proto.c 48.6 KB
Newer Older
1 2 3
/*
 *	BIRD -- Protocols
 *
4
 *	(c) 1998--2000 Martin Mares <mj@ucw.cz>
5 6 7 8
 *
 *	Can be freely distributed and used under the terms of the GNU GPL.
 */

9
#undef LOCAL_DEBUG
Martin Mareš's avatar
Martin Mareš committed
10

11 12 13 14
#include "nest/bird.h"
#include "nest/protocol.h"
#include "lib/resource.h"
#include "lib/lists.h"
15
#include "lib/event.h"
16
#include "lib/timer.h"
17
#include "lib/string.h"
18
#include "conf/conf.h"
19 20
#include "nest/route.h"
#include "nest/iface.h"
21
#include "nest/cli.h"
22
#include "filter/filter.h"
23

24
pool *proto_pool;
25
list  proto_list;
26

27
static list protocol_list;
28
struct protocol *class_to_protocol[PROTOCOL__MAX];
29

30 31
#define PD(pr, msg, args...) do { if (pr->debug & D_STATES) { log(L_TRACE "%s: " msg, pr->name , ## args); } } while(0)

32
static timer *proto_shutdown_timer;
33 34 35 36 37 38 39 40 41
static timer *gr_wait_timer;

#define GRS_NONE	0
#define GRS_INIT	1
#define GRS_ACTIVE	2
#define GRS_DONE	3

static int graceful_restart_state;
static u32 graceful_restart_locks;
42 43

static char *p_states[] = { "DOWN", "START", "UP", "STOP" };
Ondřej Zajíček's avatar
Ondřej Zajíček committed
44
static char *c_states[] = { "DOWN", "START", "UP", "FLUSHING" };
45

46 47
extern struct protocol proto_unix_iface;

48
static void proto_shutdown_loop(timer *);
49
static void proto_rethink_goal(struct proto *p);
50
static char *proto_state_name(struct proto *p);
51
static void channel_verify_limits(struct channel *c);
Ondřej Zajíček's avatar
Ondřej Zajíček committed
52
static inline void channel_reset_limit(struct channel_limit *l);
53 54


55 56
static inline int proto_is_done(struct proto *p)
{ return (p->proto_state == PS_DOWN) && (p->active_channels == 0); }
57

58 59
static inline int channel_is_active(struct channel *c)
{ return (c->channel_state == CS_START) || (c->channel_state == CS_UP); }
60 61 62 63 64

static void
proto_log_state_change(struct proto *p)
{
  if (p->debug & D_STATES)
65 66 67
  {
    char *name = proto_state_name(p);
    if (name != p->last_state_name_announced)
68
    {
69 70
      p->last_state_name_announced = name;
      PD(p, "State changed to %s", proto_state_name(p));
71
    }
72
  }
73 74
  else
    p->last_state_name_announced = NULL;
75
}
76

77

78 79
struct channel_config *
proto_cf_find_channel(struct proto_config *pc, uint net_type)
Martin Mareš's avatar
Martin Mareš committed
80
{
81 82 83 84 85 86 87
  struct channel_config *cc;

  WALK_LIST(cc, pc->channels)
    if (cc->net_type == net_type)
      return cc;

  return NULL;
Martin Mareš's avatar
Martin Mareš committed
88 89
}

90 91 92 93 94 95 96 97 98
/**
 * proto_find_channel_by_table - find channel connected to a routing table
 * @p: protocol instance
 * @t: routing table
 *
 * Returns pointer to channel or NULL
 */
struct channel *
proto_find_channel_by_table(struct proto *p, struct rtable *t)
99
{
100
  struct channel *c;
101

102 103 104
  WALK_LIST(c, p->channels)
    if (c->table == t)
      return c;
105

106
  return NULL;
107 108
}

109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
/**
 * proto_find_channel_by_name - find channel by its name
 * @p: protocol instance
 * @n: channel name
 *
 * Returns pointer to channel or NULL
 */
struct channel *
proto_find_channel_by_name(struct proto *p, const char *n)
{
  struct channel *c;

  WALK_LIST(c, p->channels)
    if (!strcmp(c->name, n))
      return c;

  return NULL;
}

128
/**
129
 * proto_add_channel - connect protocol to a routing table
130
 * @p: protocol instance
131
 * @cf: channel configuration
132
 *
133 134 135 136
 * This function creates a channel between the protocol instance @p and the
 * routing table specified in the configuration @cf, making the protocol hear
 * all changes in the table and allowing the protocol to update routes in the
 * table.
137
 *
138 139 140
 * The channel is linked in the protocol channel list and when active also in
 * the table channel list. Channels are allocated from the global resource pool
 * (@proto_pool) and they are automatically freed when the protocol is removed.
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

struct channel *
proto_add_channel(struct proto *p, struct channel_config *cf)
{
  struct channel *c = mb_allocz(proto_pool, cf->channel->channel_size);

  c->name = cf->name;
  c->channel = cf->channel;
  c->proto = p;
  c->table = cf->table->table;

  c->in_filter = cf->in_filter;
  c->out_filter = cf->out_filter;
  c->rx_limit = cf->rx_limit;
  c->in_limit = cf->in_limit;
  c->out_limit = cf->out_limit;

  c->net_type = cf->net_type;
  c->ra_mode = cf->ra_mode;
  c->preference = cf->preference;
  c->merge_limit = cf->merge_limit;
  c->in_keep_filtered = cf->in_keep_filtered;

  c->channel_state = CS_DOWN;
  c->export_state = ES_DOWN;
167
  c->last_state_change = current_time();
168
  c->last_tx_filter_change = current_time();
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
  c->reloadable = 1;

  CALL(c->channel->init, c, cf);

  add_tail(&p->channels, &c->n);

  PD(p, "Channel %s connected to table %s", c->name, c->table->name);

  return c;
}

void
proto_remove_channel(struct proto *p, struct channel *c)
{
  ASSERT(c->channel_state == CS_DOWN);

  PD(p, "Channel %s removed", c->name);

  rem_node(&c->n);
  mb_free(c);
}


static void
proto_start_channels(struct proto *p)
{
  struct channel *c;
  WALK_LIST(c, p->channels)
    if (!c->disabled)
      channel_set_state(c, CS_UP);
}

static void
proto_pause_channels(struct proto *p)
203
{
204 205 206 207 208
  struct channel *c;
  WALK_LIST(c, p->channels)
    if (!c->disabled && channel_is_active(c))
      channel_set_state(c, CS_START);
}
209

210 211 212 213 214 215 216 217
static void
proto_stop_channels(struct proto *p)
{
  struct channel *c;
  WALK_LIST(c, p->channels)
    if (!c->disabled && channel_is_active(c))
      channel_set_state(c, CS_FLUSHING);
}
218

219 220 221 222 223 224 225 226 227 228 229 230 231
static void
proto_remove_channels(struct proto *p)
{
  struct channel *c;
  WALK_LIST_FIRST(c, p->channels)
    proto_remove_channel(p, c);
}

static void
channel_schedule_feed(struct channel *c, int initial)
{
  // DBG("%s: Scheduling meal\n", p->name);
  ASSERT(c->channel_state == CS_UP);
232

233 234
  c->export_state = ES_FEEDING;
  c->refeeding = !initial;
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
  ev_schedule(c->feed_event);
}

static void
channel_feed_loop(void *ptr)
{
  struct channel *c = ptr;

  if (c->export_state != ES_FEEDING)
    return;

  if (!c->feed_active)
    if (c->proto->feed_begin)
      c->proto->feed_begin(c, !c->refeeding);

  // DBG("Feeding protocol %s continued\n", p->name);
  if (!rt_feed_channel(c))
  {
    ev_schedule(c->feed_event);
    return;
  }

  // DBG("Feeding protocol %s finished\n", p->name);
  c->export_state = ES_READY;
  // proto_log_state_change(p);

  if (c->proto->feed_end)
    c->proto->feed_end(c);
}


static void
channel_start_export(struct channel *c)
{
  ASSERT(c->channel_state == CS_UP);
  ASSERT(c->export_state == ES_DOWN);

  channel_schedule_feed(c, 1);	/* Sets ES_FEEDING */
}

static void
channel_stop_export(struct channel *c)
{
  /* Need to abort feeding */
  if (c->export_state == ES_FEEDING)
    rt_feed_channel_abort(c);

  c->export_state = ES_DOWN;
284
  c->stats.exp_routes = 0;
285 286
}

Ondřej Zajíček's avatar
Ondřej Zajíček committed
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

/* Called by protocol for reload from in_table */
void
channel_schedule_reload(struct channel *c)
{
  ASSERT(c->channel_state == CS_UP);

  rt_reload_channel_abort(c);
  ev_schedule(c->reload_event);
}

static void
channel_reload_loop(void *ptr)
{
  struct channel *c = ptr;

  if (!rt_reload_channel(c))
  {
    ev_schedule(c->reload_event);
    return;
  }
}

static void
channel_reset_import(struct channel *c)
{
  /* Need to abort feeding */
  ev_postpone(c->reload_event);
  rt_reload_channel_abort(c);

  rt_prune_sync(c->in_table, 1);
}

/* Called by protocol to activate in_table */
void
channel_setup_in_table(struct channel *c)
{
  struct rtable_config *cf = mb_allocz(c->proto->pool, sizeof(struct rtable_config));
  cf->name = "import";
  cf->addr_type = c->net_type;

  c->in_table = mb_allocz(c->proto->pool, sizeof(struct rtable));
  rt_setup(c->proto->pool, c->in_table, cf);

  c->reload_event = ev_new_init(c->proto->pool, channel_reload_loop, c);
}


335 336 337 338 339 340 341
static void
channel_do_start(struct channel *c)
{
  rt_lock_table(c->table);
  add_tail(&c->table->channels, &c->table_node);
  c->proto->active_channels++;

342
  c->feed_event = ev_new_init(c->proto->pool, channel_feed_loop, c);
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365

  channel_reset_limit(&c->rx_limit);
  channel_reset_limit(&c->in_limit);
  channel_reset_limit(&c->out_limit);

  CALL(c->channel->start, c);
}

static void
channel_do_flush(struct channel *c)
{
  rt_schedule_prune(c->table);

  c->gr_wait = 0;
  if (c->gr_lock)
    channel_graceful_restart_unlock(c);

  CALL(c->channel->shutdown, c);
}

static void
channel_do_down(struct channel *c)
{
Ondřej Zajíček's avatar
Ondřej Zajíček committed
366 367
  ASSERT(!c->feed_active && !c->reload_active);

368
  rem_node(&c->table_node);
369 370 371 372 373 374 375 376
  rt_unlock_table(c->table);
  c->proto->active_channels--;

  if ((c->stats.imp_routes + c->stats.filt_routes) != 0)
    log(L_ERR "%s: Channel %s is down but still has some routes", c->proto->name, c->name);

  memset(&c->stats, 0, sizeof(struct proto_stats));

Ondřej Zajíček's avatar
Ondřej Zajíček committed
377 378 379
  c->in_table = NULL;
  c->reload_event = NULL;

Ondřej Zajíček's avatar
Ondřej Zajíček committed
380 381
  CALL(c->channel->cleanup, c);

382 383 384 385 386 387 388 389 390 391 392
  /* Schedule protocol shutddown */
  if (proto_is_done(c->proto))
    ev_schedule(c->proto->event);
}

void
channel_set_state(struct channel *c, uint state)
{
  uint cs = c->channel_state;
  uint es = c->export_state;

Ondřej Zajíček's avatar
Ondřej Zajíček committed
393
  DBG("%s reporting channel %s state transition %s -> %s\n", c->proto->name, c->name, c_states[cs], c_states[state]);
394 395 396 397
  if (state == cs)
    return;

  c->channel_state = state;
398
  c->last_state_change = current_time();
399 400 401 402 403 404 405 406 407 408 409 410

  switch (state)
  {
  case CS_START:
    ASSERT(cs == CS_DOWN || cs == CS_UP);

    if (cs == CS_DOWN)
      channel_do_start(c);

    if (es != ES_DOWN)
      channel_stop_export(c);

Ondřej Zajíček's avatar
Ondřej Zajíček committed
411 412 413
    if (c->in_table && (cs == CS_UP))
      channel_reset_import(c);

414 415 416 417 418 419 420 421
    break;

  case CS_UP:
    ASSERT(cs == CS_DOWN || cs == CS_START);

    if (cs == CS_DOWN)
      channel_do_start(c);

422
    if (!c->gr_wait && c->proto->rt_notify)
423 424 425 426 427 428 429 430 431 432
      channel_start_export(c);

    break;

  case CS_FLUSHING:
    ASSERT(cs == CS_START || cs == CS_UP);

    if (es != ES_DOWN)
      channel_stop_export(c);

Ondřej Zajíček's avatar
Ondřej Zajíček committed
433 434 435
    if (c->in_table && (cs == CS_UP))
      channel_reset_import(c);

436 437 438 439 440 441 442 443 444 445 446 447 448
    channel_do_flush(c);
    break;

  case CS_DOWN:
    ASSERT(cs == CS_FLUSHING);

    channel_do_down(c);
    break;

  default:
    ASSERT(0);
  }
  // XXXX proto_log_state_change(c);
449 450
}

451
/**
452 453
 * channel_request_feeding - request feeding routes to the channel
 * @c: given channel
454
 *
455 456 457 458 459
 * Sometimes it is needed to send again all routes to the channel. This is
 * called feeding and can be requested by this function. This would cause
 * channel export state transition to ES_FEEDING (during feeding) and when
 * completed, it will switch back to ES_READY. This function can be called
 * even when feeding is already running, in that case it is restarted.
460
 */
461 462
void
channel_request_feeding(struct channel *c)
463
{
464
  ASSERT(c->channel_state == CS_UP);
465

466 467 468
  /* Do nothing if we are still waiting for feeding */
  if (c->export_state == ES_DOWN)
    return;
469

470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
  /* If we are already feeding, we want to restart it */
  if (c->export_state == ES_FEEDING)
  {
    /* Unless feeding is in initial state */
    if (!c->feed_active)
	return;

    rt_feed_channel_abort(c);
  }

  channel_reset_limit(&c->out_limit);

  /* Hack: reset exp_routes during refeed, and do not decrease it later */
  c->stats.exp_routes = 0;

  channel_schedule_feed(c, 0);	/* Sets ES_FEEDING */
  // proto_log_state_change(c);
}

static inline int
channel_reloadable(struct channel *c)
{
  return c->proto->reload_routes && c->reloadable;
493 494
}

495
static void
496
channel_request_reload(struct channel *c)
497
{
498
  ASSERT(c->channel_state == CS_UP);
499
  ASSERT(channel_reloadable(c));
500 501

  c->proto->reload_routes(c);
502

503 504 505 506 507 508
  /*
   * Should this be done before reload_routes() hook?
   * Perhaps, but routes are updated asynchronously.
   */
  channel_reset_limit(&c->rx_limit);
  channel_reset_limit(&c->in_limit);
509 510
}

511 512 513 514 515 516
const struct channel_class channel_basic = {
  .channel_size = sizeof(struct channel),
  .config_size = sizeof(struct channel_config)
};

void *
517
channel_config_new(const struct channel_class *cc, const char *name, uint net_type, struct proto_config *proto)
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
{
  struct channel_config *cf = NULL;
  struct rtable_config *tab = NULL;

  if (net_type)
  {
    if (!net_val_match(net_type, proto->protocol->channel_mask))
      cf_error("Unsupported channel type");

    if (proto->net_type && (net_type != proto->net_type))
      cf_error("Different channel type");

    tab = new_config->def_tables[net_type];
  }

  if (!cc)
    cc = &channel_basic;

  cf = cfg_allocz(cc->config_size);
  cf->name = name;
  cf->channel = cc;
539
  cf->parent = proto;
540 541 542 543 544 545 546 547 548 549 550 551
  cf->table = tab;
  cf->out_filter = FILTER_REJECT;

  cf->net_type = net_type;
  cf->ra_mode = RA_OPTIMAL;
  cf->preference = proto->protocol->preference;

  add_tail(&proto->channels, &cf->n);

  return cf;
}

552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571
void *
channel_config_get(const struct channel_class *cc, const char *name, uint net_type, struct proto_config *proto)
{
  struct channel_config *cf;

  /* We are using name as token, so no strcmp() */
  WALK_LIST(cf, proto->channels)
    if (cf->name == name)
    {
      /* Allow to redefine channel only if inherited from template */
      if (cf->parent == proto)
	cf_error("Multiple %s channels", name);

      cf->parent = proto;
      return cf;
    }

  return channel_config_new(cc, name, net_type, proto);
}

572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590
struct channel_config *
channel_copy_config(struct channel_config *src, struct proto_config *proto)
{
  struct channel_config *dst = cfg_alloc(src->channel->config_size);

  memcpy(dst, src, src->channel->config_size);
  add_tail(&proto->channels, &dst->n);
  CALL(src->channel->copy_config, dst, src);

  return dst;
}


static int reconfigure_type;  /* Hack to propagate type info to channel_reconfigure() */

int
channel_reconfigure(struct channel *c, struct channel_config *cf)
{
  /* FIXME: better handle these changes, also handle in_keep_filtered */
Ondřej Zajíček's avatar
Ondřej Zajíček committed
591
  if ((c->table != cf->table->table) || (cf->ra_mode && (c->ra_mode != cf->ra_mode)))
592 593
    return 0;

594 595 596
  /* Note that filter_same() requires arguments in (new, old) order */
  int import_changed = !filter_same(cf->in_filter, c->in_filter);
  int export_changed = !filter_same(cf->out_filter, c->out_filter);
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617

  if (c->preference != cf->preference)
    import_changed = 1;

  if (c->merge_limit != cf->merge_limit)
    export_changed = 1;

  /* Reconfigure channel fields */
  c->in_filter = cf->in_filter;
  c->out_filter = cf->out_filter;
  c->rx_limit = cf->rx_limit;
  c->in_limit = cf->in_limit;
  c->out_limit = cf->out_limit;

  // c->ra_mode = cf->ra_mode;
  c->merge_limit = cf->merge_limit;
  c->preference = cf->preference;
  c->in_keep_filtered = cf->in_keep_filtered;

  channel_verify_limits(c);

618 619 620
  if (export_changed)
    c->last_tx_filter_change = current_time();

Ondřej Zajíček's avatar
Ondřej Zajíček committed
621 622 623
  /* Execute channel-specific reconfigure hook */
  if (c->channel->reconfigure && !c->channel->reconfigure(c, cf))
    return 0;
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

  /* If the channel is not open, it has no routes and we cannot reload it anyways */
  if (c->channel_state != CS_UP)
    return 1;

  if (reconfigure_type == RECONFIG_SOFT)
  {
    if (import_changed)
      log(L_INFO "Channel %s.%s changed import", c->proto->name, c->name);

    if (export_changed)
      log(L_INFO "Channel %s.%s changed export", c->proto->name, c->name);

    return 1;
  }

  /* Route reload may be not supported */
  if (import_changed && !channel_reloadable(c))
    return 0;

  if (import_changed || export_changed)
    log(L_INFO "Reloading channel %s.%s", c->proto->name, c->name);

  if (import_changed)
    channel_request_reload(c);

  if (export_changed)
    channel_request_feeding(c);

  return 1;
}


int
proto_configure_channel(struct proto *p, struct channel **pc, struct channel_config *cf)
659
{
660
  struct channel *c = *pc;
661

662 663
  if (!c && cf)
  {
664 665
    /* We could add the channel, but currently it would just stay in down state
       until protocol is restarted, so it is better to force restart anyways. */
666 667 668 669 670 671 672
    if (p->proto_state != PS_DOWN)
    {
      log(L_INFO "Cannot add channel %s.%s", p->name, cf->name);
      return 0;
    }

    *pc = proto_add_channel(p, cf);
673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694
  }
  else if (c && !cf)
  {
    if (c->channel_state != CS_DOWN)
    {
      log(L_INFO "Cannot remove channel %s.%s", c->proto->name, c->name);
      return 0;
    }

    proto_remove_channel(p, c);
    *pc = NULL;
  }
  else if (c && cf)
  {
    if (!channel_reconfigure(c, cf))
    {
      log(L_INFO "Cannot reconfigure channel %s.%s", c->proto->name, c->name);
      return 0;
    }
  }

  return 1;
695 696
}

697

698
static void
699
proto_event(void *ptr)
700
{
701 702 703 704 705 706 707
  struct proto *p = ptr;

  if (p->do_start)
  {
    if_feed_baby(p);
    p->do_start = 0;
  }
708

709
  if (p->do_stop)
710
  {
711 712 713
    if (p->proto == &proto_unix_iface)
      if_flush_ifaces(p);
    p->do_stop = 0;
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
  if (proto_is_done(p))
  {
    if (p->proto->cleanup)
      p->proto->cleanup(p);

    p->active = 0;
    proto_log_state_change(p);
    proto_rethink_goal(p);
  }
}


/**
 * proto_new - create a new protocol instance
 * @c: protocol configuration
 *
 * When a new configuration has been read in, the core code starts
 * initializing all the protocol instances configured by calling their
 * init() hooks with the corresponding instance configuration. The initialization
 * code of the protocol is expected to create a new instance according to the
 * configuration by calling this function and then modifying the default settings
 * to values wanted by the protocol.
 */
void *
proto_new(struct proto_config *cf)
{
  struct proto *p = mb_allocz(proto_pool, cf->protocol->proto_size);

  p->cf = cf;
  p->debug = cf->debug;
  p->mrtdump = cf->mrtdump;
  p->name = cf->name;
  p->proto = cf->protocol;
  p->net_type = cf->net_type;
  p->disabled = cf->disabled;
  p->hash_key = random_u32();
  cf->proto = p;

  init_list(&p->channels);

  return p;
}

static struct proto *
proto_init(struct proto_config *c, node *n)
{
  struct protocol *pr = c->protocol;
  struct proto *p = pr->init(c);

  p->proto_state = PS_DOWN;
766
  p->last_state_change = current_time();
767
  p->vrf = c->vrf;
768 769
  insert_node(&p->n, n);

770
  p->event = ev_new_init(proto_pool, proto_event, p);
771 772 773 774 775 776 777 778 779 780 781 782 783 784

  PD(p, "Initializing%s", p->disabled ? " [disabled]" : "");

  return p;
}

static void
proto_start(struct proto *p)
{
  /* Here we cannot use p->cf->name since it won't survive reconfiguration */
  p->pool = rp_new(proto_pool, p->proto->name);

  if (graceful_restart_state == GRS_INIT)
    p->gr_recovery = 1;
785 786
}

787

788 789 790
/**
 * proto_config_new - create a new protocol configuration
 * @pr: protocol the configuration will belong to
791
 * @class: SYM_PROTO or SYM_TEMPLATE
792 793 794 795 796 797 798 799
 *
 * Whenever the configuration file says that a new instance
 * of a routing protocol should be created, the parser calls
 * proto_config_new() to create a configuration entry for this
 * instance (a structure staring with the &proto_config header
 * containing all the generic items followed by protocol-specific
 * ones). Also, the configuration entry gets added to the list
 * of protocol instances kept in the configuration.
800 801 802 803 804
 *
 * The function is also used to create protocol templates (when class
 * SYM_TEMPLATE is specified), the only difference is that templates
 * are not added to the list of protocol instances and therefore not
 * initialized during protos_commit()).
805
 */
806
void *
807
proto_config_new(struct protocol *pr, int class)
808
{
809
  struct proto_config *cf = cfg_allocz(pr->config_size);
810

811
  if (class == SYM_PROTO)
812 813 814 815 816 817 818 819 820 821 822 823
    add_tail(&new_config->protos, &cf->n);

  cf->global = new_config;
  cf->protocol = pr;
  cf->name = pr->name;
  cf->class = class;
  cf->debug = new_config->proto_default_debug;
  cf->mrtdump = new_config->proto_default_mrtdump;

  init_list(&cf->channels);

  return cf;
824 825
}

826

827 828 829 830 831 832 833 834 835 836 837 838 839 840
/**
 * proto_copy_config - copy a protocol configuration
 * @dest: destination protocol configuration
 * @src: source protocol configuration
 *
 * Whenever a new instance of a routing protocol is created from the
 * template, proto_copy_config() is called to copy a content of
 * the source protocol configuration to the new protocol configuration.
 * Name, class and a node in protos list of @dest are kept intact.
 * copy_config() protocol hook is used to copy protocol-specific data.
 */
void
proto_copy_config(struct proto_config *dest, struct proto_config *src)
{
841
  struct channel_config *cc;
842 843 844 845 846 847 848 849 850 851 852 853
  node old_node;
  int old_class;
  char *old_name;

  if (dest->protocol != src->protocol)
    cf_error("Can't copy configuration from a different protocol type");

  if (dest->protocol->copy_config == NULL)
    cf_error("Inheriting configuration for %s is not supported", src->protocol->name);

  DBG("Copying configuration from %s to %s\n", src->name, dest->name);

854
  /*
855 856 857 858 859 860 861 862
   * Copy struct proto_config here. Keep original node, class and name.
   * protocol-specific config copy is handled by protocol copy_config() hook
   */

  old_node = dest->n;
  old_class = dest->class;
  old_name = dest->name;

863
  memcpy(dest, src, src->protocol->config_size);
864 865 866 867

  dest->n = old_node;
  dest->class = old_class;
  dest->name = old_name;
868
  init_list(&dest->channels);
869

870 871 872 873
  WALK_LIST(cc, src->channels)
    channel_copy_config(cc, dest);

  /* FIXME: allow for undefined copy_config */
874 875 876
  dest->protocol->copy_config(dest, src);
}

Ondřej Zajíček's avatar
Ondřej Zajíček committed
877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898
void
proto_clone_config(struct symbol *sym, struct proto_config *parent)
{
  struct proto_config *cf = proto_config_new(parent->protocol, SYM_PROTO);
  proto_copy_config(cf, parent);
  cf->name = sym->name;
  cf->proto = NULL;
  cf->parent = parent;

  sym->class = cf->class;
  sym->def = cf;
}

static void
proto_undef_clone(struct symbol *sym, struct proto_config *cf)
{
  rem_node(&cf->n);

  sym->class = SYM_VOID;
  sym->def = NULL;
}

899 900 901 902 903 904 905 906
/**
 * protos_preconfig - pre-configuration processing
 * @c: new configuration
 *
 * This function calls the preconfig() hooks of all routing
 * protocols available to prepare them for reading of the new
 * configuration.
 */
907
void
908
protos_preconfig(struct config *c)
909
{
Martin Mareš's avatar
Martin Mareš committed
910 911
  struct protocol *p;

912
  init_list(&c->protos);
913
  DBG("Protocol preconfig:");
Martin Mareš's avatar
Martin Mareš committed
914
  WALK_LIST(p, protocol_list)
915 916 917 918 919 920
  {
    DBG(" %s", p->name);
    p->name_counter = 0;
    if (p->preconfig)
      p->preconfig(p, c);
  }
921
  DBG("\n");
922 923
}

924 925 926 927 928 929 930 931 932
static int
proto_reconfigure(struct proto *p, struct proto_config *oc, struct proto_config *nc, int type)
{
  /* If the protocol is DOWN, we just restart it */
  if (p->proto_state == PS_DOWN)
    return 0;

  /* If there is a too big change in core attributes, ... */
  if ((nc->protocol != oc->protocol) ||
933
      (nc->net_type != oc->net_type) ||
934
      (nc->disabled != p->disabled) ||
935
      (nc->vrf != oc->vrf))
936 937
    return 0;

938
  p->name = nc->name;
939 940
  p->debug = nc->debug;
  p->mrtdump = nc->mrtdump;
941
  reconfigure_type = type;
942 943

  /* Execute protocol specific reconfigure hook */
944
  if (!p->proto->reconfigure || !p->proto->reconfigure(p, nc))
945 946 947 948 949 950 951 952 953
    return 0;

  DBG("\t%s: same\n", oc->name);
  PD(p, "Reconfigured");
  p->cf = nc;

  return 1;
}

954 955 956 957 958 959
/**
 * protos_commit - commit new protocol configuration
 * @new: new configuration
 * @old: old configuration or %NULL if it's boot time config
 * @force_reconfig: force restart of all protocols (used for example
 * when the router ID changes)
960
 * @type: type of reconfiguration (RECONFIG_SOFT or RECONFIG_HARD)
961 962 963 964 965 966 967 968 969 970 971 972
 *
 * Scan differences between @old and @new configuration and adjust all
 * protocol instances to conform to the new configuration.
 *
 * When a protocol exists in the new configuration, but it doesn't in the
 * original one, it's immediately started. When a collision with the other
 * running protocol would arise, the new protocol will be temporarily stopped
 * by the locking mechanism.
 *
 * When a protocol exists in the old configuration, but it doesn't in the
 * new one, it's shut down and deleted after the shutdown completes.
 *
973 974 975 976 977 978 979 980
 * When a protocol exists in both configurations, the core decides
 * whether it's possible to reconfigure it dynamically - it checks all
 * the core properties of the protocol (changes in filters are ignored
 * if type is RECONFIG_SOFT) and if they match, it asks the
 * reconfigure() hook of the protocol to see if the protocol is able
 * to switch to the new configuration.  If it isn't possible, the
 * protocol is shut down and a new instance is started with the new
 * configuration after the shutdown is completed.
981
 */
982
void
983
protos_commit(struct config *new, struct config *old, int force_reconfig, int type)
984
{
985
  struct proto_config *oc, *nc;
986
  struct symbol *sym;
987 988 989
  struct proto *p;
  node *n;

990

991 992
  DBG("protos_commit:\n");
  if (old)
993 994
  {
    WALK_LIST(oc, old->protos)
995
    {
996 997
      p = oc->proto;
      sym = cf_find_symbol(new, oc->name);
Ondřej Zajíček's avatar
Ondřej Zajíček committed
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015

      /* Handle dynamic protocols */
      if (!sym && oc->parent && !new->shutdown)
      {
	struct symbol *parsym = cf_find_symbol(new, oc->parent->name);
	if (parsym && parsym->class == SYM_PROTO)
	{
	  /* This is hack, we would like to share config, but we need to copy it now */
	  new_config = new;
	  cfg_mem = new->mem;
	  conf_this_scope = new->root_scope;
	  sym = cf_get_symbol(oc->name);
	  proto_clone_config(sym, parsym->def);
	  new_config = NULL;
	  cfg_mem = NULL;
	}
      }

1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
      if (sym && sym->class == SYM_PROTO && !new->shutdown)
      {
	/* Found match, let's check if we can smoothly switch to new configuration */
	/* No need to check description */
	nc = sym->def;
	nc->proto = p;

	/* We will try to reconfigure protocol p */
	if (! force_reconfig && proto_reconfigure(p, oc, nc, type))
	  continue;

Ondřej Zajíček's avatar
Ondřej Zajíček committed
1027 1028 1029 1030 1031 1032
	if (nc->parent)
	{
	  proto_undef_clone(sym, nc);
	  goto remove;
	}

1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
	/* Unsuccessful, we will restart it */
	if (!p->disabled && !nc->disabled)
	  log(L_INFO "Restarting protocol %s", p->name);
	else if (p->disabled && !nc->disabled)
	  log(L_INFO "Enabling protocol %s", p->name);
	else if (!p->disabled && nc->disabled)
	  log(L_INFO "Disabling protocol %s", p->name);

	p->down_code = nc->disabled ? PDC_CF_DISABLE : PDC_CF_RESTART;
	p->cf_new = nc;
      }
      else if (!new->shutdown)
      {
Ondřej Zajíček's avatar
Ondřej Zajíček committed
1046
      remove:
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
	log(L_INFO "Removing protocol %s", p->name);
	p->down_code = PDC_CF_REMOVE;
	p->cf_new = NULL;
      }
      else /* global shutdown */
      {
	p->down_code = PDC_CMD_SHUTDOWN;
	p->cf_new = NULL;
      }

      p->reconfiguring = 1;
      config_add_obstacle(old);
      proto_rethink_goal(p);
Martin Mareš's avatar
Martin Mareš committed
1060
    }
1061 1062 1063
  }

  struct proto *first_dev_proto = NULL;
1064

1065
  n = NODE &(proto_list.head);
1066 1067
  WALK_LIST(nc, new->protos)
    if (!nc->proto)
1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
    {
      /* Not a first-time configuration */
      if (old)
	log(L_INFO "Adding protocol %s", nc->name);

      p = proto_init(nc, n);
      n = NODE p;

      if (p->proto == &proto_unix_iface)
	first_dev_proto = p;
    }
    else
      n = NODE nc->proto;
1081 1082

  DBG("Protocol start\n");
1083 1084

  /* Start device protocol first */
1085 1086
  if (first_dev_proto)
    proto_rethink_goal(first_dev_proto);
1087

1088 1089 1090
  /* Determine router ID for the first time - it has to be here and not in
     global_commit() because it is postponed after start of device protocol */
  if (!config->router_id)
1091 1092 1093 1094 1095
  {
    config->router_id = if_choose_router_id(config->router_id_from, 0);
    if (!config->router_id)
      die("Cannot determine router ID, please configure it manually");
  }
1096

1097 1098
  /* Start all new protocols */
  WALK_LIST_DELSAFE(p, n, proto_list)
1099
    proto_rethink_goal(p);
Martin Mareš's avatar
Martin Mareš committed
1100 1101
}

1102
static void
1103
proto_rethink_goal(struct proto *p)
1104
{
1105
  struct protocol *q;
1106
  byte goal;
1107

1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
  if (p->reconfiguring && !p->active)
  {
    struct proto_config *nc = p->cf_new;
    node *n = p->n.prev;
    DBG("%s has shut down for reconfiguration\n", p->name);
    p->cf->proto = NULL;
    config_del_obstacle(p->cf->global);
    proto_remove_channels(p);
    rem_node(&p->n);
    rfree(p->event);
1118
    mb_free(p->message);
1119 1120 1121 1122 1123
    mb_free(p);
    if (!nc)
      return;
    p = proto_init(nc, n);
  }
1124 1125

  /* Determine what state we want to reach */
1126
  if (p->disabled || p->reconfiguring)
1127
    goal = PS_DOWN;
1128
  else
1129
    goal = PS_UP;
1130 1131

  q = p->proto;
1132 1133 1134
  if (goal == PS_UP)
  {
    if (!p->active)
1135
    {
1136 1137 1138 1139 1140
      /* Going up */
      DBG("Kicking %s up\n", p->name);
      PD(p, "Starting");
      proto_start(p);
      proto_notify_state(p, (q->start ? q->start(p) : PS_UP));
1141
    }
1142 1143 1144 1145
  }
  else
  {
    if (p->proto_state == PS_START || p->proto_state == PS_UP)
1146
    {
1147 1148 1149 1150
      /* Going down */
      DBG("Kicking %s down\n", p->name);
      PD(p, "Shutting down");
      proto_notify_state(p, (q->shutdown ? q->shutdown(p) : PS_DOWN));
1151
    }
1152
  }
1153 1154
}

Ondřej Zajíček's avatar
Ondřej Zajíček committed
1155 1156 1157 1158 1159 1160 1161 1162 1163
struct proto *
proto_spawn(struct proto_config *cf, uint disabled)
{
  struct proto *p = proto_init(cf, TAIL(proto_list));
  p->disabled = disabled;
  proto_rethink_goal(p);
  return p;
}

1164

1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
/**
 * DOC: Graceful restart recovery
 *
 * Graceful restart of a router is a process when the routing plane (e.g. BIRD)
 * restarts but both the forwarding plane (e.g kernel routing table) and routing
 * neighbors keep proper routes, and therefore uninterrupted packet forwarding
 * is maintained.
 *
 * BIRD implements graceful restart recovery by deferring export of routes to
 * protocols until routing tables are refilled with the expected content. After
 * start, protocols generate routes as usual, but routes are not propagated to
 * them, until protocols report that they generated all routes. After that,
 * graceful restart recovery is finished and the export (and the initial feed)
 * to protocols is enabled.
 *
 * When graceful restart recovery need is detected during initialization, then
 * enabled protocols are marked with @gr_recovery flag before start. Such
 * protocols then decide how to proceed with graceful restart, participation is
1183
 * voluntary. Protocols could lock the recovery for each channel by function
Ondřej Zajíček's avatar
Ondřej Zajíček committed
1184
 * channel_graceful_restart_lock() (state stored in @gr_lock flag), which means
1185 1186 1187 1188 1189 1190
 * that they want to postpone the end of the recovery until they converge and
 * then unlock it. They also could set @gr_wait before advancing to %PS_UP,
 * which means that the core should defer route export to that channel until
 * the end of the recovery. This should be done by protocols that expect their
 * neigbors to keep the proper routes (kernel table, BGP sessions with BGP
 * graceful restart capability).
1191 1192 1193 1194 1195
 *
 * The graceful restart recovery is finished when either all graceful restart
 * locks are unlocked or when graceful restart wait timer fires.
 *
 */
1196

1197
static void graceful_restart_done(timer *t);
1198

1199 1200 1201 1202 1203 1204 1205
/**
 * graceful_restart_recovery - request initial graceful restart recovery
 *
 * Called by the platform initialization code if the need for recovery
 * after graceful restart is detected during boot. Have to be called
 * before protos_commit().
 */
1206 1207 1208 1209 1210 1211
void
graceful_restart_recovery(void)
{
  graceful_restart_state = GRS_INIT;
}

1212 1213 1214 1215 1216 1217 1218
/**
 * graceful_restart_init - initialize graceful restart
 *
 * When graceful restart recovery was requested, the function starts an active
 * phase of the recovery and initializes graceful restart wait timer. The
 * function have to be called after protos_commit().
 */
1219 1220 1221 1222 1223 1224 1225 1226 1227
void
graceful_restart_init(void)
{
  if (!graceful_restart_state)
    return;

  log(L_INFO "Graceful restart started");

  if (!graceful_restart_locks)
1228 1229 1230 1231
  {
    graceful_restart_done(NULL);
    return;
  }
1232 1233

  graceful_restart_state = GRS_ACTIVE;
1234 1235
  gr_wait_timer = tm_new_init(proto_pool, graceful_restart_done, NULL, 0, 0);
  tm_start(gr_wait_timer, config->gr_wait S);
1236 1237
}

1238 1239
/**
 * graceful_restart_done - finalize graceful restart
1240
 * @t: unused
1241 1242 1243 1244 1245 1246 1247
 *
 * When there are no locks on graceful restart, the functions finalizes the
 * graceful restart recovery. Protocols postponing route export until the end of
 * the recovery are awakened and the export to them is enabled. All other
 * related state is cleared. The function is also called when the graceful
 * restart wait timer fires (but there are still some locks).
 */
1248
static void
1249
graceful_restart_done(timer *t UNUSED)
1250 1251 1252 1253
{
  log(L_INFO "Graceful restart done");
  graceful_restart_state = GRS_DONE;

1254 1255 1256 1257 1258
  struct proto *p;
  WALK_LIST(p, proto_list)
  {
    if (!p->gr_recovery)
      continue;
1259

1260 1261 1262
    struct channel *c;
    WALK_LIST(c, p->channels)
    {
1263
      /* Resume postponed export of routes */
1264
      if ((c->channel_state == CS_UP) && c->gr_wait && c->proto->rt_notify)
1265
	channel_start_export(c);
1266 1267

      /* Cleanup */
1268 1269
      c->gr_wait = 0;
      c->gr_lock = 0;
1270 1271
    }

1272 1273 1274
    p->gr_recovery = 0;
  }

1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
  graceful_restart_locks = 0;
}

void
graceful_restart_show_status(void)
{
  if (graceful_restart_state != GRS_ACTIVE)
    return;

  cli_msg(-24, "Graceful restart recovery in progress");
1285
  cli_msg(-24, "  Waiting for %d channels to recover", graceful_restart_locks);
1286
  cli_msg(-24, "  Wait timer is %t/%u", tm_remains(gr_wait_timer), config->gr_wait);
1287 1288
}

1289
/**
1290 1291
 * channel_graceful_restart_lock - lock graceful restart by channel
 * @p: channel instance
1292 1293 1294
 *
 * This function allows a protocol to postpone the end of graceful restart
 * recovery until it converges. The lock is removed when the protocol calls
1295
 * channel_graceful_restart_unlock() or when the channel is closed.
1296 1297 1298 1299 1300 1301
 *
 * The function have to be called during the initial phase of graceful restart
 * recovery and only for protocols that are part of graceful restart (i.e. their
 * @gr_recovery is set), which means it should be called from protocol start
 * hooks.
 */
1302
void
1303
channel_graceful_restart_lock(struct channel *c)
1304 1305
{
  ASSERT(graceful_restart_state == GRS_INIT);
1306
  ASSERT(c->proto->gr_recovery);
1307

1308
  if (c->gr_lock)
1309 1310
    return;

1311
  c->gr_lock = 1;
1312 1313 1314
  graceful_restart_locks++;
}