summaryrefslogtreecommitdiffstats
path: root/src/vnet/interface_format.c
blob: ee332f398e977379978d4aca513a6f07a7aaaeaa (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
/*
 * Copyright (c) 2015 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.
 */
/*
 * interface_format.c: interface formatting
 *
 * Copyright (c) 2008 Eliot Dresselhaus
 *
 * Permission is hereby granted, free of charge, to any person obtaining
 * a copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sublicense, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice shall be
 * included in all copies or substantial portions of the Software.
 *
 *  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 *  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 *  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 *  NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
 *  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
 *  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
 *  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 */

#include <vnet/vnet.h>
#include <vppinfra/bitmap.h>
#include <vnet/l2/l2_input.h>
#include <vnet/l2/l2_output.h>

u8 *
format_vnet_sw_interface_flags (u8 * s, va_list * args)
{
  u32 flags = va_arg (*args, u32);

  if (flags & VNET_SW_INTERFACE_FLAG_ERROR)
    s = format (s, "error");
  else if (flags & VNET_SW_INTERFACE_FLAG_BOND_SLAVE)
    s = format (s, "bond-slave");
  else
    {
      s = format (s, "%s",
		  (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) ? "up" : "down");
      if (flags & VNET_SW_INTERFACE_FLAG_PUNT)
	s = format (s, "/punt");
    }

  return s;
}

u8 *
format_vnet_hw_interface_rx_mode (u8 * s, va_list * args)
{
  vnet_hw_interface_rx_mode mode = va_arg (*args, vnet_hw_interface_rx_mode);

  if (mode == VNET_HW_INTERFACE_RX_MODE_POLLING)
    return format (s, "polling");

  if (mode == VNET_HW_INTERFACE_RX_MODE_INTERRUPT)
    return format (s, "interrupt");

  if (mode == VNET_HW_INTERFACE_RX_MODE_ADAPTIVE)
    return format (s, "adaptive");

  return format (s, "unknown");
}

u8 *
format_vnet_hw_interface_link_speed (u8 * s, va_list * args)
{
  u32 link_speed = va_arg (*args, u32);

  if (link_speed == 0)
    return format (s, "unknown");

  if (link_speed >= 1000000)
    return format (s, "%f Gbps", (f64) link_speed / 1000000);

  if (link_speed >= 1000)
    return format (s, "%f Mbps", (f64) link_speed / 1000);

  return format (s, "%u Kbps", link_speed);
}


u8 *
format_vnet_hw_interface (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  vnet_hw_interface_t *hi = va_arg (*args, vnet_hw_interface_t *);
  vnet_hw_interface_class_t *hw_class;
  vnet_device_class_t *dev_class;
  int verbose = va_arg (*args, int);
  u32 indent;

  if (!hi)
    return format (s, "%=32s%=6s%=8s%s", "Name", "Idx", "Link", "Hardware");

  indent = format_get_indent (s);

  s = format (s, "%-32v%=6d", hi->name, hi->hw_if_index);

  if (hi->bond_info == VNET_HW_INTERFACE_BOND_INFO_SLAVE)
    s = format (s, "%=8s", "slave");
  else
    s = format (s, "%=8s",
		hi->flags & VNET_HW_INTERFACE_FLAG_LINK_UP ? "up" : "down");

  hw_class = vnet_get_hw_interface_class (vnm, hi->hw_class_index);
  dev_class = vnet_get_device_class (vnm, hi->dev_class_index);

  if (hi->bond_info && (hi->bond_info != VNET_HW_INTERFACE_BOND_INFO_SLAVE))
    {
      int hw_idx;
      s = format (s, "Slave-Idx:");
      clib_bitmap_foreach (hw_idx, hi->bond_info, s =
			   format (s, " %d", hw_idx));
    }
  else if (dev_class->format_device_name)
    s = format (s, "%U", dev_class->format_device_name, hi->dev_instance);
  else
    s = format (s, "%s%d", dev_class->name, hi->dev_instance);

  s = format (s, "\n%ULink speed: %U", format_white_space, indent + 2,
	      format_vnet_hw_interface_link_speed, hi->link_speed);

  if (verbose)
    {
      if (hw_class->format_device)
	s = format (s, "\n%U%U",
		    format_white_space, indent + 2,
		    hw_class->format_device, hi->hw_if_index, verbose);
      else
	{
	  s = format (s, "\n%U%s",
		      format_white_space, indent + 2, hw_class->name);
	  if (hw_class->format_address && vec_len (hi->hw_address) > 0)
	    s =
	      format (s, " address %U", hw_class->format_address,
		      hi->hw_address);
	}

      if (dev_class->format_device)
	s = format (s, "\n%U%U",
		    format_white_space, indent + 2,
		    dev_class->format_device, hi->dev_instance, verbose);
    }

  return s;
}

u8 *
format_vnet_sw_interface_name (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  vnet_sw_interface_t *si = va_arg (*args, vnet_sw_interface_t *);
  vnet_sw_interface_t *si_sup =
    vnet_get_sup_sw_interface (vnm, si->sw_if_index);
  vnet_hw_interface_t *hi_sup;

  ASSERT (si_sup->type == VNET_SW_INTERFACE_TYPE_HARDWARE);
  hi_sup = vnet_get_hw_interface (vnm, si_sup->hw_if_index);

  s = format (s, "%v", hi_sup->name);

  if (si->type != VNET_SW_INTERFACE_TYPE_HARDWARE)
    s = format (s, ".%d", si->sub.id);

  return s;
}

u8 *
format_vnet_sw_if_index_name (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  u32 sw_if_index = va_arg (*args, u32);
  vnet_sw_interface_t *si;

  si = vnet_get_sw_interface_safe (vnm, sw_if_index);

  if (NULL == si)
    {
      return format (s, "DELETED");
    }
  return format (s, "%U", format_vnet_sw_interface_name, vnm, si);
}

u8 *
format_vnet_hw_if_index_name (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  u32 hw_if_index = va_arg (*args, u32);
  vnet_hw_interface_t *hi;

  hi = vnet_get_hw_interface (vnm, hw_if_index);

  if (hi == 0)
    return format (s, "DELETED");

  return format (s, "%v", hi->name);
}

u8 *
format_vnet_sw_interface_cntrs (u8 * s, vnet_interface_main_t * im,
				vnet_sw_interface_t * si)
{
  u32 indent, n_printed;
  int i, j, n_counters;
  static vnet_main_t **my_vnet_mains;

  vec_reset_length (my_vnet_mains);

  indent = format_get_indent (s);
  n_printed = 0;

  {
    vlib_combined_counter_main_t *cm;
    vlib_counter_t v, vtotal;
    u8 *n = 0;

    for (i = 0; i < vec_len (vnet_mains); i++)
      {
	if (vnet_mains[i])
	  vec_add1 (my_vnet_mains, vnet_mains[i]);
      }

    if (vec_len (my_vnet_mains) == 0)
      vec_add1 (my_vnet_mains, &vnet_main);

    /* Each vnet_main_t has its own copy of the interface counters */
    n_counters = vec_len (im->combined_sw_if_counters);

    /* rx, tx counters... */
    for (j = 0; j < n_counters; j++)
      {
	vtotal.packets = 0;
	vtotal.bytes = 0;

	for (i = 0; i < vec_len (my_vnet_mains); i++)
	  {
	    im = &my_vnet_mains[i]->interface_main;
	    cm = im->combined_sw_if_counters + j;
	    vlib_get_combined_counter (cm, si->sw_if_index, &v);
	    vtotal.packets += v.packets;
	    vtotal.bytes += v.bytes;
	  }

	/* Only display non-zero counters. */
	if (vtotal.packets == 0)
	  continue;

	if (n_printed > 0)
	  s = format (s, "\n%U", format_white_space, indent);
	n_printed += 2;

	if (n)
	  _vec_len (n) = 0;
	n = format (n, "%s packets", cm->name);
	s = format (s, "%-16v%16Ld", n, vtotal.packets);

	_vec_len (n) = 0;
	n = format (n, "%s bytes", cm->name);
	s = format (s, "\n%U%-16v%16Ld",
		    format_white_space, indent, n, vtotal.bytes);
      }
    vec_free (n);
  }

  {
    vlib_simple_counter_main_t *cm;
    u64 v, vtotal;

    n_counters = vec_len (im->sw_if_counters);

    for (j = 0; j < n_counters; j++)
      {
	vtotal = 0;

	for (i = 0; i < vec_len (my_vnet_mains); i++)
	  {
	    im = &my_vnet_mains[i]->interface_main;
	    cm = im->sw_if_counters + j;

	    v = vlib_get_simple_counter (cm, si->sw_if_index);
	    vtotal += v;
	  }

	/* Only display non-zero counters. */
	if (vtotal == 0)
	  continue;

	if (n_printed > 0)
	  s = format (s, "\n%U", format_white_space, indent);
	n_printed += 1;

	s = format (s, "%-16s%16Ld", cm->name, vtotal);
      }
  }

  return s;
}

static u8 *
format_vnet_sw_interface_mtu (u8 * s, va_list * args)
{
  vnet_sw_interface_t *si = va_arg (*args, vnet_sw_interface_t *);

  return format (s, "%d/%d/%d/%d", si->mtu[VNET_MTU_L3],
		 si->mtu[VNET_MTU_IP4],
		 si->mtu[VNET_MTU_IP6], si->mtu[VNET_MTU_MPLS]);
}

u8 *
format_vnet_sw_interface (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  vnet_sw_interface_t *si = va_arg (*args, vnet_sw_interface_t *);
  vnet_interface_main_t *im = &vnm->interface_main;

  if (!si)
    return format (s, "%=32s%=5s%=10s%=21s%=16s%=16s",
		   "Name", "Idx", "State", "MTU (L3/IP4/IP6/MPLS)", "Counter",
		   "Count");

  s = format (s, "%-32U%=5d%=10U%=21U",
	      format_vnet_sw_interface_name, vnm, si, si->sw_if_index,
	      format_vnet_sw_interface_flags, si->flags,
	      format_vnet_sw_interface_mtu, si);

  s = format_vnet_sw_interface_cntrs (s, im, si);

  return s;
}

u8 *
format_vnet_sw_interface_name_override (u8 * s, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  vnet_sw_interface_t *si = va_arg (*args, vnet_sw_interface_t *);
  /* caller supplied display name for this interface */
  u8 *name = va_arg (*args, u8 *);
  vnet_interface_main_t *im = &vnm->interface_main;


  if (!si)
    return format (s, "%=32s%=5s%=16s%=16s%=16s",
		   "Name", "Idx", "State", "Counter", "Count");

  s = format (s, "%-32v%=5d%=16U",
	      name, si->sw_if_index,
	      format_vnet_sw_interface_flags, si->flags);

  s = format_vnet_sw_interface_cntrs (s, im, si);

  return s;
}

u8 *
format_vnet_buffer_flags (u8 * s, va_list * args)
{
  vlib_buffer_t *buf = va_arg (*args, vlib_buffer_t *);

#define _(a,b,c,v) if (buf->flags & VNET_BUFFER_F_##b) s = format (s, "%s ", c);
  foreach_vnet_buffer_flag;
#undef _
  return s;
}

u8 *
format_vnet_buffer_opaque (u8 * s, va_list * args)
{
  vlib_buffer_t *b = va_arg (*args, vlib_buffer_t *);
  vnet_buffer_opaque_t *o = (vnet_buffer_opaque_t *) b->opaque;
  int i;

  s = format (s, "raw: ");

  for (i = 0; i < ARRAY_LEN (b->opaque); i++)
    s = format (s, "%08x ", b->opaque[i]);

  vec_add1 (s, '\n');

  s = format (s,
	      "sw_if_index[VLIB_RX]: %d, sw_if_index[VLIB_TX]: %d",
	      o->sw_if_index[0], o->sw_if_index[1]);
  vec_add1 (s, '\n');

  s = format (s,
	      "L2 offset %d, L3 offset %d, L4 offset %d, feature arc index %d",
	      (u32) (o->l2_hdr_offset),
	      (u32) (o->l3_hdr_offset),
	      (u32) (o->l4_hdr_offset), (u32) (o->feature_arc_index));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.adj_index[VLIB_RX]: %d, ip.adj_index[VLIB_TX]: %d",
	      (u32) (o->ip.adj_index[0]), (u32) (o->ip.adj_index[1]));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.flow_hash: 0x%x, ip.save_protocol: 0x%x, ip.fib_index: %d",
	      o->ip.flow_hash, o->ip.save_protocol, o->ip.fib_index);
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.save_rewrite_length: %d, ip.rpf_id: %d",
	      o->ip.save_rewrite_length, o->ip.rpf_id);
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.icmp.type: %d ip.icmp.code: %d, ip.icmp.data: 0x%x",
	      (u32) (o->ip.icmp.type),
	      (u32) (o->ip.icmp.code), o->ip.icmp.data);
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.reass.next_index: %d, ip.reass.estimated_mtu: %d",
	      o->ip.reass.next_index, (u32) (o->ip.reass.estimated_mtu));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.reass.fragment_first: %d ip.reass.fragment_last: %d",
	      (u32) (o->ip.reass.fragment_first),
	      (u32) (o->ip.reass.fragment_last));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.reass.range_first: %d ip.reass.range_last: %d",
	      (u32) (o->ip.reass.range_first),
	      (u32) (o->ip.reass.range_last));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip.reass.next_range_bi: 0x%x, ip.reass.ip6_frag_hdr_offset: %d",
	      o->ip.reass.next_range_bi,
	      (u32) (o->ip.reass.ip6_frag_hdr_offset));
  vec_add1 (s, '\n');

  s = format (s,
	      "mpls.ttl: %d, mpls.exp: %d, mpls.first: %d, "
	      "mpls.save_rewrite_length: %d, mpls.bier.n_bytes: %d",
	      (u32) (o->mpls.ttl), (u32) (o->mpls.exp), (u32) (o->mpls.first),
	      o->mpls.save_rewrite_length, (u32) (o->mpls.bier.n_bytes));
  vec_add1 (s, '\n');

  s = format (s,
	      "l2.feature_bitmap: %08x, l2.bd_index: %d, l2.l2_len: %d, "
	      "l2.shg: %d, l2.l2fib_sn: %d, l2.bd_age: %d",
	      o->l2.feature_bitmap, (u32) (o->l2.bd_index),
	      (u32) (o->l2.l2_len), (u32) (o->l2.shg), (u32) (o->l2.l2fib_sn),
	      (u32) (o->l2.bd_age));
  vec_add1 (s, '\n');

  s = format (s,
	      "l2.feature_bitmap_input: %U, L2.feature_bitmap_output: %U",
	      format_l2_input_features, o->l2.feature_bitmap, 0,
	      format_l2_output_features, o->l2.feature_bitmap, 0);
  vec_add1 (s, '\n');

  s = format (s,
	      "l2t.next_index: %d, l2t.session_index: %d",
	      (u32) (o->l2t.next_index), o->l2t.session_index);
  vec_add1 (s, '\n');

  s = format (s,
	      "l2_classify.table_index: %d, l2_classify.opaque_index: %d, "
	      "l2_classify.hash: 0x%llx",
	      o->l2_classify.table_index,
	      o->l2_classify.opaque_index, o->l2_classify.hash);
  vec_add1 (s, '\n');

  s = format (s, "policer.index: %d", o->policer.index);
  vec_add1 (s, '\n');

  s = format (s, "ipsec.sad_index: %d", o->ipsec.sad_index);
  vec_add1 (s, '\n');

  s = format (s, "map.mtu: %d", (u32) (o->map.mtu));
  vec_add1 (s, '\n');

  s = format (s,
	      "map_t.v6.saddr: 0x%x, map_t.v6.daddr: 0x%x, "
	      "map_t.v6.frag_offset: %d, map_t.v6.l4_offset: %d",
	      o->map_t.v6.saddr,
	      o->map_t.v6.daddr,
	      (u32) (o->map_t.v6.frag_offset), (u32) (o->map_t.v6.l4_offset));
  vec_add1 (s, '\n');

  s = format (s,
	      "map_t.v6.l4_protocol: %d, map_t.checksum_offset: %d, "
	      "map_t.mtu: %d",
	      (u32) (o->map_t.v6.l4_protocol),
	      (u32) (o->map_t.checksum_offset), (u32) (o->map_t.mtu));
  vec_add1 (s, '\n');

  s = format (s,
	      "ip_frag.mtu: %d, ip_frag.next_index: %d, ip_frag.flags: 0x%x",
	      (u32) (o->ip_frag.mtu),
	      (u32) (o->ip_frag.next_index), (u32) (o->ip_frag.flags));
  vec_add1 (s, '\n');

  s = format (s, "cop.current_config_index: %d", o->cop.current_config_index);
  vec_add1 (s, '\n');

  s = format (s, "lisp.overlay_afi: %d", (u32) (o->lisp.overlay_afi));
  vec_add1 (s, '\n');

  s = format
    (s, "tcp.connection_index: %d, tcp.seq_number: %d, tcp.seq_end: %d, "
     "tcp.ack_number: %d, tcp.hdr_offset: %d, tcp.data_offset: %d",
     o->tcp.connection_index,
     o->tcp.seq_number,
     o->tcp.seq_end,
     o->tcp.ack_number,
     (u32) (o->tcp.hdr_offset), (u32) (o->tcp.data_offset));
  vec_add1 (s, '\n');

  s = format (s,
	      "tcp.data_len: %d, tcp.flags: 0x%x",
	      (u32) (o->tcp.data_len), (u32) (o->tcp.flags));
  vec_add1 (s, '\n');

  s = format (s,
	      "sctp.connection_index: %d, sctp.sid: %d, sctp.ssn: %d, "
	      "sctp.tsn: %d, sctp.hdr_offset: %d",
	      o->sctp.connection_index,
	      (u32) (o->sctp.sid),
	      (u32) (o->sctp.ssn),
	      (u32) (o->sctp.tsn), (u32) (o->sctp.hdr_offset));
  vec_add1 (s, '\n');

  s = format
    (s, "sctp.data_offset: %d, sctp.data_len: %d, sctp.subconn_idx: %d, "
     "sctp.flags: 0x%x",
     (u32) (o->sctp.data_offset),
     (u32) (o->sctp.data_len),
     (u32) (o->sctp.subconn_idx), (u32) (o->sctp.flags));
  vec_add1 (s, '\n');

  s = format (s, "snat.flags: 0x%x", o->snat.flags);
  vec_add1 (s, '\n');
  return s;
}

u8 *
format_vnet_buffer_opaque2 (u8 * s, va_list * args)
{
  vlib_buffer_t *b = va_arg (*args, vlib_buffer_t *);
  vnet_buffer_opaque2_t *o = (vnet_buffer_opaque2_t *) b->opaque2;

  int i;

  s = format (s, "raw: ");

  for (i = 0; i < ARRAY_LEN (b->opaque2); i++)
    s = format (s, "%08x ", b->opaque2[i]);
  vec_add1 (s, '\n');

  s = format (s, "qos.bits: %x, qos.source: %x",
	      (u32) (o->qos.bits), (u32) (o->qos.source));
  vec_add1 (s, '\n');
  s = format (s, "loop_counter: %d", o->loop_counter);
  vec_add1 (s, '\n');

  s = format (s, "gbp.flags: %x, gbp.sclass: %d",
	      (u32) (o->gbp.flags), (u32) (o->gbp.sclass));
  vec_add1 (s, '\n');

  s = format (s, "pg_replay_timestamp: %llu", (u32) (o->pg_replay_timestamp));
  vec_add1 (s, '\n');
  return s;
}

uword
unformat_vnet_hw_interface (unformat_input_t * input, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  u32 *hw_if_index = va_arg (*args, u32 *);
  vnet_interface_main_t *im = &vnm->interface_main;
  vnet_device_class_t *c;

  /* Try per device class functions first. */
  vec_foreach (c, im->device_classes)
  {
    if (c->unformat_device_name
	&& unformat_user (input, c->unformat_device_name, hw_if_index))
      return 1;
  }

  return unformat_user (input, unformat_hash_vec_string,
			im->hw_interface_by_name, hw_if_index);
}

uword
unformat_vnet_sw_interface (unformat_input_t * input, va_list * args)
{
  vnet_main_t *vnm = va_arg (*args, vnet_main_t *);
  u32 *result = va_arg (*args, u32 *);
  vnet_hw_interface_t *hi;
  u32 hw_if_index, id, id_specified;
  u32 sw_if_index;
  u8 *if_name = 0;
  uword *p, error = 0;

  id = ~0;
  if (unformat (input, "%_%v.%d%_", &if_name, &id)
      && ((p = hash_get (vnm->interface_main.hw_interface_by_name, if_name))))
    {
      hw_if_index = p[0];
      id_specified = 1;
    }
  else
    if (unformat (input, "%U", unformat_vnet_hw_interface, vnm, &hw_if_index))
    id_specified = 0;
  else
    goto done;

  hi = vnet_get_hw_interface (vnm, hw_if_index);
  if (!id_specified)
    {
      sw_if_index = hi->sw_if_index;
    }
  else
    {
      if (!(p = hash_get (hi->sub_interface_sw_if_index_by_id, id)))
	goto done;
      sw_if_index = p[0];
    }
  if (!vnet_sw_interface_is_api_visible (vnm, sw_if_index))
    goto done;
  *result = sw_if_index;
  error = 1;
done:
  vec_free (if_name);
  return error;
}

uword
unformat_vnet_sw_interface_flags (unformat_input_t * input, va_list * args)
{
  u32 *result = va_arg (*args, u32 *);
  u32 flags = 0;

  if (unformat (input, "up"))
    flags |= VNET_SW_INTERFACE_FLAG_ADMIN_UP;
  else if (unformat (input, "down"))
    flags &= ~VNET_SW_INTERFACE_FLAG_ADMIN_UP;
  else if (unformat (input, "punt"))
    flags |= VNET_SW_INTERFACE_FLAG_PUNT;
  else if (unformat (input, "enable"))
    flags &= ~VNET_SW_INTERFACE_FLAG_PUNT;
  else
    return 0;

  *result = flags;
  return 1;
}

uword
unformat_vnet_hw_interface_flags (unformat_input_t * input, va_list * args)
{
  u32 *result = va_arg (*args, u32 *);
  u32 flags = 0;

  if (unformat (input, "up"))
    flags |= VNET_HW_INTERFACE_FLAG_LINK_UP;
  else if (unformat (input, "down"))
    flags &= ~VNET_HW_INTERFACE_FLAG_LINK_UP;
  else
    return 0;

  *result = flags;
  return 1;
}

/*
 * fd.io coding-style-patch-verification: ON
 *
 * Local Variables:
 * eval: (c-set-style "gnu")
 * End:
 */
>app_wrk; session_t *s = 0; u64 ho_handle; /* * Find connection handle and cleanup half-open table */ ho_handle = session_lookup_half_open_handle (tc); if (ho_handle == HALF_OPEN_LOOKUP_INVALID_VALUE) { SESSION_DBG ("half-open was removed!"); return -1; } session_lookup_del_half_open (tc); /* Get the app's index from the handle we stored when opening connection * and the opaque (api_context for external apps) from transport session * index */ app_wrk = app_worker_get_if_valid (ho_handle >> 32); if (!app_wrk) return -1; opaque = tc->s_index; if (is_fail) return app_worker_connect_notify (app_wrk, s, opaque); s = session_alloc_for_connection (tc); s->session_state = SESSION_STATE_CONNECTING; s->app_wrk_index = app_wrk->wrk_index; new_si = s->session_index; new_ti = s->thread_index; if (app_worker_init_connected (app_wrk, s)) { session_free (s); app_worker_connect_notify (app_wrk, 0, opaque); return -1; } s = session_get (new_si, new_ti); s->session_state = opened_state; session_lookup_add_connection (tc, session_handle (s)); if (app_worker_connect_notify (app_wrk, s, opaque)) { s = session_get (new_si, new_ti); session_free_w_fifos (s); return -1; } return 0; } int session_stream_connect_notify (transport_connection_t * tc, u8 is_fail) { return session_stream_connect_notify_inline (tc, is_fail, SESSION_STATE_READY); } int session_ho_stream_connect_notify (transport_connection_t * tc, u8 is_fail) { return session_stream_connect_notify_inline (tc, is_fail, SESSION_STATE_OPENED); } typedef struct _session_switch_pool_args { u32 session_index; u32 thread_index; u32 new_thread_index; u32 new_session_index; } session_switch_pool_args_t; static void session_switch_pool (void *cb_args) { session_switch_pool_args_t *args = (session_switch_pool_args_t *) cb_args; session_t *s; ASSERT (args->thread_index == vlib_get_thread_index ()); s = session_get (args->session_index, args->thread_index); s->tx_fifo->master_session_index = args->new_session_index; s->tx_fifo->master_thread_index = args->new_thread_index; transport_cleanup (session_get_transport_proto (s), s->connection_index, s->thread_index); session_free (s); clib_mem_free (cb_args); } /** * Move dgram session to the right thread */ int session_dgram_connect_notify (transport_connection_t * tc, u32 old_thread_index, session_t ** new_session) { session_t *new_s; session_switch_pool_args_t *rpc_args; /* * Clone half-open session to the right thread. */ new_s = session_clone_safe (tc->s_index, old_thread_index); new_s->connection_index = tc->c_index; new_s->rx_fifo->master_session_index = new_s->session_index; new_s->rx_fifo->master_thread_index = new_s->thread_index; new_s->session_state = SESSION_STATE_READY; session_lookup_add_connection (tc, session_handle (new_s)); /* * Ask thread owning the old session to clean it up and make us the tx * fifo owner */ rpc_args = clib_mem_alloc (sizeof (*rpc_args)); rpc_args->new_session_index = new_s->session_index; rpc_args->new_thread_index = new_s->thread_index; rpc_args->session_index = tc->s_index; rpc_args->thread_index = old_thread_index; session_send_rpc_evt_to_thread (rpc_args->thread_index, session_switch_pool, rpc_args); tc->s_index = new_s->session_index; new_s->connection_index = tc->c_index; *new_session = new_s; return 0; } /** * Notification from transport that connection is being closed. * * A disconnect is sent to application but state is not removed. Once * disconnect is acknowledged by application, session disconnect is called. * Ultimately this leads to close being called on transport (passive close). */ void session_transport_closing_notify (transport_connection_t * tc) { app_worker_t *app_wrk; session_t *s; s = session_get (tc->s_index, tc->thread_index); if (s->session_state >= SESSION_STATE_TRANSPORT_CLOSING) return; s->session_state = SESSION_STATE_TRANSPORT_CLOSING; app_wrk = app_worker_get (s->app_wrk_index); app_worker_close_notify (app_wrk, s); } /** * Notification from transport that connection is being deleted * * This removes the session if it is still valid. It should be called only on * previously fully established sessions. For instance failed connects should * call stream_session_connect_notify and indicate that the connect has * failed. */ void session_transport_delete_notify (transport_connection_t * tc) { session_t *s; /* App might've been removed already */ if (!(s = session_get_if_valid (tc->s_index, tc->thread_index))) return; switch (s->session_state) { case SESSION_STATE_CREATED: /* Session was created but accept notification was not yet sent to the * app. Cleanup everything. */ session_lookup_del_session (s); session_free_w_fifos (s); break; case SESSION_STATE_ACCEPTING: case SESSION_STATE_TRANSPORT_CLOSING: case SESSION_STATE_CLOSING: case SESSION_STATE_TRANSPORT_CLOSED: /* If transport finishes or times out before we get a reply * from the app, mark transport as closed and wait for reply * before removing the session. Cleanup session table in advance * because transport will soon be closed and closed sessions * are assumed to have been removed from the lookup table */ session_lookup_del_session (s); s->session_state = SESSION_STATE_TRANSPORT_DELETED; session_cleanup_notify (s, SESSION_CLEANUP_TRANSPORT); svm_fifo_dequeue_drop_all (s->tx_fifo); break; case SESSION_STATE_APP_CLOSED: /* Cleanup lookup table as transport needs to still be valid. * Program transport close to ensure that all session events * have been cleaned up. Once transport close is called, the * session is just removed because both transport and app have * confirmed the close*/ session_lookup_del_session (s); s->session_state = SESSION_STATE_CLOSED; session_cleanup_notify (s, SESSION_CLEANUP_TRANSPORT); svm_fifo_dequeue_drop_all (s->tx_fifo); session_program_transport_close (s); break; case SESSION_STATE_TRANSPORT_DELETED: break; case SESSION_STATE_CLOSED: session_cleanup_notify (s, SESSION_CLEANUP_TRANSPORT); session_delete (s); break; default: clib_warning ("session state %u", s->session_state); session_cleanup_notify (s, SESSION_CLEANUP_TRANSPORT); session_delete (s); break; } } /** * Notification from transport that it is closed * * Should be called by transport, prior to calling delete notify, once it * knows that no more data will be exchanged. This could serve as an * early acknowledgment of an active close especially if transport delete * can be delayed a long time, e.g., tcp time-wait. */ void session_transport_closed_notify (transport_connection_t * tc) { app_worker_t *app_wrk; session_t *s; if (!(s = session_get_if_valid (tc->s_index, tc->thread_index))) return; /* Transport thinks that app requested close but it actually didn't. * Can happen for tcp if fin and rst are received in close succession. */ if (s->session_state == SESSION_STATE_READY) { session_transport_closing_notify (tc); svm_fifo_dequeue_drop_all (s->tx_fifo); s->session_state = SESSION_STATE_TRANSPORT_CLOSED; } /* If app close has not been received or has not yet resulted in * a transport close, only mark the session transport as closed */ else if (s->session_state <= SESSION_STATE_CLOSING) { s->session_state = SESSION_STATE_TRANSPORT_CLOSED; } /* If app also closed, switch to closed */ else if (s->session_state == SESSION_STATE_APP_CLOSED) s->session_state = SESSION_STATE_CLOSED; app_wrk = app_worker_get_if_valid (s->app_wrk_index); if (app_wrk) app_worker_transport_closed_notify (app_wrk, s); } /** * Notify application that connection has been reset. */ void session_transport_reset_notify (transport_connection_t * tc) { app_worker_t *app_wrk; session_t *s; s = session_get (tc->s_index, tc->thread_index); svm_fifo_dequeue_drop_all (s->tx_fifo); if (s->session_state >= SESSION_STATE_TRANSPORT_CLOSING) return; s->session_state = SESSION_STATE_TRANSPORT_CLOSING; app_wrk = app_worker_get (s->app_wrk_index); app_worker_reset_notify (app_wrk, s); } int session_stream_accept_notify (transport_connection_t * tc) { app_worker_t *app_wrk; session_t *s; s = session_get (tc->s_index, tc->thread_index); app_wrk = app_worker_get_if_valid (s->app_wrk_index); if (!app_wrk) return -1; s->session_state = SESSION_STATE_ACCEPTING; return app_worker_accept_notify (app_wrk, s); } /** * Accept a stream session. Optionally ping the server by callback. */ int session_stream_accept (transport_connection_t * tc, u32 listener_index, u32 thread_index, u8 notify) { session_t *s; int rv; s = session_alloc_for_connection (tc); s->listener_handle = ((u64) thread_index << 32) | (u64) listener_index; s->session_state = SESSION_STATE_CREATED; if ((rv = app_worker_init_accepted (s))) return rv; session_lookup_add_connection (tc, session_handle (s)); /* Shoulder-tap the server */ if (notify) { app_worker_t *app_wrk = app_worker_get (s->app_wrk_index); return app_worker_accept_notify (app_wrk, s); } return 0; } int session_open_cl (u32 app_wrk_index, session_endpoint_t * rmt, u32 opaque) { transport_connection_t *tc; transport_endpoint_cfg_t *tep; app_worker_t *app_wrk; session_handle_t sh; session_t *s; int rv; tep = session_endpoint_to_transport_cfg (rmt); rv = transport_connect (rmt->transport_proto, tep); if (rv < 0) { SESSION_DBG ("Transport failed to open connection."); return VNET_API_ERROR_SESSION_CONNECT; } tc = transport_get_half_open (rmt->transport_proto, (u32) rv); /* For dgram type of service, allocate session and fifos now */ app_wrk = app_worker_get (app_wrk_index); s = session_alloc_for_connection (tc); s->app_wrk_index = app_wrk->wrk_index; s->session_state = SESSION_STATE_OPENED; if (app_worker_init_connected (app_wrk, s)) { session_free (s); return -1; } sh = session_handle (s); session_lookup_add_connection (tc, sh); return app_worker_connect_notify (app_wrk, s, opaque); } int session_open_vc (u32 app_wrk_index, session_endpoint_t * rmt, u32 opaque) { transport_connection_t *tc; transport_endpoint_cfg_t *tep; u64 handle; int rv; tep = session_endpoint_to_transport_cfg (rmt); rv = transport_connect (rmt->transport_proto, tep); if (rv < 0) { SESSION_DBG ("Transport failed to open connection."); return VNET_API_ERROR_SESSION_CONNECT; } tc = transport_get_half_open (rmt->transport_proto, (u32) rv); /* If transport offers a stream service, only allocate session once the * connection has been established. * Add connection to half-open table and save app and tc index. The * latter is needed to help establish the connection while the former * is needed when the connect notify comes and we have to notify the * external app */ handle = (((u64) app_wrk_index) << 32) | (u64) tc->c_index; session_lookup_add_half_open (tc, handle); /* Store api_context (opaque) for when the reply comes. Not the nicest * thing but better than allocating a separate half-open pool. */ tc->s_index = opaque; if (transport_half_open_has_fifos (rmt->transport_proto)) return session_ho_stream_connect_notify (tc, 0 /* is_fail */ ); return 0; } int session_open_app (u32 app_wrk_index, session_endpoint_t * rmt, u32 opaque) { session_endpoint_cfg_t *sep = (session_endpoint_cfg_t *) rmt; transport_endpoint_cfg_t *tep_cfg = session_endpoint_to_transport_cfg (sep); sep->app_wrk_index = app_wrk_index; sep->opaque = opaque; return transport_connect (rmt->transport_proto, tep_cfg); } typedef int (*session_open_service_fn) (u32, session_endpoint_t *, u32); /* *INDENT-OFF* */ static session_open_service_fn session_open_srv_fns[TRANSPORT_N_SERVICES] = { session_open_vc, session_open_cl, session_open_app, }; /* *INDENT-ON* */ /** * Ask transport to open connection to remote transport endpoint. * * Stores handle for matching request with reply since the call can be * asynchronous. For instance, for TCP the 3-way handshake must complete * before reply comes. Session is only created once connection is established. * * @param app_index Index of the application requesting the connect * @param st Session type requested. * @param tep Remote transport endpoint * @param opaque Opaque data (typically, api_context) the application expects * on open completion. */ int session_open (u32 app_wrk_index, session_endpoint_t * rmt, u32 opaque) { transport_service_type_t tst; tst = transport_protocol_service_type (rmt->transport_proto); return session_open_srv_fns[tst] (app_wrk_index, rmt, opaque); } /** * Ask transport to listen on session endpoint. * * @param s Session for which listen will be called. Note that unlike * established sessions, listen sessions are not associated to a * thread. * @param sep Local endpoint to be listened on. */ int session_listen (session_t * ls, session_endpoint_cfg_t * sep) { transport_endpoint_t *tep; u32 tc_index, s_index; /* Transport bind/listen */ tep = session_endpoint_to_transport (sep); s_index = ls->session_index; tc_index = transport_start_listen (session_get_transport_proto (ls), s_index, tep); if (tc_index == (u32) ~ 0) return -1; /* Attach transport to session. Lookup tables are populated by the app * worker because local tables (for ct sessions) are not backed by a fib */ ls = listen_session_get (s_index); ls->connection_index = tc_index; return 0; } /** * Ask transport to stop listening on local transport endpoint. * * @param s Session to stop listening on. It must be in state LISTENING. */ int session_stop_listen (session_t * s) { transport_proto_t tp = session_get_transport_proto (s); transport_connection_t *tc; if (s->session_state != SESSION_STATE_LISTENING) return -1; tc = transport_get_listener (tp, s->connection_index); if (!tc) return VNET_API_ERROR_ADDRESS_NOT_IN_USE; if (!(tc->flags & TRANSPORT_CONNECTION_F_NO_LOOKUP)) session_lookup_del_connection (tc); transport_stop_listen (tp, s->connection_index); return 0; } /** * Initialize session closing procedure. * * Request is always sent to session node to ensure that all outstanding * requests are served before transport is notified. */ void session_close (session_t * s) { if (!s) return; if (s->session_state >= SESSION_STATE_CLOSING) { /* Session will only be removed once both app and transport * acknowledge the close */ if (s->session_state == SESSION_STATE_TRANSPORT_CLOSED || s->session_state == SESSION_STATE_TRANSPORT_DELETED) session_program_transport_close (s); return; } s->session_state = SESSION_STATE_CLOSING; session_program_transport_close (s); } /** * Notify transport the session can be disconnected. This should eventually * result in a delete notification that allows us to cleanup session state. * Called for both active/passive disconnects. * * Must be called from the session's thread. */ void session_transport_close (session_t * s) { if (s->session_state >= SESSION_STATE_APP_CLOSED) { if (s->session_state == SESSION_STATE_TRANSPORT_CLOSED) s->session_state = SESSION_STATE_CLOSED; /* If transport is already deleted, just free the session */ else if (s->session_state >= SESSION_STATE_TRANSPORT_DELETED) session_free_w_fifos (s); return; } /* If the tx queue wasn't drained, the transport can continue to try * sending the outstanding data (in closed state it cannot). It MUST however * at one point, either after sending everything or after a timeout, call * delete notify. This will finally lead to the complete cleanup of the * session. */ s->session_state = SESSION_STATE_APP_CLOSED; transport_close (session_get_transport_proto (s), s->connection_index, s->thread_index); } /** * Cleanup transport and session state. * * Notify transport of the cleanup and free the session. This should * be called only if transport reported some error and is already * closed. */ void session_transport_cleanup (session_t * s) { s->session_state = SESSION_STATE_CLOSED; /* Delete from main lookup table before we axe the the transport */ session_lookup_del_session (s); transport_cleanup (session_get_transport_proto (s), s->connection_index, s->thread_index); /* Since we called cleanup, no delete notification will come. So, make * sure the session is properly freed. */ session_free_w_fifos (s); } /** * Allocate event queues in the shared-memory segment * * That can either be a newly created memfd segment, that will need to be * mapped by all stack users, or the binary api's svm region. The latter is * assumed to be already mapped. NOTE that this assumption DOES NOT hold if * api clients bootstrap shm api over sockets (i.e. use memfd segments) and * vpp uses api svm region for event queues. */ void session_vpp_event_queues_allocate (session_main_t * smm) { u32 evt_q_length = 2048, evt_size = sizeof (session_event_t); ssvm_private_t *eqs = &smm->evt_qs_segment; api_main_t *am = &api_main; uword eqs_size = 64 << 20; pid_t vpp_pid = getpid (); void *oldheap; int i; if (smm->configured_event_queue_length) evt_q_length = smm->configured_event_queue_length; if (smm->evt_qs_use_memfd_seg) { if (smm->evt_qs_segment_size) eqs_size = smm->evt_qs_segment_size; eqs->ssvm_size = eqs_size; eqs->i_am_master = 1; eqs->my_pid = vpp_pid; eqs->name = format (0, "%s%c", "evt-qs-segment", 0); eqs->requested_va = smm->session_baseva; if (ssvm_master_init (eqs, SSVM_SEGMENT_MEMFD)) { clib_warning ("failed to initialize queue segment"); return; } } if (smm->evt_qs_use_memfd_seg) oldheap = ssvm_push_heap (eqs->sh); else oldheap = svm_push_data_heap (am->vlib_rp); for (i = 0; i < vec_len (smm->wrk); i++) { svm_msg_q_cfg_t _cfg, *cfg = &_cfg; svm_msg_q_ring_cfg_t rc[SESSION_MQ_N_RINGS] = { {evt_q_length, evt_size, 0} , {evt_q_length >> 1, 256, 0} }; cfg->consumer_pid = 0; cfg->n_rings = 2; cfg->q_nitems = evt_q_length; cfg->ring_cfgs = rc; smm->wrk[i].vpp_event_queue = svm_msg_q_alloc (cfg); if (smm->evt_qs_use_memfd_seg) { if (svm_msg_q_alloc_consumer_eventfd (smm->wrk[i].vpp_event_queue)) clib_warning ("eventfd returned"); } } if (smm->evt_qs_use_memfd_seg) ssvm_pop_heap (oldheap); else svm_pop_heap (oldheap); } ssvm_private_t * session_main_get_evt_q_segment (void) { session_main_t *smm = &session_main; if (smm->evt_qs_use_memfd_seg) return &smm->evt_qs_segment; return 0; } u64 session_segment_handle (session_t * s) { svm_fifo_t *f; if (!s->rx_fifo) return SESSION_INVALID_HANDLE; f = s->rx_fifo; return segment_manager_make_segment_handle (f->segment_manager, f->segment_index); } /* *INDENT-OFF* */ static session_fifo_rx_fn *session_tx_fns[TRANSPORT_TX_N_FNS] = { session_tx_fifo_peek_and_snd, session_tx_fifo_dequeue_and_snd, session_tx_fifo_dequeue_internal, session_tx_fifo_dequeue_and_snd }; /* *INDENT-ON* */ /** * Initialize session layer for given transport proto and ip version * * Allocates per session type (transport proto + ip version) data structures * and adds arc from session queue node to session type output node. */ void session_register_transport (transport_proto_t transport_proto, const transport_proto_vft_t * vft, u8 is_ip4, u32 output_node) { session_main_t *smm = &session_main; session_type_t session_type; u32 next_index = ~0; session_type = session_type_from_proto_and_ip (transport_proto, is_ip4); vec_validate (smm->session_type_to_next, session_type); vec_validate (smm->session_tx_fns, session_type); /* *INDENT-OFF* */ if (output_node != ~0) { foreach_vlib_main (({ next_index = vlib_node_add_next (this_vlib_main, session_queue_node.index, output_node); })); } /* *INDENT-ON* */ smm->session_type_to_next[session_type] = next_index; smm->session_tx_fns[session_type] = session_tx_fns[vft->transport_options.tx_type]; } transport_connection_t * session_get_transport (session_t * s) { if (s->session_state != SESSION_STATE_LISTENING) return transport_get_connection (session_get_transport_proto (s), s->connection_index, s->thread_index); else return transport_get_listener (session_get_transport_proto (s), s->connection_index); } void session_get_endpoint (session_t * s, transport_endpoint_t * tep, u8 is_lcl) { if (s->session_state != SESSION_STATE_LISTENING) return transport_get_endpoint (session_get_transport_proto (s), s->connection_index, s->thread_index, tep, is_lcl); else return transport_get_listener_endpoint (session_get_transport_proto (s), s->connection_index, tep, is_lcl); } transport_connection_t * listen_session_get_transport (session_t * s) { return transport_get_listener (session_get_transport_proto (s), s->connection_index); } void session_flush_frames_main_thread (vlib_main_t * vm) { ASSERT (vlib_get_thread_index () == 0); vlib_process_signal_event_mt (vm, session_queue_process_node.index, SESSION_Q_PROCESS_FLUSH_FRAMES, 0); } static clib_error_t * session_manager_main_enable (vlib_main_t * vm) { segment_manager_main_init_args_t _sm_args = { 0 }, *sm_args = &_sm_args; session_main_t *smm = &session_main; vlib_thread_main_t *vtm = vlib_get_thread_main (); u32 num_threads, preallocated_sessions_per_worker; session_worker_t *wrk; int i; num_threads = 1 /* main thread */ + vtm->n_threads; if (num_threads < 1) return clib_error_return (0, "n_thread_stacks not set"); /* Allocate cache line aligned worker contexts */ vec_validate_aligned (smm->wrk, num_threads - 1, CLIB_CACHE_LINE_BYTES); for (i = 0; i < num_threads; i++) { wrk = &smm->wrk[i]; wrk->new_head = clib_llist_make_head (wrk->event_elts, evt_list); wrk->old_head = clib_llist_make_head (wrk->event_elts, evt_list); wrk->postponed_head = clib_llist_make_head (wrk->event_elts, evt_list); wrk->disconnects_head = clib_llist_make_head (wrk->event_elts, evt_list); wrk->vm = vlib_mains[i]; wrk->last_vlib_time = vlib_time_now (vlib_mains[i]); if (num_threads > 1) clib_rwlock_init (&smm->wrk[i].peekers_rw_locks); } /* Allocate vpp event queues segment and queue */ session_vpp_event_queues_allocate (smm); /* Initialize fifo segment main baseva and timeout */ sm_args->baseva = smm->session_baseva + smm->evt_qs_segment_size; sm_args->size = smm->session_va_space_size; segment_manager_main_init (sm_args); /* Preallocate sessions */ if (smm->preallocated_sessions) { if (num_threads == 1) { pool_init_fixed (smm->wrk[0].sessions, smm->preallocated_sessions); } else { int j; preallocated_sessions_per_worker = (1.1 * (f64) smm->preallocated_sessions / (f64) (num_threads - 1)); for (j = 1; j < num_threads; j++) { pool_init_fixed (smm->wrk[j].sessions, preallocated_sessions_per_worker); } } } session_lookup_init (); app_namespaces_init (); transport_init (); smm->is_enabled = 1; /* Enable transports */ transport_enable_disable (vm, 1); transport_init_tx_pacers_period (); return 0; } void session_node_enable_disable (u8 is_en) { u8 state = is_en ? VLIB_NODE_STATE_POLLING : VLIB_NODE_STATE_DISABLED; vlib_thread_main_t *vtm = vlib_get_thread_main (); u8 have_workers = vtm->n_threads != 0; /* *INDENT-OFF* */ foreach_vlib_main (({ if (have_workers && ii == 0) { vlib_node_set_state (this_vlib_main, session_queue_process_node.index, state); if (is_en) { vlib_node_t *n = vlib_get_node (this_vlib_main, session_queue_process_node.index); vlib_start_process (this_vlib_main, n->runtime_index); } else { vlib_process_signal_event_mt (this_vlib_main, session_queue_process_node.index, SESSION_Q_PROCESS_STOP, 0); } continue; } vlib_node_set_state (this_vlib_main, session_queue_node.index, state); })); /* *INDENT-ON* */ } clib_error_t * vnet_session_enable_disable (vlib_main_t * vm, u8 is_en) { clib_error_t *error = 0; if (is_en) { if (session_main.is_enabled) return 0; error = session_manager_main_enable (vm); session_node_enable_disable (is_en); } else { session_main.is_enabled = 0; session_node_enable_disable (is_en); } return error; } clib_error_t * session_manager_main_init (vlib_main_t * vm) { session_main_t *smm = &session_main; smm->session_baseva = HIGH_SEGMENT_BASEVA; #if (HIGH_SEGMENT_BASEVA > (4ULL << 30)) smm->session_va_space_size = 128ULL << 30; smm->evt_qs_segment_size = 64 << 20; #else smm->session_va_space_size = 128 << 20; smm->evt_qs_segment_size = 1 << 20; #endif smm->is_enabled = 0; return 0; } VLIB_INIT_FUNCTION (session_manager_main_init); static clib_error_t * session_config_fn (vlib_main_t * vm, unformat_input_t * input) { session_main_t *smm = &session_main; u32 nitems; uword tmp; while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) { if (unformat (input, "event-queue-length %d", &nitems)) { if (nitems >= 2048) smm->configured_event_queue_length = nitems; else clib_warning ("event queue length %d too small, ignored", nitems); } else if (unformat (input, "preallocated-sessions %d", &smm->preallocated_sessions)) ; else if (unformat (input, "v4-session-table-buckets %d", &smm->configured_v4_session_table_buckets)) ; else if (unformat (input, "v4-halfopen-table-buckets %d", &smm->configured_v4_halfopen_table_buckets)) ; else if (unformat (input, "v6-session-table-buckets %d", &smm->configured_v6_session_table_buckets)) ; else if (unformat (input, "v6-halfopen-table-buckets %d", &smm->configured_v6_halfopen_table_buckets)) ; else if (unformat (input, "v4-session-table-memory %U", unformat_memory_size, &tmp)) { if (tmp >= 0x100000000) return clib_error_return (0, "memory size %llx (%lld) too large", tmp, tmp); smm->configured_v4_session_table_memory = tmp; } else if (unformat (input, "v4-halfopen-table-memory %U", unformat_memory_size, &tmp)) { if (tmp >= 0x100000000) return clib_error_return (0, "memory size %llx (%lld) too large", tmp, tmp); smm->configured_v4_halfopen_table_memory = tmp; } else if (unformat (input, "v6-session-table-memory %U", unformat_memory_size, &tmp)) { if (tmp >= 0x100000000) return clib_error_return (0, "memory size %llx (%lld) too large", tmp, tmp); smm->configured_v6_session_table_memory = tmp; } else if (unformat (input, "v6-halfopen-table-memory %U", unformat_memory_size, &tmp)) { if (tmp >= 0x100000000) return clib_error_return (0, "memory size %llx (%lld) too large", tmp, tmp); smm->configured_v6_halfopen_table_memory = tmp; } else if (unformat (input, "local-endpoints-table-memory %U", unformat_memory_size, &tmp)) { if (tmp >= 0x100000000) return clib_error_return (0, "memory size %llx (%lld) too large", tmp, tmp); smm->local_endpoints_table_memory = tmp; } else if (unformat (input, "local-endpoints-table-buckets %d", &smm->local_endpoints_table_buckets)) ; else if (unformat (input, "evt_qs_memfd_seg")) smm->evt_qs_use_memfd_seg = 1; else if (unformat (input, "evt_qs_seg_size %U", unformat_memory_size, &smm->evt_qs_segment_size)) ; else return clib_error_return (0, "unknown input `%U'", format_unformat_error, input); } return 0; } VLIB_CONFIG_FUNCTION (session_config_fn, "session"); /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */