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
|
// Copyright 2020 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 ipv4
import (
"fmt"
"sync/atomic"
"time"
"gvisor.dev/gvisor/pkg/tcpip"
"gvisor.dev/gvisor/pkg/tcpip/buffer"
"gvisor.dev/gvisor/pkg/tcpip/header"
"gvisor.dev/gvisor/pkg/tcpip/network/ip"
"gvisor.dev/gvisor/pkg/tcpip/stack"
)
const (
// igmpV1PresentDefault is the initial state for igmpV1Present in the
// igmpState. As per RFC 2236 Page 9 says "No IGMPv1 Router Present ... is
// the initial state."
igmpV1PresentDefault = 0
// v1RouterPresentTimeout from RFC 2236 Section 8.11, Page 18
// See note on igmpState.igmpV1Present for more detail.
v1RouterPresentTimeout = 400 * time.Second
// v1MaxRespTime from RFC 2236 Section 4, Page 5. "The IGMPv1 router
// will send General Queries with the Max Response Time set to 0. This MUST
// be interpreted as a value of 100 (10 seconds)."
//
// Note that the Max Response Time field is a value in units of deciseconds.
v1MaxRespTime = 10 * time.Second
// UnsolicitedReportIntervalMax is the maximum delay between sending
// unsolicited IGMP reports.
//
// Obtained from RFC 2236 Section 8.10, Page 19.
UnsolicitedReportIntervalMax = 10 * time.Second
)
// IGMPOptions holds options for IGMP.
type IGMPOptions struct {
// Enabled indicates whether IGMP will be performed.
//
// When enabled, IGMP may transmit IGMP report and leave messages when
// joining and leaving multicast groups respectively, and handle incoming
// IGMP packets.
//
// This field is ignored and is always assumed to be false for interfaces
// without neighbouring nodes (e.g. loopback).
Enabled bool
}
var _ ip.MulticastGroupProtocol = (*igmpState)(nil)
// igmpState is the per-interface IGMP state.
//
// igmpState.init() MUST be called after creating an IGMP state.
type igmpState struct {
// The IPv4 endpoint this igmpState is for.
ep *endpoint
genericMulticastProtocol ip.GenericMulticastProtocolState
// igmpV1Present is for maintaining compatibility with IGMPv1 Routers, from
// RFC 2236 Section 4 Page 6: "The IGMPv1 router expects Version 1
// Membership Reports in response to its Queries, and will not pay
// attention to Version 2 Membership Reports. Therefore, a state variable
// MUST be kept for each interface, describing whether the multicast
// Querier on that interface is running IGMPv1 or IGMPv2. This variable
// MUST be based upon whether or not an IGMPv1 query was heard in the last
// [Version 1 Router Present Timeout] seconds".
//
// Must be accessed with atomic operations. Holds a value of 1 when true, 0
// when false.
igmpV1Present uint32
// igmpV1Job is scheduled when this interface receives an IGMPv1 style
// message, upon expiration the igmpV1Present flag is cleared.
// igmpV1Job may not be nil once igmpState is initialized.
igmpV1Job *tcpip.Job
}
// Enabled implements ip.MulticastGroupProtocol.
func (igmp *igmpState) Enabled() bool {
// No need to perform IGMP on loopback interfaces since they don't have
// neighbouring nodes.
return igmp.ep.protocol.options.IGMP.Enabled && !igmp.ep.nic.IsLoopback() && igmp.ep.Enabled()
}
// SendReport implements ip.MulticastGroupProtocol.
//
// Precondition: igmp.ep.mu must be read locked.
func (igmp *igmpState) SendReport(groupAddress tcpip.Address) (bool, tcpip.Error) {
igmpType := header.IGMPv2MembershipReport
if igmp.v1Present() {
igmpType = header.IGMPv1MembershipReport
}
return igmp.writePacket(groupAddress, groupAddress, igmpType)
}
// SendLeave implements ip.MulticastGroupProtocol.
//
// Precondition: igmp.ep.mu must be read locked.
func (igmp *igmpState) SendLeave(groupAddress tcpip.Address) tcpip.Error {
// As per RFC 2236 Section 6, Page 8: "If the interface state says the
// Querier is running IGMPv1, this action SHOULD be skipped. If the flag
// saying we were the last host to report is cleared, this action MAY be
// skipped."
if igmp.v1Present() {
return nil
}
_, err := igmp.writePacket(header.IPv4AllRoutersGroup, groupAddress, header.IGMPLeaveGroup)
return err
}
// init sets up an igmpState struct, and is required to be called before using
// a new igmpState.
//
// Must only be called once for the lifetime of igmp.
func (igmp *igmpState) init(ep *endpoint) {
igmp.ep = ep
igmp.genericMulticastProtocol.Init(&ep.mu.RWMutex, ip.GenericMulticastProtocolOptions{
Rand: ep.protocol.stack.Rand(),
Clock: ep.protocol.stack.Clock(),
Protocol: igmp,
MaxUnsolicitedReportDelay: UnsolicitedReportIntervalMax,
AllNodesAddress: header.IPv4AllSystems,
})
igmp.igmpV1Present = igmpV1PresentDefault
igmp.igmpV1Job = ep.protocol.stack.NewJob(&ep.mu, func() {
igmp.setV1Present(false)
})
}
// handleIGMP handles an IGMP packet.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) handleIGMP(pkt *stack.PacketBuffer) {
received := igmp.ep.stats.igmp.packetsReceived
headerView, ok := pkt.Data.PullUp(header.IGMPMinimumSize)
if !ok {
received.invalid.Increment()
return
}
h := header.IGMP(headerView)
// As per RFC 1071 section 1.3,
//
// To check a checksum, the 1's complement sum is computed over the
// same set of octets, including the checksum field. If the result
// is all 1 bits (-0 in 1's complement arithmetic), the check
// succeeds.
if header.ChecksumVV(pkt.Data, 0 /* initial */) != 0xFFFF {
received.checksumErrors.Increment()
return
}
switch h.Type() {
case header.IGMPMembershipQuery:
received.membershipQuery.Increment()
if len(headerView) < header.IGMPQueryMinimumSize {
received.invalid.Increment()
return
}
igmp.handleMembershipQuery(h.GroupAddress(), h.MaxRespTime())
case header.IGMPv1MembershipReport:
received.v1MembershipReport.Increment()
if len(headerView) < header.IGMPReportMinimumSize {
received.invalid.Increment()
return
}
igmp.handleMembershipReport(h.GroupAddress())
case header.IGMPv2MembershipReport:
received.v2MembershipReport.Increment()
if len(headerView) < header.IGMPReportMinimumSize {
received.invalid.Increment()
return
}
igmp.handleMembershipReport(h.GroupAddress())
case header.IGMPLeaveGroup:
received.leaveGroup.Increment()
// As per RFC 2236 Section 6, Page 7: "IGMP messages other than Query or
// Report, are ignored in all states"
default:
// As per RFC 2236 Section 2.1 Page 3: "Unrecognized message types should
// be silently ignored. New message types may be used by newer versions of
// IGMP, by multicast routing protocols, or other uses."
received.unrecognized.Increment()
}
}
func (igmp *igmpState) v1Present() bool {
return atomic.LoadUint32(&igmp.igmpV1Present) == 1
}
func (igmp *igmpState) setV1Present(v bool) {
if v {
atomic.StoreUint32(&igmp.igmpV1Present, 1)
} else {
atomic.StoreUint32(&igmp.igmpV1Present, 0)
}
}
func (igmp *igmpState) resetV1Present() {
igmp.igmpV1Job.Cancel()
igmp.setV1Present(false)
}
// handleMembershipQuery handles a membership query.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) handleMembershipQuery(groupAddress tcpip.Address, maxRespTime time.Duration) {
// As per RFC 2236 Section 6, Page 10: If the maximum response time is zero
// then change the state to note that an IGMPv1 router is present and
// schedule the query received Job.
if maxRespTime == 0 && igmp.Enabled() {
igmp.igmpV1Job.Cancel()
igmp.igmpV1Job.Schedule(v1RouterPresentTimeout)
igmp.setV1Present(true)
maxRespTime = v1MaxRespTime
}
igmp.genericMulticastProtocol.HandleQueryLocked(groupAddress, maxRespTime)
}
// handleMembershipReport handles a membership report.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) handleMembershipReport(groupAddress tcpip.Address) {
igmp.genericMulticastProtocol.HandleReportLocked(groupAddress)
}
// writePacket assembles and sends an IGMP packet.
//
// Precondition: igmp.ep.mu must be read locked.
func (igmp *igmpState) writePacket(destAddress tcpip.Address, groupAddress tcpip.Address, igmpType header.IGMPType) (bool, tcpip.Error) {
igmpData := header.IGMP(buffer.NewView(header.IGMPReportMinimumSize))
igmpData.SetType(igmpType)
igmpData.SetGroupAddress(groupAddress)
igmpData.SetChecksum(header.IGMPCalculateChecksum(igmpData))
pkt := stack.NewPacketBuffer(stack.PacketBufferOptions{
ReserveHeaderBytes: int(igmp.ep.MaxHeaderLength()),
Data: buffer.View(igmpData).ToVectorisedView(),
})
addressEndpoint := igmp.ep.acquireOutgoingPrimaryAddressRLocked(destAddress, false /* allowExpired */)
if addressEndpoint == nil {
return false, nil
}
localAddr := addressEndpoint.AddressWithPrefix().Address
addressEndpoint.DecRef()
addressEndpoint = nil
if err := igmp.ep.addIPHeader(localAddr, destAddress, pkt, stack.NetworkHeaderParams{
Protocol: header.IGMPProtocolNumber,
TTL: header.IGMPTTL,
TOS: stack.DefaultTOS,
}, header.IPv4OptionsSerializer{
&header.IPv4SerializableRouterAlertOption{},
}); err != nil {
panic(fmt.Sprintf("failed to add IP header: %s", err))
}
sentStats := igmp.ep.stats.igmp.packetsSent
if err := igmp.ep.nic.WritePacketToRemote(header.EthernetAddressFromMulticastIPv4Address(destAddress), nil /* gso */, ProtocolNumber, pkt); err != nil {
sentStats.dropped.Increment()
return false, err
}
switch igmpType {
case header.IGMPv1MembershipReport:
sentStats.v1MembershipReport.Increment()
case header.IGMPv2MembershipReport:
sentStats.v2MembershipReport.Increment()
case header.IGMPLeaveGroup:
sentStats.leaveGroup.Increment()
default:
panic(fmt.Sprintf("unrecognized igmp type = %d", igmpType))
}
return true, nil
}
// joinGroup handles adding a new group to the membership map, setting up the
// IGMP state for the group, and sending and scheduling the required
// messages.
//
// If the group already exists in the membership map, returns
// *tcpip.ErrDuplicateAddress.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) joinGroup(groupAddress tcpip.Address) {
igmp.genericMulticastProtocol.JoinGroupLocked(groupAddress)
}
// isInGroup returns true if the specified group has been joined locally.
//
// Precondition: igmp.ep.mu must be read locked.
func (igmp *igmpState) isInGroup(groupAddress tcpip.Address) bool {
return igmp.genericMulticastProtocol.IsLocallyJoinedRLocked(groupAddress)
}
// leaveGroup handles removing the group from the membership map, cancels any
// delay timers associated with that group, and sends the Leave Group message
// if required.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) leaveGroup(groupAddress tcpip.Address) tcpip.Error {
// LeaveGroup returns false only if the group was not joined.
if igmp.genericMulticastProtocol.LeaveGroupLocked(groupAddress) {
return nil
}
return &tcpip.ErrBadLocalAddress{}
}
// softLeaveAll leaves all groups from the perspective of IGMP, but remains
// joined locally.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) softLeaveAll() {
igmp.genericMulticastProtocol.MakeAllNonMemberLocked()
}
// initializeAll attemps to initialize the IGMP state for each group that has
// been joined locally.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) initializeAll() {
igmp.genericMulticastProtocol.InitializeGroupsLocked()
}
// sendQueuedReports attempts to send any reports that are queued for sending.
//
// Precondition: igmp.ep.mu must be locked.
func (igmp *igmpState) sendQueuedReports() {
igmp.genericMulticastProtocol.SendQueuedReportsLocked()
}
|