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