1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
|
// Copyright 2018 The gVisor Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package stack_test
import (
"math"
"math/rand"
"testing"
"gvisor.dev/gvisor/pkg/tcpip"
"gvisor.dev/gvisor/pkg/tcpip/buffer"
"gvisor.dev/gvisor/pkg/tcpip/header"
"gvisor.dev/gvisor/pkg/tcpip/link/channel"
"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
"gvisor.dev/gvisor/pkg/tcpip/network/ipv6"
"gvisor.dev/gvisor/pkg/tcpip/stack"
"gvisor.dev/gvisor/pkg/tcpip/transport/udp"
"gvisor.dev/gvisor/pkg/waiter"
)
const (
stackV6Addr = "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01"
testV6Addr = "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x02"
stackAddr = "\x0a\x00\x00\x01"
stackPort = 1234
testPort = 4096
)
type testContext struct {
t *testing.T
linkEps map[tcpip.NICID]*channel.Endpoint
s *stack.Stack
ep tcpip.Endpoint
wq waiter.Queue
}
func (c *testContext) cleanup() {
if c.ep != nil {
c.ep.Close()
}
}
func (c *testContext) createV6Endpoint(v6only bool) {
var err *tcpip.Error
c.ep, err = c.s.NewEndpoint(udp.ProtocolNumber, ipv6.ProtocolNumber, &c.wq)
if err != nil {
c.t.Fatalf("NewEndpoint failed: %v", err)
}
if err := c.ep.SetSockOptBool(tcpip.V6OnlyOption, v6only); err != nil {
c.t.Fatalf("SetSockOpt failed: %v", err)
}
}
// newDualTestContextMultiNIC creates the testing context and also linkEpIDs NICs.
func newDualTestContextMultiNIC(t *testing.T, mtu uint32, linkEpIDs []tcpip.NICID) *testContext {
s := stack.New(stack.Options{
NetworkProtocols: []stack.NetworkProtocol{ipv4.NewProtocol(), ipv6.NewProtocol()},
TransportProtocols: []stack.TransportProtocol{udp.NewProtocol()}})
linkEps := make(map[tcpip.NICID]*channel.Endpoint)
for _, linkEpID := range linkEpIDs {
channelEp := channel.New(256, mtu, "")
if err := s.CreateNIC(linkEpID, channelEp); err != nil {
t.Fatalf("CreateNIC failed: %v", err)
}
linkEps[linkEpID] = channelEp
if err := s.AddAddress(linkEpID, ipv4.ProtocolNumber, stackAddr); err != nil {
t.Fatalf("AddAddress IPv4 failed: %v", err)
}
if err := s.AddAddress(linkEpID, ipv6.ProtocolNumber, stackV6Addr); err != nil {
t.Fatalf("AddAddress IPv6 failed: %v", err)
}
}
s.SetRouteTable([]tcpip.Route{
{
Destination: header.IPv4EmptySubnet,
NIC: 1,
},
{
Destination: header.IPv6EmptySubnet,
NIC: 1,
},
})
return &testContext{
t: t,
s: s,
linkEps: linkEps,
}
}
type headers struct {
srcPort uint16
dstPort uint16
}
func newPayload() []byte {
b := make([]byte, 30+rand.Intn(100))
for i := range b {
b[i] = byte(rand.Intn(256))
}
return b
}
func (c *testContext) sendV6Packet(payload []byte, h *headers, linkEpID tcpip.NICID) {
// Allocate a buffer for data and headers.
buf := buffer.NewView(header.UDPMinimumSize + header.IPv6MinimumSize + len(payload))
copy(buf[len(buf)-len(payload):], payload)
// Initialize the IP header.
ip := header.IPv6(buf)
ip.Encode(&header.IPv6Fields{
PayloadLength: uint16(header.UDPMinimumSize + len(payload)),
NextHeader: uint8(udp.ProtocolNumber),
HopLimit: 65,
SrcAddr: testV6Addr,
DstAddr: stackV6Addr,
})
// Initialize the UDP header.
u := header.UDP(buf[header.IPv6MinimumSize:])
u.Encode(&header.UDPFields{
SrcPort: h.srcPort,
DstPort: h.dstPort,
Length: uint16(header.UDPMinimumSize + len(payload)),
})
// Calculate the UDP pseudo-header checksum.
xsum := header.PseudoHeaderChecksum(udp.ProtocolNumber, testV6Addr, stackV6Addr, uint16(len(u)))
// Calculate the UDP checksum and set it.
xsum = header.Checksum(payload, xsum)
u.SetChecksum(^u.CalculateChecksum(xsum))
// Inject packet.
c.linkEps[linkEpID].InjectInbound(ipv6.ProtocolNumber, tcpip.PacketBuffer{
Data: buf.ToVectorisedView(),
})
}
func TestTransportDemuxerRegister(t *testing.T) {
for _, test := range []struct {
name string
proto tcpip.NetworkProtocolNumber
want *tcpip.Error
}{
{"failure", ipv6.ProtocolNumber, tcpip.ErrUnknownProtocol},
{"success", ipv4.ProtocolNumber, nil},
} {
t.Run(test.name, func(t *testing.T) {
s := stack.New(stack.Options{
NetworkProtocols: []stack.NetworkProtocol{ipv4.NewProtocol()},
TransportProtocols: []stack.TransportProtocol{udp.NewProtocol()},
})
var wq waiter.Queue
ep, err := s.NewEndpoint(udp.ProtocolNumber, ipv4.ProtocolNumber, &wq)
if err != nil {
t.Fatal(err)
}
tEP, ok := ep.(stack.TransportEndpoint)
if !ok {
t.Fatalf("%T does not implement stack.TransportEndpoint", ep)
}
if got, want := s.RegisterTransportEndpoint(0, []tcpip.NetworkProtocolNumber{test.proto}, udp.ProtocolNumber, stack.TransportEndpointID{}, tEP, false, 0), test.want; got != want {
t.Fatalf("s.RegisterTransportEndpoint(...) = %v, want %v", got, want)
}
})
}
}
// TestReuseBindToDevice injects varied packets on input devices and checks that
// the distribution of packets received matches expectations.
func TestDistribution(t *testing.T) {
type endpointSockopts struct {
reuse int
bindToDevice tcpip.NICID
}
for _, test := range []struct {
name string
// endpoints will received the inject packets.
endpoints []endpointSockopts
// wantedDistribution is the wanted ratio of packets received on each
// endpoint for each NIC on which packets are injected.
wantedDistributions map[tcpip.NICID][]float64
}{
{
"BindPortReuse",
// 5 endpoints that all have reuse set.
[]endpointSockopts{
{1, 0},
{1, 0},
{1, 0},
{1, 0},
{1, 0},
},
map[tcpip.NICID][]float64{
// Injected packets on dev0 get distributed evenly.
1: {0.2, 0.2, 0.2, 0.2, 0.2},
},
},
{
"BindToDevice",
// 3 endpoints with various bindings.
[]endpointSockopts{
{0, 1},
{0, 2},
{0, 3},
},
map[tcpip.NICID][]float64{
// Injected packets on dev0 go only to the endpoint bound to dev0.
1: {1, 0, 0},
// Injected packets on dev1 go only to the endpoint bound to dev1.
2: {0, 1, 0},
// Injected packets on dev2 go only to the endpoint bound to dev2.
3: {0, 0, 1},
},
},
{
"ReuseAndBindToDevice",
// 6 endpoints with various bindings.
[]endpointSockopts{
{1, 1},
{1, 1},
{1, 2},
{1, 2},
{1, 2},
{1, 0},
},
map[tcpip.NICID][]float64{
// Injected packets on dev0 get distributed among endpoints bound to
// dev0.
1: {0.5, 0.5, 0, 0, 0, 0},
// Injected packets on dev1 get distributed among endpoints bound to
// dev1 or unbound.
2: {0, 0, 1. / 3, 1. / 3, 1. / 3, 0},
// Injected packets on dev999 go only to the unbound.
1000: {0, 0, 0, 0, 0, 1},
},
},
} {
t.Run(test.name, func(t *testing.T) {
for device, wantedDistribution := range test.wantedDistributions {
t.Run(string(device), func(t *testing.T) {
var devices []tcpip.NICID
for d := range test.wantedDistributions {
devices = append(devices, d)
}
c := newDualTestContextMultiNIC(t, defaultMTU, devices)
defer c.cleanup()
c.createV6Endpoint(false)
eps := make(map[tcpip.Endpoint]int)
pollChannel := make(chan tcpip.Endpoint)
for i, endpoint := range test.endpoints {
// Try to receive the data.
wq := waiter.Queue{}
we, ch := waiter.NewChannelEntry(nil)
wq.EventRegister(&we, waiter.EventIn)
defer wq.EventUnregister(&we)
defer close(ch)
var err *tcpip.Error
ep, err := c.s.NewEndpoint(udp.ProtocolNumber, ipv6.ProtocolNumber, &wq)
if err != nil {
c.t.Fatalf("NewEndpoint failed: %v", err)
}
eps[ep] = i
go func(ep tcpip.Endpoint) {
for range ch {
pollChannel <- ep
}
}(ep)
defer ep.Close()
reusePortOption := tcpip.ReusePortOption(endpoint.reuse)
if err := ep.SetSockOpt(reusePortOption); err != nil {
c.t.Fatalf("SetSockOpt(%#v) on endpoint %d failed: %v", reusePortOption, i, err)
}
bindToDeviceOption := tcpip.BindToDeviceOption(endpoint.bindToDevice)
if err := ep.SetSockOpt(bindToDeviceOption); err != nil {
c.t.Fatalf("SetSockOpt(%#v) on endpoint %d failed: %v", bindToDeviceOption, i, err)
}
if err := ep.Bind(tcpip.FullAddress{Addr: stackV6Addr, Port: stackPort}); err != nil {
t.Fatalf("ep.Bind(...) on endpoint %d failed: %v", i, err)
}
}
npackets := 100000
nports := 10000
if got, want := len(test.endpoints), len(wantedDistribution); got != want {
t.Fatalf("got len(test.endpoints) = %d, want %d", got, want)
}
ports := make(map[uint16]tcpip.Endpoint)
stats := make(map[tcpip.Endpoint]int)
for i := 0; i < npackets; i++ {
// Send a packet.
port := uint16(i % nports)
payload := newPayload()
c.sendV6Packet(payload,
&headers{
srcPort: testPort + port,
dstPort: stackPort},
device)
var addr tcpip.FullAddress
ep := <-pollChannel
_, _, err := ep.Read(&addr)
if err != nil {
c.t.Fatalf("Read on endpoint %d failed: %v", eps[ep], err)
}
stats[ep]++
if i < nports {
ports[uint16(i)] = ep
} else {
// Check that all packets from one client are handled by the same
// socket.
if want, got := ports[port], ep; want != got {
t.Fatalf("Packet sent on port %d expected on endpoint %d but received on endpoint %d", port, eps[want], eps[got])
}
}
}
// Check that a packet distribution is as expected.
for ep, i := range eps {
wantedRatio := wantedDistribution[i]
wantedRecv := wantedRatio * float64(npackets)
actualRecv := stats[ep]
actualRatio := float64(stats[ep]) / float64(npackets)
// The deviation is less than 10%.
if math.Abs(actualRatio-wantedRatio) > 0.05 {
t.Errorf("wanted about %.0f%% (%.0f of %d) packets to arrive on endpoint %d, got %.0f%% (%d of %d)", wantedRatio*100, wantedRecv, npackets, i, actualRatio*100, actualRecv, npackets)
}
}
})
}
})
}
}
|