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