Compare commits

..
Author SHA1 Message Date
Caleb JasikandClaude Opus 5.5 a33a6356c6 udp/darwin: batch receives with recvmsg_x
ListenOut drains the socket with recvmsg_x, xnu's private batched
recvmsg (SYS_RECVMSG_X, bsd/sys/socket_private.h), up to 64 datagrams
per syscall, and calls flush once per batch instead of once per packet.
Each msghdr_x carries its own sockaddr, so this works on the shared
unconnected socket. The send side stays on sendto: xnu only batches
sendmsg_x through the stack on connected sockets, and on an unconnected
socket like this one it falls back to one sendit per message, with
msg_name documented as unsupported.

recvmsg_x is private, so ListenOut falls back to one
ReadFromUDPAddrPort per datagram, logging once, if the kernel or a
sandbox refuses it (ENOSYS, EPERM, EOPNOTSUPP) or if a batch comes back
in a shape intact msghdr_x entries can't produce: more datagrams than
headers, a length past MTU, or a sender that isn't a matching
AF_INET/AF_INET6 sockaddr. That batch is dropped rather than delivered. IPv6 sources now arrive without their
zone, matching the Linux recvmmsg path; WriteTo already drops it.

macOS raw dnclient to a Windows peer on the same gigabit LAN, iperf3
10 s, median of 3 alternating rounds against 7cfa47d:

  download, 1/4/8 streams  606/600/604 -> 869/873/863 Mbps
  download CPU             ~159% -> ~123%
  upload                   340 -> 323 Mbps
  UDP 500M loss            0.32% -> 0.31%

Download is now close to the 902 Mbps the LAN gives without a tunnel.

TestMsghdrXKernelRoundTrip and FuzzMsghdrXKernelRoundTrip pin msghdrX
against the kernel's struct msghdr_x, which the SDK doesn't ship: they
round-trip bursts through recvmsg_x on loopback and check every field
the kernel writes. recvmsg_x never reports MSG_TRUNC (soreceive_m_list
doesn't set it), so a datagram past MTU arrives cut to MTU with Flags
0, the same as ReadFromUDP. TestListenOutSingleFallback covers the
fallback loop, TestCheckMsghdrX the shape check, and the round trip
asserts the check accepts every batch the kernel returns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-23 16:30:55 -05:00
19 changed files with 927 additions and 416 deletions

No files matched your search

-3
View File
@@ -7,7 +7,6 @@ import (
"testing"
"time"
"github.com/rcrowley/go-metrics"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/config"
"github.com/slackhq/nebula/overlay/overlaytest"
@@ -22,8 +21,6 @@ func newTestLighthouse() *LightHouse {
l: test.NewLogger(),
addrMap: map[netip.Addr]*RemoteList{},
queryChan: make(chan netip.Addr, 10),
// Nothing drains queryChan here, so an 11th query is dropped.
queryDropped: metrics.NilCounter{},
}
lighthouses := []netip.Addr{}
staticList := map[netip.Addr]struct{}{}
+2 -2
View File
@@ -410,8 +410,8 @@ logging:
#try_interval: 100ms
#retries: 20
# query_buffer is the size of the buffer channel for querying lighthouses, drops are recorded in the lighthouse.query.dropped metric.
#query_buffer: 1024
# query_buffer is the size of the buffer channel for querying lighthouses
#query_buffer: 64
# trigger_buffer is the size of the buffer channel for quickly sending handshakes
# after receiving the response for lighthouse queries
+111 -98
View File
@@ -14,7 +14,6 @@ import (
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/gaissmai/bart"
@@ -29,23 +28,16 @@ type FirewallInterface interface {
AddRule(incoming bool, proto uint8, startPort int32, endPort int32, groups []string, host string, cidr, localCidr string, caName string, caSha string) error
}
// conn is a conntrack entry. Only expires changes after it is stored, a rules version change
// stores a replacement entry instead.
type conn struct {
expires atomic.Uint32 // conntrack clock second when this entry expires, see FirewallConntrack.now
Expires time.Time // Time when this conntrack entry will expire
// record why the original connection passed the firewall, so we can re-validate
// after ruleset changes
// after ruleset changes. Note, rulesVersion is a uint16 so that these two
// fields pack for free after the uint32 above
incoming bool
rulesVersion uint16
}
func newConn(expires uint32, incoming bool, rulesVersion uint16) *conn {
c := &conn{incoming: incoming, rulesVersion: rulesVersion}
c.expires.Store(expires)
return c
}
// TODO: need conntrack max tracked connections handling
type Firewall struct {
Conntrack *FirewallConntrack
@@ -87,69 +79,11 @@ type firewallMetrics struct {
droppedNoRule metrics.Counter
}
// FirewallConntrack is safe for concurrent use by every routine without a lock.
type FirewallConntrack struct {
conns sync.Map // firewall.Packet -> *conn
count atomic.Int64 // entries in conns, which sync.Map can't tell us
now atomic.Uint32 // the conntrack clock: whole seconds since epoch, kept current by Run
epoch time.Time
}
sync.Mutex
func newFirewallConntrack() *FirewallConntrack {
return &FirewallConntrack{epoch: time.Now()}
}
// Run advances the conntrack clock every second and sweeps out expired entries every minute,
// until ctx is done. Reading a shared clock is much cheaper than asking the OS on every packet.
func (ct *FirewallConntrack) Run(ctx context.Context) {
t := time.NewTicker(time.Second)
defer t.Stop()
var nextSweep uint32
for {
select {
case <-ctx.Done():
return
case <-t.C:
// Measured from epoch rather than counted, so a late tick can't slow the clock down
now := uint32(time.Since(ct.epoch) / time.Second)
ct.now.Store(now)
if now >= nextSweep {
ct.sweep(now)
nextSweep = now + 60
}
}
}
}
// sweep deletes entries that expired at or before now.
func (ct *FirewallConntrack) sweep(now uint32) {
ct.conns.Range(func(k, v any) bool {
if v.(*conn).expires.Load() <= now {
ct.delete(k.(firewall.Packet), v.(*conn))
}
return true
})
}
func (ct *FirewallConntrack) store(fp firewall.Packet, c *conn) {
if _, replaced := ct.conns.Swap(fp, c); !replaced {
ct.count.Add(1)
}
}
// delete removes fp's entry if it is still c, so a newer entry stored by another routine survives
func (ct *FirewallConntrack) delete(fp firewall.Packet, c *conn) {
if ct.conns.CompareAndDelete(fp, c) {
ct.count.Add(-1)
}
}
// clear deletes every entry
func (ct *FirewallConntrack) clear() {
ct.conns.Range(func(k, v any) bool {
ct.delete(k.(firewall.Packet), v.(*conn))
return true
})
Conns map[firewall.Packet]*conn
TimerWheel *TimerWheel[firewall.Packet]
}
// FirewallTable is the entry point for a rule, the evaluation order is:
@@ -198,10 +132,26 @@ type firewallLocalCIDR struct {
LocalCIDR *bart.Lite
}
// NewFirewall creates a new Firewall object.
// 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 {
//TODO: error on 0 duration
var tmin, tmax time.Duration
if tcpTimeout < UDPTimeout {
tmin = tcpTimeout
tmax = UDPTimeout
} else {
tmin = UDPTimeout
tmax = tcpTimeout
}
if defaultTimeout < tmin {
tmin = defaultTimeout
} else if defaultTimeout > tmax {
tmax = defaultTimeout
}
routableNetworks := new(bart.Lite)
var assignedNetworks []netip.Prefix
for _, network := range c.Networks() {
@@ -216,7 +166,10 @@ func NewFirewall(l *slog.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.Dur
}
return &Firewall{
Conntrack: newFirewallConntrack(),
Conntrack: &FirewallConntrack{
Conns: make(map[firewall.Packet]*conn),
TimerWheel: NewTimerWheel[firewall.Packet](tmin, tmax),
},
InRules: newFirewallTable(),
OutRules: newFirewallTable(),
TCPTimeout: tcpTimeout,
@@ -470,7 +423,7 @@ var ErrNoMatchingRule = errors.New("no matching rule in firewall table")
// Drop returns an error if the packet should be dropped, explaining why. It
// returns nil if the packet should not be dropped.
func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool) error {
func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache) error {
// Make sure remote address matches nebula certificate, and determine how to treat it
if h.networks == nil {
// Simple case: Certificate has one address and no unsafe networks
@@ -505,7 +458,7 @@ func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *
}
// Check if we spoke to this tuple, if we did then allow this packet
if f.inConns(fp, h, caPool) {
if f.inConns(fp, h, caPool, localCache) {
return nil
}
@@ -541,19 +494,36 @@ func (f *Firewall) Destroy() {
}
func (f *Firewall) EmitStats() {
metrics.GetOrRegisterGauge("firewall.conntrack.count", nil).Update(f.Conntrack.count.Load())
conntrack := f.Conntrack
conntrack.Lock()
conntrackCount := len(conntrack.Conns)
conntrack.Unlock()
metrics.GetOrRegisterGauge("firewall.conntrack.count", nil).Update(int64(conntrackCount))
metrics.GetOrRegisterGauge("firewall.rules.version", nil).Update(int64(f.rulesVersion))
metrics.GetOrRegisterGauge("firewall.rules.hash", nil).Update(int64(f.GetRuleHashFNV()))
}
func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool) bool {
func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache) bool {
if localCache != nil {
if _, ok := localCache[fp]; ok {
return true
}
}
conntrack := f.Conntrack
v, ok := conntrack.conns.Load(fp)
conntrack.Lock()
// Purge every time we test
ep, has := conntrack.TimerWheel.Purge()
if has {
f.evict(ep)
}
c, ok := conntrack.Conns[fp]
if !ok {
conntrack.Unlock()
return false
}
c := v.(*conn)
expires := conntrack.now.Load() + f.timeout(fp.Protocol)
if c.rulesVersion != f.rulesVersion {
// This conntrack entry was for an older rule set, validate
@@ -573,7 +543,8 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool)
"oldRulesVersion", c.rulesVersion,
)
}
conntrack.delete(fp, c)
delete(conntrack.Conns, fp)
conntrack.Unlock()
return false
}
@@ -586,35 +557,77 @@ func (f *Firewall) inConns(fp firewall.Packet, h *HostInfo, caPool *cert.CAPool)
)
}
conntrack.store(fp, newConn(expires, c.incoming, f.rulesVersion))
return true
c.rulesVersion = f.rulesVersion
}
// Most packets land in the same second as the one before, skip writing an unchanged expiry
if c.expires.Load() != expires {
c.expires.Store(expires)
switch fp.Protocol {
case iputil.IPProtocolTCP:
c.Expires = time.Now().Add(f.TCPTimeout)
case iputil.IPProtocolUDP:
c.Expires = time.Now().Add(f.UDPTimeout)
default:
c.Expires = time.Now().Add(f.DefaultTimeout)
}
conntrack.Unlock()
if localCache != nil {
localCache[fp] = struct{}{}
}
return true
}
func (f *Firewall) addConn(fp firewall.Packet, incoming bool) {
var timeout time.Duration
c := &conn{}
switch fp.Protocol {
case iputil.IPProtocolTCP:
timeout = f.TCPTimeout
case iputil.IPProtocolUDP:
timeout = f.UDPTimeout
default:
timeout = f.DefaultTimeout
}
conntrack := f.Conntrack
conntrack.Lock()
if _, ok := conntrack.Conns[fp]; !ok {
conntrack.TimerWheel.Advance(time.Now())
conntrack.TimerWheel.Add(fp, timeout)
}
// Record which rulesVersion allowed this connection, so we can retest after
// firewall reload
conntrack := f.Conntrack
conntrack.store(fp, newConn(conntrack.now.Load()+f.timeout(fp.Protocol), incoming, f.rulesVersion))
c.incoming = incoming
c.rulesVersion = f.rulesVersion
c.Expires = time.Now().Add(timeout)
conntrack.Conns[fp] = c
conntrack.Unlock()
}
// timeout returns how many seconds a conntrack entry for proto lives past its last packet
func (f *Firewall) timeout(proto uint8) uint32 {
switch proto {
case iputil.IPProtocolTCP:
return uint32(f.TCPTimeout / time.Second)
case iputil.IPProtocolUDP:
return uint32(f.UDPTimeout / time.Second)
default:
return uint32(f.DefaultTimeout / time.Second)
// 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) {
// Are we still tracking this conn?
conntrack := f.Conntrack
t, ok := conntrack.Conns[p]
if !ok {
return
}
newT := t.Expires.Sub(time.Now())
// Timeout is in the future, re-add the timer
if newT > 0 {
conntrack.TimerWheel.Advance(time.Now())
conntrack.TimerWheel.Add(p, newT)
return
}
// This conn is done
delete(conntrack.Conns, p)
}
func (ft *FirewallTable) match(p firewall.Packet, incoming bool, c *cert.CachedCertificate, caPool *cert.CAPool) bool {
+69
View File
@@ -0,0 +1,69 @@
package firewall
import (
"context"
"log/slog"
"sync/atomic"
"time"
"github.com/slackhq/nebula/logging"
)
// ConntrackCache is used as a local routine cache to know if a given flow
// has been seen in the conntrack table.
type ConntrackCache map[Packet]struct{}
type ConntrackCacheTicker struct {
cacheV uint64
cacheTick atomic.Uint64
l *slog.Logger
cache ConntrackCache
}
func NewConntrackCacheTicker(ctx context.Context, l *slog.Logger, d time.Duration) *ConntrackCacheTicker {
if d == 0 {
return nil
}
c := &ConntrackCacheTicker{
l: l,
cache: ConntrackCache{},
}
go c.tick(ctx, d)
return c
}
func (c *ConntrackCacheTicker) tick(ctx context.Context, d time.Duration) {
t := time.NewTicker(d)
defer t.Stop()
for {
select {
case <-ctx.Done():
return
case <-t.C:
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 {
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(), logging.LevelTrace) {
c.l.Log(context.Background(), logging.LevelTrace, "resetting conntrack cache", "len", ll)
}
c.cache = make(ConntrackCache, ll)
}
}
return c.cache
}
+70
View File
@@ -0,0 +1,70 @@
package firewall
import (
"bytes"
"log/slog"
"strings"
"testing"
"github.com/slackhq/nebula/logging"
"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 := range cacheLen {
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, logging.LevelTrace)
c := newFixedTicker(t, l, 3)
c.Get()
assert.Equal(t, "level=DEBUG-4 msg=\"resetting conntrack cache\" len=3\n", buf.String())
}
func TestConntrackCacheTicker_Get_JSONFormat(t *testing.T) {
buf := &bytes.Buffer{}
l := test.NewJSONLoggerWithOutput(buf, logging.LevelTrace)
c := newFixedTicker(t, l, 2)
c.Get()
assert.JSONEq(t, `{"level":"DEBUG-4","msg":"resetting conntrack cache","len":2}`, strings.TrimSpace(buf.String()))
}
func TestConntrackCacheTicker_Get_QuietBelowTrace(t *testing.T) {
buf := &bytes.Buffer{}
l := test.NewLoggerWithOutputAndLevel(buf, slog.LevelDebug)
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, logging.LevelTrace)
c := newFixedTicker(t, l, 0)
c.Get()
assert.Empty(t, buf.String())
}
-4
View File
@@ -26,10 +26,6 @@ type Packet struct {
RemotePort uint16
Protocol uint8
Fragment bool
// pad fills what would otherwise be compiler padding. A Packet with no padding hashes as one
// block of memory, which makes it a much faster map key. Always zero.
pad [2]byte
}
func (fp *Packet) Copy() *Packet {
+102 -128
View File
@@ -2,7 +2,6 @@ package nebula
import (
"bytes"
"context"
"errors"
"log/slog"
"math"
@@ -24,12 +23,39 @@ func TestNewFirewall(t *testing.T) {
l := test.NewLogger()
c := &dummyCert{}
fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c)
assert.NotNil(t, fw.Conntrack)
conntrack := fw.Conntrack
assert.NotNil(t, conntrack)
assert.NotNil(t, conntrack.Conns)
assert.NotNil(t, conntrack.TimerWheel)
assert.NotNil(t, fw.InRules)
assert.NotNil(t, fw.OutRules)
assert.Equal(t, time.Second, fw.TCPTimeout)
assert.Equal(t, time.Minute, fw.UDPTimeout)
assert.Equal(t, time.Hour, fw.DefaultTimeout)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
fw = NewFirewall(l, time.Second, time.Hour, time.Minute, c)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
fw = NewFirewall(l, time.Hour, time.Second, time.Minute, c)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
fw = NewFirewall(l, time.Hour, time.Minute, time.Second, c)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
fw = NewFirewall(l, time.Minute, time.Hour, time.Second, c)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
fw = NewFirewall(l, time.Minute, time.Second, time.Hour, c)
assert.Equal(t, time.Hour, conntrack.TimerWheel.wheelDuration)
assert.Equal(t, 3602, conntrack.TimerWheel.wheelLen)
}
func TestFirewall_AddRule(t *testing.T) {
@@ -186,44 +212,44 @@ func TestFirewall_Drop(t *testing.T) {
cp := cert.NewCAPool()
// Drop outbound
assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp))
assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp, nil))
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
// Allow outbound because conntrack
require.NoError(t, fw.Drop(p, false, &h, cp))
require.NoError(t, fw.Drop(p, false, &h, cp, nil))
// test remote mismatch
oldRemote := p.RemoteAddr
p.RemoteAddr = netip.MustParseAddr("1.2.3.10")
assert.Equal(t, fw.Drop(p, false, &h, cp), ErrInvalidRemoteIP)
assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrInvalidRemoteIP)
p.RemoteAddr = oldRemote
// ensure signer doesn't get in the way of group checks
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "", "signer-shasum"))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum-bad"))
assert.Equal(t, fw.Drop(p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
// test caSha doesn't drop on match
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "", "signer-shasum-bad"))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum"))
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
// ensure ca name doesn't get in the way of group checks
cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "ca-good", ""))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good-bad", ""))
assert.Equal(t, fw.Drop(p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
// test caName doesn't drop on match
cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "ca-good-bad", ""))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good", ""))
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
}
func TestFirewall_DropV6(t *testing.T) {
@@ -264,44 +290,44 @@ func TestFirewall_DropV6(t *testing.T) {
cp := cert.NewCAPool()
// Drop outbound
assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp))
assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp, nil))
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
// Allow outbound because conntrack
require.NoError(t, fw.Drop(p, false, &h, cp))
require.NoError(t, fw.Drop(p, false, &h, cp, nil))
// test remote mismatch
oldRemote := p.RemoteAddr
p.RemoteAddr = netip.MustParseAddr("fd12::56")
assert.Equal(t, fw.Drop(p, false, &h, cp), ErrInvalidRemoteIP)
assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrInvalidRemoteIP)
p.RemoteAddr = oldRemote
// ensure signer doesn't get in the way of group checks
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "", "signer-shasum"))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum-bad"))
assert.Equal(t, fw.Drop(p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
// test caSha doesn't drop on match
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "", "signer-shasum-bad"))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum"))
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
// ensure ca name doesn't get in the way of group checks
cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "ca-good", ""))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good-bad", ""))
assert.Equal(t, fw.Drop(p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
// test caName doesn't drop on match
cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", "", "", "ca-good-bad", ""))
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good", ""))
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
}
func BenchmarkFirewallTable_match(b *testing.B) {
@@ -508,10 +534,10 @@ func TestFirewall_Drop2(t *testing.T) {
cp := cert.NewCAPool()
// h1/c1 lacks the proper groups
require.ErrorIs(t, fw.Drop(p, true, &h1, cp), ErrNoMatchingRule)
require.ErrorIs(t, fw.Drop(p, true, &h1, cp, nil), ErrNoMatchingRule)
// c has the proper groups
resetConntrack(fw)
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
}
func TestFirewall_Drop3(t *testing.T) {
@@ -588,18 +614,18 @@ func TestFirewall_Drop3(t *testing.T) {
cp := cert.NewCAPool()
// c1 should pass because host match
require.NoError(t, fw.Drop(p, true, &h1, cp))
require.NoError(t, fw.Drop(p, true, &h1, cp, nil))
// c2 should pass because ca sha match
resetConntrack(fw)
require.NoError(t, fw.Drop(p, true, &h2, cp))
require.NoError(t, fw.Drop(p, true, &h2, cp, nil))
// c3 should fail because no match
resetConntrack(fw)
assert.Equal(t, fw.Drop(p, true, &h3, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, true, &h3, cp, nil), ErrNoMatchingRule)
// Test a remote address match
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "", "1.2.3.4/24", "", "", ""))
require.NoError(t, fw.Drop(p, true, &h1, cp))
require.NoError(t, fw.Drop(p, true, &h1, cp, nil))
}
func TestFirewall_Drop3V6(t *testing.T) {
@@ -636,7 +662,7 @@ func TestFirewall_Drop3V6(t *testing.T) {
fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
cp := cert.NewCAPool()
require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "", "fd12::34/120", "", "", ""))
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
}
func TestFirewall_DropConntrackReload(t *testing.T) {
@@ -677,12 +703,12 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
cp := cert.NewCAPool()
// Drop outbound
assert.Equal(t, fw.Drop(p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(p, true, &h, cp))
require.NoError(t, fw.Drop(p, true, &h, cp, nil))
// Allow outbound because conntrack
require.NoError(t, fw.Drop(p, false, &h, cp))
require.NoError(t, fw.Drop(p, false, &h, cp, nil))
oldFw := fw
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
@@ -691,7 +717,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
fw.rulesVersion = oldFw.rulesVersion + 1
// Allow outbound because conntrack and new rules allow port 10
require.NoError(t, fw.Drop(p, false, &h, cp))
require.NoError(t, fw.Drop(p, false, &h, cp, nil))
oldFw = fw
fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
@@ -700,7 +726,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
fw.rulesVersion = oldFw.rulesVersion + 1
// Drop outbound because conntrack doesn't match new ruleset
assert.Equal(t, fw.Drop(p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrNoMatchingRule)
}
func TestFirewall_ICMPPortBehavior(t *testing.T) {
@@ -745,12 +771,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0
p.RemotePort = 0
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(*p, true, &h, cp))
require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
//now also allow outbound
require.NoError(t, fw.Drop(*p, false, &h, cp))
require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
})
t.Run("nonzero ports", func(t *testing.T) {
@@ -758,12 +784,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0xabcd
p.RemotePort = 0x1234
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(*p, true, &h, cp))
require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
//now also allow outbound
require.NoError(t, fw.Drop(*p, false, &h, cp))
require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
})
})
@@ -775,12 +801,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0
p.RemotePort = 0
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
assert.Equal(t, fw.Drop(*p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
//now also allow outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
})
t.Run("nonzero ports, still blocked", func(t *testing.T) {
@@ -788,12 +814,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0xabcd
p.RemotePort = 0x1234
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
assert.Equal(t, fw.Drop(*p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
//now also allow outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
})
t.Run("nonzero, matching ports, still blocked", func(t *testing.T) {
@@ -801,12 +827,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 80
p.RemotePort = 80
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
assert.Equal(t, fw.Drop(*p, true, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, true, &h, cp, nil), ErrNoMatchingRule)
//now also allow outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
})
})
t.Run("Any proto, any port", func(t *testing.T) {
@@ -818,12 +844,12 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0
p.RemotePort = 0
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(*p, true, &h, cp))
require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
//now also allow outbound
require.NoError(t, fw.Drop(*p, false, &h, cp))
require.NoError(t, fw.Drop(*p, false, &h, cp, nil))
})
t.Run("nonzero ports, allowed", func(t *testing.T) {
@@ -832,15 +858,15 @@ func TestFirewall_ICMPPortBehavior(t *testing.T) {
p.LocalPort = 0xabcd
p.RemotePort = 0x1234
// Drop outbound
assert.Equal(t, fw.Drop(*p, false, &h, cp), ErrNoMatchingRule)
assert.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
// Allow inbound
resetConntrack(fw)
require.NoError(t, fw.Drop(*p, true, &h, cp))
require.NoError(t, fw.Drop(*p, true, &h, cp, nil))
//now also allow outbound
require.NoError(t, fw.Drop(*p, false, &h, cp))
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), ErrNoMatchingRule)
require.Equal(t, fw.Drop(*p, false, &h, cp, nil), ErrNoMatchingRule)
})
})
@@ -888,7 +914,7 @@ func TestFirewall_DropIPSpoofing(t *testing.T) {
Protocol: iputil.IPProtocolUDP,
Fragment: false,
}
assert.Equal(t, fw.Drop(p, true, &h1, cp), ErrInvalidRemoteIP)
assert.Equal(t, fw.Drop(p, true, &h1, cp, nil), ErrInvalidRemoteIP)
}
func TestFirewall_ConntrackSourceSpoofingAcrossPeers(t *testing.T) {
@@ -939,84 +965,20 @@ func TestFirewall_ConntrackSourceSpoofingAcrossPeers(t *testing.T) {
Protocol: iputil.IPProtocolUDP,
}
require.NoError(t, fw.Drop(flow, true, &victimHI, cp),
require.NoError(t, fw.Drop(flow, true, &victimHI, cp, nil),
"victim's own traffic from its own overlay IP must be allowed")
unseen := flow
unseen.RemotePort = 55001
assert.Equal(t, ErrInvalidRemoteIP, fw.Drop(unseen, true, &attackerHI, cp),
assert.Equal(t, ErrInvalidRemoteIP, fw.Drop(unseen, true, &attackerHI, cp, nil),
"sanity: attacker forging victim's source IP must be rejected when no conntrack entry exists")
got := fw.Drop(flow, true, &attackerHI, cp)
got := fw.Drop(flow, true, &attackerHI, cp, nil)
t.Logf("attacker replaying victim's 4-tuple: Drop returned %v (nil == packet ALLOWED == spoof succeeded)", got)
assert.Equal(t, ErrInvalidRemoteIP, got,
"SECURITY: attacker spoofed victim's overlay source IP (192.0.2.2) by reusing an existing conntrack 4-tuple; Drop returned %v instead of rejecting", got)
}
func TestFirewallConntrack_sweep(t *testing.T) {
ct := newFirewallConntrack()
live := firewall.Packet{LocalPort: 1}
dead := firewall.Packet{LocalPort: 2}
ct.store(live, newConn(11, true, 0))
ct.store(dead, newConn(10, true, 0))
assert.Equal(t, int64(2), ct.count.Load())
ct.sweep(10)
_, ok := ct.conns.Load(live)
assert.True(t, ok)
_, ok = ct.conns.Load(dead)
assert.False(t, ok)
assert.Equal(t, int64(1), ct.count.Load())
}
func TestFirewallConntrack_storeAndDeleteCount(t *testing.T) {
ct := newFirewallConntrack()
fp := firewall.Packet{LocalPort: 1}
first := newConn(10, true, 0)
ct.store(fp, first)
// Replacing an entry doesn't add to the count
second := newConn(20, true, 1)
ct.store(fp, second)
assert.Equal(t, int64(1), ct.count.Load())
// Deleting a stale entry is a no-op, the newer one stays
ct.delete(fp, first)
v, ok := ct.conns.Load(fp)
assert.True(t, ok)
assert.Same(t, second, v)
assert.Equal(t, int64(1), ct.count.Load())
ct.delete(fp, second)
_, ok = ct.conns.Load(fp)
assert.False(t, ok)
assert.Equal(t, int64(0), ct.count.Load())
ct.store(fp, first)
ct.store(firewall.Packet{LocalPort: 2}, newConn(10, true, 0))
ct.clear()
assert.Equal(t, int64(0), ct.count.Load())
}
func TestFirewallConntrack_Run(t *testing.T) {
ct := newFirewallConntrack()
ct.epoch = time.Now().Add(-time.Hour)
ct.store(firewall.Packet{LocalPort: 1}, newConn(10, true, 0))
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() {
ct.Run(ctx)
close(done)
}()
// The first tick sets the clock to an hour in and sweeps the long expired entry
require.Eventually(t, func() bool { return ct.count.Load() == 0 }, 5*time.Second, 10*time.Millisecond)
assert.GreaterOrEqual(t, ct.now.Load(), uint32(3600))
cancel()
<-done
}
// BenchmarkFirewallDropConntrackHit measures Drop on an already-established flow
// (a conntrack hit). This is the fast path that the source-IP<->cert binding
// reordering adds work to, so it quantifies the cost of moving the address checks
@@ -1025,6 +987,8 @@ func TestFirewallConntrack_Run(t *testing.T) {
// so the remote-address check is a single netip.Addr compare.
// - complex: peer cert has unsafe networks (h.networks populated), so the
// remote-address check is a BART lookup.
// - noCache/localCache: whether a per-batch ConntrackCache is supplied, which in
// the original code let the fast path skip straight past the address checks.
func BenchmarkFirewallDropConntrackHit(b *testing.B) {
l := test.NewLoggerWithOutput(&bytes.Buffer{})
@@ -1072,11 +1036,14 @@ func BenchmarkFirewallDropConntrackHit(b *testing.B) {
}
cases := []struct {
name string
host *HostInfo
name string
host *HostInfo
useCache bool
}{
{"simple", simpleHost},
{"complex", complexHost},
{"simple/noCache", simpleHost, false},
{"simple/localCache", simpleHost, true},
{"complex/noCache", complexHost, false},
{"complex/localCache", complexHost, true},
}
for _, tc := range cases {
@@ -1085,12 +1052,17 @@ func BenchmarkFirewallDropConntrackHit(b *testing.B) {
require.NoError(b, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", "", "", "", ""))
// Establish the conntrack entry so every benchmarked Drop is a hit.
require.NoError(b, fw.Drop(flow, true, tc.host, cp))
require.NoError(b, fw.Drop(flow, true, tc.host, cp, nil))
var cache firewall.ConntrackCache
if tc.useCache {
cache = firewall.ConntrackCache{}
}
b.ReportAllocs()
b.ResetTimer()
for i := 0; i < b.N; i++ {
if err := fw.Drop(flow, true, tc.host, cp); err != nil {
if err := fw.Drop(flow, true, tc.host, cp, cache); err != nil {
b.Fatal(err)
}
}
@@ -1578,7 +1550,7 @@ func (c *testcase) Test(t *testing.T, fw *Firewall) {
t.Helper()
cp := cert.NewCAPool()
resetConntrack(fw)
err := fw.Drop(c.p, true, c.h, cp)
err := fw.Drop(c.p, true, c.h, cp, nil)
if c.err == nil {
require.NoError(t, err, "failed to not drop remote address %s", c.p.RemoteAddr)
} else {
@@ -1769,5 +1741,7 @@ func (mf *mockFirewall) AddRule(incoming bool, proto uint8, startPort int32, end
}
func resetConntrack(fw *Firewall) {
fw.Conntrack.clear()
fw.Conntrack.Lock()
fw.Conntrack.Conns = map[firewall.Packet]*conn{}
fw.Conntrack.Unlock()
}
+3 -3
View File
@@ -16,7 +16,7 @@ import (
"github.com/slackhq/nebula/routing"
)
func (f *Interface) consumeInsidePacket(pkt tio.Packet, fwPacket *firewall.ParsedPacket, nb []byte, sendBatch *batch.SendBatch, rejectBuf []byte, q int) {
func (f *Interface) consumeInsidePacket(pkt tio.Packet, fwPacket *firewall.ParsedPacket, nb []byte, sendBatch *batch.SendBatch, rejectBuf []byte, q int, localCache firewall.ConntrackCache) {
// borrowed: pkt.Bytes is owned by the originating tio.Queue and is
// only valid until the next Read on that queue. Every consumer below
// (parse, self-forward, handshake cache, sendInsideMessage) reads it
@@ -109,7 +109,7 @@ func (f *Interface) consumeInsidePacket(pkt tio.Packet, fwPacket *firewall.Parse
return
}
dropReason := f.firewall.Drop(fwPacket.Packet, false, hostinfo, f.pki.GetCAPool())
dropReason := f.firewall.Drop(fwPacket.Packet, false, hostinfo, f.pki.GetCAPool(), localCache)
if dropReason == nil {
f.sendInsideMessage(hostinfo, pkt, nb, sendBatch)
} else {
@@ -380,7 +380,7 @@ func (f *Interface) sendMessageNow(t header.MessageType, st header.MessageSubTyp
}
// check if packet is in outbound fw rules
dropReason := f.firewall.Drop(fp.Packet, false, hostinfo, f.pki.GetCAPool())
dropReason := f.firewall.Drop(fp.Packet, false, hostinfo, f.pki.GetCAPool(), nil)
if dropReason != nil {
if f.l.Enabled(context.Background(), slog.LevelDebug) {
f.l.Debug("dropping cached packet",
+1 -1
View File
@@ -50,7 +50,7 @@ func newSelfForwardInterface(myAddrs ...netip.Addr) (*Interface, *capturingTun)
}
func consumeInside(f *Interface, packet []byte) {
f.consumeInsidePacket(tio.Packet{Bytes: packet}, &firewall.ParsedPacket{}, make([]byte, 12), nil, make([]byte, mtu), 0)
f.consumeInsidePacket(tio.Packet{Bytes: packet}, &firewall.ParsedPacket{}, make([]byte, 12), nil, make([]byte, mtu), 0, nil)
}
// l4Proto describes one upper-layer header for these tests: its IP next-header
+17 -18
View File
@@ -52,6 +52,8 @@ type InterfaceConfig struct {
reQueryEvery uint32
reQueryWait time.Duration
ConntrackCacheTimeout time.Duration
// CpuAffinity, when non-empty, names the CPUs each TUN reader goroutine
// should pin to. Queue i pins to CpuAffinity[i % len(CpuAffinity)] —
// shorter lists than `routines` cycle. Empty list keeps the default
@@ -109,6 +111,8 @@ type Interface struct {
rebindEpoch atomic.Uint32
version string
conntrackCacheTimeout time.Duration
ctx context.Context
writers []udp.Conn
queues []tio.Queue
@@ -223,6 +227,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
myBroadcastAddrsTable: cs.myVpnBroadcastAddrsTable,
relayManager: c.relayManager,
connectionManager: c.connectionManager,
conntrackCacheTimeout: c.ConntrackCacheTimeout,
cpuAffinity: c.CpuAffinity,
pinThreads: c.PinThreads,
@@ -356,19 +361,8 @@ type rxContext struct {
h *header.H
fwPacket *firewall.ParsedPacket
hostmapCache map[uint32]*HostInfo
// lastAllowed is the last flow the inbound firewall let in during this batch. UDP GRO hands us segments
// in runs from the same flow, so the rest of a run can skip the firewall. Cleared by endBatch.
lastAllowed firewall.Packet
lastAllowedHost *HostInfo
lhh *LightHouseHandler
}
// endBatch forgets what was learned during a batch, so nothing outlives a hostmap or firewall change by more than one batch
func (rxc *rxContext) endBatch() {
clear(rxc.hostmapCache)
rxc.lastAllowed = firewall.Packet{}
rxc.lastAllowedHost = nil
lhh *LightHouseHandler
ctCache *firewall.ConntrackCacheTicker
}
func newRxContext(f *Interface, q int) *rxContext {
@@ -380,6 +374,7 @@ func newRxContext(f *Interface, q int) *rxContext {
fwPacket: &firewall.ParsedPacket{},
hostmapCache: map[uint32]*HostInfo{},
lhh: f.lightHouse.NewRequestHandler(),
ctCache: firewall.NewConntrackCacheTicker(f.ctx, f.l, f.conntrackCacheTimeout),
}
}
@@ -401,7 +396,7 @@ func (f *Interface) listenOut(i int) {
if err := f.batchers[i].Flush(); err != nil {
f.l.Error("Failed to flush tun coalescer", "error", err)
}
rxc.endBatch()
clear(rxc.hostmapCache)
}
err := li.ListenOut(listener, flusher)
@@ -449,6 +444,8 @@ func (f *Interface) listenIn(queue tio.Queue, i int) {
fwPacket := &firewall.ParsedPacket{}
nb := make([]byte, 12, 12)
conntrackCache := firewall.NewConntrackCacheTicker(f.ctx, f.l, f.conntrackCacheTimeout)
for {
pkts, err := queue.Read()
if err != nil {
@@ -461,7 +458,7 @@ func (f *Interface) listenIn(queue tio.Queue, i int) {
}
for _, pkt := range pkts {
f.consumeInsidePacket(pkt, fwPacket, nb, sb, rejectBuf, i)
f.consumeInsidePacket(pkt, fwPacket, nb, sb, rejectBuf, i, conntrackCache.Get())
// Flush incrementally once a full sendmmsg batch has
// accumulated so the first packets of a deep read drain
// hit the wire while the rest are still being encrypted.
@@ -534,8 +531,9 @@ func (f *Interface) reloadFirewall(c *config.C) {
}
oldFw := f.firewall
// Conntrack lives as long as the interface, keep using the running one
fw.Conntrack = oldFw.Conntrack
conntrack := oldFw.Conntrack
conntrack.Lock()
defer conntrack.Unlock()
fw.rulesVersion = oldFw.rulesVersion + 1
// If rulesVersion is back to zero, we have wrapped all the way around. Be
@@ -546,7 +544,8 @@ func (f *Interface) reloadFirewall(c *config.C) {
"oldFirewallHashes", oldFw.GetRuleHashes(),
"rulesVersion", fw.rulesVersion,
)
fw.Conntrack.clear()
} else {
fw.Conntrack = conntrack
}
f.firewall = fw
+2 -1
View File
@@ -10,6 +10,7 @@ import (
"github.com/rcrowley/go-metrics"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/firewall"
"github.com/slackhq/nebula/overlay/overlaytest"
"github.com/slackhq/nebula/test"
"github.com/slackhq/nebula/udp"
@@ -42,7 +43,7 @@ func Test_emitStats_primesGauges(t *testing.T) {
hostMap: hostMap,
inside: &overlaytest.NoopTun{},
outside: &udp.NoopConn{},
firewall: &Firewall{Conntrack: newFirewallConntrack()},
firewall: &Firewall{Conntrack: &FirewallConntrack{Conns: map[firewall.Packet]*conn{}}},
lightHouse: lh,
pki: &PKI{},
handshakeManager: NewHandshakeManager(l, hostMap, lh, &udp.NoopConn{}, defaultHandshakeConfig),
+3 -12
View File
@@ -15,7 +15,6 @@ import (
"time"
"github.com/gaissmai/bart"
"github.com/rcrowley/go-metrics"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/config"
"github.com/slackhq/nebula/header"
@@ -82,8 +81,6 @@ type LightHouse struct {
updateTrigger chan struct{}
queryChan chan netip.Addr
// queryDropped counts queries QueryServer dropped because queryChan was full.
queryDropped metrics.Counter
calculatedRemotes atomic.Pointer[bart.Table[[]*calculatedRemote]] // Maps VpnAddr to []*calculatedRemote
@@ -119,8 +116,7 @@ func NewLightHouseFromConfig(ctx context.Context, l *slog.Logger, c *config.C, c
nebulaPort: nebulaPort,
punchy: p,
updateTrigger: make(chan struct{}, 1),
queryChan: make(chan netip.Addr, c.GetUint32("handshakes.query_buffer", 1024)),
queryDropped: metrics.GetOrRegisterCounter("lighthouse.query.dropped", nil),
queryChan: make(chan netip.Addr, c.GetUint32("handshakes.query_buffer", 64)),
l: l,
}
h.localAddrsFn = func(al *LocalAllowList) []netip.Addr {
@@ -516,19 +512,14 @@ func (lh *LightHouse) Query(vpnAddr netip.Addr) *RemoteList {
return nil
}
// QueryServer is asynchronous and non-blocking, no reply should be expected.
// If the query channel is full, the query is dropped.
// QueryServer is asynchronous so no reply should be expected
func (lh *LightHouse) QueryServer(vpnAddr netip.Addr) {
// Don't put lighthouse addrs in the query channel because we can't query lighthouses about lighthouses
if lh.amLighthouse || lh.IsLighthouseAddr(vpnAddr) {
return
}
select {
case lh.queryChan <- vpnAddr:
default:
lh.queryDropped.Inc(1)
}
lh.queryChan <- vpnAddr
}
func (lh *LightHouse) QueryCache(vpnAddrs []netip.Addr) *RemoteList {
-54
View File
@@ -6,9 +6,7 @@ import (
"errors"
"fmt"
"net/netip"
"sync"
"testing"
"time"
"github.com/gaissmai/bart"
"github.com/slackhq/nebula/cert"
@@ -797,55 +795,3 @@ func TestLightHouse_logLocalAddrsErr(t *testing.T) {
lh.logLocalAddrsErr(errors.New("something else"))
assert.Contains(t, out.String(), "level=WARN")
}
// heldEncWriter holds the query worker in its first GetHostInfo until release closes.
type heldEncWriter struct {
mockEncWriter
once sync.Once
held chan struct{}
release chan struct{}
}
func (w *heldEncWriter) GetHostInfo(netip.Addr) *HostInfo {
w.once.Do(func() { close(w.held) })
<-w.release
return nil
}
// QueryServer never waits on the query worker, since the packet readers call it and nothing drains the channel at
// shutdown; a query that finds the channel full is dropped and counted. The worker is held on its first query so the
// channel fills.
func TestLighthouse_QueryServerDoesNotBlock(t *testing.T) {
l := test.NewLogger()
cs := testCertState(netip.MustParsePrefix("10.128.0.1/16"))
lh1 := "10.128.0.2"
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"}}
c.Settings["handshakes"] = map[string]any{"query_buffer": 1}
lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
require.NoError(t, err)
w := &heldEncWriter{held: make(chan struct{}), release: make(chan struct{})}
lh.ifce = w
t.Cleanup(func() { close(w.release) })
lh.QueryServer(netip.MustParseAddr("10.128.0.10"))
select {
case <-w.held:
case <-time.After(5 * time.Second):
t.Fatal("the query worker never took the first query")
}
dropped := lh.queryDropped.Count()
done := make(chan struct{})
go func() {
lh.QueryServer(netip.MustParseAddr("10.128.0.11"))
lh.QueryServer(netip.MustParseAddr("10.128.0.12"))
close(done)
}()
select {
case <-done:
case <-time.After(5 * time.Second):
t.Fatal("QueryServer blocked on a full query channel")
}
assert.EqualValues(t, 1, lh.queryDropped.Count()-dropped)
}
+35 -23
View File
@@ -113,6 +113,18 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
l.Info("Using multiple routines", "routines", routines)
}
// EXPERIMENTAL
// Intentionally not documented yet while we do more testing and determine
// a good default value.
conntrackCacheTimeout := c.GetDuration("firewall.conntrack.routine_cache_timeout", 0)
if routines > 1 && !c.IsSet("firewall.conntrack.routine_cache_timeout") {
// Use a different default if we are running with multiple routines
conntrackCacheTimeout = 1 * time.Second
}
if conntrackCacheTimeout > 0 {
l.Info("Using routine-local conntrack cache", "duration", conntrackCacheTimeout)
}
var tun overlay.Device
if !configTest {
c.CatchHUP(ctx)
@@ -262,28 +274,29 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
}
ifConfig := &InterfaceConfig{
HostMap: hostMap,
Inside: tun,
Outside: udpConns[0],
pki: pki,
Firewall: fw,
DnsServer: ds,
HandshakeManager: handshakeManager,
connectionManager: connManager,
lightHouse: lightHouse,
tryPromoteEvery: c.GetUint32("counters.try_promote", defaultPromoteEvery),
reQueryEvery: c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
reQueryWait: c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
DropLocalBroadcast: c.GetBool("tun.drop_local_broadcast", false),
DropMulticast: c.GetBool("tun.drop_multicast", false),
routines: routines,
MessageMetrics: messageMetrics,
version: buildVersion,
relayManager: NewRelayManager(ctx, l, hostMap, c),
punchy: punchy,
CpuAffinity: cpuAffinity,
PinThreads: pinThreads,
l: l,
HostMap: hostMap,
Inside: tun,
Outside: udpConns[0],
pki: pki,
Firewall: fw,
DnsServer: ds,
HandshakeManager: handshakeManager,
connectionManager: connManager,
lightHouse: lightHouse,
tryPromoteEvery: c.GetUint32("counters.try_promote", defaultPromoteEvery),
reQueryEvery: c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
reQueryWait: c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
DropLocalBroadcast: c.GetBool("tun.drop_local_broadcast", false),
DropMulticast: c.GetBool("tun.drop_multicast", false),
routines: routines,
MessageMetrics: messageMetrics,
version: buildVersion,
relayManager: NewRelayManager(ctx, l, hostMap, c),
punchy: punchy,
ConntrackCacheTimeout: conntrackCacheTimeout,
CpuAffinity: cpuAffinity,
PinThreads: pinThreads,
l: l,
}
var ifce *Interface
@@ -303,7 +316,6 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
handshakeManager.f = ifce
go handshakeManager.Run(ctx)
go ifce.firewall.Conntrack.Run(ctx)
punchy.Start(ctx, ifce, hostMap, lightHouse)
}
+1 -18
View File
@@ -477,7 +477,7 @@ func (f *Interface) handleOutsideMessagePacket(hostinfo *HostInfo, messageCounte
return
}
dropReason := f.allowInbound(hostinfo, rxc)
dropReason := f.firewall.Drop(rxc.fwPacket.Packet, true, hostinfo, f.pki.GetCAPool(), rxc.ctCache.Get())
if dropReason != nil {
f.rejectOutside(out, hostinfo.ConnectionState, hostinfo, rxc.nb, rxc.scratch, rxc.q)
if f.l.Enabled(context.Background(), slog.LevelDebug) {
@@ -492,23 +492,6 @@ func (f *Interface) handleOutsideMessagePacket(hostinfo *HostInfo, messageCounte
}
}
// allowInbound runs the inbound firewall check on rxc.fwPacket. It skips the check when the packet belongs to
// the same flow from the same host as the last one allowed in this batch.
func (f *Interface) allowInbound(hostinfo *HostInfo, rxc *rxContext) error {
fp := rxc.fwPacket.Packet
if hostinfo == rxc.lastAllowedHost && fp == rxc.lastAllowed {
return nil
}
if err := f.firewall.Drop(fp, true, hostinfo, f.pki.GetCAPool()); err != nil {
return err
}
rxc.lastAllowed = fp
rxc.lastAllowedHost = hostinfo
return nil
}
func (f *Interface) maybeSendRecvError(endpoint netip.AddrPort, index uint32) {
if f.sendRecvErrorConfig.ShouldRecvError(endpoint) {
f.sendRecvError(endpoint, index)
-50
View File
@@ -6,16 +6,12 @@ import (
"net"
"net/netip"
"testing"
"time"
"github.com/gaissmai/bart"
"github.com/google/gopacket"
"github.com/google/gopacket/layers"
"github.com/slackhq/nebula/cert"
"github.com/slackhq/nebula/iputil"
"github.com/slackhq/nebula/firewall"
"github.com/slackhq/nebula/test"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"golang.org/x/net/ipv4"
@@ -845,49 +841,3 @@ func Test_newPacket_parsedFields(t *testing.T) {
assert.True(t, p.Fragment)
assert.True(t, p.FragAny)
}
func TestInterface_allowInbound(t *testing.T) {
l := test.NewLoggerWithOutput(&bytes.Buffer{})
myNetworks := new(bart.Lite)
myNetworks.Insert(netip.MustParsePrefix("192.0.2.1/24"))
me := &dummyCert{name: "me", networks: []netip.Prefix{netip.MustParsePrefix("192.0.2.1/24")}}
peer := &cert.CachedCertificate{Certificate: &dummyCert{name: "peer", networks: []netip.Prefix{netip.MustParsePrefix("192.0.2.2/24")}}}
h := &HostInfo{ConnectionState: &ConnectionState{peerCert: peer}, vpnAddrs: []netip.Addr{netip.MustParseAddr("192.0.2.2")}}
h.buildNetworks(myNetworks, peer.Certificate)
allowAll := NewFirewall(l, time.Minute, time.Minute, time.Minute, me)
require.NoError(t, allowAll.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", "", "", "", ""))
allowNone := NewFirewall(l, time.Minute, time.Minute, time.Minute, me)
f := &Interface{firewall: allowAll, pki: &PKI{}, l: l}
rxc := &rxContext{fwPacket: &firewall.ParsedPacket{}, hostmapCache: map[uint32]*HostInfo{}}
flow := firewall.Packet{
LocalAddr: netip.MustParseAddr("192.0.2.1"),
RemoteAddr: netip.MustParseAddr("192.0.2.2"),
LocalPort: 443,
RemotePort: 55000,
Protocol: iputil.IPProtocolUDP,
}
rxc.fwPacket.Packet = flow
require.NoError(t, f.allowInbound(h, rxc))
// With a firewall that allows nothing, the rest of the flow's run in this batch still passes
f.firewall = allowNone
require.NoError(t, f.allowInbound(h, rxc))
// A different flow is checked
other := flow
other.RemotePort++
rxc.fwPacket.Packet = other
require.ErrorIs(t, f.allowInbound(h, rxc), ErrNoMatchingRule)
// So is the same flow from a different host
rxc.fwPacket.Packet = flow
h2 := &HostInfo{ConnectionState: h.ConnectionState, vpnAddrs: h.vpnAddrs}
h2.buildNetworks(myNetworks, peer.Certificate)
require.ErrorIs(t, f.allowInbound(h2, rxc), ErrNoMatchingRule)
// And the same flow in the next batch
rxc.endBatch()
require.ErrorIs(t, f.allowInbound(h, rxc), ErrNoMatchingRule)
}
+124 -1
View File
@@ -180,11 +180,121 @@ func NewUDPStatsEmitter(udpConns []Conn) func() {
return func() {}
}
// recvBatch is how many datagrams one recvmsg_x call may return.
const recvBatch = 64
// msghdrX mirrors xnu's struct msghdr_x (bsd/sys/socket_private.h). recvmsg_x reports each
// datagram's length in Datalen rather than in its return value.
type msghdrX struct {
Name *byte
Namelen uint32
Iov *unix.Iovec
Iovlen int32
Control *byte
Controllen uint32
Flags int32
Datalen uint64
}
// ListenOut drains the socket with recvmsg_x, darwin's private batched recvmsg, so one syscall reads
// up to recvBatch datagrams instead of one. Sends stay on sendto: xnu batches sendmsg_x through
// the stack only on connected sockets, and on this unconnected one it loops sendit per message.
func (u *StdConn) ListenOut(r EncReader, flush func()) error {
rc, err := u.UDPConn.SyscallConn()
if err != nil {
return err
}
bufs := make([][]byte, recvBatch)
names := make([]unix.RawSockaddrInet6, recvBatch)
iovs := make([]unix.Iovec, recvBatch)
hdrs := make([]msghdrX, recvBatch)
for i := range hdrs {
bufs[i] = make([]byte, MTU)
iovs[i].Base = &bufs[i][0]
iovs[i].SetLen(MTU)
hdrs[i].Iov = &iovs[i]
hdrs[i].Iovlen = 1
hdrs[i].Name = (*byte)(unsafe.Pointer(&names[i]))
}
for {
var n int
var errno syscall.Errno
err := rc.Read(func(fd uintptr) bool {
for i := range hdrs {
hdrs[i].Namelen = unix.SizeofSockaddrInet6
hdrs[i].Flags = 0
hdrs[i].Datalen = 0
}
r0, _, e := unix.Syscall6(unix.SYS_RECVMSG_X, fd, uintptr(unsafe.Pointer(&hdrs[0])), recvBatch, 0, 0, 0)
if e == unix.EAGAIN {
return false
}
n, errno = int(r0), e
return true
})
if err != nil {
if errors.Is(err, net.ErrClosed) {
return err
}
u.l.Error("unexpected udp socket receive error", "error", err)
continue
}
if errno == unix.ENOSYS || errno == unix.EPERM || errno == unix.EOPNOTSUPP {
// recvmsg_x is private, so a kernel or sandbox (such as an iOS extension's) may refuse it.
u.l.Warn("recvmsg_x unavailable, reading one datagram per syscall", "error", errno)
return u.listenOutSingle(r, flush)
}
if errno != 0 {
if errno != unix.EINTR {
u.l.Error("unexpected udp socket receive error", "error", errno)
}
continue
}
if err := checkMsghdrX(hdrs, names, n); err != nil {
// The batch was read through a layout the kernel no longer writes, so none of it is trustworthy.
u.l.Warn("recvmsg_x returned an unexpected header, reading one datagram per syscall", "error", err)
return u.listenOutSingle(r, flush)
}
for i := 0; i < n; i++ {
addr, ok := sockaddrToAddrPort(&names[i])
if !ok {
continue
}
l := int(hdrs[i].Datalen)
r(addr, bufs[i][:l:l])
}
flush()
}
}
// checkMsghdrX rejects a recvmsg_x result that intact msghdr_x entries can't produce, which is how a change
// to xnu's private struct layout would show up.
func checkMsghdrX(hdrs []msghdrX, names []unix.RawSockaddrInet6, n int) error {
if n > len(hdrs) {
return fmt.Errorf("returned %d datagrams for %d headers", n, len(hdrs))
}
for i := range n {
if hdrs[i].Datalen > MTU {
return fmt.Errorf("datagram %d is %d bytes, past the %d byte buffer", i, hdrs[i].Datalen, MTU)
}
switch {
case names[i].Family == unix.AF_INET && hdrs[i].Namelen == unix.SizeofSockaddrInet4:
case names[i].Family == unix.AF_INET6 && hdrs[i].Namelen == unix.SizeofSockaddrInet6:
default:
return fmt.Errorf("datagram %d has address family %d with length %d", i, names[i].Family, hdrs[i].Namelen)
}
}
return nil
}
// listenOutSingle reads one datagram per recvfrom and flushes after each, for when recvmsg_x is refused or misbehaves.
func (u *StdConn) listenOutSingle(r EncReader, flush func()) error {
buffer := make([]byte, MTU)
for {
// Just read one packet at a time
n, rua, err := u.ReadFromUDPAddrPort(buffer)
if err != nil {
if errors.Is(err, net.ErrClosed) {
@@ -199,6 +309,19 @@ func (u *StdConn) ListenOut(r EncReader, flush func()) error {
}
}
func sockaddrToAddrPort(sa *unix.RawSockaddrInet6) (netip.AddrPort, bool) {
switch sa.Family {
case unix.AF_INET:
sa4 := (*unix.RawSockaddrInet4)(unsafe.Pointer(sa))
port := binary.BigEndian.Uint16((*[2]byte)(unsafe.Pointer(&sa4.Port))[:])
return netip.AddrPortFrom(netip.AddrFrom4(sa4.Addr), port), true
case unix.AF_INET6:
port := binary.BigEndian.Uint16((*[2]byte)(unsafe.Pointer(&sa.Port))[:])
return netip.AddrPortFrom(netip.AddrFrom16(sa.Addr).Unmap(), port), true
}
return netip.AddrPort{}, false
}
func (u *StdConn) SupportsMultipleReaders() bool {
return false
}
+85
View File
@@ -0,0 +1,85 @@
//go:build !e2e_testing
package udp
import (
"log/slog"
"net"
"net/netip"
"sync"
"sync/atomic"
"testing"
)
// BenchmarkListenOut measures the receive path on loopback: senders flood datagrams sized like a full tunnel
// packet, and each op is one datagram delivered to the reader. There are enough senders that the reader, not the
// senders, sets the pace. flushes/pkt is 1 when every receive is its own syscall and falls as receives batch. The
// senders share the machine, so ns/op is only comparable between runs on the same host.
func BenchmarkListenOut(b *testing.B) {
for _, tc := range []struct{ name, listen string }{
{"v4", "127.0.0.1"},
{"v6", "::1"},
} {
b.Run(tc.name, func(b *testing.B) {
c, err := NewListener(slog.New(slog.DiscardHandler), Settings{Listen: netip.AddrPortFrom(netip.MustParseAddr(tc.listen), 0)})
if err != nil {
b.Fatal(err)
}
la, err := c.LocalAddr()
if err != nil {
b.Fatal(err)
}
const size, senders = 1300, 4
txs := make([]net.Conn, senders)
for i := range txs {
if txs[i], err = net.Dial("udp", la.String()); err != nil {
b.Fatal(err)
}
defer txs[i].Close()
}
target := int64(b.N)
var got, flushes atomic.Int64
reached := make(chan struct{})
done := make(chan struct{})
go func() {
_ = c.ListenOut(func(netip.AddrPort, []byte) {
if got.Add(1) == target {
close(reached)
}
}, func() { flushes.Add(1) })
close(done)
}()
// Writes fail with ENOBUFS while the send queue is full; the flood only has to keep the receiver busy.
stop := make(chan struct{})
var sent sync.WaitGroup
for _, tx := range txs {
sent.Go(func() {
payload := make([]byte, size)
for {
select {
case <-stop:
return
default:
_, _ = tx.Write(payload)
}
}
})
}
b.SetBytes(size)
b.ResetTimer()
<-reached
b.StopTimer()
close(stop)
sent.Wait()
if err := c.Close(); err != nil {
b.Fatal(err)
}
<-done
b.ReportMetric(float64(flushes.Load())/float64(got.Load()), "flushes/pkt")
})
}
}
+302
View File
@@ -0,0 +1,302 @@
//go:build !e2e_testing
package udp
import (
"bytes"
"fmt"
"log/slog"
"net"
"net/netip"
"sync"
"syscall"
"testing"
"time"
"unsafe"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"golang.org/x/sys/unix"
)
// A burst larger than recvBatch arrives whole, with each datagram's own bytes and source address, and
// flush runs once per recvmsg_x batch rather than once per datagram.
func TestListenOutBatchesBurst(t *testing.T) {
for _, tc := range []struct{ name, listen string }{
{"v4", "127.0.0.1"},
{"v6", "::1"},
} {
t.Run(tc.name, func(t *testing.T) {
c, err := NewListener(slog.New(slog.DiscardHandler), Settings{Listen: netip.AddrPortFrom(netip.MustParseAddr(tc.listen), 0)})
require.NoError(t, err)
la, err := c.LocalAddr()
require.NoError(t, err)
tx, err := net.Dial("udp", la.String())
require.NoError(t, err)
defer tx.Close()
txAddr := netip.MustParseAddrPort(tx.LocalAddr().String())
const total = recvBatch*3 + 7
var mu sync.Mutex
var got []string
var froms []netip.AddrPort
flushes := 0
// Queued before the reader starts, so recvmsg_x sees full batches.
for i := 0; i < total; i++ {
_, err := tx.Write([]byte(fmt.Sprintf("pkt-%03d", i)))
require.NoError(t, err)
}
done := make(chan struct{})
go func() {
_ = c.ListenOut(func(addr netip.AddrPort, payload []byte) {
mu.Lock()
got = append(got, string(payload))
froms = append(froms, addr)
mu.Unlock()
}, func() {
mu.Lock()
flushes++
mu.Unlock()
})
close(done)
}()
require.Eventually(t, func() bool {
mu.Lock()
defer mu.Unlock()
return len(got) == total
}, 5*time.Second, 10*time.Millisecond)
require.NoError(t, c.Close())
<-done
for i, p := range got {
assert.Equal(t, fmt.Sprintf("pkt-%03d", i), p)
assert.Equal(t, txAddr, froms[i])
}
assert.Equal(t, (total+recvBatch-1)/recvBatch, flushes)
})
}
}
// msghdrX round-trips through xnu's recvmsg_x on a loopback socket, v4 and v6. The kernel reads Name, Namelen,
// Iov and Iovlen from each entry and writes Namelen, Flags and Datalen back, so a field at the wrong offset or a
// wrong stride between entries shows up as a wrong length, sender, or payload.
// TestListenOutSingleFallback pins the path ListenOut takes when the kernel refuses recvmsg_x.
func TestListenOutSingleFallback(t *testing.T) {
c, err := NewListener(slog.New(slog.DiscardHandler), Settings{Listen: netip.MustParseAddrPort("127.0.0.1:0")})
require.NoError(t, err)
la, err := c.LocalAddr()
require.NoError(t, err)
tx, err := net.Dial("udp", la.String())
require.NoError(t, err)
defer tx.Close()
txAddr := netip.MustParseAddrPort(tx.LocalAddr().String())
const total = 5
for i := 0; i < total; i++ {
_, err := tx.Write([]byte(fmt.Sprintf("pkt-%03d", i)))
require.NoError(t, err)
}
var mu sync.Mutex
var got []string
flushes := 0
done := make(chan struct{})
go func() {
_ = c.(*StdConn).listenOutSingle(func(addr netip.AddrPort, payload []byte) {
assert.Equal(t, txAddr, addr)
mu.Lock()
got = append(got, string(payload))
mu.Unlock()
}, func() {
mu.Lock()
flushes++
mu.Unlock()
})
close(done)
}()
require.Eventually(t, func() bool {
mu.Lock()
defer mu.Unlock()
return len(got) == total
}, 5*time.Second, 10*time.Millisecond)
require.NoError(t, c.Close())
<-done
for i, p := range got {
assert.Equal(t, fmt.Sprintf("pkt-%03d", i), p)
}
assert.Equal(t, total, flushes)
}
func TestCheckMsghdrX(t *testing.T) {
valid := func() ([]msghdrX, []unix.RawSockaddrInet6) {
hdrs := make([]msghdrX, 2)
names := make([]unix.RawSockaddrInet6, 2)
hdrs[0].Namelen, hdrs[0].Datalen = unix.SizeofSockaddrInet4, 1
names[0].Family = unix.AF_INET
hdrs[1].Namelen, hdrs[1].Datalen = unix.SizeofSockaddrInet6, MTU
names[1].Family = unix.AF_INET6
return hdrs, names
}
hdrs, names := valid()
require.NoError(t, checkMsghdrX(hdrs, names, 2))
require.NoError(t, checkMsghdrX(hdrs, names, 0))
for _, tc := range []struct {
name string
n int
mutate func([]msghdrX, []unix.RawSockaddrInet6)
}{
{"count past headers", 3, func([]msghdrX, []unix.RawSockaddrInet6) {}},
{"datalen past buffer", 2, func(h []msghdrX, _ []unix.RawSockaddrInet6) { h[1].Datalen = MTU + 1 }},
{"unknown family", 2, func(_ []msghdrX, a []unix.RawSockaddrInet6) { a[0].Family = unix.AF_UNIX }},
{"family and length disagree", 2, func(h []msghdrX, _ []unix.RawSockaddrInet6) { h[0].Namelen = unix.SizeofSockaddrInet6 }},
} {
t.Run(tc.name, func(t *testing.T) {
hdrs, names := valid()
tc.mutate(hdrs, names)
assert.Error(t, checkMsghdrX(hdrs, names, tc.n))
})
}
}
func TestMsghdrXKernelRoundTrip(t *testing.T) {
for _, v6 := range []bool{false, true} {
t.Run(fmt.Sprintf("v6=%v", v6), func(t *testing.T) {
// Loopback delivery goes through the dlil input thread, so the settle gives all three time to
// queue and come back from one call, which is what exercises the stride.
got := recvmsgXRoundTrip(t, v6, []int{1, 300, 1400}, 4, 50*time.Millisecond)
assert.Equal(t, 3, got, "largest batch from one recvmsg_x call")
})
}
}
// Any burst, of any datagram lengths up to the kernel's maxdgram, into any number of headers, comes back in order
// with every entry's fields intact, and headers past the returned count untouched.
func FuzzMsghdrXKernelRoundTrip(f *testing.F) {
f.Add(false, uint8(3), []byte{0, 1, 1, 44, 5, 120})
f.Add(true, uint8(0), []byte{0, 0, 0x23, 0x29}) // one empty datagram, then one at MTU
f.Add(false, uint8(1), []byte{0x23, 0x2a, 0x24, 0x00, 0}) // past MTU: cut to MTU, no MSG_TRUNC
f.Add(true, uint8(recvBatch-1), bytes.Repeat([]byte{0x05, 0xdc}, recvBatch))
f.Fuzz(func(t *testing.T, v6 bool, spare uint8, sizes []byte) {
// Two bytes per datagram, big-endian, folded into [0, maxdgram].
var lens []int
for i := 0; i+1 < len(sizes) && len(lens) < recvBatch; i += 2 {
lens = append(lens, (int(sizes[i])<<8|int(sizes[i+1]))%(maxDgram+1))
}
if len(lens) == 0 {
return
}
recvmsgXRoundTrip(t, v6, lens, 1+int(spare)%recvBatch, time.Millisecond)
})
}
// maxDgram is darwin's default net.inet.udp.maxdgram, the largest datagram a loopback sender may write.
const maxDgram = 9216
// recvmsgXRoundTrip sends one datagram per entry of lens to a fresh loopback socket, drains them with recvmsg_x
// into nh headers, and checks every field the kernel writes. Neighboring datagrams differ in sender, length and
// fill byte, so no field can pass by reading a neighbor's value. It returns the largest batch one call returned.
func recvmsgXRoundTrip(t *testing.T, v6 bool, lens []int, nh int, settle time.Duration) int {
t.Helper()
network, lo := "udp4", netip.MustParseAddr("127.0.0.1")
if v6 {
network, lo = "udp6", netip.IPv6Loopback()
}
rx, err := net.ListenUDP(network, net.UDPAddrFromAddrPort(netip.AddrPortFrom(lo, 0)))
require.NoError(t, err)
defer rx.Close()
txs := make([]*net.UDPConn, 3)
for i := range txs {
txs[i], err = net.DialUDP(network, nil, rx.LocalAddr().(*net.UDPAddr))
require.NoError(t, err)
defer txs[i].Close()
}
senders := make([]netip.AddrPort, len(lens))
for i, l := range lens {
tx := txs[i%len(txs)]
senders[i] = tx.LocalAddr().(*net.UDPAddr).AddrPort()
_, err := tx.Write(bytes.Repeat([]byte{byte(i + 1)}, l))
require.NoError(t, err)
}
time.Sleep(settle)
// Namelen goes in as the v6 size on both families, so the v4 size coming back proves the kernel wrote it.
const sentinel = 0xa5a5a5a5a5a5a5a5
wantNamelen := uint32(unix.SizeofSockaddrInet4)
if v6 {
wantNamelen = unix.SizeofSockaddrInet6
}
bufs := make([][]byte, nh)
names := make([]unix.RawSockaddrInet6, nh)
iovs := make([]unix.Iovec, nh)
hdrs := make([]msghdrX, nh)
for i := range hdrs {
bufs[i] = make([]byte, MTU)
iovs[i].Base = &bufs[i][0]
iovs[i].SetLen(MTU)
}
// A datagram dropped on loopback would otherwise block rc.Read forever.
require.NoError(t, rx.SetReadDeadline(time.Now().Add(5*time.Second)))
rc, err := rx.SyscallConn()
require.NoError(t, err)
largest := 0
for next := 0; next < len(lens); {
for i := range hdrs {
clear(bufs[i])
hdrs[i] = msghdrX{
Name: (*byte)(unsafe.Pointer(&names[i])),
Namelen: unix.SizeofSockaddrInet6,
Iov: &iovs[i],
Iovlen: 1,
Flags: -1,
Datalen: sentinel,
}
}
var n int
var errno syscall.Errno
require.NoError(t, rc.Read(func(fd uintptr) bool {
r0, _, e := unix.Syscall6(unix.SYS_RECVMSG_X, fd, uintptr(unsafe.Pointer(&hdrs[0])), uintptr(nh), 0, 0, 0)
n, errno = int(r0), e
return e != unix.EAGAIN
}), "received %d of %d datagrams", next, len(lens))
require.Zero(t, errno)
require.Positive(t, n)
require.LessOrEqual(t, n, min(nh, len(lens)-next))
largest = max(largest, n)
require.NoError(t, checkMsghdrX(hdrs[:nh], names[:nh], n), "ListenOut would fall back on this batch")
for i := range n {
d := next + i
h := &hdrs[i]
want := min(lens[d], MTU)
require.EqualValues(t, want, h.Datalen, "datagram %d Datalen", d)
require.Equal(t, wantNamelen, h.Namelen, "datagram %d Namelen", d)
from, ok := sockaddrToAddrPort(&names[i])
require.True(t, ok, "datagram %d family %d", d, names[i].Family)
require.Equal(t, senders[d], from, "datagram %d sender", d)
// Flags goes in as -1 and must come back 0. xnu's list receive (soreceive_m_list) never sets
// MSG_TRUNC, so a datagram past MTU arrives silently cut to MTU, as recvfrom through net.UDPConn does.
require.Zero(t, h.Flags, "datagram %d Flags", d)
require.Equal(t, bytes.Repeat([]byte{byte(d + 1)}, want), bufs[i][:want], "datagram %d payload", d)
if want < MTU {
require.Zero(t, bufs[i][want], "datagram %d wrote past Datalen", d)
}
}
for i := n; i < nh; i++ {
require.EqualValues(t, uint64(sentinel), hdrs[i].Datalen, "spare header %d Datalen", i)
require.EqualValues(t, unix.SizeofSockaddrInet6, hdrs[i].Namelen, "spare header %d Namelen", i)
}
next += n
}
return largest
}