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 allow_as_sets;                    /* Allow AS_SETs in incoming AS_PATHs */
 111   int enforce_first_as;                 /* Enable check for neighbor AS as first AS in AS_PATH */
 112   int gr_mode;                          /* Graceful restart mode (BGP_GR_*) */
 113   int llgr_mode;                        /* Long-lived graceful restart mode (BGP_LLGR_*) */
 114   int setkey;                           /* Set MD5 password to system SA/SP database */
 115   /* Times below are in seconds */
 116   unsigned gr_time;                     /* Graceful restart timeout */
 117   unsigned llgr_time;                   /* Long-lived graceful restart stale time */
 118   unsigned connect_delay_time;          /* Minimum delay between connect attempts */
 119   unsigned connect_retry_time;          /* Timeout for connect attempts */
 120   unsigned hold_time, initial_hold_time;
 121   unsigned keepalive_time;
 122   unsigned error_amnesia_time;          /* Errors are forgotten after */
 123   unsigned error_delay_time_min;        /* Time to wait after an error is detected */
 124   unsigned error_delay_time_max;
 125   unsigned disable_after_error;         /* Disable the protocol when error is detected */
 126   u32 disable_after_cease;              /* Disable it when cease is received, bitfield */
 127
 128   char *password;                       /* Password used for MD5 authentication */
 129   net_addr *remote_range;               /* Allowed neighbor range for dynamic BGP */
 130   char *dynamic_name;                   /* Name pattern for dynamic BGP */
 131   int dynamic_name_digits;              /* Minimum number of digits for dynamic names */
 132   int check_link;                       /* Use iface link state for liveness detection */
 133   int bfd;                              /* Use BFD for liveness detection */
 134 };
 135
 136 struct bgp_channel_config {
 137   struct channel_config c;
 138
 139   u32 afi;
 140   const struct bgp_af_desc *desc;
 141
 142   ip_addr next_hop_addr;                /* Local address for NEXT_HOP attribute */
 143   u8 next_hop_self;                     /* Always set next hop to local IP address (NH_*) */
 144   u8 next_hop_keep;                     /* Do not modify next hop attribute (NH_*) */
 145   u8 mandatory;                         /* Channel is mandatory in capability negotiation */
 146   u8 missing_lladdr;                    /* What we will do when we don' know link-local addr, see MLL_* */
 147   u8 gw_mode;                           /* How we compute route gateway from next_hop attr, see GW_* */
 148   u8 secondary;                         /* Accept also non-best routes (i.e. RA_ACCEPTED) */
 149   u8 gr_able;                           /* Allow full graceful restart for the channel */
 150   u8 llgr_able;                         /* Allow full long-lived GR for the channel */
 151   uint llgr_time;                       /* Long-lived graceful restart stale time */
 152   u8 ext_next_hop;                      /* Allow both IPv4 and IPv6 next hops */
 153   u8 add_path;                          /* Use ADD-PATH extension [RFC 7911] */
 154   u8 aigp;                              /* AIGP is allowed on this session */
 155   u8 aigp_originate;                    /* AIGP is originated automatically */
 156   u32 cost;                             /* IGP cost for direct next hops */
 157   u8 import_table;                      /* Use c.in_table as Adj-RIB-In */
 158   u8 export_table;                      /* Use c.out_table as Adj-RIB-Out */
 159
 160   struct rtable_config *igp_table_ip4;  /* Table for recursive IPv4 next hop lookups */
 161   struct rtable_config *igp_table_ip6;  /* Table for recursive IPv6 next hop lookups */
 162 };
 163
 164 #define BGP_PT_INTERNAL         1
 165 #define BGP_PT_EXTERNAL         2
 166
 167 #define NH_NO                   0
 168 #define NH_ALL                  1
 169 #define NH_IBGP                 2
 170 #define NH_EBGP                 3
 171
 172 #define MLL_SELF                1
 173 #define MLL_DROP                2
 174 #define MLL_IGNORE              3
 175
 176 #define GW_DIRECT               1
 177 #define GW_RECURSIVE            2
 178
 179 #define BGP_ADD_PATH_RX         1
 180 #define BGP_ADD_PATH_TX         2
 181 #define BGP_ADD_PATH_FULL       3
 182
 183 #define BGP_GR_ABLE             1
 184 #define BGP_GR_AWARE            2
 185
 186 /* For GR capability common flags */
 187 #define BGP_GRF_RESTART 0x80
 188
 189 /* For GR capability per-AF flags */
 190 #define BGP_GRF_FORWARDING 0x80
 191
 192 #define BGP_LLGR_ABLE           1
 193 #define BGP_LLGR_AWARE          2
 194
 195 #define BGP_LLGRF_FORWARDING 0x80
 196
 197 #define BGP_GRS_NONE            0       /* No GR  */
 198 #define BGP_GRS_ACTIVE          1       /* Graceful restart per RFC 4724 */
 199 #define BGP_GRS_LLGR            2       /* Long-lived GR phase (stale timer active) */
 200
 201 #define BGP_BFD_GRACEFUL        2       /* BFD down triggers graceful restart */
 202
 203
 204 struct bgp_af_caps {
 205   u32 afi;
 206   u8 ready;                             /* Multiprotocol capability, RFC 4760 */
 207   u8 gr_able;                           /* Graceful restart support, RFC 4724 */
 208   u8 gr_af_flags;                       /* Graceful restart per-AF flags */
 209   u8 llgr_able;                         /* Long-lived GR, RFC draft */
 210   u32 llgr_time;                        /* Long-lived GR stale time */
 211   u8 llgr_flags;                        /* Long-lived GR per-AF flags */
 212   u8 ext_next_hop;                      /* Extended IPv6 next hop,   RFC 5549 */
 213   u8 add_path;                          /* Multiple paths support,   RFC 7911 */
 214 };
 215
 216 struct bgp_caps {
 217   u32 as4_number;                       /* Announced ASN */
 218
 219   u8 as4_support;                       /* Four-octet AS capability, RFC 6793 */
 220   u8 ext_messages;                      /* Extended message length,  RFC draft */
 221   u8 route_refresh;                     /* Route refresh capability, RFC 2918 */
 222   u8 enhanced_refresh;                  /* Enhanced route refresh,   RFC 7313 */
 223
 224   u8 gr_aware;                          /* Graceful restart capability, RFC 4724 */
 225   u8 gr_flags;                          /* Graceful restart flags */
 226   u16 gr_time;                          /* Graceful restart time in seconds */
 227
 228   u8 llgr_aware;                        /* Long-lived GR capability, RFC draft */
 229   u8 any_ext_next_hop;                  /* Bitwise OR of per-AF ext_next_hop */
 230   u8 any_add_path;                      /* Bitwise OR of per-AF add_path */
 231
 232   u16 af_count;                         /* Number of af_data items */
 233   u16 length;                           /* Length of capabilities in OPEN msg */
 234
 235   struct bgp_af_caps af_data[0];        /* Per-AF capability data */
 236 };
 237
 238 #define WALK_AF_CAPS(caps,ac) \
 239   for (ac = caps->af_data; ac < &caps->af_data[caps->af_count]; ac++)
 240
 241
 242 struct bgp_socket {
 243   node n;                               /* Node in global bgp_sockets */
 244   sock *sk;                             /* Real listening socket */
 245   u32 uc;                               /* Use count */
 246 };
 247
 248 struct bgp_conn {
 249   struct bgp_proto *bgp;
 250   struct birdsock *sk;
 251   u8 state;                             /* State of connection state machine */
 252   u8 as4_session;                       /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
 253   u8 ext_messages;                      /* Session uses extended message length */
 254   u32 received_as;                      /* ASN received in OPEN message */
 255
 256   struct bgp_caps *local_caps;
 257   struct bgp_caps *remote_caps;
 258   timer *connect_timer;
 259   timer *hold_timer;
 260   timer *keepalive_timer;
 261   event *tx_ev;
 262   u32 packets_to_send;                  /* Bitmap of packet types to be sent */
 263   u32 channels_to_send;                 /* Bitmap of channels with packets to be sent */
 264   u8 last_channel;                      /* Channel used last time for TX */
 265   u8 last_channel_count;                /* Number of times the last channel was used in succession */
 266   int notify_code, notify_subcode, notify_size;
 267   byte *notify_data;
 268
 269   uint hold_time, keepalive_time;       /* Times calculated from my and neighbor's requirements */
 270 };
 271
 272 struct bgp_proto {
 273   struct proto p;
 274   const struct bgp_config *cf;          /* Shortcut to BGP configuration */
 275   ip_addr local_ip, remote_ip;
 276   u32 local_as, remote_as;
 277   u32 public_as;                        /* Externally visible ASN (local_as or confederation id) */
 278   u32 local_id;                         /* BGP identifier of this router */
 279   u32 remote_id;                        /* BGP identifier of the neighbor */
 280   u32 rr_cluster_id;                    /* Route reflector cluster ID */
 281   u8 start_state;                       /* Substates that partitions BS_START */
 282   u8 is_internal;                       /* Internal BGP session (local_as == remote_as) */
 283   u8 is_interior;                       /* Internal or intra-confederation BGP session */
 284   u8 as4_session;                       /* Session uses 4B AS numbers in AS_PATH (both sides support it) */
 285   u8 rr_client;                         /* Whether neighbor is RR client of me */
 286   u8 rs_client;                         /* Whether neighbor is RS client of me */
 287   u8 ipv4;                              /* Use IPv4 connection, i.e. remote_ip is IPv4 */
 288   u8 passive;                           /* Do not initiate outgoing connection */
 289   u8 route_refresh;                     /* Route refresh allowed to send [RFC 2918] */
 290   u8 enhanced_refresh;                  /* Enhanced refresh is negotiated [RFC 7313] */
 291   u8 gr_ready;                          /* Neighbor could do graceful restart */
 292   u8 llgr_ready;                        /* Neighbor could do Long-lived GR, implies gr_ready */
 293   u8 gr_active_num;                     /* Neighbor is doing GR, number of active channels */
 294   u8 channel_count;                     /* Number of active channels */
 295   u8 summary_add_path_rx;               /* Summary state of ADD_PATH RX w.r.t active channels */
 296   u32 *afi_map;                         /* Map channel index -> AFI */
 297   struct bgp_channel **channel_map;     /* Map channel index -> channel */
 298   struct bgp_conn *conn;                /* Connection we have established */
 299   struct bgp_conn outgoing_conn;        /* Outgoing connection we're working with */
 300   struct bgp_conn incoming_conn;        /* Incoming connection we have neither accepted nor rejected yet */
 301   struct object_lock *lock;             /* Lock for neighbor connection */
 302   struct neighbor *neigh;               /* Neighbor entry corresponding to remote ip, NULL if multihop */
 303   struct bgp_socket *sock;              /* Shared listening socket */
 304   struct bfd_request *bfd_req;          /* BFD request, if BFD is used */
 305   struct birdsock *postponed_sk;        /* Postponed incoming socket for dynamic BGP */
 306   ip_addr link_addr;                    /* Link-local version of local_ip */
 307   event *event;                         /* Event for respawning and shutting process */
 308   timer *startup_timer;                 /* Timer used to delay protocol startup due to previous errors (startup_delay) */
 309   timer *gr_timer;                      /* Timer waiting for reestablishment after graceful restart */
 310   int dynamic_name_counter;             /* Counter for dynamic BGP names */
 311   uint startup_delay;                   /* Delay (in seconds) of protocol startup due to previous errors */
 312   btime last_proto_error;               /* Time of last error that leads to protocol stop */
 313   u8 last_error_class;                  /* Error class of last error */
 314   u32 last_error_code;                  /* Error code of last error. BGP protocol errors
 315                                            are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
 316 };
 317
 318 struct bgp_channel {
 319   struct channel c;
 320
 321   /* Rest are BGP specific data */
 322   struct bgp_channel_config *cf;
 323
 324   u32 afi;
 325   u32 index;
 326   const struct bgp_af_desc *desc;
 327
 328   rtable *igp_table_ip4;                /* Table for recursive IPv4 next hop lookups */
 329   rtable *igp_table_ip6;                /* Table for recursive IPv6 next hop lookups */
 330
 331   /* Rest are zeroed when down */
 332   pool *pool;
 333   HASH(struct bgp_bucket) bucket_hash;  /* Hash table of route buckets */
 334   struct bgp_bucket *withdraw_bucket;   /* Withdrawn routes */
 335   list bucket_queue;                    /* Queue of buckets to send (struct bgp_bucket) */
 336
 337   HASH(struct bgp_prefix) prefix_hash;  /* Prefixes to be sent */
 338   slab *prefix_slab;                    /* Slab holding prefix nodes */
 339
 340   ip_addr next_hop_addr;                /* Local address for NEXT_HOP attribute */
 341   ip_addr link_addr;                    /* Link-local version of next_hop_addr */
 342
 343   u32 packets_to_send;                  /* Bitmap of packet types to be sent */
 344
 345   u8 ext_next_hop;                      /* Session allows both IPv4 and IPv6 next hops */
 346
 347   u8 gr_ready;                          /* Neighbor could do GR on this AF */
 348   u8 gr_active;                         /* Neighbor is doing GR (BGP_GRS_*) */
 349
 350   timer *stale_timer;                   /* Long-lived stale timer for LLGR */
 351   u32 stale_time;                       /* Stored LLGR stale time from last session */
 352
 353   u8 add_path_rx;                       /* Session expects receive of ADD-PATH extended NLRI */
 354   u8 add_path_tx;                       /* Session expects transmit of ADD-PATH extended NLRI */
 355
 356   u8 feed_state;                        /* Feed state (TX) for EoR, RR packets, see BFS_* */
 357   u8 load_state;                        /* Load state (RX) for EoR, RR packets, see BFS_* */
 358 };
 359
 360 struct bgp_prefix {
 361   node buck_node;                       /* Node in per-bucket list */
 362   struct bgp_prefix *next;              /* Node in prefix hash table */
 363   u32 hash;
 364   u32 path_id;
 365   net_addr net[0];
 366 };
 367
 368 struct bgp_bucket {
 369   node send_node;                       /* Node in send queue */
 370   struct bgp_bucket *next;              /* Node in bucket hash table */
 371   list prefixes;                        /* Prefixes in this bucket (struct bgp_prefix) */
 372   u32 hash;                             /* Hash over extended attributes */
 373   ea_list eattrs[0];                    /* Per-bucket extended attributes */
 374 };
 375
 376 struct bgp_export_state {
 377   struct bgp_proto *proto;
 378   struct bgp_channel *channel;
 379   struct linpool *pool;
 380
 381   struct bgp_proto *src;
 382   rte *route;
 383   int mpls;
 384
 385   u32 attrs_seen[1];
 386   uint err_withdraw;
 387   uint local_next_hop;
 388 };
 389
 390 struct bgp_write_state {
 391   struct bgp_proto *proto;
 392   struct bgp_channel *channel;
 393   struct linpool *pool;
 394
 395   int mp_reach;
 396   int as4_session;
 397   int add_path;
 398   int mpls;
 399
 400   eattr *mp_next_hop;
 401   const adata *mpls_labels;
 402 };
 403
 404 struct bgp_parse_state {
 405   struct bgp_proto *proto;
 406   struct bgp_channel *channel;
 407   struct linpool *pool;
 408
 409   int as4_session;
 410   int add_path;
 411   int mpls;
 412
 413   u32 attrs_seen[256/32];
 414
 415   u32 mp_reach_af;
 416   u32 mp_unreach_af;
 417
 418   uint attr_len;
 419   uint ip_reach_len;
 420   uint ip_unreach_len;
 421   uint ip_next_hop_len;
 422   uint mp_reach_len;
 423   uint mp_unreach_len;
 424   uint mp_next_hop_len;
 425
 426   byte *attrs;
 427   byte *ip_reach_nlri;
 428   byte *ip_unreach_nlri;
 429   byte *ip_next_hop_data;
 430   byte *mp_reach_nlri;
 431   byte *mp_unreach_nlri;
 432   byte *mp_next_hop_data;
 433
 434   uint err_withdraw;
 435   uint err_subcode;
 436   jmp_buf err_jmpbuf;
 437
 438   struct hostentry *hostentry;
 439   adata *mpls_labels;
 440
 441   /* Cached state for bgp_rte_update() */
 442   u32 last_id;
 443   struct rte_src *last_src;
 444   rta *cached_rta;
 445 };
 446
 447 #define BGP_PORT                179
 448 #define BGP_VERSION             4
 449 #define BGP_HEADER_LENGTH       19
 450 #define BGP_MAX_MESSAGE_LENGTH  4096
 451 #define BGP_MAX_EXT_MSG_LENGTH  65535
 452 #define BGP_RX_BUFFER_SIZE      4096
 453 #define BGP_TX_BUFFER_SIZE      4096
 454 #define BGP_RX_BUFFER_EXT_SIZE  65535
 455 #define BGP_TX_BUFFER_EXT_SIZE  65535
 456
 457 static inline int bgp_channel_is_ipv4(struct bgp_channel *c)
 458 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
 459
 460 static inline int bgp_channel_is_ipv6(struct bgp_channel *c)
 461 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
 462
 463 static inline int bgp_cc_is_ipv4(struct bgp_channel_config *c)
 464 { return BGP_AFI(c->afi) == BGP_AFI_IPV4; }
 465
 466 static inline int bgp_cc_is_ipv6(struct bgp_channel_config *c)
 467 { return BGP_AFI(c->afi) == BGP_AFI_IPV6; }
 468
 469 static inline uint bgp_max_packet_length(struct bgp_conn *conn)
 470 { return conn->ext_messages ? BGP_MAX_EXT_MSG_LENGTH : BGP_MAX_MESSAGE_LENGTH; }
 471
 472 static inline void
 473 bgp_parse_error(struct bgp_parse_state *s, uint subcode)
 474 {
 475   s->err_subcode = subcode;
 476   longjmp(s->err_jmpbuf, 1);
 477 }
 478
 479 extern struct linpool *bgp_linpool;
 480 extern struct linpool *bgp_linpool2;
 481
 482
 483 void bgp_start_timer(timer *t, uint value);
 484 void bgp_check_config(struct bgp_config *c);
 485 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
 486 void bgp_close_conn(struct bgp_conn *c);
 487 void bgp_update_startup_delay(struct bgp_proto *p);
 488 void bgp_conn_enter_openconfirm_state(struct bgp_conn *conn);
 489 void bgp_conn_enter_established_state(struct bgp_conn *conn);
 490 void bgp_conn_enter_close_state(struct bgp_conn *conn);
 491 void bgp_conn_enter_idle_state(struct bgp_conn *conn);
 492 void bgp_handle_graceful_restart(struct bgp_proto *p);
 493 void bgp_graceful_restart_done(struct bgp_channel *c);
 494 void bgp_refresh_begin(struct bgp_channel *c);
 495 void bgp_refresh_end(struct bgp_channel *c);
 496 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
 497 void bgp_stop(struct bgp_proto *p, int subcode, byte *data, uint len);
 498
 499 struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
 500 struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
 501
 502 static inline int
 503 rte_resolvable(rte *rt)
 504 {
 505   return rt->attrs->dest == RTD_UNICAST;
 506 }
 507
 508
 509 #ifdef LOCAL_DEBUG
 510 #define BGP_FORCE_DEBUG 1
 511 #else
 512 #define BGP_FORCE_DEBUG 0
 513 #endif
 514 #define BGP_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
 515         log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
 516
 517 #define BGP_TRACE_RL(rl, flags, msg, args...) do { if ((p->p.debug & flags) || BGP_FORCE_DEBUG) \
 518         log_rl(rl, L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
 519
 520
 521 /* attrs.c */
 522
 523 static inline eattr *
 524 bgp_find_attr(ea_list *attrs, uint code)
 525 {
 526   return ea_find(attrs, EA_CODE(PROTOCOL_BGP, code));
 527 }
 528
 529 eattr *
 530 bgp_set_attr(ea_list **attrs, struct linpool *pool, uint code, uint flags, uintptr_t val);
 531
 532 static inline void
 533 bgp_set_attr_u32(ea_list **to, struct linpool *pool, uint code, uint flags, u32 val)
 534 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
 535
 536 static inline void
 537 bgp_set_attr_ptr(ea_list **to, struct linpool *pool, uint code, uint flags, const struct adata *val)
 538 { bgp_set_attr(to, pool, code, flags, (uintptr_t) val); }
 539
 540 static inline void
 541 bgp_set_attr_data(ea_list **to, struct linpool *pool, uint code, uint flags, void *data, uint len)
 542 {
 543   struct adata *a = lp_alloc_adata(pool, len);
 544   memcpy(a->data, data, len);
 545   bgp_set_attr(to, pool, code, flags, (uintptr_t) a);
 546 }
 547
 548 static inline void
 549 bgp_unset_attr(ea_list **to, struct linpool *pool, uint code)
 550 { eattr *e = bgp_set_attr(to, pool, code, 0, 0); e->type = EAF_TYPE_UNDEF; }
 551
 552
 553 int bgp_encode_attrs(struct bgp_write_state *s, ea_list *attrs, byte *buf, byte *end);
 554 ea_list * bgp_decode_attrs(struct bgp_parse_state *s, byte *data, uint len);
 555 void bgp_finish_attrs(struct bgp_parse_state *s, rta *a);
 556
 557 void bgp_init_bucket_table(struct bgp_channel *c);
 558 void bgp_free_bucket_table(struct bgp_channel *c);
 559 void bgp_free_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 560 void bgp_defer_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 561 void bgp_withdraw_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 562
 563 void bgp_init_prefix_table(struct bgp_channel *c);
 564 void bgp_free_prefix_table(struct bgp_channel *c);
 565 void bgp_free_prefix(struct bgp_channel *c, struct bgp_prefix *bp);
 566
 567 int bgp_rte_better(struct rte *, struct rte *);
 568 int bgp_rte_mergable(rte *pri, rte *sec);
 569 int bgp_rte_recalculate(rtable *table, net *net, rte *new, rte *old, rte *old_best);
 570 struct rte *bgp_rte_modify_stale(struct rte *r, struct linpool *pool);
 571 void bgp_rt_notify(struct proto *P, struct channel *C, net *n, rte *new, rte *old);
 572 int bgp_preexport(struct proto *, struct rte **, struct linpool *);
 573 int bgp_get_attr(struct eattr *e, byte *buf, int buflen);
 574 void bgp_get_route_info(struct rte *, byte *buf);
 575 int bgp_total_aigp_metric_(rte *e, u64 *metric, const struct adata **ad);
 576
 577 #define BGP_AIGP_METRIC         1
 578 #define BGP_AIGP_MAX            U64(0xffffffffffffffff)
 579
 580 static inline u64
 581 bgp_total_aigp_metric(rte *r)
 582 {
 583   u64 metric = BGP_AIGP_MAX;
 584   const struct adata *ad;
 585
 586   bgp_total_aigp_metric_(r, &metric, &ad);
 587   return metric;
 588 }
 589
 590
 591 /* packets.c */
 592
 593 void bgp_dump_state_change(struct bgp_conn *conn, uint old, uint new);
 594 void bgp_prepare_capabilities(struct bgp_conn *conn);
 595 const struct bgp_af_desc *bgp_get_af_desc(u32 afi);
 596 const struct bgp_af_caps *bgp_find_af_caps(struct bgp_caps *caps, u32 afi);
 597 void bgp_schedule_packet(struct bgp_conn *conn, struct bgp_channel *c, int type);
 598 void bgp_kick_tx(void *vconn);
 599 void bgp_tx(struct birdsock *sk);
 600 int bgp_rx(struct birdsock *sk, uint size);
 601 const char * bgp_error_dsc(unsigned code, unsigned subcode);
 602 void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsigned subcode, byte *data, unsigned len);
 603
 604 void bgp_update_next_hop(struct bgp_export_state *s, eattr *a, ea_list **to);
 605
 606
 607 /* Packet types */
 608
 609 #define PKT_OPEN                0x01
 610 #define PKT_UPDATE              0x02
 611 #define PKT_NOTIFICATION        0x03
 612 #define PKT_KEEPALIVE           0x04
 613 #define PKT_ROUTE_REFRESH       0x05    /* [RFC2918] */
 614 #define PKT_BEGIN_REFRESH       0x1e    /* Dummy type for BoRR packet [RFC7313] */
 615 #define PKT_SCHEDULE_CLOSE      0x1f    /* Used internally to schedule socket close */
 616
 617 /* Attributes */
 618
 619 #define BAF_OPTIONAL            0x80
 620 #define BAF_TRANSITIVE          0x40
 621 #define BAF_PARTIAL             0x20
 622 #define BAF_EXT_LEN             0x10
 623
 624 #define BAF_DECODE_FLAGS        0x0100  /* Private flag - attribute flags are handled by the decode hook */
 625
 626 #define BA_ORIGIN               0x01    /* RFC 4271 */          /* WM */
 627 #define BA_AS_PATH              0x02                            /* WM */
 628 #define BA_NEXT_HOP             0x03                            /* WM */
 629 #define BA_MULTI_EXIT_DISC      0x04                            /* ON */
 630 #define BA_LOCAL_PREF           0x05                            /* WD */
 631 #define BA_ATOMIC_AGGR          0x06                            /* WD */
 632 #define BA_AGGREGATOR           0x07                            /* OT */
 633 #define BA_COMMUNITY            0x08    /* RFC 1997 */          /* OT */
 634 #define BA_ORIGINATOR_ID        0x09    /* RFC 4456 */          /* ON */
 635 #define BA_CLUSTER_LIST         0x0a    /* RFC 4456 */          /* ON */
 636 #define BA_MP_REACH_NLRI        0x0e    /* RFC 4760 */
 637 #define BA_MP_UNREACH_NLRI      0x0f    /* RFC 4760 */
 638 #define BA_EXT_COMMUNITY        0x10    /* RFC 4360 */
 639 #define BA_AS4_PATH             0x11    /* RFC 6793 */
 640 #define BA_AS4_AGGREGATOR       0x12    /* RFC 6793 */
 641 #define BA_AIGP                 0x1a    /* RFC 7311 */
 642 #define BA_LARGE_COMMUNITY      0x20    /* RFC 8092 */
 643
 644 /* Bird's private internal BGP attributes */
 645 #define BA_MPLS_LABEL_STACK     0xfe    /* MPLS label stack transfer attribute */
 646
 647 /* BGP connection states */
 648
 649 #define BS_IDLE                 0
 650 #define BS_CONNECT              1       /* Attempting to connect */
 651 #define BS_ACTIVE               2       /* Waiting for connection retry & listening */
 652 #define BS_OPENSENT             3
 653 #define BS_OPENCONFIRM          4
 654 #define BS_ESTABLISHED          5
 655 #define BS_CLOSE                6       /* Used during transition to BS_IDLE */
 656
 657 #define BS_MAX                  7
 658
 659 /* BGP start states
 660  *
 661  * Used in PS_START for fine-grained specification of starting state.
 662  *
 663  * When BGP protocol is started by core, it goes to BSS_PREPARE. When BGP
 664  * protocol done what is neccessary to start itself (like acquiring the lock),
 665  * it goes to BSS_CONNECT.
 666  */
 667
 668 #define BSS_PREPARE             0       /* Used before ordinary BGP started, i. e. waiting for lock */
 669 #define BSS_DELAY               1       /* Startup delay due to previous errors */
 670 #define BSS_CONNECT             2       /* Ordinary BGP connecting */
 671
 672
 673 /* BGP feed states (TX)
 674  *
 675  * RFC 4724 specifies that an initial feed should end with End-of-RIB mark.
 676  *
 677  * RFC 7313 specifies that a route refresh should be demarcated by BoRR and EoRR packets.
 678  *
 679  * These states (stored in c->feed_state) are used to keep track of these
 680  * requirements. When such feed is started, BFS_LOADING / BFS_REFRESHING is
 681  * set. When it ended, BFS_LOADED / BFS_REFRESHED is set to schedule End-of-RIB
 682  * or EoRR packet. When the packet is sent, the state returned to BFS_NONE.
 683  *
 684  * Note that when a non-demarcated feed (e.g. plain RFC 4271 initial load
 685  * without End-of-RIB or plain RFC 2918 route refresh without BoRR/EoRR
 686  * demarcation) is active, BFS_NONE is set.
 687  *
 688  * BFS_NONE, BFS_LOADING and BFS_REFRESHING are also used as load states (RX)
 689  * with correspondent semantics (-, expecting End-of-RIB, expecting EoRR).
 690  */
 691
 692 #define BFS_NONE                0       /* No feed or original non-demarcated feed */
 693 #define BFS_LOADING             1       /* Initial feed active, End-of-RIB planned */
 694 #define BFS_LOADED              2       /* Loading done, End-of-RIB marker scheduled */
 695 #define BFS_REFRESHING          3       /* Route refresh (introduced by BoRR) active */
 696 #define BFS_REFRESHED           4       /* Refresh done, EoRR packet scheduled */
 697
 698
 699 /* Error classes */
 700
 701 #define BE_NONE                 0
 702 #define BE_MISC                 1       /* Miscellaneous error */
 703 #define BE_SOCKET               2       /* Socket error */
 704 #define BE_BGP_RX               3       /* BGP protocol error notification received */
 705 #define BE_BGP_TX               4       /* BGP protocol error notification sent */
 706 #define BE_AUTO_DOWN            5       /* Automatic shutdown */
 707 #define BE_MAN_DOWN             6       /* Manual shutdown */
 708
 709 /* Misc error codes */
 710
 711 #define BEM_NEIGHBOR_LOST       1
 712 #define BEM_INVALID_NEXT_HOP    2
 713 #define BEM_INVALID_MD5         3       /* MD5 authentication kernel request failed (possibly not supported) */
 714 #define BEM_NO_SOCKET           4
 715 #define BEM_LINK_DOWN           5
 716 #define BEM_BFD_DOWN            6
 717 #define BEM_GRACEFUL_RESTART    7
 718
 719 /* Automatic shutdown error codes */
 720
 721 #define BEA_ROUTE_LIMIT_EXCEEDED 1
 722
 723 /* Well-known communities */
 724
 725 #define BGP_COMM_NO_EXPORT              0xffffff01      /* Don't export outside local AS / confed. */
 726 #define BGP_COMM_NO_ADVERTISE           0xffffff02      /* Don't export at all */
 727 #define BGP_COMM_NO_EXPORT_SUBCONFED    0xffffff03      /* NO_EXPORT even in local confederation */
 728
 729 #define BGP_COMM_LLGR_STALE             0xffff0006      /* Route is stale according to LLGR */
 730 #define BGP_COMM_NO_LLGR                0xffff0007      /* Do not treat the route according to LLGR */
 731
 732 /* Origins */
 733
 734 #define ORIGIN_IGP              0
 735 #define ORIGIN_EGP              1
 736 #define ORIGIN_INCOMPLETE       2
 737
 738
 739 #endif