]> git.ipfire.org Git - thirdparty/bird.git/blob - proto/bgp/bgp.h
1270b8f302581618504e28f740b35012e32130e6
[thirdparty/bird.git] / proto / bgp / bgp.h
1 /*
2 * BIRD -- The Border Gateway Protocol
3 *
4 * (c) 2000 Martin Mares <mj@ucw.cz>
5 * (c) 2008--2016 Ondrej Zajicek <santiago@crfreenet.org>
6 * (c) 2008--2016 CZ.NIC z.s.p.o.
7 *
8 * Can be freely distributed and used under the terms of the GNU GPL.
9 */
10
11 #ifndef _BIRD_BGP_H_
12 #define _BIRD_BGP_H_
13
14 #include <stdint.h>
15 #include <setjmp.h>
16 #include "nest/bird.h"
17 #include "nest/route.h"
18 #include "nest/bfd.h"
19 //#include "lib/lists.h"
20 #include "lib/hash.h"
21 #include "lib/socket.h"
22
23 struct linpool;
24 struct eattr;
25
26
27 /* Address families */
28
29 #define BGP_AFI_IPV4 1
30 #define BGP_AFI_IPV6 2
31
32 #define BGP_SAFI_UNICAST 1
33 #define BGP_SAFI_MULTICAST 2
34 #define BGP_SAFI_MPLS 4
35 #define BGP_SAFI_MPLS_VPN 128
36 #define BGP_SAFI_VPN_MULTICAST 129
37 #define BGP_SAFI_FLOW 133
38
39 /* Internal AF codes */
40
41 #define BGP_AF(A, B) (((u32)(A) << 16) | (u32)(B))
42 #define BGP_AFI(A) ((u32)(A) >> 16)
43 #define BGP_SAFI(A) ((u32)(A) & 0xFFFF)
44
45 #define BGP_AF_IPV4 BGP_AF( BGP_AFI_IPV4, BGP_SAFI_UNICAST )
46 #define BGP_AF_IPV6 BGP_AF( BGP_AFI_IPV6, BGP_SAFI_UNICAST )
47 #define BGP_AF_IPV4_MC BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MULTICAST )
48 #define BGP_AF_IPV6_MC BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MULTICAST )
49 #define BGP_AF_IPV4_MPLS BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MPLS )
50 #define BGP_AF_IPV6_MPLS BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MPLS )
51 #define BGP_AF_VPN4_MPLS BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MPLS_VPN )
52 #define BGP_AF_VPN6_MPLS BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MPLS_VPN )
53 #define BGP_AF_VPN4_MC BGP_AF( BGP_AFI_IPV4, BGP_SAFI_VPN_MULTICAST )
54 #define BGP_AF_VPN6_MC BGP_AF( BGP_AFI_IPV6, BGP_SAFI_VPN_MULTICAST )
55 #define BGP_AF_FLOW4 BGP_AF( BGP_AFI_IPV4, BGP_SAFI_FLOW )
56 #define BGP_AF_FLOW6 BGP_AF( BGP_AFI_IPV6, BGP_SAFI_FLOW )
57
58
59 struct bgp_write_state;
60 struct bgp_parse_state;
61 struct bgp_export_state;
62 struct bgp_bucket;
63
64 struct bgp_af_desc {
65 u32 afi;
66 u32 net;
67 u8 mpls;
68 u8 no_igp;
69 const char *name;
70 uint (*encode_nlri)(struct bgp_write_state *s, struct bgp_bucket *buck, byte *buf, uint size);
71 void (*decode_nlri)(struct bgp_parse_state *s, byte *pos, uint len, rta *a);
72 void (*update_next_hop)(struct bgp_export_state *s, eattr *nh, ea_list **to);
73 uint (*encode_next_hop)(struct bgp_write_state *s, eattr *nh, byte *buf, uint size);
74 void (*decode_next_hop)(struct bgp_parse_state *s, byte *pos, uint len, rta *a);
75 };
76
77
78 struct bgp_config {
79 struct proto_config c;
80 u32 local_as, remote_as;
81 ip_addr local_ip; /* Source address to use */
82 ip_addr remote_ip;
83 struct iface *iface; /* Interface for link-local addresses */
84 u16 local_port; /* Local listening port */
85 u16 remote_port; /* Neighbor destination port */
86 int peer_type; /* Internal or external BGP (BGP_PT_*, optional) */
87 int multihop; /* Number of hops if multihop */
88 int strict_bind; /* Bind listening socket to local address */
89 int ttl_security; /* Enable TTL security [RFC 5082] */
90 int compare_path_lengths; /* Use path lengths when selecting best route */
91 int med_metric; /* Compare MULTI_EXIT_DISC even between routes from differen ASes */
92 int igp_metric; /* Use IGP metrics when selecting best route */
93 int prefer_older; /* Prefer older routes according to RFC 5004 */
94 int deterministic_med; /* Use more complicated algo to have strict RFC 4271 MED comparison */
95 u32 default_local_pref; /* Default value for LOCAL_PREF attribute */
96 u32 default_med; /* Default value for MULTI_EXIT_DISC attribute */
97 int capabilities; /* Enable capability handshake [RFC 5492] */
98 int enable_refresh; /* Enable local support for route refresh [RFC 2918] */
99 int enable_as4; /* Enable local support for 4B AS numbers [RFC 6793] */
100 int enable_extended_messages; /* Enable local support for extended messages [draft] */
101 u32 rr_cluster_id; /* Route reflector cluster ID, if different from local ID */
102 int rr_client; /* Whether neighbor is RR client of me */
103 int rs_client; /* Whether neighbor is RS client of me */
104 u32 confederation; /* Confederation ID, or zero if confeds not active */
105 int confederation_member; /* Whether neighbor AS is member of our confederation */
106 int passive; /* Do not initiate outgoing connection */
107 int interpret_communities; /* Hardwired handling of well-known communities */
108 int allow_local_as; /* Allow that number of local ASNs in incoming AS_PATHs */
109 int allow_local_pref; /* Allow LOCAL_PREF in EBGP sessions */
110 int allow_as_sets; /* Allow AS_SETs in incoming AS_PATHs */
111 int gr_mode; /* Graceful restart mode (BGP_GR_*) */
112 int llgr_mode; /* Long-lived graceful restart mode (BGP_LLGR_*) */
113 int setkey; /* Set MD5 password to system SA/SP database */
114 /* Times below are in seconds */
115 unsigned gr_time; /* Graceful restart timeout */
116 unsigned llgr_time; /* Long-lived graceful restart stale time */
117 unsigned connect_delay_time; /* Minimum delay between connect attempts */
118 unsigned connect_retry_time; /* Timeout for connect attempts */
119 unsigned hold_time, initial_hold_time;
120 unsigned keepalive_time;
121 unsigned error_amnesia_time; /* Errors are forgotten after */
122 unsigned error_delay_time_min; /* Time to wait after an error is detected */
123 unsigned error_delay_time_max;
124 unsigned disable_after_error; /* Disable the protocol when error is detected */
125 u32 disable_after_cease; /* Disable it when cease is received, bitfield */
126
127 char *password; /* Password used for MD5 authentication */
128 net_addr *remote_range; /* Allowed neighbor range for dynamic BGP */
129 char *dynamic_name; /* Name pattern for dynamic BGP */
130 int dynamic_name_digits; /* Minimum number of digits for dynamic names */
131 int check_link; /* Use iface link state for liveness detection */
132 int bfd; /* Use BFD for liveness detection */
133 };
134
135 struct bgp_channel_config {
136 struct channel_config c;
137
138 u32 afi;
139 const struct bgp_af_desc *desc;
140
141 ip_addr next_hop_addr; /* Local address for NEXT_HOP attribute */
142 u8 next_hop_self; /* Always set next hop to local IP address (NH_*) */
143 u8 next_hop_keep; /* Do not modify next hop attribute (NH_*) */
144 u8 mandatory; /* Channel is mandatory in capability negotiation */
145 u8 missing_lladdr; /* What we will do when we don' know link-local addr, see MLL_* */
146 u8 gw_mode; /* How we compute route gateway from next_hop attr, see GW_* */
147 u8 secondary; /* Accept also non-best routes (i.e. RA_ACCEPTED) */
148 u8 gr_able; /* Allow full graceful restart for the channel */
149 u8 llgr_able; /* Allow full long-lived GR for the channel */
150 uint llgr_time; /* Long-lived graceful restart stale time */
151 u8 ext_next_hop; /* Allow both IPv4 and IPv6 next hops */
152 u8 add_path; /* Use ADD-PATH extension [RFC 7911] */
153 u8 aigp; /* AIGP is allowed on this session */
154 u8 aigp_originate; /* AIGP is originated automatically */
155 u32 cost; /* IGP cost for direct next hops */
156 u8 import_table; /* Use c.in_table as Adj-RIB-In */
157 u8 export_table; /* Use c.out_table as Adj-RIB-Out */
158
159 struct rtable_config *igp_table_ip4; /* Table for recursive IPv4 next hop lookups */
160 struct rtable_config *igp_table_ip6; /* Table for recursive IPv6 next hop lookups */
161 };
162
163 #define BGP_PT_INTERNAL 1
164 #define BGP_PT_EXTERNAL 2
165
166 #define NH_NO 0
167 #define NH_ALL 1
168 #define NH_IBGP 2
169 #define NH_EBGP 3
170
171 #define MLL_SELF 1
172 #define MLL_DROP 2
173 #define MLL_IGNORE 3
174
175 #define GW_DIRECT 1
176 #define GW_RECURSIVE 2
177
178 #define BGP_ADD_PATH_RX 1
179 #define BGP_ADD_PATH_TX 2
180 #define BGP_ADD_PATH_FULL 3
181
182 #define BGP_GR_ABLE 1
183 #define BGP_GR_AWARE 2
184
185 /* For GR capability common flags */
186 #define BGP_GRF_RESTART 0x80
187
188 /* For GR capability per-AF flags */
189 #define BGP_GRF_FORWARDING 0x80
190
191 #define BGP_LLGR_ABLE 1
192 #define BGP_LLGR_AWARE 2
193
194 #define BGP_LLGRF_FORWARDING 0x80
195
196 #define BGP_GRS_NONE 0 /* No GR */
197 #define BGP_GRS_ACTIVE 1 /* Graceful restart per RFC 4724 */
198 #define BGP_GRS_LLGR 2 /* Long-lived GR phase (stale timer active) */
199
200 #define BGP_BFD_GRACEFUL 2 /* BFD down triggers graceful restart */
201
202
203 struct bgp_af_caps {
204 u32 afi;
205 u8 ready; /* Multiprotocol capability, RFC 4760 */
206 u8 gr_able; /* Graceful restart support, RFC 4724 */
207 u8 gr_af_flags; /* Graceful restart per-AF flags */
208 u8 llgr_able; /* Long-lived GR, RFC draft */
209 u32 llgr_time; /* Long-lived GR stale time */
210 u8 llgr_flags; /* Long-lived GR per-AF flags */
211 u8 ext_next_hop; /* Extended IPv6 next hop, RFC 5549 */
212 u8 add_path; /* Multiple paths support, RFC 7911 */
213 };
214
215 struct bgp_caps {
216 u32 as4_number; /* Announced ASN */
217
218 u8 as4_support; /* Four-octet AS capability, RFC 6793 */
219 u8 ext_messages; /* Extended message length, RFC draft */
220 u8 route_refresh; /* Route refresh capability, RFC 2918 */
221 u8 enhanced_refresh; /* Enhanced route refresh, RFC 7313 */
222
223 u8 gr_aware; /* Graceful restart capability, RFC 4724 */
224 u8 gr_flags; /* Graceful restart flags */
225 u16 gr_time; /* Graceful restart time in seconds */
226
227 u8 llgr_aware; /* Long-lived GR capability, RFC draft */
228 u8 any_ext_next_hop; /* Bitwise OR of per-AF ext_next_hop */
229 u8 any_add_path; /* Bitwise OR of per-AF add_path */
230
231 u16 af_count; /* Number of af_data items */
232 u16 length; /* Length of capabilities in OPEN msg */
233
234 struct bgp_af_caps af_data[0]; /* Per-AF capability data */
235 };
236
237 #define WALK_AF_CAPS(caps,ac) \
238 for (ac = caps->af_data; ac < &caps->af_data[caps->af_count]; ac++)
239
240
241 struct bgp_socket {
242 node n; /* Node in global bgp_sockets */
243 sock *sk; /* Real listening socket */
244 u32 uc; /* Use count */
245 };
246
247 struct bgp_conn {
248 struct bgp_proto *bgp;
249 struct birdsock *sk;
250 u8 state; /* State of connection state machine */
251 u8 as4_session; /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
252 u8 ext_messages; /* Session uses extended message length */
253 u32 received_as; /* ASN received in OPEN message */
254
255 struct bgp_caps *local_caps;
256 struct bgp_caps *remote_caps;
257 timer *connect_timer;
258 timer *hold_timer;
259 timer *keepalive_timer;
260 event *tx_ev;
261 u32 packets_to_send; /* Bitmap of packet types to be sent */
262 u32 channels_to_send; /* Bitmap of channels with packets to be sent */
263 u8 last_channel; /* Channel used last time for TX */
264 u8 last_channel_count; /* Number of times the last channel was used in succession */
265 int notify_code, notify_subcode, notify_size;
266 byte *notify_data;
267
268 uint hold_time, keepalive_time; /* Times calculated from my and neighbor's requirements */
269 };
270
271 struct bgp_proto {
272 struct proto p;
273 const struct bgp_config *cf; /* Shortcut to BGP configuration */
274 ip_addr local_ip, remote_ip;
275 u32 local_as, remote_as;
276 u32 public_as; /* Externally visible ASN (local_as or confederation id) */
277 u32 local_id; /* BGP identifier of this router */
278 u32 remote_id; /* BGP identifier of the neighbor */
279 u32 rr_cluster_id; /* Route reflector cluster ID */
280 u8 start_state; /* Substates that partitions BS_START */
281 u8 is_internal; /* Internal BGP session (local_as == remote_as) */
282 u8 is_interior; /* Internal or intra-confederation BGP session */
283 u8 as4_session; /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
284 u8 rr_client; /* Whether neighbor is RR client of me */
285 u8 rs_client; /* Whether neighbor is RS client of me */
286 u8 ipv4; /* Use IPv4 connection, i.e. remote_ip is IPv4 */
287 u8 passive; /* Do not initiate outgoing connection */
288 u8 route_refresh; /* Route refresh allowed to send [RFC 2918] */
289 u8 enhanced_refresh; /* Enhanced refresh is negotiated [RFC 7313] */
290 u8 gr_ready; /* Neighbor could do graceful restart */
291 u8 llgr_ready; /* Neighbor could do Long-lived GR, implies gr_ready */
292 u8 gr_active_num; /* Neighbor is doing GR, number of active channels */
293 u8 channel_count; /* Number of active channels */
294 u8 summary_add_path_rx; /* Summary state of ADD_PATH RX w.r.t active channels */
295 u32 *afi_map; /* Map channel index -> AFI */
296 struct bgp_channel **channel_map; /* Map channel index -> channel */
297 struct bgp_conn *conn; /* Connection we have established */
298 struct bgp_conn outgoing_conn; /* Outgoing connection we're working with */
299 struct bgp_conn incoming_conn; /* Incoming connection we have neither accepted nor rejected yet */
300 struct object_lock *lock; /* Lock for neighbor connection */
301 struct neighbor *neigh; /* Neighbor entry corresponding to remote ip, NULL if multihop */
302 struct bgp_socket *sock; /* Shared listening socket */
303 struct bfd_request *bfd_req; /* BFD request, if BFD is used */
304 struct birdsock *postponed_sk; /* Postponed incoming socket for dynamic BGP */
305 ip_addr link_addr; /* Link-local version of local_ip */
306 event *event; /* Event for respawning and shutting process */
307 timer *startup_timer; /* Timer used to delay protocol startup due to previous errors (startup_delay) */
308 timer *gr_timer; /* Timer waiting for reestablishment after graceful restart */
309 int dynamic_name_counter; /* Counter for dynamic BGP names */
310 uint startup_delay; /* Delay (in seconds) of protocol startup due to previous errors */
311 btime last_proto_error; /* Time of last error that leads to protocol stop */
312 u8 last_error_class; /* Error class of last error */
313 u32 last_error_code; /* Error code of last error. BGP protocol errors
314 are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
315 };
316
317 struct bgp_channel {
318 struct channel c;
319
320 /* Rest are BGP specific data */
321 struct bgp_channel_config *cf;
322
323 u32 afi;
324 u32 index;
325 const struct bgp_af_desc *desc;
326
327 rtable *igp_table_ip4; /* Table for recursive IPv4 next hop lookups */
328 rtable *igp_table_ip6; /* Table for recursive IPv6 next hop lookups */
329
330 /* Rest are zeroed when down */
331 pool *pool;
332 HASH(struct bgp_bucket) bucket_hash; /* Hash table of route buckets */
333 struct bgp_bucket *withdraw_bucket; /* Withdrawn routes */
334 list bucket_queue; /* Queue of buckets to send (struct bgp_bucket) */
335
336 HASH(struct bgp_prefix) prefix_hash; /* Prefixes to be sent */
337 slab *prefix_slab; /* Slab holding prefix nodes */
338
339 ip_addr next_hop_addr; /* Local address for NEXT_HOP attribute */
340 ip_addr link_addr; /* Link-local version of next_hop_addr */
341
342 u32 packets_to_send; /* Bitmap of packet types to be sent */
343
344 u8 ext_next_hop; /* Session allows both IPv4 and IPv6 next hops */
345
346 u8 gr_ready; /* Neighbor could do GR on this AF */
347 u8 gr_active; /* Neighbor is doing GR (BGP_GRS_*) */
348
349 timer *stale_timer; /* Long-lived stale timer for LLGR */
350 u32 stale_time; /* Stored LLGR stale time from last session */
351
352 u8 add_path_rx; /* Session expects receive of ADD-PATH extended NLRI */
353 u8 add_path_tx; /* Session expects transmit of ADD-PATH extended NLRI */
354
355 u8 feed_state; /* Feed state (TX) for EoR, RR packets, see BFS_* */
356 u8 load_state; /* Load state (RX) for EoR, RR packets, see BFS_* */
357 };
358
359 struct bgp_prefix {
360 node buck_node; /* Node in per-bucket list */
361 struct bgp_prefix *next; /* Node in prefix hash table */
362 u32 hash;
363 u32 path_id;
364 net_addr net[0];
365 };
366
367 struct bgp_bucket {
368 node send_node; /* Node in send queue */
369 struct bgp_bucket *next; /* Node in bucket hash table */
370 list prefixes; /* Prefixes in this bucket (struct bgp_prefix) */
371 u32 hash; /* Hash over extended attributes */
372 ea_list eattrs[0]; /* Per-bucket extended attributes */
373 };
374
375 struct bgp_export_state {
376 struct bgp_proto *proto;
377 struct bgp_channel *channel;
378 struct linpool *pool;
379
380 struct bgp_proto *src;
381 rte *route;
382 int mpls;
383
384 u32 attrs_seen[1];
385 uint err_withdraw;
386 uint local_next_hop;
387 };
388
389 struct bgp_write_state {
390 struct bgp_proto *proto;
391 struct bgp_channel *channel;
392 struct linpool *pool;
393
394 int mp_reach;
395 int as4_session;
396 int add_path;
397 int mpls;
398
399 eattr *mp_next_hop;
400 const adata *mpls_labels;
401 };
402
403 struct bgp_parse_state {
404 struct bgp_proto *proto;
405 struct bgp_channel *channel;
406 struct linpool *pool;
407
408 int as4_session;
409 int add_path;
410 int mpls;
411
412 u32 attrs_seen[256/32];
413
414 u32 mp_reach_af;
415 u32 mp_unreach_af;
416
417 uint attr_len;
418 uint ip_reach_len;
419 uint ip_unreach_len;
420 uint ip_next_hop_len;
421 uint mp_reach_len;
422 uint mp_unreach_len;
423 uint mp_next_hop_len;
424
425 byte *attrs;
426 byte *ip_reach_nlri;
427 byte *ip_unreach_nlri;
428 byte *ip_next_hop_data;
429 byte *mp_reach_nlri;
430 byte *mp_unreach_nlri;
431 byte *mp_next_hop_data;
432
433 uint err_withdraw;
434 uint err_subcode;
435 jmp_buf err_jmpbuf;
436
437 struct hostentry *hostentry;
438 adata *mpls_labels;
439
440 /* Cached state for bgp_rte_update() */
441 u32 last_id;
442 struct rte_src *last_src;
443 rta *cached_rta;
444 };
445
446 #define BGP_PORT 179
447 #define BGP_VERSION 4
448 #define BGP_HEADER_LENGTH 19
449 #define BGP_MAX_MESSAGE_LENGTH 4096
450 #define BGP_MAX_EXT_MSG_LENGTH 65535
451 #define BGP_RX_BUFFER_SIZE 4096
452 #define BGP_TX_BUFFER_SIZE 4096
453 #define BGP_RX_BUFFER_EXT_SIZE 65535
454 #define BGP_TX_BUFFER_EXT_SIZE 65535
455
456 static inline int bgp_channel_is_ipv4(struct bgp_channel *c)
457 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
458
459 static inline int bgp_channel_is_ipv6(struct bgp_channel *c)
460 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
461
462 static inline int bgp_cc_is_ipv4(struct bgp_channel_config *c)
463 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
464
465 static inline int bgp_cc_is_ipv6(struct bgp_channel_config *c)
466 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
467
468 static inline uint bgp_max_packet_length(struct bgp_conn *conn)
469 { return conn->ext_messages ? BGP_MAX_EXT_MSG_LENGTH : BGP_MAX_MESSAGE_LENGTH; }
470
471 static inline void
472 bgp_parse_error(struct bgp_parse_state *s, uint subcode)
473 {
474 s->err_subcode = subcode;
475 longjmp(s->err_jmpbuf, 1);
476 }
477
478 extern struct linpool *bgp_linpool;
479 extern struct linpool *bgp_linpool2;
480
481
482 void bgp_start_timer(timer *t, uint value);
483 void bgp_check_config(struct bgp_config *c);
484 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
485 void bgp_close_conn(struct bgp_conn *c);
486 void bgp_update_startup_delay(struct bgp_proto *p);
487 void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
488 void bgp_conn_enter_established_state(struct bgp_conn *conn);
489 void bgp_conn_enter_close_state(struct bgp_conn *conn);
490 void bgp_conn_enter_idle_state(struct bgp_conn *conn);
491 void bgp_handle_graceful_restart(struct bgp_proto *p);
492 void bgp_graceful_restart_done(struct bgp_channel *c);
493 void bgp_refresh_begin(struct bgp_channel *c);
494 void bgp_refresh_end(struct bgp_channel *c);
495 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
496 void bgp_stop(struct bgp_proto *p, int subcode, byte *data, uint len);
497
498 struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
499 struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
500
501 static inline int
502 rte_resolvable(rte *rt)
503 {
504 return rt->attrs->dest == RTD_UNICAST;
505 }
506
507
508 #ifdef LOCAL_DEBUG
509 #define BGP_FORCE_DEBUG 1
510 #else
511 #define BGP_FORCE_DEBUG 0
512 #endif
513 #define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
514 log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
515
516 #define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
517 log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
518
519
520 /* attrs.c */
521
522 static inline eattr *
523 bgp_find_attr(ea_list *attrs, uint code)
524 {
525 return ea_find(attrs, EA_CODE(PROTOCOL_BGP, code));
526 }
527
528 eattr *
529 bgp_set_attr(ea_list **attrs, struct linpool *pool, uint code, uint flags, uintptr_t val);
530
531 static inline void
532 bgp_set_attr_u32(ea_list **to, struct linpool *pool, uint code, uint flags, u32 val)
533 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
534
535 static inline void
536 bgp_set_attr_ptr(ea_list **to, struct linpool *pool, uint code, uint flags, const struct adata *val)
537 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
538
539 static inline void
540 bgp_set_attr_data(ea_list **to, struct linpool *pool, uint code, uint flags, void *data, uint len)
541 {
542 struct adata *a = lp_alloc_adata(pool, len);
543 memcpy(a->data, data, len);
544 bgp_set_attr(to, pool, code, flags, (uintptr_t) a);
545 }
546
547 static inline void
548 bgp_unset_attr(ea_list **to, struct linpool *pool, uint code)
549 { eattr *e = bgp_set_attr(to, pool, code, 0, 0); e->type = EAF_TYPE_UNDEF; }
550
551
552 int bgp_encode_attrs(struct bgp_write_state *s, ea_list *attrs, byte *buf, byte *end);
553 ea_list * bgp_decode_attrs(struct bgp_parse_state *s, byte *data, uint len);
554 void bgp_finish_attrs(struct bgp_parse_state *s, rta *a);
555
556 void bgp_init_bucket_table(struct bgp_channel *c);
557 void bgp_free_bucket_table(struct bgp_channel *c);
558 void bgp_free_bucket(struct bgp_channel *c, struct bgp_bucket *b);
559 void bgp_defer_bucket(struct bgp_channel *c, struct bgp_bucket *b);
560 void bgp_withdraw_bucket(struct bgp_channel *c, struct bgp_bucket *b);
561
562 void bgp_init_prefix_table(struct bgp_channel *c);
563 void bgp_free_prefix_table(struct bgp_channel *c);
564 void bgp_free_prefix(struct bgp_channel *c, struct bgp_prefix *bp);
565
566 int bgp_rte_better(struct rte *, struct rte *);
567 int bgp_rte_mergable(rte *pri, rte *sec);
568 int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
569 struct rte *bgp_rte_modify_stale(struct rte *r, struct linpool *pool);
570 void bgp_rt_notify(struct proto *P, struct channel *C, net *n, rte *new, rte *old);
571 int bgp_preexport(struct proto *, struct rte **, struct linpool *);
572 int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
573 void bgp_get_route_info(struct rte *, byte *buf);
574 int bgp_total_aigp_metric_(rte *e, u64 *metric, const struct adata **ad);
575
576 #define BGP_AIGP_METRIC 1
577 #define BGP_AIGP_MAX U64(0xffffffffffffffff)
578
579 static inline u64
580 bgp_total_aigp_metric(rte *r)
581 {
582 u64 metric = BGP_AIGP_MAX;
583 const struct adata *ad;
584
585 bgp_total_aigp_metric_(r, &metric, &ad);
586 return metric;
587 }
588
589
590 /* packets.c */
591
592 void bgp_dump_state_change(struct bgp_conn *conn, uint old, uint new);
593 void bgp_prepare_capabilities(struct bgp_conn *conn);
594 const struct bgp_af_desc *bgp_get_af_desc(u32 afi);
595 const struct bgp_af_caps *bgp_find_af_caps(struct bgp_caps *caps, u32 afi);
596 void bgp_schedule_packet(struct bgp_conn *conn, struct bgp_channel *c, int type);
597 void bgp_kick_tx(void *vconn);
598 void bgp_tx(struct birdsock *sk);
599 int bgp_rx(struct birdsock *sk, uint size);
600 const char * bgp_error_dsc(unsigned code, unsigned subcode);
601 void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
602
603 void bgp_update_next_hop(struct bgp_export_state *s, eattr *a, ea_list **to);
604
605
606 /* Packet types */
607
608 #define PKT_OPEN 0x01
609 #define PKT_UPDATE 0x02
610 #define PKT_NOTIFICATION 0x03
611 #define PKT_KEEPALIVE 0x04
612 #define PKT_ROUTE_REFRESH 0x05 /* [RFC2918] */
613 #define PKT_BEGIN_REFRESH 0x1e /* Dummy type for BoRR packet [RFC7313] */
614 #define PKT_SCHEDULE_CLOSE 0x1f /* Used internally to schedule socket close */
615
616 /* Attributes */
617
618 #define BAF_OPTIONAL 0x80
619 #define BAF_TRANSITIVE 0x40
620 #define BAF_PARTIAL 0x20
621 #define BAF_EXT_LEN 0x10
622
623 #define BAF_DECODE_FLAGS 0x0100 /* Private flag - attribute flags are handled by the decode hook */
624
625 #define BA_ORIGIN 0x01 /* RFC 4271 */ /* WM */
626 #define BA_AS_PATH 0x02 /* WM */
627 #define BA_NEXT_HOP 0x03 /* WM */
628 #define BA_MULTI_EXIT_DISC 0x04 /* ON */
629 #define BA_LOCAL_PREF 0x05 /* WD */
630 #define BA_ATOMIC_AGGR 0x06 /* WD */
631 #define BA_AGGREGATOR 0x07 /* OT */
632 #define BA_COMMUNITY 0x08 /* RFC 1997 */ /* OT */
633 #define BA_ORIGINATOR_ID 0x09 /* RFC 4456 */ /* ON */
634 #define BA_CLUSTER_LIST 0x0a /* RFC 4456 */ /* ON */
635 #define BA_MP_REACH_NLRI 0x0e /* RFC 4760 */
636 #define BA_MP_UNREACH_NLRI 0x0f /* RFC 4760 */
637 #define BA_EXT_COMMUNITY 0x10 /* RFC 4360 */
638 #define BA_AS4_PATH 0x11 /* RFC 6793 */
639 #define BA_AS4_AGGREGATOR 0x12 /* RFC 6793 */
640 #define BA_AIGP 0x1a /* RFC 7311 */
641 #define BA_LARGE_COMMUNITY 0x20 /* RFC 8092 */
642
643 /* Bird's private internal BGP attributes */
644 #define BA_MPLS_LABEL_STACK 0xfe /* MPLS label stack transfer attribute */
645
646 /* BGP connection states */
647
648 #define BS_IDLE 0
649 #define BS_CONNECT 1 /* Attempting to connect */
650 #define BS_ACTIVE 2 /* Waiting for connection retry & listening */
651 #define BS_OPENSENT 3
652 #define BS_OPENCONFIRM 4
653 #define BS_ESTABLISHED 5
654 #define BS_CLOSE 6 /* Used during transition to BS_IDLE */
655
656 #define BS_MAX 7
657
658 /* BGP start states
659 *
660 * Used in PS_START for fine-grained specification of starting state.
661 *
662 * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP
663 * protocol done what is neccessary to start itself (like acquiring the lock),
664 * it goes to BSS_CONNECT.
665 */
666
667 #define BSS_PREPARE 0 /* Used before ordinary BGP started, i. e. waiting for lock */
668 #define BSS_DELAY 1 /* Startup delay due to previous errors */
669 #define BSS_CONNECT 2 /* Ordinary BGP connecting */
670
671
672 /* BGP feed states (TX)
673 *
674 * RFC 4724 specifies that an initial feed should end with End-of-RIB mark.
675 *
676 * RFC 7313 specifies that a route refresh should be demarcated by BoRR and EoRR packets.
677 *
678 * These states (stored in c->feed_state) are used to keep track of these
679 * requirements. When such feed is started, BFS_LOADING / BFS_REFRESHING is
680 * set. When it ended, BFS_LOADED / BFS_REFRESHED is set to schedule End-of-RIB
681 * or EoRR packet. When the packet is sent, the state returned to BFS_NONE.
682 *
683 * Note that when a non-demarcated feed (e.g. plain RFC 4271 initial load
684 * without End-of-RIB or plain RFC 2918 route refresh without BoRR/EoRR
685 * demarcation) is active, BFS_NONE is set.
686 *
687 * BFS_NONE, BFS_LOADING and BFS_REFRESHING are also used as load states (RX)
688 * with correspondent semantics (-, expecting End-of-RIB, expecting EoRR).
689 */
690
691 #define BFS_NONE 0 /* No feed or original non-demarcated feed */
692 #define BFS_LOADING 1 /* Initial feed active, End-of-RIB planned */
693 #define BFS_LOADED 2 /* Loading done, End-of-RIB marker scheduled */
694 #define BFS_REFRESHING 3 /* Route refresh (introduced by BoRR) active */
695 #define BFS_REFRESHED 4 /* Refresh done, EoRR packet scheduled */
696
697
698 /* Error classes */
699
700 #define BE_NONE 0
701 #define BE_MISC 1 /* Miscellaneous error */
702 #define BE_SOCKET 2 /* Socket error */
703 #define BE_BGP_RX 3 /* BGP protocol error notification received */
704 #define BE_BGP_TX 4 /* BGP protocol error notification sent */
705 #define BE_AUTO_DOWN 5 /* Automatic shutdown */
706 #define BE_MAN_DOWN 6 /* Manual shutdown */
707
708 /* Misc error codes */
709
710 #define BEM_NEIGHBOR_LOST 1
711 #define BEM_INVALID_NEXT_HOP 2
712 #define BEM_INVALID_MD5 3 /* MD5 authentication kernel request failed (possibly not supported) */
713 #define BEM_NO_SOCKET 4
714 #define BEM_LINK_DOWN 5
715 #define BEM_BFD_DOWN 6
716 #define BEM_GRACEFUL_RESTART 7
717
718 /* Automatic shutdown error codes */
719
720 #define BEA_ROUTE_LIMIT_EXCEEDED 1
721
722 /* Well-known communities */
723
724 #define BGP_COMM_NO_EXPORT 0xffffff01 /* Don't export outside local AS / confed. */
725 #define BGP_COMM_NO_ADVERTISE 0xffffff02 /* Don't export at all */
726 #define BGP_COMM_NO_EXPORT_SUBCONFED 0xffffff03 /* NO_EXPORT even in local confederation */
727
728 #define BGP_COMM_LLGR_STALE 0xffff0006 /* Route is stale according to LLGR */
729 #define BGP_COMM_NO_LLGR 0xffff0007 /* Do not treat the route according to LLGR */
730
731 /* Origins */
732
733 #define ORIGIN_IGP 0
734 #define ORIGIN_EGP 1
735 #define ORIGIN_INCOMPLETE 2
736
737
738 #endif