]> git.ipfire.org Git - thirdparty/bird.git/blob - proto/bgp/bgp.h
Merge branch 'bgp-grace'
[thirdparty/bird.git] / proto / bgp / bgp.h
1 /*
2 * BIRD -- The Border Gateway Protocol
3 *
4 * (c) 2000 Martin Mares <mj@ucw.cz>
5 *
6 * Can be freely distributed and used under the terms of the GNU GPL.
7 */
8
9 #ifndef _BIRD_BGP_H_
10 #define _BIRD_BGP_H_
11
12 #include <stdint.h>
13 #include "nest/route.h"
14 #include "nest/bfd.h"
15 #include "lib/hash.h"
16
17 struct linpool;
18 struct eattr;
19
20 struct bgp_config {
21 struct proto_config c;
22 u32 local_as, remote_as;
23 ip_addr remote_ip;
24 ip_addr source_addr; /* Source address to use */
25 struct iface *iface; /* Interface for link-local addresses */
26 int multihop; /* Number of hops if multihop */
27 int ttl_security; /* Enable TTL security [RFC5082] */
28 int next_hop_self; /* Always set next hop to local IP address */
29 int next_hop_keep; /* Do not touch next hop attribute */
30 int missing_lladdr; /* What we will do when we don' know link-local addr, see MLL_* */
31 int gw_mode; /* How we compute route gateway from next_hop attr, see GW_* */
32 int compare_path_lengths; /* Use path lengths when selecting best route */
33 int med_metric; /* Compare MULTI_EXIT_DISC even between routes from differen ASes */
34 int igp_metric; /* Use IGP metrics when selecting best route */
35 int prefer_older; /* Prefer older routes according to RFC 5004 */
36 int deterministic_med; /* Use more complicated algo to have strict RFC 4271 MED comparison */
37 u32 default_local_pref; /* Default value for LOCAL_PREF attribute */
38 u32 default_med; /* Default value for MULTI_EXIT_DISC attribute */
39 int capabilities; /* Enable capability handshake [RFC3392] */
40 int enable_refresh; /* Enable local support for route refresh [RFC2918] */
41 int enable_as4; /* Enable local support for 4B AS numbers [RFC4893] */
42 u32 rr_cluster_id; /* Route reflector cluster ID, if different from local ID */
43 int rr_client; /* Whether neighbor is RR client of me */
44 int rs_client; /* Whether neighbor is RS client of me */
45 int advertise_ipv4; /* Whether we should add IPv4 capability advertisement to OPEN message */
46 int passive; /* Do not initiate outgoing connection */
47 int interpret_communities; /* Hardwired handling of well-known communities */
48 int secondary; /* Accept also non-best routes (i.e. RA_ACCEPTED) */
49 int add_path; /* Use ADD-PATH extension [draft] */
50 int allow_local_as; /* Allow that number of local ASNs in incoming AS_PATHs */
51 int gr_mode; /* Graceful restart mode (BGP_GR_*) */
52 unsigned gr_time; /* Graceful restart timeout */
53 unsigned connect_retry_time;
54 unsigned hold_time, initial_hold_time;
55 unsigned keepalive_time;
56 unsigned start_delay_time; /* Minimum delay between connects */
57 unsigned error_amnesia_time; /* Errors are forgotten after */
58 unsigned error_delay_time_min; /* Time to wait after an error is detected */
59 unsigned error_delay_time_max;
60 unsigned disable_after_error; /* Disable the protocol when error is detected */
61
62 char *password; /* Password used for MD5 authentication */
63 struct rtable_config *igp_table; /* Table used for recursive next hop lookups */
64 int bfd; /* Use BFD for liveness detection */
65 };
66
67 #define MLL_SELF 1
68 #define MLL_DROP 2
69 #define MLL_IGNORE 3
70
71 #define GW_DIRECT 1
72 #define GW_RECURSIVE 2
73
74 #define ADD_PATH_RX 1
75 #define ADD_PATH_TX 2
76 #define ADD_PATH_FULL 3
77
78 #define BGP_GR_ABLE 1
79 #define BGP_GR_AWARE 2
80
81 /* For peer_gr_flags */
82 #define BGP_GRF_RESTART 0x80
83
84 /* For peer_gr_aflags */
85 #define BGP_GRF_FORWARDING 0x80
86
87
88 struct bgp_conn {
89 struct bgp_proto *bgp;
90 struct birdsock *sk;
91 unsigned int state; /* State of connection state machine */
92 struct timer *connect_retry_timer;
93 struct timer *hold_timer;
94 struct timer *keepalive_timer;
95 struct event *tx_ev;
96 int packets_to_send; /* Bitmap of packet types to be sent */
97 int notify_code, notify_subcode, notify_size;
98 byte *notify_data;
99 u32 advertised_as; /* Temporary value for AS number received */
100 int start_state; /* protocol start_state snapshot when connection established */
101 u8 peer_refresh_support; /* Peer supports route refresh [RFC2918] */
102 u8 peer_as4_support; /* Peer supports 4B AS numbers [RFC4893] */
103 u8 peer_add_path; /* Peer supports ADD-PATH [draft] */
104 u8 peer_gr_aware;
105 u8 peer_gr_able;
106 u16 peer_gr_time;
107 u8 peer_gr_flags;
108 u8 peer_gr_aflags;
109 unsigned hold_time, keepalive_time; /* Times calculated from my and neighbor's requirements */
110 };
111
112 struct bgp_proto {
113 struct proto p;
114 struct bgp_config *cf; /* Shortcut to BGP configuration */
115 u32 local_as, remote_as;
116 int start_state; /* Substates that partitions BS_START */
117 u8 is_internal; /* Internal BGP connection (local_as == remote_as) */
118 u8 as4_session; /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
119 u8 add_path_rx; /* Session expects receive of ADD-PATH extended NLRI */
120 u8 add_path_tx; /* Session expects transmit of ADD-PATH extended NLRI */
121 u32 local_id; /* BGP identifier of this router */
122 u32 remote_id; /* BGP identifier of the neighbor */
123 u32 rr_cluster_id; /* Route reflector cluster ID */
124 int rr_client; /* Whether neighbor is RR client of me */
125 int rs_client; /* Whether neighbor is RS client of me */
126 u8 gr_ready; /* Neighbor could do graceful restart */
127 u8 gr_active; /* Neighbor is doing graceful restart */
128 struct bgp_conn *conn; /* Connection we have established */
129 struct bgp_conn outgoing_conn; /* Outgoing connection we're working with */
130 struct bgp_conn incoming_conn; /* Incoming connection we have neither accepted nor rejected yet */
131 struct object_lock *lock; /* Lock for neighbor connection */
132 struct neighbor *neigh; /* Neighbor entry corresponding to remote ip, NULL if multihop */
133 struct bfd_request *bfd_req; /* BFD request, if BFD is used */
134 ip_addr source_addr; /* Local address used as an advertised next hop */
135 rtable *igp_table; /* Table used for recursive next hop lookups */
136 struct event *event; /* Event for respawning and shutting process */
137 struct timer *startup_timer; /* Timer used to delay protocol startup due to previous errors (startup_delay) */
138 struct timer *gr_timer; /* Timer waiting for reestablishment after graceful restart */
139 struct bgp_bucket **bucket_hash; /* Hash table of attribute buckets */
140 unsigned int hash_size, hash_count, hash_limit;
141 HASH(struct bgp_prefix) prefix_hash; /* Prefixes to be sent */
142 slab *prefix_slab; /* Slab holding prefix nodes */
143 list bucket_queue; /* Queue of buckets to send */
144 struct bgp_bucket *withdraw_bucket; /* Withdrawn routes */
145 unsigned send_end_mark; /* End-of-RIB mark scheduled for transmit */
146 unsigned startup_delay; /* Time to delay protocol startup by due to errors */
147 bird_clock_t last_proto_error; /* Time of last error that leads to protocol stop */
148 u8 last_error_class; /* Error class of last error */
149 u32 last_error_code; /* Error code of last error. BGP protocol errors
150 are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
151 #ifdef IPV6
152 byte *mp_reach_start, *mp_unreach_start; /* Multiprotocol BGP attribute notes */
153 unsigned mp_reach_len, mp_unreach_len;
154 ip_addr local_link; /* Link-level version of source_addr */
155 #endif
156 };
157
158 struct bgp_prefix {
159 struct {
160 ip_addr prefix;
161 int pxlen;
162 } n;
163 u32 path_id;
164 struct bgp_prefix *next;
165 node bucket_node; /* Node in per-bucket list */
166 };
167
168 struct bgp_bucket {
169 node send_node; /* Node in send queue */
170 struct bgp_bucket *hash_next, *hash_prev; /* Node in bucket hash table */
171 unsigned hash; /* Hash over extended attributes */
172 list prefixes; /* Prefixes in this buckets */
173 ea_list eattrs[0]; /* Per-bucket extended attributes */
174 };
175
176 #define BGP_PORT 179
177 #define BGP_VERSION 4
178 #define BGP_HEADER_LENGTH 19
179 #define BGP_MAX_PACKET_LENGTH 4096
180 #define BGP_RX_BUFFER_SIZE 4096
181 #define BGP_TX_BUFFER_SIZE BGP_MAX_PACKET_LENGTH
182
183 extern struct linpool *bgp_linpool;
184
185
186 void bgp_start_timer(struct timer *t, int value);
187 void bgp_check_config(struct bgp_config *c);
188 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
189 void bgp_close_conn(struct bgp_conn *c);
190 void bgp_update_startup_delay(struct bgp_proto *p);
191 void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
192 void bgp_conn_enter_established_state(struct bgp_conn *conn);
193 void bgp_conn_enter_close_state(struct bgp_conn *conn);
194 void bgp_conn_enter_idle_state(struct bgp_conn *conn);
195 void bgp_handle_graceful_restart(struct bgp_proto *p);
196 void bgp_graceful_restart_done(struct bgp_proto *p);
197 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
198 void bgp_stop(struct bgp_proto *p, unsigned subcode);
199
200 struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
201 struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
202
203
204
205 #ifdef LOCAL_DEBUG
206 #define BGP_FORCE_DEBUG 1
207 #else
208 #define BGP_FORCE_DEBUG 0
209 #endif
210 #define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
211 log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
212
213 #define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
214 log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
215
216
217 /* attrs.c */
218
219 /* Hack: although BA_NEXT_HOP attribute has type EAF_TYPE_IP_ADDRESS, in IPv6
220 * we store two addesses in it - a global address and a link local address.
221 */
222 #ifdef IPV6
223 #define NEXT_HOP_LENGTH (2*sizeof(ip_addr))
224 static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; ((ip_addr *) b)[1] = IPA_NONE; }
225 #else
226 #define NEXT_HOP_LENGTH sizeof(ip_addr)
227 static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; }
228 #endif
229
230 void bgp_attach_attr(struct ea_list **to, struct linpool *pool, unsigned attr, uintptr_t val);
231 byte *bgp_attach_attr_wa(struct ea_list **to, struct linpool *pool, unsigned attr, unsigned len);
232 struct rta *bgp_decode_attrs(struct bgp_conn *conn, byte *a, unsigned int len, struct linpool *pool, int mandatory);
233 int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
234 int bgp_rte_better(struct rte *, struct rte *);
235 int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
236 void bgp_rt_notify(struct proto *P, rtable *tbl UNUSED, net *n, rte *new, rte *old UNUSED, ea_list *attrs);
237 int bgp_import_control(struct proto *, struct rte **, struct ea_list **, struct linpool *);
238 void bgp_init_bucket_table(struct bgp_proto *);
239 void bgp_free_bucket(struct bgp_proto *p, struct bgp_bucket *buck);
240 void bgp_init_prefix_table(struct bgp_proto *p, u32 order);
241 void bgp_free_prefix(struct bgp_proto *p, struct bgp_prefix *bp);
242 unsigned int bgp_encode_attrs(struct bgp_proto *p, byte *w, ea_list *attrs, int remains);
243 void bgp_get_route_info(struct rte *, byte *buf, struct ea_list *attrs);
244
245 inline static void bgp_attach_attr_ip(struct ea_list **to, struct linpool *pool, unsigned attr, ip_addr a)
246 { *(ip_addr *) bgp_attach_attr_wa(to, pool, attr, sizeof(ip_addr)) = a; }
247
248 /* packets.c */
249
250 void mrt_dump_bgp_state_change(struct bgp_conn *conn, unsigned old, unsigned new);
251 void bgp_schedule_packet(struct bgp_conn *conn, int type);
252 void bgp_kick_tx(void *vconn);
253 void bgp_tx(struct birdsock *sk);
254 int bgp_rx(struct birdsock *sk, int size);
255 const char * bgp_error_dsc(unsigned code, unsigned subcode);
256 void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
257
258 /* Packet types */
259
260 #define PKT_OPEN 0x01
261 #define PKT_UPDATE 0x02
262 #define PKT_NOTIFICATION 0x03
263 #define PKT_KEEPALIVE 0x04
264 #define PKT_ROUTE_REFRESH 0x05
265 #define PKT_SCHEDULE_CLOSE 0x1f /* Used internally to schedule socket close */
266
267 /* Attributes */
268
269 #define BAF_OPTIONAL 0x80
270 #define BAF_TRANSITIVE 0x40
271 #define BAF_PARTIAL 0x20
272 #define BAF_EXT_LEN 0x10
273
274 #define BA_ORIGIN 0x01 /* [RFC1771] */ /* WM */
275 #define BA_AS_PATH 0x02 /* WM */
276 #define BA_NEXT_HOP 0x03 /* WM */
277 #define BA_MULTI_EXIT_DISC 0x04 /* ON */
278 #define BA_LOCAL_PREF 0x05 /* WD */
279 #define BA_ATOMIC_AGGR 0x06 /* WD */
280 #define BA_AGGREGATOR 0x07 /* OT */
281 #define BA_COMMUNITY 0x08 /* [RFC1997] */ /* OT */
282 #define BA_ORIGINATOR_ID 0x09 /* [RFC1966] */ /* ON */
283 #define BA_CLUSTER_LIST 0x0a /* ON */
284 /* We don't support these: */
285 #define BA_DPA 0x0b /* ??? */
286 #define BA_ADVERTISER 0x0c /* [RFC1863] */
287 #define BA_RCID_PATH 0x0d
288 #define BA_MP_REACH_NLRI 0x0e /* [RFC2283] */
289 #define BA_MP_UNREACH_NLRI 0x0f
290 #define BA_EXT_COMMUNITY 0x10 /* [RFC4360] */
291 #define BA_AS4_PATH 0x11 /* [RFC4893] */
292 #define BA_AS4_AGGREGATOR 0x12
293
294 /* BGP connection states */
295
296 #define BS_IDLE 0
297 #define BS_CONNECT 1 /* Attempting to connect */
298 #define BS_ACTIVE 2 /* Waiting for connection retry & listening */
299 #define BS_OPENSENT 3
300 #define BS_OPENCONFIRM 4
301 #define BS_ESTABLISHED 5
302 #define BS_CLOSE 6 /* Used during transition to BS_IDLE */
303
304 #define BS_MAX 7
305
306 /* BGP start states
307 *
308 * Used in PS_START for fine-grained specification of starting state.
309 *
310 * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP protocol
311 * done what is neccessary to start itself (like acquiring the lock), it goes to BSS_CONNECT.
312 * When some connection attempt failed because of option or capability error, it goes to
313 * BSS_CONNECT_NOCAP.
314 */
315
316 #define BSS_PREPARE 0 /* Used before ordinary BGP started, i. e. waiting for lock */
317 #define BSS_DELAY 1 /* Startup delay due to previous errors */
318 #define BSS_CONNECT 2 /* Ordinary BGP connecting */
319 #define BSS_CONNECT_NOCAP 3 /* Legacy BGP connecting (without capabilities) */
320
321 /* Error classes */
322
323 #define BE_NONE 0
324 #define BE_MISC 1 /* Miscellaneous error */
325 #define BE_SOCKET 2 /* Socket error */
326 #define BE_BGP_RX 3 /* BGP protocol error notification received */
327 #define BE_BGP_TX 4 /* BGP protocol error notification sent */
328 #define BE_AUTO_DOWN 5 /* Automatic shutdown */
329 #define BE_MAN_DOWN 6 /* Manual shutdown */
330
331 /* Misc error codes */
332
333 #define BEM_NEIGHBOR_LOST 1
334 #define BEM_INVALID_NEXT_HOP 2
335 #define BEM_INVALID_MD5 3 /* MD5 authentication kernel request failed (possibly not supported) */
336 #define BEM_NO_SOCKET 4
337 #define BEM_BFD_DOWN 5
338 #define BEM_GRACEFUL_RESTART 6
339
340 /* Automatic shutdown error codes */
341
342 #define BEA_ROUTE_LIMIT_EXCEEDED 1
343
344 /* Well-known communities */
345
346 #define BGP_COMM_NO_EXPORT 0xffffff01 /* Don't export outside local AS / confed. */
347 #define BGP_COMM_NO_ADVERTISE 0xffffff02 /* Don't export at all */
348 #define BGP_COMM_NO_EXPORT_SUBCONFED 0xffffff03 /* NO_EXPORT even in local confederation */
349
350 /* Origins */
351
352 #define ORIGIN_IGP 0
353 #define ORIGIN_EGP 1
354 #define ORIGIN_INCOMPLETE 2
355
356 /* Address families */
357
358 #define BGP_AF_IPV4 1
359 #define BGP_AF_IPV6 2
360
361 #ifdef IPV6
362 #define BGP_AF BGP_AF_IPV6
363 #else
364 #define BGP_AF BGP_AF_IPV4
365 #endif
366
367 #endif