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