avoid heap problems

.github/workflows/release.yml

@@ -24,7 +24,7 @@ jobs:
           mv build/*.tar.gz release
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v5
         with:
           name: linux-latest
           path: release
@@ -55,7 +55,7 @@ jobs:
           mv dist\windows\wintun build\dist\windows\
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v5
         with:
           name: windows-latest
           path: build
@@ -104,7 +104,7 @@ jobs:
           fi
 
       - name: Upload artifacts
-        uses: actions/upload-artifact@v6
+        uses: actions/upload-artifact@v5
         with:
           name: darwin-latest
           path: ./release/*
@@ -128,7 +128,7 @@ jobs:
 
       - name: Download artifacts
         if: ${{ env.HAS_DOCKER_CREDS == 'true' }}
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v6
         with:
           name: linux-latest
           path: artifacts
@@ -163,7 +163,7 @@ jobs:
       - uses: actions/checkout@v6
 
       - name: Download artifacts
-        uses: actions/download-artifact@v7
+        uses: actions/download-artifact@v6
         with:
           path: artifacts
 
@@ -209,11 +209,10 @@ jobs:
         id: create_release
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          GITHUB_REF_NAME: ${{ github.ref_name }}
         run: |
           cd artifacts
           gh release create \
             --verify-tag \
-            --title "Release ${GITHUB_REF_NAME}" \
-            "${GITHUB_REF_NAME}" \
+            --title "Release ${{ github.ref_name }}" \
+            "${{ github.ref_name }}" \
             SHASUM256.txt *-latest/*.zip *-latest/*.tar.gz

.github/workflows/smoke-extra.yml

@@ -18,8 +18,6 @@ jobs:
     if: github.ref == 'refs/heads/master' || contains(github.event.pull_request.labels.*.name, 'smoke-test-extra')
     name: Run extra smoke tests
     runs-on: ubuntu-latest
-    env:
-      VAGRANT_DEFAULT_PROVIDER: libvirt
     steps:
 
     - uses: actions/checkout@v6
@@ -32,13 +30,8 @@ jobs:
     - name: add hashicorp source
       run: wget -O- https://apt.releases.hashicorp.com/gpg | gpg --dearmor | sudo tee /usr/share/keyrings/hashicorp-archive-keyring.gpg && echo "deb [signed-by=/usr/share/keyrings/hashicorp-archive-keyring.gpg] https://apt.releases.hashicorp.com $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/hashicorp.list
 
-    - name: install vagrant and libvirt
-      run: |
-        sudo apt-get update && sudo apt-get install -y vagrant libvirt-daemon-system libvirt-dev
-        sudo chmod 666 /dev/kvm
-        sudo usermod -aG libvirt $(whoami)
-        sudo chmod 666 /var/run/libvirt/libvirt-sock
-        vagrant plugin install vagrant-libvirt
+    - name: install vagrant
+      run: sudo apt-get update && sudo apt-get install -y vagrant virtualbox
 
     - name: freebsd-amd64
       run: make smoke-vagrant/freebsd-amd64
@@ -49,19 +42,10 @@ jobs:
     - name: netbsd-amd64
       run: make smoke-vagrant/netbsd-amd64
 
+    - name: linux-386
+      run: make smoke-vagrant/linux-386
+
     - name: linux-amd64-ipv6disable
       run: make smoke-vagrant/linux-amd64-ipv6disable
 
-    # linux-386 runs last because it requires disabling KVM to use VirtualBox,
-    # which prevents libvirt (used by the other tests) from working after this point.
-    - name: install virtualbox for i386 test
-      run: |
-        sudo apt-get install -y virtualbox
-        sudo rmmod kvm_amd kvm_intel kvm 2>/dev/null || true
-
-    - name: linux-386
-      env:
-        VAGRANT_DEFAULT_PROVIDER: virtualbox
-      run: make smoke-vagrant/linux-386
-
     timeout-minutes: 30

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

@@ -16,10 +16,8 @@ relay:
   am_relay: true
 EOF
 
-    # TEST-NET-3 placeholder IPs; smoke-relay.sh seds them to real container IPs.
-    # Mapping: .2 lighthouse1, .3 host2, .4 host3, .5 host4.
-    export LIGHTHOUSES="192.168.100.1 203.0.113.2:4242"
-    export REMOTE_ALLOW_LIST='{"203.0.113.4/32": false, "203.0.113.5/32": false}'
+    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:
@@ -27,7 +25,7 @@ relay:
     - 192.168.100.1
 EOF
 
-    export REMOTE_ALLOW_LIST='{"203.0.113.3/32": false}'
+    export REMOTE_ALLOW_LIST='{"172.17.0.3/32": false}'
 
     HOST="host3" ../genconfig.sh >host3.yml
 

.github/workflows/smoke/build.sh

@@ -5,15 +5,9 @@ set -e -x
 rm -rf ./build
 mkdir ./build
 
-# Smoke containers run on a dedicated docker network whose subnet is allocated
-# at smoke time, not known at build time. Configs are written with TEST-NET-3
-# placeholder IPs (RFC 5737) and smoke.sh / smoke-vagrant.sh / smoke-relay.sh
-# sed the real container IPs in before starting nebula.
-#
-# Placeholder mapping (last octet == fixed container slot):
-#   203.0.113.2 -> lighthouse1, 203.0.113.3 -> host2,
-#   203.0.113.4 -> host3,       203.0.113.5 -> host4.
-LIGHTHOUSE_IP="203.0.113.2"
+# TODO: Assumes your docker bridge network is a /24, and the first container that launches will be .1
+# - We could make this better by launching the lighthouse first and then fetching what IP it is.
+NET="$(docker network inspect bridge -f '{{ range .IPAM.Config }}{{ .Subnet }}{{ end }}' | cut -d. -f1-3)"
 
 (
     cd build
@@ -31,16 +25,16 @@ LIGHTHOUSE_IP="203.0.113.2"
         ../genconfig.sh >lighthouse1.yml
 
     HOST="host2" \
-        LIGHTHOUSES="192.168.100.1 $LIGHTHOUSE_IP:4242" \
+        LIGHTHOUSES="192.168.100.1 $NET.2:4242" \
         ../genconfig.sh >host2.yml
 
     HOST="host3" \
-        LIGHTHOUSES="192.168.100.1 $LIGHTHOUSE_IP:4242" \
+        LIGHTHOUSES="192.168.100.1 $NET.2:4242" \
         INBOUND='[{"port": "any", "proto": "icmp", "group": "lighthouse"}]' \
         ../genconfig.sh >host3.yml
 
     HOST="host4" \
-        LIGHTHOUSES="192.168.100.1 $LIGHTHOUSE_IP:4242" \
+        LIGHTHOUSES="192.168.100.1 $NET.2:4242" \
         OUTBOUND='[{"port": "any", "proto": "icmp", "group": "lighthouse"}]' \
         ../genconfig.sh >host4.yml
 

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

@@ -6,8 +6,6 @@ set -o pipefail
 
 mkdir -p logs
 
-NETWORK="nebula-smoke-relay"
-
 cleanup() {
     echo
     echo " *** cleanup"
@@ -18,53 +16,22 @@ cleanup() {
     then
         docker kill lighthouse1 host2 host3 host4
     fi
-    docker network rm "$NETWORK" >/dev/null 2>&1
 }
 
 trap cleanup EXIT
 
-# Create a dedicated smoke network with an explicit subnet (required for --ip
-# below). Probe a short list of candidates so a locally-used range doesn't
-# fail the whole test — we only need one to be free.
-docker network rm "$NETWORK" >/dev/null 2>&1 || true
-for candidate in 172.30.0.0/24 172.31.0.0/24 10.98.0.0/24 10.99.0.0/24 192.168.230.0/24; do
-    if docker network create --subnet "$candidate" "$NETWORK" >/dev/null 2>&1; then
-        break
-    fi
-done
-if ! docker network inspect "$NETWORK" >/dev/null 2>&1; then
-    echo "failed to create $NETWORK: every candidate subnet is in use" >&2
-    exit 1
-fi
-
-# Derive container IPs from the network's assigned subnet. Slots: .2 lighthouse1,
-# .3 host2, .4 host3, .5 host4 — matches the placeholders in build-relay.sh.
-SUBNET="$(docker network inspect -f '{{(index .IPAM.Config 0).Subnet}}' "$NETWORK")"
-PREFIX="${SUBNET%/*}"
-PREFIX="${PREFIX%.*}"
-LIGHTHOUSE_IP="$PREFIX.2"
-HOST2_IP="$PREFIX.3"
-HOST3_IP="$PREFIX.4"
-HOST4_IP="$PREFIX.5"
-
-# Sed the placeholder TEST-NET-3 IPs in the host configs to the real ones.
-for f in build/host2.yml build/host3.yml build/host4.yml; do
-    sed "s|203\.0\.113\.|$PREFIX.|g" "$f" >"$f.tmp"
-    mv "$f.tmp" "$f"
-done
-
 docker run --name lighthouse1 --rm nebula:smoke-relay -config lighthouse1.yml -test
-docker run --name host2 --rm -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" nebula:smoke-relay -config host2.yml -test
-docker run --name host3 --rm -v "$PWD/build/host3.yml:/nebula/host3.yml:ro" nebula:smoke-relay -config host3.yml -test
-docker run --name host4 --rm -v "$PWD/build/host4.yml:/nebula/host4.yml:ro" nebula:smoke-relay -config host4.yml -test
+docker run --name host2 --rm nebula:smoke-relay -config host2.yml -test
+docker run --name host3 --rm nebula:smoke-relay -config host3.yml -test
+docker run --name host4 --rm nebula:smoke-relay -config host4.yml -test
 
-docker run --name lighthouse1 --network "$NETWORK" --ip "$LIGHTHOUSE_IP" --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]  /' &
+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
-docker run --name host2 --network "$NETWORK" --ip "$HOST2_IP" -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
+docker run --name host2 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm nebula:smoke-relay -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
 sleep 1
-docker run --name host3 --network "$NETWORK" --ip "$HOST3_IP" -v "$PWD/build/host3.yml:/nebula/host3.yml:ro" --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]  /' &
+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
-docker run --name host4 --network "$NETWORK" --ip "$HOST4_IP" -v "$PWD/build/host4.yml:/nebula/host4.yml:ro" --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]  /' &
+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
@@ -109,13 +76,7 @@ docker exec host4 sh -c 'kill 1'
 docker exec host3 sh -c 'kill 1'
 docker exec host2 sh -c 'kill 1'
 docker exec lighthouse1 sh -c 'kill 1'
-
-# Wait up to 30s for all backgrounded jobs to exit rather than relying on a
-# fixed sleep.
-for _ in $(seq 1 30); do
-    [ -z "$(jobs -r)" ] && break
-    sleep 1
-done
+sleep 5
 
 if [ "$(jobs -r)" ]
 then

.github/workflows/smoke/smoke-vagrant.sh

@@ -8,8 +8,6 @@ export VAGRANT_CWD="$PWD/vagrant-$1"
 
 mkdir -p logs
 
-NETWORK="nebula-smoke"
-
 cleanup() {
     echo
     echo " *** cleanup"
@@ -21,51 +19,21 @@ cleanup() {
         docker kill lighthouse1 host2
     fi
     vagrant destroy -f
-    docker network rm "$NETWORK" >/dev/null 2>&1
 }
 
 trap cleanup EXIT
 
-# Create a dedicated smoke network with an explicit subnet (required for --ip
-# below). Probe a short list of candidates so a locally-used range doesn't
-# fail the whole test — we only need one to be free.
-docker network rm "$NETWORK" >/dev/null 2>&1 || true
-for candidate in 172.30.0.0/24 172.31.0.0/24 10.98.0.0/24 10.99.0.0/24 192.168.230.0/24; do
-    if docker network create --subnet "$candidate" "$NETWORK" >/dev/null 2>&1; then
-        break
-    fi
-done
-if ! docker network inspect "$NETWORK" >/dev/null 2>&1; then
-    echo "failed to create $NETWORK: every candidate subnet is in use" >&2
-    exit 1
-fi
-
-# Derive container IPs from the network's assigned subnet. Slots: .2 lighthouse1,
-# .3 host2 — matches the placeholders in build.sh.
-SUBNET="$(docker network inspect -f '{{(index .IPAM.Config 0).Subnet}}' "$NETWORK")"
-PREFIX="${SUBNET%/*}"
-PREFIX="${PREFIX%.*}"
-LIGHTHOUSE_IP="$PREFIX.2"
-HOST2_IP="$PREFIX.3"
-
-# Sed the placeholder TEST-NET-3 IPs in the host configs to the real ones.
-# This must happen before `vagrant up` rsyncs build/ into the VM for host3.
-for f in build/host2.yml build/host3.yml; do
-    sed "s|203\.0\.113\.|$PREFIX.|g" "$f" >"$f.tmp"
-    mv "$f.tmp" "$f"
-done
-
 CONTAINER="nebula:${NAME:-smoke}"
 
 docker run --name lighthouse1 --rm "$CONTAINER" -config lighthouse1.yml -test
-docker run --name host2 --rm -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" "$CONTAINER" -config host2.yml -test
+docker run --name host2 --rm "$CONTAINER" -config host2.yml -test
 
 vagrant up
 vagrant ssh -c "cd /nebula && /nebula/$1-nebula -config host3.yml -test" -- -T
 
-docker run --name lighthouse1 --network "$NETWORK" --ip "$LIGHTHOUSE_IP" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
+docker run --name lighthouse1 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
 sleep 1
-docker run --name host2 --network "$NETWORK" --ip "$HOST2_IP" -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
+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
 vagrant ssh -c "cd /nebula && sudo sh -c 'echo \$\$ >/nebula/pid && exec /nebula/$1-nebula -config host3.yml'" 2>&1 -- -T | tee logs/host3 | sed -u 's/^/  [host3]  /' &
 sleep 15
@@ -128,14 +96,7 @@ vagrant ssh -c "ping -c1 192.168.100.2" -- -T
 vagrant ssh -c "sudo xargs kill </nebula/pid" -- -T
 docker exec host2 sh -c 'kill 1'
 docker exec lighthouse1 sh -c 'kill 1'
-
-# Wait up to 30s for all backgrounded jobs to exit. vagrant ssh in particular
-# takes a beat to tear down after nebula exits on the VM, so a fixed sleep is
-# racy.
-for _ in $(seq 1 30); do
-    [ -z "$(jobs -r)" ] && break
-    sleep 1
-done
+sleep 1
 
 if [ "$(jobs -r)" ]
 then

.github/workflows/smoke/smoke.sh

@@ -6,8 +6,6 @@ set -o pipefail
 
 mkdir -p logs
 
-NETWORK="nebula-smoke"
-
 cleanup() {
     echo
     echo " *** cleanup"
@@ -18,71 +16,38 @@ cleanup() {
     then
         docker kill lighthouse1 host2 host3 host4
     fi
-    docker network rm "$NETWORK" >/dev/null 2>&1
 }
 
 trap cleanup EXIT
 
-# Create a dedicated smoke network with an explicit subnet (required for --ip
-# below). Probe a short list of candidates so a locally-used range doesn't
-# fail the whole test — we only need one to be free.
-docker network rm "$NETWORK" >/dev/null 2>&1 || true
-for candidate in 172.30.0.0/24 172.31.0.0/24 10.98.0.0/24 10.99.0.0/24 192.168.230.0/24; do
-    if docker network create --subnet "$candidate" "$NETWORK" >/dev/null 2>&1; then
-        break
-    fi
-done
-if ! docker network inspect "$NETWORK" >/dev/null 2>&1; then
-    echo "failed to create $NETWORK: every candidate subnet is in use" >&2
-    exit 1
-fi
-
-# Derive container IPs from the network's assigned subnet. Slots: .2 lighthouse1,
-# .3 host2, .4 host3, .5 host4 — matches the placeholders in build.sh.
-SUBNET="$(docker network inspect -f '{{(index .IPAM.Config 0).Subnet}}' "$NETWORK")"
-PREFIX="${SUBNET%/*}"
-PREFIX="${PREFIX%.*}"
-LIGHTHOUSE_IP="$PREFIX.2"
-HOST2_IP="$PREFIX.3"
-HOST3_IP="$PREFIX.4"
-HOST4_IP="$PREFIX.5"
-
-# Sed the placeholder TEST-NET-3 IPs in the host configs to the real ones.
-# build/lighthouse1.yml has no IPs to rewrite so it's skipped.
-for f in build/host2.yml build/host3.yml build/host4.yml; do
-    sed "s|203\.0\.113\.|$PREFIX.|g" "$f" >"$f.tmp"
-    mv "$f.tmp" "$f"
-done
-
 CONTAINER="nebula:${NAME:-smoke}"
 
 docker run --name lighthouse1 --rm "$CONTAINER" -config lighthouse1.yml -test
-docker run --name host2 --rm -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" "$CONTAINER" -config host2.yml -test
-docker run --name host3 --rm -v "$PWD/build/host3.yml:/nebula/host3.yml:ro" "$CONTAINER" -config host3.yml -test
-docker run --name host4 --rm -v "$PWD/build/host4.yml:/nebula/host4.yml:ro" "$CONTAINER" -config host4.yml -test
+docker run --name host2 --rm "$CONTAINER" -config host2.yml -test
+docker run --name host3 --rm "$CONTAINER" -config host3.yml -test
+docker run --name host4 --rm "$CONTAINER" -config host4.yml -test
 
-docker run --name lighthouse1 --network "$NETWORK" --ip "$LIGHTHOUSE_IP" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
+docker run --name lighthouse1 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config lighthouse1.yml 2>&1 | tee logs/lighthouse1 | sed -u 's/^/  [lighthouse1]  /' &
 sleep 1
-docker run --name host2 --network "$NETWORK" --ip "$HOST2_IP" -v "$PWD/build/host2.yml:/nebula/host2.yml:ro" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
+docker run --name host2 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host2.yml 2>&1 | tee logs/host2 | sed -u 's/^/  [host2]  /' &
 sleep 1
-docker run --name host3 --network "$NETWORK" --ip "$HOST3_IP" -v "$PWD/build/host3.yml:/nebula/host3.yml:ro" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host3.yml 2>&1 | tee logs/host3 | sed -u 's/^/  [host3]  /' &
+docker run --name host3 --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host3.yml 2>&1 | tee logs/host3 | sed -u 's/^/  [host3]  /' &
 sleep 1
-docker run --name host4 --network "$NETWORK" --ip "$HOST4_IP" -v "$PWD/build/host4.yml:/nebula/host4.yml:ro" --device /dev/net/tun:/dev/net/tun --cap-add NET_ADMIN --rm "$CONTAINER" -config host4.yml 2>&1 | tee logs/host4 | sed -u 's/^/  [host4]  /' &
+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
-docker exec lighthouse1 tcpdump -i tun0 -q -w - -U 2>logs/lighthouse1.inside.log >logs/lighthouse1.inside.pcap &
+docker exec lighthouse1 tcpdump -i nebula1 -q -w - -U 2>logs/lighthouse1.inside.log >logs/lighthouse1.inside.pcap &
 docker exec lighthouse1 tcpdump -i eth0 -q -w - -U 2>logs/lighthouse1.outside.log >logs/lighthouse1.outside.pcap &
-docker exec host2 tcpdump -i tun0 -q -w - -U 2>logs/host2.inside.log >logs/host2.inside.pcap &
+docker exec host2 tcpdump -i nebula1 -q -w - -U 2>logs/host2.inside.log >logs/host2.inside.pcap &
 docker exec host2 tcpdump -i eth0 -q -w - -U 2>logs/host2.outside.log >logs/host2.outside.pcap &
-docker exec host3 tcpdump -i tun0 -q -w - -U 2>logs/host3.inside.log >logs/host3.inside.pcap &
+docker exec host3 tcpdump -i nebula1 -q -w - -U 2>logs/host3.inside.log >logs/host3.inside.pcap &
 docker exec host3 tcpdump -i eth0 -q -w - -U 2>logs/host3.outside.log >logs/host3.outside.pcap &
-docker exec host4 tcpdump -i tun0 -q -w - -U 2>logs/host4.inside.log >logs/host4.inside.pcap &
+docker exec host4 tcpdump -i nebula1 -q -w - -U 2>logs/host4.inside.log >logs/host4.inside.pcap &
 docker exec host4 tcpdump -i eth0 -q -w - -U 2>logs/host4.outside.log >logs/host4.outside.pcap &
 
 docker exec host2 ncat -nklv 0.0.0.0 2000 &
 docker exec host3 ncat -nklv 0.0.0.0 2000 &
-docker exec host4 ncat -e '/usr/bin/echo helloagainfromhost4' -nkluv 0.0.0.0 4000 &
 docker exec host2 ncat -e '/usr/bin/echo host2' -nkluv 0.0.0.0 3000 &
 docker exec host3 ncat -e '/usr/bin/echo host3' -nkluv 0.0.0.0 3000 &
 
@@ -154,24 +119,17 @@ echo
 echo " *** Testing conntrack"
 echo
 set -x
-
-# host4's outbound firewall only allows ICMP to the lighthouse, so host4
-# cannot initiate UDP to host2. Once host2 initiates a flow to host4:4000,
-# conntrack must let host4's listener reply on that flow. If it doesn't,
-# the echo back from host4 never reaches host2.
-docker exec host2 sh -c "(/usr/bin/echo host2; sleep 2) | ncat -nuv 192.168.100.4 4000" | grep -q helloagainfromhost4
+# host2 can ping host3 now that host3 pinged it first
+docker exec host2 ping -c1 192.168.100.3
+# host4 can ping host2 once conntrack established
+docker exec host2 ping -c1 192.168.100.4
+docker exec host4 ping -c1 192.168.100.2
 
 docker exec host4 sh -c 'kill 1'
 docker exec host3 sh -c 'kill 1'
 docker exec host2 sh -c 'kill 1'
 docker exec lighthouse1 sh -c 'kill 1'
-
-# Wait up to 30s for all backgrounded jobs to exit rather than relying on a
-# fixed sleep.
-for _ in $(seq 1 30); do
-    [ -z "$(jobs -r)" ] && break
-    sleep 1
-done
+sleep 5
 
 if [ "$(jobs -r)" ]
 then

.github/workflows/smoke/vagrant-linux-amd64-ipv6disable/Vagrantfile

@@ -1,7 +1,7 @@
 # -*- mode: ruby -*-
 # vi: set ft=ruby :
 Vagrant.configure("2") do |config|
-  config.vm.box = "bento/ubuntu-24.04"
+  config.vm.box = "ubuntu/jammy64"
 
   config.vm.synced_folder "../build", "/nebula"
 

.github/workflows/smoke/vagrant-openbsd-amd64/Vagrantfile

@@ -1,7 +1,7 @@
 # -*- mode: ruby -*-
 # vi: set ft=ruby :
 Vagrant.configure("2") do |config|
-  config.vm.box = "DefinedNet/openbsd78"
+  config.vm.box = "generic/openbsd7"
 
   config.vm.synced_folder "../build", "/nebula", type: "rsync"
 end

.github/workflows/test.yml

@@ -45,7 +45,7 @@ jobs:
     - name: Build test mobile
       run: make build-test-mobile
 
-    - uses: actions/upload-artifact@v6
+    - uses: actions/upload-artifact@v5
       with:
         name: e2e packet flow linux-latest
         path: e2e/mermaid/linux-latest
@@ -125,7 +125,7 @@ jobs:
     - name: End 2 end
       run: make e2evv
 
-    - uses: actions/upload-artifact@v6
+    - uses: actions/upload-artifact@v5
       with:
         name: e2e packet flow ${{ matrix.os }}
         path: e2e/mermaid/${{ matrix.os }}

.golangci.yaml

@@ -2,21 +2,7 @@ version: "2"
 linters:
   default: none
   enable:
-    - sloglint
     - testifylint
-  settings:
-    sloglint:
-      # Enforce key-value pair form for Info/Debug/Warn/Error/Log/With and
-      # the package-level slog equivalents. Use l.Log(ctx, level, ...) for
-      # custom levels instead of LogAttrs when you can.
-      #
-      # LogAttrs is also flagged by this rule because it takes ...slog.Attr;
-      # the few legitimate sites (where attrs is built up as a []slog.Attr)
-      # carry a //nolint:sloglint with rationale.
-      kv-only: true
-      # no-mixed-args is on by default: forbids mixing kv and attrs in one call.
-      # discard-handler is on by default (since Go 1.24): suggests
-      # slog.DiscardHandler over slog.NewTextHandler(io.Discard, nil).
   exclusions:
     generated: lax
     presets:

CHANGELOG.md

@@ -7,50 +7,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
-## [1.10.3] - 2026-02-06
-
-### Security
-
-- Fix an issue where blocklist bypass is possible when using curve P256 since the signature can have 2 valid representations.
-  Both fingerprint representations will be tested against the blocklist.
-  Any newly issued P256 based certificates will have their signature clamped to the low-s form.
-  Nebula will assert the low-s signature form when validating certificates in a future version. [GHSA-69x3-g4r3-p962](https://github.com/slackhq/nebula/security/advisories/GHSA-69x3-g4r3-p962)
-
-### Changed
-
-- Improve error reporting if nebula fails to start due to a tun device naming issue. (#1588)
-
-## [1.10.2] - 2026-01-21
-
-### Fixed
-
-- Fix panic when using `use_system_route_table` that was introduced in v1.10.1. (#1580)
-
-### Changed
-
-- Fix some typos in comments. (#1582)
-- Dependency updates. (#1581)
-
-## [1.10.1] - 2026-01-16
-
-See the [v1.10.1](https://github.com/slackhq/nebula/milestone/26?closed=1) milestone for a complete list of changes.
-
-### Fixed
-
-- Fix a bug where an unsafe route derived from the system route table could be lost on a config reload. (#1573)
-- Fix the PEM banner for ECDSA P256 public keys. (#1552)
-- Fix a regression on Windows from 1.9.x where nebula could fall back to a less performant UDP listener if 
-  non-critical ioctls failed. (#1568)
-- Fix a bug in handshake processing when a peer sends an unexpected public key. (#1566)
-
-### Added
-
-- Add a config option to control accepting `recv_error` packets which defaults to `always`. (#1569)
-
-### Changed
-
-- Various dependency updates. (#1541, #1549, #1550, #1557, #1558, #1560, #1561, #1570, #1571)
-
 ## [1.10.0] - 2025-12-04
 
 See the [v1.10.0](https://github.com/slackhq/nebula/milestone/16?closed=1) milestone for a complete list of changes.
@@ -788,10 +744,7 @@ created.)
 
 - Initial public release.
 
-[Unreleased]: https://github.com/slackhq/nebula/compare/v1.10.3...HEAD
-[1.10.3]: https://github.com/slackhq/nebula/releases/tag/v1.10.3
-[1.10.2]: https://github.com/slackhq/nebula/releases/tag/v1.10.2
-[1.10.1]: https://github.com/slackhq/nebula/releases/tag/v1.10.1
+[Unreleased]: https://github.com/slackhq/nebula/compare/v1.10.0...HEAD
 [1.10.0]: https://github.com/slackhq/nebula/releases/tag/v1.10.0
 [1.9.7]: https://github.com/slackhq/nebula/releases/tag/v1.9.7
 [1.9.6]: https://github.com/slackhq/nebula/releases/tag/v1.9.6

CODEOWNERS

@@ -1 +0,0 @@
-#ECCN:Open Source

README.md

@@ -57,7 +57,7 @@ Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for
     docker pull nebulaoss/nebula
     ```
 
-#### Mobile ([source code](https://github.com/DefinedNet/mobile_nebula))
+#### Mobile
 
 - [iOS](https://apps.apple.com/us/app/mobile-nebula/id1509587936?itsct=apps_box&itscg=30200)
 - [Android](https://play.google.com/store/apps/details?id=net.defined.mobile_nebula&pcampaignid=pcampaignidMKT-Other-global-all-co-prtnr-py-PartBadge-Mar2515-1)
@@ -76,8 +76,6 @@ Nebula was created to provide a mechanism for groups of hosts to communicate sec
 
 ## Getting started (quickly)
 
-**Don't want to manage your own PKI and lighthouses?** [Managed Nebula](https://www.defined.net/) from Defined Networking handles all of this for you.
-
 To set up a Nebula network, you'll need:
 
 #### 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.

bits.go

@@ -1,10 +1,8 @@
 package nebula
 
 import (
-	"context"
-	"log/slog"
-
 	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
 )
 
 type Bits struct {
@@ -32,7 +30,7 @@ func NewBits(bits uint64) *Bits {
 	return b
 }
 
-func (b *Bits) Check(l *slog.Logger, i uint64) bool {
+func (b *Bits) Check(l *logrus.Logger, i uint64) bool {
 	// If i is the next number, return true.
 	if i > b.current {
 		return true
@@ -44,16 +42,13 @@ func (b *Bits) Check(l *slog.Logger, i uint64) bool {
 	}
 
 	// Not within the window
-	if l.Enabled(context.Background(), slog.LevelDebug) {
-		l.Debug("rejected a packet (top)",
-			"current", b.current,
-			"incoming", i,
-		)
+	if l.Level >= logrus.DebugLevel {
+		l.Debugf("rejected a packet (top) %d %d\n", b.current, i)
 	}
 	return false
 }
 
-func (b *Bits) Update(l *slog.Logger, i uint64) bool {
+func (b *Bits) Update(l *logrus.Logger, i uint64) bool {
 	// If i is the next number, return true and update current.
 	if i == b.current+1 {
 		// Check if the oldest bit was lost since we are shifting the window by 1 and occupying it with this counter
@@ -92,13 +87,9 @@ func (b *Bits) Update(l *slog.Logger, i uint64) bool {
 	// Check to see if it's a duplicate
 	if i > b.current-b.length || i < b.length && b.current < b.length {
 		if b.current == i || b.bits[i%b.length] == true {
-			if l.Enabled(context.Background(), slog.LevelDebug) {
-				l.Debug("Receive window",
-					"accepted", false,
-					"currentCounter", b.current,
-					"incomingCounter", i,
-					"reason", "duplicate",
-				)
+			if l.Level >= logrus.DebugLevel {
+				l.WithField("receiveWindow", m{"accepted": false, "currentCounter": b.current, "incomingCounter": i, "reason": "duplicate"}).
+					Debug("Receive window")
 			}
 			b.dupeCounter.Inc(1)
 			return false
@@ -110,13 +101,12 @@ func (b *Bits) Update(l *slog.Logger, i uint64) bool {
 
 	// In all other cases, fail and don't change current.
 	b.outOfWindowCounter.Inc(1)
-	if l.Enabled(context.Background(), slog.LevelDebug) {
-		l.Debug("Receive window",
-			"accepted", false,
-			"currentCounter", b.current,
-			"incomingCounter", i,
-			"reason", "nonsense",
-		)
+	if l.Level >= logrus.DebugLevel {
+		l.WithField("accepted", false).
+			WithField("currentCounter", b.current).
+			WithField("incomingCounter", i).
+			WithField("reason", "nonsense").
+			Debug("Receive window")
 	}
 	return false
 }

boring.go

@@ -1,4 +1,5 @@
 //go:build boringcrypto
+// +build boringcrypto
 
 package nebula
 

cert/ca_pool.go

@@ -1,14 +1,11 @@
 package cert
 
 import (
-	"bufio"
-	"bytes"
-	"encoding/pem"
 	"errors"
 	"fmt"
-	"io"
 	"net/netip"
 	"slices"
+	"strings"
 	"time"
 )
 
@@ -32,46 +29,22 @@ func NewCAPool() *CAPool {
 // If the pool contains any expired certificates, an ErrExpired will be
 // returned along with the pool. The caller must handle any such errors.
 func NewCAPoolFromPEM(caPEMs []byte) (*CAPool, error) {
-	return NewCAPoolFromPEMReader(bytes.NewReader(caPEMs))
-}
-
-// NewCAPoolFromPEMReader will create a new CA pool from the provided reader.
-// The reader must contain a PEM-encoded set of nebula certificates.
-func NewCAPoolFromPEMReader(r io.Reader) (*CAPool, error) {
 	pool := NewCAPool()
-
+	var err error
 	var expired bool
-
-	scanner := bufio.NewScanner(r)
-	scanner.Split(SplitPEM)
-
-	for scanner.Scan() {
-		pemBytes := scanner.Bytes()
-
-		block, rest := pem.Decode(pemBytes)
-		if len(bytes.TrimSpace(rest)) > 0 {
-			return nil, ErrInvalidPEMBlock
+	for {
+		caPEMs, err = pool.AddCAFromPEM(caPEMs)
+		if errors.Is(err, ErrExpired) {
+			expired = true
+			err = nil
 		}
-		if block == nil {
-			return nil, ErrInvalidPEMBlock
-		}
-
-		c, err := unmarshalCertificateBlock(block)
 		if err != nil {
 			return nil, err
 		}
-
-		err = pool.AddCA(c)
-		if errors.Is(err, ErrExpired) {
-			expired = true
-			continue
-		} else if err != nil {
-			return nil, err
+		if len(caPEMs) == 0 || strings.TrimSpace(string(caPEMs)) == "" {
+			break
 		}
 	}
-	if err := scanner.Err(); err != nil {
-		return nil, ErrInvalidPEMBlock
-	}
 
 	if expired {
 		return pool, ErrExpired
@@ -168,23 +141,10 @@ func (ncp *CAPool) VerifyCertificate(now time.Time, c Certificate) (*CachedCerti
 		return nil, err
 	}
 
-	// Pre nebula v1.10.3 could generate signatures in either high or low s form and validation
-	// of signatures allowed for either. Nebula v1.10.3 and beyond clamps signature generation to low-s form
-	// but validation still allows for either. Since a change in the signature bytes affects the fingerprint, we
-	// need to test both forms until such a time comes that we enforce low-s form on signature validation.
-	fp2, err := CalculateAlternateFingerprint(c)
-	if err != nil {
-		return nil, fmt.Errorf("could not calculate alternate fingerprint to verify: %w", err)
-	}
-	if fp2 != "" && ncp.IsBlocklisted(fp2) {
-		return nil, ErrBlockListed
-	}
-
 	cc := CachedCertificate{
 		Certificate:       c,
 		InvertedGroups:    make(map[string]struct{}),
 		Fingerprint:       fp,
-		fingerprint2:      fp2,
 		signerFingerprint: signer.Fingerprint,
 	}
 
@@ -198,11 +158,6 @@ func (ncp *CAPool) VerifyCertificate(now time.Time, c Certificate) (*CachedCerti
 // VerifyCachedCertificate is the same as VerifyCertificate other than it operates on a pre-verified structure and
 // is a cheaper operation to perform as a result.
 func (ncp *CAPool) VerifyCachedCertificate(now time.Time, c *CachedCertificate) error {
-	// Check any available alternate fingerprint forms for this certificate, re P256 high-s/low-s
-	if c.fingerprint2 != "" && ncp.IsBlocklisted(c.fingerprint2) {
-		return ErrBlockListed
-	}
-
 	_, err := ncp.verify(c.Certificate, now, c.Fingerprint, c.signerFingerprint)
 	return err
 }

cert/ca_pool_test.go

@@ -1,14 +1,10 @@
 package cert
 
 import (
-	"bytes"
-	"io"
 	"net/netip"
-	"strings"
 	"testing"
 	"time"
 
-	"github.com/slackhq/nebula/cert/p256"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -115,60 +111,6 @@ k+coOv04r+zh33ISyhbsafnYduN17p2eD7CmHvHuerguXD9f32gcxo/KsFCKEjMe
 	assert.Len(t, ppppp.CAs, 1)
 }
 
-// oneByteReader wraps a reader to return at most 1 byte per Read call,
-// exercising the streaming accumulation logic in NewCAPoolFromPEMReader.
-type oneByteReader struct {
-	r io.Reader
-}
-
-func (o *oneByteReader) Read(p []byte) (int, error) {
-	if len(p) == 0 {
-		return 0, nil
-	}
-	return o.r.Read(p[:1])
-}
-
-func TestNewCAPoolFromPEMReader_EmptyReader(t *testing.T) {
-	pool, err := NewCAPoolFromPEMReader(bytes.NewReader(nil))
-	require.NoError(t, err)
-	assert.Empty(t, pool.CAs)
-
-	pool, err = NewCAPoolFromPEMReader(strings.NewReader("   \n\t\n  "))
-	require.NoError(t, err)
-	assert.Empty(t, pool.CAs)
-}
-
-func TestNewCAPoolFromPEMReader_OneByteReads(t *testing.T) {
-	ca1, _, _, pem1 := NewTestCaCert(Version2, Curve_CURVE25519, time.Now(), time.Now().Add(time.Hour), nil, nil, nil)
-	ca2, _, _, pem2 := NewTestCaCert(Version2, Curve_CURVE25519, time.Now(), time.Now().Add(time.Hour), nil, nil, nil)
-
-	bundle := append(pem1, pem2...)
-	pool, err := NewCAPoolFromPEMReader(&oneByteReader{r: bytes.NewReader(bundle)})
-	require.NoError(t, err)
-	assert.Len(t, pool.CAs, 2)
-
-	fp1, err := ca1.Fingerprint()
-	require.NoError(t, err)
-	fp2, err := ca2.Fingerprint()
-	require.NoError(t, err)
-
-	assert.Contains(t, pool.CAs, fp1)
-	assert.Contains(t, pool.CAs, fp2)
-}
-
-func TestNewCAPoolFromPEMReader_TruncatedPEM(t *testing.T) {
-	_, err := NewCAPoolFromPEMReader(strings.NewReader("-----BEGIN NEBULA CERTIFICATE-----\npartialdata"))
-	assert.ErrorIs(t, err, ErrInvalidPEMBlock)
-}
-
-func TestNewCAPoolFromPEMReader_TrailingGarbage(t *testing.T) {
-	_, _, _, pem1 := NewTestCaCert(Version2, Curve_CURVE25519, time.Now(), time.Now().Add(time.Hour), nil, nil, nil)
-
-	bundle := append(pem1, []byte("some trailing garbage")...)
-	_, err := NewCAPoolFromPEMReader(bytes.NewReader(bundle))
-	assert.ErrorIs(t, err, ErrInvalidPEMBlock)
-}
-
 func TestCertificateV1_Verify(t *testing.T) {
 	ca, _, caKey, _ := NewTestCaCert(Version1, Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, nil)
 	c, _, _, _ := NewTestCert(Version1, Curve_CURVE25519, ca, caKey, "test cert", time.Now(), time.Now().Add(5*time.Minute), nil, nil, nil)
@@ -228,15 +170,6 @@ func TestCertificateV1_VerifyP256(t *testing.T) {
 	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.EqualError(t, err, "certificate is in the block list")
 
-	// Create a copy of the cert and swap to the alternate form for the signature
-	nc := c.Copy()
-	b, err := p256.Swap(c.Signature())
-	require.NoError(t, err)
-	require.NoError(t, nc.(*certificateV1).setSignature(b))
-
-	_, err = caPool.VerifyCertificate(time.Now(), nc)
-	require.EqualError(t, err, "certificate is in the block list")
-
 	caPool.ResetCertBlocklist()
 	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.NoError(t, err)
@@ -254,7 +187,7 @@ func TestCertificateV1_VerifyP256(t *testing.T) {
 	require.NoError(t, err)
 
 	caPool = NewCAPool()
-	b, err = caPool.AddCAFromPEM(caPem)
+	b, err := caPool.AddCAFromPEM(caPem)
 	require.NoError(t, err)
 	assert.Empty(t, b)
 
@@ -263,17 +196,7 @@ func TestCertificateV1_VerifyP256(t *testing.T) {
 	})
 
 	c, _, _, _ = NewTestCert(Version1, Curve_P256, ca, caKey, "test", time.Now(), time.Now().Add(5*time.Minute), nil, nil, []string{"test1"})
-	cc, err := caPool.VerifyCertificate(time.Now(), c)
-	require.NoError(t, err)
-
-	// Reset the blocklist and block the alternate form fingerprint
-	caPool.ResetCertBlocklist()
-	caPool.BlocklistFingerprint(cc.fingerprint2)
-	err = caPool.VerifyCachedCertificate(time.Now(), cc)
-	require.EqualError(t, err, "certificate is in the block list")
-
-	caPool.ResetCertBlocklist()
-	err = caPool.VerifyCachedCertificate(time.Now(), cc)
+	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.NoError(t, err)
 }
 
@@ -471,15 +394,6 @@ func TestCertificateV2_VerifyP256(t *testing.T) {
 	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.EqualError(t, err, "certificate is in the block list")
 
-	// Create a copy of the cert and swap to the alternate form for the signature
-	nc := c.Copy()
-	b, err := p256.Swap(c.Signature())
-	require.NoError(t, err)
-	require.NoError(t, nc.(*certificateV2).setSignature(b))
-
-	_, err = caPool.VerifyCertificate(time.Now(), nc)
-	require.EqualError(t, err, "certificate is in the block list")
-
 	caPool.ResetCertBlocklist()
 	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.NoError(t, err)
@@ -497,7 +411,7 @@ func TestCertificateV2_VerifyP256(t *testing.T) {
 	require.NoError(t, err)
 
 	caPool = NewCAPool()
-	b, err = caPool.AddCAFromPEM(caPem)
+	b, err := caPool.AddCAFromPEM(caPem)
 	require.NoError(t, err)
 	assert.Empty(t, b)
 
@@ -506,17 +420,7 @@ func TestCertificateV2_VerifyP256(t *testing.T) {
 	})
 
 	c, _, _, _ = NewTestCert(Version2, Curve_P256, ca, caKey, "test", time.Now(), time.Now().Add(5*time.Minute), nil, nil, []string{"test1"})
-	cc, err := caPool.VerifyCertificate(time.Now(), c)
-	require.NoError(t, err)
-
-	// Reset the blocklist and block the alternate form fingerprint
-	caPool.ResetCertBlocklist()
-	caPool.BlocklistFingerprint(cc.fingerprint2)
-	err = caPool.VerifyCachedCertificate(time.Now(), cc)
-	require.EqualError(t, err, "certificate is in the block list")
-
-	caPool.ResetCertBlocklist()
-	err = caPool.VerifyCachedCertificate(time.Now(), cc)
+	_, err = caPool.VerifyCertificate(time.Now(), c)
 	require.NoError(t, err)
 }
 

cert/cert.go

@@ -4,8 +4,6 @@ import (
 	"fmt"
 	"net/netip"
 	"time"
-
-	"github.com/slackhq/nebula/cert/p256"
 )
 
 type Version uint8
@@ -112,9 +110,6 @@ type CachedCertificate struct {
 	InvertedGroups    map[string]struct{}
 	Fingerprint       string
 	signerFingerprint string
-
-	// A place to store a 2nd fingerprint if the certificate could have one, such as with P256
-	fingerprint2 string
 }
 
 func (cc *CachedCertificate) String() string {
@@ -124,7 +119,6 @@ func (cc *CachedCertificate) String() string {
 // Recombine will attempt to unmarshal a certificate received in a handshake.
 // Handshakes save space by placing the peers public key in a different part of the packet, we have to
 // reassemble the actual certificate structure with that in mind.
-// Implementations MUST assert the public key is not in the raw certificate bytes if the passed in public key is not empty.
 func Recombine(v Version, rawCertBytes, publicKey []byte, curve Curve) (Certificate, error) {
 	if publicKey == nil {
 		return nil, ErrNoPeerStaticKey
@@ -157,31 +151,3 @@ func Recombine(v Version, rawCertBytes, publicKey []byte, curve Curve) (Certific
 
 	return c, nil
 }
-
-// CalculateAlternateFingerprint calculates a 2nd fingerprint representation for P256 certificates
-// CAPool blocklist testing through `VerifyCertificate` and `VerifyCachedCertificate` automatically performs this step.
-func CalculateAlternateFingerprint(c Certificate) (string, error) {
-	if c.Curve() != Curve_P256 {
-		return "", nil
-	}
-
-	nc := c.Copy()
-	b, err := p256.Swap(nc.Signature())
-	if err != nil {
-		return "", err
-	}
-
-	switch v := nc.(type) {
-	case *certificateV1:
-		err = v.setSignature(b)
-	case *certificateV2:
-		err = v.setSignature(b)
-	default:
-		return "", ErrUnknownVersion
-	}
-
-	if err != nil {
-		return "", err
-	}
-	return nc.Fingerprint()
-}

cert/cert_v1.go

@@ -426,7 +426,7 @@ func unmarshalCertificateV1(b []byte, publicKey []byte) (*certificateV1, error)
 			unsafeNetworks: make([]netip.Prefix, len(rc.Details.Subnets)/2),
 			notBefore:      time.Unix(rc.Details.NotBefore, 0),
 			notAfter:       time.Unix(rc.Details.NotAfter, 0),
-			publicKey:      nil,
+			publicKey:      make([]byte, len(rc.Details.PublicKey)),
 			isCA:           rc.Details.IsCA,
 			curve:          rc.Details.Curve,
 		},
@@ -437,19 +437,12 @@ func unmarshalCertificateV1(b []byte, publicKey []byte) (*certificateV1, error)
 	copy(nc.details.groups, rc.Details.Groups)
 	nc.details.issuer = hex.EncodeToString(rc.Details.Issuer)
 
-	// If a public key is passed in as an argument, the certificate pubkey must be empty
-	// and the passed-in pubkey copied into the cert.
 	if len(publicKey) > 0 {
-		if len(rc.Details.PublicKey) != 0 {
-			return nil, ErrCertPubkeyPresent
-		}
-		nc.details.publicKey = make([]byte, len(publicKey))
-		copy(nc.details.publicKey, publicKey)
-	} else {
-		nc.details.publicKey = make([]byte, len(rc.Details.PublicKey))
-		copy(nc.details.publicKey, rc.Details.PublicKey)
+		nc.details.publicKey = publicKey
 	}
 
+	copy(nc.details.publicKey, rc.Details.PublicKey)
+
 	var ip netip.Addr
 	for i, rawIp := range rc.Details.Ips {
 		if i%2 == 0 {

cert/cert_v1_test.go

@@ -62,62 +62,6 @@ func TestCertificateV1_Marshal(t *testing.T) {
 	assert.Equal(t, nc.Groups(), nc2.Groups())
 }
 
-func TestCertificateV1_Unmarshal(t *testing.T) {
-	t.Parallel()
-	before := time.Now().Add(time.Second * -60).Round(time.Second)
-	after := time.Now().Add(time.Second * 60).Round(time.Second)
-	pubKey := []byte("1234567890abcedfghij1234567890ab")
-	invalidPubkey := []byte("00000000000000000000000000000000")
-
-	nc := certificateV1{
-		details: detailsV1{
-			name: "testing",
-			networks: []netip.Prefix{
-				mustParsePrefixUnmapped("10.1.1.1/24"),
-				mustParsePrefixUnmapped("10.1.1.2/16"),
-			},
-			unsafeNetworks: []netip.Prefix{
-				mustParsePrefixUnmapped("9.1.1.2/24"),
-				mustParsePrefixUnmapped("9.1.1.3/16"),
-			},
-			groups:    []string{"test-group1", "test-group2", "test-group3"},
-			notBefore: before,
-			notAfter:  after,
-			publicKey: pubKey,
-			isCA:      false,
-			issuer:    "1234567890abcedfghij1234567890ab",
-		},
-		signature: []byte("1234567890abcedfghij1234567890ab"),
-	}
-
-	// This certificate has a pubkey included
-	certWithPubkey, err := nc.Marshal()
-	require.NoError(t, err)
-
-	// This certificate is missing the pubkey section
-	certWithoutPubkey, err := nc.MarshalForHandshakes()
-	require.NoError(t, err)
-
-	// Cert has no pubkey and no pubkey passed in must fail to validate
-	isNil, err := unmarshalCertificateV1(certWithoutPubkey, nil)
-	require.Error(t, err)
-
-	// Cert has different pubkey than one passed in must fail
-	isNil, err = unmarshalCertificateV1(certWithPubkey, invalidPubkey)
-	require.Nil(t, isNil)
-	require.Error(t, err)
-
-	// Cert has pubkey and no pubkey argument works ok
-	_, err = unmarshalCertificateV1(certWithPubkey, nil)
-	require.NoError(t, err)
-
-	// Cert has no pubkey and valid, correctly signed pubkey passed in
-	nc2, err := unmarshalCertificateV1(certWithoutPubkey, pubKey)
-	require.NoError(t, err)
-
-	assert.Equal(t, pubKey, nc2.PublicKey())
-}
-
 func TestCertificateV1_PublicKeyPem(t *testing.T) {
 	t.Parallel()
 	before := time.Now().Add(time.Second * -60).Round(time.Second)

cert/cert_v2.go

@@ -592,13 +592,7 @@ func unmarshalCertificateV2(b []byte, publicKey []byte, curve Curve) (*certifica
 	// Maybe grab the public key
 	var rawPublicKey cryptobyte.String
 	if len(publicKey) > 0 {
-		// If a public key is passed in, then the handshake certificate must
-		// not have a public key present
-		if input.PeekASN1Tag(TagCertPublicKey) {
-			return nil, ErrCertPubkeyPresent
-		}
-		rawPublicKey = make(cryptobyte.String, len(publicKey))
-		copy(rawPublicKey, publicKey)
+		rawPublicKey = publicKey
 	} else if !input.ReadOptionalASN1(&rawPublicKey, nil, TagCertPublicKey) {
 		return nil, ErrBadFormat
 	}

cert/cert_v2_test.go

@@ -76,58 +76,6 @@ func TestCertificateV2_Marshal(t *testing.T) {
 	assert.Equal(t, nc.Groups(), nc2.Groups())
 }
 
-func TestCertificateV2_Unmarshal(t *testing.T) {
-	t.Parallel()
-	before := time.Now().Add(time.Second * -60).Round(time.Second)
-	after := time.Now().Add(time.Second * 60).Round(time.Second)
-	pubKey := []byte("1234567890abcedfghij1234567890ab")
-
-	nc := certificateV2{
-		details: detailsV2{
-			name: "testing",
-			networks: []netip.Prefix{
-				mustParsePrefixUnmapped("10.1.1.2/16"),
-				mustParsePrefixUnmapped("10.1.1.1/24"),
-			},
-			unsafeNetworks: []netip.Prefix{
-				mustParsePrefixUnmapped("9.1.1.3/16"),
-				mustParsePrefixUnmapped("9.1.1.2/24"),
-			},
-			groups:    []string{"test-group1", "test-group2", "test-group3"},
-			notBefore: before,
-			notAfter:  after,
-			isCA:      false,
-			issuer:    "1234567890abcdef1234567890abcdef",
-		},
-		signature: []byte("1234567890abcdef1234567890abcdef"),
-		publicKey: pubKey,
-	}
-
-	db, err := nc.details.Marshal()
-	require.NoError(t, err)
-	nc.rawDetails = db
-
-	certWithPubkey, err := nc.Marshal()
-	require.NoError(t, err)
-	//t.Log("Cert size:", len(b))
-	certWithoutPubkey, err := nc.MarshalForHandshakes()
-	require.NoError(t, err)
-
-	// Cert must not have a pubkey if one is passed in as an argument
-	_, err = unmarshalCertificateV2(certWithPubkey, pubKey, Curve_CURVE25519)
-	require.ErrorIs(t, err, ErrCertPubkeyPresent)
-
-	// Certs must have pubkeys
-	_, err = unmarshalCertificateV2(certWithoutPubkey, nil, Curve_CURVE25519)
-	require.ErrorIs(t, err, ErrBadFormat)
-
-	// Ensure proper unmarshal if a pubkey is passed in
-	nc2, err := unmarshalCertificateV2(certWithoutPubkey, pubKey, Curve_CURVE25519)
-	require.NoError(t, err)
-
-	assert.Equal(t, nc.PublicKey(), nc2.PublicKey())
-}
-
 func TestCertificateV2_PublicKeyPem(t *testing.T) {
 	t.Parallel()
 	before := time.Now().Add(time.Second * -60).Round(time.Second)

cert/crypto_test.go

@@ -79,7 +79,7 @@ qrlJ69wer3ZUHFXA
 	assert.Nil(t, k)
 	assert.Equal(t, rest, invalidPem)
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	curve, k, rest, err = DecryptAndUnmarshalSigningPrivateKey(passphrase, rest)
 	require.EqualError(t, err, "input did not contain a valid PEM encoded block")

cert/errors.go

@@ -21,7 +21,6 @@ var (
 	ErrPrivateKeyEncrypted        = errors.New("private key must be decrypted")
 	ErrCaNotFound                 = errors.New("could not find ca for the certificate")
 	ErrUnknownVersion             = errors.New("certificate version unrecognized")
-	ErrCertPubkeyPresent          = errors.New("certificate has unexpected pubkey present")
 
 	ErrInvalidPEMBlock                   = errors.New("input did not contain a valid PEM encoded block")
 	ErrInvalidPEMCertificateBanner       = errors.New("bytes did not contain a proper certificate banner")

cert/p256/p256.go

@@ -1,127 +0,0 @@
-package p256
-
-import (
-	"crypto/elliptic"
-	"errors"
-	"math/big"
-
-	"filippo.io/bigmod"
-
-	"golang.org/x/crypto/cryptobyte"
-	"golang.org/x/crypto/cryptobyte/asn1"
-)
-
-var halfN = new(big.Int).Rsh(elliptic.P256().Params().N, 1)
-var nMod *bigmod.Modulus
-
-func init() {
-	n, err := bigmod.NewModulus(elliptic.P256().Params().N.Bytes())
-	if err != nil {
-		panic(err)
-	}
-	nMod = n
-}
-
-func IsNormalized(sig []byte) (bool, error) {
-	r, s, err := parseSignature(sig)
-	if err != nil {
-		return false, err
-	}
-	return checkLowS(r, s), nil
-}
-
-func checkLowS(_, s []byte) bool {
-	bigS := new(big.Int).SetBytes(s)
-	// Check if S <= (N/2), because we want to include the midpoint in the set of low-s
-	return bigS.Cmp(halfN) <= 0
-}
-
-func swap(r, s []byte) ([]byte, []byte, error) {
-	var err error
-	bigS, err := bigmod.NewNat().SetBytes(s, nMod)
-	if err != nil {
-		return nil, nil, err
-	}
-	sNormalized := nMod.Nat().Sub(bigS, nMod)
-
-	result := sNormalized.Bytes(nMod)
-	for len(result) > 1 && result[0] == 0 {
-		result = result[1:]
-	}
-
-	return r, result, nil
-}
-
-func Normalize(sig []byte) ([]byte, error) {
-	r, s, err := parseSignature(sig)
-	if err != nil {
-		return nil, err
-	}
-
-	if checkLowS(r, s) {
-		return sig, nil
-	}
-
-	newR, newS, err := swap(r, s)
-	if err != nil {
-		return nil, err
-	}
-
-	return encodeSignature(newR, newS)
-}
-
-// Swap will change sig between its current form to the opposite high or low form.
-func Swap(sig []byte) ([]byte, error) {
-	r, s, err := parseSignature(sig)
-	if err != nil {
-		return nil, err
-	}
-
-	newR, newS, err := swap(r, s)
-	if err != nil {
-		return nil, err
-	}
-
-	return encodeSignature(newR, newS)
-}
-
-// parseSignature taken exactly from crypto/ecdsa/ecdsa.go
-func parseSignature(sig []byte) (r, s []byte, err error) {
-	var inner cryptobyte.String
-	input := cryptobyte.String(sig)
-	if !input.ReadASN1(&inner, asn1.SEQUENCE) ||
-		!input.Empty() ||
-		!inner.ReadASN1Integer(&r) ||
-		!inner.ReadASN1Integer(&s) ||
-		!inner.Empty() {
-		return nil, nil, errors.New("invalid ASN.1")
-	}
-	return r, s, nil
-}
-
-func encodeSignature(r, s []byte) ([]byte, error) {
-	var b cryptobyte.Builder
-	b.AddASN1(asn1.SEQUENCE, func(b *cryptobyte.Builder) {
-		addASN1IntBytes(b, r)
-		addASN1IntBytes(b, s)
-	})
-	return b.Bytes()
-}
-
-// addASN1IntBytes encodes in ASN.1 a positive integer represented as
-// a big-endian byte slice with zero or more leading zeroes.
-func addASN1IntBytes(b *cryptobyte.Builder, bytes []byte) {
-	for len(bytes) > 0 && bytes[0] == 0 {
-		bytes = bytes[1:]
-	}
-	if len(bytes) == 0 {
-		b.SetError(errors.New("invalid integer"))
-		return
-	}
-	b.AddASN1(asn1.INTEGER, func(c *cryptobyte.Builder) {
-		if bytes[0]&0x80 != 0 {
-			c.AddUint8(0)
-		}
-		c.AddBytes(bytes)
-	})
-}

cert/p256/p256_test.go

@@ -1,28 +0,0 @@
-package p256
-
-import (
-	"crypto/ecdsa"
-	"crypto/elliptic"
-	"crypto/rand"
-	"testing"
-
-	"github.com/stretchr/testify/require"
-)
-
-func TestFlipping(t *testing.T) {
-	priv, err1 := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
-	require.NoError(t, err1)
-
-	out, err := ecdsa.SignASN1(rand.Reader, priv, []byte("big chungus"))
-	require.NoError(t, err)
-
-	r, s, err := parseSignature(out)
-	require.NoError(t, err)
-
-	r, s1, err := swap(r, s)
-	require.NoError(t, err)
-	r, s2, err := swap(r, s1)
-	require.NoError(t, err)
-	require.Equal(t, s, s2)
-	require.NotEqual(t, s, s1)
-}

cert/pem.go

@@ -1,66 +1,12 @@
 package cert
 
 import (
-	"bytes"
 	"encoding/pem"
-	"errors"
 	"fmt"
 
 	"golang.org/x/crypto/ed25519"
 )
 
-var ErrTruncatedPEMBlock = errors.New("truncated PEM block")
-
-// SplitPEM is a split function for bufio.Scanner that returns each PEM block.
-func SplitPEM(data []byte, atEOF bool) (advance int, token []byte, err error) {
-	// Look for the start of a PEM block
-	start := bytes.Index(data, []byte("-----BEGIN "))
-	if start == -1 {
-		if atEOF && len(bytes.TrimSpace(data)) > 0 {
-			// Non-whitespace content with no PEM block
-			return 0, nil, ErrTruncatedPEMBlock
-		}
-		if atEOF {
-			return len(data), nil, nil
-		}
-		// Request more data
-		return 0, nil, nil
-	}
-
-	// Look for the end marker
-	endMarkerStart := bytes.Index(data[start:], []byte("-----END "))
-	if endMarkerStart == -1 {
-		if atEOF {
-			// Incomplete PEM block at EOF
-			return 0, nil, ErrTruncatedPEMBlock
-		}
-		// Need more data to find the end
-		return 0, nil, nil
-	}
-
-	// Find the actual end of the END line (after the newline)
-	endMarkerStart += start
-	endLineEnd := bytes.IndexByte(data[endMarkerStart:], '\n')
-	var end int
-	if endLineEnd == -1 {
-		if atEOF {
-			// END marker without newline at EOF - take it anyway
-			end = len(data)
-		} else {
-			// Need more data
-			return 0, nil, nil
-		}
-	} else {
-		end = endMarkerStart + endLineEnd + 1
-	}
-
-	// Extract the PEM block
-	pemBlock := data[start:end]
-
-	// Return the valid PEM block
-	return end, pemBlock, nil
-}
-
 const ( //cert banners
 	CertificateBanner   = "NEBULA CERTIFICATE"
 	CertificateV2Banner = "NEBULA CERTIFICATE V2"
@@ -91,7 +37,19 @@ func UnmarshalCertificateFromPEM(b []byte) (Certificate, []byte, error) {
 		return nil, r, ErrInvalidPEMBlock
 	}
 
-	c, err := unmarshalCertificateBlock(p)
+	var c Certificate
+	var err error
+
+	switch p.Type {
+	// Implementations must validate the resulting certificate contains valid information
+	case CertificateBanner:
+		c, err = unmarshalCertificateV1(p.Bytes, nil)
+	case CertificateV2Banner:
+		c, err = unmarshalCertificateV2(p.Bytes, nil, Curve_CURVE25519)
+	default:
+		return nil, r, ErrInvalidPEMCertificateBanner
+	}
+
 	if err != nil {
 		return nil, r, err
 	}
@@ -100,20 +58,6 @@ func UnmarshalCertificateFromPEM(b []byte) (Certificate, []byte, error) {
 
 }
 
-// unmarshalCertificateBlock decodes a single PEM block into a certificate.
-// It expects a Nebula certificate banner and returns ErrInvalidPEMCertificateBanner otherwise.
-func unmarshalCertificateBlock(block *pem.Block) (Certificate, error) {
-	switch block.Type {
-	// Implementations must validate the resulting certificate contains valid information
-	case CertificateBanner:
-		return unmarshalCertificateV1(block.Bytes, nil)
-	case CertificateV2Banner:
-		return unmarshalCertificateV2(block.Bytes, nil, Curve_CURVE25519)
-	default:
-		return nil, ErrInvalidPEMCertificateBanner
-	}
-}
-
 func marshalCertPublicKeyToPEM(c Certificate) []byte {
 	if c.IsCA() {
 		return MarshalSigningPublicKeyToPEM(c.Curve(), c.PublicKey())

cert/pem_test.go

@@ -1,88 +1,12 @@
 package cert
 
 import (
-	"bufio"
-	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
 
-func scanAll(t *testing.T, input string) ([]string, error) {
-	t.Helper()
-	scanner := bufio.NewScanner(strings.NewReader(input))
-	scanner.Split(SplitPEM)
-	var blocks []string
-	for scanner.Scan() {
-		blocks = append(blocks, scanner.Text())
-	}
-	return blocks, scanner.Err()
-}
-
-func TestSplitPEM_Single(t *testing.T) {
-	input := "-----BEGIN TEST-----\ndata\n-----END TEST-----\n"
-	blocks, err := scanAll(t, input)
-	require.NoError(t, err)
-	require.Len(t, blocks, 1)
-	require.Equal(t, input, blocks[0])
-}
-
-func TestSplitPEM_Multiple(t *testing.T) {
-	block1 := "-----BEGIN TEST-----\naaa\n-----END TEST-----\n"
-	block2 := "-----BEGIN TEST-----\nbbb\n-----END TEST-----\n"
-	blocks, err := scanAll(t, block1+block2)
-	require.NoError(t, err)
-	require.Len(t, blocks, 2)
-	require.Equal(t, block1, blocks[0])
-	require.Equal(t, block2, blocks[1])
-}
-
-func TestSplitPEM_CommentsAndWhitespaceBetweenBlocks(t *testing.T) {
-	input := "# comment\n\n-----BEGIN TEST-----\naaa\n-----END TEST-----\n\n# another comment\n\n-----BEGIN TEST-----\nbbb\n-----END TEST-----\n"
-	blocks, err := scanAll(t, input)
-	require.NoError(t, err)
-	require.Len(t, blocks, 2)
-}
-
-func TestSplitPEM_Empty(t *testing.T) {
-	blocks, err := scanAll(t, "")
-	require.NoError(t, err)
-	require.Empty(t, blocks)
-}
-
-func TestSplitPEM_WhitespaceOnly(t *testing.T) {
-	blocks, err := scanAll(t, "  \n\t\n  ")
-	require.NoError(t, err)
-	require.Empty(t, blocks)
-}
-
-func TestSplitPEM_TrailingGarbage(t *testing.T) {
-	input := "-----BEGIN TEST-----\ndata\n-----END TEST-----\ngarbage"
-	blocks, err := scanAll(t, input)
-	require.ErrorIs(t, err, ErrTruncatedPEMBlock)
-	require.Len(t, blocks, 1)
-}
-
-func TestSplitPEM_TruncatedBlock(t *testing.T) {
-	input := "-----BEGIN TEST-----\npartial data with no end"
-	_, err := scanAll(t, input)
-	require.ErrorIs(t, err, ErrTruncatedPEMBlock)
-}
-
-func TestSplitPEM_NoEndNewline(t *testing.T) {
-	input := "-----BEGIN TEST-----\ndata\n-----END TEST-----"
-	blocks, err := scanAll(t, input)
-	require.NoError(t, err)
-	require.Len(t, blocks, 1)
-	require.Equal(t, input, blocks[0])
-}
-
-func TestSplitPEM_GarbageOnly(t *testing.T) {
-	_, err := scanAll(t, "this is not PEM data")
-	require.ErrorIs(t, err, ErrTruncatedPEMBlock)
-}
-
 func TestUnmarshalCertificateFromPEM(t *testing.T) {
 	goodCert := []byte(`
 # A good cert
@@ -120,7 +44,7 @@ bzBEr00kERQxxTzTsH8cpYEgRoipvmExvg8WP8NdAJEYJosB
 	assert.Equal(t, rest, invalidPem)
 	require.EqualError(t, err, "bytes did not contain a proper certificate banner")
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	cert, rest, err = UnmarshalCertificateFromPEM(rest)
 	assert.Nil(t, cert)
@@ -182,7 +106,7 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 	assert.Equal(t, rest, invalidPem)
 	require.EqualError(t, err, "bytes did not contain a proper Ed25519/ECDSA private key banner")
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	k, rest, curve, err = UnmarshalSigningPrivateKeyFromPEM(rest)
 	assert.Nil(t, k)
@@ -244,7 +168,7 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 	assert.Equal(t, rest, invalidPem)
 	require.EqualError(t, err, "bytes did not contain a proper private key banner")
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	k, rest, curve, err = UnmarshalPrivateKeyFromPEM(rest)
 	assert.Nil(t, k)
@@ -297,7 +221,7 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 	require.EqualError(t, err, "bytes did not contain a proper public key banner")
 	assert.Equal(t, rest, invalidPem)
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	k, rest, curve, err = UnmarshalPublicKeyFromPEM(rest)
 	assert.Nil(t, k)
@@ -375,7 +299,7 @@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA=
 	require.EqualError(t, err, "bytes did not contain a proper public key banner")
 	assert.Equal(t, rest, invalidPem)
 
-	// Fail due to invalid PEM format, because
+	// Fail due to ivalid PEM format, because
 	// it's missing the requisite pre-encapsulation boundary.
 	k, rest, curve, err = UnmarshalPublicKeyFromPEM(rest)
 	assert.Nil(t, k)

cert/sign.go

@@ -9,8 +9,6 @@ import (
 	"fmt"
 	"net/netip"
 	"time"
-
-	"github.com/slackhq/nebula/cert/p256"
 )
 
 // TBSCertificate represents a certificate intended to be signed.
@@ -128,13 +126,6 @@ func (t *TBSCertificate) SignWith(signer Certificate, curve Curve, sp SignerLamb
 		return nil, err
 	}
 
-	if curve == Curve_P256 {
-		sig, err = p256.Normalize(sig)
-		if err != nil {
-			return nil, err
-		}
-	}
-
 	err = c.setSignature(sig)
 	if err != nil {
 		return nil, err

cert/sign_test.go

@@ -9,7 +9,6 @@ import (
 	"testing"
 	"time"
 
-	"github.com/slackhq/nebula/cert/p256"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -90,48 +89,3 @@ func TestCertificateV1_SignP256(t *testing.T) {
 	require.NoError(t, err)
 	assert.NotNil(t, uc)
 }
-
-func TestCertificate_SignP256_AlwaysNormalized(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")
-
-	tbs := TBSCertificate{
-		Version: Version1,
-		Name:    "testing",
-		Networks: []netip.Prefix{
-			mustParsePrefixUnmapped("10.1.1.1/24"),
-			mustParsePrefixUnmapped("10.1.1.2/16"),
-		},
-		UnsafeNetworks: []netip.Prefix{
-			mustParsePrefixUnmapped("9.1.1.2/24"),
-			mustParsePrefixUnmapped("9.1.1.3/16"),
-		},
-		Groups:    []string{"test-group1", "test-group2", "test-group3"},
-		NotBefore: before,
-		NotAfter:  after,
-		PublicKey: pubKey,
-		IsCA:      true,
-		Curve:     Curve_P256,
-	}
-
-	priv, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
-	require.NoError(t, err)
-	pub := elliptic.Marshal(elliptic.P256(), priv.PublicKey.X, priv.PublicKey.Y)
-	rawPriv := priv.D.FillBytes(make([]byte, 32))
-
-	for i := 0; i < 1000; i++ {
-		if i&1 == 1 {
-			tbs.Version = Version1
-		} else {
-			tbs.Version = Version2
-		}
-		c, err := tbs.Sign(nil, Curve_P256, rawPriv)
-		require.NoError(t, err)
-		assert.NotNil(t, c)
-		assert.True(t, c.CheckSignature(pub))
-		normie, err := p256.IsNormalized(c.Signature())
-		require.NoError(t, err)
-		assert.True(t, normie)
-	}
-}

cmd/nebula-cert/ca_test.go

@@ -200,7 +200,7 @@ func Test_ca(t *testing.T) {
 	assert.Empty(t, b)
 	assert.Len(t, lKey, 64)
 
-	// test when reading password results in an error
+	// test when reading passsword results in an error
 	os.Remove(keyF.Name())
 	os.Remove(crtF.Name())
 	ob.Reset()

cmd/nebula-cert/verify.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"io"
 	"os"
+	"strings"
 	"time"
 
 	"github.com/slackhq/nebula/cert"
@@ -39,15 +40,21 @@ func verify(args []string, out io.Writer, errOut io.Writer) error {
 		return err
 	}
 
-	caFile, err := os.Open(*vf.caPath)
+	rawCACert, err := os.ReadFile(*vf.caPath)
 	if err != nil {
 		return fmt.Errorf("error while reading ca: %w", err)
 	}
-	defer caFile.Close()
 
-	caPool, err := cert.NewCAPoolFromPEMReader(caFile)
-	if err != nil && !errors.Is(err, cert.ErrExpired) {
-		return fmt.Errorf("error while adding ca cert to pool: %w", err)
+	caPool := cert.NewCAPool()
+	for {
+		rawCACert, err = caPool.AddCAFromPEM(rawCACert)
+		if err != nil {
+			return fmt.Errorf("error while adding ca cert to pool: %w", err)
+		}
+
+		if rawCACert == nil || len(rawCACert) == 0 || strings.TrimSpace(string(rawCACert)) == "" {
+			break
+		}
 	}
 
 	rawCert, err := os.ReadFile(*vf.certPath)

cmd/nebula-cert/verify_test.go

@@ -64,7 +64,7 @@ func Test_verify(t *testing.T) {
 	err = verify([]string{"-ca", caFile.Name(), "-crt", "does_not_exist"}, ob, eb)
 	assert.Empty(t, ob.String())
 	assert.Empty(t, eb.String())
-	require.ErrorIs(t, err, cert.ErrInvalidPEMBlock)
+	require.EqualError(t, err, "error while adding ca cert to pool: input did not contain a valid PEM encoded block")
 
 	// make a ca for later
 	caPub, caPriv, _ := ed25519.GenerateKey(rand.Reader)

cmd/nebula-service/logs_generic.go

@@ -3,15 +3,8 @@
 
 package main
 
-import (
-	"log/slog"
-	"os"
+import "github.com/sirupsen/logrus"
 
-	"github.com/slackhq/nebula/logging"
-)
-
-// newPlatformLogger returns a *slog.Logger that writes to stdout. Non-Windows
-// platforms have no special sink to integrate with.
-func newPlatformLogger() *slog.Logger {
-	return logging.NewLogger(os.Stdout)
+func HookLogger(l *logrus.Logger) {
+	// Do nothing, let the logs flow to stdout/stderr
 }

cmd/nebula-service/logs_windows.go

@@ -1,86 +1,54 @@
 package main
 
 import (
-	"context"
-	"log/slog"
-	"strings"
-	"sync"
+	"fmt"
+	"io/ioutil"
+	"os"
 
-	"github.com/slackhq/nebula/logging"
+	"github.com/kardianos/service"
+	"github.com/sirupsen/logrus"
 )
 
-// newPlatformLogger returns a *slog.Logger that routes every log record
-// through the Windows service logger so records end up in the Windows
-// Event Log. All the heavy lifting (level management, format swap,
-// timestamp toggle, WithAttrs/WithGroup) comes from logging.NewHandler;
-// this file only contributes:
-//
-//   - an io.Writer that forwards each formatted line to the service
-//     logger at the current record's Event Log severity, and
-//   - a thin severityTag that embeds *logging.Handler and overrides
-//     only Handle / WithAttrs / WithGroup, so Event Viewer's severity
-//     column and severity-based filters keep working the way they did
-//     before the slog migration.
-//
-// Format (text vs json) is carried by the embedded *logging.Handler, so
-// logging.format: json in config still produces JSON lines in Event
-// Viewer, same as the pre-slog logrus setup.
-func newPlatformLogger() *slog.Logger {
-	w := &eventLogWriter{}
-	return slog.New(&severityTag{Handler: logging.NewHandler(w), w: w})
+// HookLogger routes the logrus logs through the service logger so that they end up in the Windows Event Viewer
+// logrus output will be discarded
+func HookLogger(l *logrus.Logger) {
+	l.AddHook(newLogHook(logger))
+	l.SetOutput(ioutil.Discard)
 }
 
-// eventLogWriter forwards slog-formatted lines to the Windows service
-// logger at the severity most recently stashed by severityTag.Handle.
-// The mutex serializes the stash + inner.Handle + Write cycle per record
-// across all concurrent goroutines; slog's builtin text/json handlers
-// each hold their own mutex around Write, but that only protects the
-// Write call itself, not our stash-then-handle sequence.
-type eventLogWriter struct {
-	mu    sync.Mutex
-	level slog.Level
+type logHook struct {
+	sl service.Logger
 }
 
-func (w *eventLogWriter) Write(p []byte) (int, error) {
-	line := strings.TrimRight(string(p), "\n")
-	switch {
-	case w.level >= slog.LevelError:
-		return len(p), logger.Error(line)
-	case w.level >= slog.LevelWarn:
-		return len(p), logger.Warning(line)
+func newLogHook(sl service.Logger) *logHook {
+	return &logHook{sl: sl}
+}
+
+func (h *logHook) Fire(entry *logrus.Entry) error {
+	line, err := entry.String()
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Unable to read entry, %v", err)
+		return err
+	}
+
+	switch entry.Level {
+	case logrus.PanicLevel:
+		return h.sl.Error(line)
+	case logrus.FatalLevel:
+		return h.sl.Error(line)
+	case logrus.ErrorLevel:
+		return h.sl.Error(line)
+	case logrus.WarnLevel:
+		return h.sl.Warning(line)
+	case logrus.InfoLevel:
+		return h.sl.Info(line)
+	case logrus.DebugLevel:
+		return h.sl.Info(line)
 	default:
-		return len(p), logger.Info(line)
+		return nil
 	}
 }
 
-// severityTag embeds *logging.Handler to pick up everything it does for
-// free (Enabled, SetLevel, GetLevel, SetFormat, GetFormat,
-// SetDisableTimestamp) and overrides only Handle / WithAttrs / WithGroup
-// so each record's slog.Level is stashed on the writer before formatting
-// and so derived handlers stay wrapped as severityTag rather than
-// downgrading to bare *logging.Handler.
-type severityTag struct {
-	*logging.Handler
-	w *eventLogWriter
-}
-
-func (s *severityTag) Handle(ctx context.Context, r slog.Record) error {
-	s.w.mu.Lock()
-	defer s.w.mu.Unlock()
-	s.w.level = r.Level
-	return s.Handler.Handle(ctx, r)
-}
-
-func (s *severityTag) WithAttrs(attrs []slog.Attr) slog.Handler {
-	if len(attrs) == 0 {
-		return s
-	}
-	return &severityTag{Handler: s.Handler.WithAttrs(attrs).(*logging.Handler), w: s.w}
-}
-
-func (s *severityTag) WithGroup(name string) slog.Handler {
-	if name == "" {
-		return s
-	}
-	return &severityTag{Handler: s.Handler.WithGroup(name).(*logging.Handler), w: s.w}
+func (h *logHook) Levels() []logrus.Level {
+	return logrus.AllLevels
 }

cmd/nebula-service/main.go

@@ -7,9 +7,9 @@ import (
 	"runtime/debug"
 	"strings"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/logging"
 	"github.com/slackhq/nebula/util"
 )
 
@@ -50,14 +50,9 @@ func main() {
 		os.Exit(0)
 	}
 
-	l := logging.NewLogger(os.Stdout)
-
 	if *serviceFlag != "" {
-		if err := doService(configPath, configTest, Build, serviceFlag); err != nil {
-			l.Error("Service command failed", "error", err)
-			os.Exit(1)
-		}
-		return
+		doService(configPath, configTest, Build, serviceFlag)
+		os.Exit(1)
 	}
 
 	if *configPath == "" {
@@ -66,6 +61,9 @@ func main() {
 		os.Exit(1)
 	}
 
+	l := logrus.New()
+	l.Out = os.Stdout
+
 	c := config.NewC(l)
 	err := c.Load(*configPath)
 	if err != nil {
@@ -73,16 +71,6 @@ func main() {
 		os.Exit(1)
 	}
 
-	if err := logging.ApplyConfig(l, c); err != nil {
-		fmt.Printf("failed to apply logging config: %s", err)
-		os.Exit(1)
-	}
-	c.RegisterReloadCallback(func(c *config.C) {
-		if err := logging.ApplyConfig(l, c); err != nil {
-			l.Error("Failed to reconfigure logger on reload", "error", err)
-		}
-	})
-
 	ctrl, err := nebula.Main(c, *configTest, Build, l, nil)
 	if err != nil {
 		util.LogWithContextIfNeeded("Failed to start", err, l)
@@ -90,20 +78,8 @@ func main() {
 	}
 
 	if !*configTest {
-		wait, err := ctrl.Start()
-		if err != nil {
-			util.LogWithContextIfNeeded("Error while running", err, l)
-			os.Exit(1)
-		}
-
-		go ctrl.ShutdownBlock()
-
-		if err := wait(); err != nil {
-			l.Error("Nebula stopped due to fatal error", "error", err)
-			os.Exit(2)
-		}
-
-		l.Info("Goodbye")
+		ctrl.Start()
+		ctrl.ShutdownBlock()
 	}
 
 	os.Exit(0)

cmd/nebula-service/service.go

@@ -7,9 +7,9 @@ import (
 	"path/filepath"
 
 	"github.com/kardianos/service"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/logging"
 )
 
 var logger service.Logger
@@ -25,7 +25,8 @@ func (p *program) Start(s service.Service) error {
 	// Start should not block.
 	logger.Info("Nebula service starting.")
 
-	l := newPlatformLogger()
+	l := logrus.New()
+	HookLogger(l)
 
 	c := config.NewC(l)
 	err := c.Load(*p.configPath)
@@ -33,15 +34,6 @@ func (p *program) Start(s service.Service) error {
 		return fmt.Errorf("failed to load config: %s", err)
 	}
 
-	if err := logging.ApplyConfig(l, c); err != nil {
-		return fmt.Errorf("failed to apply logging config: %s", err)
-	}
-	c.RegisterReloadCallback(func(c *config.C) {
-		if err := logging.ApplyConfig(l, c); err != nil {
-			l.Error("Failed to reconfigure logger on reload", "error", err)
-		}
-	})
-
 	p.control, err = nebula.Main(c, *p.configTest, Build, l, nil)
 	if err != nil {
 		return err
@@ -65,11 +57,11 @@ func fileExists(filename string) bool {
 	return true
 }
 
-func doService(configPath *string, configTest *bool, build string, serviceFlag *string) error {
+func doService(configPath *string, configTest *bool, build string, serviceFlag *string) {
 	if *configPath == "" {
 		ex, err := os.Executable()
 		if err != nil {
-			return err
+			panic(err)
 		}
 		*configPath = filepath.Dir(ex) + "/config.yaml"
 		if !fileExists(*configPath) {
@@ -93,16 +85,16 @@ func doService(configPath *string, configTest *bool, build string, serviceFlag *
 	// Here are what the different loggers are doing:
 	// - `log` is the standard go log utility, meant to be used while the process is still attached to stdout/stderr
 	// - `logger` is the service log utility that may be attached to a special place depending on OS (Windows will have it attached to the event log)
-	// - in program.Start we build a *slog.Logger via newPlatformLogger; on non-Windows that is a stdout-backed slog logger, on Windows it routes records through the service logger
+	// - above, in `Run` we create a `logrus.Logger` which is what nebula expects to use
 	s, err := service.New(prg, svcConfig)
 	if err != nil {
-		return err
+		log.Fatal(err)
 	}
 
 	errs := make(chan error, 5)
 	logger, err = s.Logger(errs)
 	if err != nil {
-		return err
+		log.Fatal(err)
 	}
 
 	go func() {
@@ -117,16 +109,18 @@ func doService(configPath *string, configTest *bool, build string, serviceFlag *
 
 	switch *serviceFlag {
 	case "run":
-		if err := s.Run(); err != nil {
+		err = s.Run()
+		if err != nil {
 			// Route any errors to the system logger
 			logger.Error(err)
 		}
 	default:
-		if err := service.Control(s, *serviceFlag); err != nil {
+		err := service.Control(s, *serviceFlag)
+		if err != nil {
 			log.Printf("Valid actions: %q\n", service.ControlAction)
-			return err
+			log.Fatal(err)
 		}
+		return
 	}
 
-	return nil
 }

cmd/nebula/main.go

@@ -3,13 +3,16 @@ package main
 import (
 	"flag"
 	"fmt"
+	"log"
+	"net/http"
+	_ "net/http/pprof"
 	"os"
 	"runtime/debug"
 	"strings"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/logging"
 	"github.com/slackhq/nebula/util"
 )
 
@@ -55,7 +58,8 @@ func main() {
 		os.Exit(1)
 	}
 
-	l := logging.NewLogger(os.Stdout)
+	l := logrus.New()
+	l.Out = os.Stdout
 
 	c := config.NewC(l)
 	err := c.Load(*configPath)
@@ -64,38 +68,20 @@ func main() {
 		os.Exit(1)
 	}
 
-	if err := logging.ApplyConfig(l, c); err != nil {
-		fmt.Printf("failed to apply logging config: %s", err)
-		os.Exit(1)
-	}
-	c.RegisterReloadCallback(func(c *config.C) {
-		if err := logging.ApplyConfig(l, c); err != nil {
-			l.Error("Failed to reconfigure logger on reload", "error", err)
-		}
-	})
-
 	ctrl, err := nebula.Main(c, *configTest, Build, l, nil)
 	if err != nil {
 		util.LogWithContextIfNeeded("Failed to start", err, l)
 		os.Exit(1)
 	}
 
+	go func() {
+		log.Println(http.ListenAndServe("0.0.0.0:6060", nil))
+	}()
+
 	if !*configTest {
-		wait, err := ctrl.Start()
-		if err != nil {
-			util.LogWithContextIfNeeded("Error while running", err, l)
-			os.Exit(1)
-		}
-
-		go ctrl.ShutdownBlock()
+		ctrl.Start()
 		notifyReady(l)
-
-		if err := wait(); err != nil {
-			l.Error("Nebula stopped due to fatal error", "error", err)
-			os.Exit(2)
-		}
-
-		l.Info("Goodbye")
+		ctrl.ShutdownBlock()
 	}
 
 	os.Exit(0)

cmd/nebula/notify_linux.go

@@ -1,10 +1,11 @@
 package main
 
 import (
-	"log/slog"
 	"net"
 	"os"
 	"time"
+
+	"github.com/sirupsen/logrus"
 )
 
 // SdNotifyReady tells systemd the service is ready and dependent services can now be started
@@ -12,30 +13,30 @@ import (
 // https://www.freedesktop.org/software/systemd/man/systemd.service.html
 const SdNotifyReady = "READY=1"
 
-func notifyReady(l *slog.Logger) {
+func notifyReady(l *logrus.Logger) {
 	sockName := os.Getenv("NOTIFY_SOCKET")
 	if sockName == "" {
-		l.Debug("NOTIFY_SOCKET systemd env var not set, not sending ready signal")
+		l.Debugln("NOTIFY_SOCKET systemd env var not set, not sending ready signal")
 		return
 	}
 
 	conn, err := net.DialTimeout("unixgram", sockName, time.Second)
 	if err != nil {
-		l.Error("failed to connect to systemd notification socket", "error", err)
+		l.WithError(err).Error("failed to connect to systemd notification socket")
 		return
 	}
 	defer conn.Close()
 
 	err = conn.SetWriteDeadline(time.Now().Add(time.Second))
 	if err != nil {
-		l.Error("failed to set the write deadline for the systemd notification socket", "error", err)
+		l.WithError(err).Error("failed to set the write deadline for the systemd notification socket")
 		return
 	}
 
 	if _, err = conn.Write([]byte(SdNotifyReady)); err != nil {
-		l.Error("failed to signal the systemd notification socket", "error", err)
+		l.WithError(err).Error("failed to signal the systemd notification socket")
 		return
 	}
 
-	l.Debug("notified systemd the service is ready")
+	l.Debugln("notified systemd the service is ready")
 }

cmd/nebula/notify_notlinux.go

@@ -3,8 +3,8 @@
 
 package main
 
-import "log/slog"
+import "github.com/sirupsen/logrus"
 
-func notifyReady(_ *slog.Logger) {
+func notifyReady(_ *logrus.Logger) {
 	// No init service to notify
 }

config/config.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"log/slog"
 	"math"
 	"os"
 	"os/signal"
@@ -17,6 +16,7 @@ import (
 	"time"
 
 	"dario.cat/mergo"
+	"github.com/sirupsen/logrus"
 	"go.yaml.in/yaml/v3"
 )
 
@@ -26,11 +26,11 @@ type C struct {
 	Settings    map[string]any
 	oldSettings map[string]any
 	callbacks   []func(*C)
-	l           *slog.Logger
+	l           *logrus.Logger
 	reloadLock  sync.Mutex
 }
 
-func NewC(l *slog.Logger) *C {
+func NewC(l *logrus.Logger) *C {
 	return &C{
 		Settings: make(map[string]any),
 		l:        l,
@@ -107,18 +107,12 @@ func (c *C) HasChanged(k string) bool {
 
 	newVals, err := yaml.Marshal(nv)
 	if err != nil {
-		c.l.Error("Error while marshaling new config",
-			"config_path", k,
-			"error", err,
-		)
+		c.l.WithField("config_path", k).WithError(err).Error("Error while marshaling new config")
 	}
 
 	oldVals, err := yaml.Marshal(ov)
 	if err != nil {
-		c.l.Error("Error while marshaling old config",
-			"config_path", k,
-			"error", err,
-		)
+		c.l.WithField("config_path", k).WithError(err).Error("Error while marshaling old config")
 	}
 
 	return string(newVals) != string(oldVals)
@@ -160,10 +154,7 @@ func (c *C) ReloadConfig() {
 
 	err := c.Load(c.path)
 	if err != nil {
-		c.l.Error("Error occurred while reloading config",
-			"config_path", c.path,
-			"error", err,
-		)
+		c.l.WithField("config_path", c.path).WithError(err).Error("Error occurred while reloading config")
 		return
 	}
 

connection_manager.go

@@ -5,13 +5,13 @@ import (
 	"context"
 	"encoding/binary"
 	"fmt"
-	"log/slog"
 	"net/netip"
 	"sync"
 	"sync/atomic"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
@@ -47,10 +47,10 @@ type connectionManager struct {
 
 	metricsTxPunchy metrics.Counter
 
-	l *slog.Logger
+	l *logrus.Logger
 }
 
-func newConnectionManagerFromConfig(l *slog.Logger, c *config.C, hm *HostMap, p *Punchy) *connectionManager {
+func newConnectionManagerFromConfig(l *logrus.Logger, c *config.C, hm *HostMap, p *Punchy) *connectionManager {
 	cm := &connectionManager{
 		hostMap:         hm,
 		l:               l,
@@ -85,10 +85,9 @@ func (cm *connectionManager) reload(c *config.C, initial bool) {
 		old := cm.getInactivityTimeout()
 		cm.inactivityTimeout.Store((int64)(c.GetDuration("tunnels.inactivity_timeout", 10*time.Minute)))
 		if !initial {
-			cm.l.Info("Inactivity timeout has changed",
-				"oldDuration", old,
-				"newDuration", cm.getInactivityTimeout(),
-			)
+			cm.l.WithField("oldDuration", old).
+				WithField("newDuration", cm.getInactivityTimeout()).
+				Info("Inactivity timeout has changed")
 		}
 	}
 
@@ -96,10 +95,9 @@ func (cm *connectionManager) reload(c *config.C, initial bool) {
 		old := cm.dropInactive.Load()
 		cm.dropInactive.Store(c.GetBool("tunnels.drop_inactive", false))
 		if !initial {
-			cm.l.Info("Drop inactive setting has changed",
-				"oldBool", old,
-				"newBool", cm.dropInactive.Load(),
-			)
+			cm.l.WithField("oldBool", old).
+				WithField("newBool", cm.dropInactive.Load()).
+				Info("Drop inactive setting has changed")
 		}
 	}
 }
@@ -258,7 +256,7 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 			var err error
 			index, err = AddRelay(cm.l, newhostinfo, cm.hostMap, r.PeerAddr, nil, r.Type, Requested)
 			if err != nil {
-				cm.l.Error("failed to migrate relay to new hostinfo", "error", err)
+				cm.l.WithError(err).Error("failed to migrate relay to new hostinfo")
 				continue
 			}
 			switch r.Type {
@@ -306,16 +304,16 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 
 		msg, err := req.Marshal()
 		if err != nil {
-			cm.l.Error("failed to marshal Control message to migrate relay", "error", err)
+			cm.l.WithError(err).Error("failed to marshal Control message to migrate relay")
 		} else {
 			cm.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
-			cm.l.Info("send CreateRelayRequest",
-				"relayFrom", req.RelayFromAddr,
-				"relayTo", req.RelayToAddr,
-				"initiatorRelayIndex", req.InitiatorRelayIndex,
-				"responderRelayIndex", req.ResponderRelayIndex,
-				"vpnAddrs", newhostinfo.vpnAddrs,
-			)
+			cm.l.WithFields(logrus.Fields{
+				"relayFrom":           req.RelayFromAddr,
+				"relayTo":             req.RelayToAddr,
+				"initiatorRelayIndex": req.InitiatorRelayIndex,
+				"responderRelayIndex": req.ResponderRelayIndex,
+				"vpnAddrs":            newhostinfo.vpnAddrs}).
+				Info("send CreateRelayRequest")
 		}
 	}
 }
@@ -327,7 +325,7 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 
 	hostinfo := cm.hostMap.Indexes[localIndex]
 	if hostinfo == nil {
-		cm.l.Debug("Not found in hostmap", "localIndex", localIndex)
+		cm.l.WithField("localIndex", localIndex).Debugln("Not found in hostmap")
 		return doNothing, nil, nil
 	}
 
@@ -347,10 +345,10 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 	// A hostinfo is determined alive if there is incoming traffic
 	if inTraffic {
 		decision := doNothing
-		if cm.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(cm.l).Debug("Tunnel status",
-				"tunnelCheck", m{"state": "alive", "method": "passive"},
-			)
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.l).
+				WithField("tunnelCheck", m{"state": "alive", "method": "passive"}).
+				Debug("Tunnel status")
 		}
 		hostinfo.pendingDeletion.Store(false)
 
@@ -377,9 +375,9 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 
 	if hostinfo.pendingDeletion.Load() {
 		// We have already sent a test packet and nothing was returned, this hostinfo is dead
-		hostinfo.logger(cm.l).Info("Tunnel status",
-			"tunnelCheck", m{"state": "dead", "method": "active"},
-		)
+		hostinfo.logger(cm.l).
+			WithField("tunnelCheck", m{"state": "dead", "method": "active"}).
+			Info("Tunnel status")
 
 		return deleteTunnel, hostinfo, nil
 	}
@@ -390,10 +388,10 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 			inactiveFor, isInactive := cm.isInactive(hostinfo, now)
 			if isInactive {
 				// Tunnel is inactive, tear it down
-				hostinfo.logger(cm.l).Info("Dropping tunnel due to inactivity",
-					"inactiveDuration", inactiveFor,
-					"primary", mainHostInfo,
-				)
+				hostinfo.logger(cm.l).
+					WithField("inactiveDuration", inactiveFor).
+					WithField("primary", mainHostInfo).
+					Info("Dropping tunnel due to inactivity")
 
 				return closeTunnel, hostinfo, primary
 			}
@@ -412,18 +410,18 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 			cm.sendPunch(hostinfo)
 		}
 
-		if cm.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(cm.l).Debug("Tunnel status",
-				"tunnelCheck", m{"state": "testing", "method": "active"},
-			)
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.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
 		decision = sendTestPacket
 
 	} else {
-		if cm.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(cm.l).Debug("Hostinfo sadness")
+		if cm.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(cm.l).Debugf("Hostinfo sadness")
 		}
 	}
 
@@ -495,16 +493,14 @@ func (cm *connectionManager) isInvalidCertificate(now time.Time, hostinfo *HostI
 		return false //cert is still valid! yay!
 	} else if err == cert.ErrBlockListed { //avoiding errors.Is for speed
 		// Block listed certificates should always be disconnected
-		hostinfo.logger(cm.l).Info("Remote certificate is blocked, tearing down the tunnel",
-			"error", err,
-			"fingerprint", remoteCert.Fingerprint,
-		)
+		hostinfo.logger(cm.l).WithError(err).
+			WithField("fingerprint", remoteCert.Fingerprint).
+			Info("Remote certificate is blocked, tearing down the tunnel")
 		return true
 	} else if cm.intf.disconnectInvalid.Load() {
-		hostinfo.logger(cm.l).Info("Remote certificate is no longer valid, tearing down the tunnel",
-			"error", err,
-			"fingerprint", remoteCert.Fingerprint,
-		)
+		hostinfo.logger(cm.l).WithError(err).
+			WithField("fingerprint", remoteCert.Fingerprint).
+			Info("Remote certificate is no longer valid, tearing down the tunnel")
 		return true
 	} else {
 		//if we reach here, the cert is no longer valid, but we're configured to keep tunnels from now-invalid certs open
@@ -543,11 +539,10 @@ func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
 	curCrtVersion := curCrt.Version()
 	myCrt := cs.getCertificate(curCrtVersion)
 	if myCrt == nil {
-		cm.l.Info("Re-handshaking with remote",
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"version", curCrtVersion,
-			"reason", "local certificate removed",
-		)
+		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("version", curCrtVersion).
+			WithField("reason", "local certificate removed").
+			Info("Re-handshaking with remote")
 		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 		return
 	}
@@ -555,12 +550,11 @@ func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
 	if peerCrt != nil && curCrtVersion < peerCrt.Certificate.Version() {
 		// if our certificate version is less than theirs, and we have a matching version available, rehandshake?
 		if cs.getCertificate(peerCrt.Certificate.Version()) != nil {
-			cm.l.Info("Re-handshaking with remote",
-				"vpnAddrs", hostinfo.vpnAddrs,
-				"version", curCrtVersion,
-				"peerVersion", peerCrt.Certificate.Version(),
-				"reason", "local certificate version lower than peer, attempting to correct",
-			)
+			cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+				WithField("version", curCrtVersion).
+				WithField("peerVersion", peerCrt.Certificate.Version()).
+				WithField("reason", "local certificate version lower than peer, attempting to correct").
+				Info("Re-handshaking with remote")
 			cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], func(hh *HandshakeHostInfo) {
 				hh.initiatingVersionOverride = peerCrt.Certificate.Version()
 			})
@@ -568,19 +562,17 @@ func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
 		}
 	}
 	if !bytes.Equal(curCrt.Signature(), myCrt.Signature()) {
-		cm.l.Info("Re-handshaking with remote",
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"reason", "local certificate is not current",
-		)
+		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("reason", "local certificate is not current").
+			Info("Re-handshaking with remote")
 
 		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 		return
 	}
 	if curCrtVersion < cs.initiatingVersion {
-		cm.l.Info("Re-handshaking with remote",
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"reason", "current cert version < pki.initiatingVersion",
-		)
+		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("reason", "current cert version < pki.initiatingVersion").
+			Info("Re-handshaking with remote")
 
 		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 		return

connection_manager_test.go

@@ -10,7 +10,6 @@ import (
 	"github.com/flynn/noise"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/overlay/overlaytest"
 	"github.com/slackhq/nebula/test"
 	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
@@ -53,7 +52,7 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &overlaytest.NoopTun{},
+		inside:           &test.NoopTun{},
 		outside:          &udp.NoopConn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
@@ -64,9 +63,9 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(test.NewLogger())
-	punchy := NewPunchyFromConfig(test.NewLogger(), conf)
-	nc := newConnectionManagerFromConfig(test.NewLogger(), conf, hostMap, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
 	nc.intf = ifce
 	p := []byte("")
 	nb := make([]byte, 12, 12)
@@ -136,7 +135,7 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &overlaytest.NoopTun{},
+		inside:           &test.NoopTun{},
 		outside:          &udp.NoopConn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
@@ -147,9 +146,9 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(test.NewLogger())
-	punchy := NewPunchyFromConfig(test.NewLogger(), conf)
-	nc := newConnectionManagerFromConfig(test.NewLogger(), conf, hostMap, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
 	nc.intf = ifce
 	p := []byte("")
 	nb := make([]byte, 12, 12)
@@ -221,7 +220,7 @@ func Test_NewConnectionManager_DisconnectInactive(t *testing.T) {
 	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &overlaytest.NoopTun{},
+		inside:           &test.NoopTun{},
 		outside:          &udp.NoopConn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
@@ -232,12 +231,12 @@ func Test_NewConnectionManager_DisconnectInactive(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(test.NewLogger())
+	conf := config.NewC(l)
 	conf.Settings["tunnels"] = map[string]any{
 		"drop_inactive": true,
 	}
-	punchy := NewPunchyFromConfig(test.NewLogger(), conf)
-	nc := newConnectionManagerFromConfig(test.NewLogger(), conf, hostMap, punchy)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
 	assert.True(t, nc.dropInactive.Load())
 	nc.intf = ifce
 
@@ -348,7 +347,7 @@ func Test_NewConnectionManagerTest_DisconnectInvalid(t *testing.T) {
 	lh := newTestLighthouse()
 	ifce := &Interface{
 		hostMap:          hostMap,
-		inside:           &overlaytest.NoopTun{},
+		inside:           &test.NoopTun{},
 		outside:          &udp.NoopConn{},
 		firewall:         &Firewall{},
 		lightHouse:       lh,
@@ -361,9 +360,9 @@ func Test_NewConnectionManagerTest_DisconnectInvalid(t *testing.T) {
 	ifce.disconnectInvalid.Store(true)
 
 	// Create manager
-	conf := config.NewC(test.NewLogger())
-	punchy := NewPunchyFromConfig(test.NewLogger(), conf)
-	nc := newConnectionManagerFromConfig(test.NewLogger(), conf, hostMap, punchy)
+	conf := config.NewC(l)
+	punchy := NewPunchyFromConfig(l, conf)
+	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
 	nc.intf = ifce
 	ifce.connectionManager = nc
 

connection_state.go

@@ -8,11 +8,12 @@ import (
 	"sync/atomic"
 
 	"github.com/flynn/noise"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/noiseutil"
 )
 
-const ReplayWindow = 1024 //todo I've started seeing out-of-window messages in testing?
+const ReplayWindow = 4096
 
 type ConnectionState struct {
 	eKey           *NebulaCipherState
@@ -26,7 +27,7 @@ type ConnectionState struct {
 	writeLock      sync.Mutex
 }
 
-func NewConnectionState(cs *CertState, crt cert.Certificate, initiator bool, pattern noise.HandshakePattern) (*ConnectionState, error) {
+func NewConnectionState(l *logrus.Logger, cs *CertState, crt cert.Certificate, initiator bool, pattern noise.HandshakePattern) (*ConnectionState, error) {
 	var dhFunc noise.DHFunc
 	switch crt.Curve() {
 	case cert.Curve_CURVE25519:

control.go

@@ -2,33 +2,17 @@ package nebula
 
 import (
 	"context"
-	"errors"
-	"log/slog"
 	"net/netip"
 	"os"
 	"os/signal"
-	"sync"
 	"syscall"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/overlay"
 )
 
-type RunState int
-
-const (
-	StateUnknown RunState = iota
-	StateReady
-	StateStarted
-	StateStopping
-	StateStopped
-)
-
-var ErrAlreadyStarted = errors.New("nebula is already started")
-var ErrAlreadyStopped = errors.New("nebula cannot be restarted")
-var ErrUnknownState = errors.New("nebula state is invalid")
-
 // Every interaction here needs to take extra care to copy memory and not return or use arguments "as is" when touching
 // core. This means copying IP objects, slices, de-referencing pointers and taking the actual value, etc
 
@@ -42,11 +26,8 @@ type controlHostLister interface {
 }
 
 type Control struct {
-	stateLock sync.Mutex
-	state     RunState
-
 	f                      *Interface
-	l                      *slog.Logger
+	l                      *logrus.Logger
 	ctx                    context.Context
 	cancel                 context.CancelFunc
 	sshStart               func()
@@ -68,31 +49,10 @@ type ControlHostInfo struct {
 	CurrentRelaysThroughMe []netip.Addr     `json:"currentRelaysThroughMe"`
 }
 
-// Start actually runs nebula, this is a nonblocking call.
-// The returned function blocks until nebula has fully stopped and returns the
-// first fatal reader error (if any). A nil error means nebula shut down
-// gracefully; a non-nil error means a reader hit an unexpected failure that
-// triggered the shutdown.
-func (c *Control) Start() (func() error, error) {
-	c.stateLock.Lock()
-	defer c.stateLock.Unlock()
-	switch c.state {
-	case StateReady:
-		//yay!
-	case StateStopped, StateStopping:
-		return nil, ErrAlreadyStopped
-	case StateStarted:
-		return nil, ErrAlreadyStarted
-	default:
-		return nil, ErrUnknownState
-	}
-
+// Start actually runs nebula, this is a nonblocking call. To block use Control.ShutdownBlock()
+func (c *Control) Start() {
 	// Activate the interface
-	err := c.f.activate()
-	if err != nil {
-		c.state = StateStopped
-		return nil, err
-	}
+	c.f.activate()
 
 	// Call all the delayed funcs that waited patiently for the interface to be created.
 	if c.sshStart != nil {
@@ -111,51 +71,25 @@ func (c *Control) Start() (func() error, error) {
 		c.lighthouseStart()
 	}
 
-	c.f.triggerShutdown = c.Stop
-
 	// Start reading packets.
-	out, err := c.f.run()
-	if err != nil {
-		c.state = StateStopped
-		return nil, err
-	}
-	c.state = StateStarted
-	return out, nil
-}
-
-func (c *Control) State() RunState {
-	c.stateLock.Lock()
-	defer c.stateLock.Unlock()
-	return c.state
+	c.f.run()
 }
 
 func (c *Control) Context() context.Context {
 	return c.ctx
 }
 
-// Stop is a non-blocking call that signals nebula to close all tunnels and shut down
+// Stop signals nebula to shutdown and close all tunnels, returns after the shutdown is complete
 func (c *Control) Stop() {
-	c.stateLock.Lock()
-	if c.state != StateStarted {
-		c.stateLock.Unlock()
-		// We are stopping or stopped already
-		return
-	}
-
-	c.state = StateStopping
-	c.stateLock.Unlock()
-
 	// 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.Error("Close interface failed", "error", err)
+		c.l.WithError(err).Error("Close interface failed")
 	}
-	c.stateLock.Lock()
-	c.state = StateStopped
-	c.stateLock.Unlock()
+	c.l.Info("Goodbye")
 }
 
 // ShutdownBlock will listen for and block on term and interrupt signals, calling Control.Stop() once signalled
@@ -166,7 +100,7 @@ func (c *Control) ShutdownBlock() {
 
 	rawSig := <-sigChan
 	sig := rawSig.String()
-	c.l.Info("Caught signal, shutting down", "signal", sig)
+	c.l.WithField("signal", sig).Info("Caught signal, shutting down")
 	c.Stop()
 }
 
@@ -303,10 +237,8 @@ func (c *Control) CloseAllTunnels(excludeLighthouses bool) (closed int) {
 		c.f.send(header.CloseTunnel, 0, h.ConnectionState, h, []byte{}, make([]byte, 12, 12), make([]byte, mtu))
 		c.f.closeTunnel(h)
 
-		c.l.Debug("Sending close tunnel message",
-			"vpnAddrs", h.vpnAddrs,
-			"udpAddr", h.remote,
-		)
+		c.l.WithField("vpnAddrs", h.vpnAddrs).WithField("udpAddr", h.remote).
+			Debug("Sending close tunnel message")
 		closed++
 	}
 

control_test.go

@@ -6,6 +6,7 @@ import (
 	"reflect"
 	"testing"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
@@ -78,11 +79,10 @@ func TestControl_GetHostInfoByVpnIp(t *testing.T) {
 	}, &Interface{})
 
 	c := Control{
-		state: StateReady,
 		f: &Interface{
 			hostMap: hm,
 		},
-		l: test.NewLogger(),
+		l: logrus.New(),
 	}
 
 	thi := c.GetHostInfoByVpnAddr(vpnIp, false)

control_tester.go

@@ -1,4 +1,5 @@
 //go:build e2e_testing
+// +build e2e_testing
 
 package nebula
 
@@ -79,7 +80,8 @@ func (c *Control) GetFromTun(block bool) []byte {
 
 // GetFromUDP will pull a udp packet off the udp side of nebula
 func (c *Control) GetFromUDP(block bool) *udp.Packet {
-	return c.f.outside.(*udp.TesterConn).Get(block)
+	out := c.f.outside.(*udp.TesterConn).Get(block)
+	return out
 }
 
 func (c *Control) GetUDPTxChan() <-chan *udp.Packet {

dist/wireshark/nebula.lua

@@ -84,24 +84,30 @@ end
 
 function nebula.prefs_changed()
     if default_settings.all_ports == nebula.prefs.all_ports and default_settings.port == nebula.prefs.port then
+        -- Nothing changed, bail
         return
     end
 
-    -- Remove all existing registrations
+    -- Remove our old dissector
     DissectorTable.get("udp.port"):remove_all(nebula)
 
-    if nebula.prefs.all_ports then
-        -- Register on every port for hole punch capture
+    if nebula.prefs.all_ports and default_settings.all_ports ~= nebula.prefs.all_ports then
+        default_settings.all_port = nebula.prefs.all_ports
+
         for i=0, 65535 do
             DissectorTable.get("udp.port"):add(i, nebula)
         end
-    else
-        -- Register on the configured port only
-        DissectorTable.get("udp.port"):add(nebula.prefs.port, nebula)
+
+        -- no need to establish again on specific ports
+        return
     end
 
-    default_settings.all_ports = nebula.prefs.all_ports
-    default_settings.port = nebula.prefs.port
+
+    if default_settings.all_ports ~= nebula.prefs.all_ports then
+        -- Add our new port dissector
+        default_settings.port = nebula.prefs.port
+        DissectorTable.get("udp.port"):add(default_settings.port, nebula)
+    end
 end
 
 DissectorTable.get("udp.port"):add(default_settings.port, nebula)

dns_server.go

@@ -1,249 +1,63 @@
 package nebula
 
 import (
-	"context"
 	"fmt"
-	"log/slog"
 	"net"
 	"net/netip"
 	"strconv"
 	"strings"
 	"sync"
-	"sync/atomic"
 
 	"github.com/gaissmai/bart"
 	"github.com/miekg/dns"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
 )
 
-type dnsServer struct {
+// This whole thing should be rewritten to use context
+
+var dnsR *dnsRecords
+var dnsServer *dns.Server
+var dnsAddr string
+
+type dnsRecords struct {
 	sync.RWMutex
-	l               *slog.Logger
-	ctx             context.Context
+	l               *logrus.Logger
 	dnsMap4         map[string]netip.Addr
 	dnsMap6         map[string]netip.Addr
 	hostMap         *HostMap
 	myVpnAddrsTable *bart.Lite
-
-	mux *dns.ServeMux
-
-	// enabled mirrors `lighthouse.serve_dns && lighthouse.am_lighthouse`.
-	// Start, Add, and reload consult it so callers don't need to know the
-	// gating rules. When it toggles off via reload, accumulated records are
-	// cleared so a later re-enable starts with a fresh map populated from
-	// new handshakes.
-	enabled atomic.Bool
-
-	serverMu sync.Mutex
-	server   *dns.Server
-	// started is closed once `server` has finished binding (or after
-	// ListenAndServe returns on a bind failure). Stop waits on it before
-	// calling Shutdown to avoid the miekg/dns "server not started" race
-	// where a Shutdown that arrives before bind completes is silently
-	// ignored, leaving the listener running forever.
-	started chan struct{}
-	addr    string
 }
 
-// newDnsServerFromConfig builds a dnsServer, applies the initial config, and
-// registers a reload callback. The reload callback is registered before the
-// initial config is applied, so a SIGHUP can later enable, fix, or disable
-// DNS even if the initial application failed.
-//
-// The dnsServer internally gates on `lighthouse.serve_dns &&
-// lighthouse.am_lighthouse`. Start and Add are safe to call unconditionally,
-// they no-op when DNS isn't enabled. Each Start invocation owns a ctx-cancel
-// watcher that tears the listener down on nebula shutdown. The returned
-// pointer is always non-nil, even on error.
-func newDnsServerFromConfig(ctx context.Context, l *slog.Logger, cs *CertState, hostMap *HostMap, c *config.C) (*dnsServer, error) {
-	ds := &dnsServer{
+func newDnsRecords(l *logrus.Logger, cs *CertState, hostMap *HostMap) *dnsRecords {
+	return &dnsRecords{
 		l:               l,
-		ctx:             ctx,
 		dnsMap4:         make(map[string]netip.Addr),
 		dnsMap6:         make(map[string]netip.Addr),
 		hostMap:         hostMap,
 		myVpnAddrsTable: cs.myVpnAddrsTable,
 	}
-	ds.mux = dns.NewServeMux()
-	ds.mux.HandleFunc(".", ds.handleDnsRequest)
-
-	c.RegisterReloadCallback(func(c *config.C) {
-		if err := ds.reload(c, false); err != nil {
-			ds.l.Error("Failed to reload DNS responder from config", "error", err)
-		}
-	})
-
-	if err := ds.reload(c, true); err != nil {
-		return ds, err
-	}
-	return ds, nil
 }
 
-// reload applies the latest config and reconciles the running state with it:
-//   - enabled toggled on  -> spawn a runner
-//   - enabled toggled off -> stop the runner
-//   - listen address changed (while running) -> restart on the new address
-//   - everything else     -> no-op
-//
-// On the initial call it only records configuration; Control.Start is what
-// launches the first runner via dnsStart.
-func (d *dnsServer) reload(c *config.C, initial bool) error {
-	wantsDns := c.GetBool("lighthouse.serve_dns", false)
-	amLighthouse := c.GetBool("lighthouse.am_lighthouse", false)
-	enabled := wantsDns && amLighthouse
-	newAddr := getDnsServerAddr(c)
-
-	d.serverMu.Lock()
-	running := d.server
-	runningStarted := d.started
-	sameAddr := d.addr == newAddr
-	d.addr = newAddr
-	d.enabled.Store(enabled)
-	d.serverMu.Unlock()
-
-	if initial {
-		if wantsDns && !amLighthouse {
-			d.l.Warn("DNS server refusing to run because this host is not a lighthouse.")
-		}
-		return nil
-	}
-
-	if !enabled {
-		if running != nil {
-			d.Stop()
-		}
-		// Drop any records that accumulated while enabled; a later re-enable
-		// will repopulate from fresh handshakes.
-		d.clearRecords()
-		return nil
-	}
-
-	if running == nil {
-		// Was disabled (or never started); bring it up now.
-		go d.Start()
-		return nil
-	}
-
-	if sameAddr {
-		return nil
-	}
-
-	d.shutdownServer(running, runningStarted, "reload")
-	// Old Start goroutine has now exited; bring up a fresh listener on the
-	// new address.
-	go d.Start()
-	return nil
-}
-
-// shutdownServer waits for the server to finish binding (so Shutdown actually
-// stops it rather than no-oping) and then shuts it down.
-func (d *dnsServer) shutdownServer(srv *dns.Server, started chan struct{}, reason string) {
-	if srv == nil {
-		return
-	}
-	if started != nil {
-		<-started
-	}
-	if err := srv.Shutdown(); err != nil {
-		d.l.Warn("Failed to shut down the DNS responder", "reason", reason, "error", err)
-	}
-}
-
-// Start binds and serves the DNS responder. Blocks until Stop is called or
-// the listener errors. Safe to call when DNS is disabled (returns
-// immediately). This is what Control.dnsStart points at.
-//
-// Must be invoked after the tun device is active so that lighthouse.dns.host
-// may bind to a nebula IP.
-func (d *dnsServer) Start() {
-	if !d.enabled.Load() {
-		return
-	}
-
-	started := make(chan struct{})
-	d.serverMu.Lock()
-	if d.ctx.Err() != nil {
-		d.serverMu.Unlock()
-		return
-	}
-	addr := d.addr
-	server := &dns.Server{
-		Addr:              addr,
-		Net:               "udp",
-		Handler:           d.mux,
-		NotifyStartedFunc: func() { close(started) },
-	}
-	d.server = server
-	d.started = started
-	d.serverMu.Unlock()
-
-	// Per-invocation ctx watcher. Exits when Start does, so we don't leak a
-	// watcher per reload-driven restart.
-	done := make(chan struct{})
-	go func() {
-		select {
-		case <-d.ctx.Done():
-			d.shutdownServer(server, started, "shutdown")
-		case <-done:
-		}
-	}()
-
-	d.l.Info("Starting DNS responder", "dnsListener", addr)
-	err := server.ListenAndServe()
-	close(done)
-
-	// If the listener never bound (bind error) NotifyStartedFunc never fires,
-	// so close started here to release any Stop caller waiting on it.
-	select {
-	case <-started:
-	default:
-		close(started)
-	}
-
-	if err != nil {
-		d.l.Warn("Failed to run the DNS responder", "error", err)
-	}
-}
-
-// Stop shuts down the active server, if any. Idempotent.
-func (d *dnsServer) Stop() {
-	d.serverMu.Lock()
-	srv := d.server
-	started := d.started
-	d.server = nil
-	d.started = nil
-	d.serverMu.Unlock()
-	d.shutdownServer(srv, started, "stop")
-}
-
-// Query returns the address for the given name and query type. The second
-// return value reports whether the name is known at all (in either A or AAAA),
-// which lets callers distinguish NODATA from NXDOMAIN.
-func (d *dnsServer) Query(q uint16, data string) (netip.Addr, bool) {
+func (d *dnsRecords) Query(q uint16, data string) netip.Addr {
 	data = strings.ToLower(data)
 	d.RLock()
 	defer d.RUnlock()
-	addr4, haveV4 := d.dnsMap4[data]
-	addr6, haveV6 := d.dnsMap6[data]
-	nameExists := haveV4 || haveV6
 	switch q {
 	case dns.TypeA:
-		if haveV4 {
-			return addr4, nameExists
+		if r, ok := d.dnsMap4[data]; ok {
+			return r
 		}
 	case dns.TypeAAAA:
-		if haveV6 {
-			return addr6, nameExists
+		if r, ok := d.dnsMap6[data]; ok {
+			return r
 		}
 	}
 
-	return netip.Addr{}, nameExists
+	return netip.Addr{}
 }
 
-func (d *dnsServer) QueryCert(data string) string {
-	if len(data) < 2 {
-		return ""
-	}
+func (d *dnsRecords) QueryCert(data string) string {
 	ip, err := netip.ParseAddr(data[:len(data)-1])
 	if err != nil {
 		return ""
@@ -266,19 +80,8 @@ func (d *dnsServer) QueryCert(data string) string {
 	return string(b)
 }
 
-// clearRecords drops all DNS records.
-func (d *dnsServer) clearRecords() {
-	d.Lock()
-	defer d.Unlock()
-	clear(d.dnsMap4)
-	clear(d.dnsMap6)
-}
-
 // Add adds the first IPv4 and IPv6 address that appears in `addresses` as the record for `host`
-func (d *dnsServer) Add(host string, addresses []netip.Addr) {
-	if !d.enabled.Load() {
-		return
-	}
+func (d *dnsRecords) Add(host string, addresses []netip.Addr) {
 	host = strings.ToLower(host)
 	d.Lock()
 	defer d.Unlock()
@@ -298,7 +101,7 @@ func (d *dnsServer) Add(host string, addresses []netip.Addr) {
 	}
 }
 
-func (d *dnsServer) isSelfNebulaOrLocalhost(addr string) bool {
+func (d *dnsRecords) isSelfNebulaOrLocalhost(addr string) bool {
 	a, _, _ := net.SplitHostPort(addr)
 	b, err := netip.ParseAddr(a)
 	if err != nil {
@@ -313,24 +116,13 @@ func (d *dnsServer) isSelfNebulaOrLocalhost(addr string) bool {
 	return d.myVpnAddrsTable.Contains(b)
 }
 
-func (d *dnsServer) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
-	debugEnabled := d.l.Enabled(context.Background(), slog.LevelDebug)
-	// Per RFC 2308 §2.2, a name that exists but has no record of the requested
-	// type must be answered with NOERROR and an empty answer section (NODATA),
-	// not NXDOMAIN (RFC 2308 §2.1), which is reserved for names that do not
-	// exist at all.
-	anyNameExists := false
+func (d *dnsRecords) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
 	for _, q := range m.Question {
 		switch q.Qtype {
 		case dns.TypeA, dns.TypeAAAA:
 			qType := dns.TypeToString[q.Qtype]
-			if debugEnabled {
-				d.l.Debug("DNS query", "type", qType, "name", q.Name)
-			}
-			ip, nameExists := d.Query(q.Qtype, q.Name)
-			if nameExists {
-				anyNameExists = true
-			}
+			d.l.Debugf("Query for %s %s", qType, q.Name)
+			ip := d.Query(q.Qtype, q.Name)
 			if ip.IsValid() {
 				rr, err := dns.NewRR(fmt.Sprintf("%s %s %s", q.Name, qType, ip))
 				if err == nil {
@@ -342,9 +134,7 @@ func (d *dnsServer) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
 			if !d.isSelfNebulaOrLocalhost(w.RemoteAddr().String()) {
 				return
 			}
-			if debugEnabled {
-				d.l.Debug("DNS query", "type", "TXT", "name", q.Name)
-			}
+			d.l.Debugf("Query for TXT %s", q.Name)
 			ip := d.QueryCert(q.Name)
 			if ip != "" {
 				rr, err := dns.NewRR(fmt.Sprintf("%s TXT %s", q.Name, ip))
@@ -355,12 +145,12 @@ func (d *dnsServer) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
 		}
 	}
 
-	if len(m.Answer) == 0 && !anyNameExists {
+	if len(m.Answer) == 0 {
 		m.Rcode = dns.RcodeNameError
 	}
 }
 
-func (d *dnsServer) handleDnsRequest(w dns.ResponseWriter, r *dns.Msg) {
+func (d *dnsRecords) handleDnsRequest(w dns.ResponseWriter, r *dns.Msg) {
 	m := new(dns.Msg)
 	m.SetReply(r)
 	m.Compress = false
@@ -373,6 +163,21 @@ func (d *dnsServer) handleDnsRequest(w dns.ResponseWriter, r *dns.Msg) {
 	w.WriteMsg(m)
 }
 
+func dnsMain(l *logrus.Logger, cs *CertState, hostMap *HostMap, c *config.C) func() {
+	dnsR = newDnsRecords(l, cs, hostMap)
+
+	// attach request handler func
+	dns.HandleFunc(".", dnsR.handleDnsRequest)
+
+	c.RegisterReloadCallback(func(c *config.C) {
+		reloadDns(l, c)
+	})
+
+	return func() {
+		startDns(l, c)
+	}
+}
+
 func getDnsServerAddr(c *config.C) string {
 	dnsHost := strings.TrimSpace(c.GetString("lighthouse.dns.host", ""))
 	// Old guidance was to provide the literal `[::]` in `lighthouse.dns.host` but that won't resolve.
@@ -381,3 +186,25 @@ func getDnsServerAddr(c *config.C) string {
 	}
 	return net.JoinHostPort(dnsHost, strconv.Itoa(c.GetInt("lighthouse.dns.port", 53)))
 }
+
+func startDns(l *logrus.Logger, c *config.C) {
+	dnsAddr = getDnsServerAddr(c)
+	dnsServer = &dns.Server{Addr: dnsAddr, Net: "udp"}
+	l.WithField("dnsListener", dnsAddr).Info("Starting DNS responder")
+	err := dnsServer.ListenAndServe()
+	defer dnsServer.Shutdown()
+	if err != nil {
+		l.Errorf("Failed to start server: %s\n ", err.Error())
+	}
+}
+
+func reloadDns(l *logrus.Logger, c *config.C) {
+	if dnsAddr == getDnsServerAddr(c) {
+		l.Debug("No DNS server config change detected")
+		return
+	}
+
+	l.Debug("Restarting DNS server")
+	dnsServer.Shutdown()
+	go startDns(l, c)
+}

dns_server_test.go

@@ -1,43 +1,19 @@
 package nebula
 
 import (
-	"context"
-	"log/slog"
-	"net"
 	"net/netip"
-	"strconv"
 	"testing"
-	"time"
 
 	"github.com/miekg/dns"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
 	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
 )
 
-type stubDNSWriter struct{}
-
-func (stubDNSWriter) LocalAddr() net.Addr { return &net.UDPAddr{} }
-func (stubDNSWriter) RemoteAddr() net.Addr {
-	return &net.UDPAddr{IP: net.ParseIP("127.0.0.1"), Port: 5353}
-}
-func (stubDNSWriter) Write([]byte) (int, error) { return 0, nil }
-func (stubDNSWriter) WriteMsg(*dns.Msg) error   { return nil }
-func (stubDNSWriter) Close() error              { return nil }
-func (stubDNSWriter) TsigStatus() error         { return nil }
-func (stubDNSWriter) TsigTimersOnly(bool)       {}
-func (stubDNSWriter) Hijack()                   {}
-
 func TestParsequery(t *testing.T) {
-	l := slog.New(slog.DiscardHandler)
+	l := logrus.New()
 	hostMap := &HostMap{}
-	ds := &dnsServer{
-		l:       l,
-		dnsMap4: make(map[string]netip.Addr),
-		dnsMap6: make(map[string]netip.Addr),
-		hostMap: hostMap,
-	}
-	ds.enabled.Store(true)
+	ds := newDnsRecords(l, &CertState{}, hostMap)
 	addrs := []netip.Addr{
 		netip.MustParseAddr("1.2.3.4"),
 		netip.MustParseAddr("1.2.3.5"),
@@ -45,56 +21,18 @@ func TestParsequery(t *testing.T) {
 		netip.MustParseAddr("fd01::25"),
 	}
 	ds.Add("test.com.com", addrs)
-	ds.Add("v4only.com.com", []netip.Addr{netip.MustParseAddr("1.2.3.6")})
-	ds.Add("v6only.com.com", []netip.Addr{netip.MustParseAddr("fd01::26")})
 
 	m := &dns.Msg{}
 	m.SetQuestion("test.com.com", dns.TypeA)
 	ds.parseQuery(m, nil)
 	assert.NotNil(t, m.Answer)
 	assert.Equal(t, "1.2.3.4", m.Answer[0].(*dns.A).A.String())
-	assert.Equal(t, dns.RcodeSuccess, m.Rcode)
 
 	m = &dns.Msg{}
 	m.SetQuestion("test.com.com", dns.TypeAAAA)
 	ds.parseQuery(m, nil)
 	assert.NotNil(t, m.Answer)
 	assert.Equal(t, "fd01::24", m.Answer[0].(*dns.AAAA).AAAA.String())
-	assert.Equal(t, dns.RcodeSuccess, m.Rcode)
-
-	// A known name with no record of the requested type should return NODATA
-	// (NOERROR with empty answer), not NXDOMAIN.
-	m = &dns.Msg{}
-	m.SetQuestion("v4only.com.com", dns.TypeAAAA)
-	ds.parseQuery(m, nil)
-	assert.Empty(t, m.Answer)
-	assert.Equal(t, dns.RcodeSuccess, m.Rcode)
-
-	m = &dns.Msg{}
-	m.SetQuestion("v6only.com.com", dns.TypeA)
-	ds.parseQuery(m, nil)
-	assert.Empty(t, m.Answer)
-	assert.Equal(t, dns.RcodeSuccess, m.Rcode)
-
-	// An unknown name should still return NXDOMAIN.
-	m = &dns.Msg{}
-	m.SetQuestion("unknown.com.com", dns.TypeA)
-	ds.parseQuery(m, nil)
-	assert.Empty(t, m.Answer)
-	assert.Equal(t, dns.RcodeNameError, m.Rcode)
-
-	// short lookups should not fail
-	m = &dns.Msg{}
-	m.Question = []dns.Question{{Name: "", Qtype: dns.TypeTXT, Qclass: dns.ClassINET}}
-	ds.parseQuery(m, stubDNSWriter{})
-	assert.Empty(t, m.Answer)
-	assert.Equal(t, dns.RcodeNameError, m.Rcode)
-
-	m = &dns.Msg{}
-	m.Question = []dns.Question{{Name: ".", Qtype: dns.TypeTXT, Qclass: dns.ClassINET}}
-	ds.parseQuery(m, stubDNSWriter{})
-	assert.Empty(t, m.Answer)
-	assert.Equal(t, dns.RcodeNameError, m.Rcode)
 }
 
 func Test_getDnsServerAddr(t *testing.T) {
@@ -133,208 +71,3 @@ func Test_getDnsServerAddr(t *testing.T) {
 	}
 	assert.Equal(t, "[::]:1", getDnsServerAddr(c))
 }
-
-func newTestDnsServer(t *testing.T) (*dnsServer, *config.C) {
-	t.Helper()
-	sl := slog.New(slog.DiscardHandler)
-	ds := &dnsServer{
-		l:       sl,
-		ctx:     context.Background(),
-		dnsMap4: make(map[string]netip.Addr),
-		dnsMap6: make(map[string]netip.Addr),
-		hostMap: &HostMap{},
-	}
-	ds.mux = dns.NewServeMux()
-	ds.mux.HandleFunc(".", ds.handleDnsRequest)
-	return ds, config.NewC(nil)
-}
-
-func setDnsConfig(c *config.C, host string, port string, amLighthouse, serveDns bool) {
-	c.Settings["lighthouse"] = map[string]any{
-		"am_lighthouse": amLighthouse,
-		"serve_dns":     serveDns,
-		"dns": map[string]any{
-			"host": host,
-			"port": port,
-		},
-	}
-}
-
-func TestDnsServer_reload_initial_disabled(t *testing.T) {
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", "0", true, false)
-
-	require.NoError(t, ds.reload(c, true))
-	assert.False(t, ds.enabled.Load())
-	assert.Equal(t, "127.0.0.1:0", ds.addr)
-	assert.Nil(t, ds.server)
-}
-
-func TestDnsServer_reload_initial_enabled(t *testing.T) {
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", "0", true, true)
-
-	require.NoError(t, ds.reload(c, true))
-	assert.True(t, ds.enabled.Load())
-	assert.Equal(t, "127.0.0.1:0", ds.addr)
-	// initial never starts a runner; that's Control.Start's job
-	assert.Nil(t, ds.server)
-}
-
-func TestDnsServer_reload_initial_serveDnsWithoutLighthouse(t *testing.T) {
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", "0", false, true)
-
-	require.NoError(t, ds.reload(c, true))
-	// Wants DNS but isn't a lighthouse: gated off, no runner.
-	assert.False(t, ds.enabled.Load())
-}
-
-func TestDnsServer_reload_sameAddr_noOp(t *testing.T) {
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", "0", true, true)
-
-	require.NoError(t, ds.reload(c, true))
-	// No server running yet, no addr change. Reload should not spawn anything.
-	require.NoError(t, ds.reload(c, false))
-	assert.True(t, ds.enabled.Load())
-	assert.Nil(t, ds.server)
-}
-
-func TestDnsServer_StartStop_lifecycle(t *testing.T) {
-	// Bind to a real (random) UDP port so we exercise the actual
-	// ListenAndServe + Shutdown plumbing including the started-chan race fix.
-	port := freeUDPPort(t)
-
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", port, true, true)
-	require.NoError(t, ds.reload(c, true))
-
-	done := make(chan struct{})
-	go func() {
-		ds.Start()
-		close(done)
-	}()
-
-	waitFor(t, func() bool {
-		ds.serverMu.Lock()
-		started := ds.started
-		ds.serverMu.Unlock()
-		if started == nil {
-			return false
-		}
-		select {
-		case <-started:
-			return true
-		default:
-			return false
-		}
-	})
-
-	ds.Stop()
-	select {
-	case <-done:
-	case <-time.After(5 * time.Second):
-		t.Fatal("Start did not return after Stop")
-	}
-}
-
-func TestDnsServer_Stop_beforeBind_doesNotHang(t *testing.T) {
-	// Stop called immediately after Start should not deadlock even if bind
-	// hasn't completed yet. This exercises the started-chan close-on-bind-fail
-	// path: by binding to an obviously bad port (privileged) we get a fast
-	// bind error before NotifyStartedFunc fires.
-	ds, c := newTestDnsServer(t)
-	// Use a port that should fail to bind (negative would be invalid, use a
-	// host that won't resolve to ensure listenUDP fails quickly).
-	setDnsConfig(c, "256.256.256.256", "53", true, true)
-	require.NoError(t, ds.reload(c, true))
-
-	done := make(chan struct{})
-	go func() {
-		ds.Start()
-		close(done)
-	}()
-
-	// Give Start a moment to attempt the bind and fail.
-	select {
-	case <-done:
-		// Bind failed and Start returned; Stop should be a no-op.
-	case <-time.After(time.Second):
-		t.Fatal("Start did not return after a bad bind")
-	}
-
-	stopped := make(chan struct{})
-	go func() {
-		ds.Stop()
-		close(stopped)
-	}()
-	select {
-	case <-stopped:
-	case <-time.After(time.Second):
-		t.Fatal("Stop hung after a failed bind")
-	}
-}
-
-func TestDnsServer_reload_disable_stopsRunningServer(t *testing.T) {
-	port := freeUDPPort(t)
-	ds, c := newTestDnsServer(t)
-	setDnsConfig(c, "127.0.0.1", port, true, true)
-	require.NoError(t, ds.reload(c, true))
-
-	startReturned := make(chan struct{})
-	go func() {
-		ds.Start()
-		close(startReturned)
-	}()
-	waitForBind(t, ds)
-
-	// Toggle serve_dns off; reload should shut the running server down.
-	setDnsConfig(c, "127.0.0.1", port, true, false)
-	require.NoError(t, ds.reload(c, false))
-	select {
-	case <-startReturned:
-	case <-time.After(5 * time.Second):
-		t.Fatal("Start did not return after reload disabled DNS")
-	}
-	assert.False(t, ds.enabled.Load())
-}
-
-func freeUDPPort(t *testing.T) string {
-	t.Helper()
-	conn, err := net.ListenPacket("udp", "127.0.0.1:0")
-	require.NoError(t, err)
-	port := conn.LocalAddr().(*net.UDPAddr).Port
-	require.NoError(t, conn.Close())
-	return strconv.Itoa(port)
-}
-
-func waitForBind(t *testing.T, ds *dnsServer) {
-	t.Helper()
-	waitFor(t, func() bool {
-		ds.serverMu.Lock()
-		started := ds.started
-		ds.serverMu.Unlock()
-		if started == nil {
-			return false
-		}
-		select {
-		case <-started:
-			return true
-		default:
-			return false
-		}
-	})
-}
-
-func waitFor(t *testing.T, cond func() bool) {
-	t.Helper()
-	deadline := time.Now().Add(5 * time.Second)
-	for time.Now().Before(deadline) {
-		if cond() {
-			return
-		}
-		time.Sleep(5 * time.Millisecond)
-	}
-	t.Fatal("timed out waiting for condition")
-}

e2e/handshake_manager_test.go

@@ -1,565 +0,0 @@
-//go:build e2e_testing
-// +build e2e_testing
-
-package e2e
-
-import (
-	"net/netip"
-	"testing"
-	"time"
-
-	"github.com/slackhq/nebula"
-	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/cert_test"
-	"github.com/slackhq/nebula/e2e/router"
-	"github.com/slackhq/nebula/header"
-	"github.com/slackhq/nebula/udp"
-	"github.com/stretchr/testify/assert"
-)
-
-// makeHandshakePacket creates a handshake packet with the given parameters.
-func makeHandshakePacket(from, to netip.AddrPort, subtype header.MessageSubType, remoteIndex uint32, counter uint64) *udp.Packet {
-	data := make([]byte, 200)
-	header.Encode(data, header.Version, header.Handshake, subtype, remoteIndex, counter)
-	for i := header.Len; i < len(data); i++ {
-		data[i] = byte(i)
-	}
-	return &udp.Packet{To: to, From: from, Data: data}
-}
-
-func TestHandshakeRetransmitDuplicate(t *testing.T) {
-	// Verify the responder correctly handles receiving the same msg1 multiple times
-	// (retransmission). The duplicate goes through CheckAndComplete -> ErrAlreadySeen
-	// and the cached response is resent.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	t.Log("Trigger handshake from me to them")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-
-	t.Log("Grab my msg1")
-	msg1 := myControl.GetFromUDP(true)
-
-	t.Log("Inject msg1 into them, first time")
-	theirControl.InjectUDPPacket(msg1)
-	_ = theirControl.GetFromUDP(true)
-
-	t.Log("Inject the SAME msg1 again, tests ErrAlreadySeen path")
-	theirControl.InjectUDPPacket(msg1)
-	resp2 := theirControl.GetFromUDP(true)
-	assert.NotNil(t, resp2, "should get cached response on duplicate msg1")
-
-	t.Log("Complete handshake with cached response")
-	myControl.InjectUDPPacket(resp2)
-	myControl.WaitForType(1, 0, theirControl)
-
-	t.Log("Drain cached packet and verify tunnel works")
-	cachedPacket := theirControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi"), cachedPacket, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), 80, 80)
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	t.Log("Verify only one tunnel exists on each side")
-	assert.Len(t, myControl.ListHostmapHosts(false), 1)
-	assert.Len(t, theirControl.ListHostmapHosts(false), 1)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeTruncatedPacketRecovery(t *testing.T) {
-	// Verify that a truncated handshake packet is ignored and the real
-	// packet can still complete the handshake.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	t.Log("Trigger handshake")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-
-	t.Log("Get msg1 and deliver to responder")
-	msg1 := myControl.GetFromUDP(true)
-	theirControl.InjectUDPPacket(msg1)
-
-	t.Log("Get the real response")
-	realResp := theirControl.GetFromUDP(true)
-
-	t.Log("Truncate the response and inject, should be ignored")
-	truncResp := realResp.Copy()
-	truncResp.Data = truncResp.Data[:header.Len]
-	myControl.InjectUDPPacket(truncResp)
-
-	t.Log("Verify pending handshake survived the truncated packet")
-	assert.NotEmpty(t, myControl.ListHostmapHosts(true), "pending handshake should still exist")
-
-	t.Log("Inject real response, should complete handshake")
-	myControl.InjectUDPPacket(realResp)
-	myControl.WaitForType(1, 0, theirControl)
-
-	t.Log("Drain and verify tunnel")
-	cachedPacket := theirControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi"), cachedPacket, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), 80, 80)
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeOrphanedMsg2Dropped(t *testing.T) {
-	// A msg2 arriving with no matching pending index should be silently dropped
-	// with no response sent and no state changes.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	t.Log("Complete a normal handshake")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-	r.RouteForAllUntilTxTun(theirControl)
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	t.Log("Record hostmap state")
-	myIndexes := len(myControl.ListHostmapIndexes(false))
-
-	t.Log("Inject a fake msg2 with unknown RemoteIndex")
-	myControl.InjectUDPPacket(makeHandshakePacket(theirUdpAddr, myUdpAddr, header.HandshakeIXPSK0, 0xDEADBEEF, 2))
-
-	t.Log("Verify no new indexes created")
-	assert.Equal(t, myIndexes, len(myControl.ListHostmapIndexes(false)))
-
-	t.Log("Verify no UDP response was sent")
-	time.Sleep(100 * time.Millisecond)
-	assert.Nil(t, myControl.GetFromUDP(false), "should not send a response to orphaned msg2")
-
-	t.Log("Verify existing tunnel still works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeUnknownMessageCounter(t *testing.T) {
-	// A handshake packet with an unexpected message counter should be silently
-	// dropped with no side effects and no UDP response.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, _, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	t.Log("Inject handshake with MessageCounter=3")
-	myControl.InjectUDPPacket(makeHandshakePacket(theirUdpAddr, myUdpAddr, header.HandshakeIXPSK0, 0, 3))
-
-	t.Log("Inject handshake with MessageCounter=99")
-	myControl.InjectUDPPacket(makeHandshakePacket(theirUdpAddr, myUdpAddr, header.HandshakeIXPSK0, 0, 99))
-
-	t.Log("Verify no tunnels or pending handshakes")
-	assert.Empty(t, myControl.ListHostmapHosts(false))
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-
-	t.Log("Verify no UDP response was sent")
-	time.Sleep(100 * time.Millisecond)
-	assert.Nil(t, myControl.GetFromUDP(false))
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeUnknownSubtype(t *testing.T) {
-	// A handshake packet with an unknown subtype should be silently dropped.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, _, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, _, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.Start()
-	theirControl.Start()
-
-	t.Log("Inject handshake with unknown subtype 99")
-	myControl.InjectUDPPacket(makeHandshakePacket(theirUdpAddr, myUdpAddr, header.MessageSubType(99), 0, 1))
-
-	t.Log("Verify no tunnels or pending handshakes")
-	assert.Empty(t, myControl.ListHostmapHosts(false))
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-
-	t.Log("Verify no UDP response was sent")
-	time.Sleep(100 * time.Millisecond)
-	assert.Nil(t, myControl.GetFromUDP(false))
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeLateResponse(t *testing.T) {
-	// After a handshake times out, a late response should be silently ignored
-	// with no new tunnels created.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, _, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{
-		"handshakes": m{
-			"try_interval": "200ms",
-			"retries":      2,
-		},
-	})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	t.Log("Trigger handshake from me")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-
-	t.Log("Grab msg1 but don't deliver")
-	msg1 := myControl.GetFromUDP(true)
-
-	t.Log("Wait for handshake to time out")
-	for i := 0; i < 5; i++ {
-		time.Sleep(300 * time.Millisecond)
-		myControl.GetFromUDP(false)
-	}
-
-	t.Log("Confirm no pending handshakes remain")
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-
-	t.Log("Deliver old msg1 to them, they create a tunnel")
-	theirControl.InjectUDPPacket(msg1)
-	resp := theirControl.GetFromUDP(true)
-	assert.NotNil(t, resp)
-
-	t.Log("Inject late response into me, should be ignored")
-	myControl.InjectUDPPacket(resp)
-
-	t.Log("No tunnel should exist on my side")
-	assert.Empty(t, myControl.ListHostmapHosts(false))
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeSelfConnectionRejected(t *testing.T) {
-	// Verify that a node rejects a handshake containing its own VPN IP in the
-	// peer cert. We do this by sending the initiator's own msg1 back to itself.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-
-	// Need a lighthouse entry to trigger a handshake
-	myControl.InjectLightHouseAddr(netip.MustParseAddr("10.128.0.2"), netip.MustParseAddrPort("10.0.0.2:4242"))
-
-	myControl.Start()
-
-	t.Log("Trigger handshake from me")
-	myControl.InjectTunUDPPacket(netip.MustParseAddr("10.128.0.2"), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-	msg1 := myControl.GetFromUDP(true)
-
-	t.Log("Drain any handshake retransmits before injecting")
-	time.Sleep(100 * time.Millisecond)
-	for myControl.GetFromUDP(false) != nil {
-	}
-
-	t.Log("Feed my own msg1 back to me as if it came from someone else")
-	selfMsg := msg1.Copy()
-	selfMsg.From = netip.MustParseAddrPort("10.0.0.99:4242")
-	selfMsg.To = myUdpAddr
-	myControl.InjectUDPPacket(selfMsg)
-
-	t.Log("Verify no response was sent (self-connection rejected)")
-	time.Sleep(100 * time.Millisecond)
-	// Drain any further retransmits from the original handshake, then check
-	// that none of them are a handshake response (MessageCounter=2)
-	h := &header.H{}
-	for {
-		p := myControl.GetFromUDP(false)
-		if p == nil {
-			break
-		}
-		_ = h.Parse(p.Data)
-		assert.NotEqual(t, uint64(2), h.MessageCounter,
-			"should not send a stage 2 response to self-connection")
-	}
-
-	t.Log("Verify no tunnel to myself was created")
-	assert.Nil(t, myControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false))
-
-	myControl.Stop()
-}
-
-func TestHandshakeMessageCounter0Dropped(t *testing.T) {
-	// MessageCounter=0 is not a valid handshake message and should be dropped.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, _, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	_, _, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.Start()
-
-	t.Log("Inject handshake with MessageCounter=0")
-	myControl.InjectUDPPacket(makeHandshakePacket(theirUdpAddr, myUdpAddr, header.HandshakeIXPSK0, 0, 0))
-
-	time.Sleep(100 * time.Millisecond)
-	assert.Empty(t, myControl.ListHostmapHosts(false))
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-	assert.Nil(t, myControl.GetFromUDP(false))
-
-	myControl.Stop()
-}
-
-func TestHandshakeRemoteAllowList(t *testing.T) {
-	// Verify that a handshake from a blocked underlay IP is dropped with no
-	// response and no state changes. Then verify the same packet from an
-	// allowed IP succeeds.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{
-		"lighthouse": m{
-			"remote_allow_list": m{
-				"10.0.0.0/8": true,
-				"0.0.0.0/0":  false,
-			},
-		},
-	})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	t.Log("Trigger handshake from them")
-	theirControl.InjectTunUDPPacket(myVpnIpNet[0].Addr(), 80, theirVpnIpNet[0].Addr(), 80, []byte("Hi"))
-	msg1 := theirControl.GetFromUDP(true)
-
-	t.Log("Rewrite the source to a blocked IP and inject")
-	blockedMsg := msg1.Copy()
-	blockedMsg.From = netip.MustParseAddrPort("192.168.1.1:4242")
-	myControl.InjectUDPPacket(blockedMsg)
-
-	t.Log("Verify no tunnel, no pending, no response from blocked source")
-	time.Sleep(100 * time.Millisecond)
-	assert.Empty(t, myControl.ListHostmapHosts(false))
-	assert.Empty(t, myControl.ListHostmapHosts(true))
-	assert.Nil(t, myControl.GetFromUDP(false), "should not respond to blocked source")
-
-	t.Log("Now inject the real packet from the allowed source")
-	myControl.InjectUDPPacket(msg1)
-
-	t.Log("Verify handshake completes from allowed source")
-	resp := myControl.GetFromUDP(true)
-	assert.NotNil(t, resp)
-	theirControl.InjectUDPPacket(resp)
-	theirControl.WaitForType(1, 0, myControl)
-
-	t.Log("Drain cached packet and verify tunnel works")
-	cachedPacket := myControl.GetFromTun(true)
-	assertUdpPacket(t, []byte("Hi"), cachedPacket, theirVpnIpNet[0].Addr(), myVpnIpNet[0].Addr(), 80, 80)
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeAlreadySeenPreferredRemote(t *testing.T) {
-	// When a duplicate msg1 arrives via ErrAlreadySeen, verify the tunnel
-	// remains functional and hostmap index count is stable.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	t.Log("Complete a normal handshake via the router")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi"))
-	r.RouteForAllUntilTxTun(theirControl)
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	t.Log("Record hostmap state")
-	theirIndexes := len(theirControl.ListHostmapIndexes(false))
-	hi := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-	assert.NotNil(t, hi)
-	originalRemote := hi.CurrentRemote
-
-	t.Log("Re-trigger traffic to cause a new handshake attempt (ErrAlreadySeen)")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("roam"))
-	r.RouteForAllUntilTxTun(theirControl)
-
-	t.Log("Verify tunnel still works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	t.Log("Verify remote is still valid and index count is stable")
-	hi2 := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-	assert.NotNil(t, hi2)
-	assert.Equal(t, originalRemote, hi2.CurrentRemote)
-	assert.Equal(t, theirIndexes, len(theirControl.ListHostmapIndexes(false)),
-		"no extra indexes should be created from ErrAlreadySeen")
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestHandshakeWrongResponderPacketStore(t *testing.T) {
-	// Verify that when the wrong host responds, the cached packets are
-	// transferred to the new handshake, the evil tunnel is closed, evil's
-	// address is blocked, and the correct tunnel is eventually established.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.100/24", nil)
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.99/24", nil)
-	evilControl, evilVpnIpNet, evilUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "evil", "10.128.0.2/24", nil)
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), evilUdpAddr)
-
-	r := router.NewR(t, myControl, theirControl, evilControl)
-	defer r.RenderFlow()
-
-	myControl.Start()
-	theirControl.Start()
-	evilControl.Start()
-
-	t.Log("Send multiple packets to them (cached during handshake)")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("packet1"))
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("packet2"))
-
-	t.Log("Route until evil tunnel is closed")
-	h := &header.H{}
-	r.RouteForAllExitFunc(func(p *udp.Packet, c *nebula.Control) router.ExitType {
-		if err := h.Parse(p.Data); err != nil {
-			panic(err)
-		}
-		if h.Type == header.CloseTunnel && p.To == evilUdpAddr {
-			return router.RouteAndExit
-		}
-		return router.KeepRouting
-	})
-
-	t.Log("Verify evil's address is blocked in the new pending handshake")
-	pendingHI := myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), true)
-	if pendingHI != nil {
-		assert.NotContains(t, pendingHI.RemoteAddrs, evilUdpAddr,
-			"evil's address should be blocked")
-	}
-
-	t.Log("Inject correct lighthouse addr for them")
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-
-	t.Log("Route until cached packets arrive at the real them")
-	p := r.RouteForAllUntilTxTun(theirControl)
-	assert.NotNil(t, p, "a cached packet should be delivered to the correct host")
-
-	t.Log("Verify the correct host has a tunnel")
-	assertHostInfoPair(t, myUdpAddr, theirUdpAddr, myVpnIpNet, theirVpnIpNet, myControl, theirControl)
-
-	t.Log("Verify no hostinfo artifacts from evil remain")
-	assert.Nil(t, myControl.GetHostInfoByVpnAddr(evilVpnIpNet[0].Addr(), true),
-		"no pending hostinfo for evil")
-	assert.Nil(t, myControl.GetHostInfoByVpnAddr(evilVpnIpNet[0].Addr(), false),
-		"no main hostinfo for evil")
-
-	myControl.Stop()
-	theirControl.Stop()
-	evilControl.Stop()
-}
-
-func TestHandshakeRelayComplete(t *testing.T) {
-	// Verify that a relay handshake completes correctly and relay state is
-	// properly maintained on all three nodes.
-
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, _, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{"relay": m{"use_relays": true}})
-	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "relay", "10.128.0.128/24", m{"relay": m{"am_relay": true}})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", m{"relay": m{"use_relays": true}})
-
-	myControl.InjectLightHouseAddr(relayVpnIpNet[0].Addr(), relayUdpAddr)
-	myControl.InjectRelays(theirVpnIpNet[0].Addr(), []netip.Addr{relayVpnIpNet[0].Addr()})
-	relayControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-
-	r := router.NewR(t, myControl, relayControl, theirControl)
-	defer r.RenderFlow()
-
-	myControl.Start()
-	relayControl.Start()
-	theirControl.Start()
-
-	t.Log("Trigger handshake via relay")
-	myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi via relay"))
-
-	p := r.RouteForAllUntilTxTun(theirControl)
-	assertUdpPacket(t, []byte("Hi via relay"), p, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), 80, 80)
-
-	t.Log("Verify bidirectional tunnel via relay")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	t.Log("Verify relay state on my side shows relay-to-me")
-	myHI := myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
-	assert.NotNil(t, myHI)
-	assert.NotEmpty(t, myHI.CurrentRelaysToMe, "should have relay-to-me for them")
-
-	t.Log("Verify relay state on their side shows relay-to-me")
-	theirHI := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-	assert.NotNil(t, theirHI)
-	assert.NotEmpty(t, theirHI.CurrentRelaysToMe, "should have relay-to-me for me")
-
-	t.Log("Verify relay node shows through-me relays")
-	relayHI := relayControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-	assert.NotNil(t, relayHI)
-
-	myControl.Stop()
-	relayControl.Stop()
-	theirControl.Stop()
-}
-
-// NOTE: Relay V1 cert + IPv6 rejection is not tested here because
-// InjectTunUDPPacket from a V4 node to a V6 address panics in the test
-// framework. The check is in handshake_manager.go handleOutbound relay
-// logic (lines ~304-313): if the relay host has a V1 cert and either
-// address is IPv6, the relay is skipped.
-
-// NOTE: Relay reestablishment (Disestablished state transition) is covered
-// by the existing TestReestablishRelays in handshakes_test.go.

e2e/handshakes_test.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/cert_test"
@@ -748,6 +749,7 @@ func TestStage1RaceRelays2(t *testing.T) {
 	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me     ", "10.128.0.1/24", m{"relay": m{"use_relays": true}})
 	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "relay  ", "10.128.0.128/24", m{"relay": m{"am_relay": true}})
 	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them   ", "10.128.0.2/24", 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[0].Addr(), relayUdpAddr)
@@ -769,41 +771,49 @@ func TestStage1RaceRelays2(t *testing.T) {
 	theirControl.Start()
 
 	r.Log("Get a tunnel between me and relay")
+	l.Info("Get a tunnel between me and relay")
 	assertTunnel(t, myVpnIpNet[0].Addr(), relayVpnIpNet[0].Addr(), myControl, relayControl, r)
 
 	r.Log("Get a tunnel between them and relay")
+	l.Info("Get a tunnel between them and relay")
 	assertTunnel(t, theirVpnIpNet[0].Addr(), relayVpnIpNet[0].Addr(), 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[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi from me"))
 	theirControl.InjectTunUDPPacket(myVpnIpNet[0].Addr(), 80, theirVpnIpNet[0].Addr(), 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; myControl")
+	r.Log("Wait for a packet from them to me")
+	l.Info("Wait for a packet from them to me; myControl")
 	r.RouteForAllUntilTxTun(myControl)
-	r.Log("Wait for a packet from them to me; theirControl")
+	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[0].Addr(), myVpnIpNet[0].Addr(), theirControl, myControl, r)
 
 	t.Log("Wait until we remove extra tunnels")
-	t.Logf("Waiting for hostinfos to be removed... myControl=%d theirControl=%d relayControl=%d",
-		len(myControl.GetHostmap().Indexes),
-		len(theirControl.GetHostmap().Indexes),
-		len(relayControl.GetHostmap().Indexes),
-	)
+	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)
-		t.Logf("Waiting for hostinfos to be removed... myControl=%d theirControl=%d relayControl=%d",
-			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[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
 		t.Log("Connection manager hasn't ticked yet")
 		time.Sleep(time.Second)
@@ -811,6 +821,7 @@ func TestStage1RaceRelays2(t *testing.T) {
 	}
 
 	r.Log("Assert the tunnel works")
+	l.Info("Assert the tunnel works")
 	assertTunnel(t, theirVpnIpNet[0].Addr(), myVpnIpNet[0].Addr(), theirControl, myControl, r)
 
 	myControl.Stop()
@@ -1358,81 +1369,6 @@ func TestV2NonPrimaryWithOffNetLighthouse(t *testing.T) {
 	theirControl.Stop()
 }
 
-func TestLighthouseUpdateOnReload(t *testing.T) {
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-
-	// Create the lighthouse
-	lhControl, lhVpnIpNet, lhUdpAddr, _ := newSimpleServer(cert.Version2, ca, caKey, "lh", "10.128.0.1/24", m{"lighthouse": m{"am_lighthouse": true}})
-
-	// Create a client with NO lighthouse configured and a long update interval.
-	// The initial SendUpdate at startup will be a no-op since no lighthouses are known.
-	myControl, myVpnIpNet, _, myConfig := newSimpleServer(cert.Version2, ca, caKey, "me", "10.128.0.2/24", m{
-		"lighthouse": m{
-			"interval": 600,
-			"local_allow_list": m{
-				"10.0.0.0/24": true,
-				"::/0":        false,
-			},
-		},
-	})
-
-	r := router.NewR(t, lhControl, myControl)
-	defer r.RenderFlow()
-
-	lhControl.Start()
-	myControl.Start()
-
-	// Drain any startup packets (there should be none meaningful)
-	r.FlushAll()
-
-	// Verify lighthouse has no knowledge of the client
-	assert.Nil(t, lhControl.QueryLighthouse(myVpnIpNet[0].Addr()))
-
-	// Build a new config that adds the lighthouse
-	newSettings := make(m)
-	for k, v := range myConfig.Settings {
-		newSettings[k] = v
-	}
-	newSettings["static_host_map"] = m{
-		lhVpnIpNet[0].Addr().String(): []any{lhUdpAddr.String()},
-	}
-	newSettings["lighthouse"] = m{
-		"hosts":    []any{lhVpnIpNet[0].Addr().String()},
-		"interval": 600,
-		"local_allow_list": m{
-			"10.0.0.0/24": true,
-			"::/0":        false,
-		},
-	}
-	newCfg, err := yaml.Marshal(newSettings)
-	require.NoError(t, err)
-
-	// Reload the config. The lighthouse.hosts change triggers TriggerUpdate,
-	// which wakes the update worker. It calls SendUpdate, initiating a
-	// handshake to the new lighthouse and caching the HostUpdateNotification.
-	require.NoError(t, myConfig.ReloadConfigString(string(newCfg)))
-
-	// Route until the lighthouse receives the HostUpdateNotification.
-	// This covers: handshake stage 1, stage 2, then the cached update.
-	done := make(chan struct{})
-	go func() {
-		r.RouteForAllUntilAfterMsgTypeTo(lhControl, header.LightHouse, 0)
-		close(done)
-	}()
-	select {
-	case <-done:
-	case <-time.After(5 * time.Second):
-		t.Fatal("timed out waiting for lighthouse update after config reload")
-	}
-
-	// Verify lighthouse now has the client's addresses
-	assert.NotNil(t, lhControl.QueryLighthouse(myVpnIpNet[0].Addr()))
-
-	r.RenderHostmaps("Final hostmaps", lhControl, myControl)
-	lhControl.Stop()
-	myControl.Stop()
-}
-
 func TestGoodHandshakeUnsafeDest(t *testing.T) {
 	unsafePrefix := "192.168.6.0/24"
 	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})

e2e/helpers_test.go

@@ -4,6 +4,7 @@
 package e2e
 
 import (
+	"fmt"
 	"io"
 	"net/netip"
 	"os"
@@ -11,18 +12,15 @@ import (
 	"testing"
 	"time"
 
-	"log/slog"
-
 	"dario.cat/mergo"
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
-
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/cert_test"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/e2e/router"
-	"github.com/slackhq/nebula/logging"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"go.yaml.in/yaml/v3"
@@ -134,7 +132,8 @@ func newSimpleServerWithUdpAndUnsafeNetworks(v cert.Version, caCrt cert.Certific
 			"port": udpAddr.Port(),
 		},
 		"logging": m{
-			"level": testLogLevelName(),
+			"timestamp_format": fmt.Sprintf("%v 15:04:05.000000", name),
+			"level":            l.Level.String(),
 		},
 		"timers": m{
 			"pending_deletion_interval": 2,
@@ -235,7 +234,8 @@ func newServer(caCrt []cert.Certificate, certs []cert.Certificate, key []byte, o
 			"port": udpAddr.Port(),
 		},
 		"logging": m{
-			"level": testLogLevelName(),
+			"timestamp_format": fmt.Sprintf("%v 15:04:05.000000", certs[0].Name()),
+			"level":            l.Level.String(),
 		},
 		"timers": m{
 			"pending_deletion_interval": 2,
@@ -379,32 +379,24 @@ func getAddrs(ns []netip.Prefix) []netip.Addr {
 	return a
 }
 
-func NewTestLogger() *slog.Logger {
+func NewTestLogger() *logrus.Logger {
+	l := logrus.New()
+
 	v := os.Getenv("TEST_LOGS")
 	if v == "" {
-		return slog.New(slog.NewTextHandler(io.Discard, nil))
+		l.SetOutput(io.Discard)
+		l.SetLevel(logrus.PanicLevel)
+		return l
 	}
 
-	level := slog.LevelInfo
 	switch v {
 	case "2":
-		level = slog.LevelDebug
+		l.SetLevel(logrus.DebugLevel)
 	case "3":
-		level = logging.LevelTrace
+		l.SetLevel(logrus.TraceLevel)
+	default:
+		l.SetLevel(logrus.InfoLevel)
 	}
-	return slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: level}))
-}
 
-// testLogLevelName returns the level name string accepted by logging.ApplyConfig
-// for the current TEST_LOGS setting. Kept in sync with NewTestLogger.
-func testLogLevelName() string {
-	switch os.Getenv("TEST_LOGS") {
-	case "2":
-		return "debug"
-	case "3":
-		return "trace"
-	case "":
-		return "info"
-	}
-	return "info"
+	return l
 }

e2e/tunnels_test.go

@@ -12,8 +12,6 @@ import (
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/cert_test"
 	"github.com/slackhq/nebula/e2e/router"
-	"github.com/slackhq/nebula/header"
-	"github.com/slackhq/nebula/udp"
 	"github.com/stretchr/testify/assert"
 	"gopkg.in/yaml.v3"
 )
@@ -367,106 +365,3 @@ func TestCrossStackRelaysWork(t *testing.T) {
 	//theirControl.Stop()
 	//relayControl.Stop()
 }
-
-func TestCloseTunnelAuthenticated(t *testing.T) {
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "5s"}})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "10m"}})
-
-	// Share our underlay information
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	// Start the servers
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-
-	r.Log("Assert the tunnel between me and them works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	r.Log("Close the tunnel")
-	myControl.CloseTunnel(theirVpnIpNet[0].Addr(), false)
-	r.FlushAll()
-
-	waitStart := time.Now()
-	for {
-		myIndexes := len(myControl.GetHostmap().Indexes)
-		theirIndexes := len(theirControl.GetHostmap().Indexes)
-		if myIndexes == 0 && theirIndexes == 0 {
-			break
-		}
-
-		since := time.Since(waitStart)
-		r.Logf("my tunnels: %v; their tunnels: %v; duration: %v", myIndexes, theirIndexes, since)
-		if since > time.Second*6 {
-			t.Fatal("Tunnel should have been declared inactive after 2 seconds and before 6 seconds")
-		}
-
-		time.Sleep(1 * time.Second)
-		//r.FlushAll()
-	}
-
-	r.Logf("Happy path success, tunnels were dropped within %v", time.Since(waitStart))
-
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-	r.Log("Assert another tunnel between me and them works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-	hi := myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
-	if hi == nil {
-		t.Fatal("There is no hostinfo for this tunnel")
-	}
-	myHi := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-	if myHi == nil {
-		t.Fatal("There is no hostinfo for my tunnel")
-	}
-	r.Log("It does")
-
-	buf := make([]byte, 1024)
-	hdr := header.H{
-		Version:        1,
-		Type:           header.CloseTunnel,
-		Subtype:        0,
-		Reserved:       0,
-		RemoteIndex:    hi.RemoteIndex,
-		MessageCounter: 5,
-	}
-	out, err := hdr.Encode(buf)
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	pkt := &udp.Packet{
-		To:   hi.CurrentRemote,
-		From: myHi.CurrentRemote,
-		Data: out,
-	}
-	r.InjectUDPPacket(myControl, theirControl, pkt)
-	r.Log("Injected bogus close tunnel. Let's see!")
-	waitStart = time.Now()
-	for {
-		myIndexes := len(myControl.GetHostmap().Indexes)
-		theirIndexes := len(theirControl.GetHostmap().Indexes)
-		if myIndexes == 0 {
-			t.Fatal("myIndexes should not be 0")
-		}
-		if theirIndexes == 0 {
-			t.Fatal("theirIndexes should not be 0, they should have rejected this bogus packet")
-		}
-
-		since := time.Since(waitStart)
-		r.Logf("my tunnels: %v; their tunnels: %v; duration: %v", myIndexes, theirIndexes, since)
-		if since > time.Second*4 {
-			t.Log("The tunnel would have been gone by now")
-			break
-		}
-
-		time.Sleep(1 * time.Second)
-		r.FlushAll()
-	}
-
-	myControl.Stop()
-	theirControl.Stop()
-}

examples/config.yml

@@ -144,10 +144,6 @@ listen:
   # valid values: always, never, private
   # This setting is reloadable.
   #send_recv_error: always
-  # Similar to send_recv_error, this option lets you configure if you want to accept "recv_error" packets from remote hosts.
-  # valid values: always, never, private
-  # This setting is reloadable.
-  #accept_recv_error: always
   # The so_sock option is a Linux-specific feature that allows all outgoing Nebula packets to be tagged with a specific identifier.
   # This tagging enables IP rule-based filtering. For example, it supports 0.0.0.0/0 unsafe_routes,
   # allowing for more precise routing decisions based on the packet tags. Default is 0 meaning no mark is set.
@@ -204,12 +200,6 @@ punchy:
   # Trusted SSH CA public keys. These are the public keys of the CAs that are allowed to sign SSH keys for access.
   #trusted_cas:
     #- "ssh public key string"
-  # sandbox_dir restricts file paths for profiling commands (start-cpu-profile, save-heap-profile,
-  # save-mutex-profile) to the specified directory. Relative paths will be resolved within this directory,
-  # and absolute paths outside of it will be rejected. Default is $TMP/nebula-debug.
-  # The directory is NOT automatically created.
-  # Overriding this to "" is the same as "/" and will allow overwriting any path on the host.
-  #sandbox_dir: /var/tmp/nebula-debug
 
 # EXPERIMENTAL: relay support for networks that can't establish direct connections.
 relay:
@@ -292,21 +282,24 @@ tun:
 
 # Configure logging level
 logging:
-  # trace, debug, info, warn, or error. Default is info and is reloadable.
-  # fatal and panic are accepted for backwards compatibility and map to error.
-  #NOTE: Debug and trace modes can log remotely controlled/untrusted data which can quickly fill a disk in some
-  # scenarios. Debug and trace logging are also CPU intensive and will decrease performance overall.
-  # Only enable debug or trace logging while actively investigating an issue.
+  # panic, fatal, error, warning, info, or debug. Default is info and is reloadable.
+  #NOTE: Debug mode can log remotely controlled/untrusted data which can quickly fill a disk in some
+  # scenarios. Debug logging is also CPU intensive and will decrease performance overall.
+  # Only enable debug logging while actively investigating an issue.
   level: info
-  # json or text formats currently available. Default is text.
+  # json or text formats currently available. Default is text
   format: text
-  # Disable timestamp logging. Useful when output is redirected to a logging system that already adds timestamps. Default is false.
+  # Disable timestamp logging. useful when output is redirected to logging system that already adds timestamps. Default is false
   #disable_timestamp: true
-  # Timestamps use RFC3339Nano ("2006-01-02T15:04:05.999999999Z07:00") and are not configurable.
+  # timestamp format is specified in Go time format, see:
+  #     https://golang.org/pkg/time/#pkg-constants
+  # default when `format: json`: "2006-01-02T15:04:05Z07:00" (RFC3339)
+  # default when `format: text`:
+  #     when TTY attached: seconds since beginning of execution
+  #     otherwise: "2006-01-02T15:04:05Z07:00" (RFC3339)
+  # As an example, to log as RFC3339 with millisecond precision, set to:
+  #timestamp_format: "2006-01-02T15:04:05.000Z07:00"
 
-# The stats section is reloadable. A HUP may change the backend, toggle stats
-# on or off, switch the listen/host address, or pick up new DNS for the
-# configured graphite host.
 #stats:
   #type: graphite
   #prefix: nebula
@@ -324,12 +317,10 @@ logging:
   # enables counter metrics for meta packets
   #   e.g.: `messages.tx.handshake`
   # NOTE: `message.{tx,rx}.recv_error` is always emitted
-  # Not reloadable.
   #message_metrics: false
 
   # enables detailed counter metrics for lighthouse packets
   #   e.g.: `lighthouse.rx.HostQuery`
-  # Not reloadable.
   #lighthouse_metrics: false
 
 # Handshake Manager Settings
@@ -387,8 +378,8 @@ firewall:
   # Rules are comprised of a protocol, port, and one or more of host, group, or CIDR
   # Logical evaluation is roughly: port AND proto AND (ca_sha OR ca_name) AND (host OR group OR groups OR cidr) AND (local cidr)
   # - port: Takes `0` or `any` as any, a single number `80`, a range `200-901`, or `fragment` to match second and further fragments of fragmented packets (since there is no port available).
+  #   code: same as port but makes more sense when talking about ICMP, TODO: this is not currently implemented in a way that works, use `any`
   #   proto: `any`, `tcp`, `udp`, or `icmp`
-  #          a port specification is ignored if proto is `icmp`
   #   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

examples/go_service/main.go

@@ -7,9 +7,9 @@ import (
 	"net"
 	"os"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/logging"
 	"github.com/slackhq/nebula/overlay"
 	"github.com/slackhq/nebula/service"
 )
@@ -64,7 +64,8 @@ pki:
 		return err
 	}
 
-	logger := logging.NewLogger(os.Stdout)
+	logger := logrus.New()
+	logger.Out = os.Stdout
 
 	ctrl, err := nebula.Main(&cfg, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
 	if err != nil {

firewall.go

@@ -1,13 +1,11 @@
 package nebula
 
 import (
-	"context"
 	"crypto/sha256"
 	"encoding/hex"
 	"errors"
 	"fmt"
 	"hash/fnv"
-	"log/slog"
 	"net/netip"
 	"reflect"
 	"slices"
@@ -18,6 +16,7 @@ import (
 
 	"github.com/gaissmai/bart"
 	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/firewall"
@@ -68,7 +67,7 @@ type Firewall struct {
 	incomingMetrics     firewallMetrics
 	outgoingMetrics     firewallMetrics
 
-	l *slog.Logger
+	l *logrus.Logger
 }
 
 type firewallMetrics struct {
@@ -132,7 +131,7 @@ type firewallLocalCIDR struct {
 
 // NewFirewall creates a new Firewall object. A TimerWheel is created for you from the provided timeouts.
 // The certificate provided should be the highest version loaded in memory.
-func NewFirewall(l *slog.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.Duration, c cert.Certificate) *Firewall {
+func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.Duration, c cert.Certificate) *Firewall {
 	//TODO: error on 0 duration
 	var tmin, tmax time.Duration
 
@@ -192,7 +191,7 @@ func NewFirewall(l *slog.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.Dur
 	}
 }
 
-func NewFirewallFromConfig(l *slog.Logger, cs *CertState, c *config.C) (*Firewall, error) {
+func NewFirewallFromConfig(l *logrus.Logger, cs *CertState, c *config.C) (*Firewall, error) {
 	certificate := cs.getCertificate(cert.Version2)
 	if certificate == nil {
 		certificate = cs.getCertificate(cert.Version1)
@@ -220,7 +219,7 @@ func NewFirewallFromConfig(l *slog.Logger, cs *CertState, c *config.C) (*Firewal
 	case "drop":
 		fw.InSendReject = false
 	default:
-		l.Warn("invalid firewall.inbound_action, defaulting to `drop`", "action", inboundAction)
+		l.WithField("action", inboundAction).Warn("invalid firewall.inbound_action, defaulting to `drop`")
 		fw.InSendReject = false
 	}
 
@@ -231,7 +230,7 @@ func NewFirewallFromConfig(l *slog.Logger, cs *CertState, c *config.C) (*Firewal
 	case "drop":
 		fw.OutSendReject = false
 	default:
-		l.Warn("invalid firewall.outbound_action, defaulting to `drop`", "action", outboundAction)
+		l.WithField("action", inboundAction).Warn("invalid firewall.outbound_action, defaulting to `drop`")
 		fw.OutSendReject = false
 	}
 
@@ -250,6 +249,20 @@ func NewFirewallFromConfig(l *slog.Logger, cs *CertState, c *config.C) (*Firewal
 
 // 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, cidr, localCidr, caName string, caSha string) error {
+	// 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, localIp: %v, caName: %v, caSha: %s",
+		incoming, proto, startPort, endPort, groups, host, cidr, localCidr, caName, caSha,
+	)
+	f.rules += ruleString + "\n"
+
+	direction := "incoming"
+	if !incoming {
+		direction = "outgoing"
+	}
+	f.l.WithField("firewallRule", m{"direction": direction, "proto": proto, "startPort": startPort, "endPort": endPort, "groups": groups, "host": host, "cidr": cidr, "localCidr": localCidr, "caName": caName, "caSha": caSha}).
+		Info("Firewall rule added")
+
 	var (
 		ft *FirewallTable
 		fp firewallPort
@@ -267,12 +280,6 @@ func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort
 	case firewall.ProtoUDP:
 		fp = ft.UDP
 	case firewall.ProtoICMP, firewall.ProtoICMPv6:
-		//ICMP traffic doesn't have ports, so we always coerce to "any", even if a value is provided
-		if startPort != firewall.PortAny {
-			f.l.Warn("ignoring port specification for ICMP firewall rule", "startPort", startPort)
-		}
-		startPort = firewall.PortAny
-		endPort = firewall.PortAny
 		fp = ft.ICMP
 	case firewall.ProtoAny:
 		fp = ft.AnyProto
@@ -280,21 +287,6 @@ func (f *Firewall) AddRule(incoming bool, proto uint8, startPort int32, endPort
 		return fmt.Errorf("unknown protocol %v", proto)
 	}
 
-	// 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, localIp: %v, caName: %v, caSha: %s",
-		incoming, proto, startPort, endPort, groups, host, cidr, localCidr, caName, caSha,
-	)
-	f.rules += ruleString + "\n"
-
-	direction := "incoming"
-	if !incoming {
-		direction = "outgoing"
-	}
-	f.l.Info("Firewall rule added",
-		"firewallRule", m{"direction": direction, "proto": proto, "startPort": startPort, "endPort": endPort, "groups": groups, "host": host, "cidr": cidr, "localCidr": localCidr, "caName": caName, "caSha": caSha},
-	)
-
 	return fp.addRule(f, startPort, endPort, groups, host, cidr, localCidr, caName, caSha)
 }
 
@@ -316,7 +308,7 @@ func (f *Firewall) GetRuleHashes() string {
 	return "SHA:" + f.GetRuleHash() + ",FNV:" + strconv.FormatUint(uint64(f.GetRuleHashFNV()), 10)
 }
 
-func AddFirewallRulesFromConfig(l *slog.Logger, inbound bool, c *config.C, fw FirewallInterface) error {
+func AddFirewallRulesFromConfig(l *logrus.Logger, inbound bool, c *config.C, fw FirewallInterface) error {
 	var table string
 	if inbound {
 		table = "firewall.inbound"
@@ -357,31 +349,24 @@ func AddFirewallRulesFromConfig(l *slog.Logger, inbound bool, c *config.C, fw Fi
 			sPort = r.Port
 		}
 
+		startPort, endPort, err := parsePort(sPort)
+		if err != nil {
+			return fmt.Errorf("%s rule #%v; %s %s", table, i, errPort, err)
+		}
+
 		var proto uint8
-		var startPort, endPort int32
 		switch r.Proto {
 		case "any":
 			proto = firewall.ProtoAny
-			startPort, endPort, err = parsePort(sPort)
 		case "tcp":
 			proto = firewall.ProtoTCP
-			startPort, endPort, err = parsePort(sPort)
 		case "udp":
 			proto = firewall.ProtoUDP
-			startPort, endPort, err = parsePort(sPort)
 		case "icmp":
 			proto = firewall.ProtoICMP
-			startPort = firewall.PortAny
-			endPort = firewall.PortAny
-			if sPort != "" {
-				l.Warn("ignoring port specification for ICMP firewall rule", "port", sPort)
-			}
 		default:
 			return fmt.Errorf("%s rule #%v; proto was not understood; `%s`", table, i, r.Proto)
 		}
-		if err != nil {
-			return fmt.Errorf("%s rule #%v; %s %s", table, i, errPort, err)
-		}
 
 		if r.Cidr != "" && r.Cidr != "any" {
 			_, err = netip.ParsePrefix(r.Cidr)
@@ -398,11 +383,7 @@ func AddFirewallRulesFromConfig(l *slog.Logger, inbound bool, c *config.C, fw Fi
 		}
 
 		if warning := r.sanity(); warning != nil {
-			l.Warn("firewall rule sanity check",
-				"table", table,
-				"rule", i,
-				"warning", warning,
-			)
+			l.Warnf("%s rule #%v; %s", table, i, warning)
 		}
 
 		err = fw.AddRule(inbound, proto, startPort, endPort, r.Groups, r.Host, r.Cidr, r.LocalCidr, r.CAName, r.CASha)
@@ -422,9 +403,9 @@ 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(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache) error {
+func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache, now time.Time) error {
 	// Check if we spoke to this tuple, if we did then allow this packet
-	if f.inConns(fp, h, caPool, localCache) {
+	if f.inConns(fp, h, caPool, localCache, now) {
 		return nil
 	}
 
@@ -473,7 +454,7 @@ func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *
 	}
 
 	// We always want to conntrack since it is a faster operation
-	f.addConn(fp, incoming)
+	f.addConn(fp, incoming, now)
 
 	return nil
 }
@@ -486,7 +467,7 @@ func (f *Firewall) metrics(incoming bool) firewallMetrics {
 	}
 }
 
-// Destroy cleans up any known cyclical references so the object can be freed by GC. This should be called if a new
+// Destroy cleans up any known cyclical references so the object can be free'd my GC. This should be called if a new
 // firewall object is created
 func (f *Firewall) Destroy() {
 	//TODO: clean references if/when needed
@@ -502,7 +483,7 @@ func (f *Firewall) EmitStats() {
 	metrics.GetOrRegisterGauge("firewall.rules.hash", nil).Update(int64(f.GetRuleHashFNV()))
 }
 
-func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache) bool {
+func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache, now time.Time) bool {
 	if localCache != nil {
 		if _, ok := localCache[fp]; ok {
 			return true
@@ -514,7 +495,7 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool,
 	// Purge every time we test
 	ep, has := conntrack.TimerWheel.Purge()
 	if has {
-		f.evict(ep)
+		f.evict(ep, now)
 	}
 
 	c, ok := conntrack.Conns[fp]
@@ -534,26 +515,26 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool,
 
 		// We now know which firewall table to check against
 		if !table.match(fp, c.incoming, h.ConnectionState.peerCert, caPool) {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				h.logger(f.l).Debug("dropping old conntrack entry, does not match new ruleset",
-					"fwPacket", fp,
-					"incoming", c.incoming,
-					"rulesVersion", f.rulesVersion,
-					"oldRulesVersion", c.rulesVersion,
-				)
+			if f.l.Level >= logrus.DebugLevel {
+				h.logger(f.l).
+					WithField("fwPacket", fp).
+					WithField("incoming", c.incoming).
+					WithField("rulesVersion", f.rulesVersion).
+					WithField("oldRulesVersion", c.rulesVersion).
+					Debugln("dropping old conntrack entry, does not match new ruleset")
 			}
 			delete(conntrack.Conns, fp)
 			conntrack.Unlock()
 			return false
 		}
 
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			h.logger(f.l).Debug("keeping old conntrack entry, does match new ruleset",
-				"fwPacket", fp,
-				"incoming", c.incoming,
-				"rulesVersion", f.rulesVersion,
-				"oldRulesVersion", c.rulesVersion,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			h.logger(f.l).
+				WithField("fwPacket", fp).
+				WithField("incoming", c.incoming).
+				WithField("rulesVersion", f.rulesVersion).
+				WithField("oldRulesVersion", c.rulesVersion).
+				Debugln("keeping old conntrack entry, does match new ruleset")
 		}
 
 		c.rulesVersion = f.rulesVersion
@@ -561,11 +542,11 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool,
 
 	switch fp.Protocol {
 	case firewall.ProtoTCP:
-		c.Expires = time.Now().Add(f.TCPTimeout)
+		c.Expires = now.Add(f.TCPTimeout)
 	case firewall.ProtoUDP:
-		c.Expires = time.Now().Add(f.UDPTimeout)
+		c.Expires = now.Add(f.UDPTimeout)
 	default:
-		c.Expires = time.Now().Add(f.DefaultTimeout)
+		c.Expires = now.Add(f.DefaultTimeout)
 	}
 
 	conntrack.Unlock()
@@ -577,7 +558,7 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool,
 	return true
 }
 
-func (f *Firewall) addConn(fp firewall.Packet, incoming bool) {
+func (f *Firewall) addConn(fp firewall.Packet, incoming bool, now time.Time) {
 	var timeout time.Duration
 	c := &conn{}
 
@@ -593,7 +574,7 @@ func (f *Firewall) addConn(fp firewall.Packet, incoming bool) {
 	conntrack := f.Conntrack
 	conntrack.Lock()
 	if _, ok := conntrack.Conns[fp]; !ok {
-		conntrack.TimerWheel.Advance(time.Now())
+		conntrack.TimerWheel.Advance(now)
 		conntrack.TimerWheel.Add(fp, timeout)
 	}
 
@@ -601,14 +582,14 @@ func (f *Firewall) addConn(fp firewall.Packet, incoming bool) {
 	// firewall reload
 	c.incoming = incoming
 	c.rulesVersion = f.rulesVersion
-	c.Expires = time.Now().Add(timeout)
+	c.Expires = now.Add(timeout)
 	conntrack.Conns[fp] = c
 	conntrack.Unlock()
 }
 
 // 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 firewall.Packet) {
+func (f *Firewall) evict(p firewall.Packet, now time.Time) {
 	// Are we still tracking this conn?
 	conntrack := f.Conntrack
 	t, ok := conntrack.Conns[p]
@@ -616,11 +597,11 @@ func (f *Firewall) evict(p firewall.Packet) {
 		return
 	}
 
-	newT := t.Expires.Sub(time.Now())
+	newT := t.Expires.Sub(now)
 
 	// Timeout is in the future, re-add the timer
 	if newT > 0 {
-		conntrack.TimerWheel.Advance(time.Now())
+		conntrack.TimerWheel.Advance(now)
 		conntrack.TimerWheel.Add(p, newT)
 		return
 	}
@@ -679,13 +660,6 @@ func (fp firewallPort) match(p firewall.Packet, incoming bool, c *cert.CachedCer
 		return false
 	}
 
-	// this branch is here to catch traffic from FirewallTable.Any.match and FirewallTable.ICMP.match
-	if p.Protocol == firewall.ProtoICMP || p.Protocol == firewall.ProtoICMPv6 {
-		// port numbers are re-used for connection tracking of ICMP,
-		// but we don't want to actually filter on them.
-		return fp[firewall.PortAny].match(p, c, caPool)
-	}
-
 	var port int32
 
 	if p.Fragment {
@@ -830,8 +804,10 @@ func (fr *FirewallRule) isAny(groups []string, host string, cidr string) bool {
 		return true
 	}
 
-	if slices.Contains(groups, "any") {
-		return true
+	for _, group := range groups {
+		if group == "any" {
+			return true
+		}
 	}
 
 	if host == "any" {
@@ -941,7 +917,7 @@ type rule struct {
 	CASha     string
 }
 
-func convertRule(l *slog.Logger, p any, table string, i int) (rule, error) {
+func convertRule(l *logrus.Logger, p any, table string, i int) (rule, error) {
 	r := rule{}
 
 	m, ok := p.(map[string]any)
@@ -972,10 +948,7 @@ func convertRule(l *slog.Logger, p any, table string, i int) (rule, error) {
 			return r, errors.New("group should contain a single value, an array with more than one entry was provided")
 		}
 
-		l.Warn("group was an array with a single value, converting to simple value",
-			"table", table,
-			"rule", i,
-		)
+		l.Warnf("%s rule #%v; group was an array with a single value, converting to simple value", table, i)
 		m["group"] = v[0]
 	}
 
@@ -1045,56 +1018,54 @@ func (r *rule) sanity() error {
 		}
 	}
 
-	if r.Code != "" {
-		return fmt.Errorf("code specified as [%s]. Support for 'code' will be dropped in a future release, as it has never been functional", r.Code)
-	}
-
 	//todo alert on cidr-any
 
 	return nil
 }
 
-func parsePort(s string) (int32, int32, error) {
-	var err error
-	const notAPort int32 = -2
+func parsePort(s string) (startPort, endPort int32, err error) {
 	if s == "any" {
-		return firewall.PortAny, firewall.PortAny, nil
-	}
-	if s == "fragment" {
-		return firewall.PortFragment, firewall.PortFragment, nil
-	}
-	if !strings.Contains(s, `-`) {
+		startPort = firewall.PortAny
+		endPort = firewall.PortAny
+
+	} else if s == "fragment" {
+		startPort = firewall.PortFragment
+		endPort = firewall.PortFragment
+
+	} else if strings.Contains(s, `-`) {
+		sPorts := strings.SplitN(s, `-`, 2)
+		sPorts[0] = strings.Trim(sPorts[0], " ")
+		sPorts[1] = strings.Trim(sPorts[1], " ")
+
+		if len(sPorts) != 2 || sPorts[0] == "" || sPorts[1] == "" {
+			return 0, 0, fmt.Errorf("appears to be a range but could not be parsed; `%s`", s)
+		}
+
+		rStartPort, err := strconv.Atoi(sPorts[0])
+		if err != nil {
+			return 0, 0, fmt.Errorf("beginning range was not a number; `%s`", sPorts[0])
+		}
+
+		rEndPort, err := strconv.Atoi(sPorts[1])
+		if err != nil {
+			return 0, 0, fmt.Errorf("ending range was not a number; `%s`", sPorts[1])
+		}
+
+		startPort = int32(rStartPort)
+		endPort = int32(rEndPort)
+
+		if startPort == firewall.PortAny {
+			endPort = firewall.PortAny
+		}
+
+	} else {
 		rPort, err := strconv.Atoi(s)
 		if err != nil {
-			return notAPort, notAPort, fmt.Errorf("was not a number; `%s`", s)
+			return 0, 0, fmt.Errorf("was not a number; `%s`", s)
 		}
-		return int32(rPort), int32(rPort), nil
+		startPort = int32(rPort)
+		endPort = startPort
 	}
 
-	sPorts := strings.SplitN(s, `-`, 2)
-	for i := range sPorts {
-		sPorts[i] = strings.Trim(sPorts[i], " ")
-	}
-	if len(sPorts) != 2 || sPorts[0] == "" || sPorts[1] == "" {
-		return notAPort, notAPort, fmt.Errorf("appears to be a range but could not be parsed; `%s`", s)
-	}
-
-	rStartPort, err := strconv.Atoi(sPorts[0])
-	if err != nil {
-		return notAPort, notAPort, fmt.Errorf("beginning range was not a number; `%s`", sPorts[0])
-	}
-
-	rEndPort, err := strconv.Atoi(sPorts[1])
-	if err != nil {
-		return notAPort, notAPort, fmt.Errorf("ending range was not a number; `%s`", sPorts[1])
-	}
-
-	startPort := int32(rStartPort)
-	endPort := int32(rEndPort)
-
-	if startPort == firewall.PortAny {
-		endPort = firewall.PortAny
-	}
-
-	return startPort, endPort, nil
+	return
 }

firewall/cache.go

@@ -1,10 +1,10 @@
 package firewall
 
 import (
-	"context"
-	"log/slog"
 	"sync/atomic"
 	"time"
+
+	"github.com/sirupsen/logrus"
 )
 
 // ConntrackCache is used as a local routine cache to know if a given flow
@@ -15,49 +15,41 @@ type ConntrackCacheTicker struct {
 	cacheV    uint64
 	cacheTick atomic.Uint64
 
-	l     *slog.Logger
 	cache ConntrackCache
 }
 
-func NewConntrackCacheTicker(ctx context.Context, l *slog.Logger, d time.Duration) *ConntrackCacheTicker {
+func NewConntrackCacheTicker(d time.Duration) *ConntrackCacheTicker {
 	if d == 0 {
 		return nil
 	}
 
 	c := &ConntrackCacheTicker{
-		l:     l,
 		cache: ConntrackCache{},
 	}
 
-	go c.tick(ctx, d)
+	go c.tick(d)
 
 	return c
 }
 
-func (c *ConntrackCacheTicker) tick(ctx context.Context, d time.Duration) {
-	t := time.NewTicker(d)
-	defer t.Stop()
+func (c *ConntrackCacheTicker) tick(d time.Duration) {
 	for {
-		select {
-		case <-ctx.Done():
-			return
-		case <-t.C:
-			c.cacheTick.Add(1)
-		}
+		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() ConntrackCache {
+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 c.l.Enabled(context.Background(), slog.LevelDebug) {
-				c.l.Debug("resetting conntrack cache", "len", ll)
+			if l.Level == logrus.DebugLevel {
+				l.WithField("len", ll).Debug("resetting conntrack cache")
 			}
 			c.cache = make(ConntrackCache, ll)
 		}

firewall/cache_test.go

@@ -1,69 +0,0 @@
-package firewall
-
-import (
-	"bytes"
-	"log/slog"
-	"strings"
-	"testing"
-
-	"github.com/slackhq/nebula/test"
-	"github.com/stretchr/testify/assert"
-)
-
-// The tests below pin the log format produced by ConntrackCacheTicker.Get
-// so changes cannot silently break what operators are grepping for. The
-// ticker's internal state (cache + cacheTick) is poked directly to avoid
-// racing a goroutine-driven tick in tests.
-
-func newFixedTicker(t *testing.T, l *slog.Logger, cacheLen int) *ConntrackCacheTicker {
-	t.Helper()
-	c := &ConntrackCacheTicker{
-		l:     l,
-		cache: make(ConntrackCache, cacheLen),
-	}
-	for i := 0; i < cacheLen; i++ {
-		c.cache[Packet{LocalPort: uint16(i) + 1}] = struct{}{}
-	}
-	c.cacheTick.Store(1) // cacheV starts at 0, so Get() takes the reset path
-	return c
-}
-
-func TestConntrackCacheTicker_Get_TextFormat(t *testing.T) {
-	buf := &bytes.Buffer{}
-	l := test.NewLoggerWithOutputAndLevel(buf, slog.LevelDebug)
-
-	c := newFixedTicker(t, l, 3)
-	c.Get()
-
-	assert.Equal(t, "level=DEBUG msg=\"resetting conntrack cache\" len=3\n", buf.String())
-}
-
-func TestConntrackCacheTicker_Get_JSONFormat(t *testing.T) {
-	buf := &bytes.Buffer{}
-	l := test.NewJSONLoggerWithOutput(buf, slog.LevelDebug)
-
-	c := newFixedTicker(t, l, 2)
-	c.Get()
-
-	assert.JSONEq(t, `{"level":"DEBUG","msg":"resetting conntrack cache","len":2}`, strings.TrimSpace(buf.String()))
-}
-
-func TestConntrackCacheTicker_Get_QuietBelowDebug(t *testing.T) {
-	buf := &bytes.Buffer{}
-	l := test.NewLoggerWithOutputAndLevel(buf, slog.LevelInfo)
-
-	c := newFixedTicker(t, l, 5)
-	c.Get()
-
-	assert.Empty(t, buf.String())
-}
-
-func TestConntrackCacheTicker_Get_QuietWhenCacheEmpty(t *testing.T) {
-	buf := &bytes.Buffer{}
-	l := test.NewLoggerWithOutputAndLevel(buf, slog.LevelDebug)
-
-	c := newFixedTicker(t, l, 0)
-	c.Get()
-
-	assert.Empty(t, buf.String())
-}

firewall/packet.go

@@ -22,10 +22,7 @@ const (
 type Packet struct {
 	LocalAddr  netip.Addr
 	RemoteAddr netip.Addr
-	// LocalPort is the destination port for incoming traffic, or the source port for outgoing. Zero for ICMP.
-	LocalPort uint16
-	// RemotePort is the source port for incoming traffic, or the destination port for outgoing.
-	// For ICMP, it's the "identifier". This is only used for connection tracking, actual firewall rules will not filter on ICMP identifier
+	LocalPort  uint16
 	RemotePort uint16
 	Protocol   uint8
 	Fragment   bool
@@ -49,8 +46,6 @@ func (fp Packet) MarshalJSON() ([]byte, error) {
 		proto = "tcp"
 	case ProtoICMP:
 		proto = "icmp"
-	case ProtoICMPv6:
-		proto = "icmpv6"
 	case ProtoUDP:
 		proto = "udp"
 	default:

firewall_test.go

@@ -3,13 +3,13 @@ package nebula
 import (
 	"bytes"
 	"errors"
-	"log/slog"
 	"math"
 	"net/netip"
 	"testing"
 	"time"
 
 	"github.com/gaissmai/bart"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/firewall"
@@ -58,8 +58,9 @@ func TestNewFirewall(t *testing.T) {
 }
 
 func TestFirewall_AddRule(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 
 	c := &dummyCert{}
 	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c)
@@ -86,10 +87,9 @@ func TestFirewall_AddRule(t *testing.T) {
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	require.NoError(t, fw.AddRule(true, firewall.ProtoICMP, 1, 1, []string{}, "h1", "", "", "", ""))
-	//no matter what port is given for icmp, it should end up as "any"
-	assert.Nil(t, fw.InRules.ICMP[firewall.PortAny].Any.Any)
-	assert.Empty(t, fw.InRules.ICMP[firewall.PortAny].Any.Groups)
-	assert.Contains(t, fw.InRules.ICMP[firewall.PortAny].Any.Hosts, "h1")
+	assert.Nil(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")
 
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", ti.String(), "", "", ""))
@@ -176,8 +176,9 @@ func TestFirewall_AddRule(t *testing.T) {
 }
 
 func TestFirewall_Drop(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("1.1.1.1/8"))
 	p := firewall.Packet{
@@ -252,8 +253,9 @@ func TestFirewall_Drop(t *testing.T) {
 }
 
 func TestFirewall_DropV6(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("fd00::/7"))
@@ -482,8 +484,9 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 }
 
 func TestFirewall_Drop2(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("1.1.1.1/8"))
 
@@ -540,8 +543,9 @@ func TestFirewall_Drop2(t *testing.T) {
 }
 
 func TestFirewall_Drop3(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("1.1.1.1/8"))
 
@@ -628,8 +632,9 @@ func TestFirewall_Drop3(t *testing.T) {
 }
 
 func TestFirewall_Drop3V6(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("fd00::/7"))
 
@@ -665,8 +670,9 @@ func TestFirewall_Drop3V6(t *testing.T) {
 }
 
 func TestFirewall_DropConntrackReload(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("1.1.1.1/8"))
 
@@ -728,152 +734,10 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 	assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrNoMatchingRule)
 }
 
-func TestFirewall_ICMPPortBehavior(t *testing.T) {
-	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
-	myVpnNetworksTable := new(bart.Lite)
-	myVpnNetworksTable.Insert(netip.MustParsePrefix("1.1.1.1/8"))
-
-	network := netip.MustParsePrefix("1.2.3.4/24")
-
-	c := cert.CachedCertificate{
-		Certificate: &dummyCert{
-			name:     "host1",
-			networks: []netip.Prefix{network},
-			groups:   []string{"default-group"},
-			issuer:   "signer-shasum",
-		},
-		InvertedGroups: map[string]struct{}{"default-group": {}},
-	}
-	h := HostInfo{
-		ConnectionState: &ConnectionState{
-			peerCert: &c,
-		},
-		vpnAddrs: []netip.Addr{network.Addr()},
-	}
-	h.buildNetworks(myVpnNetworksTable, c.Certificate)
-
-	cp := cert.NewCAPool()
-
-	templ := firewall.Packet{
-		LocalAddr:  netip.MustParseAddr("1.2.3.4"),
-		RemoteAddr: netip.MustParseAddr("1.2.3.4"),
-		Protocol:   firewall.ProtoICMP,
-		Fragment:   false,
-	}
-
-	t.Run("ICMP allowed", func(t *testing.T) {
-		fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
-		require.NoError(t, fw.AddRule(true, firewall.ProtoICMP, 0, 0, []string{"any"}, "", "", "", "", ""))
-		t.Run("zero ports", func(t *testing.T) {
-			p := templ.Copy()
-			p.LocalPort = 0
-			p.RemotePort = 0
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
-			//now also allow outbound
-			require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
-		})
-
-		t.Run("nonzero ports", func(t *testing.T) {
-			p := templ.Copy()
-			p.LocalPort = 0xabcd
-			p.RemotePort = 0x1234
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
-			//now also allow outbound
-			require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
-		})
-	})
-
-	t.Run("Any proto, some ports allowed", func(t *testing.T) {
-		fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
-		require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 80, 444, []string{"any"}, "", "", "", "", ""))
-		t.Run("zero ports, still blocked", func(t *testing.T) {
-			p := templ.Copy()
-			p.LocalPort = 0
-			p.RemotePort = 0
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
-			//now also allow outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-		})
-
-		t.Run("nonzero ports, still blocked", func(t *testing.T) {
-			p := templ.Copy()
-			p.LocalPort = 0xabcd
-			p.RemotePort = 0x1234
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
-			//now also allow outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-		})
-
-		t.Run("nonzero, matching ports, still blocked", func(t *testing.T) {
-			p := templ.Copy()
-			p.LocalPort = 80
-			p.RemotePort = 80
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
-			//now also allow outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-		})
-	})
-	t.Run("Any proto, any port", func(t *testing.T) {
-		fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
-		require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", "", "", "", ""))
-		t.Run("zero ports, allowed", func(t *testing.T) {
-			resetConntrack(fw)
-			p := templ.Copy()
-			p.LocalPort = 0
-			p.RemotePort = 0
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
-			//now also allow outbound
-			require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
-		})
-
-		t.Run("nonzero ports, allowed", func(t *testing.T) {
-			resetConntrack(fw)
-			p := templ.Copy()
-			p.LocalPort = 0xabcd
-			p.RemotePort = 0x1234
-			// Drop outbound
-			assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-			// Allow inbound
-			resetConntrack(fw)
-			require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
-			//now also allow outbound
-			require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
-			//different ID is blocked
-			p.RemotePort++
-			require.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
-		})
-	})
-
-}
-
 func TestFirewall_DropIPSpoofing(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 	myVpnNetworksTable := new(bart.Lite)
 	myVpnNetworksTable.Insert(netip.MustParsePrefix("192.0.2.1/24"))
 
@@ -1036,53 +900,53 @@ func TestNewFirewallFromConfig(t *testing.T) {
 	cs, err := newCertState(cert.Version2, nil, c, false, cert.Curve_CURVE25519, nil)
 	require.NoError(t, err)
 
-	conf := config.NewC(test.NewLogger())
+	conf := config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": "asdf"}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound failed to parse, should be an array of rules")
 
 	// Test both port and code
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "1", "code": "2"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.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 = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.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 = config.NewC(test.NewLogger())
-	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"code": "a", "host": "testh", "proto": "any"}}}
+	conf = config.NewC(l)
+	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"code": "a", "host": "testh"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound rule #0; code was not a number; `a`")
 
-	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "a", "host": "testh", "proto": "any"}}}
+	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "a", "host": "testh"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound rule #0; port was not a number; `a`")
 
 	// Test proto error
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"code": "1", "host": "testh"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound rule #0; proto was not understood; ``")
 
 	// Test cidr parse error
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"code": "1", "cidr": "testh", "proto": "any"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound rule #0; cidr did not parse; netip.ParsePrefix(\"testh\"): no '/'")
 
 	// Test local_cidr parse error
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"code": "1", "local_cidr": "testh", "proto": "any"}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.outbound rule #0; local_cidr did not parse; netip.ParsePrefix(\"testh\"): no '/'")
 
 	// Test both group and groups
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "group": "a", "groups": []string{"b", "c"}}}}
 	_, err = NewFirewallFromConfig(l, cs, conf)
 	require.EqualError(t, err, "firewall.inbound rule #0; only one of group or groups should be defined, both provided")
@@ -1091,35 +955,28 @@ func TestNewFirewallFromConfig(t *testing.T) {
 func TestAddFirewallRulesFromConfig(t *testing.T) {
 	l := test.NewLogger()
 	// Test adding tcp rule
-	conf := config.NewC(test.NewLogger())
+	conf := config.NewC(l)
 	mf := &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "1", "proto": "tcp", "host": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, false, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoTCP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: "", localIp: ""}, mf.lastCall)
 
 	// Test adding udp rule
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "1", "proto": "udp", "host": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, false, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoUDP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: "", localIp: ""}, mf.lastCall)
 
 	// Test adding icmp rule
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"port": "1", "proto": "icmp", "host": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, false, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoICMP, startPort: firewall.PortAny, endPort: firewall.PortAny, groups: nil, host: "a", ip: "", localIp: ""}, mf.lastCall)
-
-	// Test adding icmp rule no port
-	conf = config.NewC(test.NewLogger())
-	mf = &mockFirewall{}
-	conf.Settings["firewall"] = map[string]any{"outbound": []any{map[string]any{"proto": "icmp", "host": "a"}}}
-	require.NoError(t, AddFirewallRulesFromConfig(l, false, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoICMP, startPort: firewall.PortAny, endPort: firewall.PortAny, groups: nil, host: "a", ip: "", localIp: ""}, mf.lastCall)
+	assert.Equal(t, addRuleCall{incoming: false, proto: firewall.ProtoICMP, startPort: 1, endPort: 1, groups: nil, host: "a", ip: "", localIp: ""}, mf.lastCall)
 
 	// Test adding any rule
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "host": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
@@ -1127,14 +984,14 @@ func TestAddFirewallRulesFromConfig(t *testing.T) {
 
 	// Test adding rule with cidr
 	cidr := netip.MustParsePrefix("10.0.0.0/8")
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "cidr": cidr.String()}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: cidr.String(), localIp: ""}, mf.lastCall)
 
 	// Test adding rule with local_cidr
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "local_cidr": cidr.String()}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
@@ -1142,82 +999,82 @@ func TestAddFirewallRulesFromConfig(t *testing.T) {
 
 	// Test adding rule with cidr ipv6
 	cidr6 := netip.MustParsePrefix("fd00::/8")
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "cidr": cidr6.String()}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: cidr6.String(), localIp: ""}, mf.lastCall)
 
 	// Test adding rule with any cidr
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "cidr": "any"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: "any", localIp: ""}, mf.lastCall)
 
 	// Test adding rule with junk cidr
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "cidr": "junk/junk"}}}
 	require.EqualError(t, AddFirewallRulesFromConfig(l, true, conf, mf), "firewall.inbound rule #0; cidr did not parse; netip.ParsePrefix(\"junk/junk\"): ParseAddr(\"junk\"): unable to parse IP")
 
 	// Test adding rule with local_cidr ipv6
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "local_cidr": cidr6.String()}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: "", localIp: cidr6.String()}, mf.lastCall)
 
 	// Test adding rule with any local_cidr
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "local_cidr": "any"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, localIp: "any"}, mf.lastCall)
 
 	// Test adding rule with junk local_cidr
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "local_cidr": "junk/junk"}}}
 	require.EqualError(t, AddFirewallRulesFromConfig(l, true, conf, mf), "firewall.inbound rule #0; local_cidr did not parse; netip.ParsePrefix(\"junk/junk\"): ParseAddr(\"junk\"): unable to parse IP")
 
 	// Test adding rule with ca_sha
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "ca_sha": "12312313123"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: "", localIp: "", caSha: "12312313123"}, mf.lastCall)
 
 	// Test adding rule with ca_name
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "ca_name": "root01"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: "", localIp: "", caName: "root01"}, mf.lastCall)
 
 	// Test single group
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "group": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: "", localIp: ""}, mf.lastCall)
 
 	// Test single groups
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "groups": "a"}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a"}, ip: "", localIp: ""}, mf.lastCall)
 
 	// Test multiple AND groups
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "groups": []string{"a", "b"}}}}
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: []string{"a", "b"}, ip: "", localIp: ""}, mf.lastCall)
 
 	// Test Add error
-	conf = config.NewC(test.NewLogger())
+	conf = config.NewC(l)
 	mf = &mockFirewall{}
 	mf.nextCallReturn = errors.New("test error")
 	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "host": "a"}}}
@@ -1225,8 +1082,9 @@ func TestAddFirewallRulesFromConfig(t *testing.T) {
 }
 
 func TestFirewall_convertRule(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 
 	// Ensure group array of 1 is converted and a warning is printed
 	c := map[string]any{
@@ -1234,9 +1092,7 @@ func TestFirewall_convertRule(t *testing.T) {
 	}
 
 	r, err := convertRule(l, c, "test", 1)
-	assert.Contains(t, ob.String(), "group was an array with a single value, converting to simple value")
-	assert.Contains(t, ob.String(), "table=test")
-	assert.Contains(t, ob.String(), "rule=1")
+	assert.Contains(t, ob.String(), "test rule #1; group was an array with a single value, converting to simple value")
 	require.NoError(t, err)
 	assert.Equal(t, []string{"group1"}, r.Groups)
 
@@ -1262,8 +1118,9 @@ func TestFirewall_convertRule(t *testing.T) {
 }
 
 func TestFirewall_convertRuleSanity(t *testing.T) {
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 
 	noWarningPlease := []map[string]any{
 		{"group": "group1"},
@@ -1377,7 +1234,7 @@ type testsetup struct {
 	fw                 *Firewall
 }
 
-func newSetup(t *testing.T, l *slog.Logger, myPrefixes ...netip.Prefix) testsetup {
+func newSetup(t *testing.T, l *logrus.Logger, myPrefixes ...netip.Prefix) testsetup {
 	c := dummyCert{
 		name:     "me",
 		networks: myPrefixes,
@@ -1388,7 +1245,7 @@ func newSetup(t *testing.T, l *slog.Logger, myPrefixes ...netip.Prefix) testsetu
 	return newSetupFromCert(t, l, c)
 }
 
-func newSetupFromCert(t *testing.T, l *slog.Logger, c dummyCert) testsetup {
+func newSetupFromCert(t *testing.T, l *logrus.Logger, c dummyCert) testsetup {
 	myVpnNetworksTable := new(bart.Lite)
 	for _, prefix := range c.Networks() {
 		myVpnNetworksTable.Insert(prefix)
@@ -1405,8 +1262,9 @@ func newSetupFromCert(t *testing.T, l *slog.Logger, c dummyCert) testsetup {
 
 func TestFirewall_Drop_EnforceIPMatch(t *testing.T) {
 	t.Parallel()
+	l := test.NewLogger()
 	ob := &bytes.Buffer{}
-	l := test.NewLoggerWithOutput(ob)
+	l.SetOutput(ob)
 
 	myPrefix := netip.MustParsePrefix("1.1.1.1/8")
 	// for now, it's okay that these are all "incoming", the logic this test tries to check doesn't care about in/out

go.mod

@@ -1,10 +1,9 @@
 module github.com/slackhq/nebula
 
-go 1.25.0
+go 1.25
 
 require (
 	dario.cat/mergo v1.0.2
-	filippo.io/bigmod v0.1.0
 	github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be
 	github.com/armon/go-radix v1.0.0
 	github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432
@@ -13,26 +12,27 @@ require (
 	github.com/gogo/protobuf v1.3.2
 	github.com/google/gopacket v1.1.19
 	github.com/kardianos/service v1.2.4
-	github.com/miekg/dns v1.1.72
-	github.com/miekg/pkcs11 v1.1.2
+	github.com/miekg/dns v1.1.68
+	github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b
 	github.com/nbrownus/go-metrics-prometheus v0.0.0-20210712211119-974a6260965f
 	github.com/prometheus/client_golang v1.23.2
 	github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475
+	github.com/sirupsen/logrus v1.9.3
 	github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e
 	github.com/stefanberger/go-pkcs11uri v0.0.0-20230803200340-78284954bff6
 	github.com/stretchr/testify v1.11.1
 	github.com/vishvananda/netlink v1.3.1
 	go.yaml.in/yaml/v3 v3.0.4
-	golang.org/x/crypto v0.50.0
+	golang.org/x/crypto v0.45.0
 	golang.org/x/exp v0.0.0-20230725093048-515e97ebf090
-	golang.org/x/net v0.52.0
-	golang.org/x/sync v0.20.0
-	golang.org/x/sys v0.43.0
-	golang.org/x/term v0.42.0
+	golang.org/x/net v0.47.0
+	golang.org/x/sync v0.19.0
+	golang.org/x/sys v0.39.0
+	golang.org/x/term v0.38.0
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2
 	golang.zx2c4.com/wireguard v0.0.0-20230325221338-052af4a8072b
-	golang.zx2c4.com/wireguard/windows v0.6.1
-	google.golang.org/protobuf v1.36.11
+	golang.zx2c4.com/wireguard/windows v0.5.3
+	google.golang.org/protobuf v1.36.10
 	gopkg.in/yaml.v3 v3.0.1
 	gvisor.dev/gvisor v0.0.0-20240423190808-9d7a357edefe
 )
@@ -49,7 +49,7 @@ require (
 	github.com/prometheus/procfs v0.16.1 // indirect
 	github.com/vishvananda/netns v0.0.5 // indirect
 	go.yaml.in/yaml/v2 v2.4.2 // indirect
-	golang.org/x/mod v0.34.0 // indirect
-	golang.org/x/time v0.5.0 // indirect
-	golang.org/x/tools v0.43.0 // indirect
+	golang.org/x/mod v0.24.0 // indirect
+	golang.org/x/time v0.7.0 // indirect
+	golang.org/x/tools v0.33.0 // indirect
 )

go.sum

@@ -1,8 +1,6 @@
 cloud.google.com/go v0.34.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
 dario.cat/mergo v1.0.2 h1:85+piFYR1tMbRrLcDwR18y4UKJ3aH1Tbzi24VRW1TK8=
 dario.cat/mergo v1.0.2/go.mod h1:E/hbnu0NxMFBjpMIE34DRGLWqDy0g5FuKDhCb31ngxA=
-filippo.io/bigmod v0.1.0 h1:UNzDk7y9ADKST+axd9skUpBQeW7fG2KrTZyOE4uGQy8=
-filippo.io/bigmod v0.1.0/go.mod h1:OjOXDNlClLblvXdwgFFOQFJEocLhhtai8vGLy0JCZlI=
 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=
@@ -85,10 +83,10 @@ github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0SNc=
 github.com/kylelemons/godebug v1.1.0/go.mod h1:9/0rRGxNHcop5bhtWyNeEfOS8JIWk580+fNqagV/RAw=
 github.com/matttproud/golang_protobuf_extensions v1.0.1/go.mod h1:D8He9yQNgCq6Z5Ld7szi9bcBfOoFv/3dc6xSMkL2PC0=
-github.com/miekg/dns v1.1.72 h1:vhmr+TF2A3tuoGNkLDFK9zi36F2LS+hKTRW0Uf8kbzI=
-github.com/miekg/dns v1.1.72/go.mod h1:+EuEPhdHOsfk6Wk5TT2CzssZdqkmFhf8r+aVyDEToIs=
-github.com/miekg/pkcs11 v1.1.2 h1:/VxmeAX5qU6Q3EwafypogwWbYryHFmF2RpkJmw3m4MQ=
-github.com/miekg/pkcs11 v1.1.2/go.mod h1:XsNlhZGX73bx86s2hdc/FuaLm2CPZJemRLMA+WTFxgs=
+github.com/miekg/dns v1.1.68 h1:jsSRkNozw7G/mnmXULynzMNIsgY2dHC8LO6U6Ij2JEA=
+github.com/miekg/dns v1.1.68/go.mod h1:fujopn7TB3Pu3JM69XaawiU0wqjpL9/8xGop5UrTPps=
+github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b h1:J/AzCvg5z0Hn1rqZUJjpbzALUmkKX0Zwbc/i4fw7Sfk=
+github.com/miekg/pkcs11 v1.1.2-0.20231115102856-9078ad6b9d4b/go.mod h1:XsNlhZGX73bx86s2hdc/FuaLm2CPZJemRLMA+WTFxgs=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 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=
@@ -133,6 +131,8 @@ github.com/rogpeppe/go-internal v1.10.0/go.mod h1:UQnix2H7Ngw/k4C5ijL5+65zddjncj
 github.com/sirupsen/logrus v1.2.0/go.mod h1:LxeOpSwHxABJmUn/MG1IvRgCAasNZTLOkJPxbbu5VWo=
 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 v1.9.3 h1:dueUQJ1C2q9oE3F7wvmSGAaVtTmUizReu6fjN8uqzbQ=
+github.com/sirupsen/logrus v1.9.3/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/stefanberger/go-pkcs11uri v0.0.0-20230803200340-78284954bff6 h1:pnnLyeX7o/5aX8qUQ69P/mLojDqwda8hFOCBTmP/6hw=
@@ -162,16 +162,16 @@ golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACk
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20210322153248-0c34fe9e7dc2/go.mod h1:T9bdIzuCu7OtxOm1hfPfRQxPLYneinmdGuTeoZ9dtd4=
-golang.org/x/crypto v0.50.0 h1:zO47/JPrL6vsNkINmLoo/PH1gcxpls50DNogFvB5ZGI=
-golang.org/x/crypto v0.50.0/go.mod h1:3muZ7vA7PBCE6xgPX7nkzzjiUq87kRItoJQM1Yo8S+Q=
+golang.org/x/crypto v0.45.0 h1:jMBrvKuj23MTlT0bQEOBcAE0mjg8mK9RXFhRH6nyF3Q=
+golang.org/x/crypto v0.45.0/go.mod h1:XTGrrkGJve7CYK7J8PEww4aY7gM3qMCElcJQ8n8JdX4=
 golang.org/x/exp v0.0.0-20230725093048-515e97ebf090 h1:Di6/M8l0O2lCLc6VVRWhgCiApHV8MnQurBnFSHsQtNY=
 golang.org/x/exp v0.0.0-20230725093048-515e97ebf090/go.mod h1:FXUEEKJgO7OQYeo8N01OfiKP8RXMtf6e8aTskBGqWdc=
 golang.org/x/lint v0.0.0-20200302205851-738671d3881b/go.mod h1:3xt1FjdF8hUf6vQPIChWIBhFzV8gjjsPE/fR3IyQdNY=
 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 v0.34.0 h1:xIHgNUUnW6sYkcM5Jleh05DvLOtwc6RitGHbDk4akRI=
-golang.org/x/mod v0.34.0/go.mod h1:ykgH52iCZe79kzLLMhyCUzhMci+nQj+0XkbXpNYtVjY=
+golang.org/x/mod v0.24.0 h1:ZfthKaKaT4NrhGVZHO1/WDTwGES4De8KtWO0SIbNJMU=
+golang.org/x/mod v0.24.0/go.mod h1:IXM97Txy2VM4PJ3gI61r1YEk/gAj6zAHN3AdZt6S9Ww=
 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=
@@ -182,8 +182,8 @@ golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLL
 golang.org/x/net v0.0.0-20200625001655-4c5254603344/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
-golang.org/x/net v0.52.0 h1:He/TN1l0e4mmR3QqHMT2Xab3Aj3L9qjbhRm78/6jrW0=
-golang.org/x/net v0.52.0/go.mod h1:R1MAz7uMZxVMualyPXb+VaqGSa3LIaUqk0eEt3w36Sw=
+golang.org/x/net v0.47.0 h1:Mx+4dIFzqraBXUugkia1OOvlD6LemFo1ALMHjrXDOhY=
+golang.org/x/net v0.47.0/go.mod h1:/jNxtkgq5yWUGYkaZGqo27cfGZ1c5Nen03aYrrKpVRU=
 golang.org/x/oauth2 v0.0.0-20190226205417-e64efc72b421/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw=
 golang.org/x/sync v0.0.0-20181108010431-42b317875d0f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
@@ -191,8 +191,8 @@ golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201207232520-09787c993a3a/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
-golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sync v0.19.0 h1:vV+1eWNmZ5geRlYjzm2adRgW2/mcpevXNg50YZtPCE4=
+golang.org/x/sync v0.19.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
 golang.org/x/sys v0.0.0-20180905080454-ebe1bf3edb33/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20181116152217-5ac8a444bdc5/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@@ -206,25 +206,26 @@ golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7w
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 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 v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
-golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/sys v0.39.0 h1:CvCKL8MeisomCi6qNZ+wbb0DN9E5AATixKsvNtMoMFk=
+golang.org/x/sys v0.39.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
-golang.org/x/term v0.42.0 h1:UiKe+zDFmJobeJ5ggPwOshJIVt6/Ft0rcfrXZDLWAWY=
-golang.org/x/term v0.42.0/go.mod h1:Dq/D+snpsbazcBG5+F9Q1n2rXV8Ma+71xEjTRufARgY=
+golang.org/x/term v0.38.0 h1:PQ5pkm/rLO6HnxFR7N2lJHOZX6Kez5Y1gDSJla6jo7Q=
+golang.org/x/term v0.38.0/go.mod h1:bSEAKrOT1W+VSu9TSCMtoGEOUcKxOKgl3LE5QEF/xVg=
 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/time v0.5.0 h1:o7cqy6amK/52YcAKIPlM3a+Fpj35zvRj2TP+e1xFSfk=
-golang.org/x/time v0.5.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
+golang.org/x/time v0.7.0 h1:ntUhktv3OPE6TgYxXWv9vKvUSJyIFJlyohwbkEwPrKQ=
+golang.org/x/time v0.7.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 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 v0.43.0 h1:12BdW9CeB3Z+J/I/wj34VMl8X+fEXBxVR90JeMX5E7s=
-golang.org/x/tools v0.43.0/go.mod h1:uHkMso649BX2cZK6+RpuIPXS3ho2hZo4FVwfoy1vIk0=
+golang.org/x/tools v0.33.0 h1:4qz2S3zmRxbGIhDIAgjxvFutSvH5EfnsYrRBj0UI0bc=
+golang.org/x/tools v0.33.0/go.mod h1:CIJMaWEY88juyUfo7UbgPqbC8rU2OqfAV1h2Qp0oMYI=
 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=
@@ -233,8 +234,8 @@ golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 h1:B82qJJgjvYKsXS9jeu
 golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2/go.mod h1:deeaetjYA+DHMHg+sMSMI58GrEteJUUzzw7en6TJQcI=
 golang.zx2c4.com/wireguard v0.0.0-20230325221338-052af4a8072b h1:J1CaxgLerRR5lgx3wnr6L04cJFbWoceSK9JWBdglINo=
 golang.zx2c4.com/wireguard v0.0.0-20230325221338-052af4a8072b/go.mod h1:tqur9LnfstdR9ep2LaJT4lFUl0EjlHtge+gAjmsHUG4=
-golang.zx2c4.com/wireguard/windows v0.6.1 h1:XMaKojH1Hs/raMrmnir4n35nTvzvWj7NmSYzHn2F4qU=
-golang.zx2c4.com/wireguard/windows v0.6.1/go.mod h1:04aqInu5GYuTFvMuDw/rKBAF7mHrltW/3rekpfbbZDM=
+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=
@@ -243,8 +244,8 @@ google.golang.org/protobuf v1.20.1-0.20200309200217-e05f789c0967/go.mod h1:A+miE
 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 v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
-google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
+google.golang.org/protobuf v1.36.10 h1:AYd7cD/uASjIL6Q9LiTjz8JLcrh/88q5UObnmY3aOOE=
+google.golang.org/protobuf v1.36.10/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
 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=

handshake_ix.go

@@ -1,13 +1,11 @@
 package nebula
 
 import (
-	"bytes"
-	"context"
-	"log/slog"
 	"net/netip"
 	"time"
 
 	"github.com/flynn/noise"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/header"
 )
@@ -19,11 +17,8 @@ import (
 func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 	err := f.handshakeManager.allocateIndex(hh)
 	if err != nil {
-		f.l.Error("Failed to generate index",
-			"error", err,
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to generate index")
 		return false
 	}
 
@@ -43,32 +38,28 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 
 	crt := cs.getCertificate(v)
 	if crt == nil {
-		f.l.Error("Unable to handshake with host because no certificate is available",
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-			"certVersion", v,
-		)
+		f.l.WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).
+			WithField("certVersion", v).
+			Error("Unable to handshake with host because no certificate is available")
 		return false
 	}
 
 	crtHs := cs.getHandshakeBytes(v)
 	if crtHs == nil {
-		f.l.Error("Unable to handshake with host because no certificate handshake bytes is available",
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-			"certVersion", v,
-		)
+		f.l.WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).
+			WithField("certVersion", v).
+			Error("Unable to handshake with host because no certificate handshake bytes is available")
 		return false
 	}
 
-	ci, err := NewConnectionState(cs, crt, true, noise.HandshakeIX)
+	ci, err := NewConnectionState(f.l, cs, crt, true, noise.HandshakeIX)
 	if err != nil {
-		f.l.Error("Failed to create connection state",
-			"error", err,
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-			"certVersion", v,
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).
+			WithField("certVersion", v).
+			Error("Failed to create connection state")
 		return false
 	}
 	hh.hostinfo.ConnectionState = ci
@@ -84,12 +75,9 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 
 	hsBytes, err := hs.Marshal()
 	if err != nil {
-		f.l.Error("Failed to marshal handshake message",
-			"error", err,
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"certVersion", v,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("certVersion", v).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to marshal handshake message")
 		return false
 	}
 
@@ -97,11 +85,8 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 
 	msg, _, _, err := ci.H.WriteMessage(h, hsBytes)
 	if err != nil {
-		f.l.Error("Failed to call noise.WriteMessage",
-			"error", err,
-			"vpnAddrs", hh.hostinfo.vpnAddrs,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hh.hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).Error("Failed to call noise.WriteMessage")
 		return false
 	}
 
@@ -118,21 +103,18 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	cs := f.pki.getCertState()
 	crt := cs.GetDefaultCertificate()
 	if crt == nil {
-		f.l.Error("Unable to handshake with host because no certificate is available",
-			"from", via,
-			"handshake", m{"stage": 0, "style": "ix_psk0"},
-			"certVersion", cs.initiatingVersion,
-		)
+		f.l.WithField("from", via).
+			WithField("handshake", m{"stage": 0, "style": "ix_psk0"}).
+			WithField("certVersion", cs.initiatingVersion).
+			Error("Unable to handshake with host because no certificate is available")
 		return
 	}
 
-	ci, err := NewConnectionState(cs, crt, false, noise.HandshakeIX)
+	ci, err := NewConnectionState(f.l, cs, crt, false, noise.HandshakeIX)
 	if err != nil {
-		f.l.Error("Failed to create connection state",
-			"error", err,
-			"from", via,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Error("Failed to create connection state")
 		return
 	}
 
@@ -141,32 +123,26 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 
 	msg, _, _, err := ci.H.ReadMessage(nil, packet[header.Len:])
 	if err != nil {
-		f.l.Error("Failed to call noise.ReadMessage",
-			"error", err,
-			"from", via,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Error("Failed to call noise.ReadMessage")
 		return
 	}
 
 	hs := &NebulaHandshake{}
 	err = hs.Unmarshal(msg)
 	if err != nil || hs.Details == nil {
-		f.l.Error("Failed unmarshal handshake message",
-			"error", err,
-			"from", via,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Error("Failed unmarshal handshake message")
 		return
 	}
 
 	rc, err := cert.Recombine(cert.Version(hs.Details.CertVersion), hs.Details.Cert, ci.H.PeerStatic(), ci.Curve())
 	if err != nil {
-		f.l.Info("Handshake did not contain a certificate",
-			"error", err,
-			"from", via,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Info("Handshake did not contain a certificate")
 		return
 	}
 
@@ -177,30 +153,16 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 			fp = "<error generating certificate fingerprint>"
 		}
 
-		attrs := []slog.Attr{
-			slog.Any("error", err),
-			slog.Any("from", via),
-			slog.Any("handshake", m{"stage": 1, "style": "ix_psk0"}),
-			slog.Any("certVpnNetworks", rc.Networks()),
-			slog.String("certFingerprint", fp),
-		}
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			attrs = append(attrs, slog.Any("cert", rc))
+		e := f.l.WithError(err).WithField("from", via).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			WithField("certVpnNetworks", rc.Networks()).
+			WithField("certFingerprint", fp)
+
+		if f.l.Level >= logrus.DebugLevel {
+			e = e.WithField("cert", rc)
 		}
 
-		// LogAttrs is intentional: attrs is a pre-built []slog.Attr slice that
-		// callers grow conditionally, which has no pair-form equivalent.
-		//nolint:sloglint
-		f.l.LogAttrs(context.Background(), slog.LevelInfo, "Invalid certificate from host", attrs...)
-		return
-	}
-
-	if !bytes.Equal(remoteCert.Certificate.PublicKey(), ci.H.PeerStatic()) {
-		f.l.Info("public key mismatch between certificate and handshake",
-			"from", via,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-			"cert", remoteCert,
-		)
+		e.Info("Invalid certificate from host")
 		return
 	}
 
@@ -208,13 +170,12 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 		// We started off using the wrong certificate version, lets see if we can match the version that was sent to us
 		myCertOtherVersion := cs.getCertificate(remoteCert.Certificate.Version())
 		if myCertOtherVersion == nil {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				f.l.Debug("Might be unable to handshake with host due to missing certificate version",
-					"error", err,
-					"from", via,
-					"handshake", m{"stage": 1, "style": "ix_psk0"},
-					"cert", remoteCert,
-				)
+			if f.l.Level >= logrus.DebugLevel {
+				f.l.WithError(err).WithFields(m{
+					"from":      via,
+					"handshake": m{"stage": 1, "style": "ix_psk0"},
+					"cert":      remoteCert,
+				}).Debug("Might be unable to handshake with host due to missing certificate version")
 			}
 		} else {
 			// Record the certificate we are actually using
@@ -223,12 +184,10 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	}
 
 	if len(remoteCert.Certificate.Networks()) == 0 {
-		f.l.Info("No networks in certificate",
-			"error", err,
-			"from", via,
-			"cert", remoteCert,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("cert", remoteCert).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Info("No networks in certificate")
 		return
 	}
 
@@ -242,15 +201,12 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	vpnAddrs := make([]netip.Addr, len(vpnNetworks))
 	for i, network := range vpnNetworks {
 		if f.myVpnAddrsTable.Contains(network.Addr()) {
-			f.l.Error("Refusing to handshake with myself",
-				"vpnNetworks", vpnNetworks,
-				"from", via,
-				"certName", certName,
-				"certVersion", certVersion,
-				"fingerprint", fingerprint,
-				"issuer", issuer,
-				"handshake", m{"stage": 1, "style": "ix_psk0"},
-			)
+			f.l.WithField("vpnNetworks", vpnNetworks).WithField("from", via).
+				WithField("certName", certName).
+				WithField("certVersion", certVersion).
+				WithField("fingerprint", fingerprint).
+				WithField("issuer", issuer).
+				WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Refusing to handshake with myself")
 			return
 		}
 		vpnAddrs[i] = network.Addr()
@@ -262,28 +218,20 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	if !via.IsRelayed {
 		// We only want to apply the remote allow list for direct tunnels here
 		if !f.lightHouse.GetRemoteAllowList().AllowAll(vpnAddrs, via.UdpAddr.Addr()) {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				f.l.Debug("lighthouse.remote_allow_list denied incoming handshake",
-					"vpnAddrs", vpnAddrs,
-					"from", via,
-				)
-			}
+			f.l.WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+				Debug("lighthouse.remote_allow_list denied incoming handshake")
 			return
 		}
 	}
 
 	myIndex, err := generateIndex(f.l)
 	if err != nil {
-		f.l.Error("Failed to generate index",
-			"error", err,
-			"vpnAddrs", vpnAddrs,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to generate index")
 		return
 	}
 
@@ -301,18 +249,18 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 		},
 	}
 
-	msgRxL := f.l.With(
-		"vpnAddrs", vpnAddrs,
-		"from", via,
-		"certName", certName,
-		"certVersion", certVersion,
-		"fingerprint", fingerprint,
-		"issuer", issuer,
-		"initiatorIndex", hs.Details.InitiatorIndex,
-		"responderIndex", hs.Details.ResponderIndex,
-		"remoteIndex", h.RemoteIndex,
-		"handshake", m{"stage": 1, "style": "ix_psk0"},
-	)
+	msgRxL := f.l.WithFields(m{
+		"vpnAddrs":       vpnAddrs,
+		"from":           via,
+		"certName":       certName,
+		"certVersion":    certVersion,
+		"fingerprint":    fingerprint,
+		"issuer":         issuer,
+		"initiatorIndex": hs.Details.InitiatorIndex,
+		"responderIndex": hs.Details.ResponderIndex,
+		"remoteIndex":    h.RemoteIndex,
+		"handshake":      m{"stage": 1, "style": "ix_psk0"},
+	})
 
 	if anyVpnAddrsInCommon {
 		msgRxL.Info("Handshake message received")
@@ -324,9 +272,8 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	hs.Details.ResponderIndex = myIndex
 	hs.Details.Cert = cs.getHandshakeBytes(ci.myCert.Version())
 	if hs.Details.Cert == nil {
-		msgRxL.Error("Unable to handshake with host because no certificate handshake bytes is available",
-			"myCertVersion", ci.myCert.Version(),
-		)
+		msgRxL.WithField("myCertVersion", ci.myCert.Version()).
+			Error("Unable to handshake with host because no certificate handshake bytes is available")
 		return
 	}
 
@@ -336,43 +283,32 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 
 	hsBytes, err := hs.Marshal()
 	if err != nil {
-		f.l.Error("Failed to marshal handshake message",
-			"error", err,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Failed to marshal handshake message")
 		return
 	}
 
 	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.Error("Failed to call noise.WriteMessage",
-			"error", err,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			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.Error("Noise did not arrive at a key",
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		f.l.WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			WithField("fingerprint", fingerprint).
+			WithField("issuer", issuer).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).Error("Noise did not arrive at a key")
 		return
 	}
 
@@ -414,20 +350,13 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 			if !via.IsRelayed {
 				err := f.outside.WriteTo(msg, via.UdpAddr)
 				if err != nil {
-					f.l.Error("Failed to send handshake message",
-						"vpnAddrs", existing.vpnAddrs,
-						"from", via,
-						"handshake", m{"stage": 2, "style": "ix_psk0"},
-						"cached", true,
-						"error", err,
-					)
+					f.l.WithField("vpnAddrs", existing.vpnAddrs).WithField("from", via).
+						WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
+						WithError(err).Error("Failed to send handshake message")
 				} else {
-					f.l.Info("Handshake message sent",
-						"vpnAddrs", existing.vpnAddrs,
-						"from", via,
-						"handshake", m{"stage": 2, "style": "ix_psk0"},
-						"cached", true,
-					)
+					f.l.WithField("vpnAddrs", existing.vpnAddrs).WithField("from", via).
+						WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
+						Info("Handshake message sent")
 				}
 				return
 			} else {
@@ -437,67 +366,50 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 				}
 				hostinfo.relayState.InsertRelayTo(via.relayHI.vpnAddrs[0])
 				f.SendVia(via.relayHI, via.relay, msg, make([]byte, 12), make([]byte, mtu), false)
-				f.l.Info("Handshake message sent",
-					"vpnAddrs", existing.vpnAddrs,
-					"relay", via.relayHI.vpnAddrs[0],
-					"handshake", m{"stage": 2, "style": "ix_psk0"},
-					"cached", true,
-				)
+				f.l.WithField("vpnAddrs", existing.vpnAddrs).WithField("relay", via.relayHI.vpnAddrs[0]).
+					WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("cached", true).
+					Info("Handshake message sent")
 				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.Info("Handshake too old",
-				"vpnAddrs", vpnAddrs,
-				"from", via,
-				"certName", certName,
-				"certVersion", certVersion,
-				"oldHandshakeTime", existing.lastHandshakeTime,
-				"newHandshakeTime", hostinfo.lastHandshakeTime,
-				"fingerprint", fingerprint,
-				"issuer", issuer,
-				"initiatorIndex", hs.Details.InitiatorIndex,
-				"responderIndex", hs.Details.ResponderIndex,
-				"remoteIndex", h.RemoteIndex,
-				"handshake", m{"stage": 1, "style": "ix_psk0"},
-			)
+			f.l.WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+				WithField("certName", certName).
+				WithField("certVersion", certVersion).
+				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.SendMessageToVpnAddr(header.Test, header.TestRequest, vpnAddrs[0], []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.Error("Failed to add HostInfo due to localIndex collision",
-				"vpnAddrs", vpnAddrs,
-				"from", via,
-				"certName", certName,
-				"certVersion", certVersion,
-				"fingerprint", fingerprint,
-				"issuer", issuer,
-				"initiatorIndex", hs.Details.InitiatorIndex,
-				"responderIndex", hs.Details.ResponderIndex,
-				"remoteIndex", h.RemoteIndex,
-				"handshake", m{"stage": 1, "style": "ix_psk0"},
-				"localIndex", hostinfo.localIndexId,
-				"collision", existing.vpnAddrs,
-			)
+			f.l.WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+				WithField("certName", certName).
+				WithField("certVersion", certVersion).
+				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", existing.vpnAddrs).
+				Error("Failed to add HostInfo due to localIndex collision")
 			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.Error("Failed to add HostInfo to HostMap",
-				"error", err,
-				"vpnAddrs", vpnAddrs,
-				"from", via,
-				"certName", certName,
-				"certVersion", certVersion,
-				"fingerprint", fingerprint,
-				"issuer", issuer,
-				"initiatorIndex", hs.Details.InitiatorIndex,
-				"responderIndex", hs.Details.ResponderIndex,
-				"remoteIndex", h.RemoteIndex,
-				"handshake", m{"stage": 1, "style": "ix_psk0"},
-			)
+			f.l.WithError(err).WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+				WithField("certName", certName).
+				WithField("certVersion", certVersion).
+				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")
 			return
 		}
 	}
@@ -506,20 +418,15 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 	f.messageMetrics.Tx(header.Handshake, header.MessageSubType(msg[1]), 1)
 	if !via.IsRelayed {
 		err = f.outside.WriteTo(msg, via.UdpAddr)
-		log := f.l.With(
-			"vpnAddrs", vpnAddrs,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"initiatorIndex", hs.Details.InitiatorIndex,
-			"responderIndex", hs.Details.ResponderIndex,
-			"remoteIndex", h.RemoteIndex,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		log := f.l.WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			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"})
 		if err != nil {
-			log.Error("Failed to send handshake", "error", err)
+			log.WithError(err).Error("Failed to send handshake")
 		} else {
 			log.Info("Handshake message sent")
 		}
@@ -533,29 +440,20 @@ func ixHandshakeStage1(f *Interface, via ViaSender, packet []byte, h *header.H)
 		// it's correctly marked as working.
 		via.relayHI.relayState.UpdateRelayForByIdxState(via.remoteIdx, Established)
 		f.SendVia(via.relayHI, via.relay, msg, make([]byte, 12), make([]byte, mtu), false)
-		f.l.Info("Handshake message sent",
-			"vpnAddrs", vpnAddrs,
-			"relay", via.relayHI.vpnAddrs[0],
-			"certName", certName,
-			"certVersion", certVersion,
-			"fingerprint", fingerprint,
-			"issuer", issuer,
-			"initiatorIndex", hs.Details.InitiatorIndex,
-			"responderIndex", hs.Details.ResponderIndex,
-			"remoteIndex", h.RemoteIndex,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithField("vpnAddrs", vpnAddrs).WithField("relay", via.relayHI.vpnAddrs[0]).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			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"}).
+			Info("Handshake message sent")
 	}
 
 	f.connectionManager.AddTrafficWatch(hostinfo)
 
 	hostinfo.remotes.RefreshFromHandshake(vpnAddrs)
 
-	// Don't wait for UpdateWorker
-	if f.lightHouse.IsAnyLighthouseAddr(vpnAddrs) {
-		f.lightHouse.TriggerUpdate()
-	}
-
 	return
 }
 
@@ -572,12 +470,7 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	if !via.IsRelayed {
 		// The vpnAddr we know about is the one we tried to handshake with, use it to apply the remote allow list.
 		if !f.lightHouse.GetRemoteAllowList().AllowAll(hostinfo.vpnAddrs, via.UdpAddr.Addr()) {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				f.l.Debug("lighthouse.remote_allow_list denied incoming handshake",
-					"vpnAddrs", hostinfo.vpnAddrs,
-					"from", via,
-				)
-			}
+			f.l.WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).Debug("lighthouse.remote_allow_list denied incoming handshake")
 			return false
 		}
 	}
@@ -585,24 +478,18 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	ci := hostinfo.ConnectionState
 	msg, eKey, dKey, err := ci.H.ReadMessage(nil, packet[header.Len:])
 	if err != nil {
-		f.l.Error("Failed to call noise.ReadMessage",
-			"error", err,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-			"header", h,
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).WithField("header", h).
+			Error("Failed to call noise.ReadMessage")
 
 		// We don't want to tear down the connection on a bad ReadMessage because it could be an attacker trying
 		// to DOS us. Every other error condition after should to allow a possible good handshake to complete in the
 		// near future
 		return false
 	} else if dKey == nil || eKey == nil {
-		f.l.Error("Noise did not arrive at a key",
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+			Error("Noise did not arrive at a key")
 
 		// This should be impossible in IX but just in case, if we get here then there is no chance to recover
 		// the handshake state machine. Tear it down
@@ -612,12 +499,8 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	hs := &NebulaHandshake{}
 	err = hs.Unmarshal(msg)
 	if err != nil || hs.Details == nil {
-		f.l.Error("Failed unmarshal handshake message",
-			"error", err,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"from", via,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("vpnAddrs", hostinfo.vpnAddrs).WithField("from", via).
+			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
 		return true
@@ -625,12 +508,10 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 
 	rc, err := cert.Recombine(cert.Version(hs.Details.CertVersion), hs.Details.Cert, ci.H.PeerStatic(), ci.Curve())
 	if err != nil {
-		f.l.Info("Handshake did not contain a certificate",
-			"error", err,
-			"from", via,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+			Info("Handshake did not contain a certificate")
 		return true
 	}
 
@@ -641,41 +522,26 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 			fp = "<error generating certificate fingerprint>"
 		}
 
-		attrs := []slog.Attr{
-			slog.Any("error", err),
-			slog.Any("from", via),
-			slog.Any("vpnAddrs", hostinfo.vpnAddrs),
-			slog.Any("handshake", m{"stage": 2, "style": "ix_psk0"}),
-			slog.String("certFingerprint", fp),
-			slog.Any("certVpnNetworks", rc.Networks()),
-		}
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			attrs = append(attrs, slog.Any("cert", rc))
+		e := f.l.WithError(err).WithField("from", via).
+			WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+			WithField("certFingerprint", fp).
+			WithField("certVpnNetworks", rc.Networks())
+
+		if f.l.Level >= logrus.DebugLevel {
+			e = e.WithField("cert", rc)
 		}
 
-		// LogAttrs is intentional: attrs is a pre-built []slog.Attr slice that
-		// callers grow conditionally, which has no pair-form equivalent.
-		//nolint:sloglint
-		f.l.LogAttrs(context.Background(), slog.LevelInfo, "Invalid certificate from host", attrs...)
-		return true
-	}
-	if !bytes.Equal(remoteCert.Certificate.PublicKey(), ci.H.PeerStatic()) {
-		f.l.Info("public key mismatch between certificate and handshake",
-			"from", via,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-			"cert", remoteCert,
-		)
+		e.Info("Invalid certificate from host")
 		return true
 	}
 
 	if len(remoteCert.Certificate.Networks()) == 0 {
-		f.l.Info("No networks in certificate",
-			"error", err,
-			"from", via,
-			"vpnAddrs", hostinfo.vpnAddrs,
-			"cert", remoteCert,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithError(err).WithField("from", via).
+			WithField("vpnAddrs", hostinfo.vpnAddrs).
+			WithField("cert", remoteCert).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+			Info("No networks in certificate")
 		return true
 	}
 
@@ -716,14 +582,12 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 
 	// Ensure the right host responded
 	if !correctHostResponded {
-		f.l.Info("Incorrect host responded to handshake",
-			"intendedVpnAddrs", hostinfo.vpnAddrs,
-			"haveVpnNetworks", vpnNetworks,
-			"from", via,
-			"certName", certName,
-			"certVersion", certVersion,
-			"handshake", m{"stage": 2, "style": "ix_psk0"},
-		)
+		f.l.WithField("intendedVpnAddrs", hostinfo.vpnAddrs).WithField("haveVpnNetworks", vpnNetworks).
+			WithField("from", via).
+			WithField("certName", certName).
+			WithField("certVersion", certVersion).
+			WithField("handshake", m{"stage": 2, "style": "ix_psk0"}).
+			Info("Incorrect host responded to handshake")
 
 		// Release our old handshake from pending, it should not continue
 		f.handshakeManager.DeleteHostInfo(hostinfo)
@@ -735,11 +599,10 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 			newHH.hostinfo.remotes = hostinfo.remotes
 			newHH.hostinfo.remotes.BlockRemote(via)
 
-			f.l.Info("Blocked addresses for handshakes",
-				"blockedUdpAddrs", newHH.hostinfo.remotes.CopyBlockedRemotes(),
-				"vpnNetworks", vpnNetworks,
-				"remotes", newHH.hostinfo.remotes.CopyAddrs(f.hostMap.GetPreferredRanges()),
-			)
+			f.l.WithField("blockedUdpAddrs", newHH.hostinfo.remotes.CopyBlockedRemotes()).
+				WithField("vpnNetworks", vpnNetworks).
+				WithField("remotes", newHH.hostinfo.remotes.CopyAddrs(f.hostMap.GetPreferredRanges())).
+				Info("Blocked addresses for handshakes")
 
 			// Swap the packet store to benefit the original intended recipient
 			newHH.packetStore = hh.packetStore
@@ -757,20 +620,15 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	ci.window.Update(f.l, 2)
 
 	duration := time.Since(hh.startTime).Nanoseconds()
-	msgRxL := f.l.With(
-		"vpnAddrs", vpnAddrs,
-		"from", via,
-		"certName", certName,
-		"certVersion", certVersion,
-		"fingerprint", fingerprint,
-		"issuer", issuer,
-		"initiatorIndex", hs.Details.InitiatorIndex,
-		"responderIndex", hs.Details.ResponderIndex,
-		"remoteIndex", h.RemoteIndex,
-		"handshake", m{"stage": 2, "style": "ix_psk0"},
-		"durationNs", duration,
-		"sentCachedPackets", len(hh.packetStore),
-	)
+	msgRxL := f.l.WithField("vpnAddrs", vpnAddrs).WithField("from", via).
+		WithField("certName", certName).
+		WithField("certVersion", certVersion).
+		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).
+		WithField("sentCachedPackets", len(hh.packetStore))
 	if anyVpnAddrsInCommon {
 		msgRxL.Info("Handshake message received")
 	} else {
@@ -786,10 +644,8 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	f.handshakeManager.Complete(hostinfo, f)
 	f.connectionManager.AddTrafficWatch(hostinfo)
 
-	if f.l.Enabled(context.Background(), slog.LevelDebug) {
-		hostinfo.logger(f.l).Debug("Sending stored packets",
-			"count", len(hh.packetStore),
-		)
+	if f.l.Level >= logrus.DebugLevel {
+		hostinfo.logger(f.l).Debugf("Sending %d stored packets", len(hh.packetStore))
 	}
 
 	if len(hh.packetStore) > 0 {
@@ -804,10 +660,5 @@ func ixHandshakeStage2(f *Interface, via ViaSender, hh *HandshakeHostInfo, packe
 	hostinfo.remotes.RefreshFromHandshake(vpnAddrs)
 	f.metricHandshakes.Update(duration)
 
-	// Don't wait for UpdateWorker
-	if f.lightHouse.IsAnyLighthouseAddr(vpnAddrs) {
-		f.lightHouse.TriggerUpdate()
-	}
-
 	return false
 }

handshake_manager.go

@@ -6,13 +6,13 @@ import (
 	"crypto/rand"
 	"encoding/binary"
 	"errors"
-	"log/slog"
 	"net/netip"
 	"slices"
 	"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/udp"
@@ -59,7 +59,7 @@ type HandshakeManager struct {
 	metricInitiated        metrics.Counter
 	metricTimedOut         metrics.Counter
 	f                      *Interface
-	l                      *slog.Logger
+	l                      *logrus.Logger
 
 	// can be used to trigger outbound handshake for the given vpnIp
 	trigger chan netip.Addr
@@ -78,32 +78,32 @@ type HandshakeHostInfo struct {
 	hostinfo *HostInfo
 }
 
-func (hh *HandshakeHostInfo) cachePacket(l *slog.Logger, t header.MessageType, st header.MessageSubType, packet []byte, f packetCallback, m *cachedPacketMetrics) {
+func (hh *HandshakeHostInfo) cachePacket(l *logrus.Logger, t header.MessageType, st header.MessageSubType, packet []byte, f packetCallback, m *cachedPacketMetrics) {
 	if len(hh.packetStore) < 100 {
 		tempPacket := make([]byte, len(packet))
 		copy(tempPacket, packet)
 
 		hh.packetStore = append(hh.packetStore, &cachedPacket{t, st, f, tempPacket})
-		if l.Enabled(context.Background(), slog.LevelDebug) {
-			hh.hostinfo.logger(l).Debug("Packet store",
-				"length", len(hh.packetStore),
-				"stored", true,
-			)
+		if l.Level >= logrus.DebugLevel {
+			hh.hostinfo.logger(l).
+				WithField("length", len(hh.packetStore)).
+				WithField("stored", true).
+				Debugf("Packet store")
 		}
 
 	} else {
 		m.dropped.Inc(1)
 
-		if l.Enabled(context.Background(), slog.LevelDebug) {
-			hh.hostinfo.logger(l).Debug("Packet store",
-				"length", len(hh.packetStore),
-				"stored", false,
-			)
+		if l.Level >= logrus.DebugLevel {
+			hh.hostinfo.logger(l).
+				WithField("length", len(hh.packetStore)).
+				WithField("stored", false).
+				Debugf("Packet store")
 		}
 	}
 }
 
-func NewHandshakeManager(l *slog.Logger, mainHostMap *HostMap, lightHouse *LightHouse, outside udp.Conn, config HandshakeConfig) *HandshakeManager {
+func NewHandshakeManager(l *logrus.Logger, mainHostMap *HostMap, lightHouse *LightHouse, outside udp.Conn, config HandshakeConfig) *HandshakeManager {
 	return &HandshakeManager{
 		vpnIps:                 map[netip.Addr]*HandshakeHostInfo{},
 		indexes:                map[uint32]*HandshakeHostInfo{},
@@ -140,7 +140,7 @@ func (hm *HandshakeManager) HandleIncoming(via ViaSender, packet []byte, h *head
 	// First remote allow list check before we know the vpnIp
 	if !via.IsRelayed {
 		if !hm.lightHouse.GetRemoteAllowList().AllowUnknownVpnAddr(via.UdpAddr.Addr()) {
-			hm.l.Debug("lighthouse.remote_allow_list denied incoming handshake", "from", via)
+			hm.l.WithField("from", via).Debug("lighthouse.remote_allow_list denied incoming handshake")
 			return
 		}
 	}
@@ -183,13 +183,12 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 	hostinfo := hh.hostinfo
 	// If we are out of time, clean up
 	if hh.counter >= hm.config.retries {
-		hh.hostinfo.logger(hm.l).Info("Handshake timed out",
-			"udpAddrs", hh.hostinfo.remotes.CopyAddrs(hm.mainHostMap.GetPreferredRanges()),
-			"initiatorIndex", hh.hostinfo.localIndexId,
-			"remoteIndex", hh.hostinfo.remoteIndexId,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-			"durationNs", time.Since(hh.startTime).Nanoseconds(),
-		)
+		hh.hostinfo.logger(hm.l).WithField("udpAddrs", hh.hostinfo.remotes.CopyAddrs(hm.mainHostMap.GetPreferredRanges())).
+			WithField("initiatorIndex", hh.hostinfo.localIndexId).
+			WithField("remoteIndex", hh.hostinfo.remoteIndexId).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			WithField("durationNs", time.Since(hh.startTime).Nanoseconds()).
+			Info("Handshake timed out")
 		hm.metricTimedOut.Inc(1)
 		hm.DeleteHostInfo(hostinfo)
 		return
@@ -242,12 +241,10 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 		hm.messageMetrics.Tx(header.Handshake, header.MessageSubType(hostinfo.HandshakePacket[0][1]), 1)
 		err := hm.outside.WriteTo(hostinfo.HandshakePacket[0], addr)
 		if err != nil {
-			hostinfo.logger(hm.l).Error("Failed to send handshake message",
-				"udpAddr", addr,
-				"initiatorIndex", hostinfo.localIndexId,
-				"handshake", m{"stage": 1, "style": "ix_psk0"},
-				"error", err,
-			)
+			hostinfo.logger(hm.l).WithField("udpAddr", addr).
+				WithField("initiatorIndex", hostinfo.localIndexId).
+				WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+				WithError(err).Error("Failed to send handshake message")
 
 		} else {
 			sentTo = append(sentTo, addr)
@@ -257,21 +254,19 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 	// 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(hm.l).Info("Handshake message sent",
-			"udpAddrs", sentTo,
-			"initiatorIndex", hostinfo.localIndexId,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
-	} else if hm.l.Enabled(context.Background(), slog.LevelDebug) {
-		hostinfo.logger(hm.l).Debug("Handshake message sent",
-			"udpAddrs", sentTo,
-			"initiatorIndex", hostinfo.localIndexId,
-			"handshake", m{"stage": 1, "style": "ix_psk0"},
-		)
+		hostinfo.logger(hm.l).WithField("udpAddrs", sentTo).
+			WithField("initiatorIndex", hostinfo.localIndexId).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Info("Handshake message sent")
+	} else if hm.l.Level >= logrus.DebugLevel {
+		hostinfo.logger(hm.l).WithField("udpAddrs", sentTo).
+			WithField("initiatorIndex", hostinfo.localIndexId).
+			WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).
+			Debug("Handshake message sent")
 	}
 
 	if hm.config.useRelays && len(hostinfo.remotes.relays) > 0 {
-		hostinfo.logger(hm.l).Info("Attempt to relay through hosts", "relays", hostinfo.remotes.relays)
+		hostinfo.logger(hm.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 through the host I'm trying to connect to
@@ -286,7 +281,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 
 			relayHostInfo := hm.mainHostMap.QueryVpnAddr(relay)
 			if relayHostInfo == nil || !relayHostInfo.remote.IsValid() {
-				hostinfo.logger(hm.l).Info("Establish tunnel to relay target", "relay", relay.String())
+				hostinfo.logger(hm.l).WithField("relay", relay.String()).Info("Establish tunnel to relay target")
 				hm.f.Handshake(relay)
 				continue
 			}
@@ -297,7 +292,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 				if relayHostInfo.remote.IsValid() {
 					idx, err := AddRelay(hm.l, relayHostInfo, hm.mainHostMap, vpnIp, nil, TerminalType, Requested)
 					if err != nil {
-						hostinfo.logger(hm.l).Info("Failed to add relay to hostmap", "relay", relay.String(), "error", err)
+						hostinfo.logger(hm.l).WithField("relay", relay.String()).WithError(err).Info("Failed to add relay to hostmap")
 					}
 
 					m := NebulaControl{
@@ -331,15 +326,17 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 
 					msg, err := m.Marshal()
 					if err != nil {
-						hostinfo.logger(hm.l).Error("Failed to marshal Control message to create relay", "error", err)
+						hostinfo.logger(hm.l).
+							WithError(err).
+							Error("Failed to marshal Control message to create relay")
 					} else {
 						hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
-						hm.l.Info("send CreateRelayRequest",
-							"relayFrom", hm.f.myVpnAddrs[0],
-							"relayTo", vpnIp,
-							"initiatorRelayIndex", idx,
-							"relay", relay,
-						)
+						hm.l.WithFields(logrus.Fields{
+							"relayFrom":           hm.f.myVpnAddrs[0],
+							"relayTo":             vpnIp,
+							"initiatorRelayIndex": idx,
+							"relay":               relay}).
+							Info("send CreateRelayRequest")
 					}
 				}
 				continue
@@ -347,14 +344,14 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 
 			switch existingRelay.State {
 			case Established:
-				hostinfo.logger(hm.l).Info("Send handshake via relay", "relay", relay.String())
+				hostinfo.logger(hm.l).WithField("relay", relay.String()).Info("Send handshake via relay")
 				hm.f.SendVia(relayHostInfo, existingRelay, hostinfo.HandshakePacket[0], make([]byte, 12), make([]byte, mtu), false)
 			case Disestablished:
 				// Mark this relay as 'requested'
 				relayHostInfo.relayState.UpdateRelayForByIpState(vpnIp, Requested)
 				fallthrough
 			case Requested:
-				hostinfo.logger(hm.l).Info("Re-send CreateRelay request", "relay", relay.String())
+				hostinfo.logger(hm.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,
@@ -386,26 +383,28 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 				}
 				msg, err := m.Marshal()
 				if err != nil {
-					hostinfo.logger(hm.l).Error("Failed to marshal Control message to create relay", "error", err)
+					hostinfo.logger(hm.l).
+						WithError(err).
+						Error("Failed to marshal Control message to create relay")
 				} else {
 					// This must send over the hostinfo, not over hm.Hosts[ip]
 					hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
-					hm.l.Info("send CreateRelayRequest",
-						"relayFrom", hm.f.myVpnAddrs[0],
-						"relayTo", vpnIp,
-						"initiatorRelayIndex", existingRelay.LocalIndex,
-						"relay", relay,
-					)
+					hm.l.WithFields(logrus.Fields{
+						"relayFrom":           hm.f.myVpnAddrs[0],
+						"relayTo":             vpnIp,
+						"initiatorRelayIndex": existingRelay.LocalIndex,
+						"relay":               relay}).
+						Info("send CreateRelayRequest")
 				}
 			case PeerRequested:
 				// PeerRequested only occurs in Forwarding relays, not Terminal relays, and this is a Terminal relay case.
 				fallthrough
 			default:
-				hostinfo.logger(hm.l).Error("Relay unexpected state",
-					"vpnIp", vpnIp,
-					"state", existingRelay.State,
-					"relay", relay,
-				)
+				hostinfo.logger(hm.l).
+					WithField("vpnIp", vpnIp).
+					WithField("state", existingRelay.State).
+					WithField("relay", relay).
+					Errorf("Relay unexpected state")
 
 			}
 		}
@@ -550,10 +549,9 @@ func (hm *HandshakeManager) CheckAndComplete(hostinfo *HostInfo, handshakePacket
 	if found && existingRemoteIndex != nil && existingRemoteIndex.vpnAddrs[0] != hostinfo.vpnAddrs[0] {
 		// 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(hm.l).Info("New host shadows existing host remoteIndex",
-			"remoteIndex", hostinfo.remoteIndexId,
-			"collision", existingRemoteIndex.vpnAddrs,
-		)
+		hostinfo.logger(hm.l).
+			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", existingRemoteIndex.vpnAddrs).
+			Info("New host shadows existing host remoteIndex")
 	}
 
 	hm.mainHostMap.unlockedAddHostInfo(hostinfo, f)
@@ -573,10 +571,9 @@ func (hm *HandshakeManager) Complete(hostinfo *HostInfo, f *Interface) {
 	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(hm.l).Info("New host shadows existing host remoteIndex",
-			"remoteIndex", hostinfo.remoteIndexId,
-			"collision", existingRemoteIndex.vpnAddrs,
-		)
+		hostinfo.logger(hm.l).
+			WithField("remoteIndex", hostinfo.remoteIndexId).WithField("collision", existingRemoteIndex.vpnAddrs).
+			Info("New host shadows existing host remoteIndex")
 	}
 
 	// We need to remove from the pending hostmap first to avoid undoing work when after to the main hostmap.
@@ -593,7 +590,7 @@ func (hm *HandshakeManager) allocateIndex(hh *HandshakeHostInfo) error {
 	hm.Lock()
 	defer hm.Unlock()
 
-	for range 32 {
+	for i := 0; i < 32; i++ {
 		index, err := generateIndex(hm.l)
 		if err != nil {
 			return err
@@ -632,11 +629,10 @@ func (hm *HandshakeManager) unlockedDeleteHostInfo(hostinfo *HostInfo) {
 		hm.indexes = map[uint32]*HandshakeHostInfo{}
 	}
 
-	if hm.l.Enabled(context.Background(), slog.LevelDebug) {
-		hm.l.Debug("Pending hostmap hostInfo deleted",
-			"hostMap", m{"mapTotalSize": len(hm.vpnIps),
-				"vpnAddrs": hostinfo.vpnAddrs, "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId},
-		)
+	if hm.l.Level >= logrus.DebugLevel {
+		hm.l.WithField("hostMap", m{"mapTotalSize": len(hm.vpnIps),
+			"vpnAddrs": hostinfo.vpnAddrs, "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId}).
+			Debug("Pending hostmap hostInfo deleted")
 	}
 }
 
@@ -704,7 +700,7 @@ func (hm *HandshakeManager) EmitStats() {
 
 // Utility functions below
 
-func generateIndex(l *slog.Logger) (uint32, error) {
+func generateIndex(l *logrus.Logger) (uint32, error) {
 	b := make([]byte, 4)
 
 	// Let zero mean we don't know the ID, so don't generate zero
@@ -712,15 +708,16 @@ func generateIndex(l *slog.Logger) (uint32, error) {
 	for index == 0 {
 		_, err := rand.Read(b)
 		if err != nil {
-			l.Error("Failed to generate index", "error", err)
+			l.Errorln(err)
 			return 0, err
 		}
 
 		index = binary.BigEndian.Uint32(b)
 	}
 
-	if l.Enabled(context.Background(), slog.LevelDebug) {
-		l.Debug("Generated index", "index", index)
+	if l.Level >= logrus.DebugLevel {
+		l.WithField("index", index).
+			Debug("Generated index")
 	}
 	return index, nil
 }

hostmap.go

@@ -1,11 +1,9 @@
 package nebula
 
 import (
-	"context"
 	"encoding/json"
 	"errors"
 	"fmt"
-	"log/slog"
 	"net"
 	"net/netip"
 	"slices"
@@ -15,10 +13,10 @@ import (
 
 	"github.com/gaissmai/bart"
 	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
-	"github.com/slackhq/nebula/logging"
 )
 
 const defaultPromoteEvery = 1000       // Count of packets sent before we try moving a tunnel to a preferred underlay ip address
@@ -62,7 +60,7 @@ type HostMap struct {
 	RemoteIndexes   map[uint32]*HostInfo
 	Hosts           map[netip.Addr]*HostInfo
 	preferredRanges atomic.Pointer[[]netip.Prefix]
-	l               *slog.Logger
+	l               *logrus.Logger
 }
 
 // For synchronization, treat the pointed-to Relay struct as immutable. To edit the Relay
@@ -315,7 +313,7 @@ type cachedPacketMetrics struct {
 	dropped metrics.Counter
 }
 
-func NewHostMapFromConfig(l *slog.Logger, c *config.C) *HostMap {
+func NewHostMapFromConfig(l *logrus.Logger, c *config.C) *HostMap {
 	hm := newHostMap(l)
 
 	hm.reload(c, true)
@@ -323,12 +321,13 @@ func NewHostMapFromConfig(l *slog.Logger, c *config.C) *HostMap {
 		hm.reload(c, false)
 	})
 
-	l.Info("Main HostMap created", "preferredRanges", hm.GetPreferredRanges())
+	l.WithField("preferredRanges", hm.GetPreferredRanges()).
+		Info("Main HostMap created")
 
 	return hm
 }
 
-func newHostMap(l *slog.Logger) *HostMap {
+func newHostMap(l *logrus.Logger) *HostMap {
 	return &HostMap{
 		Indexes:       map[uint32]*HostInfo{},
 		Relays:        map[uint32]*HostInfo{},
@@ -347,10 +346,7 @@ func (hm *HostMap) reload(c *config.C, initial bool) {
 			preferredRange, err := netip.ParsePrefix(rawPreferredRange)
 
 			if err != nil {
-				hm.l.Warn("Failed to parse preferred ranges, ignoring",
-					"error", err,
-					"range", rawPreferredRanges,
-				)
+				hm.l.WithError(err).WithField("range", rawPreferredRanges).Warn("Failed to parse preferred ranges, ignoring")
 				continue
 			}
 
@@ -359,10 +355,7 @@ func (hm *HostMap) reload(c *config.C, initial bool) {
 
 		oldRanges := hm.preferredRanges.Swap(&preferredRanges)
 		if !initial {
-			hm.l.Info("preferred_ranges changed",
-				"oldPreferredRanges", *oldRanges,
-				"newPreferredRanges", preferredRanges,
-			)
+			hm.l.WithField("oldPreferredRanges", *oldRanges).WithField("newPreferredRanges", preferredRanges).Info("preferred_ranges changed")
 		}
 	}
 }
@@ -495,11 +488,10 @@ func (hm *HostMap) unlockedInnerDeleteHostInfo(hostinfo *HostInfo, addr netip.Ad
 		hm.Indexes = map[uint32]*HostInfo{}
 	}
 
-	if hm.l.Enabled(context.Background(), slog.LevelDebug) {
-		hm.l.Debug("Hostmap hostInfo deleted",
-			"hostMap", m{"mapTotalSize": len(hm.Hosts),
-				"vpnAddrs": hostinfo.vpnAddrs, "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId},
-		)
+	if hm.l.Level >= logrus.DebugLevel {
+		hm.l.WithField("hostMap", m{"mapTotalSize": len(hm.Hosts),
+			"vpnAddrs": hostinfo.vpnAddrs, "indexNumber": hostinfo.localIndexId, "remoteIndexNumber": hostinfo.remoteIndexId}).
+			Debug("Hostmap hostInfo deleted")
 	}
 
 	if isLastHostinfo {
@@ -612,9 +604,9 @@ func (hm *HostMap) queryVpnAddr(vpnIp netip.Addr, promoteIfce *Interface) *HostI
 // 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.dnsServer != nil {
+	if f.serveDns {
 		remoteCert := hostinfo.ConnectionState.peerCert
-		f.dnsServer.Add(remoteCert.Certificate.Name()+".", hostinfo.vpnAddrs)
+		dnsR.Add(remoteCert.Certificate.Name()+".", hostinfo.vpnAddrs)
 	}
 	for _, addr := range hostinfo.vpnAddrs {
 		hm.unlockedInnerAddHostInfo(addr, hostinfo, f)
@@ -623,11 +615,10 @@ func (hm *HostMap) unlockedAddHostInfo(hostinfo *HostInfo, f *Interface) {
 	hm.Indexes[hostinfo.localIndexId] = hostinfo
 	hm.RemoteIndexes[hostinfo.remoteIndexId] = hostinfo
 
-	if hm.l.Enabled(context.Background(), slog.LevelDebug) {
-		hm.l.Debug("Hostmap vpnIp added",
-			"hostMap", m{"vpnAddrs": hostinfo.vpnAddrs, "mapTotalSize": len(hm.Hosts),
-				"hostinfo": m{"existing": true, "localIndexId": hostinfo.localIndexId, "vpnAddrs": hostinfo.vpnAddrs}},
-		)
+	if hm.l.Level >= logrus.DebugLevel {
+		hm.l.WithField("hostMap", m{"vpnAddrs": hostinfo.vpnAddrs, "mapTotalSize": len(hm.Hosts),
+			"hostinfo": m{"existing": true, "localIndexId": hostinfo.localIndexId, "vpnAddrs": hostinfo.vpnAddrs}}).
+			Debug("Hostmap vpnIp added")
 	}
 }
 
@@ -793,21 +784,18 @@ func (i *HostInfo) buildNetworks(myVpnNetworksTable *bart.Lite, c cert.Certifica
 	}
 }
 
-// logger returns a derived slog.Logger with per-hostinfo fields pre-bound.
-func (i *HostInfo) logger(l *slog.Logger) *slog.Logger {
+func (i *HostInfo) logger(l *logrus.Logger) *logrus.Entry {
 	if i == nil {
-		return l
+		return logrus.NewEntry(l)
 	}
 
-	li := l.With(
-		"vpnAddrs", i.vpnAddrs,
-		"localIndex", i.localIndexId,
-		"remoteIndex", i.remoteIndexId,
-	)
+	li := l.WithField("vpnAddrs", i.vpnAddrs).
+		WithField("localIndex", i.localIndexId).
+		WithField("remoteIndex", i.remoteIndexId)
 
 	if connState := i.ConnectionState; connState != nil {
 		if peerCert := connState.peerCert; peerCert != nil {
-			li = li.With("certName", peerCert.Certificate.Name())
+			li = li.WithField("certName", peerCert.Certificate.Name())
 		}
 	}
 
@@ -816,17 +804,14 @@ func (i *HostInfo) logger(l *slog.Logger) *slog.Logger {
 
 // Utility functions
 
-func localAddrs(l *slog.Logger, allowList *LocalAllowList) []netip.Addr {
+func localAddrs(l *logrus.Logger, allowList *LocalAllowList) []netip.Addr {
 	//FIXME: This function is pretty garbage
 	var finalAddrs []netip.Addr
 	ifaces, _ := net.Interfaces()
 	for _, i := range ifaces {
 		allow := allowList.AllowName(i.Name)
-		if l.Enabled(context.Background(), logging.LevelTrace) {
-			l.Log(context.Background(), logging.LevelTrace, "localAllowList.AllowName",
-				"interfaceName", i.Name,
-				"allow", allow,
-			)
+		if l.Level >= logrus.TraceLevel {
+			l.WithField("interfaceName", i.Name).WithField("allow", allow).Trace("localAllowList.AllowName")
 		}
 
 		if !allow {
@@ -844,8 +829,8 @@ func localAddrs(l *slog.Logger, allowList *LocalAllowList) []netip.Addr {
 			}
 
 			if !addr.IsValid() {
-				if l.Enabled(context.Background(), slog.LevelDebug) {
-					l.Debug("addr was invalid", "localAddr", rawAddr)
+				if l.Level >= logrus.DebugLevel {
+					l.WithField("localAddr", rawAddr).Debug("addr was invalid")
 				}
 				continue
 			}
@@ -853,11 +838,8 @@ func localAddrs(l *slog.Logger, allowList *LocalAllowList) []netip.Addr {
 
 			if addr.IsLoopback() == false && addr.IsLinkLocalUnicast() == false {
 				isAllowed := allowList.Allow(addr)
-				if l.Enabled(context.Background(), logging.LevelTrace) {
-					l.Log(context.Background(), logging.LevelTrace, "localAllowList.Allow",
-						"localAddr", addr,
-						"allowed", isAllowed,
-					)
+				if l.Level >= logrus.TraceLevel {
+					l.WithField("localAddr", addr).WithField("allowed", isAllowed).Trace("localAllowList.Allow")
 				}
 				if !isAllowed {
 					continue

hostmap_test.go

@@ -196,7 +196,7 @@ func TestHostMap_DeleteHostInfo(t *testing.T) {
 
 func TestHostMap_reload(t *testing.T) {
 	l := test.NewLogger()
-	c := config.NewC(test.NewLogger())
+	c := config.NewC(l)
 
 	hm := NewHostMapFromConfig(l, c)
 

hostmap_tester.go

@@ -1,4 +1,5 @@
 //go:build e2e_testing
+// +build e2e_testing
 
 package nebula
 

inside.go

@@ -1,26 +1,23 @@
 package nebula
 
 import (
-	"context"
-	"log/slog"
 	"net/netip"
+	"time"
 
+	"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/overlay/batch"
+	"github.com/slackhq/nebula/packet"
 	"github.com/slackhq/nebula/routing"
 )
 
-func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet, nb []byte, sendBatch batch.TxBatcher, rejectBuf []byte, q int, localCache firewall.ConntrackCache) {
+func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet, nb []byte, out *packet.UDPPacket, q int, localCache firewall.ConntrackCache, now time.Time) {
 	err := newPacket(packet, false, fwPacket)
 	if err != nil {
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("Error while validating outbound packet",
-				"packet", packet,
-				"error", err,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("packet", packet).Debugf("Error while validating outbound packet: %s", err)
 		}
 		return
 	}
@@ -40,7 +37,7 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 		if immediatelyForwardToSelf {
 			_, err := f.readers[q].Write(packet)
 			if err != nil {
-				f.l.Error("Failed to forward to tun", "error", err)
+				f.l.WithError(err).Error("Failed to forward to tun")
 			}
 		}
 		// Otherwise, drop. On linux, we should never see these packets - Linux
@@ -58,12 +55,11 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 	})
 
 	if hostinfo == nil {
-		f.rejectInside(packet, rejectBuf, q)
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("dropping outbound packet, vpnAddr not in our vpn networks or in unsafe networks",
-				"vpnAddr", fwPacket.RemoteAddr,
-				"fwPacket", fwPacket,
-			)
+		f.rejectInside(packet, out.Payload, q) //todo vector?
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("vpnAddr", fwPacket.RemoteAddr).
+				WithField("fwPacket", fwPacket).
+				Debugln("dropping outbound packet, vpnAddr not in our vpn networks or in unsafe networks")
 		}
 		return
 	}
@@ -72,81 +68,20 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 		return
 	}
 
-	dropReason := f.firewall.Drop(*fwPacket, false, hostinfo, f.pki.GetCAPool(), localCache)
+	dropReason := f.firewall.Drop(*fwPacket, false, hostinfo, f.pki.GetCAPool(), localCache, now)
 	if dropReason == nil {
-		f.sendInsideMessage(hostinfo, packet, nb, sendBatch, rejectBuf, q)
+		f.sendNoMetricsDelayed(header.Message, 0, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, packet, nb, out, q)
 	} else {
-		f.rejectInside(packet, rejectBuf, q)
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("dropping outbound packet",
-				"fwPacket", fwPacket,
-				"reason", dropReason,
-			)
+		f.rejectInside(packet, out.Payload, q) //todo vector?
+		if f.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(f.l).
+				WithField("fwPacket", fwPacket).
+				WithField("reason", dropReason).
+				Debugln("dropping outbound packet")
 		}
 	}
 }
 
-// sendInsideMessage encrypts a firewall-approved inside packet into the
-// caller's batch slot for later sendmmsg flush. When hostinfo.remote is not
-// valid we fall through to the relay slow path via the unbatched sendNoMetrics
-// so relay behavior is unchanged.
-func (f *Interface) sendInsideMessage(hostinfo *HostInfo, p, nb []byte, sendBatch batch.TxBatcher, rejectBuf []byte, q int) {
-	ci := hostinfo.ConnectionState
-	if ci.eKey == nil {
-		return
-	}
-
-	if !hostinfo.remote.IsValid() {
-		// Slow path: relay fallback. Reuse rejectBuf as the ciphertext
-		// scratch; sendNoMetrics arranges header space for SendVia.
-		f.sendNoMetrics(header.Message, 0, ci, hostinfo, netip.AddrPort{}, p, nb, rejectBuf, q)
-		return
-	}
-
-	scratch := sendBatch.Next()
-	if scratch == nil {
-		// Batch full: bypass batching and send this packet directly so we
-		// never drop traffic on over-subscribed iterations.
-		f.sendNoMetrics(header.Message, 0, ci, hostinfo, netip.AddrPort{}, p, nb, rejectBuf, q)
-		return
-	}
-
-	if noiseutil.EncryptLockNeeded {
-		ci.writeLock.Lock()
-	}
-	c := ci.messageCounter.Add(1)
-
-	out := header.Encode(scratch, header.Version, header.Message, 0, hostinfo.remoteIndexId, c)
-	f.connectionManager.Out(hostinfo)
-
-	if 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.vpnAddrs[0])
-		hostinfo.lastRebindCount = f.rebindCount
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("Lighthouse update triggered for punch due to rebind counter",
-				"vpnAddrs", hostinfo.vpnAddrs,
-			)
-		}
-	}
-
-	out, err := ci.eKey.EncryptDanger(out, out, p, c, nb)
-	if noiseutil.EncryptLockNeeded {
-		ci.writeLock.Unlock()
-	}
-	if err != nil {
-		hostinfo.logger(f.l).Error("Failed to encrypt outgoing packet",
-			"error", err,
-			"udpAddr", hostinfo.remote,
-			"counter", c,
-		)
-		return
-	}
-
-	sendBatch.Commit(len(out), hostinfo.remote)
-}
-
 func (f *Interface) rejectInside(packet []byte, out []byte, q int) {
 	if !f.firewall.InSendReject {
 		return
@@ -159,7 +94,7 @@ func (f *Interface) rejectInside(packet []byte, out []byte, q int) {
 
 	_, err := f.readers[q].Write(out)
 	if err != nil {
-		f.l.Error("Failed to write to tun", "error", err)
+		f.l.WithError(err).Error("Failed to write to tun")
 	}
 }
 
@@ -174,11 +109,11 @@ func (f *Interface) rejectOutside(packet []byte, ci *ConnectionState, hostinfo *
 	}
 
 	if len(out) > iputil.MaxRejectPacketSize {
-		if f.l.Enabled(context.Background(), slog.LevelInfo) {
-			f.l.Info("rejectOutside: packet too big, not sending",
-				"packet", packet,
-				"outPacket", out,
-			)
+		if f.l.GetLevel() >= logrus.InfoLevel {
+			f.l.
+				WithField("packet", packet).
+				WithField("outPacket", out).
+				Info("rejectOutside: packet too big, not sending")
 		}
 		return
 	}
@@ -250,11 +185,10 @@ func (f *Interface) getOrHandshakeConsiderRouting(fwPacket *firewall.Packet, cac
 		// This would also need to interact with unsafe_route updates through reloading the config or
 		// use of the use_system_route_table option
 
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("Calculated gateway for ECMP not available, attempting other gateways",
-				"destination", destinationAddr,
-				"originalGateway", gatewayAddr,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("destination", destinationAddr).
+				WithField("originalGateway", gatewayAddr).
+				Debugln("Calculated gateway for ECMP not available, attempting other gateways")
 		}
 
 		for i := range gateways {
@@ -280,18 +214,17 @@ func (f *Interface) sendMessageNow(t header.MessageType, st header.MessageSubTyp
 	fp := &firewall.Packet{}
 	err := newPacket(p, false, fp)
 	if err != nil {
-		f.l.Warn("error while parsing outgoing packet for firewall check", "error", err)
+		f.l.Warnf("error while parsing outgoing packet for firewall check; %v", err)
 		return
 	}
 
 	// check if packet is in outbound fw rules
-	dropReason := f.firewall.Drop(*fp, false, hostinfo, f.pki.GetCAPool(), nil)
+	dropReason := f.firewall.Drop(*fp, false, hostinfo, f.pki.GetCAPool(), nil, time.Now())
 	if dropReason != nil {
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("dropping cached packet",
-				"fwPacket", fp,
-				"reason", dropReason,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("fwPacket", fp).
+				WithField("reason", dropReason).
+				Debugln("dropping cached packet")
 		}
 		return
 	}
@@ -307,10 +240,9 @@ func (f *Interface) SendMessageToVpnAddr(t header.MessageType, st header.Message
 	})
 
 	if hostInfo == nil {
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("dropping SendMessageToVpnAddr, vpnAddr not in our vpn networks or in unsafe routes",
-				"vpnAddr", vpnAddr,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("vpnAddr", vpnAddr).
+				Debugln("dropping SendMessageToVpnAddr, vpnAddr not in our vpn networks or in unsafe routes")
 		}
 		return
 	}
@@ -366,12 +298,12 @@ func (f *Interface) SendVia(via *HostInfo,
 		if noiseutil.EncryptLockNeeded {
 			via.ConnectionState.writeLock.Unlock()
 		}
-		via.logger(f.l).Error("SendVia out buffer not large enough for relay",
-			"outCap", cap(out),
-			"payloadLen", len(ad),
-			"headerLen", len(out),
-			"cipherOverhead", via.ConnectionState.eKey.Overhead(),
-		)
+		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
 	}
 
@@ -391,12 +323,12 @@ func (f *Interface) SendVia(via *HostInfo,
 		via.ConnectionState.writeLock.Unlock()
 	}
 	if err != nil {
-		via.logger(f.l).Info("Failed to EncryptDanger in sendVia", "error", err)
+		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).Info("Failed to WriteTo in sendVia", "error", err)
+		via.logger(f.l).WithError(err).Info("Failed to WriteTo in sendVia")
 	}
 	f.connectionManager.RelayUsed(relay.LocalIndex)
 }
@@ -435,10 +367,8 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 		// finally used again. This tunnel would eventually be torn down and recreated if this action didn't help.
 		f.lightHouse.QueryServer(hostinfo.vpnAddrs[0])
 		hostinfo.lastRebindCount = f.rebindCount
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			f.l.Debug("Lighthouse update triggered for punch due to rebind counter",
-				"vpnAddrs", hostinfo.vpnAddrs,
-			)
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("vpnAddrs", hostinfo.vpnAddrs).Debug("Lighthouse update triggered for punch due to rebind counter")
 		}
 	}
 
@@ -448,30 +378,24 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 		ci.writeLock.Unlock()
 	}
 	if err != nil {
-		hostinfo.logger(f.l).Error("Failed to encrypt outgoing packet",
-			"error", err,
-			"udpAddr", remote,
-			"counter", c,
-			"attemptedCounter", c,
-		)
+		hostinfo.logger(f.l).WithError(err).
+			WithField("udpAddr", remote).WithField("counter", c).
+			WithField("attemptedCounter", c).
+			Error("Failed to encrypt outgoing packet")
 		return
 	}
 
 	if remote.IsValid() {
 		err = f.writers[q].WriteTo(out, remote)
 		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to write outgoing packet",
-				"error", err,
-				"udpAddr", remote,
-			)
+			hostinfo.logger(f.l).WithError(err).
+				WithField("udpAddr", remote).Error("Failed to write outgoing packet")
 		}
 	} else if hostinfo.remote.IsValid() {
 		err = f.writers[q].WriteTo(out, hostinfo.remote)
 		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to write outgoing packet",
-				"error", err,
-				"udpAddr", remote,
-			)
+			hostinfo.logger(f.l).WithError(err).
+				WithField("udpAddr", remote).Error("Failed to write outgoing packet")
 		}
 	} else {
 		// Try to send via a relay
@@ -479,10 +403,7 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 			relayHostInfo, relay, err := f.hostMap.QueryVpnAddrsRelayFor(hostinfo.vpnAddrs, relayIP)
 			if err != nil {
 				hostinfo.relayState.DeleteRelay(relayIP)
-				hostinfo.logger(f.l).Info("sendNoMetrics failed to find HostInfo",
-					"relay", relayIP,
-					"error", err,
-				)
+				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)
@@ -490,3 +411,81 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 		}
 	}
 }
+
+func (f *Interface) sendNoMetricsDelayed(t header.MessageType, st header.MessageSubType, ci *ConnectionState, hostinfo *HostInfo, remote netip.AddrPort, p, nb []byte, out *packet.UDPPacket, q int) {
+	if ci.eKey == nil {
+		return
+	}
+	useRelay := !remote.IsValid() && !hostinfo.remote.IsValid()
+	fullOut := out.Payload
+
+	if useRelay {
+		if len(out.Payload) < 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.Payload = out.Payload[:header.Len]
+		}
+		// Save a header's worth of data at the front of the 'out' buffer.
+		out.Payload = out.Payload[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.Payload = header.Encode(out.Payload, header.Version, t, st, hostinfo.remoteIndexId, c)
+	f.connectionManager.Out(hostinfo)
+
+	// Query our LH if we haven't since the last time we've been rebound, this will cause the remote to punch against
+	// all our addrs and enable a faster roaming.
+	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.vpnAddrs[0])
+		hostinfo.lastRebindCount = f.rebindCount
+		if f.l.Level >= logrus.DebugLevel {
+			f.l.WithField("vpnAddrs", hostinfo.vpnAddrs).Debug("Lighthouse update triggered for punch due to rebind counter")
+		}
+	}
+
+	var err error
+	out.Payload, err = ci.eKey.EncryptDanger(out.Payload, out.Payload, p, c, nb)
+	if noiseutil.EncryptLockNeeded {
+		ci.writeLock.Unlock()
+	}
+	if err != nil {
+		hostinfo.logger(f.l).WithError(err).
+			WithField("udpAddr", remote).WithField("counter", c).
+			WithField("attemptedCounter", c).
+			Error("Failed to encrypt outgoing packet")
+		return
+	}
+
+	if remote.IsValid() {
+		err = f.writers[q].Prep(out, remote)
+		if err != nil {
+			hostinfo.logger(f.l).WithError(err).WithField("udpAddr", remote).Error("Failed to write outgoing packet")
+		}
+	} else if hostinfo.remote.IsValid() {
+		err = f.writers[q].Prep(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.QueryVpnAddrsRelayFor(hostinfo.vpnAddrs, relayIP)
+			if err != nil {
+				hostinfo.relayState.DeleteRelay(relayIP)
+				hostinfo.logger(f.l).WithField("relay", relayIP).WithError(err).Info("sendNoMetrics failed to find HostInfo")
+				continue
+			}
+			//todo vector!!
+			f.SendVia(relayHostInfo, relay, out.Payload, nb, fullOut[:header.Len+len(out.Payload)], true)
+			break
+		}
+	}
+}

inside_bsd.go

@@ -1,4 +1,5 @@
 //go:build darwin || dragonfly || freebsd || netbsd || openbsd
+// +build darwin dragonfly freebsd netbsd openbsd
 
 package nebula
 

inside_generic.go

@@ -1,4 +1,5 @@
 //go:build !darwin && !dragonfly && !freebsd && !netbsd && !openbsd
+// +build !darwin,!dragonfly,!freebsd,!netbsd,!openbsd
 
 package nebula
 

interface.go

@@ -4,25 +4,25 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"log/slog"
 	"net/netip"
-	"sync"
+	"os"
+	"runtime"
 	"sync/atomic"
 	"time"
 
 	"github.com/gaissmai/bart"
 	"github.com/rcrowley/go-metrics"
-
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/firewall"
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/overlay"
-	"github.com/slackhq/nebula/overlay/batch"
-	"github.com/slackhq/nebula/overlay/tio"
+	"github.com/slackhq/nebula/packet"
 	"github.com/slackhq/nebula/udp"
 )
 
 const mtu = 9001
+const batch = 1024 //todo config!
 
 type InterfaceConfig struct {
 	HostMap            *HostMap
@@ -31,7 +31,7 @@ type InterfaceConfig struct {
 	pki                *PKI
 	Cipher             string
 	Firewall           *Firewall
-	DnsServer          *dnsServer
+	ServeDns           bool
 	HandshakeManager   *HandshakeManager
 	lightHouse         *LightHouse
 	connectionManager  *connectionManager
@@ -48,7 +48,7 @@ type InterfaceConfig struct {
 	reQueryWait     time.Duration
 
 	ConntrackCacheTimeout time.Duration
-	l                     *slog.Logger
+	l                     *logrus.Logger
 }
 
 type Interface struct {
@@ -59,7 +59,7 @@ type Interface struct {
 	firewall              *Firewall
 	connectionManager     *connectionManager
 	handshakeManager      *HandshakeManager
-	dnsServer             *dnsServer
+	serveDns              bool
 	createTime            time.Time
 	lightHouse            *LightHouse
 	myBroadcastAddrsTable *bart.Lite
@@ -78,8 +78,7 @@ type Interface struct {
 	reQueryEvery    atomic.Uint32
 	reQueryWait     atomic.Int64
 
-	sendRecvErrorConfig   recvErrorConfig
-	acceptRecvErrorConfig recvErrorConfig
+	sendRecvErrorConfig sendRecvErrorConfig
 
 	// rebindCount is used to decide if an active tunnel should trigger a punch notification through a lighthouse
 	rebindCount int8
@@ -87,26 +86,20 @@ type Interface struct {
 
 	conntrackCacheTimeout time.Duration
 
-	ctx     context.Context
 	writers []udp.Conn
-	readers []tio.Queue
-	// batchers is one per tun queue, wrapping readers[i].
-	// decryptToTun sends plaintext into the batch.RxBatcher;
-	// listenOut calls its Flush at the end of each UDP recvmmsg batch.
-	batchers []batch.RxBatcher
-	wg       sync.WaitGroup
-
-	// fatalErr holds the first unexpected reader error that caused shutdown.
-	// nil means "no fatal error" (yet)
-	fatalErr atomic.Pointer[error]
-	// triggerShutdown is a function that will be run exactly once, when onFatal swaps something non-nil into fatalErr
-	triggerShutdown func()
+	readers []overlay.TunDev
 
 	metricHandshakes    metrics.Histogram
 	messageMetrics      *MessageMetrics
 	cachedPacketMetrics *cachedPacketMetrics
 
-	l *slog.Logger
+	listenInN  int
+	listenOutN int
+
+	listenInMetric  metrics.Histogram
+	listenOutMetric metrics.Histogram
+
+	l *logrus.Logger
 }
 
 type EncWriter interface {
@@ -124,34 +117,34 @@ type EncWriter interface {
 	GetCertState() *CertState
 }
 
-type recvErrorConfig uint8
+type sendRecvErrorConfig uint8
 
 const (
-	recvErrorAlways recvErrorConfig = iota
-	recvErrorNever
-	recvErrorPrivate
+	sendRecvErrorAlways sendRecvErrorConfig = iota
+	sendRecvErrorNever
+	sendRecvErrorPrivate
 )
 
-func (s recvErrorConfig) ShouldRecvError(endpoint netip.AddrPort) bool {
+func (s sendRecvErrorConfig) ShouldSendRecvError(endpoint netip.AddrPort) bool {
 	switch s {
-	case recvErrorPrivate:
+	case sendRecvErrorPrivate:
 		return endpoint.Addr().IsPrivate()
-	case recvErrorAlways:
+	case sendRecvErrorAlways:
 		return true
-	case recvErrorNever:
+	case sendRecvErrorNever:
 		return false
 	default:
-		panic(fmt.Errorf("invalid recvErrorConfig value: %d", s))
+		panic(fmt.Errorf("invalid sendRecvErrorConfig value: %d", s))
 	}
 }
 
-func (s recvErrorConfig) String() string {
+func (s sendRecvErrorConfig) String() string {
 	switch s {
-	case recvErrorAlways:
+	case sendRecvErrorAlways:
 		return "always"
-	case recvErrorNever:
+	case sendRecvErrorNever:
 		return "never"
-	case recvErrorPrivate:
+	case sendRecvErrorPrivate:
 		return "private"
 	default:
 		return fmt.Sprintf("invalid(%d)", s)
@@ -177,13 +170,12 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 
 	cs := c.pki.getCertState()
 	ifce := &Interface{
-		ctx:                   ctx,
 		pki:                   c.pki,
 		hostMap:               c.HostMap,
 		outside:               c.Outside,
 		inside:                c.Inside,
 		firewall:              c.Firewall,
-		dnsServer:             c.DnsServer,
+		serveDns:              c.ServeDns,
 		handshakeManager:      c.HandshakeManager,
 		createTime:            time.Now(),
 		lightHouse:            c.lightHouse,
@@ -192,8 +184,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 		routines:              c.routines,
 		version:               c.version,
 		writers:               make([]udp.Conn, c.routines),
-		readers:               make([]tio.Queue, c.routines),
-		batchers:              make([]batch.RxBatcher, c.routines),
+		readers:               make([]overlay.TunDev, c.routines),
 		myVpnNetworks:         cs.myVpnNetworks,
 		myVpnNetworksTable:    cs.myVpnNetworksTable,
 		myVpnAddrs:            cs.myVpnAddrs,
@@ -212,6 +203,8 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 
 		l: c.l,
 	}
+	ifce.listenInMetric = metrics.GetOrRegisterHistogram("vhost.listenIn.n", nil, metrics.NewExpDecaySample(1028, 0.015))
+	ifce.listenOutMetric = metrics.GetOrRegisterHistogram("vhost.listenOut.n", nil, metrics.NewExpDecaySample(1028, 0.015))
 
 	ifce.tryPromoteEvery.Store(c.tryPromoteEvery)
 	ifce.reQueryEvery.Store(c.reQueryEvery)
@@ -225,21 +218,18 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 // activate creates the interface on the host. After the interface is created, any
 // other services that want to bind listeners to its IP may do so successfully. However,
 // the interface isn't going to process anything until run() is called.
-func (f *Interface) activate() error {
+func (f *Interface) activate() {
 	// actually turn on tun dev
 
 	addr, err := f.outside.LocalAddr()
 	if err != nil {
-		f.l.Error("Failed to get udp listen address", "error", err)
+		f.l.WithError(err).Error("Failed to get udp listen address")
 	}
 
-	f.l.Info("Nebula interface is active",
-		"interface", f.inside.Name(),
-		"networks", f.myVpnNetworks,
-		"build", f.version,
-		"udpAddr", addr,
-		"boringcrypto", boringEnabled(),
-	)
+	f.l.WithField("interface", f.inside.Name()).WithField("networks", f.myVpnNetworks).
+		WithField("build", f.version).WithField("udpAddr", addr).
+		WithField("boringcrypto", boringEnabled()).
+		Info("Nebula interface is active")
 
 	if f.routines > 1 {
 		if !f.inside.SupportsMultiqueue() || !f.outside.SupportsMultipleReaders() {
@@ -251,145 +241,141 @@ func (f *Interface) activate() error {
 	metrics.GetOrRegisterGauge("routines", nil).Update(int64(f.routines))
 
 	// Prepare n tun queues
+	var reader overlay.TunDev = f.inside
 	for i := 0; i < f.routines; i++ {
 		if i > 0 {
-			if err = f.inside.NewMultiQueueReader(); err != nil {
-				return err
+			reader, err = f.inside.NewMultiQueueReader()
+			if err != nil {
+				f.l.Fatal(err)
 			}
 		}
-	}
-	f.readers = f.inside.Readers()
-	for i := range f.readers {
-		f.batchers[i] = batch.NewTCPCoalescer(f.readers[i])
+		f.readers[i] = reader
 	}
 
-	f.wg.Add(1) // for us to wait on Close() to return
-	if err = f.inside.Activate(); err != nil {
-		f.wg.Done()
+	if err := f.inside.Activate(); err != nil {
 		f.inside.Close()
-		return err
+		f.l.Fatal(err)
 	}
-
-	return nil
 }
 
-func (f *Interface) run() (func() error, error) {
+func (f *Interface) run() {
 	// Launch n queues to read packets from udp
 	for i := 0; i < f.routines; i++ {
-		f.wg.Go(func() {
-			f.listenOut(i)
-		})
+		go f.listenOut(i)
 	}
 
 	// Launch n queues to read packets from tun dev
 	for i := 0; i < f.routines; i++ {
-		f.wg.Go(func() {
-			f.listenIn(f.readers[i], i)
-		})
-	}
-
-	return func() error {
-		f.wg.Wait()
-		if e := f.fatalErr.Load(); e != nil {
-			return *e
-		}
-		return nil
-	}, nil
-}
-
-// onFatal stores the first fatal reader error, and calls triggerShutdown if it was the first one
-func (f *Interface) onFatal(err error) {
-	swapped := f.fatalErr.CompareAndSwap(nil, &err)
-	if !swapped {
-		return
-	}
-	if f.triggerShutdown != nil {
-		f.triggerShutdown()
+		go f.listenIn(f.readers[i], i)
 	}
 }
 
-func (f *Interface) listenOut(i int) {
+type Scratches struct {
+	h        *header.H
+	nb       []byte
+	fwPacket *firewall.Packet
+	scratch  []byte
+}
+
+func NewScratches() *Scratches {
+	return &Scratches{
+		h:        &header.H{},
+		fwPacket: &firewall.Packet{},
+		nb:       make([]byte, 12),
+		scratch:  make([]byte, udp.MTU),
+	}
+}
+
+func (f *Interface) listenOut(q int) {
+	runtime.LockOSThread()
+
 	var li udp.Conn
-	if i > 0 {
-		li = f.writers[i]
+	if q > 0 {
+		li = f.writers[q]
 	} else {
 		li = f.outside
 	}
 
-	ctCache := firewall.NewConntrackCacheTicker(f.ctx, f.l, f.conntrackCacheTimeout)
+	ctCache := firewall.NewConntrackCacheTicker(f.conntrackCacheTimeout)
 	lhh := f.lightHouse.NewRequestHandler()
-	h := &header.H{}
-	fwPacket := &firewall.Packet{}
-	nb := make([]byte, 12, 12)
 
-	coalescer := f.batchers[i]
-
-	listener := func(fromUdpAddr netip.AddrPort, payload []byte) {
-		plaintext := f.batchers[i].Reserve(len(payload))
-		f.readOutsidePackets(ViaSender{UdpAddr: fromUdpAddr}, plaintext[:0], payload, h, fwPacket, lhh, nb, i, ctCache.Get())
+	outPackets := make([]*packet.OutPacket, batch)
+	for i := 0; i < batch; i++ {
+		outPackets[i] = packet.NewOut()
 	}
 
-	flusher := func() {
-		if err := coalescer.Flush(); err != nil {
-			f.l.Error("Failed to flush tun coalescer", "error", err)
+	scratches := NewScratches()
+
+	toSend := make([][]byte, batch)
+
+	li.ListenOut(func(pkts []*packet.UDPPacket) {
+		toSend = toSend[:0]
+
+		f.readOutsidePacketsMany(pkts, outPackets, lhh, scratches, q, ctCache.Get(f.l), time.Now())
+		//we opportunistically tx, but try to also send stragglers
+		if _, err := f.readers[q].WriteMany(outPackets, q); err != nil {
+			f.l.WithError(err).Error("Failed to send packets")
 		}
-	}
+		//todo I broke this
+		//n := len(toSend)
+		//if f.l.Level == logrus.DebugLevel {
+		//	f.listenOutMetric.Update(int64(n))
+		//}
+		//f.listenOutN = n
 
-	err := li.ListenOut(listener, flusher)
-
-	if err != nil && !f.closed.Load() {
-		f.l.Error("Error while reading inbound packet, closing", "error", err)
-		f.onFatal(err)
-	}
-
-	f.l.Debug("underlay reader is done", "reader", i)
+	})
 }
 
-func (f *Interface) listenIn(reader tio.Queue, i int) {
-	rejectBuf := make([]byte, mtu)
-	sb := batch.NewSendBatch(batch.SendBatchCap, udp.MTU+32)
+func (f *Interface) listenIn(reader overlay.TunDev, queueNum int) {
+	runtime.LockOSThread()
+
 	fwPacket := &firewall.Packet{}
 	nb := make([]byte, 12, 12)
 
-	conntrackCache := firewall.NewConntrackCacheTicker(f.ctx, f.l, f.conntrackCacheTimeout)
+	conntrackCache := firewall.NewConntrackCacheTicker(f.conntrackCacheTimeout)
+
+	packets := reader.NewPacketArrays(batch)
+
+	outPackets := make([]*packet.UDPPacket, batch)
+	for i := 0; i < batch; i++ {
+		outPackets[i] = packet.New(false) //todo isv4?
+	}
 
 	for {
-		pkts, err := reader.Read()
+		n, err := reader.ReadMany(packets, queueNum)
+
+		//todo!!
 		if err != nil {
-			if !f.closed.Load() {
-				f.l.Error("Error while reading outbound packet, closing", "error", err, "reader", i)
-				f.onFatal(err)
+			if errors.Is(err, os.ErrClosed) && f.closed.Load() {
+				return
 			}
-			break
+
+			f.l.WithError(err).Error("Error while reading outbound packet")
+			// This only seems to happen when something fatal happens to the fd, so exit.
+			os.Exit(2)
 		}
 
-		sb.Reset()
-		for _, pkt := range pkts {
-			if sb.Len() >= sb.Cap() {
-				f.flushBatch(sb, i)
-				sb.Reset()
-			}
-			f.consumeInsidePacket(pkt, fwPacket, nb, sb, rejectBuf, i, conntrackCache.Get())
+		if f.l.Level == logrus.DebugLevel {
+			f.listenInMetric.Update(int64(n))
 		}
-		if sb.Len() > 0 {
-			f.flushBatch(sb, i)
+		f.listenInN = n
+
+		now := time.Now()
+		for i, pkt := range packets[:n] {
+			outPackets[i].ReadyToSend = false
+			f.consumeInsidePacket(pkt.GetPayload(), fwPacket, nb, outPackets[i], queueNum, conntrackCache.Get(f.l), now)
+			reader.RecycleRxSeg(pkt, i == (n-1), queueNum) //todo handle err?
+		}
+		_, err = f.writers[queueNum].WriteBatch(outPackets[:n])
+		if err != nil {
+			f.l.WithError(err).Error("Error while writing outbound packets")
 		}
-	}
-
-	f.l.Debug("overlay reader is done", "reader", i)
-}
-
-func (f *Interface) flushBatch(sb batch.TxBatcher, q int) {
-	bufs, dsts := sb.Get()
-	if err := f.writers[q].WriteBatch(bufs, dsts); err != nil {
-		f.l.Error("Failed to write outgoing batch", "error", err, "writer", q)
 	}
 }
 
 func (f *Interface) RegisterConfigChangeCallbacks(c *config.C) {
 	c.RegisterReloadCallback(f.reloadFirewall)
 	c.RegisterReloadCallback(f.reloadSendRecvError)
-	c.RegisterReloadCallback(f.reloadAcceptRecvError)
 	c.RegisterReloadCallback(f.reloadDisconnectInvalid)
 	c.RegisterReloadCallback(f.reloadMisc)
 
@@ -403,7 +389,7 @@ func (f *Interface) reloadDisconnectInvalid(c *config.C) {
 	if initial || c.HasChanged("pki.disconnect_invalid") {
 		f.disconnectInvalid.Store(c.GetBool("pki.disconnect_invalid", true))
 		if !initial {
-			f.l.Info("pki.disconnect_invalid changed", "value", f.disconnectInvalid.Load())
+			f.l.Infof("pki.disconnect_invalid changed to %v", f.disconnectInvalid.Load())
 		}
 	}
 }
@@ -417,7 +403,7 @@ func (f *Interface) reloadFirewall(c *config.C) {
 
 	fw, err := NewFirewallFromConfig(f.l, f.pki.getCertState(), c)
 	if err != nil {
-		f.l.Error("Error while creating firewall during reload", "error", err)
+		f.l.WithError(err).Error("Error while creating firewall during reload")
 		return
 	}
 
@@ -430,11 +416,10 @@ func (f *Interface) reloadFirewall(c *config.C) {
 	// If rulesVersion is back to zero, we have wrapped all the way around. Be
 	// safe and just reset conntrack in this case.
 	if fw.rulesVersion == 0 {
-		f.l.Warn("firewall rulesVersion has overflowed, resetting conntrack",
-			"firewallHashes", fw.GetRuleHashes(),
-			"oldFirewallHashes", oldFw.GetRuleHashes(),
-			"rulesVersion", fw.rulesVersion,
-		)
+		f.l.WithField("firewallHashes", fw.GetRuleHashes()).
+			WithField("oldFirewallHashes", oldFw.GetRuleHashes()).
+			WithField("rulesVersion", fw.rulesVersion).
+			Warn("firewall rulesVersion has overflowed, resetting conntrack")
 	} else {
 		fw.Conntrack = conntrack
 	}
@@ -442,11 +427,10 @@ func (f *Interface) reloadFirewall(c *config.C) {
 	f.firewall = fw
 
 	oldFw.Destroy()
-	f.l.Info("New firewall has been installed",
-		"firewallHashes", fw.GetRuleHashes(),
-		"oldFirewallHashes", oldFw.GetRuleHashes(),
-		"rulesVersion", fw.rulesVersion,
-	)
+	f.l.WithField("firewallHashes", fw.GetRuleHashes()).
+		WithField("oldFirewallHashes", oldFw.GetRuleHashes()).
+		WithField("rulesVersion", fw.rulesVersion).
+		Info("New firewall has been installed")
 }
 
 func (f *Interface) reloadSendRecvError(c *config.C) {
@@ -455,43 +439,21 @@ func (f *Interface) reloadSendRecvError(c *config.C) {
 
 		switch stringValue {
 		case "always":
-			f.sendRecvErrorConfig = recvErrorAlways
+			f.sendRecvErrorConfig = sendRecvErrorAlways
 		case "never":
-			f.sendRecvErrorConfig = recvErrorNever
+			f.sendRecvErrorConfig = sendRecvErrorNever
 		case "private":
-			f.sendRecvErrorConfig = recvErrorPrivate
+			f.sendRecvErrorConfig = sendRecvErrorPrivate
 		default:
 			if c.GetBool("listen.send_recv_error", true) {
-				f.sendRecvErrorConfig = recvErrorAlways
+				f.sendRecvErrorConfig = sendRecvErrorAlways
 			} else {
-				f.sendRecvErrorConfig = recvErrorNever
+				f.sendRecvErrorConfig = sendRecvErrorNever
 			}
 		}
 
-		f.l.Info("Loaded send_recv_error config", "sendRecvError", f.sendRecvErrorConfig.String())
-	}
-}
-
-func (f *Interface) reloadAcceptRecvError(c *config.C) {
-	if c.InitialLoad() || c.HasChanged("listen.accept_recv_error") {
-		stringValue := c.GetString("listen.accept_recv_error", "always")
-
-		switch stringValue {
-		case "always":
-			f.acceptRecvErrorConfig = recvErrorAlways
-		case "never":
-			f.acceptRecvErrorConfig = recvErrorNever
-		case "private":
-			f.acceptRecvErrorConfig = recvErrorPrivate
-		default:
-			if c.GetBool("listen.accept_recv_error", true) {
-				f.acceptRecvErrorConfig = recvErrorAlways
-			} else {
-				f.acceptRecvErrorConfig = recvErrorNever
-			}
-		}
-
-		f.l.Info("Loaded accept_recv_error config", "acceptRecvError", f.acceptRecvErrorConfig.String())
+		f.l.WithField("sendRecvError", f.sendRecvErrorConfig.String()).
+			Info("Loaded send_recv_error config")
 	}
 }
 
@@ -545,6 +507,11 @@ func (f *Interface) emitStats(ctx context.Context, i time.Duration) {
 			} else {
 				certMaxVersion.Update(int64(certState.v1Cert.Version()))
 			}
+			if f.l.Level != logrus.DebugLevel {
+				f.listenInMetric.Update(int64(f.listenInN))
+				f.listenOutMetric.Update(int64(f.listenOutN))
+			}
+
 		}
 	}
 }
@@ -558,23 +525,15 @@ func (f *Interface) GetCertState() *CertState {
 }
 
 func (f *Interface) Close() error {
-	var errs []error
 	f.closed.Store(true)
 
-	// Release the udp readers
-	for i, u := range f.writers {
+	for _, u := range f.writers {
 		err := u.Close()
 		if err != nil {
-			f.l.Error("Error while closing udp socket", "error", err, "writer", i)
-			errs = append(errs, err)
+			f.l.WithError(err).Error("Error while closing udp socket")
 		}
 	}
 
-	// Release the tun device (closing the tun also closes all readers)
-	closeErr := f.inside.Close()
-	if closeErr != nil {
-		errs = append(errs, closeErr)
-	}
-	f.wg.Done()
-	return errors.Join(errs...)
+	// Release the tun device
+	return f.inside.Close()
 }

lighthouse.go

@@ -5,7 +5,6 @@ import (
 	"encoding/binary"
 	"errors"
 	"fmt"
-	"log/slog"
 	"net"
 	"net/netip"
 	"slices"
@@ -16,10 +15,10 @@ import (
 
 	"github.com/gaissmai/bart"
 	"github.com/rcrowley/go-metrics"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
-	"github.com/slackhq/nebula/logging"
 	"github.com/slackhq/nebula/udp"
 	"github.com/slackhq/nebula/util"
 )
@@ -70,19 +69,18 @@ type LightHouse struct {
 	// Addr's of relays that can be used by peers to access me
 	relaysForMe atomic.Pointer[[]netip.Addr]
 
-	updateTrigger chan struct{}
-	queryChan     chan netip.Addr
+	queryChan chan netip.Addr
 
 	calculatedRemotes atomic.Pointer[bart.Table[[]*calculatedRemote]] // Maps VpnAddr to []*calculatedRemote
 
 	metrics           *MessageMetrics
 	metricHolepunchTx metrics.Counter
-	l                 *slog.Logger
+	l                 *logrus.Logger
 }
 
 // NewLightHouseFromConfig will build a Lighthouse struct from the values provided in the config object
 // addrMap should be nil unless this is during a config reload
-func NewLightHouseFromConfig(ctx context.Context, l *slog.Logger, c *config.C, cs *CertState, pc udp.Conn, p *Punchy) (*LightHouse, error) {
+func NewLightHouseFromConfig(ctx context.Context, l *logrus.Logger, c *config.C, cs *CertState, pc udp.Conn, p *Punchy) (*LightHouse, error) {
 	amLighthouse := c.GetBool("lighthouse.am_lighthouse", false)
 	nebulaPort := uint32(c.GetInt("listen.port", 0))
 	if amLighthouse && nebulaPort == 0 {
@@ -107,7 +105,6 @@ func NewLightHouseFromConfig(ctx context.Context, l *slog.Logger, c *config.C, c
 		nebulaPort:         nebulaPort,
 		punchConn:          pc,
 		punchy:             p,
-		updateTrigger:      make(chan struct{}, 1),
 		queryChan:          make(chan netip.Addr, c.GetUint32("handshakes.query_buffer", 64)),
 		l:                  l,
 	}
@@ -134,7 +131,7 @@ func NewLightHouseFromConfig(ctx context.Context, l *slog.Logger, c *config.C, c
 		case *util.ContextualError:
 			v.Log(l)
 		case error:
-			l.Error("failed to reload lighthouse", "error", err)
+			l.WithError(err).Error("failed to reload lighthouse")
 		}
 	})
 
@@ -206,10 +203,8 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 			//TODO: we could technically insert all returned addrs instead of just the first one if a dns lookup was used
 			addr := addrs[0].Unmap()
 			if lh.myVpnNetworksTable.Contains(addr) {
-				lh.l.Warn("Ignoring lighthouse.advertise_addrs report because it is within the nebula network range",
-					"addr", rawAddr,
-					"entry", i+1,
-				)
+				lh.l.WithField("addr", rawAddr).WithField("entry", i+1).
+					Warn("Ignoring lighthouse.advertise_addrs report because it is within the nebula network range")
 				continue
 			}
 
@@ -227,9 +222,7 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 		lh.interval.Store(int64(c.GetInt("lighthouse.interval", 10)))
 
 		if !initial {
-			lh.l.Info("lighthouse.interval changed",
-				"interval", lh.interval.Load(),
-			)
+			lh.l.Infof("lighthouse.interval changed to %v", lh.interval.Load())
 
 			if lh.updateCancel != nil {
 				// May not always have a running routine
@@ -323,7 +316,6 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 		if !initial {
 			//NOTE: we are not tearing down existing lighthouse connections because they might be used for non lighthouse traffic
 			lh.l.Info("lighthouse.hosts has changed")
-			lh.TriggerUpdate()
 		}
 	}
 
@@ -341,12 +333,9 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 			for _, v := range c.GetStringSlice("relay.relays", nil) {
 				configRIP, err := netip.ParseAddr(v)
 				if err != nil {
-					lh.l.Warn("Parse relay from config failed",
-						"relay", v,
-						"error", err,
-					)
+					lh.l.WithField("relay", v).WithError(err).Warn("Parse relay from config failed")
 				} else {
-					lh.l.Info("Read relay from config", "relay", v)
+					lh.l.WithField("relay", v).Info("Read relay from config")
 					relaysForMe = append(relaysForMe, configRIP)
 				}
 			}
@@ -371,10 +360,8 @@ func (lh *LightHouse) parseLighthouses(c *config.C) ([]netip.Addr, error) {
 		}
 
 		if !lh.myVpnNetworksTable.Contains(addr) {
-			lh.l.Warn("lighthouse host is not within our networks, lighthouse functionality will work but layer 3 network traffic to the lighthouse will not",
-				"vpnAddr", addr,
-				"networks", lh.myVpnNetworks,
-			)
+			lh.l.WithFields(m{"vpnAddr": addr, "networks": lh.myVpnNetworks}).
+				Warn("lighthouse host is not within our networks, lighthouse functionality will work but layer 3 network traffic to the lighthouse will not")
 		}
 		out[i] = addr
 	}
@@ -445,11 +432,8 @@ func (lh *LightHouse) loadStaticMap(c *config.C, staticList map[netip.Addr]struc
 		}
 
 		if !lh.myVpnNetworksTable.Contains(vpnAddr) {
-			lh.l.Warn("static_host_map key is not within our networks, layer 3 network traffic to this host will not work",
-				"vpnAddr", vpnAddr,
-				"networks", lh.myVpnNetworks,
-				"entry", i+1,
-			)
+			lh.l.WithFields(m{"vpnAddr": vpnAddr, "networks": lh.myVpnNetworks, "entry": i + 1}).
+				Warn("static_host_map key is not within our networks, layer 3 network traffic to this host will not work")
 		}
 
 		vals, ok := v.([]any)
@@ -550,13 +534,12 @@ func (lh *LightHouse) DeleteVpnAddrs(allVpnAddrs []netip.Addr) {
 	lh.Lock()
 	rm, ok := lh.addrMap[allVpnAddrs[0]]
 	if ok {
-		debugEnabled := lh.l.Enabled(context.Background(), slog.LevelDebug)
 		for _, addr := range allVpnAddrs {
 			srm := lh.addrMap[addr]
 			if srm == rm {
 				delete(lh.addrMap, addr)
-				if debugEnabled {
-					lh.l.Debug("deleting from lighthouse", "vpnAddr", addr)
+				if lh.l.Level >= logrus.DebugLevel {
+					lh.l.Debugf("deleting %s from lighthouse.", addr)
 				}
 			}
 		}
@@ -673,12 +656,9 @@ func (lh *LightHouse) unlockedGetRemoteList(allAddrs []netip.Addr) *RemoteList {
 
 func (lh *LightHouse) shouldAdd(vpnAddrs []netip.Addr, to netip.Addr) bool {
 	allow := lh.GetRemoteAllowList().AllowAll(vpnAddrs, to)
-	if lh.l.Enabled(context.Background(), logging.LevelTrace) {
-		lh.l.Log(context.Background(), logging.LevelTrace, "remoteAllowList.Allow",
-			"vpnAddrs", vpnAddrs,
-			"udpAddr", to,
-			"allow", allow,
-		)
+	if lh.l.Level >= logrus.TraceLevel {
+		lh.l.WithField("vpnAddrs", vpnAddrs).WithField("udpAddr", to).WithField("allow", allow).
+			Trace("remoteAllowList.Allow")
 	}
 	if !allow {
 		return false
@@ -695,12 +675,9 @@ func (lh *LightHouse) shouldAdd(vpnAddrs []netip.Addr, to netip.Addr) bool {
 func (lh *LightHouse) unlockedShouldAddV4(vpnAddr netip.Addr, to *V4AddrPort) bool {
 	udpAddr := protoV4AddrPortToNetAddrPort(to)
 	allow := lh.GetRemoteAllowList().Allow(vpnAddr, udpAddr.Addr())
-	if lh.l.Enabled(context.Background(), logging.LevelTrace) {
-		lh.l.Log(context.Background(), logging.LevelTrace, "remoteAllowList.Allow",
-			"vpnAddr", vpnAddr,
-			"udpAddr", udpAddr,
-			"allow", allow,
-		)
+	if lh.l.Level >= logrus.TraceLevel {
+		lh.l.WithField("vpnAddr", vpnAddr).WithField("udpAddr", udpAddr).WithField("allow", allow).
+			Trace("remoteAllowList.Allow")
 	}
 
 	if !allow {
@@ -718,12 +695,9 @@ func (lh *LightHouse) unlockedShouldAddV4(vpnAddr netip.Addr, to *V4AddrPort) bo
 func (lh *LightHouse) unlockedShouldAddV6(vpnAddr netip.Addr, to *V6AddrPort) bool {
 	udpAddr := protoV6AddrPortToNetAddrPort(to)
 	allow := lh.GetRemoteAllowList().Allow(vpnAddr, udpAddr.Addr())
-	if lh.l.Enabled(context.Background(), logging.LevelTrace) {
-		lh.l.Log(context.Background(), logging.LevelTrace, "remoteAllowList.Allow",
-			"vpnAddr", vpnAddr,
-			"udpAddr", udpAddr,
-			"allow", allow,
-		)
+	if lh.l.Level >= logrus.TraceLevel {
+		lh.l.WithField("vpnAddr", vpnAddr).WithField("udpAddr", udpAddr).WithField("allow", allow).
+			Trace("remoteAllowList.Allow")
 	}
 
 	if !allow {
@@ -739,14 +713,21 @@ func (lh *LightHouse) unlockedShouldAddV6(vpnAddr netip.Addr, to *V6AddrPort) bo
 
 func (lh *LightHouse) IsLighthouseAddr(vpnAddr netip.Addr) bool {
 	l := lh.GetLighthouses()
-	return slices.Contains(l, vpnAddr)
+	for i := range l {
+		if l[i] == vpnAddr {
+			return true
+		}
+	}
+	return false
 }
 
 func (lh *LightHouse) IsAnyLighthouseAddr(vpnAddrs []netip.Addr) bool {
 	l := lh.GetLighthouses()
 	for i := range vpnAddrs {
-		if slices.Contains(l, vpnAddrs[i]) {
-			return true
+		for j := range l {
+			if l[j] == vpnAddrs[i] {
+				return true
+			}
 		}
 	}
 	return false
@@ -798,10 +779,8 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 
 		if v == cert.Version1 {
 			if !addr.Is4() {
-				lh.l.Error("Can't query lighthouse for v6 address using a v1 protocol",
-					"queryVpnAddr", addr,
-					"lighthouseAddr", lhVpnAddr,
-				)
+				lh.l.WithField("queryVpnAddr", addr).WithField("lighthouseAddr", lhVpnAddr).
+					Error("Can't query lighthouse for v6 address using a v1 protocol")
 				continue
 			}
 
@@ -812,11 +791,9 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 
 				v1Query, err = msg.Marshal()
 				if err != nil {
-					lh.l.Error("Failed to marshal lighthouse v1 query payload",
-						"error", err,
-						"queryVpnAddr", addr,
-						"lighthouseAddr", lhVpnAddr,
-					)
+					lh.l.WithError(err).WithField("queryVpnAddr", addr).
+						WithField("lighthouseAddr", lhVpnAddr).
+						Error("Failed to marshal lighthouse v1 query payload")
 					continue
 				}
 			}
@@ -831,11 +808,9 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 
 				v2Query, err = msg.Marshal()
 				if err != nil {
-					lh.l.Error("Failed to marshal lighthouse v2 query payload",
-						"error", err,
-						"queryVpnAddr", addr,
-						"lighthouseAddr", lhVpnAddr,
-					)
+					lh.l.WithError(err).WithField("queryVpnAddr", addr).
+						WithField("lighthouseAddr", lhVpnAddr).
+						Error("Failed to marshal lighthouse v2 query payload")
 					continue
 				}
 			}
@@ -844,11 +819,7 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 			queried++
 
 		} else {
-			lh.l.Debug("unsupported protocol version",
-				"op", "query",
-				"queryVpnAddr", addr,
-				"version", v,
-			)
+			lh.l.Debugf("Can not query lighthouse for %v using unknown protocol version: %v", addr, v)
 			continue
 		}
 	}
@@ -877,24 +848,11 @@ func (lh *LightHouse) StartUpdateWorker() {
 				return
 			case <-clockSource.C:
 				continue
-			case <-lh.updateTrigger:
-				continue
 			}
 		}
 	}()
 }
 
-// TriggerUpdate requests an immediate lighthouse update. This is a non-blocking
-// operation intended to be called after a handshake completes with a lighthouse,
-// so the lighthouse has our current addresses without waiting for the next
-// periodic update.
-func (lh *LightHouse) TriggerUpdate() {
-	select {
-	case lh.updateTrigger <- struct{}{}:
-	default:
-	}
-}
-
 func (lh *LightHouse) SendUpdate() {
 	var v4 []*V4AddrPort
 	var v6 []*V6AddrPort
@@ -940,9 +898,8 @@ func (lh *LightHouse) SendUpdate() {
 		if v == cert.Version1 {
 			if v1Update == nil {
 				if !lh.myVpnNetworks[0].Addr().Is4() {
-					lh.l.Warn("cannot update lighthouse using v1 protocol without an IPv4 address",
-						"lighthouseAddr", lhVpnAddr,
-					)
+					lh.l.WithField("lighthouseAddr", lhVpnAddr).
+						Warn("cannot update lighthouse using v1 protocol without an IPv4 address")
 					continue
 				}
 				var relays []uint32
@@ -966,10 +923,8 @@ func (lh *LightHouse) SendUpdate() {
 
 				v1Update, err = msg.Marshal()
 				if err != nil {
-					lh.l.Error("Error while marshaling for lighthouse v1 update",
-						"error", err,
-						"lighthouseAddr", lhVpnAddr,
-					)
+					lh.l.WithError(err).WithField("lighthouseAddr", lhVpnAddr).
+						Error("Error while marshaling for lighthouse v1 update")
 					continue
 				}
 			}
@@ -995,10 +950,8 @@ func (lh *LightHouse) SendUpdate() {
 
 				v2Update, err = msg.Marshal()
 				if err != nil {
-					lh.l.Error("Error while marshaling for lighthouse v2 update",
-						"error", err,
-						"lighthouseAddr", lhVpnAddr,
-					)
+					lh.l.WithError(err).WithField("lighthouseAddr", lhVpnAddr).
+						Error("Error while marshaling for lighthouse v2 update")
 					continue
 				}
 			}
@@ -1007,10 +960,7 @@ func (lh *LightHouse) SendUpdate() {
 			updated++
 
 		} else {
-			lh.l.Debug("unsupported protocol version",
-				"op", "update",
-				"version", v,
-			)
+			lh.l.Debugf("Can not update lighthouse using unknown protocol version: %v", v)
 			continue
 		}
 	}
@@ -1024,7 +974,7 @@ type LightHouseHandler struct {
 	out  []byte
 	pb   []byte
 	meta *NebulaMeta
-	l    *slog.Logger
+	l    *logrus.Logger
 }
 
 func (lh *LightHouse) NewRequestHandler() *LightHouseHandler {
@@ -1073,19 +1023,14 @@ func (lhh *LightHouseHandler) HandleRequest(rAddr netip.AddrPort, fromVpnAddrs [
 	n := lhh.resetMeta()
 	err := n.Unmarshal(p)
 	if err != nil {
-		lhh.l.Error("Failed to unmarshal lighthouse packet",
-			"error", err,
-			"vpnAddrs", fromVpnAddrs,
-			"udpAddr", rAddr,
-		)
+		lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).WithField("udpAddr", rAddr).
+			Error("Failed to unmarshal lighthouse packet")
 		return
 	}
 
 	if n.Details == nil {
-		lhh.l.Error("Invalid lighthouse update",
-			"vpnAddrs", fromVpnAddrs,
-			"udpAddr", rAddr,
-		)
+		lhh.l.WithField("vpnAddrs", fromVpnAddrs).WithField("udpAddr", rAddr).
+			Error("Invalid lighthouse update")
 		return
 	}
 
@@ -1113,29 +1058,25 @@ func (lhh *LightHouseHandler) HandleRequest(rAddr netip.AddrPort, fromVpnAddrs [
 func (lhh *LightHouseHandler) handleHostQuery(n *NebulaMeta, fromVpnAddrs []netip.Addr, addr netip.AddrPort, w EncWriter) {
 	// Exit if we don't answer queries
 	if !lhh.lh.amLighthouse {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("I don't answer queries, but received one", "from", addr)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.Debugln("I don't answer queries, but received from: ", addr)
 		}
 		return
 	}
 
 	queryVpnAddr, useVersion, err := n.Details.GetVpnAddrAndVersion()
 	if err != nil {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("Dropping malformed HostQuery",
-				"from", fromVpnAddrs,
-				"details", n.Details,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.WithField("from", fromVpnAddrs).WithField("details", n.Details).
+				Debugln("Dropping malformed HostQuery")
 		}
 		return
 	}
 	if useVersion == cert.Version1 && queryVpnAddr.Is6() {
 		// this case really shouldn't be possible to represent, but reject it anyway.
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("invalid vpn addr for v1 handleHostQuery",
-				"vpnAddrs", fromVpnAddrs,
-				"queryVpnAddr", queryVpnAddr,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.WithField("vpnAddrs", fromVpnAddrs).WithField("queryVpnAddr", queryVpnAddr).
+				Debugln("invalid vpn addr for v1 handleHostQuery")
 		}
 		return
 	}
@@ -1160,10 +1101,7 @@ func (lhh *LightHouseHandler) handleHostQuery(n *NebulaMeta, fromVpnAddrs []neti
 	}
 
 	if err != nil {
-		lhh.l.Error("Failed to marshal lighthouse host query reply",
-			"error", err,
-			"vpnAddrs", fromVpnAddrs,
-		)
+		lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("Failed to marshal lighthouse host query reply")
 		return
 	}
 
@@ -1191,10 +1129,8 @@ func (lhh *LightHouseHandler) sendHostPunchNotification(n *NebulaMeta, fromVpnAd
 			if ok {
 				whereToPunch = newDest
 			} else {
-				if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-					lhh.l.Debug("unable to punch to host, no addresses in common",
-						"to", crt.Networks(),
-					)
+				if lhh.l.Level >= logrus.DebugLevel {
+					lhh.l.WithField("to", crt.Networks()).Debugln("unable to punch to host, no addresses in common")
 				}
 			}
 		}
@@ -1220,10 +1156,7 @@ func (lhh *LightHouseHandler) sendHostPunchNotification(n *NebulaMeta, fromVpnAd
 	}
 
 	if err != nil {
-		lhh.l.Error("Failed to marshal lighthouse host was queried for",
-			"error", err,
-			"vpnAddrs", fromVpnAddrs,
-		)
+		lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("Failed to marshal lighthouse host was queried for")
 		return
 	}
 
@@ -1265,11 +1198,8 @@ func (lhh *LightHouseHandler) coalesceAnswers(v cert.Version, c *cache, n *Nebul
 				n.Details.RelayVpnAddrs = append(n.Details.RelayVpnAddrs, netAddrToProtoAddr(r))
 			}
 		} else {
-			if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-				lhh.l.Debug("unsupported protocol version",
-					"op", "coalesceAnswers",
-					"version", v,
-				)
+			if lhh.l.Level >= logrus.DebugLevel {
+				lhh.l.WithField("version", v).Debug("unsupported protocol version")
 			}
 		}
 	}
@@ -1282,11 +1212,8 @@ func (lhh *LightHouseHandler) handleHostQueryReply(n *NebulaMeta, fromVpnAddrs [
 
 	certVpnAddr, _, err := n.Details.GetVpnAddrAndVersion()
 	if err != nil {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Error("dropping malformed HostQueryReply",
-				"error", err,
-				"vpnAddrs", fromVpnAddrs,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("dropping malformed HostQueryReply")
 		}
 		return
 	}
@@ -1311,8 +1238,8 @@ func (lhh *LightHouseHandler) handleHostQueryReply(n *NebulaMeta, fromVpnAddrs [
 
 func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVpnAddrs []netip.Addr, w EncWriter) {
 	if !lhh.lh.amLighthouse {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("I am not a lighthouse, do not take host updates", "from", fromVpnAddrs)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.Debugln("I am not a lighthouse, do not take host updates: ", fromVpnAddrs)
 		}
 		return
 	}
@@ -1335,11 +1262,8 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 
 	//Simple check that the host sent this not someone else, if detailsVpnAddr is filled
 	if detailsVpnAddr.IsValid() && !slices.Contains(fromVpnAddrs, detailsVpnAddr) {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("Host sent invalid update",
-				"vpnAddrs", fromVpnAddrs,
-				"answer", detailsVpnAddr,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.WithField("vpnAddrs", fromVpnAddrs).WithField("answer", detailsVpnAddr).Debugln("Host sent invalid update")
 		}
 		return
 	}
@@ -1361,9 +1285,7 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 	switch useVersion {
 	case cert.Version1:
 		if !fromVpnAddrs[0].Is4() {
-			lhh.l.Error("Can not send HostUpdateNotificationAck for a ipv6 vpn ip in a v1 message",
-				"vpnAddrs", fromVpnAddrs,
-			)
+			lhh.l.WithField("vpnAddrs", fromVpnAddrs).Error("Can not send HostUpdateNotificationAck for a ipv6 vpn ip in a v1 message")
 			return
 		}
 		vpnAddrB := fromVpnAddrs[0].As4()
@@ -1371,16 +1293,13 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 	case cert.Version2:
 		// do nothing, we want to send a blank message
 	default:
-		lhh.l.Error("invalid protocol version", "useVersion", useVersion)
+		lhh.l.WithField("useVersion", useVersion).Error("invalid protocol version")
 		return
 	}
 
 	ln, err := n.MarshalTo(lhh.pb)
 	if err != nil {
-		lhh.l.Error("Failed to marshal lighthouse host update ack",
-			"error", err,
-			"vpnAddrs", fromVpnAddrs,
-		)
+		lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("Failed to marshal lighthouse host update ack")
 		return
 	}
 
@@ -1397,11 +1316,8 @@ func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpn
 
 	detailsVpnAddr, _, err := n.Details.GetVpnAddrAndVersion()
 	if err != nil {
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("dropping invalid HostPunchNotification",
-				"details", n.Details,
-				"error", err,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.WithField("details", n.Details).WithError(err).Debugln("dropping invalid HostPunchNotification")
 		}
 		return
 	}
@@ -1418,11 +1334,8 @@ func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpn
 			lhh.lh.punchConn.WriteTo(empty, vpnPeer)
 		}()
 
-		if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-			lhh.l.Debug("Punching",
-				"vpnPeer", vpnPeer,
-				"logVpnAddr", logVpnAddr,
-			)
+		if lhh.l.Level >= logrus.DebugLevel {
+			lhh.l.Debugf("Punching on %v for %v", vpnPeer, logVpnAddr)
 		}
 	}
 
@@ -1447,10 +1360,8 @@ func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpn
 	if lhh.lh.punchy.GetRespond() {
 		go func() {
 			time.Sleep(lhh.lh.punchy.GetRespondDelay())
-			if lhh.l.Enabled(context.Background(), slog.LevelDebug) {
-				lhh.l.Debug("Sending a nebula test packet",
-					"vpnAddr", detailsVpnAddr,
-				)
+			if lhh.l.Level >= logrus.DebugLevel {
+				lhh.l.Debugf("Sending a nebula test packet to vpn addr %s", detailsVpnAddr)
 			}
 			//NOTE: we have to allocate a new output buffer here since we are spawning a new goroutine
 			// for each punchBack packet. We should move this into a timerwheel or a single goroutine

lighthouse_test.go

@@ -1,6 +1,7 @@
 package nebula
 
 import (
+	"context"
 	"encoding/binary"
 	"fmt"
 	"net/netip"
@@ -41,14 +42,14 @@ func Test_lhStaticMapping(t *testing.T) {
 	c := config.NewC(l)
 	c.Settings["lighthouse"] = map[string]any{"hosts": []any{lh1}}
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"1.1.1.1:4242"}}
-	_, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	_, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 
 	lh2 := "10.128.0.3"
 	c = config.NewC(l)
 	c.Settings["lighthouse"] = map[string]any{"hosts": []any{lh1, lh2}}
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"100.1.1.1:4242"}}
-	_, err = NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	_, err = NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.EqualError(t, err, "lighthouse 10.128.0.3 does not have a static_host_map entry")
 }
 
@@ -70,7 +71,7 @@ func TestReloadLighthouseInterval(t *testing.T) {
 	}
 
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"1.1.1.1:4242"}}
-	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 
@@ -98,7 +99,7 @@ func BenchmarkLighthouseHandleRequest(b *testing.B) {
 	}
 
 	c := config.NewC(l)
-	lh, err := NewLightHouseFromConfig(b.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(b, err)
 
 	hAddr := netip.MustParseAddrPort("4.5.6.7:12345")
@@ -201,7 +202,7 @@ func TestLighthouse_Memory(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	lh.ifce = &mockEncWriter{}
 	require.NoError(t, err)
 	lhh := lh.NewRequestHandler()
@@ -287,7 +288,7 @@ func TestLighthouse_reload(t *testing.T) {
 		myVpnNetworksTable: nt,
 	}
 
-	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 
 	nc := map[string]any{
@@ -522,7 +523,7 @@ func TestLighthouse_Dont_Delete_Static_Hosts(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 
@@ -588,7 +589,7 @@ func TestLighthouse_DeletesWork(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 

logger.go

@@ -0,0 +1,45 @@
+package nebula
+
+import (
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/sirupsen/logrus"
+	"github.com/slackhq/nebula/config"
+)
+
+func configLogger(l *logrus.Logger, c *config.C) 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)
+	}
+	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":
+		l.Formatter = &logrus.TextFormatter{
+			TimestampFormat:  timestampFormat,
+			FullTimestamp:    fullTimestamp,
+			DisableTimestamp: disableTimestamp,
+		}
+	case "json":
+		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
+}

logging/logger.go

@@ -1,233 +0,0 @@
-// Package logging wires the nebula runtime-reconfigurable slog handler used
-// by nebula.Main and the nebula CLI binaries. Callers build a logger with
-// NewLogger, then call ApplyConfig at startup and from a config reload
-// callback to push logging.level, logging.format, and
-// logging.disable_timestamp changes onto the logger without rebuilding it.
-package logging
-
-import (
-	"context"
-	"fmt"
-	"io"
-	"log/slog"
-	"strings"
-	"sync/atomic"
-	"time"
-)
-
-// Config is the subset of *config.C that ApplyConfig reads. Declaring it
-// here keeps the logging package from depending on config directly, which
-// would cycle through the shared test helpers (test.NewLogger imports
-// logging, and config's tests import test). *config.C satisfies this
-// interface structurally with no adapter.
-type Config interface {
-	GetString(key, def string) string
-	GetBool(key string, def bool) bool
-}
-
-// LevelTrace is a custom slog level below Debug, used when logging.level is
-// "trace". slog has no builtin trace level; the value is one step below
-// slog.LevelDebug in slog's 4-point spacing.
-const LevelTrace = slog.Level(-8)
-
-// NewLogger returns a *slog.Logger whose level, format, and timestamp
-// emission can be reconfigured at runtime via ApplyConfig and the SSH debug
-// commands. The default configuration is info-level text output so log
-// calls made before ApplyConfig runs still produce output. Timestamps
-// follow slog's default RFC3339Nano format; set logging.disable_timestamp
-// in config to suppress them.
-//
-// ApplyConfig and the SSH commands discover the reconfig surface via
-// structural type-assertion on l.Handler(), so replacement implementations
-// (tests, platform-specific sinks) need only implement the subset of
-// {SetLevel(slog.Level), SetFormat(string) error, SetDisableTimestamp(bool)}
-// they care about. Callers that pass a plain *slog.Logger without these
-// methods get a silent no-op; reconfiguration is always opt-in.
-func NewLogger(w io.Writer) *slog.Logger {
-	return slog.New(NewHandler(w))
-}
-
-// NewHandler builds the *Handler that NewLogger wraps. Exported for
-// platform-specific sinks (notably cmd/nebula-service/logs_windows.go)
-// that want to wrap the handler with extra behavior, such as tagging each
-// record with its Event Log severity, while still benefiting from all the
-// level / format / timestamp / WithAttrs machinery implemented here.
-func NewHandler(w io.Writer) *Handler {
-	root := &handlerRoot{}
-	root.level.Set(slog.LevelInfo)
-	opts := &slog.HandlerOptions{Level: &root.level}
-	return &Handler{
-		root: root,
-		text: slog.NewTextHandler(w, opts),
-		json: slog.NewJSONHandler(w, opts),
-	}
-}
-
-// handlerRoot carries the reconfiguration state shared by every logger
-// derived from a NewHandler call. All fields are consulted on the log
-// path and updated lock-free.
-type handlerRoot struct {
-	level            slog.LevelVar
-	disableTimestamp atomic.Bool
-	// jsonMode picks which of the pre-derived inner handlers Handler.Handle
-	// dispatches to. Flipping it propagates instantly to every derived logger
-	// without rebuilding or chain-replaying anything.
-	jsonMode atomic.Bool
-}
-
-// Handler is the slog.Handler returned by NewHandler. It holds two
-// pre-derived slog handlers -- one text, one json -- both built from the
-// same accumulated WithAttrs/WithGroup state. Handle picks which one to
-// dispatch to based on handlerRoot.jsonMode, so a SetFormat call takes
-// effect immediately across the whole process without having to rebuild
-// any derived loggers.
-type Handler struct {
-	root *handlerRoot
-	text slog.Handler
-	json slog.Handler
-}
-
-func (h *Handler) Enabled(_ context.Context, l slog.Level) bool {
-	return h.root.level.Level() <= l
-}
-
-func (h *Handler) Handle(ctx context.Context, r slog.Record) error {
-	if h.root.disableTimestamp.Load() {
-		r.Time = time.Time{}
-	}
-	if h.root.jsonMode.Load() {
-		return h.json.Handle(ctx, r)
-	}
-	return h.text.Handle(ctx, r)
-}
-
-func (h *Handler) WithAttrs(attrs []slog.Attr) slog.Handler {
-	if len(attrs) == 0 {
-		return h
-	}
-	return &Handler{
-		root: h.root,
-		text: h.text.WithAttrs(attrs),
-		json: h.json.WithAttrs(attrs),
-	}
-}
-
-func (h *Handler) WithGroup(name string) slog.Handler {
-	if name == "" {
-		return h
-	}
-	return &Handler{
-		root: h.root,
-		text: h.text.WithGroup(name),
-		json: h.json.WithGroup(name),
-	}
-}
-
-// SetLevel updates the effective log level. Propagates to every derived
-// logger via the shared LevelVar.
-func (h *Handler) SetLevel(level slog.Level) { h.root.level.Set(level) }
-
-// GetLevel reports the current log level.
-func (h *Handler) GetLevel() slog.Level { return h.root.level.Level() }
-
-// SetFormat flips the output format atomically. Valid formats are "text"
-// and "json". Every derived logger sees the new format on its next Handle
-// call; no rebuild or registration is required.
-func (h *Handler) SetFormat(format string) error {
-	switch format {
-	case "text":
-		h.root.jsonMode.Store(false)
-	case "json":
-		h.root.jsonMode.Store(true)
-	default:
-		return fmt.Errorf("unknown log format `%s`. possible formats: %s", format, []string{"text", "json"})
-	}
-	return nil
-}
-
-// GetFormat reports the currently selected format name.
-func (h *Handler) GetFormat() string {
-	if h.root.jsonMode.Load() {
-		return "json"
-	}
-	return "text"
-}
-
-// SetDisableTimestamp toggles whether Handle zeroes r.Time before
-// dispatching (slog's builtin text/json handlers skip emitting the time
-// attribute on a zero time).
-func (h *Handler) SetDisableTimestamp(v bool) { h.root.disableTimestamp.Store(v) }
-
-// ApplyConfig reads logging.level, logging.format, and (optionally)
-// logging.disable_timestamp from c and applies them to l. The reconfig
-// surface is discovered via structural type-assertion on l.Handler(), so
-// foreign handlers silently opt out of whichever capabilities they do not
-// implement.
-//
-// nebula.Main does NOT call this function on your behalf; callers that want
-// config-driven log level / format / timestamp updates invoke it at
-// startup and register it as a reload callback themselves. This keeps the
-// library from mutating an embedder's logger without their say-so.
-func ApplyConfig(l *slog.Logger, c Config) error {
-	h := l.Handler()
-
-	lvl, err := ParseLevel(strings.ToLower(c.GetString("logging.level", "info")))
-	if err != nil {
-		return err
-	}
-	if ls, ok := h.(interface{ SetLevel(slog.Level) }); ok {
-		ls.SetLevel(lvl)
-	}
-
-	format := strings.ToLower(c.GetString("logging.format", "text"))
-	if fs, ok := h.(interface{ SetFormat(string) error }); ok {
-		if err := fs.SetFormat(format); err != nil {
-			return err
-		}
-	}
-
-	if ts, ok := h.(interface{ SetDisableTimestamp(bool) }); ok {
-		ts.SetDisableTimestamp(c.GetBool("logging.disable_timestamp", false))
-	}
-	return nil
-}
-
-// ParseLevel converts a config-string level name ("trace", "debug", "info",
-// "warn"/"warning", "error", "fatal"/"panic") to a slog.Level. "fatal" and
-// "panic" are accepted for backwards compatibility with pre-slog configs
-// and both map to slog.LevelError.
-func ParseLevel(s string) (slog.Level, error) {
-	switch s {
-	case "trace":
-		return LevelTrace, nil
-	case "debug":
-		return slog.LevelDebug, nil
-	case "info":
-		return slog.LevelInfo, nil
-	case "warn", "warning":
-		return slog.LevelWarn, nil
-	case "error":
-		return slog.LevelError, nil
-	case "fatal", "panic":
-		return slog.LevelError, nil
-	default:
-		return 0, fmt.Errorf("not a valid logging level: %q", s)
-	}
-}
-
-// LevelName returns a human-readable name for a slog.Level matching the
-// strings accepted by ParseLevel.
-func LevelName(l slog.Level) string {
-	switch {
-	case l <= LevelTrace:
-		return "trace"
-	case l <= slog.LevelDebug:
-		return "debug"
-	case l <= slog.LevelInfo:
-		return "info"
-	case l <= slog.LevelWarn:
-		return "warn"
-	default:
-		return "error"
-	}
-}

logging/logger_bench_test.go

@@ -1,90 +0,0 @@
-package logging
-
-import (
-	"context"
-	"io"
-	"log/slog"
-	"testing"
-)
-
-// BenchmarkLogger_* compare the handler returned by NewLogger against a
-// stock slog text handler. The key thing we care about is the per-log
-// cost on a logger that has been derived via .With(), because that is the
-// shape subsystems store on their structs (HostInfo.logger(),
-// lh.l.With("subsystem", ...), etc.) and call from hot paths.
-
-func BenchmarkLogger_Stock_RootInfo(b *testing.B) {
-	l := slog.New(slog.DiscardHandler)
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		l.Info("hello", "i", i)
-	}
-}
-
-func BenchmarkLogger_Nebula_RootInfo(b *testing.B) {
-	l := NewLogger(io.Discard)
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		l.Info("hello", "i", i)
-	}
-}
-
-func BenchmarkLogger_Stock_DerivedInfo(b *testing.B) {
-	l := slog.New(slog.DiscardHandler).With(
-		"subsystem", "bench",
-		"localIndex", 1234,
-	)
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		l.Info("hello", "i", i)
-	}
-}
-
-func BenchmarkLogger_Nebula_DerivedInfo(b *testing.B) {
-	l := NewLogger(io.Discard).With(
-		"subsystem", "bench",
-		"localIndex", 1234,
-	)
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		l.Info("hello", "i", i)
-	}
-}
-
-// Gated-off-path benchmarks: mimic the typical hot-path shape
-// `if l.Enabled(ctx, slog.LevelDebug) { ... }` where the log is gated below
-// the active level. This is the dominant pattern in inside.go/outside.go and
-// what we pay on every packet.
-func BenchmarkLogger_Stock_DerivedEnabledGateMiss(b *testing.B) {
-	l := slog.New(slog.DiscardHandler).With(
-		"subsystem", "bench",
-		"localIndex", 1234,
-	)
-	ctx := context.Background()
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		if l.Enabled(ctx, slog.LevelDebug) {
-			l.Debug("hello", "i", i)
-		}
-	}
-}
-
-func BenchmarkLogger_Nebula_DerivedEnabledGateMiss(b *testing.B) {
-	l := NewLogger(io.Discard).With(
-		"subsystem", "bench",
-		"localIndex", 1234,
-	)
-	ctx := context.Background()
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		if l.Enabled(ctx, slog.LevelDebug) {
-			l.Debug("hello", "i", i)
-		}
-	}
-}

main.go

@@ -3,13 +3,13 @@ package nebula
 import (
 	"context"
 	"fmt"
-	"log/slog"
 	"net"
 	"net/netip"
 	"runtime/debug"
 	"strings"
 	"time"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/overlay"
 	"github.com/slackhq/nebula/sshd"
@@ -20,7 +20,7 @@ import (
 
 type m = map[string]any
 
-func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, deviceFactory overlay.DeviceFactory) (retcon *Control, reterr error) {
+func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logger, deviceFactory overlay.DeviceFactory) (retcon *Control, reterr error) {
 	ctx, cancel := context.WithCancel(context.Background())
 	// Automatically cancel the context if Main returns an error, to signal all created goroutines to quit.
 	defer func() {
@@ -33,6 +33,11 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		buildVersion = moduleVersion()
 	}
 
+	l := logger
+	l.Formatter = &logrus.TextFormatter{
+		FullTimestamp: true,
+	}
+
 	// Print the config if in test, the exit comes later
 	if configTest {
 		b, err := yaml.Marshal(c.Settings)
@@ -41,9 +46,21 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		}
 
 		// Print the final config
-		l.Info(string(b))
+		l.Println(string(b))
 	}
 
+	err := configLogger(l, c)
+	if err != nil {
+		return nil, util.ContextualizeIfNeeded("Failed to configure the logger", err)
+	}
+
+	c.RegisterReloadCallback(func(c *config.C) {
+		err := configLogger(l, c)
+		if err != nil {
+			l.WithError(err).Error("Failed to configure the logger")
+		}
+	})
+
 	pki, err := NewPKIFromConfig(l, c)
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Failed to load PKI from config", err)
@@ -53,9 +70,9 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Error while loading firewall rules", err)
 	}
-	l.Info("Firewall started", "firewallHashes", fw.GetRuleHashes())
+	l.WithField("firewallHashes", fw.GetRuleHashes()).Info("Firewall started")
 
-	ssh, err := sshd.NewSSHServer(l.With("subsystem", "sshd"))
+	ssh, err := sshd.NewSSHServer(l.WithField("subsystem", "sshd"))
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Error while creating SSH server", err)
 	}
@@ -64,7 +81,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 	if c.GetBool("sshd.enabled", false) {
 		sshStart, err = configSSH(l, ssh, c)
 		if err != nil {
-			l.Warn("Failed to configure sshd, ssh debugging will not be available", "error", err)
+			l.WithError(err).Warn("Failed to configure sshd, ssh debugging will not be available")
 			sshStart = nil
 		}
 	}
@@ -82,15 +99,19 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 			routines = 1
 		}
 		if routines > 1 {
-			l.Info("Using multiple routines", "routines", routines)
+			l.WithField("routines", routines).Info("Using multiple routines")
 		}
 	} else {
 		// deprecated and undocumented
 		tunQueues := c.GetInt("tun.routines", 1)
 		udpQueues := c.GetInt("listen.routines", 1)
-		routines = max(tunQueues, udpQueues)
+		if tunQueues > udpQueues {
+			routines = tunQueues
+		} else {
+			routines = udpQueues
+		}
 		if routines != 1 {
-			l.Warn("Setting tun.routines and listen.routines is deprecated. Use `routines` instead", "routines", routines)
+			l.WithField("routines", routines).Warn("Setting tun.routines and listen.routines is deprecated. Use `routines` instead")
 		}
 	}
 
@@ -103,7 +124,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		conntrackCacheTimeout = 1 * time.Second
 	}
 	if conntrackCacheTimeout > 0 {
-		l.Info("Using routine-local conntrack cache", "duration", conntrackCacheTimeout)
+		l.WithField("duration", conntrackCacheTimeout).Info("Using routine-local conntrack cache")
 	}
 
 	var tun overlay.Device
@@ -149,7 +170,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		}
 
 		for i := 0; i < routines; i++ {
-			l.Info("listening", "addr", netip.AddrPortFrom(listenHost, uint16(port)))
+			l.Infof("listening on %v", netip.AddrPortFrom(listenHost, uint16(port)))
 			udpServer, err := udp.NewListener(l, listenHost, port, routines > 1, c.GetInt("listen.batch", 64))
 			if err != nil {
 				return nil, util.NewContextualError("Failed to open udp listener", m{"queue": i}, err)
@@ -198,9 +219,13 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 	handshakeManager := NewHandshakeManager(l, hostMap, lightHouse, udpConns[0], handshakeConfig)
 	lightHouse.handshakeTrigger = handshakeManager.trigger
 
-	ds, err := newDnsServerFromConfig(ctx, l, pki.getCertState(), hostMap, c)
-	if err != nil {
-		l.Warn("Failed to start DNS responder", "error", err)
+	serveDns := false
+	if c.GetBool("lighthouse.serve_dns", false) {
+		if c.GetBool("lighthouse.am_lighthouse", false) {
+			serveDns = true
+		} else {
+			l.Warn("DNS server refusing to run because this host is not a lighthouse.")
+		}
 	}
 
 	ifConfig := &InterfaceConfig{
@@ -209,7 +234,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		Outside:               udpConns[0],
 		pki:                   pki,
 		Firewall:              fw,
-		DnsServer:             ds,
+		ServeDns:              serveDns,
 		HandshakeManager:      handshakeManager,
 		connectionManager:     connManager,
 		lightHouse:            lightHouse,
@@ -240,13 +265,12 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 		ifce.RegisterConfigChangeCallbacks(c)
 		ifce.reloadDisconnectInvalid(c)
 		ifce.reloadSendRecvError(c)
-		ifce.reloadAcceptRecvError(c)
 
 		handshakeManager.f = ifce
 		go handshakeManager.Run(ctx)
 	}
 
-	stats, err := newStatsServerFromConfig(ctx, l, c, buildVersion, configTest)
+	statsStart, err := startStats(l, c, buildVersion, configTest)
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Failed to start stats emitter", err)
 	}
@@ -259,17 +283,23 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
 
 	attachCommands(l, c, ssh, ifce)
 
+	// Start DNS server last to allow using the nebula IP as lighthouse.dns.host
+	var dnsStart func()
+	if lightHouse.amLighthouse && serveDns {
+		l.Debugln("Starting dns server")
+		dnsStart = dnsMain(l, pki.getCertState(), hostMap, c)
+	}
+
 	return &Control{
-		state:                  StateReady,
-		f:                      ifce,
-		l:                      l,
-		ctx:                    ctx,
-		cancel:                 cancel,
-		sshStart:               sshStart,
-		statsStart:             stats.Start,
-		dnsStart:               ds.Start,
-		lighthouseStart:        lightHouse.StartUpdateWorker,
-		connectionManagerStart: connManager.Start,
+		ifce,
+		l,
+		ctx,
+		cancel,
+		sshStart,
+		statsStart,
+		dnsStart,
+		lightHouse.StartUpdateWorker,
+		connManager.Start,
 	}, nil
 }
 

noise.go

@@ -15,12 +15,14 @@ type endianness interface {
 var noiseEndianness endianness = binary.BigEndian
 
 type NebulaCipherState struct {
-	c cipher.AEAD
+	c noise.Cipher
+	//k [32]byte
+	//n uint64
 }
 
 func NewNebulaCipherState(s *noise.CipherState) *NebulaCipherState {
-	x := s.Cipher()
-	return &NebulaCipherState{c: x.(cipher.AEAD)}
+	return &NebulaCipherState{c: s.Cipher()}
+
 }
 
 // EncryptDanger encrypts and authenticates a given payload.
@@ -44,7 +46,7 @@ func (s *NebulaCipherState) EncryptDanger(out, ad, plaintext []byte, n uint64, n
 		nb[2] = 0
 		nb[3] = 0
 		noiseEndianness.PutUint64(nb[4:], n)
-		out = s.c.Seal(out, nb, plaintext, ad)
+		out = s.c.(cipher.AEAD).Seal(out, nb, plaintext, ad)
 		//l.Debugf("Encryption: outlen: %d, nonce: %d, ad: %s, plainlen %d", len(out), n, ad, len(plaintext))
 		return out, nil
 	} else {
@@ -59,7 +61,7 @@ func (s *NebulaCipherState) DecryptDanger(out, ad, ciphertext []byte, n uint64,
 		nb[2] = 0
 		nb[3] = 0
 		noiseEndianness.PutUint64(nb[4:], n)
-		return s.c.Open(out, nb, ciphertext, ad)
+		return s.c.(cipher.AEAD).Open(out, nb, ciphertext, ad)
 	} else {
 		return []byte{}, nil
 	}
@@ -67,7 +69,7 @@ func (s *NebulaCipherState) DecryptDanger(out, ad, ciphertext []byte, n uint64,
 
 func (s *NebulaCipherState) Overhead() int {
 	if s != nil {
-		return s.c.Overhead()
+		return s.c.(cipher.AEAD).Overhead()
 	}
 	return 0
 }

noiseutil/boring.go

@@ -22,7 +22,7 @@ const EncryptLockNeeded = true
 // NewGCMTLS is no longer exposed in go1.19+, so we need to link it in
 // See: https://github.com/golang/go/issues/56326
 //
-// NewGCMTLS is the internal method used with boringcrypto that provides a
+// NewGCMTLS is the internal method used with boringcrypto that provices a
 // validated mode of AES-GCM which enforces the nonce is strictly
 // monotonically increasing.  This is the TLS 1.2 specification for nonce
 // generation (which also matches the method used by the Noise Protocol)

notboring.go

@@ -1,4 +1,5 @@
 //go:build !boringcrypto
+// +build !boringcrypto
 
 package nebula
 

outside.go

@@ -1,16 +1,16 @@
 package nebula
 
 import (
-	"context"
 	"encoding/binary"
 	"errors"
-	"log/slog"
 	"net/netip"
 	"time"
 
 	"github.com/google/gopacket/layers"
+	"github.com/slackhq/nebula/packet"
 	"golang.org/x/net/ipv6"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/firewall"
 	"github.com/slackhq/nebula/header"
 	"golang.org/x/net/ipv4"
@@ -20,34 +20,108 @@ const (
 	minFwPacketLen = 4
 )
 
-func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte, h *header.H, fwPacket *firewall.Packet, lhf *LightHouseHandler, nb []byte, q int, localCache firewall.ConntrackCache) {
-	err := h.Parse(packet)
+// handleRelayPackets handles relay packets. Returns false if there's nothing left to do, true for continuing to process an unwrapped TerminalType packet
+// scratch must be large enough to contain a packet to be relayed if needed
+func (f *Interface) handleRelayPackets(via ViaSender, hostinfo *HostInfo, segment []byte, scratch []byte, h *header.H, nb []byte) ([]byte, *ViaSender, bool) {
+	var err error
+	// The entire body is sent as AD, not encrypted.
+	// The packet consists of a 16-byte parsed Nebula header, Associated Data-protected payload, and a trailing 16-byte AEAD signature value.
+	// The packet is guaranteed to be at least 16 bytes at this point, b/c it got past the h.Parse() call above. If it's
+	// otherwise malformed (meaning, there is no trailing 16 byte AEAD value), then this will result in at worst a 0-length slice
+	// which will gracefully fail in the DecryptDanger call.
+	signedPayload := segment[:len(segment)-hostinfo.ConnectionState.dKey.Overhead()]
+	signatureValue := segment[len(segment)-hostinfo.ConnectionState.dKey.Overhead():]
+	scratch, err = hostinfo.ConnectionState.dKey.DecryptDanger(scratch, signedPayload, signatureValue, h.MessageCounter, nb)
 	if err != nil {
-		// Hole punch packets are 0 or 1 byte big, so lets ignore printing those errors
-		if len(packet) > 1 {
-			f.l.Info("Error while parsing inbound packet",
-				"from", via,
-				"error", err,
-				"packet", packet,
-			)
-		}
-		return
+		return nil, nil, false
+	}
+	// Successfully validated the thing. Get rid of the Relay header.
+	signedPayload = signedPayload[header.Len:]
+	// Pull the Roaming parts up here, and return in all call paths.
+	f.handleHostRoaming(hostinfo, via)
+	// Track usage of both the HostInfo and the Relay for the received & authenticated packet
+	f.connectionManager.In(hostinfo)
+	f.connectionManager.RelayUsed(h.RemoteIndex)
+
+	relay, ok := hostinfo.relayState.QueryRelayForByIdx(h.RemoteIndex)
+	if !ok {
+		// The only way this happens is if hostmap has an index to the correct HostInfo, but the HostInfo is missing
+		// its internal mapping. This should never happen.
+		hostinfo.logger(f.l).WithFields(logrus.Fields{"vpnAddrs": hostinfo.vpnAddrs, "remoteIndex": h.RemoteIndex}).Error("HostInfo missing remote relay index")
+		return nil, nil, false
 	}
 
-	//l.Error("in packet ", header, packet[HeaderLen:])
-	if !via.IsRelayed {
-		if f.myVpnNetworksTable.Contains(via.UdpAddr.Addr()) {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				f.l.Debug("Refusing to process double encrypted packet", "from", via)
+	switch relay.Type {
+	case TerminalType:
+		// If I am the target of this relay, process the unwrapped packet
+		// We need to re-write our variables to ensure this segment is correctly parsed.
+		// We could set up for a recursive call here, but this makes it easier to prove that we'll never stack-overflow
+
+		//mirrors the top of readOutsideSegment
+		err = h.Parse(signedPayload)
+		if err != nil {
+			// Hole punch packets are 0 or 1 byte big, so let's ignore printing those errors
+			if len(signedPayload) > 1 {
+				f.l.WithField("packet", segment).Infof("Error while parsing inbound packet from %s: %s", via, err)
 			}
-			return
+			return nil, nil, false
 		}
+		newVia := &ViaSender{
+			UdpAddr:   via.UdpAddr,
+			relayHI:   hostinfo,
+			remoteIdx: relay.RemoteIndex,
+			relay:     relay,
+			IsRelayed: true,
+		}
+		//continue flowing through readOutsideSegment()
+		return signedPayload, newVia, true
+	case ForwardingType:
+		// Find the target HostInfo relay object
+		targetHI, targetRelay, err := f.hostMap.QueryVpnAddrsRelayFor(hostinfo.vpnAddrs, relay.PeerAddr)
+		if err != nil {
+			hostinfo.logger(f.l).WithField("relayTo", relay.PeerAddr).WithError(err).WithField("hostinfo.vpnAddrs", hostinfo.vpnAddrs).Info("Failed to find target host info by ip")
+			return nil, nil, false
+		}
+
+		// If that relay is Established, forward the payload through it
+		if targetRelay.State == Established {
+			switch targetRelay.Type {
+			case ForwardingType:
+				// Forward this packet through the relay tunnel, and find the target HostInfo
+				f.SendVia(targetHI, targetRelay, signedPayload, nb, scratch[:0], false) //todo it would be nice to queue this up and do it later, or at least avoid a memcpy of signedPayload
+			case TerminalType:
+				hostinfo.logger(f.l).Error("Unexpected Relay Type of Terminal")
+			default:
+				hostinfo.logger(f.l).WithField("targetRelay.Type", targetRelay.Type).Error("Unexpected Relay Type")
+			}
+		} else {
+			hostinfo.logger(f.l).WithFields(logrus.Fields{"relayTo": relay.PeerAddr, "relayFrom": hostinfo.vpnAddrs[0], "targetRelayState": targetRelay.State}).Info("Unexpected target relay state")
+		}
+	}
+	return nil, nil, false
+}
+
+func (f *Interface) readOutsideSegment(via ViaSender, segment []byte, out *packet.OutPacket, lhf *LightHouseHandler, s *Scratches, q int, localCache firewall.ConntrackCache, now time.Time) {
+	h := s.h
+	err := h.Parse(segment)
+	if err != nil {
+		// Hole punch packets are 0 or 1 byte big, so let's ignore printing those errors
+		if len(segment) > 1 {
+			f.l.WithField("packet", segment).Infof("Error while parsing inbound packet from %s: %s", via, err)
+		}
+		return
 	}
 
 	var hostinfo *HostInfo
 	// verify if we've seen this index before, otherwise respond to the handshake initiation
 	if h.Type == header.Message && h.Subtype == header.MessageRelay {
 		hostinfo = f.hostMap.QueryRelayIndex(h.RemoteIndex)
+		newSegment, newVia, keepGoing := f.handleRelayPackets(via, hostinfo, segment, s.scratch, h, s.nb)
+		if !keepGoing {
+			return
+		}
+		via = *newVia
+		segment = newSegment
 	} else {
 		hostinfo = f.hostMap.QueryIndex(h.RemoteIndex)
 	}
@@ -65,85 +139,13 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 
 		switch h.Subtype {
 		case header.MessageNone:
-			if !f.decryptToTun(hostinfo, h.MessageCounter, out, packet, fwPacket, nb, q, localCache) {
+			if !f.decryptToTunDelayWrite(hostinfo, h.MessageCounter, out, segment, s.fwPacket, s.nb, q, localCache, now) {
+				out.DestroyLastSegment() //prevent a rejected segment from being used
 				return
 			}
 		case header.MessageRelay:
-			// The entire body is sent as AD, not encrypted.
-			// The packet consists of a 16-byte parsed Nebula header, Associated Data-protected payload, and a trailing 16-byte AEAD signature value.
-			// The packet is guaranteed to be at least 16 bytes at this point, b/c it got past the h.Parse() call above. If it's
-			// otherwise malformed (meaning, there is no trailing 16 byte AEAD value), then this will result in at worst a 0-length slice
-			// which will gracefully fail in the DecryptDanger call.
-			signedPayload := packet[:len(packet)-hostinfo.ConnectionState.dKey.Overhead()]
-			signatureValue := packet[len(packet)-hostinfo.ConnectionState.dKey.Overhead():]
-			out, err = hostinfo.ConnectionState.dKey.DecryptDanger(out, signedPayload, signatureValue, h.MessageCounter, nb)
-			if err != nil {
-				return
-			}
-			// Successfully validated the thing. Get rid of the Relay header.
-			signedPayload = signedPayload[header.Len:]
-			// Pull the Roaming parts up here, and return in all call paths.
-			f.handleHostRoaming(hostinfo, via)
-			// Track usage of both the HostInfo and the Relay for the received & authenticated packet
-			f.connectionManager.In(hostinfo)
-			f.connectionManager.RelayUsed(h.RemoteIndex)
-
-			relay, ok := hostinfo.relayState.QueryRelayForByIdx(h.RemoteIndex)
-			if !ok {
-				// The only way this happens is if hostmap has an index to the correct HostInfo, but the HostInfo is missing
-				// its internal mapping. This should never happen.
-				hostinfo.logger(f.l).Error("HostInfo missing remote relay index",
-					"vpnAddrs", hostinfo.vpnAddrs,
-					"remoteIndex", h.RemoteIndex,
-				)
-				return
-			}
-
-			switch relay.Type {
-			case TerminalType:
-				// If I am the target of this relay, process the unwrapped packet
-				// From this recursive point, all these variables are 'burned'. We shouldn't rely on them again.
-				via = ViaSender{
-					UdpAddr:   via.UdpAddr,
-					relayHI:   hostinfo,
-					remoteIdx: relay.RemoteIndex,
-					relay:     relay,
-					IsRelayed: true,
-				}
-				f.readOutsidePackets(via, out[:0], signedPayload, h, fwPacket, lhf, nb, q, localCache)
-				return
-			case ForwardingType:
-				// Find the target HostInfo relay object
-				targetHI, targetRelay, err := f.hostMap.QueryVpnAddrsRelayFor(hostinfo.vpnAddrs, relay.PeerAddr)
-				if err != nil {
-					hostinfo.logger(f.l).Info("Failed to find target host info by ip",
-						"relayTo", relay.PeerAddr,
-						"error", err,
-						"hostinfo.vpnAddrs", hostinfo.vpnAddrs,
-					)
-					return
-				}
-
-				// If that relay is Established, forward the payload through it
-				if targetRelay.State == Established {
-					switch targetRelay.Type {
-					case ForwardingType:
-						// Forward this packet through the relay tunnel
-						// Find the target HostInfo
-						f.SendVia(targetHI, targetRelay, signedPayload, nb, out, false)
-						return
-					case TerminalType:
-						hostinfo.logger(f.l).Error("Unexpected Relay Type of Terminal")
-					}
-				} else {
-					hostinfo.logger(f.l).Info("Unexpected target relay state",
-						"relayTo", relay.PeerAddr,
-						"relayFrom", hostinfo.vpnAddrs[0],
-						"targetRelayState", targetRelay.State,
-					)
-					return
-				}
-			}
+			f.l.Error("relayed messages cannot contain relay messages, dropping packet")
+			return
 		}
 
 	case header.LightHouse:
@@ -152,17 +154,14 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 			return
 		}
 
-		d, err := f.decrypt(hostinfo, h.MessageCounter, out, packet, h, nb)
+		d, err := f.decrypt(hostinfo, h.MessageCounter, s.scratch, segment, h, s.nb)
 		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to decrypt lighthouse packet",
-				"error", err,
-				"from", via,
-				"packet", packet,
-			)
+			hostinfo.logger(f.l).WithError(err).WithField("udpAddr", via.UdpAddr).
+				WithField("packet", segment).
+				Error("Failed to decrypt lighthouse packet")
 			return
 		}
 
-		//TODO: assert via is not relayed
 		lhf.HandleRequest(via.UdpAddr, hostinfo.vpnAddrs, d, f)
 
 		// Fallthrough to the bottom to record incoming traffic
@@ -173,13 +172,11 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 			return
 		}
 
-		d, err := f.decrypt(hostinfo, h.MessageCounter, out, packet, h, nb)
+		d, err := f.decrypt(hostinfo, h.MessageCounter, s.scratch, segment, h, s.nb)
 		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to decrypt test packet",
-				"error", err,
-				"from", via,
-				"packet", packet,
-			)
+			hostinfo.logger(f.l).WithError(err).WithField("udpAddr", via).
+				WithField("packet", segment).
+				Error("Failed to decrypt test packet")
 			return
 		}
 
@@ -187,7 +184,7 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 			// This testRequest might be from TryPromoteBest, so we should roam
 			// to the new IP address before responding
 			f.handleHostRoaming(hostinfo, via)
-			f.send(header.Test, header.TestReply, ci, hostinfo, d, nb, out)
+			f.send(header.Test, header.TestReply, ci, hostinfo, d, s.nb, s.scratch)
 		}
 
 		// Fallthrough to the bottom to record incoming traffic
@@ -197,7 +194,7 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 
 	case header.Handshake:
 		f.messageMetrics.Rx(h.Type, h.Subtype, 1)
-		f.handshakeManager.HandleIncoming(via, packet, h)
+		f.handshakeManager.HandleIncoming(via, segment, h)
 		return
 
 	case header.RecvError:
@@ -210,17 +207,9 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 		if !f.handleEncrypted(ci, via, h) {
 			return
 		}
-		_, err = f.decrypt(hostinfo, h.MessageCounter, out, packet, h, nb)
-		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to decrypt CloseTunnel packet",
-				"error", err,
-				"from", via,
-				"packet", packet,
-			)
-			return
-		}
 
-		hostinfo.logger(f.l).Info("Close tunnel received, tearing down.", "from", via)
+		hostinfo.logger(f.l).WithField("udpAddr", via).
+			Info("Close tunnel received, tearing down.")
 
 		f.closeTunnel(hostinfo)
 		return
@@ -230,13 +219,11 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 			return
 		}
 
-		d, err := f.decrypt(hostinfo, h.MessageCounter, out, packet, h, nb)
+		d, err := f.decrypt(hostinfo, h.MessageCounter, s.scratch, segment, h, s.nb)
 		if err != nil {
-			hostinfo.logger(f.l).Error("Failed to decrypt Control packet",
-				"error", err,
-				"from", via,
-				"packet", packet,
-			)
+			hostinfo.logger(f.l).WithError(err).WithField("udpAddr", via).
+				WithField("packet", segment).
+				Error("Failed to decrypt Control packet")
 			return
 		}
 
@@ -244,9 +231,7 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 
 	default:
 		f.messageMetrics.Rx(h.Type, h.Subtype, 1)
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("Unexpected packet received", "from", via)
-		}
+		hostinfo.logger(f.l).Debugf("Unexpected packet received from %s", via)
 		return
 	}
 
@@ -255,6 +240,28 @@ func (f *Interface) readOutsidePackets(via ViaSender, out []byte, packet []byte,
 	f.connectionManager.In(hostinfo)
 }
 
+func (f *Interface) readOutsidePacketsMany(packets []*packet.UDPPacket, out []*packet.OutPacket, lhf *LightHouseHandler, s *Scratches, q int, localCache firewall.ConntrackCache, now time.Time) {
+	for i, pkt := range packets {
+		via := ViaSender{UdpAddr: pkt.AddrPort()}
+
+		//l.Error("in packet ", header, packet[HeaderLen:])
+		if f.myVpnNetworksTable.Contains(via.UdpAddr.Addr()) {
+			if f.l.Level >= logrus.DebugLevel {
+				f.l.WithField("from", via).Debug("Refusing to process double encrypted packet")
+			}
+			return
+		}
+
+		for segment := range pkt.Segments() {
+			f.readOutsideSegment(via, segment, out[i], lhf, s, q, localCache, now)
+		}
+		//_, err := f.readers[q].WriteOne(out[i], false, q)
+		//if err != nil {
+		//	f.l.WithError(err).Error("Failed to write packet")
+		//}
+	}
+}
+
 // closeTunnel closes a tunnel locally, it does not send a closeTunnel packet to the remote
 func (f *Interface) closeTunnel(hostInfo *HostInfo) {
 	final := f.hostMap.DeleteHostInfo(hostInfo)
@@ -272,27 +279,20 @@ func (f *Interface) sendCloseTunnel(h *HostInfo) {
 func (f *Interface) handleHostRoaming(hostinfo *HostInfo, via ViaSender) {
 	if !via.IsRelayed && hostinfo.remote != via.UdpAddr {
 		if !f.lightHouse.GetRemoteAllowList().AllowAll(hostinfo.vpnAddrs, via.UdpAddr.Addr()) {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				hostinfo.logger(f.l).Debug("lighthouse.remote_allow_list denied roaming", "newAddr", via.UdpAddr)
-			}
+			hostinfo.logger(f.l).WithField("newAddr", via.UdpAddr).Debug("lighthouse.remote_allow_list denied roaming")
 			return
 		}
 
 		if !hostinfo.lastRoam.IsZero() && via.UdpAddr == hostinfo.lastRoamRemote && time.Since(hostinfo.lastRoam) < RoamingSuppressSeconds*time.Second {
-			if f.l.Enabled(context.Background(), slog.LevelDebug) {
-				hostinfo.logger(f.l).Debug("Suppressing roam back to previous remote",
-					"suppressSeconds", RoamingSuppressSeconds,
-					"udpAddr", hostinfo.remote,
-					"newAddr", via.UdpAddr,
-				)
+			if f.l.Level >= logrus.DebugLevel {
+				hostinfo.logger(f.l).WithField("udpAddr", hostinfo.remote).WithField("newAddr", via.UdpAddr).
+					Debugf("Suppressing roam back to previous remote for %d seconds", RoamingSuppressSeconds)
 			}
 			return
 		}
 
-		hostinfo.logger(f.l).Info("Host roamed to new udp ip/port.",
-			"udpAddr", hostinfo.remote,
-			"newAddr", via.UdpAddr,
-		)
+		hostinfo.logger(f.l).WithField("udpAddr", hostinfo.remote).WithField("newAddr", via.UdpAddr).
+			Info("Host roamed to new udp ip/port.")
 		hostinfo.lastRoam = time.Now()
 		hostinfo.lastRoamRemote = hostinfo.remote
 		hostinfo.SetRemote(via.UdpAddr)
@@ -332,11 +332,11 @@ func newPacket(data []byte, incoming bool, fp *firewall.Packet) error {
 		return ErrPacketTooShort
 	}
 
-	version := int((data[0] >> 4) & 0x0f)
-	switch version {
-	case ipv4.Version:
+	//version := int((data[0] >> 4) & 0x0f)
+	switch data[0] & 0xf0 {
+	case ipv4.Version << 4:
 		return parseV4(data, incoming, fp)
-	case ipv6.Version:
+	case ipv6.Version << 4:
 		return parseV6(data, incoming, fp)
 	}
 	return ErrUnknownIPVersion
@@ -366,29 +366,13 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 		proto := layers.IPProtocol(data[protoAt])
 
 		switch proto {
-		case layers.IPProtocolESP, layers.IPProtocolNoNextHeader:
+		case layers.IPProtocolICMPv6, layers.IPProtocolESP, layers.IPProtocolNoNextHeader:
 			fp.Protocol = uint8(proto)
 			fp.RemotePort = 0
 			fp.LocalPort = 0
 			fp.Fragment = false
 			return nil
 
-		case layers.IPProtocolICMPv6:
-			if dataLen < offset+6 {
-				return ErrIPv6PacketTooShort
-			}
-			fp.Protocol = uint8(proto)
-			fp.LocalPort = 0 //incoming vs outgoing doesn't matter for icmpv6
-			icmptype := data[offset+1]
-			switch icmptype {
-			case layers.ICMPv6TypeEchoRequest, layers.ICMPv6TypeEchoReply:
-				fp.RemotePort = binary.BigEndian.Uint16(data[offset+4 : offset+6]) //identifier
-			default:
-				fp.RemotePort = 0
-			}
-			fp.Fragment = false
-			return nil
-
 		case layers.IPProtocolTCP, layers.IPProtocolUDP:
 			if dataLen < offset+4 {
 				return ErrIPv6PacketTooShort
@@ -478,38 +462,34 @@ func parseV4(data []byte, incoming bool, fp *firewall.Packet) error {
 
 	// Accounting for a variable header length, do we have enough data for our src/dst tuples?
 	minLen := ihl
-	if !fp.Fragment {
-		if fp.Protocol == firewall.ProtoICMP {
-			minLen += minFwPacketLen + 2
-		} else {
-			minLen += minFwPacketLen
-		}
+	if !fp.Fragment && fp.Protocol != firewall.ProtoICMP {
+		minLen += minFwPacketLen
 	}
-
 	if len(data) < minLen {
 		return ErrIPv4InvalidHeaderLength
 	}
 
-	if incoming { // Firewall packets are locally oriented
+	// Firewall packets are locally oriented
+	if incoming {
 		fp.RemoteAddr, _ = netip.AddrFromSlice(data[12:16])
 		fp.LocalAddr, _ = netip.AddrFromSlice(data[16:20])
+		if fp.Fragment || fp.Protocol == firewall.ProtoICMP {
+			fp.RemotePort = 0
+			fp.LocalPort = 0
+		} else {
+			fp.RemotePort = binary.BigEndian.Uint16(data[ihl : ihl+2])
+			fp.LocalPort = binary.BigEndian.Uint16(data[ihl+2 : ihl+4])
+		}
 	} else {
 		fp.LocalAddr, _ = netip.AddrFromSlice(data[12:16])
 		fp.RemoteAddr, _ = netip.AddrFromSlice(data[16:20])
-	}
-
-	if fp.Fragment {
-		fp.RemotePort = 0
-		fp.LocalPort = 0
-	} else if fp.Protocol == firewall.ProtoICMP { //note that orientation doesn't matter on ICMP
-		fp.RemotePort = binary.BigEndian.Uint16(data[ihl+4 : ihl+6]) //identifier
-		fp.LocalPort = 0                                             //code would be uint16(data[ihl+1])
-	} else if incoming {
-		fp.RemotePort = binary.BigEndian.Uint16(data[ihl : ihl+2])  //src port
-		fp.LocalPort = binary.BigEndian.Uint16(data[ihl+2 : ihl+4]) //dst port
-	} else {
-		fp.LocalPort = binary.BigEndian.Uint16(data[ihl : ihl+2])    //src port
-		fp.RemotePort = binary.BigEndian.Uint16(data[ihl+2 : ihl+4]) //dst port
+		if fp.Fragment || fp.Protocol == firewall.ProtoICMP {
+			fp.RemotePort = 0
+			fp.LocalPort = 0
+		} else {
+			fp.LocalPort = binary.BigEndian.Uint16(data[ihl : ihl+2])
+			fp.RemotePort = binary.BigEndian.Uint16(data[ihl+2 : ihl+4])
+		}
 	}
 
 	return nil
@@ -523,64 +503,63 @@ func (f *Interface) decrypt(hostinfo *HostInfo, mc uint64, out []byte, packet []
 	}
 
 	if !hostinfo.ConnectionState.window.Update(f.l, mc) {
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("dropping out of window packet", "header", h)
-		}
+		hostinfo.logger(f.l).WithField("header", h).
+			Debugln("dropping out of window packet")
 		return nil, errors.New("out of window packet")
 	}
 
 	return out, nil
 }
 
-func (f *Interface) decryptToTun(hostinfo *HostInfo, messageCounter uint64, out []byte, packet []byte, fwPacket *firewall.Packet, nb []byte, q int, localCache firewall.ConntrackCache) bool {
+func (f *Interface) decryptToTunDelayWrite(hostinfo *HostInfo, messageCounter uint64, out *packet.OutPacket, inSegment []byte, fwPacket *firewall.Packet, nb []byte, q int, localCache firewall.ConntrackCache, now time.Time) bool {
 	var err error
 
-	out, err = hostinfo.ConnectionState.dKey.DecryptDanger(out, packet[:header.Len], packet[header.Len:], messageCounter, nb)
+	seg, err := f.readers[q].AllocSeg(out, q)
 	if err != nil {
-		hostinfo.logger(f.l).Error("Failed to decrypt packet", "error", err)
+		f.l.WithError(err).Errorln("decryptToTunDelayWrite: failed to allocate segment")
 		return false
 	}
 
-	err = newPacket(out, true, fwPacket)
+	out.SegmentPayloads[seg] = out.SegmentPayloads[seg][:0]
+	out.SegmentPayloads[seg], err = hostinfo.ConnectionState.dKey.DecryptDanger(out.SegmentPayloads[seg], inSegment[:header.Len], inSegment[header.Len:], messageCounter, nb)
 	if err != nil {
-		hostinfo.logger(f.l).Warn("Error while validating inbound packet",
-			"error", err,
-			"packet", out,
-		)
+		hostinfo.logger(f.l).WithError(err).Error("Failed to decrypt packet")
+		return false
+	}
+
+	err = newPacket(out.SegmentPayloads[seg], true, fwPacket)
+	if err != nil {
+		hostinfo.logger(f.l).WithError(err).WithField("packet", out).
+			Warnf("Error while validating inbound packet")
 		return false
 	}
 
 	if !hostinfo.ConnectionState.window.Update(f.l, messageCounter) {
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("dropping out of window packet", "fwPacket", fwPacket)
-		}
+		hostinfo.logger(f.l).WithField("fwPacket", fwPacket).
+			Debugln("dropping out of window packet")
 		return false
 	}
 
-	dropReason := f.firewall.Drop(*fwPacket, true, hostinfo, f.pki.GetCAPool(), localCache)
+	dropReason := f.firewall.Drop(*fwPacket, true, hostinfo, f.pki.GetCAPool(), localCache, now)
 	if dropReason != nil {
 		// NOTE: We give `packet` as the `out` here since we already decrypted from it and we don't need it anymore
 		// This gives us a buffer to build the reject packet in
-		f.rejectOutside(out, hostinfo.ConnectionState, hostinfo, nb, packet, q)
-		if f.l.Enabled(context.Background(), slog.LevelDebug) {
-			hostinfo.logger(f.l).Debug("dropping inbound packet",
-				"fwPacket", fwPacket,
-				"reason", dropReason,
-			)
+		f.rejectOutside(out.SegmentPayloads[seg], hostinfo.ConnectionState, hostinfo, nb, inSegment, q)
+		if f.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(f.l).WithField("fwPacket", fwPacket).
+				WithField("reason", dropReason).
+				Debugln("dropping inbound packet")
 		}
 		return false
 	}
 
 	f.connectionManager.In(hostinfo)
-	err = f.batchers[q].Commit(out)
-	if err != nil {
-		f.l.Error("Failed to write to tun", "error", err)
-	}
+	out.Segments[seg] = out.Segments[seg][:len(out.SegmentHeaders[seg])+len(out.SegmentPayloads[seg])]
 	return true
 }
 
 func (f *Interface) maybeSendRecvError(endpoint netip.AddrPort, index uint32) {
-	if f.sendRecvErrorConfig.ShouldRecvError(endpoint) {
+	if f.sendRecvErrorConfig.ShouldSendRecvError(endpoint) {
 		f.sendRecvError(endpoint, index)
 	}
 }
@@ -590,41 +569,28 @@ func (f *Interface) sendRecvError(endpoint netip.AddrPort, index uint32) {
 
 	b := header.Encode(make([]byte, header.Len), header.Version, header.RecvError, 0, index, 0)
 	_ = f.outside.WriteTo(b, endpoint)
-	if f.l.Enabled(context.Background(), slog.LevelDebug) {
-		f.l.Debug("Recv error sent",
-			"index", index,
-			"udpAddr", endpoint,
-		)
+	if f.l.Level >= logrus.DebugLevel {
+		f.l.WithField("index", index).
+			WithField("udpAddr", endpoint).
+			Debug("Recv error sent")
 	}
 }
 
 func (f *Interface) handleRecvError(addr netip.AddrPort, h *header.H) {
-	if !f.acceptRecvErrorConfig.ShouldRecvError(addr) {
-		f.l.Debug("Recv error received, ignoring",
-			"index", h.RemoteIndex,
-			"udpAddr", addr,
-		)
-		return
-	}
-
-	if f.l.Enabled(context.Background(), slog.LevelDebug) {
-		f.l.Debug("Recv error received",
-			"index", h.RemoteIndex,
-			"udpAddr", addr,
-		)
+	if f.l.Level >= logrus.DebugLevel {
+		f.l.WithField("index", h.RemoteIndex).
+			WithField("udpAddr", addr).
+			Debug("Recv error received")
 	}
 
 	hostinfo := f.hostMap.QueryReverseIndex(h.RemoteIndex)
 	if hostinfo == nil {
-		f.l.Debug("Did not find remote index in main hostmap", "remoteIndex", h.RemoteIndex)
+		f.l.WithField("remoteIndex", h.RemoteIndex).Debugln("Did not find remote index in main hostmap")
 		return
 	}
 
 	if hostinfo.remote.IsValid() && hostinfo.remote != addr {
-		f.l.Info("Someone spoofing recv_errors?",
-			"addr", addr,
-			"hostinfoRemote", hostinfo.remote,
-		)
+		f.l.Infoln("Someone spoofing recv_errors? ", addr, hostinfo.remote)
 		return
 	}
 

outside_test.go

@@ -155,7 +155,6 @@ func Test_newPacket_v6(t *testing.T) {
 	// next layer, missing length byte
 	err = newPacket(buffer.Bytes()[:49], true, p)
 	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
-	err = nil
 
 	// A good ICMP packet
 	ip = layers.IPv6{
@@ -166,26 +165,20 @@ func Test_newPacket_v6(t *testing.T) {
 		DstIP:      net.IPv6linklocalallnodes,
 	}
 
-	icmp := layers.ICMPv6{
-		TypeCode: layers.ICMPv6TypeEchoRequest,
-		Checksum: 0x1234,
-	}
+	icmp := layers.ICMPv6{}
 
 	buffer.Clear()
-	require.NoError(t, gopacket.SerializeLayers(buffer, opt, &ip, &icmp))
-	require.Error(t, newPacket(buffer.Bytes(), true, p))
-
-	buffer.Clear()
-	echo := layers.ICMPv6Echo{
-		Identifier: 0xabcd,
-		SeqNumber:  1234,
+	err = gopacket.SerializeLayers(buffer, opt, &ip, &icmp)
+	if err != nil {
+		panic(err)
 	}
-	require.NoError(t, gopacket.SerializeLayers(buffer, opt, &ip, &icmp, &echo))
-	require.NoError(t, newPacket(buffer.Bytes(), true, p))
+
+	err = newPacket(buffer.Bytes(), true, p)
+	require.NoError(t, err)
 	assert.Equal(t, uint8(layers.IPProtocolICMPv6), p.Protocol)
 	assert.Equal(t, netip.MustParseAddr("ff02::2"), p.RemoteAddr)
 	assert.Equal(t, netip.MustParseAddr("ff02::1"), p.LocalAddr)
-	assert.Equal(t, uint16(0xabcd), p.RemotePort)
+	assert.Equal(t, uint16(0), p.RemotePort)
 	assert.Equal(t, uint16(0), p.LocalPort)
 	assert.False(t, p.Fragment)
 
@@ -581,7 +574,7 @@ func BenchmarkParseV6(b *testing.B) {
 	}
 
 	evilBytes := buffer.Bytes()
-	for range 200 {
+	for i := 0; i < 200; i++ {
 		evilBytes = append(evilBytes, hopHeader...)
 	}
 	evilBytes = append(evilBytes, lastHopHeader...)

overlay/batch/batch.go

@@ -1,33 +0,0 @@
-package batch
-
-import "net/netip"
-
-type RxBatcher interface {
-	// Reserve creates a pkt to borrow
-	Reserve(sz int) []byte
-	// Commit borrows pkt. The caller must keep pkt valid until the next Flush
-	Commit(pkt []byte) error
-	// Flush emits every queued packet in arrival order. Returns the
-	// first error observed; keeps draining so one bad packet doesn't hold up
-	// the rest. After Flush returns, borrowed payload slices may be recycled.
-	Flush() error
-}
-
-type TxBatcher interface {
-	// Next returns a zero-length slice with slotCap capacity over the next unused
-	// slot's backing bytes. The caller writes into the returned slice and then
-	// calls Commit with the final length and destination. Next returns nil when
-	// the batch is full.
-	Next() []byte
-	// Commit records the slot just returned by Next as a packet of length n
-	// destined for dst.
-	Commit(n int, dst netip.AddrPort)
-	// Reset clears committed slots; backing storage is retained for reuse.
-	Reset()
-	// Len returns the number of committed packets.
-	Len() int
-	// Cap returns the maximum number of slots in the batch.
-	Cap() int
-	// Get returns the buffers needed to send the batch
-	Get() ([][]byte, []netip.AddrPort)
-}

overlay/batch/passthrough.go

@@ -1,57 +0,0 @@
-package batch
-
-import (
-	"io"
-
-	"github.com/slackhq/nebula/udp"
-)
-
-// Passthrough is a RxBatcher that doesn't batch anything, it just accumulates and then sends packets.
-type Passthrough struct {
-	out     io.Writer
-	slots   [][]byte
-	backing []byte
-	cursor  int
-}
-
-func NewPassthrough(w io.Writer) *Passthrough {
-	const baseNumSlots = 128
-	return &Passthrough{
-		out:     w,
-		slots:   make([][]byte, 0, baseNumSlots),
-		backing: make([]byte, 0, baseNumSlots*udp.MTU),
-	}
-}
-
-func (p *Passthrough) Reserve(sz int) []byte {
-	if len(p.backing)+sz > cap(p.backing) {
-		// Grow: allocate a fresh backing. Already-committed slices still
-		// reference the old array and remain valid until Flush drops them.
-		newCap := max(cap(p.backing)*2, sz)
-		p.backing = make([]byte, 0, newCap)
-	}
-	start := len(p.backing)
-	p.backing = p.backing[:start+sz]
-	return p.backing[start : start+sz : start+sz] //return zero length, sz-cap slice
-}
-
-func (p *Passthrough) Commit(pkt []byte) error {
-	p.slots = append(p.slots, pkt)
-	return nil
-}
-
-func (p *Passthrough) Flush() error {
-	var firstErr error
-	for _, s := range p.slots {
-		_, err := p.out.Write(s)
-		if err != nil && firstErr == nil {
-			firstErr = err
-		}
-	}
-	for i := range p.slots {
-		p.slots[i] = nil
-	}
-	p.slots = p.slots[:0]
-	p.backing = p.backing[:0]
-	return firstErr
-}

overlay/batch/tcp_coalesce.go

@@ -1,506 +0,0 @@
-package batch
-
-import (
-	"bytes"
-	"encoding/binary"
-	"io"
-
-	"github.com/slackhq/nebula/overlay/tio"
-)
-
-// ipProtoTCP is the IANA protocol number for TCP. Hardcoded instead of
-// reaching for golang.org/x/sys/unix — that package doesn't define the
-// constant on Windows, which would break cross-compiles even though this
-// file runs unchanged on every platform.
-const ipProtoTCP = 6
-
-// tcpCoalesceBufSize caps total bytes per superpacket. Mirrors the kernel's
-// sk_gso_max_size of ~64KiB; anything beyond this would be rejected anyway.
-const tcpCoalesceBufSize = 65535
-
-// tcpCoalesceMaxSegs caps how many segments we'll coalesce into a single
-// superpacket. Keeping this well below the kernel's TSO ceiling bounds
-// latency.
-const tcpCoalesceMaxSegs = 64
-
-// tcpCoalesceHdrCap is the scratch space we copy a seed's IP+TCP header
-// into. IPv6 (40) + TCP with full options (60) = 100 bytes.
-const tcpCoalesceHdrCap = 100
-
-// initialSlots is the starting capacity of the slot pool. One flow per
-// packet is the worst case so this matches a typical UDP recvmmsg batch.
-const initialSlots = 64
-
-// flowKey identifies a TCP flow by {src, dst, sport, dport, family}.
-// Comparable, so linear scans over the slot list stay tight.
-type flowKey struct {
-	src, dst     [16]byte
-	sport, dport uint16
-	isV6         bool
-}
-
-// coalesceSlot is one entry in the coalescer's ordered event queue. When
-// passthrough is true the slot holds a single borrowed packet that must be
-// emitted verbatim (non-TCP, non-admissible TCP, or oversize seed). When
-// passthrough is false the slot is an in-progress coalesced superpacket:
-// hdrBuf is a mutable copy of the seed's IP+TCP header (we patch total
-// length and pseudo-header partial at flush), and payIovs are *borrowed*
-// slices from the caller's plaintext buffers — no payload is ever copied.
-// The caller (listenOut) must keep those buffers alive until Flush.
-type coalesceSlot struct {
-	passthrough bool
-	rawPkt      []byte // borrowed when passthrough
-
-	fk       flowKey
-	hdrBuf   [tcpCoalesceHdrCap]byte
-	hdrLen   int
-	ipHdrLen int
-	isV6     bool
-	gsoSize  int
-	numSeg   int
-	totalPay int
-	nextSeq  uint32
-	// psh closes the chain: set when the last-accepted segment had PSH or
-	// was sub-gsoSize. No further appends after that.
-	psh     bool
-	payIovs [][]byte
-}
-
-// TCPCoalescer accumulates adjacent in-flow TCP data segments across
-// multiple concurrent flows and emits each flow's run as a single TSO
-// superpacket via tio.GSOWriter. All output — coalesced or not — is
-// deferred until Flush so arrival order is preserved on the wire. Owns
-// no locks; one coalescer per TUN write queue.
-type TCPCoalescer struct {
-	plainW io.Writer
-	gsoW   tio.GSOWriter // nil when the queue doesn't support TSO
-
-	// slots is the ordered event queue. Flush walks it once and emits each
-	// entry as either a WriteGSO (coalesced) or a plainW.Write (passthrough).
-	slots []*coalesceSlot
-	// openSlots maps a flow key to its most recent non-sealed slot, so new
-	// segments can extend an in-progress superpacket in O(1). Slots are
-	// removed from this map when they close (PSH or short-last-segment),
-	// when a non-admissible packet for that flow arrives, or in Flush.
-	openSlots map[flowKey]*coalesceSlot
-	pool      []*coalesceSlot // free list for reuse
-
-	backing []byte
-}
-
-func NewTCPCoalescer(w io.Writer) *TCPCoalescer {
-	c := &TCPCoalescer{
-		plainW:    w,
-		slots:     make([]*coalesceSlot, 0, initialSlots),
-		openSlots: make(map[flowKey]*coalesceSlot, initialSlots),
-		pool:      make([]*coalesceSlot, 0, initialSlots),
-		backing:   make([]byte, 0, initialSlots*65535),
-	}
-	if gw, ok := w.(tio.GSOWriter); ok && gw.GSOSupported() {
-		c.gsoW = gw
-	}
-	return c
-}
-
-// parsedTCP holds the fields extracted from a single parse so later steps
-// (admission, slot lookup, canAppend) don't re-walk the header.
-type parsedTCP struct {
-	fk        flowKey
-	ipHdrLen  int
-	tcpHdrLen int
-	hdrLen    int
-	payLen    int
-	seq       uint32
-	flags     byte
-}
-
-// parseTCPBase extracts the flow key and IP/TCP offsets for any TCP packet,
-// regardless of whether it's admissible for coalescing. Returns ok=false
-// for non-TCP or malformed input. Accepts IPv4 (no options, no fragmentation)
-// and IPv6 (no extension headers).
-func parseTCPBase(pkt []byte) (parsedTCP, bool) {
-	var p parsedTCP
-	if len(pkt) < 20 {
-		return p, false
-	}
-	v := pkt[0] >> 4
-	switch v {
-	case 4:
-		ihl := int(pkt[0]&0x0f) * 4
-		if ihl != 20 {
-			return p, false
-		}
-		if pkt[9] != ipProtoTCP {
-			return p, false
-		}
-		// Reject actual fragmentation (MF or non-zero frag offset).
-		if binary.BigEndian.Uint16(pkt[6:8])&0x3fff != 0 {
-			return p, false
-		}
-		totalLen := int(binary.BigEndian.Uint16(pkt[2:4]))
-		if totalLen > len(pkt) || totalLen < ihl {
-			return p, false
-		}
-		p.ipHdrLen = 20
-		p.fk.isV6 = false
-		copy(p.fk.src[:4], pkt[12:16])
-		copy(p.fk.dst[:4], pkt[16:20])
-		pkt = pkt[:totalLen]
-	case 6:
-		if len(pkt) < 40 {
-			return p, false
-		}
-		if pkt[6] != ipProtoTCP {
-			return p, false
-		}
-		payloadLen := int(binary.BigEndian.Uint16(pkt[4:6]))
-		if 40+payloadLen > len(pkt) {
-			return p, false
-		}
-		p.ipHdrLen = 40
-		p.fk.isV6 = true
-		copy(p.fk.src[:], pkt[8:24])
-		copy(p.fk.dst[:], pkt[24:40])
-		pkt = pkt[:40+payloadLen]
-	default:
-		return p, false
-	}
-
-	if len(pkt) < p.ipHdrLen+20 {
-		return p, false
-	}
-	tcpOff := int(pkt[p.ipHdrLen+12]>>4) * 4
-	if tcpOff < 20 || tcpOff > 60 {
-		return p, false
-	}
-	if len(pkt) < p.ipHdrLen+tcpOff {
-		return p, false
-	}
-	p.tcpHdrLen = tcpOff
-	p.hdrLen = p.ipHdrLen + tcpOff
-	p.payLen = len(pkt) - p.hdrLen
-	p.seq = binary.BigEndian.Uint32(pkt[p.ipHdrLen+4 : p.ipHdrLen+8])
-	p.flags = pkt[p.ipHdrLen+13]
-	p.fk.sport = binary.BigEndian.Uint16(pkt[p.ipHdrLen : p.ipHdrLen+2])
-	p.fk.dport = binary.BigEndian.Uint16(pkt[p.ipHdrLen+2 : p.ipHdrLen+4])
-	return p, true
-}
-
-// coalesceable reports whether a parsed TCP segment is eligible for
-// coalescing. Accepts only ACK or ACK|PSH with a non-empty payload.
-func (p parsedTCP) coalesceable() bool {
-	const ack = 0x10
-	const psh = 0x08
-	if p.flags&^(ack|psh) != 0 || p.flags&ack == 0 {
-		return false
-	}
-	return p.payLen > 0
-}
-
-func (c *TCPCoalescer) Reserve(sz int) []byte {
-	if len(c.backing)+sz > cap(c.backing) {
-		// Grow: allocate a fresh backing. Already-committed slices still
-		// reference the old array and remain valid until Flush drops them.
-		newCap := max(cap(c.backing)*2, sz)
-		c.backing = make([]byte, 0, newCap)
-	}
-	start := len(c.backing)
-	c.backing = c.backing[:start+sz]
-	return c.backing[start : start+sz : start+sz] //return zero length, sz-cap slice
-}
-
-// Commit borrows pkt. The caller must keep pkt valid until the next Flush,
-// whether or not the packet was coalesced — passthrough (non-admissible)
-// packets are queued and written at Flush time, not synchronously.
-func (c *TCPCoalescer) Commit(pkt []byte) error {
-	if c.gsoW == nil {
-		c.addPassthrough(pkt)
-		return nil
-	}
-
-	info, ok := parseTCPBase(pkt)
-	if !ok {
-		// Non-TCP or malformed — can't possibly collide with an open flow.
-		c.addPassthrough(pkt)
-		return nil
-	}
-	if !info.coalesceable() {
-		// TCP but not admissible (SYN/FIN/RST/URG/CWR/ECE or zero-payload).
-		// Seal this flow's open slot so later in-flow packets don't extend
-		// it and accidentally reorder past this passthrough.
-		delete(c.openSlots, info.fk)
-		c.addPassthrough(pkt)
-		return nil
-	}
-
-	if open := c.openSlots[info.fk]; open != nil {
-		if c.canAppend(open, pkt, info) {
-			c.appendPayload(open, pkt, info)
-			if open.psh {
-				delete(c.openSlots, info.fk)
-			}
-			return nil
-		}
-		// Can't extend — seal it and fall through to seed a fresh slot.
-		delete(c.openSlots, info.fk)
-	}
-	c.seed(pkt, info)
-	return nil
-}
-
-// Flush emits every queued event in arrival order. Coalesced slots go out
-// via WriteGSO; passthrough slots go out via plainW.Write. Returns the
-// first error observed; keeps draining so one bad packet doesn't hold up
-// the rest. After Flush returns, borrowed payload slices may be recycled.
-func (c *TCPCoalescer) Flush() error {
-	var first error
-	for _, s := range c.slots {
-		var err error
-		if s.passthrough {
-			_, err = c.plainW.Write(s.rawPkt)
-		} else {
-			err = c.flushSlot(s)
-		}
-		if err != nil && first == nil {
-			first = err
-		}
-		c.release(s)
-	}
-	for i := range c.slots {
-		c.slots[i] = nil
-	}
-	c.slots = c.slots[:0]
-	for k := range c.openSlots {
-		delete(c.openSlots, k)
-	}
-
-	c.backing = c.backing[:0]
-	return first
-}
-
-func (c *TCPCoalescer) addPassthrough(pkt []byte) {
-	s := c.take()
-	s.passthrough = true
-	s.rawPkt = pkt
-	c.slots = append(c.slots, s)
-}
-
-func (c *TCPCoalescer) seed(pkt []byte, info parsedTCP) {
-	if info.hdrLen > tcpCoalesceHdrCap || info.hdrLen+info.payLen > tcpCoalesceBufSize {
-		// Pathological shape — can't fit our scratch, emit as-is.
-		c.addPassthrough(pkt)
-		return
-	}
-	s := c.take()
-	s.passthrough = false
-	s.rawPkt = nil
-	copy(s.hdrBuf[:], pkt[:info.hdrLen])
-	s.hdrLen = info.hdrLen
-	s.ipHdrLen = info.ipHdrLen
-	s.isV6 = info.fk.isV6
-	s.fk = info.fk
-	s.gsoSize = info.payLen
-	s.numSeg = 1
-	s.totalPay = info.payLen
-	s.nextSeq = info.seq + uint32(info.payLen)
-	s.psh = info.flags&0x08 != 0
-	s.payIovs = append(s.payIovs[:0], pkt[info.hdrLen:info.hdrLen+info.payLen])
-	c.slots = append(c.slots, s)
-	if !s.psh {
-		c.openSlots[info.fk] = s
-	}
-}
-
-// canAppend reports whether info's packet extends the slot's seed: same
-// header shape and stable contents, adjacent seq, not oversized, chain not
-// closed.
-func (c *TCPCoalescer) canAppend(s *coalesceSlot, pkt []byte, info parsedTCP) bool {
-	if s.psh {
-		return false
-	}
-	if info.hdrLen != s.hdrLen {
-		return false
-	}
-	if info.seq != s.nextSeq {
-		return false
-	}
-	if s.numSeg >= tcpCoalesceMaxSegs {
-		return false
-	}
-	if info.payLen > s.gsoSize {
-		return false
-	}
-	if s.hdrLen+s.totalPay+info.payLen > tcpCoalesceBufSize {
-		return false
-	}
-	if !headersMatch(s.hdrBuf[:s.hdrLen], pkt[:info.hdrLen], s.isV6, s.ipHdrLen) {
-		return false
-	}
-	return true
-}
-
-func (c *TCPCoalescer) appendPayload(s *coalesceSlot, pkt []byte, info parsedTCP) {
-	s.payIovs = append(s.payIovs, pkt[info.hdrLen:info.hdrLen+info.payLen])
-	s.numSeg++
-	s.totalPay += info.payLen
-	s.nextSeq = info.seq + uint32(info.payLen)
-	if info.flags&0x08 != 0 {
-		// Propagate PSH into the seed header so kernel TSO sets it on the
-		// last segment. Without this the sender's push signal is dropped.
-		s.hdrBuf[s.ipHdrLen+13] |= 0x08
-	}
-	if info.payLen < s.gsoSize || info.flags&0x08 != 0 {
-		s.psh = true
-	}
-}
-
-func (c *TCPCoalescer) take() *coalesceSlot {
-	if n := len(c.pool); n > 0 {
-		s := c.pool[n-1]
-		c.pool[n-1] = nil
-		c.pool = c.pool[:n-1]
-		return s
-	}
-	return &coalesceSlot{}
-}
-
-func (c *TCPCoalescer) release(s *coalesceSlot) {
-	s.passthrough = false
-	s.rawPkt = nil
-	for i := range s.payIovs {
-		s.payIovs[i] = nil
-	}
-	s.payIovs = s.payIovs[:0]
-	s.numSeg = 0
-	s.totalPay = 0
-	s.psh = false
-	c.pool = append(c.pool, s)
-}
-
-// flushSlot patches the header and calls WriteGSO. Does not remove the
-// slot from c.slots.
-func (c *TCPCoalescer) flushSlot(s *coalesceSlot) error {
-	total := s.hdrLen + s.totalPay
-	l4Len := total - s.ipHdrLen
-	hdr := s.hdrBuf[:s.hdrLen]
-
-	if s.isV6 {
-		binary.BigEndian.PutUint16(hdr[4:6], uint16(l4Len))
-	} else {
-		binary.BigEndian.PutUint16(hdr[2:4], uint16(total))
-		hdr[10] = 0
-		hdr[11] = 0
-		binary.BigEndian.PutUint16(hdr[10:12], ipv4HdrChecksum(hdr[:s.ipHdrLen]))
-	}
-
-	var psum uint32
-	if s.isV6 {
-		psum = pseudoSumIPv6(hdr[8:24], hdr[24:40], ipProtoTCP, l4Len)
-	} else {
-		psum = pseudoSumIPv4(hdr[12:16], hdr[16:20], ipProtoTCP, l4Len)
-	}
-	tcsum := s.ipHdrLen + 16
-	binary.BigEndian.PutUint16(hdr[tcsum:tcsum+2], foldOnceNoInvert(psum))
-
-	return c.gsoW.WriteGSO(hdr[:s.ipHdrLen], hdr[s.ipHdrLen:], s.payIovs)
-}
-
-// headersMatch compares two IP+TCP header prefixes for byte-for-byte
-// equality on every field that must be identical across coalesced
-// segments. Size/IPID/IPCsum/seq/flags/tcpCsum are masked out.
-func headersMatch(a, b []byte, isV6 bool, ipHdrLen int) bool {
-	if len(a) != len(b) {
-		return false
-	}
-	if isV6 {
-		// IPv6: bytes [0:4] = version/TC/flow-label, [6:8] = next_hdr/hop,
-		// [8:40] = src+dst. Skip [4:6] payload length.
-		if !bytes.Equal(a[0:4], b[0:4]) {
-			return false
-		}
-		if !bytes.Equal(a[6:40], b[6:40]) {
-			return false
-		}
-	} else {
-		// IPv4: [0:2] version/IHL/TOS, [6:10] flags/fragoff/TTL/proto,
-		// [12:20] src+dst. Skip [2:4] total len, [4:6] id, [10:12] csum.
-		if !bytes.Equal(a[0:2], b[0:2]) {
-			return false
-		}
-		if !bytes.Equal(a[6:10], b[6:10]) {
-			return false
-		}
-		if !bytes.Equal(a[12:20], b[12:20]) {
-			return false
-		}
-	}
-	// TCP: compare [0:4] ports, [8:13] ack+dataoff, [14:16] window,
-	// [18:tcpHdrLen] options (incl. urgent).
-	tcp := ipHdrLen
-	if !bytes.Equal(a[tcp:tcp+4], b[tcp:tcp+4]) {
-		return false
-	}
-	if !bytes.Equal(a[tcp+8:tcp+13], b[tcp+8:tcp+13]) {
-		return false
-	}
-	if !bytes.Equal(a[tcp+14:tcp+16], b[tcp+14:tcp+16]) {
-		return false
-	}
-	if !bytes.Equal(a[tcp+18:], b[tcp+18:]) {
-		return false
-	}
-	return true
-}
-
-// ipv4HdrChecksum computes the IPv4 header checksum over hdr (which must
-// already have its checksum field zeroed) and returns the folded/inverted
-// 16-bit value to store.
-func ipv4HdrChecksum(hdr []byte) uint16 {
-	var sum uint32
-	for i := 0; i+1 < len(hdr); i += 2 {
-		sum += uint32(binary.BigEndian.Uint16(hdr[i : i+2]))
-	}
-	if len(hdr)%2 == 1 {
-		sum += uint32(hdr[len(hdr)-1]) << 8
-	}
-	for sum>>16 != 0 {
-		sum = (sum & 0xffff) + (sum >> 16)
-	}
-	return ^uint16(sum)
-}
-
-// pseudoSumIPv4 / pseudoSumIPv6 build the TCP pseudo-header partial sum
-// expected by the virtio NEEDS_CSUM kernel path: the 32-bit accumulator
-// before folding.
-func pseudoSumIPv4(src, dst []byte, proto byte, l4Len int) uint32 {
-	var sum uint32
-	sum += uint32(binary.BigEndian.Uint16(src[0:2]))
-	sum += uint32(binary.BigEndian.Uint16(src[2:4]))
-	sum += uint32(binary.BigEndian.Uint16(dst[0:2]))
-	sum += uint32(binary.BigEndian.Uint16(dst[2:4]))
-	sum += uint32(proto)
-	sum += uint32(l4Len)
-	return sum
-}
-
-func pseudoSumIPv6(src, dst []byte, proto byte, l4Len int) uint32 {
-	var sum uint32
-	for i := 0; i < 16; i += 2 {
-		sum += uint32(binary.BigEndian.Uint16(src[i : i+2]))
-		sum += uint32(binary.BigEndian.Uint16(dst[i : i+2]))
-	}
-	sum += uint32(l4Len >> 16)
-	sum += uint32(l4Len & 0xffff)
-	sum += uint32(proto)
-	return sum
-}
-
-// foldOnceNoInvert folds the 32-bit accumulator to 16 bits and returns it
-// unchanged (no one's complement). This is what virtio NEEDS_CSUM wants in
-// the L4 checksum field — the kernel will add the payload sum and invert.
-func foldOnceNoInvert(sum uint32) uint16 {
-	for sum>>16 != 0 {
-		sum = (sum & 0xffff) + (sum >> 16)
-	}
-	return uint16(sum)
-}

overlay/batch/tcp_coalesce_test.go

@@ -1,618 +0,0 @@
-package batch
-
-import (
-	"encoding/binary"
-	"testing"
-)
-
-// fakeTunWriter records plain Writes and WriteGSO calls without touching a
-// real TUN fd. WriteGSO records the IP header, transport header, and
-// borrowed payload fragments separately so tests can inspect each.
-type fakeTunWriter struct {
-	gsoEnabled bool
-	writes     [][]byte
-	gsoWrites  []fakeGSOWrite
-}
-
-// fakeGSOWrite captures one WriteGSO call. hdr is the concatenation of the
-// IP and transport headers (in that order), gsoSize / isV6 / csumStart are
-// derived from the call so existing assertions keep working unchanged.
-type fakeGSOWrite struct {
-	hdr       []byte
-	pays      [][]byte
-	gsoSize   uint16
-	isV6      bool
-	csumStart uint16
-}
-
-// total returns hdrLen + sum of pay lens.
-func (g fakeGSOWrite) total() int {
-	n := len(g.hdr)
-	for _, p := range g.pays {
-		n += len(p)
-	}
-	return n
-}
-
-// payLen sums the pays.
-func (g fakeGSOWrite) payLen() int {
-	var n int
-	for _, p := range g.pays {
-		n += len(p)
-	}
-	return n
-}
-
-func (w *fakeTunWriter) Write(p []byte) (int, error) {
-	buf := make([]byte, len(p))
-	copy(buf, p)
-	w.writes = append(w.writes, buf)
-	return len(p), nil
-}
-
-func (w *fakeTunWriter) WriteGSO(hdr []byte, transportHdr []byte, pays [][]byte) error {
-	hcopy := make([]byte, len(hdr)+len(transportHdr))
-	copy(hcopy, hdr)
-	copy(hcopy[len(hdr):], transportHdr)
-	paysCopy := make([][]byte, len(pays))
-	for i, p := range pays {
-		pc := make([]byte, len(p))
-		copy(pc, p)
-		paysCopy[i] = pc
-	}
-	var gsoSize uint16
-	if len(pays) > 1 {
-		gsoSize = uint16(len(pays[0]))
-	}
-	isV6 := len(hdr) > 0 && hdr[0]>>4 == 6
-	w.gsoWrites = append(w.gsoWrites, fakeGSOWrite{
-		hdr:       hcopy,
-		pays:      paysCopy,
-		gsoSize:   gsoSize,
-		isV6:      isV6,
-		csumStart: uint16(len(hdr)),
-	})
-	return nil
-}
-
-func (w *fakeTunWriter) GSOSupported() bool { return w.gsoEnabled }
-
-// buildTCPv4 constructs a minimal IPv4+TCP packet with the given payload,
-// seq, and flags. Assumes no IP options and a 20-byte TCP header.
-func buildTCPv4(seq uint32, flags byte, payload []byte) []byte {
-	return buildTCPv4Ports(1000, 2000, seq, flags, payload)
-}
-
-// buildTCPv4Ports is buildTCPv4 with caller-specified ports so tests can
-// build distinct flows.
-func buildTCPv4Ports(sport, dport uint16, seq uint32, flags byte, payload []byte) []byte {
-	const ipHdrLen = 20
-	const tcpHdrLen = 20
-	total := ipHdrLen + tcpHdrLen + len(payload)
-	pkt := make([]byte, total)
-
-	pkt[0] = 0x45
-	pkt[1] = 0x00
-	binary.BigEndian.PutUint16(pkt[2:4], uint16(total))
-	binary.BigEndian.PutUint16(pkt[4:6], 0)
-	binary.BigEndian.PutUint16(pkt[6:8], 0x4000)
-	pkt[8] = 64
-	pkt[9] = ipProtoTCP
-	copy(pkt[12:16], []byte{10, 0, 0, 1})
-	copy(pkt[16:20], []byte{10, 0, 0, 2})
-
-	binary.BigEndian.PutUint16(pkt[20:22], sport)
-	binary.BigEndian.PutUint16(pkt[22:24], dport)
-	binary.BigEndian.PutUint32(pkt[24:28], seq)
-	binary.BigEndian.PutUint32(pkt[28:32], 12345)
-	pkt[32] = 0x50
-	pkt[33] = flags
-	binary.BigEndian.PutUint16(pkt[34:36], 0xffff)
-
-	copy(pkt[40:], payload)
-	return pkt
-}
-
-const (
-	tcpAck    = 0x10
-	tcpPsh    = 0x08
-	tcpSyn    = 0x02
-	tcpFin    = 0x01
-	tcpAckPsh = tcpAck | tcpPsh
-)
-
-func TestCoalescerPassthroughWhenGSOUnavailable(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: false}
-	c := NewTCPCoalescer(w)
-	pkt := buildTCPv4(1000, tcpAck, []byte("hello"))
-	if err := c.Commit(pkt); err != nil {
-		t.Fatal(err)
-	}
-	// No sync write — passthrough is deferred to Flush.
-	if len(w.writes) != 0 || len(w.gsoWrites) != 0 {
-		t.Fatalf("no Add-time writes: got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.writes) != 1 || len(w.gsoWrites) != 0 {
-		t.Fatalf("want single plain write, got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerNonTCPPassthrough(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pkt := make([]byte, 28)
-	pkt[0] = 0x45
-	binary.BigEndian.PutUint16(pkt[2:4], 28)
-	pkt[9] = 1
-	copy(pkt[12:16], []byte{10, 0, 0, 1})
-	copy(pkt[16:20], []byte{10, 0, 0, 2})
-	if err := c.Commit(pkt); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.writes) != 1 || len(w.gsoWrites) != 0 {
-		t.Fatalf("ICMP should pass through unchanged")
-	}
-}
-
-func TestCoalescerSeedThenFlushAlone(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pkt := buildTCPv4(1000, tcpAck, make([]byte, 1000))
-	if err := c.Commit(pkt); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.writes) != 0 || len(w.gsoWrites) != 0 {
-		t.Fatalf("unexpected output before flush")
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// Single-segment flush now goes through WriteGSO with GSO_NONE
-	// (virtio NEEDS_CSUM lets the kernel fill in the L4 csum).
-	if len(w.gsoWrites) != 1 || len(w.writes) != 0 {
-		t.Fatalf("single-seg flush: writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-	g := w.gsoWrites[0]
-	if g.total() != 40+1000 {
-		t.Errorf("super total=%d want %d", g.total(), 40+1000)
-	}
-	if g.payLen() != 1000 {
-		t.Errorf("payLen=%d want 1000", g.payLen())
-	}
-}
-
-func TestCoalescerCoalescesAdjacentACKs(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	if err := c.Commit(buildTCPv4(1000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(2200, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(3400, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.gsoWrites) != 1 {
-		t.Fatalf("want 1 gso write, got %d (plain=%d)", len(w.gsoWrites), len(w.writes))
-	}
-	g := w.gsoWrites[0]
-	if g.gsoSize != 1200 {
-		t.Errorf("gsoSize=%d want 1200", g.gsoSize)
-	}
-	if len(g.hdr) != 40 {
-		t.Errorf("hdrLen=%d want 40", len(g.hdr))
-	}
-	if g.csumStart != 20 {
-		t.Errorf("csumStart=%d want 20", g.csumStart)
-	}
-	if len(g.pays) != 3 {
-		t.Errorf("pay count=%d want 3", len(g.pays))
-	}
-	if g.total() != 40+3*1200 {
-		t.Errorf("superpacket len=%d want %d", g.total(), 40+3*1200)
-	}
-	if tot := binary.BigEndian.Uint16(g.hdr[2:4]); int(tot) != g.total() {
-		t.Errorf("ip total_length=%d want %d", tot, g.total())
-	}
-}
-
-func TestCoalescerRejectsSeqGap(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	if err := c.Commit(buildTCPv4(1000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(3000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// Each packet flushes as its own single-segment WriteGSO now.
-	if len(w.gsoWrites) != 2 || len(w.writes) != 0 {
-		t.Fatalf("seq gap: want 2 gso writes got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerRejectsFlagMismatch(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	if err := c.Commit(buildTCPv4(1000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	// SYN|ACK is non-admissible. Must flush matching flow's slot (gso)
-	// and then plain-write the SYN packet itself.
-	syn := buildTCPv4(2200, tcpSyn|tcpAck, pay)
-	if err := c.Commit(syn); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.writes) != 1 || len(w.gsoWrites) != 1 {
-		t.Fatalf("flag mismatch: want 1 plain + 1 gso, got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerRejectsFIN(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	fin := buildTCPv4(1000, tcpAck|tcpFin, []byte("x"))
-	if err := c.Commit(fin); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// FIN isn't admissible — passthrough as plain, no slot, no gso.
-	if len(w.writes) != 1 || len(w.gsoWrites) != 0 {
-		t.Fatalf("FIN should be passthrough, got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerShortLastSegmentClosesChain(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	full := make([]byte, 1200)
-	half := make([]byte, 500)
-	if err := c.Commit(buildTCPv4(1000, tcpAck, full)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(2200, tcpAck, half)); err != nil {
-		t.Fatal(err)
-	}
-	// Chain now closed; next packet seeds a new slot on the same flow
-	// after flushing the old one.
-	if err := c.Commit(buildTCPv4(2700, tcpAck, full)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// Expect two gso writes: the first two packets coalesced, then the
-	// third flushed alone (single-seg via GSO_NONE).
-	if len(w.gsoWrites) != 2 {
-		t.Fatalf("want 2 gso writes got %d", len(w.gsoWrites))
-	}
-	if len(w.writes) != 0 {
-		t.Fatalf("want 0 plain writes got %d", len(w.writes))
-	}
-	if w.gsoWrites[0].gsoSize != 1200 {
-		t.Errorf("gsoSize=%d want 1200", w.gsoWrites[0].gsoSize)
-	}
-	if got, want := w.gsoWrites[0].total(), 40+1200+500; got != want {
-		t.Errorf("super len=%d want %d", got, want)
-	}
-}
-
-func TestCoalescerPSHFinalizesChain(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	if err := c.Commit(buildTCPv4(1000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(2200, tcpAckPsh, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(3400, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// First two coalesce; the third seeds a fresh slot that flushes alone.
-	if len(w.gsoWrites) != 2 {
-		t.Fatalf("want 2 gso writes got %d", len(w.gsoWrites))
-	}
-	if len(w.writes) != 0 {
-		t.Fatalf("want 0 plain writes got %d", len(w.writes))
-	}
-}
-
-// TestCoalescerPropagatesPSHFromAppended ensures that when an appended
-// segment carries PSH (or is short, sealing the chain), the PSH bit ends
-// up in the emitted superpacket's TCP flags. The kernel TSO path keeps
-// PSH only on the last segment iff the input header has it set; if the
-// coalescer drops it the sender's push signal never reaches the receiver.
-func TestCoalescerPropagatesPSHFromAppended(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	// Seed has no PSH; second segment carries PSH and seals the chain.
-	if err := c.Commit(buildTCPv4(1000, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4(2200, tcpAckPsh, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(0); err != nil {
-		t.Fatal(err)
-	}
-	if len(w.gsoWrites) != 1 {
-		t.Fatalf("want 1 gso write got %d", len(w.gsoWrites))
-	}
-	g := w.gsoWrites[0]
-	const ipHdrLen = 20
-	flags := g.hdr[ipHdrLen+13]
-	if flags&tcpPsh == 0 {
-		t.Fatalf("PSH lost from coalesced superpacket: flags=0x%02x", flags)
-	}
-	if flags&tcpAck == 0 {
-		t.Fatalf("ACK missing from coalesced superpacket: flags=0x%02x", flags)
-	}
-}
-
-func TestCoalescerRejectsDifferentFlow(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-	p1 := buildTCPv4(1000, tcpAck, pay)
-	p2 := buildTCPv4(2200, tcpAck, pay)
-	binary.BigEndian.PutUint16(p2[20:22], 9999)
-	if err := c.Commit(p1); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(p2); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// Two independent flows, each flushes its own single-segment WriteGSO.
-	if len(w.gsoWrites) != 2 || len(w.writes) != 0 {
-		t.Fatalf("diff flow: want 2 gso writes got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerRejectsIPOptions(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 500)
-	pkt := buildTCPv4(1000, tcpAck, pay)
-	// Bump IHL to 6 to simulate 4 bytes of IP options. Don't actually add
-	// bytes — parser should bail before it matters.
-	pkt[0] = 0x46
-	if err := c.Commit(pkt); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	// Non-admissible parse → passthrough as plain.
-	if len(w.writes) != 1 || len(w.gsoWrites) != 0 {
-		t.Fatalf("IP options should passthrough, got writes=%d gso=%d", len(w.writes), len(w.gsoWrites))
-	}
-}
-
-func TestCoalescerCapBySegments(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 512)
-	seq := uint32(1000)
-	for i := 0; i < tcpCoalesceMaxSegs+5; i++ {
-		if err := c.Commit(buildTCPv4(seq, tcpAck, pay)); err != nil {
-			t.Fatal(err)
-		}
-		seq += uint32(len(pay))
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	for _, g := range w.gsoWrites {
-		segs := len(g.pays)
-		if segs > tcpCoalesceMaxSegs {
-			t.Fatalf("super exceeded seg cap: %d > %d", segs, tcpCoalesceMaxSegs)
-		}
-	}
-}
-
-// TestCoalescerMultipleFlowsInSameBatch proves two interleaved bulk TCP
-// flows coalesce independently in a single Flush.
-func TestCoalescerMultipleFlowsInSameBatch(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-
-	// Flow A: sport 1000. Flow B: sport 3000.
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 100, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 500, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 1300, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 1700, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 2500, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 2900, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-
-	if len(w.gsoWrites) != 2 {
-		t.Fatalf("want 2 gso writes (one per flow), got %d", len(w.gsoWrites))
-	}
-	if len(w.writes) != 0 {
-		t.Fatalf("want no plain writes, got %d", len(w.writes))
-	}
-	// Each superpacket should carry 3 segments.
-	for i, g := range w.gsoWrites {
-		if len(g.pays) != 3 {
-			t.Errorf("gso[%d]: segs=%d want 3", i, len(g.pays))
-		}
-		if g.gsoSize != 1200 {
-			t.Errorf("gso[%d]: gsoSize=%d want 1200", i, g.gsoSize)
-		}
-	}
-	// Verify each superpacket carries the source port it was seeded with.
-	seenSports := map[uint16]bool{}
-	for _, g := range w.gsoWrites {
-		sp := binary.BigEndian.Uint16(g.hdr[20:22])
-		seenSports[sp] = true
-	}
-	if !seenSports[1000] || !seenSports[3000] {
-		t.Errorf("expected superpackets for sports 1000 and 3000, got %v", seenSports)
-	}
-}
-
-// TestCoalescerPreservesArrivalOrder confirms that with passthrough and
-// coalesced events both queued, Flush emits them in Add order rather than
-// writing passthrough packets synchronously.
-func TestCoalescerPreservesArrivalOrder(t *testing.T) {
-	w := &orderedFakeWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	// Sequence: coalesceable TCP, ICMP (passthrough), coalesceable TCP on
-	// a different flow. Expected emit order: gso(X), plain(ICMP), gso(Y).
-	pay := make([]byte, 1200)
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 100, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	icmp := make([]byte, 28)
-	icmp[0] = 0x45
-	binary.BigEndian.PutUint16(icmp[2:4], 28)
-	icmp[9] = 1
-	copy(icmp[12:16], []byte{10, 0, 0, 1})
-	copy(icmp[16:20], []byte{10, 0, 0, 3})
-	if err := c.Commit(icmp); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 500, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	// Nothing should have hit the writer synchronously.
-	if len(w.events) != 0 {
-		t.Fatalf("Add emitted events synchronously: %v", w.events)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-	if got, want := w.events, []string{"gso", "plain", "gso"}; !stringSliceEq(got, want) {
-		t.Fatalf("flush order=%v want %v", got, want)
-	}
-}
-
-// orderedFakeWriter records only the sequence of call types so tests can
-// assert arrival order without inspecting bytes.
-type orderedFakeWriter struct {
-	gsoEnabled bool
-	events     []string
-}
-
-func (w *orderedFakeWriter) Write(p []byte) (int, error) {
-	w.events = append(w.events, "plain")
-	return len(p), nil
-}
-
-func (w *orderedFakeWriter) WriteGSO(hdr []byte, transportHdr []byte, pays [][]byte) error {
-	w.events = append(w.events, "gso")
-	return nil
-}
-
-func (w *orderedFakeWriter) GSOSupported() bool { return w.gsoEnabled }
-
-func stringSliceEq(a, b []string) bool {
-	if len(a) != len(b) {
-		return false
-	}
-	for i := range a {
-		if a[i] != b[i] {
-			return false
-		}
-	}
-	return true
-}
-
-// TestCoalescerInterleavedFlowsPreserveOrdering checks that a non-admissible
-// packet (SYN) mid-flow only flushes its own flow, not others.
-func TestCoalescerInterleavedFlowsPreserveOrdering(t *testing.T) {
-	w := &fakeTunWriter{gsoEnabled: true}
-	c := NewTCPCoalescer(w)
-	pay := make([]byte, 1200)
-
-	// Flow A two segments.
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 100, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(1000, 2000, 1300, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	// Flow B two segments.
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 500, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 1700, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	// Flow A SYN (non-admissible) — must flush only flow A's slot.
-	syn := buildTCPv4Ports(1000, 2000, 9999, tcpSyn|tcpAck, pay)
-	if err := c.Commit(syn); err != nil {
-		t.Fatal(err)
-	}
-	// Flow B continues — should still be coalesced with its seed.
-	if err := c.Commit(buildTCPv4Ports(3000, 2000, 2900, tcpAck, pay)); err != nil {
-		t.Fatal(err)
-	}
-	if err := c.Flush(); err != nil {
-		t.Fatal(err)
-	}
-
-	// Expected:
-	//   - 1 gso for flow A (first 2 segments)
-	//   - 1 plain for flow A SYN
-	//   - 1 gso for flow B (3 segments)
-	if len(w.gsoWrites) != 2 {
-		t.Fatalf("want 2 gso writes, got %d", len(w.gsoWrites))
-	}
-	if len(w.writes) != 1 {
-		t.Fatalf("want 1 plain write (SYN), got %d", len(w.writes))
-	}
-	// Find the 3-segment gso (flow B) and the 2-segment gso (flow A).
-	var segCounts []int
-	for _, g := range w.gsoWrites {
-		segCounts = append(segCounts, len(g.pays))
-	}
-	if !(segCounts[0] == 2 && segCounts[1] == 3) && !(segCounts[0] == 3 && segCounts[1] == 2) {
-		t.Errorf("unexpected segment counts: %v (want 2 and 3)", segCounts)
-	}
-}

overlay/batch/tx_batch.go

@@ -1,61 +0,0 @@
-package batch
-
-import "net/netip"
-
-const SendBatchCap = 128
-
-// SendBatch accumulates encrypted UDP packets for potential TX offloading.
-// One SendBatch is owned by each listenIn goroutine; no locking is needed.
-// The backing storage holds up to batchCap packets of slotCap bytes each;
-// bufs and dsts are parallel slices of committed slots.
-type SendBatch struct {
-	bufs     [][]byte
-	dsts     []netip.AddrPort
-	backing  []byte
-	slotCap  int
-	batchCap int
-	nextSlot int
-}
-
-func NewSendBatch(batchCap, slotCap int) *SendBatch {
-	return &SendBatch{
-		bufs:     make([][]byte, 0, batchCap),
-		dsts:     make([]netip.AddrPort, 0, batchCap),
-		backing:  make([]byte, batchCap*slotCap),
-		slotCap:  slotCap,
-		batchCap: batchCap,
-	}
-}
-
-func (b *SendBatch) Next() []byte {
-	if b.nextSlot >= b.batchCap {
-		return nil
-	}
-	start := b.nextSlot * b.slotCap
-	return b.backing[start : start : start+b.slotCap] //set len to 0 but cap to slotCap
-}
-
-func (b *SendBatch) Commit(n int, dst netip.AddrPort) {
-	start := b.nextSlot * b.slotCap
-	b.bufs = append(b.bufs, b.backing[start:start+n])
-	b.dsts = append(b.dsts, dst)
-	b.nextSlot++
-}
-
-func (b *SendBatch) Reset() {
-	b.bufs = b.bufs[:0]
-	b.dsts = b.dsts[:0]
-	b.nextSlot = 0
-}
-
-func (b *SendBatch) Len() int {
-	return len(b.bufs)
-}
-
-func (b *SendBatch) Cap() int {
-	return b.batchCap
-}
-
-func (b *SendBatch) Get() ([][]byte, []netip.AddrPort) {
-	return b.bufs, b.dsts
-}

overlay/batch/tx_batch_test.go

@@ -1,69 +0,0 @@
-package batch
-
-import (
-	"net/netip"
-	"testing"
-)
-
-func TestSendBatchBookkeeping(t *testing.T) {
-	b := NewSendBatch(4, 32)
-	if b.Len() != 0 || b.Cap() != 4 {
-		t.Fatalf("fresh batch: len=%d cap=%d", b.Len(), b.Cap())
-	}
-
-	ap := netip.MustParseAddrPort("10.0.0.1:4242")
-	for i := 0; i < 4; i++ {
-		slot := b.Next()
-		if slot == nil {
-			t.Fatalf("slot %d: Next returned nil before cap", i)
-		}
-		if cap(slot) != 32 || len(slot) != 0 {
-			t.Fatalf("slot %d: got len=%d cap=%d want len=0 cap=32", i, len(slot), cap(slot))
-		}
-		// Write a marker byte.
-		slot = append(slot, byte(i), byte(i+1), byte(i+2))
-		b.Commit(len(slot), ap)
-	}
-	if b.Next() != nil {
-		t.Fatalf("Next should return nil when full")
-	}
-	if b.Len() != 4 {
-		t.Fatalf("Len=%d want 4", b.Len())
-	}
-	for i, buf := range b.bufs {
-		if len(buf) != 3 || buf[0] != byte(i) {
-			t.Errorf("buf %d: %x", i, buf)
-		}
-		if b.dsts[i] != ap {
-			t.Errorf("dst %d: got %v want %v", i, b.dsts[i], ap)
-		}
-	}
-
-	// Reset returns empty and Next works again.
-	b.Reset()
-	if b.Len() != 0 {
-		t.Fatalf("after Reset Len=%d want 0", b.Len())
-	}
-	slot := b.Next()
-	if slot == nil || cap(slot) != 32 {
-		t.Fatalf("after Reset Next nil or wrong cap: %v cap=%d", slot == nil, cap(slot))
-	}
-}
-
-func TestSendBatchSlotsDoNotOverlap(t *testing.T) {
-	b := NewSendBatch(3, 8)
-	ap := netip.MustParseAddrPort("10.0.0.1:80")
-
-	// Fill three slots, each with its own sentinel byte.
-	for i := 0; i < 3; i++ {
-		s := b.Next()
-		s = append(s, byte(0xA0+i), byte(0xB0+i))
-		b.Commit(len(s), ap)
-	}
-
-	for i, buf := range b.bufs {
-		if buf[0] != byte(0xA0+i) || buf[1] != byte(0xB0+i) {
-			t.Errorf("slot %d corrupted: %x", i, buf)
-		}
-	}
-}

overlay/device.go

@@ -1,24 +1,17 @@
 package overlay
 
 import (
-	"io"
 	"net/netip"
 
-	"github.com/slackhq/nebula/overlay/tio"
 	"github.com/slackhq/nebula/routing"
 )
 
-// defaultBatchBufSize is the per-Queue scratch size for Read on backends
-// that don't do TSO segmentation. 65535 covers any single IP packet.
-const defaultBatchBufSize = 65535
-
 type Device interface {
-	io.Closer
+	TunDev
 	Activate() error
 	Networks() []netip.Prefix
 	Name() string
 	RoutesFor(netip.Addr) routing.Gateways
-	SupportsMultiqueue() bool //todo remove?
-	NewMultiQueueReader() error
-	Readers() []tio.Queue
+	SupportsMultiqueue() bool
+	NewMultiQueueReader() (TunDev, error)
 }

overlay/eventfd/eventfd.go

@@ -0,0 +1,91 @@
+package eventfd
+
+import (
+	"encoding/binary"
+	"syscall"
+
+	"golang.org/x/sys/unix"
+)
+
+type EventFD struct {
+	fd  int
+	buf [8]byte
+}
+
+func New() (EventFD, error) {
+	fd, err := unix.Eventfd(0, unix.EFD_NONBLOCK)
+	if err != nil {
+		return EventFD{}, err
+	}
+	return EventFD{
+		fd:  fd,
+		buf: [8]byte{},
+	}, nil
+}
+
+func (e *EventFD) Kick() error {
+	binary.LittleEndian.PutUint64(e.buf[:], 1) //is this right???
+	_, err := syscall.Write(int(e.fd), e.buf[:])
+	return err
+}
+
+func (e *EventFD) Close() error {
+	if e.fd != 0 {
+		return unix.Close(e.fd)
+	}
+	return nil
+}
+
+func (e *EventFD) FD() int {
+	return e.fd
+}
+
+type Epoll struct {
+	fd     int
+	buf    [8]byte
+	events []syscall.EpollEvent
+}
+
+func NewEpoll() (Epoll, error) {
+	fd, err := unix.EpollCreate1(0)
+	if err != nil {
+		return Epoll{}, err
+	}
+	return Epoll{
+		fd:     fd,
+		buf:    [8]byte{},
+		events: make([]syscall.EpollEvent, 1),
+	}, nil
+}
+
+func (ep *Epoll) AddEvent(fdToAdd int) error {
+	event := syscall.EpollEvent{
+		Events: syscall.EPOLLIN,
+		Fd:     int32(fdToAdd),
+	}
+	return syscall.EpollCtl(ep.fd, syscall.EPOLL_CTL_ADD, fdToAdd, &event)
+}
+
+func (ep *Epoll) Block() (int, error) {
+	n, err := syscall.EpollWait(ep.fd, ep.events, -1)
+	if err != nil {
+		//goland:noinspection GoDirectComparisonOfErrors
+		if err == syscall.EINTR {
+			return 0, nil //??
+		}
+		return -1, err
+	}
+	return n, nil
+}
+
+func (ep *Epoll) Clear() error {
+	_, err := syscall.Read(int(ep.events[0].Fd), ep.buf[:])
+	return err
+}
+
+func (ep *Epoll) Close() error {
+	if ep.fd != 0 {
+		return unix.Close(ep.fd)
+	}
+	return nil
+}

overlay/overlaytest/noop.go

@@ -1,56 +0,0 @@
-// Package overlaytest provides fakes of overlay.Device for tests that do
-// not want to touch a real tun device or route table.
-package overlaytest
-
-import (
-	"errors"
-	"net/netip"
-
-	"github.com/slackhq/nebula/overlay/tio"
-	"github.com/slackhq/nebula/routing"
-)
-
-// NoopTun is an overlay.Device that silently discards every read and write.
-// Useful in tests that need to construct a nebula Interface but do not
-// exercise the datapath.
-type NoopTun struct{}
-
-func (NoopTun) RoutesFor(addr netip.Addr) routing.Gateways {
-	return routing.Gateways{}
-}
-
-func (NoopTun) Activate() error {
-	return nil
-}
-
-func (NoopTun) Networks() []netip.Prefix {
-	return []netip.Prefix{}
-}
-
-func (NoopTun) Name() string {
-	return "noop"
-}
-
-func (NoopTun) Read() ([][]byte, error) {
-	return nil, nil
-}
-
-func (NoopTun) Write([]byte) (int, error) {
-	return 0, nil
-}
-
-func (NoopTun) SupportsMultiqueue() bool {
-	return false
-}
-
-func (NoopTun) NewMultiQueueReader() error {
-	return errors.New("unsupported")
-}
-
-func (NoopTun) Readers() []tio.Queue {
-	return []tio.Queue{NoopTun{}}
-}
-
-func (NoopTun) Close() error {
-	return nil
-}

overlay/packets.go

@@ -0,0 +1,36 @@
+package overlay
+
+//import (
+//	"github.com/slackhq/nebula/util/virtio"
+//)
+
+//type VirtIOPacket struct {
+//	Payload   []byte
+//	Header    virtio.NetHdr
+//	Chains    []uint16
+//	ChainRefs [][]byte
+//}
+//
+//func NewVIO() *VirtIOPacket {
+//	out := new(VirtIOPacket)
+//	out.Payload = nil
+//	out.ChainRefs = make([][]byte, 0, 4)
+//	out.Chains = make([]uint16, 0, 8)
+//	return out
+//}
+//
+//func (v *VirtIOPacket) Reset() {
+//	v.Payload = nil
+//	v.ChainRefs = v.ChainRefs[:0]
+//	v.Chains = v.Chains[:0]
+//}
+
+// TunPacket is formerly VirtIOPacket
+type TunPacket interface {
+	SetPayload([]byte)
+	GetPayload() []byte
+}
+type OutPacket interface {
+	SetPayload([]byte)
+	GetPayload() []byte
+}

overlay/route.go

@@ -2,7 +2,6 @@ package overlay
 
 import (
 	"fmt"
-	"log/slog"
 	"math"
 	"net"
 	"net/netip"
@@ -10,6 +9,7 @@ import (
 	"strconv"
 
 	"github.com/gaissmai/bart"
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/routing"
 )
@@ -48,14 +48,11 @@ func (r Route) String() string {
 	return s
 }
 
-func makeRouteTree(l *slog.Logger, routes []Route, allowMTU bool) (*bart.Table[routing.Gateways], error) {
+func makeRouteTree(l *logrus.Logger, routes []Route, allowMTU bool) (*bart.Table[routing.Gateways], error) {
 	routeTree := new(bart.Table[routing.Gateways])
 	for _, r := range routes {
 		if !allowMTU && r.MTU > 0 {
-			l.Warn("route MTU is not supported on this platform",
-				"goos", runtime.GOOS,
-				"route", r,
-			)
+			l.WithField("route", r).Warnf("route MTU is not supported in %s", runtime.GOOS)
 		}
 
 		gateways := r.Via

overlay/route_test.go

@@ -295,7 +295,7 @@ func Test_makeRouteTree(t *testing.T) {
 	routes, err := parseUnsafeRoutes(c, []netip.Prefix{n})
 	require.NoError(t, err)
 	assert.Len(t, routes, 2)
-	routeTree, err := makeRouteTree(test.NewLogger(), routes, true)
+	routeTree, err := makeRouteTree(l, routes, true)
 	require.NoError(t, err)
 
 	ip, err := netip.ParseAddr("1.0.0.2")
@@ -367,7 +367,7 @@ func Test_makeMultipathUnsafeRouteTree(t *testing.T) {
 	routes, err := parseUnsafeRoutes(c, []netip.Prefix{n})
 	require.NoError(t, err)
 	assert.Len(t, routes, 3)
-	routeTree, err := makeRouteTree(test.NewLogger(), routes, true)
+	routeTree, err := makeRouteTree(l, routes, true)
 	require.NoError(t, err)
 
 	ip, err := netip.ParseAddr("192.168.86.1")

overlay/tio/container_gso_linux.go

@@ -1,70 +0,0 @@
-package tio
-
-import (
-	"encoding/binary"
-	"errors"
-	"fmt"
-
-	"golang.org/x/sys/unix"
-)
-
-type offloadContainer struct {
-	pq []*Offload
-	// pqi is exactly the same as pq, but stored as the interface type
-	pqi        []Queue
-	shutdownFd int
-}
-
-func NewOffloadContainer() (Container, error) {
-	shutdownFd, err := unix.Eventfd(0, unix.EFD_NONBLOCK|unix.EFD_CLOEXEC)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create eventfd: %w", err)
-	}
-
-	out := &offloadContainer{
-		pq:         []*Offload{},
-		pqi:        []Queue{},
-		shutdownFd: shutdownFd,
-	}
-
-	return out, nil
-}
-
-func (c *offloadContainer) Queues() []Queue {
-	return c.pqi
-}
-
-func (c *offloadContainer) Add(fd int) error {
-	x, err := newOffload(fd, c.shutdownFd)
-	if err != nil {
-		return err
-	}
-	c.pq = append(c.pq, x)
-	c.pqi = append(c.pqi, x)
-
-	return nil
-}
-
-func (c *offloadContainer) wakeForShutdown() error {
-	var buf [8]byte
-	binary.NativeEndian.PutUint64(buf[:], 1)
-	_, err := unix.Write(c.shutdownFd, buf[:])
-	return err
-}
-
-func (c *offloadContainer) Close() error {
-	errs := []error{}
-
-	// Signal all readers blocked in poll to wake up and exit
-	if err := c.wakeForShutdown(); err != nil {
-		errs = append(errs, err)
-	}
-
-	for _, x := range c.pq {
-		if err := x.Close(); err != nil {
-			errs = append(errs, err)
-		}
-	}
-
-	return errors.Join(errs...)
-}

overlay/tio/container_poll_linux.go

@@ -1,69 +0,0 @@
-package tio
-
-import (
-	"encoding/binary"
-	"errors"
-	"fmt"
-
-	"golang.org/x/sys/unix"
-)
-
-type pollContainer struct {
-	pq []*Poll
-	// pqi is exactly the same as pq, but stored as the interface type
-	pqi        []Queue
-	shutdownFd int
-}
-
-func NewPollContainer() (Container, error) {
-	shutdownFd, err := unix.Eventfd(0, unix.EFD_NONBLOCK|unix.EFD_CLOEXEC)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create eventfd: %w", err)
-	}
-
-	out := &pollContainer{
-		pq:         []*Poll{},
-		pqi:        []Queue{},
-		shutdownFd: shutdownFd,
-	}
-
-	return out, nil
-}
-
-func (c *pollContainer) Queues() []Queue {
-	return c.pqi
-}
-
-func (c *pollContainer) Add(fd int) error {
-	x, err := newPoll(fd, c.shutdownFd)
-	if err != nil {
-		return err
-	}
-	c.pq = append(c.pq, x)
-	c.pqi = append(c.pqi, x)
-
-	return nil
-}
-
-func (c *pollContainer) wakeForShutdown() error {
-	var buf [8]byte
-	binary.NativeEndian.PutUint64(buf[:], 1)
-	_, err := unix.Write(int(c.shutdownFd), buf[:])
-	return err
-}
-
-func (c *pollContainer) Close() error {
-	errs := []error{}
-
-	if err := c.wakeForShutdown(); err != nil {
-		errs = append(errs, err)
-	}
-
-	for _, x := range c.pq {
-		if err := x.Close(); err != nil {
-			errs = append(errs, err)
-		}
-	}
-
-	return errors.Join(errs...)
-}

overlay/tio/tio.go

@@ -1,67 +0,0 @@
-package tio
-
-import (
-	"io"
-)
-
-// defaultBatchBufSize is the per-Queue scratch size for Read on backends
-// that don't do TSO segmentation. 65535 covers any single IP packet.
-const defaultBatchBufSize = 65535
-
-// Container holds one or many Queue objects and helps close them in an orderly way
-type Container interface {
-	io.Closer
-	Queues() []Queue
-
-	// Add takes a tun fd, adds it to the container, and prepares it for use as a Queue
-	Add(fd int) error
-
-	io.Closer
-}
-
-// Queue is a readable/writable Poll queue. One Queue is driven by a single
-// read goroutine plus concurrent writers (see Write / WriteReject below).
-type Queue interface {
-	io.Closer
-
-	// Read returns one or more packets. The returned slices are borrowed
-	// from the Queue's internal buffer and are only valid until the next
-	// Read or Close on this Queue - callers must encrypt or copy each
-	// slice before the next call. Not safe for concurrent Reads.
-	Read() ([][]byte, error)
-
-	// Write emits a single packet on the plaintext (outside→inside)
-	// delivery path. Not safe for concurrent Writes.
-	Write(p []byte) (int, error)
-}
-
-// GSOWriter is implemented by Queues that can emit a TCP TSO superpacket
-// assembled from a header prefix plus one or more borrowed payload
-// fragments, in a single vectored write (writev with a leading
-// virtio_net_hdr). This lets the coalescer avoid copying payload bytes
-// between the caller's decrypt buffer and the TUN. Backends without GSO
-// support return false from GSOSupported and coalescing is skipped.
-//
-// hdr contains the IPv4/IPv6 + TCP header prefix (mutable - callers will
-// have filled in total length and pseudo-header partial). pays are
-// non-overlapping payload fragments whose concatenation is the full
-// superpacket payload; they are read-only from the writer's perspective
-// and must remain valid until the call returns. gsoSize is the MSS:
-// every segment except possibly the last is exactly that many bytes.
-// csumStart is the byte offset where the TCP header begins within hdr.
-//
-// # TODO fold into Queue
-//
-// hdr's TCP checksum field must already hold the pseudo-header partial
-// sum (single-fold, not inverted), per virtio NEEDS_CSUM semantics.
-type GSOWriter interface {
-	// WriteGSO emits a TCP TSO superpacket in a single writev. hdr is the
-	// IPv4/IPv6 + TCP header prefix (already finalized — total length, IP csum,
-	// and TCP pseudo-header partial set by the caller). pays are payload
-	// fragments whose concatenation forms the full coalesced payload; each
-	// slice is read-only and must stay valid until return.
-	// every segment in pays except possibly the last is exactly the same size.
-	// csumStart is the byte offset where the TCP header begins within hdr.
-	WriteGSO(hdr []byte, transportHdr []byte, pays [][]byte) error
-	GSOSupported() bool
-}

overlay/tio/tio_gso_linux.go

@@ -1,353 +0,0 @@
-package tio
-
-import (
-	"fmt"
-	"io"
-	"os"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"unsafe"
-
-	"golang.org/x/sys/unix"
-)
-
-// Space for segmented output. Worst case is many small segments, each paying
-// an IP+TCP header. Should be a multiple of 64KiB.
-// const tunSegBufSize = 0xffff * 8 TODO larger? config?
-const tunSegBufSize = 131072
-
-// tunSegBufCap is the total size we allocate for the per-reader segment
-// buffer. It is sized as one worst-case TSO superpacket (tunSegBufSize) plus
-// the same again as drain headroom so a Read wake can accumulate
-// additional packets after an initial big read without overflowing.
-const tunSegBufCap = tunSegBufSize * 2
-
-// tunDrainCap caps how many packets a single Read will accumulate via
-// the post-wake drain loop. Sized to soak up a burst of small ACKs while
-// bounding how much work a single caller holds before handing off.
-const tunDrainCap = 64 //256
-
-// gsoInitialPayIovs is the starting capacity (in payload fragments) of
-// Offload.gsoIovs. Sized to cover the default coalesce segment cap without
-// any reallocations.
-const gsoInitialPayIovs = 66
-
-// validVnetHdr is the 10-byte virtio_net_hdr we prepend to every non-GSO TUN
-// write. Only flag set is VIRTIO_NET_HDR_F_DATA_VALID, which marks the skb
-// CHECKSUM_UNNECESSARY so the receiving network stack skips L4 checks
-// verification. All packets that reach the plain Write paths
-// already carry a valid L4 checksum (either supplied by a remote peer whose
-// ciphertext we AEAD-authenticated, or produced by finishChecksum during TSO
-// segmentation, or built locally by CreateRejectPacket), so trusting them is
-// safe.
-var validVnetHdr = [virtioNetHdrLen]byte{unix.VIRTIO_NET_HDR_F_DATA_VALID}
-
-// Offload wraps a TUN file descriptor with poll-based reads. The FD provided will be changed to non-blocking.
-// A shared eventfd allows Close to wake all readers blocked in poll.
-type Offload struct {
-	fd         int
-	shutdownFd int
-	readPoll   [2]unix.PollFd
-	writePoll  [2]unix.PollFd
-	writeLock  sync.Mutex //there's more than one potential write source per-routine, so we need this to protect writePoll
-	closed     atomic.Bool
-	readBuf    []byte   // scratch for a single raw read (virtio hdr + superpacket)
-	segBuf     []byte   // backing store for segmented output
-	segOff     int      // cursor into segBuf for the current Read drain
-	pending    [][]byte // segments returned from the most recent Read
-
-	// gsoHdrBuf is a per-queue 10-byte scratch for the virtio_net_hdr emitted
-	// by WriteGSO. Separate from validVnetHdr so a concurrent non-GSO Write on
-	// another queue never observes a half-written header.
-	gsoHdrBuf [virtioNetHdrLen]byte
-	// gsoIovs is the writev iovec scratch for WriteGSO. Sized to hold the
-	// virtio header + IP/TCP header + up to gsoInitialPayIovs payload
-	// fragments; grown on demand if a coalescer pushes more.
-	gsoIovs []unix.Iovec
-}
-
-func newOffload(fd int, shutdownFd int) (*Offload, error) {
-	if err := unix.SetNonblock(fd, true); err != nil {
-		return nil, fmt.Errorf("failed to set tun fd non-blocking: %w", err)
-	}
-
-	out := &Offload{
-		fd:         fd,
-		shutdownFd: shutdownFd,
-		closed:     atomic.Bool{},
-		readBuf:    make([]byte, virtioNetHdrLen+tunReadBufSize),
-		readPoll: [2]unix.PollFd{
-			{Fd: int32(fd), Events: unix.POLLIN},
-			{Fd: int32(shutdownFd), Events: unix.POLLIN},
-		},
-		writePoll: [2]unix.PollFd{
-			{Fd: int32(fd), Events: unix.POLLOUT},
-			{Fd: int32(shutdownFd), Events: unix.POLLIN},
-		},
-		writeLock: sync.Mutex{},
-
-		segBuf:  make([]byte, tunSegBufCap),
-		gsoIovs: make([]unix.Iovec, 2, 2+gsoInitialPayIovs),
-	}
-
-	out.gsoIovs[0].Base = &out.gsoHdrBuf[0]
-	out.gsoIovs[0].SetLen(virtioNetHdrLen)
-
-	return out, nil
-}
-
-func (r *Offload) blockOnRead() error {
-	const problemFlags = unix.POLLHUP | unix.POLLNVAL | unix.POLLERR
-	var err error
-	for {
-		_, err = unix.Poll(r.readPoll[:], -1)
-		if err != unix.EINTR {
-			break
-		}
-	}
-	//always reset these!
-	tunEvents := r.readPoll[0].Revents
-	shutdownEvents := r.readPoll[1].Revents
-	r.readPoll[0].Revents = 0
-	r.readPoll[1].Revents = 0
-	//do the err check before trusting the potentially bogus bits we just got
-	if err != nil {
-		return err
-	}
-	if shutdownEvents&(unix.POLLIN|problemFlags) != 0 {
-		return os.ErrClosed
-	} else if tunEvents&problemFlags != 0 {
-		return os.ErrClosed
-	}
-	return nil
-}
-
-func (r *Offload) blockOnWrite() error {
-	const problemFlags = unix.POLLHUP | unix.POLLNVAL | unix.POLLERR
-	var err error
-	for {
-		_, err = unix.Poll(r.writePoll[:], -1)
-		if err != unix.EINTR {
-			break
-		}
-	}
-	//always reset these!
-	r.writeLock.Lock()
-	tunEvents := r.writePoll[0].Revents
-	shutdownEvents := r.writePoll[1].Revents
-	r.writePoll[0].Revents = 0
-	r.writePoll[1].Revents = 0
-	r.writeLock.Unlock()
-	//do the err check before trusting the potentially bogus bits we just got
-	if err != nil {
-		return err
-	}
-	if shutdownEvents&(unix.POLLIN|problemFlags) != 0 {
-		return os.ErrClosed
-	} else if tunEvents&problemFlags != 0 {
-		return os.ErrClosed
-	}
-	return nil
-}
-
-func (r *Offload) readRaw(buf []byte) (int, error) {
-	for {
-		if n, err := unix.Read(r.fd, buf); err == nil {
-			return n, nil
-		} else if err == unix.EAGAIN {
-			if err = r.blockOnRead(); err != nil {
-				return 0, err
-			}
-			continue
-		} else if err == unix.EINTR {
-			continue
-		} else if err == unix.EBADF {
-			return 0, os.ErrClosed
-		} else {
-			return 0, err
-		}
-	}
-}
-
-// Read reads one or more superpackets from the tun and returns the
-// resulting packets. The first read blocks via poll; once the fd is known
-// readable we drain additional packets non-blocking until the kernel queue
-// is empty (EAGAIN), we've collected tunDrainCap packets, or we're out of
-// segBuf headroom. This amortizes the poll wake over bursts of small
-// packets (e.g. TCP ACKs). Slices point into the Offload's internal buffers
-// and are only valid until the next Read or Close on this Queue.
-func (r *Offload) Read() ([][]byte, error) {
-	r.pending = r.pending[:0]
-	r.segOff = 0
-
-	// Initial (blocking) read. Retry on decode errors so a single bad
-	// packet does not stall the reader.
-	for {
-		n, err := r.readRaw(r.readBuf)
-		if err != nil {
-			return nil, err
-		}
-		if err := r.decodeRead(n); err != nil {
-			// Drop and read again — a bad packet should not kill the reader.
-			continue
-		}
-		break
-	}
-
-	// Drain: non-blocking reads until the kernel queue is empty, the drain
-	// cap is reached, or segBuf no longer has room for another worst-case
-	// superpacket.
-	for len(r.pending) < tunDrainCap && tunSegBufCap-r.segOff >= tunSegBufSize {
-		n, err := unix.Read(r.fd, r.readBuf)
-		if err != nil {
-			// EAGAIN / EINTR / anything else: stop draining. We already
-			// have a valid batch from the first read.
-			break
-		}
-		if n <= 0 {
-			break
-		}
-		if err := r.decodeRead(n); err != nil {
-			// Drop this packet and stop the drain; we'd rather hand off
-			// what we have than keep spinning here.
-			break
-		}
-	}
-
-	return r.pending, nil
-}
-
-// decodeRead decodes the virtio header plus payload in r.readBuf[:n], appends
-// the segments to r.pending, and advances r.segOff by the total scratch used.
-func (r *Offload) decodeRead(n int) error {
-	if n < virtioNetHdrLen {
-		return fmt.Errorf("short tun read: %d < %d", n, virtioNetHdrLen)
-	}
-	var hdr VirtioNetHdr
-	hdr.decode(r.readBuf[:virtioNetHdrLen])
-	before := len(r.pending)
-	if err := segmentInto(r.readBuf[virtioNetHdrLen:n], hdr, &r.pending, r.segBuf[r.segOff:]); err != nil {
-		return err
-	}
-	for k := before; k < len(r.pending); k++ {
-		r.segOff += len(r.pending[k])
-	}
-	return nil
-}
-
-func (r *Offload) Write(buf []byte) (int, error) {
-	iovs := [2]unix.Iovec{
-		{Base: &validVnetHdr[0]},
-		{Base: &buf[0]},
-	}
-	iovs[0].SetLen(virtioNetHdrLen)
-	iovs[1].SetLen(len(buf))
-	return r.writeWithScratch(buf, &iovs)
-}
-
-func (r *Offload) writeWithScratch(buf []byte, iovs *[2]unix.Iovec) (int, error) {
-	if len(buf) == 0 {
-		return 0, nil
-	}
-	// Point the payload iovec at the caller's buffer. iovs[0] is pre-wired
-	// to validVnetHdr during Offload construction so we don't rebuild it here.
-	iovs[1].Base = &buf[0]
-	iovs[1].SetLen(len(buf))
-	return r.rawWrite(unsafe.Slice(&iovs[0], len(iovs)))
-}
-
-func (r *Offload) rawWrite(iovs []unix.Iovec) (int, error) {
-	for {
-		n, _, errno := syscall.Syscall(unix.SYS_WRITEV, uintptr(r.fd), uintptr(unsafe.Pointer(&iovs[0])), uintptr(len(iovs)))
-		if errno == 0 {
-			if int(n) < virtioNetHdrLen {
-				return 0, io.ErrShortWrite
-			}
-			return int(n) - virtioNetHdrLen, nil
-		}
-		if errno == unix.EAGAIN {
-			if err := r.blockOnWrite(); err != nil {
-				return 0, err
-			}
-			continue
-		}
-		if errno == unix.EINTR {
-			continue
-		}
-		if errno == unix.EBADF {
-			return 0, os.ErrClosed
-		}
-		return 0, errno
-	}
-}
-
-// GSOSupported reports whether this queue was opened with IFF_VNET_HDR and
-// can accept WriteGSO. When false, callers should fall back to per-segment
-// Write calls.
-func (r *Offload) GSOSupported() bool { return true }
-
-func (r *Offload) WriteGSO(hdr []byte, transportHdr []byte, pays [][]byte) error {
-	if len(hdr) == 0 || len(pays) == 0 || len(transportHdr) == 0 {
-		return nil
-	}
-	vhdr := VirtioNetHdr{
-		Flags:      unix.VIRTIO_NET_HDR_F_NEEDS_CSUM,
-		HdrLen:     uint16(len(hdr) + len(transportHdr)),
-		GSOSize:    uint16(len(pays[0])),
-		CsumStart:  uint16(len(hdr)),
-		CsumOffset: 16, // TCP checksum field lives 16 bytes into the TCP header
-	}
-	if len(pays) > 1 {
-		ipVer := hdr[0] >> 4
-		if ipVer == 6 {
-			vhdr.GSOType = unix.VIRTIO_NET_HDR_GSO_TCPV6
-		} else if ipVer == 4 {
-			vhdr.GSOType = unix.VIRTIO_NET_HDR_GSO_TCPV4
-		} else {
-			vhdr.GSOType = unix.VIRTIO_NET_HDR_GSO_NONE
-			vhdr.GSOSize = 0
-		}
-	} else {
-		vhdr.GSOType = unix.VIRTIO_NET_HDR_GSO_NONE
-		vhdr.GSOSize = 0
-	}
-	vhdr.encode(r.gsoHdrBuf[:])
-
-	// Build the iovec array: [virtio_hdr, hdr, transportHdr, pays...]. r.gsoIovs[0] is
-	// wired to gsoHdrBuf at construction and never changes.
-	need := 3 + len(pays)
-	if cap(r.gsoIovs) < need {
-		grown := make([]unix.Iovec, need)
-		grown[0] = r.gsoIovs[0]
-		r.gsoIovs = grown
-	} else {
-		r.gsoIovs = r.gsoIovs[:need]
-	}
-	r.gsoIovs[1].Base = &hdr[0]
-	r.gsoIovs[1].SetLen(len(hdr))
-	r.gsoIovs[2].Base = &transportHdr[0]
-	r.gsoIovs[2].SetLen(len(transportHdr))
-	for i, p := range pays {
-		r.gsoIovs[3+i].Base = &p[0]
-		r.gsoIovs[3+i].SetLen(len(p))
-	}
-
-	_, err := r.rawWrite(r.gsoIovs)
-	return err
-}
-
-func (r *Offload) Close() error {
-	if r.closed.Swap(true) {
-		return nil
-	}
-
-	//shutdownFd is owned by the container, so we should not close it
-	var err error
-	if r.fd >= 0 {
-		err = unix.Close(r.fd)
-		r.fd = -1
-	}
-
-	return err
-}