aboutsummaryrefslogtreecommitdiffstats
path: root/src/vnet/adj/adj_nbr.c
blob: d289bb6a2a880e08aa9e2d4fccfbc1280bd8429b (plain)
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
/*
 * Copyright (c) 2016 Cisco and/or its affiliates.
 * 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.
 */

#include <vnet/adj/adj_nbr.h>
#include <vnet/adj/adj_internal.h>
#include <vnet/ethernet/arp_packet.h>
#include <vnet/fib/fib_walk.h>

/*
 * Vector Hash tables of neighbour (traditional) adjacencies
 *  Key: interface(for the vector index), address (and its proto),
 *       link-type/ether-type.
 */
static uword **adj_nbr_tables[FIB_PROTOCOL_IP_MAX];

typedef struct adj_nbr_key_t_
{
    ip46_address_t ank_ip;
    u64 ank_linkt;
} adj_nbr_key_t;

#define ADJ_NBR_SET_KEY(_key, _lt, _nh)         \
{						\
    ip46_address_copy(&(_key).ank_ip, (_nh));   \
    _key.ank_linkt = (_lt);			\
}

#define ADJ_NBR_ITF_OK(_proto, _itf)			\
    (((_itf) < vec_len(adj_nbr_tables[_proto])) &&	\
     (NULL != adj_nbr_tables[_proto][(_itf)]))

#define ADJ_NBR_ASSERT_NH_PROTO(nh_proto, err)          \
  do {                                                  \
      ASSERT (nh_proto < FIB_PROTOCOL_IP_MAX);          \
      const fib_protocol_t nh_proto__ = (nh_proto);     \
      if (nh_proto__ >= FIB_PROTOCOL_IP_MAX)            \
        {                                               \
          clib_warning ("BUG: protocol %d > %d\n",      \
                        (int)nh_proto__,                \
                        FIB_PROTOCOL_IP_MAX);           \
          return err;                                   \
        }                                               \
  } while (0)

static void
adj_nbr_insert (fib_protocol_t nh_proto,
		vnet_link_t link_type,
		const ip46_address_t *nh_addr,
		u32 sw_if_index,
		adj_index_t adj_index)
{
    adj_nbr_key_t kv;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto,);

    if (sw_if_index >= vec_len(adj_nbr_tables[nh_proto]))
    {
	vec_validate(adj_nbr_tables[nh_proto], sw_if_index);
    }
    if (NULL == adj_nbr_tables[nh_proto][sw_if_index])
    {
	adj_nbr_tables[nh_proto][sw_if_index] =
	    hash_create_mem(0, sizeof(adj_nbr_key_t), sizeof(adj_index_t));
    }

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    hash_set_mem_alloc (&adj_nbr_tables[nh_proto][sw_if_index],
                        &kv, adj_index);
}

void
adj_nbr_remove (adj_index_t ai,
                fib_protocol_t nh_proto,
		vnet_link_t link_type,
		const ip46_address_t *nh_addr,
		u32 sw_if_index)
{
    adj_nbr_key_t kv;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto,);

    if (!ADJ_NBR_ITF_OK(nh_proto, sw_if_index))
	return;

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    hash_unset_mem_free(&adj_nbr_tables[nh_proto][sw_if_index], &kv);

    if (0 == hash_elts(adj_nbr_tables[nh_proto][sw_if_index]))
    {
        hash_free(adj_nbr_tables[nh_proto][sw_if_index]);
    }
}

typedef struct adj_nbr_get_n_adjs_walk_ctx_t_
{
    vnet_link_t linkt;
    u32 count;
} adj_nbr_get_n_adjs_walk_ctx_t;

static adj_walk_rc_t
adj_nbr_get_n_adjs_walk (adj_index_t ai,
                         void *data)
{
    adj_nbr_get_n_adjs_walk_ctx_t *ctx = data;
    const ip_adjacency_t *adj;

    adj = adj_get(ai);

    if (ctx->linkt == adj->ia_link)
        ctx->count++;

    return (ADJ_WALK_RC_CONTINUE);
}

u32
adj_nbr_get_n_adjs (vnet_link_t link_type, u32 sw_if_index)
{
    adj_nbr_get_n_adjs_walk_ctx_t ctx = {
        .linkt = link_type,
    };
    fib_protocol_t fproto;

    FOR_EACH_FIB_IP_PROTOCOL(fproto)
    {
        adj_nbr_walk (sw_if_index,
                      fproto,
                      adj_nbr_get_n_adjs_walk,
                      &ctx);
    }

    return (ctx.count);
}

adj_index_t
adj_nbr_find (fib_protocol_t nh_proto,
	      vnet_link_t link_type,
	      const ip46_address_t *nh_addr,
	      u32 sw_if_index)
{
    adj_nbr_key_t kv;
    uword *p;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto, ADJ_INDEX_INVALID);

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    if (!ADJ_NBR_ITF_OK(nh_proto, sw_if_index))
	return (ADJ_INDEX_INVALID);

    p = hash_get_mem(adj_nbr_tables[nh_proto][sw_if_index], &kv);

    if (p)
    {
	return (p[0]);
    }
    return (ADJ_INDEX_INVALID);
}

static inline u32
adj_get_nd_node (fib_protocol_t proto)
{
    switch (proto) {
    case FIB_PROTOCOL_IP4:
	return (ip4_arp_node.index);
    case FIB_PROTOCOL_IP6:
	return (ip6_discover_neighbor_node.index);
    case FIB_PROTOCOL_MPLS:
	break;
    }
    ASSERT(0);
    return (ip4_arp_node.index);
}

/**
 * @brief Check and set feature flags if o/p interface has any o/p features.
 */
static void
adj_nbr_evaluate_feature (adj_index_t ai)
{
    ip_adjacency_t *adj;
    vnet_feature_main_t *fm = &feature_main;
    i16 feature_count;
    u8 arc_index;
    u32 sw_if_index;

    adj = adj_get(ai);

    switch (adj->ia_link)
    {
    case VNET_LINK_IP4:
        arc_index = ip4_main.lookup_main.output_feature_arc_index;
        break;
    case VNET_LINK_IP6:
        arc_index = ip6_main.lookup_main.output_feature_arc_index;
        break;
    case VNET_LINK_MPLS:
        arc_index = mpls_main.output_feature_arc_index;
        break;
    default:
        return;
    }

    sw_if_index = adj->rewrite_header.sw_if_index;
    if (vec_len(fm->feature_count_by_sw_if_index[arc_index]) > sw_if_index)
    {
        feature_count = fm->feature_count_by_sw_if_index[arc_index][sw_if_index];
        if (feature_count > 0)
        {
            vnet_feature_config_main_t *cm;

            adj->rewrite_header.flags |= VNET_REWRITE_HAS_FEATURES;
            cm = &fm->feature_config_mains[arc_index];

            adj->ia_cfg_index = vec_elt (cm->config_index_by_sw_if_index,
                                         sw_if_index);
        }
    }
    return;
}

static ip_adjacency_t*
adj_nbr_alloc (fib_protocol_t nh_proto,
	       vnet_link_t link_type,
	       const ip46_address_t *nh_addr,
	       u32 sw_if_index)
{
    ip_adjacency_t *adj;

    adj = adj_alloc(nh_proto);

    adj_nbr_insert(nh_proto, link_type, nh_addr,
		   sw_if_index,
		   adj_get_index(adj));

    /*
     * since we just added the ADJ we have no rewrite string for it,
     * so its for ARP
     */
    adj->lookup_next_index = IP_LOOKUP_NEXT_ARP;
    adj->sub_type.nbr.next_hop = *nh_addr;
    adj->ia_link = link_type;
    adj->ia_nh_proto = nh_proto;
    adj->rewrite_header.sw_if_index = sw_if_index;
    vnet_rewrite_update_mtu(vnet_get_main(), adj->ia_link,
                            &adj->rewrite_header);

    adj_nbr_evaluate_feature (adj_get_index(adj));
    return (adj);
}

void
adj_nbr_set_mtu (adj_index_t adj_index, u16 mtu)
{
    ip_adjacency_t *adj;

    ASSERT(ADJ_INDEX_INVALID != adj_index);

    adj = adj_get(adj_index);

    if (0 == mtu)
        vnet_rewrite_update_mtu(vnet_get_main(), adj->ia_link,
                                &adj->rewrite_header);
    else
    {
        vnet_rewrite_update_mtu(vnet_get_main(), adj->ia_link,
                                &adj->rewrite_header);
        adj->rewrite_header.max_l3_packet_bytes =
            clib_min (adj->rewrite_header.max_l3_packet_bytes, mtu);
    }
}

/*
 * adj_nbr_add_or_lock
 *
 * Add an adjacency for the neighbour requested.
 *
 * The key for an adj is:
 *   - the Next-hops protocol (i.e. v4 or v6)
 *   - the address of the next-hop
 *   - the interface the next-hop is reachable through
 */
adj_index_t
adj_nbr_add_or_lock (fib_protocol_t nh_proto,
		     vnet_link_t link_type,
		     const ip46_address_t *nh_addr,
		     u32 sw_if_index)
{
    adj_index_t adj_index;

    adj_index = adj_nbr_find(nh_proto, link_type, nh_addr, sw_if_index);

    if (ADJ_INDEX_INVALID == adj_index)
    {
        ip_adjacency_t *adj;
	vnet_main_t *vnm;

	vnm = vnet_get_main();
	adj = adj_nbr_alloc(nh_proto, link_type, nh_addr, sw_if_index);
	adj_index = adj_get_index(adj);
	adj_lock(adj_index);

        if (ip46_address_is_equal(&ADJ_BCAST_ADDR, nh_addr))
        {
            adj->lookup_next_index = IP_LOOKUP_NEXT_BCAST;
        }

	vnet_rewrite_init(vnm, sw_if_index, link_type,
			  adj_get_nd_node(nh_proto),
			  vnet_tx_node_index_for_sw_interface(vnm, sw_if_index),
			  &adj->rewrite_header);

	/*
	 * we need a rewrite where the destination IP address is converted
	 * to the appropriate link-layer address. This is interface specific.
	 * So ask the interface to do it.
	 */
	vnet_update_adjacency_for_sw_interface(vnm, sw_if_index, adj_index);
        adj_delegate_adj_created(adj_get(adj_index));
    }
    else
    {
	adj_lock(adj_index);
    }

    return (adj_index);
}

adj_index_t
adj_nbr_add_or_lock_w_rewrite (fib_protocol_t nh_proto,
			       vnet_link_t link_type,
			       const ip46_address_t *nh_addr,
			       u32 sw_if_index,
			       u8 *rewrite)
{
    adj_index_t adj_index;

    adj_index = adj_nbr_find(nh_proto, link_type, nh_addr, sw_if_index);

    if (ADJ_INDEX_INVALID == adj_index)
    {
        ip_adjacency_t *adj;

        adj = adj_nbr_alloc(nh_proto, link_type, nh_addr, sw_if_index);
	adj->rewrite_header.sw_if_index = sw_if_index;
        adj_index = adj_get_index(adj);
    }

    adj_lock(adj_index);
    adj_nbr_update_rewrite(adj_index,
			   ADJ_NBR_REWRITE_FLAG_COMPLETE,
			   rewrite);

    adj_delegate_adj_created(adj_get(adj_index));

    return (adj_index);
}

/**
 * adj_nbr_update_rewrite
 *
 * Update the adjacency's rewrite string. A NULL string implies the
 * rewrite is reset (i.e. when ARP/ND entry is gone).
 * NB: the adj being updated may be handling traffic in the DP.
 */
void
adj_nbr_update_rewrite (adj_index_t adj_index,
			adj_nbr_rewrite_flag_t flags,
			u8 *rewrite)
{
    ip_adjacency_t *adj;

    ASSERT(ADJ_INDEX_INVALID != adj_index);

    adj = adj_get(adj_index);

    if (flags & ADJ_NBR_REWRITE_FLAG_COMPLETE)
    {
	/*
	 * update the adj's rewrite string and build the arc
	 * from the rewrite node to the interface's TX node
	 */
	adj_nbr_update_rewrite_internal(adj, IP_LOOKUP_NEXT_REWRITE,
					adj_get_rewrite_node(adj->ia_link),
					vnet_tx_node_index_for_sw_interface(
					    vnet_get_main(),
					    adj->rewrite_header.sw_if_index),
					rewrite);
    }
    else
    {
	adj_nbr_update_rewrite_internal(adj, IP_LOOKUP_NEXT_ARP,
					adj_get_nd_node(adj->ia_nh_proto),
					vnet_tx_node_index_for_sw_interface(
					    vnet_get_main(),
					    adj->rewrite_header.sw_if_index),
					rewrite);
    }
}

/**
 * adj_nbr_update_rewrite_internal
 *
 * Update the adjacency's rewrite string. A NULL string implies the
 * rewrite is reset (i.e. when ARP/ND entry is gone).
 * NB: the adj being updated may be handling traffic in the DP.
 */
void
adj_nbr_update_rewrite_internal (ip_adjacency_t *adj,
				 ip_lookup_next_t adj_next_index,
				 u32 this_node,
				 u32 next_node,
				 u8 *rewrite)
{
    ip_adjacency_t *walk_adj;
    adj_index_t walk_ai, ai;
    vlib_main_t * vm;
    u32 old_next;
    int do_walk;

    vm = vlib_get_main();
    old_next = adj->lookup_next_index;

    ai = walk_ai = adj_get_index(adj);
    if (VNET_LINK_MPLS == adj->ia_link)
    {
        /*
         * The link type MPLS has no children in the control plane graph, it only
         * has children in the data-plane graph. The backwalk is up the former.
         * So we need to walk from its IP cousin.
         */
        walk_ai = adj_nbr_find(adj->ia_nh_proto,
                               fib_proto_to_link(adj->ia_nh_proto),
                               &adj->sub_type.nbr.next_hop,
                               adj->rewrite_header.sw_if_index);
    }

    /*
     * Don't call the walk re-entrantly
     */
    if (ADJ_INDEX_INVALID != walk_ai)
    {
        walk_adj = adj_get(walk_ai);
        if (ADJ_FLAG_SYNC_WALK_ACTIVE & walk_adj->ia_flags)
        {
            do_walk = 0;
        }
        else
        {
            /*
             * Prevent re-entrant walk of the same adj
             */
            walk_adj->ia_flags |= ADJ_FLAG_SYNC_WALK_ACTIVE;
            do_walk = 1;
        }
    }
    else
    {
        do_walk = 0;
    }

    /*
     * lock the adjacencies that are affected by updates this walk will provoke.
     * Since the aim of the walk is to update children to link to a different
     * DPO, this adj will no longer be in use and its lock count will drop to 0.
     * We don't want it to be deleted as part of this endeavour.
     */
    adj_lock(ai);
    adj_lock(walk_ai);

    /*
     * Updating a rewrite string is not atomic;
     *  - the rewrite string is too long to write in one instruction
     *  - when swapping from incomplete to complete, we also need to update
     *    the VLIB graph next-index of the adj.
     * ideally we would only want to suspend forwarding via this adj whilst we
     * do this, but we do not have that level of granularity - it's suspend all
     * worker threads or nothing.
     * The other choices are:
     *  - to mark the adj down and back walk so child load-balances drop this adj
     *    from the set.
     *  - update the next_node index of this adj to point to error-drop
     * both of which will mean for MAC change we will drop for this adj
     * which is not acceptable. However, when the adj changes type (from
     * complete to incomplete and vice-versa) the child DPOs, which have the
     * VLIB graph next node index, will be sending packets to the wrong graph
     * node. So from the options above, updating the next_node of the adj to
     * be drop will work, but it relies on each graph node v4/v6/mpls, rewrite/
     * arp/midchain always be valid w.r.t. a mis-match of adj type and node type
     * (i.e. a rewrite adj in the arp node). This is not enforceable. Getting it
     * wrong will lead to hard to find bugs since its a race condition. So we
     * choose the more reliable method of updating the children to use the drop,
     * then switching adj's type, then updating the children again. Did I mention
     * that this doesn't happen often...
     * So we need to distinguish between the two cases:
     *  1 - mac change
     *  2 - adj type change
     */
    if (do_walk &&
        old_next != adj_next_index &&
        ADJ_INDEX_INVALID != walk_ai)
    {
        /*
         * the adj is changing type. we need to fix all children so that they
         * stack momentarily on a drop, while the adj changes. If we don't do
         * this  the children will send packets to a VLIB graph node that does
         * not correspond to the adj's type - and it goes downhill from there.
         */
	fib_node_back_walk_ctx_t bw_ctx = {
	    .fnbw_reason = FIB_NODE_BW_REASON_FLAG_ADJ_DOWN,
            /*
             * force this walk to be synchronous. if we don't and a node in the graph
             * (a heavily shared path-list) chooses to back-ground the walk (make it
             * async) then it will pause and we will do the adj update below, before
             * all the children are updated. not good.
             */
            .fnbw_flags = FIB_NODE_BW_FLAG_FORCE_SYNC,
	};

	fib_walk_sync(FIB_NODE_TYPE_ADJ, walk_ai, &bw_ctx);
	/*
	 * fib_walk_sync may allocate a new adjacency and potentially cuase a
	 * realloc for adj_pool. When that happens, adj pointer is no longer
	 * valid here. We refresh the adj pointer accordingly.
	 */
	adj = adj_get (ai);
    }

    /*
     * If we are just updating the MAC string of the adj (which we also can't
     * do atomically), then we need to stop packets switching through the adj.
     * We can't do that on a per-adj basis, so it's all the packets.
     * If we are updating the type, and we walked back to the children above,
     * then this barrier serves to flush the queues/frames.
     */
    vlib_worker_thread_barrier_sync(vm);

    adj->lookup_next_index = adj_next_index;
    adj->ia_node_index = this_node;

    if (NULL != rewrite)
    {
	/*
	 * new rewrite provided.
	 * fill in the adj's rewrite string, and build the VLIB graph arc.
	 */
	vnet_rewrite_set_data_internal(&adj->rewrite_header,
				       sizeof(adj->rewrite_data),
				       rewrite,
				       vec_len(rewrite));
	vec_free(rewrite);
    }
    else
    {
	vnet_rewrite_clear_data_internal(&adj->rewrite_header,
					 sizeof(adj->rewrite_data));
    }
    adj->rewrite_header.next_index = vlib_node_add_next(vlib_get_main(),
                                                        this_node,
                                                        next_node);

    /*
     * done with the rewrite update - let the workers loose.
     */
    vlib_worker_thread_barrier_release(vm);

    if (do_walk &&
        (old_next != adj->lookup_next_index) &&
        (ADJ_INDEX_INVALID != walk_ai))
    {
        /*
         * backwalk to the children so they can stack on the now updated
         * adjacency
         */
        fib_node_back_walk_ctx_t bw_ctx = {
	    .fnbw_reason = FIB_NODE_BW_REASON_FLAG_ADJ_UPDATE,
	};

	fib_walk_sync(FIB_NODE_TYPE_ADJ, walk_ai, &bw_ctx);
    }
    /*
     * Prevent re-entrant walk of the same adj
     */
    if (do_walk)
    {
        walk_adj = adj_get(walk_ai);
        walk_adj->ia_flags &= ~ADJ_FLAG_SYNC_WALK_ACTIVE;
    }

    adj_delegate_adj_modified(adj_get(ai));
    adj_unlock(ai);
    adj_unlock(walk_ai);
}

u32
adj_nbr_db_size (void)
{
    fib_protocol_t proto;
    u32 sw_if_index = 0;
    u64 count = 0;

    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
    {
	vec_foreach_index(sw_if_index, adj_nbr_tables[proto])
	{
	    if (NULL != adj_nbr_tables[proto][sw_if_index])
	    {
		count += hash_elts(adj_nbr_tables[proto][sw_if_index]);
	    }
	}
    }
    return (count);
}

/**
 * @brief Walk all adjacencies on a link for a given next-hop protocol
 */
void
adj_nbr_walk (u32 sw_if_index,
	      fib_protocol_t adj_nh_proto,
	      adj_walk_cb_t cb,
	      void *ctx)
{
    adj_index_t ai, *ais, *aip;
    adj_nbr_key_t *key;

    ADJ_NBR_ASSERT_NH_PROTO (adj_nh_proto,);

    if (!ADJ_NBR_ITF_OK(adj_nh_proto, sw_if_index))
	return;

    ais = NULL;

    /* elements may be removed from the table during the walk, so
     * collect the set first then process them */
    hash_foreach_mem (key, ai, adj_nbr_tables[adj_nh_proto][sw_if_index],
    ({
        vec_add1(ais, ai);
    }));

    vec_foreach(aip, ais)
    {
        /* An adj may be deleted during the walk so check first */
        if (!pool_is_free_index(adj_pool, *aip))
            cb(*aip, ctx);
    }
    vec_free(ais);
}

/**
 * @brief Walk adjacencies on a link with a given v4 next-hop.
 * that is visit the adjacencies with different link types.
 */
void
adj_nbr_walk_nh4 (u32 sw_if_index,
		 const ip4_address_t *addr,
		 adj_walk_cb_t cb,
		 void *ctx)
{
    if (!ADJ_NBR_ITF_OK(FIB_PROTOCOL_IP4, sw_if_index))
	return;

    ip46_address_t nh = {
	.ip4 = *addr,
    };
    vnet_link_t linkt;
    adj_index_t ai;

    FOR_EACH_VNET_LINK(linkt)
    {
        ai = adj_nbr_find (FIB_PROTOCOL_IP4, linkt, &nh, sw_if_index);

        if (INDEX_INVALID != ai)
            cb(ai, ctx);
    }
}

/**
 * @brief Walk adjacencies on a link with a given v6 next-hop.
 * that is visit the adjacencies with different link types.
 */
void
adj_nbr_walk_nh6 (u32 sw_if_index,
		 const ip6_address_t *addr,
		 adj_walk_cb_t cb,
		 void *ctx)
{
    if (!ADJ_NBR_ITF_OK(FIB_PROTOCOL_IP6, sw_if_index))
	return;

    ip46_address_t nh = {
	.ip6 = *addr,
    };
    vnet_link_t linkt;
    adj_index_t ai;

    FOR_EACH_VNET_LINK(linkt)
    {
        ai = adj_nbr_find (FIB_PROTOCOL_IP6, linkt, &nh, sw_if_index);

        if (INDEX_INVALID != ai)
            cb(ai, ctx);
    }
}

/**
 * @brief Walk adjacencies on a link with a given next-hop.
 * that is visit the adjacencies with different link types.
 */
void
adj_nbr_walk_nh (u32 sw_if_index,
		 fib_protocol_t adj_nh_proto,
		 const ip46_address_t *nh,
		 adj_walk_cb_t cb,
		 void *ctx)
{
    ADJ_NBR_ASSERT_NH_PROTO (adj_nh_proto,);

    if (!ADJ_NBR_ITF_OK(adj_nh_proto, sw_if_index))
	return;

    switch (adj_nh_proto)
    {
    case FIB_PROTOCOL_IP4:
        adj_nbr_walk_nh4(sw_if_index, &nh->ip4, cb, ctx);
        break; 
    case FIB_PROTOCOL_IP6:
        adj_nbr_walk_nh6(sw_if_index, &nh->ip6, cb, ctx);
        break;
    case FIB_PROTOCOL_MPLS:
        ASSERT(0);
        break;
    }
}

/**
 * Flags associated with the interface state walks
 */
typedef enum adj_nbr_interface_flags_t_
{
    ADJ_NBR_INTERFACE_UP = (1 << 0),
} adj_nbr_interface_flags_t;

/**
 * Context for the state change walk of the DB
 */
typedef struct adj_nbr_interface_state_change_ctx_t_
{
    /**
     * Flags on the interface
     */
    adj_nbr_interface_flags_t flags;
} adj_nbr_interface_state_change_ctx_t;

static adj_walk_rc_t
adj_nbr_interface_state_change_one (adj_index_t ai,
                                    void *arg)
{
    /*
     * Back walk the graph to inform the forwarding entries
     * that this interface state has changed. Do this synchronously
     * since this is the walk that provides convergence
     */
    adj_nbr_interface_state_change_ctx_t *ctx = arg;
    fib_node_back_walk_ctx_t bw_ctx = {
	.fnbw_reason = ((ctx->flags & ADJ_NBR_INTERFACE_UP) ?
                        FIB_NODE_BW_REASON_FLAG_INTERFACE_UP :
                        FIB_NODE_BW_REASON_FLAG_INTERFACE_DOWN),
        /*
         * the force sync applies only as far as the first fib_entry.
         * And it's the fib_entry's we need to converge away from
         * the adjacencies on the now down link
         */
        .fnbw_flags = (!(ctx->flags & ADJ_NBR_INTERFACE_UP) ?
                       FIB_NODE_BW_FLAG_FORCE_SYNC :
                       FIB_NODE_BW_FLAG_NONE),
    };
    ip_adjacency_t *adj;

    adj_lock (ai);

    adj = adj_get(ai);

    adj->ia_flags |= ADJ_FLAG_SYNC_WALK_ACTIVE;
    fib_walk_sync(FIB_NODE_TYPE_ADJ, ai, &bw_ctx);
    adj->ia_flags &= ~ADJ_FLAG_SYNC_WALK_ACTIVE;

    adj_unlock (ai);
    return (ADJ_WALK_RC_CONTINUE);
}

/**
 * @brief Registered function for SW interface state changes
 */
static clib_error_t *
adj_nbr_sw_interface_state_change (vnet_main_t * vnm,
                                   u32 sw_if_index,
                                   u32 flags)
{
    fib_protocol_t proto;

    /*
     * walk each adj on the interface and trigger a walk from that adj
     */
    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
    {
	adj_nbr_interface_state_change_ctx_t ctx = {
	    .flags = ((flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) ?
                      ADJ_NBR_INTERFACE_UP :
                      0),
	};

	adj_nbr_walk(sw_if_index, proto,
		     adj_nbr_interface_state_change_one,
		     &ctx);
    }

    return (NULL);
}

VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION_PRIO(
    adj_nbr_sw_interface_state_change,
    VNET_ITF_FUNC_PRIORITY_HIGH);

/**
 * @brief Invoked on each SW interface of a HW interface when the
 * HW interface state changes
 */
static walk_rc_t
adj_nbr_hw_sw_interface_state_change (vnet_main_t * vnm,
                                      u32 sw_if_index,
                                      void *arg)
{
    adj_nbr_interface_state_change_ctx_t *ctx = arg;
    fib_protocol_t proto;

    /*
     * walk each adj on the interface and trigger a walk from that adj
     */
    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
    {
	adj_nbr_walk(sw_if_index, proto,
		     adj_nbr_interface_state_change_one,
		     ctx);
    }
    return (WALK_CONTINUE);
}

/**
 * @brief Registered callback for HW interface state changes
 */
static clib_error_t *
adj_nbr_hw_interface_state_change (vnet_main_t * vnm,
                                   u32 hw_if_index,
                                   u32 flags)
{
    /*
     * walk SW interface on the HW
     */
    adj_nbr_interface_state_change_ctx_t ctx = {
        .flags = ((flags & VNET_HW_INTERFACE_FLAG_LINK_UP) ?
                  ADJ_NBR_INTERFACE_UP :
                  0),
    };

    vnet_hw_interface_walk_sw(vnm, hw_if_index,
                              adj_nbr_hw_sw_interface_state_change,
                              &ctx);

    return (NULL);
}

VNET_HW_INTERFACE_LINK_UP_DOWN_FUNCTION_PRIO(
    adj_nbr_hw_interface_state_change,
    VNET_ITF_FUNC_PRIORITY_HIGH);

static adj_walk_rc_t
adj_nbr_interface_delete_one (adj_index_t ai,
			      void *arg)
{
    /*
     * Back walk the graph to inform the forwarding entries
     * that this interface has been deleted.
     */
    fib_node_back_walk_ctx_t bw_ctx = {
	.fnbw_reason = FIB_NODE_BW_REASON_FLAG_INTERFACE_DELETE,
    };
    ip_adjacency_t *adj;

    adj_lock(ai);

    adj = adj_get(ai);

    adj->ia_flags |= ADJ_FLAG_SYNC_WALK_ACTIVE;
    fib_walk_sync(FIB_NODE_TYPE_ADJ, ai, &bw_ctx);
    adj->ia_flags &= ~ADJ_FLAG_SYNC_WALK_ACTIVE;

    adj_unlock(ai);
    return (ADJ_WALK_RC_CONTINUE);
}

/**
 * adj_nbr_interface_add_del
 *
 * Registered to receive interface Add and delete notifications
 */
static clib_error_t *
adj_nbr_interface_add_del (vnet_main_t * vnm,
			   u32 sw_if_index,
			   u32 is_add)
{
    fib_protocol_t proto;

    if (is_add)
    {
	/*
	 * not interested in interface additions. we will not back walk
	 * to resolve paths through newly added interfaces. Why? The control
	 * plane should have the brains to add interfaces first, then routes.
	 * So the case where there are paths with a interface that matches
	 * one just created is the case where the path resolved through an
	 * interface that was deleted, and still has not been removed. The
	 * new interface added, is NO GUARANTEE that the interface being
	 * added now, even though it may have the same sw_if_index, is the
	 * same interface that the path needs. So tough!
	 * If the control plane wants these routes to resolve it needs to
	 * remove and add them again.
	 */
	return (NULL);
    }

    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
    {
	adj_nbr_walk(sw_if_index, proto,
		     adj_nbr_interface_delete_one,
		     NULL);
    }

    return (NULL);
}

VNET_SW_INTERFACE_ADD_DEL_FUNCTION(adj_nbr_interface_add_del);


static adj_walk_rc_t
adj_nbr_ethernet_mac_change_one (adj_index_t ai,
                                 void *arg)
{
    vnet_update_adjacency_for_sw_interface(vnet_get_main(),
                                           adj_get_sw_if_index(ai),
                                           ai);

    return (ADJ_WALK_RC_CONTINUE);
}

/**
 * Callback function invoked when an interface's MAC Address changes
 */
static void
adj_nbr_ethernet_change_mac (ethernet_main_t * em,
                             u32 sw_if_index, uword opaque)
{
    fib_protocol_t proto;

    FOR_EACH_FIB_IP_PROTOCOL(proto)
    {
	adj_nbr_walk(sw_if_index, proto,
		     adj_nbr_ethernet_mac_change_one,
		     NULL);
    }
}

static adj_walk_rc_t
adj_nbr_show_one (adj_index_t ai,
		  void *arg)
{
    vlib_cli_output (arg, "[@%d]  %U",
                     ai,
                     format_ip_adjacency, ai,
		     FORMAT_IP_ADJACENCY_NONE);

    return (ADJ_WALK_RC_CONTINUE);
}

static clib_error_t *
adj_nbr_show (vlib_main_t * vm,
	      unformat_input_t * input,
	      vlib_cli_command_t * cmd)
{
    adj_index_t ai = ADJ_INDEX_INVALID;
    ip46_address_t nh = ip46_address_initializer;
    u32 sw_if_index = ~0;

    while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
	if (unformat (input, "%U",
                      unformat_vnet_sw_interface, vnet_get_main(),
                      &sw_if_index))
	    ;
	else if (unformat (input, "%U",
			   unformat_ip46_address, &nh, IP46_TYPE_ANY))
	    ;
	else if (unformat (input, "%d", &ai))
	    ;
	else
	    break;
    }

    if (ADJ_INDEX_INVALID != ai)
    {
	vlib_cli_output (vm, "[@%d] %U",
                         ai,
                         format_ip_adjacency, ai,
			 FORMAT_IP_ADJACENCY_DETAIL);
    }
    else if (~0 != sw_if_index)
    {
	fib_protocol_t proto;

        if (ip46_address_is_zero(&nh))
        {
            for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
            {
                adj_nbr_walk(sw_if_index, proto,
                             adj_nbr_show_one,
                             vm);
            }
        }
        else
        {
            proto = (ip46_address_is_ip4(&nh) ?
                     FIB_PROTOCOL_IP4 :
                     FIB_PROTOCOL_IP6);
            adj_nbr_walk_nh(sw_if_index, proto, &nh,
                            adj_nbr_show_one,
                            vm);
        }
    }
    else
    {
	fib_protocol_t proto;

	for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
	{
	    vec_foreach_index(sw_if_index, adj_nbr_tables[proto])
	    {
		adj_nbr_walk(sw_if_index, proto,
			     adj_nbr_show_one,
			     vm);
	    }
	}
    }

    return 0;
}

/*?
 * Show all neighbour adjacencies.
 * @cliexpar
 * @cliexstart{sh adj nbr}
 * [@2] ipv4 via 1.0.0.2 loop0: IP4: 00:00:22:aa:bb:cc -> 00:00:11:aa:bb:cc
 * [@3] mpls via 1.0.0.2 loop0: MPLS_UNICAST: 00:00:22:aa:bb:cc -> 00:00:11:aa:bb:cc
 * [@4] ipv4 via 1.0.0.3 loop0: IP4: 00:00:22:aa:bb:cc -> 00:00:11:aa:bb:cc
 * [@5] mpls via 1.0.0.3 loop0: MPLS_UNICAST: 00:00:22:aa:bb:cc -> 00:00:11:aa:bb:cc
 * @cliexend
 ?*/
VLIB_CLI_COMMAND (ip4_show_fib_command, static) = {
    .path = "show adj nbr",
    .short_help = "show adj nbr [<adj_index>] [interface]",
    .function = adj_nbr_show,
};

u8*
format_adj_nbr_incomplete (u8* s, va_list *ap)
{
    index_t index = va_arg(*ap, index_t);
    CLIB_UNUSED(u32 indent) = va_arg(*ap, u32);
    vnet_main_t * vnm = vnet_get_main();
    ip_adjacency_t * adj = adj_get(index);

    s = format (s, "arp-%U", format_vnet_link, adj->ia_link);
    s = format (s, ": via %U",
                format_ip46_address, &adj->sub_type.nbr.next_hop,
		adj_proto_to_46(adj->ia_nh_proto));
    s = format (s, " %U",
                format_vnet_sw_if_index_name,
                vnm, adj->rewrite_header.sw_if_index);

    return (s);
}

u8*
format_adj_nbr (u8* s, va_list *ap)
{
    index_t index = va_arg(*ap, index_t);
    CLIB_UNUSED(u32 indent) = va_arg(*ap, u32);
    ip_adjacency_t * adj = adj_get(index);

    s = format (s, "%U", format_vnet_link, adj->ia_link);
    s = format (s, " via %U ",
		format_ip46_address, &adj->sub_type.nbr.next_hop,
		adj_proto_to_46(adj->ia_nh_proto));
    s = format (s, "%U",
		format_vnet_rewrite,
		&adj->rewrite_header, sizeof (adj->rewrite_data), 0);

    return (s);
}

static void
adj_dpo_lock (dpo_id_t *dpo)
{
    adj_lock(dpo->dpoi_index);
}
static void
adj_dpo_unlock (dpo_id_t *dpo)
{
    adj_unlock(dpo->dpoi_index);
}

static void
adj_mem_show (void)
{
    fib_show_memory_usage("Adjacency",
			  pool_elts(adj_pool),
			  pool_len(adj_pool),
			  sizeof(ip_adjacency_t));
}

const static dpo_vft_t adj_nbr_dpo_vft = {
    .dv_lock = adj_dpo_lock,
    .dv_unlock = adj_dpo_unlock,
    .dv_format = format_adj_nbr,
    .dv_mem_show = adj_mem_show,
    .dv_get_urpf = adj_dpo_get_urpf,
    .dv_get_mtu = adj_dpo_get_mtu,
};
const static dpo_vft_t adj_nbr_incompl_dpo_vft = {
    .dv_lock = adj_dpo_lock,
    .dv_unlock = adj_dpo_unlock,
    .dv_format = format_adj_nbr_incomplete,
    .dv_get_urpf = adj_dpo_get_urpf,
    .dv_get_mtu = adj_dpo_get_mtu,
};

/**
 * @brief The per-protocol VLIB graph nodes that are assigned to an adjacency
 *        object.
 *
 * this means that these graph nodes are ones from which a nbr is the
 * parent object in the DPO-graph.
 */
const static char* const nbr_ip4_nodes[] =
{
    "ip4-rewrite",
    NULL,
};
const static char* const nbr_ip6_nodes[] =
{
    "ip6-rewrite",
    NULL,
};
const static char* const nbr_mpls_nodes[] =
{
    "mpls-output",
    NULL,
};
const static char* const nbr_ethernet_nodes[] =
{
    "adj-l2-rewrite",
    NULL,
};
const static char* const * const nbr_nodes[DPO_PROTO_NUM] =
{
    [DPO_PROTO_IP4]  = nbr_ip4_nodes,
    [DPO_PROTO_IP6]  = nbr_ip6_nodes,
    [DPO_PROTO_MPLS] = nbr_mpls_nodes,
    [DPO_PROTO_ETHERNET] = nbr_ethernet_nodes,
};

const static char* const nbr_incomplete_ip4_nodes[] =
{
    "ip4-arp",
    NULL,
};
const static char* const nbr_incomplete_ip6_nodes[] =
{
    "ip6-discover-neighbor",
    NULL,
};
const static char* const nbr_incomplete_mpls_nodes[] =
{
    "mpls-adj-incomplete",
    NULL,
};

const static char* const * const nbr_incomplete_nodes[DPO_PROTO_NUM] =
{
    [DPO_PROTO_IP4]  = nbr_incomplete_ip4_nodes,
    [DPO_PROTO_IP6]  = nbr_incomplete_ip6_nodes,
    [DPO_PROTO_MPLS] = nbr_incomplete_mpls_nodes,
};

void
adj_nbr_module_init (void)
{
    dpo_register(DPO_ADJACENCY,
                 &adj_nbr_dpo_vft,
                 nbr_nodes);
    dpo_register(DPO_ADJACENCY_INCOMPLETE,
                 &adj_nbr_incompl_dpo_vft,
                 nbr_incomplete_nodes);

    ethernet_address_change_ctx_t ctx = {
        .function = adj_nbr_ethernet_change_mac,
        .function_opaque = 0,
    };
    vec_add1 (ethernet_main.address_change_callbacks, ctx);
}