nest/route.h

   1 /*
   2  *      BIRD Internet Routing Daemon -- Routing Table
   3  *
   4  *      (c) 1998--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_ROUTE_H_
  10 #define _BIRD_ROUTE_H_
  11
  12 #include "lib/lists.h"
  13 #include "lib/bitmap.h"
  14 #include "lib/resource.h"
  15 #include "lib/net.h"
  16
  17 struct ea_list;
  18 struct protocol;
  19 struct proto;
  20 struct rte_src;
  21 struct symbol;
  22 struct timer;
  23 struct fib;
  24 struct filter;
  25 struct f_trie;
  26 struct f_trie_walk_state;
  27 struct cli;
  28
  29 /*
  30  *      Generic data structure for storing network prefixes. Also used
  31  *      for the master routing table. Currently implemented as a hash
  32  *      table.
  33  *
  34  *      Available operations:
  35  *              - insertion of new entry
  36  *              - deletion of entry
  37  *              - searching for entry by network prefix
  38  *              - asynchronous retrieval of fib contents
  39  */
  40
  41 struct fib_node {
  42   struct fib_node *next;                /* Next in hash chain */
  43   struct fib_iterator *readers;         /* List of readers of this node */
  44   net_addr addr[0];
  45 };
  46
  47 struct fib_iterator {                   /* See lib/slists.h for an explanation */
  48   struct fib_iterator *prev, *next;     /* Must be synced with struct fib_node! */
  49   byte efef;                            /* 0xff to distinguish between iterator and node */
  50   byte pad[3];
  51   struct fib_node *node;                /* Or NULL if freshly merged */
  52   uint hash;
  53 };
  54
  55 typedef void (*fib_init_fn)(struct fib *, void *);
  56
  57 struct fib {
  58   pool *fib_pool;                       /* Pool holding all our data */
  59   slab *fib_slab;                       /* Slab holding all fib nodes */
  60   struct fib_node **hash_table;         /* Node hash table */
  61   uint hash_size;                       /* Number of hash table entries (a power of two) */
  62   uint hash_order;                      /* Binary logarithm of hash_size */
  63   uint hash_shift;                      /* 32 - hash_order */
  64   uint addr_type;                       /* Type of address data stored in fib (NET_*) */
  65   uint node_size;                       /* FIB node size, 0 for nonuniform */
  66   uint node_offset;                     /* Offset of fib_node struct inside of user data */
  67   uint entries;                         /* Number of entries */
  68   uint entries_min, entries_max;        /* Entry count limits (else start rehashing) */
  69   fib_init_fn init;                     /* Constructor */
  70 };
  71
  72 static inline void * fib_node_to_user(struct fib *f, struct fib_node *e)
  73 { return e ? (void *) ((char *) e - f->node_offset) : NULL; }
  74
  75 static inline struct fib_node * fib_user_to_node(struct fib *f, void *e)
  76 { return e ? (void *) ((char *) e + f->node_offset) : NULL; }
  77
  78 void fib_init(struct fib *f, pool *p, uint addr_type, uint node_size, uint node_offset, uint hash_order, fib_init_fn init);
  79 void *fib_find(struct fib *, const net_addr *); /* Find or return NULL if doesn't exist */
  80 void *fib_get_chain(struct fib *f, const net_addr *a); /* Find first node in linked list from hash table */
  81 void *fib_get(struct fib *, const net_addr *);  /* Find or create new if nonexistent */
  82 void *fib_route(struct fib *, const net_addr *); /* Longest-match routing lookup */
  83 void fib_delete(struct fib *, void *);  /* Remove fib entry */
  84 void fib_free(struct fib *);            /* Destroy the fib */
  85 void fib_check(struct fib *);           /* Consistency check for debugging */
  86
  87 void fit_init(struct fib_iterator *, struct fib *); /* Internal functions, don't call */
  88 struct fib_node *fit_get(struct fib *, struct fib_iterator *);
  89 void fit_put(struct fib_iterator *, struct fib_node *);
  90 void fit_put_next(struct fib *f, struct fib_iterator *i, struct fib_node *n, uint hpos);
  91 void fit_put_end(struct fib_iterator *i);
  92 void fit_copy(struct fib *f, struct fib_iterator *dst, struct fib_iterator *src);
  93
  94
  95 #define FIB_WALK(fib, type, z) do {                             \
  96         struct fib_node *fn_, **ff_ = (fib)->hash_table;        \
  97         uint count_ = (fib)->hash_size;                         \
  98         type *z;                                                \
  99         while (count_--)                                        \
 100           for (fn_ = *ff_++; z = fib_node_to_user(fib, fn_); fn_=fn_->next)
 101
 102 #define FIB_WALK_END } while (0)
 103
 104 #define FIB_ITERATE_INIT(it, fib) fit_init(it, fib)
 105
 106 #define FIB_ITERATE_START(fib, it, type, z) do {                \
 107         struct fib_node *fn_ = fit_get(fib, it);                \
 108         uint count_ = (fib)->hash_size;                         \
 109         uint hpos_ = (it)->hash;                                \
 110         type *z;                                                \
 111         for(;;) {                                               \
 112           if (!fn_)                                             \
 113             {                                                   \
 114                if (++hpos_ >= count_)                           \
 115                  break;                                         \
 116                fn_ = (fib)->hash_table[hpos_];                  \
 117                continue;                                        \
 118             }                                                   \
 119           z = fib_node_to_user(fib, fn_);
 120
 121 #define FIB_ITERATE_END fn_ = fn_->next; } } while(0)
 122
 123 #define FIB_ITERATE_PUT(it) fit_put(it, fn_)
 124
 125 #define FIB_ITERATE_PUT_NEXT(it, fib) fit_put_next(fib, it, fn_, hpos_)
 126
 127 #define FIB_ITERATE_PUT_END(it) fit_put_end(it)
 128
 129 #define FIB_ITERATE_UNLINK(it, fib) fit_get(fib, it)
 130
 131 #define FIB_ITERATE_COPY(dst, src, fib) fit_copy(fib, dst, src)
 132
 133
 134 /*
 135  *      Master Routing Tables. Generally speaking, each of them contains a FIB
 136  *      with each entry pointing to a list of route entries representing routes
 137  *      to given network (with the selected one at the head).
 138  *
 139  *      Each of the RTE's contains variable data (the preference and protocol-dependent
 140  *      metrics) and a pointer to a route attribute block common for many routes).
 141  *
 142  *      It's guaranteed that there is at most one RTE for every (prefix,proto) pair.
 143  */
 144
 145 struct rtable_config {
 146   node n;
 147   char *name;
 148   struct rtable *table;
 149   struct proto_config *krt_attached;    /* Kernel syncer attached to this table */
 150   uint addr_type;                       /* Type of address data stored in table (NET_*) */
 151   uint gc_threshold;                    /* Maximum number of operations before GC is run */
 152   uint gc_period;                       /* Approximate time between two consecutive GC runs */
 153   byte sorted;                          /* Routes of network are sorted according to rte_better() */
 154   byte internal;                        /* Internal table of a protocol */
 155   byte trie_used;                       /* Rtable has attached trie */
 156   btime min_settle_time;                /* Minimum settle time for notifications */
 157   btime max_settle_time;                /* Maximum settle time for notifications */
 158 };
 159
 160 typedef struct rtable {
 161   resource r;
 162   node n;                               /* Node in list of all tables */
 163   pool *rp;                             /* Resource pool to allocate everything from, including itself */
 164   struct fib fib;
 165   struct f_trie *trie;                  /* Trie of prefixes defined in fib */
 166   char *name;                           /* Name of this table */
 167   list channels;                        /* List of attached channels (struct channel) */
 168   uint addr_type;                       /* Type of address data stored in table (NET_*) */
 169   int pipe_busy;                        /* Pipe loop detection */
 170   int use_count;                        /* Number of protocols using this table */
 171   u32 rt_count;                         /* Number of routes in the table */
 172
 173   byte internal;                        /* Internal table of a protocol */
 174
 175   struct hmap id_map;
 176   struct hostcache *hostcache;
 177   struct rtable_config *config;         /* Configuration of this table */
 178   struct config *deleted;               /* Table doesn't exist in current configuration,
 179                                          * delete as soon as use_count becomes 0 and remove
 180                                          * obstacle from this routing table.
 181                                          */
 182   struct event *rt_event;               /* Routing table event */
 183   struct timer *prune_timer;            /* Timer for periodic pruning / GC */
 184   btime last_rt_change;                 /* Last time when route changed */
 185   btime base_settle_time;               /* Start time of rtable settling interval */
 186   btime gc_time;                        /* Time of last GC */
 187   uint gc_counter;                      /* Number of operations since last GC */
 188   byte prune_state;                     /* Table prune state, 1 -> scheduled, 2-> running */
 189   byte prune_trie;                      /* Prune prefix trie during next table prune */
 190   byte hcu_scheduled;                   /* Hostcache update is scheduled */
 191   byte nhu_state;                       /* Next Hop Update state */
 192   struct fib_iterator prune_fit;        /* Rtable prune FIB iterator */
 193   struct fib_iterator nhu_fit;          /* Next Hop Update FIB iterator */
 194   struct f_trie *trie_new;              /* New prefix trie defined during pruning */
 195   struct f_trie *trie_old;              /* Old prefix trie waiting to be freed */
 196   u32 trie_lock_count;                  /* Prefix trie locked by walks */
 197   u32 trie_old_lock_count;              /* Old prefix trie locked by walks */
 198
 199   list subscribers;                     /* Subscribers for notifications */
 200   struct timer *settle_timer;           /* Settle time for notifications */
 201   list flowspec_links;                  /* List of flowspec links, src for NET_IPx and dst for NET_FLOWx */
 202   struct f_trie *flowspec_trie;         /* Trie for evaluation of flowspec notifications */
 203 } rtable;
 204
 205 struct rt_subscription {
 206   node n;
 207   rtable *tab;
 208   void (*hook)(struct rt_subscription *b);
 209   void *data;
 210 };
 211
 212 struct rt_flowspec_link {
 213   node n;
 214   rtable *src;
 215   rtable *dst;
 216   u32 uc;
 217 };
 218
 219 #define NHU_CLEAN       0
 220 #define NHU_SCHEDULED   1
 221 #define NHU_RUNNING     2
 222 #define NHU_DIRTY       3
 223
 224 typedef struct network {
 225   struct rte *routes;                   /* Available routes for this network */
 226   struct fib_node n;                    /* FIB flags reserved for kernel syncer */
 227 } net;
 228
 229 struct hostcache {
 230   slab *slab;                           /* Slab holding all hostentries */
 231   struct hostentry **hash_table;        /* Hash table for hostentries */
 232   unsigned hash_order, hash_shift;
 233   unsigned hash_max, hash_min;
 234   unsigned hash_items;
 235   linpool *lp;                          /* Linpool for trie */
 236   struct f_trie *trie;                  /* Trie of prefixes that might affect hostentries */
 237   list hostentries;                     /* List of all hostentries */
 238   byte update_hostcache;
 239 };
 240
 241 struct hostentry {
 242   node ln;
 243   ip_addr addr;                         /* IP address of host, part of key */
 244   ip_addr link;                         /* (link-local) IP address of host, used as gw
 245                                            if host is directly attached */
 246   struct rtable *tab;                   /* Dependent table, part of key */
 247   struct hostentry *next;               /* Next in hash chain */
 248   unsigned hash_key;                    /* Hash key */
 249   unsigned uc;                          /* Use count */
 250   struct rta *src;                      /* Source rta entry */
 251   byte dest;                            /* Chosen route destination type (RTD_...) */
 252   byte nexthop_linkable;                /* Nexthop list is completely non-device */
 253   u32 igp_metric;                       /* Chosen route IGP metric */
 254 };
 255
 256 typedef struct rte {
 257   struct rte *next;
 258   net *net;                             /* Network this RTE belongs to */
 259   struct rte_src *src;                  /* Route source that created the route */
 260   struct channel *sender;               /* Channel used to send the route to the routing table */
 261   struct rta *attrs;                    /* Attributes of this route */
 262   u32 id;                               /* Table specific route id */
 263   byte flags;                           /* Flags (REF_...) */
 264   byte pflags;                          /* Protocol-specific flags */
 265   btime lastmod;                        /* Last modified */
 266 } rte;
 267
 268 #define REF_COW         1               /* Copy this rte on write */
 269 #define REF_FILTERED    2               /* Route is rejected by import filter */
 270 #define REF_STALE       4               /* Route is stale in a refresh cycle */
 271 #define REF_DISCARD     8               /* Route is scheduled for discard */
 272 #define REF_MODIFY      16              /* Route is scheduled for modify */
 273
 274 /* Route is valid for propagation (may depend on other flags in the future), accepts NULL */
 275 static inline int rte_is_valid(rte *r) { return r && !(r->flags & REF_FILTERED); }
 276
 277 /* Route just has REF_FILTERED flag */
 278 static inline int rte_is_filtered(rte *r) { return !!(r->flags & REF_FILTERED); }
 279
 280
 281 /* Types of route announcement, also used as flags */
 282 #define RA_UNDEF        0               /* Undefined RA type */
 283 #define RA_OPTIMAL      1               /* Announcement of optimal route change */
 284 #define RA_ACCEPTED     2               /* Announcement of first accepted route */
 285 #define RA_ANY          3               /* Announcement of any route change */
 286 #define RA_MERGED       4               /* Announcement of optimal route merged with next ones */
 287
 288 /* Return value of preexport() callback */
 289 #define RIC_ACCEPT      1               /* Accepted by protocol */
 290 #define RIC_PROCESS     0               /* Process it through import filter */
 291 #define RIC_REJECT      -1              /* Rejected by protocol */
 292 #define RIC_DROP        -2              /* Silently dropped by protocol */
 293
 294 extern list routing_tables;
 295 struct config;
 296
 297 void rt_init(void);
 298 void rt_preconfig(struct config *);
 299 void rt_postconfig(struct config *);
 300 void rt_commit(struct config *new, struct config *old);
 301 void rt_lock_table(rtable *);
 302 void rt_unlock_table(rtable *);
 303 struct f_trie * rt_lock_trie(rtable *tab);
 304 void rt_unlock_trie(rtable *tab, struct f_trie *trie);
 305 void rt_subscribe(rtable *tab, struct rt_subscription *s);
 306 void rt_unsubscribe(struct rt_subscription *s);
 307 void rt_flowspec_link(rtable *src, rtable *dst);
 308 void rt_flowspec_unlink(rtable *src, rtable *dst);
 309 rtable *rt_setup(pool *, struct rtable_config *);
 310 static inline void rt_shutdown(rtable *r) { rfree(r->rp); }
 311
 312 static inline net *net_find(rtable *tab, const net_addr *addr) { return (net *) fib_find(&tab->fib, addr); }
 313 static inline net *net_find_valid(rtable *tab, const net_addr *addr)
 314 { net *n = net_find(tab, addr); return (n && rte_is_valid(n->routes)) ? n : NULL; }
 315 static inline net *net_get(rtable *tab, const net_addr *addr) { return (net *) fib_get(&tab->fib, addr); }
 316 net *net_get(rtable *tab, const net_addr *addr);
 317 net *net_route(rtable *tab, const net_addr *n);
 318 int net_roa_check(rtable *tab, const net_addr *n, u32 asn);
 319 rte *rte_find(net *net, struct rte_src *src);
 320 rte *rte_get_temp(struct rta *, struct rte_src *src);
 321 void rte_update2(struct channel *c, const net_addr *n, rte *new, struct rte_src *src);
 322 /* rte_update() moved to protocol.h to avoid dependency conflicts */
 323 int rt_examine(rtable *t, net_addr *a, struct channel *c, const struct filter *filter);
 324 rte *rt_export_merged(struct channel *c, net *net, rte **rt_free, linpool *pool, int silent);
 325 void rt_refresh_begin(rtable *t, struct channel *c);
 326 void rt_refresh_end(rtable *t, struct channel *c);
 327 void rt_modify_stale(rtable *t, struct channel *c);
 328 void rt_schedule_prune(rtable *t);
 329 void rte_dump(rte *);
 330 void rte_free(rte *);
 331 rte *rte_do_cow(rte *);
 332 static inline rte * rte_cow(rte *r) { return (r->flags & REF_COW) ? rte_do_cow(r) : r; }
 333 rte *rte_cow_rta(rte *r, linpool *lp);
 334 void rt_dump(rtable *);
 335 void rt_dump_all(void);
 336 int rt_feed_channel(struct channel *c);
 337 void rt_feed_channel_abort(struct channel *c);
 338 int rte_update_in(struct channel *c, const net_addr *n, rte *new, struct rte_src *src);
 339 int rt_reload_channel(struct channel *c);
 340 void rt_reload_channel_abort(struct channel *c);
 341 void rt_prune_sync(rtable *t, int all);
 342 int rte_update_out(struct channel *c, const net_addr *n, rte *new, rte *old0, int refeed);
 343 struct rtable_config *rt_new_table(struct symbol *s, uint addr_type);
 344
 345 static inline int rt_is_ip(rtable *tab)
 346 { return (tab->addr_type == NET_IP4) || (tab->addr_type == NET_IP6); }
 347
 348 static inline int rt_is_vpn(rtable *tab)
 349 { return (tab->addr_type == NET_VPN4) || (tab->addr_type == NET_VPN6); }
 350
 351 static inline int rt_is_roa(rtable *tab)
 352 { return (tab->addr_type == NET_ROA4) || (tab->addr_type == NET_ROA6); }
 353
 354 static inline int rt_is_flow(rtable *tab)
 355 { return (tab->addr_type == NET_FLOW4) || (tab->addr_type == NET_FLOW6); }
 356
 357
 358 /* Default limit for ECMP next hops, defined in sysdep code */
 359 extern const int rt_default_ecmp;
 360
 361 struct rt_show_data_rtable {
 362   node n;
 363   rtable *table;
 364   struct channel *export_channel;
 365 };
 366
 367 struct rt_show_data {
 368   net_addr *addr;
 369   list tables;
 370   struct rt_show_data_rtable *tab;      /* Iterator over table list */
 371   struct rt_show_data_rtable *last_table; /* Last table in output */
 372   struct fib_iterator fit;              /* Iterator over networks in table */
 373   struct f_trie_walk_state *walk_state; /* Iterator over networks in trie */
 374   struct f_trie *walk_lock;             /* Locked trie for walking */
 375   int verbose, tables_defined_by;
 376   const struct filter *filter;
 377   struct proto *show_protocol;
 378   struct proto *export_protocol;
 379   struct channel *export_channel;
 380   struct config *running_on_config;
 381   struct krt_proto *kernel;
 382   int export_mode, addr_mode, primary_only, filtered, stats;
 383
 384   int table_open;                       /* Iteration (fit) is open */
 385   int trie_walk;                        /* Current table is iterated using trie */
 386   int net_counter, rt_counter, show_counter, table_counter;
 387   int net_counter_last, rt_counter_last, show_counter_last;
 388 };
 389
 390 void rt_show(struct rt_show_data *);
 391 struct rt_show_data_rtable * rt_show_add_table(struct rt_show_data *d, rtable *t);
 392
 393 /* Value of table definition mode in struct rt_show_data */
 394 #define RSD_TDB_DEFAULT   0             /* no table specified */
 395 #define RSD_TDB_INDIRECT  0             /* show route ... protocol P ... */
 396 #define RSD_TDB_ALL       RSD_TDB_SET                   /* show route ... table all ... */
 397 #define RSD_TDB_DIRECT    RSD_TDB_SET | RSD_TDB_NMN     /* show route ... table X table Y ... */
 398
 399 #define RSD_TDB_SET       0x1           /* internal: show empty tables */
 400 #define RSD_TDB_NMN       0x2           /* internal: need matching net */
 401
 402 /* Value of addr_mode */
 403 #define RSD_ADDR_EQUAL  1               /* Exact query - show route <addr> */
 404 #define RSD_ADDR_FOR    2               /* Longest prefix match - show route for <addr> */
 405 #define RSD_ADDR_IN     3               /* Interval query - show route in <addr> */
 406
 407 /* Value of export_mode in struct rt_show_data */
 408 #define RSEM_NONE       0               /* Export mode not used */
 409 #define RSEM_PREEXPORT  1               /* Routes ready for export, before filtering */
 410 #define RSEM_EXPORT     2               /* Routes accepted by export filter */
 411 #define RSEM_NOEXPORT   3               /* Routes rejected by export filter */
 412 #define RSEM_EXPORTED   4               /* Routes marked in export map */
 413
 414 /*
 415  *      Route Attributes
 416  *
 417  *      Beware: All standard BGP attributes must be represented here instead
 418  *      of making them local to the route. This is needed to ensure proper
 419  *      construction of BGP route attribute lists.
 420  */
 421
 422 /* Nexthop structure */
 423 struct nexthop {
 424   ip_addr gw;                           /* Next hop */
 425   struct iface *iface;                  /* Outgoing interface */
 426   struct nexthop *next;
 427   byte flags;
 428   byte weight;
 429   byte labels_orig;                     /* Number of labels before hostentry was applied */
 430   byte labels;                          /* Number of all labels */
 431   u32 label[0];
 432 };
 433
 434 #define RNF_ONLINK              0x1     /* Gateway is onlink regardless of IP ranges */
 435
 436
 437 struct rte_src {
 438   struct rte_src *next;                 /* Hash chain */
 439   struct proto *proto;                  /* Protocol the source is based on */
 440   u32 private_id;                       /* Private ID, assigned by the protocol */
 441   u32 global_id;                        /* Globally unique ID of the source */
 442   unsigned uc;                          /* Use count */
 443 };
 444
 445
 446 typedef struct rta {
 447   struct rta *next, **pprev;            /* Hash chain */
 448   u32 uc;                               /* Use count */
 449   u32 hash_key;                         /* Hash over important fields */
 450   struct ea_list *eattrs;               /* Extended Attribute chain */
 451   struct hostentry *hostentry;          /* Hostentry for recursive next-hops */
 452   ip_addr from;                         /* Advertising router */
 453   u32 igp_metric;                       /* IGP metric to next hop (for iBGP routes) */
 454   u16 cached:1;                         /* Are attributes cached? */
 455   u16 source:7;                         /* Route source (RTS_...) */
 456   u16 scope:4;                          /* Route scope (SCOPE_... -- see ip.h) */
 457   u16 dest:4;                           /* Route destination type (RTD_...) */
 458   word pref;
 459   struct nexthop nh;                    /* Next hop */
 460 } rta;
 461
 462 #define RTS_STATIC 1                    /* Normal static route */
 463 #define RTS_INHERIT 2                   /* Route inherited from kernel */
 464 #define RTS_DEVICE 3                    /* Device route */
 465 #define RTS_STATIC_DEVICE 4             /* Static device route */
 466 #define RTS_REDIRECT 5                  /* Learned via redirect */
 467 #define RTS_RIP 6                       /* RIP route */
 468 #define RTS_OSPF 7                      /* OSPF route */
 469 #define RTS_OSPF_IA 8                   /* OSPF inter-area route */
 470 #define RTS_OSPF_EXT1 9                 /* OSPF external route type 1 */
 471 #define RTS_OSPF_EXT2 10                /* OSPF external route type 2 */
 472 #define RTS_BGP 11                      /* BGP route */
 473 #define RTS_PIPE 12                     /* Inter-table wormhole */
 474 #define RTS_BABEL 13                    /* Babel route */
 475 #define RTS_RPKI 14                     /* Route Origin Authorization */
 476 #define RTS_PERF 15                     /* Perf checker */
 477 #define RTS_MAX 16
 478
 479 #define RTD_NONE 0                      /* Undefined next hop */
 480 #define RTD_UNICAST 1                   /* Next hop is neighbor router */
 481 #define RTD_BLACKHOLE 2                 /* Silently drop packets */
 482 #define RTD_UNREACHABLE 3               /* Reject as unreachable */
 483 #define RTD_PROHIBIT 4                  /* Administratively prohibited */
 484 #define RTD_MAX 5
 485
 486 #define IGP_METRIC_UNKNOWN 0x80000000   /* Default igp_metric used when no other
 487                                            protocol-specific metric is availabe */
 488
 489
 490 extern const char * rta_dest_names[RTD_MAX];
 491
 492 static inline const char *rta_dest_name(uint n)
 493 { return (n < RTD_MAX) ? rta_dest_names[n] : "???"; }
 494
 495 /* Route has regular, reachable nexthop (i.e. not RTD_UNREACHABLE and like) */
 496 static inline int rte_is_reachable(rte *r)
 497 { return r->attrs->dest == RTD_UNICAST; }
 498
 499
 500 /*
 501  *      Extended Route Attributes
 502  */
 503
 504 typedef struct eattr {
 505   word id;                              /* EA_CODE(PROTOCOL_..., protocol-dependent ID) */
 506   byte flags;                           /* Protocol-dependent flags */
 507   byte type:5;                          /* Attribute type */
 508   byte originated:1;                    /* The attribute has originated locally */
 509   byte fresh:1;                         /* An uncached attribute (e.g. modified in export filter) */
 510   byte undef:1;                         /* Explicitly undefined */
 511   union {
 512     uintptr_t data;
 513     const struct adata *ptr;            /* Attribute data elsewhere */
 514   } u;
 515 } eattr;
 516
 517
 518 #define EA_CODE(proto,id) (((proto) << 8) | (id))
 519 #define EA_ID(ea) ((ea) & 0xff)
 520 #define EA_PROTO(ea) ((ea) >> 8)
 521 #define EA_CUSTOM(id) ((id) | EA_CUSTOM_BIT)
 522 #define EA_IS_CUSTOM(ea) ((ea) & EA_CUSTOM_BIT)
 523 #define EA_CUSTOM_ID(ea) ((ea) & ~EA_CUSTOM_BIT)
 524
 525 const char *ea_custom_name(uint ea);
 526
 527 #define EA_GEN_IGP_METRIC EA_CODE(PROTOCOL_NONE, 0)
 528
 529 #define EA_CODE_MASK 0xffff
 530 #define EA_CUSTOM_BIT 0x8000
 531 #define EA_ALLOW_UNDEF 0x10000          /* ea_find: allow EAF_TYPE_UNDEF */
 532 #define EA_BIT(n) ((n) << 24)           /* Used in bitfield accessors */
 533 #define EA_BIT_GET(ea) ((ea) >> 24)
 534
 535 #define EAF_TYPE_MASK 0x1f              /* Mask with this to get type */
 536 #define EAF_TYPE_INT 0x01               /* 32-bit unsigned integer number */
 537 #define EAF_TYPE_OPAQUE 0x02            /* Opaque byte string (not filterable) */
 538 #define EAF_TYPE_IP_ADDRESS 0x04        /* IP address */
 539 #define EAF_TYPE_ROUTER_ID 0x05         /* Router ID (IPv4 address) */
 540 #define EAF_TYPE_AS_PATH 0x06           /* BGP AS path (encoding per RFC 1771:4.3) */
 541 #define EAF_TYPE_BITFIELD 0x09          /* 32-bit embedded bitfield */
 542 #define EAF_TYPE_INT_SET 0x0a           /* Set of u32's (e.g., a community list) */
 543 #define EAF_TYPE_EC_SET 0x0e            /* Set of pairs of u32's - ext. community list */
 544 #define EAF_TYPE_LC_SET 0x12            /* Set of triplets of u32's - large community list */
 545 #define EAF_TYPE_IFACE 0x16             /* Interface pointer stored in adata */
 546 #define EAF_EMBEDDED 0x01               /* Data stored in eattr.u.data (part of type spec) */
 547 #define EAF_VAR_LENGTH 0x02             /* Attribute length is variable (part of type spec) */
 548
 549 typedef struct adata {
 550   uint length;                          /* Length of data */
 551   byte data[0];
 552 } adata;
 553
 554 extern const adata null_adata;          /* adata of length 0 */
 555
 556 static inline struct adata *
 557 lp_alloc_adata(struct linpool *pool, uint len)
 558 {
 559   struct adata *ad = lp_alloc(pool, sizeof(struct adata) + len);
 560   ad->length = len;
 561   return ad;
 562 }
 563
 564 static inline int adata_same(const struct adata *a, const struct adata *b)
 565 { return (a->length == b->length && !memcmp(a->data, b->data, a->length)); }
 566
 567
 568 typedef struct ea_list {
 569   struct ea_list *next;                 /* In case we have an override list */
 570   byte flags;                           /* Flags: EALF_... */
 571   byte rfu;
 572   word count;                           /* Number of attributes */
 573   eattr attrs[0];                       /* Attribute definitions themselves */
 574 } ea_list;
 575
 576 #define EALF_SORTED 1                   /* Attributes are sorted by code */
 577 #define EALF_BISECT 2                   /* Use interval bisection for searching */
 578 #define EALF_CACHED 4                   /* Attributes belonging to cached rta */
 579
 580 struct rte_src *rt_find_source(struct proto *p, u32 id);
 581 struct rte_src *rt_get_source(struct proto *p, u32 id);
 582 static inline void rt_lock_source(struct rte_src *src) { src->uc++; }
 583 static inline void rt_unlock_source(struct rte_src *src) { src->uc--; }
 584 void rt_prune_sources(void);
 585
 586 struct ea_walk_state {
 587   ea_list *eattrs;                      /* Ccurrent ea_list, initially set by caller */
 588   eattr *ea;                            /* Current eattr, initially NULL */
 589   u32 visited[4];                       /* Bitfield, limiting max to 128 */
 590 };
 591
 592 eattr *ea_find(ea_list *, unsigned ea);
 593 eattr *ea_walk(struct ea_walk_state *s, uint id, uint max);
 594 uintptr_t ea_get_int(ea_list *, unsigned ea, uintptr_t def);
 595 void ea_dump(ea_list *);
 596 void ea_sort(ea_list *);                /* Sort entries in all sub-lists */
 597 unsigned ea_scan(ea_list *);            /* How many bytes do we need for merged ea_list */
 598 void ea_merge(ea_list *from, ea_list *to); /* Merge sub-lists to allocated buffer */
 599 int ea_same(ea_list *x, ea_list *y);    /* Test whether two ea_lists are identical */
 600 uint ea_hash(ea_list *e);       /* Calculate 16-bit hash value */
 601 ea_list *ea_append(ea_list *to, ea_list *what);
 602 void ea_format_bitfield(const struct eattr *a, byte *buf, int bufsize, const char **names, int min, int max);
 603
 604 #define ea_normalize(ea) do { \
 605   if (ea->next) { \
 606     ea_list *t = alloca(ea_scan(ea)); \
 607     ea_merge(ea, t); \
 608     ea = t; \
 609   } \
 610   ea_sort(ea); \
 611   if (ea->count == 0) \
 612     ea = NULL; \
 613 } while(0) \
 614
 615 struct ea_one_attr_list {
 616   ea_list l;
 617   eattr a;
 618 };
 619
 620 static inline eattr *
 621 ea_set_attr(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, uintptr_t val)
 622 {
 623   struct ea_one_attr_list *ea = lp_alloc(pool, sizeof(*ea));
 624   *ea = (struct ea_one_attr_list) {
 625     .l.flags = EALF_SORTED,
 626     .l.count = 1,
 627     .l.next = *to,
 628
 629     .a.id = id,
 630     .a.type = type,
 631     .a.flags = flags,
 632   };
 633
 634   if (type & EAF_EMBEDDED)
 635     ea->a.u.data = val;
 636   else
 637     ea->a.u.ptr = (struct adata *) val;
 638
 639   *to = &ea->l;
 640
 641   return &ea->a;
 642 }
 643
 644 static inline void
 645 ea_unset_attr(ea_list **to, struct linpool *pool, _Bool local, uint code)
 646 {
 647   struct ea_one_attr_list *ea = lp_alloc(pool, sizeof(*ea));
 648   *ea = (struct ea_one_attr_list) {
 649     .l.flags = EALF_SORTED,
 650     .l.count = 1,
 651     .l.next = *to,
 652     .a.id = code,
 653     .a.fresh = local,
 654     .a.originated = local,
 655     .a.undef = 1,
 656   };
 657
 658   *to = &ea->l;
 659 }
 660
 661 static inline void
 662 ea_set_attr_u32(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, u32 val)
 663 { ea_set_attr(to, pool, id, flags, type, (uintptr_t) val); }
 664
 665 static inline void
 666 ea_set_attr_ptr(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, struct adata *val)
 667 { ea_set_attr(to, pool, id, flags, type, (uintptr_t) val); }
 668
 669 static inline void
 670 ea_set_attr_data(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, void *data, uint len)
 671 {
 672   struct adata *a = lp_alloc_adata(pool, len);
 673   memcpy(a->data, data, len);
 674   ea_set_attr(to, pool, id, flags, type, (uintptr_t) a);
 675 }
 676
 677
 678 #define NEXTHOP_MAX_SIZE (sizeof(struct nexthop) + sizeof(u32)*MPLS_MAX_LABEL_STACK)
 679
 680 static inline size_t nexthop_size(const struct nexthop *nh)
 681 { return sizeof(struct nexthop) + sizeof(u32)*nh->labels; }
 682 int nexthop__same(struct nexthop *x, struct nexthop *y); /* Compare multipath nexthops */
 683 static inline int nexthop_same(struct nexthop *x, struct nexthop *y)
 684 { return (x == y) || nexthop__same(x, y); }
 685 struct nexthop *nexthop_merge(struct nexthop *x, struct nexthop *y, int rx, int ry, int max, linpool *lp);
 686 struct nexthop *nexthop_sort(struct nexthop *x);
 687 static inline void nexthop_link(struct rta *a, struct nexthop *from)
 688 { memcpy(&a->nh, from, nexthop_size(from)); }
 689 void nexthop_insert(struct nexthop **n, struct nexthop *y);
 690 int nexthop_is_sorted(struct nexthop *x);
 691
 692 void rta_init(void);
 693 static inline size_t rta_size(const rta *a) { return sizeof(rta) + sizeof(u32)*a->nh.labels; }
 694 #define RTA_MAX_SIZE (sizeof(rta) + sizeof(u32)*MPLS_MAX_LABEL_STACK)
 695 rta *rta_lookup(rta *);                 /* Get rta equivalent to this one, uc++ */
 696 static inline int rta_is_cached(rta *r) { return r->cached; }
 697 static inline rta *rta_clone(rta *r) { r->uc++; return r; }
 698 void rta__free(rta *r);
 699 static inline void rta_free(rta *r) { if (r && !--r->uc) rta__free(r); }
 700 rta *rta_do_cow(rta *o, linpool *lp);
 701 static inline rta * rta_cow(rta *r, linpool *lp) { return rta_is_cached(r) ? rta_do_cow(r, lp) : r; }
 702 void rta_dump(rta *);
 703 void rta_dump_all(void);
 704 void rta_show(struct cli *, rta *);
 705
 706 u32 rt_get_igp_metric(rte *rt);
 707 struct hostentry * rt_get_hostentry(rtable *tab, ip_addr a, ip_addr ll, rtable *dep);
 708 void rta_apply_hostentry(rta *a, struct hostentry *he, mpls_label_stack *mls);
 709
 710 static inline void
 711 rta_set_recursive_next_hop(rtable *dep, rta *a, rtable *tab, ip_addr gw, ip_addr ll, mpls_label_stack *mls)
 712 {
 713   rta_apply_hostentry(a, rt_get_hostentry(tab, gw, ll, dep), mls);
 714 }
 715
 716 /*
 717  * rta_set_recursive_next_hop() acquires hostentry from hostcache and fills
 718  * rta->hostentry field.  New hostentry has zero use count. Cached rta locks its
 719  * hostentry (increases its use count), uncached rta does not lock it. Hostentry
 720  * with zero use count is removed asynchronously during host cache update,
 721  * therefore it is safe to hold such hostentry temorarily. Hostentry holds a
 722  * lock for a 'source' rta, mainly to share multipath nexthops.
 723  *
 724  * There is no need to hold a lock for hostentry->dep table, because that table
 725  * contains routes responsible for that hostentry, and therefore is non-empty if
 726  * given hostentry has non-zero use count. If the hostentry has zero use count,
 727  * the entry is removed before dep is referenced.
 728  *
 729  * The protocol responsible for routes with recursive next hops should hold a
 730  * lock for a 'source' table governing that routes (argument tab to
 731  * rta_set_recursive_next_hop()), because its routes reference hostentries
 732  * (through rta) related to the governing table. When all such routes are
 733  * removed, rtas are immediately removed achieving zero uc. Then the 'source'
 734  * table lock could be immediately released, although hostentries may still
 735  * exist - they will be freed together with the 'source' table.
 736  */
 737
 738 static inline void rt_lock_hostentry(struct hostentry *he) { if (he) he->uc++; }
 739 static inline void rt_unlock_hostentry(struct hostentry *he) { if (he) he->uc--; }
 740
 741 int rt_flowspec_check(rtable *tab_ip, rtable *tab_flow, const net_addr *n, rta *a, int interior);
 742
 743
 744 /*
 745  *      Default protocol preferences
 746  */
 747
 748 #define DEF_PREF_DIRECT         240     /* Directly connected */
 749 #define DEF_PREF_STATIC         200     /* Static route */
 750 #define DEF_PREF_OSPF           150     /* OSPF intra-area, inter-area and type 1 external routes */
 751 #define DEF_PREF_BABEL          130     /* Babel */
 752 #define DEF_PREF_RIP            120     /* RIP */
 753 #define DEF_PREF_BGP            100     /* BGP */
 754 #define DEF_PREF_RPKI           100     /* RPKI */
 755 #define DEF_PREF_INHERITED      10      /* Routes inherited from other routing daemons */
 756
 757 /*
 758  *      Route Origin Authorization
 759  */
 760
 761 #define ROA_UNKNOWN     0
 762 #define ROA_VALID       1
 763 #define ROA_INVALID     2
 764
 765 #endif