tailscale/util/linuxfw/nftables.go
Maisem Ali c3a8e63100 util/linuxfw: add additional nftable detection logic
We were previously using the netlink API to see if there are chains/rules that
already exist. This works fine in environments where there is either full
nftable support or no support at all. However, we have identified certain
environments which have partial nftable support and the only feasible way of
detecting such an environment is to try to create some of the chains that we
need.

This adds a check to create a dummy postrouting chain which is immediately
deleted. The goal of the check is to ensure we are able to use nftables and
that it won't error out later. This check is only done in the path where we
detected that the system has no preexisting nftable rules.

Updates #5621
Updates #8555
Updates #8762

Signed-off-by: Maisem Ali <maisem@tailscale.com>
2023-10-18 13:39:55 -07:00

293 lines
7.1 KiB
Go

// Copyright (c) Tailscale Inc & AUTHORS
// SPDX-License-Identifier: BSD-3-Clause
// TODO(#8502): add support for more architectures
//go:build linux && (arm64 || amd64)
package linuxfw
import (
"fmt"
"sort"
"strings"
"github.com/google/nftables"
"github.com/google/nftables/expr"
"github.com/google/nftables/xt"
"github.com/josharian/native"
"golang.org/x/sys/unix"
"tailscale.com/types/logger"
"tailscale.com/util/cmpx"
)
// DebugNetfilter prints debug information about netfilter rules to the
// provided log function.
func DebugNetfilter(logf logger.Logf) error {
conn, err := nftables.New()
if err != nil {
return err
}
chains, err := conn.ListChains()
if err != nil {
return fmt.Errorf("cannot list chains: %w", err)
}
if len(chains) == 0 {
logf("netfilter: no chains")
return nil
}
for _, chain := range chains {
logf("netfilter: table=%s chain=%s", chain.Table.Name, chain.Name)
rules, err := conn.GetRules(chain.Table, chain)
if err != nil {
continue
}
sort.Slice(rules, func(i, j int) bool {
return rules[i].Position < rules[j].Position
})
for i, rule := range rules {
logf("netfilter: rule[%d]: pos=%d flags=%d", i, rule.Position, rule.Flags)
for _, ex := range rule.Exprs {
switch v := ex.(type) {
case *expr.Meta:
key := cmpx.Or(metaKeyNames[v.Key], "UNKNOWN")
logf("netfilter: Meta: key=%s source_register=%v register=%d", key, v.SourceRegister, v.Register)
case *expr.Cmp:
op := cmpx.Or(cmpOpNames[v.Op], "UNKNOWN")
logf("netfilter: Cmp: op=%s register=%d data=%s", op, v.Register, formatMaybePrintable(v.Data))
case *expr.Counter:
// don't print
case *expr.Verdict:
kind := cmpx.Or(verdictNames[v.Kind], "UNKNOWN")
logf("netfilter: Verdict: kind=%s data=%s", kind, v.Chain)
case *expr.Target:
logf("netfilter: Target: name=%s info=%s", v.Name, printTargetInfo(v.Name, v.Info))
case *expr.Match:
logf("netfilter: Match: name=%s info=%+v", v.Name, printMatchInfo(v.Name, v.Info))
case *expr.Payload:
logf("netfilter: Payload: op=%s src=%d dst=%d base=%s offset=%d len=%d",
payloadOperationTypeNames[v.OperationType],
v.SourceRegister, v.DestRegister,
payloadBaseNames[v.Base],
v.Offset, v.Len)
// TODO(andrew): csum
case *expr.Bitwise:
var xor string
for _, b := range v.Xor {
if b != 0 {
xor = fmt.Sprintf(" xor=%v", v.Xor)
break
}
}
logf("netfilter: Bitwise: src=%d dst=%d len=%d mask=%v%s",
v.SourceRegister, v.DestRegister, v.Len, v.Mask, xor)
default:
logf("netfilter: unknown %T: %+v", v, v)
}
}
}
}
return nil
}
// detectNetfilter returns the number of nftables rules present in the system.
func detectNetfilter() (int, error) {
// Frist try creating a dummy postrouting chain. Emperically, we have
// noticed that on some devices there is partial nftables support and the
// kernel rejects some chains that are valid on other devices. This is a
// workaround to detect that case.
//
// This specifically allows us to run in on GKE nodes using COS images which
// have partial nftables support (as of 2023-10-18). When we try to create a
// dummy postrouting chain, we get an error like:
// add chain: conn.Receive: netlink receive: no such file or directory
nft, err := newNfTablesRunner(logger.Discard)
if err != nil {
return 0, FWModeNotSupportedError{
Mode: FirewallModeNfTables,
Err: fmt.Errorf("cannot create nftables runner: %w", err),
}
}
if err := nft.createDummyPostroutingChains(); err != nil {
return 0, FWModeNotSupportedError{
Mode: FirewallModeNfTables,
Err: err,
}
}
conn, err := nftables.New()
if err != nil {
return 0, FWModeNotSupportedError{
Mode: FirewallModeNfTables,
Err: err,
}
}
chains, err := conn.ListChains()
if err != nil {
return 0, FWModeNotSupportedError{
Mode: FirewallModeNfTables,
Err: fmt.Errorf("cannot list chains: %w", err),
}
}
var validRules int
for _, chain := range chains {
rules, err := conn.GetRules(chain.Table, chain)
if err != nil {
continue
}
validRules += len(rules)
}
return validRules, nil
}
func printMatchInfo(name string, info xt.InfoAny) string {
var sb strings.Builder
sb.WriteString(`{`)
var handled bool = true
switch v := info.(type) {
// TODO(andrew): we should support these common types
//case *xt.ConntrackMtinfo3:
//case *xt.ConntrackMtinfo2:
case *xt.Tcp:
fmt.Fprintf(&sb, "Src:%s Dst:%s", formatPortRange(v.SrcPorts), formatPortRange(v.DstPorts))
if v.Option != 0 {
fmt.Fprintf(&sb, " Option:%d", v.Option)
}
if v.FlagsMask != 0 {
fmt.Fprintf(&sb, " FlagsMask:%d", v.FlagsMask)
}
if v.FlagsCmp != 0 {
fmt.Fprintf(&sb, " FlagsCmp:%d", v.FlagsCmp)
}
if v.InvFlags != 0 {
fmt.Fprintf(&sb, " InvFlags:%d", v.InvFlags)
}
case *xt.Udp:
fmt.Fprintf(&sb, "Src:%s Dst:%s", formatPortRange(v.SrcPorts), formatPortRange(v.DstPorts))
if v.InvFlags != 0 {
fmt.Fprintf(&sb, " InvFlags:%d", v.InvFlags)
}
case *xt.AddrType:
var sprefix, dprefix string
if v.InvertSource {
sprefix = "!"
}
if v.InvertDest {
dprefix = "!"
}
// TODO(andrew): translate source/dest
fmt.Fprintf(&sb, "Source:%s%d Dest:%s%d", sprefix, v.Source, dprefix, v.Dest)
case *xt.AddrTypeV1:
// TODO(andrew): translate source/dest
fmt.Fprintf(&sb, "Source:%d Dest:%d", v.Source, v.Dest)
var flags []string
for flag, name := range addrTypeFlagNames {
if v.Flags&flag != 0 {
flags = append(flags, name)
}
}
if len(flags) > 0 {
sort.Strings(flags)
fmt.Fprintf(&sb, "Flags:%s", strings.Join(flags, ","))
}
default:
handled = false
}
if handled {
sb.WriteString(`}`)
return sb.String()
}
unknown, ok := info.(*xt.Unknown)
if !ok {
return fmt.Sprintf("(%T)%+v", info, info)
}
data := []byte(*unknown)
// Things where upstream has no type
handled = true
switch name {
case "pkttype":
if len(data) != 8 {
handled = false
break
}
pkttype := int(native.Endian.Uint32(data[0:4]))
invert := int(native.Endian.Uint32(data[4:8]))
var invertPrefix string
if invert != 0 {
invertPrefix = "!"
}
pkttypeName := packetTypeNames[pkttype]
if pkttypeName != "" {
fmt.Fprintf(&sb, "PktType:%s%s", invertPrefix, pkttypeName)
} else {
fmt.Fprintf(&sb, "PktType:%s%d", invertPrefix, pkttype)
}
default:
handled = true
}
if !handled {
return fmt.Sprintf("(%T)%+v", info, info)
}
sb.WriteString(`}`)
return sb.String()
}
func printTargetInfo(name string, info xt.InfoAny) string {
var sb strings.Builder
sb.WriteString(`{`)
unknown, ok := info.(*xt.Unknown)
if !ok {
return fmt.Sprintf("(%T)%+v", info, info)
}
data := []byte(*unknown)
// Things where upstream has no type
switch name {
case "LOG":
if len(data) != 32 {
fmt.Fprintf(&sb, `Error:"bad size; want 32, got %d"`, len(data))
break
}
level := data[0]
logflags := data[1]
prefix := unix.ByteSliceToString(data[2:])
fmt.Fprintf(&sb, "Level:%d LogFlags:%d Prefix:%q", level, logflags, prefix)
default:
return fmt.Sprintf("(%T)%+v", info, info)
}
sb.WriteString(`}`)
return sb.String()
}