]> git.ipfire.org Git - thirdparty/bird.git/blob - proto/bgp/bgp.h
BFD protocol, ready for release.
[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
15 struct linpool;
16 struct eattr;
17 struct bfd_request;
18
19 struct bgp_config {
20 struct proto_config c;
21 u32 local_as, remote_as;
22 ip_addr remote_ip;
23 ip_addr source_addr; /* Source address to use */
24 struct iface *iface; /* Interface for link-local addresses */
25 int multihop; /* Number of hops if multihop */
26 int ttl_security; /* Enable TTL security [RFC5082] */
27 int next_hop_self; /* Always set next hop to local IP address */
28 int next_hop_keep; /* Do not touch next hop attribute */
29 int missing_lladdr; /* What we will do when we don' know link-local addr, see MLL_* */
30 int gw_mode; /* How we compute route gateway from next_hop attr, see GW_* */
31 int compare_path_lengths; /* Use path lengths when selecting best route */
32 int med_metric; /* Compare MULTI_EXIT_DISC even between routes from differen ASes */
33 int igp_metric; /* Use IGP metrics when selecting best route */
34 int prefer_older; /* Prefer older routes according to RFC 5004 */
35 int deterministic_med; /* Use more complicated algo to have strict RFC 4271 MED comparison */
36 u32 default_local_pref; /* Default value for LOCAL_PREF attribute */
37 u32 default_med; /* Default value for MULTI_EXIT_DISC attribute */
38 int capabilities; /* Enable capability handshake [RFC3392] */
39 int enable_refresh; /* Enable local support for route refresh [RFC2918] */
40 int enable_as4; /* Enable local support for 4B AS numbers [RFC4893] */
41 u32 rr_cluster_id; /* Route reflector cluster ID, if different from local ID */
42 int rr_client; /* Whether neighbor is RR client of me */
43 int rs_client; /* Whether neighbor is RS client of me */
44 int advertise_ipv4; /* Whether we should add IPv4 capability advertisement to OPEN message */
45 int passive; /* Do not initiate outgoing connection */
46 int interpret_communities; /* Hardwired handling of well-known communities */
47 int secondary; /* Accept also non-best routes (i.e. RA_ACCEPTED) */
48 unsigned connect_retry_time;
49 unsigned hold_time, initial_hold_time;
50 unsigned keepalive_time;
51 unsigned start_delay_time; /* Minimum delay between connects */
52 unsigned error_amnesia_time; /* Errors are forgotten after */
53 unsigned error_delay_time_min; /* Time to wait after an error is detected */
54 unsigned error_delay_time_max;
55 unsigned disable_after_error; /* Disable the protocol when error is detected */
56
57 char *password; /* Password used for MD5 authentication */
58 struct rtable_config *igp_table; /* Table used for recursive next hop lookups */
59 int bfd; /* Use BFD for liveness detection */
60 };
61
62 #define MLL_SELF 1
63 #define MLL_DROP 2
64 #define MLL_IGNORE 3
65
66 #define GW_DIRECT 1
67 #define GW_RECURSIVE 2
68
69 struct bgp_conn {
70 struct bgp_proto *bgp;
71 struct birdsock *sk;
72 unsigned int state; /* State of connection state machine */
73 struct timer *connect_retry_timer;
74 struct timer *hold_timer;
75 struct timer *keepalive_timer;
76 struct event *tx_ev;
77 int packets_to_send; /* Bitmap of packet types to be sent */
78 int notify_code, notify_subcode, notify_size;
79 byte *notify_data;
80 u32 advertised_as; /* Temporary value for AS number received */
81 int start_state; /* protocol start_state snapshot when connection established */
82 int want_as4_support; /* Connection tries to establish AS4 session */
83 int peer_as4_support; /* Peer supports 4B AS numbers [RFC4893] */
84 int peer_refresh_support; /* Peer supports route refresh [RFC2918] */
85 unsigned hold_time, keepalive_time; /* Times calculated from my and neighbor's requirements */
86 };
87
88 struct bgp_proto {
89 struct proto p;
90 struct bgp_config *cf; /* Shortcut to BGP configuration */
91 u32 local_as, remote_as;
92 int start_state; /* Substates that partitions BS_START */
93 int is_internal; /* Internal BGP connection (local_as == remote_as) */
94 int as4_session; /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
95 u32 local_id; /* BGP identifier of this router */
96 u32 remote_id; /* BGP identifier of the neighbor */
97 u32 rr_cluster_id; /* Route reflector cluster ID */
98 int rr_client; /* Whether neighbor is RR client of me */
99 int rs_client; /* Whether neighbor is RS client of me */
100 struct bgp_conn *conn; /* Connection we have established */
101 struct bgp_conn outgoing_conn; /* Outgoing connection we're working with */
102 struct bgp_conn incoming_conn; /* Incoming connection we have neither accepted nor rejected yet */
103 struct object_lock *lock; /* Lock for neighbor connection */
104 struct neighbor *neigh; /* Neighbor entry corresponding to remote ip, NULL if multihop */
105 struct bfd_request *bfd_req; /* BFD request, if BFD is used */
106 ip_addr source_addr; /* Local address used as an advertised next hop */
107 rtable *igp_table; /* Table used for recursive next hop lookups */
108 struct event *event; /* Event for respawning and shutting process */
109 struct timer *startup_timer; /* Timer used to delay protocol startup due to previous errors (startup_delay) */
110 struct bgp_bucket **bucket_hash; /* Hash table of attribute buckets */
111 unsigned int hash_size, hash_count, hash_limit;
112 struct fib prefix_fib; /* Prefixes to be sent */
113 list bucket_queue; /* Queue of buckets to send */
114 struct bgp_bucket *withdraw_bucket; /* Withdrawn routes */
115 unsigned startup_delay; /* Time to delay protocol startup by due to errors */
116 bird_clock_t last_proto_error; /* Time of last error that leads to protocol stop */
117 u8 last_error_class; /* Error class of last error */
118 u32 last_error_code; /* Error code of last error. BGP protocol errors
119 are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
120 #ifdef IPV6
121 byte *mp_reach_start, *mp_unreach_start; /* Multiprotocol BGP attribute notes */
122 unsigned mp_reach_len, mp_unreach_len;
123 ip_addr local_link; /* Link-level version of source_addr */
124 #endif
125 };
126
127 struct bgp_prefix {
128 struct fib_node n; /* Node in prefix fib */
129 node bucket_node; /* Node in per-bucket list */
130 };
131
132 struct bgp_bucket {
133 node send_node; /* Node in send queue */
134 struct bgp_bucket *hash_next, *hash_prev; /* Node in bucket hash table */
135 unsigned hash; /* Hash over extended attributes */
136 list prefixes; /* Prefixes in this buckets */
137 ea_list eattrs[0]; /* Per-bucket extended attributes */
138 };
139
140 #define BGP_PORT 179
141 #define BGP_VERSION 4
142 #define BGP_HEADER_LENGTH 19
143 #define BGP_MAX_PACKET_LENGTH 4096
144 #define BGP_RX_BUFFER_SIZE 4096
145 #define BGP_TX_BUFFER_SIZE BGP_MAX_PACKET_LENGTH
146
147 extern struct linpool *bgp_linpool;
148
149
150 void bgp_start_timer(struct timer *t, int value);
151 void bgp_check_config(struct bgp_config *c);
152 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
153 void bgp_close_conn(struct bgp_conn *c);
154 void bgp_update_startup_delay(struct bgp_proto *p);
155 void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
156 void bgp_conn_enter_established_state(struct bgp_conn *conn);
157 void bgp_conn_enter_close_state(struct bgp_conn *conn);
158 void bgp_conn_enter_idle_state(struct bgp_conn *conn);
159 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
160 void bgp_stop(struct bgp_proto *p, unsigned subcode);
161
162
163
164 #ifdef LOCAL_DEBUG
165 #define BGP_FORCE_DEBUG 1
166 #else
167 #define BGP_FORCE_DEBUG 0
168 #endif
169 #define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
170 log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
171
172 #define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
173 log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
174
175
176 /* attrs.c */
177
178 /* Hack: although BA_NEXT_HOP attribute has type EAF_TYPE_IP_ADDRESS, in IPv6
179 * we store two addesses in it - a global address and a link local address.
180 */
181 #ifdef IPV6
182 #define NEXT_HOP_LENGTH (2*sizeof(ip_addr))
183 static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; ((ip_addr *) b)[1] = IPA_NONE; }
184 #else
185 #define NEXT_HOP_LENGTH sizeof(ip_addr)
186 static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; }
187 #endif
188
189 void bgp_attach_attr(struct ea_list **to, struct linpool *pool, unsigned attr, uintptr_t val);
190 byte *bgp_attach_attr_wa(struct ea_list **to, struct linpool *pool, unsigned attr, unsigned len);
191 struct rta *bgp_decode_attrs(struct bgp_conn *conn, byte *a, unsigned int len, struct linpool *pool, int mandatory);
192 int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
193 int bgp_rte_better(struct rte *, struct rte *);
194 int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
195 void bgp_rt_notify(struct proto *P, rtable *tbl UNUSED, net *n, rte *new, rte *old UNUSED, ea_list *attrs);
196 int bgp_import_control(struct proto *, struct rte **, struct ea_list **, struct linpool *);
197 void bgp_attr_init(struct bgp_proto *);
198 unsigned int bgp_encode_attrs(struct bgp_proto *p, byte *w, ea_list *attrs, int remains);
199 void bgp_free_bucket(struct bgp_proto *p, struct bgp_bucket *buck);
200 void bgp_get_route_info(struct rte *, byte *buf, struct ea_list *attrs);
201
202 inline static void bgp_attach_attr_ip(struct ea_list **to, struct linpool *pool, unsigned attr, ip_addr a)
203 { *(ip_addr *) bgp_attach_attr_wa(to, pool, attr, sizeof(ip_addr)) = a; }
204
205 /* packets.c */
206
207 void mrt_dump_bgp_state_change(struct bgp_conn *conn, unsigned old, unsigned new);
208 void bgp_schedule_packet(struct bgp_conn *conn, int type);
209 void bgp_kick_tx(void *vconn);
210 void bgp_tx(struct birdsock *sk);
211 int bgp_rx(struct birdsock *sk, int size);
212 const char * bgp_error_dsc(unsigned code, unsigned subcode);
213 void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
214
215 /* Packet types */
216
217 #define PKT_OPEN 0x01
218 #define PKT_UPDATE 0x02
219 #define PKT_NOTIFICATION 0x03
220 #define PKT_KEEPALIVE 0x04
221 #define PKT_ROUTE_REFRESH 0x05
222 #define PKT_SCHEDULE_CLOSE 0x1f /* Used internally to schedule socket close */
223
224 /* Attributes */
225
226 #define BAF_OPTIONAL 0x80
227 #define BAF_TRANSITIVE 0x40
228 #define BAF_PARTIAL 0x20
229 #define BAF_EXT_LEN 0x10
230
231 #define BA_ORIGIN 0x01 /* [RFC1771] */ /* WM */
232 #define BA_AS_PATH 0x02 /* WM */
233 #define BA_NEXT_HOP 0x03 /* WM */
234 #define BA_MULTI_EXIT_DISC 0x04 /* ON */
235 #define BA_LOCAL_PREF 0x05 /* WD */
236 #define BA_ATOMIC_AGGR 0x06 /* WD */
237 #define BA_AGGREGATOR 0x07 /* OT */
238 #define BA_COMMUNITY 0x08 /* [RFC1997] */ /* OT */
239 #define BA_ORIGINATOR_ID 0x09 /* [RFC1966] */ /* ON */
240 #define BA_CLUSTER_LIST 0x0a /* ON */
241 /* We don't support these: */
242 #define BA_DPA 0x0b /* ??? */
243 #define BA_ADVERTISER 0x0c /* [RFC1863] */
244 #define BA_RCID_PATH 0x0d
245 #define BA_MP_REACH_NLRI 0x0e /* [RFC2283] */
246 #define BA_MP_UNREACH_NLRI 0x0f
247 #define BA_EXT_COMMUNITY 0x10 /* [RFC4360] */
248 #define BA_AS4_PATH 0x11 /* [RFC4893] */
249 #define BA_AS4_AGGREGATOR 0x12
250
251 /* BGP connection states */
252
253 #define BS_IDLE 0
254 #define BS_CONNECT 1 /* Attempting to connect */
255 #define BS_ACTIVE 2 /* Waiting for connection retry & listening */
256 #define BS_OPENSENT 3
257 #define BS_OPENCONFIRM 4
258 #define BS_ESTABLISHED 5
259 #define BS_CLOSE 6 /* Used during transition to BS_IDLE */
260
261 #define BS_MAX 7
262
263 /* BGP start states
264 *
265 * Used in PS_START for fine-grained specification of starting state.
266 *
267 * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP protocol
268 * done what is neccessary to start itself (like acquiring the lock), it goes to BSS_CONNECT.
269 * When some connection attempt failed because of option or capability error, it goes to
270 * BSS_CONNECT_NOCAP.
271 */
272
273 #define BSS_PREPARE 0 /* Used before ordinary BGP started, i. e. waiting for lock */
274 #define BSS_DELAY 1 /* Startup delay due to previous errors */
275 #define BSS_CONNECT 2 /* Ordinary BGP connecting */
276 #define BSS_CONNECT_NOCAP 3 /* Legacy BGP connecting (without capabilities) */
277
278 /* Error classes */
279
280 #define BE_NONE 0
281 #define BE_MISC 1 /* Miscellaneous error */
282 #define BE_SOCKET 2 /* Socket error */
283 #define BE_BGP_RX 3 /* BGP protocol error notification received */
284 #define BE_BGP_TX 4 /* BGP protocol error notification sent */
285 #define BE_AUTO_DOWN 5 /* Automatic shutdown */
286 #define BE_MAN_DOWN 6 /* Manual shutdown */
287
288 /* Misc error codes */
289
290 #define BEM_NEIGHBOR_LOST 1
291 #define BEM_INVALID_NEXT_HOP 2
292 #define BEM_INVALID_MD5 3 /* MD5 authentication kernel request failed (possibly not supported) */
293 #define BEM_NO_SOCKET 4
294 #define BEM_BFD_DOWN 5
295
296 /* Automatic shutdown error codes */
297
298 #define BEA_ROUTE_LIMIT_EXCEEDED 1
299
300 /* Well-known communities */
301
302 #define BGP_COMM_NO_EXPORT 0xffffff01 /* Don't export outside local AS / confed. */
303 #define BGP_COMM_NO_ADVERTISE 0xffffff02 /* Don't export at all */
304 #define BGP_COMM_NO_EXPORT_SUBCONFED 0xffffff03 /* NO_EXPORT even in local confederation */
305
306 /* Origins */
307
308 #define ORIGIN_IGP 0
309 #define ORIGIN_EGP 1
310 #define ORIGIN_INCOMPLETE 2
311
312 /* Address families */
313
314 #define BGP_AF_IPV4 1
315 #define BGP_AF_IPV6 2
316
317 #ifdef IPV6
318 #define BGP_AF BGP_AF_IPV6
319 #else
320 #define BGP_AF BGP_AF_IPV4
321 #endif
322
323 #endif