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 filter;
24 struct cli;
25
26 /*
27 * Generic data structure for storing network prefixes. Also used
28 * for the master routing table. Currently implemented as a hash
29 * table.
30 *
31 * Available operations:
32 * - insertion of new entry
33 * - deletion of entry
34 * - searching for entry by network prefix
35 * - asynchronous retrieval of fib contents
36 */
37
38 struct fib_node {
39 struct fib_node *next; /* Next in hash chain */
40 struct fib_iterator *readers; /* List of readers of this node */
41 net_addr addr[0];
42 };
43
44 struct fib_iterator { /* See lib/slists.h for an explanation */
45 struct fib_iterator *prev, *next; /* Must be synced with struct fib_node! */
46 byte efef; /* 0xff to distinguish between iterator and node */
47 byte pad[3];
48 struct fib_node *node; /* Or NULL if freshly merged */
49 uint hash;
50 };
51
52 typedef void (*fib_init_fn)(void *);
53
54 struct fib {
55 pool *fib_pool; /* Pool holding all our data */
56 slab *fib_slab; /* Slab holding all fib nodes */
57 struct fib_node **hash_table; /* Node hash table */
58 uint hash_size; /* Number of hash table entries (a power of two) */
59 uint hash_order; /* Binary logarithm of hash_size */
60 uint hash_shift; /* 32 - hash_order */
61 uint addr_type; /* Type of address data stored in fib (NET_*) */
62 uint node_size; /* FIB node size, 0 for nonuniform */
63 uint node_offset; /* Offset of fib_node struct inside of user data */
64 uint entries; /* Number of entries */
65 uint entries_min, entries_max; /* Entry count limits (else start rehashing) */
66 fib_init_fn init; /* Constructor */
67 };
68
fib_node_to_user(struct fib * f,struct fib_node * e)69 static inline void * fib_node_to_user(struct fib *f, struct fib_node *e)
70 { return e ? (void *) ((char *) e - f->node_offset) : NULL; }
71
fib_user_to_node(struct fib * f,void * e)72 static inline struct fib_node * fib_user_to_node(struct fib *f, void *e)
73 { return e ? (void *) ((char *) e + f->node_offset) : NULL; }
74
75 void fib_init(struct fib *f, pool *p, uint addr_type, uint node_size, uint node_offset, uint hash_order, fib_init_fn init);
76 void *fib_find(struct fib *, const net_addr *); /* Find or return NULL if doesn't exist */
77 void *fib_get_chain(struct fib *f, const net_addr *a); /* Find first node in linked list from hash table */
78 void *fib_get(struct fib *, const net_addr *); /* Find or create new if nonexistent */
79 void *fib_route(struct fib *, const net_addr *); /* Longest-match routing lookup */
80 void fib_delete(struct fib *, void *); /* Remove fib entry */
81 void fib_free(struct fib *); /* Destroy the fib */
82 void fib_check(struct fib *); /* Consistency check for debugging */
83
84 void fit_init(struct fib_iterator *, struct fib *); /* Internal functions, don't call */
85 struct fib_node *fit_get(struct fib *, struct fib_iterator *);
86 void fit_put(struct fib_iterator *, struct fib_node *);
87 void fit_put_next(struct fib *f, struct fib_iterator *i, struct fib_node *n, uint hpos);
88 void fit_put_end(struct fib_iterator *i);
89 void fit_copy(struct fib *f, struct fib_iterator *dst, struct fib_iterator *src);
90
91
92 #define FIB_WALK(fib, type, z) do { \
93 struct fib_node *fn_, **ff_ = (fib)->hash_table; \
94 uint count_ = (fib)->hash_size; \
95 type *z; \
96 while (count_--) \
97 for (fn_ = *ff_++; z = fib_node_to_user(fib, fn_); fn_=fn_->next)
98
99 #define FIB_WALK_END } while (0)
100
101 #define FIB_ITERATE_INIT(it, fib) fit_init(it, fib)
102
103 #define FIB_ITERATE_START(fib, it, type, z) do { \
104 struct fib_node *fn_ = fit_get(fib, it); \
105 uint count_ = (fib)->hash_size; \
106 uint hpos_ = (it)->hash; \
107 type *z; \
108 for(;;) { \
109 if (!fn_) \
110 { \
111 if (++hpos_ >= count_) \
112 break; \
113 fn_ = (fib)->hash_table[hpos_]; \
114 continue; \
115 } \
116 z = fib_node_to_user(fib, fn_);
117
118 #define FIB_ITERATE_END fn_ = fn_->next; } } while(0)
119
120 #define FIB_ITERATE_PUT(it) fit_put(it, fn_)
121
122 #define FIB_ITERATE_PUT_NEXT(it, fib) fit_put_next(fib, it, fn_, hpos_)
123
124 #define FIB_ITERATE_PUT_END(it) fit_put_end(it)
125
126 #define FIB_ITERATE_UNLINK(it, fib) fit_get(fib, it)
127
128 #define FIB_ITERATE_COPY(dst, src, fib) fit_copy(fib, dst, src)
129
130
131 /*
132 * Master Routing Tables. Generally speaking, each of them contains a FIB
133 * with each entry pointing to a list of route entries representing routes
134 * to given network (with the selected one at the head).
135 *
136 * Each of the RTE's contains variable data (the preference and protocol-dependent
137 * metrics) and a pointer to a route attribute block common for many routes).
138 *
139 * It's guaranteed that there is at most one RTE for every (prefix,proto) pair.
140 */
141
142 struct rtable_config {
143 node n;
144 char *name;
145 struct rtable *table;
146 struct proto_config *krt_attached; /* Kernel syncer attached to this table */
147 uint addr_type; /* Type of address data stored in table (NET_*) */
148 int gc_max_ops; /* Maximum number of operations before GC is run */
149 int gc_min_time; /* Minimum time between two consecutive GC runs */
150 byte sorted; /* Routes of network are sorted according to rte_better() */
151 btime min_settle_time; /* Minimum settle time for notifications */
152 btime max_settle_time; /* Maximum settle time for notifications */
153 };
154
155 typedef struct rtable {
156 node n; /* Node in list of all tables */
157 struct fib fib;
158 char *name; /* Name of this table */
159 list channels; /* List of attached channels (struct channel) */
160 uint addr_type; /* Type of address data stored in table (NET_*) */
161 int pipe_busy; /* Pipe loop detection */
162 int use_count; /* Number of protocols using this table */
163 u32 rt_count; /* Number of routes in the table */
164 struct hmap id_map;
165 struct hostcache *hostcache;
166 struct rtable_config *config; /* Configuration of this table */
167 struct config *deleted; /* Table doesn't exist in current configuration,
168 * delete as soon as use_count becomes 0 and remove
169 * obstacle from this routing table.
170 */
171 struct event *rt_event; /* Routing table event */
172 btime last_rt_change; /* Last time when route changed */
173 btime base_settle_time; /* Start time of rtable settling interval */
174 btime gc_time; /* Time of last GC */
175 int gc_counter; /* Number of operations since last GC */
176 byte prune_state; /* Table prune state, 1 -> scheduled, 2-> running */
177 byte hcu_scheduled; /* Hostcache update is scheduled */
178 byte nhu_state; /* Next Hop Update state */
179 struct fib_iterator prune_fit; /* Rtable prune FIB iterator */
180 struct fib_iterator nhu_fit; /* Next Hop Update FIB iterator */
181
182 list subscribers; /* Subscribers for notifications */
183 struct timer *settle_timer; /* Settle time for notifications */
184 } rtable;
185
186 struct rt_subscription {
187 node n;
188 rtable *tab;
189 void (*hook)(struct rt_subscription *b);
190 void *data;
191 };
192
193 #define NHU_CLEAN 0
194 #define NHU_SCHEDULED 1
195 #define NHU_RUNNING 2
196 #define NHU_DIRTY 3
197
198 typedef struct network {
199 struct rte *routes; /* Available routes for this network */
200 struct fib_node n; /* FIB flags reserved for kernel syncer */
201 } net;
202
203 struct hostcache {
204 slab *slab; /* Slab holding all hostentries */
205 struct hostentry **hash_table; /* Hash table for hostentries */
206 unsigned hash_order, hash_shift;
207 unsigned hash_max, hash_min;
208 unsigned hash_items;
209 linpool *lp; /* Linpool for trie */
210 struct f_trie *trie; /* Trie of prefixes that might affect hostentries */
211 list hostentries; /* List of all hostentries */
212 byte update_hostcache;
213 };
214
215 struct hostentry {
216 node ln;
217 ip_addr addr; /* IP address of host, part of key */
218 ip_addr link; /* (link-local) IP address of host, used as gw
219 if host is directly attached */
220 struct rtable *tab; /* Dependent table, part of key */
221 struct hostentry *next; /* Next in hash chain */
222 unsigned hash_key; /* Hash key */
223 unsigned uc; /* Use count */
224 struct rta *src; /* Source rta entry */
225 byte dest; /* Chosen route destination type (RTD_...) */
226 byte nexthop_linkable; /* Nexthop list is completely non-device */
227 u32 igp_metric; /* Chosen route IGP metric */
228 };
229
230 typedef struct rte {
231 struct rte *next;
232 net *net; /* Network this RTE belongs to */
233 struct channel *sender; /* Channel used to send the route to the routing table */
234 struct rta *attrs; /* Attributes of this route */
235 u32 id; /* Table specific route id */
236 byte flags; /* Flags (REF_...) */
237 byte pflags; /* Protocol-specific flags */
238 word pref; /* Route preference */
239 btime lastmod; /* Last modified */
240 union { /* Protocol-dependent data (metrics etc.) */
241 #ifdef CONFIG_RIP
242 struct {
243 struct iface *from; /* Incoming iface */
244 u8 metric; /* RIP metric */
245 u16 tag; /* External route tag */
246 } rip;
247 #endif
248 #ifdef CONFIG_OSPF
249 struct {
250 u32 metric1, metric2; /* OSPF Type 1 and Type 2 metrics */
251 u32 tag; /* External route tag */
252 u32 router_id; /* Router that originated this route */
253 } ospf;
254 #endif
255 #ifdef CONFIG_BGP
256 struct {
257 u8 suppressed; /* Used for deterministic MED comparison */
258 s8 stale; /* Route is LLGR_STALE, -1 if unknown */
259 } bgp;
260 #endif
261 #ifdef CONFIG_BABEL
262 struct {
263 u16 seqno; /* Babel seqno */
264 u16 metric; /* Babel metric */
265 u64 router_id; /* Babel router id */
266 } babel;
267 #endif
268 struct { /* Routes generated by krt sync (both temporary and inherited ones) */
269 s8 src; /* Alleged route source (see krt.h) */
270 u8 proto; /* Kernel source protocol ID */
271 u8 seen; /* Seen during last scan */
272 u8 best; /* Best route in network, propagated to core */
273 u32 metric; /* Kernel metric */
274 } krt;
275 } u;
276 } rte;
277
278 #define REF_COW 1 /* Copy this rte on write */
279 #define REF_FILTERED 2 /* Route is rejected by import filter */
280 #define REF_STALE 4 /* Route is stale in a refresh cycle */
281 #define REF_DISCARD 8 /* Route is scheduled for discard */
282 #define REF_MODIFY 16 /* Route is scheduled for modify */
283
284 /* Route is valid for propagation (may depend on other flags in the future), accepts NULL */
rte_is_valid(rte * r)285 static inline int rte_is_valid(rte *r) { return r && !(r->flags & REF_FILTERED); }
286
287 /* Route just has REF_FILTERED flag */
rte_is_filtered(rte * r)288 static inline int rte_is_filtered(rte *r) { return !!(r->flags & REF_FILTERED); }
289
290
291 /* Types of route announcement, also used as flags */
292 #define RA_UNDEF 0 /* Undefined RA type */
293 #define RA_OPTIMAL 1 /* Announcement of optimal route change */
294 #define RA_ACCEPTED 2 /* Announcement of first accepted route */
295 #define RA_ANY 3 /* Announcement of any route change */
296 #define RA_MERGED 4 /* Announcement of optimal route merged with next ones */
297
298 /* Return value of preexport() callback */
299 #define RIC_ACCEPT 1 /* Accepted by protocol */
300 #define RIC_PROCESS 0 /* Process it through import filter */
301 #define RIC_REJECT -1 /* Rejected by protocol */
302 #define RIC_DROP -2 /* Silently dropped by protocol */
303
304 extern list routing_tables;
305 struct config;
306
307 void rt_init(void);
308 void rt_preconfig(struct config *);
309 void rt_commit(struct config *new, struct config *old);
310 void rt_lock_table(rtable *);
311 void rt_unlock_table(rtable *);
312 void rt_subscribe(rtable *tab, struct rt_subscription *s);
313 void rt_unsubscribe(struct rt_subscription *s);
314 void rt_setup(pool *, rtable *, struct rtable_config *);
net_find(rtable * tab,const net_addr * addr)315 static inline net *net_find(rtable *tab, const net_addr *addr) { return (net *) fib_find(&tab->fib, addr); }
net_find_valid(rtable * tab,const net_addr * addr)316 static inline net *net_find_valid(rtable *tab, const net_addr *addr)
317 { net *n = net_find(tab, addr); return (n && rte_is_valid(n->routes)) ? n : NULL; }
net_get(rtable * tab,const net_addr * addr)318 static inline net *net_get(rtable *tab, const net_addr *addr) { return (net *) fib_get(&tab->fib, addr); }
319 void *net_route(rtable *tab, const net_addr *n);
320 int net_roa_check(rtable *tab, const net_addr *n, u32 asn);
321 rte *rte_find(net *net, struct rte_src *src);
322 rte *rte_get_temp(struct rta *);
323 void rte_update2(struct channel *c, const net_addr *n, rte *new, struct rte_src *src);
324 /* rte_update() moved to protocol.h to avoid dependency conflicts */
325 int rt_examine(rtable *t, net_addr *a, struct proto *p, const struct filter *filter);
326 rte *rt_export_merged(struct channel *c, net *net, rte **rt_free, linpool *pool, int silent);
327 void rt_refresh_begin(rtable *t, struct channel *c);
328 void rt_refresh_end(rtable *t, struct channel *c);
329 void rt_modify_stale(rtable *t, struct channel *c);
330 void rt_schedule_prune(rtable *t);
331 void rte_dump(rte *);
332 void rte_free(rte *);
333 rte *rte_do_cow(rte *);
rte_cow(rte * r)334 static inline rte * rte_cow(rte *r) { return (r->flags & REF_COW) ? rte_do_cow(r) : r; }
335 rte *rte_cow_rta(rte *r, linpool *lp);
336 void rte_init_tmp_attrs(struct rte *r, linpool *lp, uint max);
337 void rte_make_tmp_attr(struct rte *r, uint id, uint type, uintptr_t val);
338 void rte_make_tmp_attrs(struct rte **r, struct linpool *pool, struct rta **old_attrs);
339 uintptr_t rte_store_tmp_attr(struct rte *r, uint id);
340 void rt_dump(rtable *);
341 void rt_dump_all(void);
342 int rt_feed_channel(struct channel *c);
343 void rt_feed_channel_abort(struct channel *c);
344 int rte_update_in(struct channel *c, const net_addr *n, rte *new, struct rte_src *src);
345 int rt_reload_channel(struct channel *c);
346 void rt_reload_channel_abort(struct channel *c);
347 void rt_prune_sync(rtable *t, int all);
348 int rte_update_out(struct channel *c, const net_addr *n, rte *new, rte *old0, int refeed);
349 struct rtable_config *rt_new_table(struct symbol *s, uint addr_type);
350
351
352 /* Default limit for ECMP next hops, defined in sysdep code */
353 extern const int rt_default_ecmp;
354
355 struct rt_show_data_rtable {
356 node n;
357 rtable *table;
358 struct channel *export_channel;
359 };
360
361 struct rt_show_data {
362 net_addr *addr;
363 list tables;
364 struct rt_show_data_rtable *tab; /* Iterator over table list */
365 struct rt_show_data_rtable *last_table; /* Last table in output */
366 struct fib_iterator fit; /* Iterator over networks in table */
367 int verbose, tables_defined_by;
368 const struct filter *filter;
369 struct proto *show_protocol;
370 struct proto *export_protocol;
371 struct channel *export_channel;
372 struct config *running_on_config;
373 struct krt_proto *kernel;
374 int export_mode, primary_only, filtered, stats, show_for;
375
376 int table_open; /* Iteration (fit) is open */
377 int net_counter, rt_counter, show_counter, table_counter;
378 int net_counter_last, rt_counter_last, show_counter_last;
379 };
380
381 void rt_show(struct rt_show_data *);
382 struct rt_show_data_rtable * rt_show_add_table(struct rt_show_data *d, rtable *t);
383
384 /* Value of table definition mode in struct rt_show_data */
385 #define RSD_TDB_DEFAULT 0 /* no table specified */
386 #define RSD_TDB_INDIRECT 0 /* show route ... protocol P ... */
387 #define RSD_TDB_ALL RSD_TDB_SET /* show route ... table all ... */
388 #define RSD_TDB_DIRECT RSD_TDB_SET | RSD_TDB_NMN /* show route ... table X table Y ... */
389
390 #define RSD_TDB_SET 0x1 /* internal: show empty tables */
391 #define RSD_TDB_NMN 0x2 /* internal: need matching net */
392
393 /* Value of export_mode in struct rt_show_data */
394 #define RSEM_NONE 0 /* Export mode not used */
395 #define RSEM_PREEXPORT 1 /* Routes ready for export, before filtering */
396 #define RSEM_EXPORT 2 /* Routes accepted by export filter */
397 #define RSEM_NOEXPORT 3 /* Routes rejected by export filter */
398 #define RSEM_EXPORTED 4 /* Routes marked in export map */
399
400 /*
401 * Route Attributes
402 *
403 * Beware: All standard BGP attributes must be represented here instead
404 * of making them local to the route. This is needed to ensure proper
405 * construction of BGP route attribute lists.
406 */
407
408 /* Nexthop structure */
409 struct nexthop {
410 ip_addr gw; /* Next hop */
411 struct iface *iface; /* Outgoing interface */
412 struct nexthop *next;
413 byte flags;
414 byte weight;
415 byte labels_orig; /* Number of labels before hostentry was applied */
416 byte labels; /* Number of all labels */
417 u32 label[0];
418 };
419
420 #define RNF_ONLINK 0x1 /* Gateway is onlink regardless of IP ranges */
421
422
423 struct rte_src {
424 struct rte_src *next; /* Hash chain */
425 struct proto *proto; /* Protocol the source is based on */
426 u32 private_id; /* Private ID, assigned by the protocol */
427 u32 global_id; /* Globally unique ID of the source */
428 unsigned uc; /* Use count */
429 };
430
431
432 typedef struct rta {
433 struct rta *next, **pprev; /* Hash chain */
434 u32 uc; /* Use count */
435 u32 hash_key; /* Hash over important fields */
436 struct ea_list *eattrs; /* Extended Attribute chain */
437 struct rte_src *src; /* Route source that created the route */
438 struct hostentry *hostentry; /* Hostentry for recursive next-hops */
439 ip_addr from; /* Advertising router */
440 u32 igp_metric; /* IGP metric to next hop (for iBGP routes) */
441 u8 source; /* Route source (RTS_...) */
442 u8 scope; /* Route scope (SCOPE_... -- see ip.h) */
443 u8 dest; /* Route destination type (RTD_...) */
444 u8 aflags;
445 struct nexthop nh; /* Next hop */
446 } rta;
447
448 #define RTS_DUMMY 0 /* Dummy route to be removed soon */
449 #define RTS_STATIC 1 /* Normal static route */
450 #define RTS_INHERIT 2 /* Route inherited from kernel */
451 #define RTS_DEVICE 3 /* Device route */
452 #define RTS_STATIC_DEVICE 4 /* Static device route */
453 #define RTS_REDIRECT 5 /* Learned via redirect */
454 #define RTS_RIP 6 /* RIP route */
455 #define RTS_OSPF 7 /* OSPF route */
456 #define RTS_OSPF_IA 8 /* OSPF inter-area route */
457 #define RTS_OSPF_EXT1 9 /* OSPF external route type 1 */
458 #define RTS_OSPF_EXT2 10 /* OSPF external route type 2 */
459 #define RTS_BGP 11 /* BGP route */
460 #define RTS_PIPE 12 /* Inter-table wormhole */
461 #define RTS_BABEL 13 /* Babel route */
462 #define RTS_RPKI 14 /* Route Origin Authorization */
463 #define RTS_PERF 15 /* Perf checker */
464 #define RTS_MAX 16
465
466 #define RTC_UNICAST 0
467 #define RTC_BROADCAST 1
468 #define RTC_MULTICAST 2
469 #define RTC_ANYCAST 3 /* IPv6 Anycast */
470
471 #define RTD_NONE 0 /* Undefined next hop */
472 #define RTD_UNICAST 1 /* Next hop is neighbor router */
473 #define RTD_BLACKHOLE 2 /* Silently drop packets */
474 #define RTD_UNREACHABLE 3 /* Reject as unreachable */
475 #define RTD_PROHIBIT 4 /* Administratively prohibited */
476 #define RTD_MAX 5
477
478 #define RTAF_CACHED 1 /* This is a cached rta */
479
480 #define IGP_METRIC_UNKNOWN 0x80000000 /* Default igp_metric used when no other
481 protocol-specific metric is availabe */
482
483
484 extern const char * rta_dest_names[RTD_MAX];
485
rta_dest_name(uint n)486 static inline const char *rta_dest_name(uint n)
487 { return (n < RTD_MAX) ? rta_dest_names[n] : "???"; }
488
489 /* Route has regular, reachable nexthop (i.e. not RTD_UNREACHABLE and like) */
rte_is_reachable(rte * r)490 static inline int rte_is_reachable(rte *r)
491 { return r->attrs->dest == RTD_UNICAST; }
492
493
494 /*
495 * Extended Route Attributes
496 */
497
498 typedef struct eattr {
499 word id; /* EA_CODE(PROTOCOL_..., protocol-dependent ID) */
500 byte flags; /* Protocol-dependent flags */
501 byte type; /* Attribute type and several flags (EAF_...) */
502 union {
503 u32 data;
504 const struct adata *ptr; /* Attribute data elsewhere */
505 } u;
506 } eattr;
507
508
509 #define EA_CODE(proto,id) (((proto) << 8) | (id))
510 #define EA_ID(ea) ((ea) & 0xff)
511 #define EA_PROTO(ea) ((ea) >> 8)
512 #define EA_ID_FLAG(ea) (1 << EA_ID(ea))
513 #define EA_CUSTOM(id) ((id) | EA_CUSTOM_BIT)
514 #define EA_IS_CUSTOM(ea) ((ea) & EA_CUSTOM_BIT)
515 #define EA_CUSTOM_ID(ea) ((ea) & ~EA_CUSTOM_BIT)
516
517 const char *ea_custom_name(uint ea);
518
519 #define EA_GEN_IGP_METRIC EA_CODE(PROTOCOL_NONE, 0)
520
521 #define EA_CODE_MASK 0xffff
522 #define EA_CUSTOM_BIT 0x8000
523 #define EA_ALLOW_UNDEF 0x10000 /* ea_find: allow EAF_TYPE_UNDEF */
524 #define EA_BIT(n) ((n) << 24) /* Used in bitfield accessors */
525 #define EA_BIT_GET(ea) ((ea) >> 24)
526
527 #define EAF_TYPE_MASK 0x1f /* Mask with this to get type */
528 #define EAF_TYPE_INT 0x01 /* 32-bit unsigned integer number */
529 #define EAF_TYPE_OPAQUE 0x02 /* Opaque byte string (not filterable) */
530 #define EAF_TYPE_IP_ADDRESS 0x04 /* IP address */
531 #define EAF_TYPE_ROUTER_ID 0x05 /* Router ID (IPv4 address) */
532 #define EAF_TYPE_AS_PATH 0x06 /* BGP AS path (encoding per RFC 1771:4.3) */
533 #define EAF_TYPE_BITFIELD 0x09 /* 32-bit embedded bitfield */
534 #define EAF_TYPE_INT_SET 0x0a /* Set of u32's (e.g., a community list) */
535 #define EAF_TYPE_EC_SET 0x0e /* Set of pairs of u32's - ext. community list */
536 #define EAF_TYPE_LC_SET 0x12 /* Set of triplets of u32's - large community list */
537 #define EAF_TYPE_UNDEF 0x1f /* `force undefined' entry */
538 #define EAF_EMBEDDED 0x01 /* Data stored in eattr.u.data (part of type spec) */
539 #define EAF_VAR_LENGTH 0x02 /* Attribute length is variable (part of type spec) */
540 #define EAF_ORIGINATED 0x20 /* The attribute has originated locally */
541 #define EAF_FRESH 0x40 /* An uncached attribute (e.g. modified in export filter) */
542
543 typedef struct adata {
544 uint length; /* Length of data */
545 byte data[0];
546 } adata;
547
548 extern const adata null_adata; /* adata of length 0 */
549
550 static inline struct adata *
lp_alloc_adata(struct linpool * pool,uint len)551 lp_alloc_adata(struct linpool *pool, uint len)
552 {
553 struct adata *ad = lp_alloc(pool, sizeof(struct adata) + len);
554 ad->length = len;
555 return ad;
556 }
557
adata_same(const struct adata * a,const struct adata * b)558 static inline int adata_same(const struct adata *a, const struct adata *b)
559 { return (a->length == b->length && !memcmp(a->data, b->data, a->length)); }
560
561
562 typedef struct ea_list {
563 struct ea_list *next; /* In case we have an override list */
564 byte flags; /* Flags: EALF_... */
565 byte rfu;
566 word count; /* Number of attributes */
567 eattr attrs[0]; /* Attribute definitions themselves */
568 } ea_list;
569
570 #define EALF_SORTED 1 /* Attributes are sorted by code */
571 #define EALF_BISECT 2 /* Use interval bisection for searching */
572 #define EALF_CACHED 4 /* Attributes belonging to cached rta */
573 #define EALF_TEMP 8 /* Temporary ea_list added by make_tmp_attrs hooks */
574
575 struct rte_src *rt_find_source(struct proto *p, u32 id);
576 struct rte_src *rt_get_source(struct proto *p, u32 id);
rt_lock_source(struct rte_src * src)577 static inline void rt_lock_source(struct rte_src *src) { src->uc++; }
rt_unlock_source(struct rte_src * src)578 static inline void rt_unlock_source(struct rte_src *src) { src->uc--; }
579 void rt_prune_sources(void);
580
581 struct ea_walk_state {
582 ea_list *eattrs; /* Ccurrent ea_list, initially set by caller */
583 eattr *ea; /* Current eattr, initially NULL */
584 u32 visited[4]; /* Bitfield, limiting max to 128 */
585 };
586
587 eattr *ea_find(ea_list *, unsigned ea);
588 eattr *ea_walk(struct ea_walk_state *s, uint id, uint max);
589 int ea_get_int(ea_list *, unsigned ea, int def);
590 void ea_dump(ea_list *);
591 void ea_sort(ea_list *); /* Sort entries in all sub-lists */
592 unsigned ea_scan(ea_list *); /* How many bytes do we need for merged ea_list */
593 void ea_merge(ea_list *from, ea_list *to); /* Merge sub-lists to allocated buffer */
594 int ea_same(ea_list *x, ea_list *y); /* Test whether two ea_lists are identical */
595 uint ea_hash(ea_list *e); /* Calculate 16-bit hash value */
596 ea_list *ea_append(ea_list *to, ea_list *what);
597 void ea_format_bitfield(const struct eattr *a, byte *buf, int bufsize, const char **names, int min, int max);
598
599 #define ea_normalize(ea) do { \
600 if (ea->next) { \
601 ea_list *t = alloca(ea_scan(ea)); \
602 ea_merge(ea, t); \
603 ea = t; \
604 } \
605 ea_sort(ea); \
606 if (ea->count == 0) \
607 ea = NULL; \
608 } while(0) \
609
610 static inline eattr *
ea_set_attr(ea_list ** to,struct linpool * pool,uint id,uint flags,uint type,uintptr_t val)611 ea_set_attr(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, uintptr_t val)
612 {
613 ea_list *a = lp_alloc(pool, sizeof(ea_list) + sizeof(eattr));
614 eattr *e = &a->attrs[0];
615
616 a->flags = EALF_SORTED;
617 a->count = 1;
618 a->next = *to;
619 *to = a;
620
621 e->id = id;
622 e->type = type;
623 e->flags = flags;
624
625 if (type & EAF_EMBEDDED)
626 e->u.data = (u32) val;
627 else
628 e->u.ptr = (struct adata *) val;
629
630 return e;
631 }
632
633 static inline void
ea_set_attr_u32(ea_list ** to,struct linpool * pool,uint id,uint flags,uint type,u32 val)634 ea_set_attr_u32(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, u32 val)
635 { ea_set_attr(to, pool, id, flags, type, (uintptr_t) val); }
636
637 static inline void
ea_set_attr_ptr(ea_list ** to,struct linpool * pool,uint id,uint flags,uint type,struct adata * val)638 ea_set_attr_ptr(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, struct adata *val)
639 { ea_set_attr(to, pool, id, flags, type, (uintptr_t) val); }
640
641 static inline void
ea_set_attr_data(ea_list ** to,struct linpool * pool,uint id,uint flags,uint type,void * data,uint len)642 ea_set_attr_data(ea_list **to, struct linpool *pool, uint id, uint flags, uint type, void *data, uint len)
643 {
644 struct adata *a = lp_alloc_adata(pool, len);
645 memcpy(a->data, data, len);
646 ea_set_attr(to, pool, id, flags, type, (uintptr_t) a);
647 }
648
649
650 #define NEXTHOP_MAX_SIZE (sizeof(struct nexthop) + sizeof(u32)*MPLS_MAX_LABEL_STACK)
651
nexthop_size(const struct nexthop * nh)652 static inline size_t nexthop_size(const struct nexthop *nh)
653 { return sizeof(struct nexthop) + sizeof(u32)*nh->labels; }
654 int nexthop__same(struct nexthop *x, struct nexthop *y); /* Compare multipath nexthops */
nexthop_same(struct nexthop * x,struct nexthop * y)655 static inline int nexthop_same(struct nexthop *x, struct nexthop *y)
656 { return (x == y) || nexthop__same(x, y); }
657 struct nexthop *nexthop_merge(struct nexthop *x, struct nexthop *y, int rx, int ry, int max, linpool *lp);
658 struct nexthop *nexthop_sort(struct nexthop *x);
nexthop_link(struct rta * a,struct nexthop * from)659 static inline void nexthop_link(struct rta *a, struct nexthop *from)
660 { memcpy(&a->nh, from, nexthop_size(from)); }
661 void nexthop_insert(struct nexthop **n, struct nexthop *y);
662 int nexthop_is_sorted(struct nexthop *x);
663
664 void rta_init(void);
rta_size(const rta * a)665 static inline size_t rta_size(const rta *a) { return sizeof(rta) + sizeof(u32)*a->nh.labels; }
666 #define RTA_MAX_SIZE (sizeof(rta) + sizeof(u32)*MPLS_MAX_LABEL_STACK)
667 rta *rta_lookup(rta *); /* Get rta equivalent to this one, uc++ */
rta_is_cached(rta * r)668 static inline int rta_is_cached(rta *r) { return r->aflags & RTAF_CACHED; }
rta_clone(rta * r)669 static inline rta *rta_clone(rta *r) { r->uc++; return r; }
670 void rta__free(rta *r);
rta_free(rta * r)671 static inline void rta_free(rta *r) { if (r && !--r->uc) rta__free(r); }
672 rta *rta_do_cow(rta *o, linpool *lp);
rta_cow(rta * r,linpool * lp)673 static inline rta * rta_cow(rta *r, linpool *lp) { return rta_is_cached(r) ? rta_do_cow(r, lp) : r; }
674 void rta_dump(rta *);
675 void rta_dump_all(void);
676 void rta_show(struct cli *, rta *);
677
678 u32 rt_get_igp_metric(rte *rt);
679 struct hostentry * rt_get_hostentry(rtable *tab, ip_addr a, ip_addr ll, rtable *dep);
680 void rta_apply_hostentry(rta *a, struct hostentry *he, mpls_label_stack *mls);
681
682 static inline void
rta_set_recursive_next_hop(rtable * dep,rta * a,rtable * tab,ip_addr gw,ip_addr ll,mpls_label_stack * mls)683 rta_set_recursive_next_hop(rtable *dep, rta *a, rtable *tab, ip_addr gw, ip_addr ll, mpls_label_stack *mls)
684 {
685 rta_apply_hostentry(a, rt_get_hostentry(tab, gw, ll, dep), mls);
686 }
687
688 /*
689 * rta_set_recursive_next_hop() acquires hostentry from hostcache and fills
690 * rta->hostentry field. New hostentry has zero use count. Cached rta locks its
691 * hostentry (increases its use count), uncached rta does not lock it. Hostentry
692 * with zero use count is removed asynchronously during host cache update,
693 * therefore it is safe to hold such hostentry temorarily. Hostentry holds a
694 * lock for a 'source' rta, mainly to share multipath nexthops.
695 *
696 * There is no need to hold a lock for hostentry->dep table, because that table
697 * contains routes responsible for that hostentry, and therefore is non-empty if
698 * given hostentry has non-zero use count. If the hostentry has zero use count,
699 * the entry is removed before dep is referenced.
700 *
701 * The protocol responsible for routes with recursive next hops should hold a
702 * lock for a 'source' table governing that routes (argument tab to
703 * rta_set_recursive_next_hop()), because its routes reference hostentries
704 * (through rta) related to the governing table. When all such routes are
705 * removed, rtas are immediately removed achieving zero uc. Then the 'source'
706 * table lock could be immediately released, although hostentries may still
707 * exist - they will be freed together with the 'source' table.
708 */
709
rt_lock_hostentry(struct hostentry * he)710 static inline void rt_lock_hostentry(struct hostentry *he) { if (he) he->uc++; }
rt_unlock_hostentry(struct hostentry * he)711 static inline void rt_unlock_hostentry(struct hostentry *he) { if (he) he->uc--; }
712
713 /*
714 * Default protocol preferences
715 */
716
717 #define DEF_PREF_DIRECT 240 /* Directly connected */
718 #define DEF_PREF_STATIC 200 /* Static route */
719 #define DEF_PREF_OSPF 150 /* OSPF intra-area, inter-area and type 1 external routes */
720 #define DEF_PREF_BABEL 130 /* Babel */
721 #define DEF_PREF_RIP 120 /* RIP */
722 #define DEF_PREF_BGP 100 /* BGP */
723 #define DEF_PREF_RPKI 100 /* RPKI */
724 #define DEF_PREF_INHERITED 10 /* Routes inherited from other routing daemons */
725
726 /*
727 * Route Origin Authorization
728 */
729
730 #define ROA_UNKNOWN 0
731 #define ROA_VALID 1
732 #define ROA_INVALID 2
733
734 #endif
735