proto/bgp/bgp.h

   1 /*
   2  *      BIRD -- The Border Gateway Protocol
   3  *
   4  *      (c) 2000 Martin Mares <mj@ucw.cz>
   5  *      (c) 2008--2016 Ondrej Zajicek <santiago@crfreenet.org>
   6  *      (c) 2008--2016 CZ.NIC z.s.p.o.
   7  *
   8  *      Can be freely distributed and used under the terms of the GNU GPL.
   9  */
  10
  11 #ifndef _BIRD_BGP_H_
  12 #define _BIRD_BGP_H_
  13
  14 #include <stdint.h>
  15 #include <setjmp.h>
  16 #include "nest/bird.h"
  17 #include "nest/route.h"
  18 #include "nest/bfd.h"
  19 //#include "lib/lists.h"
  20 #include "lib/hash.h"
  21 #include "lib/socket.h"
  22
  23 struct linpool;
  24 struct eattr;
  25
  26
  27 /* Address families */
  28
  29 #define BGP_AFI_IPV4            1
  30 #define BGP_AFI_IPV6            2
  31
  32 #define BGP_SAFI_UNICAST        1
  33 #define BGP_SAFI_MULTICAST      2
  34 #define BGP_SAFI_MPLS           4
  35 #define BGP_SAFI_MPLS_VPN       128
  36 #define BGP_SAFI_VPN_MULTICAST  129
  37 #define BGP_SAFI_FLOW           133
  38
  39 /* Internal AF codes */
  40
  41 #define BGP_AF(A, B)            (((u32)(A) << 16) | (u32)(B))
  42 #define BGP_AFI(A)              ((u32)(A) >> 16)
  43 #define BGP_SAFI(A)             ((u32)(A) & 0xFFFF)
  44
  45 #define BGP_AF_IPV4             BGP_AF( BGP_AFI_IPV4, BGP_SAFI_UNICAST )
  46 #define BGP_AF_IPV6             BGP_AF( BGP_AFI_IPV6, BGP_SAFI_UNICAST )
  47 #define BGP_AF_IPV4_MC          BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MULTICAST )
  48 #define BGP_AF_IPV6_MC          BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MULTICAST )
  49 #define BGP_AF_IPV4_MPLS        BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MPLS )
  50 #define BGP_AF_IPV6_MPLS        BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MPLS )
  51 #define BGP_AF_VPN4_MPLS        BGP_AF( BGP_AFI_IPV4, BGP_SAFI_MPLS_VPN )
  52 #define BGP_AF_VPN6_MPLS        BGP_AF( BGP_AFI_IPV6, BGP_SAFI_MPLS_VPN )
  53 #define BGP_AF_VPN4_MC          BGP_AF( BGP_AFI_IPV4, BGP_SAFI_VPN_MULTICAST )
  54 #define BGP_AF_VPN6_MC          BGP_AF( BGP_AFI_IPV6, BGP_SAFI_VPN_MULTICAST )
  55 #define BGP_AF_FLOW4            BGP_AF( BGP_AFI_IPV4, BGP_SAFI_FLOW )
  56 #define BGP_AF_FLOW6            BGP_AF( BGP_AFI_IPV6, BGP_SAFI_FLOW )
  57
  58
  59 struct bgp_write_state;
  60 struct bgp_parse_state;
  61 struct bgp_export_state;
  62 struct bgp_bucket;
  63
  64 struct bgp_af_desc {
  65   u32 afi;
  66   u32 net;
  67   u8 mpls;
  68   u8 no_igp;
  69   const char *name;
  70   uint (*encode_nlri)(struct bgp_write_state *s, struct bgp_bucket *buck, byte *buf, uint size);
  71   void (*decode_nlri)(struct bgp_parse_state *s, byte *pos, uint len, rta *a);
  72   void (*update_next_hop)(struct bgp_export_state *s, eattr *nh, ea_list **to);
  73   uint (*encode_next_hop)(struct bgp_write_state *s, eattr *nh, byte *buf, uint size);
  74   void (*decode_next_hop)(struct bgp_parse_state *s, byte *pos, uint len, rta *a);
  75 };
  76
  77
  78 struct bgp_config {
  79   struct proto_config c;
  80   u32 local_as, remote_as;
  81   ip_addr local_ip;                     /* Source address to use */
  82   ip_addr remote_ip;
  83   struct iface *iface;                  /* Interface for link-local addresses */
  84   u16 local_port;                       /* Local listening port */
  85   u16 remote_port;                      /* Neighbor destination port */
  86   int peer_type;                        /* Internal or external BGP (BGP_PT_*, optional) */
  87   int multihop;                         /* Number of hops if multihop */
  88   int strict_bind;                      /* Bind listening socket to local address */
  89   int ttl_security;                     /* Enable TTL security [RFC 5082] */
  90   int compare_path_lengths;             /* Use path lengths when selecting best route */
  91   int med_metric;                       /* Compare MULTI_EXIT_DISC even between routes from differen ASes */
  92   int igp_metric;                       /* Use IGP metrics when selecting best route */
  93   int prefer_older;                     /* Prefer older routes according to RFC 5004 */
  94   int deterministic_med;                /* Use more complicated algo to have strict RFC 4271 MED comparison */
  95   u32 default_local_pref;               /* Default value for LOCAL_PREF attribute */
  96   u32 default_med;                      /* Default value for MULTI_EXIT_DISC attribute */
  97   int capabilities;                     /* Enable capability handshake [RFC 5492] */
  98   int enable_refresh;                   /* Enable local support for route refresh [RFC 2918] */
  99   int enable_as4;                       /* Enable local support for 4B AS numbers [RFC 6793] */
 100   int enable_extended_messages;         /* Enable local support for extended messages [draft] */
 101   u32 rr_cluster_id;                    /* Route reflector cluster ID, if different from local ID */
 102   int rr_client;                        /* Whether neighbor is RR client of me */
 103   int rs_client;                        /* Whether neighbor is RS client of me */
 104   u32 confederation;                    /* Confederation ID, or zero if confeds not active */
 105   int confederation_member;             /* Whether neighbor AS is member of our confederation */
 106   int passive;                          /* Do not initiate outgoing connection */
 107   int interpret_communities;            /* Hardwired handling of well-known communities */
 108   int allow_local_as;                   /* Allow that number of local ASNs in incoming AS_PATHs */
 109   int allow_local_pref;                 /* Allow LOCAL_PREF in EBGP sessions */
 110   int gr_mode;                          /* Graceful restart mode (BGP_GR_*) */
 111   int llgr_mode;                        /* Long-lived graceful restart mode (BGP_LLGR_*) */
 112   int setkey;                           /* Set MD5 password to system SA/SP database */
 113   /* Times below are in seconds */
 114   unsigned gr_time;                     /* Graceful restart timeout */
 115   unsigned llgr_time;                   /* Long-lived graceful restart stale time */
 116   unsigned connect_delay_time;          /* Minimum delay between connect attempts */
 117   unsigned connect_retry_time;          /* Timeout for connect attempts */
 118   unsigned hold_time, initial_hold_time;
 119   unsigned keepalive_time;
 120   unsigned error_amnesia_time;          /* Errors are forgotten after */
 121   unsigned error_delay_time_min;        /* Time to wait after an error is detected */
 122   unsigned error_delay_time_max;
 123   unsigned disable_after_error;         /* Disable the protocol when error is detected */
 124   u32 disable_after_cease;              /* Disable it when cease is received, bitfield */
 125
 126   char *password;                       /* Password used for MD5 authentication */
 127   int check_link;                       /* Use iface link state for liveness detection */
 128   int bfd;                              /* Use BFD for liveness detection */
 129 };
 130
 131 struct bgp_channel_config {
 132   struct channel_config c;
 133
 134   u32 afi;
 135   const struct bgp_af_desc *desc;
 136
 137   ip_addr next_hop_addr;                /* Local address for NEXT_HOP attribute */
 138   u8 next_hop_self;                     /* Always set next hop to local IP address (NH_*) */
 139   u8 next_hop_keep;                     /* Do not modify next hop attribute (NH_*) */
 140   u8 mandatory;                         /* Channel is mandatory in capability negotiation */
 141   u8 missing_lladdr;                    /* What we will do when we don' know link-local addr, see MLL_* */
 142   u8 gw_mode;                           /* How we compute route gateway from next_hop attr, see GW_* */
 143   u8 secondary;                         /* Accept also non-best routes (i.e. RA_ACCEPTED) */
 144   u8 gr_able;                           /* Allow full graceful restart for the channel */
 145   u8 llgr_able;                         /* Allow full long-lived GR for the channel */
 146   uint llgr_time;                       /* Long-lived graceful restart stale time */
 147   u8 ext_next_hop;                      /* Allow both IPv4 and IPv6 next hops */
 148   u8 add_path;                          /* Use ADD-PATH extension [RFC 7911] */
 149   u8 import_table;                      /* Use c.in_table as Adj-RIB-In */
 150
 151   uint rest[0];                         /* Remaining items are reconfigured separately */
 152   struct rtable_config *igp_table_ip4;  /* Table for recursive IPv4 next hop lookups */
 153   struct rtable_config *igp_table_ip6;  /* Table for recursive IPv6 next hop lookups */
 154 };
 155
 156 #define BGP_PT_INTERNAL         1
 157 #define BGP_PT_EXTERNAL         2
 158
 159 #define NH_NO                   0
 160 #define NH_ALL                  1
 161 #define NH_IBGP                 2
 162 #define NH_EBGP                 3
 163
 164 #define MLL_SELF                1
 165 #define MLL_DROP                2
 166 #define MLL_IGNORE              3
 167
 168 #define GW_DIRECT               1
 169 #define GW_RECURSIVE            2
 170
 171 #define BGP_ADD_PATH_RX         1
 172 #define BGP_ADD_PATH_TX         2
 173 #define BGP_ADD_PATH_FULL       3
 174
 175 #define BGP_GR_ABLE             1
 176 #define BGP_GR_AWARE            2
 177
 178 /* For GR capability common flags */
 179 #define BGP_GRF_RESTART 0x80
 180
 181 /* For GR capability per-AF flags */
 182 #define BGP_GRF_FORWARDING 0x80
 183
 184 #define BGP_LLGR_ABLE           1
 185 #define BGP_LLGR_AWARE          2
 186
 187 #define BGP_LLGRF_FORWARDING 0x80
 188
 189 #define BGP_GRS_NONE            0       /* No GR  */
 190 #define BGP_GRS_ACTIVE          1       /* Graceful restart per RFC 4724 */
 191 #define BGP_GRS_LLGR            2       /* Long-lived GR phase (stale timer active) */
 192
 193 #define BGP_BFD_GRACEFUL        2       /* BFD down triggers graceful restart */
 194
 195
 196 struct bgp_af_caps {
 197   u32 afi;
 198   u8 ready;                             /* Multiprotocol capability, RFC 4760 */
 199   u8 gr_able;                           /* Graceful restart support, RFC 4724 */
 200   u8 gr_af_flags;                       /* Graceful restart per-AF flags */
 201   u8 llgr_able;                         /* Long-lived GR, RFC draft */
 202   u32 llgr_time;                        /* Long-lived GR stale time */
 203   u8 llgr_flags;                        /* Long-lived GR per-AF flags */
 204   u8 ext_next_hop;                      /* Extended IPv6 next hop,   RFC 5549 */
 205   u8 add_path;                          /* Multiple paths support,   RFC 7911 */
 206 };
 207
 208 struct bgp_caps {
 209   u32 as4_number;                       /* Announced ASN */
 210
 211   u8 as4_support;                       /* Four-octet AS capability, RFC 6793 */
 212   u8 ext_messages;                      /* Extended message length,  RFC draft */
 213   u8 route_refresh;                     /* Route refresh capability, RFC 2918 */
 214   u8 enhanced_refresh;                  /* Enhanced route refresh,   RFC 7313 */
 215
 216   u8 gr_aware;                          /* Graceful restart capability, RFC 4724 */
 217   u8 gr_flags;                          /* Graceful restart flags */
 218   u16 gr_time;                          /* Graceful restart time in seconds */
 219
 220   u8 llgr_aware;                        /* Long-lived GR capability, RFC draft */
 221   u8 any_ext_next_hop;                  /* Bitwise OR of per-AF ext_next_hop */
 222   u8 any_add_path;                      /* Bitwise OR of per-AF add_path */
 223
 224   u16 af_count;                         /* Number of af_data items */
 225   u16 length;                           /* Length of capabilities in OPEN msg */
 226
 227   struct bgp_af_caps af_data[0];        /* Per-AF capability data */
 228 };
 229
 230 #define WALK_AF_CAPS(caps,ac) \
 231   for (ac = caps->af_data; ac < &caps->af_data[caps->af_count]; ac++)
 232
 233
 234 struct bgp_socket {
 235   node n;                               /* Node in global bgp_sockets */
 236   sock *sk;                             /* Real listening socket */
 237   u32 uc;                               /* Use count */
 238 };
 239
 240 struct bgp_conn {
 241   struct bgp_proto *bgp;
 242   struct birdsock *sk;
 243   u8 state;                             /* State of connection state machine */
 244   u8 as4_session;                       /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
 245   u8 ext_messages;                      /* Session uses extended message length */
 246   u32 received_as;                      /* ASN received in OPEN message */
 247
 248   struct bgp_caps *local_caps;
 249   struct bgp_caps *remote_caps;
 250   timer *connect_timer;
 251   timer *hold_timer;
 252   timer *keepalive_timer;
 253   event *tx_ev;
 254   u32 packets_to_send;                  /* Bitmap of packet types to be sent */
 255   u32 channels_to_send;                 /* Bitmap of channels with packets to be sent */
 256   u8 last_channel;                      /* Channel used last time for TX */
 257   u8 last_channel_count;                /* Number of times the last channel was used in succession */
 258   int notify_code, notify_subcode, notify_size;
 259   byte *notify_data;
 260
 261   uint hold_time, keepalive_time;       /* Times calculated from my and neighbor's requirements */
 262 };
 263
 264 struct bgp_proto {
 265   struct proto p;
 266   const struct bgp_config *cf;          /* Shortcut to BGP configuration */
 267   ip_addr local_ip, remote_ip;
 268   u32 local_as, remote_as;
 269   u32 public_as;                        /* Externally visible ASN (local_as or confederation id) */
 270   u32 local_id;                         /* BGP identifier of this router */
 271   u32 remote_id;                        /* BGP identifier of the neighbor */
 272   u32 rr_cluster_id;                    /* Route reflector cluster ID */
 273   int start_state;                      /* Substates that partitions BS_START */
 274   u8 is_internal;                       /* Internal BGP session (local_as == remote_as) */
 275   u8 is_interior;                       /* Internal or intra-confederation BGP session */
 276   u8 as4_session;                       /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
 277   u8 rr_client;                         /* Whether neighbor is RR client of me */
 278   u8 rs_client;                         /* Whether neighbor is RS client of me */
 279   u8 route_refresh;                     /* Route refresh allowed to send [RFC 2918] */
 280   u8 enhanced_refresh;                  /* Enhanced refresh is negotiated [RFC 7313] */
 281   u8 gr_ready;                          /* Neighbor could do graceful restart */
 282   u8 llgr_ready;                        /* Neighbor could do Long-lived GR, implies gr_ready */
 283   u8 gr_active_num;                     /* Neighbor is doing GR, number of active channels */
 284   u8 channel_count;                     /* Number of active channels */
 285   u8 summary_add_path_rx;               /* Summary state of ADD_PATH RX w.r.t active channels */
 286   u32 *afi_map;                         /* Map channel index -> AFI */
 287   struct bgp_channel **channel_map;     /* Map channel index -> channel */
 288   struct bgp_conn *conn;                /* Connection we have established */
 289   struct bgp_conn outgoing_conn;        /* Outgoing connection we're working with */
 290   struct bgp_conn incoming_conn;        /* Incoming connection we have neither accepted nor rejected yet */
 291   struct object_lock *lock;             /* Lock for neighbor connection */
 292   struct neighbor *neigh;               /* Neighbor entry corresponding to remote ip, NULL if multihop */
 293   struct bgp_socket *sock;              /* Shared listening socket */
 294   struct bfd_request *bfd_req;          /* BFD request, if BFD is used */
 295   ip_addr link_addr;                    /* Link-local version of local_ip */
 296   event *event;                         /* Event for respawning and shutting process */
 297   timer *startup_timer;                 /* Timer used to delay protocol startup due to previous errors (startup_delay) */
 298   timer *gr_timer;                      /* Timer waiting for reestablishment after graceful restart */
 299   uint startup_delay;                   /* Delay (in seconds) of protocol startup due to previous errors */
 300   btime last_proto_error;               /* Time of last error that leads to protocol stop */
 301   u8 last_error_class;                  /* Error class of last error */
 302   u32 last_error_code;                  /* Error code of last error. BGP protocol errors
 303                                            are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
 304 };
 305
 306 struct bgp_channel {
 307   struct channel c;
 308
 309   /* Rest are BGP specific data */
 310   struct bgp_channel_config *cf;
 311
 312   u32 afi;
 313   u32 index;
 314   const struct bgp_af_desc *desc;
 315
 316   rtable *igp_table_ip4;                /* Table for recursive IPv4 next hop lookups */
 317   rtable *igp_table_ip6;                /* Table for recursive IPv6 next hop lookups */
 318
 319   /* Rest are zeroed when down */
 320   pool *pool;
 321   HASH(struct bgp_bucket) bucket_hash;  /* Hash table of route buckets */
 322   struct bgp_bucket *withdraw_bucket;   /* Withdrawn routes */
 323   list bucket_queue;                    /* Queue of buckets to send (struct bgp_bucket) */
 324
 325   HASH(struct bgp_prefix) prefix_hash;  /* Prefixes to be sent */
 326   slab *prefix_slab;                    /* Slab holding prefix nodes */
 327
 328   ip_addr next_hop_addr;                /* Local address for NEXT_HOP attribute */
 329   ip_addr link_addr;                    /* Link-local version of next_hop_addr */
 330
 331   u32 packets_to_send;                  /* Bitmap of packet types to be sent */
 332
 333   u8 ext_next_hop;                      /* Session allows both IPv4 and IPv6 next hops */
 334
 335   u8 gr_ready;                          /* Neighbor could do GR on this AF */
 336   u8 gr_active;                         /* Neighbor is doing GR (BGP_GRS_*) */
 337
 338   timer *stale_timer;                   /* Long-lived stale timer for LLGR */
 339   u32 stale_time;                       /* Stored LLGR stale time from last session */
 340
 341   u8 add_path_rx;                       /* Session expects receive of ADD-PATH extended NLRI */
 342   u8 add_path_tx;                       /* Session expects transmit of ADD-PATH extended NLRI */
 343
 344   u8 feed_state;                        /* Feed state (TX) for EoR, RR packets, see BFS_* */
 345   u8 load_state;                        /* Load state (RX) for EoR, RR packets, see BFS_* */
 346 };
 347
 348 struct bgp_prefix {
 349   node buck_node;                       /* Node in per-bucket list */
 350   struct bgp_prefix *next;              /* Node in prefix hash table */
 351   u32 hash;
 352   u32 path_id;
 353   net_addr net[0];
 354 };
 355
 356 struct bgp_bucket {
 357   node send_node;                       /* Node in send queue */
 358   struct bgp_bucket *next;              /* Node in bucket hash table */
 359   list prefixes;                        /* Prefixes in this bucket (struct bgp_prefix) */
 360   u32 hash;                             /* Hash over extended attributes */
 361   ea_list eattrs[0];                    /* Per-bucket extended attributes */
 362 };
 363
 364 struct bgp_export_state {
 365   struct bgp_proto *proto;
 366   struct bgp_channel *channel;
 367   struct linpool *pool;
 368
 369   struct bgp_proto *src;
 370   rte *route;
 371   int mpls;
 372
 373   u32 attrs_seen[1];
 374   uint err_withdraw;
 375 };
 376
 377 struct bgp_write_state {
 378   struct bgp_proto *proto;
 379   struct bgp_channel *channel;
 380   struct linpool *pool;
 381
 382   int mp_reach;
 383   int as4_session;
 384   int add_path;
 385   int mpls;
 386
 387   eattr *mp_next_hop;
 388   adata *mpls_labels;
 389 };
 390
 391 struct bgp_parse_state {
 392   struct bgp_proto *proto;
 393   struct bgp_channel *channel;
 394   struct linpool *pool;
 395
 396   int as4_session;
 397   int add_path;
 398   int mpls;
 399
 400   u32 attrs_seen[256/32];
 401
 402   u32 mp_reach_af;
 403   u32 mp_unreach_af;
 404
 405   uint attr_len;
 406   uint ip_reach_len;
 407   uint ip_unreach_len;
 408   uint ip_next_hop_len;
 409   uint mp_reach_len;
 410   uint mp_unreach_len;
 411   uint mp_next_hop_len;
 412
 413   byte *attrs;
 414   byte *ip_reach_nlri;
 415   byte *ip_unreach_nlri;
 416   byte *ip_next_hop_data;
 417   byte *mp_reach_nlri;
 418   byte *mp_unreach_nlri;
 419   byte *mp_next_hop_data;
 420
 421   uint err_withdraw;
 422   uint err_subcode;
 423   jmp_buf err_jmpbuf;
 424
 425   struct hostentry *hostentry;
 426   adata *mpls_labels;
 427
 428   /* Cached state for bgp_rte_update() */
 429   u32 last_id;
 430   struct rte_src *last_src;
 431   rta *cached_rta;
 432 };
 433
 434 #define BGP_PORT                179
 435 #define BGP_VERSION             4
 436 #define BGP_HEADER_LENGTH       19
 437 #define BGP_MAX_MESSAGE_LENGTH  4096
 438 #define BGP_MAX_EXT_MSG_LENGTH  65535
 439 #define BGP_RX_BUFFER_SIZE      4096
 440 #define BGP_TX_BUFFER_SIZE      4096
 441 #define BGP_RX_BUFFER_EXT_SIZE  65535
 442 #define BGP_TX_BUFFER_EXT_SIZE  65535
 443
 444 static inline int bgp_channel_is_ipv4(struct bgp_channel *c)
 445 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
 446
 447 static inline int bgp_channel_is_ipv6(struct bgp_channel *c)
 448 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
 449
 450 static inline int bgp_cc_is_ipv4(struct bgp_channel_config *c)
 451 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
 452
 453 static inline int bgp_cc_is_ipv6(struct bgp_channel_config *c)
 454 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
 455
 456 static inline uint bgp_max_packet_length(struct bgp_conn *conn)
 457 { return conn->ext_messages ? BGP_MAX_EXT_MSG_LENGTH : BGP_MAX_MESSAGE_LENGTH; }
 458
 459 static inline void
 460 bgp_parse_error(struct bgp_parse_state *s, uint subcode)
 461 {
 462   s->err_subcode = subcode;
 463   longjmp(s->err_jmpbuf, 1);
 464 }
 465
 466 extern struct linpool *bgp_linpool;
 467 extern struct linpool *bgp_linpool2;
 468
 469
 470 void bgp_start_timer(timer *t, uint value);
 471 void bgp_check_config(struct bgp_config *c);
 472 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
 473 void bgp_close_conn(struct bgp_conn *c);
 474 void bgp_update_startup_delay(struct bgp_proto *p);
 475 void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
 476 void bgp_conn_enter_established_state(struct bgp_conn *conn);
 477 void bgp_conn_enter_close_state(struct bgp_conn *conn);
 478 void bgp_conn_enter_idle_state(struct bgp_conn *conn);
 479 void bgp_handle_graceful_restart(struct bgp_proto *p);
 480 void bgp_graceful_restart_done(struct bgp_channel *c);
 481 void bgp_refresh_begin(struct bgp_channel *c);
 482 void bgp_refresh_end(struct bgp_channel *c);
 483 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
 484 void bgp_stop(struct bgp_proto *p, uint subcode, byte *data, uint len);
 485
 486 struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
 487 struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
 488
 489
 490
 491 #ifdef LOCAL_DEBUG
 492 #define BGP_FORCE_DEBUG 1
 493 #else
 494 #define BGP_FORCE_DEBUG 0
 495 #endif
 496 #define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
 497         log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
 498
 499 #define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
 500         log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
 501
 502
 503 /* attrs.c */
 504
 505 static inline eattr *
 506 bgp_find_attr(ea_list *attrs, uint code)
 507 {
 508   return ea_find(attrs, EA_CODE(PROTOCOL_BGP, code));
 509 }
 510
 511 eattr *
 512 bgp_set_attr(ea_list **attrs, struct linpool *pool, uint code, uint flags, uintptr_t val);
 513
 514 static inline void
 515 bgp_set_attr_u32(ea_list **to, struct linpool *pool, uint code, uint flags, u32 val)
 516 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
 517
 518 static inline void
 519 bgp_set_attr_ptr(ea_list **to, struct linpool *pool, uint code, uint flags, struct adata *val)
 520 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
 521
 522 static inline void
 523 bgp_set_attr_data(ea_list **to, struct linpool *pool, uint code, uint flags, void *data, uint len)
 524 {
 525   struct adata *a = lp_alloc_adata(pool, len);
 526   memcpy(a->data, data, len);
 527   bgp_set_attr(to, pool, code, flags, (uintptr_t) a);
 528 }
 529
 530 static inline void
 531 bgp_unset_attr(ea_list **to, struct linpool *pool, uint code)
 532 { eattr *e = bgp_set_attr(to, pool, code, 0, 0); e->type = EAF_TYPE_UNDEF; }
 533
 534
 535 int bgp_encode_attrs(struct bgp_write_state *s, ea_list *attrs, byte *buf, byte *end);
 536 ea_list * bgp_decode_attrs(struct bgp_parse_state *s, byte *data, uint len);
 537
 538 void bgp_init_bucket_table(struct bgp_channel *c);
 539 void bgp_free_bucket_table(struct bgp_channel *c);
 540 void bgp_free_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 541 void bgp_defer_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 542 void bgp_withdraw_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 543
 544 void bgp_init_prefix_table(struct bgp_channel *c);
 545 void bgp_free_prefix_table(struct bgp_channel *c);
 546 void bgp_free_prefix(struct bgp_channel *c, struct bgp_prefix *bp);
 547
 548 int bgp_rte_better(struct rte *, struct rte *);
 549 int bgp_rte_mergable(rte *pri, rte *sec);
 550 int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
 551 struct rte *bgp_rte_modify_stale(struct rte *r, struct linpool *pool);
 552 void bgp_rt_notify(struct proto *P, struct channel *C, net *n, rte *new, rte *old);
 553 int bgp_preexport(struct proto *, struct rte **, struct linpool *);
 554 int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
 555 void bgp_get_route_info(struct rte *, byte *buf);
 556
 557
 558 /* packets.c */
 559
 560 void bgp_dump_state_change(struct bgp_conn *conn, uint old, uint new);
 561 void bgp_prepare_capabilities(struct bgp_conn *conn);
 562 const struct bgp_af_desc *bgp_get_af_desc(u32 afi);
 563 const struct bgp_af_caps *bgp_find_af_caps(struct bgp_caps *caps, u32 afi);
 564 void bgp_schedule_packet(struct bgp_conn *conn, struct bgp_channel *c, int type);
 565 void bgp_kick_tx(void *vconn);
 566 void bgp_tx(struct birdsock *sk);
 567 int bgp_rx(struct birdsock *sk, uint size);
 568 const char * bgp_error_dsc(unsigned code, unsigned subcode);
 569 void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
 570
 571 void bgp_update_next_hop(struct bgp_export_state *s, eattr *a, ea_list **to);
 572
 573
 574 /* Packet types */
 575
 576 #define PKT_OPEN                0x01
 577 #define PKT_UPDATE              0x02
 578 #define PKT_NOTIFICATION        0x03
 579 #define PKT_KEEPALIVE           0x04
 580 #define PKT_ROUTE_REFRESH       0x05    /* [RFC2918] */
 581 #define PKT_BEGIN_REFRESH       0x1e    /* Dummy type for BoRR packet [RFC7313] */
 582 #define PKT_SCHEDULE_CLOSE      0x1f    /* Used internally to schedule socket close */
 583
 584 /* Attributes */
 585
 586 #define BAF_OPTIONAL            0x80
 587 #define BAF_TRANSITIVE          0x40
 588 #define BAF_PARTIAL             0x20
 589 #define BAF_EXT_LEN             0x10
 590
 591 #define BA_ORIGIN               0x01    /* RFC 4271 */          /* WM */
 592 #define BA_AS_PATH              0x02                            /* WM */
 593 #define BA_NEXT_HOP             0x03                            /* WM */
 594 #define BA_MULTI_EXIT_DISC      0x04                            /* ON */
 595 #define BA_LOCAL_PREF           0x05                            /* WD */
 596 #define BA_ATOMIC_AGGR          0x06                            /* WD */
 597 #define BA_AGGREGATOR           0x07                            /* OT */
 598 #define BA_COMMUNITY            0x08    /* RFC 1997 */          /* OT */
 599 #define BA_ORIGINATOR_ID        0x09    /* RFC 4456 */          /* ON */
 600 #define BA_CLUSTER_LIST         0x0a    /* RFC 4456 */          /* ON */
 601 #define BA_MP_REACH_NLRI        0x0e    /* RFC 4760 */
 602 #define BA_MP_UNREACH_NLRI      0x0f    /* RFC 4760 */
 603 #define BA_EXT_COMMUNITY        0x10    /* RFC 4360 */
 604 #define BA_AS4_PATH             0x11    /* RFC 6793 */
 605 #define BA_AS4_AGGREGATOR       0x12    /* RFC 6793 */
 606 #define BA_LARGE_COMMUNITY      0x20    /* RFC 8092 */
 607
 608 /* Bird's private internal BGP attributes */
 609 #define BA_MPLS_LABEL_STACK     0xfe    /* MPLS label stack transfer attribute */
 610
 611 /* BGP connection states */
 612
 613 #define BS_IDLE                 0
 614 #define BS_CONNECT              1       /* Attempting to connect */
 615 #define BS_ACTIVE               2       /* Waiting for connection retry & listening */
 616 #define BS_OPENSENT             3
 617 #define BS_OPENCONFIRM          4
 618 #define BS_ESTABLISHED          5
 619 #define BS_CLOSE                6       /* Used during transition to BS_IDLE */
 620
 621 #define BS_MAX                  7
 622
 623 /* BGP start states
 624  *
 625  * Used in PS_START for fine-grained specification of starting state.
 626  *
 627  * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP
 628  * protocol done what is neccessary to start itself (like acquiring the lock),
 629  * it goes to BSS_CONNECT.
 630  */
 631
 632 #define BSS_PREPARE             0       /* Used before ordinary BGP started, i. e. waiting for lock */
 633 #define BSS_DELAY               1       /* Startup delay due to previous errors */
 634 #define BSS_CONNECT             2       /* Ordinary BGP connecting */
 635
 636
 637 /* BGP feed states (TX)
 638  *
 639  * RFC 4724 specifies that an initial feed should end with End-of-RIB mark.
 640  *
 641  * RFC 7313 specifies that a route refresh should be demarcated by BoRR and EoRR packets.
 642  *
 643  * These states (stored in c->feed_state) are used to keep track of these
 644  * requirements. When such feed is started, BFS_LOADING / BFS_REFRESHING is
 645  * set. When it ended, BFS_LOADED / BFS_REFRESHED is set to schedule End-of-RIB
 646  * or EoRR packet. When the packet is sent, the state returned to BFS_NONE.
 647  *
 648  * Note that when a non-demarcated feed (e.g. plain RFC 4271 initial load
 649  * without End-of-RIB or plain RFC 2918 route refresh without BoRR/EoRR
 650  * demarcation) is active, BFS_NONE is set.
 651  *
 652  * BFS_NONE, BFS_LOADING and BFS_REFRESHING are also used as load states (RX)
 653  * with correspondent semantics (-, expecting End-of-RIB, expecting EoRR).
 654  */
 655
 656 #define BFS_NONE                0       /* No feed or original non-demarcated feed */
 657 #define BFS_LOADING             1       /* Initial feed active, End-of-RIB planned */
 658 #define BFS_LOADED              2       /* Loading done, End-of-RIB marker scheduled */
 659 #define BFS_REFRESHING          3       /* Route refresh (introduced by BoRR) active */
 660 #define BFS_REFRESHED           4       /* Refresh done, EoRR packet scheduled */
 661
 662
 663 /* Error classes */
 664
 665 #define BE_NONE                 0
 666 #define BE_MISC                 1       /* Miscellaneous error */
 667 #define BE_SOCKET               2       /* Socket error */
 668 #define BE_BGP_RX               3       /* BGP protocol error notification received */
 669 #define BE_BGP_TX               4       /* BGP protocol error notification sent */
 670 #define BE_AUTO_DOWN            5       /* Automatic shutdown */
 671 #define BE_MAN_DOWN             6       /* Manual shutdown */
 672
 673 /* Misc error codes */
 674
 675 #define BEM_NEIGHBOR_LOST       1
 676 #define BEM_INVALID_NEXT_HOP    2
 677 #define BEM_INVALID_MD5         3       /* MD5 authentication kernel request failed (possibly not supported) */
 678 #define BEM_NO_SOCKET           4
 679 #define BEM_LINK_DOWN           5
 680 #define BEM_BFD_DOWN            6
 681 #define BEM_GRACEFUL_RESTART    7
 682
 683 /* Automatic shutdown error codes */
 684
 685 #define BEA_ROUTE_LIMIT_EXCEEDED 1
 686
 687 /* Well-known communities */
 688
 689 #define BGP_COMM_NO_EXPORT              0xffffff01      /* Don't export outside local AS / confed. */
 690 #define BGP_COMM_NO_ADVERTISE           0xffffff02      /* Don't export at all */
 691 #define BGP_COMM_NO_EXPORT_SUBCONFED    0xffffff03      /* NO_EXPORT even in local confederation */
 692
 693 #define BGP_COMM_LLGR_STALE             0xffff0006      /* Route is stale according to LLGR */
 694 #define BGP_COMM_NO_LLGR                0xffff0007      /* Do not treat the route according to LLGR */
 695
 696 /* Origins */
 697
 698 #define ORIGIN_IGP              0
 699 #define ORIGIN_EGP              1
 700 #define ORIGIN_INCOMPLETE       2
 701
 702
 703 #endif