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
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
|
// 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
import (
"log"
"reflect"
"sort"
"strings"
"sync/atomic"
"gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/tcpip"
"gvisor.dev/gvisor/pkg/tcpip/buffer"
"gvisor.dev/gvisor/pkg/tcpip/header"
)
var ipv4BroadcastAddr = tcpip.ProtocolAddress{
Protocol: header.IPv4ProtocolNumber,
AddressWithPrefix: tcpip.AddressWithPrefix{
Address: header.IPv4Broadcast,
PrefixLen: 8 * header.IPv4AddressSize,
},
}
// NIC represents a "network interface card" to which the networking stack is
// attached.
type NIC struct {
stack *Stack
id tcpip.NICID
name string
linkEP LinkEndpoint
context NICContext
stats NICStats
mu struct {
sync.RWMutex
enabled bool
spoofing bool
promiscuous bool
primary map[tcpip.NetworkProtocolNumber][]*referencedNetworkEndpoint
endpoints map[NetworkEndpointID]*referencedNetworkEndpoint
addressRanges []tcpip.Subnet
mcastJoins map[NetworkEndpointID]int32
// packetEPs is protected by mu, but the contained PacketEndpoint
// values are not.
packetEPs map[tcpip.NetworkProtocolNumber][]PacketEndpoint
ndp ndpState
}
}
// NICStats includes transmitted and received stats.
type NICStats struct {
Tx DirectionStats
Rx DirectionStats
DisabledRx DirectionStats
}
func makeNICStats() NICStats {
var s NICStats
tcpip.InitStatCounters(reflect.ValueOf(&s).Elem())
return s
}
// DirectionStats includes packet and byte counts.
type DirectionStats struct {
Packets *tcpip.StatCounter
Bytes *tcpip.StatCounter
}
// PrimaryEndpointBehavior is an enumeration of an endpoint's primacy behavior.
type PrimaryEndpointBehavior int
const (
// CanBePrimaryEndpoint indicates the endpoint can be used as a primary
// endpoint for new connections with no local address. This is the
// default when calling NIC.AddAddress.
CanBePrimaryEndpoint PrimaryEndpointBehavior = iota
// FirstPrimaryEndpoint indicates the endpoint should be the first
// primary endpoint considered. If there are multiple endpoints with
// this behavior, the most recently-added one will be first.
FirstPrimaryEndpoint
// NeverPrimaryEndpoint indicates the endpoint should never be a
// primary endpoint.
NeverPrimaryEndpoint
)
// newNIC returns a new NIC using the default NDP configurations from stack.
func newNIC(stack *Stack, id tcpip.NICID, name string, ep LinkEndpoint, ctx NICContext) *NIC {
// TODO(b/141011931): Validate a LinkEndpoint (ep) is valid. For
// example, make sure that the link address it provides is a valid
// unicast ethernet address.
// TODO(b/143357959): RFC 8200 section 5 requires that IPv6 endpoints
// observe an MTU of at least 1280 bytes. Ensure that this requirement
// of IPv6 is supported on this endpoint's LinkEndpoint.
nic := &NIC{
stack: stack,
id: id,
name: name,
linkEP: ep,
context: ctx,
stats: makeNICStats(),
}
nic.mu.primary = make(map[tcpip.NetworkProtocolNumber][]*referencedNetworkEndpoint)
nic.mu.endpoints = make(map[NetworkEndpointID]*referencedNetworkEndpoint)
nic.mu.mcastJoins = make(map[NetworkEndpointID]int32)
nic.mu.packetEPs = make(map[tcpip.NetworkProtocolNumber][]PacketEndpoint)
nic.mu.ndp = ndpState{
nic: nic,
configs: stack.ndpConfigs,
dad: make(map[tcpip.Address]dadState),
defaultRouters: make(map[tcpip.Address]defaultRouterState),
onLinkPrefixes: make(map[tcpip.Subnet]onLinkPrefixState),
autoGenAddresses: make(map[tcpip.Address]autoGenAddressState),
}
// Register supported packet endpoint protocols.
for _, netProto := range header.Ethertypes {
nic.mu.packetEPs[netProto] = []PacketEndpoint{}
}
for _, netProto := range stack.networkProtocols {
nic.mu.packetEPs[netProto.Number()] = []PacketEndpoint{}
}
nic.linkEP.Attach(nic)
return nic
}
// enabled returns true if n is enabled.
func (n *NIC) enabled() bool {
n.mu.RLock()
enabled := n.mu.enabled
n.mu.RUnlock()
return enabled
}
// disable disables n.
//
// It undoes the work done by enable.
func (n *NIC) disable() *tcpip.Error {
n.mu.RLock()
enabled := n.mu.enabled
n.mu.RUnlock()
if !enabled {
return nil
}
n.mu.Lock()
defer n.mu.Unlock()
if !n.mu.enabled {
return nil
}
// TODO(b/147015577): Should Routes that are currently bound to n be
// invalidated? Currently, Routes will continue to work when a NIC is enabled
// again, and applications may not know that the underlying NIC was ever
// disabled.
if _, ok := n.stack.networkProtocols[header.IPv6ProtocolNumber]; ok {
n.mu.ndp.stopSolicitingRouters()
n.mu.ndp.cleanupState(false /* hostOnly */)
// Stop DAD for all the unicast IPv6 endpoints that are in the
// permanentTentative state.
for _, r := range n.mu.endpoints {
if addr := r.ep.ID().LocalAddress; r.getKind() == permanentTentative && header.IsV6UnicastAddress(addr) {
n.mu.ndp.stopDuplicateAddressDetection(addr)
}
}
// The NIC may have already left the multicast group.
if err := n.leaveGroupLocked(header.IPv6AllNodesMulticastAddress); err != nil && err != tcpip.ErrBadLocalAddress {
return err
}
}
if _, ok := n.stack.networkProtocols[header.IPv4ProtocolNumber]; ok {
// The address may have already been removed.
if err := n.removePermanentAddressLocked(ipv4BroadcastAddr.AddressWithPrefix.Address); err != nil && err != tcpip.ErrBadLocalAddress {
return err
}
}
n.mu.enabled = false
return nil
}
// enable enables n.
//
// If the stack has IPv6 enabled, enable will join the IPv6 All-Nodes Multicast
// address (ff02::1), start DAD for permanent addresses, and start soliciting
// routers if the stack is not operating as a router. If the stack is also
// configured to auto-generate a link-local address, one will be generated.
func (n *NIC) enable() *tcpip.Error {
n.mu.RLock()
enabled := n.mu.enabled
n.mu.RUnlock()
if enabled {
return nil
}
n.mu.Lock()
defer n.mu.Unlock()
if n.mu.enabled {
return nil
}
n.mu.enabled = true
// Create an endpoint to receive broadcast packets on this interface.
if _, ok := n.stack.networkProtocols[header.IPv4ProtocolNumber]; ok {
if _, err := n.addAddressLocked(ipv4BroadcastAddr, NeverPrimaryEndpoint, permanent, static, false /* deprecated */); err != nil {
return err
}
}
// Join the IPv6 All-Nodes Multicast group if the stack is configured to
// use IPv6. This is required to ensure that this node properly receives
// and responds to the various NDP messages that are destined to the
// all-nodes multicast address. An example is the Neighbor Advertisement
// when we perform Duplicate Address Detection, or Router Advertisement
// when we do Router Discovery. See RFC 4862, section 5.4.2 and RFC 4861
// section 4.2 for more information.
//
// Also auto-generate an IPv6 link-local address based on the NIC's
// link address if it is configured to do so. Note, each interface is
// required to have IPv6 link-local unicast address, as per RFC 4291
// section 2.1.
_, ok := n.stack.networkProtocols[header.IPv6ProtocolNumber]
if !ok {
return nil
}
// Join the All-Nodes multicast group before starting DAD as responses to DAD
// (NDP NS) messages may be sent to the All-Nodes multicast group if the
// source address of the NDP NS is the unspecified address, as per RFC 4861
// section 7.2.4.
if err := n.joinGroupLocked(header.IPv6ProtocolNumber, header.IPv6AllNodesMulticastAddress); err != nil {
return err
}
// Perform DAD on the all the unicast IPv6 endpoints that are in the permanent
// state.
//
// Addresses may have aleady completed DAD but in the time since the NIC was
// last enabled, other devices may have acquired the same addresses.
for _, r := range n.mu.endpoints {
addr := r.ep.ID().LocalAddress
if k := r.getKind(); (k != permanent && k != permanentTentative) || !header.IsV6UnicastAddress(addr) {
continue
}
r.setKind(permanentTentative)
if err := n.mu.ndp.startDuplicateAddressDetection(addr, r); err != nil {
return err
}
}
// Do not auto-generate an IPv6 link-local address for loopback devices.
if n.stack.autoGenIPv6LinkLocal && !n.isLoopback() {
// The valid and preferred lifetime is infinite for the auto-generated
// link-local address.
n.mu.ndp.doSLAAC(header.IPv6LinkLocalPrefix.Subnet(), header.NDPInfiniteLifetime, header.NDPInfiniteLifetime)
}
// If we are operating as a router, then do not solicit routers since we
// won't process the RAs anyways.
//
// Routers do not process Router Advertisements (RA) the same way a host
// does. That is, routers do not learn from RAs (e.g. on-link prefixes
// and default routers). Therefore, soliciting RAs from other routers on
// a link is unnecessary for routers.
if !n.stack.forwarding {
n.mu.ndp.startSolicitingRouters()
}
return nil
}
// remove detaches NIC from the link endpoint, and marks existing referenced
// network endpoints expired. This guarantees no packets between this NIC and
// the network stack.
func (n *NIC) remove() *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
// Detach from link endpoint, so no packet comes in.
n.linkEP.Attach(nil)
// Remove permanent and permanentTentative addresses, so no packet goes out.
var errs []*tcpip.Error
for nid, ref := range n.mu.endpoints {
switch ref.getKind() {
case permanentTentative, permanent:
if err := n.removePermanentAddressLocked(nid.LocalAddress); err != nil {
errs = append(errs, err)
}
}
}
if len(errs) > 0 {
return errs[0]
}
return nil
}
// becomeIPv6Router transitions n into an IPv6 router.
//
// When transitioning into an IPv6 router, host-only state (NDP discovered
// routers, discovered on-link prefixes, and auto-generated addresses) will
// be cleaned up/invalidated and NDP router solicitations will be stopped.
func (n *NIC) becomeIPv6Router() {
n.mu.Lock()
defer n.mu.Unlock()
n.mu.ndp.cleanupState(true /* hostOnly */)
n.mu.ndp.stopSolicitingRouters()
}
// becomeIPv6Host transitions n into an IPv6 host.
//
// When transitioning into an IPv6 host, NDP router solicitations will be
// started.
func (n *NIC) becomeIPv6Host() {
n.mu.Lock()
defer n.mu.Unlock()
n.mu.ndp.startSolicitingRouters()
}
// setPromiscuousMode enables or disables promiscuous mode.
func (n *NIC) setPromiscuousMode(enable bool) {
n.mu.Lock()
n.mu.promiscuous = enable
n.mu.Unlock()
}
func (n *NIC) isPromiscuousMode() bool {
n.mu.RLock()
rv := n.mu.promiscuous
n.mu.RUnlock()
return rv
}
func (n *NIC) isLoopback() bool {
return n.linkEP.Capabilities()&CapabilityLoopback != 0
}
// setSpoofing enables or disables address spoofing.
func (n *NIC) setSpoofing(enable bool) {
n.mu.Lock()
n.mu.spoofing = enable
n.mu.Unlock()
}
// primaryEndpoint will return the first non-deprecated endpoint if such an
// endpoint exists for the given protocol and remoteAddr. If no non-deprecated
// endpoint exists, the first deprecated endpoint will be returned.
//
// If an IPv6 primary endpoint is requested, Source Address Selection (as
// defined by RFC 6724 section 5) will be performed.
func (n *NIC) primaryEndpoint(protocol tcpip.NetworkProtocolNumber, remoteAddr tcpip.Address) *referencedNetworkEndpoint {
if protocol == header.IPv6ProtocolNumber && remoteAddr != "" {
return n.primaryIPv6Endpoint(remoteAddr)
}
n.mu.RLock()
defer n.mu.RUnlock()
var deprecatedEndpoint *referencedNetworkEndpoint
for _, r := range n.mu.primary[protocol] {
if !r.isValidForOutgoingRLocked() {
continue
}
if !r.deprecated {
if r.tryIncRef() {
// r is not deprecated, so return it immediately.
//
// If we kept track of a deprecated endpoint, decrement its reference
// count since it was incremented when we decided to keep track of it.
if deprecatedEndpoint != nil {
deprecatedEndpoint.decRefLocked()
deprecatedEndpoint = nil
}
return r
}
} else if deprecatedEndpoint == nil && r.tryIncRef() {
// We prefer an endpoint that is not deprecated, but we keep track of r in
// case n doesn't have any non-deprecated endpoints.
//
// If we end up finding a more preferred endpoint, r's reference count
// will be decremented when such an endpoint is found.
deprecatedEndpoint = r
}
}
// n doesn't have any valid non-deprecated endpoints, so return
// deprecatedEndpoint (which may be nil if n doesn't have any valid deprecated
// endpoints either).
return deprecatedEndpoint
}
// ipv6AddrCandidate is an IPv6 candidate for Source Address Selection (RFC
// 6724 section 5).
type ipv6AddrCandidate struct {
ref *referencedNetworkEndpoint
scope header.IPv6AddressScope
}
// primaryIPv6Endpoint returns an IPv6 endpoint following Source Address
// Selection (RFC 6724 section 5).
//
// Note, only rules 1-3 are followed.
//
// remoteAddr must be a valid IPv6 address.
func (n *NIC) primaryIPv6Endpoint(remoteAddr tcpip.Address) *referencedNetworkEndpoint {
n.mu.RLock()
defer n.mu.RUnlock()
primaryAddrs := n.mu.primary[header.IPv6ProtocolNumber]
if len(primaryAddrs) == 0 {
return nil
}
// Create a candidate set of available addresses we can potentially use as a
// source address.
cs := make([]ipv6AddrCandidate, 0, len(primaryAddrs))
for _, r := range primaryAddrs {
// If r is not valid for outgoing connections, it is not a valid endpoint.
if !r.isValidForOutgoingRLocked() {
continue
}
addr := r.ep.ID().LocalAddress
scope, err := header.ScopeForIPv6Address(addr)
if err != nil {
// Should never happen as we got r from the primary IPv6 endpoint list and
// ScopeForIPv6Address only returns an error if addr is not an IPv6
// address.
log.Fatalf("header.ScopeForIPv6Address(%s): %s", addr, err)
}
cs = append(cs, ipv6AddrCandidate{
ref: r,
scope: scope,
})
}
remoteScope, err := header.ScopeForIPv6Address(remoteAddr)
if err != nil {
// primaryIPv6Endpoint should never be called with an invalid IPv6 address.
log.Fatalf("header.ScopeForIPv6Address(%s): %s", remoteAddr, err)
}
// Sort the addresses as per RFC 6724 section 5 rules 1-3.
//
// TODO(b/146021396): Implement rules 4-8 of RFC 6724 section 5.
sort.Slice(cs, func(i, j int) bool {
sa := cs[i]
sb := cs[j]
// Prefer same address as per RFC 6724 section 5 rule 1.
if sa.ref.ep.ID().LocalAddress == remoteAddr {
return true
}
if sb.ref.ep.ID().LocalAddress == remoteAddr {
return false
}
// Prefer appropriate scope as per RFC 6724 section 5 rule 2.
if sa.scope < sb.scope {
return sa.scope >= remoteScope
} else if sb.scope < sa.scope {
return sb.scope < remoteScope
}
// Avoid deprecated addresses as per RFC 6724 section 5 rule 3.
if saDep, sbDep := sa.ref.deprecated, sb.ref.deprecated; saDep != sbDep {
// If sa is not deprecated, it is preferred over sb.
return sbDep
}
// sa and sb are equal, return the endpoint that is closest to the front of
// the primary endpoint list.
return i < j
})
// Return the most preferred address that can have its reference count
// incremented.
for _, c := range cs {
if r := c.ref; r.tryIncRef() {
return r
}
}
return nil
}
// hasPermanentAddrLocked returns true if n has a permanent (including currently
// tentative) address, addr.
func (n *NIC) hasPermanentAddrLocked(addr tcpip.Address) bool {
ref, ok := n.mu.endpoints[NetworkEndpointID{addr}]
if !ok {
return false
}
kind := ref.getKind()
return kind == permanent || kind == permanentTentative
}
type getRefBehaviour int
const (
// spoofing indicates that the NIC's spoofing flag should be observed when
// getting a NIC's referenced network endpoint.
spoofing getRefBehaviour = iota
// promiscuous indicates that the NIC's promiscuous flag should be observed
// when getting a NIC's referenced network endpoint.
promiscuous
// forceSpoofing indicates that the NIC should be assumed to be spoofing,
// regardless of what the NIC's spoofing flag is when getting a NIC's
// referenced network endpoint.
forceSpoofing
)
func (n *NIC) getRef(protocol tcpip.NetworkProtocolNumber, dst tcpip.Address) *referencedNetworkEndpoint {
return n.getRefOrCreateTemp(protocol, dst, CanBePrimaryEndpoint, promiscuous)
}
// findEndpoint finds the endpoint, if any, with the given address.
func (n *NIC) findEndpoint(protocol tcpip.NetworkProtocolNumber, address tcpip.Address, peb PrimaryEndpointBehavior) *referencedNetworkEndpoint {
return n.getRefOrCreateTemp(protocol, address, peb, spoofing)
}
// getRefEpOrCreateTemp returns the referenced network endpoint for the given
// protocol and address.
//
// If none exists a temporary one may be created if we are in promiscuous mode
// or spoofing. Promiscuous mode will only be checked if promiscuous is true.
// Similarly, spoofing will only be checked if spoofing is true.
func (n *NIC) getRefOrCreateTemp(protocol tcpip.NetworkProtocolNumber, address tcpip.Address, peb PrimaryEndpointBehavior, tempRef getRefBehaviour) *referencedNetworkEndpoint {
id := NetworkEndpointID{address}
n.mu.RLock()
var spoofingOrPromiscuous bool
switch tempRef {
case spoofing:
spoofingOrPromiscuous = n.mu.spoofing
case promiscuous:
spoofingOrPromiscuous = n.mu.promiscuous
case forceSpoofing:
spoofingOrPromiscuous = true
}
if ref, ok := n.mu.endpoints[id]; ok {
// An endpoint with this id exists, check if it can be used and return it.
switch ref.getKind() {
case permanentExpired:
if !spoofingOrPromiscuous {
n.mu.RUnlock()
return nil
}
fallthrough
case temporary, permanent:
if ref.tryIncRef() {
n.mu.RUnlock()
return ref
}
}
}
// A usable reference was not found, create a temporary one if requested by
// the caller or if the address is found in the NIC's subnets.
createTempEP := spoofingOrPromiscuous
if !createTempEP {
for _, sn := range n.mu.addressRanges {
// Skip the subnet address.
if address == sn.ID() {
continue
}
// For now just skip the broadcast address, until we support it.
// FIXME(b/137608825): Add support for sending/receiving directed
// (subnet) broadcast.
if address == sn.Broadcast() {
continue
}
if sn.Contains(address) {
createTempEP = true
break
}
}
}
n.mu.RUnlock()
if !createTempEP {
return nil
}
// Try again with the lock in exclusive mode. If we still can't get the
// endpoint, create a new "temporary" endpoint. It will only exist while
// there's a route through it.
n.mu.Lock()
if ref, ok := n.mu.endpoints[id]; ok {
// No need to check the type as we are ok with expired endpoints at this
// point.
if ref.tryIncRef() {
n.mu.Unlock()
return ref
}
// tryIncRef failing means the endpoint is scheduled to be removed once the
// lock is released. Remove it here so we can create a new (temporary) one.
// The removal logic waiting for the lock handles this case.
n.removeEndpointLocked(ref)
}
// Add a new temporary endpoint.
netProto, ok := n.stack.networkProtocols[protocol]
if !ok {
n.mu.Unlock()
return nil
}
ref, _ := n.addAddressLocked(tcpip.ProtocolAddress{
Protocol: protocol,
AddressWithPrefix: tcpip.AddressWithPrefix{
Address: address,
PrefixLen: netProto.DefaultPrefixLen(),
},
}, peb, temporary, static, false)
n.mu.Unlock()
return ref
}
// addAddressLocked adds a new protocolAddress to n.
//
// If n already has the address in a non-permanent state, and the kind given is
// permanent, that address will be promoted in place and its properties set to
// the properties provided. Otherwise, it returns tcpip.ErrDuplicateAddress.
func (n *NIC) addAddressLocked(protocolAddress tcpip.ProtocolAddress, peb PrimaryEndpointBehavior, kind networkEndpointKind, configType networkEndpointConfigType, deprecated bool) (*referencedNetworkEndpoint, *tcpip.Error) {
// TODO(b/141022673): Validate IP addresses before adding them.
// Sanity check.
id := NetworkEndpointID{LocalAddress: protocolAddress.AddressWithPrefix.Address}
if ref, ok := n.mu.endpoints[id]; ok {
// Endpoint already exists.
if kind != permanent {
return nil, tcpip.ErrDuplicateAddress
}
switch ref.getKind() {
case permanentTentative, permanent:
// The NIC already have a permanent endpoint with that address.
return nil, tcpip.ErrDuplicateAddress
case permanentExpired, temporary:
// Promote the endpoint to become permanent and respect the new peb,
// configType and deprecated status.
if ref.tryIncRef() {
// TODO(b/147748385): Perform Duplicate Address Detection when promoting
// an IPv6 endpoint to permanent.
ref.setKind(permanent)
ref.deprecated = deprecated
ref.configType = configType
refs := n.mu.primary[ref.protocol]
for i, r := range refs {
if r == ref {
switch peb {
case CanBePrimaryEndpoint:
return ref, nil
case FirstPrimaryEndpoint:
if i == 0 {
return ref, nil
}
n.mu.primary[r.protocol] = append(refs[:i], refs[i+1:]...)
case NeverPrimaryEndpoint:
n.mu.primary[r.protocol] = append(refs[:i], refs[i+1:]...)
return ref, nil
}
}
}
n.insertPrimaryEndpointLocked(ref, peb)
return ref, nil
}
// tryIncRef failing means the endpoint is scheduled to be removed once
// the lock is released. Remove it here so we can create a new
// (permanent) one. The removal logic waiting for the lock handles this
// case.
n.removeEndpointLocked(ref)
}
}
netProto, ok := n.stack.networkProtocols[protocolAddress.Protocol]
if !ok {
return nil, tcpip.ErrUnknownProtocol
}
// Create the new network endpoint.
ep, err := netProto.NewEndpoint(n.id, protocolAddress.AddressWithPrefix, n.stack, n, n.linkEP, n.stack)
if err != nil {
return nil, err
}
isIPv6Unicast := protocolAddress.Protocol == header.IPv6ProtocolNumber && header.IsV6UnicastAddress(protocolAddress.AddressWithPrefix.Address)
// If the address is an IPv6 address and it is a permanent address,
// mark it as tentative so it goes through the DAD process if the NIC is
// enabled. If the NIC is not enabled, DAD will be started when the NIC is
// enabled.
if isIPv6Unicast && kind == permanent {
kind = permanentTentative
}
ref := &referencedNetworkEndpoint{
refs: 1,
ep: ep,
nic: n,
protocol: protocolAddress.Protocol,
kind: kind,
configType: configType,
deprecated: deprecated,
}
// Set up cache if link address resolution exists for this protocol.
if n.linkEP.Capabilities()&CapabilityResolutionRequired != 0 {
if _, ok := n.stack.linkAddrResolvers[protocolAddress.Protocol]; ok {
ref.linkCache = n.stack
}
}
// If we are adding an IPv6 unicast address, join the solicited-node
// multicast address.
if isIPv6Unicast {
snmc := header.SolicitedNodeAddr(protocolAddress.AddressWithPrefix.Address)
if err := n.joinGroupLocked(protocolAddress.Protocol, snmc); err != nil {
return nil, err
}
}
n.mu.endpoints[id] = ref
n.insertPrimaryEndpointLocked(ref, peb)
// If we are adding a tentative IPv6 address, start DAD if the NIC is enabled.
if isIPv6Unicast && kind == permanentTentative && n.mu.enabled {
if err := n.mu.ndp.startDuplicateAddressDetection(protocolAddress.AddressWithPrefix.Address, ref); err != nil {
return nil, err
}
}
return ref, nil
}
// AddAddress adds a new address to n, so that it starts accepting packets
// targeted at the given address (and network protocol).
func (n *NIC) AddAddress(protocolAddress tcpip.ProtocolAddress, peb PrimaryEndpointBehavior) *tcpip.Error {
// Add the endpoint.
n.mu.Lock()
_, err := n.addAddressLocked(protocolAddress, peb, permanent, static, false /* deprecated */)
n.mu.Unlock()
return err
}
// AllAddresses returns all addresses (primary and non-primary) associated with
// this NIC.
func (n *NIC) AllAddresses() []tcpip.ProtocolAddress {
n.mu.RLock()
defer n.mu.RUnlock()
addrs := make([]tcpip.ProtocolAddress, 0, len(n.mu.endpoints))
for nid, ref := range n.mu.endpoints {
// Don't include tentative, expired or temporary endpoints to
// avoid confusion and prevent the caller from using those.
switch ref.getKind() {
case permanentExpired, temporary:
continue
}
addrs = append(addrs, tcpip.ProtocolAddress{
Protocol: ref.protocol,
AddressWithPrefix: tcpip.AddressWithPrefix{
Address: nid.LocalAddress,
PrefixLen: ref.ep.PrefixLen(),
},
})
}
return addrs
}
// PrimaryAddresses returns the primary addresses associated with this NIC.
func (n *NIC) PrimaryAddresses() []tcpip.ProtocolAddress {
n.mu.RLock()
defer n.mu.RUnlock()
var addrs []tcpip.ProtocolAddress
for proto, list := range n.mu.primary {
for _, ref := range list {
// Don't include tentative, expired or tempory endpoints
// to avoid confusion and prevent the caller from using
// those.
switch ref.getKind() {
case permanentTentative, permanentExpired, temporary:
continue
}
addrs = append(addrs, tcpip.ProtocolAddress{
Protocol: proto,
AddressWithPrefix: tcpip.AddressWithPrefix{
Address: ref.ep.ID().LocalAddress,
PrefixLen: ref.ep.PrefixLen(),
},
})
}
}
return addrs
}
// primaryAddress returns the primary address associated with this NIC.
//
// primaryAddress will return the first non-deprecated address if such an
// address exists. If no non-deprecated address exists, the first deprecated
// address will be returned.
func (n *NIC) primaryAddress(proto tcpip.NetworkProtocolNumber) tcpip.AddressWithPrefix {
n.mu.RLock()
defer n.mu.RUnlock()
list, ok := n.mu.primary[proto]
if !ok {
return tcpip.AddressWithPrefix{}
}
var deprecatedEndpoint *referencedNetworkEndpoint
for _, ref := range list {
// Don't include tentative, expired or tempory endpoints to avoid confusion
// and prevent the caller from using those.
switch ref.getKind() {
case permanentTentative, permanentExpired, temporary:
continue
}
if !ref.deprecated {
return tcpip.AddressWithPrefix{
Address: ref.ep.ID().LocalAddress,
PrefixLen: ref.ep.PrefixLen(),
}
}
if deprecatedEndpoint == nil {
deprecatedEndpoint = ref
}
}
if deprecatedEndpoint != nil {
return tcpip.AddressWithPrefix{
Address: deprecatedEndpoint.ep.ID().LocalAddress,
PrefixLen: deprecatedEndpoint.ep.PrefixLen(),
}
}
return tcpip.AddressWithPrefix{}
}
// AddAddressRange adds a range of addresses to n, so that it starts accepting
// packets targeted at the given addresses and network protocol. The range is
// given by a subnet address, and all addresses contained in the subnet are
// used except for the subnet address itself and the subnet's broadcast
// address.
func (n *NIC) AddAddressRange(protocol tcpip.NetworkProtocolNumber, subnet tcpip.Subnet) {
n.mu.Lock()
n.mu.addressRanges = append(n.mu.addressRanges, subnet)
n.mu.Unlock()
}
// RemoveAddressRange removes the given address range from n.
func (n *NIC) RemoveAddressRange(subnet tcpip.Subnet) {
n.mu.Lock()
// Use the same underlying array.
tmp := n.mu.addressRanges[:0]
for _, sub := range n.mu.addressRanges {
if sub != subnet {
tmp = append(tmp, sub)
}
}
n.mu.addressRanges = tmp
n.mu.Unlock()
}
// AddressRanges returns the Subnets associated with this NIC.
func (n *NIC) AddressRanges() []tcpip.Subnet {
n.mu.RLock()
defer n.mu.RUnlock()
sns := make([]tcpip.Subnet, 0, len(n.mu.addressRanges)+len(n.mu.endpoints))
for nid := range n.mu.endpoints {
sn, err := tcpip.NewSubnet(nid.LocalAddress, tcpip.AddressMask(strings.Repeat("\xff", len(nid.LocalAddress))))
if err != nil {
// This should never happen as the mask has been carefully crafted to
// match the address.
panic("Invalid endpoint subnet: " + err.Error())
}
sns = append(sns, sn)
}
return append(sns, n.mu.addressRanges...)
}
// insertPrimaryEndpointLocked adds r to n's primary endpoint list as required
// by peb.
//
// n MUST be locked.
func (n *NIC) insertPrimaryEndpointLocked(r *referencedNetworkEndpoint, peb PrimaryEndpointBehavior) {
switch peb {
case CanBePrimaryEndpoint:
n.mu.primary[r.protocol] = append(n.mu.primary[r.protocol], r)
case FirstPrimaryEndpoint:
n.mu.primary[r.protocol] = append([]*referencedNetworkEndpoint{r}, n.mu.primary[r.protocol]...)
}
}
func (n *NIC) removeEndpointLocked(r *referencedNetworkEndpoint) {
id := *r.ep.ID()
// Nothing to do if the reference has already been replaced with a different
// one. This happens in the case where 1) this endpoint's ref count hit zero
// and was waiting (on the lock) to be removed and 2) the same address was
// re-added in the meantime by removing this endpoint from the list and
// adding a new one.
if n.mu.endpoints[id] != r {
return
}
if r.getKind() == permanent {
panic("Reference count dropped to zero before being removed")
}
delete(n.mu.endpoints, id)
refs := n.mu.primary[r.protocol]
for i, ref := range refs {
if ref == r {
n.mu.primary[r.protocol] = append(refs[:i], refs[i+1:]...)
break
}
}
r.ep.Close()
}
func (n *NIC) removeEndpoint(r *referencedNetworkEndpoint) {
n.mu.Lock()
n.removeEndpointLocked(r)
n.mu.Unlock()
}
func (n *NIC) removePermanentAddressLocked(addr tcpip.Address) *tcpip.Error {
r, ok := n.mu.endpoints[NetworkEndpointID{addr}]
if !ok {
return tcpip.ErrBadLocalAddress
}
kind := r.getKind()
if kind != permanent && kind != permanentTentative {
return tcpip.ErrBadLocalAddress
}
isIPv6Unicast := r.protocol == header.IPv6ProtocolNumber && header.IsV6UnicastAddress(addr)
if isIPv6Unicast {
// If we are removing a tentative IPv6 unicast address, stop
// DAD.
if kind == permanentTentative {
n.mu.ndp.stopDuplicateAddressDetection(addr)
}
// If we are removing an address generated via SLAAC, cleanup
// its SLAAC resources and notify the integrator.
if r.configType == slaac {
n.mu.ndp.cleanupAutoGenAddrResourcesAndNotify(addr)
}
}
r.setKind(permanentExpired)
if !r.decRefLocked() {
// The endpoint still has references to it.
return nil
}
// At this point the endpoint is deleted.
// If we are removing an IPv6 unicast address, leave the solicited-node
// multicast address.
if isIPv6Unicast {
snmc := header.SolicitedNodeAddr(addr)
if err := n.leaveGroupLocked(snmc); err != nil {
return err
}
}
return nil
}
// RemoveAddress removes an address from n.
func (n *NIC) RemoveAddress(addr tcpip.Address) *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
return n.removePermanentAddressLocked(addr)
}
// joinGroup adds a new endpoint for the given multicast address, if none
// exists yet. Otherwise it just increments its count.
func (n *NIC) joinGroup(protocol tcpip.NetworkProtocolNumber, addr tcpip.Address) *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
return n.joinGroupLocked(protocol, addr)
}
// joinGroupLocked adds a new endpoint for the given multicast address, if none
// exists yet. Otherwise it just increments its count. n MUST be locked before
// joinGroupLocked is called.
func (n *NIC) joinGroupLocked(protocol tcpip.NetworkProtocolNumber, addr tcpip.Address) *tcpip.Error {
// TODO(b/143102137): When implementing MLD, make sure MLD packets are
// not sent unless a valid link-local address is available for use on n
// as an MLD packet's source address must be a link-local address as
// outlined in RFC 3810 section 5.
id := NetworkEndpointID{addr}
joins := n.mu.mcastJoins[id]
if joins == 0 {
netProto, ok := n.stack.networkProtocols[protocol]
if !ok {
return tcpip.ErrUnknownProtocol
}
if _, err := n.addAddressLocked(tcpip.ProtocolAddress{
Protocol: protocol,
AddressWithPrefix: tcpip.AddressWithPrefix{
Address: addr,
PrefixLen: netProto.DefaultPrefixLen(),
},
}, NeverPrimaryEndpoint, permanent, static, false /* deprecated */); err != nil {
return err
}
}
n.mu.mcastJoins[id] = joins + 1
return nil
}
// leaveGroup decrements the count for the given multicast address, and when it
// reaches zero removes the endpoint for this address.
func (n *NIC) leaveGroup(addr tcpip.Address) *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
return n.leaveGroupLocked(addr)
}
// leaveGroupLocked decrements the count for the given multicast address, and
// when it reaches zero removes the endpoint for this address. n MUST be locked
// before leaveGroupLocked is called.
func (n *NIC) leaveGroupLocked(addr tcpip.Address) *tcpip.Error {
id := NetworkEndpointID{addr}
joins := n.mu.mcastJoins[id]
switch joins {
case 0:
// There are no joins with this address on this NIC.
return tcpip.ErrBadLocalAddress
case 1:
// This is the last one, clean up.
if err := n.removePermanentAddressLocked(addr); err != nil {
return err
}
}
n.mu.mcastJoins[id] = joins - 1
return nil
}
// isInGroup returns true if n has joined the multicast group addr.
func (n *NIC) isInGroup(addr tcpip.Address) bool {
n.mu.RLock()
joins := n.mu.mcastJoins[NetworkEndpointID{addr}]
n.mu.RUnlock()
return joins != 0
}
func handlePacket(protocol tcpip.NetworkProtocolNumber, dst, src tcpip.Address, localLinkAddr, remotelinkAddr tcpip.LinkAddress, ref *referencedNetworkEndpoint, pkt tcpip.PacketBuffer) {
r := makeRoute(protocol, dst, src, localLinkAddr, ref, false /* handleLocal */, false /* multicastLoop */)
r.RemoteLinkAddress = remotelinkAddr
ref.ep.HandlePacket(&r, pkt)
ref.decRef()
}
// DeliverNetworkPacket finds the appropriate network protocol endpoint and
// hands the packet over for further processing. This function is called when
// the NIC receives a packet from the link endpoint.
// Note that the ownership of the slice backing vv is retained by the caller.
// This rule applies only to the slice itself, not to the items of the slice;
// the ownership of the items is not retained by the caller.
func (n *NIC) DeliverNetworkPacket(linkEP LinkEndpoint, remote, local tcpip.LinkAddress, protocol tcpip.NetworkProtocolNumber, pkt tcpip.PacketBuffer) {
n.mu.RLock()
enabled := n.mu.enabled
// If the NIC is not yet enabled, don't receive any packets.
if !enabled {
n.mu.RUnlock()
n.stats.DisabledRx.Packets.Increment()
n.stats.DisabledRx.Bytes.IncrementBy(uint64(pkt.Data.Size()))
return
}
n.stats.Rx.Packets.Increment()
n.stats.Rx.Bytes.IncrementBy(uint64(pkt.Data.Size()))
netProto, ok := n.stack.networkProtocols[protocol]
if !ok {
n.mu.RUnlock()
n.stack.stats.UnknownProtocolRcvdPackets.Increment()
return
}
// If no local link layer address is provided, assume it was sent
// directly to this NIC.
if local == "" {
local = n.linkEP.LinkAddress()
}
// Are any packet sockets listening for this network protocol?
packetEPs := n.mu.packetEPs[protocol]
// Check whether there are packet sockets listening for every protocol.
// If we received a packet with protocol EthernetProtocolAll, then the
// previous for loop will have handled it.
if protocol != header.EthernetProtocolAll {
packetEPs = append(packetEPs, n.mu.packetEPs[header.EthernetProtocolAll]...)
}
n.mu.RUnlock()
for _, ep := range packetEPs {
ep.HandlePacket(n.id, local, protocol, pkt.Clone())
}
if netProto.Number() == header.IPv4ProtocolNumber || netProto.Number() == header.IPv6ProtocolNumber {
n.stack.stats.IP.PacketsReceived.Increment()
}
if len(pkt.Data.First()) < netProto.MinimumPacketSize() {
n.stack.stats.MalformedRcvdPackets.Increment()
return
}
src, dst := netProto.ParseAddresses(pkt.Data.First())
if n.stack.handleLocal && !n.isLoopback() && n.getRef(protocol, src) != nil {
// The source address is one of our own, so we never should have gotten a
// packet like this unless handleLocal is false. Loopback also calls this
// function even though the packets didn't come from the physical interface
// so don't drop those.
n.stack.stats.IP.InvalidSourceAddressesReceived.Increment()
return
}
if ref := n.getRef(protocol, dst); ref != nil {
handlePacket(protocol, dst, src, linkEP.LinkAddress(), remote, ref, pkt)
return
}
// This NIC doesn't care about the packet. Find a NIC that cares about the
// packet and forward it to the NIC.
//
// TODO: Should we be forwarding the packet even if promiscuous?
if n.stack.Forwarding() {
r, err := n.stack.FindRoute(0, "", dst, protocol, false /* multicastLoop */)
if err != nil {
n.stack.stats.IP.InvalidDestinationAddressesReceived.Increment()
return
}
defer r.Release()
r.LocalLinkAddress = n.linkEP.LinkAddress()
r.RemoteLinkAddress = remote
// Found a NIC.
n := r.ref.nic
n.mu.RLock()
ref, ok := n.mu.endpoints[NetworkEndpointID{dst}]
ok = ok && ref.isValidForOutgoingRLocked() && ref.tryIncRef()
n.mu.RUnlock()
if ok {
r.RemoteAddress = src
// TODO(b/123449044): Update the source NIC as well.
ref.ep.HandlePacket(&r, pkt)
ref.decRef()
} else {
// n doesn't have a destination endpoint.
// Send the packet out of n.
pkt.Header = buffer.NewPrependableFromView(pkt.Data.First())
pkt.Data.RemoveFirst()
// TODO(b/128629022): use route.WritePacket.
if err := n.linkEP.WritePacket(&r, nil /* gso */, protocol, pkt); err != nil {
r.Stats().IP.OutgoingPacketErrors.Increment()
} else {
n.stats.Tx.Packets.Increment()
n.stats.Tx.Bytes.IncrementBy(uint64(pkt.Header.UsedLength() + pkt.Data.Size()))
}
}
return
}
// If a packet socket handled the packet, don't treat it as invalid.
if len(packetEPs) == 0 {
n.stack.stats.IP.InvalidDestinationAddressesReceived.Increment()
}
}
// DeliverTransportPacket delivers the packets to the appropriate transport
// protocol endpoint.
func (n *NIC) DeliverTransportPacket(r *Route, protocol tcpip.TransportProtocolNumber, pkt tcpip.PacketBuffer) {
state, ok := n.stack.transportProtocols[protocol]
if !ok {
n.stack.stats.UnknownProtocolRcvdPackets.Increment()
return
}
transProto := state.proto
// Raw socket packets are delivered based solely on the transport
// protocol number. We do not inspect the payload to ensure it's
// validly formed.
n.stack.demux.deliverRawPacket(r, protocol, pkt)
if len(pkt.Data.First()) < transProto.MinimumPacketSize() {
n.stack.stats.MalformedRcvdPackets.Increment()
return
}
srcPort, dstPort, err := transProto.ParsePorts(pkt.Data.First())
if err != nil {
n.stack.stats.MalformedRcvdPackets.Increment()
return
}
id := TransportEndpointID{dstPort, r.LocalAddress, srcPort, r.RemoteAddress}
if n.stack.demux.deliverPacket(r, protocol, pkt, id) {
return
}
// Try to deliver to per-stack default handler.
if state.defaultHandler != nil {
if state.defaultHandler(r, id, pkt) {
return
}
}
// We could not find an appropriate destination for this packet, so
// deliver it to the global handler.
if !transProto.HandleUnknownDestinationPacket(r, id, pkt) {
n.stack.stats.MalformedRcvdPackets.Increment()
}
}
// DeliverTransportControlPacket delivers control packets to the appropriate
// transport protocol endpoint.
func (n *NIC) DeliverTransportControlPacket(local, remote tcpip.Address, net tcpip.NetworkProtocolNumber, trans tcpip.TransportProtocolNumber, typ ControlType, extra uint32, pkt tcpip.PacketBuffer) {
state, ok := n.stack.transportProtocols[trans]
if !ok {
return
}
transProto := state.proto
// ICMPv4 only guarantees that 8 bytes of the transport protocol will
// be present in the payload. We know that the ports are within the
// first 8 bytes for all known transport protocols.
if len(pkt.Data.First()) < 8 {
return
}
srcPort, dstPort, err := transProto.ParsePorts(pkt.Data.First())
if err != nil {
return
}
id := TransportEndpointID{srcPort, local, dstPort, remote}
if n.stack.demux.deliverControlPacket(n, net, trans, typ, extra, pkt, id) {
return
}
}
// ID returns the identifier of n.
func (n *NIC) ID() tcpip.NICID {
return n.id
}
// Name returns the name of n.
func (n *NIC) Name() string {
return n.name
}
// Stack returns the instance of the Stack that owns this NIC.
func (n *NIC) Stack() *Stack {
return n.stack
}
// LinkEndpoint returns the link endpoint of n.
func (n *NIC) LinkEndpoint() LinkEndpoint {
return n.linkEP
}
// isAddrTentative returns true if addr is tentative on n.
//
// Note that if addr is not associated with n, then this function will return
// false. It will only return true if the address is associated with the NIC
// AND it is tentative.
func (n *NIC) isAddrTentative(addr tcpip.Address) bool {
n.mu.RLock()
defer n.mu.RUnlock()
ref, ok := n.mu.endpoints[NetworkEndpointID{addr}]
if !ok {
return false
}
return ref.getKind() == permanentTentative
}
// dupTentativeAddrDetected attempts to inform n that a tentative addr
// is a duplicate on a link.
//
// dupTentativeAddrDetected will delete the tentative address if it exists.
func (n *NIC) dupTentativeAddrDetected(addr tcpip.Address) *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
ref, ok := n.mu.endpoints[NetworkEndpointID{addr}]
if !ok {
return tcpip.ErrBadAddress
}
if ref.getKind() != permanentTentative {
return tcpip.ErrInvalidEndpointState
}
return n.removePermanentAddressLocked(addr)
}
// setNDPConfigs sets the NDP configurations for n.
//
// Note, if c contains invalid NDP configuration values, it will be fixed to
// use default values for the erroneous values.
func (n *NIC) setNDPConfigs(c NDPConfigurations) {
c.validate()
n.mu.Lock()
n.mu.ndp.configs = c
n.mu.Unlock()
}
// handleNDPRA handles an NDP Router Advertisement message that arrived on n.
func (n *NIC) handleNDPRA(ip tcpip.Address, ra header.NDPRouterAdvert) {
n.mu.Lock()
defer n.mu.Unlock()
n.mu.ndp.handleRA(ip, ra)
}
type networkEndpointKind int32
const (
// A permanentTentative endpoint is a permanent address that is not yet
// considered to be fully bound to an interface in the traditional
// sense. That is, the address is associated with a NIC, but packets
// destined to the address MUST NOT be accepted and MUST be silently
// dropped, and the address MUST NOT be used as a source address for
// outgoing packets. For IPv6, addresses will be of this kind until
// NDP's Duplicate Address Detection has resolved, or be deleted if
// the process results in detecting a duplicate address.
permanentTentative networkEndpointKind = iota
// A permanent endpoint is created by adding a permanent address (vs. a
// temporary one) to the NIC. Its reference count is biased by 1 to avoid
// removal when no route holds a reference to it. It is removed by explicitly
// removing the permanent address from the NIC.
permanent
// An expired permanent endpoint is a permanent endpoint that had its address
// removed from the NIC, and it is waiting to be removed once no more routes
// hold a reference to it. This is achieved by decreasing its reference count
// by 1. If its address is re-added before the endpoint is removed, its type
// changes back to permanent and its reference count increases by 1 again.
permanentExpired
// A temporary endpoint is created for spoofing outgoing packets, or when in
// promiscuous mode and accepting incoming packets that don't match any
// permanent endpoint. Its reference count is not biased by 1 and the
// endpoint is removed immediately when no more route holds a reference to
// it. A temporary endpoint can be promoted to permanent if its address
// is added permanently.
temporary
)
func (n *NIC) registerPacketEndpoint(netProto tcpip.NetworkProtocolNumber, ep PacketEndpoint) *tcpip.Error {
n.mu.Lock()
defer n.mu.Unlock()
eps, ok := n.mu.packetEPs[netProto]
if !ok {
return tcpip.ErrNotSupported
}
n.mu.packetEPs[netProto] = append(eps, ep)
return nil
}
func (n *NIC) unregisterPacketEndpoint(netProto tcpip.NetworkProtocolNumber, ep PacketEndpoint) {
n.mu.Lock()
defer n.mu.Unlock()
eps, ok := n.mu.packetEPs[netProto]
if !ok {
return
}
for i, epOther := range eps {
if epOther == ep {
n.mu.packetEPs[netProto] = append(eps[:i], eps[i+1:]...)
return
}
}
}
type networkEndpointConfigType int32
const (
// A statically configured endpoint is an address that was added by
// some user-specified action (adding an explicit address, joining a
// multicast group).
static networkEndpointConfigType = iota
// A slaac configured endpoint is an IPv6 endpoint that was
// added by SLAAC as per RFC 4862 section 5.5.3.
slaac
)
type referencedNetworkEndpoint struct {
ep NetworkEndpoint
nic *NIC
protocol tcpip.NetworkProtocolNumber
// linkCache is set if link address resolution is enabled for this
// protocol. Set to nil otherwise.
linkCache LinkAddressCache
// refs is counting references held for this endpoint. When refs hits zero it
// triggers the automatic removal of the endpoint from the NIC.
refs int32
// networkEndpointKind must only be accessed using {get,set}Kind().
kind networkEndpointKind
// configType is the method that was used to configure this endpoint.
// This must never change except during endpoint creation and promotion to
// permanent.
configType networkEndpointConfigType
// deprecated indicates whether or not the endpoint should be considered
// deprecated. That is, when deprecated is true, other endpoints that are not
// deprecated should be preferred.
deprecated bool
}
func (r *referencedNetworkEndpoint) getKind() networkEndpointKind {
return networkEndpointKind(atomic.LoadInt32((*int32)(&r.kind)))
}
func (r *referencedNetworkEndpoint) setKind(kind networkEndpointKind) {
atomic.StoreInt32((*int32)(&r.kind), int32(kind))
}
// isValidForOutgoing returns true if the endpoint can be used to send out a
// packet. It requires the endpoint to not be marked expired (i.e., its address
// has been removed), or the NIC to be in spoofing mode.
func (r *referencedNetworkEndpoint) isValidForOutgoing() bool {
r.nic.mu.RLock()
defer r.nic.mu.RUnlock()
return r.isValidForOutgoingRLocked()
}
// isValidForOutgoingRLocked returns true if the endpoint can be used to send
// out a packet. It requires the endpoint to not be marked expired (i.e., its
// address has been removed), or the NIC to be in spoofing mode.
//
// r's NIC must be read locked.
func (r *referencedNetworkEndpoint) isValidForOutgoingRLocked() bool {
return r.nic.mu.enabled && (r.getKind() != permanentExpired || r.nic.mu.spoofing)
}
// decRef decrements the ref count and cleans up the endpoint once it reaches
// zero.
func (r *referencedNetworkEndpoint) decRef() {
if atomic.AddInt32(&r.refs, -1) == 0 {
r.nic.removeEndpoint(r)
}
}
// decRefLocked is the same as decRef but assumes that the NIC.mu mutex is
// locked. Returns true if the endpoint was removed.
func (r *referencedNetworkEndpoint) decRefLocked() bool {
if atomic.AddInt32(&r.refs, -1) == 0 {
r.nic.removeEndpointLocked(r)
return true
}
return false
}
// incRef increments the ref count. It must only be called when the caller is
// known to be holding a reference to the endpoint, otherwise tryIncRef should
// be used.
func (r *referencedNetworkEndpoint) incRef() {
atomic.AddInt32(&r.refs, 1)
}
// tryIncRef attempts to increment the ref count from n to n+1, but only if n is
// not zero. That is, it will increment the count if the endpoint is still
// alive, and do nothing if it has already been clean up.
func (r *referencedNetworkEndpoint) tryIncRef() bool {
for {
v := atomic.LoadInt32(&r.refs)
if v == 0 {
return false
}
if atomic.CompareAndSwapInt32(&r.refs, v, v+1) {
return true
}
}
}
// stack returns the Stack instance that owns the underlying endpoint.
func (r *referencedNetworkEndpoint) stack() *Stack {
return r.nic.stack
}
|