aboutsummaryrefslogtreecommitdiffstats
path: root/src/plugins/hs_apps/echo_client.c
blob: d641a9ec14e7a94ca9f4b12f6eb7648dfc4bdbd5 (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
/*
 * echo_client.c - vpp built-in echo client code
 *
 * Copyright (c) 2017-2019 by 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/vnet.h>
#include <vlibapi/api.h>
#include <vlibmemory/api.h>
#include <hs_apps/echo_client.h>

echo_client_main_t echo_client_main;

#define ECHO_CLIENT_DBG (0)
#define DBG(_fmt, _args...)			\
    if (ECHO_CLIENT_DBG) 				\
      clib_warning (_fmt, ##_args)

static void
signal_evt_to_cli_i (int *code)
{
  echo_client_main_t *ecm = &echo_client_main;
  ASSERT (vlib_get_thread_index () == 0);
  vlib_process_signal_event (ecm->vlib_main, ecm->cli_node_index, *code, 0);
}

static void
signal_evt_to_cli (int code)
{
  if (vlib_get_thread_index () != 0)
    vl_api_rpc_call_main_thread (signal_evt_to_cli_i, (u8 *) & code,
				 sizeof (code));
  else
    signal_evt_to_cli_i (&code);
}

static void
send_data_chunk (echo_client_main_t * ecm, eclient_session_t * s)
{
  u8 *test_data = ecm->connect_test_data;
  int test_buf_len, test_buf_offset, rv;
  u32 bytes_this_chunk;

  test_buf_len = vec_len (test_data);
  ASSERT (test_buf_len > 0);
  test_buf_offset = s->bytes_sent % test_buf_len;
  bytes_this_chunk = clib_min (test_buf_len - test_buf_offset,
			       s->bytes_to_send);

  if (!ecm->is_dgram)
    {
      if (ecm->no_copy)
	{
	  svm_fifo_t *f = s->data.tx_fifo;
	  rv = clib_min (svm_fifo_max_enqueue_prod (f), bytes_this_chunk);
	  svm_fifo_enqueue_nocopy (f, rv);
	  session_send_io_evt_to_thread_custom (
	    &f->shr->master_session_index, s->thread_index, SESSION_IO_EVT_TX);
	}
      else
	rv = app_send_stream (&s->data, test_data + test_buf_offset,
			      bytes_this_chunk, 0);
    }
  else
    {
      svm_fifo_t *f = s->data.tx_fifo;
      u32 max_enqueue = svm_fifo_max_enqueue_prod (f);

      if (max_enqueue < sizeof (session_dgram_hdr_t))
	return;

      max_enqueue -= sizeof (session_dgram_hdr_t);

      if (ecm->no_copy)
	{
	  session_dgram_hdr_t hdr;
	  app_session_transport_t *at = &s->data.transport;

	  rv = clib_min (max_enqueue, bytes_this_chunk);

	  hdr.data_length = rv;
	  hdr.data_offset = 0;
	  clib_memcpy_fast (&hdr.rmt_ip, &at->rmt_ip,
			    sizeof (ip46_address_t));
	  hdr.is_ip4 = at->is_ip4;
	  hdr.rmt_port = at->rmt_port;
	  clib_memcpy_fast (&hdr.lcl_ip, &at->lcl_ip,
			    sizeof (ip46_address_t));
	  hdr.lcl_port = at->lcl_port;
	  svm_fifo_enqueue (f, sizeof (hdr), (u8 *) & hdr);
	  svm_fifo_enqueue_nocopy (f, rv);
	  session_send_io_evt_to_thread_custom (
	    &f->shr->master_session_index, s->thread_index, SESSION_IO_EVT_TX);
	}
      else
	{
	  bytes_this_chunk = clib_min (bytes_this_chunk, max_enqueue);
	  rv = app_send_dgram (&s->data, test_data + test_buf_offset,
			       bytes_this_chunk, 0);
	}
    }

  /* If we managed to enqueue data... */
  if (rv > 0)
    {
      /* Account for it... */
      s->bytes_to_send -= rv;
      s->bytes_sent += rv;

      if (ECHO_CLIENT_DBG)
	{
          /* *INDENT-OFF* */
          ELOG_TYPE_DECLARE (e) =
            {
              .format = "tx-enq: xfer %d bytes, sent %u remain %u",
              .format_args = "i4i4i4",
            };
          /* *INDENT-ON* */
	  struct
	  {
	    u32 data[3];
	  } *ed;
	  ed = ELOG_DATA (&vlib_global_main.elog_main, e);
	  ed->data[0] = rv;
	  ed->data[1] = s->bytes_sent;
	  ed->data[2] = s->bytes_to_send;
	}
    }
}

static void
receive_data_chunk (echo_client_main_t * ecm, eclient_session_t * s)
{
  svm_fifo_t *rx_fifo = s->data.rx_fifo;
  u32 thread_index = vlib_get_thread_index ();
  int n_read, i;

  if (ecm->test_bytes)
    {
      if (!ecm->is_dgram)
	n_read = app_recv_stream (&s->data, ecm->rx_buf[thread_index],
				  vec_len (ecm->rx_buf[thread_index]));
      else
	n_read = app_recv_dgram (&s->data, ecm->rx_buf[thread_index],
				 vec_len (ecm->rx_buf[thread_index]));
    }
  else
    {
      n_read = svm_fifo_max_dequeue_cons (rx_fifo);
      svm_fifo_dequeue_drop (rx_fifo, n_read);
    }

  if (n_read > 0)
    {
      if (ECHO_CLIENT_DBG)
	{
          /* *INDENT-OFF* */
          ELOG_TYPE_DECLARE (e) =
            {
              .format = "rx-deq: %d bytes",
              .format_args = "i4",
            };
          /* *INDENT-ON* */
	  struct
	  {
	    u32 data[1];
	  } *ed;
	  ed = ELOG_DATA (&vlib_global_main.elog_main, e);
	  ed->data[0] = n_read;
	}

      if (ecm->test_bytes)
	{
	  for (i = 0; i < n_read; i++)
	    {
	      if (ecm->rx_buf[thread_index][i]
		  != ((s->bytes_received + i) & 0xff))
		{
		  clib_warning ("read %d error at byte %lld, 0x%x not 0x%x",
				n_read, s->bytes_received + i,
				ecm->rx_buf[thread_index][i],
				((s->bytes_received + i) & 0xff));
		  ecm->test_failed = 1;
		}
	    }
	}
      ASSERT (n_read <= s->bytes_to_receive);
      s->bytes_to_receive -= n_read;
      s->bytes_received += n_read;
    }
}

static uword
echo_client_node_fn (vlib_main_t * vm, vlib_node_runtime_t * node,
		     vlib_frame_t * frame)
{
  echo_client_main_t *ecm = &echo_client_main;
  int my_thread_index = vlib_get_thread_index ();
  eclient_session_t *sp;
  int i;
  int delete_session;
  u32 *connection_indices;
  u32 *connections_this_batch;
  u32 nconnections_this_batch;

  connection_indices = ecm->connection_index_by_thread[my_thread_index];
  connections_this_batch =
    ecm->connections_this_batch_by_thread[my_thread_index];

  if ((ecm->run_test != ECHO_CLIENTS_RUNNING) ||
      ((vec_len (connection_indices) == 0)
       && vec_len (connections_this_batch) == 0))
    return 0;

  /* Grab another pile of connections */
  if (PREDICT_FALSE (vec_len (connections_this_batch) == 0))
    {
      nconnections_this_batch =
	clib_min (ecm->connections_per_batch, vec_len (connection_indices));

      ASSERT (nconnections_this_batch > 0);
      vec_validate (connections_this_batch, nconnections_this_batch - 1);
      clib_memcpy_fast (connections_this_batch,
			connection_indices + vec_len (connection_indices)
			- nconnections_this_batch,
			nconnections_this_batch * sizeof (u32));
      _vec_len (connection_indices) -= nconnections_this_batch;
    }

  if (PREDICT_FALSE (ecm->prev_conns != ecm->connections_per_batch
		     && ecm->prev_conns == vec_len (connections_this_batch)))
    {
      ecm->repeats++;
      ecm->prev_conns = vec_len (connections_this_batch);
      if (ecm->repeats == 500000)
	{
	  clib_warning ("stuck clients");
	}
    }
  else
    {
      ecm->prev_conns = vec_len (connections_this_batch);
      ecm->repeats = 0;
    }

  for (i = 0; i < vec_len (connections_this_batch); i++)
    {
      delete_session = 1;

      sp = pool_elt_at_index (ecm->sessions, connections_this_batch[i]);

      if (sp->bytes_to_send > 0)
	{
	  send_data_chunk (ecm, sp);
	  delete_session = 0;
	}
      if (sp->bytes_to_receive > 0)
	{
	  delete_session = 0;
	}
      if (PREDICT_FALSE (delete_session == 1))
	{
	  session_t *s;

	  clib_atomic_fetch_add (&ecm->tx_total, sp->bytes_sent);
	  clib_atomic_fetch_add (&ecm->rx_total, sp->bytes_received);
	  s = session_get_from_handle_if_valid (sp->vpp_session_handle);

	  if (s)
	    {
	      vnet_disconnect_args_t _a, *a = &_a;
	      a->handle = session_handle (s);
	      a->app_index = ecm->app_index;
	      vnet_disconnect_session (a);

	      vec_delete (connections_this_batch, 1, i);
	      i--;
	      clib_atomic_fetch_add (&ecm->ready_connections, -1);
	    }
	  else
	    {
	      clib_warning ("session AWOL?");
	      vec_delete (connections_this_batch, 1, i);
	    }

	  /* Kick the debug CLI process */
	  if (ecm->ready_connections == 0)
	    {
	      signal_evt_to_cli (2);
	    }
	}
    }

  ecm->connection_index_by_thread[my_thread_index] = connection_indices;
  ecm->connections_this_batch_by_thread[my_thread_index] =
    connections_this_batch;
  return 0;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (echo_clients_node) =
{
  .function = echo_client_node_fn,
  .name = "echo-clients",
  .type = VLIB_NODE_TYPE_INPUT,
  .state = VLIB_NODE_STATE_DISABLED,
};
/* *INDENT-ON* */

static int
echo_clients_init (vlib_main_t * vm)
{
  echo_client_main_t *ecm = &echo_client_main;
  vlib_thread_main_t *vtm = vlib_get_thread_main ();
  u32 num_threads;
  int i;

  num_threads = 1 /* main thread */  + vtm->n_threads;

  /* Init test data. Big buffer */
  vec_validate (ecm->connect_test_data, 4 * 1024 * 1024 - 1);
  for (i = 0; i < vec_len (ecm->connect_test_data); i++)
    ecm->connect_test_data[i] = i & 0xff;

  vec_validate (ecm->rx_buf, num_threads - 1);
  for (i = 0; i < num_threads; i++)
    vec_validate (ecm->rx_buf[i], vec_len (ecm->connect_test_data) - 1);

  ecm->is_init = 1;

  vec_validate (ecm->connection_index_by_thread, vtm->n_vlib_mains);
  vec_validate (ecm->connections_this_batch_by_thread, vtm->n_vlib_mains);
  vec_validate (ecm->quic_session_index_by_thread, vtm->n_vlib_mains);
  vec_validate (ecm->vpp_event_queue, vtm->n_vlib_mains);

  return 0;
}

static int
quic_echo_clients_qsession_connected_callback (u32 app_index, u32 api_context,
					       session_t * s,
					       session_error_t err)
{
  echo_client_main_t *ecm = &echo_client_main;
  vnet_connect_args_t *a = 0;
  int rv;
  u8 thread_index = vlib_get_thread_index ();
  session_endpoint_cfg_t sep = SESSION_ENDPOINT_CFG_NULL;
  u32 stream_n;
  session_handle_t handle;

  DBG ("QUIC Connection handle %d", session_handle (s));

  vec_validate (a, 1);
  a->uri = (char *) ecm->connect_uri;
  if (parse_uri (a->uri, &sep))
    return -1;
  sep.parent_handle = handle = session_handle (s);

  for (stream_n = 0; stream_n < ecm->quic_streams; stream_n++)
    {
      clib_memset (a, 0, sizeof (*a));
      a->app_index = ecm->app_index;
      a->api_context = -1 - api_context;
      clib_memcpy (&a->sep_ext, &sep, sizeof (sep));

      DBG ("QUIC opening stream %d", stream_n);
      if ((rv = vnet_connect (a)))
	{
	  clib_error ("Stream session %d opening failed: %d", stream_n, rv);
	  return -1;
	}
      DBG ("QUIC stream %d connected", stream_n);
    }
  /*
   * 's' is no longer valid, its underlying pool could have been moved in
   * vnet_connect()
   */
  vec_add1 (ecm->quic_session_index_by_thread[thread_index], handle);
  vec_free (a);
  return 0;
}

static int
quic_echo_clients_session_connected_callback (u32 app_index, u32 api_context,
					      session_t * s,
					      session_error_t err)
{
  echo_client_main_t *ecm = &echo_client_main;
  eclient_session_t *session;
  u32 session_index;
  u8 thread_index;

  if (PREDICT_FALSE (ecm->run_test != ECHO_CLIENTS_STARTING))
    return -1;

  if (err)
    {
      clib_warning ("connection %d failed!", api_context);
      ecm->run_test = ECHO_CLIENTS_EXITING;
      signal_evt_to_cli (-1);
      return 0;
    }

  if (s->listener_handle == SESSION_INVALID_HANDLE)
    return quic_echo_clients_qsession_connected_callback (app_index,
							  api_context, s,
							  err);
  DBG ("STREAM Connection callback %d", api_context);

  thread_index = s->thread_index;
  ASSERT (thread_index == vlib_get_thread_index ()
	  || session_transport_service_type (s) == TRANSPORT_SERVICE_CL);

  if (!ecm->vpp_event_queue[thread_index])
    ecm->vpp_event_queue[thread_index] =
      session_main_get_vpp_event_queue (thread_index);

  /*
   * Setup session
   */
  clib_spinlock_lock_if_init (&ecm->sessions_lock);
  pool_get (ecm->sessions, session);
  clib_spinlock_unlock_if_init (&ecm->sessions_lock);

  clib_memset (session, 0, sizeof (*session));
  session_index = session - ecm->sessions;
  session->bytes_to_send = ecm->bytes_to_send;
  session->bytes_to_receive = ecm->no_return ? 0ULL : ecm->bytes_to_send;
  session->data.rx_fifo = s->rx_fifo;
  session->data.rx_fifo->shr->client_session_index = session_index;
  session->data.tx_fifo = s->tx_fifo;
  session->data.tx_fifo->shr->client_session_index = session_index;
  session->data.vpp_evt_q = ecm->vpp_event_queue[thread_index];
  session->vpp_session_handle = session_handle (s);

  if (ecm->is_dgram)
    {
      transport_connection_t *tc;
      tc = session_get_transport (s);
      clib_memcpy_fast (&session->data.transport, tc,
			sizeof (session->data.transport));
      session->data.is_dgram = 1;
    }

  vec_add1 (ecm->connection_index_by_thread[thread_index], session_index);
  clib_atomic_fetch_add (&ecm->ready_connections, 1);
  if (ecm->ready_connections == ecm->expected_connections)
    {
      ecm->run_test = ECHO_CLIENTS_RUNNING;
      /* Signal the CLI process that the action is starting... */
      signal_evt_to_cli (1);
    }

  return 0;
}

static int
echo_clients_session_connected_callback (u32 app_index, u32 api_context,
					 session_t * s, session_error_t err)
{
  echo_client_main_t *ecm = &echo_client_main;
  eclient_session_t *session;
  u32 session_index;
  u8 thread_index;

  if (PREDICT_FALSE (ecm->run_test != ECHO_CLIENTS_STARTING))
    return -1;

  if (err)
    {
      clib_warning ("connection %d failed!", api_context);
      ecm->run_test = ECHO_CLIENTS_EXITING;
      signal_evt_to_cli (-1);
      return 0;
    }

  thread_index = s->thread_index;
  ASSERT (thread_index == vlib_get_thread_index ()
	  || session_transport_service_type (s) == TRANSPORT_SERVICE_CL);

  if (!ecm->vpp_event_queue[thread_index])
    ecm->vpp_event_queue[thread_index] =
      session_main_get_vpp_event_queue (thread_index);

  /*
   * Setup session
   */
  clib_spinlock_lock_if_init (&ecm->sessions_lock);
  pool_get (ecm->sessions, session);
  clib_spinlock_unlock_if_init (&ecm->sessions_lock);

  clib_memset (session, 0, sizeof (*session));
  session_index = session - ecm->sessions;
  session->bytes_to_send = ecm->bytes_to_send;
  session->bytes_to_receive = ecm->no_return ? 0ULL : ecm->bytes_to_send;
  session->data.rx_fifo = s->rx_fifo;
  session->data.rx_fifo->shr->client_session_index = session_index;
  session->data.tx_fifo = s->tx_fifo;
  session->data.tx_fifo->shr->client_session_index = session_index;
  session->data.vpp_evt_q = ecm->vpp_event_queue[thread_index];
  session->vpp_session_handle = session_handle (s);

  if (ecm->is_dgram)
    {
      transport_connection_t *tc;
      tc = session_get_transport (s);
      clib_memcpy_fast (&session->data.transport, tc,
			sizeof (session->data.transport));
      session->data.is_dgram = 1;
    }

  vec_add1 (ecm->connection_index_by_thread[thread_index], session_index);
  clib_atomic_fetch_add (&ecm->ready_connections, 1);
  if (ecm->ready_connections == ecm->expected_connections)
    {
      ecm->run_test = ECHO_CLIENTS_RUNNING;
      /* Signal the CLI process that the action is starting... */
      signal_evt_to_cli (1);
    }

  return 0;
}

static void
echo_clients_session_reset_callback (session_t * s)
{
  echo_client_main_t *ecm = &echo_client_main;
  vnet_disconnect_args_t _a = { 0 }, *a = &_a;

  if (s->session_state == SESSION_STATE_READY)
    clib_warning ("Reset active connection %U", format_session, s, 2);

  a->handle = session_handle (s);
  a->app_index = ecm->app_index;
  vnet_disconnect_session (a);
  return;
}

static int
echo_clients_session_create_callback (session_t * s)
{
  return 0;
}

static void
echo_clients_session_disconnect_callback (session_t * s)
{
  echo_client_main_t *ecm = &echo_client_main;
  vnet_disconnect_args_t _a = { 0 }, *a = &_a;
  a->handle = session_handle (s);
  a->app_index = ecm->app_index;
  vnet_disconnect_session (a);
  return;
}

void
echo_clients_session_disconnect (session_t * s)
{
  echo_client_main_t *ecm = &echo_client_main;
  vnet_disconnect_args_t _a = { 0 }, *a = &_a;
  a->handle = session_handle (s);
  a->app_index = ecm->app_index;
  vnet_disconnect_session (a);
}

static int
echo_clients_rx_callback (session_t * s)
{
  echo_client_main_t *ecm = &echo_client_main;
  eclient_session_t *sp;

  if (PREDICT_FALSE (ecm->run_test != ECHO_CLIENTS_RUNNING))
    {
      echo_clients_session_disconnect (s);
      return -1;
    }

  sp =
    pool_elt_at_index (ecm->sessions, s->rx_fifo->shr->client_session_index);
  receive_data_chunk (ecm, sp);

  if (svm_fifo_max_dequeue_cons (s->rx_fifo))
    {
      if (svm_fifo_set_event (s->rx_fifo))
	session_send_io_evt_to_thread (s->rx_fifo, SESSION_IO_EVT_BUILTIN_RX);
    }
  return 0;
}

int
echo_client_add_segment_callback (u32 client_index, u64 segment_handle)
{
  /* New heaps may be added */
  return 0;
}

/* *INDENT-OFF* */
static session_cb_vft_t echo_clients = {
  .session_reset_callback = echo_clients_session_reset_callback,
  .session_connected_callback = echo_clients_session_connected_callback,
  .session_accept_callback = echo_clients_session_create_callback,
  .session_disconnect_callback = echo_clients_session_disconnect_callback,
  .builtin_app_rx_callback = echo_clients_rx_callback,
  .add_segment_callback = echo_client_add_segment_callback
};
/* *INDENT-ON* */

static clib_error_t *
echo_clients_attach (u8 * appns_id, u64 appns_flags, u64 appns_secret)
{
  vnet_app_add_cert_key_pair_args_t _ck_pair, *ck_pair = &_ck_pair;
  u32 prealloc_fifos, segment_size = 256 << 20;
  echo_client_main_t *ecm = &echo_client_main;
  vnet_app_attach_args_t _a, *a = &_a;
  u64 options[18];
  int rv;

  clib_memset (a, 0, sizeof (*a));
  clib_memset (options, 0, sizeof (options));

  a->api_client_index = ~0;
  a->name = format (0, "echo_client");
  if (ecm->transport_proto == TRANSPORT_PROTO_QUIC)
    echo_clients.session_connected_callback =
      quic_echo_clients_session_connected_callback;
  a->session_cb_vft = &echo_clients;

  prealloc_fifos = ecm->prealloc_fifos ? ecm->expected_connections : 1;

  if (ecm->private_segment_size)
    segment_size = ecm->private_segment_size;

  options[APP_OPTIONS_ACCEPT_COOKIE] = 0x12345678;
  options[APP_OPTIONS_SEGMENT_SIZE] = segment_size;
  options[APP_OPTIONS_ADD_SEGMENT_SIZE] = segment_size;
  options[APP_OPTIONS_RX_FIFO_SIZE] = ecm->fifo_size;
  options[APP_OPTIONS_TX_FIFO_SIZE] = ecm->fifo_size;
  options[APP_OPTIONS_PRIVATE_SEGMENT_COUNT] = ecm->private_segment_count;
  options[APP_OPTIONS_PREALLOC_FIFO_PAIRS] = prealloc_fifos;
  options[APP_OPTIONS_FLAGS] = APP_OPTIONS_FLAGS_IS_BUILTIN;
  options[APP_OPTIONS_TLS_ENGINE] = ecm->tls_engine;
  options[APP_OPTIONS_PCT_FIRST_ALLOC] = 100;
  if (appns_id)
    {
      options[APP_OPTIONS_FLAGS] |= appns_flags;
      options[APP_OPTIONS_NAMESPACE_SECRET] = appns_secret;
    }
  a->options = options;
  a->namespace_id = appns_id;

  if ((rv = vnet_application_attach (a)))
    return clib_error_return (0, "attach returned %d", rv);

  ecm->app_index = a->app_index;
  vec_free (a->name);

  clib_memset (ck_pair, 0, sizeof (*ck_pair));
  ck_pair->cert = (u8 *) test_srv_crt_rsa;
  ck_pair->key = (u8 *) test_srv_key_rsa;
  ck_pair->cert_len = test_srv_crt_rsa_len;
  ck_pair->key_len = test_srv_key_rsa_len;
  vnet_app_add_cert_key_pair (ck_pair);
  ecm->ckpair_index = ck_pair->index;

  return 0;
}

static int
echo_clients_detach ()
{
  echo_client_main_t *ecm = &echo_client_main;
  vnet_app_detach_args_t _da, *da = &_da;
  int rv;

  da->app_index = ecm->app_index;
  da->api_client_index = ~0;
  rv = vnet_application_detach (da);
  ecm->test_client_attached = 0;
  ecm->app_index = ~0;
  vnet_app_del_cert_key_pair (ecm->ckpair_index);

  return rv;
}

static void *
echo_client_thread_fn (void *arg)
{
  return 0;
}

/** Start a transmit thread */
int
echo_clients_start_tx_pthread (echo_client_main_t * ecm)
{
  if (ecm->client_thread_handle == 0)
    {
      int rv = pthread_create (&ecm->client_thread_handle,
			       NULL /*attr */ ,
			       echo_client_thread_fn, 0);
      if (rv)
	{
	  ecm->client_thread_handle = 0;
	  return -1;
	}
    }
  return 0;
}

static int
echo_client_transport_needs_crypto (transport_proto_t proto)
{
  return proto == TRANSPORT_PROTO_TLS || proto == TRANSPORT_PROTO_DTLS ||
	 proto == TRANSPORT_PROTO_QUIC;
}

clib_error_t *
echo_clients_connect (vlib_main_t * vm, u32 n_clients)
{
  session_endpoint_cfg_t sep = SESSION_ENDPOINT_CFG_NULL;
  echo_client_main_t *ecm = &echo_client_main;
  vnet_connect_args_t _a, *a = &_a;
  int i, rv;

  clib_memset (a, 0, sizeof (*a));

  if (parse_uri ((char *) ecm->connect_uri, &sep))
    return clib_error_return (0, "invalid uri");

  for (i = 0; i < n_clients; i++)
    {
      clib_memcpy (&a->sep_ext, &sep, sizeof (sep));
      a->api_context = i;
      a->app_index = ecm->app_index;
      if (echo_client_transport_needs_crypto (a->sep_ext.transport_proto))
	{
	  session_endpoint_alloc_ext_cfg (&a->sep_ext,
					  TRANSPORT_ENDPT_EXT_CFG_CRYPTO);
	  a->sep_ext.ext_cfg->crypto.ckpair_index = ecm->ckpair_index;
	}

      vlib_worker_thread_barrier_sync (vm);
      rv = vnet_connect (a);
      if (a->sep_ext.ext_cfg)
	clib_mem_free (a->sep_ext.ext_cfg);
      if (rv)
	{
	  vlib_worker_thread_barrier_release (vm);
	  return clib_error_return (0, "connect returned: %d", rv);
	}
      vlib_worker_thread_barrier_release (vm);

      /* Crude pacing for call setups  */
      if ((i % 16) == 0)
	vlib_process_suspend (vm, 100e-6);
      ASSERT (i + 1 >= ecm->ready_connections);
      while (i + 1 - ecm->ready_connections > 128)
	vlib_process_suspend (vm, 1e-3);
    }
  return 0;
}

#define ec_cli_output(_fmt, _args...) 			\
  if (!ecm->no_output)  				\
    vlib_cli_output(vm, _fmt, ##_args)

static clib_error_t *
echo_clients_command_fn (vlib_main_t * vm,
			 unformat_input_t * input, vlib_cli_command_t * cmd)
{
  echo_client_main_t *ecm = &echo_client_main;
  vlib_thread_main_t *thread_main = vlib_get_thread_main ();
  u64 tmp, total_bytes, appns_flags = 0, appns_secret = 0;
  session_endpoint_cfg_t sep = SESSION_ENDPOINT_CFG_NULL;
  f64 test_timeout = 20.0, syn_timeout = 20.0, delta;
  char *default_uri = "tcp://6.0.1.1/1234";
  u8 *appns_id = 0, barrier_acq_needed = 0;
  int preallocate_sessions = 0, i, rv;
  uword *event_data = 0, event_type;
  f64 time_before_connects;
  u32 n_clients = 1;
  char *transfer_type;
  clib_error_t *error = 0;

  ecm->quic_streams = 1;
  ecm->bytes_to_send = 8192;
  ecm->no_return = 0;
  ecm->fifo_size = 64 << 10;
  ecm->connections_per_batch = 1000;
  ecm->private_segment_count = 0;
  ecm->private_segment_size = 0;
  ecm->no_output = 0;
  ecm->test_bytes = 0;
  ecm->test_failed = 0;
  ecm->vlib_main = vm;
  ecm->tls_engine = CRYPTO_ENGINE_OPENSSL;
  ecm->no_copy = 0;
  ecm->run_test = ECHO_CLIENTS_STARTING;

  if (vlib_num_workers ())
    {
      /* The request came over the binary api and the inband cli handler
       * is not mp_safe. Drop the barrier to make sure the workers are not
       * blocked.
       */
      if (vlib_thread_is_main_w_barrier ())
	{
	  barrier_acq_needed = 1;
	  vlib_worker_thread_barrier_release (vm);
	}
      /*
       * There's a good chance that both the client and the server echo
       * apps will be enabled so make sure the session queue node polls on
       * the main thread as connections will probably be established on it.
       */
      vlib_node_set_state (vm, session_queue_node.index,
			   VLIB_NODE_STATE_POLLING);
    }

  if (thread_main->n_vlib_mains > 1)
    clib_spinlock_init (&ecm->sessions_lock);
  vec_free (ecm->connect_uri);

  while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (input, "uri %s", &ecm->connect_uri))
	;
      else if (unformat (input, "nclients %d", &n_clients))
	;
      else if (unformat (input, "quic-streams %d", &ecm->quic_streams))
	;
      else if (unformat (input, "mbytes %lld", &tmp))
	ecm->bytes_to_send = tmp << 20;
      else if (unformat (input, "gbytes %lld", &tmp))
	ecm->bytes_to_send = tmp << 30;
      else if (unformat (input, "bytes %lld", &ecm->bytes_to_send))
	;
      else if (unformat (input, "test-timeout %f", &test_timeout))
	;
      else if (unformat (input, "syn-timeout %f", &syn_timeout))
	;
      else if (unformat (input, "no-return"))
	ecm->no_return = 1;
      else if (unformat (input, "fifo-size %d", &ecm->fifo_size))
	ecm->fifo_size <<= 10;
      else if (unformat (input, "private-segment-count %d",
			 &ecm->private_segment_count))
	;
      else if (unformat (input, "private-segment-size %U",
			 unformat_memory_size, &tmp))
	{
	  if (tmp >= 0x100000000ULL)
	    {
	      error = clib_error_return (
		0, "private segment size %lld (%llu) too large", tmp, tmp);
	      goto cleanup;
	    }
	  ecm->private_segment_size = tmp;
	}
      else if (unformat (input, "preallocate-fifos"))
	ecm->prealloc_fifos = 1;
      else if (unformat (input, "preallocate-sessions"))
	preallocate_sessions = 1;
      else
	if (unformat (input, "client-batch %d", &ecm->connections_per_batch))
	;
      else if (unformat (input, "appns %_%v%_", &appns_id))
	;
      else if (unformat (input, "all-scope"))
	appns_flags |= (APP_OPTIONS_FLAGS_USE_GLOBAL_SCOPE
			| APP_OPTIONS_FLAGS_USE_LOCAL_SCOPE);
      else if (unformat (input, "local-scope"))
	appns_flags = APP_OPTIONS_FLAGS_USE_LOCAL_SCOPE;
      else if (unformat (input, "global-scope"))
	appns_flags = APP_OPTIONS_FLAGS_USE_GLOBAL_SCOPE;
      else if (unformat (input, "secret %lu", &appns_secret))
	;
      else if (unformat (input, "no-output"))
	ecm->no_output = 1;
      else if (unformat (input, "test-bytes"))
	ecm->test_bytes = 1;
      else if (unformat (input, "tls-engine %d", &ecm->tls_engine))
	;
      else
	{
	  error = clib_error_return (0, "failed: unknown input `%U'",
				     format_unformat_error, input);
	  goto cleanup;
	}
    }

  /* Store cli process node index for signalling */
  ecm->cli_node_index =
    vlib_get_current_process (vm)->node_runtime.node_index;

  if (ecm->is_init == 0)
    {
      if (echo_clients_init (vm))
	{
	  error = clib_error_return (0, "failed init");
	  goto cleanup;
	}
    }


  ecm->ready_connections = 0;
  ecm->expected_connections = n_clients * ecm->quic_streams;
  ecm->rx_total = 0;
  ecm->tx_total = 0;

  if (!ecm->connect_uri)
    {
      clib_warning ("No uri provided. Using default: %s", default_uri);
      ecm->connect_uri = format (0, "%s%c", default_uri, 0);
    }

  if ((rv = parse_uri ((char *) ecm->connect_uri, &sep)))
    {
      error = clib_error_return (0, "Uri parse error: %d", rv);
      goto cleanup;
    }
  ecm->transport_proto = sep.transport_proto;
  ecm->is_dgram = (sep.transport_proto == TRANSPORT_PROTO_UDP);

#if ECHO_CLIENT_PTHREAD
  echo_clients_start_tx_pthread ();
#endif

  vlib_worker_thread_barrier_sync (vm);
  vnet_session_enable_disable (vm, 1 /* turn on session and transports */ );
  vlib_worker_thread_barrier_release (vm);

  if (ecm->test_client_attached == 0)
    {
      if ((error = echo_clients_attach (appns_id, appns_flags, appns_secret)))
	{
	  vec_free (appns_id);
	  clib_error_report (error);
	  goto cleanup;
	}
      vec_free (appns_id);
    }
  ecm->test_client_attached = 1;

  /* Turn on the builtin client input nodes */
  for (i = 0; i < thread_main->n_vlib_mains; i++)
    vlib_node_set_state (vlib_get_main_by_index (i), echo_clients_node.index,
			 VLIB_NODE_STATE_POLLING);

  if (preallocate_sessions)
    pool_init_fixed (ecm->sessions, 1.1 * n_clients);

  /* Fire off connect requests */
  time_before_connects = vlib_time_now (vm);
  if ((error = echo_clients_connect (vm, n_clients)))
    {
      goto cleanup;
    }

  /* Park until the sessions come up, or ten seconds elapse... */
  vlib_process_wait_for_event_or_clock (vm, syn_timeout);
  event_type = vlib_process_get_events (vm, &event_data);
  switch (event_type)
    {
    case ~0:
      ec_cli_output ("Timeout with only %d sessions active...",
		     ecm->ready_connections);
      error = clib_error_return (0, "failed: syn timeout with %d sessions",
				 ecm->ready_connections);
      goto cleanup;

    case 1:
      delta = vlib_time_now (vm) - time_before_connects;
      if (delta != 0.0)
	ec_cli_output ("%d three-way handshakes in %.2f seconds %.2f/s",
		       n_clients, delta, ((f64) n_clients) / delta);

      ecm->test_start_time = vlib_time_now (ecm->vlib_main);
      ec_cli_output ("Test started at %.6f", ecm->test_start_time);
      break;

    default:
      ec_cli_output ("unexpected event(1): %d", event_type);
      error = clib_error_return (0, "failed: unexpected event(1): %d",
				 event_type);
      goto cleanup;
    }

  /* Now wait for the sessions to finish... */
  vlib_process_wait_for_event_or_clock (vm, test_timeout);
  event_type = vlib_process_get_events (vm, &event_data);
  switch (event_type)
    {
    case ~0:
      ec_cli_output ("Timeout with %d sessions still active...",
		     ecm->ready_connections);
      error = clib_error_return (0, "failed: timeout with %d sessions",
				 ecm->ready_connections);
      goto cleanup;

    case 2:
      ecm->test_end_time = vlib_time_now (vm);
      ec_cli_output ("Test finished at %.6f", ecm->test_end_time);
      break;

    default:
      ec_cli_output ("unexpected event(2): %d", event_type);
      error = clib_error_return (0, "failed: unexpected event(2): %d",
				 event_type);
      goto cleanup;
    }

  delta = ecm->test_end_time - ecm->test_start_time;
  if (delta != 0.0)
    {
      total_bytes = (ecm->no_return ? ecm->tx_total : ecm->rx_total);
      transfer_type = ecm->no_return ? "half-duplex" : "full-duplex";
      ec_cli_output ("%lld bytes (%lld mbytes, %lld gbytes) in %.2f seconds",
		     total_bytes, total_bytes / (1ULL << 20),
		     total_bytes / (1ULL << 30), delta);
      ec_cli_output ("%.2f bytes/second %s", ((f64) total_bytes) / (delta),
		     transfer_type);
      ec_cli_output ("%.4f gbit/second %s",
		     (((f64) total_bytes * 8.0) / delta / 1e9),
		     transfer_type);
    }
  else
    {
      ec_cli_output ("zero delta-t?");
      error = clib_error_return (0, "failed: zero delta-t");
      goto cleanup;
    }

  if (ecm->test_bytes && ecm->test_failed)
    error = clib_error_return (0, "failed: test bytes");

cleanup:
  ecm->run_test = ECHO_CLIENTS_EXITING;
  vlib_process_wait_for_event_or_clock (vm, 10e-3);
  for (i = 0; i < vec_len (ecm->connection_index_by_thread); i++)
    {
      vec_reset_length (ecm->connection_index_by_thread[i]);
      vec_reset_length (ecm->connections_this_batch_by_thread[i]);
      vec_reset_length (ecm->quic_session_index_by_thread[i]);
    }

  pool_free (ecm->sessions);

  /* Detach the application, so we can use different fifo sizes next time */
  if (ecm->test_client_attached)
    {
      if (echo_clients_detach ())
	{
	  error = clib_error_return (0, "failed: app detach");
	  ec_cli_output ("WARNING: app detach failed...");
	}
    }
  if (error)
    ec_cli_output ("test failed");
  vec_free (ecm->connect_uri);
  clib_spinlock_free (&ecm->sessions_lock);

  if (barrier_acq_needed)
    vlib_worker_thread_barrier_sync (vm);

  return error;
}

/* *INDENT-OFF* */
VLIB_CLI_COMMAND (echo_clients_command, static) =
{
  .path = "test echo clients",
  .short_help = "test echo clients [nclients %d][[m|g]bytes <bytes>]"
      "[test-timeout <time>][syn-timeout <time>][no-return][fifo-size <size>]"
      "[private-segment-count <count>][private-segment-size <bytes>[m|g]]"
      "[preallocate-fifos][preallocate-sessions][client-batch <batch-size>]"
      "[uri <tcp://ip/port>][test-bytes][no-output]",
  .function = echo_clients_command_fn,
  .is_mp_safe = 1,
};
/* *INDENT-ON* */

clib_error_t *
echo_clients_main_init (vlib_main_t * vm)
{
  echo_client_main_t *ecm = &echo_client_main;
  ecm->is_init = 0;
  return 0;
}

VLIB_INIT_FUNCTION (echo_clients_main_init);

/*
 * fd.io coding-style-patch-verification: ON
 *
 * Local Variables:
 * eval: (c-set-style "gnu")
 * End:
 */
class="n">FIB_SOURCE_RR); path->fp_via_fib = FIB_NODE_INDEX_INVALID; } break; case FIB_PATH_TYPE_BIER_FMASK: bier_fmask_child_remove(path->fp_via_bier_fmask, path->fp_sibling); break; case FIB_PATH_TYPE_BIER_IMP: bier_imp_unlock(path->fp_dpo.dpoi_index); break; case FIB_PATH_TYPE_BIER_TABLE: bier_table_ecmp_unlock(path->fp_via_bier_tbl); break; case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: case FIB_PATH_TYPE_ATTACHED: if (dpo_is_adj(&path->fp_dpo)) adj_child_remove(path->fp_dpo.dpoi_index, path->fp_sibling); break; case FIB_PATH_TYPE_UDP_ENCAP: udp_encap_unlock(path->fp_dpo.dpoi_index); break; case FIB_PATH_TYPE_EXCLUSIVE: dpo_reset(&path->exclusive.fp_ex_dpo); break; case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_DEAG: case FIB_PATH_TYPE_DVR: /* * these hold only the path's DPO, which is reset below. */ break; } /* * release the adj we were holding and pick up the * drop just in case. */ dpo_reset(&path->fp_dpo); path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; return; } static fib_forward_chain_type_t fib_path_to_chain_type (const fib_path_t *path) { if (DPO_PROTO_MPLS == path->fp_nh_proto) { if (FIB_PATH_TYPE_RECURSIVE == path->fp_type && MPLS_EOS == path->recursive.fp_nh.fp_eos) { return (FIB_FORW_CHAIN_TYPE_MPLS_EOS); } else { return (FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS); } } else { return (fib_forw_chain_type_from_dpo_proto(path->fp_nh_proto)); } } /* * fib_path_back_walk_notify * * A back walk has reach this path. */ static fib_node_back_walk_rc_t fib_path_back_walk_notify (fib_node_t *node, fib_node_back_walk_ctx_t *ctx) { fib_path_t *path; path = fib_path_from_fib_node(node); FIB_PATH_DBG(path, "bw:%U", format_fib_node_bw_reason, ctx->fnbw_reason); switch (path->fp_type) { case FIB_PATH_TYPE_RECURSIVE: if (FIB_NODE_BW_REASON_FLAG_EVALUATE & ctx->fnbw_reason) { /* * modify the recursive adjacency to use the new forwarding * of the via-fib. * this update is visible to packets in flight in the DP. */ fib_path_recursive_adj_update( path, fib_path_to_chain_type(path), &path->fp_dpo); } if ((FIB_NODE_BW_REASON_FLAG_ADJ_UPDATE & ctx->fnbw_reason) || (FIB_NODE_BW_REASON_FLAG_ADJ_MTU & ctx->fnbw_reason) || (FIB_NODE_BW_REASON_FLAG_ADJ_DOWN & ctx->fnbw_reason)) { /* * ADJ updates (complete<->incomplete) do not need to propagate to * recursive entries. * The only reason its needed as far back as here, is that the adj * and the incomplete adj are a different DPO type, so the LBs need * to re-stack. * If this walk was quashed in the fib_entry, then any non-fib_path * children (like tunnels that collapse out the LB when they stack) * would not see the update. */ return (FIB_NODE_BACK_WALK_CONTINUE); } break; case FIB_PATH_TYPE_BIER_FMASK: if (FIB_NODE_BW_REASON_FLAG_EVALUATE & ctx->fnbw_reason) { /* * update to use the BIER fmask's new forwading */ fib_path_bier_fmask_update(path, &path->fp_dpo); } if ((FIB_NODE_BW_REASON_FLAG_ADJ_UPDATE & ctx->fnbw_reason) || (FIB_NODE_BW_REASON_FLAG_ADJ_DOWN & ctx->fnbw_reason)) { /* * ADJ updates (complete<->incomplete) do not need to propagate to * recursive entries. * The only reason its needed as far back as here, is that the adj * and the incomplete adj are a different DPO type, so the LBs need * to re-stack. * If this walk was quashed in the fib_entry, then any non-fib_path * children (like tunnels that collapse out the LB when they stack) * would not see the update. */ return (FIB_NODE_BACK_WALK_CONTINUE); } break; case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: /* FIXME comment * ADJ_UPDATE backwalk pass silently through here and up to * the path-list when the multipath adj collapse occurs. * The reason we do this is that the assumtption is that VPP * runs in an environment where the Control-Plane is remote * and hence reacts slowly to link up down. In order to remove * this down link from the ECMP set quickly, we back-walk. * VPP also has dedicated CPUs, so we are not stealing resources * from the CP to do so. */ if (FIB_NODE_BW_REASON_FLAG_INTERFACE_UP & ctx->fnbw_reason) { if (path->fp_oper_flags & FIB_PATH_OPER_FLAG_RESOLVED) { /* * alreday resolved. no need to walk back again */ return (FIB_NODE_BACK_WALK_CONTINUE); } path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RESOLVED; } if (FIB_NODE_BW_REASON_FLAG_INTERFACE_DOWN & ctx->fnbw_reason) { if (!(path->fp_oper_flags & FIB_PATH_OPER_FLAG_RESOLVED)) { /* * alreday unresolved. no need to walk back again */ return (FIB_NODE_BACK_WALK_CONTINUE); } path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; } if (FIB_NODE_BW_REASON_FLAG_INTERFACE_DELETE & ctx->fnbw_reason) { /* * The interface this path resolves through has been deleted. * This will leave the path in a permanent drop state. The route * needs to be removed and readded (and hence the path-list deleted) * before it can forward again. */ fib_path_unresolve(path); path->fp_oper_flags |= FIB_PATH_OPER_FLAG_DROP; } if (FIB_NODE_BW_REASON_FLAG_ADJ_UPDATE & ctx->fnbw_reason) { /* * restack the DPO to pick up the correct DPO sub-type */ dpo_id_t tmp = DPO_INVALID; uword if_is_up; if_is_up = vnet_sw_interface_is_up( vnet_get_main(), path->attached_next_hop.fp_interface); dpo_copy (&tmp, &path->fp_dpo); path = fib_path_attached_next_hop_get_adj( path, dpo_proto_to_link(path->fp_nh_proto), &tmp); dpo_copy(&path->fp_dpo, &tmp); dpo_reset(&tmp); path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; if (if_is_up && adj_is_up(path->fp_dpo.dpoi_index)) { path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RESOLVED; } if (!if_is_up) { /* * If the interface is not up there is no reason to walk * back to children. if we did they would only evalute * that this path is unresolved and hence it would * not contribute the adjacency - so it would be wasted * CPU time. */ return (FIB_NODE_BACK_WALK_CONTINUE); } } if (FIB_NODE_BW_REASON_FLAG_ADJ_DOWN & ctx->fnbw_reason) { if (!(path->fp_oper_flags & FIB_PATH_OPER_FLAG_RESOLVED)) { /* * alreday unresolved. no need to walk back again */ return (FIB_NODE_BACK_WALK_CONTINUE); } /* * the adj has gone down. the path is no longer resolved. */ path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; } break; case FIB_PATH_TYPE_ATTACHED: case FIB_PATH_TYPE_DVR: /* * FIXME; this could schedule a lower priority walk, since attached * routes are not usually in ECMP configurations so the backwalk to * the FIB entry does not need to be high priority */ if (FIB_NODE_BW_REASON_FLAG_INTERFACE_UP & ctx->fnbw_reason) { path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RESOLVED; } if (FIB_NODE_BW_REASON_FLAG_INTERFACE_DOWN & ctx->fnbw_reason) { path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; } if (FIB_NODE_BW_REASON_FLAG_INTERFACE_DELETE & ctx->fnbw_reason) { fib_path_unresolve(path); path->fp_oper_flags |= FIB_PATH_OPER_FLAG_DROP; } if (FIB_NODE_BW_REASON_FLAG_INTERFACE_BIND & ctx->fnbw_reason) { /* bind walks should appear here and pass silently up to * to the fib_entry */ } break; case FIB_PATH_TYPE_UDP_ENCAP: { dpo_id_t via_dpo = DPO_INVALID; /* * hope for the best - clear if restrictions apply. */ path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RESOLVED; udp_encap_contribute_forwarding(path->udp_encap.fp_udp_encap_id, path->fp_nh_proto, &via_dpo); /* * If this path is contributing a drop, then it's not resolved */ if (dpo_is_drop(&via_dpo) || load_balance_is_drop(&via_dpo)) { path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; } /* * update the path's contributed DPO */ dpo_copy(&path->fp_dpo, &via_dpo); dpo_reset(&via_dpo); break; } case FIB_PATH_TYPE_INTF_RX: ASSERT(0); case FIB_PATH_TYPE_DEAG: /* * FIXME When VRF delete is allowed this will need a poke. */ case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_EXCLUSIVE: case FIB_PATH_TYPE_BIER_TABLE: case FIB_PATH_TYPE_BIER_IMP: /* * these path types have no parents. so to be * walked from one is unexpected. */ ASSERT(0); break; } /* * propagate the backwalk further to the path-list */ fib_path_list_back_walk(path->fp_pl_index, ctx); return (FIB_NODE_BACK_WALK_CONTINUE); } static void fib_path_memory_show (void) { fib_show_memory_usage("Path", pool_elts(fib_path_pool), pool_len(fib_path_pool), sizeof(fib_path_t)); } /* * The FIB path's graph node virtual function table */ static const fib_node_vft_t fib_path_vft = { .fnv_get = fib_path_get_node, .fnv_last_lock = fib_path_last_lock_gone, .fnv_back_walk = fib_path_back_walk_notify, .fnv_mem_show = fib_path_memory_show, }; static fib_path_cfg_flags_t fib_path_route_flags_to_cfg_flags (const fib_route_path_t *rpath) { fib_path_cfg_flags_t cfg_flags = FIB_PATH_CFG_FLAG_NONE; if (rpath->frp_flags & FIB_ROUTE_PATH_POP_PW_CW) cfg_flags |= FIB_PATH_CFG_FLAG_POP_PW_CW; if (rpath->frp_flags & FIB_ROUTE_PATH_RESOLVE_VIA_HOST) cfg_flags |= FIB_PATH_CFG_FLAG_RESOLVE_HOST; if (rpath->frp_flags & FIB_ROUTE_PATH_RESOLVE_VIA_ATTACHED) cfg_flags |= FIB_PATH_CFG_FLAG_RESOLVE_ATTACHED; if (rpath->frp_flags & FIB_ROUTE_PATH_LOCAL) cfg_flags |= FIB_PATH_CFG_FLAG_LOCAL; if (rpath->frp_flags & FIB_ROUTE_PATH_ATTACHED) cfg_flags |= FIB_PATH_CFG_FLAG_ATTACHED; if (rpath->frp_flags & FIB_ROUTE_PATH_INTF_RX) cfg_flags |= FIB_PATH_CFG_FLAG_INTF_RX; if (rpath->frp_flags & FIB_ROUTE_PATH_RPF_ID) cfg_flags |= FIB_PATH_CFG_FLAG_RPF_ID; if (rpath->frp_flags & FIB_ROUTE_PATH_EXCLUSIVE) cfg_flags |= FIB_PATH_CFG_FLAG_EXCLUSIVE; if (rpath->frp_flags & FIB_ROUTE_PATH_DROP) cfg_flags |= FIB_PATH_CFG_FLAG_DROP; if (rpath->frp_flags & FIB_ROUTE_PATH_SOURCE_LOOKUP) cfg_flags |= FIB_PATH_CFG_FLAG_DEAG_SRC; if (rpath->frp_flags & FIB_ROUTE_PATH_ICMP_UNREACH) cfg_flags |= FIB_PATH_CFG_FLAG_ICMP_UNREACH; if (rpath->frp_flags & FIB_ROUTE_PATH_ICMP_PROHIBIT) cfg_flags |= FIB_PATH_CFG_FLAG_ICMP_PROHIBIT; if (rpath->frp_flags & FIB_ROUTE_PATH_GLEAN) cfg_flags |= FIB_PATH_CFG_FLAG_GLEAN; return (cfg_flags); } /* * fib_path_create * * Create and initialise a new path object. * return the index of the path. */ fib_node_index_t fib_path_create (fib_node_index_t pl_index, const fib_route_path_t *rpath) { fib_path_t *path; pool_get(fib_path_pool, path); clib_memset(path, 0, sizeof(*path)); fib_node_init(&path->fp_node, FIB_NODE_TYPE_PATH); dpo_reset(&path->fp_dpo); path->fp_pl_index = pl_index; path->fp_nh_proto = rpath->frp_proto; path->fp_via_fib = FIB_NODE_INDEX_INVALID; path->fp_weight = rpath->frp_weight; if (0 == path->fp_weight) { /* * a weight of 0 is a meaningless value. We could either reject it, and thus force * clients to always use 1, or we can accept it and fixup approrpiately. */ path->fp_weight = 1; } path->fp_preference = rpath->frp_preference; path->fp_cfg_flags = fib_path_route_flags_to_cfg_flags(rpath); /* * deduce the path's tpye from the parementers and save what is needed. */ if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_LOCAL) { path->fp_type = FIB_PATH_TYPE_RECEIVE; path->receive.fp_interface = rpath->frp_sw_if_index; path->receive.fp_addr = rpath->frp_addr; } else if (rpath->frp_flags & FIB_ROUTE_PATH_UDP_ENCAP) { path->fp_type = FIB_PATH_TYPE_UDP_ENCAP; path->udp_encap.fp_udp_encap_id = rpath->frp_udp_encap_id; } else if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_INTF_RX) { path->fp_type = FIB_PATH_TYPE_INTF_RX; path->intf_rx.fp_interface = rpath->frp_sw_if_index; } else if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_RPF_ID) { path->fp_type = FIB_PATH_TYPE_DEAG; path->deag.fp_tbl_id = rpath->frp_fib_index; path->deag.fp_rpf_id = rpath->frp_rpf_id; } else if (rpath->frp_flags & FIB_ROUTE_PATH_BIER_FMASK) { path->fp_type = FIB_PATH_TYPE_BIER_FMASK; path->bier_fmask.fp_bier_fmask = rpath->frp_bier_fmask; } else if (rpath->frp_flags & FIB_ROUTE_PATH_BIER_IMP) { path->fp_type = FIB_PATH_TYPE_BIER_IMP; path->bier_imp.fp_bier_imp = rpath->frp_bier_imp; } else if (rpath->frp_flags & FIB_ROUTE_PATH_BIER_TABLE) { path->fp_type = FIB_PATH_TYPE_BIER_TABLE; path->bier_table.fp_bier_tbl = rpath->frp_bier_tbl; } else if (rpath->frp_flags & FIB_ROUTE_PATH_DEAG) { path->fp_type = FIB_PATH_TYPE_DEAG; path->deag.fp_tbl_id = rpath->frp_fib_index; } else if (rpath->frp_flags & FIB_ROUTE_PATH_DVR) { path->fp_type = FIB_PATH_TYPE_DVR; path->dvr.fp_interface = rpath->frp_sw_if_index; } else if (rpath->frp_flags & FIB_ROUTE_PATH_EXCLUSIVE) { path->fp_type = FIB_PATH_TYPE_EXCLUSIVE; dpo_copy(&path->exclusive.fp_ex_dpo, &rpath->dpo); } else if ((path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_PROHIBIT) || (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_UNREACH) || (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_DROP)) { path->fp_type = FIB_PATH_TYPE_SPECIAL; } else if ((path->fp_cfg_flags & FIB_PATH_CFG_FLAG_CLASSIFY)) { path->fp_type = FIB_PATH_TYPE_SPECIAL; path->classify.fp_classify_table_id = rpath->frp_classify_table_id; } else if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_GLEAN) { path->fp_type = FIB_PATH_TYPE_ATTACHED; path->attached.fp_interface = rpath->frp_sw_if_index; path->attached.fp_connected = rpath->frp_connected; } else if (~0 != rpath->frp_sw_if_index) { if (ip46_address_is_zero(&rpath->frp_addr)) { path->fp_type = FIB_PATH_TYPE_ATTACHED; path->attached.fp_interface = rpath->frp_sw_if_index; } else { path->fp_type = FIB_PATH_TYPE_ATTACHED_NEXT_HOP; path->attached_next_hop.fp_interface = rpath->frp_sw_if_index; path->attached_next_hop.fp_nh = rpath->frp_addr; } } else { if (ip46_address_is_zero(&rpath->frp_addr)) { if (~0 == rpath->frp_fib_index) { path->fp_type = FIB_PATH_TYPE_SPECIAL; } else { path->fp_type = FIB_PATH_TYPE_DEAG; path->deag.fp_tbl_id = rpath->frp_fib_index; path->deag.fp_rpf_id = ~0; } } else { path->fp_type = FIB_PATH_TYPE_RECURSIVE; if (DPO_PROTO_MPLS == path->fp_nh_proto) { path->recursive.fp_nh.fp_local_label = rpath->frp_local_label; path->recursive.fp_nh.fp_eos = rpath->frp_eos; } else { path->recursive.fp_nh.fp_ip = rpath->frp_addr; } path->recursive.fp_tbl_id = rpath->frp_fib_index; } } FIB_PATH_DBG(path, "create"); return (fib_path_get_index(path)); } /* * fib_path_create_special * * Create and initialise a new path object. * return the index of the path. */ fib_node_index_t fib_path_create_special (fib_node_index_t pl_index, dpo_proto_t nh_proto, fib_path_cfg_flags_t flags, const dpo_id_t *dpo) { fib_path_t *path; pool_get(fib_path_pool, path); clib_memset(path, 0, sizeof(*path)); fib_node_init(&path->fp_node, FIB_NODE_TYPE_PATH); dpo_reset(&path->fp_dpo); path->fp_pl_index = pl_index; path->fp_weight = 1; path->fp_preference = 0; path->fp_nh_proto = nh_proto; path->fp_via_fib = FIB_NODE_INDEX_INVALID; path->fp_cfg_flags = flags; if (FIB_PATH_CFG_FLAG_DROP & flags) { path->fp_type = FIB_PATH_TYPE_SPECIAL; } else if (FIB_PATH_CFG_FLAG_LOCAL & flags) { path->fp_type = FIB_PATH_TYPE_RECEIVE; path->attached.fp_interface = FIB_NODE_INDEX_INVALID; } else { path->fp_type = FIB_PATH_TYPE_EXCLUSIVE; ASSERT(NULL != dpo); dpo_copy(&path->exclusive.fp_ex_dpo, dpo); } return (fib_path_get_index(path)); } /* * fib_path_copy * * Copy a path. return index of new path. */ fib_node_index_t fib_path_copy (fib_node_index_t path_index, fib_node_index_t path_list_index) { fib_path_t *path, *orig_path; pool_get(fib_path_pool, path); orig_path = fib_path_get(path_index); ASSERT(NULL != orig_path); clib_memcpy(path, orig_path, sizeof(*path)); FIB_PATH_DBG(path, "create-copy:%d", path_index); /* * reset the dynamic section */ fib_node_init(&path->fp_node, FIB_NODE_TYPE_PATH); path->fp_oper_flags = FIB_PATH_OPER_FLAG_NONE; path->fp_pl_index = path_list_index; path->fp_via_fib = FIB_NODE_INDEX_INVALID; clib_memset(&path->fp_dpo, 0, sizeof(path->fp_dpo)); dpo_reset(&path->fp_dpo); if (path->fp_type == FIB_PATH_TYPE_EXCLUSIVE) { clib_memset(&path->exclusive.fp_ex_dpo, 0, sizeof(dpo_id_t)); dpo_copy(&path->exclusive.fp_ex_dpo, &orig_path->exclusive.fp_ex_dpo); } return (fib_path_get_index(path)); } /* * fib_path_destroy * * destroy a path that is no longer required */ void fib_path_destroy (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(NULL != path); FIB_PATH_DBG(path, "destroy"); fib_path_unresolve(path); fib_node_deinit(&path->fp_node); pool_put(fib_path_pool, path); } /* * fib_path_destroy * * destroy a path that is no longer required */ uword fib_path_hash (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (hash_memory(STRUCT_MARK_PTR(path, path_hash_start), (STRUCT_OFFSET_OF(fib_path_t, path_hash_end) - STRUCT_OFFSET_OF(fib_path_t, path_hash_start)), 0)); } /* * fib_path_cmp_i * * Compare two paths for equivalence. */ static int fib_path_cmp_i (const fib_path_t *path1, const fib_path_t *path2) { int res; res = 1; /* * paths of different types and protocol are not equal. * different weights and/or preference only are the same path. */ if (path1->fp_type != path2->fp_type) { res = (path1->fp_type - path2->fp_type); } else if (path1->fp_nh_proto != path2->fp_nh_proto) { res = (path1->fp_nh_proto - path2->fp_nh_proto); } else { /* * both paths are of the same type. * consider each type and its attributes in turn. */ switch (path1->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: res = ip46_address_cmp(&path1->attached_next_hop.fp_nh, &path2->attached_next_hop.fp_nh); if (0 == res) { res = (path1->attached_next_hop.fp_interface - path2->attached_next_hop.fp_interface); } break; case FIB_PATH_TYPE_ATTACHED: res = (path1->attached.fp_interface - path2->attached.fp_interface); break; case FIB_PATH_TYPE_RECURSIVE: res = ip46_address_cmp(&path1->recursive.fp_nh.fp_ip, &path2->recursive.fp_nh.fp_ip); if (0 == res) { res = (path1->recursive.fp_tbl_id - path2->recursive.fp_tbl_id); } break; case FIB_PATH_TYPE_BIER_FMASK: res = (path1->bier_fmask.fp_bier_fmask - path2->bier_fmask.fp_bier_fmask); break; case FIB_PATH_TYPE_BIER_IMP: res = (path1->bier_imp.fp_bier_imp - path2->bier_imp.fp_bier_imp); break; case FIB_PATH_TYPE_BIER_TABLE: res = bier_table_id_cmp(&path1->bier_table.fp_bier_tbl, &path2->bier_table.fp_bier_tbl); break; case FIB_PATH_TYPE_DEAG: res = (path1->deag.fp_tbl_id - path2->deag.fp_tbl_id); if (0 == res) { res = (path1->deag.fp_rpf_id - path2->deag.fp_rpf_id); } break; case FIB_PATH_TYPE_INTF_RX: res = (path1->intf_rx.fp_interface - path2->intf_rx.fp_interface); break; case FIB_PATH_TYPE_UDP_ENCAP: res = (path1->udp_encap.fp_udp_encap_id - path2->udp_encap.fp_udp_encap_id); break; case FIB_PATH_TYPE_DVR: res = (path1->dvr.fp_interface - path2->dvr.fp_interface); break; case FIB_PATH_TYPE_EXCLUSIVE: res = dpo_cmp(&path1->exclusive.fp_ex_dpo, &path2->exclusive.fp_ex_dpo); break; case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_RECEIVE: res = 0; break; } } return (res); } /* * fib_path_cmp_for_sort * * Compare two paths for equivalence. Used during path sorting. * As usual 0 means equal. */ int fib_path_cmp_for_sort (void * v1, void * v2) { fib_node_index_t *pi1 = v1, *pi2 = v2; fib_path_t *path1, *path2; path1 = fib_path_get(*pi1); path2 = fib_path_get(*pi2); /* * when sorting paths we want the highest preference paths * first, so that the choices set built is in prefernce order */ if (path1->fp_preference != path2->fp_preference) { return (path1->fp_preference - path2->fp_preference); } return (fib_path_cmp_i(path1, path2)); } /* * fib_path_cmp * * Compare two paths for equivalence. */ int fib_path_cmp (fib_node_index_t pi1, fib_node_index_t pi2) { fib_path_t *path1, *path2; path1 = fib_path_get(pi1); path2 = fib_path_get(pi2); return (fib_path_cmp_i(path1, path2)); } int fib_path_cmp_w_route_path (fib_node_index_t path_index, const fib_route_path_t *rpath) { fib_path_t *path; int res; path = fib_path_get(path_index); res = 1; if (path->fp_weight != rpath->frp_weight) { res = (path->fp_weight - rpath->frp_weight); } else { /* * both paths are of the same type. * consider each type and its attributes in turn. */ switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: res = ip46_address_cmp(&path->attached_next_hop.fp_nh, &rpath->frp_addr); if (0 == res) { res = (path->attached_next_hop.fp_interface - rpath->frp_sw_if_index); } break; case FIB_PATH_TYPE_ATTACHED: res = (path->attached.fp_interface - rpath->frp_sw_if_index); break; case FIB_PATH_TYPE_RECURSIVE: if (DPO_PROTO_MPLS == path->fp_nh_proto) { res = path->recursive.fp_nh.fp_local_label - rpath->frp_local_label; if (res == 0) { res = path->recursive.fp_nh.fp_eos - rpath->frp_eos; } } else { res = ip46_address_cmp(&path->recursive.fp_nh.fp_ip, &rpath->frp_addr); } if (0 == res) { res = (path->recursive.fp_tbl_id - rpath->frp_fib_index); } break; case FIB_PATH_TYPE_BIER_FMASK: res = (path->bier_fmask.fp_bier_fmask - rpath->frp_bier_fmask); break; case FIB_PATH_TYPE_BIER_IMP: res = (path->bier_imp.fp_bier_imp - rpath->frp_bier_imp); break; case FIB_PATH_TYPE_BIER_TABLE: res = bier_table_id_cmp(&path->bier_table.fp_bier_tbl, &rpath->frp_bier_tbl); break; case FIB_PATH_TYPE_INTF_RX: res = (path->intf_rx.fp_interface - rpath->frp_sw_if_index); break; case FIB_PATH_TYPE_UDP_ENCAP: res = (path->udp_encap.fp_udp_encap_id - rpath->frp_udp_encap_id); break; case FIB_PATH_TYPE_DEAG: res = (path->deag.fp_tbl_id - rpath->frp_fib_index); if (0 == res) { res = (path->deag.fp_rpf_id - rpath->frp_rpf_id); } break; case FIB_PATH_TYPE_DVR: res = (path->dvr.fp_interface - rpath->frp_sw_if_index); break; case FIB_PATH_TYPE_EXCLUSIVE: res = dpo_cmp(&path->exclusive.fp_ex_dpo, &rpath->dpo); break; case FIB_PATH_TYPE_RECEIVE: if (rpath->frp_flags & FIB_ROUTE_PATH_LOCAL) { res = 0; } else { res = 1; } break; case FIB_PATH_TYPE_SPECIAL: res = 0; break; } } return (res); } /* * fib_path_recursive_loop_detect * * A forward walk of the FIB object graph to detect for a cycle/loop. This * walk is initiated when an entry is linking to a new path list or from an old. * The entry vector passed contains all the FIB entrys that are children of this * path (it is all the entries encountered on the walk so far). If this vector * contains the entry this path resolve via, then a loop is about to form. * The loop must be allowed to form, since we need the dependencies in place * so that we can track when the loop breaks. * However, we MUST not produce a loop in the forwarding graph (else packets * would loop around the switch path until the loop breaks), so we mark recursive * paths as looped so that they do not contribute forwarding information. * By marking the path as looped, an etry such as; * X/Y * via a.a.a.a (looped) * via b.b.b.b (not looped) * can still forward using the info provided by b.b.b.b only */ int fib_path_recursive_loop_detect (fib_node_index_t path_index, fib_node_index_t **entry_indicies) { fib_path_t *path; path = fib_path_get(path_index); /* * the forced drop path is never looped, cos it is never resolved. */ if (fib_path_is_permanent_drop(path)) { return (0); } switch (path->fp_type) { case FIB_PATH_TYPE_RECURSIVE: { fib_node_index_t *entry_index, *entries; int looped = 0; entries = *entry_indicies; vec_foreach(entry_index, entries) { if (*entry_index == path->fp_via_fib) { /* * the entry that is about to link to this path-list (or * one of this path-list's children) is the same entry that * this recursive path resolves through. this is a cycle. * abort the walk. */ looped = 1; break; } } if (looped) { FIB_PATH_DBG(path, "recursive loop formed"); path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RECURSIVE_LOOP; dpo_copy(&path->fp_dpo, drop_dpo_get(path->fp_nh_proto)); } else { /* * no loop here yet. keep forward walking the graph. */ if (fib_entry_recursive_loop_detect(path->fp_via_fib, entry_indicies)) { FIB_PATH_DBG(path, "recursive loop formed"); path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RECURSIVE_LOOP; } else { FIB_PATH_DBG(path, "recursive loop cleared"); path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RECURSIVE_LOOP; } } break; } case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: case FIB_PATH_TYPE_ATTACHED: if (dpo_is_adj(&path->fp_dpo) && adj_recursive_loop_detect(path->fp_dpo.dpoi_index, entry_indicies)) { FIB_PATH_DBG(path, "recursive loop formed"); path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RECURSIVE_LOOP; } else { FIB_PATH_DBG(path, "recursive loop cleared"); path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RECURSIVE_LOOP; } break; case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_DEAG: case FIB_PATH_TYPE_DVR: case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_UDP_ENCAP: case FIB_PATH_TYPE_EXCLUSIVE: case FIB_PATH_TYPE_BIER_FMASK: case FIB_PATH_TYPE_BIER_TABLE: case FIB_PATH_TYPE_BIER_IMP: /* * these path types cannot be part of a loop, since they are the leaves * of the graph. */ break; } return (fib_path_is_looped(path_index)); } int fib_path_resolve (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); /* * hope for the best. */ path->fp_oper_flags |= FIB_PATH_OPER_FLAG_RESOLVED; /* * the forced drop path resolves via the drop adj */ if (fib_path_is_permanent_drop(path)) { dpo_copy(&path->fp_dpo, drop_dpo_get(path->fp_nh_proto)); path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; return (fib_path_is_resolved(path_index)); } switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: fib_path_attached_next_hop_set(path); break; case FIB_PATH_TYPE_ATTACHED: { dpo_id_t tmp = DPO_INVALID; /* * path->attached.fp_interface */ if (!vnet_sw_interface_is_up(vnet_get_main(), path->attached.fp_interface)) { path->fp_oper_flags &= ~FIB_PATH_OPER_FLAG_RESOLVED; } fib_path_attached_get_adj(path, dpo_proto_to_link(path->fp_nh_proto), &tmp); /* * re-fetch after possible mem realloc */ path = fib_path_get(path_index); dpo_copy(&path->fp_dpo, &tmp); /* * become a child of the adjacency so we receive updates * when the interface state changes */ if (dpo_is_adj(&path->fp_dpo)) { path->fp_sibling = adj_child_add(path->fp_dpo.dpoi_index, FIB_NODE_TYPE_PATH, fib_path_get_index(path)); } dpo_reset(&tmp); break; } case FIB_PATH_TYPE_RECURSIVE: { /* * Create a RR source entry in the table for the address * that this path recurses through. * This resolve action is recursive, hence we may create * more paths in the process. more creates mean maybe realloc * of this path. */ fib_node_index_t fei; fib_prefix_t pfx; ASSERT(FIB_NODE_INDEX_INVALID == path->fp_via_fib); if (DPO_PROTO_MPLS == path->fp_nh_proto) { fib_prefix_from_mpls_label(path->recursive.fp_nh.fp_local_label, path->recursive.fp_nh.fp_eos, &pfx); } else { ASSERT(!ip46_address_is_zero(&path->recursive.fp_nh.fp_ip)); fib_protocol_t fp = (ip46_address_is_ip4(&path->recursive.fp_nh.fp_ip) ? FIB_PROTOCOL_IP4 : FIB_PROTOCOL_IP6); fib_prefix_from_ip46_addr(fp, &path->recursive.fp_nh.fp_ip, &pfx); } fib_table_lock(path->recursive.fp_tbl_id, dpo_proto_to_fib(path->fp_nh_proto), FIB_SOURCE_RR); fei = fib_table_entry_special_add(path->recursive.fp_tbl_id, &pfx, FIB_SOURCE_RR, FIB_ENTRY_FLAG_NONE); path = fib_path_get(path_index); path->fp_via_fib = fei; /* * become a dependent child of the entry so the path is * informed when the forwarding for the entry changes. */ path->fp_sibling = fib_entry_child_add(path->fp_via_fib, FIB_NODE_TYPE_PATH, fib_path_get_index(path)); /* * create and configure the IP DPO */ fib_path_recursive_adj_update( path, fib_path_to_chain_type(path), &path->fp_dpo); break; } case FIB_PATH_TYPE_BIER_FMASK: { /* * become a dependent child of the entry so the path is * informed when the forwarding for the entry changes. */ path->fp_sibling = bier_fmask_child_add(path->bier_fmask.fp_bier_fmask, FIB_NODE_TYPE_PATH, fib_path_get_index(path)); path->fp_via_bier_fmask = path->bier_fmask.fp_bier_fmask; fib_path_bier_fmask_update(path, &path->fp_dpo); break; } case FIB_PATH_TYPE_BIER_IMP: bier_imp_lock(path->bier_imp.fp_bier_imp); bier_imp_contribute_forwarding(path->bier_imp.fp_bier_imp, DPO_PROTO_IP4, &path->fp_dpo); break; case FIB_PATH_TYPE_BIER_TABLE: { /* * Find/create the BIER table to link to */ ASSERT(FIB_NODE_INDEX_INVALID == path->fp_via_bier_tbl); path->fp_via_bier_tbl = bier_table_ecmp_create_and_lock(&path->bier_table.fp_bier_tbl); bier_table_contribute_forwarding(path->fp_via_bier_tbl, &path->fp_dpo); break; } case FIB_PATH_TYPE_SPECIAL: if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_PROHIBIT) { ip_null_dpo_add_and_lock (path->fp_nh_proto, IP_NULL_ACTION_SEND_ICMP_PROHIBIT, &path->fp_dpo); } else if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_UNREACH) { ip_null_dpo_add_and_lock (path->fp_nh_proto, IP_NULL_ACTION_SEND_ICMP_UNREACH, &path->fp_dpo); } else if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_CLASSIFY) { dpo_set (&path->fp_dpo, DPO_CLASSIFY, path->fp_nh_proto, classify_dpo_create (path->fp_nh_proto, path->classify.fp_classify_table_id)); } else { /* * Resolve via the drop */ dpo_copy(&path->fp_dpo, drop_dpo_get(path->fp_nh_proto)); } break; case FIB_PATH_TYPE_DEAG: { if (DPO_PROTO_BIER == path->fp_nh_proto) { bier_disp_table_contribute_forwarding(path->deag.fp_tbl_id, &path->fp_dpo); } else { /* * Resolve via a lookup DPO. * FIXME. control plane should add routes with a table ID */ lookup_input_t input; lookup_cast_t cast; cast = (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_RPF_ID ? LOOKUP_MULTICAST : LOOKUP_UNICAST); input = (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_DEAG_SRC ? LOOKUP_INPUT_SRC_ADDR : LOOKUP_INPUT_DST_ADDR); lookup_dpo_add_or_lock_w_fib_index(path->deag.fp_tbl_id, path->fp_nh_proto, cast, input, LOOKUP_TABLE_FROM_CONFIG, &path->fp_dpo); } break; } case FIB_PATH_TYPE_DVR: dvr_dpo_add_or_lock(path->dvr.fp_interface, path->fp_nh_proto, &path->fp_dpo); break; case FIB_PATH_TYPE_RECEIVE: /* * Resolve via a receive DPO. */ receive_dpo_add_or_lock(path->fp_nh_proto, path->receive.fp_interface, &path->receive.fp_addr, &path->fp_dpo); break; case FIB_PATH_TYPE_UDP_ENCAP: udp_encap_lock(path->udp_encap.fp_udp_encap_id); udp_encap_contribute_forwarding(path->udp_encap.fp_udp_encap_id, path->fp_nh_proto, &path->fp_dpo); break; case FIB_PATH_TYPE_INTF_RX: { /* * Resolve via a receive DPO. */ interface_rx_dpo_add_or_lock(path->fp_nh_proto, path->intf_rx.fp_interface, &path->fp_dpo); break; } case FIB_PATH_TYPE_EXCLUSIVE: /* * Resolve via the user provided DPO */ dpo_copy(&path->fp_dpo, &path->exclusive.fp_ex_dpo); break; } return (fib_path_is_resolved(path_index)); } u32 fib_path_get_resolving_interface (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: return (path->attached_next_hop.fp_interface); case FIB_PATH_TYPE_ATTACHED: return (path->attached.fp_interface); case FIB_PATH_TYPE_RECEIVE: return (path->receive.fp_interface); case FIB_PATH_TYPE_RECURSIVE: if (fib_path_is_resolved(path_index)) { return (fib_entry_get_resolving_interface(path->fp_via_fib)); } break; case FIB_PATH_TYPE_DVR: return (path->dvr.fp_interface); case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_UDP_ENCAP: case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_DEAG: case FIB_PATH_TYPE_EXCLUSIVE: case FIB_PATH_TYPE_BIER_FMASK: case FIB_PATH_TYPE_BIER_TABLE: case FIB_PATH_TYPE_BIER_IMP: break; } return (dpo_get_urpf(&path->fp_dpo)); } index_t fib_path_get_resolving_index (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: case FIB_PATH_TYPE_ATTACHED: case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_DEAG: case FIB_PATH_TYPE_DVR: case FIB_PATH_TYPE_EXCLUSIVE: break; case FIB_PATH_TYPE_UDP_ENCAP: return (path->udp_encap.fp_udp_encap_id); case FIB_PATH_TYPE_RECURSIVE: return (path->fp_via_fib); case FIB_PATH_TYPE_BIER_FMASK: return (path->bier_fmask.fp_bier_fmask); case FIB_PATH_TYPE_BIER_TABLE: return (path->fp_via_bier_tbl); case FIB_PATH_TYPE_BIER_IMP: return (path->bier_imp.fp_bier_imp); } return (~0); } adj_index_t fib_path_get_adj (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); if (dpo_is_adj(&path->fp_dpo)) { return (path->fp_dpo.dpoi_index); } return (ADJ_INDEX_INVALID); } u16 fib_path_get_weight (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); return (path->fp_weight); } u16 fib_path_get_preference (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); return (path->fp_preference); } u32 fib_path_get_rpf_id (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); if (FIB_PATH_CFG_FLAG_RPF_ID & path->fp_cfg_flags) { return (path->deag.fp_rpf_id); } return (~0); } /** * @brief Contribute the path's adjacency to the list passed. * By calling this function over all paths, recursively, a child * can construct its full set of forwarding adjacencies, and hence its * uRPF list. */ void fib_path_contribute_urpf (fib_node_index_t path_index, index_t urpf) { fib_path_t *path; path = fib_path_get(path_index); /* * resolved and unresolved paths contribute to the RPF list. */ switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: fib_urpf_list_append(urpf, path->attached_next_hop.fp_interface); break; case FIB_PATH_TYPE_ATTACHED: fib_urpf_list_append(urpf, path->attached.fp_interface); break; case FIB_PATH_TYPE_RECURSIVE: if (FIB_NODE_INDEX_INVALID != path->fp_via_fib && !fib_path_is_looped(path_index)) { /* * there's unresolved due to constraints, and there's unresolved * due to ain't got no via. can't do nowt w'out via. */ fib_entry_contribute_urpf(path->fp_via_fib, urpf); } break; case FIB_PATH_TYPE_EXCLUSIVE: case FIB_PATH_TYPE_SPECIAL: { /* * these path types may link to an adj, if that's what * the clinet gave */ u32 rpf_sw_if_index; rpf_sw_if_index = dpo_get_urpf(&path->fp_dpo); if (~0 != rpf_sw_if_index) { fib_urpf_list_append(urpf, rpf_sw_if_index); } break; } case FIB_PATH_TYPE_DVR: fib_urpf_list_append(urpf, path->dvr.fp_interface); break; case FIB_PATH_TYPE_UDP_ENCAP: fib_urpf_list_append(urpf, path->udp_encap.fp_udp_encap_id); break; case FIB_PATH_TYPE_DEAG: case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_BIER_FMASK: case FIB_PATH_TYPE_BIER_TABLE: case FIB_PATH_TYPE_BIER_IMP: /* * these path types don't link to an adj */ break; } } void fib_path_stack_mpls_disp (fib_node_index_t path_index, dpo_proto_t payload_proto, fib_mpls_lsp_mode_t mode, dpo_id_t *dpo) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: { dpo_id_t tmp = DPO_INVALID; dpo_copy(&tmp, dpo); mpls_disp_dpo_create(payload_proto, ~0, mode, &tmp, dpo); dpo_reset(&tmp); break; } case FIB_PATH_TYPE_DEAG: { dpo_id_t tmp = DPO_INVALID; dpo_copy(&tmp, dpo); mpls_disp_dpo_create(payload_proto, path->deag.fp_rpf_id, mode, &tmp, dpo); dpo_reset(&tmp); break; } case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_ATTACHED: case FIB_PATH_TYPE_RECURSIVE: case FIB_PATH_TYPE_INTF_RX: case FIB_PATH_TYPE_UDP_ENCAP: case FIB_PATH_TYPE_EXCLUSIVE: case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_BIER_FMASK: case FIB_PATH_TYPE_BIER_TABLE: case FIB_PATH_TYPE_BIER_IMP: case FIB_PATH_TYPE_DVR: break; } if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_POP_PW_CW) { dpo_id_t tmp = DPO_INVALID; dpo_copy(&tmp, dpo); pw_cw_dpo_create(&tmp, dpo); dpo_reset(&tmp); } } void fib_path_contribute_forwarding (fib_node_index_t path_index, fib_forward_chain_type_t fct, dpo_proto_t payload_proto, dpo_id_t *dpo) { fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); /* * The DPO stored in the path was created when the path was resolved. * This then represents the path's 'native' protocol; IP. * For all others will need to go find something else. */ if (fib_path_to_chain_type(path) == fct) { dpo_copy(dpo, &path->fp_dpo); } else { switch (path->fp_type) { case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: switch (fct) { case FIB_FORW_CHAIN_TYPE_MPLS_EOS: { dpo_id_t tmp = DPO_INVALID; dpo_copy (&tmp, dpo); path = fib_path_attached_next_hop_get_adj( path, dpo_proto_to_link(payload_proto), &tmp); dpo_copy (dpo, &tmp); dpo_reset(&tmp); break; } case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: { dpo_id_t tmp = DPO_INVALID; dpo_copy (&tmp, dpo); path = fib_path_attached_next_hop_get_adj( path, fib_forw_chain_type_to_link_type(fct), &tmp); dpo_copy (dpo, &tmp); dpo_reset(&tmp); break; } case FIB_FORW_CHAIN_TYPE_BIER: break; } break; case FIB_PATH_TYPE_RECURSIVE: switch (fct) { case FIB_FORW_CHAIN_TYPE_MPLS_EOS: case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: case FIB_FORW_CHAIN_TYPE_BIER: fib_path_recursive_adj_update(path, fct, dpo); break; case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: ASSERT(0); break; } break; case FIB_PATH_TYPE_BIER_TABLE: switch (fct) { case FIB_FORW_CHAIN_TYPE_BIER: bier_table_contribute_forwarding(path->fp_via_bier_tbl, dpo); break; case FIB_FORW_CHAIN_TYPE_MPLS_EOS: case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: ASSERT(0); break; } break; case FIB_PATH_TYPE_BIER_FMASK: switch (fct) { case FIB_FORW_CHAIN_TYPE_BIER: fib_path_bier_fmask_update(path, dpo); break; case FIB_FORW_CHAIN_TYPE_MPLS_EOS: case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: ASSERT(0); break; } break; case FIB_PATH_TYPE_BIER_IMP: bier_imp_contribute_forwarding(path->bier_imp.fp_bier_imp, fib_forw_chain_type_to_dpo_proto(fct), dpo); break; case FIB_PATH_TYPE_DEAG: switch (fct) { case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: lookup_dpo_add_or_lock_w_table_id(MPLS_FIB_DEFAULT_TABLE_ID, DPO_PROTO_MPLS, LOOKUP_UNICAST, LOOKUP_INPUT_DST_ADDR, LOOKUP_TABLE_FROM_CONFIG, dpo); break; case FIB_FORW_CHAIN_TYPE_MPLS_EOS: case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: dpo_copy(dpo, &path->fp_dpo); break; case FIB_FORW_CHAIN_TYPE_BIER: break; case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: ASSERT(0); break; } break; case FIB_PATH_TYPE_EXCLUSIVE: dpo_copy(dpo, &path->exclusive.fp_ex_dpo); break; case FIB_PATH_TYPE_ATTACHED: switch (fct) { case FIB_FORW_CHAIN_TYPE_MPLS_EOS: /* * End of stack traffic via an attacehd path (a glean) * must forace an IP lookup so that the IP packet can * match against any installed adj-fibs */ lookup_dpo_add_or_lock_w_fib_index( fib_table_get_index_for_sw_if_index( dpo_proto_to_fib(payload_proto), path->attached.fp_interface), payload_proto, LOOKUP_UNICAST, LOOKUP_INPUT_DST_ADDR, LOOKUP_TABLE_FROM_CONFIG, dpo); break; case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS: case FIB_FORW_CHAIN_TYPE_UNICAST_IP4: case FIB_FORW_CHAIN_TYPE_UNICAST_IP6: case FIB_FORW_CHAIN_TYPE_ETHERNET: case FIB_FORW_CHAIN_TYPE_NSH: case FIB_FORW_CHAIN_TYPE_BIER: fib_path_attached_get_adj(path, fib_forw_chain_type_to_link_type(fct), dpo); break; case FIB_FORW_CHAIN_TYPE_MCAST_IP4: case FIB_FORW_CHAIN_TYPE_MCAST_IP6: { adj_index_t ai; /* * Create the adj needed for sending IP multicast traffic */ if (vnet_sw_interface_is_p2p(vnet_get_main(), path->attached.fp_interface)) { /* * point-2-point interfaces do not require a glean, since * there is nothing to ARP. Install a rewrite/nbr adj instead */ ai = adj_nbr_add_or_lock(dpo_proto_to_fib(path->fp_nh_proto), fib_forw_chain_type_to_link_type(fct), &zero_addr, path->attached.fp_interface); } else { ai = adj_mcast_add_or_lock(dpo_proto_to_fib(path->fp_nh_proto), fib_forw_chain_type_to_link_type(fct), path->attached.fp_interface); } dpo_set(dpo, DPO_ADJACENCY, fib_forw_chain_type_to_dpo_proto(fct), ai); adj_unlock(ai); } break; } break; case FIB_PATH_TYPE_INTF_RX: /* * Create the adj needed for sending IP multicast traffic */ interface_rx_dpo_add_or_lock(payload_proto, path->intf_rx.fp_interface, dpo); break; case FIB_PATH_TYPE_UDP_ENCAP: udp_encap_contribute_forwarding(path->udp_encap.fp_udp_encap_id, path->fp_nh_proto, dpo); break; case FIB_PATH_TYPE_RECEIVE: case FIB_PATH_TYPE_SPECIAL: case FIB_PATH_TYPE_DVR: dpo_copy(dpo, &path->fp_dpo); break; } } } load_balance_path_t * fib_path_append_nh_for_multipath_hash (fib_node_index_t path_index, fib_forward_chain_type_t fct, dpo_proto_t payload_proto, load_balance_path_t *hash_key) { load_balance_path_t *mnh; fib_path_t *path; path = fib_path_get(path_index); ASSERT(path); vec_add2(hash_key, mnh, 1); mnh->path_weight = path->fp_weight; mnh->path_index = path_index; if (fib_path_is_resolved(path_index)) { fib_path_contribute_forwarding(path_index, fct, payload_proto, &mnh->path_dpo); } else { dpo_copy(&mnh->path_dpo, drop_dpo_get(fib_forw_chain_type_to_dpo_proto(fct))); } return (hash_key); } int fib_path_is_recursive_constrained (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return ((FIB_PATH_TYPE_RECURSIVE == path->fp_type) && ((path->fp_cfg_flags & FIB_PATH_CFG_FLAG_RESOLVE_ATTACHED) || (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_RESOLVE_HOST))); } int fib_path_is_exclusive (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (FIB_PATH_TYPE_EXCLUSIVE == path->fp_type); } int fib_path_is_deag (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (FIB_PATH_TYPE_DEAG == path->fp_type); } int fib_path_is_resolved (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (dpo_id_is_valid(&path->fp_dpo) && (path->fp_oper_flags & FIB_PATH_OPER_FLAG_RESOLVED) && !fib_path_is_looped(path_index) && !fib_path_is_permanent_drop(path)); } int fib_path_is_looped (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (path->fp_oper_flags & FIB_PATH_OPER_FLAG_RECURSIVE_LOOP); } fib_path_list_walk_rc_t fib_path_encode (fib_node_index_t path_list_index, fib_node_index_t path_index, const fib_path_ext_t *path_ext, void *args) { fib_path_encode_ctx_t *ctx = args; fib_route_path_t *rpath; fib_path_t *path; path = fib_path_get(path_index); if (!path) return (FIB_PATH_LIST_WALK_CONTINUE); vec_add2(ctx->rpaths, rpath, 1); rpath->frp_weight = path->fp_weight; rpath->frp_preference = path->fp_preference; rpath->frp_proto = path->fp_nh_proto; rpath->frp_sw_if_index = ~0; rpath->frp_fib_index = 0; switch (path->fp_type) { case FIB_PATH_TYPE_RECEIVE: rpath->frp_addr = path->receive.fp_addr; rpath->frp_sw_if_index = path->receive.fp_interface; rpath->frp_flags |= FIB_ROUTE_PATH_LOCAL; break; case FIB_PATH_TYPE_ATTACHED: rpath->frp_sw_if_index = path->attached.fp_interface; break; case FIB_PATH_TYPE_ATTACHED_NEXT_HOP: rpath->frp_sw_if_index = path->attached_next_hop.fp_interface; rpath->frp_addr = path->attached_next_hop.fp_nh; break; case FIB_PATH_TYPE_BIER_FMASK: rpath->frp_bier_fmask = path->bier_fmask.fp_bier_fmask; break; case FIB_PATH_TYPE_SPECIAL: break; case FIB_PATH_TYPE_DEAG: rpath->frp_fib_index = path->deag.fp_tbl_id; if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_RPF_ID) { rpath->frp_flags |= FIB_ROUTE_PATH_RPF_ID; } break; case FIB_PATH_TYPE_RECURSIVE: rpath->frp_addr = path->recursive.fp_nh.fp_ip; rpath->frp_fib_index = path->recursive.fp_tbl_id; break; case FIB_PATH_TYPE_DVR: rpath->frp_sw_if_index = path->dvr.fp_interface; rpath->frp_flags |= FIB_ROUTE_PATH_DVR; break; case FIB_PATH_TYPE_UDP_ENCAP: rpath->frp_udp_encap_id = path->udp_encap.fp_udp_encap_id; rpath->frp_flags |= FIB_ROUTE_PATH_UDP_ENCAP; break; case FIB_PATH_TYPE_INTF_RX: rpath->frp_sw_if_index = path->receive.fp_interface; rpath->frp_flags |= FIB_ROUTE_PATH_INTF_RX; break; case FIB_PATH_TYPE_EXCLUSIVE: rpath->frp_flags |= FIB_ROUTE_PATH_EXCLUSIVE; default: break; } if (path_ext && path_ext->fpe_type == FIB_PATH_EXT_MPLS) { rpath->frp_label_stack = path_ext->fpe_path.frp_label_stack; } if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_DROP) rpath->frp_flags |= FIB_ROUTE_PATH_DROP; if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_UNREACH) rpath->frp_flags |= FIB_ROUTE_PATH_ICMP_UNREACH; if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_ICMP_PROHIBIT) rpath->frp_flags |= FIB_ROUTE_PATH_ICMP_PROHIBIT; return (FIB_PATH_LIST_WALK_CONTINUE); } dpo_proto_t fib_path_get_proto (fib_node_index_t path_index) { fib_path_t *path; path = fib_path_get(path_index); return (path->fp_nh_proto); } void fib_path_module_init (void) { fib_node_register_type (FIB_NODE_TYPE_PATH, &fib_path_vft); fib_path_logger = vlib_log_register_class ("fib", "path"); } static clib_error_t * show_fib_path_command (vlib_main_t * vm, unformat_input_t * input, vlib_cli_command_t * cmd) { fib_node_index_t pi; fib_path_t *path; if (unformat (input, "%d", &pi)) { /* * show one in detail */ if (!pool_is_free_index(fib_path_pool, pi)) { path = fib_path_get(pi); u8 *s = format(NULL, "%U", format_fib_path, pi, 1, FIB_PATH_FORMAT_FLAGS_NONE); s = format(s, "\n children:"); s = fib_node_children_format(path->fp_node.fn_children, s); vlib_cli_output (vm, "%v", s); vec_free(s); } else { vlib_cli_output (vm, "path %d invalid", pi); } } else { vlib_cli_output (vm, "FIB Paths"); pool_foreach_index (pi, fib_path_pool) { vlib_cli_output (vm, "%U", format_fib_path, pi, 0, FIB_PATH_FORMAT_FLAGS_NONE); } } return (NULL); } VLIB_CLI_COMMAND (show_fib_path, static) = { .path = "show fib paths", .function = show_fib_path_command, .short_help = "show fib paths", };