+
+## [1.6.0] - 2022-06-30
+
+### Added
+
+- Experimental: nebula clients can be configured to act as relays for other nebula clients.
+  Primarily useful when stubborn NATs make a direct tunnel impossible. (#678)
+
+- Configuration option to report manually specified `ip:port`s to lighthouses. (#650)
+
+- Windows arm64 build. (#638)
+
+- `punchy` and most `lighthouse` config options now support hot reloading. (#649)
+
+### Changed
+
+- Build against go 1.18. (#656)
+
+- Promoted `routines` config from experimental to supported feature. (#702)
+
+- Dependencies updated. (#664)
+
+### Fixed
+
+- Packets destined for the same host that sent it will be returned on MacOS.
+  This matches the default behavior of other operating systems. (#501)
+
+- `unsafe_route` configuration will no longer crash on Windows. (#648)
+
+- A few panics that were introduced in 1.5.x. (#657, #658, #675)
+
+### Security
+
+- You can set `listen.send_recv_error` to control the conditions in which
+  `recv_error` messages are sent. Sending these messages can expose the fact
+  that Nebula is running on a host, but it speeds up re-handshaking. (#670)
+
+### Removed
+
+- `x509` config stanza support has been removed. (#685)
+
+## [1.5.2] - 2021-12-14
+
+### Added
+
+- Warn when a non lighthouse node does not have lighthouse hosts configured. (#587)
+
+### Changed
+
+- No longer fatals if expired CA certificates are present in `pki.ca`, as long as 1 valid CA is present. (#599)
+
+- `nebula-cert` will now enforce ipv4 addresses. (#604)
+
+- Warn on macOS if an unsafe route cannot be created due to a collision with an
+  existing route. (#610)
+
+- Warn if you set a route MTU on platforms where we don't support it. (#611)
+
+### Fixed
+
+- Rare race condition when tearing down a tunnel due to `recv_error` and sending packets on another thread. (#590)
+
+- Bug in `routes` and `unsafe_routes` handling that was introduced in 1.5.0. (#595)
+
+- `-test` mode no longer results in a crash. (#602)
+
+### Removed
+
+- `x509.ca` config alias for `pki.ca`. (#604)
+
+### Security
+
+- Upgraded `golang.org/x/crypto` to address an issue which allowed unauthenticated clients to cause a panic in SSH
+  servers. (#603)
+
+## 1.5.1 - 2021-12-13
+
+(This release was skipped due to discovering #610 and #611 after the tag was
+created.)
+
+## [1.5.0] - 2021-11-11
+
+### Added
+
+- SSH `print-cert` has a new `-raw` flag to get the PEM representation of a certificate. (#483)
+
+- New build architecture: Linux `riscv64`. (#542)
+
+- New experimental config option `remote_allow_ranges`. (#540)
+
+- New config option `pki.disconnect_invalid` that will tear down tunnels when they become invalid (through expiry or
+  removal of root trust). Default is `false`. Note, this will not currently recognize if a remote has changed
+  certificates since the last handshake. (#370)
+
+- New config option `unsafe_routes.<route>.metric` will set a metric for a specific unsafe route. It's useful if you have
+  more than one identical route and want to prefer one against the other. (#353)
+
+### Changed
+
+- Build against go 1.17. (#553)
+
+- Build with `CGO_ENABLED=0` set, to create more portable binaries. This could
+  have an effect on DNS resolution if you rely on anything non-standard. (#421)
+
+- Windows now uses the [wintun](https://www.wintun.net/) driver which does not require installation. This driver
+  is a large improvement over the TAP driver that was used in previous versions. If you had a previous version
+  of `nebula` running, you will want to disable the tap driver in Control Panel, or uninstall the `tap0901` driver
+  before running this version. (#289)
+
+- Darwin binaries are now universal (works on both amd64 and arm64), signed, and shipped in a notarized zip file.
+  `nebula-darwin.zip` will be the only darwin release artifact. (#571)
+
+- Darwin uses syscalls and AF_ROUTE to configure the routing table, instead of
+  using `/sbin/route`. Setting `tun.dev` is now allowed on Darwin as well, it
+  must be in the format `utun[0-9]+` or it will be ignored. (#163)
+
+### Deprecated
+
+- The `preferred_ranges` option has been supported as a replacement for
+  `local_range` since v1.0.0. It has now been documented and `local_range`
+  has been officially deprecated. (#541)
+
+### Fixed
+
+- Valid recv_error packets were incorrectly marked as "spoofing" and ignored. (#482)
+
+- SSH server handles single `exec` requests correctly. (#483)
+
+- Signing a certificate with `nebula-cert sign` now verifies that the supplied
+  ca-key matches the ca-crt. (#503)
+
+- If `preferred_ranges` (or the deprecated `local_range`) is configured, we
+  will immediately switch to a preferred remote address after the reception of
+  a handshake packet (instead of waiting until 1,000 packets have been sent).
+  (#532)
+
+- A race condition when `punchy.respond` is enabled and ensures the correct
+  vpn ip is sent a punch back response in highly queried node. (#566)
+
+- Fix a rare crash during handshake due to a race condition. (#535)
+
 ## [1.4.0] - 2021-05-11
 
 ### Added
@@ -16,13 +259,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Experimental: Nebula can now do work on more than 2 cpu cores in send and receive paths via
   the new `routines` config option. (#382, #391, #395)
-  
+
 - ICMP ping requests can be responded to when the `tun.disabled` is `true`.
   This is useful so that you can "ping" a lighthouse running in this mode. (#342)
 
 - Run smoke tests via `make smoke-docker`. (#287)
 
-- More reported stats, udp memory use on linux, build version (when using Prometheus), firewall, 
+- More reported stats, udp memory use on linux, build version (when using Prometheus), firewall,
   handshake, and cached packet stats. (#390, #405, #450, #453)
 
 - IPv6 support for the underlay network. (#369)
@@ -35,7 +278,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Example systemd unit file now better arranged startup order when using `sshd`
   and other fixes. (#317, #412, #438)
-  
+
 - Reduced memory utilization/garbage collection. (#320, #323, #340)
 
 - Reduced CPU utilization. (#329)
@@ -245,7 +488,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Initial public release.
 
-[Unreleased]: https://github.com/slackhq/nebula/compare/v1.4.0...HEAD
+[Unreleased]: https://github.com/slackhq/nebula/compare/v1.7.2...HEAD
+[1.7.2]: https://github.com/slackhq/nebula/releases/tag/v1.7.2
+[1.7.1]: https://github.com/slackhq/nebula/releases/tag/v1.7.1
+[1.7.0]: https://github.com/slackhq/nebula/releases/tag/v1.7.0
+[1.6.1]: https://github.com/slackhq/nebula/releases/tag/v1.6.1
+[1.6.0]: https://github.com/slackhq/nebula/releases/tag/v1.6.0
+[1.5.2]: https://github.com/slackhq/nebula/releases/tag/v1.5.2
+[1.5.0]: https://github.com/slackhq/nebula/releases/tag/v1.5.0
 [1.4.0]: https://github.com/slackhq/nebula/releases/tag/v1.4.0
 [1.3.0]: https://github.com/slackhq/nebula/releases/tag/v1.3.0
 [1.2.0]: https://github.com/slackhq/nebula/releases/tag/v1.2.0

LOGGING.md

@@ -0,0 +1,38 @@
+### Logging conventions
+
+A log message (the string/format passed to `Info`, `Error`, `Debug` etc, as well as their `Sprintf` counterparts) should
+be a descriptive message about the event and may contain specific identifying characteristics. Regardless of the
+level of detail in the message identifying characteristics should always be included via `WithField`, `WithFields` or
+`WithError`
+
+If an error is being logged use `l.WithError(err)` so that there is better discoverability about the event as well
+as the specific error condition.
+
+#### Common fields
+
+- `cert` - a `cert.NebulaCertificate` object, do not `.String()` this manually, `logrus` will marshal objects properly
+  for the formatter it is using.
+- `fingerprint` - a single `NebeulaCertificate` hex encoded fingerprint
+- `fingerprints` - an array of `NebulaCertificate` hex encoded fingerprints
+- `fwPacket` - a FirewallPacket object
+- `handshake` - an object containing:
+    - `stage` - the current stage counter
+    - `style` - noise handshake style `ix_psk0`, `xx`, etc
+- `header` - a nebula header object
+- `udpAddr` - a `net.UDPAddr` object
+- `udpIp` - a udp ip address
+- `vpnIp` - vpn ip of the host (remote or local)
+- `relay` - the vpnIp of the relay host that is or should be handling the relay packet
+- `relayFrom` - The vpnIp of the initial sender of the relayed packet 
+- `relayTo` - The vpnIp of the final destination of a relayed packet
+
+#### Example:
+
+```
+l.WithError(err).
+    WithField("vpnIp", IntIp(hostinfo.hostId)).
+    WithField("udpAddr", addr).
+    WithField("handshake", m{"stage": 1, "style": "ix"}).
+    WithField("cert", remoteCert).
+    Info("Invalid certificate from host")
+```

Makefile

@@ -1,7 +1,9 @@
-GOMINVERSION = 1.16
+GOMINVERSION = 1.20
 NEBULA_CMD_PATH = "./cmd/nebula"
 GO111MODULE = on
 export GO111MODULE
+CGO_ENABLED = 0
+export CGO_ENABLED
 
 # Set up OS specific bits
 ifeq ($(OS),Windows_NT)
@@ -39,13 +41,15 @@ ALL_LINUX = linux-amd64 \
 	linux-mipsle \
 	linux-mips64 \
 	linux-mips64le \
-	linux-mips-softfloat
+	linux-mips-softfloat \
+	linux-riscv64
 
 ALL = $(ALL_LINUX) \
 	darwin-amd64 \
 	darwin-arm64 \
 	freebsd-amd64 \
-	windows-amd64
+	windows-amd64 \
+	windows-arm64
 
 e2e:
 	$(TEST_ENV) go test -tags=e2e_testing -count=1 $(TEST_FLAGS) ./e2e
@@ -62,6 +66,9 @@ e2evvv: e2ev
 e2evvvv: TEST_ENV += TEST_LOGS=3
 e2evvvv: e2ev
 
+e2e-bench: TEST_FLAGS = -bench=. -benchmem -run=^$
+e2e-bench: e2e
+
 all: $(ALL:%=build/%/nebula) $(ALL:%=build/%/nebula-cert)
 
 release: $(ALL:%=build/nebula-%.tar.gz)
@@ -70,17 +77,25 @@ release-linux: $(ALL_LINUX:%=build/nebula-%.tar.gz)
 
 release-freebsd: build/nebula-freebsd-amd64.tar.gz
 
+release-boringcrypto: build/nebula-linux-$(shell go env GOARCH)-boringcrypto.tar.gz
+
 BUILD_ARGS = -trimpath
 
 bin-windows: build/windows-amd64/nebula.exe build/windows-amd64/nebula-cert.exe
 	mv $? .
 
+bin-windows-arm64: build/windows-arm64/nebula.exe build/windows-arm64/nebula-cert.exe
+	mv $? .
+
 bin-darwin: build/darwin-amd64/nebula build/darwin-amd64/nebula-cert
 	mv $? .
 
 bin-freebsd: build/freebsd-amd64/nebula build/freebsd-amd64/nebula-cert
 	mv $? .
 
+bin-boringcrypto: build/linux-$(shell go env GOARCH)-boringcrypto/nebula build/linux-$(shell go env GOARCH)-boringcrypto/nebula-cert
+	mv $? .
+
 bin:
 	go build $(BUILD_ARGS) -ldflags "$(LDFLAGS)" -o ./nebula${NEBULA_CMD_SUFFIX} ${NEBULA_CMD_PATH}
 	go build $(BUILD_ARGS) -ldflags "$(LDFLAGS)" -o ./nebula-cert${NEBULA_CMD_SUFFIX} ./cmd/nebula-cert
@@ -95,6 +110,10 @@ build/linux-mips-%: GOENV += GOMIPS=$(word 3, $(subst -, ,$*))
 # Build an extra small binary for mips-softfloat
 build/linux-mips-softfloat/%: LDFLAGS += -s -w
 
+# boringcrypto
+build/linux-amd64-boringcrypto/%: GOENV += GOEXPERIMENT=boringcrypto CGO_ENABLED=1
+build/linux-arm64-boringcrypto/%: GOENV += GOEXPERIMENT=boringcrypto CGO_ENABLED=1
+
 build/%/nebula: .FORCE
 	GOOS=$(firstword $(subst -, , $*)) \
 		GOARCH=$(word 2, $(subst -, ,$*)) $(GOENV) \
@@ -123,6 +142,9 @@ vet:
 test:
 	go test -v ./...
 
+test-boringcrypto:
+	GOEXPERIMENT=boringcrypto CGO_ENABLED=1 go test -v ./...
+
 test-cov-html:
 	go test -coverprofile=coverage.out
 	go tool cover -html=coverage.out
@@ -160,6 +182,12 @@ bin-docker: bin build/linux-amd64/nebula build/linux-amd64/nebula-cert
 smoke-docker: bin-docker
 	cd .github/workflows/smoke/ && ./build.sh
 	cd .github/workflows/smoke/ && ./smoke.sh
+	cd .github/workflows/smoke/ && NAME="smoke-p256" CURVE="P256" ./build.sh
+	cd .github/workflows/smoke/ && NAME="smoke-p256" ./smoke.sh
+
+smoke-relay-docker: bin-docker
+	cd .github/workflows/smoke/ && ./build-relay.sh
+	cd .github/workflows/smoke/ && ./smoke-relay.sh
 
 smoke-docker-race: BUILD_ARGS = -race
 smoke-docker-race: smoke-docker

README.md

@@ -1,4 +1,4 @@
-## What is Nebula?
+## What is Nebula?
 Nebula is a scalable overlay networking tool with a focus on performance, simplicity and security.
 It lets you seamlessly connect computers anywhere in the world. Nebula is portable, and runs on Linux, OSX, Windows, iOS, and Android.
 It can be used to connect a small number of computers, but is also able to connect tens of thousands of computers.
@@ -8,21 +8,39 @@ and tunneling, and each of those individual pieces existed before Nebula in vari
 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.
 
+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/enQtOTA5MDI4NDg3MTg4LTkwY2EwNTI4NzQyMzc0M2ZlODBjNWI3NTY1MzhiOThiMmZlZjVkMTI0NGY4YTMyNjUwMWEyNzNkZTJmYzQxOGU)
+You can also join the NebulaOSS Slack group [here](https://join.slack.com/t/nebulaoss/shared_invite/enQtOTA5MDI4NDg3MTg4LTkwY2EwNTI4NzQyMzc0M2ZlODBjNWI3NTY1MzhiOThiMmZlZjVkMTI0NGY4YTMyNjUwMWEyNzNkZTJmYzQxOGU).
 
 ## Supported Platforms
 
 #### Desktop and Server
 
-Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for downloads
+Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for downloads or see the [Distribution Packages](https://github.com/slackhq/nebula#distribution-packages) section.
 
-- Linux - 64 and 32 bit, arm, and others 
+- Linux - 64 and 32 bit, arm, and others
 - Windows
 - MacOS
 - Freebsd
 
+#### Distribution Packages
+
+- [Arch Linux](https://archlinux.org/packages/community/x86_64/nebula/)
+    ```
+    $ sudo pacman -S nebula
+    ```
+- [Fedora Linux](https://src.fedoraproject.org/rpms/nebula)
+    ```
+    $ sudo dnf install nebula
+    ```
+
+- [macOS Homebrew](https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/nebula.rb)
+    ```
+    $ brew install nebula
+    ```
+
 #### Mobile
 
 - [iOS](https://apps.apple.com/us/app/mobile-nebula/id1509587936?itsct=apps_box&itscg=30200)
@@ -36,15 +54,15 @@ Nebula's user-defined groups allow for provider agnostic traffic filtering betwe
 Discovery nodes 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 key exchange, and AES-256-GCM in its default configuration.
+Nebula uses Elliptic-curve Diffie-Hellman (`ECDH`) key exchange and `AES-256-GCM` in its default configuration.
 
-Nebula was created to provide a mechanism for groups hosts to communicate securely, even across the internet, while enabling expressive firewall definitions similar in style to cloud security groups.
+Nebula was created to provide a mechanism for groups of hosts to communicate securely, even across the internet, while enabling expressive firewall definitions similar in style to cloud security groups.
 
 ## Getting started (quickly)
 
 To set up a Nebula network, you'll need:
 
-#### 1. The [Nebula binaries](https://github.com/slackhq/nebula/releases) for your specific platform. Specifically you'll need `nebula-cert` and the specific nebula binary for each platform you use.
+#### 1. The [Nebula binaries](https://github.com/slackhq/nebula/releases) or [Distribution Packages](https://github.com/slackhq/nebula#distribution-packages) for your specific platform. Specifically you'll need `nebula-cert` and the specific nebula binary for each platform you use.
 
 #### 2. (Optional, but you really should..) At least one discovery node with a routable IP address, which we call a lighthouse.
 
@@ -79,13 +97,13 @@ Download a copy of the nebula [example configuration](https://github.com/slackhq
 
 #### 6. Copy nebula credentials, configuration, and binaries to each host
 
-For each host, copy the nebula binary to the host, along with `config.yaml` from step 5, and the files `ca.crt`, `{host}.crt`, and `{host}.key` from step 4.
+For each host, copy the nebula binary to the host, along with `config.yml` from step 5, and the files `ca.crt`, `{host}.crt`, and `{host}.key` from step 4.
 
 **DO NOT COPY `ca.key` TO INDIVIDUAL NODES.**
 
 #### 7. Run nebula on each host
 ```
-./nebula -config /path/to/config.yaml
+./nebula -config /path/to/config.yml
 ```
 
 ## Building Nebula from source
@@ -100,6 +118,17 @@ To build nebula for a specific platform (ex, Windows):
 
 See the [Makefile](Makefile) for more details on build targets
 
+## Curve P256 and BoringCrypto
+
+The default curve used for cryptographic handshakes and signatures is Curve25519. This is the recommended setting for most users. If your deployment has certain compliance requirements, you have the option of creating your CA using `nebula-cert ca -curve P256` to use NIST Curve P256. The CA will then sign certificates using ECDSA P256, and any hosts using these certificates will use P256 for ECDH handshakes.
+
+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
+
+This is not the recommended default deployment, but may be useful based on your compliance requirements.
+
 ## Credits
 
 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.

SECURITY.md

@@ -0,0 +1,12 @@
+Security Policy
+===============
+
+Reporting a Vulnerability
+-------------------------
+
+If you believe you have found a security vulnerability with Nebula, please let
+us know right away. We will investigate all reports and do our best to quickly
+fix valid issues.
+
+You can submit your report on [HackerOne](https://hackerone.com/slack) and our
+security team will respond as soon as possible.

allow_list.go

@@ -4,11 +4,27 @@ import (
 	"fmt"
 	"net"
 	"regexp"
+
+	"github.com/slackhq/nebula/cidr"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/iputil"
 )
 
 type AllowList struct {
 	// The values of this cidrTree are `bool`, signifying allow/deny
-	cidrTree *CIDR6Tree
+	cidrTree *cidr.Tree6
+}
+
+type RemoteAllowList struct {
+	AllowList *AllowList
+
+	// Inside Range Specific, keys of this tree are inside CIDRs and values
+	// are *AllowList
+	insideAllowLists *cidr.Tree6
+}
+
+type LocalAllowList struct {
+	AllowList *AllowList
 
 	// To avoid ambiguity, all rules must be true, or all rules must be false.
 	nameRules []AllowListNameRule
@@ -19,6 +35,223 @@ type AllowListNameRule struct {
 	Allow bool
 }
 
+func NewLocalAllowListFromConfig(c *config.C, k string) (*LocalAllowList, error) {
+	var nameRules []AllowListNameRule
+	handleKey := func(key string, value interface{}) (bool, error) {
+		if key == "interfaces" {
+			var err error
+			nameRules, err = getAllowListInterfaces(k, value)
+			if err != nil {
+				return false, err
+			}
+
+			return true, nil
+		}
+		return false, nil
+	}
+
+	al, err := newAllowListFromConfig(c, k, handleKey)
+	if err != nil {
+		return nil, err
+	}
+	return &LocalAllowList{AllowList: al, nameRules: nameRules}, nil
+}
+
+func NewRemoteAllowListFromConfig(c *config.C, k, rangesKey string) (*RemoteAllowList, error) {
+	al, err := newAllowListFromConfig(c, k, nil)
+	if err != nil {
+		return nil, err
+	}
+	remoteAllowRanges, err := getRemoteAllowRanges(c, rangesKey)
+	if err != nil {
+		return nil, err
+	}
+	return &RemoteAllowList{AllowList: al, insideAllowLists: remoteAllowRanges}, nil
+}
+
+// If the handleKey func returns true, the rest of the parsing is skipped
+// for this key. This allows parsing of special values like `interfaces`.
+func newAllowListFromConfig(c *config.C, k string, handleKey func(key string, value interface{}) (bool, error)) (*AllowList, error) {
+	r := c.Get(k)
+	if r == nil {
+		return nil, nil
+	}
+
+	return newAllowList(k, r, handleKey)
+}
+
+// If the handleKey func returns true, the rest of the parsing is skipped
+// for this key. This allows parsing of special values like `interfaces`.
+func newAllowList(k string, raw interface{}, handleKey func(key string, value interface{}) (bool, error)) (*AllowList, error) {
+	rawMap, ok := raw.(map[interface{}]interface{})
+	if !ok {
+		return nil, fmt.Errorf("config `%s` has invalid type: %T", k, raw)
+	}
+
+	tree := cidr.NewTree6()
+
+	// Keep track of the rules we have added for both ipv4 and ipv6
+	type allowListRules struct {
+		firstValue     bool
+		allValuesMatch bool
+		defaultSet     bool
+		allValues      bool
+	}
+
+	rules4 := allowListRules{firstValue: true, allValuesMatch: true, defaultSet: false}
+	rules6 := allowListRules{firstValue: true, allValuesMatch: true, defaultSet: false}
+
+	for rawKey, rawValue := range rawMap {
+		rawCIDR, ok := rawKey.(string)
+		if !ok {
+			return nil, fmt.Errorf("config `%s` has invalid key (type %T): %v", k, rawKey, rawKey)
+		}
+
+		if handleKey != nil {
+			handled, err := handleKey(rawCIDR, rawValue)
+			if err != nil {
+				return nil, err
+			}
+			if handled {
+				continue
+			}
+		}
+
+		value, ok := rawValue.(bool)
+		if !ok {
+			return nil, fmt.Errorf("config `%s` has invalid value (type %T): %v", k, rawValue, rawValue)
+		}
+
+		_, ipNet, err := net.ParseCIDR(rawCIDR)
+		if err != nil {
+			return nil, fmt.Errorf("config `%s` has invalid CIDR: %s", k, rawCIDR)
+		}
+
+		// TODO: should we error on duplicate CIDRs in the config?
+		tree.AddCIDR(ipNet, value)
+
+		maskBits, maskSize := ipNet.Mask.Size()
+
+		var rules *allowListRules
+		if maskSize == 32 {
+			rules = &rules4
+		} else {
+			rules = &rules6
+		}
+
+		if rules.firstValue {
+			rules.allValues = value
+			rules.firstValue = false
+		} else {
+			if value != rules.allValues {
+				rules.allValuesMatch = false
+			}
+		}
+
+		// Check if this is 0.0.0.0/0 or ::/0
+		if maskBits == 0 {
+			rules.defaultSet = true
+		}
+	}
+
+	if !rules4.defaultSet {
+		if rules4.allValuesMatch {
+			_, zeroCIDR, _ := net.ParseCIDR("0.0.0.0/0")
+			tree.AddCIDR(zeroCIDR, !rules4.allValues)
+		} else {
+			return nil, fmt.Errorf("config `%s` contains both true and false rules, but no default set for 0.0.0.0/0", k)
+		}
+	}
+
+	if !rules6.defaultSet {
+		if rules6.allValuesMatch {
+			_, zeroCIDR, _ := net.ParseCIDR("::/0")
+			tree.AddCIDR(zeroCIDR, !rules6.allValues)
+		} else {
+			return nil, fmt.Errorf("config `%s` contains both true and false rules, but no default set for ::/0", k)
+		}
+	}
+
+	return &AllowList{cidrTree: tree}, nil
+}
+
+func getAllowListInterfaces(k string, v interface{}) ([]AllowListNameRule, error) {
+	var nameRules []AllowListNameRule
+
+	rawRules, ok := v.(map[interface{}]interface{})
+	if !ok {
+		return nil, fmt.Errorf("config `%s.interfaces` is invalid (type %T): %v", k, v, v)
+	}
+
+	firstEntry := true
+	var allValues bool
+	for rawName, rawAllow := range rawRules {
+		name, ok := rawName.(string)
+		if !ok {
+			return nil, fmt.Errorf("config `%s.interfaces` has invalid key (type %T): %v", k, rawName, rawName)
+		}
+		allow, ok := rawAllow.(bool)
+		if !ok {
+			return nil, fmt.Errorf("config `%s.interfaces` has invalid value (type %T): %v", k, rawAllow, rawAllow)
+		}
+
+		nameRE, err := regexp.Compile("^" + name + "$")
+		if err != nil {
+			return nil, fmt.Errorf("config `%s.interfaces` has invalid key: %s: %v", k, name, err)
+		}
+
+		nameRules = append(nameRules, AllowListNameRule{
+			Name:  nameRE,
+			Allow: allow,
+		})
+
+		if firstEntry {
+			allValues = allow
+			firstEntry = false
+		} else {
+			if allow != allValues {
+				return nil, fmt.Errorf("config `%s.interfaces` values must all be the same true/false value", k)
+			}
+		}
+	}
+
+	return nameRules, nil
+}
+
+func getRemoteAllowRanges(c *config.C, k string) (*cidr.Tree6, error) {
+	value := c.Get(k)
+	if value == nil {
+		return nil, nil
+	}
+
+	remoteAllowRanges := cidr.NewTree6()
+
+	rawMap, ok := value.(map[interface{}]interface{})
+	if !ok {
+		return nil, fmt.Errorf("config `%s` has invalid type: %T", k, value)
+	}
+	for rawKey, rawValue := range rawMap {
+		rawCIDR, ok := rawKey.(string)
+		if !ok {
+			return nil, fmt.Errorf("config `%s` has invalid key (type %T): %v", k, rawKey, rawKey)
+		}
+
+		allowList, err := newAllowList(fmt.Sprintf("%s.%s", k, rawCIDR), rawValue, nil)
+		if err != nil {
+			return nil, err
+		}
+
+		_, ipNet, err := net.ParseCIDR(rawCIDR)
+		if err != nil {
+			return nil, fmt.Errorf("config `%s` has invalid CIDR: %s", k, rawCIDR)
+		}
+
+		remoteAllowRanges.AddCIDR(ipNet, allowList)
+	}
+
+	return remoteAllowRanges, nil
+}
+
 func (al *AllowList) Allow(ip net.IP) bool {
 	if al == nil {
 		return true
@@ -33,7 +266,7 @@ func (al *AllowList) Allow(ip net.IP) bool {
 	}
 }
 
-func (al *AllowList) AllowIpV4(ip uint32) bool {
+func (al *AllowList) AllowIpV4(ip iputil.VpnIp) bool {
 	if al == nil {
 		return true
 	}
@@ -61,7 +294,14 @@ func (al *AllowList) AllowIpV6(hi, lo uint64) bool {
 	}
 }
 
-func (al *AllowList) AllowName(name string) bool {
+func (al *LocalAllowList) Allow(ip net.IP) bool {
+	if al == nil {
+		return true
+	}
+	return al.AllowList.Allow(ip)
+}
+
+func (al *LocalAllowList) AllowName(name string) bool {
 	if al == nil || len(al.nameRules) == 0 {
 		return true
 	}
@@ -75,3 +315,47 @@ func (al *AllowList) AllowName(name string) bool {
 	// If no rules match, return the default, which is the inverse of the rules
 	return !al.nameRules[0].Allow
 }
+
+func (al *RemoteAllowList) AllowUnknownVpnIp(ip net.IP) bool {
+	if al == nil {
+		return true
+	}
+	return al.AllowList.Allow(ip)
+}
+
+func (al *RemoteAllowList) Allow(vpnIp iputil.VpnIp, ip net.IP) bool {
+	if !al.getInsideAllowList(vpnIp).Allow(ip) {
+		return false
+	}
+	return al.AllowList.Allow(ip)
+}
+
+func (al *RemoteAllowList) AllowIpV4(vpnIp iputil.VpnIp, ip iputil.VpnIp) bool {
+	if al == nil {
+		return true
+	}
+	if !al.getInsideAllowList(vpnIp).AllowIpV4(ip) {
+		return false
+	}
+	return al.AllowList.AllowIpV4(ip)
+}
+
+func (al *RemoteAllowList) AllowIpV6(vpnIp iputil.VpnIp, hi, lo uint64) bool {
+	if al == nil {
+		return true
+	}
+	if !al.getInsideAllowList(vpnIp).AllowIpV6(hi, lo) {
+		return false
+	}
+	return al.AllowList.AllowIpV6(hi, lo)
+}
+
+func (al *RemoteAllowList) getInsideAllowList(vpnIp iputil.VpnIp) *AllowList {
+	if al.insideAllowLists != nil {
+		inside := al.insideAllowLists.MostSpecificContainsIpV4(vpnIp)
+		if inside != nil {
+			return inside.(*AllowList)
+		}
+	}
+	return nil
+}

allow_list_test.go

@@ -5,21 +5,110 @@ import (
 	"regexp"
 	"testing"
 
+	"github.com/slackhq/nebula/cidr"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
 )
 
+func TestNewAllowListFromConfig(t *testing.T) {
+	l := test.NewLogger()
+	c := config.NewC(l)
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"192.168.0.0": true,
+	}
+	r, err := newAllowListFromConfig(c, "allowlist", nil)
+	assert.EqualError(t, err, "config `allowlist` has invalid CIDR: 192.168.0.0")
+	assert.Nil(t, r)
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"192.168.0.0/16": "abc",
+	}
+	r, err = newAllowListFromConfig(c, "allowlist", nil)
+	assert.EqualError(t, err, "config `allowlist` has invalid value (type string): abc")
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"192.168.0.0/16": true,
+		"10.0.0.0/8":     false,
+	}
+	r, err = newAllowListFromConfig(c, "allowlist", nil)
+	assert.EqualError(t, err, "config `allowlist` contains both true and false rules, but no default set for 0.0.0.0/0")
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"0.0.0.0/0":      true,
+		"10.0.0.0/8":     false,
+		"10.42.42.0/24":  true,
+		"fd00::/8":       true,
+		"fd00:fd00::/16": false,
+	}
+	r, err = newAllowListFromConfig(c, "allowlist", nil)
+	assert.EqualError(t, err, "config `allowlist` contains both true and false rules, but no default set for ::/0")
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"0.0.0.0/0":     true,
+		"10.0.0.0/8":    false,
+		"10.42.42.0/24": true,
+	}
+	r, err = newAllowListFromConfig(c, "allowlist", nil)
+	if assert.NoError(t, err) {
+		assert.NotNil(t, r)
+	}
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"0.0.0.0/0":      true,
+		"10.0.0.0/8":     false,
+		"10.42.42.0/24":  true,
+		"::/0":           false,
+		"fd00::/8":       true,
+		"fd00:fd00::/16": false,
+	}
+	r, err = newAllowListFromConfig(c, "allowlist", nil)
+	if assert.NoError(t, err) {
+		assert.NotNil(t, r)
+	}
+
+	// Test interface names
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"interfaces": map[interface{}]interface{}{
+			`docker.*`: "foo",
+		},
+	}
+	lr, err := NewLocalAllowListFromConfig(c, "allowlist")
+	assert.EqualError(t, err, "config `allowlist.interfaces` has invalid value (type string): foo")
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"interfaces": map[interface{}]interface{}{
+			`docker.*`: false,
+			`eth.*`:    true,
+		},
+	}
+	lr, err = NewLocalAllowListFromConfig(c, "allowlist")
+	assert.EqualError(t, err, "config `allowlist.interfaces` values must all be the same true/false value")
+
+	c.Settings["allowlist"] = map[interface{}]interface{}{
+		"interfaces": map[interface{}]interface{}{
+			`docker.*`: false,
+		},
+	}
+	lr, err = NewLocalAllowListFromConfig(c, "allowlist")
+	if assert.NoError(t, err) {
+		assert.NotNil(t, lr)
+	}
+}
+
 func TestAllowList_Allow(t *testing.T) {
 	assert.Equal(t, true, ((*AllowList)(nil)).Allow(net.ParseIP("1.1.1.1")))
 
-	tree := NewCIDR6Tree()
-	tree.AddCIDR(getCIDR("0.0.0.0/0"), true)
-	tree.AddCIDR(getCIDR("10.0.0.0/8"), false)
-	tree.AddCIDR(getCIDR("10.42.42.42/32"), true)
-	tree.AddCIDR(getCIDR("10.42.0.0/16"), true)
-	tree.AddCIDR(getCIDR("10.42.42.0/24"), true)
-	tree.AddCIDR(getCIDR("10.42.42.0/24"), false)
-	tree.AddCIDR(getCIDR("::1/128"), true)
-	tree.AddCIDR(getCIDR("::2/128"), false)
+	tree := cidr.NewTree6()
+	tree.AddCIDR(cidr.Parse("0.0.0.0/0"), true)
+	tree.AddCIDR(cidr.Parse("10.0.0.0/8"), false)
+	tree.AddCIDR(cidr.Parse("10.42.42.42/32"), true)
+	tree.AddCIDR(cidr.Parse("10.42.0.0/16"), true)
+	tree.AddCIDR(cidr.Parse("10.42.42.0/24"), true)
+	tree.AddCIDR(cidr.Parse("10.42.42.0/24"), false)
+	tree.AddCIDR(cidr.Parse("::1/128"), true)
+	tree.AddCIDR(cidr.Parse("::2/128"), false)
 	al := &AllowList{cidrTree: tree}
 
 	assert.Equal(t, true, al.Allow(net.ParseIP("1.1.1.1")))
@@ -31,14 +120,14 @@ func TestAllowList_Allow(t *testing.T) {
 	assert.Equal(t, false, al.Allow(net.ParseIP("::2")))
 }
 
-func TestAllowList_AllowName(t *testing.T) {
-	assert.Equal(t, true, ((*AllowList)(nil)).AllowName("docker0"))
+func TestLocalAllowList_AllowName(t *testing.T) {
+	assert.Equal(t, true, ((*LocalAllowList)(nil)).AllowName("docker0"))
 
 	rules := []AllowListNameRule{
 		{Name: regexp.MustCompile("^docker.*$"), Allow: false},
 		{Name: regexp.MustCompile("^tun.*$"), Allow: false},
 	}
-	al := &AllowList{nameRules: rules}
+	al := &LocalAllowList{nameRules: rules}
 
 	assert.Equal(t, false, al.AllowName("docker0"))
 	assert.Equal(t, false, al.AllowName("tun0"))
@@ -48,7 +137,7 @@ func TestAllowList_AllowName(t *testing.T) {
 		{Name: regexp.MustCompile("^eth.*$"), Allow: true},
 		{Name: regexp.MustCompile("^ens.*$"), Allow: true},
 	}
-	al = &AllowList{nameRules: rules}
+	al = &LocalAllowList{nameRules: rules}
 
 	assert.Equal(t, false, al.AllowName("docker0"))
 	assert.Equal(t, true, al.AllowName("eth0"))

bits_test.go

@@ -3,11 +3,12 @@ package nebula
 import (
 	"testing"
 
+	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
 )
 
 func TestBits(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	b := NewBits(10)
 
 	// make sure it is the right size
@@ -75,7 +76,7 @@ func TestBits(t *testing.T) {
 }
 
 func TestBitsDupeCounter(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	b := NewBits(10)
 	b.lostCounter.Clear()
 	b.dupeCounter.Clear()
@@ -100,7 +101,7 @@ func TestBitsDupeCounter(t *testing.T) {
 }
 
 func TestBitsOutOfWindowCounter(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	b := NewBits(10)
 	b.lostCounter.Clear()
 	b.dupeCounter.Clear()
@@ -130,7 +131,7 @@ func TestBitsOutOfWindowCounter(t *testing.T) {
 }
 
 func TestBitsLostCounter(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	b := NewBits(10)
 	b.lostCounter.Clear()
 	b.dupeCounter.Clear()

boring.go

@@ -0,0 +1,8 @@
+//go:build boringcrypto
+// +build boringcrypto
+
+package nebula
+
+import "crypto/boring"
+
+var boringEnabled = boring.Enabled

calculated_remote.go

@@ -0,0 +1,143 @@
+package nebula
+
+import (
+	"fmt"
+	"math"
+	"net"
+	"strconv"
+
+	"github.com/slackhq/nebula/cidr"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/iputil"
+)
+
+// This allows us to "guess" what the remote might be for a host while we wait
+// for the lighthouse response. See "lighthouse.calculated_remotes" in the
+// example config file.
+type calculatedRemote struct {
+	ipNet  net.IPNet
+	maskIP iputil.VpnIp
+	mask   iputil.VpnIp
+	port   uint32
+}
+
+func newCalculatedRemote(ipNet *net.IPNet, port int) (*calculatedRemote, error) {
+	// Ensure this is an IPv4 mask that we expect
+	ones, bits := ipNet.Mask.Size()
+	if ones == 0 || bits != 32 {
+		return nil, fmt.Errorf("invalid mask: %v", ipNet)
+	}
+	if port < 0 || port > math.MaxUint16 {
+		return nil, fmt.Errorf("invalid port: %d", port)
+	}
+
+	return &calculatedRemote{
+		ipNet:  *ipNet,
+		maskIP: iputil.Ip2VpnIp(ipNet.IP),
+		mask:   iputil.Ip2VpnIp(ipNet.Mask),
+		port:   uint32(port),
+	}, nil
+}
+
+func (c *calculatedRemote) String() string {
+	return fmt.Sprintf("CalculatedRemote(mask=%v port=%d)", c.ipNet, c.port)
+}
+
+func (c *calculatedRemote) Apply(ip iputil.VpnIp) *Ip4AndPort {
+	// Combine the masked bytes of the "mask" IP with the unmasked bytes
+	// of the overlay IP
+	masked := (c.maskIP & c.mask) | (ip & ^c.mask)
+
+	return &Ip4AndPort{Ip: uint32(masked), Port: c.port}
+}
+
+func NewCalculatedRemotesFromConfig(c *config.C, k string) (*cidr.Tree4, error) {
+	value := c.Get(k)
+	if value == nil {
+		return nil, nil
+	}
+
+	calculatedRemotes := cidr.NewTree4()
+
+	rawMap, ok := value.(map[any]any)
+	if !ok {
+		return nil, fmt.Errorf("config `%s` has invalid type: %T", k, value)
+	}
+	for rawKey, rawValue := range rawMap {
+		rawCIDR, ok := rawKey.(string)
+		if !ok {
+			return nil, fmt.Errorf("config `%s` has invalid key (type %T): %v", k, rawKey, rawKey)
+		}
+
+		_, ipNet, err := net.ParseCIDR(rawCIDR)
+		if err != nil {
+			return nil, fmt.Errorf("config `%s` has invalid CIDR: %s", k, rawCIDR)
+		}
+
+		entry, err := newCalculatedRemotesListFromConfig(rawValue)
+		if err != nil {
+			return nil, fmt.Errorf("config '%s.%s': %w", k, rawCIDR, err)
+		}
+
+		calculatedRemotes.AddCIDR(ipNet, entry)
+	}
+
+	return calculatedRemotes, nil
+}
+
+func newCalculatedRemotesListFromConfig(raw any) ([]*calculatedRemote, error) {
+	rawList, ok := raw.([]any)
+	if !ok {
+		return nil, fmt.Errorf("calculated_remotes entry has invalid type: %T", raw)
+	}
+
+	var l []*calculatedRemote
+	for _, e := range rawList {
+		c, err := newCalculatedRemotesEntryFromConfig(e)
+		if err != nil {
+			return nil, fmt.Errorf("calculated_remotes entry: %w", err)
+		}
+		l = append(l, c)
+	}
+
+	return l, nil
+}
+
+func newCalculatedRemotesEntryFromConfig(raw any) (*calculatedRemote, error) {
+	rawMap, ok := raw.(map[any]any)
+	if !ok {
+		return nil, fmt.Errorf("invalid type: %T", raw)
+	}
+
+	rawValue := rawMap["mask"]
+	if rawValue == nil {
+		return nil, fmt.Errorf("missing mask: %v", rawMap)
+	}
+	rawMask, ok := rawValue.(string)
+	if !ok {
+		return nil, fmt.Errorf("invalid mask (type %T): %v", rawValue, rawValue)
+	}
+	_, ipNet, err := net.ParseCIDR(rawMask)
+	if err != nil {
+		return nil, fmt.Errorf("invalid mask: %s", rawMask)
+	}
+
+	var port int
+	rawValue = rawMap["port"]
+	if rawValue == nil {
+		return nil, fmt.Errorf("missing port: %v", rawMap)
+	}
+	switch v := rawValue.(type) {
+	case int:
+		port = v
+	case string:
+		port, err = strconv.Atoi(v)
+		if err != nil {
+			return nil, fmt.Errorf("invalid port: %s: %w", v, err)
+		}
+	default:
+		return nil, fmt.Errorf("invalid port (type %T): %v", rawValue, rawValue)
+	}
+
+	return newCalculatedRemote(ipNet, port)
+}

calculated_remote_test.go

@@ -0,0 +1,27 @@
+package nebula
+
+import (
+	"net"
+	"testing"
+
+	"github.com/slackhq/nebula/iputil"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCalculatedRemoteApply(t *testing.T) {
+	_, ipNet, err := net.ParseCIDR("192.168.1.0/24")
+	require.NoError(t, err)
+
+	c, err := newCalculatedRemote(ipNet, 4242)
+	require.NoError(t, err)
+
+	input := iputil.Ip2VpnIp([]byte{10, 0, 10, 182})
+
+	expected := &Ip4AndPort{
+		Ip:   uint32(iputil.Ip2VpnIp([]byte{192, 168, 1, 182})),
+		Port: 4242,
+	}
+
+	assert.Equal(t, expected, c.Apply(input))
+}

cert.go

@@ -9,6 +9,7 @@ import (
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/config"
 )
 
 type CertState struct {
@@ -45,16 +46,11 @@ func NewCertState(certificate *cert.NebulaCertificate, privateKey []byte) (*Cert
 	return cs, nil
 }
 
-func NewCertStateFromConfig(c *Config) (*CertState, error) {
+func NewCertStateFromConfig(c *config.C) (*CertState, error) {
 	var pemPrivateKey []byte
 	var err error
 
 	privPathOrPEM := c.GetString("pki.key", "")
-	if privPathOrPEM == "" {
-		// Support backwards compat with the old x509
-		//TODO: remove after this is rolled out everywhere - NB 2018/02/23
-		privPathOrPEM = c.GetString("x509.key", "")
-	}
 
 	if privPathOrPEM == "" {
 		return nil, errors.New("no pki.key path or PEM data provided")
@@ -70,7 +66,7 @@ func NewCertStateFromConfig(c *Config) (*CertState, error) {
 		}
 	}
 
-	rawKey, _, err := cert.UnmarshalX25519PrivateKey(pemPrivateKey)
+	rawKey, _, curve, err := cert.UnmarshalPrivateKey(pemPrivateKey)
 	if err != nil {
 		return nil, fmt.Errorf("error while unmarshaling pki.key %s: %s", privPathOrPEM, err)
 	}
@@ -78,11 +74,6 @@ func NewCertStateFromConfig(c *Config) (*CertState, error) {
 	var rawCert []byte
 
 	pubPathOrPEM := c.GetString("pki.cert", "")
-	if pubPathOrPEM == "" {
-		// Support backwards compat with the old x509
-		//TODO: remove after this is rolled out everywhere - NB 2018/02/23
-		pubPathOrPEM = c.GetString("x509.cert", "")
-	}
 
 	if pubPathOrPEM == "" {
 		return nil, errors.New("no pki.cert path or PEM data provided")
@@ -111,31 +102,25 @@ func NewCertStateFromConfig(c *Config) (*CertState, error) {
 		return nil, fmt.Errorf("no IPs encoded in certificate")
 	}
 
-	if err = nebulaCert.VerifyPrivateKey(rawKey); err != nil {
+	if err = nebulaCert.VerifyPrivateKey(curve, rawKey); err != nil {
 		return nil, fmt.Errorf("private key is not a pair with public key in nebula cert")
 	}
 
 	return NewCertState(nebulaCert, rawKey)
 }
 
-func loadCAFromConfig(l *logrus.Logger, c *Config) (*cert.NebulaCAPool, error) {
+func loadCAFromConfig(l *logrus.Logger, c *config.C) (*cert.NebulaCAPool, error) {
 	var rawCA []byte
 	var err error
 
 	caPathOrPEM := c.GetString("pki.ca", "")
-	if caPathOrPEM == "" {
-		// Support backwards compat with the old x509
-		//TODO: remove after this is rolled out everywhere - NB 2018/02/23
-		caPathOrPEM = c.GetString("x509.ca", "")
-	}
-
 	if caPathOrPEM == "" {
 		return nil, errors.New("no pki.ca path or PEM data provided")
 	}
 
 	if strings.Contains(caPathOrPEM, "-----BEGIN") {
 		rawCA = []byte(caPathOrPEM)
-		caPathOrPEM = "<inline>"
+
 	} else {
 		rawCA, err = ioutil.ReadFile(caPathOrPEM)
 		if err != nil {
@@ -144,18 +129,32 @@ func loadCAFromConfig(l *logrus.Logger, c *Config) (*cert.NebulaCAPool, error) {
 	}
 
 	CAs, err := cert.NewCAPoolFromBytes(rawCA)
-	if err != nil {
+	if errors.Is(err, cert.ErrExpired) {
+		var expired int
+		for _, cert := range CAs.CAs {
+			if cert.Expired(time.Now()) {
+				expired++
+				l.WithField("cert", cert).Warn("expired certificate present in CA pool")
+			}
+		}
+
+		if expired >= len(CAs.CAs) {
+			return nil, errors.New("no valid CA certificates present")
+		}
+
+	} else if err != nil {
 		return nil, fmt.Errorf("error while adding CA certificate to CA trust store: %s", err)
 	}
 
 	for _, fp := range c.GetStringSlice("pki.blocklist", []string{}) {
-		l.WithField("fingerprint", fp).Infof("Blocklisting cert")
+		l.WithField("fingerprint", fp).Info("Blocklisting cert")
 		CAs.BlocklistFingerprint(fp)
 	}
 
-	// Support deprecated config for at leaast one minor release to allow for migrations
+	// Support deprecated config for at least one minor release to allow for migrations
+	//TODO: remove in 2022 or later
 	for _, fp := range c.GetStringSlice("pki.blacklist", []string{}) {
-		l.WithField("fingerprint", fp).Infof("Blocklisting cert")
+		l.WithField("fingerprint", fp).Info("Blocklisting cert")
 		l.Warn("pki.blacklist is deprecated and will not be supported in a future release. Please migrate your config to use pki.blocklist")
 		CAs.BlocklistFingerprint(fp)
 	}

cert/ca.go

@@ -1,6 +1,7 @@
 package cert
 
 import (
+	"errors"
 	"fmt"
 	"strings"
 	"time"
@@ -21,19 +22,32 @@ func NewCAPool() *NebulaCAPool {
 	return &ca
 }
 
+// NewCAPoolFromBytes will create a new CA pool from the provided
+// input bytes, which must be a PEM-encoded set of nebula certificates.
+// If the pool contains any expired certificates, an ErrExpired will be
+// returned along with the pool. The caller must handle any such errors.
 func NewCAPoolFromBytes(caPEMs []byte) (*NebulaCAPool, error) {
 	pool := NewCAPool()
 	var err error
+	var expired bool
 	for {
 		caPEMs, err = pool.AddCACertificate(caPEMs)
+		if errors.Is(err, ErrExpired) {
+			expired = true
+			err = nil
+		}
 		if err != nil {
 			return nil, err
 		}
-		if caPEMs == nil || len(caPEMs) == 0 || strings.TrimSpace(string(caPEMs)) == "" {
+		if len(caPEMs) == 0 || strings.TrimSpace(string(caPEMs)) == "" {
 			break
 		}
 	}
 
+	if expired {
+		return pool, ErrExpired
+	}
+
 	return pool, nil
 }
 
@@ -47,15 +61,11 @@ func (ncp *NebulaCAPool) AddCACertificate(pemBytes []byte) ([]byte, error) {
 	}
 
 	if !c.Details.IsCA {
-		return pemBytes, fmt.Errorf("provided certificate was not a CA; %s", c.Details.Name)
+		return pemBytes, fmt.Errorf("%s: %w", c.Details.Name, ErrNotCA)
 	}
 
 	if !c.CheckSignature(c.Details.PublicKey) {
-		return pemBytes, fmt.Errorf("provided certificate was not self signed; %s", c.Details.Name)
-	}
-
-	if c.Expired(time.Now()) {
-		return pemBytes, fmt.Errorf("provided CA certificate is expired; %s", c.Details.Name)
+		return pemBytes, fmt.Errorf("%s: %w", c.Details.Name, ErrNotSelfSigned)
 	}
 
 	sum, err := c.Sha256Sum()
@@ -64,6 +74,10 @@ func (ncp *NebulaCAPool) AddCACertificate(pemBytes []byte) ([]byte, error) {
 	}
 
 	ncp.CAs[sum] = c
+	if c.Expired(time.Now()) {
+		return pemBytes, fmt.Errorf("%s: %w", c.Details.Name, ErrExpired)
+	}
+
 	return pemBytes, nil
 }
 
@@ -77,9 +91,15 @@ func (ncp *NebulaCAPool) ResetCertBlocklist() {
 	ncp.certBlocklist = make(map[string]struct{})
 }
 
-// IsBlocklisted returns true if the fingerprint fails to generate or has been explicitly blocklisted
+// NOTE: This uses an internal cache for Sha256Sum() that will not be invalidated
+// automatically if you manually change any fields in the NebulaCertificate.
 func (ncp *NebulaCAPool) IsBlocklisted(c *NebulaCertificate) bool {
-	h, err := c.Sha256Sum()
+	return ncp.isBlocklistedWithCache(c, false)
+}
+
+// IsBlocklisted returns true if the fingerprint fails to generate or has been explicitly blocklisted
+func (ncp *NebulaCAPool) isBlocklistedWithCache(c *NebulaCertificate, useCache bool) bool {
+	h, err := c.sha256SumWithCache(useCache)
 	if err != nil {
 		return true
 	}

cert/cert.go

@@ -2,35 +2,55 @@ package cert
 
 import (
 	"bytes"
-	"crypto"
+	"crypto/ecdh"
+	"crypto/ecdsa"
+	"crypto/ed25519"
+	"crypto/elliptic"
 	"crypto/rand"
 	"crypto/sha256"
 	"encoding/binary"
 	"encoding/hex"
 	"encoding/json"
 	"encoding/pem"
+	"errors"
 	"fmt"
+	"math"
+	"math/big"
 	"net"
+	"sync/atomic"
 	"time"
 
-	"github.com/golang/protobuf/proto"
 	"golang.org/x/crypto/curve25519"
-	"golang.org/x/crypto/ed25519"
+	"google.golang.org/protobuf/proto"
 )
 
 const publicKeyLen = 32
 
 const (
-	CertBanner              = "NEBULA CERTIFICATE"
-	X25519PrivateKeyBanner  = "NEBULA X25519 PRIVATE KEY"
-	X25519PublicKeyBanner   = "NEBULA X25519 PUBLIC KEY"
-	Ed25519PrivateKeyBanner = "NEBULA ED25519 PRIVATE KEY"
-	Ed25519PublicKeyBanner  = "NEBULA ED25519 PUBLIC KEY"
+	CertBanner                       = "NEBULA CERTIFICATE"
+	X25519PrivateKeyBanner           = "NEBULA X25519 PRIVATE KEY"
+	X25519PublicKeyBanner            = "NEBULA X25519 PUBLIC KEY"
+	EncryptedEd25519PrivateKeyBanner = "NEBULA ED25519 ENCRYPTED PRIVATE KEY"
+	Ed25519PrivateKeyBanner          = "NEBULA ED25519 PRIVATE KEY"
+	Ed25519PublicKeyBanner           = "NEBULA ED25519 PUBLIC KEY"
+
+	P256PrivateKeyBanner               = "NEBULA P256 PRIVATE KEY"
+	P256PublicKeyBanner                = "NEBULA P256 PUBLIC KEY"
+	EncryptedECDSAP256PrivateKeyBanner = "NEBULA ECDSA P256 ENCRYPTED PRIVATE KEY"
+	ECDSAP256PrivateKeyBanner          = "NEBULA ECDSA P256 PRIVATE KEY"
 )
 
 type NebulaCertificate struct {
 	Details   NebulaCertificateDetails
 	Signature []byte
+
+	// the cached hex string of the calculated sha256sum
+	// for VerifyWithCache
+	sha256sum atomic.Pointer[string]
+
+	// the cached public key bytes if they were verified as the signer
+	// for VerifyWithCache
+	signatureVerified atomic.Pointer[[]byte]
 }
 
 type NebulaCertificateDetails struct {
@@ -46,10 +66,25 @@ type NebulaCertificateDetails struct {
 
 	// Map of groups for faster lookup
 	InvertedGroups map[string]struct{}
+
+	Curve Curve
+}
+
+type NebulaEncryptedData struct {
+	EncryptionMetadata NebulaEncryptionMetadata
+	Ciphertext         []byte
+}
+
+type NebulaEncryptionMetadata struct {
+	EncryptionAlgorithm string
+	Argon2Parameters    Argon2Parameters
 }
 
 type m map[string]interface{}
 
+// Returned if we try to unmarshal an encrypted private key without a passphrase
+var ErrPrivateKeyEncrypted = errors.New("private key must be decrypted")
+
 // UnmarshalNebulaCertificate will unmarshal a protobuf byte representation of a nebula cert
 func UnmarshalNebulaCertificate(b []byte) (*NebulaCertificate, error) {
 	if len(b) == 0 {
@@ -84,6 +119,7 @@ func UnmarshalNebulaCertificate(b []byte) (*NebulaCertificate, error) {
 			PublicKey:      make([]byte, len(rc.Details.PublicKey)),
 			IsCA:           rc.Details.IsCA,
 			InvertedGroups: make(map[string]struct{}),
+			Curve:          rc.Details.Curve,
 		},
 		Signature: make([]byte, len(rc.Signature)),
 	}
@@ -134,6 +170,28 @@ func UnmarshalNebulaCertificateFromPEM(b []byte) (*NebulaCertificate, []byte, er
 	return nc, r, err
 }
 
+func MarshalPrivateKey(curve Curve, b []byte) []byte {
+	switch curve {
+	case Curve_CURVE25519:
+		return pem.EncodeToMemory(&pem.Block{Type: X25519PrivateKeyBanner, Bytes: b})
+	case Curve_P256:
+		return pem.EncodeToMemory(&pem.Block{Type: P256PrivateKeyBanner, Bytes: b})
+	default:
+		return nil
+	}
+}
+
+func MarshalSigningPrivateKey(curve Curve, b []byte) []byte {
+	switch curve {
+	case Curve_CURVE25519:
+		return pem.EncodeToMemory(&pem.Block{Type: Ed25519PrivateKeyBanner, Bytes: b})
+	case Curve_P256:
+		return pem.EncodeToMemory(&pem.Block{Type: ECDSAP256PrivateKeyBanner, Bytes: b})
+	default:
+		return nil
+	}
+}
+
 // MarshalX25519PrivateKey is a simple helper to PEM encode an X25519 private key
 func MarshalX25519PrivateKey(b []byte) []byte {
 	return pem.EncodeToMemory(&pem.Block{Type: X25519PrivateKeyBanner, Bytes: b})
@@ -144,6 +202,87 @@ func MarshalEd25519PrivateKey(key ed25519.PrivateKey) []byte {
 	return pem.EncodeToMemory(&pem.Block{Type: Ed25519PrivateKeyBanner, Bytes: key})
 }
 
+func UnmarshalPrivateKey(b []byte) ([]byte, []byte, Curve, error) {
+	k, r := pem.Decode(b)
+	if k == nil {
+		return nil, r, 0, fmt.Errorf("input did not contain a valid PEM encoded block")
+	}
+	var expectedLen int
+	var curve Curve
+	switch k.Type {
+	case X25519PrivateKeyBanner:
+		expectedLen = 32
+		curve = Curve_CURVE25519
+	case P256PrivateKeyBanner:
+		expectedLen = 32
+		curve = Curve_P256
+	default:
+		return nil, r, 0, fmt.Errorf("bytes did not contain a proper nebula private key banner")
+	}
+	if len(k.Bytes) != expectedLen {
+		return nil, r, 0, fmt.Errorf("key was not %d bytes, is invalid %s private key", expectedLen, curve)
+	}
+	return k.Bytes, r, curve, nil
+}
+
+func UnmarshalSigningPrivateKey(b []byte) ([]byte, []byte, Curve, error) {
+	k, r := pem.Decode(b)
+	if k == nil {
+		return nil, r, 0, fmt.Errorf("input did not contain a valid PEM encoded block")
+	}
+	var curve Curve
+	switch k.Type {
+	case EncryptedEd25519PrivateKeyBanner:
+		return nil, nil, Curve_CURVE25519, ErrPrivateKeyEncrypted
+	case EncryptedECDSAP256PrivateKeyBanner:
+		return nil, nil, Curve_P256, ErrPrivateKeyEncrypted
+	case Ed25519PrivateKeyBanner:
+		curve = Curve_CURVE25519
+		if len(k.Bytes) != ed25519.PrivateKeySize {
+			return nil, r, 0, fmt.Errorf("key was not %d bytes, is invalid Ed25519 private key", ed25519.PrivateKeySize)
+		}
+	case ECDSAP256PrivateKeyBanner:
+		curve = Curve_P256
+		if len(k.Bytes) != 32 {
+			return nil, r, 0, fmt.Errorf("key was not 32 bytes, is invalid ECDSA P256 private key")
+		}
+	default:
+		return nil, r, 0, fmt.Errorf("bytes did not contain a proper nebula Ed25519/ECDSA private key banner")
+	}
+	return k.Bytes, r, curve, nil
+}
+
+// EncryptAndMarshalSigningPrivateKey is a simple helper to encrypt and PEM encode a private key
+func EncryptAndMarshalSigningPrivateKey(curve Curve, b []byte, passphrase []byte, kdfParams *Argon2Parameters) ([]byte, error) {
+	ciphertext, err := aes256Encrypt(passphrase, kdfParams, b)
+	if err != nil {
+		return nil, err
+	}
+
+	b, err = proto.Marshal(&RawNebulaEncryptedData{
+		EncryptionMetadata: &RawNebulaEncryptionMetadata{
+			EncryptionAlgorithm: "AES-256-GCM",
+			Argon2Parameters: &RawNebulaArgon2Parameters{
+				Version:     kdfParams.version,
+				Memory:      kdfParams.Memory,
+				Parallelism: uint32(kdfParams.Parallelism),
+				Iterations:  kdfParams.Iterations,
+				Salt:        kdfParams.salt,
+			},
+		},
+		Ciphertext: ciphertext,
+	})
+
+	switch curve {
+	case Curve_CURVE25519:
+		return pem.EncodeToMemory(&pem.Block{Type: EncryptedEd25519PrivateKeyBanner, Bytes: b}), nil
+	case Curve_P256:
+		return pem.EncodeToMemory(&pem.Block{Type: EncryptedECDSAP256PrivateKeyBanner, Bytes: b}), nil
+	default:
+		return nil, fmt.Errorf("invalid curve: %v", curve)
+	}
+}
+
 // UnmarshalX25519PrivateKey will try to pem decode an X25519 private key, returning any other bytes b
 // or an error on failure
 func UnmarshalX25519PrivateKey(b []byte) ([]byte, []byte, error) {
@@ -168,9 +307,13 @@ func UnmarshalEd25519PrivateKey(b []byte) (ed25519.PrivateKey, []byte, error) {
 	if k == nil {
 		return nil, r, fmt.Errorf("input did not contain a valid PEM encoded block")
 	}
-	if k.Type != Ed25519PrivateKeyBanner {
+
+	if k.Type == EncryptedEd25519PrivateKeyBanner {
+		return nil, r, ErrPrivateKeyEncrypted
+	} else if k.Type != Ed25519PrivateKeyBanner {
 		return nil, r, fmt.Errorf("bytes did not contain a proper nebula Ed25519 private key banner")
 	}
+
 	if len(k.Bytes) != ed25519.PrivateKeySize {
 		return nil, r, fmt.Errorf("key was not 64 bytes, is invalid ed25519 private key")
 	}
@@ -178,6 +321,126 @@ func UnmarshalEd25519PrivateKey(b []byte) (ed25519.PrivateKey, []byte, error) {
 	return k.Bytes, r, nil
 }
 
+// UnmarshalNebulaCertificate will unmarshal a protobuf byte representation of a nebula cert into its
+// protobuf-generated struct.
+func UnmarshalNebulaEncryptedData(b []byte) (*NebulaEncryptedData, error) {
+	if len(b) == 0 {
+		return nil, fmt.Errorf("nil byte array")
+	}
+	var rned RawNebulaEncryptedData
+	err := proto.Unmarshal(b, &rned)
+	if err != nil {
+		return nil, err
+	}
+
+	if rned.EncryptionMetadata == nil {
+		return nil, fmt.Errorf("encoded EncryptionMetadata was nil")
+	}
+
+	if rned.EncryptionMetadata.Argon2Parameters == nil {
+		return nil, fmt.Errorf("encoded Argon2Parameters was nil")
+	}
+
+	params, err := unmarshalArgon2Parameters(rned.EncryptionMetadata.Argon2Parameters)
+	if err != nil {
+		return nil, err
+	}
+
+	ned := NebulaEncryptedData{
+		EncryptionMetadata: NebulaEncryptionMetadata{
+			EncryptionAlgorithm: rned.EncryptionMetadata.EncryptionAlgorithm,
+			Argon2Parameters:    *params,
+		},
+		Ciphertext: rned.Ciphertext,
+	}
+
+	return &ned, nil
+}
+
+func unmarshalArgon2Parameters(params *RawNebulaArgon2Parameters) (*Argon2Parameters, error) {
+	if params.Version < math.MinInt32 || params.Version > math.MaxInt32 {
+		return nil, fmt.Errorf("Argon2Parameters Version must be at least %d and no more than %d", math.MinInt32, math.MaxInt32)
+	}
+	if params.Memory <= 0 || params.Memory > math.MaxUint32 {
+		return nil, fmt.Errorf("Argon2Parameters Memory must be be greater than 0 and no more than %d KiB", uint32(math.MaxUint32))
+	}
+	if params.Parallelism <= 0 || params.Parallelism > math.MaxUint8 {
+		return nil, fmt.Errorf("Argon2Parameters Parallelism must be be greater than 0 and no more than %d", math.MaxUint8)
+	}
+	if params.Iterations <= 0 || params.Iterations > math.MaxUint32 {
+		return nil, fmt.Errorf("-argon-iterations must be be greater than 0 and no more than %d", uint32(math.MaxUint32))
+	}
+
+	return &Argon2Parameters{
+		version:     rune(params.Version),
+		Memory:      uint32(params.Memory),
+		Parallelism: uint8(params.Parallelism),
+		Iterations:  uint32(params.Iterations),
+		salt:        params.Salt,
+	}, nil
+
+}
+
+// DecryptAndUnmarshalSigningPrivateKey will try to pem decode and decrypt an Ed25519/ECDSA private key with
+// the given passphrase, returning any other bytes b or an error on failure
+func DecryptAndUnmarshalSigningPrivateKey(passphrase, b []byte) (Curve, []byte, []byte, error) {
+	var curve Curve
+
+	k, r := pem.Decode(b)
+	if k == nil {
+		return curve, nil, r, fmt.Errorf("input did not contain a valid PEM encoded block")
+	}
+
+	switch k.Type {
+	case EncryptedEd25519PrivateKeyBanner:
+		curve = Curve_CURVE25519
+	case EncryptedECDSAP256PrivateKeyBanner:
+		curve = Curve_P256
+	default:
+		return curve, nil, r, fmt.Errorf("bytes did not contain a proper nebula encrypted Ed25519/ECDSA private key banner")
+	}
+
+	ned, err := UnmarshalNebulaEncryptedData(k.Bytes)
+	if err != nil {
+		return curve, nil, r, err
+	}
+
+	var bytes []byte
+	switch ned.EncryptionMetadata.EncryptionAlgorithm {
+	case "AES-256-GCM":
+		bytes, err = aes256Decrypt(passphrase, &ned.EncryptionMetadata.Argon2Parameters, ned.Ciphertext)
+		if err != nil {
+			return curve, nil, r, err
+		}
+	default:
+		return curve, nil, r, fmt.Errorf("unsupported encryption algorithm: %s", ned.EncryptionMetadata.EncryptionAlgorithm)
+	}
+
+	switch curve {
+	case Curve_CURVE25519:
+		if len(bytes) != ed25519.PrivateKeySize {
+			return curve, nil, r, fmt.Errorf("key was not %d bytes, is invalid ed25519 private key", ed25519.PrivateKeySize)
+		}
+	case Curve_P256:
+		if len(bytes) != 32 {
+			return curve, nil, r, fmt.Errorf("key was not 32 bytes, is invalid ECDSA P256 private key")
+		}
+	}
+
+	return curve, bytes, r, nil
+}
+
+func MarshalPublicKey(curve Curve, b []byte) []byte {
+	switch curve {
+	case Curve_CURVE25519:
+		return pem.EncodeToMemory(&pem.Block{Type: X25519PublicKeyBanner, Bytes: b})
+	case Curve_P256:
+		return pem.EncodeToMemory(&pem.Block{Type: P256PublicKeyBanner, Bytes: b})
+	default:
+		return nil
+	}
+}
+
 // MarshalX25519PublicKey is a simple helper to PEM encode an X25519 public key
 func MarshalX25519PublicKey(b []byte) []byte {
 	return pem.EncodeToMemory(&pem.Block{Type: X25519PublicKeyBanner, Bytes: b})
@@ -188,6 +451,30 @@ func MarshalEd25519PublicKey(key ed25519.PublicKey) []byte {
 	return pem.EncodeToMemory(&pem.Block{Type: Ed25519PublicKeyBanner, Bytes: key})
 }
 
+func UnmarshalPublicKey(b []byte) ([]byte, []byte, Curve, error) {
+	k, r := pem.Decode(b)
+	if k == nil {
+		return nil, r, 0, fmt.Errorf("input did not contain a valid PEM encoded block")
+	}
+	var expectedLen int
+	var curve Curve
+	switch k.Type {
+	case X25519PublicKeyBanner:
+		expectedLen = 32
+		curve = Curve_CURVE25519
+	case P256PublicKeyBanner:
+		// Uncompressed
+		expectedLen = 65
+		curve = Curve_P256
+	default:
+		return nil, r, 0, fmt.Errorf("bytes did not contain a proper nebula public key banner")
+	}
+	if len(k.Bytes) != expectedLen {
+		return nil, r, 0, fmt.Errorf("key was not %d bytes, is invalid %s public key", expectedLen, curve)
+	}
+	return k.Bytes, r, curve, nil
+}
+
 // UnmarshalX25519PublicKey will try to pem decode an X25519 public key, returning any other bytes b
 // or an error on failure
 func UnmarshalX25519PublicKey(b []byte) ([]byte, []byte, error) {
@@ -223,27 +510,86 @@ func UnmarshalEd25519PublicKey(b []byte) (ed25519.PublicKey, []byte, error) {
 }
 
 // Sign signs a nebula cert with the provided private key
-func (nc *NebulaCertificate) Sign(key ed25519.PrivateKey) error {
+func (nc *NebulaCertificate) Sign(curve Curve, key []byte) error {
+	if curve != nc.Details.Curve {
+		return fmt.Errorf("curve in cert and private key supplied don't match")
+	}
+
 	b, err := proto.Marshal(nc.getRawDetails())
 	if err != nil {
 		return err
 	}
 
-	sig, err := key.Sign(rand.Reader, b, crypto.Hash(0))
-	if err != nil {
-		return err
+	var sig []byte
+
+	switch curve {
+	case Curve_CURVE25519:
+		signer := ed25519.PrivateKey(key)
+		sig = ed25519.Sign(signer, b)
+	case Curve_P256:
+		signer := &ecdsa.PrivateKey{
+			PublicKey: ecdsa.PublicKey{
+				Curve: elliptic.P256(),
+			},
+			// ref: https://github.com/golang/go/blob/go1.19/src/crypto/x509/sec1.go#L95
+			D: new(big.Int).SetBytes(key),
+		}
+		// ref: https://github.com/golang/go/blob/go1.19/src/crypto/x509/sec1.go#L119
+		signer.X, signer.Y = signer.Curve.ScalarBaseMult(key)
+
+		// We need to hash first for ECDSA
+		// - https://pkg.go.dev/crypto/ecdsa#SignASN1
+		hashed := sha256.Sum256(b)
+		sig, err = ecdsa.SignASN1(rand.Reader, signer, hashed[:])
+		if err != nil {
+			return err
+		}
+	default:
+		return fmt.Errorf("invalid curve: %s", nc.Details.Curve)
 	}
+
 	nc.Signature = sig
 	return nil
 }
 
 // CheckSignature verifies the signature against the provided public key
-func (nc *NebulaCertificate) CheckSignature(key ed25519.PublicKey) bool {
+func (nc *NebulaCertificate) CheckSignature(key []byte) bool {
 	b, err := proto.Marshal(nc.getRawDetails())
 	if err != nil {
 		return false
 	}
-	return ed25519.Verify(key, b, nc.Signature)
+	switch nc.Details.Curve {
+	case Curve_CURVE25519:
+		return ed25519.Verify(ed25519.PublicKey(key), b, nc.Signature)
+	case Curve_P256:
+		x, y := elliptic.Unmarshal(elliptic.P256(), key)
+		pubKey := &ecdsa.PublicKey{Curve: elliptic.P256(), X: x, Y: y}
+		hashed := sha256.Sum256(b)
+		return ecdsa.VerifyASN1(pubKey, hashed[:], nc.Signature)
+	default:
+		return false
+	}
+}
+
+// NOTE: This uses an internal cache that will not be invalidated automatically
+// if you manually change any fields in the NebulaCertificate.
+func (nc *NebulaCertificate) checkSignatureWithCache(key []byte, useCache bool) bool {
+	if !useCache {
+		return nc.CheckSignature(key)
+	}
+
+	if v := nc.signatureVerified.Load(); v != nil {
+		return bytes.Equal(*v, key)
+	}
+
+	verified := nc.CheckSignature(key)
+	if verified {
+		keyCopy := make([]byte, len(key))
+		copy(keyCopy, key)
+		nc.signatureVerified.Store(&keyCopy)
+	}
+
+	return verified
 }
 
 // Expired will return true if the nebula cert is too young or too old compared to the provided time, otherwise false
@@ -253,8 +599,27 @@ func (nc *NebulaCertificate) Expired(t time.Time) bool {
 
 // Verify will ensure a certificate is good in all respects (expiry, group membership, signature, cert blocklist, etc)
 func (nc *NebulaCertificate) Verify(t time.Time, ncp *NebulaCAPool) (bool, error) {
-	if ncp.IsBlocklisted(nc) {
-		return false, fmt.Errorf("certificate has been blocked")
+	return nc.verify(t, ncp, false)
+}
+
+// VerifyWithCache will ensure a certificate is good in all respects (expiry, group membership, signature, cert blocklist, etc)
+//
+// NOTE: This uses an internal cache that will not be invalidated automatically
+// if you manually change any fields in the NebulaCertificate.
+func (nc *NebulaCertificate) VerifyWithCache(t time.Time, ncp *NebulaCAPool) (bool, error) {
+	return nc.verify(t, ncp, true)
+}
+
+// ResetCache resets the cache used by VerifyWithCache.
+func (nc *NebulaCertificate) ResetCache() {
+	nc.sha256sum.Store(nil)
+	nc.signatureVerified.Store(nil)
+}
+
+// Verify will ensure a certificate is good in all respects (expiry, group membership, signature, cert blocklist, etc)
+func (nc *NebulaCertificate) verify(t time.Time, ncp *NebulaCAPool, useCache bool) (bool, error) {
+	if ncp.isBlocklistedWithCache(nc, useCache) {
+		return false, ErrBlockListed
 	}
 
 	signer, err := ncp.GetCAForCert(nc)
@@ -263,15 +628,15 @@ func (nc *NebulaCertificate) Verify(t time.Time, ncp *NebulaCAPool) (bool, error
 	}
 
 	if signer.Expired(t) {
-		return false, fmt.Errorf("root certificate is expired")
+		return false, ErrRootExpired
 	}
 
 	if nc.Expired(t) {
-		return false, fmt.Errorf("certificate is expired")
+		return false, ErrExpired
 	}
 
-	if !nc.CheckSignature(signer.Details.PublicKey) {
-		return false, fmt.Errorf("certificate signature did not match")
+	if !nc.checkSignatureWithCache(signer.Details.PublicKey, useCache) {
+		return false, ErrSignatureMismatch
 	}
 
 	if err := nc.CheckRootConstrains(signer); err != nil {
@@ -324,13 +689,57 @@ func (nc *NebulaCertificate) CheckRootConstrains(signer *NebulaCertificate) erro
 }
 
 // VerifyPrivateKey checks that the public key in the Nebula certificate and a supplied private key match
-func (nc *NebulaCertificate) VerifyPrivateKey(key []byte) error {
-	var dst, key32 [32]byte
-	copy(key32[:], key)
-	curve25519.ScalarBaseMult(&dst, &key32)
-	if !bytes.Equal(dst[:], nc.Details.PublicKey) {
+func (nc *NebulaCertificate) VerifyPrivateKey(curve Curve, key []byte) error {
+	if curve != nc.Details.Curve {
+		return fmt.Errorf("curve in cert and private key supplied don't match")
+	}
+	if nc.Details.IsCA {
+		switch curve {
+		case Curve_CURVE25519:
+			// the call to PublicKey below will panic slice bounds out of range otherwise
+			if len(key) != ed25519.PrivateKeySize {
+				return fmt.Errorf("key was not 64 bytes, is invalid ed25519 private key")
+			}
+
+			if !ed25519.PublicKey(nc.Details.PublicKey).Equal(ed25519.PrivateKey(key).Public()) {
+				return fmt.Errorf("public key in cert and private key supplied don't match")
+			}
+		case Curve_P256:
+			privkey, err := ecdh.P256().NewPrivateKey(key)
+			if err != nil {
+				return fmt.Errorf("cannot parse private key as P256")
+			}
+			pub := privkey.PublicKey().Bytes()
+			if !bytes.Equal(pub, nc.Details.PublicKey) {
+				return fmt.Errorf("public key in cert and private key supplied don't match")
+			}
+		default:
+			return fmt.Errorf("invalid curve: %s", curve)
+		}
+		return nil
+	}
+
+	var pub []byte
+	switch curve {
+	case Curve_CURVE25519:
+		var err error
+		pub, err = curve25519.X25519(key, curve25519.Basepoint)
+		if err != nil {
+			return err
+		}
+	case Curve_P256:
+		privkey, err := ecdh.P256().NewPrivateKey(key)
+		if err != nil {
+			return err
+		}
+		pub = privkey.PublicKey().Bytes()
+	default:
+		return fmt.Errorf("invalid curve: %s", curve)
+	}
+	if !bytes.Equal(pub, nc.Details.PublicKey) {
 		return fmt.Errorf("public key in cert and private key supplied don't match")
 	}
+
 	return nil
 }
 
@@ -379,6 +788,7 @@ func (nc *NebulaCertificate) String() string {
 	s += fmt.Sprintf("\t\tIs CA: %v\n", nc.Details.IsCA)
 	s += fmt.Sprintf("\t\tIssuer: %s\n", nc.Details.Issuer)
 	s += fmt.Sprintf("\t\tPublic key: %x\n", nc.Details.PublicKey)
+	s += fmt.Sprintf("\t\tCurve: %s\n", nc.Details.Curve)
 	s += "\t}\n"
 	fp, err := nc.Sha256Sum()
 	if err == nil {
@@ -399,6 +809,7 @@ func (nc *NebulaCertificate) getRawDetails() *RawNebulaCertificateDetails {
 		NotAfter:  nc.Details.NotAfter.Unix(),
 		PublicKey: make([]byte, len(nc.Details.PublicKey)),
 		IsCA:      nc.Details.IsCA,
+		Curve:     nc.Details.Curve,
 	}
 
 	for _, ipNet := range nc.Details.Ips {
@@ -447,6 +858,25 @@ func (nc *NebulaCertificate) Sha256Sum() (string, error) {
 	return hex.EncodeToString(sum[:]), nil
 }
 
+// NOTE: This uses an internal cache that will not be invalidated automatically
+// if you manually change any fields in the NebulaCertificate.
+func (nc *NebulaCertificate) sha256SumWithCache(useCache bool) (string, error) {
+	if !useCache {
+		return nc.Sha256Sum()
+	}
+
+	if s := nc.sha256sum.Load(); s != nil {
+		return *s, nil
+	}
+	s, err := nc.Sha256Sum()
+	if err != nil {
+		return s, err
+	}
+
+	nc.sha256sum.Store(&s)
+	return s, nil
+}
+
 func (nc *NebulaCertificate) MarshalJSON() ([]byte, error) {
 	toString := func(ips []*net.IPNet) []string {
 		s := []string{}
@@ -468,6 +898,7 @@ func (nc *NebulaCertificate) MarshalJSON() ([]byte, error) {
 			"publicKey": fmt.Sprintf("%x", nc.Details.PublicKey),
 			"isCa":      nc.Details.IsCA,
 			"issuer":    nc.Details.Issuer,
+			"curve":     nc.Details.Curve.String(),
 		},
 		"fingerprint": fp,
 		"signature":   fmt.Sprintf("%x", nc.Signature),

cert/cert.pb.go

@@ -1,7 +1,7 @@
 // Code generated by protoc-gen-go. DO NOT EDIT.
 // versions:
-// 	protoc-gen-go v1.26.0
-// 	protoc        v3.14.0
+// 	protoc-gen-go v1.30.0
+// 	protoc        v3.21.5
 // source: cert.proto
 
 package cert
@@ -20,6 +20,52 @@ const (
 	_ = protoimpl.EnforceVersion(protoimpl.MaxVersion - 20)
 )
 
+type Curve int32
+
+const (
+	Curve_CURVE25519 Curve = 0
+	Curve_P256       Curve = 1
+)
+
+// Enum value maps for Curve.
+var (
+	Curve_name = map[int32]string{
+		0: "CURVE25519",
+		1: "P256",
+	}
+	Curve_value = map[string]int32{
+		"CURVE25519": 0,
+		"P256":       1,
+	}
+)
+
+func (x Curve) Enum() *Curve {
+	p := new(Curve)
+	*p = x
+	return p
+}
+
+func (x Curve) String() string {
+	return protoimpl.X.EnumStringOf(x.Descriptor(), protoreflect.EnumNumber(x))
+}
+
+func (Curve) Descriptor() protoreflect.EnumDescriptor {
+	return file_cert_proto_enumTypes[0].Descriptor()
+}
+
+func (Curve) Type() protoreflect.EnumType {
+	return &file_cert_proto_enumTypes[0]
+}
+
+func (x Curve) Number() protoreflect.EnumNumber {
+	return protoreflect.EnumNumber(x)
+}
+
+// Deprecated: Use Curve.Descriptor instead.
+func (Curve) EnumDescriptor() ([]byte, []int) {
+	return file_cert_proto_rawDescGZIP(), []int{0}
+}
+
 type RawNebulaCertificate struct {
 	state         protoimpl.MessageState
 	sizeCache     protoimpl.SizeCache
@@ -91,6 +137,7 @@ type RawNebulaCertificateDetails struct {
 	IsCA      bool     `protobuf:"varint,8,opt,name=IsCA,proto3" json:"IsCA,omitempty"`
 	// sha-256 of the issuer certificate, if this field is blank the cert is self-signed
 	Issuer []byte `protobuf:"bytes,9,opt,name=Issuer,proto3" json:"Issuer,omitempty"`
+	Curve  Curve  `protobuf:"varint,100,opt,name=curve,proto3,enum=cert.Curve" json:"curve,omitempty"`
 }
 
 func (x *RawNebulaCertificateDetails) Reset() {
@@ -188,6 +235,202 @@ func (x *RawNebulaCertificateDetails) GetIssuer() []byte {
 	return nil
 }
 
+func (x *RawNebulaCertificateDetails) GetCurve() Curve {
+	if x != nil {
+		return x.Curve
+	}
+	return Curve_CURVE25519
+}
+
+type RawNebulaEncryptedData struct {
+	state         protoimpl.MessageState
+	sizeCache     protoimpl.SizeCache
+	unknownFields protoimpl.UnknownFields
+
+	EncryptionMetadata *RawNebulaEncryptionMetadata `protobuf:"bytes,1,opt,name=EncryptionMetadata,proto3" json:"EncryptionMetadata,omitempty"`
+	Ciphertext         []byte                       `protobuf:"bytes,2,opt,name=Ciphertext,proto3" json:"Ciphertext,omitempty"`
+}
+
+func (x *RawNebulaEncryptedData) Reset() {
+	*x = RawNebulaEncryptedData{}
+	if protoimpl.UnsafeEnabled {
+		mi := &file_cert_proto_msgTypes[2]
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		ms.StoreMessageInfo(mi)
+	}
+}
+
+func (x *RawNebulaEncryptedData) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*RawNebulaEncryptedData) ProtoMessage() {}
+
+func (x *RawNebulaEncryptedData) ProtoReflect() protoreflect.Message {
+	mi := &file_cert_proto_msgTypes[2]
+	if protoimpl.UnsafeEnabled && x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use RawNebulaEncryptedData.ProtoReflect.Descriptor instead.
+func (*RawNebulaEncryptedData) Descriptor() ([]byte, []int) {
+	return file_cert_proto_rawDescGZIP(), []int{2}
+}
+
+func (x *RawNebulaEncryptedData) GetEncryptionMetadata() *RawNebulaEncryptionMetadata {
+	if x != nil {
+		return x.EncryptionMetadata
+	}
+	return nil
+}
+
+func (x *RawNebulaEncryptedData) GetCiphertext() []byte {
+	if x != nil {
+		return x.Ciphertext
+	}
+	return nil
+}
+
+type RawNebulaEncryptionMetadata struct {
+	state         protoimpl.MessageState
+	sizeCache     protoimpl.SizeCache
+	unknownFields protoimpl.UnknownFields
+
+	EncryptionAlgorithm string                     `protobuf:"bytes,1,opt,name=EncryptionAlgorithm,proto3" json:"EncryptionAlgorithm,omitempty"`
+	Argon2Parameters    *RawNebulaArgon2Parameters `protobuf:"bytes,2,opt,name=Argon2Parameters,proto3" json:"Argon2Parameters,omitempty"`
+}
+
+func (x *RawNebulaEncryptionMetadata) Reset() {
+	*x = RawNebulaEncryptionMetadata{}
+	if protoimpl.UnsafeEnabled {
+		mi := &file_cert_proto_msgTypes[3]
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		ms.StoreMessageInfo(mi)
+	}
+}
+
+func (x *RawNebulaEncryptionMetadata) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*RawNebulaEncryptionMetadata) ProtoMessage() {}
+
+func (x *RawNebulaEncryptionMetadata) ProtoReflect() protoreflect.Message {
+	mi := &file_cert_proto_msgTypes[3]
+	if protoimpl.UnsafeEnabled && x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use RawNebulaEncryptionMetadata.ProtoReflect.Descriptor instead.
+func (*RawNebulaEncryptionMetadata) Descriptor() ([]byte, []int) {
+	return file_cert_proto_rawDescGZIP(), []int{3}
+}
+
+func (x *RawNebulaEncryptionMetadata) GetEncryptionAlgorithm() string {
+	if x != nil {
+		return x.EncryptionAlgorithm
+	}
+	return ""
+}
+
+func (x *RawNebulaEncryptionMetadata) GetArgon2Parameters() *RawNebulaArgon2Parameters {
+	if x != nil {
+		return x.Argon2Parameters
+	}
+	return nil
+}
+
+type RawNebulaArgon2Parameters struct {
+	state         protoimpl.MessageState
+	sizeCache     protoimpl.SizeCache
+	unknownFields protoimpl.UnknownFields
+
+	Version     int32  `protobuf:"varint,1,opt,name=version,proto3" json:"version,omitempty"` // rune in Go
+	Memory      uint32 `protobuf:"varint,2,opt,name=memory,proto3" json:"memory,omitempty"`
+	Parallelism uint32 `protobuf:"varint,4,opt,name=parallelism,proto3" json:"parallelism,omitempty"` // uint8 in Go
+	Iterations  uint32 `protobuf:"varint,3,opt,name=iterations,proto3" json:"iterations,omitempty"`
+	Salt        []byte `protobuf:"bytes,5,opt,name=salt,proto3" json:"salt,omitempty"`
+}
+
+func (x *RawNebulaArgon2Parameters) Reset() {
+	*x = RawNebulaArgon2Parameters{}
+	if protoimpl.UnsafeEnabled {
+		mi := &file_cert_proto_msgTypes[4]
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		ms.StoreMessageInfo(mi)
+	}
+}
+
+func (x *RawNebulaArgon2Parameters) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*RawNebulaArgon2Parameters) ProtoMessage() {}
+
+func (x *RawNebulaArgon2Parameters) ProtoReflect() protoreflect.Message {
+	mi := &file_cert_proto_msgTypes[4]
+	if protoimpl.UnsafeEnabled && x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use RawNebulaArgon2Parameters.ProtoReflect.Descriptor instead.
+func (*RawNebulaArgon2Parameters) Descriptor() ([]byte, []int) {
+	return file_cert_proto_rawDescGZIP(), []int{4}
+}
+
+func (x *RawNebulaArgon2Parameters) GetVersion() int32 {
+	if x != nil {
+		return x.Version
+	}
+	return 0
+}
+
+func (x *RawNebulaArgon2Parameters) GetMemory() uint32 {
+	if x != nil {
+		return x.Memory
+	}
+	return 0
+}
+
+func (x *RawNebulaArgon2Parameters) GetParallelism() uint32 {
+	if x != nil {
+		return x.Parallelism
+	}
+	return 0
+}
+
+func (x *RawNebulaArgon2Parameters) GetIterations() uint32 {
+	if x != nil {
+		return x.Iterations
+	}
+	return 0
+}
+
+func (x *RawNebulaArgon2Parameters) GetSalt() []byte {
+	if x != nil {
+		return x.Salt
+	}
+	return nil
+}
+
 var File_cert_proto protoreflect.FileDescriptor
 
 var file_cert_proto_rawDesc = []byte{
@@ -199,7 +442,7 @@ var file_cert_proto_rawDesc = []byte{
 	0x69, 0x66, 0x69, 0x63, 0x61, 0x74, 0x65, 0x44, 0x65, 0x74, 0x61, 0x69, 0x6c, 0x73, 0x52, 0x07,
 	0x44, 0x65, 0x74, 0x61, 0x69, 0x6c, 0x73, 0x12, 0x1c, 0x0a, 0x09, 0x53, 0x69, 0x67, 0x6e, 0x61,
 	0x74, 0x75, 0x72, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x09, 0x53, 0x69, 0x67, 0x6e,
-	0x61, 0x74, 0x75, 0x72, 0x65, 0x22, 0xf9, 0x01, 0x0a, 0x1b, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62,
+	0x61, 0x74, 0x75, 0x72, 0x65, 0x22, 0x9c, 0x02, 0x0a, 0x1b, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62,
 	0x75, 0x6c, 0x61, 0x43, 0x65, 0x72, 0x74, 0x69, 0x66, 0x69, 0x63, 0x61, 0x74, 0x65, 0x44, 0x65,
 	0x74, 0x61, 0x69, 0x6c, 0x73, 0x12, 0x12, 0x0a, 0x04, 0x4e, 0x61, 0x6d, 0x65, 0x18, 0x01, 0x20,
 	0x01, 0x28, 0x09, 0x52, 0x04, 0x4e, 0x61, 0x6d, 0x65, 0x12, 0x10, 0x0a, 0x03, 0x49, 0x70, 0x73,
@@ -215,9 +458,43 @@ var file_cert_proto_rawDesc = []byte{
 	0x69, 0x63, 0x4b, 0x65, 0x79, 0x12, 0x12, 0x0a, 0x04, 0x49, 0x73, 0x43, 0x41, 0x18, 0x08, 0x20,
 	0x01, 0x28, 0x08, 0x52, 0x04, 0x49, 0x73, 0x43, 0x41, 0x12, 0x16, 0x0a, 0x06, 0x49, 0x73, 0x73,
 	0x75, 0x65, 0x72, 0x18, 0x09, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x06, 0x49, 0x73, 0x73, 0x75, 0x65,
-	0x72, 0x42, 0x20, 0x5a, 0x1e, 0x67, 0x69, 0x74, 0x68, 0x75, 0x62, 0x2e, 0x63, 0x6f, 0x6d, 0x2f,
-	0x73, 0x6c, 0x61, 0x63, 0x6b, 0x68, 0x71, 0x2f, 0x6e, 0x65, 0x62, 0x75, 0x6c, 0x61, 0x2f, 0x63,
-	0x65, 0x72, 0x74, 0x62, 0x06, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x33,
+	0x72, 0x12, 0x21, 0x0a, 0x05, 0x63, 0x75, 0x72, 0x76, 0x65, 0x18, 0x64, 0x20, 0x01, 0x28, 0x0e,
+	0x32, 0x0b, 0x2e, 0x63, 0x65, 0x72, 0x74, 0x2e, 0x43, 0x75, 0x72, 0x76, 0x65, 0x52, 0x05, 0x63,
+	0x75, 0x72, 0x76, 0x65, 0x22, 0x8b, 0x01, 0x0a, 0x16, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62, 0x75,
+	0x6c, 0x61, 0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x65, 0x64, 0x44, 0x61, 0x74, 0x61, 0x12,
+	0x51, 0x0a, 0x12, 0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x4d, 0x65, 0x74,
+	0x61, 0x64, 0x61, 0x74, 0x61, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x21, 0x2e, 0x63, 0x65,
+	0x72, 0x74, 0x2e, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62, 0x75, 0x6c, 0x61, 0x45, 0x6e, 0x63, 0x72,
+	0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x4d, 0x65, 0x74, 0x61, 0x64, 0x61, 0x74, 0x61, 0x52, 0x12,
+	0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x4d, 0x65, 0x74, 0x61, 0x64, 0x61,
+	0x74, 0x61, 0x12, 0x1e, 0x0a, 0x0a, 0x43, 0x69, 0x70, 0x68, 0x65, 0x72, 0x74, 0x65, 0x78, 0x74,
+	0x18, 0x02, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x0a, 0x43, 0x69, 0x70, 0x68, 0x65, 0x72, 0x74, 0x65,
+	0x78, 0x74, 0x22, 0x9c, 0x01, 0x0a, 0x1b, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62, 0x75, 0x6c, 0x61,
+	0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x4d, 0x65, 0x74, 0x61, 0x64, 0x61,
+	0x74, 0x61, 0x12, 0x30, 0x0a, 0x13, 0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e,
+	0x41, 0x6c, 0x67, 0x6f, 0x72, 0x69, 0x74, 0x68, 0x6d, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52,
+	0x13, 0x45, 0x6e, 0x63, 0x72, 0x79, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x41, 0x6c, 0x67, 0x6f, 0x72,
+	0x69, 0x74, 0x68, 0x6d, 0x12, 0x4b, 0x0a, 0x10, 0x41, 0x72, 0x67, 0x6f, 0x6e, 0x32, 0x50, 0x61,
+	0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x1f,
+	0x2e, 0x63, 0x65, 0x72, 0x74, 0x2e, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62, 0x75, 0x6c, 0x61, 0x41,
+	0x72, 0x67, 0x6f, 0x6e, 0x32, 0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x52,
+	0x10, 0x41, 0x72, 0x67, 0x6f, 0x6e, 0x32, 0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72,
+	0x73, 0x22, 0xa3, 0x01, 0x0a, 0x19, 0x52, 0x61, 0x77, 0x4e, 0x65, 0x62, 0x75, 0x6c, 0x61, 0x41,
+	0x72, 0x67, 0x6f, 0x6e, 0x32, 0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x12,
+	0x18, 0x0a, 0x07, 0x76, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x18, 0x01, 0x20, 0x01, 0x28, 0x05,
+	0x52, 0x07, 0x76, 0x65, 0x72, 0x73, 0x69, 0x6f, 0x6e, 0x12, 0x16, 0x0a, 0x06, 0x6d, 0x65, 0x6d,
+	0x6f, 0x72, 0x79, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x06, 0x6d, 0x65, 0x6d, 0x6f, 0x72,
+	0x79, 0x12, 0x20, 0x0a, 0x0b, 0x70, 0x61, 0x72, 0x61, 0x6c, 0x6c, 0x65, 0x6c, 0x69, 0x73, 0x6d,
+	0x18, 0x04, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x0b, 0x70, 0x61, 0x72, 0x61, 0x6c, 0x6c, 0x65, 0x6c,
+	0x69, 0x73, 0x6d, 0x12, 0x1e, 0x0a, 0x0a, 0x69, 0x74, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e,
+	0x73, 0x18, 0x03, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x0a, 0x69, 0x74, 0x65, 0x72, 0x61, 0x74, 0x69,
+	0x6f, 0x6e, 0x73, 0x12, 0x12, 0x0a, 0x04, 0x73, 0x61, 0x6c, 0x74, 0x18, 0x05, 0x20, 0x01, 0x28,
+	0x0c, 0x52, 0x04, 0x73, 0x61, 0x6c, 0x74, 0x2a, 0x21, 0x0a, 0x05, 0x43, 0x75, 0x72, 0x76, 0x65,
+	0x12, 0x0e, 0x0a, 0x0a, 0x43, 0x55, 0x52, 0x56, 0x45, 0x32, 0x35, 0x35, 0x31, 0x39, 0x10, 0x00,
+	0x12, 0x08, 0x0a, 0x04, 0x50, 0x32, 0x35, 0x36, 0x10, 0x01, 0x42, 0x20, 0x5a, 0x1e, 0x67, 0x69,
+	0x74, 0x68, 0x75, 0x62, 0x2e, 0x63, 0x6f, 0x6d, 0x2f, 0x73, 0x6c, 0x61, 0x63, 0x6b, 0x68, 0x71,
+	0x2f, 0x6e, 0x65, 0x62, 0x75, 0x6c, 0x61, 0x2f, 0x63, 0x65, 0x72, 0x74, 0x62, 0x06, 0x70, 0x72,
+	0x6f, 0x74, 0x6f, 0x33,
 }
 
 var (
@@ -232,18 +509,26 @@ func file_cert_proto_rawDescGZIP() []byte {
 	return file_cert_proto_rawDescData
 }
 
-var file_cert_proto_msgTypes = make([]protoimpl.MessageInfo, 2)
+var file_cert_proto_enumTypes = make([]protoimpl.EnumInfo, 1)
+var file_cert_proto_msgTypes = make([]protoimpl.MessageInfo, 5)
 var file_cert_proto_goTypes = []interface{}{
-	(*RawNebulaCertificate)(nil),        // 0: cert.RawNebulaCertificate
-	(*RawNebulaCertificateDetails)(nil), // 1: cert.RawNebulaCertificateDetails
+	(Curve)(0),                          // 0: cert.Curve
+	(*RawNebulaCertificate)(nil),        // 1: cert.RawNebulaCertificate
+	(*RawNebulaCertificateDetails)(nil), // 2: cert.RawNebulaCertificateDetails
+	(*RawNebulaEncryptedData)(nil),      // 3: cert.RawNebulaEncryptedData
+	(*RawNebulaEncryptionMetadata)(nil), // 4: cert.RawNebulaEncryptionMetadata
+	(*RawNebulaArgon2Parameters)(nil),   // 5: cert.RawNebulaArgon2Parameters
 }
 var file_cert_proto_depIdxs = []int32{
-	1, // 0: cert.RawNebulaCertificate.Details:type_name -> cert.RawNebulaCertificateDetails
-	1, // [1:1] is the sub-list for method output_type
-	1, // [1:1] is the sub-list for method input_type
-	1, // [1:1] is the sub-list for extension type_name
-	1, // [1:1] is the sub-list for extension extendee
-	0, // [0:1] is the sub-list for field type_name
+	2, // 0: cert.RawNebulaCertificate.Details:type_name -> cert.RawNebulaCertificateDetails
+	0, // 1: cert.RawNebulaCertificateDetails.curve:type_name -> cert.Curve
+	4, // 2: cert.RawNebulaEncryptedData.EncryptionMetadata:type_name -> cert.RawNebulaEncryptionMetadata
+	5, // 3: cert.RawNebulaEncryptionMetadata.Argon2Parameters:type_name -> cert.RawNebulaArgon2Parameters
+	4, // [4:4] is the sub-list for method output_type
+	4, // [4:4] is the sub-list for method input_type
+	4, // [4:4] is the sub-list for extension type_name
+	4, // [4:4] is the sub-list for extension extendee
+	0, // [0:4] is the sub-list for field type_name
 }
 
 func init() { file_cert_proto_init() }
@@ -276,19 +561,56 @@ func file_cert_proto_init() {
 				return nil
 			}
 		}
+		file_cert_proto_msgTypes[2].Exporter = func(v interface{}, i int) interface{} {
+			switch v := v.(*RawNebulaEncryptedData); i {
+			case 0:
+				return &v.state
+			case 1:
+				return &v.sizeCache
+			case 2:
+				return &v.unknownFields
+			default:
+				return nil
+			}
+		}
+		file_cert_proto_msgTypes[3].Exporter = func(v interface{}, i int) interface{} {
+			switch v := v.(*RawNebulaEncryptionMetadata); i {
+			case 0:
+				return &v.state
+			case 1:
+				return &v.sizeCache
+			case 2:
+				return &v.unknownFields
+			default:
+				return nil
+			}
+		}
+		file_cert_proto_msgTypes[4].Exporter = func(v interface{}, i int) interface{} {
+			switch v := v.(*RawNebulaArgon2Parameters); i {
+			case 0:
+				return &v.state
+			case 1:
+				return &v.sizeCache
+			case 2:
+				return &v.unknownFields
+			default:
+				return nil
+			}
+		}
 	}
 	type x struct{}
 	out := protoimpl.TypeBuilder{
 		File: protoimpl.DescBuilder{
 			GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
 			RawDescriptor: file_cert_proto_rawDesc,
-			NumEnums:      0,
-			NumMessages:   2,
+			NumEnums:      1,
+			NumMessages:   5,
 			NumExtensions: 0,
 			NumServices:   0,
 		},
 		GoTypes:           file_cert_proto_goTypes,
 		DependencyIndexes: file_cert_proto_depIdxs,
+		EnumInfos:         file_cert_proto_enumTypes,
 		MessageInfos:      file_cert_proto_msgTypes,
 	}.Build()
 	File_cert_proto = out.File

cert/cert.proto

@@ -5,6 +5,11 @@ option go_package = "github.com/slackhq/nebula/cert";
 
 //import "google/protobuf/timestamp.proto";
 
+enum Curve {
+  CURVE25519 = 0;
+  P256 = 1;
+}
+
 message RawNebulaCertificate {
     RawNebulaCertificateDetails Details = 1;
     bytes Signature = 2;
@@ -26,4 +31,24 @@ message RawNebulaCertificateDetails {
 
     // sha-256 of the issuer certificate, if this field is blank the cert is self-signed
     bytes Issuer = 9;
-}
+
+    Curve curve = 100;
+}
+
+message RawNebulaEncryptedData {
+	RawNebulaEncryptionMetadata EncryptionMetadata = 1;
+	bytes Ciphertext = 2;
+}
+
+message RawNebulaEncryptionMetadata {
+	string EncryptionAlgorithm = 1;
+	RawNebulaArgon2Parameters Argon2Parameters = 2;
+}
+
+message RawNebulaArgon2Parameters {
+	int32 version = 1; // rune in Go
+	uint32 memory = 2;
+	uint32 parallelism = 4; // uint8 in Go
+	uint32 iterations = 3;
+	bytes salt = 5;
+}

cert/cert_test.go

@@ -1,6 +1,9 @@
 package cert
 
 import (
+	"crypto/ecdh"
+	"crypto/ecdsa"
+	"crypto/elliptic"
 	"crypto/rand"
 	"fmt"
 	"io"
@@ -8,11 +11,11 @@ import (
 	"testing"
 	"time"
 
-	"github.com/golang/protobuf/proto"
-	"github.com/slackhq/nebula/util"
+	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
 	"golang.org/x/crypto/curve25519"
 	"golang.org/x/crypto/ed25519"
+	"google.golang.org/protobuf/proto"
 )
 
 func TestMarshalingNebulaCertificate(t *testing.T) {
@@ -101,7 +104,49 @@ func TestNebulaCertificate_Sign(t *testing.T) {
 	pub, priv, err := ed25519.GenerateKey(rand.Reader)
 	assert.Nil(t, err)
 	assert.False(t, nc.CheckSignature(pub))
-	assert.Nil(t, nc.Sign(priv))
+	assert.Nil(t, nc.Sign(Curve_CURVE25519, priv))
+	assert.True(t, nc.CheckSignature(pub))
+
+	_, err = nc.Marshal()
+	assert.Nil(t, err)
+	//t.Log("Cert size:", len(b))
+}
+
+func TestNebulaCertificate_SignP256(t *testing.T) {
+	before := time.Now().Add(time.Second * -60).Round(time.Second)
+	after := time.Now().Add(time.Second * 60).Round(time.Second)
+	pubKey := []byte("01234567890abcedfghij1234567890ab1234567890abcedfghij1234567890ab")
+
+	nc := NebulaCertificate{
+		Details: NebulaCertificateDetails{
+			Name: "testing",
+			Ips: []*net.IPNet{
+				{IP: net.ParseIP("10.1.1.1"), Mask: net.IPMask(net.ParseIP("255.255.255.0"))},
+				{IP: net.ParseIP("10.1.1.2"), Mask: net.IPMask(net.ParseIP("255.255.0.0"))},
+				{IP: net.ParseIP("10.1.1.3"), Mask: net.IPMask(net.ParseIP("255.0.255.0"))},
+			},
+			Subnets: []*net.IPNet{
+				{IP: net.ParseIP("9.1.1.1"), Mask: net.IPMask(net.ParseIP("255.0.255.0"))},
+				{IP: net.ParseIP("9.1.1.2"), Mask: net.IPMask(net.ParseIP("255.255.255.0"))},
+				{IP: net.ParseIP("9.1.1.3"), Mask: net.IPMask(net.ParseIP("255.255.0.0"))},
+			},
+			Groups:    []string{"test-group1", "test-group2", "test-group3"},
+			NotBefore: before,
+			NotAfter:  after,
+			PublicKey: pubKey,
+			IsCA:      false,
+			Curve:     Curve_P256,
+			Issuer:    "1234567890abcedfghij1234567890ab",
+		},
+	}
+
+	priv, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	pub := elliptic.Marshal(elliptic.P256(), priv.PublicKey.X, priv.PublicKey.Y)
+	rawPriv := priv.D.FillBytes(make([]byte, 32))
+
+	assert.Nil(t, err)
+	assert.False(t, nc.CheckSignature(pub))
+	assert.Nil(t, nc.Sign(Curve_P256, rawPriv))
 	assert.True(t, nc.CheckSignature(pub))
 
 	_, err = nc.Marshal()
@@ -153,7 +198,7 @@ func TestNebulaCertificate_MarshalJSON(t *testing.T) {
 	assert.Nil(t, err)
 	assert.Equal(
 		t,
-		"{\"details\":{\"groups\":[\"test-group1\",\"test-group2\",\"test-group3\"],\"ips\":[\"10.1.1.1/24\",\"10.1.1.2/16\",\"10.1.1.3/ff00ff00\"],\"isCa\":false,\"issuer\":\"1234567890abcedfghij1234567890ab\",\"name\":\"testing\",\"notAfter\":\"0000-11-30T02:00:00Z\",\"notBefore\":\"0000-11-30T01:00:00Z\",\"publicKey\":\"313233343536373839306162636564666768696a313233343536373839306162\",\"subnets\":[\"9.1.1.1/ff00ff00\",\"9.1.1.2/24\",\"9.1.1.3/16\"]},\"fingerprint\":\"26cb1c30ad7872c804c166b5150fa372f437aa3856b04edb4334b4470ec728e4\",\"signature\":\"313233343536373839306162636564666768696a313233343536373839306162\"}",
+		"{\"details\":{\"curve\":\"CURVE25519\",\"groups\":[\"test-group1\",\"test-group2\",\"test-group3\"],\"ips\":[\"10.1.1.1/24\",\"10.1.1.2/16\",\"10.1.1.3/ff00ff00\"],\"isCa\":false,\"issuer\":\"1234567890abcedfghij1234567890ab\",\"name\":\"testing\",\"notAfter\":\"0000-11-30T02:00:00Z\",\"notBefore\":\"0000-11-30T01:00:00Z\",\"publicKey\":\"313233343536373839306162636564666768696a313233343536373839306162\",\"subnets\":[\"9.1.1.1/ff00ff00\",\"9.1.1.2/24\",\"9.1.1.3/16\"]},\"fingerprint\":\"26cb1c30ad7872c804c166b5150fa372f437aa3856b04edb4334b4470ec728e4\",\"signature\":\"313233343536373839306162636564666768696a313233343536373839306162\"}",
 		string(b),
 	)
 }
@@ -177,7 +222,7 @@ func TestNebulaCertificate_Verify(t *testing.T) {
 
 	v, err := c.Verify(time.Now(), caPool)
 	assert.False(t, v)
-	assert.EqualError(t, err, "certificate has been blocked")
+	assert.EqualError(t, err, "certificate is in the block list")
 
 	caPool.ResetCertBlocklist()
 	v, err = c.Verify(time.Now(), caPool)
@@ -217,6 +262,65 @@ func TestNebulaCertificate_Verify(t *testing.T) {
 	assert.Nil(t, err)
 }
 
+func TestNebulaCertificate_VerifyP256(t *testing.T) {
+	ca, _, caKey, err := newTestCaCertP256(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+
+	c, _, _, err := newTestCert(ca, caKey, time.Now(), time.Now().Add(5*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+
+	h, err := ca.Sha256Sum()
+	assert.Nil(t, err)
+
+	caPool := NewCAPool()
+	caPool.CAs[h] = ca
+
+	f, err := c.Sha256Sum()
+	assert.Nil(t, err)
+	caPool.BlocklistFingerprint(f)
+
+	v, err := c.Verify(time.Now(), caPool)
+	assert.False(t, v)
+	assert.EqualError(t, err, "certificate is in the block list")
+
+	caPool.ResetCertBlocklist()
+	v, err = c.Verify(time.Now(), caPool)
+	assert.True(t, v)
+	assert.Nil(t, err)
+
+	v, err = c.Verify(time.Now().Add(time.Hour*1000), caPool)
+	assert.False(t, v)
+	assert.EqualError(t, err, "root certificate is expired")
+
+	c, _, _, err = newTestCert(ca, caKey, time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+	v, err = c.Verify(time.Now().Add(time.Minute*6), caPool)
+	assert.False(t, v)
+	assert.EqualError(t, err, "certificate is expired")
+
+	// Test group assertion
+	ca, _, caKey, err = newTestCaCertP256(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{"test1", "test2"})
+	assert.Nil(t, err)
+
+	caPem, err := ca.MarshalToPEM()
+	assert.Nil(t, err)
+
+	caPool = NewCAPool()
+	caPool.AddCACertificate(caPem)
+
+	c, _, _, err = newTestCert(ca, caKey, time.Now(), time.Now().Add(5*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{"test1", "bad"})
+	assert.Nil(t, err)
+	v, err = c.Verify(time.Now(), caPool)
+	assert.False(t, v)
+	assert.EqualError(t, err, "certificate contained a group not present on the signing ca: bad")
+
+	c, _, _, err = newTestCert(ca, caKey, time.Now(), time.Now().Add(5*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{"test1"})
+	assert.Nil(t, err)
+	v, err = c.Verify(time.Now(), caPool)
+	assert.True(t, v)
+	assert.Nil(t, err)
+}
+
 func TestNebulaCertificate_Verify_IPs(t *testing.T) {
 	_, caIp1, _ := net.ParseCIDR("10.0.0.0/16")
 	_, caIp2, _ := net.ParseCIDR("192.168.0.0/24")
@@ -375,16 +479,43 @@ func TestNebulaCertificate_Verify_Subnets(t *testing.T) {
 	assert.Nil(t, err)
 }
 
-func TestNebulaVerifyPrivateKey(t *testing.T) {
+func TestNebulaCertificate_VerifyPrivateKey(t *testing.T) {
 	ca, _, caKey, err := newTestCaCert(time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
 	assert.Nil(t, err)
+	err = ca.VerifyPrivateKey(Curve_CURVE25519, caKey)
+	assert.Nil(t, err)
+
+	_, _, caKey2, err := newTestCaCert(time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+	err = ca.VerifyPrivateKey(Curve_CURVE25519, caKey2)
+	assert.NotNil(t, err)
 
 	c, _, priv, err := newTestCert(ca, caKey, time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
-	err = c.VerifyPrivateKey(priv)
+	err = c.VerifyPrivateKey(Curve_CURVE25519, priv)
 	assert.Nil(t, err)
 
 	_, priv2 := x25519Keypair()
-	err = c.VerifyPrivateKey(priv2)
+	err = c.VerifyPrivateKey(Curve_CURVE25519, priv2)
+	assert.NotNil(t, err)
+}
+
+func TestNebulaCertificate_VerifyPrivateKeyP256(t *testing.T) {
+	ca, _, caKey, err := newTestCaCertP256(time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+	err = ca.VerifyPrivateKey(Curve_P256, caKey)
+	assert.Nil(t, err)
+
+	_, _, caKey2, err := newTestCaCertP256(time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
+	assert.Nil(t, err)
+	err = ca.VerifyPrivateKey(Curve_P256, caKey2)
+	assert.NotNil(t, err)
+
+	c, _, priv, err := newTestCert(ca, caKey, time.Time{}, time.Time{}, []*net.IPNet{}, []*net.IPNet{}, []string{})
+	err = c.VerifyPrivateKey(Curve_P256, priv)
+	assert.Nil(t, err)
+
+	_, priv2 := p256Keypair()
+	err = c.VerifyPrivateKey(Curve_P256, priv2)
 	assert.NotNil(t, err)
 }
 
@@ -422,6 +553,25 @@ BVG+oJpAoqokUBbI4U0N8CSfpUABEkB/Pm5A2xyH/nc8mg/wvGUWG3pZ7nHzaDMf
 8/phAUt+FLzqTECzQKisYswKvE3pl9mbEYKbOdIHrxdIp95mo4sF
 -----END NEBULA CERTIFICATE-----
 
+`
+
+	expired := `
+# expired certificate
+-----BEGIN NEBULA CERTIFICATE-----
+CjkKB2V4cGlyZWQouPmWjQYwufmWjQY6ILCRaoCkJlqHgv5jfDN4lzLHBvDzaQm4
+vZxfu144hmgjQAESQG4qlnZi8DncvD/LDZnLgJHOaX1DWCHHEh59epVsC+BNgTie
+WH1M9n4O7cFtGlM6sJJOS+rCVVEJ3ABS7+MPdQs=
+-----END NEBULA CERTIFICATE-----
+`
+
+	p256 := `
+# p256 certificate
+-----BEGIN NEBULA CERTIFICATE-----
+CmYKEG5lYnVsYSBQMjU2IHRlc3Qo4s+7mgYw4tXrsAc6QQRkaW2jFmllYvN4+/k2
+6tctO9sPT3jOx8ES6M1nIqOhpTmZeabF/4rELDqPV4aH5jfJut798DUXql0FlF8H
+76gvQAGgBgESRzBFAiEAib0/te6eMiZOKD8gdDeloMTS0wGuX2t0C7TFdUhAQzgC
+IBNWYMep3ysx9zCgknfG5dKtwGTaqF++BWKDYdyl34KX
+-----END NEBULA CERTIFICATE-----
 `
 
 	rootCA := NebulaCertificate{
@@ -436,6 +586,12 @@ BVG+oJpAoqokUBbI4U0N8CSfpUABEkB/Pm5A2xyH/nc8mg/wvGUWG3pZ7nHzaDMf
 		},
 	}
 
+	rootCAP256 := NebulaCertificate{
+		Details: NebulaCertificateDetails{
+			Name: "nebula P256 test",
+		},
+	}
+
 	p, err := NewCAPoolFromBytes([]byte(noNewLines))
 	assert.Nil(t, err)
 	assert.Equal(t, p.CAs[string("c9bfaf7ce8e84b2eeda2e27b469f4b9617bde192efd214b68891ecda6ed49522")].Details.Name, rootCA.Details.Name)
@@ -445,6 +601,24 @@ BVG+oJpAoqokUBbI4U0N8CSfpUABEkB/Pm5A2xyH/nc8mg/wvGUWG3pZ7nHzaDMf
 	assert.Nil(t, err)
 	assert.Equal(t, pp.CAs[string("c9bfaf7ce8e84b2eeda2e27b469f4b9617bde192efd214b68891ecda6ed49522")].Details.Name, rootCA.Details.Name)
 	assert.Equal(t, pp.CAs[string("5c9c3f23e7ee7fe97637cbd3a0a5b854154d1d9aaaf7b566a51f4a88f76b64cd")].Details.Name, rootCA01.Details.Name)
+
+	// expired cert, no valid certs
+	ppp, err := NewCAPoolFromBytes([]byte(expired))
+	assert.Equal(t, ErrExpired, err)
+	assert.Equal(t, ppp.CAs[string("152070be6bb19bc9e3bde4c2f0e7d8f4ff5448b4c9856b8eccb314fade0229b0")].Details.Name, "expired")
+
+	// expired cert, with valid certs
+	pppp, err := NewCAPoolFromBytes(append([]byte(expired), noNewLines...))
+	assert.Equal(t, ErrExpired, err)
+	assert.Equal(t, pppp.CAs[string("c9bfaf7ce8e84b2eeda2e27b469f4b9617bde192efd214b68891ecda6ed49522")].Details.Name, rootCA.Details.Name)
+	assert.Equal(t, pppp.CAs[string("5c9c3f23e7ee7fe97637cbd3a0a5b854154d1d9aaaf7b566a51f4a88f76b64cd")].Details.Name, rootCA01.Details.Name)
+	assert.Equal(t, pppp.CAs[string("152070be6bb19bc9e3bde4c2f0e7d8f4ff5448b4c9856b8eccb314fade0229b0")].Details.Name, "expired")
+	assert.Equal(t, len(pppp.CAs), 3)
+
+	ppppp, err := NewCAPoolFromBytes([]byte(p256))
+	assert.Nil(t, err)
+	assert.Equal(t, ppppp.CAs[string("a7938893ec8c4ef769b06d7f425e5e46f7a7f5ffa49c3bcf4a86b608caba9159")].Details.Name, rootCAP256.Details.Name)
+	assert.Equal(t, len(ppppp.CAs), 1)
 }
 
 func appendByteSlices(b ...[]byte) []byte {
@@ -500,11 +674,16 @@ bzBEr00kERQxxTzTsH8cpYEgRoipvmExvg8WP8NdAJEYJosB
 	assert.EqualError(t, err, "input did not contain a valid PEM encoded block")
 }
 
-func TestUnmarshalEd25519PrivateKey(t *testing.T) {
+func TestUnmarshalSigningPrivateKey(t *testing.T) {
 	privKey := []byte(`# A good key
 -----BEGIN NEBULA ED25519 PRIVATE KEY-----
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA==
 -----END NEBULA ED25519 PRIVATE KEY-----
+`)
+	privP256Key := []byte(`# A good key
+-----BEGIN NEBULA ECDSA P256 PRIVATE KEY-----
+AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
+-----END NEBULA ECDSA P256 PRIVATE KEY-----
 `)
 	shortKey := []byte(`# A short key
 -----BEGIN NEBULA ED25519 PRIVATE KEY-----
@@ -521,39 +700,139 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA==
 -END NEBULA ED25519 PRIVATE KEY-----`)
 
-	keyBundle := appendByteSlices(privKey, shortKey, invalidBanner, invalidPem)
+	keyBundle := appendByteSlices(privKey, privP256Key, shortKey, invalidBanner, invalidPem)
 
 	// Success test case
-	k, rest, err := UnmarshalEd25519PrivateKey(keyBundle)
+	k, rest, curve, err := UnmarshalSigningPrivateKey(keyBundle)
 	assert.Len(t, k, 64)
+	assert.Equal(t, rest, appendByteSlices(privP256Key, shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_CURVE25519, curve)
+	assert.Nil(t, err)
+
+	// Success test case
+	k, rest, curve, err = UnmarshalSigningPrivateKey(rest)
+	assert.Len(t, k, 32)
 	assert.Equal(t, rest, appendByteSlices(shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_P256, curve)
 	assert.Nil(t, err)
 
 	// Fail due to short key
-	k, rest, err = UnmarshalEd25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalSigningPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, appendByteSlices(invalidBanner, invalidPem))
-	assert.EqualError(t, err, "key was not 64 bytes, is invalid ed25519 private key")
+	assert.EqualError(t, err, "key was not 64 bytes, is invalid Ed25519 private key")
 
 	// Fail due to invalid banner
-	k, rest, err = UnmarshalEd25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalSigningPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
-	assert.EqualError(t, err, "bytes did not contain a proper nebula Ed25519 private key banner")
+	assert.EqualError(t, err, "bytes did not contain a proper nebula Ed25519/ECDSA private key banner")
 
 	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
-	k, rest, err = UnmarshalEd25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalSigningPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
 	assert.EqualError(t, err, "input did not contain a valid PEM encoded block")
 }
 
-func TestUnmarshalX25519PrivateKey(t *testing.T) {
+func TestDecryptAndUnmarshalSigningPrivateKey(t *testing.T) {
+	passphrase := []byte("DO NOT USE THIS KEY")
+	privKey := []byte(`# A good key
+-----BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
+CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCognnjujd67Vsv99p22wfAjQaDT
+oCMW1mdjkU3gACKNW4MSXOWR9Sts4C81yk1RUku2gvGKs3TB9LYoklLsIizSYOLl
++Vs//O1T0I1Xbml2XBAROsb/VSoDln/6LMqR4B6fn6B3GOsLBBqRI8daDl9lRMPB
+qrlJ69wer3ZUHFXA
+-----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=
+-----END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
+`)
+	invalidBanner := []byte(`# Invalid banner (not encrypted)
+-----BEGIN NEBULA ED25519 PRIVATE KEY-----
+bWRp2CTVFhW9HD/qCd28ltDgK3w8VXSeaEYczDWos8sMUBqDb9jP3+NYwcS4lURG
+XgLvodMXZJuaFPssp+WwtA==
+-----END NEBULA ED25519 PRIVATE KEY-----
+`)
+	invalidPem := []byte(`# Not a valid PEM format
+-BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
+CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCognnjujd67Vsv99p22wfAjQaDT
+oCMW1mdjkU3gACKNW4MSXOWR9Sts4C81yk1RUku2gvGKs3TB9LYoklLsIizSYOLl
++Vs//O1T0I1Xbml2XBAROsb/VSoDln/6LMqR4B6fn6B3GOsLBBqRI8daDl9lRMPB
+qrlJ69wer3ZUHFXA
+-END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
+`)
+
+	keyBundle := appendByteSlices(privKey, shortKey, invalidBanner, invalidPem)
+
+	// Success test case
+	curve, k, rest, err := DecryptAndUnmarshalSigningPrivateKey(passphrase, keyBundle)
+	assert.Nil(t, err)
+	assert.Equal(t, Curve_CURVE25519, curve)
+	assert.Len(t, k, 64)
+	assert.Equal(t, rest, appendByteSlices(shortKey, invalidBanner, invalidPem))
+
+	// Fail due to short key
+	curve, k, rest, err = DecryptAndUnmarshalSigningPrivateKey(passphrase, rest)
+	assert.EqualError(t, err, "key was not 64 bytes, is invalid ed25519 private key")
+	assert.Nil(t, k)
+	assert.Equal(t, rest, appendByteSlices(invalidBanner, invalidPem))
+
+	// Fail due to invalid banner
+	curve, k, rest, err = DecryptAndUnmarshalSigningPrivateKey(passphrase, rest)
+	assert.EqualError(t, err, "bytes did not contain a proper nebula encrypted Ed25519/ECDSA private key banner")
+	assert.Nil(t, k)
+	assert.Equal(t, rest, invalidPem)
+
+	// Fail due to ivalid PEM format, because
+	// it's missing the requisite pre-encapsulation boundary.
+	curve, k, rest, err = DecryptAndUnmarshalSigningPrivateKey(passphrase, rest)
+	assert.EqualError(t, err, "input did not contain a valid PEM encoded block")
+	assert.Nil(t, k)
+	assert.Equal(t, rest, invalidPem)
+
+	// Fail due to invalid passphrase
+	curve, k, rest, err = DecryptAndUnmarshalSigningPrivateKey([]byte("invalid passphrase"), privKey)
+	assert.EqualError(t, err, "invalid passphrase or corrupt private key")
+	assert.Nil(t, k)
+	assert.Equal(t, rest, []byte{})
+}
+
+func TestEncryptAndMarshalSigningPrivateKey(t *testing.T) {
+	// Having proved that decryption works correctly above, we can test the
+	// encryption function produces a value which can be decrypted
+	passphrase := []byte("passphrase")
+	bytes := []byte("AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA")
+	kdfParams := NewArgon2Parameters(64*1024, 4, 3)
+	key, err := EncryptAndMarshalSigningPrivateKey(Curve_CURVE25519, bytes, passphrase, kdfParams)
+	assert.Nil(t, err)
+
+	// Verify the "key" can be decrypted successfully
+	curve, k, rest, err := DecryptAndUnmarshalSigningPrivateKey(passphrase, key)
+	assert.Len(t, k, 64)
+	assert.Equal(t, Curve_CURVE25519, curve)
+	assert.Equal(t, rest, []byte{})
+	assert.Nil(t, err)
+
+	// EncryptAndMarshalEd25519PrivateKey does not create any errors itself
+}
+
+func TestUnmarshalPrivateKey(t *testing.T) {
 	privKey := []byte(`# A good key
 -----BEGIN NEBULA X25519 PRIVATE KEY-----
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 -----END NEBULA X25519 PRIVATE KEY-----
+`)
+	privP256Key := []byte(`# A good key
+-----BEGIN NEBULA P256 PRIVATE KEY-----
+AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
+-----END NEBULA P256 PRIVATE KEY-----
 `)
 	shortKey := []byte(`# A short key
 -----BEGIN NEBULA X25519 PRIVATE KEY-----
@@ -570,29 +849,37 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 -END NEBULA X25519 PRIVATE KEY-----`)
 
-	keyBundle := appendByteSlices(privKey, shortKey, invalidBanner, invalidPem)
+	keyBundle := appendByteSlices(privKey, privP256Key, shortKey, invalidBanner, invalidPem)
 
 	// Success test case
-	k, rest, err := UnmarshalX25519PrivateKey(keyBundle)
+	k, rest, curve, err := UnmarshalPrivateKey(keyBundle)
+	assert.Len(t, k, 32)
+	assert.Equal(t, rest, appendByteSlices(privP256Key, shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_CURVE25519, curve)
+	assert.Nil(t, err)
+
+	// Success test case
+	k, rest, curve, err = UnmarshalPrivateKey(rest)
 	assert.Len(t, k, 32)
 	assert.Equal(t, rest, appendByteSlices(shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_P256, curve)
 	assert.Nil(t, err)
 
 	// Fail due to short key
-	k, rest, err = UnmarshalX25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, appendByteSlices(invalidBanner, invalidPem))
-	assert.EqualError(t, err, "key was not 32 bytes, is invalid X25519 private key")
+	assert.EqualError(t, err, "key was not 32 bytes, is invalid CURVE25519 private key")
 
 	// Fail due to invalid banner
-	k, rest, err = UnmarshalX25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
-	assert.EqualError(t, err, "bytes did not contain a proper nebula X25519 private key banner")
+	assert.EqualError(t, err, "bytes did not contain a proper nebula private key banner")
 
 	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
-	k, rest, err = UnmarshalX25519PrivateKey(rest)
+	k, rest, curve, err = UnmarshalPrivateKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
 	assert.EqualError(t, err, "input did not contain a valid PEM encoded block")
@@ -652,6 +939,12 @@ func TestUnmarshalX25519PublicKey(t *testing.T) {
 -----BEGIN NEBULA X25519 PUBLIC KEY-----
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 -----END NEBULA X25519 PUBLIC KEY-----
+`)
+	pubP256Key := []byte(`# A good key
+-----BEGIN NEBULA P256 PUBLIC KEY-----
+AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
+AAAAAAAAAAAAAAAAAAAAAAA=
+-----END NEBULA P256 PUBLIC KEY-----
 `)
 	shortKey := []byte(`# A short key
 -----BEGIN NEBULA X25519 PUBLIC KEY-----
@@ -668,29 +961,37 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 -END NEBULA X25519 PUBLIC KEY-----`)
 
-	keyBundle := appendByteSlices(pubKey, shortKey, invalidBanner, invalidPem)
+	keyBundle := appendByteSlices(pubKey, pubP256Key, shortKey, invalidBanner, invalidPem)
 
 	// Success test case
-	k, rest, err := UnmarshalX25519PublicKey(keyBundle)
+	k, rest, curve, err := UnmarshalPublicKey(keyBundle)
 	assert.Equal(t, len(k), 32)
 	assert.Nil(t, err)
+	assert.Equal(t, rest, appendByteSlices(pubP256Key, shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_CURVE25519, curve)
+
+	// Success test case
+	k, rest, curve, err = UnmarshalPublicKey(rest)
+	assert.Equal(t, len(k), 65)
+	assert.Nil(t, err)
 	assert.Equal(t, rest, appendByteSlices(shortKey, invalidBanner, invalidPem))
+	assert.Equal(t, Curve_P256, curve)
 
 	// Fail due to short key
-	k, rest, err = UnmarshalX25519PublicKey(rest)
+	k, rest, curve, err = UnmarshalPublicKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, appendByteSlices(invalidBanner, invalidPem))
-	assert.EqualError(t, err, "key was not 32 bytes, is invalid X25519 public key")
+	assert.EqualError(t, err, "key was not 32 bytes, is invalid CURVE25519 public key")
 
 	// Fail due to invalid banner
-	k, rest, err = UnmarshalX25519PublicKey(rest)
+	k, rest, curve, err = UnmarshalPublicKey(rest)
 	assert.Nil(t, k)
-	assert.EqualError(t, err, "bytes did not contain a proper nebula X25519 public key banner")
+	assert.EqualError(t, err, "bytes did not contain a proper nebula public key banner")
 	assert.Equal(t, rest, invalidPem)
 
 	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
-	k, rest, err = UnmarshalX25519PublicKey(rest)
+	k, rest, curve, err = UnmarshalPublicKey(rest)
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
 	assert.EqualError(t, err, "input did not contain a valid PEM encoded block")
@@ -745,7 +1046,7 @@ func TestNebulaCertificate_Copy(t *testing.T) {
 	assert.Nil(t, err)
 	cc := c.Copy()
 
-	util.AssertDeepCopyEqual(t, c, cc)
+	test.AssertDeepCopyEqual(t, c, cc)
 }
 
 func TestUnmarshalNebulaCertificate(t *testing.T) {
@@ -787,13 +1088,56 @@ func newTestCaCert(before, after time.Time, ips, subnets []*net.IPNet, groups []
 		nc.Details.Groups = groups
 	}
 
-	err = nc.Sign(priv)
+	err = nc.Sign(Curve_CURVE25519, priv)
 	if err != nil {
 		return nil, nil, nil, err
 	}
 	return nc, pub, priv, nil
 }
 
+func newTestCaCertP256(before, after time.Time, ips, subnets []*net.IPNet, groups []string) (*NebulaCertificate, []byte, []byte, error) {
+	priv, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	pub := elliptic.Marshal(elliptic.P256(), priv.PublicKey.X, priv.PublicKey.Y)
+	rawPriv := priv.D.FillBytes(make([]byte, 32))
+
+	if before.IsZero() {
+		before = time.Now().Add(time.Second * -60).Round(time.Second)
+	}
+	if after.IsZero() {
+		after = time.Now().Add(time.Second * 60).Round(time.Second)
+	}
+
+	nc := &NebulaCertificate{
+		Details: NebulaCertificateDetails{
+			Name:           "test ca",
+			NotBefore:      time.Unix(before.Unix(), 0),
+			NotAfter:       time.Unix(after.Unix(), 0),
+			PublicKey:      pub,
+			IsCA:           true,
+			Curve:          Curve_P256,
+			InvertedGroups: make(map[string]struct{}),
+		},
+	}
+
+	if len(ips) > 0 {
+		nc.Details.Ips = ips
+	}
+
+	if len(subnets) > 0 {
+		nc.Details.Subnets = subnets
+	}
+
+	if len(groups) > 0 {
+		nc.Details.Groups = groups
+	}
+
+	err = nc.Sign(Curve_P256, rawPriv)
+	if err != nil {
+		return nil, nil, nil, err
+	}
+	return nc, pub, rawPriv, nil
+}
+
 func newTestCert(ca *NebulaCertificate, key []byte, before, after time.Time, ips, subnets []*net.IPNet, groups []string) (*NebulaCertificate, []byte, []byte, error) {
 	issuer, err := ca.Sha256Sum()
 	if err != nil {
@@ -827,7 +1171,16 @@ func newTestCert(ca *NebulaCertificate, key []byte, before, after time.Time, ips
 		}
 	}
 
-	pub, rawPriv := x25519Keypair()
+	var pub, rawPriv []byte
+
+	switch ca.Details.Curve {
+	case Curve_CURVE25519:
+		pub, rawPriv = x25519Keypair()
+	case Curve_P256:
+		pub, rawPriv = p256Keypair()
+	default:
+		return nil, nil, nil, fmt.Errorf("unknown curve: %v", ca.Details.Curve)
+	}
 
 	nc := &NebulaCertificate{
 		Details: NebulaCertificateDetails{
@@ -839,12 +1192,13 @@ func newTestCert(ca *NebulaCertificate, key []byte, before, after time.Time, ips
 			NotAfter:       time.Unix(after.Unix(), 0),
 			PublicKey:      pub,
 			IsCA:           false,
+			Curve:          ca.Details.Curve,
 			Issuer:         issuer,
 			InvertedGroups: make(map[string]struct{}),
 		},
 	}
 
-	err = nc.Sign(key)
+	err = nc.Sign(ca.Details.Curve, key)
 	if err != nil {
 		return nil, nil, nil, err
 	}
@@ -853,10 +1207,24 @@ func newTestCert(ca *NebulaCertificate, key []byte, before, after time.Time, ips
 }
 
 func x25519Keypair() ([]byte, []byte) {
-	var pubkey, privkey [32]byte
-	if _, err := io.ReadFull(rand.Reader, privkey[:]); err != nil {
+	privkey := make([]byte, 32)
+	if _, err := io.ReadFull(rand.Reader, privkey); err != nil {
 		panic(err)
 	}
-	curve25519.ScalarBaseMult(&pubkey, &privkey)
-	return pubkey[:], privkey[:]
+
+	pubkey, err := curve25519.X25519(privkey, curve25519.Basepoint)
+	if err != nil {
+		panic(err)
+	}
+
+	return pubkey, privkey
+}
+
+func p256Keypair() ([]byte, []byte) {
+	privkey, err := ecdh.P256().GenerateKey(rand.Reader)
+	if err != nil {
+		panic(err)
+	}
+	pubkey := privkey.PublicKey()
+	return pubkey.Bytes(), privkey.Bytes()
 }

cert/crypto.go

@@ -0,0 +1,140 @@
+package cert
+
+import (
+	"crypto/aes"
+	"crypto/cipher"
+	"crypto/rand"
+	"fmt"
+	"io"
+
+	"golang.org/x/crypto/argon2"
+)
+
+// KDF factors
+type Argon2Parameters struct {
+	version     rune
+	Memory      uint32 // KiB
+	Parallelism uint8
+	Iterations  uint32
+	salt        []byte
+}
+
+// Returns a new Argon2Parameters object with current version set
+func NewArgon2Parameters(memory uint32, parallelism uint8, iterations uint32) *Argon2Parameters {
+	return &Argon2Parameters{
+		version:     argon2.Version,
+		Memory:      memory, // KiB
+		Parallelism: parallelism,
+		Iterations:  iterations,
+	}
+}
+
+// Encrypts data using AES-256-GCM and the Argon2id key derivation function
+func aes256Encrypt(passphrase []byte, kdfParams *Argon2Parameters, data []byte) ([]byte, error) {
+	key, err := aes256DeriveKey(passphrase, kdfParams)
+	if err != nil {
+		return nil, err
+	}
+
+	// this should never happen, but since this dictates how our calls into the
+	// aes package behave and could be catastraphic, let's sanity check this
+	if len(key) != 32 {
+		return nil, fmt.Errorf("invalid AES-256 key length (%d) - cowardly refusing to encrypt", len(key))
+	}
+
+	block, err := aes.NewCipher(key)
+	if err != nil {
+		return nil, err
+	}
+
+	gcm, err := cipher.NewGCM(block)
+	if err != nil {
+		return nil, err
+	}
+
+	nonce := make([]byte, gcm.NonceSize())
+	if _, err := io.ReadFull(rand.Reader, nonce); err != nil {
+		return nil, err
+	}
+
+	ciphertext := gcm.Seal(nil, nonce, data, nil)
+	blob := joinNonceCiphertext(nonce, ciphertext)
+
+	return blob, nil
+}
+
+// Decrypts data using AES-256-GCM and the Argon2id key derivation function
+// Expects the data to include an Argon2id parameter string before the encrypted data
+func aes256Decrypt(passphrase []byte, kdfParams *Argon2Parameters, data []byte) ([]byte, error) {
+	key, err := aes256DeriveKey(passphrase, kdfParams)
+	if err != nil {
+		return nil, err
+	}
+
+	block, err := aes.NewCipher(key)
+	if err != nil {
+		return nil, err
+	}
+
+	gcm, err := cipher.NewGCM(block)
+
+	nonce, ciphertext, err := splitNonceCiphertext(data, gcm.NonceSize())
+	if err != nil {
+		return nil, err
+	}
+
+	plaintext, err := gcm.Open(nil, nonce, ciphertext, nil)
+	if err != nil {
+		return nil, fmt.Errorf("invalid passphrase or corrupt private key")
+	}
+
+	return plaintext, nil
+}
+
+func aes256DeriveKey(passphrase []byte, params *Argon2Parameters) ([]byte, error) {
+	if params.salt == nil {
+		params.salt = make([]byte, 32)
+		if _, err := rand.Read(params.salt); err != nil {
+			return nil, err
+		}
+	}
+
+	// keySize of 32 bytes will result in AES-256 encryption
+	key, err := deriveKey(passphrase, 32, params)
+	if err != nil {
+		return nil, err
+	}
+
+	return key, nil
+}
+
+// Derives a key from a passphrase using Argon2id
+func deriveKey(passphrase []byte, keySize uint32, params *Argon2Parameters) ([]byte, error) {
+	if params.version != argon2.Version {
+		return nil, fmt.Errorf("incompatible Argon2 version: %d", params.version)
+	}
+
+	if params.salt == nil {
+		return nil, fmt.Errorf("salt must be set in argon2Parameters")
+	} else if len(params.salt) < 16 {
+		return nil, fmt.Errorf("salt must be at least 128  bits")
+	}
+
+	key := argon2.IDKey(passphrase, params.salt, params.Iterations, params.Memory, params.Parallelism, keySize)
+
+	return key, nil
+}
+
+// Prepends nonce to ciphertext
+func joinNonceCiphertext(nonce []byte, ciphertext []byte) []byte {
+	return append(nonce, ciphertext...)
+}
+
+// Splits nonce from ciphertext
+func splitNonceCiphertext(blob []byte, nonceSize int) ([]byte, []byte, error) {
+	if len(blob) <= nonceSize {
+		return nil, nil, fmt.Errorf("invalid ciphertext blob - blob shorter than nonce length")
+	}
+
+	return blob[:nonceSize], blob[nonceSize:], nil
+}

cert/crypto_test.go

@@ -0,0 +1,25 @@
+package cert
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"golang.org/x/crypto/argon2"
+)
+
+func TestNewArgon2Parameters(t *testing.T) {
+	p := NewArgon2Parameters(64*1024, 4, 3)
+	assert.EqualValues(t, &Argon2Parameters{
+		version:     argon2.Version,
+		Memory:      64 * 1024,
+		Parallelism: 4,
+		Iterations:  3,
+	}, p)
+	p = NewArgon2Parameters(2*1024*1024, 2, 1)
+	assert.EqualValues(t, &Argon2Parameters{
+		version:     argon2.Version,
+		Memory:      2 * 1024 * 1024,
+		Parallelism: 2,
+		Iterations:  1,
+	}, p)
+}

cert/errors.go

@@ -0,0 +1,14 @@
+package cert
+
+import (
+	"errors"
+)
+
+var (
+	ErrRootExpired       = errors.New("root certificate is expired")
+	ErrExpired           = errors.New("certificate is expired")
+	ErrNotCA             = errors.New("certificate is not a CA")
+	ErrNotSelfSigned     = errors.New("certificate is not self-signed")
+	ErrBlockListed       = errors.New("certificate is in the block list")
+	ErrSignatureMismatch = errors.New("certificate signature did not match")
+)

cidr/parse.go

@@ -0,0 +1,10 @@
+package cidr
+
+import "net"
+
+// Parse is a convenience function that returns only the IPNet
+// This function ignores errors since it is primarily a test helper, the result could be nil
+func Parse(s string) *net.IPNet {
+	_, c, _ := net.ParseCIDR(s)
+	return c
+}

cidr/tree4.go

@@ -1,39 +1,46 @@
-package nebula
+package cidr
 
 import (
-	"encoding/binary"
-	"fmt"
 	"net"
+
+	"github.com/slackhq/nebula/iputil"
 )
 
-type CIDRNode struct {
-	left   *CIDRNode
-	right  *CIDRNode
-	parent *CIDRNode
+type Node struct {
+	left   *Node
+	right  *Node
+	parent *Node
 	value  interface{}
 }
 
-type CIDRTree struct {
-	root *CIDRNode
+type entry struct {
+	CIDR  *net.IPNet
+	Value *interface{}
+}
+
+type Tree4 struct {
+	root *Node
+	list []entry
 }
 
 const (
-	startbit = uint32(0x80000000)
+	startbit = iputil.VpnIp(0x80000000)
 )
 
-func NewCIDRTree() *CIDRTree {
-	tree := new(CIDRTree)
-	tree.root = &CIDRNode{}
+func NewTree4() *Tree4 {
+	tree := new(Tree4)
+	tree.root = &Node{}
+	tree.list = []entry{}
 	return tree
 }
 
-func (tree *CIDRTree) AddCIDR(cidr *net.IPNet, val interface{}) {
+func (tree *Tree4) AddCIDR(cidr *net.IPNet, val interface{}) {
 	bit := startbit
 	node := tree.root
 	next := tree.root
 
-	ip := ip2int(cidr.IP)
-	mask := ip2int(cidr.Mask)
+	ip := iputil.Ip2VpnIp(cidr.IP)
+	mask := iputil.Ip2VpnIp(cidr.Mask)
 
 	// Find our last ancestor in the tree
 	for bit&mask != 0 {
@@ -53,13 +60,22 @@ func (tree *CIDRTree) AddCIDR(cidr *net.IPNet, val interface{}) {
 
 	// We already have this range so update the value
 	if next != nil {
+		addCIDR := cidr.String()
+		for i, v := range tree.list {
+			if addCIDR == v.CIDR.String() {
+				tree.list = append(tree.list[:i], tree.list[i+1:]...)
+				break
+			}
+		}
+
+		tree.list = append(tree.list, entry{CIDR: cidr, Value: &val})
 		node.value = val
 		return
 	}
 
 	// Build up the rest of the tree we don't already have
 	for bit&mask != 0 {
-		next = &CIDRNode{}
+		next = &Node{}
 		next.parent = node
 
 		if ip&bit != 0 {
@@ -74,10 +90,11 @@ func (tree *CIDRTree) AddCIDR(cidr *net.IPNet, val interface{}) {
 
 	// Final node marks our cidr, set the value
 	node.value = val
+	tree.list = append(tree.list, entry{CIDR: cidr, Value: &val})
 }
 
-// Finds the first match, which may be the least specific
-func (tree *CIDRTree) Contains(ip uint32) (value interface{}) {
+// Contains finds the first match, which may be the least specific
+func (tree *Tree4) Contains(ip iputil.VpnIp) (value interface{}) {
 	bit := startbit
 	node := tree.root
 
@@ -99,8 +116,8 @@ func (tree *CIDRTree) Contains(ip uint32) (value interface{}) {
 	return value
 }
 
-// Finds the most specific match
-func (tree *CIDRTree) MostSpecificContains(ip uint32) (value interface{}) {
+// MostSpecificContains finds the most specific match
+func (tree *Tree4) MostSpecificContains(ip iputil.VpnIp) (value interface{}) {
 	bit := startbit
 	node := tree.root
 
@@ -121,8 +138,8 @@ func (tree *CIDRTree) MostSpecificContains(ip uint32) (value interface{}) {
 	return value
 }
 
-// Finds the most specific match
-func (tree *CIDRTree) Match(ip uint32) (value interface{}) {
+// Match finds the most specific match
+func (tree *Tree4) Match(ip iputil.VpnIp) (value interface{}) {
 	bit := startbit
 	node := tree.root
 	lastNode := node
@@ -144,26 +161,7 @@ func (tree *CIDRTree) Match(ip uint32) (value interface{}) {
 	return value
 }
 
-// A helper type to avoid converting to IP when logging
-type IntIp uint32
-
-func (ip IntIp) String() string {
-	return fmt.Sprintf("%v", int2ip(uint32(ip)))
-}
-
-func (ip IntIp) MarshalJSON() ([]byte, error) {
-	return []byte(fmt.Sprintf("\"%s\"", int2ip(uint32(ip)).String())), nil
-}
-
-func ip2int(ip []byte) uint32 {
-	if len(ip) == 16 {
-		return binary.BigEndian.Uint32(ip[12:16])
-	}
-	return binary.BigEndian.Uint32(ip)
-}
-
-func int2ip(nn uint32) net.IP {
-	ip := make(net.IP, 4)
-	binary.BigEndian.PutUint32(ip, nn)
-	return ip
+// List will return all CIDRs and their current values. Do not modify the contents!
+func (tree *Tree4) List() []entry {
+	return tree.list
 }

cidr/tree4_test.go

@@ -0,0 +1,167 @@
+package cidr
+
+import (
+	"net"
+	"testing"
+
+	"github.com/slackhq/nebula/iputil"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestCIDRTree_List(t *testing.T) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("1.0.0.0/16"), "1")
+	tree.AddCIDR(Parse("1.0.0.0/8"), "2")
+	tree.AddCIDR(Parse("1.0.0.0/16"), "3")
+	tree.AddCIDR(Parse("1.0.0.0/16"), "4")
+	list := tree.List()
+	assert.Len(t, list, 2)
+	assert.Equal(t, "1.0.0.0/8", list[0].CIDR.String())
+	assert.Equal(t, "2", *list[0].Value)
+	assert.Equal(t, "1.0.0.0/16", list[1].CIDR.String())
+	assert.Equal(t, "4", *list[1].Value)
+}
+
+func TestCIDRTree_Contains(t *testing.T) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("1.0.0.0/8"), "1")
+	tree.AddCIDR(Parse("2.1.0.0/16"), "2")
+	tree.AddCIDR(Parse("3.1.1.0/24"), "3")
+	tree.AddCIDR(Parse("4.1.1.0/24"), "4a")
+	tree.AddCIDR(Parse("4.1.1.1/32"), "4b")
+	tree.AddCIDR(Parse("4.1.2.1/32"), "4c")
+	tree.AddCIDR(Parse("254.0.0.0/4"), "5")
+
+	tests := []struct {
+		Result interface{}
+		IP     string
+	}{
+		{"1", "1.0.0.0"},
+		{"1", "1.255.255.255"},
+		{"2", "2.1.0.0"},
+		{"2", "2.1.255.255"},
+		{"3", "3.1.1.0"},
+		{"3", "3.1.1.255"},
+		{"4a", "4.1.1.255"},
+		{"4a", "4.1.1.1"},
+		{"5", "240.0.0.0"},
+		{"5", "255.255.255.255"},
+		{nil, "239.0.0.0"},
+		{nil, "4.1.2.2"},
+	}
+
+	for _, tt := range tests {
+		assert.Equal(t, tt.Result, tree.Contains(iputil.Ip2VpnIp(net.ParseIP(tt.IP))))
+	}
+
+	tree = NewTree4()
+	tree.AddCIDR(Parse("1.1.1.1/0"), "cool")
+	assert.Equal(t, "cool", tree.Contains(iputil.Ip2VpnIp(net.ParseIP("0.0.0.0"))))
+	assert.Equal(t, "cool", tree.Contains(iputil.Ip2VpnIp(net.ParseIP("255.255.255.255"))))
+}
+
+func TestCIDRTree_MostSpecificContains(t *testing.T) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("1.0.0.0/8"), "1")
+	tree.AddCIDR(Parse("2.1.0.0/16"), "2")
+	tree.AddCIDR(Parse("3.1.1.0/24"), "3")
+	tree.AddCIDR(Parse("4.1.1.0/24"), "4a")
+	tree.AddCIDR(Parse("4.1.1.0/30"), "4b")
+	tree.AddCIDR(Parse("4.1.1.1/32"), "4c")
+	tree.AddCIDR(Parse("254.0.0.0/4"), "5")
+
+	tests := []struct {
+		Result interface{}
+		IP     string
+	}{
+		{"1", "1.0.0.0"},
+		{"1", "1.255.255.255"},
+		{"2", "2.1.0.0"},
+		{"2", "2.1.255.255"},
+		{"3", "3.1.1.0"},
+		{"3", "3.1.1.255"},
+		{"4a", "4.1.1.255"},
+		{"4b", "4.1.1.2"},
+		{"4c", "4.1.1.1"},
+		{"5", "240.0.0.0"},
+		{"5", "255.255.255.255"},
+		{nil, "239.0.0.0"},
+		{nil, "4.1.2.2"},
+	}
+
+	for _, tt := range tests {
+		assert.Equal(t, tt.Result, tree.MostSpecificContains(iputil.Ip2VpnIp(net.ParseIP(tt.IP))))
+	}
+
+	tree = NewTree4()
+	tree.AddCIDR(Parse("1.1.1.1/0"), "cool")
+	assert.Equal(t, "cool", tree.MostSpecificContains(iputil.Ip2VpnIp(net.ParseIP("0.0.0.0"))))
+	assert.Equal(t, "cool", tree.MostSpecificContains(iputil.Ip2VpnIp(net.ParseIP("255.255.255.255"))))
+}
+
+func TestCIDRTree_Match(t *testing.T) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("4.1.1.0/32"), "1a")
+	tree.AddCIDR(Parse("4.1.1.1/32"), "1b")
+
+	tests := []struct {
+		Result interface{}
+		IP     string
+	}{
+		{"1a", "4.1.1.0"},
+		{"1b", "4.1.1.1"},
+	}
+
+	for _, tt := range tests {
+		assert.Equal(t, tt.Result, tree.Match(iputil.Ip2VpnIp(net.ParseIP(tt.IP))))
+	}
+
+	tree = NewTree4()
+	tree.AddCIDR(Parse("1.1.1.1/0"), "cool")
+	assert.Equal(t, "cool", tree.Contains(iputil.Ip2VpnIp(net.ParseIP("0.0.0.0"))))
+	assert.Equal(t, "cool", tree.Contains(iputil.Ip2VpnIp(net.ParseIP("255.255.255.255"))))
+}
+
+func BenchmarkCIDRTree_Contains(b *testing.B) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("1.1.0.0/16"), "1")
+	tree.AddCIDR(Parse("1.2.1.1/32"), "1")
+	tree.AddCIDR(Parse("192.2.1.1/32"), "1")
+	tree.AddCIDR(Parse("172.2.1.1/32"), "1")
+
+	ip := iputil.Ip2VpnIp(net.ParseIP("1.2.1.1"))
+	b.Run("found", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			tree.Contains(ip)
+		}
+	})
+
+	ip = iputil.Ip2VpnIp(net.ParseIP("1.2.1.255"))
+	b.Run("not found", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			tree.Contains(ip)
+		}
+	})
+}
+
+func BenchmarkCIDRTree_Match(b *testing.B) {
+	tree := NewTree4()
+	tree.AddCIDR(Parse("1.1.0.0/16"), "1")
+	tree.AddCIDR(Parse("1.2.1.1/32"), "1")
+	tree.AddCIDR(Parse("192.2.1.1/32"), "1")
+	tree.AddCIDR(Parse("172.2.1.1/32"), "1")
+
+	ip := iputil.Ip2VpnIp(net.ParseIP("1.2.1.1"))
+	b.Run("found", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			tree.Match(ip)
+		}
+	})
+
+	ip = iputil.Ip2VpnIp(net.ParseIP("1.2.1.255"))
+	b.Run("not found", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			tree.Match(ip)
+		}
+	})
+}

cidr/tree6.go

@@ -1,26 +1,27 @@
-package nebula
+package cidr
 
 import (
-	"encoding/binary"
 	"net"
+
+	"github.com/slackhq/nebula/iputil"
 )
 
 const startbit6 = uint64(1 << 63)
 
-type CIDR6Tree struct {
-	root4 *CIDRNode
-	root6 *CIDRNode
+type Tree6 struct {
+	root4 *Node
+	root6 *Node
 }
 
-func NewCIDR6Tree() *CIDR6Tree {
-	tree := new(CIDR6Tree)
-	tree.root4 = &CIDRNode{}
-	tree.root6 = &CIDRNode{}
+func NewTree6() *Tree6 {
+	tree := new(Tree6)
+	tree.root4 = &Node{}
+	tree.root6 = &Node{}
 	return tree
 }
 
-func (tree *CIDR6Tree) AddCIDR(cidr *net.IPNet, val interface{}) {
-	var node, next *CIDRNode
+func (tree *Tree6) AddCIDR(cidr *net.IPNet, val interface{}) {
+	var node, next *Node
 
 	cidrIP, ipv4 := isIPV4(cidr.IP)
 	if ipv4 {
@@ -33,8 +34,8 @@ func (tree *CIDR6Tree) AddCIDR(cidr *net.IPNet, val interface{}) {
 	}
 
 	for i := 0; i < len(cidrIP); i += 4 {
-		ip := binary.BigEndian.Uint32(cidrIP[i : i+4])
-		mask := binary.BigEndian.Uint32(cidr.Mask[i : i+4])
+		ip := iputil.Ip2VpnIp(cidrIP[i : i+4])
+		mask := iputil.Ip2VpnIp(cidr.Mask[i : i+4])
 		bit := startbit
 
 		// Find our last ancestor in the tree
@@ -55,7 +56,7 @@ func (tree *CIDR6Tree) AddCIDR(cidr *net.IPNet, val interface{}) {
 
 		// Build up the rest of the tree we don't already have
 		for bit&mask != 0 {
-			next = &CIDRNode{}
+			next = &Node{}
 			next.parent = node
 
 			if ip&bit != 0 {
@@ -74,8 +75,8 @@ func (tree *CIDR6Tree) AddCIDR(cidr *net.IPNet, val interface{}) {
 }
 
 // Finds the most specific match
-func (tree *CIDR6Tree) MostSpecificContains(ip net.IP) (value interface{}) {
-	var node *CIDRNode
+func (tree *Tree6) MostSpecificContains(ip net.IP) (value interface{}) {
+	var node *Node
 
 	wholeIP, ipv4 := isIPV4(ip)
 	if ipv4 {
@@ -85,7 +86,7 @@ func (tree *CIDR6Tree) MostSpecificContains(ip net.IP) (value interface{}) {
 	}
 
 	for i := 0; i < len(wholeIP); i += 4 {
-		ip := ip2int(wholeIP[i : i+4])
+		ip := iputil.Ip2VpnIp(wholeIP[i : i+4])
 		bit := startbit
 
 		for node != nil {
@@ -110,7 +111,7 @@ func (tree *CIDR6Tree) MostSpecificContains(ip net.IP) (value interface{}) {
 	return value
 }
 
-func (tree *CIDR6Tree) MostSpecificContainsIpV4(ip uint32) (value interface{}) {
+func (tree *Tree6) MostSpecificContainsIpV4(ip iputil.VpnIp) (value interface{}) {
 	bit := startbit
 	node := tree.root4
 
@@ -131,7 +132,7 @@ func (tree *CIDR6Tree) MostSpecificContainsIpV4(ip uint32) (value interface{}) {
 	return value
 }
 
-func (tree *CIDR6Tree) MostSpecificContainsIpV6(hi, lo uint64) (value interface{}) {
+func (tree *Tree6) MostSpecificContainsIpV6(hi, lo uint64) (value interface{}) {
 	ip := hi
 	node := tree.root6
 

cidr/tree6_test.go

@@ -1,6 +1,7 @@
-package nebula
+package cidr
 
 import (
+	"encoding/binary"
 	"net"
 	"testing"
 
@@ -8,17 +9,17 @@ import (
 )
 
 func TestCIDR6Tree_MostSpecificContains(t *testing.T) {
-	tree := NewCIDR6Tree()
-	tree.AddCIDR(getCIDR("1.0.0.0/8"), "1")
-	tree.AddCIDR(getCIDR("2.1.0.0/16"), "2")
-	tree.AddCIDR(getCIDR("3.1.1.0/24"), "3")
-	tree.AddCIDR(getCIDR("4.1.1.1/24"), "4a")
-	tree.AddCIDR(getCIDR("4.1.1.1/30"), "4b")
-	tree.AddCIDR(getCIDR("4.1.1.1/32"), "4c")
-	tree.AddCIDR(getCIDR("254.0.0.0/4"), "5")
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/64"), "6a")
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/80"), "6b")
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/96"), "6c")
+	tree := NewTree6()
+	tree.AddCIDR(Parse("1.0.0.0/8"), "1")
+	tree.AddCIDR(Parse("2.1.0.0/16"), "2")
+	tree.AddCIDR(Parse("3.1.1.0/24"), "3")
+	tree.AddCIDR(Parse("4.1.1.1/24"), "4a")
+	tree.AddCIDR(Parse("4.1.1.1/30"), "4b")
+	tree.AddCIDR(Parse("4.1.1.1/32"), "4c")
+	tree.AddCIDR(Parse("254.0.0.0/4"), "5")
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/64"), "6a")
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/80"), "6b")
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/96"), "6c")
 
 	tests := []struct {
 		Result interface{}
@@ -46,9 +47,9 @@ func TestCIDR6Tree_MostSpecificContains(t *testing.T) {
 		assert.Equal(t, tt.Result, tree.MostSpecificContains(net.ParseIP(tt.IP)))
 	}
 
-	tree = NewCIDR6Tree()
-	tree.AddCIDR(getCIDR("1.1.1.1/0"), "cool")
-	tree.AddCIDR(getCIDR("::/0"), "cool6")
+	tree = NewTree6()
+	tree.AddCIDR(Parse("1.1.1.1/0"), "cool")
+	tree.AddCIDR(Parse("::/0"), "cool6")
 	assert.Equal(t, "cool", tree.MostSpecificContains(net.ParseIP("0.0.0.0")))
 	assert.Equal(t, "cool", tree.MostSpecificContains(net.ParseIP("255.255.255.255")))
 	assert.Equal(t, "cool6", tree.MostSpecificContains(net.ParseIP("::")))
@@ -56,10 +57,10 @@ func TestCIDR6Tree_MostSpecificContains(t *testing.T) {
 }
 
 func TestCIDR6Tree_MostSpecificContainsIpV6(t *testing.T) {
-	tree := NewCIDR6Tree()
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/64"), "6a")
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/80"), "6b")
-	tree.AddCIDR(getCIDR("1:2:0:4:5:0:0:0/96"), "6c")
+	tree := NewTree6()
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/64"), "6a")
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/80"), "6b")
+	tree.AddCIDR(Parse("1:2:0:4:5:0:0:0/96"), "6c")
 
 	tests := []struct {
 		Result interface{}
@@ -71,7 +72,10 @@ func TestCIDR6Tree_MostSpecificContainsIpV6(t *testing.T) {
 	}
 
 	for _, tt := range tests {
-		ip := NewIp6AndPort(net.ParseIP(tt.IP), 0)
-		assert.Equal(t, tt.Result, tree.MostSpecificContainsIpV6(ip.Hi, ip.Lo))
+		ip := net.ParseIP(tt.IP)
+		hi := binary.BigEndian.Uint64(ip[:8])
+		lo := binary.BigEndian.Uint64(ip[8:])
+
+		assert.Equal(t, tt.Result, tree.MostSpecificContainsIpV6(hi, lo))
 	}
 }

cidr_radix_test.go

@@ -1,157 +0,0 @@
-package nebula
-
-import (
-	"net"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-)
-
-func TestCIDRTree_Contains(t *testing.T) {
-	tree := NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.0.0.0/8"), "1")
-	tree.AddCIDR(getCIDR("2.1.0.0/16"), "2")
-	tree.AddCIDR(getCIDR("3.1.1.0/24"), "3")
-	tree.AddCIDR(getCIDR("4.1.1.0/24"), "4a")
-	tree.AddCIDR(getCIDR("4.1.1.1/32"), "4b")
-	tree.AddCIDR(getCIDR("4.1.2.1/32"), "4c")
-	tree.AddCIDR(getCIDR("254.0.0.0/4"), "5")
-
-	tests := []struct {
-		Result interface{}
-		IP     string
-	}{
-		{"1", "1.0.0.0"},
-		{"1", "1.255.255.255"},
-		{"2", "2.1.0.0"},
-		{"2", "2.1.255.255"},
-		{"3", "3.1.1.0"},
-		{"3", "3.1.1.255"},
-		{"4a", "4.1.1.255"},
-		{"4a", "4.1.1.1"},
-		{"5", "240.0.0.0"},
-		{"5", "255.255.255.255"},
-		{nil, "239.0.0.0"},
-		{nil, "4.1.2.2"},
-	}
-
-	for _, tt := range tests {
-		assert.Equal(t, tt.Result, tree.Contains(ip2int(net.ParseIP(tt.IP))))
-	}
-
-	tree = NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.1.1.1/0"), "cool")
-	assert.Equal(t, "cool", tree.Contains(ip2int(net.ParseIP("0.0.0.0"))))
-	assert.Equal(t, "cool", tree.Contains(ip2int(net.ParseIP("255.255.255.255"))))
-}
-
-func TestCIDRTree_MostSpecificContains(t *testing.T) {
-	tree := NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.0.0.0/8"), "1")
-	tree.AddCIDR(getCIDR("2.1.0.0/16"), "2")
-	tree.AddCIDR(getCIDR("3.1.1.0/24"), "3")
-	tree.AddCIDR(getCIDR("4.1.1.0/24"), "4a")
-	tree.AddCIDR(getCIDR("4.1.1.0/30"), "4b")
-	tree.AddCIDR(getCIDR("4.1.1.1/32"), "4c")
-	tree.AddCIDR(getCIDR("254.0.0.0/4"), "5")
-
-	tests := []struct {
-		Result interface{}
-		IP     string
-	}{
-		{"1", "1.0.0.0"},
-		{"1", "1.255.255.255"},
-		{"2", "2.1.0.0"},
-		{"2", "2.1.255.255"},
-		{"3", "3.1.1.0"},
-		{"3", "3.1.1.255"},
-		{"4a", "4.1.1.255"},
-		{"4b", "4.1.1.2"},
-		{"4c", "4.1.1.1"},
-		{"5", "240.0.0.0"},
-		{"5", "255.255.255.255"},
-		{nil, "239.0.0.0"},
-		{nil, "4.1.2.2"},
-	}
-
-	for _, tt := range tests {
-		assert.Equal(t, tt.Result, tree.MostSpecificContains(ip2int(net.ParseIP(tt.IP))))
-	}
-
-	tree = NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.1.1.1/0"), "cool")
-	assert.Equal(t, "cool", tree.MostSpecificContains(ip2int(net.ParseIP("0.0.0.0"))))
-	assert.Equal(t, "cool", tree.MostSpecificContains(ip2int(net.ParseIP("255.255.255.255"))))
-}
-
-func TestCIDRTree_Match(t *testing.T) {
-	tree := NewCIDRTree()
-	tree.AddCIDR(getCIDR("4.1.1.0/32"), "1a")
-	tree.AddCIDR(getCIDR("4.1.1.1/32"), "1b")
-
-	tests := []struct {
-		Result interface{}
-		IP     string
-	}{
-		{"1a", "4.1.1.0"},
-		{"1b", "4.1.1.1"},
-	}
-
-	for _, tt := range tests {
-		assert.Equal(t, tt.Result, tree.Match(ip2int(net.ParseIP(tt.IP))))
-	}
-
-	tree = NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.1.1.1/0"), "cool")
-	assert.Equal(t, "cool", tree.Contains(ip2int(net.ParseIP("0.0.0.0"))))
-	assert.Equal(t, "cool", tree.Contains(ip2int(net.ParseIP("255.255.255.255"))))
-}
-
-func BenchmarkCIDRTree_Contains(b *testing.B) {
-	tree := NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.1.0.0/16"), "1")
-	tree.AddCIDR(getCIDR("1.2.1.1/32"), "1")
-	tree.AddCIDR(getCIDR("192.2.1.1/32"), "1")
-	tree.AddCIDR(getCIDR("172.2.1.1/32"), "1")
-
-	ip := ip2int(net.ParseIP("1.2.1.1"))
-	b.Run("found", func(b *testing.B) {
-		for i := 0; i < b.N; i++ {
-			tree.Contains(ip)
-		}
-	})
-
-	ip = ip2int(net.ParseIP("1.2.1.255"))
-	b.Run("not found", func(b *testing.B) {
-		for i := 0; i < b.N; i++ {
-			tree.Contains(ip)
-		}
-	})
-}
-
-func BenchmarkCIDRTree_Match(b *testing.B) {
-	tree := NewCIDRTree()
-	tree.AddCIDR(getCIDR("1.1.0.0/16"), "1")
-	tree.AddCIDR(getCIDR("1.2.1.1/32"), "1")
-	tree.AddCIDR(getCIDR("192.2.1.1/32"), "1")
-	tree.AddCIDR(getCIDR("172.2.1.1/32"), "1")
-
-	ip := ip2int(net.ParseIP("1.2.1.1"))
-	b.Run("found", func(b *testing.B) {
-		for i := 0; i < b.N; i++ {
-			tree.Match(ip)
-		}
-	})
-
-	ip = ip2int(net.ParseIP("1.2.1.255"))
-	b.Run("not found", func(b *testing.B) {
-		for i := 0; i < b.N; i++ {
-			tree.Match(ip)
-		}
-	})
-}
-
-func getCIDR(s string) *net.IPNet {
-	_, c, _ := net.ParseCIDR(s)
-	return c
-}

cmd/nebula-cert/ca.go

@@ -1,11 +1,14 @@
 package main
 
 import (
+	"crypto/ecdsa"
+	"crypto/elliptic"
 	"crypto/rand"
 	"flag"
 	"fmt"
 	"io"
 	"io/ioutil"
+	"math"
 	"net"
 	"os"
 	"strings"
@@ -17,15 +20,21 @@ import (
 )
 
 type caFlags struct {
-	set         *flag.FlagSet
-	name        *string
-	duration    *time.Duration
-	outKeyPath  *string
-	outCertPath *string
-	outQRPath   *string
-	groups      *string
-	ips         *string
-	subnets     *string
+	set              *flag.FlagSet
+	name             *string
+	duration         *time.Duration
+	outKeyPath       *string
+	outCertPath      *string
+	outQRPath        *string
+	groups           *string
+	ips              *string
+	subnets          *string
+	argonMemory      *uint
+	argonIterations  *uint
+	argonParallelism *uint
+	encryption       *bool
+
+	curve *string
 }
 
 func newCaFlags() *caFlags {
@@ -37,12 +46,31 @@ func newCaFlags() *caFlags {
 	cf.outCertPath = cf.set.String("out-crt", "ca.crt", "Optional: path to write the certificate to")
 	cf.outQRPath = cf.set.String("out-qr", "", "Optional: output a qr code image (png) of the certificate")
 	cf.groups = cf.set.String("groups", "", "Optional: comma separated list of groups. This will limit which groups subordinate certs can use")
-	cf.ips = cf.set.String("ips", "", "Optional: comma separated list of ip and network in CIDR notation. This will limit which ip addresses and networks subordinate certs can use")
-	cf.subnets = cf.set.String("subnets", "", "Optional: comma separated list of ip and network in CIDR notation. This will limit which subnet addresses and networks subordinate certs can use")
+	cf.ips = cf.set.String("ips", "", "Optional: comma separated list of ipv4 address and network in CIDR notation. This will limit which ipv4 addresses and networks subordinate certs can use for ip addresses")
+	cf.subnets = cf.set.String("subnets", "", "Optional: comma separated list of ipv4 address and network in CIDR notation. This will limit which ipv4 addresses and networks subordinate certs can use in subnets")
+	cf.argonMemory = cf.set.Uint("argon-memory", 2*1024*1024, "Optional: Argon2 memory parameter (in KiB) used for encrypted private key passphrase")
+	cf.argonParallelism = cf.set.Uint("argon-parallelism", 4, "Optional: Argon2 parallelism parameter used for encrypted private key passphrase")
+	cf.argonIterations = cf.set.Uint("argon-iterations", 1, "Optional: Argon2 iterations parameter used for encrypted private key passphrase")
+	cf.encryption = cf.set.Bool("encrypt", false, "Optional: prompt for passphrase and write out-key in an encrypted format")
+	cf.curve = cf.set.String("curve", "25519", "EdDSA/ECDSA Curve (25519, P256)")
 	return &cf
 }
 
-func ca(args []string, out io.Writer, errOut io.Writer) error {
+func parseArgonParameters(memory uint, parallelism uint, iterations uint) (*cert.Argon2Parameters, error) {
+	if memory <= 0 || memory > math.MaxUint32 {
+		return nil, newHelpErrorf("-argon-memory must be be greater than 0 and no more than %d KiB", uint32(math.MaxUint32))
+	}
+	if parallelism <= 0 || parallelism > math.MaxUint8 {
+		return nil, newHelpErrorf("-argon-parallelism must be be greater than 0 and no more than %d", math.MaxUint8)
+	}
+	if iterations <= 0 || iterations > math.MaxUint32 {
+		return nil, newHelpErrorf("-argon-iterations must be be greater than 0 and no more than %d", uint32(math.MaxUint32))
+	}
+
+	return cert.NewArgon2Parameters(uint32(memory), uint8(parallelism), uint32(iterations)), nil
+}
+
+func ca(args []string, out io.Writer, errOut io.Writer, pr PasswordReader) error {
 	cf := newCaFlags()
 	err := cf.set.Parse(args)
 	if err != nil {
@@ -58,6 +86,12 @@ func ca(args []string, out io.Writer, errOut io.Writer) error {
 	if err := mustFlagString("out-crt", cf.outCertPath); err != nil {
 		return err
 	}
+	var kdfParams *cert.Argon2Parameters
+	if *cf.encryption {
+		if kdfParams, err = parseArgonParameters(*cf.argonMemory, *cf.argonParallelism, *cf.argonIterations); err != nil {
+			return err
+		}
+	}
 
 	if *cf.duration <= 0 {
 		return &helpError{"-duration must be greater than 0"}
@@ -82,6 +116,9 @@ func ca(args []string, out io.Writer, errOut io.Writer) error {
 				if err != nil {
 					return newHelpErrorf("invalid ip definition: %s", err)
 				}
+				if ip.To4() == nil {
+					return newHelpErrorf("invalid ip definition: can only be ipv4, have %s", rs)
+				}
 
 				ipNet.IP = ip
 				ips = append(ips, ipNet)
@@ -98,14 +135,55 @@ func ca(args []string, out io.Writer, errOut io.Writer) error {
 				if err != nil {
 					return newHelpErrorf("invalid subnet definition: %s", err)
 				}
+				if s.IP.To4() == nil {
+					return newHelpErrorf("invalid subnet definition: can only be ipv4, have %s", rs)
+				}
 				subnets = append(subnets, s)
 			}
 		}
 	}
 
-	pub, rawPriv, err := ed25519.GenerateKey(rand.Reader)
-	if err != nil {
-		return fmt.Errorf("error while generating ed25519 keys: %s", err)
+	var passphrase []byte
+	if *cf.encryption {
+		for i := 0; i < 5; i++ {
+			out.Write([]byte("Enter passphrase: "))
+			passphrase, err = pr.ReadPassword()
+
+			if err == ErrNoTerminal {
+				return fmt.Errorf("out-key must be encrypted interactively")
+			} else if err != nil {
+				return fmt.Errorf("error reading passphrase: %s", err)
+			}
+
+			if len(passphrase) > 0 {
+				break
+			}
+		}
+
+		if len(passphrase) == 0 {
+			return fmt.Errorf("no passphrase specified, remove -encrypt flag to write out-key in plaintext")
+		}
+	}
+
+	var curve cert.Curve
+	var pub, rawPriv []byte
+	switch *cf.curve {
+	case "25519", "X25519", "Curve25519", "CURVE25519":
+		curve = cert.Curve_CURVE25519
+		pub, rawPriv, err = ed25519.GenerateKey(rand.Reader)
+		if err != nil {
+			return fmt.Errorf("error while generating ed25519 keys: %s", err)
+		}
+	case "P256":
+		var key *ecdsa.PrivateKey
+		curve = cert.Curve_P256
+		key, err = ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+		if err != nil {
+			return fmt.Errorf("error while generating ecdsa keys: %s", err)
+		}
+		// ref: https://github.com/golang/go/blob/go1.19/src/crypto/x509/sec1.go#L60
+		rawPriv = key.D.FillBytes(make([]byte, 32))
+		pub = elliptic.Marshal(elliptic.P256(), key.X, key.Y)
 	}
 
 	nc := cert.NebulaCertificate{
@@ -118,6 +196,7 @@ func ca(args []string, out io.Writer, errOut io.Writer) error {
 			NotAfter:  time.Now().Add(*cf.duration),
 			PublicKey: pub,
 			IsCA:      true,
+			Curve:     curve,
 		},
 	}
 
@@ -129,12 +208,22 @@ func ca(args []string, out io.Writer, errOut io.Writer) error {
 		return fmt.Errorf("refusing to overwrite existing CA cert: %s", *cf.outCertPath)
 	}
 
-	err = nc.Sign(rawPriv)
+	err = nc.Sign(curve, rawPriv)
 	if err != nil {
 		return fmt.Errorf("error while signing: %s", err)
 	}
 
-	err = ioutil.WriteFile(*cf.outKeyPath, cert.MarshalEd25519PrivateKey(rawPriv), 0600)
+	if *cf.encryption {
+		b, err := cert.EncryptAndMarshalSigningPrivateKey(curve, rawPriv, passphrase, kdfParams)
+		if err != nil {
+			return fmt.Errorf("error while encrypting out-key: %s", err)
+		}
+
+		err = ioutil.WriteFile(*cf.outKeyPath, b, 0600)
+	} else {
+		err = ioutil.WriteFile(*cf.outKeyPath, cert.MarshalSigningPrivateKey(curve, rawPriv), 0600)
+	}
+
 	if err != nil {
 		return fmt.Errorf("error while writing out-key: %s", err)
 	}

cmd/nebula-cert/ca_test.go

@@ -1,11 +1,15 @@
+//go:build !windows
 // +build !windows
 
 package main
 
 import (
 	"bytes"
+	"encoding/pem"
+	"errors"
 	"io/ioutil"
 	"os"
+	"strings"
 	"testing"
 	"time"
 
@@ -25,12 +29,22 @@ func Test_caHelp(t *testing.T) {
 	assert.Equal(
 		t,
 		"Usage of "+os.Args[0]+" ca <flags>: create a self signed certificate authority\n"+
+			"  -argon-iterations uint\n"+
+			"    \tOptional: Argon2 iterations parameter used for encrypted private key passphrase (default 1)\n"+
+			"  -argon-memory uint\n"+
+			"    \tOptional: Argon2 memory parameter (in KiB) used for encrypted private key passphrase (default 2097152)\n"+
+			"  -argon-parallelism uint\n"+
+			"    \tOptional: Argon2 parallelism parameter used for encrypted private key passphrase (default 4)\n"+
+			"  -curve string\n"+
+			"    \tEdDSA/ECDSA Curve (25519, P256) (default \"25519\")\n"+
 			"  -duration duration\n"+
 			"    \tOptional: amount of time the certificate should be valid for. Valid time units are seconds: \"s\", minutes: \"m\", hours: \"h\" (default 8760h0m0s)\n"+
+			"  -encrypt\n"+
+			"    \tOptional: prompt for passphrase and write out-key in an encrypted format\n"+
 			"  -groups string\n"+
 			"    \tOptional: comma separated list of groups. This will limit which groups subordinate certs can use\n"+
 			"  -ips string\n"+
-			"    \tOptional: comma separated list of ip and network in CIDR notation. This will limit which ip addresses and networks subordinate certs can use\n"+
+			"    \tOptional: comma separated list of ipv4 address and network in CIDR notation. This will limit which ipv4 addresses and networks subordinate certs can use for ip addresses\n"+
 			"  -name string\n"+
 			"    \tRequired: name of the certificate authority\n"+
 			"  -out-crt string\n"+
@@ -40,7 +54,7 @@ func Test_caHelp(t *testing.T) {
 			"  -out-qr string\n"+
 			"    \tOptional: output a qr code image (png) of the certificate\n"+
 			"  -subnets string\n"+
-			"    \tOptional: comma separated list of ip and network in CIDR notation. This will limit which subnet addresses and networks subordinate certs can use\n",
+			"    \tOptional: comma separated list of ipv4 address and network in CIDR notation. This will limit which ipv4 addresses and networks subordinate certs can use in subnets\n",
 		ob.String(),
 	)
 }
@@ -49,8 +63,38 @@ func Test_ca(t *testing.T) {
 	ob := &bytes.Buffer{}
 	eb := &bytes.Buffer{}
 
+	nopw := &StubPasswordReader{
+		password: []byte(""),
+		err:      nil,
+	}
+
+	errpw := &StubPasswordReader{
+		password: []byte(""),
+		err:      errors.New("stub error"),
+	}
+
+	passphrase := []byte("DO NOT USE THIS KEY")
+	testpw := &StubPasswordReader{
+		password: passphrase,
+		err:      nil,
+	}
+
+	pwPromptOb := "Enter passphrase: "
+
 	// required args
-	assertHelpError(t, ca([]string{"-out-key", "nope", "-out-crt", "nope", "duration", "100m"}, ob, eb), "-name is required")
+	assertHelpError(t, ca(
+		[]string{"-out-key", "nope", "-out-crt", "nope", "duration", "100m"}, ob, eb, nopw,
+	), "-name is required")
+	assert.Equal(t, "", ob.String())
+	assert.Equal(t, "", eb.String())
+
+	// ipv4 only ips
+	assertHelpError(t, ca([]string{"-name", "ipv6", "-ips", "100::100/100"}, ob, eb, nopw), "invalid ip definition: can only be ipv4, have 100::100/100")
+	assert.Equal(t, "", ob.String())
+	assert.Equal(t, "", eb.String())
+
+	// ipv4 only subnets
+	assertHelpError(t, ca([]string{"-name", "ipv6", "-subnets", "100::100/100"}, ob, eb, nopw), "invalid subnet definition: can only be ipv4, have 100::100/100")
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 
@@ -58,7 +102,7 @@ func Test_ca(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args := []string{"-name", "test", "-duration", "100m", "-out-crt", "/do/not/write/pleasecrt", "-out-key", "/do/not/write/pleasekey"}
-	assert.EqualError(t, ca(args, ob, eb), "error while writing out-key: open /do/not/write/pleasekey: "+NoSuchDirError)
+	assert.EqualError(t, ca(args, ob, eb, nopw), "error while writing out-key: open /do/not/write/pleasekey: "+NoSuchDirError)
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 
@@ -71,7 +115,7 @@ func Test_ca(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-name", "test", "-duration", "100m", "-out-crt", "/do/not/write/pleasecrt", "-out-key", keyF.Name()}
-	assert.EqualError(t, ca(args, ob, eb), "error while writing out-crt: open /do/not/write/pleasecrt: "+NoSuchDirError)
+	assert.EqualError(t, ca(args, ob, eb, nopw), "error while writing out-crt: open /do/not/write/pleasecrt: "+NoSuchDirError)
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 
@@ -85,7 +129,7 @@ func Test_ca(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-name", "test", "-duration", "100m", "-groups", "1,,   2    ,        ,,,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
-	assert.Nil(t, ca(args, ob, eb))
+	assert.Nil(t, ca(args, ob, eb, nopw))
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 
@@ -111,19 +155,67 @@ func Test_ca(t *testing.T) {
 	assert.Equal(t, "", lCrt.Details.Issuer)
 	assert.True(t, lCrt.CheckSignature(lCrt.Details.PublicKey))
 
+	// test encrypted key
+	os.Remove(keyF.Name())
+	os.Remove(crtF.Name())
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-encrypt", "-name", "test", "-duration", "100m", "-groups", "1,2,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
+	assert.Nil(t, ca(args, ob, eb, testpw))
+	assert.Equal(t, pwPromptOb, ob.String())
+	assert.Equal(t, "", eb.String())
+
+	// read encrypted key file and verify default params
+	rb, _ = ioutil.ReadFile(keyF.Name())
+	k, _ := pem.Decode(rb)
+	ned, err := cert.UnmarshalNebulaEncryptedData(k.Bytes)
+	assert.Nil(t, err)
+	// we won't know salt in advance, so just check start of string
+	assert.Equal(t, uint32(2*1024*1024), ned.EncryptionMetadata.Argon2Parameters.Memory)
+	assert.Equal(t, uint8(4), ned.EncryptionMetadata.Argon2Parameters.Parallelism)
+	assert.Equal(t, uint32(1), ned.EncryptionMetadata.Argon2Parameters.Iterations)
+
+	// verify the key is valid and decrypt-able
+	var curve cert.Curve
+	curve, lKey, b, err = cert.DecryptAndUnmarshalSigningPrivateKey(passphrase, rb)
+	assert.Equal(t, cert.Curve_CURVE25519, curve)
+	assert.Nil(t, err)
+	assert.Len(t, b, 0)
+	assert.Len(t, lKey, 64)
+
+	// test when reading passsword results in an error
+	os.Remove(keyF.Name())
+	os.Remove(crtF.Name())
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-encrypt", "-name", "test", "-duration", "100m", "-groups", "1,2,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
+	assert.Error(t, ca(args, ob, eb, errpw))
+	assert.Equal(t, pwPromptOb, ob.String())
+	assert.Equal(t, "", eb.String())
+
+	// test when user fails to enter a password
+	os.Remove(keyF.Name())
+	os.Remove(crtF.Name())
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-encrypt", "-name", "test", "-duration", "100m", "-groups", "1,2,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
+	assert.EqualError(t, ca(args, ob, eb, nopw), "no passphrase specified, remove -encrypt flag to write out-key in plaintext")
+	assert.Equal(t, strings.Repeat(pwPromptOb, 5), ob.String()) // prompts 5 times before giving up
+	assert.Equal(t, "", eb.String())
+
 	// create valid cert/key for overwrite tests
 	os.Remove(keyF.Name())
 	os.Remove(crtF.Name())
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-name", "test", "-duration", "100m", "-groups", "1,,   2    ,        ,,,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
-	assert.Nil(t, ca(args, ob, eb))
+	assert.Nil(t, ca(args, ob, eb, nopw))
 
 	// test that we won't overwrite existing certificate file
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-name", "test", "-duration", "100m", "-groups", "1,,   2    ,        ,,,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
-	assert.EqualError(t, ca(args, ob, eb), "refusing to overwrite existing CA key: "+keyF.Name())
+	assert.EqualError(t, ca(args, ob, eb, nopw), "refusing to overwrite existing CA key: "+keyF.Name())
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 
@@ -132,7 +224,7 @@ func Test_ca(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-name", "test", "-duration", "100m", "-groups", "1,,   2    ,        ,,,3,4,5", "-out-crt", crtF.Name(), "-out-key", keyF.Name()}
-	assert.EqualError(t, ca(args, ob, eb), "refusing to overwrite existing CA cert: "+crtF.Name())
+	assert.EqualError(t, ca(args, ob, eb, nopw), "refusing to overwrite existing CA cert: "+crtF.Name())
 	assert.Equal(t, "", ob.String())
 	assert.Equal(t, "", eb.String())
 	os.Remove(keyF.Name())

cmd/nebula-cert/keygen.go

@@ -14,6 +14,8 @@ type keygenFlags struct {
 	set        *flag.FlagSet
 	outKeyPath *string
 	outPubPath *string
+
+	curve *string
 }
 
 func newKeygenFlags() *keygenFlags {
@@ -21,6 +23,7 @@ func newKeygenFlags() *keygenFlags {
 	cf.set.Usage = func() {}
 	cf.outPubPath = cf.set.String("out-pub", "", "Required: path to write the public key to")
 	cf.outKeyPath = cf.set.String("out-key", "", "Required: path to write the private key to")
+	cf.curve = cf.set.String("curve", "25519", "ECDH Curve (25519, P256)")
 	return &cf
 }
 
@@ -38,14 +41,25 @@ func keygen(args []string, out io.Writer, errOut io.Writer) error {
 		return err
 	}
 
-	pub, rawPriv := x25519Keypair()
+	var pub, rawPriv []byte
+	var curve cert.Curve
+	switch *cf.curve {
+	case "25519", "X25519", "Curve25519", "CURVE25519":
+		pub, rawPriv = x25519Keypair()
+		curve = cert.Curve_CURVE25519
+	case "P256":
+		pub, rawPriv = p256Keypair()
+		curve = cert.Curve_P256
+	default:
+		return fmt.Errorf("invalid curve: %s", *cf.curve)
+	}
 
-	err = ioutil.WriteFile(*cf.outKeyPath, cert.MarshalX25519PrivateKey(rawPriv), 0600)
+	err = ioutil.WriteFile(*cf.outKeyPath, cert.MarshalPrivateKey(curve, rawPriv), 0600)
 	if err != nil {
 		return fmt.Errorf("error while writing out-key: %s", err)
 	}
 
-	err = ioutil.WriteFile(*cf.outPubPath, cert.MarshalX25519PublicKey(pub), 0600)
+	err = ioutil.WriteFile(*cf.outPubPath, cert.MarshalPublicKey(curve, pub), 0600)
 	if err != nil {
 		return fmt.Errorf("error while writing out-pub: %s", err)
 	}

cmd/nebula-cert/keygen_test.go

@@ -22,6 +22,8 @@ func Test_keygenHelp(t *testing.T) {
 	assert.Equal(
 		t,
 		"Usage of "+os.Args[0]+" keygen <flags>: create a public/private key pair. the public key can be passed to `nebula-cert sign`\n"+
+			"  -curve string\n"+
+			"    \tECDH Curve (25519, P256) (default \"25519\")\n"+
 			"  -out-key string\n"+
 			"    \tRequired: path to write the private key to\n"+
 			"  -out-pub string\n"+

cmd/nebula-cert/main.go

@@ -62,11 +62,11 @@ func main() {
 
 	switch args[0] {
 	case "ca":
-		err = ca(args[1:], os.Stdout, os.Stderr)
+		err = ca(args[1:], os.Stdout, os.Stderr, StdinPasswordReader{})
 	case "keygen":
 		err = keygen(args[1:], os.Stdout, os.Stderr)
 	case "sign":
-		err = signCert(args[1:], os.Stdout, os.Stderr)
+		err = signCert(args[1:], os.Stdout, os.Stderr, StdinPasswordReader{})
 	case "print":
 		err = printCert(args[1:], os.Stdout, os.Stderr)
 	case "verify":
@@ -127,6 +127,8 @@ func help(err string, out io.Writer) {
 	fmt.Fprintln(out, "    "+signSummary())
 	fmt.Fprintln(out, "    "+printSummary())
 	fmt.Fprintln(out, "    "+verifySummary())
+	fmt.Fprintln(out, "")
+	fmt.Fprintf(out, "  To see usage for a given mode, use %s <mode> -h\n", os.Args[0])
 }
 
 func mustFlagString(name string, val *string) error {

cmd/nebula-cert/main_test.go

@@ -22,7 +22,9 @@ func Test_help(t *testing.T) {
 		"    " + keygenSummary() + "\n" +
 		"    " + signSummary() + "\n" +
 		"    " + printSummary() + "\n" +
-		"    " + verifySummary() + "\n"
+		"    " + verifySummary() + "\n" +
+		"\n" +
+		"  To see usage for a given mode, use " + os.Args[0] + " <mode> -h\n"
 
 	ob := &bytes.Buffer{}
 

cmd/nebula-cert/passwords.go

@@ -0,0 +1,28 @@
+package main
+
+import (
+	"errors"
+	"fmt"
+	"os"
+
+	"golang.org/x/term"
+)
+
+var ErrNoTerminal = errors.New("cannot read password from nonexistent terminal")
+
+type PasswordReader interface {
+	ReadPassword() ([]byte, error)
+}
+
+type StdinPasswordReader struct{}
+
+func (pr StdinPasswordReader) ReadPassword() ([]byte, error) {
+	if !term.IsTerminal(int(os.Stdin.Fd())) {
+		return nil, ErrNoTerminal
+	}
+
+	password, err := term.ReadPassword(int(os.Stdin.Fd()))
+	fmt.Println()
+
+	return password, err
+}

cmd/nebula-cert/passwords_test.go

@@ -0,0 +1,10 @@
+package main
+
+type StubPasswordReader struct {
+	password []byte
+	err      error
+}
+
+func (pr *StubPasswordReader) ReadPassword() ([]byte, error) {
+	return pr.password, pr.err
+}

cmd/nebula-cert/print_test.go

@@ -87,7 +87,7 @@ func Test_printCert(t *testing.T) {
 	assert.Nil(t, err)
 	assert.Equal(
 		t,
-		"NebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\nNebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\nNebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\n",
+		"NebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t\tCurve: CURVE25519\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\nNebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t\tCurve: CURVE25519\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\nNebulaCertificate {\n\tDetails {\n\t\tName: test\n\t\tIps: []\n\t\tSubnets: []\n\t\tGroups: [\n\t\t\t\"hi\"\n\t\t]\n\t\tNot before: 0001-01-01 00:00:00 +0000 UTC\n\t\tNot After: 0001-01-01 00:00:00 +0000 UTC\n\t\tIs CA: false\n\t\tIssuer: \n\t\tPublic key: 0102030405060708090001020304050607080900010203040506070809000102\n\t\tCurve: CURVE25519\n\t}\n\tFingerprint: cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\n\tSignature: 0102030405060708090001020304050607080900010203040506070809000102\n}\n",
 		ob.String(),
 	)
 	assert.Equal(t, "", eb.String())
@@ -115,7 +115,7 @@ func Test_printCert(t *testing.T) {
 	assert.Nil(t, err)
 	assert.Equal(
 		t,
-		"{\"details\":{\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n{\"details\":{\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n{\"details\":{\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n",
+		"{\"details\":{\"curve\":\"CURVE25519\",\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n{\"details\":{\"curve\":\"CURVE25519\",\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n{\"details\":{\"curve\":\"CURVE25519\",\"groups\":[\"hi\"],\"ips\":[],\"isCa\":false,\"issuer\":\"\",\"name\":\"test\",\"notAfter\":\"0001-01-01T00:00:00Z\",\"notBefore\":\"0001-01-01T00:00:00Z\",\"publicKey\":\"0102030405060708090001020304050607080900010203040506070809000102\",\"subnets\":[]},\"fingerprint\":\"cc3492c0e9c48f17547f5987ea807462ebb3451e622590a10bb3763c344c82bd\",\"signature\":\"0102030405060708090001020304050607080900010203040506070809000102\"}\n",
 		ob.String(),
 	)
 	assert.Equal(t, "", eb.String())

cmd/nebula-cert/sign.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"crypto/ecdh"
 	"crypto/rand"
 	"flag"
 	"fmt"
@@ -37,19 +38,19 @@ func newSignFlags() *signFlags {
 	sf.caKeyPath = sf.set.String("ca-key", "ca.key", "Optional: path to the signing CA key")
 	sf.caCertPath = sf.set.String("ca-crt", "ca.crt", "Optional: path to the signing CA cert")
 	sf.name = sf.set.String("name", "", "Required: name of the cert, usually a hostname")
-	sf.ip = sf.set.String("ip", "", "Required: ip and network in CIDR notation to assign the cert")
+	sf.ip = sf.set.String("ip", "", "Required: ipv4 address and network in CIDR notation to assign the cert")
 	sf.duration = sf.set.Duration("duration", 0, "Optional: how long the cert should be valid for. The default is 1 second before the signing cert expires. Valid time units are seconds: \"s\", minutes: \"m\", hours: \"h\"")
 	sf.inPubPath = sf.set.String("in-pub", "", "Optional (if out-key not set): path to read a previously generated public key")
 	sf.outKeyPath = sf.set.String("out-key", "", "Optional (if in-pub not set): path to write the private key to")
 	sf.outCertPath = sf.set.String("out-crt", "", "Optional: path to write the certificate to")
 	sf.outQRPath = sf.set.String("out-qr", "", "Optional: output a qr code image (png) of the certificate")
 	sf.groups = sf.set.String("groups", "", "Optional: comma separated list of groups")
-	sf.subnets = sf.set.String("subnets", "", "Optional: comma separated list of subnet this cert can serve for")
+	sf.subnets = sf.set.String("subnets", "", "Optional: comma separated list of ipv4 address and network in CIDR notation. Subnets this cert can serve for")
 	return &sf
 
 }
 
-func signCert(args []string, out io.Writer, errOut io.Writer) error {
+func signCert(args []string, out io.Writer, errOut io.Writer, pr PasswordReader) error {
 	sf := newSignFlags()
 	err := sf.set.Parse(args)
 	if err != nil {
@@ -77,8 +78,37 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 		return fmt.Errorf("error while reading ca-key: %s", err)
 	}
 
-	caKey, _, err := cert.UnmarshalEd25519PrivateKey(rawCAKey)
-	if err != nil {
+	var curve cert.Curve
+	var caKey []byte
+
+	// naively attempt to decode the private key as though it is not encrypted
+	caKey, _, curve, err = cert.UnmarshalSigningPrivateKey(rawCAKey)
+	if err == cert.ErrPrivateKeyEncrypted {
+		// ask for a passphrase until we get one
+		var passphrase []byte
+		for i := 0; i < 5; i++ {
+			out.Write([]byte("Enter passphrase: "))
+			passphrase, err = pr.ReadPassword()
+
+			if err == ErrNoTerminal {
+				return fmt.Errorf("ca-key is encrypted and must be decrypted interactively")
+			} else if err != nil {
+				return fmt.Errorf("error reading password: %s", err)
+			}
+
+			if len(passphrase) > 0 {
+				break
+			}
+		}
+		if len(passphrase) == 0 {
+			return fmt.Errorf("cannot open encrypted ca-key without passphrase")
+		}
+
+		curve, caKey, _, err = cert.DecryptAndUnmarshalSigningPrivateKey(passphrase, rawCAKey)
+		if err != nil {
+			return fmt.Errorf("error while parsing encrypted ca-key: %s", err)
+		}
+	} else if err != nil {
 		return fmt.Errorf("error while parsing ca-key: %s", err)
 	}
 
@@ -92,6 +122,10 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 		return fmt.Errorf("error while parsing ca-crt: %s", err)
 	}
 
+	if err := caCert.VerifyPrivateKey(curve, caKey); err != nil {
+		return fmt.Errorf("refusing to sign, root certificate does not match private key")
+	}
+
 	issuer, err := caCert.Sha256Sum()
 	if err != nil {
 		return fmt.Errorf("error while getting -ca-crt fingerprint: %s", err)
@@ -110,6 +144,9 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 	if err != nil {
 		return newHelpErrorf("invalid ip definition: %s", err)
 	}
+	if ip.To4() == nil {
+		return newHelpErrorf("invalid ip definition: can only be ipv4, have %s", *sf.ip)
+	}
 	ipNet.IP = ip
 
 	groups := []string{}
@@ -131,6 +168,9 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 				if err != nil {
 					return newHelpErrorf("invalid subnet definition: %s", err)
 				}
+				if s.IP.To4() == nil {
+					return newHelpErrorf("invalid subnet definition: can only be ipv4, have %s", rs)
+				}
 				subnets = append(subnets, s)
 			}
 		}
@@ -142,12 +182,16 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 		if err != nil {
 			return fmt.Errorf("error while reading in-pub: %s", err)
 		}
-		pub, _, err = cert.UnmarshalX25519PublicKey(rawPub)
+		var pubCurve cert.Curve
+		pub, _, pubCurve, err = cert.UnmarshalPublicKey(rawPub)
 		if err != nil {
 			return fmt.Errorf("error while parsing in-pub: %s", err)
 		}
+		if pubCurve != curve {
+			return fmt.Errorf("curve of in-pub does not match ca")
+		}
 	} else {
-		pub, rawPriv = x25519Keypair()
+		pub, rawPriv = newKeypair(curve)
 	}
 
 	nc := cert.NebulaCertificate{
@@ -161,6 +205,7 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 			PublicKey: pub,
 			IsCA:      false,
 			Issuer:    issuer,
+			Curve:     curve,
 		},
 	}
 
@@ -180,7 +225,7 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 		return fmt.Errorf("refusing to overwrite existing cert: %s", *sf.outCertPath)
 	}
 
-	err = nc.Sign(caKey)
+	err = nc.Sign(curve, caKey)
 	if err != nil {
 		return fmt.Errorf("error while signing: %s", err)
 	}
@@ -190,7 +235,7 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 			return fmt.Errorf("refusing to overwrite existing key: %s", *sf.outKeyPath)
 		}
 
-		err = ioutil.WriteFile(*sf.outKeyPath, cert.MarshalX25519PrivateKey(rawPriv), 0600)
+		err = ioutil.WriteFile(*sf.outKeyPath, cert.MarshalPrivateKey(curve, rawPriv), 0600)
 		if err != nil {
 			return fmt.Errorf("error while writing out-key: %s", err)
 		}
@@ -221,13 +266,38 @@ func signCert(args []string, out io.Writer, errOut io.Writer) error {
 	return nil
 }
 
+func newKeypair(curve cert.Curve) ([]byte, []byte) {
+	switch curve {
+	case cert.Curve_CURVE25519:
+		return x25519Keypair()
+	case cert.Curve_P256:
+		return p256Keypair()
+	default:
+		return nil, nil
+	}
+}
+
 func x25519Keypair() ([]byte, []byte) {
-	var pubkey, privkey [32]byte
-	if _, err := io.ReadFull(rand.Reader, privkey[:]); err != nil {
+	privkey := make([]byte, 32)
+	if _, err := io.ReadFull(rand.Reader, privkey); err != nil {
 		panic(err)
 	}
-	curve25519.ScalarBaseMult(&pubkey, &privkey)
-	return pubkey[:], privkey[:]
+
+	pubkey, err := curve25519.X25519(privkey, curve25519.Basepoint)
+	if err != nil {
+		panic(err)
+	}
+
+	return pubkey, privkey
+}
+
+func p256Keypair() ([]byte, []byte) {
+	privkey, err := ecdh.P256().GenerateKey(rand.Reader)
+	if err != nil {
+		panic(err)
+	}
+	pubkey := privkey.PublicKey()
+	return pubkey.Bytes(), privkey.Bytes()
 }
 
 func signSummary() string {

cmd/nebula-cert/sign_test.go

@@ -1,3 +1,4 @@
+//go:build !windows
 // +build !windows
 
 package main
@@ -5,6 +6,7 @@ package main
 import (
 	"bytes"
 	"crypto/rand"
+	"errors"
 	"io/ioutil"
 	"os"
 	"testing"
@@ -38,7 +40,7 @@ func Test_signHelp(t *testing.T) {
 			"  -in-pub string\n"+
 			"    \tOptional (if out-key not set): path to read a previously generated public key\n"+
 			"  -ip string\n"+
-			"    \tRequired: ip and network in CIDR notation to assign the cert\n"+
+			"    \tRequired: ipv4 address and network in CIDR notation to assign the cert\n"+
 			"  -name string\n"+
 			"    \tRequired: name of the cert, usually a hostname\n"+
 			"  -out-crt string\n"+
@@ -48,7 +50,7 @@ func Test_signHelp(t *testing.T) {
 			"  -out-qr string\n"+
 			"    \tOptional: output a qr code image (png) of the certificate\n"+
 			"  -subnets string\n"+
-			"    \tOptional: comma separated list of subnet this cert can serve for\n",
+			"    \tOptional: comma separated list of ipv4 address and network in CIDR notation. Subnets this cert can serve for\n",
 		ob.String(),
 	)
 }
@@ -57,18 +59,39 @@ func Test_signCert(t *testing.T) {
 	ob := &bytes.Buffer{}
 	eb := &bytes.Buffer{}
 
-	// required args
+	nopw := &StubPasswordReader{
+		password: []byte(""),
+		err:      nil,
+	}
 
-	assertHelpError(t, signCert([]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-ip", "1.1.1.1/24", "-out-key", "nope", "-out-crt", "nope"}, ob, eb), "-name is required")
+	errpw := &StubPasswordReader{
+		password: []byte(""),
+		err:      errors.New("stub error"),
+	}
+
+	passphrase := []byte("DO NOT USE THIS KEY")
+	testpw := &StubPasswordReader{
+		password: passphrase,
+		err:      nil,
+	}
+
+	// required args
+	assertHelpError(t, signCert(
+		[]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-ip", "1.1.1.1/24", "-out-key", "nope", "-out-crt", "nope"}, ob, eb, nopw,
+	), "-name is required")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
-	assertHelpError(t, signCert([]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-name", "test", "-out-key", "nope", "-out-crt", "nope"}, ob, eb), "-ip is required")
+	assertHelpError(t, signCert(
+		[]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-name", "test", "-out-key", "nope", "-out-crt", "nope"}, ob, eb, nopw,
+	), "-ip is required")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
 	// cannot set -in-pub and -out-key
-	assertHelpError(t, signCert([]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-name", "test", "-in-pub", "nope", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope"}, ob, eb), "cannot set both -in-pub and -out-key")
+	assertHelpError(t, signCert(
+		[]string{"-ca-crt", "./nope", "-ca-key", "./nope", "-name", "test", "-in-pub", "nope", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope"}, ob, eb, nopw,
+	), "cannot set both -in-pub and -out-key")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -76,7 +99,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args := []string{"-ca-crt", "./nope", "-ca-key", "./nope", "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while reading ca-key: open ./nope: "+NoSuchFileError)
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while reading ca-key: open ./nope: "+NoSuchFileError)
 
 	// failed to unmarshal key
 	ob.Reset()
@@ -86,7 +109,7 @@ func Test_signCert(t *testing.T) {
 	defer os.Remove(caKeyF.Name())
 
 	args = []string{"-ca-crt", "./nope", "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while parsing ca-key: input did not contain a valid PEM encoded block")
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while parsing ca-key: input did not contain a valid PEM encoded block")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -98,7 +121,7 @@ func Test_signCert(t *testing.T) {
 
 	// failed to read cert
 	args = []string{"-ca-crt", "./nope", "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while reading ca-crt: open ./nope: "+NoSuchFileError)
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while reading ca-crt: open ./nope: "+NoSuchFileError)
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -110,7 +133,7 @@ func Test_signCert(t *testing.T) {
 	defer os.Remove(caCrtF.Name())
 
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while parsing ca-crt: input did not contain a valid PEM encoded block")
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while parsing ca-crt: input did not contain a valid PEM encoded block")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -129,7 +152,7 @@ func Test_signCert(t *testing.T) {
 
 	// failed to read pub
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-in-pub", "./nope", "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while reading in-pub: open ./nope: "+NoSuchFileError)
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while reading in-pub: open ./nope: "+NoSuchFileError)
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -141,7 +164,7 @@ func Test_signCert(t *testing.T) {
 	defer os.Remove(inPubF.Name())
 
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-in-pub", inPubF.Name(), "-duration", "100m"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while parsing in-pub: input did not contain a valid PEM encoded block")
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while parsing in-pub: input did not contain a valid PEM encoded block")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -155,7 +178,14 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "a1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
-	assertHelpError(t, signCert(args, ob, eb), "invalid ip definition: invalid CIDR address: a1.1.1.1/24")
+	assertHelpError(t, signCert(args, ob, eb, nopw), "invalid ip definition: invalid CIDR address: a1.1.1.1/24")
+	assert.Empty(t, ob.String())
+	assert.Empty(t, eb.String())
+
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "100::100/100", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m"}
+	assertHelpError(t, signCert(args, ob, eb, nopw), "invalid ip definition: can only be ipv4, have 100::100/100")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -163,7 +193,28 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m", "-subnets", "a"}
-	assertHelpError(t, signCert(args, ob, eb), "invalid subnet definition: invalid CIDR address: a")
+	assertHelpError(t, signCert(args, ob, eb, nopw), "invalid subnet definition: invalid CIDR address: a")
+	assert.Empty(t, ob.String())
+	assert.Empty(t, eb.String())
+
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m", "-subnets", "100::100/100"}
+	assertHelpError(t, signCert(args, ob, eb, nopw), "invalid subnet definition: can only be ipv4, have 100::100/100")
+	assert.Empty(t, ob.String())
+	assert.Empty(t, eb.String())
+
+	// mismatched ca key
+	_, caPriv2, _ := ed25519.GenerateKey(rand.Reader)
+	caKeyF2, err := ioutil.TempFile("", "sign-cert-2.key")
+	assert.Nil(t, err)
+	defer os.Remove(caKeyF2.Name())
+	caKeyF2.Write(cert.MarshalEd25519PrivateKey(caPriv2))
+
+	ob.Reset()
+	eb.Reset()
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF2.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "nope", "-out-key", "nope", "-duration", "100m", "-subnets", "a"}
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "refusing to sign, root certificate does not match private key")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -171,7 +222,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "/do/not/write/pleasecrt", "-out-key", "/do/not/write/pleasekey", "-duration", "100m", "-subnets", "10.1.1.1/32"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while writing out-key: open /do/not/write/pleasekey: "+NoSuchDirError)
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while writing out-key: open /do/not/write/pleasekey: "+NoSuchDirError)
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -184,7 +235,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", "/do/not/write/pleasecrt", "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32"}
-	assert.EqualError(t, signCert(args, ob, eb), "error while writing out-crt: open /do/not/write/pleasecrt: "+NoSuchDirError)
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "error while writing out-crt: open /do/not/write/pleasecrt: "+NoSuchDirError)
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 	os.Remove(keyF.Name())
@@ -198,7 +249,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.Nil(t, signCert(args, ob, eb))
+	assert.Nil(t, signCert(args, ob, eb, nopw))
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -240,7 +291,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-in-pub", inPubF.Name(), "-duration", "100m", "-groups", "1"}
-	assert.Nil(t, signCert(args, ob, eb))
+	assert.Nil(t, signCert(args, ob, eb, nopw))
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -255,7 +306,7 @@ func Test_signCert(t *testing.T) {
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "1000m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.EqualError(t, signCert(args, ob, eb), "refusing to sign, root certificate constraints violated: certificate expires after signing certificate")
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "refusing to sign, root certificate constraints violated: certificate expires after signing certificate")
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -263,14 +314,14 @@ func Test_signCert(t *testing.T) {
 	os.Remove(keyF.Name())
 	os.Remove(crtF.Name())
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.Nil(t, signCert(args, ob, eb))
+	assert.Nil(t, signCert(args, ob, eb, nopw))
 
 	// test that we won't overwrite existing key file
 	os.Remove(crtF.Name())
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.EqualError(t, signCert(args, ob, eb), "refusing to overwrite existing key: "+keyF.Name())
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "refusing to overwrite existing key: "+keyF.Name())
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
 
@@ -278,14 +329,83 @@ func Test_signCert(t *testing.T) {
 	os.Remove(keyF.Name())
 	os.Remove(crtF.Name())
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.Nil(t, signCert(args, ob, eb))
+	assert.Nil(t, signCert(args, ob, eb, nopw))
 
 	// test that we won't overwrite existing certificate file
 	os.Remove(keyF.Name())
 	ob.Reset()
 	eb.Reset()
 	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
-	assert.EqualError(t, signCert(args, ob, eb), "refusing to overwrite existing cert: "+crtF.Name())
+	assert.EqualError(t, signCert(args, ob, eb, nopw), "refusing to overwrite existing cert: "+crtF.Name())
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
+
+	// create valid cert/key using encrypted CA key
+	os.Remove(caKeyF.Name())
+	os.Remove(caCrtF.Name())
+	os.Remove(keyF.Name())
+	os.Remove(crtF.Name())
+	ob.Reset()
+	eb.Reset()
+
+	caKeyF, err = ioutil.TempFile("", "sign-cert.key")
+	assert.Nil(t, err)
+	defer os.Remove(caKeyF.Name())
+
+	caCrtF, err = ioutil.TempFile("", "sign-cert.crt")
+	assert.Nil(t, err)
+	defer os.Remove(caCrtF.Name())
+
+	// generate the encrypted key
+	caPub, caPriv, _ = ed25519.GenerateKey(rand.Reader)
+	kdfParams := cert.NewArgon2Parameters(64*1024, 4, 3)
+	b, _ = cert.EncryptAndMarshalSigningPrivateKey(cert.Curve_CURVE25519, caPriv, passphrase, kdfParams)
+	caKeyF.Write(b)
+
+	ca = cert.NebulaCertificate{
+		Details: cert.NebulaCertificateDetails{
+			Name:      "ca",
+			NotBefore: time.Now(),
+			NotAfter:  time.Now().Add(time.Minute * 200),
+			PublicKey: caPub,
+			IsCA:      true,
+		},
+	}
+	b, _ = ca.MarshalToPEM()
+	caCrtF.Write(b)
+
+	// test with the proper password
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
+	assert.Nil(t, signCert(args, ob, eb, testpw))
+	assert.Equal(t, "Enter passphrase: ", ob.String())
+	assert.Empty(t, eb.String())
+
+	// test with the wrong password
+	ob.Reset()
+	eb.Reset()
+
+	testpw.password = []byte("invalid password")
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
+	assert.Error(t, signCert(args, ob, eb, testpw))
+	assert.Equal(t, "Enter passphrase: ", ob.String())
+	assert.Empty(t, eb.String())
+
+	// test with the user not entering a password
+	ob.Reset()
+	eb.Reset()
+
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
+	assert.Error(t, signCert(args, ob, eb, nopw))
+	// normally the user hitting enter on the prompt would add newlines between these
+	assert.Equal(t, "Enter passphrase: Enter passphrase: Enter passphrase: Enter passphrase: Enter passphrase: ", ob.String())
+	assert.Empty(t, eb.String())
+
+	// test an error condition
+	ob.Reset()
+	eb.Reset()
+
+	args = []string{"-ca-crt", caCrtF.Name(), "-ca-key", caKeyF.Name(), "-name", "test", "-ip", "1.1.1.1/24", "-out-crt", crtF.Name(), "-out-key", keyF.Name(), "-duration", "100m", "-subnets", "10.1.1.1/32, ,   10.2.2.2/32   ,   ,  ,, 10.5.5.5/32", "-groups", "1,,   2    ,        ,,,3,4,5"}
+	assert.Error(t, signCert(args, ob, eb, errpw))
+	assert.Equal(t, "Enter passphrase: ", ob.String())
+	assert.Empty(t, eb.String())
 }

cmd/nebula-cert/verify_test.go

@@ -72,12 +72,12 @@ func Test_verify(t *testing.T) {
 		Details: cert.NebulaCertificateDetails{
 			Name:      "test-ca",
 			NotBefore: time.Now().Add(time.Hour * -1),
-			NotAfter:  time.Now().Add(time.Hour),
+			NotAfter:  time.Now().Add(time.Hour * 2),
 			PublicKey: caPub,
 			IsCA:      true,
 		},
 	}
-	ca.Sign(caPriv)
+	ca.Sign(cert.Curve_CURVE25519, caPriv)
 	b, _ := ca.MarshalToPEM()
 	caFile.Truncate(0)
 	caFile.Seek(0, 0)
@@ -117,7 +117,7 @@ func Test_verify(t *testing.T) {
 		},
 	}
 
-	crt.Sign(badPriv)
+	crt.Sign(cert.Curve_CURVE25519, badPriv)
 	b, _ = crt.MarshalToPEM()
 	certFile.Truncate(0)
 	certFile.Seek(0, 0)
@@ -129,7 +129,7 @@ func Test_verify(t *testing.T) {
 	assert.EqualError(t, err, "certificate signature did not match")
 
 	// verified cert at path
-	crt.Sign(caPriv)
+	crt.Sign(cert.Curve_CURVE25519, caPriv)
 	b, _ = crt.MarshalToPEM()
 	certFile.Truncate(0)
 	certFile.Seek(0, 0)

cmd/nebula-service/logs_generic.go

@@ -1,3 +1,4 @@
+//go:build !windows
 // +build !windows
 
 package main

cmd/nebula-service/main.go

@@ -7,11 +7,13 @@ import (
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/util"
 )
 
 // A version string that can be set with
 //
-//     -ldflags "-X main.Build=SOMEVERSION"
+//	-ldflags "-X main.Build=SOMEVERSION"
 //
 // at compile-time.
 var Build string
@@ -49,17 +51,17 @@ func main() {
 	l := logrus.New()
 	l.Out = os.Stdout
 
-	config := nebula.NewConfig(l)
-	err := config.Load(*configPath)
+	c := config.NewC(l)
+	err := c.Load(*configPath)
 	if err != nil {
 		fmt.Printf("failed to load config: %s", err)
 		os.Exit(1)
 	}
 
-	c, err := nebula.Main(config, *configTest, Build, l, nil)
+	ctrl, err := nebula.Main(c, *configTest, Build, l, nil)
 
 	switch v := err.(type) {
-	case nebula.ContextualError:
+	case util.ContextualError:
 		v.Log(l)
 		os.Exit(1)
 	case error:
@@ -68,8 +70,8 @@ func main() {
 	}
 
 	if !*configTest {
-		c.Start()
-		c.ShutdownBlock()
+		ctrl.Start()
+		ctrl.ShutdownBlock()
 	}
 
 	os.Exit(0)

cmd/nebula-service/service.go

@@ -9,6 +9,7 @@ import (
 	"github.com/kardianos/service"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/config"
 )
 
 var logger service.Logger
@@ -27,13 +28,13 @@ func (p *program) Start(s service.Service) error {
 	l := logrus.New()
 	HookLogger(l)
 
-	config := nebula.NewConfig(l)
-	err := config.Load(*p.configPath)
+	c := config.NewC(l)
+	err := c.Load(*p.configPath)
 	if err != nil {
 		return fmt.Errorf("failed to load config: %s", err)
 	}
 
-	p.control, err = nebula.Main(config, *p.configTest, Build, l, nil)
+	p.control, err = nebula.Main(c, *p.configTest, Build, l, nil)
 	if err != nil {
 		return err
 	}
@@ -48,6 +49,14 @@ func (p *program) Stop(s service.Service) error {
 	return nil
 }
 
+func fileExists(filename string) bool {
+	_, err := os.Stat(filename)
+	if os.IsNotExist(err) {
+		return false
+	}
+	return true
+}
+
 func doService(configPath *string, configTest *bool, build string, serviceFlag *string) {
 	if *configPath == "" {
 		ex, err := os.Executable()
@@ -55,6 +64,9 @@ func doService(configPath *string, configTest *bool, build string, serviceFlag *
 			panic(err)
 		}
 		*configPath = filepath.Dir(ex) + "/config.yaml"
+		if !fileExists(*configPath) {
+			*configPath = filepath.Dir(ex) + "/config.yml"
+		}
 	}
 
 	svcConfig := &service.Config{

cmd/nebula/main.go

@@ -7,11 +7,13 @@ import (
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/util"
 )
 
 // A version string that can be set with
 //
-//     -ldflags "-X main.Build=SOMEVERSION"
+//	-ldflags "-X main.Build=SOMEVERSION"
 //
 // at compile-time.
 var Build string
@@ -43,17 +45,17 @@ func main() {
 	l := logrus.New()
 	l.Out = os.Stdout
 
-	config := nebula.NewConfig(l)
-	err := config.Load(*configPath)
+	c := config.NewC(l)
+	err := c.Load(*configPath)
 	if err != nil {
 		fmt.Printf("failed to load config: %s", err)
 		os.Exit(1)
 	}
 
-	c, err := nebula.Main(config, *configTest, Build, l, nil)
+	ctrl, err := nebula.Main(c, *configTest, Build, l, nil)
 
 	switch v := err.(type) {
-	case nebula.ContextualError:
+	case util.ContextualError:
 		v.Log(l)
 		os.Exit(1)
 	case error:
@@ -62,8 +64,8 @@ func main() {
 	}
 
 	if !*configTest {
-		c.Start()
-		c.ShutdownBlock()
+		ctrl.Start()
+		ctrl.ShutdownBlock()
 	}
 
 	os.Exit(0)

config/config.go

@@ -1,17 +1,17 @@
-package nebula
+package config
 
 import (
+	"context"
 	"errors"
 	"fmt"
 	"io/ioutil"
-	"net"
 	"os"
 	"os/signal"
 	"path/filepath"
-	"regexp"
 	"sort"
 	"strconv"
 	"strings"
+	"sync"
 	"syscall"
 	"time"
 
@@ -20,24 +20,25 @@ import (
 	"gopkg.in/yaml.v2"
 )
 
-type Config struct {
+type C struct {
 	path        string
 	files       []string
 	Settings    map[interface{}]interface{}
 	oldSettings map[interface{}]interface{}
-	callbacks   []func(*Config)
+	callbacks   []func(*C)
 	l           *logrus.Logger
+	reloadLock  sync.Mutex
 }
 
-func NewConfig(l *logrus.Logger) *Config {
-	return &Config{
+func NewC(l *logrus.Logger) *C {
+	return &C{
 		Settings: make(map[interface{}]interface{}),
 		l:        l,
 	}
 }
 
 // Load will find all yaml files within path and load them in lexical order
-func (c *Config) Load(path string) error {
+func (c *C) Load(path string) error {
 	c.path = path
 	c.files = make([]string, 0)
 
@@ -60,7 +61,7 @@ func (c *Config) Load(path string) error {
 	return nil
 }
 
-func (c *Config) LoadString(raw string) error {
+func (c *C) LoadString(raw string) error {
 	if raw == "" {
 		return errors.New("Empty configuration")
 	}
@@ -71,16 +72,21 @@ func (c *Config) LoadString(raw string) error {
 // here should decide if they need to make a change to the current process before making the change. HasChanged can be
 // used to help decide if a change is necessary.
 // These functions should return quickly or spawn their own go routine if they will take a while
-func (c *Config) RegisterReloadCallback(f func(*Config)) {
+func (c *C) RegisterReloadCallback(f func(*C)) {
 	c.callbacks = append(c.callbacks, f)
 }
 
+// InitialLoad returns true if this is the first load of the config, and ReloadConfig has not been called yet.
+func (c *C) InitialLoad() bool {
+	return c.oldSettings == nil
+}
+
 // HasChanged checks if the underlying structure of the provided key has changed after a config reload. The value of
 // k in both the old and new settings will be serialized, the result of the string comparison is returned.
 // If k is an empty string the entire config is tested.
 // It's important to note that this is very rudimentary and susceptible to configuration ordering issues indicating
 // there is change when there actually wasn't any.
-func (c *Config) HasChanged(k string) bool {
+func (c *C) HasChanged(k string) bool {
 	if c.oldSettings == nil {
 		return false
 	}
@@ -114,19 +120,29 @@ func (c *Config) HasChanged(k string) bool {
 
 // CatchHUP will listen for the HUP signal in a go routine and reload all configs found in the
 // original path provided to Load. The old settings are shallow copied for change detection after the reload.
-func (c *Config) CatchHUP() {
+func (c *C) CatchHUP(ctx context.Context) {
 	ch := make(chan os.Signal, 1)
 	signal.Notify(ch, syscall.SIGHUP)
 
 	go func() {
-		for range ch {
-			c.l.Info("Caught HUP, reloading config")
-			c.ReloadConfig()
+		for {
+			select {
+			case <-ctx.Done():
+				signal.Stop(ch)
+				close(ch)
+				return
+			case <-ch:
+				c.l.Info("Caught HUP, reloading config")
+				c.ReloadConfig()
+			}
 		}
 	}()
 }
 
-func (c *Config) ReloadConfig() {
+func (c *C) ReloadConfig() {
+	c.reloadLock.Lock()
+	defer c.reloadLock.Unlock()
+
 	c.oldSettings = make(map[interface{}]interface{})
 	for k, v := range c.Settings {
 		c.oldSettings[k] = v
@@ -143,8 +159,29 @@ func (c *Config) ReloadConfig() {
 	}
 }
 
+func (c *C) ReloadConfigString(raw string) error {
+	c.reloadLock.Lock()
+	defer c.reloadLock.Unlock()
+
+	c.oldSettings = make(map[interface{}]interface{})
+	for k, v := range c.Settings {
+		c.oldSettings[k] = v
+	}
+
+	err := c.LoadString(raw)
+	if err != nil {
+		return err
+	}
+
+	for _, v := range c.callbacks {
+		v(c)
+	}
+
+	return nil
+}
+
 // GetString will get the string for k or return the default d if not found or invalid
-func (c *Config) GetString(k, d string) string {
+func (c *C) GetString(k, d string) string {
 	r := c.Get(k)
 	if r == nil {
 		return d
@@ -154,7 +191,7 @@ func (c *Config) GetString(k, d string) string {
 }
 
 // GetStringSlice will get the slice of strings for k or return the default d if not found or invalid
-func (c *Config) GetStringSlice(k string, d []string) []string {
+func (c *C) GetStringSlice(k string, d []string) []string {
 	r := c.Get(k)
 	if r == nil {
 		return d
@@ -174,7 +211,7 @@ func (c *Config) GetStringSlice(k string, d []string) []string {
 }
 
 // GetMap will get the map for k or return the default d if not found or invalid
-func (c *Config) GetMap(k string, d map[interface{}]interface{}) map[interface{}]interface{} {
+func (c *C) GetMap(k string, d map[interface{}]interface{}) map[interface{}]interface{} {
 	r := c.Get(k)
 	if r == nil {
 		return d
@@ -189,7 +226,7 @@ func (c *Config) GetMap(k string, d map[interface{}]interface{}) map[interface{}
 }
 
 // GetInt will get the int for k or return the default d if not found or invalid
-func (c *Config) GetInt(k string, d int) int {
+func (c *C) GetInt(k string, d int) int {
 	r := c.GetString(k, strconv.Itoa(d))
 	v, err := strconv.Atoi(r)
 	if err != nil {
@@ -200,7 +237,7 @@ func (c *Config) GetInt(k string, d int) int {
 }
 
 // GetBool will get the bool for k or return the default d if not found or invalid
-func (c *Config) GetBool(k string, d bool) bool {
+func (c *C) GetBool(k string, d bool) bool {
 	r := strings.ToLower(c.GetString(k, fmt.Sprintf("%v", d)))
 	v, err := strconv.ParseBool(r)
 	if err != nil {
@@ -217,7 +254,7 @@ func (c *Config) GetBool(k string, d bool) bool {
 }
 
 // GetDuration will get the duration for k or return the default d if not found or invalid
-func (c *Config) GetDuration(k string, d time.Duration) time.Duration {
+func (c *C) GetDuration(k string, d time.Duration) time.Duration {
 	r := c.GetString(k, "")
 	v, err := time.ParseDuration(r)
 	if err != nil {
@@ -226,160 +263,15 @@ func (c *Config) GetDuration(k string, d time.Duration) time.Duration {
 	return v
 }
 
-func (c *Config) GetAllowList(k string, allowInterfaces bool) (*AllowList, error) {
-	r := c.Get(k)
-	if r == nil {
-		return nil, nil
-	}
-
-	rawMap, ok := r.(map[interface{}]interface{})
-	if !ok {
-		return nil, fmt.Errorf("config `%s` has invalid type: %T", k, r)
-	}
-
-	tree := NewCIDR6Tree()
-	var nameRules []AllowListNameRule
-
-	// Keep track of the rules we have added for both ipv4 and ipv6
-	type allowListRules struct {
-		firstValue     bool
-		allValuesMatch bool
-		defaultSet     bool
-		allValues      bool
-	}
-	rules4 := allowListRules{firstValue: true, allValuesMatch: true, defaultSet: false}
-	rules6 := allowListRules{firstValue: true, allValuesMatch: true, defaultSet: false}
-
-	for rawKey, rawValue := range rawMap {
-		rawCIDR, ok := rawKey.(string)
-		if !ok {
-			return nil, fmt.Errorf("config `%s` has invalid key (type %T): %v", k, rawKey, rawKey)
-		}
-
-		// Special rule for interface names
-		if rawCIDR == "interfaces" {
-			if !allowInterfaces {
-				return nil, fmt.Errorf("config `%s` does not support `interfaces`", k)
-			}
-			var err error
-			nameRules, err = c.getAllowListInterfaces(k, rawValue)
-			if err != nil {
-				return nil, err
-			}
-
-			continue
-		}
-
-		value, ok := rawValue.(bool)
-		if !ok {
-			return nil, fmt.Errorf("config `%s` has invalid value (type %T): %v", k, rawValue, rawValue)
-		}
-
-		_, cidr, err := net.ParseCIDR(rawCIDR)
-		if err != nil {
-			return nil, fmt.Errorf("config `%s` has invalid CIDR: %s", k, rawCIDR)
-		}
-
-		// TODO: should we error on duplicate CIDRs in the config?
-		tree.AddCIDR(cidr, value)
-
-		maskBits, maskSize := cidr.Mask.Size()
-
-		var rules *allowListRules
-		if maskSize == 32 {
-			rules = &rules4
-		} else {
-			rules = &rules6
-		}
-
-		if rules.firstValue {
-			rules.allValues = value
-			rules.firstValue = false
-		} else {
-			if value != rules.allValues {
-				rules.allValuesMatch = false
-			}
-		}
-
-		// Check if this is 0.0.0.0/0 or ::/0
-		if maskBits == 0 {
-			rules.defaultSet = true
-		}
-	}
-
-	if !rules4.defaultSet {
-		if rules4.allValuesMatch {
-			_, zeroCIDR, _ := net.ParseCIDR("0.0.0.0/0")
-			tree.AddCIDR(zeroCIDR, !rules4.allValues)
-		} else {
-			return nil, fmt.Errorf("config `%s` contains both true and false rules, but no default set for 0.0.0.0/0", k)
-		}
-	}
-
-	if !rules6.defaultSet {
-		if rules6.allValuesMatch {
-			_, zeroCIDR, _ := net.ParseCIDR("::/0")
-			tree.AddCIDR(zeroCIDR, !rules6.allValues)
-		} else {
-			return nil, fmt.Errorf("config `%s` contains both true and false rules, but no default set for ::/0", k)
-		}
-	}
-
-	return &AllowList{cidrTree: tree, nameRules: nameRules}, nil
-}
-
-func (c *Config) getAllowListInterfaces(k string, v interface{}) ([]AllowListNameRule, error) {
-	var nameRules []AllowListNameRule
-
-	rawRules, ok := v.(map[interface{}]interface{})
-	if !ok {
-		return nil, fmt.Errorf("config `%s.interfaces` is invalid (type %T): %v", k, v, v)
-	}
-
-	firstEntry := true
-	var allValues bool
-	for rawName, rawAllow := range rawRules {
-		name, ok := rawName.(string)
-		if !ok {
-			return nil, fmt.Errorf("config `%s.interfaces` has invalid key (type %T): %v", k, rawName, rawName)
-		}
-		allow, ok := rawAllow.(bool)
-		if !ok {
-			return nil, fmt.Errorf("config `%s.interfaces` has invalid value (type %T): %v", k, rawAllow, rawAllow)
-		}
-
-		nameRE, err := regexp.Compile("^" + name + "$")
-		if err != nil {
-			return nil, fmt.Errorf("config `%s.interfaces` has invalid key: %s: %v", k, name, err)
-		}
-
-		nameRules = append(nameRules, AllowListNameRule{
-			Name:  nameRE,
-			Allow: allow,
-		})
-
-		if firstEntry {
-			allValues = allow
-			firstEntry = false
-		} else {
-			if allow != allValues {
-				return nil, fmt.Errorf("config `%s.interfaces` values must all be the same true/false value", k)
-			}
-		}
-	}
-
-	return nameRules, nil
-}
-
-func (c *Config) Get(k string) interface{} {
+func (c *C) Get(k string) interface{} {
 	return c.get(k, c.Settings)
 }
 
-func (c *Config) IsSet(k string) bool {
+func (c *C) IsSet(k string) bool {
 	return c.get(k, c.Settings) != nil
 }
 
-func (c *Config) get(k string, v interface{}) interface{} {
+func (c *C) get(k string, v interface{}) interface{} {
 	parts := strings.Split(k, ".")
 	for _, p := range parts {
 		m, ok := v.(map[interface{}]interface{})
@@ -398,7 +290,7 @@ func (c *Config) get(k string, v interface{}) interface{} {
 
 // direct signifies if this is the config path directly specified by the user,
 // versus a file/dir found by recursing into that path
-func (c *Config) resolve(path string, direct bool) error {
+func (c *C) resolve(path string, direct bool) error {
 	i, err := os.Stat(path)
 	if err != nil {
 		return nil
@@ -424,7 +316,7 @@ func (c *Config) resolve(path string, direct bool) error {
 	return nil
 }
 
-func (c *Config) addFile(path string, direct bool) error {
+func (c *C) addFile(path string, direct bool) error {
 	ext := filepath.Ext(path)
 
 	if !direct && ext != ".yaml" && ext != ".yml" {
@@ -440,7 +332,7 @@ func (c *Config) addFile(path string, direct bool) error {
 	return nil
 }
 
-func (c *Config) parseRaw(b []byte) error {
+func (c *C) parseRaw(b []byte) error {
 	var m map[interface{}]interface{}
 
 	err := yaml.Unmarshal(b, &m)
@@ -452,7 +344,7 @@ func (c *Config) parseRaw(b []byte) error {
 	return nil
 }
 
-func (c *Config) parse() error {
+func (c *C) parse() error {
 	var m map[interface{}]interface{}
 
 	for _, path := range c.files {
@@ -495,38 +387,3 @@ func readDirNames(path string) ([]string, error) {
 	sort.Strings(paths)
 	return paths, nil
 }
-
-func configLogger(c *Config) error {
-	// set up our logging level
-	logLevel, err := logrus.ParseLevel(strings.ToLower(c.GetString("logging.level", "info")))
-	if err != nil {
-		return fmt.Errorf("%s; possible levels: %s", err, logrus.AllLevels)
-	}
-	c.l.SetLevel(logLevel)
-
-	disableTimestamp := c.GetBool("logging.disable_timestamp", false)
-	timestampFormat := c.GetString("logging.timestamp_format", "")
-	fullTimestamp := (timestampFormat != "")
-	if timestampFormat == "" {
-		timestampFormat = time.RFC3339
-	}
-
-	logFormat := strings.ToLower(c.GetString("logging.format", "text"))
-	switch logFormat {
-	case "text":
-		c.l.Formatter = &logrus.TextFormatter{
-			TimestampFormat:  timestampFormat,
-			FullTimestamp:    fullTimestamp,
-			DisableTimestamp: disableTimestamp,
-		}
-	case "json":
-		c.l.Formatter = &logrus.JSONFormatter{
-			TimestampFormat:  timestampFormat,
-			DisableTimestamp: disableTimestamp,
-		}
-	default:
-		return fmt.Errorf("unknown log format `%s`. possible formats: %s", logFormat, []string{"text", "json"})
-	}
-
-	return nil
-}

config/config_test.go

@@ -1,4 +1,4 @@
-package nebula
+package config
 
 import (
 	"io/ioutil"
@@ -7,19 +7,23 @@ import (
 	"testing"
 	"time"
 
+	"github.com/imdario/mergo"
+	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"gopkg.in/yaml.v2"
 )
 
 func TestConfig_Load(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	dir, err := ioutil.TempDir("", "config-test")
 	// invalid yaml
-	c := NewConfig(l)
+	c := NewC(l)
 	ioutil.WriteFile(filepath.Join(dir, "01.yaml"), []byte(" invalid yaml"), 0644)
 	assert.EqualError(t, c.Load(dir), "yaml: unmarshal errors:\n  line 1: cannot unmarshal !!str `invalid...` into map[interface {}]interface {}")
 
 	// simple multi config merge
-	c = NewConfig(l)
+	c = NewC(l)
 	os.RemoveAll(dir)
 	os.Mkdir(dir, 0755)
 
@@ -41,9 +45,9 @@ func TestConfig_Load(t *testing.T) {
 }
 
 func TestConfig_Get(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	// test simple type
-	c := NewConfig(l)
+	c := NewC(l)
 	c.Settings["firewall"] = map[interface{}]interface{}{"outbound": "hi"}
 	assert.Equal(t, "hi", c.Get("firewall.outbound"))
 
@@ -57,15 +61,15 @@ func TestConfig_Get(t *testing.T) {
 }
 
 func TestConfig_GetStringSlice(t *testing.T) {
-	l := NewTestLogger()
-	c := NewConfig(l)
+	l := test.NewLogger()
+	c := NewC(l)
 	c.Settings["slice"] = []interface{}{"one", "two"}
 	assert.Equal(t, []string{"one", "two"}, c.GetStringSlice("slice", []string{}))
 }
 
 func TestConfig_GetBool(t *testing.T) {
-	l := NewTestLogger()
-	c := NewConfig(l)
+	l := test.NewLogger()
+	c := NewC(l)
 	c.Settings["bool"] = true
 	assert.Equal(t, true, c.GetBool("bool", false))
 
@@ -91,116 +95,22 @@ func TestConfig_GetBool(t *testing.T) {
 	assert.Equal(t, false, c.GetBool("bool", true))
 }
 
-func TestConfig_GetAllowList(t *testing.T) {
-	l := NewTestLogger()
-	c := NewConfig(l)
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"192.168.0.0": true,
-	}
-	r, err := c.GetAllowList("allowlist", false)
-	assert.EqualError(t, err, "config `allowlist` has invalid CIDR: 192.168.0.0")
-	assert.Nil(t, r)
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"192.168.0.0/16": "abc",
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	assert.EqualError(t, err, "config `allowlist` has invalid value (type string): abc")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"192.168.0.0/16": true,
-		"10.0.0.0/8":     false,
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	assert.EqualError(t, err, "config `allowlist` contains both true and false rules, but no default set for 0.0.0.0/0")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"0.0.0.0/0":      true,
-		"10.0.0.0/8":     false,
-		"10.42.42.0/24":  true,
-		"fd00::/8":       true,
-		"fd00:fd00::/16": false,
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	assert.EqualError(t, err, "config `allowlist` contains both true and false rules, but no default set for ::/0")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"0.0.0.0/0":     true,
-		"10.0.0.0/8":    false,
-		"10.42.42.0/24": true,
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	if assert.NoError(t, err) {
-		assert.NotNil(t, r)
-	}
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"0.0.0.0/0":      true,
-		"10.0.0.0/8":     false,
-		"10.42.42.0/24":  true,
-		"::/0":           false,
-		"fd00::/8":       true,
-		"fd00:fd00::/16": false,
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	if assert.NoError(t, err) {
-		assert.NotNil(t, r)
-	}
-
-	// Test interface names
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"interfaces": map[interface{}]interface{}{
-			`docker.*`: false,
-		},
-	}
-	r, err = c.GetAllowList("allowlist", false)
-	assert.EqualError(t, err, "config `allowlist` does not support `interfaces`")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"interfaces": map[interface{}]interface{}{
-			`docker.*`: "foo",
-		},
-	}
-	r, err = c.GetAllowList("allowlist", true)
-	assert.EqualError(t, err, "config `allowlist.interfaces` has invalid value (type string): foo")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"interfaces": map[interface{}]interface{}{
-			`docker.*`: false,
-			`eth.*`:    true,
-		},
-	}
-	r, err = c.GetAllowList("allowlist", true)
-	assert.EqualError(t, err, "config `allowlist.interfaces` values must all be the same true/false value")
-
-	c.Settings["allowlist"] = map[interface{}]interface{}{
-		"interfaces": map[interface{}]interface{}{
-			`docker.*`: false,
-		},
-	}
-	r, err = c.GetAllowList("allowlist", true)
-	if assert.NoError(t, err) {
-		assert.NotNil(t, r)
-	}
-}
-
 func TestConfig_HasChanged(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	// No reload has occurred, return false
-	c := NewConfig(l)
+	c := NewC(l)
 	c.Settings["test"] = "hi"
 	assert.False(t, c.HasChanged(""))
 
 	// Test key change
-	c = NewConfig(l)
+	c = NewC(l)
 	c.Settings["test"] = "hi"
 	c.oldSettings = map[interface{}]interface{}{"test": "no"}
 	assert.True(t, c.HasChanged("test"))
 	assert.True(t, c.HasChanged(""))
 
 	// No key change
-	c = NewConfig(l)
+	c = NewC(l)
 	c.Settings["test"] = "hi"
 	c.oldSettings = map[interface{}]interface{}{"test": "hi"}
 	assert.False(t, c.HasChanged("test"))
@@ -208,13 +118,13 @@ func TestConfig_HasChanged(t *testing.T) {
 }
 
 func TestConfig_ReloadConfig(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	done := make(chan bool, 1)
 	dir, err := ioutil.TempDir("", "config-test")
 	assert.Nil(t, err)
 	ioutil.WriteFile(filepath.Join(dir, "01.yaml"), []byte("outer:\n  inner: hi"), 0644)
 
-	c := NewConfig(l)
+	c := NewC(l)
 	assert.Nil(t, c.Load(dir))
 
 	assert.False(t, c.HasChanged("outer.inner"))
@@ -223,7 +133,7 @@ func TestConfig_ReloadConfig(t *testing.T) {
 
 	ioutil.WriteFile(filepath.Join(dir, "01.yaml"), []byte("outer:\n  inner: ho"), 0644)
 
-	c.RegisterReloadCallback(func(c *Config) {
+	c.RegisterReloadCallback(func(c *C) {
 		done <- true
 	})
 
@@ -240,3 +150,77 @@ func TestConfig_ReloadConfig(t *testing.T) {
 	}
 
 }
+
+// Ensure mergo merges are done the way we expect.
+// This is needed to test for potential regressions, like:
+// - https://github.com/imdario/mergo/issues/187
+func TestConfig_MergoMerge(t *testing.T) {
+	configs := [][]byte{
+		[]byte(`
+listen:
+  port: 1234
+`),
+		[]byte(`
+firewall:
+  inbound:
+    - port: 443
+      proto: tcp
+      groups:
+        - server
+    - port: 443
+      proto: tcp
+      groups:
+        - webapp
+`),
+		[]byte(`
+listen:
+  host: 0.0.0.0
+  port: 4242
+firewall:
+  outbound:
+    - port: any
+      proto: any
+      host: any
+  inbound:
+    - port: any
+      proto: icmp
+      host: any
+`),
+	}
+
+	var m map[any]any
+
+	// merge the same way config.parse() merges
+	for _, b := range configs {
+		var nm map[any]any
+		err := yaml.Unmarshal(b, &nm)
+		require.NoError(t, err)
+
+		// We need to use WithAppendSlice so that firewall rules in separate
+		// files are appended together
+		err = mergo.Merge(&nm, m, mergo.WithAppendSlice)
+		m = nm
+		require.NoError(t, err)
+	}
+
+	t.Logf("Merged Config: %#v", m)
+	mYaml, err := yaml.Marshal(m)
+	require.NoError(t, err)
+	t.Logf("Merged Config as YAML:\n%s", mYaml)
+
+	// If a bug is present, some items might be replaced instead of merged like we expect
+	expected := map[any]any{
+		"firewall": map[any]any{
+			"inbound": []any{
+				map[any]any{"host": "any", "port": "any", "proto": "icmp"},
+				map[any]any{"groups": []any{"server"}, "port": 443, "proto": "tcp"},
+				map[any]any{"groups": []any{"webapp"}, "port": 443, "proto": "tcp"}},
+			"outbound": []any{
+				map[any]any{"host": "any", "port": "any", "proto": "any"}}},
+		"listen": map[any]any{
+			"host": "0.0.0.0",
+			"port": 4242,
+		},
+	}
+	assert.Equal(t, expected, m)
+}

connection_manager.go

@@ -1,258 +1,490 @@
 package nebula
 
 import (
+	"bytes"
+	"context"
 	"sync"
 	"time"
 
+	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 )
 
-// TODO: incount and outcount are intended as a shortcut to locking the mutexes for every single packet
-// and something like every 10 packets we could lock, send 10, then unlock for a moment
+type trafficDecision int
+
+const (
+	doNothing      trafficDecision = 0
+	deleteTunnel   trafficDecision = 1 // delete the hostinfo on our side, do not notify the remote
+	closeTunnel    trafficDecision = 2 // delete the hostinfo and notify the remote
+	swapPrimary    trafficDecision = 3
+	migrateRelays  trafficDecision = 4
+	tryRehandshake trafficDecision = 5
+)
 
 type connectionManager struct {
-	hostMap      *HostMap
-	in           map[uint32]struct{}
-	inLock       *sync.RWMutex
-	inCount      int
-	out          map[uint32]struct{}
-	outLock      *sync.RWMutex
-	outCount     int
-	TrafficTimer *SystemTimerWheel
-	intf         *Interface
+	in     map[uint32]struct{}
+	inLock *sync.RWMutex
 
-	pendingDeletion      map[uint32]int
-	pendingDeletionLock  *sync.RWMutex
-	pendingDeletionTimer *SystemTimerWheel
+	out     map[uint32]struct{}
+	outLock *sync.RWMutex
 
-	checkInterval           int
-	pendingDeletionInterval int
+	// relayUsed holds which relay localIndexs are in use
+	relayUsed     map[uint32]struct{}
+	relayUsedLock *sync.RWMutex
+
+	hostMap                 *HostMap
+	trafficTimer            *LockingTimerWheel[uint32]
+	intf                    *Interface
+	pendingDeletion         map[uint32]struct{}
+	punchy                  *Punchy
+	checkInterval           time.Duration
+	pendingDeletionInterval time.Duration
+	metricsTxPunchy         metrics.Counter
 
 	l *logrus.Logger
-	// I wanted to call one matLock
 }
 
-func newConnectionManager(l *logrus.Logger, intf *Interface, checkInterval, pendingDeletionInterval int) *connectionManager {
+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
+	}
+
 	nc := &connectionManager{
 		hostMap:                 intf.hostMap,
 		in:                      make(map[uint32]struct{}),
 		inLock:                  &sync.RWMutex{},
-		inCount:                 0,
 		out:                     make(map[uint32]struct{}),
 		outLock:                 &sync.RWMutex{},
-		outCount:                0,
-		TrafficTimer:            NewSystemTimerWheel(time.Millisecond*500, time.Second*60),
+		relayUsed:               make(map[uint32]struct{}),
+		relayUsedLock:           &sync.RWMutex{},
+		trafficTimer:            NewLockingTimerWheel[uint32](time.Millisecond*500, max),
 		intf:                    intf,
-		pendingDeletion:         make(map[uint32]int),
-		pendingDeletionLock:     &sync.RWMutex{},
-		pendingDeletionTimer:    NewSystemTimerWheel(time.Millisecond*500, time.Second*60),
+		pendingDeletion:         make(map[uint32]struct{}),
 		checkInterval:           checkInterval,
 		pendingDeletionInterval: pendingDeletionInterval,
+		punchy:                  punchy,
+		metricsTxPunchy:         metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
 		l:                       l,
 	}
-	nc.Start()
+
+	nc.Start(ctx)
 	return nc
 }
 
-func (n *connectionManager) In(ip uint32) {
+func (n *connectionManager) In(localIndex uint32) {
 	n.inLock.RLock()
 	// If this already exists, return
-	if _, ok := n.in[ip]; ok {
+	if _, ok := n.in[localIndex]; ok {
 		n.inLock.RUnlock()
 		return
 	}
 	n.inLock.RUnlock()
 	n.inLock.Lock()
-	n.in[ip] = struct{}{}
+	n.in[localIndex] = struct{}{}
 	n.inLock.Unlock()
 }
 
-func (n *connectionManager) Out(ip uint32) {
+func (n *connectionManager) Out(localIndex uint32) {
 	n.outLock.RLock()
 	// If this already exists, return
-	if _, ok := n.out[ip]; ok {
+	if _, ok := n.out[localIndex]; ok {
 		n.outLock.RUnlock()
 		return
 	}
 	n.outLock.RUnlock()
 	n.outLock.Lock()
-	// double check since we dropped the lock temporarily
-	if _, ok := n.out[ip]; ok {
+	n.out[localIndex] = struct{}{}
+	n.outLock.Unlock()
+}
+
+func (n *connectionManager) RelayUsed(localIndex uint32) {
+	n.relayUsedLock.RLock()
+	// If this already exists, return
+	if _, ok := n.relayUsed[localIndex]; ok {
+		n.relayUsedLock.RUnlock()
+		return
+	}
+	n.relayUsedLock.RUnlock()
+	n.relayUsedLock.Lock()
+	n.relayUsed[localIndex] = struct{}{}
+	n.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()
+	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
 	}
-	n.out[ip] = struct{}{}
-	n.AddTrafficWatch(ip, n.checkInterval)
+	n.out[localIndex] = struct{}{}
+	n.trafficTimer.Add(localIndex, n.checkInterval)
 	n.outLock.Unlock()
 }
 
-func (n *connectionManager) CheckIn(vpnIP uint32) bool {
-	n.inLock.RLock()
-	if _, ok := n.in[vpnIP]; ok {
-		n.inLock.RUnlock()
-		return true
-	}
-	n.inLock.RUnlock()
-	return false
+func (n *connectionManager) Start(ctx context.Context) {
+	go n.Run(ctx)
 }
 
-func (n *connectionManager) ClearIP(ip uint32) {
-	n.inLock.Lock()
-	n.outLock.Lock()
-	delete(n.in, ip)
-	delete(n.out, ip)
-	n.inLock.Unlock()
-	n.outLock.Unlock()
-}
+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)
+	defer clockSource.Stop()
 
-func (n *connectionManager) ClearPendingDeletion(ip uint32) {
-	n.pendingDeletionLock.Lock()
-	delete(n.pendingDeletion, ip)
-	n.pendingDeletionLock.Unlock()
-}
-
-func (n *connectionManager) AddPendingDeletion(ip uint32) {
-	n.pendingDeletionLock.Lock()
-	if _, ok := n.pendingDeletion[ip]; ok {
-		n.pendingDeletion[ip] += 1
-	} else {
-		n.pendingDeletion[ip] = 0
-	}
-	n.pendingDeletionTimer.Add(ip, time.Second*time.Duration(n.pendingDeletionInterval))
-	n.pendingDeletionLock.Unlock()
-}
-
-func (n *connectionManager) checkPendingDeletion(ip uint32) bool {
-	n.pendingDeletionLock.RLock()
-	if _, ok := n.pendingDeletion[ip]; ok {
-
-		n.pendingDeletionLock.RUnlock()
-		return true
-	}
-	n.pendingDeletionLock.RUnlock()
-	return false
-}
-
-func (n *connectionManager) AddTrafficWatch(vpnIP uint32, seconds int) {
-	n.TrafficTimer.Add(vpnIP, time.Second*time.Duration(seconds))
-}
-
-func (n *connectionManager) Start() {
-	go n.Run()
-}
-
-func (n *connectionManager) Run() {
-	clockSource := time.Tick(500 * time.Millisecond)
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
 
-	for now := range clockSource {
-		n.HandleMonitorTick(now, p, nb, out)
-		n.HandleDeletionTick(now)
+	for {
+		select {
+		case <-ctx.Done():
+			return
+
+		case now := <-clockSource.C:
+			n.trafficTimer.Advance(now)
+			for {
+				localIndex, has := n.trafficTimer.Purge()
+				if !has {
+					break
+				}
+
+				n.doTrafficCheck(localIndex, p, nb, out, now)
+			}
+		}
 	}
 }
 
-func (n *connectionManager) HandleMonitorTick(now time.Time, p, nb, out []byte) {
-	n.TrafficTimer.advance(now)
-	for {
-		ep := n.TrafficTimer.Purge()
-		if ep == nil {
-			break
+func (n *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
+	decision, hostinfo, primary := n.makeTrafficDecision(localIndex, p, nb, out, now)
+
+	switch decision {
+	case deleteTunnel:
+		if n.hostMap.DeleteHostInfo(hostinfo) {
+			// Only clearing the lighthouse cache if this is the last hostinfo for this vpn ip in the hostmap
+			n.intf.lightHouse.DeleteVpnIp(hostinfo.vpnIp)
 		}
 
-		vpnIP := ep.(uint32)
+	case closeTunnel:
+		n.intf.sendCloseTunnel(hostinfo)
+		n.intf.closeTunnel(hostinfo)
 
-		// Check for traffic coming back in from this host.
-		traf := n.CheckIn(vpnIP)
+	case swapPrimary:
+		n.swapPrimary(hostinfo, primary)
 
-		// If we saw incoming packets from this ip, just return
-		if traf {
-			if n.l.Level >= logrus.DebugLevel {
-				n.l.WithField("vpnIp", IntIp(vpnIP)).
-					WithField("tunnelCheck", m{"state": "alive", "method": "passive"}).
-					Debug("Tunnel status")
+	case migrateRelays:
+		n.migrateRelayUsed(hostinfo, primary)
+
+	case tryRehandshake:
+		n.tryRehandshake(hostinfo)
+	}
+
+	n.resetRelayTrafficCheck(hostinfo)
+}
+
+func (n *connectionManager) resetRelayTrafficCheck(hostinfo *HostInfo) {
+	if hostinfo != nil {
+		n.relayUsedLock.Lock()
+		defer n.relayUsedLock.Unlock()
+		// No need to migrate any relays, delete usage info now.
+		for _, idx := range hostinfo.relayState.CopyRelayForIdxs() {
+			delete(n.relayUsed, idx)
+		}
+	}
+}
+
+func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo) {
+	relayFor := oldhostinfo.relayState.CopyAllRelayFor()
+
+	for _, r := range relayFor {
+		existing, ok := newhostinfo.relayState.QueryRelayForByIp(r.PeerIp)
+
+		var index uint32
+		var relayFrom iputil.VpnIp
+		var relayTo iputil.VpnIp
+		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 = newhostinfo.vpnIp
+				relayTo = existing.PeerIp
+			case ForwardingType:
+				relayFrom = existing.PeerIp
+				relayTo = newhostinfo.vpnIp
+			default:
+				// should never happen
+			}
+		case !ok:
+			n.relayUsedLock.RLock()
+			if _, relayUsed := n.relayUsed[r.LocalIndex]; !relayUsed {
+				// The relay hasn't been used; don't migrate it.
+				n.relayUsedLock.RUnlock()
+				continue
+			}
+			n.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.PeerIp, nil, r.Type, Requested)
+			if err != nil {
+				n.l.WithError(err).Error("failed to migrate relay to new hostinfo")
+				continue
+			}
+			switch r.Type {
+			case TerminalType:
+				relayFrom = newhostinfo.vpnIp
+				relayTo = r.PeerIp
+			case ForwardingType:
+				relayFrom = r.PeerIp
+				relayTo = newhostinfo.vpnIp
+			default:
+				// should never happen
 			}
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
-			continue
 		}
 
-		// If we didn't we may need to probe or destroy the conn
-		hostinfo, err := n.hostMap.QueryVpnIP(vpnIP)
+		// Send a CreateRelayRequest to the peer.
+		req := NebulaControl{
+			Type:                NebulaControl_CreateRelayRequest,
+			InitiatorRelayIndex: index,
+			RelayFromIp:         uint32(relayFrom),
+			RelayToIp:           uint32(relayTo),
+		}
+		msg, err := req.Marshal()
 		if err != nil {
-			n.l.Debugf("Not found in hostmap: %s", IntIp(vpnIP))
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
-			continue
-		}
-
-		hostinfo.logger(n.l).
-			WithField("tunnelCheck", m{"state": "testing", "method": "active"}).
-			Debug("Tunnel status")
-
-		if hostinfo != nil && hostinfo.ConnectionState != nil {
-			// Send a test packet to trigger an authenticated tunnel test, this should suss out any lingering tunnel issues
-			n.intf.SendMessageToVpnIp(test, testRequest, vpnIP, p, nb, out)
-
+			n.l.WithError(err).Error("failed to marshal Control message to migrate relay")
 		} else {
-			hostinfo.logger(n.l).Debugf("Hostinfo sadness: %s", IntIp(vpnIP))
+			n.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
+			n.l.WithFields(logrus.Fields{
+				"relayFrom":           iputil.VpnIp(req.RelayFromIp),
+				"relayTo":             iputil.VpnIp(req.RelayToIp),
+				"initiatorRelayIndex": req.InitiatorRelayIndex,
+				"responderRelayIndex": req.ResponderRelayIndex,
+				"vpnIp":               newhostinfo.vpnIp}).
+				Info("send CreateRelayRequest")
 		}
-		n.AddPendingDeletion(vpnIP)
 	}
-
 }
 
-func (n *connectionManager) HandleDeletionTick(now time.Time) {
-	n.pendingDeletionTimer.advance(now)
-	for {
-		ep := n.pendingDeletionTimer.Purge()
-		if ep == nil {
-			break
-		}
+func (n *connectionManager) makeTrafficDecision(localIndex uint32, p, nb, out []byte, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
+	n.hostMap.RLock()
+	defer n.hostMap.RUnlock()
 
-		vpnIP := ep.(uint32)
+	hostinfo := n.hostMap.Indexes[localIndex]
+	if hostinfo == nil {
+		n.l.WithField("localIndex", localIndex).Debugf("Not found in hostmap")
+		delete(n.pendingDeletion, localIndex)
+		return doNothing, nil, nil
+	}
 
-		// If we saw incoming packets from this ip, just return
-		traf := n.CheckIn(vpnIP)
-		if traf {
-			n.l.WithField("vpnIp", IntIp(vpnIP)).
-				WithField("tunnelCheck", m{"state": "alive", "method": "active"}).
-				Debug("Tunnel status")
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
-			continue
-		}
+	if n.isInvalidCertificate(now, hostinfo) {
+		delete(n.pendingDeletion, hostinfo.localIndexId)
+		return closeTunnel, hostinfo, nil
+	}
 
-		hostinfo, err := n.hostMap.QueryVpnIP(vpnIP)
-		if err != nil {
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
-			n.l.Debugf("Not found in hostmap: %s", IntIp(vpnIP))
-			continue
-		}
+	primary := n.hostMap.Hosts[hostinfo.vpnIp]
+	mainHostInfo := true
+	if primary != nil && primary != hostinfo {
+		mainHostInfo = false
+	}
 
-		// If it comes around on deletion wheel and hasn't resolved itself, delete
-		if n.checkPendingDeletion(vpnIP) {
-			cn := ""
-			if hostinfo.ConnectionState != nil && hostinfo.ConnectionState.peerCert != nil {
-				cn = hostinfo.ConnectionState.peerCert.Details.Name
-			}
+	// Check for traffic on this hostinfo
+	inTraffic, outTraffic := n.getAndResetTrafficCheck(localIndex)
+
+	// A hostinfo is determined alive if there is incoming traffic
+	if inTraffic {
+		decision := doNothing
+		if n.l.Level >= logrus.DebugLevel {
 			hostinfo.logger(n.l).
-				WithField("tunnelCheck", m{"state": "dead", "method": "active"}).
-				WithField("certName", cn).
-				Info("Tunnel status")
+				WithField("tunnelCheck", m{"state": "alive", "method": "passive"}).
+				Debug("Tunnel status")
+		}
+		delete(n.pendingDeletion, hostinfo.localIndexId)
+
+		if mainHostInfo {
+			decision = tryRehandshake
 
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
-			// TODO: This is only here to let tests work. Should do proper mocking
-			if n.intf.lightHouse != nil {
-				n.intf.lightHouse.DeleteVpnIP(vpnIP)
-			}
-			n.hostMap.DeleteHostInfo(hostinfo)
 		} else {
-			n.ClearIP(vpnIP)
-			n.ClearPendingDeletion(vpnIP)
+			if n.shouldSwapPrimary(hostinfo, primary) {
+				decision = swapPrimary
+			} else {
+				// migrate the relays to the primary, if in use.
+				decision = migrateRelays
+			}
+		}
+
+		n.trafficTimer.Add(hostinfo.localIndexId, n.checkInterval)
+
+		if !outTraffic {
+			// Send a punch packet to keep the NAT state alive
+			n.sendPunch(hostinfo)
+		}
+
+		return decision, hostinfo, primary
+	}
+
+	if _, ok := n.pendingDeletion[hostinfo.localIndexId]; ok {
+		// We have already sent a test packet and nothing was returned, this hostinfo is dead
+		hostinfo.logger(n.l).
+			WithField("tunnelCheck", m{"state": "dead", "method": "active"}).
+			Info("Tunnel status")
+
+		delete(n.pendingDeletion, hostinfo.localIndexId)
+		return deleteTunnel, hostinfo, nil
+	}
+
+	if hostinfo != nil && hostinfo.ConnectionState != nil && mainHostInfo {
+		if !outTraffic {
+			// 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)
+			return doNothing, nil, nil
+
+		}
+
+		if n.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)
+		}
+
+		if n.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(n.l).
+				WithField("tunnelCheck", m{"state": "testing", "method": "active"}).
+				Debug("Tunnel status")
+		}
+
+		// Send a test packet to trigger an authenticated tunnel test, this should suss out any lingering tunnel issues
+		n.intf.SendMessageToHostInfo(header.Test, header.TestRequest, hostinfo, p, nb, out)
+
+	} else {
+		if n.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(n.l).Debugf("Hostinfo sadness")
+		}
+	}
+
+	n.pendingDeletion[hostinfo.localIndexId] = struct{}{}
+	n.trafficTimer.Add(hostinfo.localIndexId, n.pendingDeletionInterval)
+	return doNothing, nil, nil
+}
+
+func (n *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.
+
+	if current.vpnIp < n.intf.myVpnIp {
+		// Only one side should flip primary because if both flip then we may never resolve to a single tunnel.
+		// vpn ip is static across all tunnels for this host pair so lets use that to determine who is flipping.
+		// The remotes vpn ip is lower than mine. I will not flip.
+		return false
+	}
+
+	certState := n.intf.certState.Load()
+	return bytes.Equal(current.ConnectionState.certState.certificate.Signature, certState.certificate.Signature)
+}
+
+func (n *connectionManager) swapPrimary(current, primary *HostInfo) {
+	n.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.vpnIp] == primary {
+		n.hostMap.unlockedMakePrimary(current)
+	}
+	n.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 {
+	remoteCert := hostinfo.GetCert()
+	if remoteCert == nil {
+		return false
+	}
+
+	valid, err := remoteCert.VerifyWithCache(now, n.intf.caPool)
+	if valid {
+		return false
+	}
+
+	if !n.intf.disconnectInvalid && err != cert.ErrBlockListed {
+		// Block listed certificates should always be disconnected
+		return false
+	}
+
+	fingerprint, _ := remoteCert.Sha256Sum()
+	hostinfo.logger(n.l).WithError(err).
+		WithField("fingerprint", fingerprint).
+		Info("Remote certificate is no longer valid, tearing down the tunnel")
+
+	return true
+}
+
+func (n *connectionManager) sendPunch(hostinfo *HostInfo) {
+	if !n.punchy.GetPunch() {
+		// Punching is disabled
+		return
+	}
+
+	if n.punchy.GetTargetEverything() {
+		hostinfo.remotes.ForEach(n.hostMap.preferredRanges, func(addr *udp.Addr, preferred bool) {
+			n.metricsTxPunchy.Inc(1)
+			n.intf.outside.WriteTo([]byte{1}, addr)
+		})
+
+	} else if hostinfo.remote != nil {
+		n.metricsTxPunchy.Inc(1)
+		n.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
+	}
+}
+
+func (n *connectionManager) tryRehandshake(hostinfo *HostInfo) {
+	certState := n.intf.certState.Load()
+	if bytes.Equal(hostinfo.ConnectionState.certState.certificate.Signature, certState.certificate.Signature) {
+		return
+	}
+
+	n.l.WithField("vpnIp", hostinfo.vpnIp).
+		WithField("reason", "local certificate is not current").
+		Info("Re-handshaking with remote")
+
+	//TODO: this is copied from getOrHandshake to keep the extra checks out of the hot path, figure it out
+	newHostinfo := n.intf.handshakeManager.AddVpnIp(hostinfo.vpnIp, n.intf.initHostInfo)
+	if !newHostinfo.HandshakeReady {
+		ixHandshakeStage0(n.intf, newHostinfo.vpnIp, newHostinfo)
+	}
+
+	//If this is a static host, we don't need to wait for the HostQueryReply
+	//We can trigger the handshake right now
+	if _, ok := n.intf.lightHouse.GetStaticHostList()[hostinfo.vpnIp]; ok {
+		select {
+		case n.intf.handshakeManager.trigger <- hostinfo.vpnIp:
+		default:
 		}
 	}
 }

connection_manager_test.go

@@ -1,23 +1,44 @@
 package nebula
 
 import (
+	"context"
+	"crypto/ed25519"
+	"crypto/rand"
 	"net"
 	"testing"
 	"time"
 
 	"github.com/flynn/noise"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/test"
+	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
 )
 
-var vpnIP uint32
+var vpnIp iputil.VpnIp
+
+func newTestLighthouse() *LightHouse {
+	lh := &LightHouse{
+		l:       test.NewLogger(),
+		addrMap: map[iputil.VpnIp]*RemoteList{},
+	}
+	lighthouses := map[iputil.VpnIp]struct{}{}
+	staticList := map[iputil.VpnIp]struct{}{}
+
+	lh.lighthouses.Store(&lighthouses)
+	lh.staticList.Store(&staticList)
+
+	return lh
+}
 
 func Test_NewConnectionManagerTest(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	//_, tuncidr, _ := net.ParseCIDR("1.1.1.1/24")
 	_, vpncidr, _ := net.ParseCIDR("172.1.1.1/24")
 	_, localrange, _ := net.ParseCIDR("10.1.1.1/24")
-	vpnIP = ip2int(net.ParseIP("172.1.1.2"))
+	vpnIp = iputil.Ip2VpnIp(net.ParseIP("172.1.1.2"))
 	preferredRanges := []*net.IPNet{localrange}
 
 	// Very incomplete mock objects
@@ -29,59 +50,71 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 		rawCertificateNoKey: []byte{},
 	}
 
-	lh := NewLightHouse(l, false, &net.IPNet{IP: net.IP{0, 0, 0, 0}, Mask: net.IPMask{0, 0, 0, 0}}, []uint32{}, 1000, 0, &udpConn{}, false, 1, false)
+	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &Tun{},
-		outside:          &udpConn{},
-		certState:        cs,
+		inside:           &test.NoopTun{},
+		outside:          &udp.Conn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
-		handshakeManager: NewHandshakeManager(l, vpncidr, preferredRanges, hostMap, lh, &udpConn{}, defaultHandshakeConfig),
+		handshakeManager: NewHandshakeManager(l, vpncidr, preferredRanges, hostMap, lh, &udp.Conn{}, defaultHandshakeConfig),
 		l:                l,
 	}
-	now := time.Now()
+	ifce.certState.Store(cs)
 
 	// Create manager
-	nc := newConnectionManager(l, ifce, 5, 10)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
-	nc.HandleMonitorTick(now, p, nb, out)
+
 	// Add an ip we have established a connection w/ to hostmap
-	hostinfo := nc.hostMap.AddVpnIP(vpnIP)
+	hostinfo := &HostInfo{
+		vpnIp:         vpnIp,
+		localIndexId:  1099,
+		remoteIndexId: 9901,
+	}
 	hostinfo.ConnectionState = &ConnectionState{
 		certState: cs,
 		H:         &noise.HandshakeState{},
 	}
+	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
-	// We saw traffic out to vpnIP
-	nc.Out(vpnIP)
-	assert.NotContains(t, nc.pendingDeletion, vpnIP)
-	assert.Contains(t, nc.hostMap.Hosts, vpnIP)
-	// Move ahead 5s. Nothing should happen
-	next_tick := now.Add(5 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// Move ahead 6s. We haven't heard back
-	next_tick = now.Add(6 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// This host should now be up for deletion
-	assert.Contains(t, nc.pendingDeletion, vpnIP)
-	assert.Contains(t, nc.hostMap.Hosts, vpnIP)
-	// Move ahead some more
-	next_tick = now.Add(45 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// The host should be evicted
-	assert.NotContains(t, nc.pendingDeletion, vpnIP)
-	assert.NotContains(t, nc.hostMap.Hosts, vpnIP)
+	// We saw traffic out to vpnIp
+	nc.Out(hostinfo.localIndexId)
+	nc.In(hostinfo.localIndexId)
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnIp)
+	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.out, 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)
+
+	// Do another traffic check tick, this host should be pending deletion now
+	nc.Out(hostinfo.localIndexId)
+	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.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnIp)
+
+	// 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.vpnIp)
+	assert.NotContains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 }
 
 func Test_NewConnectionManagerTest2(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	//_, tuncidr, _ := net.ParseCIDR("1.1.1.1/24")
 	_, vpncidr, _ := net.ParseCIDR("172.1.1.1/24")
 	_, localrange, _ := net.ParseCIDR("10.1.1.1/24")
@@ -96,55 +129,163 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 		rawCertificateNoKey: []byte{},
 	}
 
-	lh := NewLightHouse(l, false, &net.IPNet{IP: net.IP{0, 0, 0, 0}, Mask: net.IPMask{0, 0, 0, 0}}, []uint32{}, 1000, 0, &udpConn{}, false, 1, false)
+	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &Tun{},
-		outside:          &udpConn{},
-		certState:        cs,
+		inside:           &test.NoopTun{},
+		outside:          &udp.Conn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
-		handshakeManager: NewHandshakeManager(l, vpncidr, preferredRanges, hostMap, lh, &udpConn{}, defaultHandshakeConfig),
+		handshakeManager: NewHandshakeManager(l, vpncidr, preferredRanges, hostMap, lh, &udp.Conn{}, defaultHandshakeConfig),
 		l:                l,
 	}
-	now := time.Now()
+	ifce.certState.Store(cs)
 
 	// Create manager
-	nc := newConnectionManager(l, ifce, 5, 10)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
-	nc.HandleMonitorTick(now, p, nb, out)
+
 	// Add an ip we have established a connection w/ to hostmap
-	hostinfo := nc.hostMap.AddVpnIP(vpnIP)
+	hostinfo := &HostInfo{
+		vpnIp:         vpnIp,
+		localIndexId:  1099,
+		remoteIndexId: 9901,
+	}
 	hostinfo.ConnectionState = &ConnectionState{
 		certState: cs,
 		H:         &noise.HandshakeState{},
 	}
+	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
-	// We saw traffic out to vpnIP
-	nc.Out(vpnIP)
-	assert.NotContains(t, nc.pendingDeletion, vpnIP)
-	assert.Contains(t, nc.hostMap.Hosts, vpnIP)
-	// Move ahead 5s. Nothing should happen
-	next_tick := now.Add(5 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// Move ahead 6s. We haven't heard back
-	next_tick = now.Add(6 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// This host should now be up for deletion
-	assert.Contains(t, nc.pendingDeletion, vpnIP)
-	assert.Contains(t, nc.hostMap.Hosts, vpnIP)
-	// We heard back this time
-	nc.In(vpnIP)
-	// Move ahead some more
-	next_tick = now.Add(45 * time.Second)
-	nc.HandleMonitorTick(next_tick, p, nb, out)
-	nc.HandleDeletionTick(next_tick)
-	// The host should be evicted
-	assert.NotContains(t, nc.pendingDeletion, vpnIP)
-	assert.Contains(t, nc.hostMap.Hosts, vpnIP)
+	// We saw traffic out to vpnIp
+	nc.Out(hostinfo.localIndexId)
+	nc.In(hostinfo.localIndexId)
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.vpnIp)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnIp)
+	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)
+
+	// Do another traffic check tick, this host should be pending deletion now
+	nc.Out(hostinfo.localIndexId)
+	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.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnIp)
+
+	// We saw traffic, should no longer be pending deletion
+	nc.In(hostinfo.localIndexId)
+	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.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
+	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnIp)
+}
+
+// Check if we can disconnect the peer.
+// Validate if the peer's certificate is invalid (expired, etc.)
+// Disconnect only if disconnectInvalid: true is set.
+func Test_NewConnectionManagerTest_DisconnectInvalid(t *testing.T) {
+	now := time.Now()
+	l := test.NewLogger()
+	ipNet := net.IPNet{
+		IP:   net.IPv4(172, 1, 1, 2),
+		Mask: net.IPMask{255, 255, 255, 0},
+	}
+	_, vpncidr, _ := net.ParseCIDR("172.1.1.1/24")
+	_, localrange, _ := net.ParseCIDR("10.1.1.1/24")
+	preferredRanges := []*net.IPNet{localrange}
+	hostMap := NewHostMap(l, "test", vpncidr, preferredRanges)
+
+	// Generate keys for CA and peer's cert.
+	pubCA, privCA, _ := ed25519.GenerateKey(rand.Reader)
+	caCert := cert.NebulaCertificate{
+		Details: cert.NebulaCertificateDetails{
+			Name:      "ca",
+			NotBefore: now,
+			NotAfter:  now.Add(1 * time.Hour),
+			IsCA:      true,
+			PublicKey: pubCA,
+		},
+	}
+	caCert.Sign(cert.Curve_CURVE25519, privCA)
+	ncp := &cert.NebulaCAPool{
+		CAs: cert.NewCAPool().CAs,
+	}
+	ncp.CAs["ca"] = &caCert
+
+	pubCrt, _, _ := ed25519.GenerateKey(rand.Reader)
+	peerCert := cert.NebulaCertificate{
+		Details: cert.NebulaCertificateDetails{
+			Name:      "host",
+			Ips:       []*net.IPNet{&ipNet},
+			Subnets:   []*net.IPNet{},
+			NotBefore: now,
+			NotAfter:  now.Add(60 * time.Second),
+			PublicKey: pubCrt,
+			IsCA:      false,
+			Issuer:    "ca",
+		},
+	}
+	peerCert.Sign(cert.Curve_CURVE25519, privCA)
+
+	cs := &CertState{
+		rawCertificate:      []byte{},
+		privateKey:          []byte{},
+		certificate:         &cert.NebulaCertificate{},
+		rawCertificateNoKey: []byte{},
+	}
+
+	lh := newTestLighthouse()
+	ifce := &Interface{
+		hostMap:           hostMap,
+		inside:            &test.NoopTun{},
+		outside:           &udp.Conn{},
+		firewall:          &Firewall{},
+		lightHouse:        lh,
+		handshakeManager:  NewHandshakeManager(l, vpncidr, preferredRanges, hostMap, lh, &udp.Conn{}, defaultHandshakeConfig),
+		l:                 l,
+		disconnectInvalid: true,
+		caPool:            ncp,
+	}
+	ifce.certState.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)
+	ifce.connectionManager = nc
+	hostinfo, _ := nc.hostMap.AddVpnIp(vpnIp, nil)
+	hostinfo.ConnectionState = &ConnectionState{
+		certState: cs,
+		peerCert:  &peerCert,
+		H:         &noise.HandshakeState{},
+	}
+
+	// Move ahead 45s.
+	// Check if to disconnect with invalid certificate.
+	// Should be alive.
+	nextTick := now.Add(45 * time.Second)
+	invalid := nc.isInvalidCertificate(nextTick, hostinfo)
+	assert.False(t, invalid)
+
+	// Move ahead 61s.
+	// Check if to disconnect with invalid certificate.
+	// Should be disconnected.
+	nextTick = now.Add(61 * time.Second)
+	invalid = nc.isInvalidCertificate(nextTick, hostinfo)
+	assert.True(t, invalid)
 }

connection_state.go

@@ -9,31 +9,43 @@ import (
 	"github.com/flynn/noise"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/noiseutil"
 )
 
 const ReplayWindow = 1024
 
 type ConnectionState struct {
-	eKey                 *NebulaCipherState
-	dKey                 *NebulaCipherState
-	H                    *noise.HandshakeState
-	certState            *CertState
-	peerCert             *cert.NebulaCertificate
-	initiator            bool
-	atomicMessageCounter uint64
-	window               *Bits
-	queueLock            sync.Mutex
-	writeLock            sync.Mutex
-	ready                bool
+	eKey           *NebulaCipherState
+	dKey           *NebulaCipherState
+	H              *noise.HandshakeState
+	certState      *CertState
+	peerCert       *cert.NebulaCertificate
+	initiator      bool
+	messageCounter atomic.Uint64
+	window         *Bits
+	queueLock      sync.Mutex
+	writeLock      sync.Mutex
+	ready          bool
 }
 
 func (f *Interface) newConnectionState(l *logrus.Logger, initiator bool, pattern noise.HandshakePattern, psk []byte, pskStage int) *ConnectionState {
-	cs := noise.NewCipherSuite(noise.DH25519, noise.CipherAESGCM, noise.HashSHA256)
+	var dhFunc noise.DHFunc
+	curCertState := f.certState.Load()
+
+	switch curCertState.certificate.Details.Curve {
+	case cert.Curve_CURVE25519:
+		dhFunc = noise.DH25519
+	case cert.Curve_P256:
+		dhFunc = noiseutil.DHP256
+	default:
+		l.Errorf("invalid curve: %s", curCertState.certificate.Details.Curve)
+		return nil
+	}
+	cs := noise.NewCipherSuite(dhFunc, noiseutil.CipherAESGCM, noise.HashSHA256)
 	if f.cipher == "chachapoly" {
-		cs = noise.NewCipherSuite(noise.DH25519, noise.CipherChaChaPoly, noise.HashSHA256)
+		cs = noise.NewCipherSuite(dhFunc, noise.CipherChaChaPoly, noise.HashSHA256)
 	}
 
-	curCertState := f.certState
 	static := noise.DHKey{Private: curCertState.privateKey, Public: curCertState.publicKey}
 
 	b := NewBits(ReplayWindow)
@@ -70,7 +82,7 @@ func (cs *ConnectionState) MarshalJSON() ([]byte, error) {
 	return json.Marshal(m{
 		"certificate":     cs.peerCert,
 		"initiator":       cs.initiator,
-		"message_counter": atomic.LoadUint64(&cs.atomicMessageCounter),
+		"message_counter": cs.messageCounter.Load(),
 		"ready":           cs.ready,
 	})
 }

control.go

@@ -1,14 +1,17 @@
 package nebula
 
 import (
+	"context"
 	"net"
 	"os"
 	"os/signal"
-	"sync/atomic"
 	"syscall"
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 )
 
 // Every interaction here needs to take extra care to copy memory and not return or use arguments "as is" when touching
@@ -17,20 +20,23 @@ import (
 type Control struct {
 	f          *Interface
 	l          *logrus.Logger
+	cancel     context.CancelFunc
 	sshStart   func()
 	statsStart func()
 	dnsStart   func()
 }
 
 type ControlHostInfo struct {
-	VpnIP          net.IP                  `json:"vpnIp"`
-	LocalIndex     uint32                  `json:"localIndex"`
-	RemoteIndex    uint32                  `json:"remoteIndex"`
-	RemoteAddrs    []*udpAddr              `json:"remoteAddrs"`
-	CachedPackets  int                     `json:"cachedPackets"`
-	Cert           *cert.NebulaCertificate `json:"cert"`
-	MessageCounter uint64                  `json:"messageCounter"`
-	CurrentRemote  *udpAddr                `json:"currentRemote"`
+	VpnIp                  net.IP                  `json:"vpnIp"`
+	LocalIndex             uint32                  `json:"localIndex"`
+	RemoteIndex            uint32                  `json:"remoteIndex"`
+	RemoteAddrs            []*udp.Addr             `json:"remoteAddrs"`
+	CachedPackets          int                     `json:"cachedPackets"`
+	Cert                   *cert.NebulaCertificate `json:"cert"`
+	MessageCounter         uint64                  `json:"messageCounter"`
+	CurrentRemote          *udp.Addr               `json:"currentRemote"`
+	CurrentRelaysToMe      []iputil.VpnIp          `json:"currentRelaysToMe"`
+	CurrentRelaysThroughMe []iputil.VpnIp          `json:"currentRelaysThroughMe"`
 }
 
 // Start actually runs nebula, this is a nonblocking call. To block use Control.ShutdownBlock()
@@ -55,14 +61,20 @@ func (c *Control) Start() {
 
 // Stop signals nebula to shutdown, returns after the shutdown is complete
 func (c *Control) Stop() {
-	//TODO: stop tun and udp routines, the lock on hostMap effectively does that though
+	// Stop the handshakeManager (and other services), to prevent new tunnels from
+	// being created while we're shutting them all down.
+	c.cancel()
+
 	c.CloseAllTunnels(false)
+	if err := c.f.Close(); err != nil {
+		c.l.WithError(err).Error("Close interface failed")
+	}
 	c.l.Info("Goodbye")
 }
 
 // ShutdownBlock will listen for and block on term and interrupt signals, calling Control.Stop() once signalled
 func (c *Control) ShutdownBlock() {
-	sigChan := make(chan os.Signal)
+	sigChan := make(chan os.Signal, 1)
 	signal.Notify(sigChan, syscall.SIGTERM)
 	signal.Notify(sigChan, syscall.SIGINT)
 
@@ -83,17 +95,26 @@ func (c *Control) RebindUDPServer() {
 	c.f.rebindCount++
 }
 
-// ListHostmap returns details about the actual or pending (handshaking) hostmap
-func (c *Control) ListHostmap(pendingMap bool) []ControlHostInfo {
+// ListHostmapHosts returns details about the actual or pending (handshaking) hostmap by vpn ip
+func (c *Control) ListHostmapHosts(pendingMap bool) []ControlHostInfo {
 	if pendingMap {
-		return listHostMap(c.f.handshakeManager.pendingHostMap)
+		return listHostMapHosts(c.f.handshakeManager.pendingHostMap)
 	} else {
-		return listHostMap(c.f.hostMap)
+		return listHostMapHosts(c.f.hostMap)
 	}
 }
 
-// GetHostInfoByVpnIP returns a single tunnels hostInfo, or nil if not found
-func (c *Control) GetHostInfoByVpnIP(vpnIP uint32, pending bool) *ControlHostInfo {
+// ListHostmapIndexes returns details about the actual or pending (handshaking) hostmap by local index id
+func (c *Control) ListHostmapIndexes(pendingMap bool) []ControlHostInfo {
+	if pendingMap {
+		return listHostMapIndexes(c.f.handshakeManager.pendingHostMap)
+	} else {
+		return listHostMapIndexes(c.f.hostMap)
+	}
+}
+
+// GetHostInfoByVpnIp returns a single tunnels hostInfo, or nil if not found
+func (c *Control) GetHostInfoByVpnIp(vpnIp iputil.VpnIp, pending bool) *ControlHostInfo {
 	var hm *HostMap
 	if pending {
 		hm = c.f.handshakeManager.pendingHostMap
@@ -101,7 +122,7 @@ func (c *Control) GetHostInfoByVpnIP(vpnIP uint32, pending bool) *ControlHostInf
 		hm = c.f.hostMap
 	}
 
-	h, err := hm.QueryVpnIP(vpnIP)
+	h, err := hm.QueryVpnIp(vpnIp)
 	if err != nil {
 		return nil
 	}
@@ -111,8 +132,8 @@ func (c *Control) GetHostInfoByVpnIP(vpnIP uint32, pending bool) *ControlHostInf
 }
 
 // SetRemoteForTunnel forces a tunnel to use a specific remote
-func (c *Control) SetRemoteForTunnel(vpnIP uint32, addr udpAddr) *ControlHostInfo {
-	hostInfo, err := c.f.hostMap.QueryVpnIP(vpnIP)
+func (c *Control) SetRemoteForTunnel(vpnIp iputil.VpnIp, addr udp.Addr) *ControlHostInfo {
+	hostInfo, err := c.f.hostMap.QueryVpnIp(vpnIp)
 	if err != nil {
 		return nil
 	}
@@ -123,26 +144,25 @@ func (c *Control) SetRemoteForTunnel(vpnIP uint32, addr udpAddr) *ControlHostInf
 }
 
 // CloseTunnel closes a fully established tunnel. If localOnly is false it will notify the remote end as well.
-func (c *Control) CloseTunnel(vpnIP uint32, localOnly bool) bool {
-	hostInfo, err := c.f.hostMap.QueryVpnIP(vpnIP)
+func (c *Control) CloseTunnel(vpnIp iputil.VpnIp, localOnly bool) bool {
+	hostInfo, err := c.f.hostMap.QueryVpnIp(vpnIp)
 	if err != nil {
 		return false
 	}
 
 	if !localOnly {
 		c.f.send(
-			closeTunnel,
+			header.CloseTunnel,
 			0,
 			hostInfo.ConnectionState,
 			hostInfo,
-			hostInfo.remote,
 			[]byte{},
 			make([]byte, 12, 12),
 			make([]byte, mtu),
 		)
 	}
 
-	c.f.closeTunnel(hostInfo, false)
+	c.f.closeTunnel(hostInfo)
 	return true
 }
 
@@ -150,38 +170,64 @@ func (c *Control) CloseTunnel(vpnIP uint32, localOnly bool) bool {
 // the int returned is a count of tunnels closed
 func (c *Control) CloseAllTunnels(excludeLighthouses bool) (closed int) {
 	//TODO: this is probably better as a function in ConnectionManager or HostMap directly
-	c.f.hostMap.Lock()
-	for _, h := range c.f.hostMap.Hosts {
+	lighthouses := c.f.lightHouse.GetLighthouses()
+
+	shutdown := func(h *HostInfo) {
 		if excludeLighthouses {
-			if _, ok := c.f.lightHouse.lighthouses[h.hostId]; ok {
-				continue
+			if _, ok := lighthouses[h.vpnIp]; ok {
+				return
 			}
 		}
+		c.f.send(header.CloseTunnel, 0, h.ConnectionState, h, []byte{}, make([]byte, 12, 12), make([]byte, mtu))
+		c.f.closeTunnel(h)
 
-		if h.ConnectionState.ready {
-			c.f.send(closeTunnel, 0, h.ConnectionState, h, h.remote, []byte{}, make([]byte, 12, 12), make([]byte, mtu))
-			c.f.closeTunnel(h, true)
+		c.l.WithField("vpnIp", h.vpnIp).WithField("udpAddr", h.remote).
+			Debug("Sending close tunnel message")
+		closed++
+	}
 
-			c.l.WithField("vpnIp", IntIp(h.hostId)).WithField("udpAddr", h.remote).
-				Debug("Sending close tunnel message")
-			closed++
+	// Learn which hosts are being used as relays, so we can shut them down last.
+	relayingHosts := map[iputil.VpnIp]*HostInfo{}
+	// Grab the hostMap lock to access the Relays map
+	c.f.hostMap.Lock()
+	for _, relayingHost := range c.f.hostMap.Relays {
+		relayingHosts[relayingHost.vpnIp] = relayingHost
+	}
+	c.f.hostMap.Unlock()
+
+	hostInfos := []*HostInfo{}
+	// Grab the hostMap lock to access the Hosts map
+	c.f.hostMap.Lock()
+	for _, relayHost := range c.f.hostMap.Indexes {
+		if _, ok := relayingHosts[relayHost.vpnIp]; !ok {
+			hostInfos = append(hostInfos, relayHost)
 		}
 	}
 	c.f.hostMap.Unlock()
+
+	for _, h := range hostInfos {
+		shutdown(h)
+	}
+	for _, h := range relayingHosts {
+		shutdown(h)
+	}
 	return
 }
 
 func copyHostInfo(h *HostInfo, preferredRanges []*net.IPNet) ControlHostInfo {
+
 	chi := ControlHostInfo{
-		VpnIP:         int2ip(h.hostId),
-		LocalIndex:    h.localIndexId,
-		RemoteIndex:   h.remoteIndexId,
-		RemoteAddrs:   h.remotes.CopyAddrs(preferredRanges),
-		CachedPackets: len(h.packetStore),
+		VpnIp:                  h.vpnIp.ToIP(),
+		LocalIndex:             h.localIndexId,
+		RemoteIndex:            h.remoteIndexId,
+		RemoteAddrs:            h.remotes.CopyAddrs(preferredRanges),
+		CachedPackets:          len(h.packetStore),
+		CurrentRelaysToMe:      h.relayState.CopyRelayIps(),
+		CurrentRelaysThroughMe: h.relayState.CopyRelayForIps(),
 	}
 
 	if h.ConnectionState != nil {
-		chi.MessageCounter = atomic.LoadUint64(&h.ConnectionState.atomicMessageCounter)
+		chi.MessageCounter = h.ConnectionState.messageCounter.Load()
 	}
 
 	if c := h.GetCert(); c != nil {
@@ -195,7 +241,7 @@ func copyHostInfo(h *HostInfo, preferredRanges []*net.IPNet) ControlHostInfo {
 	return chi
 }
 
-func listHostMap(hm *HostMap) []ControlHostInfo {
+func listHostMapHosts(hm *HostMap) []ControlHostInfo {
 	hm.RLock()
 	hosts := make([]ControlHostInfo, len(hm.Hosts))
 	i := 0
@@ -207,3 +253,16 @@ func listHostMap(hm *HostMap) []ControlHostInfo {
 
 	return hosts
 }
+
+func listHostMapIndexes(hm *HostMap) []ControlHostInfo {
+	hm.RLock()
+	hosts := make([]ControlHostInfo, len(hm.Indexes))
+	i := 0
+	for _, v := range hm.Indexes {
+		hosts[i] = copyHostInfo(v, hm.preferredRanges)
+		i++
+	}
+	hm.RUnlock()
+
+	return hosts
+}

control_test.go

@@ -8,17 +8,19 @@ import (
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/util"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/test"
+	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
 )
 
-func TestControl_GetHostInfoByVpnIP(t *testing.T) {
-	l := NewTestLogger()
+func TestControl_GetHostInfoByVpnIp(t *testing.T) {
+	l := test.NewLogger()
 	// Special care must be taken to re-use all objects provided to the hostmap and certificate in the expectedInfo object
 	// To properly ensure we are not exposing core memory to the caller
 	hm := NewHostMap(l, "test", &net.IPNet{}, make([]*net.IPNet, 0))
-	remote1 := NewUDPAddr(int2ip(100), 4444)
-	remote2 := NewUDPAddr(net.ParseIP("1:2:3:4:5:6:7:8"), 4444)
+	remote1 := udp.NewAddr(net.ParseIP("0.0.0.100"), 4444)
+	remote2 := udp.NewAddr(net.ParseIP("1:2:3:4:5:6:7:8"), 4444)
 	ipNet := net.IPNet{
 		IP:   net.IPv4(1, 2, 3, 4),
 		Mask: net.IPMask{255, 255, 255, 0},
@@ -45,10 +47,10 @@ func TestControl_GetHostInfoByVpnIP(t *testing.T) {
 		Signature: []byte{1, 2, 1, 2, 1, 3},
 	}
 
-	remotes := NewRemoteList()
+	remotes := NewRemoteList(nil)
 	remotes.unlockedPrependV4(0, NewIp4AndPort(remote1.IP, uint32(remote1.Port)))
 	remotes.unlockedPrependV6(0, NewIp6AndPort(remote2.IP, uint32(remote2.Port)))
-	hm.Add(ip2int(ipNet.IP), &HostInfo{
+	hm.Add(iputil.Ip2VpnIp(ipNet.IP), &HostInfo{
 		remote:  remote1,
 		remotes: remotes,
 		ConnectionState: &ConnectionState{
@@ -56,10 +58,15 @@ func TestControl_GetHostInfoByVpnIP(t *testing.T) {
 		},
 		remoteIndexId: 200,
 		localIndexId:  201,
-		hostId:        ip2int(ipNet.IP),
+		vpnIp:         iputil.Ip2VpnIp(ipNet.IP),
+		relayState: RelayState{
+			relays:        map[iputil.VpnIp]struct{}{},
+			relayForByIp:  map[iputil.VpnIp]*Relay{},
+			relayForByIdx: map[uint32]*Relay{},
+		},
 	})
 
-	hm.Add(ip2int(ipNet2.IP), &HostInfo{
+	hm.Add(iputil.Ip2VpnIp(ipNet2.IP), &HostInfo{
 		remote:  remote1,
 		remotes: remotes,
 		ConnectionState: &ConnectionState{
@@ -67,7 +74,12 @@ func TestControl_GetHostInfoByVpnIP(t *testing.T) {
 		},
 		remoteIndexId: 200,
 		localIndexId:  201,
-		hostId:        ip2int(ipNet2.IP),
+		vpnIp:         iputil.Ip2VpnIp(ipNet2.IP),
+		relayState: RelayState{
+			relays:        map[iputil.VpnIp]struct{}{},
+			relayForByIp:  map[iputil.VpnIp]*Relay{},
+			relayForByIdx: map[uint32]*Relay{},
+		},
 	})
 
 	c := Control{
@@ -77,26 +89,28 @@ func TestControl_GetHostInfoByVpnIP(t *testing.T) {
 		l: logrus.New(),
 	}
 
-	thi := c.GetHostInfoByVpnIP(ip2int(ipNet.IP), false)
+	thi := c.GetHostInfoByVpnIp(iputil.Ip2VpnIp(ipNet.IP), false)
 
 	expectedInfo := ControlHostInfo{
-		VpnIP:          net.IPv4(1, 2, 3, 4).To4(),
-		LocalIndex:     201,
-		RemoteIndex:    200,
-		RemoteAddrs:    []*udpAddr{remote2, remote1},
-		CachedPackets:  0,
-		Cert:           crt.Copy(),
-		MessageCounter: 0,
-		CurrentRemote:  NewUDPAddr(int2ip(100), 4444),
+		VpnIp:                  net.IPv4(1, 2, 3, 4).To4(),
+		LocalIndex:             201,
+		RemoteIndex:            200,
+		RemoteAddrs:            []*udp.Addr{remote2, remote1},
+		CachedPackets:          0,
+		Cert:                   crt.Copy(),
+		MessageCounter:         0,
+		CurrentRemote:          udp.NewAddr(net.ParseIP("0.0.0.100"), 4444),
+		CurrentRelaysToMe:      []iputil.VpnIp{},
+		CurrentRelaysThroughMe: []iputil.VpnIp{},
 	}
 
 	// Make sure we don't have any unexpected fields
-	assertFields(t, []string{"VpnIP", "LocalIndex", "RemoteIndex", "RemoteAddrs", "CachedPackets", "Cert", "MessageCounter", "CurrentRemote"}, thi)
-	util.AssertDeepCopyEqual(t, &expectedInfo, thi)
+	assertFields(t, []string{"VpnIp", "LocalIndex", "RemoteIndex", "RemoteAddrs", "CachedPackets", "Cert", "MessageCounter", "CurrentRemote", "CurrentRelaysToMe", "CurrentRelaysThroughMe"}, thi)
+	test.AssertDeepCopyEqual(t, &expectedInfo, thi)
 
 	// Make sure we don't panic if the host info doesn't have a cert yet
 	assert.NotPanics(t, func() {
-		thi = c.GetHostInfoByVpnIP(ip2int(ipNet2.IP), false)
+		thi = c.GetHostInfoByVpnIp(iputil.Ip2VpnIp(ipNet2.IP), false)
 	})
 }
 

control_tester.go

@@ -1,3 +1,4 @@
+//go:build e2e_testing
 // +build e2e_testing
 
 package nebula
@@ -5,14 +6,20 @@ package nebula
 import (
 	"net"
 
+	"github.com/slackhq/nebula/cert"
+
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/overlay"
+	"github.com/slackhq/nebula/udp"
 )
 
-// WaitForTypeByIndex will pipe all messages from this control device into the pipeTo control device
+// WaitForType will pipe all messages from this control device into the pipeTo control device
 // returning after a message matching the criteria has been piped
-func (c *Control) WaitForType(msgType NebulaMessageType, subType NebulaMessageSubType, pipeTo *Control) {
-	h := &Header{}
+func (c *Control) WaitForType(msgType header.MessageType, subType header.MessageSubType, pipeTo *Control) {
+	h := &header.H{}
 	for {
 		p := c.f.outside.Get(true)
 		if err := h.Parse(p.Data); err != nil {
@@ -27,8 +34,8 @@ func (c *Control) WaitForType(msgType NebulaMessageType, subType NebulaMessageSu
 
 // WaitForTypeByIndex is similar to WaitForType except it adds an index check
 // Useful if you have many nodes communicating and want to wait to find a specific nodes packet
-func (c *Control) WaitForTypeByIndex(toIndex uint32, msgType NebulaMessageType, subType NebulaMessageSubType, pipeTo *Control) {
-	h := &Header{}
+func (c *Control) WaitForTypeByIndex(toIndex uint32, msgType header.MessageType, subType header.MessageSubType, pipeTo *Control) {
+	h := &header.H{}
 	for {
 		p := c.f.outside.Get(true)
 		if err := h.Parse(p.Data); err != nil {
@@ -45,12 +52,12 @@ func (c *Control) WaitForTypeByIndex(toIndex uint32, msgType NebulaMessageType,
 // This is necessary if you did not configure static hosts or are not running a lighthouse
 func (c *Control) InjectLightHouseAddr(vpnIp net.IP, toAddr *net.UDPAddr) {
 	c.f.lightHouse.Lock()
-	remoteList := c.f.lightHouse.unlockedGetRemoteList(ip2int(vpnIp))
+	remoteList := c.f.lightHouse.unlockedGetRemoteList(iputil.Ip2VpnIp(vpnIp))
 	remoteList.Lock()
 	defer remoteList.Unlock()
 	c.f.lightHouse.Unlock()
 
-	iVpnIp := ip2int(vpnIp)
+	iVpnIp := iputil.Ip2VpnIp(vpnIp)
 	if v4 := toAddr.IP.To4(); v4 != nil {
 		remoteList.unlockedPrependV4(iVpnIp, NewIp4AndPort(v4, uint32(toAddr.Port)))
 	} else {
@@ -58,26 +65,44 @@ func (c *Control) InjectLightHouseAddr(vpnIp net.IP, toAddr *net.UDPAddr) {
 	}
 }
 
+// InjectRelays will push relayVpnIps into the local lighthouse cache for the vpnIp
+// This is necessary to inform an initiator of possible relays for communicating with a responder
+func (c *Control) InjectRelays(vpnIp net.IP, relayVpnIps []net.IP) {
+	c.f.lightHouse.Lock()
+	remoteList := c.f.lightHouse.unlockedGetRemoteList(iputil.Ip2VpnIp(vpnIp))
+	remoteList.Lock()
+	defer remoteList.Unlock()
+	c.f.lightHouse.Unlock()
+
+	iVpnIp := iputil.Ip2VpnIp(vpnIp)
+	uVpnIp := []uint32{}
+	for _, rVPnIp := range relayVpnIps {
+		uVpnIp = append(uVpnIp, uint32(iputil.Ip2VpnIp(rVPnIp)))
+	}
+
+	remoteList.unlockedSetRelay(iVpnIp, iVpnIp, uVpnIp)
+}
+
 // GetFromTun will pull a packet off the tun side of nebula
 func (c *Control) GetFromTun(block bool) []byte {
-	return c.f.inside.(*Tun).Get(block)
+	return c.f.inside.(*overlay.TestTun).Get(block)
 }
 
 // GetFromUDP will pull a udp packet off the udp side of nebula
-func (c *Control) GetFromUDP(block bool) *UdpPacket {
+func (c *Control) GetFromUDP(block bool) *udp.Packet {
 	return c.f.outside.Get(block)
 }
 
-func (c *Control) GetUDPTxChan() <-chan *UdpPacket {
-	return c.f.outside.txPackets
+func (c *Control) GetUDPTxChan() <-chan *udp.Packet {
+	return c.f.outside.TxPackets
 }
 
 func (c *Control) GetTunTxChan() <-chan []byte {
-	return c.f.inside.(*Tun).txPackets
+	return c.f.inside.(*overlay.TestTun).TxPackets
 }
 
 // InjectUDPPacket will inject a packet into the udp side of nebula
-func (c *Control) InjectUDPPacket(p *UdpPacket) {
+func (c *Control) InjectUDPPacket(p *udp.Packet) {
 	c.f.outside.Send(p)
 }
 
@@ -87,7 +112,7 @@ func (c *Control) InjectTunUDPPacket(toIp net.IP, toPort uint16, fromPort uint16
 		Version:  4,
 		TTL:      64,
 		Protocol: layers.IPProtocolUDP,
-		SrcIP:    c.f.inside.CidrNet().IP,
+		SrcIP:    c.f.inside.Cidr().IP,
 		DstIP:    toIp,
 	}
 
@@ -110,15 +135,19 @@ func (c *Control) InjectTunUDPPacket(toIp net.IP, toPort uint16, fromPort uint16
 		panic(err)
 	}
 
-	c.f.inside.(*Tun).Send(buffer.Bytes())
+	c.f.inside.(*overlay.TestTun).Send(buffer.Bytes())
+}
+
+func (c *Control) GetVpnIp() iputil.VpnIp {
+	return c.f.myVpnIp
 }
 
 func (c *Control) GetUDPAddr() string {
-	return c.f.outside.addr.String()
+	return c.f.outside.Addr.String()
 }
 
 func (c *Control) KillPendingTunnel(vpnIp net.IP) bool {
-	hostinfo, ok := c.f.handshakeManager.pendingHostMap.Hosts[ip2int(vpnIp)]
+	hostinfo, ok := c.f.handshakeManager.pendingHostMap.Hosts[iputil.Ip2VpnIp(vpnIp)]
 	if !ok {
 		return false
 	}
@@ -126,3 +155,25 @@ func (c *Control) KillPendingTunnel(vpnIp net.IP) bool {
 	c.f.handshakeManager.pendingHostMap.DeleteHostInfo(hostinfo)
 	return true
 }
+
+func (c *Control) GetHostmap() *HostMap {
+	return c.f.hostMap
+}
+
+func (c *Control) GetCert() *cert.NebulaCertificate {
+	return c.f.certState.Load().certificate
+}
+
+func (c *Control) ReHandshake(vpnIp iputil.VpnIp) {
+	hostinfo := c.f.handshakeManager.AddVpnIp(vpnIp, c.f.initHostInfo)
+	ixHandshakeStage0(c.f, vpnIp, hostinfo)
+
+	// If this is a static host, we don't need to wait for the HostQueryReply
+	// We can trigger the handshake right now
+	if _, ok := c.f.lightHouse.GetStaticHostList()[hostinfo.vpnIp]; ok {
+		select {
+		case c.f.handshakeManager.trigger <- hostinfo.vpnIp:
+		default:
+		}
+	}
+}

dist/arch/nebula.service

@@ -1,6 +1,6 @@
 [Unit]
-Description=nebula
-Wants=basic.target network-online.target
+Description=Nebula overlay networking tool
+Wants=basic.target network-online.target nss-lookup.target time-sync.target
 After=basic.target network.target network-online.target
 
 [Service]

dist/fedora/nebula.service

@@ -1,15 +1,14 @@
 [Unit]
 Description=Nebula overlay networking tool
-
+Wants=basic.target network-online.target nss-lookup.target time-sync.target
 After=basic.target network.target network-online.target
 Before=sshd.service
-Wants=basic.target network-online.target
 
 [Service]
+SyslogIdentifier=nebula
 ExecReload=/bin/kill -HUP $MAINPID
 ExecStart=/usr/bin/nebula -config /etc/nebula/config.yml
 Restart=always
-SyslogIdentifier=nebula
 
 [Install]
 WantedBy=multi-user.target

dist/windows/wintun/LICENSE.txt

@@ -0,0 +1,84 @@
+Prebuilt Binaries License
+-------------------------
+
+1. DEFINITIONS. "Software" means the precise contents of the "wintun.dll"
+   files that are included in the .zip file that contains this document as
+   downloaded from wintun.net/builds.
+
+2. LICENSE GRANT. WireGuard LLC grants to you a non-exclusive and
+   non-transferable right to use Software for lawful purposes under certain
+   obligations and limited rights as set forth in this agreement.
+
+3. RESTRICTIONS. Software is owned and copyrighted by WireGuard LLC. It is
+   licensed, not sold. Title to Software and all associated intellectual
+   property rights are retained by WireGuard. You must not:
+   a. reverse engineer, decompile, disassemble, extract from, or otherwise
+      modify the Software;
+   b. modify or create derivative work based upon Software in whole or in
+      parts, except insofar as only the API interfaces of the "wintun.h" file
+      distributed alongside the Software (the "Permitted API") are used;
+   c. remove any proprietary notices, labels, or copyrights from the Software;
+   d. resell, redistribute, lease, rent, transfer, sublicense, or otherwise
+      transfer rights of the Software without the prior written consent of
+      WireGuard LLC, except insofar as the Software is distributed alongside
+      other software that uses the Software only via the Permitted API;
+   e. use the name of WireGuard LLC, the WireGuard project, the Wintun
+      project, or the names of its contributors to endorse or promote products
+      derived from the Software without specific prior written consent.
+
+4. LIMITED WARRANTY. THE SOFTWARE IS PROVIDED "AS IS" AND WITHOUT WARRANTY OF
+   ANY KIND. WIREGUARD LLC HEREBY EXCLUDES AND DISCLAIMS ALL IMPLIED OR
+   STATUTORY WARRANTIES, INCLUDING ANY WARRANTIES OF MERCHANTABILITY, FITNESS
+   FOR A PARTICULAR PURPOSE, QUALITY, NON-INFRINGEMENT, TITLE, RESULTS,
+   EFFORTS, OR QUIET ENJOYMENT. THERE IS NO WARRANTY THAT THE PRODUCT WILL BE
+   ERROR-FREE OR WILL FUNCTION WITHOUT INTERRUPTION. YOU ASSUME THE ENTIRE
+   RISK FOR THE RESULTS OBTAINED USING THE PRODUCT. TO THE EXTENT THAT
+   WIREGUARD LLC MAY NOT DISCLAIM ANY WARRANTY AS A MATTER OF APPLICABLE LAW,
+   THE SCOPE AND DURATION OF SUCH WARRANTY WILL BE THE MINIMUM PERMITTED UNDER
+   SUCH LAW. ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND
+   WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR
+   A PARTICULAR PURPOSE OR NON-INFRINGEMENT ARE DISCLAIMED, EXCEPT TO THE
+   EXTENT THAT THESE DISCLAIMERS ARE HELD TO BE LEGALLY INVALID.
+
+5. LIMITATION OF LIABILITY. To the extent not prohibited by law, in no event
+   WireGuard LLC or any third-party-developer will be liable for any lost
+   revenue, profit or data or for special, indirect, consequential, incidental
+   or punitive damages, however caused regardless of the theory of liability,
+   arising out of or related to the use of or inability to use Software, even
+   if WireGuard LLC has been advised of the possibility of such damages.
+   Solely you are responsible for determining the appropriateness of using
+   Software and accept full responsibility for all risks associated with its
+   exercise of rights under this agreement, including but not limited to the
+   risks and costs of program errors, compliance with applicable laws, damage
+   to or loss of data, programs or equipment, and unavailability or
+   interruption of operations. The foregoing limitations will apply even if
+   the above stated warranty fails of its essential purpose. You acknowledge,
+   that it is in the nature of software that software is complex and not
+   completely free of errors. In no event shall WireGuard LLC or any
+   third-party-developer be liable to you under any theory for any damages
+   suffered by you or any user of Software or for any special, incidental,
+   indirect, consequential or similar damages (including without limitation
+   damages for loss of business profits, business interruption, loss of
+   business information or any other pecuniary loss) arising out of the use or
+   inability to use Software, even if WireGuard LLC has been advised of the
+   possibility of such damages and regardless of the legal or quitable theory
+   (contract, tort, or otherwise) upon which the claim is based.
+
+6. TERMINATION. This agreement is affected until terminated. You may
+   terminate this agreement at any time. This agreement will terminate
+   immediately without notice from WireGuard LLC if you fail to comply with
+   the terms and conditions of this agreement. Upon termination, you must
+   delete Software and all copies of Software and cease all forms of
+   distribution of Software.
+
+7. SEVERABILITY. If any provision of this agreement is held to be
+   unenforceable, this agreement will remain in effect with the provision
+   omitted, unless omission would frustrate the intent of the parties, in
+   which case this agreement will immediately terminate.
+
+8. RESERVATION OF RIGHTS. All rights not expressly granted in this agreement
+   are reserved by WireGuard LLC. For example, WireGuard LLC reserves the
+   right at any time to cease development of Software, to alter distribution
+   details, features, specifications, capabilities, functions, licensing
+   terms, release dates, APIs, ABIs, general availability, or other
+   characteristics of the Software.

dist/windows/wintun/README.md

@@ -0,0 +1,339 @@
+# [Wintun Network Adapter](https://www.wintun.net/)
+### TUN Device Driver for Windows
+
+This is a layer 3 TUN driver for Windows 7, 8, 8.1, and 10. Originally created for [WireGuard](https://www.wireguard.com/), it is intended to be useful to a wide variety of projects that require layer 3 tunneling devices with implementations primarily in userspace.
+
+## Installation
+
+Wintun is deployed as a platform-specific `wintun.dll` file. Install the `wintun.dll` file side-by-side with your application. Download the dll from [wintun.net](https://www.wintun.net/), alongside the header file for your application described below.
+
+## Usage
+
+Include the [`wintun.h` file](https://git.zx2c4.com/wintun/tree/api/wintun.h) in your project simply by copying it there and dynamically load the `wintun.dll` using [`LoadLibraryEx()`](https://docs.microsoft.com/en-us/windows/win32/api/libloaderapi/nf-libloaderapi-loadlibraryexa) and [`GetProcAddress()`](https://docs.microsoft.com/en-us/windows/win32/api/libloaderapi/nf-libloaderapi-getprocaddress) to resolve each function, using the typedefs provided in the header file. The [`InitializeWintun` function in the example.c code](https://git.zx2c4.com/wintun/tree/example/example.c) provides this in a function that you can simply copy and paste.
+
+With the library setup, Wintun can then be used by first creating an adapter, configuring it, and then setting its status to "up". Adapters have names (e.g. "OfficeNet") and types (e.g. "Wintun").
+
+```C
+WINTUN_ADAPTER_HANDLE Adapter1 = WintunCreateAdapter(L"OfficeNet", L"Wintun", &SomeFixedGUID1);
+WINTUN_ADAPTER_HANDLE Adapter2 = WintunCreateAdapter(L"HomeNet", L"Wintun", &SomeFixedGUID2);
+WINTUN_ADAPTER_HANDLE Adapter3 = WintunCreateAdapter(L"Data Center", L"Wintun", &SomeFixedGUID3);
+```
+
+After creating an adapter, we can use it by starting a session:
+
+```C
+WINTUN_SESSION_HANDLE Session = WintunStartSession(Adapter2, 0x400000);
+```
+
+Then, the `WintunAllocateSendPacket` and `WintunSendPacket` functions can be used for sending packets ([used by `SendPackets` in the example.c code](https://git.zx2c4.com/wintun/tree/example/example.c)):
+
+```C
+BYTE *OutgoingPacket = WintunAllocateSendPacket(Session, PacketDataSize);
+if (OutgoingPacket)
+{
+    memcpy(OutgoingPacket, PacketData, PacketDataSize);
+    WintunSendPacket(Session, OutgoingPacket);
+}
+else if (GetLastError() != ERROR_BUFFER_OVERFLOW) // Silently drop packets if the ring is full
+    Log(L"Packet write failed");
+```
+
+And the `WintunReceivePacket` and `WintunReleaseReceivePacket` functions can be used for receiving packets ([used by `ReceivePackets` in the example.c code](https://git.zx2c4.com/wintun/tree/example/example.c)):
+
+```C
+for (;;)
+{
+    DWORD IncomingPacketSize;
+    BYTE *IncomingPacket = WintunReceivePacket(Session, &IncomingPacketSize);
+    if (IncomingPacket)
+    {
+        DoSomethingWithPacket(IncomingPacket, IncomingPacketSize);
+        WintunReleaseReceivePacket(Session, IncomingPacket);
+    }
+    else if (GetLastError() == ERROR_NO_MORE_ITEMS)
+        WaitForSingleObject(WintunGetReadWaitEvent(Session), INFINITE);
+    else
+    {
+        Log(L"Packet read failed");
+        break;
+    }
+}
+```
+
+Some high performance use cases may want to spin on `WintunReceivePackets` for a number of cycles before falling back to waiting on the read-wait event.
+
+You are **highly encouraged** to read the [**example.c short example**](https://git.zx2c4.com/wintun/tree/example/example.c) to see how to put together a simple userspace network tunnel.
+
+The various functions and definitions are [documented in the reference below](#Reference).
+
+## Reference
+
+### Macro Definitions
+
+#### WINTUN\_MAX\_POOL
+
+`#define WINTUN_MAX_POOL   256`
+
+Maximum pool name length including zero terminator
+
+#### WINTUN\_MIN\_RING\_CAPACITY
+
+`#define WINTUN_MIN_RING_CAPACITY   0x20000 /* 128kiB */`
+
+Minimum ring capacity.
+
+#### WINTUN\_MAX\_RING\_CAPACITY
+
+`#define WINTUN_MAX_RING_CAPACITY   0x4000000 /* 64MiB */`
+
+Maximum ring capacity.
+
+#### WINTUN\_MAX\_IP\_PACKET\_SIZE
+
+`#define WINTUN_MAX_IP_PACKET_SIZE   0xFFFF`
+
+Maximum IP packet size
+
+### Typedefs
+
+#### WINTUN\_ADAPTER\_HANDLE
+
+`typedef void* WINTUN_ADAPTER_HANDLE`
+
+A handle representing Wintun adapter
+
+#### WINTUN\_ENUM\_CALLBACK
+
+`typedef BOOL(* WINTUN_ENUM_CALLBACK) (WINTUN_ADAPTER_HANDLE Adapter, LPARAM Param)`
+
+Called by WintunEnumAdapters for each adapter in the pool.
+
+**Parameters**
+
+- *Adapter*: Adapter handle, which will be freed when this function returns.
+- *Param*: An application-defined value passed to the WintunEnumAdapters.
+
+**Returns**
+
+Non-zero to continue iterating adapters; zero to stop.
+
+#### WINTUN\_LOGGER\_CALLBACK
+
+`typedef void(* WINTUN_LOGGER_CALLBACK) (WINTUN_LOGGER_LEVEL Level, DWORD64 Timestamp, const WCHAR *Message)`
+
+Called by internal logger to report diagnostic messages
+
+**Parameters**
+
+- *Level*: Message level.
+- *Timestamp*: Message timestamp in in 100ns intervals since 1601-01-01 UTC.
+- *Message*: Message text.
+
+#### WINTUN\_SESSION\_HANDLE
+
+`typedef void* WINTUN_SESSION_HANDLE`
+
+A handle representing Wintun session
+
+### Enumeration Types
+
+#### WINTUN\_LOGGER\_LEVEL
+
+`enum WINTUN_LOGGER_LEVEL`
+
+Determines the level of logging, passed to WINTUN\_LOGGER\_CALLBACK.
+
+- *WINTUN\_LOG\_INFO*: Informational
+- *WINTUN\_LOG\_WARN*: Warning
+- *WINTUN\_LOG\_ERR*: Error
+
+Enumerator
+
+### Functions
+
+#### WintunCreateAdapter()
+
+`WINTUN_ADAPTER_HANDLE WintunCreateAdapter (const WCHAR * Name, const WCHAR * TunnelType, const GUID * RequestedGUID)`
+
+Creates a new Wintun adapter.
+
+**Parameters**
+
+- *Name*: The requested name of the adapter. Zero-terminated string of up to MAX\_ADAPTER\_NAME-1 characters.
+- *Name*: Name of the adapter tunnel type. Zero-terminated string of up to MAX\_ADAPTER\_NAME-1 characters.
+- *RequestedGUID*: The GUID of the created network adapter, which then influences NLA generation deterministically. If it is set to NULL, the GUID is chosen by the system at random, and hence a new NLA entry is created for each new adapter. It is called "requested" GUID because the API it uses is completely undocumented, and so there could be minor interesting complications with its usage.
+
+**Returns**
+
+If the function succeeds, the return value is the adapter handle. Must be released with WintunCloseAdapter. If the function fails, the return value is NULL. To get extended error information, call GetLastError.
+
+#### WintunOpenAdapter()
+
+`WINTUN_ADAPTER_HANDLE WintunOpenAdapter (const WCHAR * Name)`
+
+Opens an existing Wintun adapter.
+
+**Parameters**
+
+- *Name*: The requested name of the adapter. Zero-terminated string of up to MAX\_ADAPTER\_NAME-1 characters.
+
+**Returns**
+
+If the function succeeds, the return value is adapter handle. Must be released with WintunCloseAdapter. If the function fails, the return value is NULL. To get extended error information, call GetLastError.
+
+#### WintunCloseAdapter()
+
+`void WintunCloseAdapter (WINTUN_ADAPTER_HANDLE Adapter)`
+
+Releases Wintun adapter resources and, if adapter was created with WintunCreateAdapter, removes adapter.
+
+**Parameters**
+
+- *Adapter*: Adapter handle obtained with WintunCreateAdapter or WintunOpenAdapter.
+
+#### WintunDeleteDriver()
+
+`BOOL WintunDeleteDriver ()`
+
+Deletes the Wintun driver if there are no more adapters in use.
+
+**Returns**
+
+If the function succeeds, the return value is nonzero. If the function fails, the return value is zero. To get extended error information, call GetLastError.
+
+#### WintunGetAdapterLuid()
+
+`void WintunGetAdapterLuid (WINTUN_ADAPTER_HANDLE Adapter, NET_LUID * Luid)`
+
+Returns the LUID of the adapter.
+
+**Parameters**
+
+- *Adapter*: Adapter handle obtained with WintunOpenAdapter or WintunCreateAdapter
+- *Luid*: Pointer to LUID to receive adapter LUID.
+
+#### WintunGetRunningDriverVersion()
+
+`DWORD WintunGetRunningDriverVersion (void )`
+
+Determines the version of the Wintun driver currently loaded.
+
+**Returns**
+
+If the function succeeds, the return value is the version number. If the function fails, the return value is zero. To get extended error information, call GetLastError. Possible errors include the following: ERROR\_FILE\_NOT\_FOUND Wintun not loaded
+
+#### WintunSetLogger()
+
+`void WintunSetLogger (WINTUN_LOGGER_CALLBACK NewLogger)`
+
+Sets logger callback function.
+
+**Parameters**
+
+- *NewLogger*: Pointer to callback function to use as a new global logger. NewLogger may be called from various threads concurrently. Should the logging require serialization, you must handle serialization in NewLogger. Set to NULL to disable.
+
+#### WintunStartSession()
+
+`WINTUN_SESSION_HANDLE WintunStartSession (WINTUN_ADAPTER_HANDLE Adapter, DWORD Capacity)`
+
+Starts Wintun session.
+
+**Parameters**
+
+- *Adapter*: Adapter handle obtained with WintunOpenAdapter or WintunCreateAdapter
+- *Capacity*: Rings capacity. Must be between WINTUN\_MIN\_RING\_CAPACITY and WINTUN\_MAX\_RING\_CAPACITY (incl.) Must be a power of two.
+
+**Returns**
+
+Wintun session handle. Must be released with WintunEndSession. If the function fails, the return value is NULL. To get extended error information, call GetLastError.
+
+#### WintunEndSession()
+
+`void WintunEndSession (WINTUN_SESSION_HANDLE Session)`
+
+Ends Wintun session.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+
+#### WintunGetReadWaitEvent()
+
+`HANDLE WintunGetReadWaitEvent (WINTUN_SESSION_HANDLE Session)`
+
+Gets Wintun session's read-wait event handle.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+
+**Returns**
+
+Pointer to receive event handle to wait for available data when reading. Should WintunReceivePackets return ERROR\_NO\_MORE\_ITEMS (after spinning on it for a while under heavy load), wait for this event to become signaled before retrying WintunReceivePackets. Do not call CloseHandle on this event - it is managed by the session.
+
+#### WintunReceivePacket()
+
+`BYTE* WintunReceivePacket (WINTUN_SESSION_HANDLE Session, DWORD * PacketSize)`
+
+Retrieves one or packet. After the packet content is consumed, call WintunReleaseReceivePacket with Packet returned from this function to release internal buffer. This function is thread-safe.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+- *PacketSize*: Pointer to receive packet size.
+
+**Returns**
+
+Pointer to layer 3 IPv4 or IPv6 packet. Client may modify its content at will. If the function fails, the return value is NULL. To get extended error information, call GetLastError. Possible errors include the following: ERROR\_HANDLE\_EOF Wintun adapter is terminating; ERROR\_NO\_MORE\_ITEMS Wintun buffer is exhausted; ERROR\_INVALID\_DATA Wintun buffer is corrupt
+
+#### WintunReleaseReceivePacket()
+
+`void WintunReleaseReceivePacket (WINTUN_SESSION_HANDLE Session, const BYTE * Packet)`
+
+Releases internal buffer after the received packet has been processed by the client. This function is thread-safe.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+- *Packet*: Packet obtained with WintunReceivePacket
+
+#### WintunAllocateSendPacket()
+
+`BYTE* WintunAllocateSendPacket (WINTUN_SESSION_HANDLE Session, DWORD PacketSize)`
+
+Allocates memory for a packet to send. After the memory is filled with packet data, call WintunSendPacket to send and release internal buffer. WintunAllocateSendPacket is thread-safe and the WintunAllocateSendPacket order of calls define the packet sending order.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+- *PacketSize*: Exact packet size. Must be less or equal to WINTUN\_MAX\_IP\_PACKET\_SIZE.
+
+**Returns**
+
+Returns pointer to memory where to prepare layer 3 IPv4 or IPv6 packet for sending. If the function fails, the return value is NULL. To get extended error information, call GetLastError. Possible errors include the following: ERROR\_HANDLE\_EOF Wintun adapter is terminating; ERROR\_BUFFER\_OVERFLOW Wintun buffer is full;
+
+#### WintunSendPacket()
+
+`void WintunSendPacket (WINTUN_SESSION_HANDLE Session, const BYTE * Packet)`
+
+Sends the packet and releases internal buffer. WintunSendPacket is thread-safe, but the WintunAllocateSendPacket order of calls define the packet sending order. This means the packet is not guaranteed to be sent in the WintunSendPacket yet.
+
+**Parameters**
+
+- *Session*: Wintun session handle obtained with WintunStartSession
+- *Packet*: Packet obtained with WintunAllocateSendPacket
+
+## Building
+
+**Do not distribute drivers or files named "Wintun", as they will most certainly clash with official deployments. Instead distribute [`wintun.dll` as downloaded from wintun.net](https://www.wintun.net).**
+
+General requirements:
+
+- [Visual Studio 2019](https://visualstudio.microsoft.com/downloads/) with Windows SDK
+- [Windows Driver Kit](https://docs.microsoft.com/en-us/windows-hardware/drivers/download-the-wdk)
+
+`wintun.sln` may be opened in Visual Studio for development and building. Be sure to run `bcdedit /set testsigning on` and then reboot before to enable unsigned driver loading. The default run sequence (F5) in Visual Studio will build the example project and its dependencies.
+
+## License
+
+The entire contents of [the repository](https://git.zx2c4.com/wintun/), including all documentation and example code, is "Copyright © 2018-2021 WireGuard LLC. All Rights Reserved." Source code is licensed under the [GPLv2](COPYING). Prebuilt binaries from [wintun.net](https://www.wintun.net/) are released under a more permissive license suitable for more forms of software contained inside of the .zip files distributed there.

dist/windows/wintun/include/wintun.h

@@ -0,0 +1,270 @@
+/* SPDX-License-Identifier: GPL-2.0 OR MIT
+ *
+ * Copyright (C) 2018-2021 WireGuard LLC. All Rights Reserved.
+ */
+
+#pragma once
+
+#include <winsock2.h>
+#include <windows.h>
+#include <ipexport.h>
+#include <ifdef.h>
+#include <ws2ipdef.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef ALIGNED
+#    if defined(_MSC_VER)
+#        define ALIGNED(n) __declspec(align(n))
+#    elif defined(__GNUC__)
+#        define ALIGNED(n) __attribute__((aligned(n)))
+#    else
+#        error "Unable to define ALIGNED"
+#    endif
+#endif
+
+/* MinGW is missing this one, unfortunately. */
+#ifndef _Post_maybenull_
+#    define _Post_maybenull_
+#endif
+
+#pragma warning(push)
+#pragma warning(disable : 4324) /* structure was padded due to alignment specifier */
+
+/**
+ * A handle representing Wintun adapter
+ */
+typedef struct _WINTUN_ADAPTER *WINTUN_ADAPTER_HANDLE;
+
+/**
+ * Creates a new Wintun adapter.
+ *
+ * @param Name          The requested name of the adapter. Zero-terminated string of up to MAX_ADAPTER_NAME-1
+ *                      characters.
+ *
+ * @param TunnelType    Name of the adapter tunnel type. Zero-terminated string of up to MAX_ADAPTER_NAME-1
+ *                      characters.
+ *
+ * @param RequestedGUID The GUID of the created network adapter, which then influences NLA generation deterministically.
+ *                      If it is set to NULL, the GUID is chosen by the system at random, and hence a new NLA entry is
+ *                      created for each new adapter. It is called "requested" GUID because the API it uses is
+ *                      completely undocumented, and so there could be minor interesting complications with its usage.
+ *
+ * @return If the function succeeds, the return value is the adapter handle. Must be released with
+ * WintunCloseAdapter. If the function fails, the return value is NULL. To get extended error information, call
+ * GetLastError.
+ */
+typedef _Must_inspect_result_
+_Return_type_success_(return != NULL)
+_Post_maybenull_
+WINTUN_ADAPTER_HANDLE(WINAPI WINTUN_CREATE_ADAPTER_FUNC)
+(_In_z_ LPCWSTR Name, _In_z_ LPCWSTR TunnelType, _In_opt_ const GUID *RequestedGUID);
+
+/**
+ * Opens an existing Wintun adapter.
+ *
+ * @param Name          The requested name of the adapter. Zero-terminated string of up to MAX_ADAPTER_NAME-1
+ *                      characters.
+ *
+ * @return If the function succeeds, the return value is the adapter handle. Must be released with
+ * WintunCloseAdapter. If the function fails, the return value is NULL. To get extended error information, call
+ * GetLastError.
+ */
+typedef _Must_inspect_result_
+_Return_type_success_(return != NULL)
+_Post_maybenull_
+WINTUN_ADAPTER_HANDLE(WINAPI WINTUN_OPEN_ADAPTER_FUNC)(_In_z_ LPCWSTR Name);
+
+/**
+ * Releases Wintun adapter resources and, if adapter was created with WintunCreateAdapter, removes adapter.
+ *
+ * @param Adapter       Adapter handle obtained with WintunCreateAdapter or WintunOpenAdapter.
+ */
+typedef VOID(WINAPI WINTUN_CLOSE_ADAPTER_FUNC)(_In_opt_ WINTUN_ADAPTER_HANDLE Adapter);
+
+/**
+ * Deletes the Wintun driver if there are no more adapters in use.
+ *
+ * @return If the function succeeds, the return value is nonzero. If the function fails, the return value is zero. To
+ *         get extended error information, call GetLastError.
+ */
+typedef _Return_type_success_(return != FALSE)
+BOOL(WINAPI WINTUN_DELETE_DRIVER_FUNC)(VOID);
+
+/**
+ * Returns the LUID of the adapter.
+ *
+ * @param Adapter       Adapter handle obtained with WintunCreateAdapter or WintunOpenAdapter
+ *
+ * @param Luid          Pointer to LUID to receive adapter LUID.
+ */
+typedef VOID(WINAPI WINTUN_GET_ADAPTER_LUID_FUNC)(_In_ WINTUN_ADAPTER_HANDLE Adapter, _Out_ NET_LUID *Luid);
+
+/**
+ * Determines the version of the Wintun driver currently loaded.
+ *
+ * @return If the function succeeds, the return value is the version number. If the function fails, the return value is
+ *         zero. To get extended error information, call GetLastError. Possible errors include the following:
+ *         ERROR_FILE_NOT_FOUND  Wintun not loaded
+ */
+typedef _Return_type_success_(return != 0)
+DWORD(WINAPI WINTUN_GET_RUNNING_DRIVER_VERSION_FUNC)(VOID);
+
+/**
+ * Determines the level of logging, passed to WINTUN_LOGGER_CALLBACK.
+ */
+typedef enum
+{
+    WINTUN_LOG_INFO, /**< Informational */
+    WINTUN_LOG_WARN, /**< Warning */
+    WINTUN_LOG_ERR   /**< Error */
+} WINTUN_LOGGER_LEVEL;
+
+/**
+ * Called by internal logger to report diagnostic messages
+ *
+ * @param Level         Message level.
+ *
+ * @param Timestamp     Message timestamp in in 100ns intervals since 1601-01-01 UTC.
+ *
+ * @param Message       Message text.
+ */
+typedef VOID(CALLBACK *WINTUN_LOGGER_CALLBACK)(
+    _In_ WINTUN_LOGGER_LEVEL Level,
+    _In_ DWORD64 Timestamp,
+    _In_z_ LPCWSTR Message);
+
+/**
+ * Sets logger callback function.
+ *
+ * @param NewLogger     Pointer to callback function to use as a new global logger. NewLogger may be called from various
+ *                      threads concurrently. Should the logging require serialization, you must handle serialization in
+ *                      NewLogger. Set to NULL to disable.
+ */
+typedef VOID(WINAPI WINTUN_SET_LOGGER_FUNC)(_In_ WINTUN_LOGGER_CALLBACK NewLogger);
+
+/**
+ * Minimum ring capacity.
+ */
+#define WINTUN_MIN_RING_CAPACITY 0x20000 /* 128kiB */
+
+/**
+ * Maximum ring capacity.
+ */
+#define WINTUN_MAX_RING_CAPACITY 0x4000000 /* 64MiB */
+
+/**
+ * A handle representing Wintun session
+ */
+typedef struct _TUN_SESSION *WINTUN_SESSION_HANDLE;
+
+/**
+ * Starts Wintun session.
+ *
+ * @param Adapter       Adapter handle obtained with WintunOpenAdapter or WintunCreateAdapter
+ *
+ * @param Capacity      Rings capacity. Must be between WINTUN_MIN_RING_CAPACITY and WINTUN_MAX_RING_CAPACITY (incl.)
+ *                      Must be a power of two.
+ *
+ * @return Wintun session handle. Must be released with WintunEndSession. If the function fails, the return value is
+ *         NULL. To get extended error information, call GetLastError.
+ */
+typedef _Must_inspect_result_
+_Return_type_success_(return != NULL)
+_Post_maybenull_
+WINTUN_SESSION_HANDLE(WINAPI WINTUN_START_SESSION_FUNC)(_In_ WINTUN_ADAPTER_HANDLE Adapter, _In_ DWORD Capacity);
+
+/**
+ * Ends Wintun session.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ */
+typedef VOID(WINAPI WINTUN_END_SESSION_FUNC)(_In_ WINTUN_SESSION_HANDLE Session);
+
+/**
+ * Gets Wintun session's read-wait event handle.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ *
+ * @return Pointer to receive event handle to wait for available data when reading. Should
+ *         WintunReceivePackets return ERROR_NO_MORE_ITEMS (after spinning on it for a while under heavy
+ *         load), wait for this event to become signaled before retrying WintunReceivePackets. Do not call
+ *         CloseHandle on this event - it is managed by the session.
+ */
+typedef HANDLE(WINAPI WINTUN_GET_READ_WAIT_EVENT_FUNC)(_In_ WINTUN_SESSION_HANDLE Session);
+
+/**
+ * Maximum IP packet size
+ */
+#define WINTUN_MAX_IP_PACKET_SIZE 0xFFFF
+
+/**
+ * Retrieves one or packet. After the packet content is consumed, call WintunReleaseReceivePacket with Packet returned
+ * from this function to release internal buffer. This function is thread-safe.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ *
+ * @param PacketSize    Pointer to receive packet size.
+ *
+ * @return Pointer to layer 3 IPv4 or IPv6 packet. Client may modify its content at will. If the function fails, the
+ *         return value is NULL. To get extended error information, call GetLastError. Possible errors include the
+ *         following:
+ *         ERROR_HANDLE_EOF     Wintun adapter is terminating;
+ *         ERROR_NO_MORE_ITEMS  Wintun buffer is exhausted;
+ *         ERROR_INVALID_DATA   Wintun buffer is corrupt
+ */
+typedef _Must_inspect_result_
+_Return_type_success_(return != NULL)
+_Post_maybenull_
+_Post_writable_byte_size_(*PacketSize)
+BYTE *(WINAPI WINTUN_RECEIVE_PACKET_FUNC)(_In_ WINTUN_SESSION_HANDLE Session, _Out_ DWORD *PacketSize);
+
+/**
+ * Releases internal buffer after the received packet has been processed by the client. This function is thread-safe.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ *
+ * @param Packet        Packet obtained with WintunReceivePacket
+ */
+typedef VOID(
+    WINAPI WINTUN_RELEASE_RECEIVE_PACKET_FUNC)(_In_ WINTUN_SESSION_HANDLE Session, _In_ const BYTE *Packet);
+
+/**
+ * Allocates memory for a packet to send. After the memory is filled with packet data, call WintunSendPacket to send
+ * and release internal buffer. WintunAllocateSendPacket is thread-safe and the WintunAllocateSendPacket order of
+ * calls define the packet sending order.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ *
+ * @param PacketSize    Exact packet size. Must be less or equal to WINTUN_MAX_IP_PACKET_SIZE.
+ *
+ * @return Returns pointer to memory where to prepare layer 3 IPv4 or IPv6 packet for sending. If the function fails,
+ *         the return value is NULL. To get extended error information, call GetLastError. Possible errors include the
+ *         following:
+ *         ERROR_HANDLE_EOF       Wintun adapter is terminating;
+ *         ERROR_BUFFER_OVERFLOW  Wintun buffer is full;
+ */
+typedef _Must_inspect_result_
+_Return_type_success_(return != NULL)
+_Post_maybenull_
+_Post_writable_byte_size_(PacketSize)
+BYTE *(WINAPI WINTUN_ALLOCATE_SEND_PACKET_FUNC)(_In_ WINTUN_SESSION_HANDLE Session, _In_ DWORD PacketSize);
+
+/**
+ * Sends the packet and releases internal buffer. WintunSendPacket is thread-safe, but the WintunAllocateSendPacket
+ * order of calls define the packet sending order. This means the packet is not guaranteed to be sent in the
+ * WintunSendPacket yet.
+ *
+ * @param Session       Wintun session handle obtained with WintunStartSession
+ *
+ * @param Packet        Packet obtained with WintunAllocateSendPacket
+ */
+typedef VOID(WINAPI WINTUN_SEND_PACKET_FUNC)(_In_ WINTUN_SESSION_HANDLE Session, _In_ const BYTE *Packet);
+
+#pragma warning(pop)
+
+#ifdef __cplusplus
+}
+#endif

dns_server.go

@@ -4,10 +4,13 @@ import (
 	"fmt"
 	"net"
 	"strconv"
+	"strings"
 	"sync"
 
 	"github.com/miekg/dns"
 	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/iputil"
 )
 
 // This whole thing should be rewritten to use context
@@ -31,11 +34,10 @@ func newDnsRecords(hostMap *HostMap) *dnsRecords {
 
 func (d *dnsRecords) Query(data string) string {
 	d.RLock()
-	if r, ok := d.dnsMap[data]; ok {
-		d.RUnlock()
+	defer d.RUnlock()
+	if r, ok := d.dnsMap[strings.ToLower(data)]; ok {
 		return r
 	}
-	d.RUnlock()
 	return ""
 }
 
@@ -44,8 +46,8 @@ func (d *dnsRecords) QueryCert(data string) string {
 	if ip == nil {
 		return ""
 	}
-	iip := ip2int(ip)
-	hostinfo, err := d.hostMap.QueryVpnIP(iip)
+	iip := iputil.Ip2VpnIp(ip)
+	hostinfo, err := d.hostMap.QueryVpnIp(iip)
 	if err != nil {
 		return ""
 	}
@@ -60,8 +62,8 @@ func (d *dnsRecords) QueryCert(data string) string {
 
 func (d *dnsRecords) Add(host, data string) {
 	d.Lock()
-	d.dnsMap[host] = data
-	d.Unlock()
+	defer d.Unlock()
+	d.dnsMap[strings.ToLower(host)] = data
 }
 
 func parseQuery(l *logrus.Logger, m *dns.Msg, w dns.ResponseWriter) {
@@ -109,7 +111,7 @@ func handleDnsRequest(l *logrus.Logger, w dns.ResponseWriter, r *dns.Msg) {
 	w.WriteMsg(m)
 }
 
-func dnsMain(l *logrus.Logger, hostMap *HostMap, c *Config) func() {
+func dnsMain(l *logrus.Logger, hostMap *HostMap, c *config.C) func() {
 	dnsR = newDnsRecords(hostMap)
 
 	// attach request handler func
@@ -117,7 +119,7 @@ func dnsMain(l *logrus.Logger, hostMap *HostMap, c *Config) func() {
 		handleDnsRequest(l, w, r)
 	})
 
-	c.RegisterReloadCallback(func(c *Config) {
+	c.RegisterReloadCallback(func(c *config.C) {
 		reloadDns(l, c)
 	})
 
@@ -126,14 +128,14 @@ func dnsMain(l *logrus.Logger, hostMap *HostMap, c *Config) func() {
 	}
 }
 
-func getDnsServerAddr(c *Config) string {
+func getDnsServerAddr(c *config.C) string {
 	return c.GetString("lighthouse.dns.host", "") + ":" + strconv.Itoa(c.GetInt("lighthouse.dns.port", 53))
 }
 
-func startDns(l *logrus.Logger, c *Config) {
+func startDns(l *logrus.Logger, c *config.C) {
 	dnsAddr = getDnsServerAddr(c)
 	dnsServer = &dns.Server{Addr: dnsAddr, Net: "udp"}
-	l.WithField("dnsListener", dnsAddr).Infof("Starting DNS responder")
+	l.WithField("dnsListener", dnsAddr).Info("Starting DNS responder")
 	err := dnsServer.ListenAndServe()
 	defer dnsServer.Shutdown()
 	if err != nil {
@@ -141,7 +143,7 @@ func startDns(l *logrus.Logger, c *Config) {
 	}
 }
 
-func reloadDns(l *logrus.Logger, c *Config) {
+func reloadDns(l *logrus.Logger, c *config.C) {
 	if dnsAddr == getDnsServerAddr(c) {
 		l.Debug("No DNS server config change detected")
 		return

e2e/handshakes_test.go

@@ -1,31 +1,62 @@
+//go:build e2e_testing
 // +build e2e_testing
 
 package e2e
 
 import (
+	"fmt"
 	"net"
 	"testing"
 	"time"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/e2e/router"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
+	"gopkg.in/yaml.v2"
 )
 
+func BenchmarkHotPath(b *testing.B) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, _, _, _ := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
+
+	// Put their info in our lighthouse
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	r := router.NewR(b, myControl, theirControl)
+	r.CancelFlowLogs()
+
+	for n := 0; n < b.N; n++ {
+		myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+		_ = r.RouteForAllUntilTxTun(theirControl)
+	}
+
+	myControl.Stop()
+	theirControl.Stop()
+}
+
 func TestGoodHandshake(t *testing.T) {
 	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
-	myControl, myVpnIp, myUdpAddr := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 1})
-	theirControl, theirVpnIp, theirUdpAddr := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
 
 	// Put their info in our lighthouse
-	myControl.InjectLightHouseAddr(theirVpnIp, theirUdpAddr)
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
 
 	// Start the servers
 	myControl.Start()
 	theirControl.Start()
 
 	t.Log("Send a udp packet through to begin standing up the tunnel, this should come out the other side")
-	myControl.InjectTunUDPPacket(theirVpnIp, 80, 80, []byte("Hi from me"))
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
 
 	t.Log("Have them consume my stage 0 packet. They have a tunnel now")
 	theirControl.InjectUDPPacket(myControl.GetFromUDP(true))
@@ -36,7 +67,7 @@ func TestGoodHandshake(t *testing.T) {
 	t.Log("I consume a garbage packet with a proper nebula header for our tunnel")
 	// this should log a statement and get ignored, allowing the real handshake packet to complete the tunnel
 	badPacket := stage1Packet.Copy()
-	badPacket.Data = badPacket.Data[:len(badPacket.Data)-nebula.HeaderLen]
+	badPacket.Data = badPacket.Data[:len(badPacket.Data)-header.Len]
 	myControl.InjectUDPPacket(badPacket)
 
 	t.Log("Have me consume their real stage 1 packet. I have a tunnel now")
@@ -46,15 +77,18 @@ func TestGoodHandshake(t *testing.T) {
 	myControl.WaitForType(1, 0, theirControl)
 
 	t.Log("Make sure our host infos are correct")
-	assertHostInfoPair(t, myUdpAddr, theirUdpAddr, myVpnIp, theirVpnIp, myControl, theirControl)
+	assertHostInfoPair(t, myUdpAddr, theirUdpAddr, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl)
 
 	t.Log("Get that cached packet and make sure it looks right")
 	myCachedPacket := theirControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIp, theirVpnIp, 80, 80)
+	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
 
 	t.Log("Do a bidirectional tunnel test")
-	assertTunnel(t, myVpnIp, theirVpnIp, myControl, theirControl, router.NewR(myControl, theirControl))
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
 
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
 	myControl.Stop()
 	theirControl.Stop()
 	//TODO: assert hostmaps
@@ -66,18 +100,19 @@ func TestWrongResponderHandshake(t *testing.T) {
 	// The IPs here are chosen on purpose:
 	// The current remote handling will sort by preference, public, and then lexically.
 	// So we need them to have a higher address than evil (we could apply a preference though)
-	myControl, myVpnIp, myUdpAddr := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 100})
-	theirControl, theirVpnIp, theirUdpAddr := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 99})
-	evilControl, evilVpnIp, evilUdpAddr := newSimpleServer(ca, caKey, "evil", net.IP{10, 0, 0, 2})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 100}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 99}, nil)
+	evilControl, evilVpnIp, evilUdpAddr, _ := newSimpleServer(ca, caKey, "evil", net.IP{10, 0, 0, 2}, nil)
 
 	// Add their real udp addr, which should be tried after evil.
-	myControl.InjectLightHouseAddr(theirVpnIp, theirUdpAddr)
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
 
 	// Put the evil udp addr in for their vpn Ip, this is a case of being lied to by the lighthouse.
-	myControl.InjectLightHouseAddr(theirVpnIp, evilUdpAddr)
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, evilUdpAddr)
 
 	// Build a router so we don't have to reason who gets which packet
-	r := router.NewR(myControl, theirControl, evilControl)
+	r := router.NewR(t, myControl, theirControl, evilControl)
+	defer r.RenderFlow()
 
 	// Start the servers
 	myControl.Start()
@@ -85,9 +120,9 @@ func TestWrongResponderHandshake(t *testing.T) {
 	evilControl.Start()
 
 	t.Log("Start the handshake process, we will route until we see our cached packet get sent to them")
-	myControl.InjectTunUDPPacket(theirVpnIp, 80, 80, []byte("Hi from me"))
-	r.RouteForAllExitFunc(func(p *nebula.UdpPacket, c *nebula.Control) router.ExitType {
-		h := &nebula.Header{}
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+	r.RouteForAllExitFunc(func(p *udp.Packet, c *nebula.Control) router.ExitType {
+		h := &header.H{}
 		err := h.Parse(p.Data)
 		if err != nil {
 			panic(err)
@@ -104,78 +139,732 @@ func TestWrongResponderHandshake(t *testing.T) {
 
 	t.Log("My cached packet should be received by them")
 	myCachedPacket := theirControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIp, theirVpnIp, 80, 80)
+	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
 
 	t.Log("Test the tunnel with them")
-	assertHostInfoPair(t, myUdpAddr, theirUdpAddr, myVpnIp, theirVpnIp, myControl, theirControl)
-	assertTunnel(t, myVpnIp, theirVpnIp, myControl, theirControl, r)
+	assertHostInfoPair(t, myUdpAddr, theirUdpAddr, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl)
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
 
 	t.Log("Flush all packets from all controllers")
 	r.FlushAll()
 
 	t.Log("Ensure ensure I don't have any hostinfo artifacts from evil")
-	assert.Nil(t, myControl.GetHostInfoByVpnIP(ip2int(evilVpnIp), true), "My pending hostmap should not contain evil")
-	assert.Nil(t, myControl.GetHostInfoByVpnIP(ip2int(evilVpnIp), false), "My main hostmap should not contain evil")
+	assert.Nil(t, myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(evilVpnIp.IP), true), "My pending hostmap should not contain evil")
+	assert.Nil(t, myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(evilVpnIp.IP), false), "My main hostmap should not contain evil")
 	//NOTE: if evil lost the handshake race it may still have a tunnel since me would reject the handshake since the tunnel is complete
 
 	//TODO: assert hostmaps for everyone
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl, evilControl)
 	t.Log("Success!")
 	myControl.Stop()
 	theirControl.Stop()
 }
 
-func Test_Case1_Stage1Race(t *testing.T) {
+func TestStage1Race(t *testing.T) {
+	// This tests ensures that two hosts handshaking with each other at the same time will allow traffic to flow
+	// But will eventually collapse down to a single tunnel
+
 	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
-	myControl, myVpnIp, myUdpAddr := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 1})
-	theirControl, theirVpnIp, theirUdpAddr := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
 
 	// Put their info in our lighthouse and vice versa
-	myControl.InjectLightHouseAddr(theirVpnIp, theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIp, myUdpAddr)
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
 
 	// Build a router so we don't have to reason who gets which packet
-	r := router.NewR(myControl, theirControl)
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
 
 	// Start the servers
 	myControl.Start()
 	theirControl.Start()
 
 	t.Log("Trigger a handshake to start on both me and them")
-	myControl.InjectTunUDPPacket(theirVpnIp, 80, 80, []byte("Hi from me"))
-	theirControl.InjectTunUDPPacket(myVpnIp, 80, 80, []byte("Hi from them"))
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+	theirControl.InjectTunUDPPacket(myVpnIpNet.IP, 80, 80, []byte("Hi from them"))
 
 	t.Log("Get both stage 1 handshake packets")
 	myHsForThem := myControl.GetFromUDP(true)
 	theirHsForMe := theirControl.GetFromUDP(true)
 
-	t.Log("Now inject both stage 1 handshake packets")
-	myControl.InjectUDPPacket(theirHsForMe)
-	theirControl.InjectUDPPacket(myHsForThem)
-	//TODO: they should win, grab their index for me and make sure I use it in the end.
+	r.Log("Now inject both stage 1 handshake packets")
+	r.InjectUDPPacket(theirControl, myControl, theirHsForMe)
+	r.InjectUDPPacket(myControl, theirControl, myHsForThem)
 
-	t.Log("They should not have a stage 2 (won the race) but I should send one")
-	theirControl.InjectUDPPacket(myControl.GetFromUDP(true))
+	r.Log("Route until they receive a message packet")
+	myCachedPacket := r.RouteForAllUntilTxTun(theirControl)
+	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
 
-	t.Log("Route for me until I send a message packet to them")
-	myControl.WaitForType(1, 0, theirControl)
+	r.Log("Their cached packet should be received by me")
+	theirCachedPacket := r.RouteForAllUntilTxTun(myControl)
+	assertUdpPacket(t, []byte("Hi from them"), theirCachedPacket, theirVpnIpNet.IP, myVpnIpNet.IP, 80, 80)
 
-	t.Log("My cached packet should be received by them")
-	myCachedPacket := theirControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi from me"), myCachedPacket, myVpnIp, theirVpnIp, 80, 80)
+	r.Log("Do a bidirectional tunnel test")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
 
-	t.Log("Route for them until I send a message packet to me")
-	theirControl.WaitForType(1, 0, myControl)
+	myHostmapHosts := myControl.ListHostmapHosts(false)
+	myHostmapIndexes := myControl.ListHostmapIndexes(false)
+	theirHostmapHosts := theirControl.ListHostmapHosts(false)
+	theirHostmapIndexes := theirControl.ListHostmapIndexes(false)
 
-	t.Log("Their cached packet should be received by me")
-	theirCachedPacket := myControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi from them"), theirCachedPacket, theirVpnIp, myVpnIp, 80, 80)
+	// We should have two tunnels on both sides
+	assert.Len(t, myHostmapHosts, 1)
+	assert.Len(t, theirHostmapHosts, 1)
+	assert.Len(t, myHostmapIndexes, 2)
+	assert.Len(t, theirHostmapIndexes, 2)
 
-	t.Log("Do a bidirectional tunnel test")
-	assertTunnel(t, myVpnIp, theirVpnIp, myControl, theirControl, r)
+	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
+
+	r.Log("Spin until connection manager tears down a tunnel")
+
+	for len(myControl.GetHostmap().Indexes)+len(theirControl.GetHostmap().Indexes) > 2 {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		t.Log("Connection manager hasn't ticked yet")
+		time.Sleep(time.Second)
+	}
+
+	myFinalHostmapHosts := myControl.ListHostmapHosts(false)
+	myFinalHostmapIndexes := myControl.ListHostmapIndexes(false)
+	theirFinalHostmapHosts := theirControl.ListHostmapHosts(false)
+	theirFinalHostmapIndexes := theirControl.ListHostmapIndexes(false)
+
+	// We should only have a single tunnel now on both sides
+	assert.Len(t, myFinalHostmapHosts, 1)
+	assert.Len(t, theirFinalHostmapHosts, 1)
+	assert.Len(t, myFinalHostmapIndexes, 1)
+	assert.Len(t, theirFinalHostmapIndexes, 1)
+
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+	myControl.Stop()
+	theirControl.Stop()
+}
+
+func TestUncleanShutdownRaceLoser(t *testing.T) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	r.Log("Trigger a handshake from me to them")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+
+	p := r.RouteForAllUntilTxTun(theirControl)
+	assertUdpPacket(t, []byte("Hi from me"), p, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
+
+	r.Log("Nuke my hostmap")
+	myHostmap := myControl.GetHostmap()
+	myHostmap.Hosts = map[iputil.VpnIp]*nebula.HostInfo{}
+	myHostmap.Indexes = map[uint32]*nebula.HostInfo{}
+	myHostmap.RemoteIndexes = map[uint32]*nebula.HostInfo{}
+
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me again"))
+	p = r.RouteForAllUntilTxTun(theirControl)
+	assertUdpPacket(t, []byte("Hi from me again"), p, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
+
+	r.Log("Assert the tunnel works")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+
+	r.Log("Wait for the dead index to go away")
+	start := len(theirControl.GetHostmap().Indexes)
+	for {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		if len(theirControl.GetHostmap().Indexes) < start {
+			break
+		}
+		time.Sleep(time.Second)
+	}
+
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+}
+
+func TestUncleanShutdownRaceWinner(t *testing.T) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	r.Log("Trigger a handshake from me to them")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+
+	p := r.RouteForAllUntilTxTun(theirControl)
+	assertUdpPacket(t, []byte("Hi from me"), p, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+
+	r.Log("Nuke my hostmap")
+	theirHostmap := theirControl.GetHostmap()
+	theirHostmap.Hosts = map[iputil.VpnIp]*nebula.HostInfo{}
+	theirHostmap.Indexes = map[uint32]*nebula.HostInfo{}
+	theirHostmap.RemoteIndexes = map[uint32]*nebula.HostInfo{}
+
+	theirControl.InjectTunUDPPacket(myVpnIpNet.IP, 80, 80, []byte("Hi from them again"))
+	p = r.RouteForAllUntilTxTun(myControl)
+	assertUdpPacket(t, []byte("Hi from them again"), p, theirVpnIpNet.IP, myVpnIpNet.IP, 80, 80)
+	r.RenderHostmaps("Derp hostmaps", myControl, theirControl)
+
+	r.Log("Assert the tunnel works")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+
+	r.Log("Wait for the dead index to go away")
+	start := len(myControl.GetHostmap().Indexes)
+	for {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		if len(myControl.GetHostmap().Indexes) < start {
+			break
+		}
+		time.Sleep(time.Second)
+	}
+
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+}
+
+func TestRelays(t *testing.T) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, _, _ := newSimpleServer(ca, caKey, "me     ", net.IP{10, 0, 0, 1}, m{"relay": m{"use_relays": true}})
+	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(ca, caKey, "relay  ", net.IP{10, 0, 0, 128}, m{"relay": m{"am_relay": true}})
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them   ", net.IP{10, 0, 0, 2}, m{"relay": m{"use_relays": true}})
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+	myControl.InjectRelays(theirVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+	relayControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, relayControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	relayControl.Start()
+	theirControl.Start()
+
+	t.Log("Trigger a handshake from me to them via the relay")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+
+	p := r.RouteForAllUntilTxTun(theirControl)
+	r.Log("Assert the tunnel works")
+	assertUdpPacket(t, []byte("Hi from me"), p, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
+	r.RenderHostmaps("Final hostmaps", myControl, relayControl, theirControl)
+	//TODO: assert we actually used the relay even though it should be impossible for a tunnel to have occurred without it
+}
+
+func TestStage1RaceRelays(t *testing.T) {
+	//NOTE: this is a race between me and relay resulting in a full tunnel from me to them via relay
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me     ", net.IP{10, 0, 0, 1}, m{"relay": m{"use_relays": true}})
+	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(ca, caKey, "relay  ", net.IP{10, 0, 0, 128}, m{"relay": m{"am_relay": true}})
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them   ", net.IP{10, 0, 0, 2}, m{"relay": m{"use_relays": true}})
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+	theirControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+
+	myControl.InjectRelays(theirVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+	theirControl.InjectRelays(myVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+
+	relayControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	relayControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, relayControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	relayControl.Start()
+	theirControl.Start()
+
+	r.Log("Get a tunnel between me and relay")
+	assertTunnel(t, myVpnIpNet.IP, relayVpnIpNet.IP, myControl, relayControl, r)
+
+	r.Log("Get a tunnel between them and relay")
+	assertTunnel(t, theirVpnIpNet.IP, relayVpnIpNet.IP, theirControl, relayControl, r)
+
+	r.Log("Trigger a handshake from both them and me via relay to them and me")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+	theirControl.InjectTunUDPPacket(myVpnIpNet.IP, 80, 80, []byte("Hi from them"))
+
+	r.Log("Wait for a packet from them to me")
+	p := r.RouteForAllUntilTxTun(myControl)
+	_ = p
 
 	myControl.Stop()
 	theirControl.Stop()
-	//TODO: assert hostmaps
+	relayControl.Stop()
+	//
+	////TODO: assert hostmaps
 }
 
+func TestStage1RaceRelays2(t *testing.T) {
+	//NOTE: this is a race between me and relay resulting in a full tunnel from me to them via relay
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me     ", net.IP{10, 0, 0, 1}, m{"relay": m{"use_relays": true}})
+	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(ca, caKey, "relay  ", net.IP{10, 0, 0, 128}, m{"relay": m{"am_relay": true}})
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them   ", net.IP{10, 0, 0, 2}, m{"relay": m{"use_relays": true}})
+	l := NewTestLogger()
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+	theirControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+
+	myControl.InjectRelays(theirVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+	theirControl.InjectRelays(myVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+
+	relayControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	relayControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, relayControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	relayControl.Start()
+	theirControl.Start()
+
+	r.Log("Get a tunnel between me and relay")
+	l.Info("Get a tunnel between me and relay")
+	assertTunnel(t, myVpnIpNet.IP, relayVpnIpNet.IP, myControl, relayControl, r)
+
+	r.Log("Get a tunnel between them and relay")
+	l.Info("Get a tunnel between them and relay")
+	assertTunnel(t, theirVpnIpNet.IP, relayVpnIpNet.IP, theirControl, relayControl, r)
+
+	r.Log("Trigger a handshake from both them and me via relay to them and me")
+	l.Info("Trigger a handshake from both them and me via relay to them and me")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+	theirControl.InjectTunUDPPacket(myVpnIpNet.IP, 80, 80, []byte("Hi from them"))
+
+	//r.RouteUntilAfterMsgType(myControl, header.Control, header.MessageNone)
+	//r.RouteUntilAfterMsgType(theirControl, header.Control, header.MessageNone)
+
+	r.Log("Wait for a packet from them to me")
+	l.Info("Wait for a packet from them to me; myControl")
+	r.RouteForAllUntilTxTun(myControl)
+	l.Info("Wait for a packet from them to me; theirControl")
+	r.RouteForAllUntilTxTun(theirControl)
+
+	r.Log("Assert the tunnel works")
+	l.Info("Assert the tunnel works")
+	assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+
+	t.Log("Wait until we remove extra tunnels")
+	l.Info("Wait until we remove extra tunnels")
+	l.WithFields(
+		logrus.Fields{
+			"myControl":    len(myControl.GetHostmap().Indexes),
+			"theirControl": len(theirControl.GetHostmap().Indexes),
+			"relayControl": len(relayControl.GetHostmap().Indexes),
+		}).Info("Waiting for hostinfos to be removed...")
+	hostInfos := len(myControl.GetHostmap().Indexes) + len(theirControl.GetHostmap().Indexes) + len(relayControl.GetHostmap().Indexes)
+	retries := 60
+	for hostInfos > 6 && retries > 0 {
+		hostInfos = len(myControl.GetHostmap().Indexes) + len(theirControl.GetHostmap().Indexes) + len(relayControl.GetHostmap().Indexes)
+		l.WithFields(
+			logrus.Fields{
+				"myControl":    len(myControl.GetHostmap().Indexes),
+				"theirControl": len(theirControl.GetHostmap().Indexes),
+				"relayControl": len(relayControl.GetHostmap().Indexes),
+			}).Info("Waiting for hostinfos to be removed...")
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		t.Log("Connection manager hasn't ticked yet")
+		time.Sleep(time.Second)
+		retries--
+	}
+
+	r.Log("Assert the tunnel works")
+	l.Info("Assert the tunnel works")
+	assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+
+	myControl.Stop()
+	theirControl.Stop()
+	relayControl.Stop()
+
+	//
+	////TODO: assert hostmaps
+}
+func TestRehandshakingRelays(t *testing.T) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, _, _ := newSimpleServer(ca, caKey, "me     ", net.IP{10, 0, 0, 1}, m{"relay": m{"use_relays": true}})
+	relayControl, relayVpnIpNet, relayUdpAddr, relayConfig := newSimpleServer(ca, caKey, "relay  ", net.IP{10, 0, 0, 128}, m{"relay": m{"am_relay": true}})
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them   ", net.IP{10, 0, 0, 2}, m{"relay": m{"use_relays": true}})
+
+	// Teach my how to get to the relay and that their can be reached via the relay
+	myControl.InjectLightHouseAddr(relayVpnIpNet.IP, relayUdpAddr)
+	myControl.InjectRelays(theirVpnIpNet.IP, []net.IP{relayVpnIpNet.IP})
+	relayControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, relayControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	relayControl.Start()
+	theirControl.Start()
+
+	t.Log("Trigger a handshake from me to them via the relay")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+
+	p := r.RouteForAllUntilTxTun(theirControl)
+	r.Log("Assert the tunnel works")
+	assertUdpPacket(t, []byte("Hi from me"), p, myVpnIpNet.IP, theirVpnIpNet.IP, 80, 80)
+	r.RenderHostmaps("working hostmaps", myControl, relayControl, theirControl)
+
+	// When I update the certificate for the relay, both me and them will have 2 host infos for the relay,
+	// and the main host infos will not have any relay state to handle the me<->relay<->them tunnel.
+	r.Log("Renew relay certificate and spin until me and them sees it")
+	_, _, myNextPrivKey, myNextPEM := newTestCert(ca, caKey, "relay", time.Now(), time.Now().Add(5*time.Minute), relayVpnIpNet, nil, []string{"new group"})
+
+	caB, err := ca.MarshalToPEM()
+	if err != nil {
+		panic(err)
+	}
+
+	relayConfig.Settings["pki"] = m{
+		"ca":   string(caB),
+		"cert": string(myNextPEM),
+		"key":  string(myNextPrivKey),
+	}
+	rc, err := yaml.Marshal(relayConfig.Settings)
+	assert.NoError(t, err)
+	relayConfig.ReloadConfigString(string(rc))
+
+	for {
+		r.Log("Assert the tunnel works between myVpnIpNet and relayVpnIpNet")
+		assertTunnel(t, myVpnIpNet.IP, relayVpnIpNet.IP, myControl, relayControl, r)
+		c := myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(relayVpnIpNet.IP), false)
+		if len(c.Cert.Details.Groups) != 0 {
+			// We have a new certificate now
+			r.Log("Certificate between my and relay is updated!")
+			break
+		}
+
+		time.Sleep(time.Second)
+	}
+
+	for {
+		r.Log("Assert the tunnel works between theirVpnIpNet and relayVpnIpNet")
+		assertTunnel(t, theirVpnIpNet.IP, relayVpnIpNet.IP, theirControl, relayControl, r)
+		c := theirControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(relayVpnIpNet.IP), false)
+		if len(c.Cert.Details.Groups) != 0 {
+			// We have a new certificate now
+			r.Log("Certificate between their and relay is updated!")
+			break
+		}
+
+		time.Sleep(time.Second)
+	}
+
+	r.Log("Assert the relay tunnel still works")
+	assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+	r.RenderHostmaps("working hostmaps", myControl, relayControl, theirControl)
+	// We should have two hostinfos on all sides
+	for len(myControl.GetHostmap().Indexes) != 2 {
+		t.Logf("Waiting for myControl hostinfos (%v != 2) to get cleaned up from lack of use...", len(myControl.GetHostmap().Indexes))
+		r.Log("Assert the relay tunnel still works")
+		assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+		r.Log("yupitdoes")
+		time.Sleep(time.Second)
+	}
+	t.Logf("myControl hostinfos got cleaned up!")
+	for len(theirControl.GetHostmap().Indexes) != 2 {
+		t.Logf("Waiting for theirControl hostinfos (%v != 2) to get cleaned up from lack of use...", len(theirControl.GetHostmap().Indexes))
+		r.Log("Assert the relay tunnel still works")
+		assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+		r.Log("yupitdoes")
+		time.Sleep(time.Second)
+	}
+	t.Logf("theirControl hostinfos got cleaned up!")
+	for len(relayControl.GetHostmap().Indexes) != 2 {
+		t.Logf("Waiting for relayControl hostinfos (%v != 2) to get cleaned up from lack of use...", len(relayControl.GetHostmap().Indexes))
+		r.Log("Assert the relay tunnel still works")
+		assertTunnel(t, theirVpnIpNet.IP, myVpnIpNet.IP, theirControl, myControl, r)
+		r.Log("yupitdoes")
+		time.Sleep(time.Second)
+	}
+	t.Logf("relayControl hostinfos got cleaned up!")
+}
+
+func TestRehandshaking(t *testing.T) {
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, myConfig := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 2}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, theirConfig := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 1}, nil)
+
+	// Put their info in our lighthouse and vice versa
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	t.Log("Stand up a tunnel between me and them")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+
+	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
+
+	r.Log("Renew my certificate and spin until their sees it")
+	_, _, myNextPrivKey, myNextPEM := newTestCert(ca, caKey, "me", time.Now(), time.Now().Add(5*time.Minute), myVpnIpNet, nil, []string{"new group"})
+
+	caB, err := ca.MarshalToPEM()
+	if err != nil {
+		panic(err)
+	}
+
+	myConfig.Settings["pki"] = m{
+		"ca":   string(caB),
+		"cert": string(myNextPEM),
+		"key":  string(myNextPrivKey),
+	}
+	rc, err := yaml.Marshal(myConfig.Settings)
+	assert.NoError(t, err)
+	myConfig.ReloadConfigString(string(rc))
+
+	for {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		c := theirControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(myVpnIpNet.IP), false)
+		if len(c.Cert.Details.Groups) != 0 {
+			// We have a new certificate now
+			break
+		}
+
+		time.Sleep(time.Second)
+	}
+
+	// Flip their firewall to only allowing the new group to catch the tunnels reverting incorrectly
+	rc, err = yaml.Marshal(theirConfig.Settings)
+	assert.NoError(t, err)
+	var theirNewConfig m
+	assert.NoError(t, yaml.Unmarshal(rc, &theirNewConfig))
+	theirFirewall := theirNewConfig["firewall"].(map[interface{}]interface{})
+	theirFirewall["inbound"] = []m{{
+		"proto": "any",
+		"port":  "any",
+		"group": "new group",
+	}}
+	rc, err = yaml.Marshal(theirNewConfig)
+	assert.NoError(t, err)
+	theirConfig.ReloadConfigString(string(rc))
+
+	r.Log("Spin until there is only 1 tunnel")
+	for len(myControl.GetHostmap().Indexes)+len(theirControl.GetHostmap().Indexes) > 2 {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		t.Log("Connection manager hasn't ticked yet")
+		time.Sleep(time.Second)
+	}
+
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+	myFinalHostmapHosts := myControl.ListHostmapHosts(false)
+	myFinalHostmapIndexes := myControl.ListHostmapIndexes(false)
+	theirFinalHostmapHosts := theirControl.ListHostmapHosts(false)
+	theirFinalHostmapIndexes := theirControl.ListHostmapIndexes(false)
+
+	// Make sure the correct tunnel won
+	c := theirControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(myVpnIpNet.IP), false)
+	assert.Contains(t, c.Cert.Details.Groups, "new group")
+
+	// We should only have a single tunnel now on both sides
+	assert.Len(t, myFinalHostmapHosts, 1)
+	assert.Len(t, theirFinalHostmapHosts, 1)
+	assert.Len(t, myFinalHostmapIndexes, 1)
+	assert.Len(t, theirFinalHostmapIndexes, 1)
+
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+
+	myControl.Stop()
+	theirControl.Stop()
+}
+
+func TestRehandshakingLoser(t *testing.T) {
+	// The purpose of this test is that the race loser renews their certificate and rehandshakes. The final tunnel
+	// Should be the one with the new certificate
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, myConfig := newSimpleServer(ca, caKey, "me  ", net.IP{10, 0, 0, 2}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, theirConfig := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 1}, nil)
+
+	// Put their info in our lighthouse and vice versa
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Build a router so we don't have to reason who gets which packet
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	t.Log("Stand up a tunnel between me and them")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+
+	tt1 := myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(theirVpnIpNet.IP), false)
+	tt2 := theirControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(myVpnIpNet.IP), false)
+	fmt.Println(tt1.LocalIndex, tt2.LocalIndex)
+
+	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
+
+	r.Log("Renew their certificate and spin until mine sees it")
+	_, _, theirNextPrivKey, theirNextPEM := newTestCert(ca, caKey, "them", time.Now(), time.Now().Add(5*time.Minute), theirVpnIpNet, nil, []string{"their new group"})
+
+	caB, err := ca.MarshalToPEM()
+	if err != nil {
+		panic(err)
+	}
+
+	theirConfig.Settings["pki"] = m{
+		"ca":   string(caB),
+		"cert": string(theirNextPEM),
+		"key":  string(theirNextPrivKey),
+	}
+	rc, err := yaml.Marshal(theirConfig.Settings)
+	assert.NoError(t, err)
+	theirConfig.ReloadConfigString(string(rc))
+
+	for {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		theirCertInMe := myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(theirVpnIpNet.IP), false)
+
+		_, theirNewGroup := theirCertInMe.Cert.Details.InvertedGroups["their new group"]
+		if theirNewGroup {
+			break
+		}
+
+		time.Sleep(time.Second)
+	}
+
+	// Flip my firewall to only allowing the new group to catch the tunnels reverting incorrectly
+	rc, err = yaml.Marshal(myConfig.Settings)
+	assert.NoError(t, err)
+	var myNewConfig m
+	assert.NoError(t, yaml.Unmarshal(rc, &myNewConfig))
+	theirFirewall := myNewConfig["firewall"].(map[interface{}]interface{})
+	theirFirewall["inbound"] = []m{{
+		"proto": "any",
+		"port":  "any",
+		"group": "their new group",
+	}}
+	rc, err = yaml.Marshal(myNewConfig)
+	assert.NoError(t, err)
+	myConfig.ReloadConfigString(string(rc))
+
+	r.Log("Spin until there is only 1 tunnel")
+	for len(myControl.GetHostmap().Indexes)+len(theirControl.GetHostmap().Indexes) > 2 {
+		assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+		t.Log("Connection manager hasn't ticked yet")
+		time.Sleep(time.Second)
+	}
+
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+	myFinalHostmapHosts := myControl.ListHostmapHosts(false)
+	myFinalHostmapIndexes := myControl.ListHostmapIndexes(false)
+	theirFinalHostmapHosts := theirControl.ListHostmapHosts(false)
+	theirFinalHostmapIndexes := theirControl.ListHostmapIndexes(false)
+
+	// Make sure the correct tunnel won
+	theirCertInMe := myControl.GetHostInfoByVpnIp(iputil.Ip2VpnIp(theirVpnIpNet.IP), false)
+	assert.Contains(t, theirCertInMe.Cert.Details.Groups, "their new group")
+
+	// We should only have a single tunnel now on both sides
+	assert.Len(t, myFinalHostmapHosts, 1)
+	assert.Len(t, theirFinalHostmapHosts, 1)
+	assert.Len(t, myFinalHostmapIndexes, 1)
+	assert.Len(t, theirFinalHostmapIndexes, 1)
+
+	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
+	myControl.Stop()
+	theirControl.Stop()
+}
+
+func TestRaceRegression(t *testing.T) {
+	// This test forces stage 1, stage 2, stage 1 to be received by me from them
+	// We had a bug where we were not finding the duplicate handshake and responding to the final stage 1 which
+	// caused a cross-linked hostinfo
+	ca, _, caKey, _ := newTestCaCert(time.Now(), time.Now().Add(10*time.Minute), []*net.IPNet{}, []*net.IPNet{}, []string{})
+	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(ca, caKey, "me", net.IP{10, 0, 0, 1}, nil)
+	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(ca, caKey, "them", net.IP{10, 0, 0, 2}, nil)
+
+	// Put their info in our lighthouse
+	myControl.InjectLightHouseAddr(theirVpnIpNet.IP, theirUdpAddr)
+	theirControl.InjectLightHouseAddr(myVpnIpNet.IP, myUdpAddr)
+
+	// Start the servers
+	myControl.Start()
+	theirControl.Start()
+
+	//them rx stage:1 initiatorIndex=642843150 responderIndex=0
+	//me rx   stage:1 initiatorIndex=120607833 responderIndex=0
+	//them rx stage:1 initiatorIndex=642843150 responderIndex=0
+	//me rx   stage:2 initiatorIndex=642843150 responderIndex=3701775874
+	//me rx   stage:1 initiatorIndex=120607833 responderIndex=0
+	//them rx stage:2 initiatorIndex=120607833 responderIndex=4209862089
+
+	t.Log("Start both handshakes")
+	myControl.InjectTunUDPPacket(theirVpnIpNet.IP, 80, 80, []byte("Hi from me"))
+	theirControl.InjectTunUDPPacket(myVpnIpNet.IP, 80, 80, []byte("Hi from them"))
+
+	t.Log("Get both stage 1")
+	myStage1ForThem := myControl.GetFromUDP(true)
+	theirStage1ForMe := theirControl.GetFromUDP(true)
+
+	t.Log("Inject them in a special way")
+	theirControl.InjectUDPPacket(myStage1ForThem)
+	myControl.InjectUDPPacket(theirStage1ForMe)
+	theirControl.InjectUDPPacket(myStage1ForThem)
+
+	//TODO: ensure stage 2
+	t.Log("Get both stage 2")
+	myStage2ForThem := myControl.GetFromUDP(true)
+	theirStage2ForMe := theirControl.GetFromUDP(true)
+
+	t.Log("Inject them in a special way again")
+	myControl.InjectUDPPacket(theirStage2ForMe)
+	myControl.InjectUDPPacket(theirStage1ForMe)
+	theirControl.InjectUDPPacket(myStage2ForThem)
+
+	r := router.NewR(t, myControl, theirControl)
+	defer r.RenderFlow()
+
+	t.Log("Flush the packets")
+	r.RouteForAllUntilTxTun(myControl)
+	r.RouteForAllUntilTxTun(theirControl)
+	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
+
+	t.Log("Make sure the tunnel still works")
+	assertTunnel(t, myVpnIpNet.IP, theirVpnIpNet.IP, myControl, theirControl, r)
+
+	myControl.Stop()
+	theirControl.Stop()
+}
+
+//TODO: test
+// Race winner renews and handshakes
+// Race loser renews and handshakes
+// Does race winner repin the cert to old?
 //TODO: add a test with many lies

e2e/helpers_test.go

@@ -1,13 +1,12 @@
+//go:build e2e_testing
 // +build e2e_testing
 
 package e2e
 
 import (
 	"crypto/rand"
-	"encoding/binary"
 	"fmt"
 	"io"
-	"io/ioutil"
 	"net"
 	"os"
 	"testing"
@@ -15,10 +14,13 @@ import (
 
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
+	"github.com/imdario/mergo"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/e2e/router"
+	"github.com/slackhq/nebula/iputil"
 	"github.com/stretchr/testify/assert"
 	"golang.org/x/crypto/curve25519"
 	"golang.org/x/crypto/ed25519"
@@ -28,7 +30,7 @@ import (
 type m map[string]interface{}
 
 // newSimpleServer creates a nebula instance with many assumptions
-func newSimpleServer(caCrt *cert.NebulaCertificate, caKey []byte, name string, udpIp net.IP) (*nebula.Control, net.IP, *net.UDPAddr) {
+func newSimpleServer(caCrt *cert.NebulaCertificate, caKey []byte, name string, udpIp net.IP, overrides m) (*nebula.Control, *net.IPNet, *net.UDPAddr, *config.C) {
 	l := NewTestLogger()
 
 	vpnIpNet := &net.IPNet{IP: make([]byte, len(udpIp)), Mask: net.IPMask{255, 255, 255, 0}}
@@ -75,22 +77,35 @@ func newSimpleServer(caCrt *cert.NebulaCertificate, caKey []byte, name string, u
 			"timestamp_format": fmt.Sprintf("%v 15:04:05.000000", name),
 			"level":            l.Level.String(),
 		},
+		"timers": m{
+			"pending_deletion_interval": 2,
+			"connection_alive_interval": 2,
+		},
 	}
+
+	if overrides != nil {
+		err = mergo.Merge(&overrides, mc, mergo.WithAppendSlice)
+		if err != nil {
+			panic(err)
+		}
+		mc = overrides
+	}
+
 	cb, err := yaml.Marshal(mc)
 	if err != nil {
 		panic(err)
 	}
 
-	config := nebula.NewConfig(l)
-	config.LoadString(string(cb))
+	c := config.NewC(l)
+	c.LoadString(string(cb))
 
-	control, err := nebula.Main(config, false, "e2e-test", l, nil)
+	control, err := nebula.Main(c, false, "e2e-test", l, nil)
 
 	if err != nil {
 		panic(err)
 	}
 
-	return control, vpnIpNet.IP, &udpAddr
+	return control, vpnIpNet, &udpAddr, c
 }
 
 // newTestCaCert will generate a CA cert
@@ -126,7 +141,7 @@ func newTestCaCert(before, after time.Time, ips, subnets []*net.IPNet, groups []
 		nc.Details.Groups = groups
 	}
 
-	err = nc.Sign(priv)
+	err = nc.Sign(cert.Curve_CURVE25519, priv)
 	if err != nil {
 		panic(err)
 	}
@@ -172,7 +187,7 @@ func newTestCert(ca *cert.NebulaCertificate, key []byte, name string, before, af
 		},
 	}
 
-	err = nc.Sign(key)
+	err = nc.Sign(ca.Details.Curve, key)
 	if err != nil {
 		panic(err)
 	}
@@ -186,25 +201,17 @@ func newTestCert(ca *cert.NebulaCertificate, key []byte, name string, before, af
 }
 
 func x25519Keypair() ([]byte, []byte) {
-	var pubkey, privkey [32]byte
-	if _, err := io.ReadFull(rand.Reader, privkey[:]); err != nil {
+	privkey := make([]byte, 32)
+	if _, err := io.ReadFull(rand.Reader, privkey); err != nil {
 		panic(err)
 	}
-	curve25519.ScalarBaseMult(&pubkey, &privkey)
-	return pubkey[:], privkey[:]
-}
 
-func ip2int(ip []byte) uint32 {
-	if len(ip) == 16 {
-		return binary.BigEndian.Uint32(ip[12:16])
+	pubkey, err := curve25519.X25519(privkey, curve25519.Basepoint)
+	if err != nil {
+		panic(err)
 	}
-	return binary.BigEndian.Uint32(ip)
-}
 
-func int2ip(nn uint32) net.IP {
-	ip := make(net.IP, 4)
-	binary.BigEndian.PutUint32(ip, nn)
-	return ip
+	return pubkey, privkey
 }
 
 type doneCb func()
@@ -228,26 +235,26 @@ func deadline(t *testing.T, seconds time.Duration) doneCb {
 func assertTunnel(t *testing.T, vpnIpA, vpnIpB net.IP, controlA, controlB *nebula.Control, r *router.R) {
 	// Send a packet from them to me
 	controlB.InjectTunUDPPacket(vpnIpA, 80, 90, []byte("Hi from B"))
-	bPacket := r.RouteUntilTxTun(controlB, controlA)
+	bPacket := r.RouteForAllUntilTxTun(controlA)
 	assertUdpPacket(t, []byte("Hi from B"), bPacket, vpnIpB, vpnIpA, 90, 80)
 
 	// And once more from me to them
 	controlA.InjectTunUDPPacket(vpnIpB, 80, 90, []byte("Hello from A"))
-	aPacket := r.RouteUntilTxTun(controlA, controlB)
+	aPacket := r.RouteForAllUntilTxTun(controlB)
 	assertUdpPacket(t, []byte("Hello from A"), aPacket, vpnIpA, vpnIpB, 90, 80)
 }
 
 func assertHostInfoPair(t *testing.T, addrA, addrB *net.UDPAddr, vpnIpA, vpnIpB net.IP, controlA, controlB *nebula.Control) {
 	// Get both host infos
-	hBinA := controlA.GetHostInfoByVpnIP(ip2int(vpnIpB), false)
-	assert.NotNil(t, hBinA, "Host B was not found by vpnIP in controlA")
+	hBinA := controlA.GetHostInfoByVpnIp(iputil.Ip2VpnIp(vpnIpB), false)
+	assert.NotNil(t, hBinA, "Host B was not found by vpnIp in controlA")
 
-	hAinB := controlB.GetHostInfoByVpnIP(ip2int(vpnIpA), false)
-	assert.NotNil(t, hAinB, "Host A was not found by vpnIP in controlB")
+	hAinB := controlB.GetHostInfoByVpnIp(iputil.Ip2VpnIp(vpnIpA), false)
+	assert.NotNil(t, hAinB, "Host A was not found by vpnIp in controlB")
 
 	// Check that both vpn and real addr are correct
-	assert.Equal(t, vpnIpB, hBinA.VpnIP, "Host B VpnIp is wrong in control A")
-	assert.Equal(t, vpnIpA, hAinB.VpnIP, "Host A VpnIp is wrong in control B")
+	assert.Equal(t, vpnIpB, hBinA.VpnIp, "Host B VpnIp is wrong in control A")
+	assert.Equal(t, vpnIpA, hAinB.VpnIp, "Host A VpnIp is wrong in control B")
 
 	assert.Equal(t, addrB.IP.To16(), hBinA.CurrentRemote.IP.To16(), "Host B remote ip is wrong in control A")
 	assert.Equal(t, addrA.IP.To16(), hAinB.CurrentRemote.IP.To16(), "Host A remote ip is wrong in control B")
@@ -300,7 +307,8 @@ func NewTestLogger() *logrus.Logger {
 
 	v := os.Getenv("TEST_LOGS")
 	if v == "" {
-		l.SetOutput(ioutil.Discard)
+		l.SetOutput(io.Discard)
+		l.SetLevel(logrus.PanicLevel)
 		return l
 	}
 

e2e/router/hostmap.go

@@ -0,0 +1,145 @@
+//go:build e2e_testing
+// +build e2e_testing
+
+package router
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+
+	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/iputil"
+)
+
+type edge struct {
+	from string
+	to   string
+	dual bool
+}
+
+func renderHostmaps(controls ...*nebula.Control) string {
+	var lines []*edge
+	r := "graph TB\n"
+	for _, c := range controls {
+		sr, se := renderHostmap(c)
+		r += sr
+		for _, e := range se {
+			add := true
+
+			// Collapse duplicate edges into a bi-directionally connected edge
+			for _, ge := range lines {
+				if e.to == ge.from && e.from == ge.to {
+					add = false
+					ge.dual = true
+					break
+				}
+			}
+
+			if add {
+				lines = append(lines, e)
+			}
+		}
+	}
+
+	for _, line := range lines {
+		if line.dual {
+			r += fmt.Sprintf("\t%v <--> %v\n", line.from, line.to)
+		} else {
+			r += fmt.Sprintf("\t%v --> %v\n", line.from, line.to)
+		}
+
+	}
+
+	return r
+}
+
+func renderHostmap(c *nebula.Control) (string, []*edge) {
+	var lines []string
+	var globalLines []*edge
+
+	clusterName := strings.Trim(c.GetCert().Details.Name, " ")
+	clusterVpnIp := c.GetCert().Details.Ips[0].IP
+	r := fmt.Sprintf("\tsubgraph %s[\"%s (%s)\"]\n", clusterName, clusterName, clusterVpnIp)
+
+	hm := c.GetHostmap()
+	hm.RLock()
+	defer hm.RUnlock()
+
+	// Draw the vpn to index nodes
+	r += fmt.Sprintf("\t\tsubgraph %s.hosts[\"Hosts (vpn ip to index)\"]\n", clusterName)
+	hosts := sortedHosts(hm.Hosts)
+	for _, vpnIp := range hosts {
+		hi := hm.Hosts[vpnIp]
+		r += fmt.Sprintf("\t\t\t%v.%v[\"%v\"]\n", clusterName, vpnIp, vpnIp)
+		lines = append(lines, fmt.Sprintf("%v.%v --> %v.%v", clusterName, vpnIp, clusterName, hi.GetLocalIndex()))
+
+		rs := hi.GetRelayState()
+		for _, relayIp := range rs.CopyRelayIps() {
+			lines = append(lines, fmt.Sprintf("%v.%v --> %v.%v", clusterName, vpnIp, clusterName, relayIp))
+		}
+
+		for _, relayIp := range rs.CopyRelayForIdxs() {
+			lines = append(lines, fmt.Sprintf("%v.%v --> %v.%v", clusterName, vpnIp, clusterName, relayIp))
+		}
+	}
+	r += "\t\tend\n"
+
+	// Draw the relay hostinfos
+	if len(hm.Relays) > 0 {
+		r += fmt.Sprintf("\t\tsubgraph %s.relays[\"Relays (relay index to hostinfo)\"]\n", clusterName)
+		for relayIndex, hi := range hm.Relays {
+			r += fmt.Sprintf("\t\t\t%v.%v[\"%v\"]\n", clusterName, relayIndex, relayIndex)
+			lines = append(lines, fmt.Sprintf("%v.%v --> %v.%v", clusterName, relayIndex, clusterName, hi.GetLocalIndex()))
+		}
+		r += "\t\tend\n"
+	}
+
+	// Draw the local index to relay or remote index nodes
+	r += fmt.Sprintf("\t\tsubgraph indexes.%s[\"Indexes (index to hostinfo)\"]\n", clusterName)
+	indexes := sortedIndexes(hm.Indexes)
+	for _, idx := range indexes {
+		hi, ok := hm.Indexes[idx]
+		if ok {
+			r += fmt.Sprintf("\t\t\t%v.%v[\"%v (%v)\"]\n", clusterName, idx, idx, hi.GetVpnIp())
+			remoteClusterName := strings.Trim(hi.GetCert().Details.Name, " ")
+			globalLines = append(globalLines, &edge{from: fmt.Sprintf("%v.%v", clusterName, idx), to: fmt.Sprintf("%v.%v", remoteClusterName, hi.GetRemoteIndex())})
+			_ = hi
+		}
+	}
+	r += "\t\tend\n"
+
+	// Add the edges inside this host
+	for _, line := range lines {
+		r += fmt.Sprintf("\t\t%v\n", line)
+	}
+
+	r += "\tend\n"
+	return r, globalLines
+}
+
+func sortedHosts(hosts map[iputil.VpnIp]*nebula.HostInfo) []iputil.VpnIp {
+	keys := make([]iputil.VpnIp, 0, len(hosts))
+	for key := range hosts {
+		keys = append(keys, key)
+	}
+
+	sort.SliceStable(keys, func(i, j int) bool {
+		return keys[i] > keys[j]
+	})
+
+	return keys
+}
+
+func sortedIndexes(indexes map[uint32]*nebula.HostInfo) []uint32 {
+	keys := make([]uint32, 0, len(indexes))
+	for key := range indexes {
+		keys = append(keys, key)
+	}
+
+	sort.SliceStable(keys, func(i, j int) bool {
+		return keys[i] > keys[j]
+	})
+
+	return keys
+}

e2e/router/router.go

@@ -1,15 +1,29 @@
+//go:build e2e_testing
 // +build e2e_testing
 
 package router
 
 import (
+	"context"
 	"fmt"
 	"net"
+	"os"
+	"path/filepath"
 	"reflect"
+	"sort"
 	"strconv"
+	"strings"
 	"sync"
+	"testing"
+	"time"
 
+	"github.com/google/gopacket"
+	"github.com/google/gopacket/layers"
 	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
+	"golang.org/x/exp/maps"
 )
 
 type R struct {
@@ -25,38 +39,125 @@ type R struct {
 	// map[from address + ":" + to address] => ip:port to rewrite in the udp packet to receiver
 	outNat map[string]net.UDPAddr
 
+	// A map of vpn ip to the nebula control it belongs to
+	vpnControls map[iputil.VpnIp]*nebula.Control
+
+	ignoreFlows []ignoreFlow
+	flow        []flowEntry
+
+	// A set of additional mermaid graphs to draw in the flow log markdown file
+	// Currently consisting only of hostmap renders
+	additionalGraphs []mermaidGraph
+
 	// All interactions are locked to help serialize behavior
 	sync.Mutex
+
+	fn           string
+	cancelRender context.CancelFunc
+	t            testing.TB
+}
+
+type ignoreFlow struct {
+	tun         NullBool
+	messageType header.MessageType
+	subType     header.MessageSubType
+	//from
+	//to
+}
+
+type mermaidGraph struct {
+	title   string
+	content string
+}
+
+type NullBool struct {
+	HasValue bool
+	IsTrue   bool
+}
+
+type flowEntry struct {
+	note   string
+	packet *packet
+}
+
+type packet struct {
+	from   *nebula.Control
+	to     *nebula.Control
+	packet *udp.Packet
+	tun    bool // a packet pulled off a tun device
+	rx     bool // the packet was received by a udp device
+}
+
+func (p *packet) WasReceived() {
+	if p != nil {
+		p.rx = true
+	}
 }
 
 type ExitType int
 
 const (
-	// Keeps routing, the function will get called again on the next packet
+	// KeepRouting the function will get called again on the next packet
 	KeepRouting ExitType = 0
-	// Does not route this packet and exits immediately
+	// ExitNow does not route this packet and exits immediately
 	ExitNow ExitType = 1
-	// Routes this packet and exits immediately afterwards
+	// RouteAndExit routes this packet and exits immediately afterwards
 	RouteAndExit ExitType = 2
 )
 
-type ExitFunc func(packet *nebula.UdpPacket, receiver *nebula.Control) ExitType
+type ExitFunc func(packet *udp.Packet, receiver *nebula.Control) ExitType
 
-func NewR(controls ...*nebula.Control) *R {
-	r := &R{
-		controls: make(map[string]*nebula.Control),
-		inNat:    make(map[string]*nebula.Control),
-		outNat:   make(map[string]net.UDPAddr),
+// NewR creates a new router to pass packets in a controlled fashion between the provided controllers.
+// The packet flow will be recorded in a file within the mermaid directory under the same name as the test.
+// Renders will occur automatically, roughly every 100ms, until a call to RenderFlow() is made
+func NewR(t testing.TB, controls ...*nebula.Control) *R {
+	ctx, cancel := context.WithCancel(context.Background())
+
+	if err := os.MkdirAll("mermaid", 0755); err != nil {
+		panic(err)
 	}
 
+	r := &R{
+		controls:     make(map[string]*nebula.Control),
+		vpnControls:  make(map[iputil.VpnIp]*nebula.Control),
+		inNat:        make(map[string]*nebula.Control),
+		outNat:       make(map[string]net.UDPAddr),
+		flow:         []flowEntry{},
+		ignoreFlows:  []ignoreFlow{},
+		fn:           filepath.Join("mermaid", fmt.Sprintf("%s.md", t.Name())),
+		t:            t,
+		cancelRender: cancel,
+	}
+
+	// Try to remove our render file
+	os.Remove(r.fn)
+
 	for _, c := range controls {
 		addr := c.GetUDPAddr()
 		if _, ok := r.controls[addr]; ok {
 			panic("Duplicate listen address: " + addr)
 		}
+
+		r.vpnControls[c.GetVpnIp()] = c
 		r.controls[addr] = c
 	}
 
+	// Spin the renderer in case we go nuts and the test never completes
+	go func() {
+		clockSource := time.NewTicker(time.Millisecond * 100)
+		defer clockSource.Stop()
+
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			case <-clockSource.C:
+				r.renderHostmaps("clock tick")
+				r.renderFlow()
+			}
+		}
+	}()
+
 	return r
 }
 
@@ -75,10 +176,227 @@ func (r *R) AddRoute(ip net.IP, port uint16, c *nebula.Control) {
 	r.inNat[inAddr] = c
 }
 
+// RenderFlow renders the packet flow seen up until now and stops further automatic renders from happening.
+func (r *R) RenderFlow() {
+	r.cancelRender()
+	r.renderFlow()
+}
+
+// CancelFlowLogs stops flow logs from being tracked and destroys any logs already collected
+func (r *R) CancelFlowLogs() {
+	r.cancelRender()
+	r.flow = nil
+}
+
+func (r *R) renderFlow() {
+	if r.flow == nil {
+		return
+	}
+
+	f, err := os.OpenFile(r.fn, os.O_CREATE|os.O_TRUNC|os.O_RDWR, 0644)
+	if err != nil {
+		panic(err)
+	}
+
+	var participants = map[string]struct{}{}
+	var participantsVals []string
+
+	fmt.Fprintln(f, "```mermaid")
+	fmt.Fprintln(f, "sequenceDiagram")
+
+	// Assemble participants
+	for _, e := range r.flow {
+		if e.packet == nil {
+			continue
+		}
+
+		addr := e.packet.from.GetUDPAddr()
+		if _, ok := participants[addr]; ok {
+			continue
+		}
+		participants[addr] = struct{}{}
+		sanAddr := strings.Replace(addr, ":", "-", 1)
+		participantsVals = append(participantsVals, sanAddr)
+		fmt.Fprintf(
+			f, "    participant %s as Nebula: %s<br/>UDP: %s\n",
+			sanAddr, e.packet.from.GetVpnIp(), sanAddr,
+		)
+	}
+
+	if len(participantsVals) > 2 {
+		// Get the first and last participantVals for notes
+		participantsVals = []string{participantsVals[0], participantsVals[len(participantsVals)-1]}
+	}
+
+	// Print packets
+	h := &header.H{}
+	for _, e := range r.flow {
+		if e.packet == nil {
+			//fmt.Fprintf(f, "    note over %s: %s\n", strings.Join(participantsVals, ", "), e.note)
+			continue
+		}
+
+		p := e.packet
+		if p.tun {
+			fmt.Fprintln(f, r.formatUdpPacket(p))
+
+		} else {
+			if err := h.Parse(p.packet.Data); err != nil {
+				panic(err)
+			}
+
+			line := "--x"
+			if p.rx {
+				line = "->>"
+			}
+
+			fmt.Fprintf(f,
+				"    %s%s%s: %s(%s), index %v, counter: %v\n",
+				strings.Replace(p.from.GetUDPAddr(), ":", "-", 1),
+				line,
+				strings.Replace(p.to.GetUDPAddr(), ":", "-", 1),
+				h.TypeName(), h.SubTypeName(), h.RemoteIndex, h.MessageCounter,
+			)
+		}
+	}
+	fmt.Fprintln(f, "```")
+
+	for _, g := range r.additionalGraphs {
+		fmt.Fprintf(f, "## %s\n", g.title)
+		fmt.Fprintln(f, "```mermaid")
+		fmt.Fprintln(f, g.content)
+		fmt.Fprintln(f, "```")
+	}
+}
+
+// IgnoreFlow tells the router to stop recording future flows that matches the provided criteria.
+// messageType and subType will target nebula underlay packets while tun will target nebula overlay packets
+// NOTE: This is a very broad system, if you set tun to true then no more tun traffic will be rendered
+func (r *R) IgnoreFlow(messageType header.MessageType, subType header.MessageSubType, tun NullBool) {
+	r.Lock()
+	defer r.Unlock()
+	r.ignoreFlows = append(r.ignoreFlows, ignoreFlow{
+		tun,
+		messageType,
+		subType,
+	})
+}
+
+func (r *R) RenderHostmaps(title string, controls ...*nebula.Control) {
+	r.Lock()
+	defer r.Unlock()
+
+	s := renderHostmaps(controls...)
+	if len(r.additionalGraphs) > 0 {
+		lastGraph := r.additionalGraphs[len(r.additionalGraphs)-1]
+		if lastGraph.content == s && lastGraph.title == title {
+			// Ignore this rendering if it matches the last rendering added
+			// This is useful if you want to track rendering changes
+			return
+		}
+	}
+
+	r.additionalGraphs = append(r.additionalGraphs, mermaidGraph{
+		title:   title,
+		content: s,
+	})
+}
+
+func (r *R) renderHostmaps(title string) {
+	c := maps.Values(r.controls)
+	sort.SliceStable(c, func(i, j int) bool {
+		return c[i].GetVpnIp() > c[j].GetVpnIp()
+	})
+
+	s := renderHostmaps(c...)
+	if len(r.additionalGraphs) > 0 {
+		lastGraph := r.additionalGraphs[len(r.additionalGraphs)-1]
+		if lastGraph.content == s {
+			// Ignore this rendering if it matches the last rendering added
+			// This is useful if you want to track rendering changes
+			return
+		}
+	}
+
+	r.additionalGraphs = append(r.additionalGraphs, mermaidGraph{
+		title:   title,
+		content: s,
+	})
+}
+
+// InjectFlow can be used to record packet flow if the test is handling the routing on its own.
+// The packet is assumed to have been received
+func (r *R) InjectFlow(from, to *nebula.Control, p *udp.Packet) {
+	r.Lock()
+	defer r.Unlock()
+	r.unlockedInjectFlow(from, to, p, false)
+}
+
+func (r *R) Log(arg ...any) {
+	if r.flow == nil {
+		return
+	}
+
+	r.Lock()
+	r.flow = append(r.flow, flowEntry{note: fmt.Sprint(arg...)})
+	r.t.Log(arg...)
+	r.Unlock()
+}
+
+func (r *R) Logf(format string, arg ...any) {
+	if r.flow == nil {
+		return
+	}
+
+	r.Lock()
+	r.flow = append(r.flow, flowEntry{note: fmt.Sprintf(format, arg...)})
+	r.t.Logf(format, arg...)
+	r.Unlock()
+}
+
+// unlockedInjectFlow is used by the router to record a packet has been transmitted, the packet is returned and
+// should be marked as received AFTER it has been placed on the receivers channel.
+// If flow logs have been disabled this function will return nil
+func (r *R) unlockedInjectFlow(from, to *nebula.Control, p *udp.Packet, tun bool) *packet {
+	if r.flow == nil {
+		return nil
+	}
+
+	r.renderHostmaps(fmt.Sprintf("Packet %v", len(r.flow)))
+
+	if len(r.ignoreFlows) > 0 {
+		var h header.H
+		err := h.Parse(p.Data)
+		if err != nil {
+			panic(err)
+		}
+
+		for _, i := range r.ignoreFlows {
+			if !tun {
+				if i.messageType == h.Type && i.subType == h.Subtype {
+					return nil
+				}
+			} else if i.tun.HasValue && i.tun.IsTrue {
+				return nil
+			}
+		}
+	}
+
+	fp := &packet{
+		from:   from,
+		to:     to,
+		packet: p.Copy(),
+		tun:    tun,
+	}
+
+	r.flow = append(r.flow, flowEntry{packet: fp})
+	return fp
+}
+
 // OnceFrom will route a single packet from sender then return
 // If the router doesn't have the nebula controller for that address, we panic
 func (r *R) OnceFrom(sender *nebula.Control) {
-	r.RouteExitFunc(sender, func(*nebula.UdpPacket, *nebula.Control) ExitType {
+	r.RouteExitFunc(sender, func(*udp.Packet, *nebula.Control) ExitType {
 		return RouteAndExit
 	})
 }
@@ -93,6 +411,11 @@ func (r *R) RouteUntilTxTun(sender *nebula.Control, receiver *nebula.Control) []
 		select {
 		// Maybe we already have something on the tun for us
 		case b := <-tunTx:
+			r.Lock()
+			np := udp.Packet{Data: make([]byte, len(b))}
+			copy(np.Data, b)
+			r.unlockedInjectFlow(receiver, receiver, &np, true)
+			r.Unlock()
 			return b
 
 		// Nope, lets push the sender along
@@ -105,20 +428,80 @@ func (r *R) RouteUntilTxTun(sender *nebula.Control, receiver *nebula.Control) []
 				r.Unlock()
 				panic("No control for udp tx")
 			}
-
+			fp := r.unlockedInjectFlow(sender, c, p, false)
 			c.InjectUDPPacket(p)
+			fp.WasReceived()
 			r.Unlock()
 		}
 	}
 }
 
+// RouteForAllUntilTxTun will route for everyone and return when a packet is seen on receivers tun
+// If the router doesn't have the nebula controller for that address, we panic
+func (r *R) RouteForAllUntilTxTun(receiver *nebula.Control) []byte {
+	sc := make([]reflect.SelectCase, len(r.controls)+1)
+	cm := make([]*nebula.Control, len(r.controls)+1)
+
+	i := 0
+	sc[i] = reflect.SelectCase{
+		Dir:  reflect.SelectRecv,
+		Chan: reflect.ValueOf(receiver.GetTunTxChan()),
+		Send: reflect.Value{},
+	}
+	cm[i] = receiver
+
+	i++
+	for _, c := range r.controls {
+		sc[i] = reflect.SelectCase{
+			Dir:  reflect.SelectRecv,
+			Chan: reflect.ValueOf(c.GetUDPTxChan()),
+			Send: reflect.Value{},
+		}
+
+		cm[i] = c
+		i++
+	}
+
+	for {
+		x, rx, _ := reflect.Select(sc)
+		r.Lock()
+
+		if x == 0 {
+			// we are the tun tx, we can exit
+			p := rx.Interface().([]byte)
+			np := udp.Packet{Data: make([]byte, len(p))}
+			copy(np.Data, p)
+
+			r.unlockedInjectFlow(cm[x], cm[x], &np, true)
+			r.Unlock()
+			return p
+
+		} else {
+			// we are a udp tx, route and continue
+			p := rx.Interface().(*udp.Packet)
+			outAddr := cm[x].GetUDPAddr()
+
+			inAddr := net.JoinHostPort(p.ToIp.String(), fmt.Sprintf("%v", p.ToPort))
+			c := r.getControl(outAddr, inAddr, p)
+			if c == nil {
+				r.Unlock()
+				panic("No control for udp tx")
+			}
+			fp := r.unlockedInjectFlow(cm[x], c, p, false)
+			c.InjectUDPPacket(p)
+			fp.WasReceived()
+		}
+		r.Unlock()
+	}
+}
+
 // RouteExitFunc will call the whatDo func with each udp packet from sender.
 // whatDo can return:
 //   - exitNow: the packet will not be routed and this call will return immediately
 //   - routeAndExit: this call will return immediately after routing the last packet from sender
 //   - keepRouting: the packet will be routed and whatDo will be called again on the next packet from sender
 func (r *R) RouteExitFunc(sender *nebula.Control, whatDo ExitFunc) {
-	h := &nebula.Header{}
+	h := &header.H{}
 	for {
 		p := sender.GetFromUDP(true)
 		r.Lock()
@@ -141,12 +524,16 @@ func (r *R) RouteExitFunc(sender *nebula.Control, whatDo ExitFunc) {
 			return
 
 		case RouteAndExit:
+			fp := r.unlockedInjectFlow(sender, receiver, p, false)
 			receiver.InjectUDPPacket(p)
+			fp.WasReceived()
 			r.Unlock()
 			return
 
 		case KeepRouting:
+			fp := r.unlockedInjectFlow(sender, receiver, p, false)
 			receiver.InjectUDPPacket(p)
+			fp.WasReceived()
 
 		default:
 			panic(fmt.Sprintf("Unknown exitFunc return: %v", e))
@@ -158,9 +545,9 @@ func (r *R) RouteExitFunc(sender *nebula.Control, whatDo ExitFunc) {
 
 // RouteUntilAfterMsgType will route for sender until a message type is seen and sent from sender
 // If the router doesn't have the nebula controller for that address, we panic
-func (r *R) RouteUntilAfterMsgType(sender *nebula.Control, msgType nebula.NebulaMessageType, subType nebula.NebulaMessageSubType) {
-	h := &nebula.Header{}
-	r.RouteExitFunc(sender, func(p *nebula.UdpPacket, r *nebula.Control) ExitType {
+func (r *R) RouteUntilAfterMsgType(sender *nebula.Control, msgType header.MessageType, subType header.MessageSubType) {
+	h := &header.H{}
+	r.RouteExitFunc(sender, func(p *udp.Packet, r *nebula.Control) ExitType {
 		if err := h.Parse(p.Data); err != nil {
 			panic(err)
 		}
@@ -172,6 +559,34 @@ func (r *R) RouteUntilAfterMsgType(sender *nebula.Control, msgType nebula.Nebula
 	})
 }
 
+func (r *R) RouteForAllUntilAfterMsgTypeTo(receiver *nebula.Control, msgType header.MessageType, subType header.MessageSubType) {
+	h := &header.H{}
+	r.RouteForAllExitFunc(func(p *udp.Packet, r *nebula.Control) ExitType {
+		if r != receiver {
+			return KeepRouting
+		}
+
+		if err := h.Parse(p.Data); err != nil {
+			panic(err)
+		}
+
+		if h.Type == msgType && h.Subtype == subType {
+			return RouteAndExit
+		}
+
+		return KeepRouting
+	})
+}
+
+func (r *R) InjectUDPPacket(sender, receiver *nebula.Control, packet *udp.Packet) {
+	r.Lock()
+	defer r.Unlock()
+
+	fp := r.unlockedInjectFlow(sender, receiver, packet, false)
+	receiver.InjectUDPPacket(packet)
+	fp.WasReceived()
+}
+
 // RouteForUntilAfterToAddr will route for sender and return only after it sees and sends a packet destined for toAddr
 // finish can be any of the exitType values except `keepRouting`, the default value is `routeAndExit`
 // If the router doesn't have the nebula controller for that address, we panic
@@ -180,7 +595,7 @@ func (r *R) RouteForUntilAfterToAddr(sender *nebula.Control, toAddr *net.UDPAddr
 		finish = RouteAndExit
 	}
 
-	r.RouteExitFunc(sender, func(p *nebula.UdpPacket, r *nebula.Control) ExitType {
+	r.RouteExitFunc(sender, func(p *udp.Packet, r *nebula.Control) ExitType {
 		if p.ToIp.Equal(toAddr.IP) && p.ToPort == uint16(toAddr.Port) {
 			return finish
 		}
@@ -214,7 +629,7 @@ func (r *R) RouteForAllExitFunc(whatDo ExitFunc) {
 		x, rx, _ := reflect.Select(sc)
 		r.Lock()
 
-		p := rx.Interface().(*nebula.UdpPacket)
+		p := rx.Interface().(*udp.Packet)
 
 		outAddr := cm[x].GetUDPAddr()
 		inAddr := net.JoinHostPort(p.ToIp.String(), fmt.Sprintf("%v", p.ToPort))
@@ -231,12 +646,16 @@ func (r *R) RouteForAllExitFunc(whatDo ExitFunc) {
 			return
 
 		case RouteAndExit:
+			fp := r.unlockedInjectFlow(cm[x], receiver, p, false)
 			receiver.InjectUDPPacket(p)
+			fp.WasReceived()
 			r.Unlock()
 			return
 
 		case KeepRouting:
+			fp := r.unlockedInjectFlow(cm[x], receiver, p, false)
 			receiver.InjectUDPPacket(p)
+			fp.WasReceived()
 
 		default:
 			panic(fmt.Sprintf("Unknown exitFunc return: %v", e))
@@ -276,7 +695,7 @@ func (r *R) FlushAll() {
 		}
 		r.Lock()
 
-		p := rx.Interface().(*nebula.UdpPacket)
+		p := rx.Interface().(*udp.Packet)
 
 		outAddr := cm[x].GetUDPAddr()
 		inAddr := net.JoinHostPort(p.ToIp.String(), fmt.Sprintf("%v", p.ToPort))
@@ -291,7 +710,7 @@ func (r *R) FlushAll() {
 
 // getControl performs or seeds NAT translation and returns the control for toAddr, p from fields may change
 // This is an internal router function, the caller must hold the lock
-func (r *R) getControl(fromAddr, toAddr string, p *nebula.UdpPacket) *nebula.Control {
+func (r *R) getControl(fromAddr, toAddr string, p *udp.Packet) *nebula.Control {
 	if newAddr, ok := r.outNat[fromAddr+":"+toAddr]; ok {
 		p.FromIp = newAddr.IP
 		p.FromPort = uint16(newAddr.Port)
@@ -318,3 +737,31 @@ func (r *R) getControl(fromAddr, toAddr string, p *nebula.UdpPacket) *nebula.Con
 
 	return r.controls[toAddr]
 }
+
+func (r *R) formatUdpPacket(p *packet) string {
+	packet := gopacket.NewPacket(p.packet.Data, layers.LayerTypeIPv4, gopacket.Lazy)
+	v4 := packet.Layer(layers.LayerTypeIPv4).(*layers.IPv4)
+	if v4 == nil {
+		panic("not an ipv4 packet")
+	}
+
+	from := "unknown"
+	if c, ok := r.vpnControls[iputil.Ip2VpnIp(v4.SrcIP)]; ok {
+		from = c.GetUDPAddr()
+	}
+
+	udp := packet.Layer(layers.LayerTypeUDP).(*layers.UDP)
+	if udp == nil {
+		panic("not a udp packet")
+	}
+
+	data := packet.ApplicationLayer()
+	return fmt.Sprintf(
+		"    %s-->>%s: src port: %v<br/>dest port: %v<br/>data: \"%v\"\n",
+		strings.Replace(from, ":", "-", 1),
+		strings.Replace(p.to.GetUDPAddr(), ":", "-", 1),
+		udp.SrcPort,
+		udp.DstPort,
+		string(data.Payload()),
+	)
+}

examples/config.yml

@@ -7,9 +7,11 @@ pki:
   ca: /etc/nebula/ca.crt
   cert: /etc/nebula/host.crt
   key: /etc/nebula/host.key
-  #blocklist is a list of certificate fingerprints that we will refuse to talk to
+  # blocklist is a list of certificate fingerprints that we will refuse to talk to
   #blocklist:
   #  - c99d4e650533b92061b09918e838a5a0a6aaee21eed1d12fd937682865936c72
+  # disconnect_invalid is a toggle to force a client to be disconnected if the certificate is expired or invalid.
+  #disconnect_invalid: false
 
 # 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.
@@ -45,8 +47,9 @@ lighthouse:
   # allowed. You can provide CIDRs here with `true` to allow and `false` to
   # deny. The most specific CIDR rule applies to each remote. If all rules are
   # "allow", the default will be "deny", and vice-versa. If both "allow" and
-  # "deny" rules are present, then you MUST set a rule for "0.0.0.0/0" as the
-  # default.
+  # "deny" IPv4 rules are present, then you MUST set a rule for "0.0.0.0/0" as
+  # the default. Similarly if both "allow" and "deny" IPv6 rules are present,
+  # then you MUST set a rule for "::/0" as the default.
   #remote_allow_list:
     # Example to block IPs from this subnet from being used for remote IPs.
     #"172.16.0.0/12": false
@@ -56,6 +59,14 @@ lighthouse:
     #"10.0.0.0/8": false
     #"10.42.42.0/24": true
 
+  # EXPERIMENTAL: This option may change or disappear in the future.
+  # Optionally allows the definition of remote_allow_list blocks
+  # specific to an inside VPN IP CIDR.
+  #remote_allow_ranges:
+    # This rule would only allow only private IPs for this VPN range
+    #"10.42.42.0/24":
+      #"192.168.0.0/16": true
+
   # local_allow_list allows you to filter which local IP addresses we advertise
   # to the lighthouses. This uses the same logic as `remote_allow_list`, but
   # additionally, you can specify an `interfaces` map of regular expressions
@@ -71,10 +82,32 @@ lighthouse:
     # Example to only advertise this subnet to the lighthouse.
     #"10.0.0.0/8": true
 
+  # advertise_addrs are routable addresses that will be included along with discovered addresses to report to the
+  # lighthouse, the format is "ip:port". `port` can be `0`, in which case the actual listening port will be used in its
+  # place, useful if `listen.port` is set to 0.
+  # This option is mainly useful when there are static ip addresses the host can be reached at that nebula can not
+  # typically discover on its own. Examples being port forwarding or multiple paths to the internet.
+  #advertise_addrs:
+    #- "1.1.1.1:4242"
+    #- "1.2.3.4:0" # port will be replaced with the real listening port
+
+  # EXPERIMENTAL: This option may change or disappear in the future.
+  # This setting allows us to "guess" what the remote might be for a host
+  # while we wait for the lighthouse response.
+  #calculated_remotes:
+    # For any Nebula IPs in 10.0.10.0/24, this will apply the mask and add
+    # the calculated IP as an initial remote (while we wait for the response
+    # from the lighthouse). Both CIDRs must have the same mask size.
+    # For example, Nebula IP 10.0.10.123 will have a calculated remote of
+    # 192.168.1.123
+    #10.0.10.0/24:
+      #- mask: 192.168.1.0/24
+      #  port: 4242
+
 # Port Nebula will be listening on. The default here is 4242. For a lighthouse node, the port should be defined,
 # however using port 0 will dynamically assign a port and is recommended for roaming nodes.
 listen:
-  # To listen on both any ipv4 and ipv6 use "[::]"
+  # To listen on both any ipv4 and ipv6 use "::"
   host: 0.0.0.0
   port: 4242
   # Sets the max number of packets to pull from the kernel for each syscall (under systems that support recvmmsg)
@@ -86,14 +119,18 @@ listen:
   # max, net.core.rmem_max and net.core.wmem_max
   #read_buffer: 10485760
   #write_buffer: 10485760
+  # By default, Nebula replies to packets it has no tunnel for with a "recv_error" packet. This packet helps speed up reconnection
+  # in the case that Nebula on either side did not shut down cleanly. This response can be abused as a way to discover if Nebula is running
+  # on a host though. This option lets you configure if you want to send "recv_error" packets always, never, or only to private network remotes.
+  # valid values: always, never, private
+  # This setting is reloadable.
+  #send_recv_error: always
 
-# EXPERIMENTAL: This option is currently only supported on linux and may
-# change in future minor releases.
-#
 # Routines is the number of thread pairs to run that consume from the tun and UDP queues.
 # Currently, this defaults to 1 which means we have 1 tun queue reader and 1
 # UDP queue reader. Setting this above one will set IFF_MULTI_QUEUE on the tun
 # device and SO_REUSEPORT on the UDP socket to allow multiple queues.
+# This option is only supported on Linux.
 #routines: 1
 
 punchy:
@@ -105,16 +142,21 @@ punchy:
   # Default is false
   #respond: true
 
-  # delays a punch response for misbehaving NATs, default is 1 second, respond must be true to take effect
+  # delays a punch response for misbehaving NATs, default is 1 second.
   #delay: 1s
 
+  # set the delay before attempting punchy.respond. Default is 5 seconds. respond must be true to take effect.
+  #respond_delay: 5s
+
 # Cipher allows you to choose between the available ciphers for your network. Options are chachapoly or aes
 # IMPORTANT: this value must be identical on ALL NODES/LIGHTHOUSES. We do not/will not support use of different ciphers simultaneously!
-#cipher: chachapoly
+#cipher: aes
 
-# Local range is used to define a hint about the local network range, which speeds up discovering the fastest
+# Preferred ranges is used to define a hint about the local network ranges, which speeds up discovering the fastest
 # path to a network adjacent nebula node.
-#local_range: "172.16.0.0/24"
+# NOTE: the previous option "local_range" only allowed definition of a single range
+# and has been deprecated for "preferred_ranges"
+#preferred_ranges: ["172.16.0.0/24"]
 
 # sshd can expose informational and administrative functions via ssh this is a
 #sshd:
@@ -132,11 +174,27 @@ punchy:
       #keys:
         #- "ssh public key string"
 
+# EXPERIMENTAL: relay support for networks that can't establish direct connections.
+relay:
+  # Relays are a list of Nebula IP's that peers can use to relay packets to me.
+  # IPs in this list must have am_relay set to true in their configs, otherwise
+  # they will reject relay requests.
+  #relays:
+    #- 192.168.100.1
+    #- <other Nebula VPN IPs of hosts used as relays to access me>
+  # Set am_relay to true to permit other hosts to list my IP in their relays config. Default false.
+  am_relay: false
+  # Set use_relays to false to prevent this instance from attempting to establish connections through relays.
+  # default true
+  use_relays: true
+
 # Configure the private interface. Note: addr is baked into the nebula certificate
 tun:
   # When tun is disabled, a lighthouse can be started without a local tun interface (and therefore without root)
   disabled: false
-  # Name of the device
+  # Name of the device. If not set, a default will be chosen by the OS.
+  # For macOS: if set, must be in the form `utun[0-9]+`.
+  # For FreeBSD: Required to be set, must be in the form `tun[0-9]+`.
   dev: nebula1
   # Toggles forwarding of local broadcast packets, the address of which depends on the ip/mask encoded in pki.cert
   drop_local_broadcast: false
@@ -146,18 +204,28 @@ tun:
   tx_queue: 500
   # Default MTU for every packet, safe setting is (and the default) 1300 for internet based traffic
   mtu: 1300
+
   # Route based MTU overrides, you have known vpn ip paths that can support larger MTUs you can increase/decrease them here
   routes:
     #- mtu: 8800
     #  route: 10.0.0.0/16
+
   # Unsafe routes allows you to route traffic over nebula to non-nebula nodes
   # Unsafe routes should be avoided unless you have hosts/services that cannot run nebula
   # NOTE: The nebula certificate of the "via" node *MUST* have the "route" defined as a subnet in its certificate
+  # `mtu`: will default to tun mtu if this option is not specified
+  # `metric`: will default to 0 if this option is not specified
+  # `install`: will default to true, controls whether this route is installed in the systems routing table.
   unsafe_routes:
     #- route: 172.16.1.0/24
     #  via: 192.168.100.99
-    #  mtu: 1300 #mtu will default to tun mtu if this option is not sepcified
+    #  mtu: 1300
+    #  metric: 100
+    #  install: true
 
+  # 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
 
 # TODO
 # Configure logging level
@@ -200,7 +268,7 @@ logging:
   #   e.g.: `lighthouse.rx.HostQuery`
   #lighthouse_metrics: false
 
-# Handshake Manger Settings
+# Handshake Manager Settings
 #handshakes:
   # Handshakes are sent to all known addresses at each interval with a linear backoff,
   # Wait try_interval after the 1st attempt, 2 * try_interval after the 2nd, etc, until the handshake is older than timeout
@@ -214,11 +282,19 @@ logging:
 
 # Nebula security group configuration
 firewall:
+  # Action to take when a packet is not allowed by the firewall rules.
+  # Can be one of:
+  #   `drop` (default): silently drop the packet.
+  #   `reject`: send a reject reply.
+  #     - For TCP, this will be a RST "Connection Reset" packet.
+  #     - For other protocols, this will be an ICMP port unreachable packet.
+  outbound_action: drop
+  inbound_action: drop
+
   conntrack:
     tcp_timeout: 12m
     udp_timeout: 3m
     default_timeout: 10m
-    max_connections: 100000
 
   # The firewall is default deny. There is no way to write a deny rule.
   # Rules are comprised of a protocol, port, and one or more of host, group, or CIDR
@@ -229,7 +305,8 @@ firewall:
   #   host: `any` or a literal hostname, ie `test-host`
   #   group: `any` or a literal group name, ie `default-group`
   #   groups: Same as group but accepts a list of values. Multiple values are AND'd together and a certificate would have to contain all groups to pass
-  #   cidr: a CIDR, `0.0.0.0/0` is any.
+  #   cidr: a remote CIDR, `0.0.0.0/0` is any.
+  #   local_cidr: a local CIDR, `0.0.0.0/0` is any. This could be used to filter destinations when using unsafe_routes.
   #   ca_name: An issuing CA name
   #   ca_sha: An issuing CA shasum
 

examples/quickstart-vagrant/README.md

@@ -119,7 +119,6 @@ firewall:
     tcp_timeout: 12m
     udp_timeout: 3m
     default_timeout: 10m
-    max_connections: 100,000
 
   inbound:
     - proto: icmp

examples/quickstart-vagrant/ansible/roles/nebula/files/systemd.nebula.service

@@ -1,7 +1,8 @@
 [Unit]
-Description=nebula
-Wants=basic.target
-After=basic.target network.target
+Description=Nebula overlay networking tool
+Wants=basic.target network-online.target nss-lookup.target time-sync.target
+After=basic.target network.target network-online.target
+Before=sshd.service
 
 [Service]
 SyslogIdentifier=nebula

examples/quickstart-vagrant/ansible/roles/nebula/templates/config.yml.j2

@@ -65,7 +65,6 @@ firewall:
     tcp_timeout: 12m
     udp_timeout: 3m
     default_timeout: 10m
-    max_connections: 100,000
 
   inbound:
     - proto: icmp

examples/service_scripts/nebula.plist

@@ -0,0 +1,34 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+    <dict>
+        <key>KeepAlive</key>
+        <true/>
+        <key>Label</key>
+        <string>net.defined.nebula</string>
+        <key>WorkingDirectory</key>
+        <string>/Users/{username}/.local/bin/nebula</string>
+        <key>LimitLoadToSessionType</key>
+        <array>
+            <string>Aqua</string>
+            <string>Background</string>
+            <string>LoginWindow</string>
+            <string>StandardIO</string>
+            <string>System</string>
+        </array>
+        <key>ProgramArguments</key>
+        <array>
+            <string>./nebula</string>
+            <string>-config</string>
+            <string>./config.yml</string>
+        </array>
+        <key>RunAtLoad</key>
+        <true/>
+        <key>StandardErrorPath</key>
+        <string>./nebula.log</string>
+        <key>StandardOutPath</key>
+        <string>./nebula.log</string>
+        <key>UserName</key>
+        <string>root</string>
+    </dict>
+</plist>

examples/service_scripts/nebula.service

@@ -1,7 +1,7 @@
 [Unit]
-Description=nebula
-Wants=basic.target
-After=basic.target network.target
+Description=Nebula overlay networking tool
+Wants=basic.target network-online.target nss-lookup.target time-sync.target
+After=basic.target network.target network-online.target
 Before=sshd.service
 
 [Service]

firewall.go

@@ -4,7 +4,6 @@ import (
 	"crypto/sha256"
 	"encoding/binary"
 	"encoding/hex"
-	"encoding/json"
 	"errors"
 	"fmt"
 	"net"
@@ -12,29 +11,21 @@ import (
 	"strconv"
 	"strings"
 	"sync"
-	"sync/atomic"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
-)
-
-const (
-	fwProtoAny  = 0 // When we want to handle HOPOPT (0) we can change this, if ever
-	fwProtoTCP  = 6
-	fwProtoUDP  = 17
-	fwProtoICMP = 1
-
-	fwPortAny      = 0  // Special value for matching `port: any`
-	fwPortFragment = -1 // Special value for matching `port: fragment`
+	"github.com/slackhq/nebula/cidr"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/firewall"
 )
 
 const tcpACK = 0x10
 const tcpFIN = 0x01
 
 type FirewallInterface interface {
-	AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, caName string, caSha string) error
+	AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, localIp *net.IPNet, caName string, caSha string) error
 }
 
 type conn struct {
@@ -56,6 +47,9 @@ type Firewall struct {
 	InRules  *FirewallTable
 	OutRules *FirewallTable
 
+	InSendReject  bool
+	OutSendReject bool
+
 	//TODO: we should have many more options for TCP, an option for ICMP, and mimic the kernel a bit better
 	// https://www.kernel.org/doc/Documentation/networking/nf_conntrack-sysctl.txt
 	TCPTimeout     time.Duration //linux: 5 days max
@@ -63,7 +57,7 @@ type Firewall struct {
 	DefaultTimeout time.Duration //linux: 600s
 
 	// Used to ensure we don't emit local packets for ips we don't own
-	localIps *CIDRTree
+	localIps *cidr.Tree4
 
 	rules        string
 	rulesVersion uint16
@@ -85,8 +79,8 @@ type firewallMetrics struct {
 type FirewallConntrack struct {
 	sync.Mutex
 
-	Conns      map[FirewallPacket]*conn
-	TimerWheel *TimerWheel
+	Conns      map[firewall.Packet]*conn
+	TimerWheel *TimerWheel[firewall.Packet]
 }
 
 type FirewallTable struct {
@@ -112,59 +106,18 @@ type FirewallCA struct {
 }
 
 type FirewallRule struct {
-	// Any makes Hosts, Groups, and CIDR irrelevant
-	Any    bool
-	Hosts  map[string]struct{}
-	Groups [][]string
-	CIDR   *CIDRTree
+	// Any makes Hosts, Groups, CIDR and LocalCIDR irrelevant
+	Any       bool
+	Hosts     map[string]struct{}
+	Groups    [][]string
+	CIDR      *cidr.Tree4
+	LocalCIDR *cidr.Tree4
 }
 
 // Even though ports are uint16, int32 maps are faster for lookup
 // Plus we can use `-1` for fragment rules
 type firewallPort map[int32]*FirewallCA
 
-type FirewallPacket struct {
-	LocalIP    uint32
-	RemoteIP   uint32
-	LocalPort  uint16
-	RemotePort uint16
-	Protocol   uint8
-	Fragment   bool
-}
-
-func (fp *FirewallPacket) Copy() *FirewallPacket {
-	return &FirewallPacket{
-		LocalIP:    fp.LocalIP,
-		RemoteIP:   fp.RemoteIP,
-		LocalPort:  fp.LocalPort,
-		RemotePort: fp.RemotePort,
-		Protocol:   fp.Protocol,
-		Fragment:   fp.Fragment,
-	}
-}
-
-func (fp FirewallPacket) MarshalJSON() ([]byte, error) {
-	var proto string
-	switch fp.Protocol {
-	case fwProtoTCP:
-		proto = "tcp"
-	case fwProtoICMP:
-		proto = "icmp"
-	case fwProtoUDP:
-		proto = "udp"
-	default:
-		proto = fmt.Sprintf("unknown %v", fp.Protocol)
-	}
-	return json.Marshal(m{
-		"LocalIP":    int2ip(fp.LocalIP).String(),
-		"RemoteIP":   int2ip(fp.RemoteIP).String(),
-		"LocalPort":  fp.LocalPort,
-		"RemotePort": fp.RemotePort,
-		"Protocol":   proto,
-		"Fragment":   fp.Fragment,
-	})
-}
-
 // NewFirewall creates a new Firewall object. A TimerWheel is created for you from the provided timeouts.
 func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.Duration, c *cert.NebulaCertificate) *Firewall {
 	//TODO: error on 0 duration
@@ -184,7 +137,7 @@ func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.D
 		max = defaultTimeout
 	}
 
-	localIps := NewCIDRTree()
+	localIps := cidr.NewTree4()
 	for _, ip := range c.Details.Ips {
 		localIps.AddCIDR(&net.IPNet{IP: ip.IP, Mask: net.IPMask{255, 255, 255, 255}}, struct{}{})
 	}
@@ -195,8 +148,8 @@ func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.D
 
 	return &Firewall{
 		Conntrack: &FirewallConntrack{
-			Conns:      make(map[FirewallPacket]*conn),
-			TimerWheel: NewTimerWheel(min, max),
+			Conns:      make(map[firewall.Packet]*conn),
+			TimerWheel: NewTimerWheel[firewall.Packet](min, max),
 		},
 		InRules:        newFirewallTable(),
 		OutRules:       newFirewallTable(),
@@ -220,7 +173,7 @@ func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.D
 	}
 }
 
-func NewFirewallFromConfig(l *logrus.Logger, nc *cert.NebulaCertificate, c *Config) (*Firewall, error) {
+func NewFirewallFromConfig(l *logrus.Logger, nc *cert.NebulaCertificate, c *config.C) (*Firewall, error) {
 	fw := NewFirewall(
 		l,
 		c.GetDuration("firewall.conntrack.tcp_timeout", time.Minute*12),
@@ -230,6 +183,28 @@ func NewFirewallFromConfig(l *logrus.Logger, nc *cert.NebulaCertificate, c *Conf
 		//TODO: max_connections
 	)
 
+	inboundAction := c.GetString("firewall.inbound_action", "drop")
+	switch inboundAction {
+	case "reject":
+		fw.InSendReject = true
+	case "drop":
+		fw.InSendReject = false
+	default:
+		l.WithField("action", inboundAction).Warn("invalid firewall.inbound_action, defaulting to `drop`")
+		fw.InSendReject = false
+	}
+
+	outboundAction := c.GetString("firewall.outbound_action", "drop")
+	switch outboundAction {
+	case "reject":
+		fw.OutSendReject = true
+	case "drop":
+		fw.OutSendReject = false
+	default:
+		l.WithField("action", inboundAction).Warn("invalid firewall.outbound_action, defaulting to `drop`")
+		fw.OutSendReject = false
+	}
+
 	err := AddFirewallRulesFromConfig(l, false, c, fw)
 	if err != nil {
 		return nil, err
@@ -244,18 +219,22 @@ func NewFirewallFromConfig(l *logrus.Logger, nc *cert.NebulaCertificate, c *Conf
 }
 
 // AddRule properly creates the in memory rule structure for a firewall table.
-func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, caName string, caSha string) error {
+func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, localIp *net.IPNet, caName string, caSha string) error {
 	// Under gomobile, stringing a nil pointer with fmt causes an abort in debug mode for iOS
 	// https://github.com/golang/go/issues/14131
 	sIp := ""
 	if ip != nil {
 		sIp = ip.String()
 	}
+	lIp := ""
+	if localIp != nil {
+		lIp = localIp.String()
+	}
 
 	// We need this rule string because we generate a hash. Removing this will break firewall reload.
 	ruleString := fmt.Sprintf(
-		"incoming: %v, proto: %v, startPort: %v, endPort: %v, groups: %v, host: %v, ip: %v, caName: %v, caSha: %s",
-		incoming, proto, startPort, endPort, groups, host, sIp, caName, caSha,
+		"incoming: %v, proto: %v, startPort: %v, endPort: %v, groups: %v, host: %v, ip: %v, localIp: %v, caName: %v, caSha: %s",
+		incoming, proto, startPort, endPort, groups, host, sIp, lIp, caName, caSha,
 	)
 	f.rules += ruleString + "\n"
 
@@ -263,7 +242,7 @@ func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort
 	if !incoming {
 		direction = "outgoing"
 	}
-	f.l.WithField("firewallRule", m{"direction": direction, "proto": proto, "startPort": startPort, "endPort": endPort, "groups": groups, "host": host, "ip": sIp, "caName": caName, "caSha": caSha}).
+	f.l.WithField("firewallRule", m{"direction": direction, "proto": proto, "startPort": startPort, "endPort": endPort, "groups": groups, "host": host, "ip": sIp, "localIp": lIp, "caName": caName, "caSha": caSha}).
 		Info("Firewall rule added")
 
 	var (
@@ -278,19 +257,19 @@ func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort
 	}
 
 	switch proto {
-	case fwProtoTCP:
+	case firewall.ProtoTCP:
 		fp = ft.TCP
-	case fwProtoUDP:
+	case firewall.ProtoUDP:
 		fp = ft.UDP
-	case fwProtoICMP:
+	case firewall.ProtoICMP:
 		fp = ft.ICMP
-	case fwProtoAny:
+	case firewall.ProtoAny:
 		fp = ft.AnyProto
 	default:
 		return fmt.Errorf("unknown protocol %v", proto)
 	}
 
-	return fp.addRule(startPort, endPort, groups, host, ip, caName, caSha)
+	return fp.addRule(startPort, endPort, groups, host, ip, localIp, caName, caSha)
 }
 
 // GetRuleHash returns a hash representation of all inbound and outbound rules
@@ -299,7 +278,7 @@ func (f *Firewall) GetRuleHash() string {
 	return hex.EncodeToString(sum[:])
 }
 
-func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, config *Config, fw FirewallInterface) error {
+func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, c *config.C, fw FirewallInterface) error {
 	var table string
 	if inbound {
 		table = "firewall.inbound"
@@ -307,7 +286,7 @@ func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, config *Config,
 		table = "firewall.outbound"
 	}
 
-	r := config.Get(table)
+	r := c.Get(table)
 	if r == nil {
 		return nil
 	}
@@ -328,8 +307,8 @@ func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, config *Config,
 			return fmt.Errorf("%s rule #%v; only one of port or code should be provided", table, i)
 		}
 
-		if r.Host == "" && len(r.Groups) == 0 && r.Group == "" && r.Cidr == "" && r.CAName == "" && r.CASha == "" {
-			return fmt.Errorf("%s rule #%v; at least one of host, group, cidr, ca_name, or ca_sha must be provided", table, i)
+		if r.Host == "" && len(r.Groups) == 0 && r.Group == "" && r.Cidr == "" && r.LocalCidr == "" && r.CAName == "" && r.CASha == "" {
+			return fmt.Errorf("%s rule #%v; at least one of host, group, cidr, local_cidr, ca_name, or ca_sha must be provided", table, i)
 		}
 
 		if len(r.Groups) > 0 {
@@ -362,13 +341,13 @@ func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, config *Config,
 		var proto uint8
 		switch r.Proto {
 		case "any":
-			proto = fwProtoAny
+			proto = firewall.ProtoAny
 		case "tcp":
-			proto = fwProtoTCP
+			proto = firewall.ProtoTCP
 		case "udp":
-			proto = fwProtoUDP
+			proto = firewall.ProtoUDP
 		case "icmp":
-			proto = fwProtoICMP
+			proto = firewall.ProtoICMP
 		default:
 			return fmt.Errorf("%s rule #%v; proto was not understood; `%s`", table, i, r.Proto)
 		}
@@ -381,7 +360,15 @@ func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, config *Config,
 			}
 		}
 
-		err = fw.AddRule(inbound, proto, startPort, endPort, groups, r.Host, cidr, r.CAName, r.CASha)
+		var localCidr *net.IPNet
+		if r.LocalCidr != "" {
+			_, localCidr, err = net.ParseCIDR(r.LocalCidr)
+			if err != nil {
+				return fmt.Errorf("%s rule #%v; local_cidr did not parse; %s", table, i, err)
+			}
+		}
+
+		err = fw.AddRule(inbound, proto, startPort, endPort, groups, r.Host, cidr, localCidr, r.CAName, r.CASha)
 		if err != nil {
 			return fmt.Errorf("%s rule #%v; `%s`", table, i, err)
 		}
@@ -396,7 +383,7 @@ var ErrNoMatchingRule = errors.New("no matching rule in firewall table")
 
 // Drop returns an error if the packet should be dropped, explaining why. It
 // returns nil if the packet should not be dropped.
-func (f *Firewall) Drop(packet []byte, fp FirewallPacket, incoming bool, h *HostInfo, caPool *cert.NebulaCAPool, localCache ConntrackCache) error {
+func (f *Firewall) Drop(packet []byte, fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.NebulaCAPool, localCache firewall.ConntrackCache) error {
 	// Check if we spoke to this tuple, if we did then allow this packet
 	if f.inConns(packet, fp, incoming, h, caPool, localCache) {
 		return nil
@@ -410,7 +397,7 @@ func (f *Firewall) Drop(packet []byte, fp FirewallPacket, incoming bool, h *Host
 		}
 	} else {
 		// Simple case: Certificate has one IP and no subnets
-		if fp.RemoteIP != h.hostId {
+		if fp.RemoteIP != h.vpnIp {
 			f.metrics(incoming).droppedRemoteIP.Inc(1)
 			return ErrInvalidRemoteIP
 		}
@@ -462,7 +449,7 @@ func (f *Firewall) EmitStats() {
 	metrics.GetOrRegisterGauge("firewall.rules.version", nil).Update(int64(f.rulesVersion))
 }
 
-func (f *Firewall) inConns(packet []byte, fp FirewallPacket, incoming bool, h *HostInfo, caPool *cert.NebulaCAPool, localCache ConntrackCache) bool {
+func (f *Firewall) inConns(packet []byte, fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.NebulaCAPool, localCache firewall.ConntrackCache) bool {
 	if localCache != nil {
 		if _, ok := localCache[fp]; ok {
 			return true
@@ -520,14 +507,14 @@ func (f *Firewall) inConns(packet []byte, fp FirewallPacket, incoming bool, h *H
 	}
 
 	switch fp.Protocol {
-	case fwProtoTCP:
+	case firewall.ProtoTCP:
 		c.Expires = time.Now().Add(f.TCPTimeout)
 		if incoming {
 			f.checkTCPRTT(c, packet)
 		} else {
 			setTCPRTTTracking(c, packet)
 		}
-	case fwProtoUDP:
+	case firewall.ProtoUDP:
 		c.Expires = time.Now().Add(f.UDPTimeout)
 	default:
 		c.Expires = time.Now().Add(f.DefaultTimeout)
@@ -542,17 +529,17 @@ func (f *Firewall) inConns(packet []byte, fp FirewallPacket, incoming bool, h *H
 	return true
 }
 
-func (f *Firewall) addConn(packet []byte, fp FirewallPacket, incoming bool) {
+func (f *Firewall) addConn(packet []byte, fp firewall.Packet, incoming bool) {
 	var timeout time.Duration
 	c := &conn{}
 
 	switch fp.Protocol {
-	case fwProtoTCP:
+	case firewall.ProtoTCP:
 		timeout = f.TCPTimeout
 		if !incoming {
 			setTCPRTTTracking(c, packet)
 		}
-	case fwProtoUDP:
+	case firewall.ProtoUDP:
 		timeout = f.UDPTimeout
 	default:
 		timeout = f.DefaultTimeout
@@ -561,6 +548,7 @@ func (f *Firewall) addConn(packet []byte, fp FirewallPacket, incoming bool) {
 	conntrack := f.Conntrack
 	conntrack.Lock()
 	if _, ok := conntrack.Conns[fp]; !ok {
+		conntrack.TimerWheel.Advance(time.Now())
 		conntrack.TimerWheel.Add(fp, timeout)
 	}
 
@@ -575,7 +563,7 @@ func (f *Firewall) addConn(packet []byte, fp FirewallPacket, incoming bool) {
 
 // Evict checks if a conntrack entry has expired, if so it is removed, if not it is re-added to the wheel
 // Caller must own the connMutex lock!
-func (f *Firewall) evict(p FirewallPacket) {
+func (f *Firewall) evict(p firewall.Packet) {
 	//TODO: report a stat if the tcp rtt tracking was never resolved?
 	// Are we still tracking this conn?
 	conntrack := f.Conntrack
@@ -588,6 +576,7 @@ func (f *Firewall) evict(p FirewallPacket) {
 
 	// Timeout is in the future, re-add the timer
 	if newT > 0 {
+		conntrack.TimerWheel.Advance(time.Now())
 		conntrack.TimerWheel.Add(p, newT)
 		return
 	}
@@ -596,21 +585,21 @@ func (f *Firewall) evict(p FirewallPacket) {
 	delete(conntrack.Conns, p)
 }
 
-func (ft *FirewallTable) match(p FirewallPacket, incoming bool, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
+func (ft *FirewallTable) match(p firewall.Packet, incoming bool, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
 	if ft.AnyProto.match(p, incoming, c, caPool) {
 		return true
 	}
 
 	switch p.Protocol {
-	case fwProtoTCP:
+	case firewall.ProtoTCP:
 		if ft.TCP.match(p, incoming, c, caPool) {
 			return true
 		}
-	case fwProtoUDP:
+	case firewall.ProtoUDP:
 		if ft.UDP.match(p, incoming, c, caPool) {
 			return true
 		}
-	case fwProtoICMP:
+	case firewall.ProtoICMP:
 		if ft.ICMP.match(p, incoming, c, caPool) {
 			return true
 		}
@@ -619,7 +608,7 @@ func (ft *FirewallTable) match(p FirewallPacket, incoming bool, c *cert.NebulaCe
 	return false
 }
 
-func (fp firewallPort) addRule(startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, caName string, caSha string) error {
+func (fp firewallPort) addRule(startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, localIp *net.IPNet, caName string, caSha string) error {
 	if startPort > endPort {
 		return fmt.Errorf("start port was lower than end port")
 	}
@@ -632,7 +621,7 @@ func (fp firewallPort) addRule(startPort int32, endPort int32, groups []string,
 			}
 		}
 
-		if err := fp[i].addRule(groups, host, ip, caName, caSha); err != nil {
+		if err := fp[i].addRule(groups, host, ip, localIp, caName, caSha); err != nil {
 			return err
 		}
 	}
@@ -640,7 +629,7 @@ func (fp firewallPort) addRule(startPort int32, endPort int32, groups []string,
 	return nil
 }
 
-func (fp firewallPort) match(p FirewallPacket, incoming bool, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
+func (fp firewallPort) match(p firewall.Packet, incoming bool, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
 	// We don't have any allowed ports, bail
 	if fp == nil {
 		return false
@@ -649,7 +638,7 @@ func (fp firewallPort) match(p FirewallPacket, incoming bool, c *cert.NebulaCert
 	var port int32
 
 	if p.Fragment {
-		port = fwPortFragment
+		port = firewall.PortFragment
 	} else if incoming {
 		port = int32(p.LocalPort)
 	} else {
@@ -660,15 +649,16 @@ func (fp firewallPort) match(p FirewallPacket, incoming bool, c *cert.NebulaCert
 		return true
 	}
 
-	return fp[fwPortAny].match(p, c, caPool)
+	return fp[firewall.PortAny].match(p, c, caPool)
 }
 
-func (fc *FirewallCA) addRule(groups []string, host string, ip *net.IPNet, caName, caSha string) error {
+func (fc *FirewallCA) addRule(groups []string, host string, ip, localIp *net.IPNet, caName, caSha string) error {
 	fr := func() *FirewallRule {
 		return &FirewallRule{
-			Hosts:  make(map[string]struct{}),
-			Groups: make([][]string, 0),
-			CIDR:   NewCIDRTree(),
+			Hosts:     make(map[string]struct{}),
+			Groups:    make([][]string, 0),
+			CIDR:      cidr.NewTree4(),
+			LocalCIDR: cidr.NewTree4(),
 		}
 	}
 
@@ -677,14 +667,14 @@ func (fc *FirewallCA) addRule(groups []string, host string, ip *net.IPNet, caNam
 			fc.Any = fr()
 		}
 
-		return fc.Any.addRule(groups, host, ip)
+		return fc.Any.addRule(groups, host, ip, localIp)
 	}
 
 	if caSha != "" {
 		if _, ok := fc.CAShas[caSha]; !ok {
 			fc.CAShas[caSha] = fr()
 		}
-		err := fc.CAShas[caSha].addRule(groups, host, ip)
+		err := fc.CAShas[caSha].addRule(groups, host, ip, localIp)
 		if err != nil {
 			return err
 		}
@@ -694,7 +684,7 @@ func (fc *FirewallCA) addRule(groups []string, host string, ip *net.IPNet, caNam
 		if _, ok := fc.CANames[caName]; !ok {
 			fc.CANames[caName] = fr()
 		}
-		err := fc.CANames[caName].addRule(groups, host, ip)
+		err := fc.CANames[caName].addRule(groups, host, ip, localIp)
 		if err != nil {
 			return err
 		}
@@ -703,7 +693,7 @@ func (fc *FirewallCA) addRule(groups []string, host string, ip *net.IPNet, caNam
 	return nil
 }
 
-func (fc *FirewallCA) match(p FirewallPacket, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
+func (fc *FirewallCA) match(p firewall.Packet, c *cert.NebulaCertificate, caPool *cert.NebulaCAPool) bool {
 	if fc == nil {
 		return false
 	}
@@ -726,17 +716,18 @@ func (fc *FirewallCA) match(p FirewallPacket, c *cert.NebulaCertificate, caPool
 	return fc.CANames[s.Details.Name].match(p, c)
 }
 
-func (fr *FirewallRule) addRule(groups []string, host string, ip *net.IPNet) error {
+func (fr *FirewallRule) addRule(groups []string, host string, ip *net.IPNet, localIp *net.IPNet) error {
 	if fr.Any {
 		return nil
 	}
 
-	if fr.isAny(groups, host, ip) {
+	if fr.isAny(groups, host, ip, localIp) {
 		fr.Any = true
 		// If it's any we need to wipe out any pre-existing rules to save on memory
 		fr.Groups = make([][]string, 0)
 		fr.Hosts = make(map[string]struct{})
-		fr.CIDR = NewCIDRTree()
+		fr.CIDR = cidr.NewTree4()
+		fr.LocalCIDR = cidr.NewTree4()
 	} else {
 		if len(groups) > 0 {
 			fr.Groups = append(fr.Groups, groups)
@@ -749,13 +740,17 @@ func (fr *FirewallRule) addRule(groups []string, host string, ip *net.IPNet) err
 		if ip != nil {
 			fr.CIDR.AddCIDR(ip, struct{}{})
 		}
+
+		if localIp != nil {
+			fr.LocalCIDR.AddCIDR(localIp, struct{}{})
+		}
 	}
 
 	return nil
 }
 
-func (fr *FirewallRule) isAny(groups []string, host string, ip *net.IPNet) bool {
-	if len(groups) == 0 && host == "" && ip == nil {
+func (fr *FirewallRule) isAny(groups []string, host string, ip, localIp *net.IPNet) bool {
+	if len(groups) == 0 && host == "" && ip == nil && localIp == nil {
 		return true
 	}
 
@@ -773,10 +768,14 @@ func (fr *FirewallRule) isAny(groups []string, host string, ip *net.IPNet) bool
 		return true
 	}
 
+	if localIp != nil && localIp.Contains(net.IPv4(0, 0, 0, 0)) {
+		return true
+	}
+
 	return false
 }
 
-func (fr *FirewallRule) match(p FirewallPacket, c *cert.NebulaCertificate) bool {
+func (fr *FirewallRule) match(p firewall.Packet, c *cert.NebulaCertificate) bool {
 	if fr == nil {
 		return false
 	}
@@ -814,20 +813,25 @@ func (fr *FirewallRule) match(p FirewallPacket, c *cert.NebulaCertificate) bool
 		return true
 	}
 
+	if fr.LocalCIDR != nil && fr.LocalCIDR.Contains(p.LocalIP) != nil {
+		return true
+	}
+
 	// No host, group, or cidr matched, bye bye
 	return false
 }
 
 type rule struct {
-	Port   string
-	Code   string
-	Proto  string
-	Host   string
-	Group  string
-	Groups []string
-	Cidr   string
-	CAName string
-	CASha  string
+	Port      string
+	Code      string
+	Proto     string
+	Host      string
+	Group     string
+	Groups    []string
+	Cidr      string
+	LocalCidr string
+	CAName    string
+	CASha     string
 }
 
 func convertRule(l *logrus.Logger, p interface{}, table string, i int) (rule, error) {
@@ -851,6 +855,7 @@ func convertRule(l *logrus.Logger, p interface{}, table string, i int) (rule, er
 	r.Proto = toString("proto", m)
 	r.Host = toString("host", m)
 	r.Cidr = toString("cidr", m)
+	r.LocalCidr = toString("local_cidr", m)
 	r.CAName = toString("ca_name", m)
 	r.CASha = toString("ca_sha", m)
 
@@ -885,12 +890,12 @@ func convertRule(l *logrus.Logger, p interface{}, table string, i int) (rule, er
 
 func parsePort(s string) (startPort, endPort int32, err error) {
 	if s == "any" {
-		startPort = fwPortAny
-		endPort = fwPortAny
+		startPort = firewall.PortAny
+		endPort = firewall.PortAny
 
 	} else if s == "fragment" {
-		startPort = fwPortFragment
-		endPort = fwPortFragment
+		startPort = firewall.PortFragment
+		endPort = firewall.PortFragment
 
 	} else if strings.Contains(s, `-`) {
 		sPorts := strings.SplitN(s, `-`, 2)
@@ -914,8 +919,8 @@ func parsePort(s string) (startPort, endPort int32, err error) {
 		startPort = int32(rStartPort)
 		endPort = int32(rEndPort)
 
-		if startPort == fwPortAny {
-			endPort = fwPortAny
+		if startPort == firewall.PortAny {
+			endPort = firewall.PortAny
 		}
 
 	} else {
@@ -930,7 +935,7 @@ func parsePort(s string) (startPort, endPort int32, err error) {
 	return
 }
 
-//TODO: write tests for these
+// TODO: write tests for these
 func setTCPRTTTracking(c *conn, p []byte) {
 	if c.Seq != 0 {
 		return
@@ -968,54 +973,3 @@ func (f *Firewall) checkTCPRTT(c *conn, p []byte) bool {
 	c.Seq = 0
 	return true
 }
-
-// ConntrackCache is used as a local routine cache to know if a given flow
-// has been seen in the conntrack table.
-type ConntrackCache map[FirewallPacket]struct{}
-
-type ConntrackCacheTicker struct {
-	cacheV    uint64
-	cacheTick uint64
-
-	cache ConntrackCache
-}
-
-func NewConntrackCacheTicker(d time.Duration) *ConntrackCacheTicker {
-	if d == 0 {
-		return nil
-	}
-
-	c := &ConntrackCacheTicker{
-		cache: ConntrackCache{},
-	}
-
-	go c.tick(d)
-
-	return c
-}
-
-func (c *ConntrackCacheTicker) tick(d time.Duration) {
-	for {
-		time.Sleep(d)
-		atomic.AddUint64(&c.cacheTick, 1)
-	}
-}
-
-// Get checks if the cache ticker has moved to the next version before returning
-// the map. If it has moved, we reset the map.
-func (c *ConntrackCacheTicker) Get(l *logrus.Logger) ConntrackCache {
-	if c == nil {
-		return nil
-	}
-	if tick := atomic.LoadUint64(&c.cacheTick); tick != c.cacheV {
-		c.cacheV = tick
-		if ll := len(c.cache); ll > 0 {
-			if l.Level == logrus.DebugLevel {
-				l.WithField("len", ll).Debug("resetting conntrack cache")
-			}
-			c.cache = make(ConntrackCache, ll)
-		}
-	}
-
-	return c.cache
-}

firewall/cache.go

@@ -0,0 +1,59 @@
+package firewall
+
+import (
+	"sync/atomic"
+	"time"
+
+	"github.com/sirupsen/logrus"
+)
+
+// ConntrackCache is used as a local routine cache to know if a given flow
+// has been seen in the conntrack table.
+type ConntrackCache map[Packet]struct{}
+
+type ConntrackCacheTicker struct {
+	cacheV    uint64
+	cacheTick atomic.Uint64
+
+	cache ConntrackCache
+}
+
+func NewConntrackCacheTicker(d time.Duration) *ConntrackCacheTicker {
+	if d == 0 {
+		return nil
+	}
+
+	c := &ConntrackCacheTicker{
+		cache: ConntrackCache{},
+	}
+
+	go c.tick(d)
+
+	return c
+}
+
+func (c *ConntrackCacheTicker) tick(d time.Duration) {
+	for {
+		time.Sleep(d)
+		c.cacheTick.Add(1)
+	}
+}
+
+// Get checks if the cache ticker has moved to the next version before returning
+// the map. If it has moved, we reset the map.
+func (c *ConntrackCacheTicker) Get(l *logrus.Logger) ConntrackCache {
+	if c == nil {
+		return nil
+	}
+	if tick := c.cacheTick.Load(); tick != c.cacheV {
+		c.cacheV = tick
+		if ll := len(c.cache); ll > 0 {
+			if l.Level == logrus.DebugLevel {
+				l.WithField("len", ll).Debug("resetting conntrack cache")
+			}
+			c.cache = make(ConntrackCache, ll)
+		}
+	}
+
+	return c.cache
+}

firewall/packet.go

@@ -0,0 +1,62 @@
+package firewall
+
+import (
+	"encoding/json"
+	"fmt"
+
+	"github.com/slackhq/nebula/iputil"
+)
+
+type m map[string]interface{}
+
+const (
+	ProtoAny  = 0 // When we want to handle HOPOPT (0) we can change this, if ever
+	ProtoTCP  = 6
+	ProtoUDP  = 17
+	ProtoICMP = 1
+
+	PortAny      = 0  // Special value for matching `port: any`
+	PortFragment = -1 // Special value for matching `port: fragment`
+)
+
+type Packet struct {
+	LocalIP    iputil.VpnIp
+	RemoteIP   iputil.VpnIp
+	LocalPort  uint16
+	RemotePort uint16
+	Protocol   uint8
+	Fragment   bool
+}
+
+func (fp *Packet) Copy() *Packet {
+	return &Packet{
+		LocalIP:    fp.LocalIP,
+		RemoteIP:   fp.RemoteIP,
+		LocalPort:  fp.LocalPort,
+		RemotePort: fp.RemotePort,
+		Protocol:   fp.Protocol,
+		Fragment:   fp.Fragment,
+	}
+}
+
+func (fp Packet) MarshalJSON() ([]byte, error) {
+	var proto string
+	switch fp.Protocol {
+	case ProtoTCP:
+		proto = "tcp"
+	case ProtoICMP:
+		proto = "icmp"
+	case ProtoUDP:
+		proto = "udp"
+	default:
+		proto = fmt.Sprintf("unknown %v", fp.Protocol)
+	}
+	return json.Marshal(m{
+		"LocalIP":    fp.LocalIP.String(),
+		"RemoteIP":   fp.RemoteIP.String(),
+		"LocalPort":  fp.LocalPort,
+		"RemotePort": fp.RemotePort,
+		"Protocol":   proto,
+		"Fragment":   fp.Fragment,
+	})
+}

firewall_test.go

@@ -11,11 +11,15 @@ import (
 
 	"github.com/rcrowley/go-metrics"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/config"
+	"github.com/slackhq/nebula/firewall"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
 )
 
 func TestNewFirewall(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	c := &cert.NebulaCertificate{}
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	conntrack := fw.Conntrack
@@ -30,31 +34,31 @@ func TestNewFirewall(t *testing.T) {
 
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 
 	fw = NewFirewall(l, time.Second, time.Hour, time.Minute, c)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 
 	fw = NewFirewall(l, time.Hour, time.Second, time.Minute, c)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 
 	fw = NewFirewall(l, time.Hour, time.Minute, time.Second, c)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 
 	fw = NewFirewall(l, time.Minute, time.Hour, time.Second, c)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 
 	fw = NewFirewall(l, time.Minute, time.Second, time.Hour, c)
 	assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
-	assert.Equal(t, 3601, conntrack.TimerWheel.wheelLen)
+	assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
 }
 
 func TestFirewall_AddRule(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
@@ -65,93 +69,89 @@ func TestFirewall_AddRule(t *testing.T) {
 
 	_, ti, _ := net.ParseCIDR("1.2.3.4/32")
 
-	assert.Nil(t, fw.AddRule(true, fwProtoTCP, 1, 1, []string{}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoTCP, 1, 1, []string{}, "", nil, nil, "", ""))
 	// An empty rule is any
 	assert.True(t, fw.InRules.TCP[1].Any.Any)
 	assert.Empty(t, fw.InRules.TCP[1].Any.Groups)
 	assert.Empty(t, fw.InRules.TCP[1].Any.Hosts)
-	assert.Nil(t, fw.InRules.TCP[1].Any.CIDR.root.left)
-	assert.Nil(t, fw.InRules.TCP[1].Any.CIDR.root.right)
-	assert.Nil(t, fw.InRules.TCP[1].Any.CIDR.root.value)
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(true, fwProtoUDP, 1, 1, []string{"g1"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoUDP, 1, 1, []string{"g1"}, "", nil, nil, "", ""))
 	assert.False(t, fw.InRules.UDP[1].Any.Any)
 	assert.Contains(t, fw.InRules.UDP[1].Any.Groups[0], "g1")
 	assert.Empty(t, fw.InRules.UDP[1].Any.Hosts)
-	assert.Nil(t, fw.InRules.UDP[1].Any.CIDR.root.left)
-	assert.Nil(t, fw.InRules.UDP[1].Any.CIDR.root.right)
-	assert.Nil(t, fw.InRules.UDP[1].Any.CIDR.root.value)
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(true, fwProtoICMP, 1, 1, []string{}, "h1", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoICMP, 1, 1, []string{}, "h1", nil, nil, "", ""))
 	assert.False(t, fw.InRules.ICMP[1].Any.Any)
 	assert.Empty(t, fw.InRules.ICMP[1].Any.Groups)
 	assert.Contains(t, fw.InRules.ICMP[1].Any.Hosts, "h1")
-	assert.Nil(t, fw.InRules.ICMP[1].Any.CIDR.root.left)
-	assert.Nil(t, fw.InRules.ICMP[1].Any.CIDR.root.right)
-	assert.Nil(t, fw.InRules.ICMP[1].Any.CIDR.root.value)
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 1, 1, []string{}, "", ti, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", ti, nil, "", ""))
 	assert.False(t, fw.OutRules.AnyProto[1].Any.Any)
 	assert.Empty(t, fw.OutRules.AnyProto[1].Any.Groups)
 	assert.Empty(t, fw.OutRules.AnyProto[1].Any.Hosts)
-	assert.NotNil(t, fw.OutRules.AnyProto[1].Any.CIDR.Match(ip2int(ti.IP)))
+	assert.NotNil(t, fw.OutRules.AnyProto[1].Any.CIDR.Match(iputil.Ip2VpnIp(ti.IP)))
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(true, fwProtoUDP, 1, 1, []string{"g1"}, "", nil, "ca-name", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", nil, ti, "", ""))
+	assert.False(t, fw.OutRules.AnyProto[1].Any.Any)
+	assert.Empty(t, fw.OutRules.AnyProto[1].Any.Groups)
+	assert.Empty(t, fw.OutRules.AnyProto[1].Any.Hosts)
+	assert.NotNil(t, fw.OutRules.AnyProto[1].Any.LocalCIDR.Match(iputil.Ip2VpnIp(ti.IP)))
+
+	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoUDP, 1, 1, []string{"g1"}, "", nil, nil, "ca-name", ""))
 	assert.Contains(t, fw.InRules.UDP[1].CANames, "ca-name")
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(true, fwProtoUDP, 1, 1, []string{"g1"}, "", nil, "", "ca-sha"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoUDP, 1, 1, []string{"g1"}, "", nil, nil, "", "ca-sha"))
 	assert.Contains(t, fw.InRules.UDP[1].CAShas, "ca-sha")
 
 	// Set any and clear fields
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 0, 0, []string{"g1", "g2"}, "h1", ti, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{"g1", "g2"}, "h1", ti, ti, "", ""))
 	assert.Equal(t, []string{"g1", "g2"}, fw.OutRules.AnyProto[0].Any.Groups[0])
 	assert.Contains(t, fw.OutRules.AnyProto[0].Any.Hosts, "h1")
-	assert.NotNil(t, fw.OutRules.AnyProto[0].Any.CIDR.Match(ip2int(ti.IP)))
+	assert.NotNil(t, fw.OutRules.AnyProto[0].Any.CIDR.Match(iputil.Ip2VpnIp(ti.IP)))
+	assert.NotNil(t, fw.OutRules.AnyProto[0].Any.LocalCIDR.Match(iputil.Ip2VpnIp(ti.IP)))
 
 	// run twice just to make sure
 	//TODO: these ANY rules should clear the CA firewall portion
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 0, 0, []string{"any"}, "", nil, "", ""))
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 0, 0, []string{}, "any", nil, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{"any"}, "", nil, nil, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{}, "any", nil, nil, "", ""))
 	assert.True(t, fw.OutRules.AnyProto[0].Any.Any)
 	assert.Empty(t, fw.OutRules.AnyProto[0].Any.Groups)
 	assert.Empty(t, fw.OutRules.AnyProto[0].Any.Hosts)
-	assert.Nil(t, fw.OutRules.AnyProto[0].Any.CIDR.root.left)
-	assert.Nil(t, fw.OutRules.AnyProto[0].Any.CIDR.root.right)
-	assert.Nil(t, fw.OutRules.AnyProto[0].Any.CIDR.root.value)
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 0, 0, []string{}, "any", nil, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{}, "any", nil, nil, "", ""))
 	assert.True(t, fw.OutRules.AnyProto[0].Any.Any)
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	_, anyIp, _ := net.ParseCIDR("0.0.0.0/0")
-	assert.Nil(t, fw.AddRule(false, fwProtoAny, 0, 0, []string{}, "", anyIp, "", ""))
+	assert.Nil(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{}, "", anyIp, nil, "", ""))
 	assert.True(t, fw.OutRules.AnyProto[0].Any.Any)
 
 	// Test error conditions
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	assert.Error(t, fw.AddRule(true, math.MaxUint8, 0, 0, []string{}, "", nil, "", ""))
-	assert.Error(t, fw.AddRule(true, fwProtoAny, 10, 0, []string{}, "", nil, "", ""))
+	assert.Error(t, fw.AddRule(true, math.MaxUint8, 0, 0, []string{}, "", nil, nil, "", ""))
+	assert.Error(t, fw.AddRule(true, firewall.ProtoAny, 10, 0, []string{}, "", nil, nil, "", ""))
 }
 
 func TestFirewall_Drop(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
-	p := FirewallPacket{
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		10,
-		90,
-		fwProtoUDP,
-		false,
+	p := firewall.Packet{
+		LocalIP:    iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		RemoteIP:   iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		LocalPort:  10,
+		RemotePort: 90,
+		Protocol:   firewall.ProtoUDP,
+		Fragment:   false,
 	}
 
 	ipNet := net.IPNet{
@@ -172,12 +172,12 @@ func TestFirewall_Drop(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h.CreateRemoteCIDR(&c)
 
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"any"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", nil, nil, "", ""))
 	cp := cert.NewCAPool()
 
 	// Drop outbound
@@ -190,34 +190,34 @@ func TestFirewall_Drop(t *testing.T) {
 
 	// test remote mismatch
 	oldRemote := p.RemoteIP
-	p.RemoteIP = ip2int(net.IPv4(1, 2, 3, 10))
+	p.RemoteIP = iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 10))
 	assert.Equal(t, fw.Drop([]byte{}, p, false, &h, cp, nil), ErrInvalidRemoteIP)
 	p.RemoteIP = oldRemote
 
 	// ensure signer doesn't get in the way of group checks
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"nope"}, "", nil, "", "signer-shasum"))
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"default-group"}, "", nil, "", "signer-shasum-bad"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", nil, nil, "", "signer-shasum"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", nil, nil, "", "signer-shasum-bad"))
 	assert.Equal(t, fw.Drop([]byte{}, p, true, &h, cp, nil), ErrNoMatchingRule)
 
 	// test caSha doesn't drop on match
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"nope"}, "", nil, "", "signer-shasum-bad"))
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"default-group"}, "", nil, "", "signer-shasum"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", nil, nil, "", "signer-shasum-bad"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", nil, nil, "", "signer-shasum"))
 	assert.NoError(t, fw.Drop([]byte{}, p, true, &h, cp, nil))
 
 	// ensure ca name doesn't get in the way of group checks
 	cp.CAs["signer-shasum"] = &cert.NebulaCertificate{Details: cert.NebulaCertificateDetails{Name: "ca-good"}}
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"nope"}, "", nil, "ca-good", ""))
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"default-group"}, "", nil, "ca-good-bad", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", nil, nil, "ca-good", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", nil, nil, "ca-good-bad", ""))
 	assert.Equal(t, fw.Drop([]byte{}, p, true, &h, cp, nil), ErrNoMatchingRule)
 
 	// test caName doesn't drop on match
 	cp.CAs["signer-shasum"] = &cert.NebulaCertificate{Details: cert.NebulaCertificateDetails{Name: "ca-good"}}
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"nope"}, "", nil, "ca-good-bad", ""))
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"default-group"}, "", nil, "ca-good", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", nil, nil, "ca-good-bad", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", nil, nil, "ca-good", ""))
 	assert.NoError(t, fw.Drop([]byte{}, p, true, &h, cp, nil))
 }
 
@@ -227,24 +227,24 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 	}
 
 	_, n, _ := net.ParseCIDR("172.1.1.1/32")
-	_ = ft.TCP.addRule(10, 10, []string{"good-group"}, "good-host", n, "", "")
-	_ = ft.TCP.addRule(10, 10, []string{"good-group2"}, "good-host", n, "", "")
-	_ = ft.TCP.addRule(10, 10, []string{"good-group3"}, "good-host", n, "", "")
-	_ = ft.TCP.addRule(10, 10, []string{"good-group4"}, "good-host", n, "", "")
-	_ = ft.TCP.addRule(10, 10, []string{"good-group, good-group1"}, "good-host", n, "", "")
+	_ = ft.TCP.addRule(10, 10, []string{"good-group"}, "good-host", n, n, "", "")
+	_ = ft.TCP.addRule(10, 10, []string{"good-group2"}, "good-host", n, n, "", "")
+	_ = ft.TCP.addRule(10, 10, []string{"good-group3"}, "good-host", n, n, "", "")
+	_ = ft.TCP.addRule(10, 10, []string{"good-group4"}, "good-host", n, n, "", "")
+	_ = ft.TCP.addRule(10, 10, []string{"good-group, good-group1"}, "good-host", n, n, "", "")
 	cp := cert.NewCAPool()
 
 	b.Run("fail on proto", func(b *testing.B) {
 		c := &cert.NebulaCertificate{}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoUDP}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoUDP}, true, c, cp)
 		}
 	})
 
 	b.Run("fail on port", func(b *testing.B) {
 		c := &cert.NebulaCertificate{}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 1}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 1}, true, c, cp)
 		}
 	})
 
@@ -258,7 +258,7 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			},
 		}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 10}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10}, true, c, cp)
 		}
 	})
 
@@ -270,7 +270,7 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			},
 		}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 10}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10}, true, c, cp)
 		}
 	})
 
@@ -282,12 +282,12 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			},
 		}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 10}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10}, true, c, cp)
 		}
 	})
 
 	b.Run("pass on ip", func(b *testing.B) {
-		ip := ip2int(net.IPv4(172, 1, 1, 1))
+		ip := iputil.Ip2VpnIp(net.IPv4(172, 1, 1, 1))
 		c := &cert.NebulaCertificate{
 			Details: cert.NebulaCertificateDetails{
 				InvertedGroups: map[string]struct{}{"nope": {}},
@@ -295,14 +295,12 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			},
 		}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 10, RemoteIP: ip}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10, RemoteIP: ip}, true, c, cp)
 		}
 	})
 
-	_ = ft.TCP.addRule(0, 0, []string{"good-group"}, "good-host", n, "", "")
-
-	b.Run("pass on ip with any port", func(b *testing.B) {
-		ip := ip2int(net.IPv4(172, 1, 1, 1))
+	b.Run("pass on local ip", func(b *testing.B) {
+		ip := iputil.Ip2VpnIp(net.IPv4(172, 1, 1, 1))
 		c := &cert.NebulaCertificate{
 			Details: cert.NebulaCertificateDetails{
 				InvertedGroups: map[string]struct{}{"nope": {}},
@@ -310,23 +308,51 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			},
 		}
 		for n := 0; n < b.N; n++ {
-			ft.match(FirewallPacket{Protocol: fwProtoTCP, LocalPort: 100, RemoteIP: ip}, true, c, cp)
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10, LocalIP: ip}, true, c, cp)
+		}
+	})
+
+	_ = ft.TCP.addRule(0, 0, []string{"good-group"}, "good-host", n, n, "", "")
+
+	b.Run("pass on ip with any port", func(b *testing.B) {
+		ip := iputil.Ip2VpnIp(net.IPv4(172, 1, 1, 1))
+		c := &cert.NebulaCertificate{
+			Details: cert.NebulaCertificateDetails{
+				InvertedGroups: map[string]struct{}{"nope": {}},
+				Name:           "good-host",
+			},
+		}
+		for n := 0; n < b.N; n++ {
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, RemoteIP: ip}, true, c, cp)
+		}
+	})
+
+	b.Run("pass on local ip with any port", func(b *testing.B) {
+		ip := iputil.Ip2VpnIp(net.IPv4(172, 1, 1, 1))
+		c := &cert.NebulaCertificate{
+			Details: cert.NebulaCertificateDetails{
+				InvertedGroups: map[string]struct{}{"nope": {}},
+				Name:           "good-host",
+			},
+		}
+		for n := 0; n < b.N; n++ {
+			ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalIP: ip}, true, c, cp)
 		}
 	})
 }
 
 func TestFirewall_Drop2(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
-	p := FirewallPacket{
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		10,
-		90,
-		fwProtoUDP,
-		false,
+	p := firewall.Packet{
+		LocalIP:    iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		RemoteIP:   iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		LocalPort:  10,
+		RemotePort: 90,
+		Protocol:   firewall.ProtoUDP,
+		Fragment:   false,
 	}
 
 	ipNet := net.IPNet{
@@ -345,7 +371,7 @@ func TestFirewall_Drop2(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h.CreateRemoteCIDR(&c)
 
@@ -364,7 +390,7 @@ func TestFirewall_Drop2(t *testing.T) {
 	h1.CreateRemoteCIDR(&c1)
 
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"default-group", "test-group"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group", "test-group"}, "", nil, nil, "", ""))
 	cp := cert.NewCAPool()
 
 	// h1/c1 lacks the proper groups
@@ -375,17 +401,17 @@ func TestFirewall_Drop2(t *testing.T) {
 }
 
 func TestFirewall_Drop3(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
-	p := FirewallPacket{
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		1,
-		1,
-		fwProtoUDP,
-		false,
+	p := firewall.Packet{
+		LocalIP:    iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		RemoteIP:   iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		LocalPort:  1,
+		RemotePort: 1,
+		Protocol:   firewall.ProtoUDP,
+		Fragment:   false,
 	}
 
 	ipNet := net.IPNet{
@@ -411,7 +437,7 @@ func TestFirewall_Drop3(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c1,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h1.CreateRemoteCIDR(&c1)
 
@@ -426,7 +452,7 @@ func TestFirewall_Drop3(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c2,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h2.CreateRemoteCIDR(&c2)
 
@@ -441,13 +467,13 @@ func TestFirewall_Drop3(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c3,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h3.CreateRemoteCIDR(&c3)
 
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 1, 1, []string{}, "host1", nil, "", ""))
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 1, 1, []string{}, "", nil, "", "signer-sha"))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "host1", nil, nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "", nil, nil, "", "signer-sha"))
 	cp := cert.NewCAPool()
 
 	// c1 should pass because host match
@@ -461,17 +487,17 @@ func TestFirewall_Drop3(t *testing.T) {
 }
 
 func TestFirewall_DropConntrackReload(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
-	p := FirewallPacket{
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		ip2int(net.IPv4(1, 2, 3, 4)),
-		10,
-		90,
-		fwProtoUDP,
-		false,
+	p := firewall.Packet{
+		LocalIP:    iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		RemoteIP:   iputil.Ip2VpnIp(net.IPv4(1, 2, 3, 4)),
+		LocalPort:  10,
+		RemotePort: 90,
+		Protocol:   firewall.ProtoUDP,
+		Fragment:   false,
 	}
 
 	ipNet := net.IPNet{
@@ -492,12 +518,12 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 		ConnectionState: &ConnectionState{
 			peerCert: &c,
 		},
-		hostId: ip2int(ipNet.IP),
+		vpnIp: iputil.Ip2VpnIp(ipNet.IP),
 	}
 	h.CreateRemoteCIDR(&c)
 
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 0, 0, []string{"any"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", nil, nil, "", ""))
 	cp := cert.NewCAPool()
 
 	// Drop outbound
@@ -510,7 +536,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 
 	oldFw := fw
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 10, 10, []string{"any"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 10, 10, []string{"any"}, "", nil, nil, "", ""))
 	fw.Conntrack = oldFw.Conntrack
 	fw.rulesVersion = oldFw.rulesVersion + 1
 
@@ -519,7 +545,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 
 	oldFw = fw
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	assert.Nil(t, fw.AddRule(true, fwProtoAny, 11, 11, []string{"any"}, "", nil, "", ""))
+	assert.Nil(t, fw.AddRule(true, firewall.ProtoAny, 11, 11, []string{"any"}, "", nil, nil, "", ""))
 	fw.Conntrack = oldFw.Conntrack
 	fw.rulesVersion = oldFw.rulesVersion + 1
 
@@ -643,28 +669,28 @@ func Test_parsePort(t *testing.T) {
 }
 
 func TestNewFirewallFromConfig(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	// Test a bad rule definition
 	c := &cert.NebulaCertificate{}
-	conf := NewConfig(l)
+	conf := config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": "asdf"}
 	_, err := NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.outbound failed to parse, should be an array of rules")
 
 	// Test both port and code
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"port": "1", "code": "2"}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.outbound rule #0; only one of port or code should be provided")
 
 	// Test missing host, group, cidr, ca_name and ca_sha
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
-	assert.EqualError(t, err, "firewall.outbound rule #0; at least one of host, group, cidr, ca_name, or ca_sha must be provided")
+	assert.EqualError(t, err, "firewall.outbound rule #0; at least one of host, group, cidr, local_cidr, ca_name, or ca_sha must be provided")
 
 	// Test code/port error
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"code": "a", "host": "testh"}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.outbound rule #0; code was not a number; `a`")
@@ -674,91 +700,112 @@ func TestNewFirewallFromConfig(t *testing.T) {
 	assert.EqualError(t, err, "firewall.outbound rule #0; port was not a number; `a`")
 
 	// Test proto error
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"code": "1", "host": "testh"}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.outbound rule #0; proto was not understood; ``")
 
 	// Test cidr parse error
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"code": "1", "cidr": "testh", "proto": "any"}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.outbound rule #0; cidr did not parse; invalid CIDR address: testh")
 
+	// Test local_cidr parse error
+	conf = config.NewC(l)
+	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"code": "1", "local_cidr": "testh", "proto": "any"}}}
+	_, err = NewFirewallFromConfig(l, c, conf)
+	assert.EqualError(t, err, "firewall.outbound rule #0; local_cidr did not parse; invalid CIDR address: testh")
+
 	// Test both group and groups
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "group": "a", "groups": []string{"b", "c"}}}}
 	_, err = NewFirewallFromConfig(l, c, conf)
 	assert.EqualError(t, err, "firewall.inbound rule #0; only one of group or groups should be defined, both provided")
 }
 
 func TestAddFirewallRulesFromConfig(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	// Test adding tcp rule
-	conf := NewConfig(l)
+	conf := config.NewC(l)
 	mf := &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "tcp", "host": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, false, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: false, proto: fwProtoTCP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoTCP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test adding udp rule
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "udp", "host": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, false, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: false, proto: fwProtoUDP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoUDP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test adding icmp rule
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"outbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "icmp", "host": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, false, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: false, proto: fwProtoICMP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoICMP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test adding any rule
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "host": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, host: "a", ip: nil, localIp: nil}, mf.lastCall)
+
+	// Test adding rule with cidr
+	cidr := &net.IPNet{IP: net.ParseIP("10.0.0.0").To4(), Mask: net.IPv4Mask(255, 0, 0, 0)}
+	conf = config.NewC(l)
+	mf = &mockFirewall{}
+	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "cidr": cidr.String()}}}
+	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: cidr, localIp: nil}, mf.lastCall)
+
+	// Test adding rule with local_cidr
+	conf = config.NewC(l)
+	mf = &mockFirewall{}
+	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "local_cidr": cidr.String()}}}
+	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: nil, localIp: cidr}, mf.lastCall)
 
 	// Test adding rule with ca_sha
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "ca_sha": "12312313123"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: nil, ip: nil, caSha: "12312313123"}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: nil, localIp: nil, caSha: "12312313123"}, mf.lastCall)
 
 	// Test adding rule with ca_name
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "ca_name": "root01"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: nil, ip: nil, caName: "root01"}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: nil, localIp: nil, caName: "root01"}, mf.lastCall)
 
 	// Test single group
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "group": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test single groups
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "groups": "a"}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test multiple AND groups
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "groups": []string{"a", "b"}}}}
 	assert.Nil(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: fwProtoAny, startPort: 1, endPort: 1, groups: []string{"a", "b"}, ip: nil}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a", "b"}, ip: nil, localIp: nil}, mf.lastCall)
 
 	// Test Add error
-	conf = NewConfig(l)
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	mf.nextCallReturn = errors.New("test error")
 	conf.Settings["firewall"] = map[interface{}]interface{}{"inbound": []interface{}{map[interface{}]interface{}{"port": "1", "proto": "any", "host": "a"}}}
@@ -857,7 +904,7 @@ func TestTCPRTTTracking(t *testing.T) {
 }
 
 func TestFirewall_convertRule(t *testing.T) {
-	l := NewTestLogger()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
 	l.SetOutput(ob)
 
@@ -900,6 +947,7 @@ type addRuleCall struct {
 	groups    []string
 	host      string
 	ip        *net.IPNet
+	localIp   *net.IPNet
 	caName    string
 	caSha     string
 }
@@ -909,7 +957,7 @@ type mockFirewall struct {
 	nextCallReturn error
 }
 
-func (mf *mockFirewall) AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, caName string, caSha string) error {
+func (mf *mockFirewall) AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, ip *net.IPNet, localIp *net.IPNet, caName string, caSha string) error {
 	mf.lastCall = addRuleCall{
 		incoming:  incoming,
 		proto:     proto,
@@ -918,6 +966,7 @@ func (mf *mockFirewall) AddRule(incoming bool, proto uint8, startPort int32, end
 		groups:    groups,
 		host:      host,
 		ip:        ip,
+		localIp:   localIp,
 		caName:    caName,
 		caSha:     caSha,
 	}
@@ -929,6 +978,6 @@ func (mf *mockFirewall) AddRule(incoming bool, proto uint8, startPort int32, end
 
 func resetConntrack(fw *Firewall) {
 	fw.Conntrack.Lock()
-	fw.Conntrack.Conns = map[FirewallPacket]*conn{}
+	fw.Conntrack.Conns = map[firewall.Packet]*conn{}
 	fw.Conntrack.Unlock()
 }

go.mod

@@ -1,35 +1,49 @@
 module github.com/slackhq/nebula
 
-go 1.16
+go 1.20
 
 require (
-	github.com/anmitsu/go-shlex v0.0.0-20161002113705-648efa622239
+	github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be
 	github.com/armon/go-radix v1.0.0
-	github.com/cespare/xxhash/v2 v2.1.1 // indirect
 	github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432
-	github.com/flynn/go-shlex v0.0.0-20150515145356-3f9db97f8568 // indirect
-	github.com/flynn/noise v0.0.0-20210331153838-4bdb43be3117
+	github.com/flynn/noise v1.0.0
 	github.com/gogo/protobuf v1.3.2
-	github.com/golang/protobuf v1.5.0
 	github.com/google/gopacket v1.1.19
-	github.com/imdario/mergo v0.3.8
-	github.com/kardianos/service v1.1.0
-	github.com/konsorten/go-windows-terminal-sequences v1.0.2 // indirect
-	github.com/miekg/dns v1.1.25
-	github.com/nbrownus/go-metrics-prometheus v0.0.0-20180622211546-6e6d5173d99c
-	github.com/prometheus/client_golang v1.2.1
-	github.com/prometheus/client_model v0.0.0-20191202183732-d1d2010b5bee // indirect
-	github.com/prometheus/procfs v0.0.8 // indirect
-	github.com/rcrowley/go-metrics v0.0.0-20190826022208-cac0b30c2563
-	github.com/sirupsen/logrus v1.4.2
+	github.com/imdario/mergo v0.3.15
+	github.com/kardianos/service v1.2.2
+	github.com/miekg/dns v1.1.54
+	github.com/nbrownus/go-metrics-prometheus v0.0.0-20210712211119-974a6260965f
+	github.com/prometheus/client_golang v1.15.1
+	github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475
+	github.com/sirupsen/logrus v1.9.0
 	github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e
-	github.com/songgao/water v0.0.0-20190725173103-fd331bda3f4b
-	github.com/stretchr/testify v1.6.1
-	github.com/vishvananda/netlink v1.0.1-0.20190522153524-00009fb8606a
-	github.com/vishvananda/netns v0.0.0-20191106174202-0a2b9b5464df // indirect
-	golang.org/x/crypto v0.0.0-20210322153248-0c34fe9e7dc2
-	golang.org/x/net v0.0.0-20210226172049-e18ecbb05110
-	golang.org/x/sys v0.0.0-20201119102817-f84b799fce68
-	google.golang.org/protobuf v1.26.0
-	gopkg.in/yaml.v2 v2.2.7
+	github.com/songgao/water v0.0.0-20200317203138-2b4b6d7c09d8
+	github.com/stretchr/testify v1.8.2
+	github.com/vishvananda/netlink v1.1.0
+	golang.org/x/crypto v0.8.0
+	golang.org/x/exp v0.0.0-20230425010034-47ecfdc1ba53
+	golang.org/x/net v0.9.0
+	golang.org/x/sys v0.8.0
+	golang.org/x/term v0.8.0
+	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2
+	golang.zx2c4.com/wireguard/windows v0.5.3
+	google.golang.org/protobuf v1.30.0
+	gopkg.in/yaml.v2 v2.4.0
+)
+
+require (
+	github.com/beorn7/perks v1.0.1 // indirect
+	github.com/cespare/xxhash/v2 v2.2.0 // indirect
+	github.com/davecgh/go-spew v1.1.1 // indirect
+	github.com/golang/protobuf v1.5.3 // indirect
+	github.com/matttproud/golang_protobuf_extensions v1.0.4 // indirect
+	github.com/pmezard/go-difflib v1.0.0 // indirect
+	github.com/prometheus/client_model v0.4.0 // indirect
+	github.com/prometheus/common v0.42.0 // indirect
+	github.com/prometheus/procfs v0.9.0 // indirect
+	github.com/rogpeppe/go-internal v1.10.0 // indirect
+	github.com/vishvananda/netns v0.0.4 // indirect
+	golang.org/x/mod v0.10.0 // indirect
+	golang.org/x/tools v0.8.0 // indirect
+	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,31 +1,33 @@
+cloud.google.com/go v0.34.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
 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=
 github.com/alecthomas/units v0.0.0-20190717042225-c3de453c63f4/go.mod h1:ybxpYRFXyAe+OPACYpWeL0wqObRcbAqCMya13uyzqw0=
-github.com/anmitsu/go-shlex v0.0.0-20161002113705-648efa622239 h1:kFOfPq6dUM1hTo4JG6LR5AXSUEsOjtdm0kw0FtQtMJA=
-github.com/anmitsu/go-shlex v0.0.0-20161002113705-648efa622239/go.mod h1:2FmKhYUyUczH0OGQWaF5ceTx0UBShxjsH6f8oGKYe2c=
+github.com/alecthomas/units v0.0.0-20190924025748-f65c72e2690d/go.mod h1:rBZYJk541a8SKzHPHnH3zbiI+7dagKZ0cgpgrD7Fyho=
+github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be h1:9AeTilPcZAjCFIImctFaOjnTIavg87rW78vTPkQqLI8=
+github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be/go.mod h1:ySMOLuWl6zY27l47sB3qLNK6tF2fkHG55UZxx8oIVo4=
 github.com/armon/go-radix v1.0.0 h1:F4z6KzEeeQIMeLFa97iZU6vupzoecKdU5TX24SNppXI=
 github.com/armon/go-radix v1.0.0/go.mod h1:ufUuZ+zHj4x4TnLV4JWEpy2hxWSpsRywHrMgIH9cCH8=
 github.com/beorn7/perks v0.0.0-20180321164747-3a771d992973/go.mod h1:Dwedo/Wpr24TaqPxmxbtue+5NUziq4I4S80YR8gNf3Q=
 github.com/beorn7/perks v1.0.0/go.mod h1:KWe93zE9D1o94FZ5RNwFwVgaQK1VOXiVxmqh+CedLV8=
 github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
 github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
-github.com/cespare/xxhash/v2 v2.1.0/go.mod h1:dgIUBU3pDso/gPgZ1osOZ0iQf77oPR28Tjxl5dIMyVM=
-github.com/cespare/xxhash/v2 v2.1.1 h1:6MnRN8NT7+YBpUIWxHtefFZOKTAPgGjpQSxqLNn0+qY=
 github.com/cespare/xxhash/v2 v2.1.1/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/cespare/xxhash/v2 v2.2.0 h1:DC2CZ1Ep5Y4k3ZQ899DldepgrayRUGE6BBZ/cd9Cj44=
+github.com/cespare/xxhash/v2 v2.2.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
 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=
 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/go-shlex v0.0.0-20150515145356-3f9db97f8568 h1:BHsljHzVlRcyQhjrss6TZTdY2VfCqZPbv5k3iBFa2ZQ=
-github.com/flynn/go-shlex v0.0.0-20150515145356-3f9db97f8568/go.mod h1:xEzjJPgXI435gkrCt3MPfRiAkVrwSbHsst4LCFVfpJc=
-github.com/flynn/noise v0.0.0-20210331153838-4bdb43be3117 h1:Dxhvhray2DpvNnrZEnoGG5rz238fUeQTh4sdzTr+d1U=
-github.com/flynn/noise v0.0.0-20210331153838-4bdb43be3117/go.mod h1:xbMo+0i6+IGbYdJhF31t2eR1BIU0CYc12+BNAKwUTag=
+github.com/flynn/noise v1.0.0 h1:DlTHqmzmvcEiKj+4RYo/imoswx/4r6iBlCMfVtrMXpQ=
+github.com/flynn/noise v1.0.0/go.mod h1:xbMo+0i6+IGbYdJhF31t2eR1BIU0CYc12+BNAKwUTag=
 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=
 github.com/go-logfmt/logfmt v0.3.0/go.mod h1:Qt1PoO58o5twSAckw1HlFXLmHsOX5/0LbT9GBnD5lWE=
 github.com/go-logfmt/logfmt v0.4.0/go.mod h1:3RMwSq7FuexP4Kalkev3ejPJsZTpXXBr9+V4qmtdjCk=
+github.com/go-logfmt/logfmt v0.5.0/go.mod h1:wCYkCAKZfumFQihp8CzCvQ3paCTfi41vtzG1KdI/P7A=
 github.com/go-stack/stack v1.8.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
 github.com/gogo/protobuf v1.1.1/go.mod h1:r8qH/GZQm5c6nD/R0oafs1akxWv10x8SbQlK7atdtwQ=
 github.com/gogo/protobuf v1.3.2 h1:Ov1cvc58UF3b5XjBnZv7+opcTcQFZebYjWzi34vdm4Q=
@@ -33,150 +35,215 @@ github.com/gogo/protobuf v1.3.2/go.mod h1:P1XiOD3dCwIKUDQYPy72D8LYyHL2YPYrpS2s69
 github.com/golang/protobuf v1.2.0/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
 github.com/golang/protobuf v1.3.1/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
 github.com/golang/protobuf v1.3.2/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
-github.com/golang/protobuf v1.5.0 h1:LUVKkCeviFUMKqHa4tXIIij/lbhnMbP7Fn5wKdKkRh4=
+github.com/golang/protobuf v1.4.0-rc.1/go.mod h1:ceaxUfeHdC40wWswd/P6IGgMaK3YpKi5j83Wpe3EHw8=
+github.com/golang/protobuf v1.4.0-rc.1.0.20200221234624-67d41d38c208/go.mod h1:xKAWHe0F5eneWXFV3EuXVDTCmh+JuBKY0li0aMyXATA=
+github.com/golang/protobuf v1.4.0-rc.2/go.mod h1:LlEzMj4AhA7rCAGe4KMBDvJI+AwstrUpVNzEA03Pprs=
+github.com/golang/protobuf v1.4.0-rc.4.0.20200313231945-b860323f09d0/go.mod h1:WU3c8KckQ9AFe+yFwt9sWVRKCVIyN9cPHBJSNnbL67w=
+github.com/golang/protobuf v1.4.0/go.mod h1:jodUvKwWbYaEsadDk5Fwe5c77LiNKVO9IDvqG2KuDX0=
+github.com/golang/protobuf v1.4.2/go.mod h1:oDoupMAO8OvCJWAcko0GGGIgR6R6ocIYbsSw735rRwI=
+github.com/golang/protobuf v1.4.3/go.mod h1:oDoupMAO8OvCJWAcko0GGGIgR6R6ocIYbsSw735rRwI=
 github.com/golang/protobuf v1.5.0/go.mod h1:FsONVRAS9T7sI+LIUmWTfcYkHO4aIWwzhcaSAoJOfIk=
+github.com/golang/protobuf v1.5.3 h1:KhyjKVUg7Usr/dYsdSqoFveMYd5ko72D+zANwlG1mmg=
+github.com/golang/protobuf v1.5.3/go.mod h1:XVQd3VNwM+JqD3oG2Ue2ip4fOMUkwXdXDdiuN0vRsmY=
 github.com/google/go-cmp v0.3.0/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
 github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMywk6iLU=
-github.com/google/go-cmp v0.5.5 h1:Khx7svrCpmxxtHBq5j2mp/xVjsi8hQMfNLvJFAlrGgU=
+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 v0.5.9 h1:O2Tfq5qg4qc4AmwVlvv0oLiVAGB7enBSJ2x2DqQFi38=
 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=
-github.com/imdario/mergo v0.3.8 h1:CGgOkSJeqMRmt0D9XLWExdT4m4F1vd3FV3VPt+0VxkQ=
-github.com/imdario/mergo v0.3.8/go.mod h1:2EnlNZ0deacrJVfApfmtdGgDfMuh/nq6Ok1EcJh5FfA=
+github.com/imdario/mergo v0.3.15 h1:M8XP7IuFNsqUx6VPK2P9OSmsYsI/YFaGil0uD21V3dM=
+github.com/imdario/mergo v0.3.15/go.mod h1:WBLT9ZmE3lPoWsEzCh9LPo3TiwVN+ZKEjmz+hD27ysY=
+github.com/jpillora/backoff v1.0.0/go.mod h1:J/6gKK9jxlEcS3zixgDgUAsiuZ7yrSoa/FX5e0EB2j4=
 github.com/json-iterator/go v1.1.6/go.mod h1:+SdeFBvtyEkXs7REEP0seUULqWtbJapLOCVDaaPEHmU=
-github.com/json-iterator/go v1.1.7/go.mod h1:KdQUCv79m/52Kvf8AW2vK1V8akMuk1QjK/uOdHXbAo4=
+github.com/json-iterator/go v1.1.10/go.mod h1:KdQUCv79m/52Kvf8AW2vK1V8akMuk1QjK/uOdHXbAo4=
+github.com/json-iterator/go v1.1.11/go.mod h1:KdQUCv79m/52Kvf8AW2vK1V8akMuk1QjK/uOdHXbAo4=
 github.com/julienschmidt/httprouter v1.2.0/go.mod h1:SYymIcj16QtmaHHD7aYtjjsJG7VTCxuUUipMqKk8s4w=
-github.com/kardianos/service v1.1.0 h1:QV2SiEeWK42P0aEmGcsAgjApw/lRxkwopvT+Gu6t1/0=
-github.com/kardianos/service v1.1.0/go.mod h1:RrJI2xn5vve/r32U5suTbeaSGoMU6GbNPoj36CVYcHc=
+github.com/julienschmidt/httprouter v1.3.0/go.mod h1:JR6WtHb+2LUe8TCKY3cZOxFyyO8IZAc4RVcycCCAKdM=
+github.com/kardianos/service v1.2.2 h1:ZvePhAHfvo0A7Mftk/tEzqEZ7Q4lgnR8sGz4xu1YX60=
+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/konsorten/go-windows-terminal-sequences v1.0.1/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
-github.com/konsorten/go-windows-terminal-sequences v1.0.2 h1:DB17ag19krx9CFsz4o3enTrPXyIXCl+2iCXH/aMAp9s=
-github.com/konsorten/go-windows-terminal-sequences v1.0.2/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=
-github.com/kr/pretty v0.2.1 h1:Fmg33tUaq4/8ym9TJN1x7sLJnHVwhP33CNkpYV/7rwI=
+github.com/kr/pretty v0.1.0/go.mod h1:dAy3ld7l9f0ibDNOQOHHMYYIIbhfbHSm3C4ZsoJORNo=
 github.com/kr/pretty v0.2.1/go.mod h1:ipq/a2n7PKx3OHsz4KJII5eveXtPO4qwEXGdVfWzfnI=
+github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
 github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
 github.com/kr/text v0.1.0 h1:45sCR5RtlFHMR4UwH9sdQ5TC8v0qDQCHnXt+kaKSTVE=
 github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
-github.com/matttproud/golang_protobuf_extensions v1.0.1 h1:4hp9jkHxhMHkqkrB3Ix0jegS5sx/RkqARlsWZ6pIwiU=
 github.com/matttproud/golang_protobuf_extensions v1.0.1/go.mod h1:D8He9yQNgCq6Z5Ld7szi9bcBfOoFv/3dc6xSMkL2PC0=
-github.com/miekg/dns v1.1.25 h1:dFwPR6SfLtrSwgDcIq2bcU/gVutB4sNApq2HBdqcakg=
-github.com/miekg/dns v1.1.25/go.mod h1:bPDLeHnStXmXAq1m/Ch/hvfNHr14JKNPMBo3VZKjuso=
+github.com/matttproud/golang_protobuf_extensions v1.0.4 h1:mmDVorXM7PCGKw94cs5zkfA9PSy5pEvNWRP0ET0TIVo=
+github.com/matttproud/golang_protobuf_extensions v1.0.4/go.mod h1:BSXmuO+STAnVfrANrmjBb36TMTDstsz7MSK+HVaYKv4=
+github.com/miekg/dns v1.1.54 h1:5jon9mWcb0sFJGpnI99tOMhCPyJ+RPVz5b63MQG0VWI=
+github.com/miekg/dns v1.1.54/go.mod h1:uInx36IzPl7FYnDcMeVWxj9byh7DutNykX4G9Sj60FY=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 github.com/modern-go/reflect2 v0.0.0-20180701023420-4b7aa43c6742/go.mod h1:bx2lNnkwVCuqBIxFjflWJWanXIb3RllmbCylyMrvgv0=
 github.com/modern-go/reflect2 v1.0.1/go.mod h1:bx2lNnkwVCuqBIxFjflWJWanXIb3RllmbCylyMrvgv0=
 github.com/mwitkow/go-conntrack v0.0.0-20161129095857-cc309e4a2223/go.mod h1:qRWi+5nqEBWmkhHvq77mSJWrCKwh8bxhgT7d/eI7P4U=
-github.com/nbrownus/go-metrics-prometheus v0.0.0-20180622211546-6e6d5173d99c h1:G/mfx/MWYuaaGlHkZQBBXFAJiYnRt/GaOVxnRHjlxg4=
-github.com/nbrownus/go-metrics-prometheus v0.0.0-20180622211546-6e6d5173d99c/go.mod h1:1yMri853KAI2pPAUnESjaqZj9JeImOUM+6A4GuuPmTs=
+github.com/mwitkow/go-conntrack v0.0.0-20190716064945-2f068394615f/go.mod h1:qRWi+5nqEBWmkhHvq77mSJWrCKwh8bxhgT7d/eI7P4U=
+github.com/nbrownus/go-metrics-prometheus v0.0.0-20210712211119-974a6260965f h1:8dM0ilqKL0Uzl42GABzzC4Oqlc3kGRILz0vgoff7nwg=
+github.com/nbrownus/go-metrics-prometheus v0.0.0-20210712211119-974a6260965f/go.mod h1:nwPd6pDNId/Xi16qtKrFHrauSwMNuvk+zcjk89wrnlA=
 github.com/pkg/errors v0.8.0/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/pkg/errors v0.8.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
+github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/prometheus/client_golang v0.9.1/go.mod h1:7SWBe2y4D6OKWSNQJUaRYU/AaXPKyh/dDVn+NZz0KFw=
 github.com/prometheus/client_golang v1.0.0/go.mod h1:db9x61etRT2tGnBNRi70OPL5FsnadC4Ky3P0J6CfImo=
-github.com/prometheus/client_golang v1.2.1 h1:JnMpQc6ppsNgw9QPAGF6Dod479itz7lvlsMzzNayLOI=
-github.com/prometheus/client_golang v1.2.1/go.mod h1:XMU6Z2MjaRKVu/dC1qupJI9SiNkDYzz3xecMgSW/F+U=
+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 v1.15.1 h1:8tXpTmJbyH5lydzFPoxSIJ0J46jdh3tylbvM1xCv0LI=
+github.com/prometheus/client_golang v1.15.1/go.mod h1:e9yaBhRPU2pPNsZwE+JdQl0KEt1N9XgF6zxWmaC0xOk=
 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 v0.0.0-20190812154241-14fe0d1b01d4/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
-github.com/prometheus/client_model v0.0.0-20191202183732-d1d2010b5bee h1:iBZPTYkGLvdu6+A5TsMUJQkQX9Ad4aCEnSQtdxPuTCQ=
-github.com/prometheus/client_model v0.0.0-20191202183732-d1d2010b5bee/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
+github.com/prometheus/client_model v0.2.0/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
+github.com/prometheus/client_model v0.4.0 h1:5lQXD3cAg1OXBf4Wq03gTrXHeaV0TQvGfUooCfx1yqY=
+github.com/prometheus/client_model v0.4.0/go.mod h1:oMQmHW1/JoDwqLtg57MGgP/Fb1CJEYF2imWWhWtMkYU=
 github.com/prometheus/common v0.4.1/go.mod h1:TNfzLD0ON7rHzMJeJkieUDPYmFC7Snx/y86RQel1bk4=
-github.com/prometheus/common v0.7.0 h1:L+1lyG48J1zAQXA3RBX/nG/B3gjlHq0zTt2tlbJLyCY=
-github.com/prometheus/common v0.7.0/go.mod h1:DjGbpBbp5NYNiECxcL/VnbXCCaQpKd3tt26CguLLsqA=
+github.com/prometheus/common v0.10.0/go.mod h1:Tlit/dnDKsSWFlCLTWaA1cyBgKHSMdTB80sz/V91rCo=
+github.com/prometheus/common v0.26.0/go.mod h1:M7rCNAaPfAosfx8veZJCuw84e35h3Cfd9VFqTh1DIvc=
+github.com/prometheus/common v0.42.0 h1:EKsfXEYo4JpWMHH5cg+KOUWeuJSov1Id8zGR8eeI1YM=
+github.com/prometheus/common v0.42.0/go.mod h1:xBwqVerjNdUDjgODMpudtOMwlOwf2SaTr1yjz4b7Zbc=
 github.com/prometheus/procfs v0.0.0-20181005140218-185b4288413d/go.mod h1:c3At6R/oaqEKCNdg8wHV1ftS6bRYblBhIjjI8uT2IGk=
 github.com/prometheus/procfs v0.0.2/go.mod h1:TjEm7ze935MbeOT/UhFTIMYKhuLP4wbCsTZCD3I8kEA=
-github.com/prometheus/procfs v0.0.5/go.mod h1:4A/X28fw3Fc593LaREMrKMqOKvUAntwMDaekg4FpcdQ=
-github.com/prometheus/procfs v0.0.8 h1:+fpWZdT24pJBiqJdAwYBjPSk+5YmQzYNPYzQsdzLkt8=
-github.com/prometheus/procfs v0.0.8/go.mod h1:7Qr8sr6344vo1JqZ6HhLceV9o3AJ1Ff+GxbHq6oeK9A=
-github.com/rcrowley/go-metrics v0.0.0-20190826022208-cac0b30c2563 h1:dY6ETXrvDG7Sa4vE8ZQG4yqWg6UnOcbqTAahkV813vQ=
-github.com/rcrowley/go-metrics v0.0.0-20190826022208-cac0b30c2563/go.mod h1:bCqnVzQkZxMG4s8nGwiZ5l3QUCyqpo9Y+/ZMZ9VjZe4=
+github.com/prometheus/procfs v0.1.3/go.mod h1:lV6e/gmhEcM9IjHGsFOCxxuZ+z1YqCvr4OA4YeYWdaU=
+github.com/prometheus/procfs v0.6.0/go.mod h1:cz+aTbrPOrUb4q7XlbU9ygM+/jj0fzG6c1xBZuNvfVA=
+github.com/prometheus/procfs v0.9.0 h1:wzCHvIvM5SxWqYvwgVL7yJY8Lz3PKn49KQtpgMYJfhI=
+github.com/prometheus/procfs v0.9.0/go.mod h1:+pB4zwohETzFnmlpe6yd2lSc+0/46IYZRB/chUwxUZY=
+github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475 h1:N/ElC8H3+5XpJzTSTfLsJV/mx9Q9g7kxmchpfZyxgzM=
+github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475/go.mod h1:bCqnVzQkZxMG4s8nGwiZ5l3QUCyqpo9Y+/ZMZ9VjZe4=
+github.com/rogpeppe/go-internal v1.10.0 h1:TMyTOH3F/DB16zRVcYyreMH6GnZZrwQVAoYjRBZyWFQ=
+github.com/rogpeppe/go-internal v1.10.0/go.mod h1:UQnix2H7Ngw/k4C5ijL5+65zddjncjaFoBhdsK/akog=
 github.com/sirupsen/logrus v1.2.0/go.mod h1:LxeOpSwHxABJmUn/MG1IvRgCAasNZTLOkJPxbbu5VWo=
-github.com/sirupsen/logrus v1.4.2 h1:SPIRibHv4MatM3XXNO2BJeFLZwZ2LvZgfQ5+UNI2im4=
 github.com/sirupsen/logrus v1.4.2/go.mod h1:tLMulIdttU9McNUspp0xgXVQah82FyeX6MwdIuYE2rE=
+github.com/sirupsen/logrus v1.6.0/go.mod h1:7uNnSEd1DgxDLC74fIahvMZmmYsHGZGEOFrfsX/uA88=
+github.com/sirupsen/logrus v1.9.0 h1:trlNQbNUG3OdDrDil03MCb1H2o9nJ1x4/5LYw7byDE0=
+github.com/sirupsen/logrus v1.9.0/go.mod h1:naHLuLoDiP4jHNo9R0sCBMtWGeIprob74mVsIT4qYEQ=
 github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e h1:MRM5ITcdelLK2j1vwZ3Je0FKVCfqOLp5zO6trqMLYs0=
 github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e/go.mod h1:XV66xRDqSt+GTGFMVlhk3ULuV0y9ZmzeVGR4mloJI3M=
-github.com/songgao/water v0.0.0-20190725173103-fd331bda3f4b h1:+y4hCMc/WKsDbAPsOQZgBSaSZ26uh2afyaWeVg/3s/c=
-github.com/songgao/water v0.0.0-20190725173103-fd331bda3f4b/go.mod h1:P5HUIBuIWKbyjl083/loAegFkfbFNx5i2qEP4CNbm7E=
+github.com/songgao/water v0.0.0-20200317203138-2b4b6d7c09d8 h1:TG/diQgUe0pntT/2D9tmUCz4VNwm9MfrtPr0SU2qSX8=
+github.com/songgao/water v0.0.0-20200317203138-2b4b6d7c09d8/go.mod h1:P5HUIBuIWKbyjl083/loAegFkfbFNx5i2qEP4CNbm7E=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
 github.com/stretchr/objx v0.1.1/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/objx v0.4.0/go.mod h1:YvHI0jy2hoMjB+UWwv71VJQ9isScKT/TqJzVSSt89Yw=
+github.com/stretchr/objx v0.5.0/go.mod h1:Yh+to48EsGEfYuaHDzXPcE3xhTkx73EhmCGUpEOglKo=
 github.com/stretchr/testify v1.2.2/go.mod h1:a8OnRcib4nhh0OaRAV+Yts87kKdq0PP7pXfy6kDkUVs=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
-github.com/stretchr/testify v1.6.1 h1:hDPOHmpOpP40lSULcqw7IrRb/u7w6RpDC9399XyoNd0=
-github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
-github.com/vishvananda/netlink v1.0.1-0.20190522153524-00009fb8606a h1:Bt1IVPhiCDMqwGrc2nnbIN4QKvJGx6SK2NzWBmW00ao=
-github.com/vishvananda/netlink v1.0.1-0.20190522153524-00009fb8606a/go.mod h1:+SR5DhBJrl6ZM7CoCKvpw5BKroDKQ+PJqOg65H/2ktk=
-github.com/vishvananda/netns v0.0.0-20191106174202-0a2b9b5464df h1:OviZH7qLw/7ZovXvuNyL3XQl8UFofeikI1NW1Gypu7k=
+github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
+github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
+github.com/stretchr/testify v1.8.2 h1:+h33VjcLVPDHtOdpUCuF+7gSuG3yGIftsP1YvFihtJ8=
+github.com/stretchr/testify v1.8.2/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4=
+github.com/vishvananda/netlink v1.1.0 h1:1iyaYNBLmP6L0220aDnYQpo1QEV4t4hJ+xEEhhJH8j0=
+github.com/vishvananda/netlink v1.1.0/go.mod h1:cTgwzPIzzgDAYoQrMm0EdrjRUBkTqKYppBueQtXaqoE=
 github.com/vishvananda/netns v0.0.0-20191106174202-0a2b9b5464df/go.mod h1:JP3t17pCcGlemwknint6hfoeCVQrEMVwxRLRjXpq+BU=
+github.com/vishvananda/netns v0.0.4 h1:Oeaw1EM2JMxD51g9uhtC0D7erkIjgmj8+JZc26m1YX8=
+github.com/vishvananda/netns v0.0.4/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=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.0.0-20190923035154-9ee001bba392/go.mod h1:/lpIB1dKB+9EgE3H3cr1v9wB50oz8l4C4h62xy7jSTY=
 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 h1:It14KIkyBFYkHkwZ7k45minvA9aorojkyjGk9KJ5B/w=
 golang.org/x/crypto v0.0.0-20210322153248-0c34fe9e7dc2/go.mod h1:T9bdIzuCu7OtxOm1hfPfRQxPLYneinmdGuTeoZ9dtd4=
+golang.org/x/crypto v0.8.0 h1:pd9TJtTueMTVQXzk8E2XESSMQDj/U7OUu0PqJqPXQjQ=
+golang.org/x/crypto v0.8.0/go.mod h1:mRqEX+O9/h5TFCrQhkgjo2yKi0yYA+9ecGkdQoHrywE=
+golang.org/x/exp v0.0.0-20230425010034-47ecfdc1ba53 h1:5llv2sWeaMSnA3w2kS57ouQQ4pudlXrR0dCgw51QK9o=
+golang.org/x/exp v0.0.0-20230425010034-47ecfdc1ba53/go.mod h1:V1LtkGg67GoY2N1AnLN78QLrzxkLyJw7RJb1gzOOz9w=
 golang.org/x/lint v0.0.0-20200302205851-738671d3881b/go.mod h1:3xt1FjdF8hUf6vQPIChWIBhFzV8gjjsPE/fR3IyQdNY=
 golang.org/x/mod v0.1.1-0.20191105210325-c90efee705ee/go.mod h1:QqPTAvyqsEbceGzBzNggFXnrqF1CaUcvgkdR5Ot7KZg=
 golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/mod v0.3.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
+golang.org/x/mod v0.10.0 h1:lFO9qtOdlre5W1jxS3r/4szv2/6iXxScdzjoBMXNhYk=
+golang.org/x/mod v0.10.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
+golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20181114220301-adae6a3d119a/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
+golang.org/x/net v0.0.0-20190108225652-1e06a53dbb7e/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190613194153-d28f0bde5980/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
-golang.org/x/net v0.0.0-20190923162816-aa69164e4478/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
+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 h1:qWPm9rbaAMKs8Bq/9LRpbMqxWRVUAQwMI9fVrssnTfw=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
+golang.org/x/net v0.9.0 h1:aWJ/m6xSmxWBx+V0XRHTlrYrPG56jKsLdTFmsSsCzOM=
+golang.org/x/net v0.9.0/go.mod h1:d48xBJpPfHeWQsugry2m+kC02ZBRGRgulfHnEXEuWns=
+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=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9 h1:SQFwaSi55rU7vdNs9Yr0Z324VNlrF+0wMqRXT4St8ck=
 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 v0.1.0 h1:wsuoTGHzEhffawBOhz5CYhcrV4IdKZbEyZjBMuTp12o=
 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 v0.0.0-20190204203706-41f3e6584952/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190422165155-953cdadca894/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190922100055-0a153f010e69/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190924154521-2837fb4f24fe/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20191010194322-b09406accb47/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20190606203320-7fc4e5ec1444/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20200106162015-b016eb3dc98e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20200323222414-85ca7c5b95cd/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20200615200032-f1bc736245b1/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20200625212154-ddb9806d33ae/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20201119102817-f84b799fce68 h1:nxC68pudNYkKU6jWhgrqdreuFiOQWj1Fs7T3VrH4Pjw=
+golang.org/x/sys v0.0.0-20201015000850-e3ed0017c211/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1 h1:v+OssWQX+hTHEmOBgwxdZxK4zHq3yOs8F9J7mk0PY8E=
+golang.org/x/sys v0.0.0-20210124154548-22da62e12c0c/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20210603081109-ebe580a85c40/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220715151400-c0bba94af5f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.8.0 h1:EBmGv8NaZBZTWvrbjNoL6HVt+IVy3QDQpJs7VRIw3tU=
+golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
+golang.org/x/term v0.8.0 h1:n5xxQn2i3PC0yLAbjTpNT85q/Kgzcr2gIoX9OrJUols=
+golang.org/x/term v0.8.0/go.mod h1:xPskH00ivmX89bAKVGSKKtLOWNx2+17Eiy94tnKShWo=
 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=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.0.0-20190907020128-2ca718005c18/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20200130002326-2f3ba24bd6e7/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
 golang.org/x/tools v0.0.0-20200619180055-7c47624df98f/go.mod h1:EkVYQZoAsY45+roYkvgYkIh4xh/qjgUK9TdY2XT94GE=
 golang.org/x/tools v0.0.0-20210106214847-113979e3529a/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
+golang.org/x/tools v0.8.0 h1:vSDcovVPld282ceKgDimkRSC8kpaH1dgyc9UMzlt84Y=
+golang.org/x/tools v0.8.0/go.mod h1:JxBZ99ISMI5ViVkT1tr6tdNmXeTrcpVSD3vZ1RsRdN4=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 h1:go1bK/D/BFZV2I8cIQd1NKEZ+0owSTG1fDTci4IqFcE=
 golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
+golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 h1:B82qJJgjvYKsXS9jeunTOisW56dUokqW/FOteYJJ/yg=
+golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2/go.mod h1:deeaetjYA+DHMHg+sMSMI58GrEteJUUzzw7en6TJQcI=
+golang.zx2c4.com/wireguard/windows v0.5.3 h1:On6j2Rpn3OEMXqBq00QEDC7bWSZrPIHKIus8eIuExIE=
+golang.zx2c4.com/wireguard/windows v0.5.3/go.mod h1:9TEe8TJmtwyQebdFwAkEWOPr3prrtqm+REGFifP60hI=
+google.golang.org/appengine v1.4.0/go.mod h1:xpcJRLb0r/rnEns0DIKYYv+WjYCduHsrkT7/EB5XEv4=
+google.golang.org/protobuf v0.0.0-20200109180630-ec00e32a8dfd/go.mod h1:DFci5gLYBciE7Vtevhsrf46CRTquxDuWsQurQQe4oz8=
+google.golang.org/protobuf v0.0.0-20200221191635-4d8936d0db64/go.mod h1:kwYJMbMJ01Woi6D6+Kah6886xMZcty6N08ah7+eCXa0=
+google.golang.org/protobuf v0.0.0-20200228230310-ab0ca4ff8a60/go.mod h1:cfTl7dwQJ+fmap5saPgwCLgHXTUD7jkjRqWcaiX5VyM=
+google.golang.org/protobuf v1.20.1-0.20200309200217-e05f789c0967/go.mod h1:A+miEFZTKqfCUM6K7xSMQL9OKL/b6hQv+e19PK+JZNE=
+google.golang.org/protobuf v1.21.0/go.mod h1:47Nbq4nVaFHyn7ilMalzfO3qCViNmqZ2kzikPIcrTAo=
+google.golang.org/protobuf v1.23.0/go.mod h1:EGpADcykh3NcUnDUJcl1+ZksZNG86OlYog2l/sGQquU=
 google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
-google.golang.org/protobuf v1.26.0 h1:bxAC2xTBsZGibn2RTntX0oH50xLsqy1OxA9tTL3p/lk=
 google.golang.org/protobuf v1.26.0/go.mod h1:9q0QmTI4eRPtz6boOQmLYwt+qCgq0jsYwAQnmE0givc=
+google.golang.org/protobuf v1.30.0 h1:kPPoIgf3TsEvrm0PFe15JQ+570QVxYzEvvHqChK+cng=
+google.golang.org/protobuf v1.30.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
 gopkg.in/alecthomas/kingpin.v2 v2.2.6/go.mod h1:FMv+mEhP44yOT+4EoQTLFTRgOQ1FBLkstjWtayDeSgw=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/check.v1 v1.0.0-20190902080502-41f04d3bba15/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
 gopkg.in/yaml.v2 v2.2.1/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
-gopkg.in/yaml.v2 v2.2.7 h1:VUgggvou5XRW9mHwD/yXxIYSMtY0zoKQf/v226p2nyo=
-gopkg.in/yaml.v2 v2.2.7/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
-gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c h1:dUUwHk2QECo/6vqA44rthZ8ie2QXMNeKRTHCNY2nXvo=
+gopkg.in/yaml.v2 v2.2.4/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
+gopkg.in/yaml.v2 v2.2.5/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
+gopkg.in/yaml.v2 v2.3.0/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
+gopkg.in/yaml.v2 v2.4.0 h1:D8xgwECY7CYvx+Y2n4sBz93Jn9JRvxdiyyo8CTfuKaY=
+gopkg.in/yaml.v2 v2.4.0/go.mod h1:RDklbk79AGWmwhnvt/jBztapEOGDOx6ZbXqjP6csGnQ=
 gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
+gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
+gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

handshake.go

@@ -1,24 +1,27 @@
 package nebula
 
-const (
-	handshakeIXPSK0 = 0
-	handshakeXXPSK0 = 1
+import (
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/udp"
 )
 
-func HandleIncomingHandshake(f *Interface, addr *udpAddr, packet []byte, h *Header, hostinfo *HostInfo) {
-	if !f.lightHouse.remoteAllowList.Allow(addr.IP) {
-		f.l.WithField("udpAddr", addr).Debug("lighthouse.remote_allow_list denied incoming handshake")
-		return
+func HandleIncomingHandshake(f *Interface, addr *udp.Addr, via *ViaSender, packet []byte, h *header.H, hostinfo *HostInfo) {
+	// First remote allow list check before we know the vpnIp
+	if addr != nil {
+		if !f.lightHouse.GetRemoteAllowList().AllowUnknownVpnIp(addr.IP) {
+			f.l.WithField("udpAddr", addr).Debug("lighthouse.remote_allow_list denied incoming handshake")
+			return
+		}
 	}
 
 	switch h.Subtype {
-	case handshakeIXPSK0:
+	case header.HandshakeIXPSK0:
 		switch h.MessageCounter {
 		case 1:
-			ixHandshakeStage1(f, addr, packet, h)
+			ixHandshakeStage1(f, addr, via, packet, h)
 		case 2:
 			newHostinfo, _ := f.handshakeManager.QueryIndex(h.RemoteIndex)
-			tearDown := ixHandshakeStage2(f, addr, newHostinfo, packet, h)
+			tearDown := ixHandshakeStage2(f, addr, via, newHostinfo, packet, h)
 			if tearDown && newHostinfo != nil {
 				f.handshakeManager.DeleteHostInfo(newHostinfo)
 			}

handshake_ix.go

@@ -1,18 +1,19 @@
 package nebula
 
 import (
-	"sync/atomic"
 	"time"
 
 	"github.com/flynn/noise"
-	"github.com/golang/protobuf/proto"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 )
 
 // NOISE IX Handshakes
 
 // This function constructs a handshake packet, but does not actually send it
 // Sending is done by the handshake manager
-func ixHandshakeStage0(f *Interface, vpnIp uint32, hostinfo *HostInfo) {
+func ixHandshakeStage0(f *Interface, vpnIp iputil.VpnIp, hostinfo *HostInfo) {
 	// This queries the lighthouse if we don't know a remote for the host
 	// We do it here to provoke the lighthouse to preempt our timer wheel and trigger the stage 1 packet to send
 	// more quickly, effect is a quicker handshake.
@@ -22,7 +23,7 @@ func ixHandshakeStage0(f *Interface, vpnIp uint32, hostinfo *HostInfo) {
 
 	err := f.handshakeManager.AddIndexHostInfo(hostinfo)
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(vpnIp)).
+		f.l.WithError(err).WithField("vpnIp", vpnIp).
 			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to generate index")
 		return
 	}
@@ -40,20 +41,20 @@ func ixHandshakeStage0(f *Interface, vpnIp uint32, hostinfo *HostInfo) {
 	hs := &NebulaHandshake{
 		Details: hsProto,
 	}
-	hsBytes, err = proto.Marshal(hs)
+	hsBytes, err = hs.Marshal()
 
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(vpnIp)).
+		f.l.WithError(err).WithField("vpnIp", vpnIp).
 			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to marshal handshake message")
 		return
 	}
 
-	header := HeaderEncode(make([]byte, HeaderLen), Version, uint8(handshake), handshakeIXPSK0, 0, 1)
-	atomic.AddUint64(&ci.atomicMessageCounter, 1)
+	h := header.Encode(make([]byte, header.Len), header.Version, header.Handshake, header.HandshakeIXPSK0, 0, 1)
+	ci.messageCounter.Add(1)
 
-	msg, _, _, err := ci.H.WriteMessage(header, hsBytes)
+	msg, _, _, err := ci.H.WriteMessage(h, hsBytes)
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(vpnIp)).
+		f.l.WithError(err).WithField("vpnIp", vpnIp).
 			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to call noise.WriteMessage")
 		return
 	}
@@ -67,12 +68,12 @@ func ixHandshakeStage0(f *Interface, vpnIp uint32, hostinfo *HostInfo) {
 	hostinfo.handshakeStart = time.Now()
 }
 
-func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
+func ixHandshakeStage1(f *Interface, addr *udp.Addr, via *ViaSender, packet []byte, h *header.H) {
 	ci := f.newConnectionState(f.l, false, noise.HandshakeIX, []byte{}, 0)
 	// Mark packet 1 as seen so it doesn't show up as missed
 	ci.window.Update(f.l, 1)
 
-	msg, _, _, err := ci.H.ReadMessage(nil, packet[HeaderLen:])
+	msg, _, _, err := ci.H.ReadMessage(nil, packet[header.Len:])
 	if err != nil {
 		f.l.WithError(err).WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to call noise.ReadMessage")
@@ -80,7 +81,7 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 	}
 
 	hs := &NebulaHandshake{}
-	err = proto.Unmarshal(msg, hs)
+	err = hs.Unmarshal(msg)
 	/*
 		l.Debugln("GOT INDEX: ", hs.Details.InitiatorIndex)
 	*/
@@ -97,23 +98,33 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 			Info("Invalid certificate from host")
 		return
 	}
-	vpnIP := ip2int(remoteCert.Details.Ips[0].IP)
+	vpnIp := iputil.Ip2VpnIp(remoteCert.Details.Ips[0].IP)
 	certName := remoteCert.Details.Name
 	fingerprint, _ := remoteCert.Sha256Sum()
+	issuer := remoteCert.Details.Issuer
 
-	if vpnIP == ip2int(f.certState.certificate.Details.Ips[0].IP) {
-		f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+	if vpnIp == f.myVpnIp {
+		f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Refusing to handshake with myself")
 		return
 	}
 
+	if addr != nil {
+		if !f.lightHouse.GetRemoteAllowList().Allow(vpnIp, addr.IP) {
+			f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).Debug("lighthouse.remote_allow_list denied incoming handshake")
+			return
+		}
+	}
+
 	myIndex, err := generateIndex(f.l)
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to generate index")
 		return
 	}
@@ -122,17 +133,23 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 		ConnectionState:   ci,
 		localIndexId:      myIndex,
 		remoteIndexId:     hs.Details.InitiatorIndex,
-		hostId:            vpnIP,
+		vpnIp:             vpnIp,
 		HandshakePacket:   make(map[uint8][]byte, 0),
 		lastHandshakeTime: hs.Details.Time,
+		relayState: RelayState{
+			relays:        map[iputil.VpnIp]struct{}{},
+			relayForByIp:  map[iputil.VpnIp]*Relay{},
+			relayForByIdx: map[uint32]*Relay{},
+		},
 	}
 
 	hostinfo.Lock()
 	defer hostinfo.Unlock()
 
-	f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+	f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 		WithField("certName", certName).
 		WithField("fingerprint", fingerprint).
+		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"}).
 		Info("Handshake message received")
@@ -142,33 +159,36 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 	// Update the time in case their clock is way off from ours
 	hs.Details.Time = uint64(time.Now().UnixNano())
 
-	hsBytes, err := proto.Marshal(hs)
+	hsBytes, err := hs.Marshal()
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to marshal handshake message")
 		return
 	}
 
-	header := HeaderEncode(make([]byte, HeaderLen), Version, uint8(handshake), handshakeIXPSK0, hs.Details.InitiatorIndex, 2)
-	msg, dKey, eKey, err := ci.H.WriteMessage(header, hsBytes)
+	nh := header.Encode(make([]byte, header.Len), header.Version, header.Handshake, header.HandshakeIXPSK0, hs.Details.InitiatorIndex, 2)
+	msg, dKey, eKey, err := ci.H.WriteMessage(nh, hsBytes)
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to call noise.WriteMessage")
 		return
 	} else if dKey == nil || eKey == nil {
-		f.l.WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Noise did not arrive at a key")
 		return
 	}
 
-	hostinfo.HandshakePacket[0] = make([]byte, len(packet[HeaderLen:]))
-	copy(hostinfo.HandshakePacket[0], packet[HeaderLen:])
+	hostinfo.HandshakePacket[0] = make([]byte, len(packet[header.Len:]))
+	copy(hostinfo.HandshakePacket[0], packet[header.Len:])
 
 	// Regardless of whether you are the sender or receiver, you should arrive here
 	// and complete standing up the connection.
@@ -183,68 +203,87 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 	ci.dKey = NewNebulaCipherState(dKey)
 	ci.eKey = NewNebulaCipherState(eKey)
 
-	hostinfo.remotes = f.lightHouse.QueryCache(vpnIP)
+	hostinfo.remotes = f.lightHouse.QueryCache(vpnIp)
 	hostinfo.SetRemote(addr)
 	hostinfo.CreateRemoteCIDR(remoteCert)
 
-	// Only overwrite existing record if we should win the handshake race
-	overwrite := vpnIP > ip2int(f.certState.certificate.Details.Ips[0].IP)
-	existing, err := f.handshakeManager.CheckAndComplete(hostinfo, 0, overwrite, f)
+	existing, err := f.handshakeManager.CheckAndComplete(hostinfo, 0, f)
 	if err != nil {
 		switch err {
 		case ErrAlreadySeen:
+			// Update remote if preferred (Note we have to switch to locking
+			// the existing hostinfo, and then switch back so the defer Unlock
+			// higher in this function still works)
+			hostinfo.Unlock()
+			existing.Lock()
+			// Update remote if preferred
+			if existing.SetRemoteIfPreferred(f.hostMap, addr) {
+				// Send a test packet to ensure the other side has also switched to
+				// the preferred remote
+				f.SendMessageToVpnIp(header.Test, header.TestRequest, vpnIp, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
+			}
+			existing.Unlock()
+			hostinfo.Lock()
+
 			msg = existing.HandshakePacket[2]
-			f.messageMetrics.Tx(handshake, NebulaMessageSubType(msg[1]), 1)
-			err := f.outside.WriteTo(msg, addr)
-			if err != nil {
-				f.l.WithField("vpnIp", IntIp(existing.hostId)).WithField("udpAddr", addr).
-					WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
-					WithError(err).Error("Failed to send handshake message")
+			f.messageMetrics.Tx(header.Handshake, header.MessageSubType(msg[1]), 1)
+			if addr != nil {
+				err := f.outside.WriteTo(msg, addr)
+				if err != nil {
+					f.l.WithField("vpnIp", existing.vpnIp).WithField("udpAddr", addr).
+						WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
+						WithError(err).Error("Failed to send handshake message")
+				} else {
+					f.l.WithField("vpnIp", existing.vpnIp).WithField("udpAddr", addr).
+						WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
+						Info("Handshake message sent")
+				}
+				return
 			} else {
-				f.l.WithField("vpnIp", IntIp(existing.hostId)).WithField("udpAddr", addr).
+				if via == nil {
+					f.l.Error("Handshake send failed: both addr and via are nil.")
+					return
+				}
+				hostinfo.relayState.InsertRelayTo(via.relayHI.vpnIp)
+				f.SendVia(via.relayHI, via.relay, msg, make([]byte, 12), make([]byte, mtu), false)
+				f.l.WithField("vpnIp", existing.vpnIp).WithField("relay", via.relayHI.vpnIp).
 					WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
 					Info("Handshake message sent")
+				return
 			}
-			return
 		case ErrExistingHostInfo:
 			// This means there was an existing tunnel and this handshake was older than the one we are currently based on
-			f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+			f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 				WithField("certName", certName).
 				WithField("oldHandshakeTime", existing.lastHandshakeTime).
 				WithField("newHandshakeTime", hostinfo.lastHandshakeTime).
 				WithField("fingerprint", fingerprint).
+				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"}).
 				Info("Handshake too old")
 
 			// Send a test packet to trigger an authenticated tunnel test, this should suss out any lingering tunnel issues
-			f.SendMessageToVpnIp(test, testRequest, vpnIP, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
+			f.SendMessageToVpnIp(header.Test, header.TestRequest, vpnIp, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
 			return
 		case ErrLocalIndexCollision:
 			// This means we failed to insert because of collision on localIndexId. Just let the next handshake packet retry
-			f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+			f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 				WithField("certName", certName).
 				WithField("fingerprint", fingerprint).
+				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("localIndex", hostinfo.localIndexId).WithField("collision", IntIp(existing.hostId)).
+				WithField("localIndex", hostinfo.localIndexId).WithField("collision", existing.vpnIp).
 				Error("Failed to add HostInfo due to localIndex collision")
 			return
-		case ErrExistingHandshake:
-			// We have a race where both parties think they are an initiator and this tunnel lost, let the other one finish
-			f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
-				WithField("certName", certName).
-				WithField("fingerprint", fingerprint).
-				WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
-				WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
-				Error("Prevented a pending handshake race")
-			return
 		default:
 			// Shouldn't happen, but just in case someone adds a new error type to CheckAndComplete
 			// And we forget to update it here
-			f.l.WithError(err).WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+			f.l.WithError(err).WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 				WithField("certName", certName).
 				WithField("fingerprint", fingerprint).
+				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"}).
 				Error("Failed to add HostInfo to HostMap")
@@ -253,31 +292,51 @@ func ixHandshakeStage1(f *Interface, addr *udpAddr, packet []byte, h *Header) {
 	}
 
 	// Do the send
-	f.messageMetrics.Tx(handshake, NebulaMessageSubType(msg[1]), 1)
-	err = f.outside.WriteTo(msg, addr)
-	if err != nil {
-		f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
-			WithField("certName", certName).
-			WithField("fingerprint", fingerprint).
-			WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
-			WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
-			WithError(err).Error("Failed to send handshake")
+	f.messageMetrics.Tx(header.Handshake, header.MessageSubType(msg[1]), 1)
+	if addr != nil {
+		err = f.outside.WriteTo(msg, addr)
+		if err != nil {
+			f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
+				WithField("certName", certName).
+				WithField("fingerprint", fingerprint).
+				WithField("issuer", issuer).
+				WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
+				WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+				WithError(err).Error("Failed to send handshake")
+		} else {
+			f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
+				WithField("certName", certName).
+				WithField("fingerprint", fingerprint).
+				WithField("issuer", issuer).
+				WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
+				WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+				WithField("sentCachedPackets", len(hostinfo.packetStore)).
+				Info("Handshake message sent")
+		}
 	} else {
-		f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+		if via == nil {
+			f.l.Error("Handshake send failed: both addr and via are nil.")
+			return
+		}
+		hostinfo.relayState.InsertRelayTo(via.relayHI.vpnIp)
+		f.SendVia(via.relayHI, via.relay, msg, make([]byte, 12), make([]byte, mtu), false)
+		f.l.WithField("vpnIp", vpnIp).WithField("relay", via.relayHI.vpnIp).
 			WithField("certName", certName).
 			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
 			WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
 			WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 			WithField("sentCachedPackets", len(hostinfo.packetStore)).
 			Info("Handshake message sent")
 	}
 
+	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
 	hostinfo.handshakeComplete(f.l, f.cachedPacketMetrics)
 
 	return
 }
 
-func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet []byte, h *Header) bool {
+func ixHandshakeStage2(f *Interface, addr *udp.Addr, via *ViaSender, hostinfo *HostInfo, packet []byte, h *header.H) bool {
 	if hostinfo == nil {
 		// Nothing here to tear down, got a bogus stage 2 packet
 		return true
@@ -286,21 +345,33 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 	hostinfo.Lock()
 	defer hostinfo.Unlock()
 
+	if addr != nil {
+		if !f.lightHouse.GetRemoteAllowList().Allow(hostinfo.vpnIp, addr.IP) {
+			f.l.WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).Debug("lighthouse.remote_allow_list denied incoming handshake")
+			return false
+		}
+	}
+
 	ci := hostinfo.ConnectionState
 	if ci.ready {
-		f.l.WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("header", h).
 			Info("Handshake is already complete")
 
-		//TODO: evaluate addr for preference, if we handshook with a less preferred addr we can correct quickly here
+		// Update remote if preferred
+		if hostinfo.SetRemoteIfPreferred(f.hostMap, addr) {
+			// Send a test packet to ensure the other side has also switched to
+			// the preferred remote
+			f.SendMessageToVpnIp(header.Test, header.TestRequest, hostinfo.vpnIp, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
+		}
 
 		// We already have a complete tunnel, there is nothing that can be done by processing further stage 1 packets
 		return false
 	}
 
-	msg, eKey, dKey, err := ci.H.ReadMessage(nil, packet[HeaderLen:])
+	msg, eKey, dKey, err := ci.H.ReadMessage(nil, packet[header.Len:])
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("header", h).
 			Error("Failed to call noise.ReadMessage")
 
@@ -309,7 +380,7 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 		// near future
 		return false
 	} else if dKey == nil || eKey == nil {
-		f.l.WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 			Error("Noise did not arrive at a key")
 
@@ -319,9 +390,9 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 	}
 
 	hs := &NebulaHandshake{}
-	err = proto.Unmarshal(msg, hs)
+	err = hs.Unmarshal(msg)
 	if err != nil || hs.Details == nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).Error("Failed unmarshal handshake message")
 
 		// The handshake state machine is complete, if things break now there is no chance to recover. Tear down and start again
@@ -330,7 +401,7 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 
 	remoteCert, err := RecombineCertAndValidate(ci.H, hs.Details.Cert, f.caPool)
 	if err != nil {
-		f.l.WithError(err).WithField("vpnIp", IntIp(hostinfo.hostId)).WithField("udpAddr", addr).
+		f.l.WithError(err).WithField("vpnIp", hostinfo.vpnIp).WithField("udpAddr", addr).
 			WithField("cert", remoteCert).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 			Error("Invalid certificate from host")
 
@@ -338,13 +409,14 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 		return true
 	}
 
-	vpnIP := ip2int(remoteCert.Details.Ips[0].IP)
+	vpnIp := iputil.Ip2VpnIp(remoteCert.Details.Ips[0].IP)
 	certName := remoteCert.Details.Name
 	fingerprint, _ := remoteCert.Sha256Sum()
+	issuer := remoteCert.Details.Issuer
 
 	// Ensure the right host responded
-	if vpnIP != hostinfo.hostId {
-		f.l.WithField("intendedVpnIp", IntIp(hostinfo.hostId)).WithField("haveVpnIp", IntIp(vpnIP)).
+	if vpnIp != hostinfo.vpnIp {
+		f.l.WithField("intendedVpnIp", hostinfo.vpnIp).WithField("haveVpnIp", vpnIp).
 			WithField("udpAddr", addr).WithField("certName", certName).
 			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 			Info("Incorrect host responded to handshake")
@@ -354,7 +426,7 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 
 		// Create a new hostinfo/handshake for the intended vpn ip
 		//TODO: this adds it to the timer wheel in a way that aggressively retries
-		newHostInfo := f.getOrHandshake(hostinfo.hostId)
+		newHostInfo := f.getOrHandshake(hostinfo.vpnIp)
 		newHostInfo.Lock()
 
 		// Block the current used address
@@ -362,9 +434,9 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 		newHostInfo.remotes.BlockRemote(addr)
 
 		// Get the correct remote list for the host we did handshake with
-		hostinfo.remotes = f.lightHouse.QueryCache(vpnIP)
+		hostinfo.remotes = f.lightHouse.QueryCache(vpnIp)
 
-		f.l.WithField("blockedUdpAddrs", newHostInfo.remotes.CopyBlockedRemotes()).WithField("vpnIp", IntIp(vpnIP)).
+		f.l.WithField("blockedUdpAddrs", newHostInfo.remotes.CopyBlockedRemotes()).WithField("vpnIp", vpnIp).
 			WithField("remotes", newHostInfo.remotes.CopyAddrs(f.hostMap.preferredRanges)).
 			Info("Blocked addresses for handshakes")
 
@@ -375,7 +447,7 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 		hostinfo.ConnectionState.queueLock.Unlock()
 
 		// Finally, put the correct vpn ip in the host info, tell them to close the tunnel, and return true to tear down
-		hostinfo.hostId = vpnIP
+		hostinfo.vpnIp = vpnIp
 		f.sendCloseTunnel(hostinfo)
 		newHostInfo.Unlock()
 
@@ -386,9 +458,10 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 	ci.window.Update(f.l, 2)
 
 	duration := time.Since(hostinfo.handshakeStart).Nanoseconds()
-	f.l.WithField("vpnIp", IntIp(vpnIP)).WithField("udpAddr", addr).
+	f.l.WithField("vpnIp", vpnIp).WithField("udpAddr", addr).
 		WithField("certName", certName).
 		WithField("fingerprint", fingerprint).
+		WithField("issuer", issuer).
 		WithField("initiatorIndex", hs.Details.InitiatorIndex).WithField("responderIndex", hs.Details.ResponderIndex).
 		WithField("remoteIndex", h.RemoteIndex).WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
 		WithField("durationNs", duration).
@@ -404,14 +477,18 @@ func ixHandshakeStage2(f *Interface, addr *udpAddr, hostinfo *HostInfo, packet [
 	ci.eKey = NewNebulaCipherState(eKey)
 
 	// Make sure the current udpAddr being used is set for responding
-	hostinfo.SetRemote(addr)
+	if addr != nil {
+		hostinfo.SetRemote(addr)
+	} else {
+		hostinfo.relayState.InsertRelayTo(via.relayHI.vpnIp)
+	}
 
 	// Build up the radix for the firewall if we have subnets in the cert
 	hostinfo.CreateRemoteCIDR(remoteCert)
 
 	// Complete our handshake and update metrics, this will replace any existing tunnels for this vpnIp
-	//TODO: Complete here does not do a race avoidance, it will just take the new tunnel. Is this ok?
 	f.handshakeManager.Complete(hostinfo, f)
+	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
 	hostinfo.handshakeComplete(f.l, f.cachedPacketMetrics)
 	f.metricHandshakes.Update(duration)
 

handshake_manager.go

@@ -2,6 +2,7 @@ package nebula
 
 import (
 	"bytes"
+	"context"
 	"crypto/rand"
 	"encoding/binary"
 	"errors"
@@ -10,12 +11,16 @@ import (
 
 	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 )
 
 const (
 	DefaultHandshakeTryInterval   = time.Millisecond * 100
 	DefaultHandshakeRetries       = 10
 	DefaultHandshakeTriggerBuffer = 64
+	DefaultUseRelays              = true
 )
 
 var (
@@ -23,6 +28,7 @@ var (
 		tryInterval:   DefaultHandshakeTryInterval,
 		retries:       DefaultHandshakeRetries,
 		triggerBuffer: DefaultHandshakeTriggerBuffer,
+		useRelays:     DefaultUseRelays,
 	}
 )
 
@@ -30,6 +36,7 @@ type HandshakeConfig struct {
 	tryInterval   time.Duration
 	retries       int
 	triggerBuffer int
+	useRelays     bool
 
 	messageMetrics *MessageMetrics
 }
@@ -38,27 +45,27 @@ type HandshakeManager struct {
 	pendingHostMap         *HostMap
 	mainHostMap            *HostMap
 	lightHouse             *LightHouse
-	outside                *udpConn
+	outside                *udp.Conn
 	config                 HandshakeConfig
-	OutboundHandshakeTimer *SystemTimerWheel
+	OutboundHandshakeTimer *LockingTimerWheel[iputil.VpnIp]
 	messageMetrics         *MessageMetrics
 	metricInitiated        metrics.Counter
 	metricTimedOut         metrics.Counter
 	l                      *logrus.Logger
 
-	// can be used to trigger outbound handshake for the given vpnIP
-	trigger chan uint32
+	// can be used to trigger outbound handshake for the given vpnIp
+	trigger chan iputil.VpnIp
 }
 
-func NewHandshakeManager(l *logrus.Logger, tunCidr *net.IPNet, preferredRanges []*net.IPNet, mainHostMap *HostMap, lightHouse *LightHouse, outside *udpConn, config HandshakeConfig) *HandshakeManager {
+func NewHandshakeManager(l *logrus.Logger, tunCidr *net.IPNet, preferredRanges []*net.IPNet, mainHostMap *HostMap, lightHouse *LightHouse, outside *udp.Conn, config HandshakeConfig) *HandshakeManager {
 	return &HandshakeManager{
 		pendingHostMap:         NewHostMap(l, "pending", tunCidr, preferredRanges),
 		mainHostMap:            mainHostMap,
 		lightHouse:             lightHouse,
 		outside:                outside,
 		config:                 config,
-		trigger:                make(chan uint32, config.triggerBuffer),
-		OutboundHandshakeTimer: NewSystemTimerWheel(config.tryInterval, hsTimeout(config.retries, config.tryInterval)),
+		trigger:                make(chan iputil.VpnIp, config.triggerBuffer),
+		OutboundHandshakeTimer: NewLockingTimerWheel[iputil.VpnIp](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),
@@ -66,33 +73,35 @@ func NewHandshakeManager(l *logrus.Logger, tunCidr *net.IPNet, preferredRanges [
 	}
 }
 
-func (c *HandshakeManager) Run(f EncWriter) {
-	clockSource := time.Tick(c.config.tryInterval)
+func (c *HandshakeManager) Run(ctx context.Context, f EncWriter) {
+	clockSource := time.NewTicker(c.config.tryInterval)
+	defer clockSource.Stop()
+
 	for {
 		select {
+		case <-ctx.Done():
+			return
 		case vpnIP := <-c.trigger:
-			c.l.WithField("vpnIp", IntIp(vpnIP)).Debug("HandshakeManager: triggered")
 			c.handleOutbound(vpnIP, f, true)
-		case now := <-clockSource:
+		case now := <-clockSource.C:
 			c.NextOutboundHandshakeTimerTick(now, f)
 		}
 	}
 }
 
 func (c *HandshakeManager) NextOutboundHandshakeTimerTick(now time.Time, f EncWriter) {
-	c.OutboundHandshakeTimer.advance(now)
+	c.OutboundHandshakeTimer.Advance(now)
 	for {
-		ep := c.OutboundHandshakeTimer.Purge()
-		if ep == nil {
+		vpnIp, has := c.OutboundHandshakeTimer.Purge()
+		if !has {
 			break
 		}
-		vpnIP := ep.(uint32)
-		c.handleOutbound(vpnIP, f, false)
+		c.handleOutbound(vpnIp, f, false)
 	}
 }
 
-func (c *HandshakeManager) handleOutbound(vpnIP uint32, f EncWriter, lighthouseTriggered bool) {
-	hostinfo, err := c.pendingHostMap.QueryVpnIP(vpnIP)
+func (c *HandshakeManager) handleOutbound(vpnIp iputil.VpnIp, f EncWriter, lighthouseTriggered bool) {
+	hostinfo, err := c.pendingHostMap.QueryVpnIp(vpnIp)
 	if err != nil {
 		return
 	}
@@ -110,7 +119,7 @@ func (c *HandshakeManager) handleOutbound(vpnIP uint32, f EncWriter, lighthouseT
 	if !hostinfo.HandshakeReady {
 		// There is currently a slight race in getOrHandshake due to ConnectionState not being part of the HostInfo directly
 		// Our hostinfo here was added to the pending map and the wheel may have ticked to us before we created ConnectionState
-		c.OutboundHandshakeTimer.Add(vpnIP, c.config.tryInterval*time.Duration(hostinfo.HandshakeCounter))
+		c.OutboundHandshakeTimer.Add(vpnIp, c.config.tryInterval*time.Duration(hostinfo.HandshakeCounter))
 		return
 	}
 
@@ -127,32 +136,40 @@ func (c *HandshakeManager) handleOutbound(vpnIP uint32, f EncWriter, lighthouseT
 		return
 	}
 
-	// We only care about a lighthouse trigger before the first handshake transmit attempt. This is a very specific
-	// optimization for a fast lighthouse reply
-	//TODO: it would feel better to do this once, anytime, as our delay increases over time
-	if lighthouseTriggered && hostinfo.HandshakeCounter > 0 {
+	// Get a remotes object if we don't already have one.
+	// This is mainly to protect us as this should never be the case
+	// NB ^ This comment doesn't jive. It's how the thing gets initialized.
+	// It's the common path. Should it update every time, in case a future LH query/queries give us more info?
+	if hostinfo.remotes == nil {
+		hostinfo.remotes = c.lightHouse.QueryCache(vpnIp)
+	}
+
+	remotes := hostinfo.remotes.CopyAddrs(c.pendingHostMap.preferredRanges)
+	remotesHaveChanged := !udp.AddrSlice(remotes).Equal(hostinfo.HandshakeLastRemotes)
+
+	// We only care about a lighthouse trigger if we have new remotes to send to.
+	// This is a very specific optimization for a fast lighthouse reply.
+	if lighthouseTriggered && !remotesHaveChanged {
 		// If we didn't return here a lighthouse could cause us to aggressively send handshakes
 		return
 	}
 
-	// Get a remotes object if we don't already have one.
-	// This is mainly to protect us as this should never be the case
-	if hostinfo.remotes == nil {
-		hostinfo.remotes = c.lightHouse.QueryCache(vpnIP)
-	}
+	hostinfo.HandshakeLastRemotes = remotes
 
-	//TODO: this will generate a load of queries for hosts with only 1 ip (i'm not using a lighthouse, static mapped)
-	if hostinfo.remotes.Len(c.pendingHostMap.preferredRanges) <= 1 {
+	// TODO: this will generate a load of queries for hosts with only 1 ip
+	// (such as ones registered to the lighthouse with only a private IP)
+	// So we only do it one time after attempting 5 handshakes already.
+	if len(remotes) <= 1 && hostinfo.HandshakeCounter == 5 {
 		// If we only have 1 remote it is highly likely our query raced with the other host registered within the lighthouse
-		// Our vpnIP here has a tunnel with a lighthouse but has yet to send a host update packet there so we only know about
+		// Our vpnIp here has a tunnel with a lighthouse but has yet to send a host update packet there so we only know about
 		// the learned public ip for them. Query again to short circuit the promotion counter
-		c.lightHouse.QueryServer(vpnIP, f)
+		c.lightHouse.QueryServer(vpnIp, f)
 	}
 
-	// Send a the handshake to all known ips, stage 2 takes care of assigning the hostinfo.remote based on the first to reply
-	var sentTo []*udpAddr
-	hostinfo.remotes.ForEach(c.pendingHostMap.preferredRanges, func(addr *udpAddr, _ bool) {
-		c.messageMetrics.Tx(handshake, NebulaMessageSubType(hostinfo.HandshakePacket[0][1]), 1)
+	// Send the handshake to all known ips, stage 2 takes care of assigning the hostinfo.remote based on the first to reply
+	var sentTo []*udp.Addr
+	hostinfo.remotes.ForEach(c.pendingHostMap.preferredRanges, func(addr *udp.Addr, _ bool) {
+		c.messageMetrics.Tx(header.Handshake, header.MessageSubType(hostinfo.HandshakePacket[0][1]), 1)
 		err = c.outside.WriteTo(hostinfo.HandshakePacket[0], addr)
 		if err != nil {
 			hostinfo.logger(c.l).WithField("udpAddr", addr).
@@ -165,12 +182,102 @@ func (c *HandshakeManager) handleOutbound(vpnIP uint32, f EncWriter, lighthouseT
 		}
 	})
 
-	// 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
-	if len(sentTo) > 0 {
+	// 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,
+	// so only log when the list of remotes has changed
+	if remotesHaveChanged {
 		hostinfo.logger(c.l).WithField("udpAddrs", sentTo).
 			WithField("initiatorIndex", hostinfo.localIndexId).
 			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
 			Info("Handshake message sent")
+	} else if c.l.IsLevelEnabled(logrus.DebugLevel) {
+		hostinfo.logger(c.l).WithField("udpAddrs", sentTo).
+			WithField("initiatorIndex", hostinfo.localIndexId).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Debug("Handshake message sent")
+	}
+
+	if c.config.useRelays && len(hostinfo.remotes.relays) > 0 {
+		hostinfo.logger(c.l).WithField("relays", hostinfo.remotes.relays).Info("Attempt to relay through hosts")
+		// Send a RelayRequest to all known Relay IP's
+		for _, relay := range hostinfo.remotes.relays {
+			// Don't relay to myself, and don't relay through the host I'm trying to connect to
+			if *relay == vpnIp || *relay == c.lightHouse.myVpnIp {
+				continue
+			}
+			relayHostInfo, err := c.mainHostMap.QueryVpnIp(*relay)
+			if err != nil || relayHostInfo.remote == nil {
+				hostinfo.logger(c.l).WithError(err).WithField("relay", relay.String()).Info("Establish tunnel to relay target")
+				f.Handshake(*relay)
+				continue
+			}
+			// Check the relay HostInfo to see if we already established a relay through it
+			if existingRelay, ok := relayHostInfo.relayState.QueryRelayForByIp(vpnIp); ok {
+				switch existingRelay.State {
+				case Established:
+					hostinfo.logger(c.l).WithField("relay", relay.String()).Info("Send handshake via relay")
+					f.SendVia(relayHostInfo, existingRelay, hostinfo.HandshakePacket[0], make([]byte, 12), make([]byte, mtu), false)
+				case Requested:
+					hostinfo.logger(c.l).WithField("relay", relay.String()).Info("Re-send CreateRelay request")
+					// Re-send the CreateRelay request, in case the previous one was lost.
+					m := NebulaControl{
+						Type:                NebulaControl_CreateRelayRequest,
+						InitiatorRelayIndex: existingRelay.LocalIndex,
+						RelayFromIp:         uint32(c.lightHouse.myVpnIp),
+						RelayToIp:           uint32(vpnIp),
+					}
+					msg, err := m.Marshal()
+					if err != nil {
+						hostinfo.logger(c.l).
+							WithError(err).
+							Error("Failed to marshal Control message to create relay")
+					} else {
+						// This must send over the hostinfo, not over hm.Hosts[ip]
+						f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
+						c.l.WithFields(logrus.Fields{
+							"relayFrom":           c.lightHouse.myVpnIp,
+							"relayTo":             vpnIp,
+							"initiatorRelayIndex": existingRelay.LocalIndex,
+							"relay":               *relay}).
+							Info("send CreateRelayRequest")
+					}
+				default:
+					hostinfo.logger(c.l).
+						WithField("vpnIp", vpnIp).
+						WithField("state", existingRelay.State).
+						WithField("relay", relayHostInfo.vpnIp).
+						Errorf("Relay unexpected state")
+				}
+			} else {
+				// No relays exist or requested yet.
+				if relayHostInfo.remote != nil {
+					idx, err := AddRelay(c.l, relayHostInfo, c.mainHostMap, vpnIp, nil, TerminalType, Requested)
+					if err != nil {
+						hostinfo.logger(c.l).WithField("relay", relay.String()).WithError(err).Info("Failed to add relay to hostmap")
+					}
+
+					m := NebulaControl{
+						Type:                NebulaControl_CreateRelayRequest,
+						InitiatorRelayIndex: idx,
+						RelayFromIp:         uint32(c.lightHouse.myVpnIp),
+						RelayToIp:           uint32(vpnIp),
+					}
+					msg, err := m.Marshal()
+					if err != nil {
+						hostinfo.logger(c.l).
+							WithError(err).
+							Error("Failed to marshal Control message to create relay")
+					} else {
+						f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
+						c.l.WithFields(logrus.Fields{
+							"relayFrom":           c.lightHouse.myVpnIp,
+							"relayTo":             vpnIp,
+							"initiatorRelayIndex": idx,
+							"relay":               *relay}).
+							Info("send CreateRelayRequest")
+					}
+				}
+			}
+		}
 	}
 
 	// Increment the counter to increase our delay, linear backoff
@@ -178,18 +285,17 @@ func (c *HandshakeManager) handleOutbound(vpnIP uint32, f EncWriter, lighthouseT
 
 	// If a lighthouse triggered this attempt then we are still in the timer wheel and do not need to re-add
 	if !lighthouseTriggered {
-		//TODO: feel like we dupe handshake real fast in a tight loop, why?
-		c.OutboundHandshakeTimer.Add(vpnIP, c.config.tryInterval*time.Duration(hostinfo.HandshakeCounter))
+		c.OutboundHandshakeTimer.Add(vpnIp, c.config.tryInterval*time.Duration(hostinfo.HandshakeCounter))
 	}
 }
 
-func (c *HandshakeManager) AddVpnIP(vpnIP uint32) *HostInfo {
-	hostinfo := c.pendingHostMap.AddVpnIP(vpnIP)
-	// We lock here and use an array to insert items to prevent locking the
-	// main receive thread for very long by waiting to add items to the pending map
-	//TODO: what lock?
-	c.OutboundHandshakeTimer.Add(vpnIP, c.config.tryInterval)
-	c.metricInitiated.Inc(1)
+func (c *HandshakeManager) AddVpnIp(vpnIp iputil.VpnIp, init func(*HostInfo)) *HostInfo {
+	hostinfo, created := c.pendingHostMap.AddVpnIp(vpnIp, init)
+
+	if created {
+		c.OutboundHandshakeTimer.Add(vpnIp, c.config.tryInterval)
+		c.metricInitiated.Inc(1)
+	}
 
 	return hostinfo
 }
@@ -198,36 +304,40 @@ var (
 	ErrExistingHostInfo    = errors.New("existing hostinfo")
 	ErrAlreadySeen         = errors.New("already seen")
 	ErrLocalIndexCollision = errors.New("local index collision")
-	ErrExistingHandshake   = errors.New("existing handshake")
 )
 
 // CheckAndComplete checks for any conflicts in the main and pending hostmap
 // before adding hostinfo to main. If err is nil, it was added. Otherwise err will be:
-
+//
 // ErrAlreadySeen if we already have an entry in the hostmap that has seen the
 // exact same handshake packet
 //
 // ErrExistingHostInfo if we already have an entry in the hostmap for this
-// VpnIP and the new handshake was older than the one we currently have
+// VpnIp and the new handshake was older than the one we currently have
 //
 // ErrLocalIndexCollision if we already have an entry in the main or pending
 // hostmap for the hostinfo.localIndexId.
-func (c *HandshakeManager) CheckAndComplete(hostinfo *HostInfo, handshakePacket uint8, overwrite bool, f *Interface) (*HostInfo, error) {
+func (c *HandshakeManager) CheckAndComplete(hostinfo *HostInfo, handshakePacket uint8, f *Interface) (*HostInfo, error) {
 	c.pendingHostMap.Lock()
 	defer c.pendingHostMap.Unlock()
 	c.mainHostMap.Lock()
 	defer c.mainHostMap.Unlock()
 
 	// Check if we already have a tunnel with this vpn ip
-	existingHostInfo, found := c.mainHostMap.Hosts[hostinfo.hostId]
+	existingHostInfo, found := c.mainHostMap.Hosts[hostinfo.vpnIp]
 	if found && existingHostInfo != nil {
-		// Is it just a delayed handshake packet?
-		if bytes.Equal(hostinfo.HandshakePacket[handshakePacket], existingHostInfo.HandshakePacket[handshakePacket]) {
-			return existingHostInfo, ErrAlreadySeen
+		testHostInfo := existingHostInfo
+		for testHostInfo != nil {
+			// Is it just a delayed handshake packet?
+			if bytes.Equal(hostinfo.HandshakePacket[handshakePacket], testHostInfo.HandshakePacket[handshakePacket]) {
+				return testHostInfo, ErrAlreadySeen
+			}
+
+			testHostInfo = testHostInfo.next
 		}
 
 		// Is this a newer handshake?
-		if existingHostInfo.lastHandshakeTime >= hostinfo.lastHandshakeTime {
+		if existingHostInfo.lastHandshakeTime >= hostinfo.lastHandshakeTime && !existingHostInfo.ConnectionState.initiator {
 			return existingHostInfo, ErrExistingHostInfo
 		}
 
@@ -247,69 +357,39 @@ func (c *HandshakeManager) CheckAndComplete(hostinfo *HostInfo, handshakePacket
 	}
 
 	existingRemoteIndex, found := c.mainHostMap.RemoteIndexes[hostinfo.remoteIndexId]
-	if found && existingRemoteIndex != nil && existingRemoteIndex.hostId != hostinfo.hostId {
+	if found && existingRemoteIndex != nil && existingRemoteIndex.vpnIp != hostinfo.vpnIp {
 		// We have a collision, but this can happen since we can't control
 		// the remote ID. Just log about the situation as a note.
 		hostinfo.logger(c.l).
-			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", IntIp(existingRemoteIndex.hostId)).
+			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", existingRemoteIndex.vpnIp).
 			Info("New host shadows existing host remoteIndex")
 	}
 
-	// Check if we are also handshaking with this vpn ip
-	pendingHostInfo, found := c.pendingHostMap.Hosts[hostinfo.hostId]
-	if found && pendingHostInfo != nil {
-		if !overwrite {
-			// We won, let our pending handshake win
-			return pendingHostInfo, ErrExistingHandshake
-		}
-
-		// We lost, take this handshake and move any cached packets over so they get sent
-		pendingHostInfo.ConnectionState.queueLock.Lock()
-		hostinfo.packetStore = append(hostinfo.packetStore, pendingHostInfo.packetStore...)
-		c.pendingHostMap.unlockedDeleteHostInfo(pendingHostInfo)
-		pendingHostInfo.ConnectionState.queueLock.Unlock()
-		pendingHostInfo.logger(c.l).Info("Handshake race lost, replacing pending handshake with completed tunnel")
-	}
-
-	if existingHostInfo != nil {
-		// We are going to overwrite this entry, so remove the old references
-		delete(c.mainHostMap.Hosts, existingHostInfo.hostId)
-		delete(c.mainHostMap.Indexes, existingHostInfo.localIndexId)
-		delete(c.mainHostMap.RemoteIndexes, existingHostInfo.remoteIndexId)
-	}
-
-	c.mainHostMap.addHostInfo(hostinfo, f)
+	c.mainHostMap.unlockedAddHostInfo(hostinfo, f)
 	return existingHostInfo, nil
 }
 
 // Complete is a simpler version of CheckAndComplete when we already know we
 // won't have a localIndexId collision because we already have an entry in the
-// pendingHostMap
+// pendingHostMap. An existing hostinfo is returned if there was one.
 func (c *HandshakeManager) Complete(hostinfo *HostInfo, f *Interface) {
 	c.pendingHostMap.Lock()
 	defer c.pendingHostMap.Unlock()
 	c.mainHostMap.Lock()
 	defer c.mainHostMap.Unlock()
 
-	existingHostInfo, found := c.mainHostMap.Hosts[hostinfo.hostId]
-	if found && existingHostInfo != nil {
-		// We are going to overwrite this entry, so remove the old references
-		delete(c.mainHostMap.Hosts, existingHostInfo.hostId)
-		delete(c.mainHostMap.Indexes, existingHostInfo.localIndexId)
-		delete(c.mainHostMap.RemoteIndexes, existingHostInfo.remoteIndexId)
-	}
-
 	existingRemoteIndex, found := c.mainHostMap.RemoteIndexes[hostinfo.remoteIndexId]
 	if found && existingRemoteIndex != nil {
 		// We have a collision, but this can happen since we can't control
 		// the remote ID. Just log about the situation as a note.
 		hostinfo.logger(c.l).
-			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", IntIp(existingRemoteIndex.hostId)).
+			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", existingRemoteIndex.vpnIp).
 			Info("New host shadows existing host remoteIndex")
 	}
 
-	c.mainHostMap.addHostInfo(hostinfo, f)
+	// We need to remove from the pending hostmap first to avoid undoing work when after to the main hostmap.
 	c.pendingHostMap.unlockedDeleteHostInfo(hostinfo)
+	c.mainHostMap.unlockedAddHostInfo(hostinfo, f)
 }
 
 // AddIndexHostInfo generates a unique localIndexId for this HostInfo

handshake_manager_test.go

@@ -5,26 +5,43 @@ import (
 	"testing"
 	"time"
 
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/test"
+	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
 )
 
-func Test_NewHandshakeManagerVpnIP(t *testing.T) {
-	l := NewTestLogger()
+func Test_NewHandshakeManagerVpnIp(t *testing.T) {
+	l := test.NewLogger()
 	_, tuncidr, _ := net.ParseCIDR("172.1.1.1/24")
 	_, vpncidr, _ := net.ParseCIDR("172.1.1.1/24")
 	_, localrange, _ := net.ParseCIDR("10.1.1.1/24")
-	ip := ip2int(net.ParseIP("172.1.1.2"))
+	ip := iputil.Ip2VpnIp(net.ParseIP("172.1.1.2"))
 	preferredRanges := []*net.IPNet{localrange}
 	mw := &mockEncWriter{}
 	mainHM := NewHostMap(l, "test", vpncidr, preferredRanges)
+	lh := newTestLighthouse()
 
-	blah := NewHandshakeManager(l, tuncidr, preferredRanges, mainHM, &LightHouse{}, &udpConn{}, defaultHandshakeConfig)
+	blah := NewHandshakeManager(l, tuncidr, preferredRanges, mainHM, lh, &udp.Conn{}, defaultHandshakeConfig)
 
 	now := time.Now()
 	blah.NextOutboundHandshakeTimerTick(now, mw)
 
-	i := blah.AddVpnIP(ip)
-	i.remotes = NewRemoteList()
+	var initCalled bool
+	initFunc := func(*HostInfo) {
+		initCalled = true
+	}
+
+	i := blah.AddVpnIp(ip, initFunc)
+	assert.True(t, initCalled)
+
+	initCalled = false
+	i2 := blah.AddVpnIp(ip, initFunc)
+	assert.False(t, initCalled)
+	assert.Same(t, i, i2)
+
+	i.remotes = NewRemoteList(nil)
 	i.HandshakeReady = true
 
 	// Adding something to pending should not affect the main hostmap
@@ -49,48 +66,8 @@ func Test_NewHandshakeManagerVpnIP(t *testing.T) {
 	assert.NotContains(t, blah.pendingHostMap.Hosts, ip)
 }
 
-func Test_NewHandshakeManagerTrigger(t *testing.T) {
-	l := NewTestLogger()
-	_, tuncidr, _ := net.ParseCIDR("172.1.1.1/24")
-	_, vpncidr, _ := net.ParseCIDR("172.1.1.1/24")
-	_, localrange, _ := net.ParseCIDR("10.1.1.1/24")
-	ip := ip2int(net.ParseIP("172.1.1.2"))
-	preferredRanges := []*net.IPNet{localrange}
-	mw := &mockEncWriter{}
-	mainHM := NewHostMap(l, "test", vpncidr, preferredRanges)
-	lh := &LightHouse{addrMap: make(map[uint32]*RemoteList), l: l}
-
-	blah := NewHandshakeManager(l, tuncidr, preferredRanges, mainHM, lh, &udpConn{}, defaultHandshakeConfig)
-
-	now := time.Now()
-	blah.NextOutboundHandshakeTimerTick(now, mw)
-
-	assert.Equal(t, 0, testCountTimerWheelEntries(blah.OutboundHandshakeTimer))
-
-	hi := blah.AddVpnIP(ip)
-	hi.HandshakeReady = true
-	assert.Equal(t, 1, testCountTimerWheelEntries(blah.OutboundHandshakeTimer))
-	assert.Equal(t, 0, hi.HandshakeCounter, "Should not have attempted a handshake yet")
-
-	// Trigger the same method the channel will but, this should set our remotes pointer
-	blah.handleOutbound(ip, mw, true)
-	assert.Equal(t, 1, hi.HandshakeCounter, "Trigger should have done a handshake attempt")
-	assert.NotNil(t, hi.remotes, "Manager should have set my remotes pointer")
-
-	// Make sure the trigger doesn't double schedule the timer entry
-	assert.Equal(t, 1, testCountTimerWheelEntries(blah.OutboundHandshakeTimer))
-
-	uaddr := NewUDPAddrFromString("10.1.1.1:4242")
-	hi.remotes.unlockedPrependV4(ip, NewIp4AndPort(uaddr.IP, uint32(uaddr.Port)))
-
-	// We now have remotes but only the first trigger should have pushed things forward
-	blah.handleOutbound(ip, mw, true)
-	assert.Equal(t, 1, hi.HandshakeCounter, "Trigger should have not done a handshake attempt")
-	assert.Equal(t, 1, testCountTimerWheelEntries(blah.OutboundHandshakeTimer))
-}
-
-func testCountTimerWheelEntries(tw *SystemTimerWheel) (c int) {
-	for _, i := range tw.wheel {
+func testCountTimerWheelEntries(tw *LockingTimerWheel[iputil.VpnIp]) (c int) {
+	for _, i := range tw.t.wheel {
 		n := i.Head
 		for n != nil {
 			c++
@@ -103,6 +80,16 @@ func testCountTimerWheelEntries(tw *SystemTimerWheel) (c int) {
 type mockEncWriter struct {
 }
 
-func (mw *mockEncWriter) SendMessageToVpnIp(t NebulaMessageType, st NebulaMessageSubType, vpnIp uint32, p, nb, out []byte) {
+func (mw *mockEncWriter) SendMessageToVpnIp(t header.MessageType, st header.MessageSubType, vpnIp iputil.VpnIp, p, nb, out []byte) {
 	return
 }
+
+func (mw *mockEncWriter) SendVia(via *HostInfo, relay *Relay, ad, nb, out []byte, nocopy bool) {
+	return
+}
+
+func (mw *mockEncWriter) SendMessageToHostInfo(t header.MessageType, st header.MessageSubType, hostinfo *HostInfo, p, nb, out []byte) {
+	return
+}
+
+func (mw *mockEncWriter) Handshake(vpnIP iputil.VpnIp) {}

header/header.go

@@ -1,4 +1,4 @@
-package nebula
+package header
 
 import (
 	"encoding/binary"
@@ -19,82 +19,89 @@ import (
 // |-----------------------------------------------------------------------|
 // |                               payload...                              |
 
-const (
-	Version   uint8 = 1
-	HeaderLen       = 16
-)
-
-type NebulaMessageType uint8
-type NebulaMessageSubType uint8
+type m map[string]interface{}
 
 const (
-	handshake   NebulaMessageType = 0
-	message     NebulaMessageType = 1
-	recvError   NebulaMessageType = 2
-	lightHouse  NebulaMessageType = 3
-	test        NebulaMessageType = 4
-	closeTunnel NebulaMessageType = 5
-
-	//TODO These are deprecated as of 06/12/2018 - NB
-	testRemote      NebulaMessageType = 6
-	testRemoteReply NebulaMessageType = 7
+	Version uint8 = 1
+	Len           = 16
 )
 
-var typeMap = map[NebulaMessageType]string{
-	handshake:   "handshake",
-	message:     "message",
-	recvError:   "recvError",
-	lightHouse:  "lightHouse",
-	test:        "test",
-	closeTunnel: "closeTunnel",
+type MessageType uint8
+type MessageSubType uint8
 
-	//TODO These are deprecated as of 06/12/2018 - NB
-	testRemote:      "testRemote",
-	testRemoteReply: "testRemoteReply",
+const (
+	Handshake   MessageType = 0
+	Message     MessageType = 1
+	RecvError   MessageType = 2
+	LightHouse  MessageType = 3
+	Test        MessageType = 4
+	CloseTunnel MessageType = 5
+	Control     MessageType = 6
+)
+
+var typeMap = map[MessageType]string{
+	Handshake:   "handshake",
+	Message:     "message",
+	RecvError:   "recvError",
+	LightHouse:  "lightHouse",
+	Test:        "test",
+	CloseTunnel: "closeTunnel",
+	Control:     "control",
 }
 
 const (
-	testRequest NebulaMessageSubType = 0
-	testReply   NebulaMessageSubType = 1
+	MessageNone  MessageSubType = 0
+	MessageRelay MessageSubType = 1
 )
 
-var eHeaderTooShort = errors.New("header is too short")
+const (
+	TestRequest MessageSubType = 0
+	TestReply   MessageSubType = 1
+)
 
-var subTypeTestMap = map[NebulaMessageSubType]string{
-	testRequest: "testRequest",
-	testReply:   "testReply",
+const (
+	HandshakeIXPSK0 MessageSubType = 0
+	HandshakeXXPSK0 MessageSubType = 1
+)
+
+var ErrHeaderTooShort = errors.New("header is too short")
+
+var subTypeTestMap = map[MessageSubType]string{
+	TestRequest: "testRequest",
+	TestReply:   "testReply",
 }
 
-var subTypeNoneMap = map[NebulaMessageSubType]string{0: "none"}
+var subTypeNoneMap = map[MessageSubType]string{0: "none"}
 
-var subTypeMap = map[NebulaMessageType]*map[NebulaMessageSubType]string{
-	message:     &subTypeNoneMap,
-	recvError:   &subTypeNoneMap,
-	lightHouse:  &subTypeNoneMap,
-	test:        &subTypeTestMap,
-	closeTunnel: &subTypeNoneMap,
-	handshake: {
-		handshakeIXPSK0: "ix_psk0",
+var subTypeMap = map[MessageType]*map[MessageSubType]string{
+	Message: {
+		MessageNone:  "none",
+		MessageRelay: "relay",
 	},
-	//TODO: these are deprecated
-	testRemote:      &subTypeNoneMap,
-	testRemoteReply: &subTypeNoneMap,
+	RecvError:   &subTypeNoneMap,
+	LightHouse:  &subTypeNoneMap,
+	Test:        &subTypeTestMap,
+	CloseTunnel: &subTypeNoneMap,
+	Handshake: {
+		HandshakeIXPSK0: "ix_psk0",
+	},
+	Control: &subTypeNoneMap,
 }
 
-type Header struct {
+type H struct {
 	Version        uint8
-	Type           NebulaMessageType
-	Subtype        NebulaMessageSubType
+	Type           MessageType
+	Subtype        MessageSubType
 	Reserved       uint16
 	RemoteIndex    uint32
 	MessageCounter uint64
 }
 
-// HeaderEncode uses the provided byte array to encode the provided header values into.
+// Encode uses the provided byte array to encode the provided header values into.
 // Byte array must be capped higher than HeaderLen or this will panic
-func HeaderEncode(b []byte, v uint8, t uint8, st uint8, ri uint32, c uint64) []byte {
-	b = b[:HeaderLen]
-	b[0] = byte(v<<4 | (t & 0x0f))
+func Encode(b []byte, v uint8, t MessageType, st MessageSubType, ri uint32, c uint64) []byte {
+	b = b[:Len]
+	b[0] = v<<4 | byte(t&0x0f)
 	b[1] = byte(st)
 	binary.BigEndian.PutUint16(b[2:4], 0)
 	binary.BigEndian.PutUint32(b[4:8], ri)
@@ -103,7 +110,7 @@ func HeaderEncode(b []byte, v uint8, t uint8, st uint8, ri uint32, c uint64) []b
 }
 
 // String creates a readable string representation of a header
-func (h *Header) String() string {
+func (h *H) String() string {
 	if h == nil {
 		return "<nil>"
 	}
@@ -112,7 +119,7 @@ func (h *Header) String() string {
 }
 
 // MarshalJSON creates a json string representation of a header
-func (h *Header) MarshalJSON() ([]byte, error) {
+func (h *H) MarshalJSON() ([]byte, error) {
 	return json.Marshal(m{
 		"version":        h.Version,
 		"type":           h.TypeName(),
@@ -124,24 +131,24 @@ func (h *Header) MarshalJSON() ([]byte, error) {
 }
 
 // Encode turns header into bytes
-func (h *Header) Encode(b []byte) ([]byte, error) {
+func (h *H) Encode(b []byte) ([]byte, error) {
 	if h == nil {
 		return nil, errors.New("nil header")
 	}
 
-	return HeaderEncode(b, h.Version, uint8(h.Type), uint8(h.Subtype), h.RemoteIndex, h.MessageCounter), nil
+	return Encode(b, h.Version, h.Type, h.Subtype, h.RemoteIndex, h.MessageCounter), nil
 }
 
 // Parse is a helper function to parses given bytes into new Header struct
-func (h *Header) Parse(b []byte) error {
-	if len(b) < HeaderLen {
-		return eHeaderTooShort
+func (h *H) Parse(b []byte) error {
+	if len(b) < Len {
+		return ErrHeaderTooShort
 	}
 	// get upper 4 bytes
 	h.Version = uint8((b[0] >> 4) & 0x0f)
 	// get lower 4 bytes
-	h.Type = NebulaMessageType(b[0] & 0x0f)
-	h.Subtype = NebulaMessageSubType(b[1])
+	h.Type = MessageType(b[0] & 0x0f)
+	h.Subtype = MessageSubType(b[1])
 	h.Reserved = binary.BigEndian.Uint16(b[2:4])
 	h.RemoteIndex = binary.BigEndian.Uint32(b[4:8])
 	h.MessageCounter = binary.BigEndian.Uint64(b[8:16])
@@ -149,12 +156,12 @@ func (h *Header) Parse(b []byte) error {
 }
 
 // TypeName will transform the headers message type into a human string
-func (h *Header) TypeName() string {
+func (h *H) TypeName() string {
 	return TypeName(h.Type)
 }
 
 // TypeName will transform a nebula message type into a human string
-func TypeName(t NebulaMessageType) string {
+func TypeName(t MessageType) string {
 	if n, ok := typeMap[t]; ok {
 		return n
 	}
@@ -163,12 +170,12 @@ func TypeName(t NebulaMessageType) string {
 }
 
 // SubTypeName will transform the headers message sub type into a human string
-func (h *Header) SubTypeName() string {
+func (h *H) SubTypeName() string {
 	return SubTypeName(h.Type, h.Subtype)
 }
 
 // SubTypeName will transform a nebula message sub type into a human string
-func SubTypeName(t NebulaMessageType, s NebulaMessageSubType) string {
+func SubTypeName(t MessageType, s MessageSubType) string {
 	if n, ok := subTypeMap[t]; ok {
 		if x, ok := (*n)[s]; ok {
 			return x
@@ -179,8 +186,8 @@ func SubTypeName(t NebulaMessageType, s NebulaMessageSubType) string {
 }
 
 // NewHeader turns bytes into a header
-func NewHeader(b []byte) (*Header, error) {
-	h := new(Header)
+func NewHeader(b []byte) (*H, error) {
+	h := new(H)
 	if err := h.Parse(b); err != nil {
 		return nil, err
 	}

header/header_test.go

@@ -0,0 +1,120 @@
+package header
+
+import (
+	"reflect"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+type headerTest struct {
+	expectedBytes []byte
+	*H
+}
+
+// 0001 0010 00010010
+var headerBigEndianTests = []headerTest{{
+	expectedBytes: []byte{0x54, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xa, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x9},
+	// 1010 0000
+	H: &H{
+		// 1111 1+2+4+8 = 15
+		Version:        5,
+		Type:           4,
+		Subtype:        0,
+		Reserved:       0,
+		RemoteIndex:    10,
+		MessageCounter: 9,
+	},
+},
+}
+
+func TestEncode(t *testing.T) {
+	for _, tt := range headerBigEndianTests {
+		b, err := tt.Encode(make([]byte, Len))
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		assert.Equal(t, tt.expectedBytes, b)
+	}
+}
+
+func TestParse(t *testing.T) {
+	for _, tt := range headerBigEndianTests {
+		b := tt.expectedBytes
+		parsedHeader := &H{}
+		parsedHeader.Parse(b)
+
+		if !reflect.DeepEqual(tt.H, parsedHeader) {
+			t.Fatalf("got %#v; want %#v", parsedHeader, tt.H)
+		}
+	}
+}
+
+func TestTypeName(t *testing.T) {
+	assert.Equal(t, "test", TypeName(Test))
+	assert.Equal(t, "test", (&H{Type: Test}).TypeName())
+
+	assert.Equal(t, "unknown", TypeName(99))
+	assert.Equal(t, "unknown", (&H{Type: 99}).TypeName())
+}
+
+func TestSubTypeName(t *testing.T) {
+	assert.Equal(t, "testRequest", SubTypeName(Test, TestRequest))
+	assert.Equal(t, "testRequest", (&H{Type: Test, Subtype: TestRequest}).SubTypeName())
+
+	assert.Equal(t, "unknown", SubTypeName(99, TestRequest))
+	assert.Equal(t, "unknown", (&H{Type: 99, Subtype: TestRequest}).SubTypeName())
+
+	assert.Equal(t, "unknown", SubTypeName(Test, 99))
+	assert.Equal(t, "unknown", (&H{Type: Test, Subtype: 99}).SubTypeName())
+
+	assert.Equal(t, "none", SubTypeName(Message, 0))
+	assert.Equal(t, "none", (&H{Type: Message, Subtype: 0}).SubTypeName())
+}
+
+func TestTypeMap(t *testing.T) {
+	// Force people to document this stuff
+	assert.Equal(t, map[MessageType]string{
+		Handshake:   "handshake",
+		Message:     "message",
+		RecvError:   "recvError",
+		LightHouse:  "lightHouse",
+		Test:        "test",
+		CloseTunnel: "closeTunnel",
+		Control:     "control",
+	}, typeMap)
+
+	assert.Equal(t, map[MessageType]*map[MessageSubType]string{
+		Message: {
+			MessageNone:  "none",
+			MessageRelay: "relay",
+		},
+		RecvError:   &subTypeNoneMap,
+		LightHouse:  &subTypeNoneMap,
+		Test:        &subTypeTestMap,
+		CloseTunnel: &subTypeNoneMap,
+		Handshake: {
+			HandshakeIXPSK0: "ix_psk0",
+		},
+		Control: &subTypeNoneMap,
+	}, subTypeMap)
+}
+
+func TestHeader_String(t *testing.T) {
+	assert.Equal(
+		t,
+		"ver=100 type=test subtype=testRequest reserved=0x63 remoteindex=98 messagecounter=97",
+		(&H{100, Test, TestRequest, 99, 98, 97}).String(),
+	)
+}
+
+func TestHeader_MarshalJSON(t *testing.T) {
+	b, err := (&H{100, Test, TestRequest, 99, 98, 97}).MarshalJSON()
+	assert.Nil(t, err)
+	assert.Equal(
+		t,
+		"{\"messageCounter\":97,\"remoteIndex\":98,\"reserved\":99,\"subType\":\"testRequest\",\"type\":\"test\",\"version\":100}",
+		string(b),
+	)
+}

header_test.go

@@ -1,119 +0,0 @@
-package nebula
-
-import (
-	"reflect"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-)
-
-type headerTest struct {
-	expectedBytes []byte
-	*Header
-}
-
-// 0001 0010 00010010
-var headerBigEndianTests = []headerTest{{
-	expectedBytes: []byte{0x54, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0xa, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x9},
-	// 1010 0000
-	Header: &Header{
-		// 1111 1+2+4+8 = 15
-		Version:        5,
-		Type:           4,
-		Subtype:        0,
-		Reserved:       0,
-		RemoteIndex:    10,
-		MessageCounter: 9,
-	},
-},
-}
-
-func TestEncode(t *testing.T) {
-	for _, tt := range headerBigEndianTests {
-		b, err := tt.Encode(make([]byte, HeaderLen))
-		if err != nil {
-			t.Fatal(err)
-		}
-
-		assert.Equal(t, tt.expectedBytes, b)
-	}
-}
-
-func TestParse(t *testing.T) {
-	for _, tt := range headerBigEndianTests {
-		b := tt.expectedBytes
-		parsedHeader := &Header{}
-		parsedHeader.Parse(b)
-
-		if !reflect.DeepEqual(tt.Header, parsedHeader) {
-			t.Fatalf("got %#v; want %#v", parsedHeader, tt.Header)
-		}
-	}
-}
-
-func TestTypeName(t *testing.T) {
-	assert.Equal(t, "test", TypeName(test))
-	assert.Equal(t, "test", (&Header{Type: test}).TypeName())
-
-	assert.Equal(t, "unknown", TypeName(99))
-	assert.Equal(t, "unknown", (&Header{Type: 99}).TypeName())
-}
-
-func TestSubTypeName(t *testing.T) {
-	assert.Equal(t, "testRequest", SubTypeName(test, testRequest))
-	assert.Equal(t, "testRequest", (&Header{Type: test, Subtype: testRequest}).SubTypeName())
-
-	assert.Equal(t, "unknown", SubTypeName(99, testRequest))
-	assert.Equal(t, "unknown", (&Header{Type: 99, Subtype: testRequest}).SubTypeName())
-
-	assert.Equal(t, "unknown", SubTypeName(test, 99))
-	assert.Equal(t, "unknown", (&Header{Type: test, Subtype: 99}).SubTypeName())
-
-	assert.Equal(t, "none", SubTypeName(message, 0))
-	assert.Equal(t, "none", (&Header{Type: message, Subtype: 0}).SubTypeName())
-}
-
-func TestTypeMap(t *testing.T) {
-	// Force people to document this stuff
-	assert.Equal(t, map[NebulaMessageType]string{
-		handshake:       "handshake",
-		message:         "message",
-		recvError:       "recvError",
-		lightHouse:      "lightHouse",
-		test:            "test",
-		closeTunnel:     "closeTunnel",
-		testRemote:      "testRemote",
-		testRemoteReply: "testRemoteReply",
-	}, typeMap)
-
-	assert.Equal(t, map[NebulaMessageType]*map[NebulaMessageSubType]string{
-		message:     &subTypeNoneMap,
-		recvError:   &subTypeNoneMap,
-		lightHouse:  &subTypeNoneMap,
-		test:        &subTypeTestMap,
-		closeTunnel: &subTypeNoneMap,
-		handshake: {
-			handshakeIXPSK0: "ix_psk0",
-		},
-		testRemote:      &subTypeNoneMap,
-		testRemoteReply: &subTypeNoneMap,
-	}, subTypeMap)
-}
-
-func TestHeader_String(t *testing.T) {
-	assert.Equal(
-		t,
-		"ver=100 type=test subtype=testRequest reserved=0x63 remoteindex=98 messagecounter=97",
-		(&Header{100, test, testRequest, 99, 98, 97}).String(),
-	)
-}
-
-func TestHeader_MarshalJSON(t *testing.T) {
-	b, err := (&Header{100, test, testRequest, 99, 98, 97}).MarshalJSON()
-	assert.Nil(t, err)
-	assert.Equal(
-		t,
-		"{\"messageCounter\":97,\"remoteIndex\":98,\"reserved\":99,\"subType\":\"testRequest\",\"type\":\"test\",\"version\":100}",
-		string(b),
-	)
-}

hostmap.go

@@ -11,48 +11,211 @@ import (
 	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
+	"github.com/slackhq/nebula/cidr"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/udp"
 )
 
-//const ProbeLen = 100
+// const ProbeLen = 100
 const PromoteEvery = 1000
 const ReQueryEvery = 5000
 const MaxRemotes = 10
 
+// MaxHostInfosPerVpnIp is the max number of hostinfos we will track for a given vpn ip
+// 5 allows for an initial handshake and each host pair re-handshaking twice
+const MaxHostInfosPerVpnIp = 5
+
 // How long we should prevent roaming back to the previous IP.
 // This helps prevent flapping due to packets already in flight
 const RoamingSuppressSeconds = 2
 
+const (
+	Requested = iota
+	PeerRequested
+	Established
+)
+
+const (
+	Unknowntype = iota
+	ForwardingType
+	TerminalType
+)
+
+type Relay struct {
+	Type        int
+	State       int
+	LocalIndex  uint32
+	RemoteIndex uint32
+	PeerIp      iputil.VpnIp
+}
+
 type HostMap struct {
 	sync.RWMutex    //Because we concurrently read and write to our maps
 	name            string
 	Indexes         map[uint32]*HostInfo
+	Relays          map[uint32]*HostInfo // Maps a Relay IDX to a Relay HostInfo object
 	RemoteIndexes   map[uint32]*HostInfo
-	Hosts           map[uint32]*HostInfo
+	Hosts           map[iputil.VpnIp]*HostInfo
 	preferredRanges []*net.IPNet
 	vpnCIDR         *net.IPNet
-	unsafeRoutes    *CIDRTree
 	metricsEnabled  bool
 	l               *logrus.Logger
 }
 
+// For synchronization, treat the pointed-to Relay struct as immutable. To edit the Relay
+// 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 {
+	sync.RWMutex
+
+	relays        map[iputil.VpnIp]struct{} // Set of VpnIp's of Hosts to use as relays to access this peer
+	relayForByIp  map[iputil.VpnIp]*Relay   // Maps VpnIps of peers for which this HostInfo is a relay to some Relay info
+	relayForByIdx map[uint32]*Relay         // Maps a local index to some Relay info
+}
+
+func (rs *RelayState) DeleteRelay(ip iputil.VpnIp) {
+	rs.Lock()
+	defer rs.Unlock()
+	delete(rs.relays, ip)
+}
+
+func (rs *RelayState) CopyAllRelayFor() []*Relay {
+	rs.RLock()
+	defer rs.RUnlock()
+	ret := make([]*Relay, 0, len(rs.relayForByIdx))
+	for _, r := range rs.relayForByIdx {
+		ret = append(ret, r)
+	}
+	return ret
+}
+
+func (rs *RelayState) GetRelayForByIp(ip iputil.VpnIp) (*Relay, bool) {
+	rs.RLock()
+	defer rs.RUnlock()
+	r, ok := rs.relayForByIp[ip]
+	return r, ok
+}
+
+func (rs *RelayState) InsertRelayTo(ip iputil.VpnIp) {
+	rs.Lock()
+	defer rs.Unlock()
+	rs.relays[ip] = struct{}{}
+}
+
+func (rs *RelayState) CopyRelayIps() []iputil.VpnIp {
+	rs.RLock()
+	defer rs.RUnlock()
+	ret := make([]iputil.VpnIp, 0, len(rs.relays))
+	for ip := range rs.relays {
+		ret = append(ret, ip)
+	}
+	return ret
+}
+
+func (rs *RelayState) CopyRelayForIps() []iputil.VpnIp {
+	rs.RLock()
+	defer rs.RUnlock()
+	currentRelays := make([]iputil.VpnIp, 0, len(rs.relayForByIp))
+	for relayIp := range rs.relayForByIp {
+		currentRelays = append(currentRelays, relayIp)
+	}
+	return currentRelays
+}
+
+func (rs *RelayState) CopyRelayForIdxs() []uint32 {
+	rs.RLock()
+	defer rs.RUnlock()
+	ret := make([]uint32, 0, len(rs.relayForByIdx))
+	for i := range rs.relayForByIdx {
+		ret = append(ret, i)
+	}
+	return ret
+}
+
+func (rs *RelayState) RemoveRelay(localIdx uint32) (iputil.VpnIp, bool) {
+	rs.Lock()
+	defer rs.Unlock()
+	r, ok := rs.relayForByIdx[localIdx]
+	if !ok {
+		return iputil.VpnIp(0), false
+	}
+	delete(rs.relayForByIdx, localIdx)
+	delete(rs.relayForByIp, r.PeerIp)
+	return r.PeerIp, true
+}
+
+func (rs *RelayState) CompleteRelayByIP(vpnIp iputil.VpnIp, remoteIdx uint32) bool {
+	rs.Lock()
+	defer rs.Unlock()
+	r, ok := rs.relayForByIp[vpnIp]
+	if !ok {
+		return false
+	}
+	newRelay := *r
+	newRelay.State = Established
+	newRelay.RemoteIndex = remoteIdx
+	rs.relayForByIdx[r.LocalIndex] = &newRelay
+	rs.relayForByIp[r.PeerIp] = &newRelay
+	return true
+}
+
+func (rs *RelayState) CompleteRelayByIdx(localIdx uint32, remoteIdx uint32) (*Relay, bool) {
+	rs.Lock()
+	defer rs.Unlock()
+	r, ok := rs.relayForByIdx[localIdx]
+	if !ok {
+		return nil, false
+	}
+	newRelay := *r
+	newRelay.State = Established
+	newRelay.RemoteIndex = remoteIdx
+	rs.relayForByIdx[r.LocalIndex] = &newRelay
+	rs.relayForByIp[r.PeerIp] = &newRelay
+	return &newRelay, true
+}
+
+func (rs *RelayState) QueryRelayForByIp(vpnIp iputil.VpnIp) (*Relay, bool) {
+	rs.RLock()
+	defer rs.RUnlock()
+	r, ok := rs.relayForByIp[vpnIp]
+	return r, ok
+}
+
+func (rs *RelayState) QueryRelayForByIdx(idx uint32) (*Relay, bool) {
+	rs.RLock()
+	defer rs.RUnlock()
+	r, ok := rs.relayForByIdx[idx]
+	return r, ok
+}
+
+func (rs *RelayState) InsertRelay(ip iputil.VpnIp, idx uint32, r *Relay) {
+	rs.Lock()
+	defer rs.Unlock()
+	rs.relayForByIp[ip] = r
+	rs.relayForByIdx[idx] = r
+}
+
 type HostInfo struct {
 	sync.RWMutex
 
-	remote            *udpAddr
-	remotes           *RemoteList
-	promoteCounter    uint32
-	ConnectionState   *ConnectionState
-	handshakeStart    time.Time        //todo: this an entry in the handshake manager
-	HandshakeReady    bool             //todo: being in the manager means you are ready
-	HandshakeCounter  int              //todo: another handshake manager entry
-	HandshakeComplete bool             //todo: this should go away in favor of ConnectionState.ready
-	HandshakePacket   map[uint8][]byte //todo: this is other handshake manager entry
-	packetStore       []*cachedPacket  //todo: this is other handshake manager entry
-	remoteIndexId     uint32
-	localIndexId      uint32
-	hostId            uint32
-	recvError         int
-	remoteCidr        *CIDRTree
+	remote               *udp.Addr
+	remotes              *RemoteList
+	promoteCounter       atomic.Uint32
+	ConnectionState      *ConnectionState
+	handshakeStart       time.Time        //todo: this an entry in the handshake manager
+	HandshakeReady       bool             //todo: being in the manager means you are ready
+	HandshakeCounter     int              //todo: another handshake manager entry
+	HandshakeLastRemotes []*udp.Addr      //todo: another handshake manager entry, which remotes we sent to last time
+	HandshakeComplete    bool             //todo: this should go away in favor of ConnectionState.ready
+	HandshakePacket      map[uint8][]byte //todo: this is other handshake manager entry
+	packetStore          []*cachedPacket  //todo: this is other handshake manager entry
+	remoteIndexId        uint32
+	localIndexId         uint32
+	vpnIp                iputil.VpnIp
+	recvError            int
+	remoteCidr           *cidr.Tree4
+	relayState           RelayState
 
 	// lastRebindCount is the other side of Interface.rebindCount, if these values don't match then we need to ask LH
 	// for a punch from the remote end of this tunnel. The goal being to prime their conntrack for our traffic just like
@@ -65,17 +228,27 @@ type HostInfo struct {
 	lastHandshakeTime uint64
 
 	lastRoam       time.Time
-	lastRoamRemote *udpAddr
+	lastRoamRemote *udp.Addr
+
+	// 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
+}
+
+type ViaSender struct {
+	relayHI   *HostInfo // relayHI is the host info object of the relay
+	remoteIdx uint32    // remoteIdx is the index included in the header of the received packet
+	relay     *Relay    // relay contains the rest of the relay information, including the PeerIP of the host trying to communicate with us.
 }
 
 type cachedPacket struct {
-	messageType    NebulaMessageType
-	messageSubType NebulaMessageSubType
+	messageType    header.MessageType
+	messageSubType header.MessageSubType
 	callback       packetCallback
 	packet         []byte
 }
 
-type packetCallback func(t NebulaMessageType, st NebulaMessageSubType, h *HostInfo, p, nb, out []byte)
+type packetCallback func(t header.MessageType, st header.MessageSubType, h *HostInfo, p, nb, out []byte)
 
 type cachedPacketMetrics struct {
 	sent    metrics.Counter
@@ -83,17 +256,18 @@ type cachedPacketMetrics struct {
 }
 
 func NewHostMap(l *logrus.Logger, name string, vpnCIDR *net.IPNet, preferredRanges []*net.IPNet) *HostMap {
-	h := map[uint32]*HostInfo{}
+	h := map[iputil.VpnIp]*HostInfo{}
 	i := map[uint32]*HostInfo{}
 	r := map[uint32]*HostInfo{}
+	relays := map[uint32]*HostInfo{}
 	m := HostMap{
 		name:            name,
 		Indexes:         i,
+		Relays:          relays,
 		RemoteIndexes:   r,
 		Hosts:           h,
 		preferredRanges: preferredRanges,
 		vpnCIDR:         vpnCIDR,
-		unsafeRoutes:    NewCIDRTree(),
 		l:               l,
 	}
 	return &m
@@ -105,16 +279,29 @@ func (hm *HostMap) EmitStats(name string) {
 	hostLen := len(hm.Hosts)
 	indexLen := len(hm.Indexes)
 	remoteIndexLen := len(hm.RemoteIndexes)
+	relaysLen := len(hm.Relays)
 	hm.RUnlock()
 
 	metrics.GetOrRegisterGauge("hostmap."+name+".hosts", nil).Update(int64(hostLen))
 	metrics.GetOrRegisterGauge("hostmap."+name+".indexes", nil).Update(int64(indexLen))
 	metrics.GetOrRegisterGauge("hostmap."+name+".remoteIndexes", nil).Update(int64(remoteIndexLen))
+	metrics.GetOrRegisterGauge("hostmap."+name+".relayIndexes", nil).Update(int64(relaysLen))
 }
 
-func (hm *HostMap) GetIndexByVpnIP(vpnIP uint32) (uint32, error) {
+func (hm *HostMap) RemoveRelay(localIdx uint32) {
+	hm.Lock()
+	_, ok := hm.Relays[localIdx]
+	if !ok {
+		hm.Unlock()
+		return
+	}
+	delete(hm.Relays, localIdx)
+	hm.Unlock()
+}
+
+func (hm *HostMap) GetIndexByVpnIp(vpnIp iputil.VpnIp) (uint32, error) {
 	hm.RLock()
-	if i, ok := hm.Hosts[vpnIP]; ok {
+	if i, ok := hm.Hosts[vpnIp]; ok {
 		index := i.localIndexId
 		hm.RUnlock()
 		return index, nil
@@ -123,44 +310,35 @@ func (hm *HostMap) GetIndexByVpnIP(vpnIP uint32) (uint32, error) {
 	return 0, errors.New("vpn IP not found")
 }
 
-func (hm *HostMap) Add(ip uint32, hostinfo *HostInfo) {
+func (hm *HostMap) Add(ip iputil.VpnIp, hostinfo *HostInfo) {
 	hm.Lock()
 	hm.Hosts[ip] = hostinfo
 	hm.Unlock()
 }
 
-func (hm *HostMap) AddVpnIP(vpnIP uint32) *HostInfo {
-	h := &HostInfo{}
+func (hm *HostMap) AddVpnIp(vpnIp iputil.VpnIp, init func(hostinfo *HostInfo)) (hostinfo *HostInfo, created bool) {
 	hm.RLock()
-	if _, ok := hm.Hosts[vpnIP]; !ok {
+	if h, ok := hm.Hosts[vpnIp]; !ok {
 		hm.RUnlock()
 		h = &HostInfo{
-			promoteCounter:  0,
-			hostId:          vpnIP,
+			vpnIp:           vpnIp,
 			HandshakePacket: make(map[uint8][]byte, 0),
+			relayState: RelayState{
+				relays:        map[iputil.VpnIp]struct{}{},
+				relayForByIp:  map[iputil.VpnIp]*Relay{},
+				relayForByIdx: map[uint32]*Relay{},
+			},
+		}
+		if init != nil {
+			init(h)
 		}
 		hm.Lock()
-		hm.Hosts[vpnIP] = h
+		hm.Hosts[vpnIp] = h
 		hm.Unlock()
-		return h
+		return h, true
 	} else {
-		h = hm.Hosts[vpnIP]
 		hm.RUnlock()
-		return h
-	}
-}
-
-func (hm *HostMap) DeleteVpnIP(vpnIP uint32) {
-	hm.Lock()
-	delete(hm.Hosts, vpnIP)
-	if len(hm.Hosts) == 0 {
-		hm.Hosts = map[uint32]*HostInfo{}
-	}
-	hm.Unlock()
-
-	if hm.l.Level >= logrus.DebugLevel {
-		hm.l.WithField("hostMap", m{"mapName": hm.name, "vpnIp": IntIp(vpnIP), "mapTotalSize": len(hm.Hosts)}).
-			Debug("Hostmap vpnIp deleted")
+		return h, false
 	}
 }
 
@@ -173,50 +351,13 @@ func (hm *HostMap) addRemoteIndexHostInfo(index uint32, h *HostInfo) {
 
 	if hm.l.Level > logrus.DebugLevel {
 		hm.l.WithField("hostMap", m{"mapName": hm.name, "indexNumber": index, "mapTotalSize": len(hm.Indexes),
-			"hostinfo": m{"existing": true, "localIndexId": h.localIndexId, "hostId": IntIp(h.hostId)}}).
+			"hostinfo": m{"existing": true, "localIndexId": h.localIndexId, "hostId": h.vpnIp}}).
 			Debug("Hostmap remoteIndex added")
 	}
 }
 
-func (hm *HostMap) AddVpnIPHostInfo(vpnIP uint32, h *HostInfo) {
-	hm.Lock()
-	h.hostId = vpnIP
-	hm.Hosts[vpnIP] = h
-	hm.Indexes[h.localIndexId] = h
-	hm.RemoteIndexes[h.remoteIndexId] = h
-	hm.Unlock()
-
-	if hm.l.Level > logrus.DebugLevel {
-		hm.l.WithField("hostMap", m{"mapName": hm.name, "vpnIp": IntIp(vpnIP), "mapTotalSize": len(hm.Hosts),
-			"hostinfo": m{"existing": true, "localIndexId": h.localIndexId, "hostId": IntIp(h.hostId)}}).
-			Debug("Hostmap vpnIp added")
-	}
-}
-
-// This is only called in pendingHostmap, to cleanup an inbound handshake
-func (hm *HostMap) DeleteIndex(index uint32) {
-	hm.Lock()
-	hostinfo, ok := hm.Indexes[index]
-	if ok {
-		delete(hm.Indexes, index)
-		delete(hm.RemoteIndexes, hostinfo.remoteIndexId)
-
-		// Check if we have an entry under hostId that matches the same hostinfo
-		// instance. Clean it up as well if we do.
-		hostinfo2, ok := hm.Hosts[hostinfo.hostId]
-		if ok && hostinfo2 == hostinfo {
-			delete(hm.Hosts, hostinfo.hostId)
-		}
-	}
-	hm.Unlock()
-
-	if hm.l.Level >= logrus.DebugLevel {
-		hm.l.WithField("hostMap", m{"mapName": hm.name, "indexNumber": index, "mapTotalSize": len(hm.Indexes)}).
-			Debug("Hostmap index deleted")
-	}
-}
-
-// This is used to cleanup on recv_error
+// DeleteReverseIndex is used to clean up on recv_error
+// This function should only ever be called on the pending hostmap
 func (hm *HostMap) DeleteReverseIndex(index uint32) {
 	hm.Lock()
 	hostinfo, ok := hm.RemoteIndexes[index]
@@ -227,9 +368,9 @@ func (hm *HostMap) DeleteReverseIndex(index uint32) {
 		// Check if we have an entry under hostId that matches the same hostinfo
 		// instance. Clean it up as well if we do (they might not match in pendingHostmap)
 		var hostinfo2 *HostInfo
-		hostinfo2, ok = hm.Hosts[hostinfo.hostId]
+		hostinfo2, ok = hm.Hosts[hostinfo.vpnIp]
 		if ok && hostinfo2 == hostinfo {
-			delete(hm.Hosts, hostinfo.hostId)
+			delete(hm.Hosts, hostinfo.vpnIp)
 		}
 	}
 	hm.Unlock()
@@ -240,45 +381,113 @@ func (hm *HostMap) DeleteReverseIndex(index uint32) {
 	}
 }
 
-func (hm *HostMap) DeleteHostInfo(hostinfo *HostInfo) {
+// DeleteHostInfo will fully unlink the hostinfo and return true if it was the final hostinfo for this vpn ip
+func (hm *HostMap) DeleteHostInfo(hostinfo *HostInfo) bool {
+	// Delete the host itself, ensuring it's not modified anymore
+	hm.Lock()
+	// If we have a previous or next hostinfo then we are not the last one for this vpn ip
+	final := (hostinfo.next == nil && hostinfo.prev == nil)
+	hm.unlockedDeleteHostInfo(hostinfo)
+	hm.Unlock()
+
+	return final
+}
+
+func (hm *HostMap) DeleteRelayIdx(localIdx uint32) {
 	hm.Lock()
 	defer hm.Unlock()
-	hm.unlockedDeleteHostInfo(hostinfo)
+	delete(hm.RemoteIndexes, localIdx)
+}
+
+func (hm *HostMap) MakePrimary(hostinfo *HostInfo) {
+	hm.Lock()
+	defer hm.Unlock()
+	hm.unlockedMakePrimary(hostinfo)
+}
+
+func (hm *HostMap) unlockedMakePrimary(hostinfo *HostInfo) {
+	oldHostinfo := hm.Hosts[hostinfo.vpnIp]
+	if oldHostinfo == hostinfo {
+		return
+	}
+
+	if hostinfo.prev != nil {
+		hostinfo.prev.next = hostinfo.next
+	}
+
+	if hostinfo.next != nil {
+		hostinfo.next.prev = hostinfo.prev
+	}
+
+	hm.Hosts[hostinfo.vpnIp] = hostinfo
+
+	if oldHostinfo == nil {
+		return
+	}
+
+	hostinfo.next = oldHostinfo
+	oldHostinfo.prev = hostinfo
+	hostinfo.prev = nil
 }
 
 func (hm *HostMap) unlockedDeleteHostInfo(hostinfo *HostInfo) {
-	// Check if this same hostId is in the hostmap with a different instance.
-	// This could happen if we have an entry in the pending hostmap with different
-	// index values than the one in the main hostmap.
-	hostinfo2, ok := hm.Hosts[hostinfo.hostId]
-	if ok && hostinfo2 != hostinfo {
-		delete(hm.Hosts, hostinfo2.hostId)
-		delete(hm.Indexes, hostinfo2.localIndexId)
-		delete(hm.RemoteIndexes, hostinfo2.remoteIndexId)
+	primary, ok := hm.Hosts[hostinfo.vpnIp]
+	if ok && primary == hostinfo {
+		// The vpnIp pointer points to the same hostinfo as the local index id, we can remove it
+		delete(hm.Hosts, hostinfo.vpnIp)
+		if len(hm.Hosts) == 0 {
+			hm.Hosts = map[iputil.VpnIp]*HostInfo{}
+		}
+
+		if hostinfo.next != nil {
+			// We had more than 1 hostinfo at this vpnip, promote the next in the list to primary
+			hm.Hosts[hostinfo.vpnIp] = hostinfo.next
+			// It is primary, there is no previous hostinfo now
+			hostinfo.next.prev = nil
+		}
+
+	} else {
+		// Relink if we were in the middle of multiple hostinfos for this vpn ip
+		if hostinfo.prev != nil {
+			hostinfo.prev.next = hostinfo.next
+		}
+
+		if hostinfo.next != nil {
+			hostinfo.next.prev = hostinfo.prev
+		}
 	}
 
-	delete(hm.Hosts, hostinfo.hostId)
-	if len(hm.Hosts) == 0 {
-		hm.Hosts = map[uint32]*HostInfo{}
+	hostinfo.next = nil
+	hostinfo.prev = nil
+
+	// The remote index uses index ids outside our control so lets make sure we are only removing
+	// the remote index pointer here if it points to the hostinfo we are deleting
+	hostinfo2, ok := hm.RemoteIndexes[hostinfo.remoteIndexId]
+	if ok && hostinfo2 == hostinfo {
+		delete(hm.RemoteIndexes, hostinfo.remoteIndexId)
+		if len(hm.RemoteIndexes) == 0 {
+			hm.RemoteIndexes = map[uint32]*HostInfo{}
+		}
 	}
+
 	delete(hm.Indexes, hostinfo.localIndexId)
 	if len(hm.Indexes) == 0 {
 		hm.Indexes = map[uint32]*HostInfo{}
 	}
-	delete(hm.RemoteIndexes, hostinfo.remoteIndexId)
-	if len(hm.RemoteIndexes) == 0 {
-		hm.RemoteIndexes = map[uint32]*HostInfo{}
-	}
 
 	if hm.l.Level >= logrus.DebugLevel {
 		hm.l.WithField("hostMap", m{"mapName": hm.name, "mapTotalSize": len(hm.Hosts),
-			"vpnIp": IntIp(hostinfo.hostId), "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId}).
+			"vpnIp": hostinfo.vpnIp, "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId}).
 			Debug("Hostmap hostInfo deleted")
 	}
+
+	for _, localRelayIdx := range hostinfo.relayState.CopyRelayForIdxs() {
+		delete(hm.Relays, localRelayIdx)
+	}
 }
 
 func (hm *HostMap) QueryIndex(index uint32) (*HostInfo, error) {
-	//TODO: we probably just want ot return bool instead of error, or at least a static error
+	//TODO: we probably just want to return bool instead of error, or at least a static error
 	hm.RLock()
 	if h, ok := hm.Indexes[index]; ok {
 		hm.RUnlock()
@@ -289,6 +498,31 @@ func (hm *HostMap) QueryIndex(index uint32) (*HostInfo, error) {
 	}
 }
 
+// Retrieves a HostInfo by Index. Returns whether the HostInfo is primary at time of query.
+// This helper exists so that the hostinfo.prev pointer can be read while the hostmap lock is held.
+func (hm *HostMap) QueryIndexIsPrimary(index uint32) (*HostInfo, bool, error) {
+	//TODO: we probably just want to return bool instead of error, or at least a static error
+	hm.RLock()
+	if h, ok := hm.Indexes[index]; ok {
+		hm.RUnlock()
+		return h, h.prev == nil, nil
+	} else {
+		hm.RUnlock()
+		return nil, false, errors.New("unable to find index")
+	}
+}
+func (hm *HostMap) QueryRelayIndex(index uint32) (*HostInfo, error) {
+	//TODO: we probably just want to return bool instead of error, or at least a static error
+	hm.RLock()
+	if h, ok := hm.Relays[index]; ok {
+		hm.RUnlock()
+		return h, nil
+	} else {
+		hm.RUnlock()
+		return nil, errors.New("unable to find index")
+	}
+}
+
 func (hm *HostMap) QueryReverseIndex(index uint32) (*HostInfo, error) {
 	hm.RLock()
 	if h, ok := hm.RemoteIndexes[index]; ok {
@@ -300,17 +534,35 @@ func (hm *HostMap) QueryReverseIndex(index uint32) (*HostInfo, error) {
 	}
 }
 
-func (hm *HostMap) QueryVpnIP(vpnIp uint32) (*HostInfo, error) {
-	return hm.queryVpnIP(vpnIp, nil)
+func (hm *HostMap) QueryVpnIp(vpnIp iputil.VpnIp) (*HostInfo, error) {
+	return hm.queryVpnIp(vpnIp, nil)
 }
 
-// PromoteBestQueryVpnIP will attempt to lazily switch to the best remote every
+func (hm *HostMap) QueryVpnIpRelayFor(targetIp, relayHostIp iputil.VpnIp) (*HostInfo, *Relay, error) {
+	hm.RLock()
+	defer hm.RUnlock()
+
+	h, ok := hm.Hosts[relayHostIp]
+	if !ok {
+		return nil, nil, errors.New("unable to find host")
+	}
+	for h != nil {
+		r, ok := h.relayState.QueryRelayForByIp(targetIp)
+		if ok && r.State == Established {
+			return h, r, nil
+		}
+		h = h.next
+	}
+	return nil, nil, errors.New("unable to find host with relay")
+}
+
+// PromoteBestQueryVpnIp will attempt to lazily switch to the best remote every
 // `PromoteEvery` calls to this function for a given host.
-func (hm *HostMap) PromoteBestQueryVpnIP(vpnIp uint32, ifce *Interface) (*HostInfo, error) {
-	return hm.queryVpnIP(vpnIp, ifce)
+func (hm *HostMap) PromoteBestQueryVpnIp(vpnIp iputil.VpnIp, ifce *Interface) (*HostInfo, error) {
+	return hm.queryVpnIp(vpnIp, ifce)
 }
 
-func (hm *HostMap) queryVpnIP(vpnIp uint32, promoteIfce *Interface) (*HostInfo, error) {
+func (hm *HostMap) queryVpnIp(vpnIp iputil.VpnIp, promoteIfce *Interface) (*HostInfo, error) {
 	hm.RLock()
 	if h, ok := hm.Hosts[vpnIp]; ok {
 		hm.RUnlock()
@@ -326,118 +578,80 @@ func (hm *HostMap) queryVpnIP(vpnIp uint32, promoteIfce *Interface) (*HostInfo,
 	return nil, errors.New("unable to find host")
 }
 
-func (hm *HostMap) queryUnsafeRoute(ip uint32) uint32 {
-	r := hm.unsafeRoutes.MostSpecificContains(ip)
-	if r != nil {
-		return r.(uint32)
-	} else {
-		return 0
-	}
-}
-
-// We already have the hm Lock when this is called, so make sure to not call
-// any other methods that might try to grab it again
-func (hm *HostMap) addHostInfo(hostinfo *HostInfo, f *Interface) {
+// unlockedAddHostInfo assumes you have a write-lock and will add a hostinfo object to the hostmap Indexes and RemoteIndexes maps.
+// If an entry exists for the Hosts table (vpnIp -> hostinfo) then the provided hostinfo will be made primary
+func (hm *HostMap) unlockedAddHostInfo(hostinfo *HostInfo, f *Interface) {
 	if f.serveDns {
 		remoteCert := hostinfo.ConnectionState.peerCert
 		dnsR.Add(remoteCert.Details.Name+".", remoteCert.Details.Ips[0].IP.String())
 	}
 
-	hm.Hosts[hostinfo.hostId] = hostinfo
+	existing := hm.Hosts[hostinfo.vpnIp]
+	hm.Hosts[hostinfo.vpnIp] = hostinfo
+
+	if existing != nil {
+		hostinfo.next = existing
+		existing.prev = hostinfo
+	}
+
 	hm.Indexes[hostinfo.localIndexId] = hostinfo
 	hm.RemoteIndexes[hostinfo.remoteIndexId] = hostinfo
 
 	if hm.l.Level >= logrus.DebugLevel {
-		hm.l.WithField("hostMap", m{"mapName": hm.name, "vpnIp": IntIp(hostinfo.hostId), "mapTotalSize": len(hm.Hosts),
-			"hostinfo": m{"existing": true, "localIndexId": hostinfo.localIndexId, "hostId": IntIp(hostinfo.hostId)}}).
+		hm.l.WithField("hostMap", m{"mapName": hm.name, "vpnIp": hostinfo.vpnIp, "mapTotalSize": len(hm.Hosts),
+			"hostinfo": m{"existing": true, "localIndexId": hostinfo.localIndexId, "hostId": hostinfo.vpnIp}}).
 			Debug("Hostmap vpnIp added")
 	}
-}
 
-// punchList assembles a list of all non nil RemoteList pointer entries in this hostmap
-// The caller can then do the its work outside of the read lock
-func (hm *HostMap) punchList(rl []*RemoteList) []*RemoteList {
-	hm.RLock()
-	defer hm.RUnlock()
-
-	for _, v := range hm.Hosts {
-		if v.remotes != nil {
-			rl = append(rl, v.remotes)
+	i := 1
+	check := hostinfo
+	for check != nil {
+		if i > MaxHostInfosPerVpnIp {
+			hm.unlockedDeleteHostInfo(check)
 		}
+		check = check.next
+		i++
 	}
-	return rl
-}
-
-// Punchy iterates through the result of punchList() to assemble all known addresses and sends a hole punch packet to them
-func (hm *HostMap) Punchy(conn *udpConn) {
-	var metricsTxPunchy metrics.Counter
-	if hm.metricsEnabled {
-		metricsTxPunchy = metrics.GetOrRegisterCounter("messages.tx.punchy", nil)
-	} else {
-		metricsTxPunchy = metrics.NilCounter{}
-	}
-
-	var remotes []*RemoteList
-	b := []byte{1}
-	for {
-		remotes = hm.punchList(remotes[:0])
-		for _, rl := range remotes {
-			//TODO: CopyAddrs generates garbage but ForEach locks for the work here, figure out which way is better
-			for _, addr := range rl.CopyAddrs(hm.preferredRanges) {
-				metricsTxPunchy.Inc(1)
-				conn.WriteTo(b, addr)
-			}
-		}
-		time.Sleep(time.Second * 10)
-	}
-}
-
-func (hm *HostMap) addUnsafeRoutes(routes *[]route) {
-	for _, r := range *routes {
-		hm.l.WithField("route", r.route).WithField("via", r.via).Warn("Adding UNSAFE Route")
-		hm.unsafeRoutes.AddCIDR(r.route, ip2int(*r.via))
-	}
-}
-
-func (i *HostInfo) BindConnectionState(cs *ConnectionState) {
-	i.ConnectionState = cs
 }
 
 // TryPromoteBest handles re-querying lighthouses and probing for better paths
 // NOTE: It is an error to call this if you are a lighthouse since they should not roam clients!
 func (i *HostInfo) TryPromoteBest(preferredRanges []*net.IPNet, ifce *Interface) {
-	c := atomic.AddUint32(&i.promoteCounter, 1)
+	c := i.promoteCounter.Add(1)
 	if c%PromoteEvery == 0 {
 		// The lock here is currently protecting i.remote access
 		i.RLock()
-		defer i.RUnlock()
+		remote := i.remote
+		i.RUnlock()
 
 		// return early if we are already on a preferred remote
-		rIP := i.remote.IP
-		for _, l := range preferredRanges {
-			if l.Contains(rIP) {
-				return
+		if remote != nil {
+			rIP := remote.IP
+			for _, l := range preferredRanges {
+				if l.Contains(rIP) {
+					return
+				}
 			}
 		}
 
-		i.remotes.ForEach(preferredRanges, func(addr *udpAddr, preferred bool) {
-			if addr == nil || !preferred {
+		i.remotes.ForEach(preferredRanges, func(addr *udp.Addr, preferred bool) {
+			if remote != nil && (addr == nil || !preferred) {
 				return
 			}
 
 			// Try to send a test packet to that host, this should
 			// cause it to detect a roaming event and switch remotes
-			ifce.send(test, testRequest, i.ConnectionState, i, addr, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
+			ifce.sendTo(header.Test, header.TestRequest, i.ConnectionState, i, addr, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
 		})
 	}
 
 	// Re query our lighthouses for new remotes occasionally
 	if c%ReQueryEvery == 0 && ifce.lightHouse != nil {
-		ifce.lightHouse.QueryServer(i.hostId, ifce)
+		ifce.lightHouse.QueryServer(i.vpnIp, ifce)
 	}
 }
 
-func (i *HostInfo) cachePacket(l *logrus.Logger, t NebulaMessageType, st NebulaMessageSubType, packet []byte, f packetCallback, m *cachedPacketMetrics) {
+func (i *HostInfo) cachePacket(l *logrus.Logger, t header.MessageType, st header.MessageSubType, packet []byte, f packetCallback, m *cachedPacketMetrics) {
 	//TODO: return the error so we can log with more context
 	if len(i.packetStore) < 100 {
 		tempPacket := make([]byte, len(packet))
@@ -470,7 +684,7 @@ func (i *HostInfo) handshakeComplete(l *logrus.Logger, m *cachedPacketMetrics) {
 	i.HandshakeComplete = true
 	//TODO: this should be managed by the handshake state machine to set it based on how many handshake were seen.
 	// Clamping it to 2 gets us out of the woods for now
-	atomic.StoreUint64(&i.ConnectionState.atomicMessageCounter, 2)
+	i.ConnectionState.messageCounter.Store(2)
 
 	if l.Level >= logrus.DebugLevel {
 		i.logger(l).Debugf("Sending %d stored packets", len(i.packetStore))
@@ -489,7 +703,6 @@ func (i *HostInfo) handshakeComplete(l *logrus.Logger, m *cachedPacketMetrics) {
 	i.packetStore = make([]*cachedPacket, 0)
 	i.ConnectionState.ready = true
 	i.ConnectionState.queueLock.Unlock()
-	i.ConnectionState.certState = nil
 }
 
 func (i *HostInfo) GetCert() *cert.NebulaCertificate {
@@ -499,16 +712,52 @@ func (i *HostInfo) GetCert() *cert.NebulaCertificate {
 	return nil
 }
 
-func (i *HostInfo) SetRemote(remote *udpAddr) {
+func (i *HostInfo) SetRemote(remote *udp.Addr) {
 	// We copy here because we likely got this remote from a source that reuses the object
 	if !i.remote.Equals(remote) {
 		i.remote = remote.Copy()
-		i.remotes.LearnRemote(i.hostId, remote.Copy())
+		i.remotes.LearnRemote(i.vpnIp, remote.Copy())
 	}
 }
 
-func (i *HostInfo) ClearConnectionState() {
-	i.ConnectionState = nil
+// SetRemoteIfPreferred returns true if the remote was changed. The lastRoam
+// time on the HostInfo will also be updated.
+func (i *HostInfo) SetRemoteIfPreferred(hm *HostMap, newRemote *udp.Addr) bool {
+	if newRemote == nil {
+		// relays have nil udp Addrs
+		return false
+	}
+	currentRemote := i.remote
+	if currentRemote == nil {
+		i.SetRemote(newRemote)
+		return true
+	}
+
+	// NOTE: We do this loop here instead of calling `isPreferred` in
+	// remote_list.go so that we only have to loop over preferredRanges once.
+	newIsPreferred := false
+	for _, l := range hm.preferredRanges {
+		// return early if we are already on a preferred remote
+		if l.Contains(currentRemote.IP) {
+			return false
+		}
+
+		if l.Contains(newRemote.IP) {
+			newIsPreferred = true
+		}
+	}
+
+	if newIsPreferred {
+		// Consider this a roaming event
+		i.lastRoam = time.Now()
+		i.lastRoamRemote = currentRemote.Copy()
+
+		i.SetRemote(newRemote)
+
+		return true
+	}
+
+	return false
 }
 
 func (i *HostInfo) RecvErrorExceeded() bool {
@@ -525,7 +774,7 @@ func (i *HostInfo) CreateRemoteCIDR(c *cert.NebulaCertificate) {
 		return
 	}
 
-	remoteCidr := NewCIDRTree()
+	remoteCidr := cidr.NewTree4()
 	for _, ip := range c.Details.Ips {
 		remoteCidr.AddCIDR(&net.IPNet{IP: ip.IP, Mask: net.IPMask{255, 255, 255, 255}}, struct{}{})
 	}
@@ -541,7 +790,9 @@ func (i *HostInfo) logger(l *logrus.Logger) *logrus.Entry {
 		return logrus.NewEntry(l)
 	}
 
-	li := l.WithField("vpnIp", IntIp(i.hostId))
+	li := l.WithField("vpnIp", i.vpnIp).
+		WithField("localIndex", i.localIndexId).
+		WithField("remoteIndex", i.remoteIndexId)
 
 	if connState := i.ConnectionState; connState != nil {
 		if peerCert := connState.peerCert; peerCert != nil {
@@ -552,41 +803,9 @@ func (i *HostInfo) logger(l *logrus.Logger) *logrus.Entry {
 	return li
 }
 
-//########################
-
-/*
-
-func (hm *HostMap) DebugRemotes(vpnIp uint32) string {
-	s := "\n"
-	for _, h := range hm.Hosts {
-		for _, r := range h.Remotes {
-			s += fmt.Sprintf("%s : %d ## %v\n", r.addr.IP.String(), r.addr.Port, r.probes)
-		}
-	}
-	return s
-}
-
-func (i *HostInfo) HandleReply(addr *net.UDPAddr, counter int) {
-	for _, r := range i.Remotes {
-		if r.addr.IP.Equal(addr.IP) && r.addr.Port == addr.Port {
-			r.ProbeReceived(counter)
-		}
-	}
-}
-
-func (i *HostInfo) Probes() []*Probe {
-	p := []*Probe{}
-	for _, d := range i.Remotes {
-		p = append(p, &Probe{Addr: d.addr, Counter: d.Probe()})
-	}
-	return p
-}
-
-*/
-
 // Utility functions
 
-func localIps(l *logrus.Logger, allowList *AllowList) *[]net.IP {
+func localIps(l *logrus.Logger, allowList *LocalAllowList) *[]net.IP {
 	//FIXME: This function is pretty garbage
 	var ips []net.IP
 	ifaces, _ := net.Interfaces()

hostmap_test.go

@@ -1 +1,207 @@
 package nebula
+
+import (
+	"net"
+	"testing"
+
+	"github.com/slackhq/nebula/test"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestHostMap_MakePrimary(t *testing.T) {
+	l := test.NewLogger()
+	hm := NewHostMap(
+		l, "test",
+		&net.IPNet{
+			IP:   net.IP{10, 0, 0, 1},
+			Mask: net.IPMask{255, 255, 255, 0},
+		},
+		[]*net.IPNet{},
+	)
+
+	f := &Interface{}
+
+	h1 := &HostInfo{vpnIp: 1, localIndexId: 1}
+	h2 := &HostInfo{vpnIp: 1, localIndexId: 2}
+	h3 := &HostInfo{vpnIp: 1, localIndexId: 3}
+	h4 := &HostInfo{vpnIp: 1, localIndexId: 4}
+
+	hm.unlockedAddHostInfo(h4, f)
+	hm.unlockedAddHostInfo(h3, f)
+	hm.unlockedAddHostInfo(h2, f)
+	hm.unlockedAddHostInfo(h1, f)
+
+	// Make sure we go h1 -> h2 -> h3 -> h4
+	prim, _ := hm.QueryVpnIp(1)
+	assert.Equal(t, h1.localIndexId, prim.localIndexId)
+	assert.Equal(t, h2.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h1.localIndexId, h2.prev.localIndexId)
+	assert.Equal(t, h3.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h2.localIndexId, h3.prev.localIndexId)
+	assert.Equal(t, h4.localIndexId, h3.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h4.prev.localIndexId)
+	assert.Nil(t, h4.next)
+
+	// Swap h3/middle to primary
+	hm.MakePrimary(h3)
+
+	// Make sure we go h3 -> h1 -> h2 -> h4
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h3.localIndexId, prim.localIndexId)
+	assert.Equal(t, h1.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h2.localIndexId, h1.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h1.prev.localIndexId)
+	assert.Equal(t, h4.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h1.localIndexId, h2.prev.localIndexId)
+	assert.Equal(t, h2.localIndexId, h4.prev.localIndexId)
+	assert.Nil(t, h4.next)
+
+	// Swap h4/tail to primary
+	hm.MakePrimary(h4)
+
+	// Make sure we go h4 -> h3 -> h1 -> h2
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h4.localIndexId, prim.localIndexId)
+	assert.Equal(t, h3.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h1.localIndexId, h3.next.localIndexId)
+	assert.Equal(t, h4.localIndexId, h3.prev.localIndexId)
+	assert.Equal(t, h2.localIndexId, h1.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h1.prev.localIndexId)
+	assert.Equal(t, h1.localIndexId, h2.prev.localIndexId)
+	assert.Nil(t, h2.next)
+
+	// Swap h4 again should be no-op
+	hm.MakePrimary(h4)
+
+	// Make sure we go h4 -> h3 -> h1 -> h2
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h4.localIndexId, prim.localIndexId)
+	assert.Equal(t, h3.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h1.localIndexId, h3.next.localIndexId)
+	assert.Equal(t, h4.localIndexId, h3.prev.localIndexId)
+	assert.Equal(t, h2.localIndexId, h1.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h1.prev.localIndexId)
+	assert.Equal(t, h1.localIndexId, h2.prev.localIndexId)
+	assert.Nil(t, h2.next)
+}
+
+func TestHostMap_DeleteHostInfo(t *testing.T) {
+	l := test.NewLogger()
+	hm := NewHostMap(
+		l, "test",
+		&net.IPNet{
+			IP:   net.IP{10, 0, 0, 1},
+			Mask: net.IPMask{255, 255, 255, 0},
+		},
+		[]*net.IPNet{},
+	)
+
+	f := &Interface{}
+
+	h1 := &HostInfo{vpnIp: 1, localIndexId: 1}
+	h2 := &HostInfo{vpnIp: 1, localIndexId: 2}
+	h3 := &HostInfo{vpnIp: 1, localIndexId: 3}
+	h4 := &HostInfo{vpnIp: 1, localIndexId: 4}
+	h5 := &HostInfo{vpnIp: 1, localIndexId: 5}
+	h6 := &HostInfo{vpnIp: 1, localIndexId: 6}
+
+	hm.unlockedAddHostInfo(h6, f)
+	hm.unlockedAddHostInfo(h5, f)
+	hm.unlockedAddHostInfo(h4, f)
+	hm.unlockedAddHostInfo(h3, f)
+	hm.unlockedAddHostInfo(h2, f)
+	hm.unlockedAddHostInfo(h1, f)
+
+	// h6 should be deleted
+	assert.Nil(t, h6.next)
+	assert.Nil(t, h6.prev)
+	_, err := hm.QueryIndex(h6.localIndexId)
+	assert.Error(t, err)
+
+	// Make sure we go h1 -> h2 -> h3 -> h4 -> h5
+	prim, _ := hm.QueryVpnIp(1)
+	assert.Equal(t, h1.localIndexId, prim.localIndexId)
+	assert.Equal(t, h2.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h1.localIndexId, h2.prev.localIndexId)
+	assert.Equal(t, h3.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h2.localIndexId, h3.prev.localIndexId)
+	assert.Equal(t, h4.localIndexId, h3.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h4.prev.localIndexId)
+	assert.Equal(t, h5.localIndexId, h4.next.localIndexId)
+	assert.Equal(t, h4.localIndexId, h5.prev.localIndexId)
+	assert.Nil(t, h5.next)
+
+	// Delete primary
+	hm.DeleteHostInfo(h1)
+	assert.Nil(t, h1.prev)
+	assert.Nil(t, h1.next)
+
+	// Make sure we go h2 -> h3 -> h4 -> h5
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h2.localIndexId, prim.localIndexId)
+	assert.Equal(t, h3.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h3.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h2.localIndexId, h3.prev.localIndexId)
+	assert.Equal(t, h4.localIndexId, h3.next.localIndexId)
+	assert.Equal(t, h3.localIndexId, h4.prev.localIndexId)
+	assert.Equal(t, h5.localIndexId, h4.next.localIndexId)
+	assert.Equal(t, h4.localIndexId, h5.prev.localIndexId)
+	assert.Nil(t, h5.next)
+
+	// Delete in the middle
+	hm.DeleteHostInfo(h3)
+	assert.Nil(t, h3.prev)
+	assert.Nil(t, h3.next)
+
+	// Make sure we go h2 -> h4 -> h5
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h2.localIndexId, prim.localIndexId)
+	assert.Equal(t, h4.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h4.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h2.localIndexId, h4.prev.localIndexId)
+	assert.Equal(t, h5.localIndexId, h4.next.localIndexId)
+	assert.Equal(t, h4.localIndexId, h5.prev.localIndexId)
+	assert.Nil(t, h5.next)
+
+	// Delete the tail
+	hm.DeleteHostInfo(h5)
+	assert.Nil(t, h5.prev)
+	assert.Nil(t, h5.next)
+
+	// Make sure we go h2 -> h4
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h2.localIndexId, prim.localIndexId)
+	assert.Equal(t, h4.localIndexId, prim.next.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Equal(t, h4.localIndexId, h2.next.localIndexId)
+	assert.Equal(t, h2.localIndexId, h4.prev.localIndexId)
+	assert.Nil(t, h4.next)
+
+	// Delete the head
+	hm.DeleteHostInfo(h2)
+	assert.Nil(t, h2.prev)
+	assert.Nil(t, h2.next)
+
+	// Make sure we only have h4
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Equal(t, h4.localIndexId, prim.localIndexId)
+	assert.Nil(t, prim.prev)
+	assert.Nil(t, prim.next)
+	assert.Nil(t, h4.next)
+
+	// Delete the only item
+	hm.DeleteHostInfo(h4)
+	assert.Nil(t, h4.prev)
+	assert.Nil(t, h4.next)
+
+	// Make sure we have nil
+	prim, _ = hm.QueryVpnIp(1)
+	assert.Nil(t, prim)
+}

hostmap_tester.go

@@ -0,0 +1,24 @@
+//go:build e2e_testing
+// +build e2e_testing
+
+package nebula
+
+// This file contains functions used to export information to the e2e testing framework
+
+import "github.com/slackhq/nebula/iputil"
+
+func (i *HostInfo) GetVpnIp() iputil.VpnIp {
+	return i.vpnIp
+}
+
+func (i *HostInfo) GetLocalIndex() uint32 {
+	return i.localIndexId
+}
+
+func (i *HostInfo) GetRemoteIndex() uint32 {
+	return i.remoteIndexId
+}
+
+func (i *HostInfo) GetRelayState() *RelayState {
+	return &i.relayState
+}

inside.go

@@ -1,16 +1,21 @@
 package nebula
 
 import (
-	"sync/atomic"
-
 	"github.com/flynn/noise"
 	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/firewall"
+	"github.com/slackhq/nebula/header"
+	"github.com/slackhq/nebula/iputil"
+	"github.com/slackhq/nebula/noiseutil"
+	"github.com/slackhq/nebula/udp"
 )
 
-func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *FirewallPacket, nb, out []byte, q int, localCache ConntrackCache) {
+func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet, nb, out []byte, q int, localCache firewall.ConntrackCache) {
 	err := newPacket(packet, false, fwPacket)
 	if err != nil {
-		f.l.WithField("packet", packet).Debugf("Error while validating outbound packet: %s", err)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("packet", packet).Debugf("Error while validating outbound packet: %s", err)
+		}
 		return
 	}
 
@@ -19,8 +24,19 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *FirewallPacket,
 		return
 	}
 
-	// Ignore packets from self to self
 	if fwPacket.RemoteIP == f.myVpnIp {
+		// Immediately forward packets from self to self.
+		// This should only happen on Darwin-based and FreeBSD hosts, which
+		// routes packets from the Nebula IP to the Nebula IP through the Nebula
+		// TUN device.
+		if immediatelyForwardToSelf {
+			_, err := f.readers[q].Write(packet)
+			if err != nil {
+				f.l.WithError(err).Error("Failed to forward to tun")
+			}
+		}
+		// Otherwise, drop. On linux, we should never see these packets - Linux
+		// routes packets from the nebula IP to the nebula IP through the loopback device.
 		return
 	}
 
@@ -31,8 +47,9 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *FirewallPacket,
 
 	hostinfo := f.getOrHandshake(fwPacket.RemoteIP)
 	if hostinfo == nil {
+		f.rejectInside(packet, out, q)
 		if f.l.Level >= logrus.DebugLevel {
-			f.l.WithField("vpnIp", IntIp(fwPacket.RemoteIP)).
+			f.l.WithField("vpnIp", fwPacket.RemoteIP).
 				WithField("fwPacket", fwPacket).
 				Debugln("dropping outbound packet, vpnIp not in our CIDR or in unsafe routes")
 		}
@@ -40,12 +57,12 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *FirewallPacket,
 	}
 	ci := hostinfo.ConnectionState
 
-	if ci.ready == false {
+	if !ci.ready {
 		// Because we might be sending stored packets, lock here to stop new things going to
 		// the packet queue.
 		ci.queueLock.Lock()
 		if !ci.ready {
-			hostinfo.cachePacket(f.l, message, 0, packet, f.sendMessageNow, f.cachedPacketMetrics)
+			hostinfo.cachePacket(f.l, header.Message, 0, packet, f.sendMessageNow, f.cachedPacketMetrics)
 			ci.queueLock.Unlock()
 			return
 		}
@@ -54,31 +71,63 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *FirewallPacket,
 
 	dropReason := f.firewall.Drop(packet, *fwPacket, false, hostinfo, f.caPool, localCache)
 	if dropReason == nil {
-		f.sendNoMetrics(message, 0, ci, hostinfo, hostinfo.remote, packet, nb, out, q)
+		f.sendNoMetrics(header.Message, 0, ci, hostinfo, nil, packet, nb, out, q)
 
-	} else if f.l.Level >= logrus.DebugLevel {
-		hostinfo.logger(f.l).
-			WithField("fwPacket", fwPacket).
-			WithField("reason", dropReason).
-			Debugln("dropping outbound packet")
+	} else {
+		f.rejectInside(packet, out, q)
+		if f.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(f.l).
+				WithField("fwPacket", fwPacket).
+				WithField("reason", dropReason).
+				Debugln("dropping outbound packet")
+		}
 	}
 }
 
+func (f *Interface) rejectInside(packet []byte, out []byte, q int) {
+	if !f.firewall.InSendReject {
+		return
+	}
+
+	out = iputil.CreateRejectPacket(packet, out)
+	_, err := f.readers[q].Write(out)
+	if err != nil {
+		f.l.WithError(err).Error("Failed to write to tun")
+	}
+}
+
+func (f *Interface) rejectOutside(packet []byte, ci *ConnectionState, hostinfo *HostInfo, nb, out []byte, q int) {
+	if !f.firewall.OutSendReject {
+		return
+	}
+
+	// Use some out buffer space to build the packet before encryption
+	// Need 40 bytes for the reject packet (20 byte ipv4 header, 20 byte tcp rst packet)
+	// Leave 100 bytes for the encrypted packet (60 byte Nebula header, 40 byte reject packet)
+	out = out[:140]
+	outPacket := iputil.CreateRejectPacket(packet, out[100:])
+	f.sendNoMetrics(header.Message, 0, ci, hostinfo, nil, outPacket, nb, out, q)
+}
+
+func (f *Interface) Handshake(vpnIp iputil.VpnIp) {
+	f.getOrHandshake(vpnIp)
+}
+
 // getOrHandshake returns nil if the vpnIp is not routable
-func (f *Interface) getOrHandshake(vpnIp uint32) *HostInfo {
-	if f.hostMap.vpnCIDR.Contains(int2ip(vpnIp)) == false {
-		vpnIp = f.hostMap.queryUnsafeRoute(vpnIp)
+func (f *Interface) getOrHandshake(vpnIp iputil.VpnIp) *HostInfo {
+	if !ipMaskContains(f.lightHouse.myVpnIp, f.lightHouse.myVpnZeros, vpnIp) {
+		vpnIp = f.inside.RouteFor(vpnIp)
 		if vpnIp == 0 {
 			return nil
 		}
 	}
-	hostinfo, err := f.hostMap.PromoteBestQueryVpnIP(vpnIp, f)
+	hostinfo, err := f.hostMap.PromoteBestQueryVpnIp(vpnIp, f)
 
 	//if err != nil || hostinfo.ConnectionState == nil {
 	if err != nil {
-		hostinfo, err = f.handshakeManager.pendingHostMap.QueryVpnIP(vpnIp)
+		hostinfo, err = f.handshakeManager.pendingHostMap.QueryVpnIp(vpnIp)
 		if err != nil {
-			hostinfo = f.handshakeManager.AddVpnIP(vpnIp)
+			hostinfo = f.handshakeManager.AddVpnIp(vpnIp, f.initHostInfo)
 		}
 	}
 	ci := hostinfo.ConnectionState
@@ -97,16 +146,6 @@ func (f *Interface) getOrHandshake(vpnIp uint32) *HostInfo {
 		return hostinfo
 	}
 
-	if ci == nil {
-		// if we don't have a connection state, then send a handshake initiation
-		ci = f.newConnectionState(f.l, true, noise.HandshakeIX, []byte{}, 0)
-		// FIXME: Maybe make XX selectable, but probably not since psk makes it nearly pointless for us.
-		//ci = f.newConnectionState(true, noise.HandshakeXX, []byte{}, 0)
-		hostinfo.ConnectionState = ci
-	} else if ci.eKey == nil {
-		// if we don't have any state at all, create it
-	}
-
 	// If we have already created the handshake packet, we don't want to call the function at all.
 	if !hostinfo.HandshakeReady {
 		ixHandshakeStage0(f, vpnIp, hostinfo)
@@ -115,7 +154,13 @@ func (f *Interface) getOrHandshake(vpnIp uint32) *HostInfo {
 
 		// If this is a static host, we don't need to wait for the HostQueryReply
 		// We can trigger the handshake right now
-		if _, ok := f.lightHouse.staticList[vpnIp]; ok {
+		_, doTrigger := f.lightHouse.GetStaticHostList()[vpnIp]
+		if !doTrigger {
+			// Add any calculated remotes, and trigger early handshake if one found
+			doTrigger = f.lightHouse.addCalculatedRemotes(vpnIp)
+		}
+
+		if doTrigger {
 			select {
 			case f.handshakeManager.trigger <- vpnIp:
 			default:
@@ -126,8 +171,14 @@ func (f *Interface) getOrHandshake(vpnIp uint32) *HostInfo {
 	return hostinfo
 }
 
-func (f *Interface) sendMessageNow(t NebulaMessageType, st NebulaMessageSubType, hostInfo *HostInfo, p, nb, out []byte) {
-	fp := &FirewallPacket{}
+// initHostInfo is the init function to pass to (*HandshakeManager).AddVpnIP that
+// will create the initial Noise ConnectionState
+func (f *Interface) initHostInfo(hostinfo *HostInfo) {
+	hostinfo.ConnectionState = f.newConnectionState(f.l, true, noise.HandshakeIX, []byte{}, 0)
+}
+
+func (f *Interface) sendMessageNow(t header.MessageType, st header.MessageSubType, hostinfo *HostInfo, p, nb, out []byte) {
+	fp := &firewall.Packet{}
 	err := newPacket(p, false, fp)
 	if err != nil {
 		f.l.Warnf("error while parsing outgoing packet for firewall check; %v", err)
@@ -135,7 +186,7 @@ func (f *Interface) sendMessageNow(t NebulaMessageType, st NebulaMessageSubType,
 	}
 
 	// check if packet is in outbound fw rules
-	dropReason := f.firewall.Drop(p, *fp, false, hostInfo, f.caPool, nil)
+	dropReason := f.firewall.Drop(p, *fp, false, hostinfo, f.caPool, nil)
 	if dropReason != nil {
 		if f.l.Level >= logrus.DebugLevel {
 			f.l.WithField("fwPacket", fp).
@@ -145,15 +196,15 @@ func (f *Interface) sendMessageNow(t NebulaMessageType, st NebulaMessageSubType,
 		return
 	}
 
-	f.sendNoMetrics(message, st, hostInfo.ConnectionState, hostInfo, hostInfo.remote, p, nb, out, 0)
+	f.sendNoMetrics(header.Message, st, hostinfo.ConnectionState, hostinfo, nil, p, nb, out, 0)
 }
 
 // SendMessageToVpnIp handles real ip:port lookup and sends to the current best known address for vpnIp
-func (f *Interface) SendMessageToVpnIp(t NebulaMessageType, st NebulaMessageSubType, vpnIp uint32, p, nb, out []byte) {
+func (f *Interface) SendMessageToVpnIp(t header.MessageType, st header.MessageSubType, vpnIp iputil.VpnIp, p, nb, out []byte) {
 	hostInfo := f.getOrHandshake(vpnIp)
 	if hostInfo == nil {
 		if f.l.Level >= logrus.DebugLevel {
-			f.l.WithField("vpnIp", IntIp(vpnIp)).
+			f.l.WithField("vpnIp", vpnIp).
 				Debugln("dropping SendMessageToVpnIp, vpnIp not in our CIDR or in unsafe routes")
 		}
 		return
@@ -164,56 +215,140 @@ func (f *Interface) SendMessageToVpnIp(t NebulaMessageType, st NebulaMessageSubT
 		// the packet queue.
 		hostInfo.ConnectionState.queueLock.Lock()
 		if !hostInfo.ConnectionState.ready {
-			hostInfo.cachePacket(f.l, t, st, p, f.sendMessageToVpnIp, f.cachedPacketMetrics)
+			hostInfo.cachePacket(f.l, t, st, p, f.SendMessageToHostInfo, f.cachedPacketMetrics)
 			hostInfo.ConnectionState.queueLock.Unlock()
 			return
 		}
 		hostInfo.ConnectionState.queueLock.Unlock()
 	}
 
-	f.sendMessageToVpnIp(t, st, hostInfo, p, nb, out)
-	return
+	f.SendMessageToHostInfo(t, st, hostInfo, p, nb, out)
 }
 
-func (f *Interface) sendMessageToVpnIp(t NebulaMessageType, st NebulaMessageSubType, hostInfo *HostInfo, p, nb, out []byte) {
-	f.send(t, st, hostInfo.ConnectionState, hostInfo, hostInfo.remote, p, nb, out)
+func (f *Interface) SendMessageToHostInfo(t header.MessageType, st header.MessageSubType, hi *HostInfo, p, nb, out []byte) {
+	f.send(t, st, hi.ConnectionState, hi, p, nb, out)
 }
 
-func (f *Interface) send(t NebulaMessageType, st NebulaMessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote *udpAddr, p, nb, out []byte) {
+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, nil, p, nb, out, 0)
+}
+
+func (f *Interface) sendTo(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote *udp.Addr, p, nb, out []byte) {
 	f.messageMetrics.Tx(t, st, 1)
 	f.sendNoMetrics(t, st, ci, hostinfo, remote, p, nb, out, 0)
 }
 
-func (f *Interface) sendNoMetrics(t NebulaMessageType, st NebulaMessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote *udpAddr, p, nb, out []byte, q int) {
+// sendVia sends a payload through a Relay tunnel. No authentication or encryption is done
+// to the payload for the ultimate target host, making this a useful method for sending
+// handshake messages to peers through relay tunnels.
+// via is the HostInfo through which the message is relayed.
+// ad is the plaintext data to authenticate, but not encrypt
+// nb is a buffer used to store the nonce value, re-used for performance reasons.
+// out is a buffer used to store the result of the Encrypt operation
+// q indicates which writer to use to send the packet.
+func (f *Interface) SendVia(via *HostInfo,
+	relay *Relay,
+	ad,
+	nb,
+	out []byte,
+	nocopy bool,
+) {
+	if noiseutil.EncryptLockNeeded {
+		// NOTE: for goboring AESGCMTLS we need to lock because of the nonce check
+		via.ConnectionState.writeLock.Lock()
+	}
+	c := via.ConnectionState.messageCounter.Add(1)
+
+	out = header.Encode(out, header.Version, header.Message, header.MessageRelay, relay.RemoteIndex, c)
+	f.connectionManager.Out(via.localIndexId)
+
+	// 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.
+	if len(out)+len(ad)+via.ConnectionState.eKey.Overhead() > cap(out) {
+		if noiseutil.EncryptLockNeeded {
+			via.ConnectionState.writeLock.Unlock()
+		}
+		via.logger(f.l).
+			WithField("outCap", cap(out)).
+			WithField("payloadLen", len(ad)).
+			WithField("headerLen", len(out)).
+			WithField("cipherOverhead", via.ConnectionState.eKey.Overhead()).
+			Error("SendVia out buffer not large enough for relay")
+		return
+	}
+
+	// The header bytes are written to the 'out' slice; Grow the slice to hold the header and associated data payload.
+	offset := len(out)
+	out = out[:offset+len(ad)]
+
+	// In one call path, the associated data _is_ already stored in out. In other call paths, the associated data must
+	// be copied into 'out'.
+	if !nocopy {
+		copy(out[offset:], ad)
+	}
+
+	var err error
+	out, err = via.ConnectionState.eKey.EncryptDanger(out, out, nil, c, nb)
+	if noiseutil.EncryptLockNeeded {
+		via.ConnectionState.writeLock.Unlock()
+	}
+	if err != nil {
+		via.logger(f.l).WithError(err).Info("Failed to EncryptDanger in sendVia")
+		return
+	}
+	err = f.writers[0].WriteTo(out, via.remote)
+	if err != nil {
+		via.logger(f.l).WithError(err).Info("Failed to WriteTo in sendVia")
+	}
+	f.connectionManager.RelayUsed(relay.LocalIndex)
+}
+
+func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote *udp.Addr, p, nb, out []byte, q int) {
 	if ci.eKey == nil {
 		//TODO: log warning
 		return
 	}
+	useRelay := remote == nil && hostinfo.remote == nil
+	fullOut := out
 
-	var err error
-	//TODO: enable if we do more than 1 tun queue
-	//ci.writeLock.Lock()
-	c := atomic.AddUint64(&ci.atomicMessageCounter, 1)
+	if useRelay {
+		if len(out) < header.Len {
+			// out always has a capacity of mtu, but not always a length greater than the header.Len.
+			// Grow it to make sure the next operation works.
+			out = out[:header.Len]
+		}
+		// Save a header's worth of data at the front of the 'out' buffer.
+		out = out[header.Len:]
+	}
+
+	if noiseutil.EncryptLockNeeded {
+		// NOTE: for goboring AESGCMTLS we need to lock because of the nonce check
+		ci.writeLock.Lock()
+	}
+	c := ci.messageCounter.Add(1)
 
 	//l.WithField("trace", string(debug.Stack())).Error("out Header ", &Header{Version, t, st, 0, hostinfo.remoteIndexId, c}, p)
-	out = HeaderEncode(out, Version, uint8(t), uint8(st), hostinfo.remoteIndexId, c)
-	f.connectionManager.Out(hostinfo.hostId)
+	out = header.Encode(out, header.Version, t, st, hostinfo.remoteIndexId, c)
+	f.connectionManager.Out(hostinfo.localIndexId)
 
 	// 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 IPs and enable a faster roaming.
-	if t != closeTunnel && hostinfo.lastRebindCount != f.rebindCount {
+	if t != header.CloseTunnel && hostinfo.lastRebindCount != f.rebindCount {
 		//NOTE: there is an update hole if a tunnel isn't used and exactly 256 rebinds occur before the tunnel is
 		// finally used again. This tunnel would eventually be torn down and recreated if this action didn't help.
-		f.lightHouse.QueryServer(hostinfo.hostId, f)
+		f.lightHouse.QueryServer(hostinfo.vpnIp, f)
 		hostinfo.lastRebindCount = f.rebindCount
 		if f.l.Level >= logrus.DebugLevel {
-			f.l.WithField("vpnIp", hostinfo.hostId).Debug("Lighthouse update triggered for punch due to rebind counter")
+			f.l.WithField("vpnIp", hostinfo.vpnIp).Debug("Lighthouse update triggered for punch due to rebind counter")
 		}
 	}
 
+	var err error
 	out, err = ci.eKey.EncryptDanger(out, out, p, c, nb)
-	//TODO: see above note on lock
-	//ci.writeLock.Unlock()
+	if noiseutil.EncryptLockNeeded {
+		ci.writeLock.Unlock()
+	}
 	if err != nil {
 		hostinfo.logger(f.l).WithError(err).
 			WithField("udpAddr", remote).WithField("counter", c).
@@ -222,19 +357,34 @@ func (f *Interface) sendNoMetrics(t NebulaMessageType, st NebulaMessageSubType,
 		return
 	}
 
-	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")
+	if remote != nil {
+		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 != nil {
+		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")
+		}
+	} else {
+		// Try to send via a relay
+		for _, relayIP := range hostinfo.relayState.CopyRelayIps() {
+			relayHostInfo, relay, err := f.hostMap.QueryVpnIpRelayFor(hostinfo.vpnIp, relayIP)
+			if err != nil {
+				hostinfo.relayState.DeleteRelay(relayIP)
+				hostinfo.logger(f.l).WithField("relay", relayIP).WithError(err).Info("sendNoMetrics failed to find HostInfo")
+				continue
+			}
+			f.SendVia(relayHostInfo, relay, out, nb, fullOut[:header.Len+len(out)], true)
+			break
+		}
 	}
-	return
 }
 
-func isMulticast(ip uint32) bool {
+func isMulticast(ip iputil.VpnIp) bool {
 	// Class D multicast
-	if (((ip >> 24) & 0xff) & 0xf0) == 0xe0 {
-		return true
-	}
-
-	return false
+	return (((ip >> 24) & 0xff) & 0xf0) == 0xe0
 }