You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
tailscale/net/tstun/wrap_test.go

577 lines
14 KiB
Go

// Copyright (c) 2020 Tailscale Inc & AUTHORS All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package tstun
import (
"bytes"
"context"
"encoding/binary"
"fmt"
"net/netip"
"strconv"
"strings"
"testing"
"unsafe"
"github.com/google/go-cmp/cmp"
"github.com/google/go-cmp/cmp/cmpopts"
"github.com/tailscale/wireguard-go/tun/tuntest"
"go4.org/mem"
"go4.org/netipx"
"tailscale.com/disco"
"tailscale.com/net/connstats"
"tailscale.com/net/netaddr"
"tailscale.com/net/packet"
"tailscale.com/tstest"
"tailscale.com/tstime/mono"
"tailscale.com/types/ipproto"
"tailscale.com/types/key"
"tailscale.com/types/logger"
"tailscale.com/types/netlogtype"
"tailscale.com/wgengine/filter"
)
net/packet: remove the custom IP4/IP6 types in favor of netaddr.IP. Upstream netaddr has a change that makes it alloc-free, so it's safe to use in hot codepaths. This gets rid of one of the many IP types in our codebase. Performance is currently worse across the board. This is likely due in part to netaddr.IP being a larger value type (4b -> 24b for IPv4, 16b -> 24b for IPv6), and in other part due to missing low-hanging fruit optimizations in netaddr. However, the regression is less bad than it looks at first glance, because we'd micro-optimized packet.IP* in the past few weeks. This change drops us back to roughly where we were at the 1.2 release, but with the benefit of a significant code and architectural simplification. name old time/op new time/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 12.2ns ± 5% 29.7ns ± 2% +142.32% (p=0.008 n=5+5) Decode/tcp6-8 12.6ns ± 3% 65.1ns ± 2% +418.47% (p=0.008 n=5+5) Decode/udp4-8 11.8ns ± 3% 30.5ns ± 2% +157.94% (p=0.008 n=5+5) Decode/udp6-8 27.1ns ± 1% 65.7ns ± 2% +142.36% (p=0.016 n=4+5) Decode/icmp4-8 24.6ns ± 2% 30.5ns ± 2% +23.65% (p=0.016 n=4+5) Decode/icmp6-8 22.9ns ±51% 65.5ns ± 2% +186.19% (p=0.008 n=5+5) Decode/igmp-8 18.1ns ±44% 30.2ns ± 1% +66.89% (p=0.008 n=5+5) Decode/unknown-8 20.8ns ± 1% 10.6ns ± 9% -49.11% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 30.5ns ± 1% 77.9ns ± 3% +155.01% (p=0.008 n=5+5) Filter/tcp4_syn_in-8 43.7ns ± 3% 123.0ns ± 3% +181.72% (p=0.008 n=5+5) Filter/tcp4_syn_out-8 24.5ns ± 2% 45.7ns ± 6% +86.22% (p=0.008 n=5+5) Filter/udp4_in-8 64.8ns ± 1% 210.0ns ± 2% +223.87% (p=0.008 n=5+5) Filter/udp4_out-8 119ns ± 0% 278ns ± 0% +133.78% (p=0.016 n=4+5) Filter/icmp6-8 40.3ns ± 2% 204.4ns ± 4% +407.70% (p=0.008 n=5+5) Filter/tcp6_syn_in-8 35.3ns ± 3% 199.2ns ± 2% +464.95% (p=0.008 n=5+5) Filter/tcp6_syn_out-8 32.8ns ± 2% 81.0ns ± 2% +147.10% (p=0.008 n=5+5) Filter/udp6_in-8 106ns ± 2% 290ns ± 2% +174.48% (p=0.008 n=5+5) Filter/udp6_out-8 184ns ± 2% 314ns ± 3% +70.43% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/tstun goos:linux goarch:amd64 Write-8 9.02ns ± 3% 8.92ns ± 1% ~ (p=0.421 n=5+5) name old alloc/op new alloc/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00B 0.00B ~ (all equal) Decode/tcp6-8 0.00B 0.00B ~ (all equal) Decode/udp4-8 0.00B 0.00B ~ (all equal) Decode/udp6-8 0.00B 0.00B ~ (all equal) Decode/icmp4-8 0.00B 0.00B ~ (all equal) Decode/icmp6-8 0.00B 0.00B ~ (all equal) Decode/igmp-8 0.00B 0.00B ~ (all equal) Decode/unknown-8 0.00B 0.00B ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp4_in-8 0.00B 0.00B ~ (all equal) Filter/udp4_out-8 16.0B ± 0% 64.0B ± 0% +300.00% (p=0.008 n=5+5) Filter/icmp6-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp6_in-8 0.00B 0.00B ~ (all equal) Filter/udp6_out-8 48.0B ± 0% 64.0B ± 0% +33.33% (p=0.008 n=5+5) name old allocs/op new allocs/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00 0.00 ~ (all equal) Decode/tcp6-8 0.00 0.00 ~ (all equal) Decode/udp4-8 0.00 0.00 ~ (all equal) Decode/udp6-8 0.00 0.00 ~ (all equal) Decode/icmp4-8 0.00 0.00 ~ (all equal) Decode/icmp6-8 0.00 0.00 ~ (all equal) Decode/igmp-8 0.00 0.00 ~ (all equal) Decode/unknown-8 0.00 0.00 ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp4_in-8 0.00 0.00 ~ (all equal) Filter/udp4_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Filter/icmp6-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp6_in-8 0.00 0.00 ~ (all equal) Filter/udp6_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
func udp4(src, dst string, sport, dport uint16) []byte {
sip, err := netip.ParseAddr(src)
net/packet: remove the custom IP4/IP6 types in favor of netaddr.IP. Upstream netaddr has a change that makes it alloc-free, so it's safe to use in hot codepaths. This gets rid of one of the many IP types in our codebase. Performance is currently worse across the board. This is likely due in part to netaddr.IP being a larger value type (4b -> 24b for IPv4, 16b -> 24b for IPv6), and in other part due to missing low-hanging fruit optimizations in netaddr. However, the regression is less bad than it looks at first glance, because we'd micro-optimized packet.IP* in the past few weeks. This change drops us back to roughly where we were at the 1.2 release, but with the benefit of a significant code and architectural simplification. name old time/op new time/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 12.2ns ± 5% 29.7ns ± 2% +142.32% (p=0.008 n=5+5) Decode/tcp6-8 12.6ns ± 3% 65.1ns ± 2% +418.47% (p=0.008 n=5+5) Decode/udp4-8 11.8ns ± 3% 30.5ns ± 2% +157.94% (p=0.008 n=5+5) Decode/udp6-8 27.1ns ± 1% 65.7ns ± 2% +142.36% (p=0.016 n=4+5) Decode/icmp4-8 24.6ns ± 2% 30.5ns ± 2% +23.65% (p=0.016 n=4+5) Decode/icmp6-8 22.9ns ±51% 65.5ns ± 2% +186.19% (p=0.008 n=5+5) Decode/igmp-8 18.1ns ±44% 30.2ns ± 1% +66.89% (p=0.008 n=5+5) Decode/unknown-8 20.8ns ± 1% 10.6ns ± 9% -49.11% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 30.5ns ± 1% 77.9ns ± 3% +155.01% (p=0.008 n=5+5) Filter/tcp4_syn_in-8 43.7ns ± 3% 123.0ns ± 3% +181.72% (p=0.008 n=5+5) Filter/tcp4_syn_out-8 24.5ns ± 2% 45.7ns ± 6% +86.22% (p=0.008 n=5+5) Filter/udp4_in-8 64.8ns ± 1% 210.0ns ± 2% +223.87% (p=0.008 n=5+5) Filter/udp4_out-8 119ns ± 0% 278ns ± 0% +133.78% (p=0.016 n=4+5) Filter/icmp6-8 40.3ns ± 2% 204.4ns ± 4% +407.70% (p=0.008 n=5+5) Filter/tcp6_syn_in-8 35.3ns ± 3% 199.2ns ± 2% +464.95% (p=0.008 n=5+5) Filter/tcp6_syn_out-8 32.8ns ± 2% 81.0ns ± 2% +147.10% (p=0.008 n=5+5) Filter/udp6_in-8 106ns ± 2% 290ns ± 2% +174.48% (p=0.008 n=5+5) Filter/udp6_out-8 184ns ± 2% 314ns ± 3% +70.43% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/tstun goos:linux goarch:amd64 Write-8 9.02ns ± 3% 8.92ns ± 1% ~ (p=0.421 n=5+5) name old alloc/op new alloc/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00B 0.00B ~ (all equal) Decode/tcp6-8 0.00B 0.00B ~ (all equal) Decode/udp4-8 0.00B 0.00B ~ (all equal) Decode/udp6-8 0.00B 0.00B ~ (all equal) Decode/icmp4-8 0.00B 0.00B ~ (all equal) Decode/icmp6-8 0.00B 0.00B ~ (all equal) Decode/igmp-8 0.00B 0.00B ~ (all equal) Decode/unknown-8 0.00B 0.00B ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp4_in-8 0.00B 0.00B ~ (all equal) Filter/udp4_out-8 16.0B ± 0% 64.0B ± 0% +300.00% (p=0.008 n=5+5) Filter/icmp6-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp6_in-8 0.00B 0.00B ~ (all equal) Filter/udp6_out-8 48.0B ± 0% 64.0B ± 0% +33.33% (p=0.008 n=5+5) name old allocs/op new allocs/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00 0.00 ~ (all equal) Decode/tcp6-8 0.00 0.00 ~ (all equal) Decode/udp4-8 0.00 0.00 ~ (all equal) Decode/udp6-8 0.00 0.00 ~ (all equal) Decode/icmp4-8 0.00 0.00 ~ (all equal) Decode/icmp6-8 0.00 0.00 ~ (all equal) Decode/igmp-8 0.00 0.00 ~ (all equal) Decode/unknown-8 0.00 0.00 ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp4_in-8 0.00 0.00 ~ (all equal) Filter/udp4_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Filter/icmp6-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp6_in-8 0.00 0.00 ~ (all equal) Filter/udp6_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
if err != nil {
panic(err)
}
dip, err := netip.ParseAddr(dst)
net/packet: remove the custom IP4/IP6 types in favor of netaddr.IP. Upstream netaddr has a change that makes it alloc-free, so it's safe to use in hot codepaths. This gets rid of one of the many IP types in our codebase. Performance is currently worse across the board. This is likely due in part to netaddr.IP being a larger value type (4b -> 24b for IPv4, 16b -> 24b for IPv6), and in other part due to missing low-hanging fruit optimizations in netaddr. However, the regression is less bad than it looks at first glance, because we'd micro-optimized packet.IP* in the past few weeks. This change drops us back to roughly where we were at the 1.2 release, but with the benefit of a significant code and architectural simplification. name old time/op new time/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 12.2ns ± 5% 29.7ns ± 2% +142.32% (p=0.008 n=5+5) Decode/tcp6-8 12.6ns ± 3% 65.1ns ± 2% +418.47% (p=0.008 n=5+5) Decode/udp4-8 11.8ns ± 3% 30.5ns ± 2% +157.94% (p=0.008 n=5+5) Decode/udp6-8 27.1ns ± 1% 65.7ns ± 2% +142.36% (p=0.016 n=4+5) Decode/icmp4-8 24.6ns ± 2% 30.5ns ± 2% +23.65% (p=0.016 n=4+5) Decode/icmp6-8 22.9ns ±51% 65.5ns ± 2% +186.19% (p=0.008 n=5+5) Decode/igmp-8 18.1ns ±44% 30.2ns ± 1% +66.89% (p=0.008 n=5+5) Decode/unknown-8 20.8ns ± 1% 10.6ns ± 9% -49.11% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 30.5ns ± 1% 77.9ns ± 3% +155.01% (p=0.008 n=5+5) Filter/tcp4_syn_in-8 43.7ns ± 3% 123.0ns ± 3% +181.72% (p=0.008 n=5+5) Filter/tcp4_syn_out-8 24.5ns ± 2% 45.7ns ± 6% +86.22% (p=0.008 n=5+5) Filter/udp4_in-8 64.8ns ± 1% 210.0ns ± 2% +223.87% (p=0.008 n=5+5) Filter/udp4_out-8 119ns ± 0% 278ns ± 0% +133.78% (p=0.016 n=4+5) Filter/icmp6-8 40.3ns ± 2% 204.4ns ± 4% +407.70% (p=0.008 n=5+5) Filter/tcp6_syn_in-8 35.3ns ± 3% 199.2ns ± 2% +464.95% (p=0.008 n=5+5) Filter/tcp6_syn_out-8 32.8ns ± 2% 81.0ns ± 2% +147.10% (p=0.008 n=5+5) Filter/udp6_in-8 106ns ± 2% 290ns ± 2% +174.48% (p=0.008 n=5+5) Filter/udp6_out-8 184ns ± 2% 314ns ± 3% +70.43% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/tstun goos:linux goarch:amd64 Write-8 9.02ns ± 3% 8.92ns ± 1% ~ (p=0.421 n=5+5) name old alloc/op new alloc/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00B 0.00B ~ (all equal) Decode/tcp6-8 0.00B 0.00B ~ (all equal) Decode/udp4-8 0.00B 0.00B ~ (all equal) Decode/udp6-8 0.00B 0.00B ~ (all equal) Decode/icmp4-8 0.00B 0.00B ~ (all equal) Decode/icmp6-8 0.00B 0.00B ~ (all equal) Decode/igmp-8 0.00B 0.00B ~ (all equal) Decode/unknown-8 0.00B 0.00B ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp4_in-8 0.00B 0.00B ~ (all equal) Filter/udp4_out-8 16.0B ± 0% 64.0B ± 0% +300.00% (p=0.008 n=5+5) Filter/icmp6-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp6_in-8 0.00B 0.00B ~ (all equal) Filter/udp6_out-8 48.0B ± 0% 64.0B ± 0% +33.33% (p=0.008 n=5+5) name old allocs/op new allocs/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00 0.00 ~ (all equal) Decode/tcp6-8 0.00 0.00 ~ (all equal) Decode/udp4-8 0.00 0.00 ~ (all equal) Decode/udp6-8 0.00 0.00 ~ (all equal) Decode/icmp4-8 0.00 0.00 ~ (all equal) Decode/icmp6-8 0.00 0.00 ~ (all equal) Decode/igmp-8 0.00 0.00 ~ (all equal) Decode/unknown-8 0.00 0.00 ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp4_in-8 0.00 0.00 ~ (all equal) Filter/udp4_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Filter/icmp6-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp6_in-8 0.00 0.00 ~ (all equal) Filter/udp6_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
if err != nil {
panic(err)
}
header := &packet.UDP4Header{
IP4Header: packet.IP4Header{
net/packet: remove the custom IP4/IP6 types in favor of netaddr.IP. Upstream netaddr has a change that makes it alloc-free, so it's safe to use in hot codepaths. This gets rid of one of the many IP types in our codebase. Performance is currently worse across the board. This is likely due in part to netaddr.IP being a larger value type (4b -> 24b for IPv4, 16b -> 24b for IPv6), and in other part due to missing low-hanging fruit optimizations in netaddr. However, the regression is less bad than it looks at first glance, because we'd micro-optimized packet.IP* in the past few weeks. This change drops us back to roughly where we were at the 1.2 release, but with the benefit of a significant code and architectural simplification. name old time/op new time/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 12.2ns ± 5% 29.7ns ± 2% +142.32% (p=0.008 n=5+5) Decode/tcp6-8 12.6ns ± 3% 65.1ns ± 2% +418.47% (p=0.008 n=5+5) Decode/udp4-8 11.8ns ± 3% 30.5ns ± 2% +157.94% (p=0.008 n=5+5) Decode/udp6-8 27.1ns ± 1% 65.7ns ± 2% +142.36% (p=0.016 n=4+5) Decode/icmp4-8 24.6ns ± 2% 30.5ns ± 2% +23.65% (p=0.016 n=4+5) Decode/icmp6-8 22.9ns ±51% 65.5ns ± 2% +186.19% (p=0.008 n=5+5) Decode/igmp-8 18.1ns ±44% 30.2ns ± 1% +66.89% (p=0.008 n=5+5) Decode/unknown-8 20.8ns ± 1% 10.6ns ± 9% -49.11% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 30.5ns ± 1% 77.9ns ± 3% +155.01% (p=0.008 n=5+5) Filter/tcp4_syn_in-8 43.7ns ± 3% 123.0ns ± 3% +181.72% (p=0.008 n=5+5) Filter/tcp4_syn_out-8 24.5ns ± 2% 45.7ns ± 6% +86.22% (p=0.008 n=5+5) Filter/udp4_in-8 64.8ns ± 1% 210.0ns ± 2% +223.87% (p=0.008 n=5+5) Filter/udp4_out-8 119ns ± 0% 278ns ± 0% +133.78% (p=0.016 n=4+5) Filter/icmp6-8 40.3ns ± 2% 204.4ns ± 4% +407.70% (p=0.008 n=5+5) Filter/tcp6_syn_in-8 35.3ns ± 3% 199.2ns ± 2% +464.95% (p=0.008 n=5+5) Filter/tcp6_syn_out-8 32.8ns ± 2% 81.0ns ± 2% +147.10% (p=0.008 n=5+5) Filter/udp6_in-8 106ns ± 2% 290ns ± 2% +174.48% (p=0.008 n=5+5) Filter/udp6_out-8 184ns ± 2% 314ns ± 3% +70.43% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/tstun goos:linux goarch:amd64 Write-8 9.02ns ± 3% 8.92ns ± 1% ~ (p=0.421 n=5+5) name old alloc/op new alloc/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00B 0.00B ~ (all equal) Decode/tcp6-8 0.00B 0.00B ~ (all equal) Decode/udp4-8 0.00B 0.00B ~ (all equal) Decode/udp6-8 0.00B 0.00B ~ (all equal) Decode/icmp4-8 0.00B 0.00B ~ (all equal) Decode/icmp6-8 0.00B 0.00B ~ (all equal) Decode/igmp-8 0.00B 0.00B ~ (all equal) Decode/unknown-8 0.00B 0.00B ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp4_in-8 0.00B 0.00B ~ (all equal) Filter/udp4_out-8 16.0B ± 0% 64.0B ± 0% +300.00% (p=0.008 n=5+5) Filter/icmp6-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp6_in-8 0.00B 0.00B ~ (all equal) Filter/udp6_out-8 48.0B ± 0% 64.0B ± 0% +33.33% (p=0.008 n=5+5) name old allocs/op new allocs/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00 0.00 ~ (all equal) Decode/tcp6-8 0.00 0.00 ~ (all equal) Decode/udp4-8 0.00 0.00 ~ (all equal) Decode/udp6-8 0.00 0.00 ~ (all equal) Decode/icmp4-8 0.00 0.00 ~ (all equal) Decode/icmp6-8 0.00 0.00 ~ (all equal) Decode/igmp-8 0.00 0.00 ~ (all equal) Decode/unknown-8 0.00 0.00 ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp4_in-8 0.00 0.00 ~ (all equal) Filter/udp4_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Filter/icmp6-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp6_in-8 0.00 0.00 ~ (all equal) Filter/udp6_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
Src: sip,
Dst: dip,
IPID: 0,
},
SrcPort: sport,
DstPort: dport,
}
return packet.Generate(header, []byte("udp_payload"))
}
func tcp4syn(src, dst string, sport, dport uint16) []byte {
sip, err := netip.ParseAddr(src)
if err != nil {
panic(err)
}
dip, err := netip.ParseAddr(dst)
if err != nil {
panic(err)
}
ipHeader := packet.IP4Header{
IPProto: ipproto.TCP,
Src: sip,
Dst: dip,
IPID: 0,
}
tcpHeader := make([]byte, 20)
binary.BigEndian.PutUint16(tcpHeader[0:], sport)
binary.BigEndian.PutUint16(tcpHeader[2:], dport)
tcpHeader[13] |= 2 // SYN
both := packet.Generate(ipHeader, tcpHeader)
// 20 byte IP4 + 20 byte TCP
binary.BigEndian.PutUint16(both[2:4], 40)
return both
}
func nets(nets ...string) (ret []netip.Prefix) {
for _, s := range nets {
if i := strings.IndexByte(s, '/'); i == -1 {
ip, err := netip.ParseAddr(s)
if err != nil {
panic(err)
}
bits := uint8(32)
if ip.Is6() {
bits = 128
}
ret = append(ret, netip.PrefixFrom(ip, int(bits)))
} else {
pfx, err := netip.ParsePrefix(s)
if err != nil {
panic(err)
}
ret = append(ret, pfx)
}
}
return ret
}
func ports(s string) filter.PortRange {
if s == "*" {
return filter.PortRange{First: 0, Last: 65535}
}
var fs, ls string
i := strings.IndexByte(s, '-')
if i == -1 {
fs = s
ls = fs
} else {
fs = s[:i]
ls = s[i+1:]
}
first, err := strconv.ParseInt(fs, 10, 16)
if err != nil {
panic(fmt.Sprintf("invalid NetPortRange %q", s))
}
last, err := strconv.ParseInt(ls, 10, 16)
if err != nil {
panic(fmt.Sprintf("invalid NetPortRange %q", s))
}
return filter.PortRange{First: uint16(first), Last: uint16(last)}
}
func netports(netPorts ...string) (ret []filter.NetPortRange) {
for _, s := range netPorts {
i := strings.LastIndexByte(s, ':')
if i == -1 {
panic(fmt.Sprintf("invalid NetPortRange %q", s))
}
npr := filter.NetPortRange{
Net: nets(s[:i])[0],
Ports: ports(s[i+1:]),
}
ret = append(ret, npr)
}
return ret
}
func setfilter(logf logger.Logf, tun *Wrapper) {
protos := []ipproto.Proto{
ipproto.TCP,
ipproto.UDP,
}
matches := []filter.Match{
{IPProto: protos, Srcs: nets("5.6.7.8"), Dsts: netports("1.2.3.4:89-90")},
{IPProto: protos, Srcs: nets("1.2.3.4"), Dsts: netports("5.6.7.8:98")},
}
var sb netipx.IPSetBuilder
sb.AddPrefix(netip.MustParsePrefix("1.2.0.0/16"))
ipSet, _ := sb.IPSet()
tun.SetFilter(filter.New(matches, ipSet, ipSet, nil, logf))
}
func newChannelTUN(logf logger.Logf, secure bool) (*tuntest.ChannelTUN, *Wrapper) {
chtun := tuntest.NewChannelTUN()
tun := Wrap(logf, chtun.TUN())
if secure {
setfilter(logf, tun)
} else {
tun.disableFilter = true
}
return chtun, tun
}
func newFakeTUN(logf logger.Logf, secure bool) (*fakeTUN, *Wrapper) {
ftun := NewFake()
tun := Wrap(logf, ftun)
if secure {
setfilter(logf, tun)
} else {
tun.disableFilter = true
}
return ftun.(*fakeTUN), tun
}
func TestReadAndInject(t *testing.T) {
chtun, tun := newChannelTUN(t.Logf, false)
defer tun.Close()
const size = 2 // all payloads have this size
written := []string{"w0", "w1"}
injected := []string{"i0", "i1"}
go func() {
for _, packet := range written {
payload := []byte(packet)
chtun.Outbound <- payload
}
}()
for _, packet := range injected {
go func(packet string) {
payload := []byte(packet)
err := tun.InjectOutbound(payload)
if err != nil {
t.Errorf("%s: error: %v", packet, err)
}
}(packet)
}
var buf [MaxPacketSize]byte
var seen = make(map[string]bool)
sizes := make([]int, 1)
// We expect the same packets back, in no particular order.
for i := 0; i < len(written)+len(injected); i++ {
packet := buf[:]
buffs := [][]byte{packet}
numPackets, err := tun.Read(buffs, sizes, 0)
if err != nil {
t.Errorf("read %d: error: %v", i, err)
}
if numPackets != 1 {
t.Fatalf("read %d packets, expected %d", numPackets, 1)
}
packet = packet[:sizes[0]]
packetLen := len(packet)
if packetLen != size {
t.Errorf("read %d: got size %d; want %d", i, packetLen, size)
}
got := string(packet)
t.Logf("read %d: got %s", i, got)
seen[got] = true
}
for _, packet := range written {
if !seen[packet] {
t.Errorf("%s not received", packet)
}
}
for _, packet := range injected {
if !seen[packet] {
t.Errorf("%s not received", packet)
}
}
}
func TestWriteAndInject(t *testing.T) {
chtun, tun := newChannelTUN(t.Logf, false)
defer tun.Close()
const size = 2 // all payloads have this size
written := []string{"w0", "w1"}
injected := []string{"i0", "i1"}
go func() {
for _, packet := range written {
payload := []byte(packet)
_, err := tun.Write([][]byte{payload}, 0)
if err != nil {
t.Errorf("%s: error: %v", packet, err)
}
}
}()
for _, packet := range injected {
go func(packet string) {
payload := []byte(packet)
err := tun.InjectInboundCopy(payload)
if err != nil {
t.Errorf("%s: error: %v", packet, err)
}
}(packet)
}
seen := make(map[string]bool)
// We expect the same packets back, in no particular order.
for i := 0; i < len(written)+len(injected); i++ {
packet := <-chtun.Inbound
got := string(packet)
t.Logf("read %d: got %s", i, got)
seen[got] = true
}
for _, packet := range written {
if !seen[packet] {
t.Errorf("%s not received", packet)
}
}
for _, packet := range injected {
if !seen[packet] {
t.Errorf("%s not received", packet)
}
}
}
func TestFilter(t *testing.T) {
chtun, tun := newChannelTUN(t.Logf, true)
defer tun.Close()
type direction int
const (
in direction = iota
out
)
tests := []struct {
name string
dir direction
drop bool
data []byte
}{
{"junk_in", in, true, []byte("\x45not a valid IPv4 packet")},
{"junk_out", out, true, []byte("\x45not a valid IPv4 packet")},
net/packet: remove the custom IP4/IP6 types in favor of netaddr.IP. Upstream netaddr has a change that makes it alloc-free, so it's safe to use in hot codepaths. This gets rid of one of the many IP types in our codebase. Performance is currently worse across the board. This is likely due in part to netaddr.IP being a larger value type (4b -> 24b for IPv4, 16b -> 24b for IPv6), and in other part due to missing low-hanging fruit optimizations in netaddr. However, the regression is less bad than it looks at first glance, because we'd micro-optimized packet.IP* in the past few weeks. This change drops us back to roughly where we were at the 1.2 release, but with the benefit of a significant code and architectural simplification. name old time/op new time/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 12.2ns ± 5% 29.7ns ± 2% +142.32% (p=0.008 n=5+5) Decode/tcp6-8 12.6ns ± 3% 65.1ns ± 2% +418.47% (p=0.008 n=5+5) Decode/udp4-8 11.8ns ± 3% 30.5ns ± 2% +157.94% (p=0.008 n=5+5) Decode/udp6-8 27.1ns ± 1% 65.7ns ± 2% +142.36% (p=0.016 n=4+5) Decode/icmp4-8 24.6ns ± 2% 30.5ns ± 2% +23.65% (p=0.016 n=4+5) Decode/icmp6-8 22.9ns ±51% 65.5ns ± 2% +186.19% (p=0.008 n=5+5) Decode/igmp-8 18.1ns ±44% 30.2ns ± 1% +66.89% (p=0.008 n=5+5) Decode/unknown-8 20.8ns ± 1% 10.6ns ± 9% -49.11% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 30.5ns ± 1% 77.9ns ± 3% +155.01% (p=0.008 n=5+5) Filter/tcp4_syn_in-8 43.7ns ± 3% 123.0ns ± 3% +181.72% (p=0.008 n=5+5) Filter/tcp4_syn_out-8 24.5ns ± 2% 45.7ns ± 6% +86.22% (p=0.008 n=5+5) Filter/udp4_in-8 64.8ns ± 1% 210.0ns ± 2% +223.87% (p=0.008 n=5+5) Filter/udp4_out-8 119ns ± 0% 278ns ± 0% +133.78% (p=0.016 n=4+5) Filter/icmp6-8 40.3ns ± 2% 204.4ns ± 4% +407.70% (p=0.008 n=5+5) Filter/tcp6_syn_in-8 35.3ns ± 3% 199.2ns ± 2% +464.95% (p=0.008 n=5+5) Filter/tcp6_syn_out-8 32.8ns ± 2% 81.0ns ± 2% +147.10% (p=0.008 n=5+5) Filter/udp6_in-8 106ns ± 2% 290ns ± 2% +174.48% (p=0.008 n=5+5) Filter/udp6_out-8 184ns ± 2% 314ns ± 3% +70.43% (p=0.016 n=4+5) pkg:tailscale.com/wgengine/tstun goos:linux goarch:amd64 Write-8 9.02ns ± 3% 8.92ns ± 1% ~ (p=0.421 n=5+5) name old alloc/op new alloc/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00B 0.00B ~ (all equal) Decode/tcp6-8 0.00B 0.00B ~ (all equal) Decode/udp4-8 0.00B 0.00B ~ (all equal) Decode/udp6-8 0.00B 0.00B ~ (all equal) Decode/icmp4-8 0.00B 0.00B ~ (all equal) Decode/icmp6-8 0.00B 0.00B ~ (all equal) Decode/igmp-8 0.00B 0.00B ~ (all equal) Decode/unknown-8 0.00B 0.00B ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp4_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp4_in-8 0.00B 0.00B ~ (all equal) Filter/udp4_out-8 16.0B ± 0% 64.0B ± 0% +300.00% (p=0.008 n=5+5) Filter/icmp6-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_in-8 0.00B 0.00B ~ (all equal) Filter/tcp6_syn_out-8 0.00B 0.00B ~ (all equal) Filter/udp6_in-8 0.00B 0.00B ~ (all equal) Filter/udp6_out-8 48.0B ± 0% 64.0B ± 0% +33.33% (p=0.008 n=5+5) name old allocs/op new allocs/op delta pkg:tailscale.com/net/packet goos:linux goarch:amd64 Decode/tcp4-8 0.00 0.00 ~ (all equal) Decode/tcp6-8 0.00 0.00 ~ (all equal) Decode/udp4-8 0.00 0.00 ~ (all equal) Decode/udp6-8 0.00 0.00 ~ (all equal) Decode/icmp4-8 0.00 0.00 ~ (all equal) Decode/icmp6-8 0.00 0.00 ~ (all equal) Decode/igmp-8 0.00 0.00 ~ (all equal) Decode/unknown-8 0.00 0.00 ~ (all equal) pkg:tailscale.com/wgengine/filter goos:linux goarch:amd64 Filter/icmp4-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp4_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp4_in-8 0.00 0.00 ~ (all equal) Filter/udp4_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Filter/icmp6-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_in-8 0.00 0.00 ~ (all equal) Filter/tcp6_syn_out-8 0.00 0.00 ~ (all equal) Filter/udp6_in-8 0.00 0.00 ~ (all equal) Filter/udp6_out-8 1.00 ± 0% 1.00 ± 0% ~ (all equal) Signed-off-by: David Anderson <danderson@tailscale.com>
4 years ago
{"bad_port_in", in, true, udp4("5.6.7.8", "1.2.3.4", 22, 22)},
{"bad_port_out", out, false, udp4("1.2.3.4", "5.6.7.8", 22, 22)},
{"bad_ip_in", in, true, udp4("8.1.1.1", "1.2.3.4", 89, 89)},
{"bad_ip_out", out, false, udp4("1.2.3.4", "8.1.1.1", 98, 98)},
{"good_packet_in", in, false, udp4("5.6.7.8", "1.2.3.4", 89, 89)},
{"good_packet_out", out, false, udp4("1.2.3.4", "5.6.7.8", 98, 98)},
}
// A reader on the other end of the tun.
go func() {
var recvbuf []byte
for {
select {
case <-tun.closed:
return
case recvbuf = <-chtun.Inbound:
// continue
}
for _, tt := range tests {
if tt.drop && bytes.Equal(recvbuf, tt.data) {
t.Errorf("did not drop %s", tt.name)
}
}
}
}()
var buf [MaxPacketSize]byte
stats := connstats.NewStatistics(0, 0, nil)
defer stats.Shutdown(context.Background())
tun.SetStatistics(stats)
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
var n int
var err error
var filtered bool
sizes := make([]int, 1)
tunStats, _ := stats.TestExtract()
if len(tunStats) > 0 {
t.Errorf("connstats.Statistics.Extract = %v, want {}", stats)
}
if tt.dir == in {
// Use the side effect of updating the last
// activity atomic to determine whether the
// data was actually filtered.
// If it stays zero, nothing made it through
// to the wrapped TUN.
tun.lastActivityAtomic.StoreAtomic(0)
_, err = tun.Write([][]byte{tt.data}, 0)
filtered = tun.lastActivityAtomic.LoadAtomic() == 0
} else {
chtun.Outbound <- tt.data
n, err = tun.Read([][]byte{buf[:]}, sizes, 0)
// In the read direction, errors are fatal, so we return n = 0 instead.
filtered = (n == 0)
}
if err != nil {
t.Errorf("got err %v; want nil", err)
}
if filtered {
if !tt.drop {
t.Errorf("got drop; want accept")
}
} else {
if tt.drop {
t.Errorf("got accept; want drop")
}
}
got, _ := stats.TestExtract()
want := map[netlogtype.Connection]netlogtype.Counts{}
if !tt.drop {
var p packet.Parsed
p.Decode(tt.data)
switch tt.dir {
case in:
conn := netlogtype.Connection{Proto: ipproto.UDP, Src: p.Dst, Dst: p.Src}
want[conn] = netlogtype.Counts{RxPackets: 1, RxBytes: uint64(len(tt.data))}
case out:
conn := netlogtype.Connection{Proto: ipproto.UDP, Src: p.Src, Dst: p.Dst}
want[conn] = netlogtype.Counts{TxPackets: 1, TxBytes: uint64(len(tt.data))}
}
}
if diff := cmp.Diff(got, want, cmpopts.EquateEmpty()); diff != "" {
t.Errorf("stats.TestExtract (-got +want):\n%s", diff)
}
})
}
}
func TestAllocs(t *testing.T) {
ftun, tun := newFakeTUN(t.Logf, false)
defer tun.Close()
buf := [][]byte{[]byte{0x00}}
err := tstest.MinAllocsPerRun(t, 0, func() {
_, err := ftun.Write(buf, 0)
if err != nil {
t.Errorf("write: error: %v", err)
return
}
})
if err != nil {
t.Error(err)
}
}
func TestClose(t *testing.T) {
ftun, tun := newFakeTUN(t.Logf, false)
data := [][]byte{udp4("1.2.3.4", "5.6.7.8", 98, 98)}
_, err := ftun.Write(data, 0)
if err != nil {
t.Error(err)
}
tun.Close()
_, err = ftun.Write(data, 0)
if err == nil {
t.Error("Expected error from ftun.Write() after Close()")
}
}
func BenchmarkWrite(b *testing.B) {
b.ReportAllocs()
ftun, tun := newFakeTUN(b.Logf, true)
defer tun.Close()
packet := [][]byte{udp4("5.6.7.8", "1.2.3.4", 89, 89)}
for i := 0; i < b.N; i++ {
_, err := ftun.Write(packet, 0)
if err != nil {
b.Errorf("err = %v; want nil", err)
}
}
}
func TestAtomic64Alignment(t *testing.T) {
off := unsafe.Offsetof(Wrapper{}.lastActivityAtomic)
if off%8 != 0 {
t.Errorf("offset %v not 8-byte aligned", off)
}
c := new(Wrapper)
c.lastActivityAtomic.StoreAtomic(mono.Now())
}
func TestPeerAPIBypass(t *testing.T) {
wrapperWithPeerAPI := &Wrapper{
PeerAPIPort: func(ip netip.Addr) (port uint16, ok bool) {
if ip == netip.MustParseAddr("100.64.1.2") {
return 60000, true
}
return
},
}
tests := []struct {
name string
w *Wrapper
filter *filter.Filter
pkt []byte
want filter.Response
}{
{
name: "reject_nil_filter",
w: &Wrapper{
PeerAPIPort: func(netip.Addr) (port uint16, ok bool) {
return 60000, true
},
},
pkt: tcp4syn("1.2.3.4", "100.64.1.2", 1234, 60000),
want: filter.Drop,
},
{
name: "reject_with_filter",
w: &Wrapper{},
filter: filter.NewAllowNone(logger.Discard, new(netipx.IPSet)),
pkt: tcp4syn("1.2.3.4", "100.64.1.2", 1234, 60000),
want: filter.Drop,
},
{
name: "peerapi_bypass_filter",
w: wrapperWithPeerAPI,
filter: filter.NewAllowNone(logger.Discard, new(netipx.IPSet)),
pkt: tcp4syn("1.2.3.4", "100.64.1.2", 1234, 60000),
want: filter.Accept,
},
{
name: "peerapi_dont_bypass_filter_wrong_port",
w: wrapperWithPeerAPI,
filter: filter.NewAllowNone(logger.Discard, new(netipx.IPSet)),
pkt: tcp4syn("1.2.3.4", "100.64.1.2", 1234, 60001),
want: filter.Drop,
},
{
name: "peerapi_dont_bypass_filter_wrong_dst_ip",
w: wrapperWithPeerAPI,
filter: filter.NewAllowNone(logger.Discard, new(netipx.IPSet)),
pkt: tcp4syn("1.2.3.4", "100.64.1.3", 1234, 60000),
want: filter.Drop,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
p := new(packet.Parsed)
p.Decode(tt.pkt)
tt.w.SetFilter(tt.filter)
tt.w.disableTSMPRejected = true
if got := tt.w.filterIn(p); got != tt.want {
t.Errorf("got = %v; want %v", got, tt.want)
}
})
}
}
// Issue 1526: drop disco frames from ourselves.
func TestFilterDiscoLoop(t *testing.T) {
var memLog tstest.MemLogger
discoPub := key.DiscoPublicFromRaw32(mem.B([]byte{1: 1, 2: 2, 31: 0}))
tw := &Wrapper{logf: memLog.Logf, limitedLogf: memLog.Logf}
tw.SetDiscoKey(discoPub)
uh := packet.UDP4Header{
IP4Header: packet.IP4Header{
IPProto: ipproto.UDP,
Src: netaddr.IPv4(1, 2, 3, 4),
Dst: netaddr.IPv4(5, 6, 7, 8),
},
SrcPort: 9,
DstPort: 10,
}
discobs := discoPub.Raw32()
discoPayload := fmt.Sprintf("%s%s%s", disco.Magic, discobs[:], [disco.NonceLen]byte{})
pkt := make([]byte, uh.Len()+len(discoPayload))
uh.Marshal(pkt)
copy(pkt[uh.Len():], discoPayload)
p := new(packet.Parsed)
p.Decode(pkt)
got := tw.filterIn(p)
if got != filter.DropSilently {
t.Errorf("got %v; want DropSilently", got)
}
if got, want := memLog.String(), "[unexpected] received self disco in packet over tstun; dropping\n"; got != want {
t.Errorf("log output mismatch\n got: %q\nwant: %q\n", got, want)
}
memLog.Reset()
pp := new(packet.Parsed)
pp.Decode(pkt)
got = tw.filterOut(pp)
if got != filter.DropSilently {
t.Errorf("got %v; want DropSilently", got)
}
if got, want := memLog.String(), "[unexpected] received self disco out packet over tstun; dropping\n"; got != want {
t.Errorf("log output mismatch\n got: %q\nwant: %q\n", got, want)
}
}