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
|
// SPDX-License-Identifier: GPL-2.0-or-later
/*
* Addpath TX ID selection, and related utilities
* Copyright (C) 2018 Amazon.com, Inc. or its affiliates
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "bgp_addpath.h"
#include "bgp_route.h"
#include "bgp_open.h"
#include "bgp_packet.h"
static const struct bgp_addpath_strategy_names strat_names[BGP_ADDPATH_MAX] = {
{
.config_name = "addpath-tx-all-paths",
.human_name = "All",
.human_description = "Advertise all paths via addpath",
.type_json_name = "addpathTxAllPaths",
.id_json_name = "addpathTxIdAll"
},
{
.config_name = "addpath-tx-bestpath-per-AS",
.human_name = "Best-Per-AS",
.human_description = "Advertise bestpath per AS via addpath",
.type_json_name = "addpathTxBestpathPerAS",
.id_json_name = "addpathTxIdBestPerAS"
},
{
.config_name = "addpath-tx-best-selected",
.human_name = "Best-Selected",
.human_description = "Advertise best N selected paths via addpath",
.type_json_name = "addpathTxBestSelectedPaths",
.id_json_name = "addpathTxIdBestSelected"
},
};
static const struct bgp_addpath_strategy_names unknown_names = {
.config_name = "addpath-tx-unknown",
.human_name = "Unknown-Addpath-Strategy",
.human_description = "Unknown Addpath Strategy",
.type_json_name = "addpathTxUnknown",
.id_json_name = "addpathTxIdUnknown"
};
/*
* Returns a structure full of strings associated with an addpath type. Will
* never return null.
*/
const struct bgp_addpath_strategy_names *
bgp_addpath_names(enum bgp_addpath_strat strat)
{
if (strat < BGP_ADDPATH_MAX)
return &(strat_names[strat]);
else
return &unknown_names;
};
/*
* Returns if any peer is transmitting addpaths for a given afi/safi.
*/
bool bgp_addpath_is_addpath_used(struct bgp_addpath_bgp_data *d, afi_t afi,
safi_t safi)
{
return d->total_peercount[afi][safi] > 0;
}
/*
* Initialize the BGP instance level data for addpath.
*/
void bgp_addpath_init_bgp_data(struct bgp_addpath_bgp_data *d)
{
safi_t safi;
afi_t afi;
int i;
FOREACH_AFI_SAFI (afi, safi) {
for (i = 0; i < BGP_ADDPATH_MAX; i++) {
d->id_allocators[afi][safi][i] = NULL;
d->peercount[afi][safi][i] = 0;
}
d->total_peercount[afi][safi] = 0;
}
}
/*
* Free up resources associated with BGP route info structures.
*/
void bgp_addpath_free_info_data(struct bgp_addpath_info_data *d,
struct bgp_addpath_node_data *nd)
{
int i;
for (i = 0; i < BGP_ADDPATH_MAX; i++) {
if (d->addpath_tx_id[i] != IDALLOC_INVALID)
idalloc_free_to_pool(&nd->free_ids[i],
d->addpath_tx_id[i]);
}
}
/*
* Return the addpath ID used to send a particular route, to a particular peer,
* in a particular AFI/SAFI.
*/
uint32_t bgp_addpath_id_for_peer(struct peer *peer, afi_t afi, safi_t safi,
struct bgp_addpath_info_data *d)
{
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
if (peer->addpath_type[afi][safi] < BGP_ADDPATH_MAX)
return d->addpath_tx_id[peer->addpath_type[afi][safi]];
else
return IDALLOC_INVALID;
}
/*
* Returns true if the path has an assigned addpath ID for any of the addpath
* strategies.
*/
bool bgp_addpath_info_has_ids(struct bgp_addpath_info_data *d)
{
int i;
for (i = 0; i < BGP_ADDPATH_MAX; i++)
if (d->addpath_tx_id[i] != 0)
return true;
return false;
}
/*
* Releases any ID's associated with the BGP prefix.
*/
void bgp_addpath_free_node_data(struct bgp_addpath_bgp_data *bd,
struct bgp_addpath_node_data *nd, afi_t afi,
safi_t safi)
{
int i;
for (i = 0; i < BGP_ADDPATH_MAX; i++) {
idalloc_drain_pool(bd->id_allocators[afi][safi][i],
&(nd->free_ids[i]));
}
}
/*
* Check to see if the addpath strategy requires DMED to be configured to work.
*/
bool bgp_addpath_dmed_required(int strategy)
{
return strategy == BGP_ADDPATH_BEST_PER_AS;
}
/*
* Return true if this is a path we should advertise due to a
* configured addpath-tx knob
*/
bool bgp_addpath_tx_path(enum bgp_addpath_strat strat, struct bgp_path_info *pi)
{
switch (strat) {
case BGP_ADDPATH_NONE:
return false;
case BGP_ADDPATH_ALL:
return true;
case BGP_ADDPATH_BEST_PER_AS:
if (CHECK_FLAG(pi->flags, BGP_PATH_DMED_SELECTED))
return true;
else
return false;
case BGP_ADDPATH_BEST_SELECTED:
return true;
case BGP_ADDPATH_MAX:
return false;
}
assert(!"Reached end of function we should never hit");
}
static void bgp_addpath_flush_type_rn(struct bgp *bgp, afi_t afi, safi_t safi,
enum bgp_addpath_strat addpath_type,
struct bgp_dest *dest)
{
struct bgp_path_info *pi;
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
idalloc_drain_pool(
bgp->tx_addpath.id_allocators[afi][safi][addpath_type],
&(dest->tx_addpath.free_ids[addpath_type]));
for (pi = bgp_dest_get_bgp_path_info(dest); pi; pi = pi->next) {
if (pi->tx_addpath.addpath_tx_id[addpath_type]
!= IDALLOC_INVALID) {
idalloc_free(
bgp->tx_addpath
.id_allocators[afi][safi][addpath_type],
pi->tx_addpath.addpath_tx_id[addpath_type]);
pi->tx_addpath.addpath_tx_id[addpath_type] =
IDALLOC_INVALID;
}
}
}
/*
* Purge all addpath ID's on a BGP instance associated with the addpath
* strategy, and afi/safi combination. This lets us let go of all memory held to
* track ID numbers associated with an addpath type not in use. Since
* post-bestpath ID processing is skipped for types not used, this is the only
* chance to free this data.
*/
static void bgp_addpath_flush_type(struct bgp *bgp, afi_t afi, safi_t safi,
enum bgp_addpath_strat addpath_type)
{
struct bgp_dest *dest, *ndest;
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
for (dest = bgp_table_top(bgp->rib[afi][safi]); dest;
dest = bgp_route_next(dest)) {
if (safi == SAFI_MPLS_VPN) {
struct bgp_table *table;
table = bgp_dest_get_bgp_table_info(dest);
if (!table)
continue;
for (ndest = bgp_table_top(table); ndest;
ndest = bgp_route_next(ndest))
bgp_addpath_flush_type_rn(bgp, afi, safi,
addpath_type, ndest);
} else {
bgp_addpath_flush_type_rn(bgp, afi, safi, addpath_type,
dest);
}
}
idalloc_destroy(bgp->tx_addpath.id_allocators[afi][safi][addpath_type]);
bgp->tx_addpath.id_allocators[afi][safi][addpath_type] = NULL;
}
/*
* Allocate an Addpath ID for the given type on a path, if necessary.
*/
static void bgp_addpath_populate_path(struct id_alloc *allocator,
struct bgp_path_info *path,
enum bgp_addpath_strat addpath_type)
{
if (bgp_addpath_tx_path(addpath_type, path)) {
path->tx_addpath.addpath_tx_id[addpath_type] =
idalloc_allocate(allocator);
}
}
/*
* Compute addpath ID's on a BGP instance associated with the addpath strategy,
* and afi/safi combination. Since we won't waste the time computing addpath IDs
* for unused strategies, the first time a peer is configured to use a strategy,
* we have to backfill the data.
* In labeled-unicast, addpath allocations SHOULD be done in unicast SAFI.
*/
static void bgp_addpath_populate_type(struct bgp *bgp, afi_t afi, safi_t safi,
enum bgp_addpath_strat addpath_type)
{
struct bgp_dest *dest, *ndest;
char buf[200];
struct id_alloc *allocator;
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
snprintf(buf, sizeof(buf), "Addpath ID Allocator %s:%d/%d",
bgp_addpath_names(addpath_type)->config_name, (int)afi,
(int)safi);
buf[sizeof(buf) - 1] = '\0';
zlog_info("Computing addpath IDs for addpath type %s",
bgp_addpath_names(addpath_type)->human_name);
bgp->tx_addpath.id_allocators[afi][safi][addpath_type] =
idalloc_new(buf);
idalloc_reserve(bgp->tx_addpath.id_allocators[afi][safi][addpath_type],
BGP_ADDPATH_TX_ID_FOR_DEFAULT_ORIGINATE);
allocator = bgp->tx_addpath.id_allocators[afi][safi][addpath_type];
for (dest = bgp_table_top(bgp->rib[afi][safi]); dest;
dest = bgp_route_next(dest)) {
struct bgp_path_info *bi;
if (safi == SAFI_MPLS_VPN) {
struct bgp_table *table;
table = bgp_dest_get_bgp_table_info(dest);
if (!table)
continue;
for (ndest = bgp_table_top(table); ndest;
ndest = bgp_route_next(ndest))
for (bi = bgp_dest_get_bgp_path_info(ndest); bi;
bi = bi->next)
bgp_addpath_populate_path(allocator, bi,
addpath_type);
} else {
for (bi = bgp_dest_get_bgp_path_info(dest); bi;
bi = bi->next)
bgp_addpath_populate_path(allocator, bi,
addpath_type);
}
}
}
/*
* Handle updates to a peer or group's addpath strategy. If after adjusting
* counts a addpath strategy is in use for the first time, or no longer in use,
* the IDs for that strategy will be populated or flushed.
*/
void bgp_addpath_type_changed(struct bgp *bgp)
{
afi_t afi;
safi_t safi;
struct listnode *node, *nnode;
struct peer *peer;
int peer_count[AFI_MAX][SAFI_MAX][BGP_ADDPATH_MAX];
enum bgp_addpath_strat type;
FOREACH_AFI_SAFI(afi, safi) {
for (type=0; type<BGP_ADDPATH_MAX; type++) {
peer_count[afi][safi][type] = 0;
}
bgp->tx_addpath.total_peercount[afi][safi] = 0;
}
for (ALL_LIST_ELEMENTS(bgp->peer, node, nnode, peer)) {
FOREACH_AFI_SAFI(afi, safi) {
type = peer->addpath_type[afi][safi];
if (type != BGP_ADDPATH_NONE) {
peer_count[afi][safi][type] += 1;
bgp->tx_addpath.total_peercount[afi][safi] += 1;
}
}
}
FOREACH_AFI_SAFI(afi, safi) {
for (type=0; type<BGP_ADDPATH_MAX; type++) {
int old = bgp->tx_addpath.peercount[afi][safi][type];
int new = peer_count[afi][safi][type];
bgp->tx_addpath.peercount[afi][safi][type] = new;
if (old == 0 && new != 0) {
bgp_addpath_populate_type(bgp, afi, safi,
type);
} else if (old != 0 && new == 0) {
bgp_addpath_flush_type(bgp, afi, safi, type);
}
}
}
}
int bgp_addpath_capability_action(enum bgp_addpath_strat addpath_type,
uint8_t paths)
{
int action = CAPABILITY_ACTION_UNSET;
switch (addpath_type) {
case BGP_ADDPATH_ALL:
case BGP_ADDPATH_BEST_PER_AS:
action = CAPABILITY_ACTION_SET;
break;
case BGP_ADDPATH_BEST_SELECTED:
if (paths)
action = CAPABILITY_ACTION_SET;
else
action = CAPABILITY_ACTION_UNSET;
break;
case BGP_ADDPATH_NONE:
case BGP_ADDPATH_MAX:
action = CAPABILITY_ACTION_UNSET;
break;
}
return action;
}
/*
* Change the addpath type assigned to a peer, or peer group. In addition to
* adjusting the counts, peer sessions will be reset as needed to make the
* change take effect.
*/
void bgp_addpath_set_peer_type(struct peer *peer, afi_t afi, safi_t safi,
enum bgp_addpath_strat addpath_type,
uint8_t paths)
{
struct bgp *bgp = peer->bgp;
enum bgp_addpath_strat old_type;
struct listnode *node, *nnode;
struct peer *tmp_peer;
struct peer_group *group;
int action = bgp_addpath_capability_action(addpath_type, paths);
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
peer->addpath_best_selected[afi][safi] = paths;
old_type = peer->addpath_type[afi][safi];
if (addpath_type == old_type)
return;
if (addpath_type == BGP_ADDPATH_NONE && peer->group &&
!CHECK_FLAG(peer->sflags, PEER_STATUS_GROUP)) {
/* A "no" config on a group member inherits group */
addpath_type = peer->group->conf->addpath_type[afi][safi];
}
peer->addpath_type[afi][safi] = addpath_type;
bgp_addpath_type_changed(bgp);
if (addpath_type != BGP_ADDPATH_NONE) {
if (bgp_addpath_dmed_required(addpath_type)) {
if (!CHECK_FLAG(bgp->flags,
BGP_FLAG_DETERMINISTIC_MED)) {
zlog_warn(
"%s: enabling bgp deterministic-med, this is required for addpath-tx-bestpath-per-AS",
peer->host);
SET_FLAG(bgp->flags,
BGP_FLAG_DETERMINISTIC_MED);
bgp_recalculate_all_bestpaths(bgp);
}
}
}
zlog_info("Resetting peer %s%pBP due to change in addpath config",
CHECK_FLAG(peer->sflags, PEER_STATUS_GROUP) ? "group " : "",
peer);
if (CHECK_FLAG(peer->sflags, PEER_STATUS_GROUP)) {
group = peer->group;
/* group will be null as peer_group_delete calls peer_delete on
* group->conf. That peer_delete will eventuallly end up here
* if the group was configured to tx addpaths.
*/
if (group != NULL) {
for (ALL_LIST_ELEMENTS(group->peer, node, nnode,
tmp_peer)) {
if (tmp_peer->addpath_type[afi][safi] ==
old_type) {
bgp_addpath_set_peer_type(
tmp_peer, afi, safi,
addpath_type, paths);
}
}
}
} else {
if (!CHECK_FLAG(peer->cap, PEER_CAP_DYNAMIC_RCV) &&
!CHECK_FLAG(peer->cap, PEER_CAP_DYNAMIC_ADV))
peer_change_action(peer, afi, safi, peer_change_reset);
}
bgp_capability_send(peer, afi, safi, CAPABILITY_CODE_ADDPATH, action);
}
/*
* Intended to run after bestpath. This function will take TX IDs from paths
* that no longer need them, and give them to paths that do. This prevents
* best-per-as updates from needing to do a separate withdraw and update just to
* swap out which path is sent.
*/
void bgp_addpath_update_ids(struct bgp *bgp, struct bgp_dest *bn, afi_t afi,
safi_t safi)
{
int i;
struct bgp_path_info *pi;
struct id_alloc_pool **pool_ptr;
if (safi == SAFI_LABELED_UNICAST)
safi = SAFI_UNICAST;
for (i = 0; i < BGP_ADDPATH_MAX; i++) {
struct id_alloc *alloc =
bgp->tx_addpath.id_allocators[afi][safi][i];
pool_ptr = &(bn->tx_addpath.free_ids[i]);
if (bgp->tx_addpath.peercount[afi][safi][i] == 0)
continue;
/* Free Unused IDs back to the pool.*/
for (pi = bgp_dest_get_bgp_path_info(bn); pi; pi = pi->next) {
if (pi->tx_addpath.addpath_tx_id[i] != IDALLOC_INVALID
&& !bgp_addpath_tx_path(i, pi)) {
idalloc_free_to_pool(pool_ptr,
pi->tx_addpath.addpath_tx_id[i]);
pi->tx_addpath.addpath_tx_id[i] =
IDALLOC_INVALID;
}
}
/* Give IDs to paths that need them (pulling from the pool) */
for (pi = bgp_dest_get_bgp_path_info(bn); pi; pi = pi->next) {
if (pi->tx_addpath.addpath_tx_id[i] == IDALLOC_INVALID
&& bgp_addpath_tx_path(i, pi)) {
pi->tx_addpath.addpath_tx_id[i] =
idalloc_allocate_prefer_pool(
alloc, pool_ptr);
}
}
/* Free any IDs left in the pool to the main allocator */
idalloc_drain_pool(alloc, pool_ptr);
}
}
|