Statistics
| Branch: | Revision:

iof-bird-daemon / proto / bgp / bgp.h @ 6cf72d7a

History | View | Annotate | Download (14.6 KB)

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
  u16 remote_port;                         /* Neighbor destination port */
27
  int multihop;                                /* Number of hops if multihop */
28
  int ttl_security;                        /* Enable TTL security [RFC5082] */
29
  int next_hop_self;                        /* Always set next hop to local IP address */
30
  int next_hop_keep;                        /* Do not touch next hop attribute */
31
  int missing_lladdr;                        /* What we will do when we don' know link-local addr, see MLL_* */
32
  int gw_mode;                                /* How we compute route gateway from next_hop attr, see GW_* */
33
  int compare_path_lengths;                /* Use path lengths when selecting best route */
34
  int med_metric;                        /* Compare MULTI_EXIT_DISC even between routes from differen ASes */
35
  int igp_metric;                        /* Use IGP metrics when selecting best route */
36
  int prefer_older;                        /* Prefer older routes according to RFC 5004 */
37
  int deterministic_med;                /* Use more complicated algo to have strict RFC 4271 MED comparison */
38
  u32 default_local_pref;                /* Default value for LOCAL_PREF attribute */
39
  u32 default_med;                        /* Default value for MULTI_EXIT_DISC attribute */
40
  int capabilities;                        /* Enable capability handshake [RFC3392] */
41
  int enable_refresh;                        /* Enable local support for route refresh [RFC2918] */
42
  int enable_as4;                        /* Enable local support for 4B AS numbers [RFC4893] */
43
  u32 rr_cluster_id;                        /* Route reflector cluster ID, if different from local ID */
44
  int rr_client;                        /* Whether neighbor is RR client of me */
45
  int rs_client;                        /* Whether neighbor is RS client of me */
46
  int advertise_ipv4;                        /* Whether we should add IPv4 capability advertisement to OPEN message */
47
  int passive;                                /* Do not initiate outgoing connection */
48
  int interpret_communities;                /* Hardwired handling of well-known communities */
49
  int secondary;                        /* Accept also non-best routes (i.e. RA_ACCEPTED) */
50
  int add_path;                                /* Use ADD-PATH extension [draft] */
51
  int allow_local_as;                        /* Allow that number of local ASNs in incoming AS_PATHs */
52
  int gr_mode;                                /* Graceful restart mode (BGP_GR_*) */
53
  unsigned gr_time;                        /* Graceful restart timeout */
54
  unsigned connect_delay_time;                /* Minimum delay between connect attempts */
55
  unsigned connect_retry_time;                /* Timeout for connect attempts */
56
  unsigned hold_time, initial_hold_time;
57
  unsigned keepalive_time;
58
  unsigned error_amnesia_time;                /* Errors are forgotten after */
59
  unsigned error_delay_time_min;        /* Time to wait after an error is detected */
60
  unsigned error_delay_time_max;
61
  unsigned disable_after_error;                /* Disable the protocol when error is detected */
62

    
63
  char *password;                        /* Password used for MD5 authentication */
64
  struct rtable_config *igp_table;        /* Table used for recursive next hop lookups */
65
  int check_link;                        /* Use iface link state for liveness detection */
66
  int bfd;                                /* Use BFD for liveness detection */
67
};
68

    
69
#define MLL_SELF 1
70
#define MLL_DROP 2
71
#define MLL_IGNORE 3
72

    
73
#define GW_DIRECT 1
74
#define GW_RECURSIVE 2
75

    
76
#define ADD_PATH_RX 1
77
#define ADD_PATH_TX 2
78
#define ADD_PATH_FULL 3
79

    
80
#define BGP_GR_ABLE 1
81
#define BGP_GR_AWARE 2
82

    
83
/* For peer_gr_flags */
84
#define BGP_GRF_RESTART 0x80
85

    
86
/* For peer_gr_aflags */
87
#define BGP_GRF_FORWARDING 0x80
88

    
89

    
90
struct bgp_conn {
91
  struct bgp_proto *bgp;
92
  struct birdsock *sk;
93
  unsigned int state;                        /* State of connection state machine */
94
  struct timer *connect_retry_timer;
95
  struct timer *hold_timer;
96
  struct timer *keepalive_timer;
97
  struct event *tx_ev;
98
  int packets_to_send;                        /* Bitmap of packet types to be sent */
99
  int notify_code, notify_subcode, notify_size;
100
  byte *notify_data;
101
  u32 advertised_as;                        /* Temporary value for AS number received */
102
  int start_state;                        /* protocol start_state snapshot when connection established */
103
  u8 peer_refresh_support;                /* Peer supports route refresh [RFC2918] */
104
  u8 peer_as4_support;                        /* Peer supports 4B AS numbers [RFC4893] */
105
  u8 peer_add_path;                        /* Peer supports ADD-PATH [draft] */
106
  u8 peer_gr_aware;
107
  u8 peer_gr_able;
108
  u16 peer_gr_time;
109
  u8 peer_gr_flags;
110
  u8 peer_gr_aflags;
111
  unsigned hold_time, keepalive_time;        /* Times calculated from my and neighbor's requirements */
112
};
113

    
114
struct bgp_proto {
115
  struct proto p;
116
  struct bgp_config *cf;                /* Shortcut to BGP configuration */
117
  u32 local_as, remote_as;
118
  int start_state;                        /* Substates that partitions BS_START */
119
  u8 is_internal;                        /* Internal BGP connection (local_as == remote_as) */
120
  u8 as4_session;                        /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
121
  u8 add_path_rx;                        /* Session expects receive of ADD-PATH extended NLRI */
122
  u8 add_path_tx;                        /* Session expects transmit of ADD-PATH extended NLRI */
123
  u32 local_id;                                /* BGP identifier of this router */
124
  u32 remote_id;                        /* BGP identifier of the neighbor */
125
  u32 rr_cluster_id;                        /* Route reflector cluster ID */
126
  int rr_client;                        /* Whether neighbor is RR client of me */
127
  int rs_client;                        /* Whether neighbor is RS client of me */
128
  u8 gr_ready;                                /* Neighbor could do graceful restart */
129
  u8 gr_active;                                /* Neighbor is doing graceful restart */
130
  struct bgp_conn *conn;                /* Connection we have established */
131
  struct bgp_conn outgoing_conn;        /* Outgoing connection we're working with */
132
  struct bgp_conn incoming_conn;        /* Incoming connection we have neither accepted nor rejected yet */
133
  struct object_lock *lock;                /* Lock for neighbor connection */
134
  struct neighbor *neigh;                /* Neighbor entry corresponding to remote ip, NULL if multihop */
135
  struct bfd_request *bfd_req;                /* BFD request, if BFD is used */
136
  ip_addr source_addr;                        /* Local address used as an advertised next hop */
137
  rtable *igp_table;                        /* Table used for recursive next hop lookups */
138
  struct event *event;                        /* Event for respawning and shutting process */
139
  struct timer *startup_timer;                /* Timer used to delay protocol startup due to previous errors (startup_delay) */
140
  struct timer *gr_timer;                /* Timer waiting for reestablishment after graceful restart */
141
  struct bgp_bucket **bucket_hash;        /* Hash table of attribute buckets */
142
  unsigned int hash_size, hash_count, hash_limit;
143
  HASH(struct bgp_prefix) prefix_hash;        /* Prefixes to be sent */
144
  slab *prefix_slab;                        /* Slab holding prefix nodes */
145
  list bucket_queue;                        /* Queue of buckets to send */
146
  struct bgp_bucket *withdraw_bucket;        /* Withdrawn routes */
147
  unsigned send_end_mark;                /* End-of-RIB mark scheduled for transmit */
148
  unsigned startup_delay;                /* Time to delay protocol startup by due to errors */
149
  bird_clock_t last_proto_error;        /* Time of last error that leads to protocol stop */
150
  u8 last_error_class;                         /* Error class of last error */
151
  u32 last_error_code;                        /* Error code of last error. BGP protocol errors
152
                                           are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
153
#ifdef IPV6
154
  byte *mp_reach_start, *mp_unreach_start; /* Multiprotocol BGP attribute notes */
155
  unsigned mp_reach_len, mp_unreach_len;
156
  ip_addr local_link;                        /* Link-level version of source_addr */
157
#endif
158
};
159

    
160
struct bgp_prefix {
161
  struct {
162
    ip_addr prefix;
163
    int pxlen;
164
  } n;
165
  u32 path_id;
166
  struct bgp_prefix *next;
167
  node bucket_node;                        /* Node in per-bucket list */
168
};
169

    
170
struct bgp_bucket {
171
  node send_node;                        /* Node in send queue */
172
  struct bgp_bucket *hash_next, *hash_prev;        /* Node in bucket hash table */
173
  unsigned hash;                        /* Hash over extended attributes */
174
  list prefixes;                        /* Prefixes in this buckets */
175
  ea_list eattrs[0];                        /* Per-bucket extended attributes */
176
};
177

    
178
#define BGP_PORT                179
179
#define BGP_VERSION                4
180
#define BGP_HEADER_LENGTH        19
181
#define BGP_MAX_PACKET_LENGTH        4096
182
#define BGP_RX_BUFFER_SIZE        4096
183
#define BGP_TX_BUFFER_SIZE        BGP_MAX_PACKET_LENGTH
184

    
185
extern struct linpool *bgp_linpool;
186

    
187

    
188
void bgp_start_timer(struct timer *t, int value);
189
void bgp_check_config(struct bgp_config *c);
190
void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
191
void bgp_close_conn(struct bgp_conn *c);
192
void bgp_update_startup_delay(struct bgp_proto *p);
193
void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
194
void bgp_conn_enter_established_state(struct bgp_conn *conn);
195
void bgp_conn_enter_close_state(struct bgp_conn *conn);
196
void bgp_conn_enter_idle_state(struct bgp_conn *conn);
197
void bgp_handle_graceful_restart(struct bgp_proto *p);
198
void bgp_graceful_restart_done(struct bgp_proto *p);
199
void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
200
void bgp_stop(struct bgp_proto *p, unsigned subcode);
201

    
202
struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
203
struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
204

    
205

    
206

    
207
#ifdef LOCAL_DEBUG
208
#define BGP_FORCE_DEBUG 1
209
#else
210
#define BGP_FORCE_DEBUG 0
211
#endif
212
#define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
213
        log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
214

    
215
#define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
216
        log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
217

    
218

    
219
/* attrs.c */
220

    
221
/* Hack: although BA_NEXT_HOP attribute has type EAF_TYPE_IP_ADDRESS, in IPv6
222
 * we store two addesses in it - a global address and a link local address.
223
 */
224
#ifdef IPV6
225
#define NEXT_HOP_LENGTH (2*sizeof(ip_addr))
226
static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; ((ip_addr *) b)[1] = IPA_NONE; }
227
#else
228
#define NEXT_HOP_LENGTH sizeof(ip_addr)
229
static inline void set_next_hop(byte *b, ip_addr addr) { ((ip_addr *) b)[0] = addr; }
230
#endif
231

    
232
void bgp_attach_attr(struct ea_list **to, struct linpool *pool, unsigned attr, uintptr_t val);
233
byte *bgp_attach_attr_wa(struct ea_list **to, struct linpool *pool, unsigned attr, unsigned len);
234
struct rta *bgp_decode_attrs(struct bgp_conn *conn, byte *a, unsigned int len, struct linpool *pool, int mandatory);
235
int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
236
int bgp_rte_better(struct rte *, struct rte *);
237
int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
238
void bgp_rt_notify(struct proto *P, rtable *tbl UNUSED, net *n, rte *new, rte *old UNUSED, ea_list *attrs);
239
int bgp_import_control(struct proto *, struct rte **, struct ea_list **, struct linpool *);
240
void bgp_init_bucket_table(struct bgp_proto *);
241
void bgp_free_bucket(struct bgp_proto *p, struct bgp_bucket *buck);
242
void bgp_init_prefix_table(struct bgp_proto *p, u32 order);
243
void bgp_free_prefix(struct bgp_proto *p, struct bgp_prefix *bp);
244
unsigned int bgp_encode_attrs(struct bgp_proto *p, byte *w, ea_list *attrs, int remains);
245
void bgp_get_route_info(struct rte *, byte *buf, struct ea_list *attrs);
246

    
247
inline static void bgp_attach_attr_ip(struct ea_list **to, struct linpool *pool, unsigned attr, ip_addr a)
248
{ *(ip_addr *) bgp_attach_attr_wa(to, pool, attr, sizeof(ip_addr)) = a; }
249

    
250
/* packets.c */
251

    
252
void mrt_dump_bgp_state_change(struct bgp_conn *conn, unsigned old, unsigned new);
253
void bgp_schedule_packet(struct bgp_conn *conn, int type);
254
void bgp_kick_tx(void *vconn);
255
void bgp_tx(struct birdsock *sk);
256
int bgp_rx(struct birdsock *sk, int size);
257
const char * bgp_error_dsc(unsigned code, unsigned subcode);
258
void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
259

    
260
/* Packet types */
261

    
262
#define PKT_OPEN                0x01
263
#define PKT_UPDATE                0x02
264
#define PKT_NOTIFICATION        0x03
265
#define PKT_KEEPALIVE                0x04
266
#define PKT_ROUTE_REFRESH        0x05
267
#define PKT_SCHEDULE_CLOSE        0x1f        /* Used internally to schedule socket close */
268

    
269
/* Attributes */
270

    
271
#define BAF_OPTIONAL                0x80
272
#define BAF_TRANSITIVE                0x40
273
#define BAF_PARTIAL                0x20
274
#define BAF_EXT_LEN                0x10
275

    
276
#define BA_ORIGIN                0x01        /* [RFC1771] */                /* WM */
277
#define BA_AS_PATH                0x02                                /* WM */
278
#define BA_NEXT_HOP                0x03                                /* WM */
279
#define BA_MULTI_EXIT_DISC        0x04                                /* ON */
280
#define BA_LOCAL_PREF                0x05                                /* WD */
281
#define BA_ATOMIC_AGGR                0x06                                /* WD */
282
#define BA_AGGREGATOR                0x07                                /* OT */
283
#define BA_COMMUNITY                0x08        /* [RFC1997] */                /* OT */
284
#define BA_ORIGINATOR_ID        0x09        /* [RFC1966] */                /* ON */
285
#define BA_CLUSTER_LIST                0x0a                                /* ON */
286
/* We don't support these: */
287
#define BA_DPA                        0x0b        /* ??? */
288
#define BA_ADVERTISER                0x0c        /* [RFC1863] */
289
#define BA_RCID_PATH                0x0d
290
#define BA_MP_REACH_NLRI        0x0e        /* [RFC2283] */
291
#define BA_MP_UNREACH_NLRI        0x0f
292
#define BA_EXT_COMMUNITY        0x10        /* [RFC4360] */
293
#define BA_AS4_PATH             0x11    /* [RFC4893] */
294
#define BA_AS4_AGGREGATOR       0x12
295

    
296
/* BGP connection states */
297

    
298
#define BS_IDLE                        0
299
#define BS_CONNECT                1        /* Attempting to connect */
300
#define BS_ACTIVE                2        /* Waiting for connection retry & listening */
301
#define BS_OPENSENT                3
302
#define BS_OPENCONFIRM                4
303
#define BS_ESTABLISHED                5
304
#define BS_CLOSE                6        /* Used during transition to BS_IDLE */
305

    
306
#define BS_MAX                        7
307

    
308
/* BGP start states
309
 * 
310
 * Used in PS_START for fine-grained specification of starting state.
311
 *
312
 * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP protocol
313
 * done what is neccessary to start itself (like acquiring the lock), it goes to BSS_CONNECT.
314
 * When some connection attempt failed because of option or capability error, it goes to
315
 * BSS_CONNECT_NOCAP.
316
 */
317

    
318
#define BSS_PREPARE                0        /* Used before ordinary BGP started, i. e. waiting for lock */
319
#define BSS_DELAY                1        /* Startup delay due to previous errors */
320
#define BSS_CONNECT                2        /* Ordinary BGP connecting */
321
#define BSS_CONNECT_NOCAP        3        /* Legacy BGP connecting (without capabilities) */
322

    
323
/* Error classes */
324

    
325
#define BE_NONE                        0
326
#define BE_MISC                        1        /* Miscellaneous error */
327
#define BE_SOCKET                2        /* Socket error */
328
#define BE_BGP_RX                3        /* BGP protocol error notification received */
329
#define BE_BGP_TX                4        /* BGP protocol error notification sent */
330
#define BE_AUTO_DOWN                5        /* Automatic shutdown */
331
#define BE_MAN_DOWN                6        /* Manual shutdown */
332

    
333
/* Misc error codes */
334

    
335
#define BEM_NEIGHBOR_LOST        1
336
#define BEM_INVALID_NEXT_HOP        2
337
#define BEM_INVALID_MD5                3        /* MD5 authentication kernel request failed (possibly not supported) */
338
#define BEM_NO_SOCKET                4
339
#define BEM_LINK_DOWN                5
340
#define BEM_BFD_DOWN                6
341
#define BEM_GRACEFUL_RESTART        7
342

    
343
/* Automatic shutdown error codes */
344

    
345
#define BEA_ROUTE_LIMIT_EXCEEDED 1
346

    
347
/* Well-known communities */
348

    
349
#define BGP_COMM_NO_EXPORT                0xffffff01        /* Don't export outside local AS / confed. */
350
#define BGP_COMM_NO_ADVERTISE                0xffffff02        /* Don't export at all */
351
#define BGP_COMM_NO_EXPORT_SUBCONFED        0xffffff03        /* NO_EXPORT even in local confederation */
352

    
353
/* Origins */
354

    
355
#define ORIGIN_IGP                0
356
#define ORIGIN_EGP                1
357
#define ORIGIN_INCOMPLETE        2
358

    
359
/* Address families */
360

    
361
#define BGP_AF_IPV4                1
362
#define BGP_AF_IPV6                2
363

    
364
#ifdef IPV6
365
#define BGP_AF BGP_AF_IPV6
366
#else
367
#define BGP_AF BGP_AF_IPV4
368
#endif
369

    
370
#endif