1 /* 2 * Linux INET6 implementation 3 * 4 * Authors: 5 * Pedro Roque <roque@di.fc.ul.pt> 6 * 7 * This program is free software; you can redistribute it and/or 8 * modify it under the terms of the GNU General Public License 9 * as published by the Free Software Foundation; either version 10 * 2 of the License, or (at your option) any later version. 11 */ 12 13 #ifndef _IP6_FIB_H 14 #define _IP6_FIB_H 15 16 #include <linux/ipv6_route.h> 17 #include <linux/rtnetlink.h> 18 #include <linux/spinlock.h> 19 #include <linux/notifier.h> 20 #include <net/dst.h> 21 #include <net/flow.h> 22 #include <net/ip_fib.h> 23 #include <net/netlink.h> 24 #include <net/inetpeer.h> 25 #include <net/fib_notifier.h> 26 27 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 28 #define FIB6_TABLE_HASHSZ 256 29 #else 30 #define FIB6_TABLE_HASHSZ 1 31 #endif 32 33 #define RT6_DEBUG 2 34 35 #if RT6_DEBUG >= 3 36 #define RT6_TRACE(x...) pr_debug(x) 37 #else 38 #define RT6_TRACE(x...) do { ; } while (0) 39 #endif 40 41 struct rt6_info; 42 struct fib6_info; 43 44 struct fib6_config { 45 u32 fc_table; 46 u32 fc_metric; 47 int fc_dst_len; 48 int fc_src_len; 49 int fc_ifindex; 50 u32 fc_flags; 51 u32 fc_protocol; 52 u16 fc_type; /* only 8 bits are used */ 53 u16 fc_delete_all_nh : 1, 54 fc_ignore_dev_down:1, 55 __unused : 14; 56 57 struct in6_addr fc_dst; 58 struct in6_addr fc_src; 59 struct in6_addr fc_prefsrc; 60 struct in6_addr fc_gateway; 61 62 unsigned long fc_expires; 63 struct nlattr *fc_mx; 64 int fc_mx_len; 65 int fc_mp_len; 66 struct nlattr *fc_mp; 67 68 struct nl_info fc_nlinfo; 69 struct nlattr *fc_encap; 70 u16 fc_encap_type; 71 }; 72 73 struct fib6_node { 74 struct fib6_node __rcu *parent; 75 struct fib6_node __rcu *left; 76 struct fib6_node __rcu *right; 77 #ifdef CONFIG_IPV6_SUBTREES 78 struct fib6_node __rcu *subtree; 79 #endif 80 struct fib6_info __rcu *leaf; 81 82 __u16 fn_bit; /* bit key */ 83 __u16 fn_flags; 84 int fn_sernum; 85 struct fib6_info __rcu *rr_ptr; 86 struct rcu_head rcu; 87 }; 88 89 struct fib6_gc_args { 90 int timeout; 91 int more; 92 }; 93 94 #ifndef CONFIG_IPV6_SUBTREES 95 #define FIB6_SUBTREE(fn) NULL 96 #else 97 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1)) 98 #endif 99 100 /* 101 * routing information 102 * 103 */ 104 105 struct rt6key { 106 struct in6_addr addr; 107 int plen; 108 }; 109 110 struct fib6_table; 111 112 struct rt6_exception_bucket { 113 struct hlist_head chain; 114 int depth; 115 }; 116 117 struct rt6_exception { 118 struct hlist_node hlist; 119 struct rt6_info *rt6i; 120 unsigned long stamp; 121 struct rcu_head rcu; 122 }; 123 124 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10 125 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT) 126 #define FIB6_MAX_DEPTH 5 127 128 struct fib6_nh { 129 struct fib_nh_common nh_common; 130 131 #ifdef CONFIG_IPV6_ROUTER_PREF 132 unsigned long last_probe; 133 #endif 134 }; 135 136 struct fib6_info { 137 struct fib6_table *fib6_table; 138 struct fib6_info __rcu *fib6_next; 139 struct fib6_node __rcu *fib6_node; 140 141 /* Multipath routes: 142 * siblings is a list of fib6_info that have the the same metric/weight, 143 * destination, but not the same gateway. nsiblings is just a cache 144 * to speed up lookup. 145 */ 146 struct list_head fib6_siblings; 147 unsigned int fib6_nsiblings; 148 149 refcount_t fib6_ref; 150 unsigned long expires; 151 struct dst_metrics *fib6_metrics; 152 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1] 153 154 struct rt6key fib6_dst; 155 u32 fib6_flags; 156 struct rt6key fib6_src; 157 struct rt6key fib6_prefsrc; 158 159 struct rt6_info * __percpu *rt6i_pcpu; 160 struct rt6_exception_bucket __rcu *rt6i_exception_bucket; 161 162 u32 fib6_metric; 163 u8 fib6_protocol; 164 u8 fib6_type; 165 u8 exception_bucket_flushed:1, 166 should_flush:1, 167 dst_nocount:1, 168 dst_nopolicy:1, 169 dst_host:1, 170 unused:3; 171 172 struct fib6_nh fib6_nh; 173 struct rcu_head rcu; 174 }; 175 176 struct rt6_info { 177 struct dst_entry dst; 178 struct fib6_info __rcu *from; 179 180 struct rt6key rt6i_dst; 181 struct rt6key rt6i_src; 182 struct in6_addr rt6i_gateway; 183 struct inet6_dev *rt6i_idev; 184 u32 rt6i_flags; 185 186 struct list_head rt6i_uncached; 187 struct uncached_list *rt6i_uncached_list; 188 189 /* more non-fragment space at head required */ 190 unsigned short rt6i_nfheader_len; 191 }; 192 193 struct fib6_result { 194 struct fib6_nh *nh; 195 struct fib6_info *f6i; 196 u32 fib6_flags; 197 u8 fib6_type; 198 struct rt6_info *rt6; 199 }; 200 201 #define for_each_fib6_node_rt_rcu(fn) \ 202 for (rt = rcu_dereference((fn)->leaf); rt; \ 203 rt = rcu_dereference(rt->fib6_next)) 204 205 #define for_each_fib6_walker_rt(w) \ 206 for (rt = (w)->leaf; rt; \ 207 rt = rcu_dereference_protected(rt->fib6_next, 1)) 208 209 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst) 210 { 211 return ((struct rt6_info *)dst)->rt6i_idev; 212 } 213 214 static inline void fib6_clean_expires(struct fib6_info *f6i) 215 { 216 f6i->fib6_flags &= ~RTF_EXPIRES; 217 f6i->expires = 0; 218 } 219 220 static inline void fib6_set_expires(struct fib6_info *f6i, 221 unsigned long expires) 222 { 223 f6i->expires = expires; 224 f6i->fib6_flags |= RTF_EXPIRES; 225 } 226 227 static inline bool fib6_check_expired(const struct fib6_info *f6i) 228 { 229 if (f6i->fib6_flags & RTF_EXPIRES) 230 return time_after(jiffies, f6i->expires); 231 return false; 232 } 233 234 /* Function to safely get fn->sernum for passed in rt 235 * and store result in passed in cookie. 236 * Return true if we can get cookie safely 237 * Return false if not 238 */ 239 static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i, 240 u32 *cookie) 241 { 242 struct fib6_node *fn; 243 bool status = false; 244 245 fn = rcu_dereference(f6i->fib6_node); 246 247 if (fn) { 248 *cookie = fn->fn_sernum; 249 /* pairs with smp_wmb() in fib6_update_sernum_upto_root() */ 250 smp_rmb(); 251 status = true; 252 } 253 254 return status; 255 } 256 257 static inline u32 rt6_get_cookie(const struct rt6_info *rt) 258 { 259 struct fib6_info *from; 260 u32 cookie = 0; 261 262 rcu_read_lock(); 263 264 from = rcu_dereference(rt->from); 265 if (from && (rt->rt6i_flags & RTF_PCPU || 266 unlikely(!list_empty(&rt->rt6i_uncached)))) 267 fib6_get_cookie_safe(from, &cookie); 268 269 rcu_read_unlock(); 270 271 return cookie; 272 } 273 274 static inline void ip6_rt_put(struct rt6_info *rt) 275 { 276 /* dst_release() accepts a NULL parameter. 277 * We rely on dst being first structure in struct rt6_info 278 */ 279 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0); 280 dst_release(&rt->dst); 281 } 282 283 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags); 284 void fib6_info_destroy_rcu(struct rcu_head *head); 285 286 static inline void fib6_info_hold(struct fib6_info *f6i) 287 { 288 refcount_inc(&f6i->fib6_ref); 289 } 290 291 static inline bool fib6_info_hold_safe(struct fib6_info *f6i) 292 { 293 return refcount_inc_not_zero(&f6i->fib6_ref); 294 } 295 296 static inline void fib6_info_release(struct fib6_info *f6i) 297 { 298 if (f6i && refcount_dec_and_test(&f6i->fib6_ref)) 299 call_rcu(&f6i->rcu, fib6_info_destroy_rcu); 300 } 301 302 enum fib6_walk_state { 303 #ifdef CONFIG_IPV6_SUBTREES 304 FWS_S, 305 #endif 306 FWS_L, 307 FWS_R, 308 FWS_C, 309 FWS_U 310 }; 311 312 struct fib6_walker { 313 struct list_head lh; 314 struct fib6_node *root, *node; 315 struct fib6_info *leaf; 316 enum fib6_walk_state state; 317 unsigned int skip; 318 unsigned int count; 319 int (*func)(struct fib6_walker *); 320 void *args; 321 }; 322 323 struct rt6_statistics { 324 __u32 fib_nodes; /* all fib6 nodes */ 325 __u32 fib_route_nodes; /* intermediate nodes */ 326 __u32 fib_rt_entries; /* rt entries in fib table */ 327 __u32 fib_rt_cache; /* cached rt entries in exception table */ 328 __u32 fib_discarded_routes; /* total number of routes delete */ 329 330 /* The following stats are not protected by any lock */ 331 atomic_t fib_rt_alloc; /* total number of routes alloced */ 332 atomic_t fib_rt_uncache; /* rt entries in uncached list */ 333 }; 334 335 #define RTN_TL_ROOT 0x0001 336 #define RTN_ROOT 0x0002 /* tree root node */ 337 #define RTN_RTINFO 0x0004 /* node with valid routing info */ 338 339 /* 340 * priority levels (or metrics) 341 * 342 */ 343 344 345 struct fib6_table { 346 struct hlist_node tb6_hlist; 347 u32 tb6_id; 348 spinlock_t tb6_lock; 349 struct fib6_node tb6_root; 350 struct inet_peer_base tb6_peers; 351 unsigned int flags; 352 unsigned int fib_seq; 353 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0) 354 }; 355 356 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC 357 #define RT6_TABLE_MAIN RT_TABLE_MAIN 358 #define RT6_TABLE_DFLT RT6_TABLE_MAIN 359 #define RT6_TABLE_INFO RT6_TABLE_MAIN 360 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN 361 362 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 363 #define FIB6_TABLE_MIN 1 364 #define FIB6_TABLE_MAX RT_TABLE_MAX 365 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL 366 #else 367 #define FIB6_TABLE_MIN RT_TABLE_MAIN 368 #define FIB6_TABLE_MAX FIB6_TABLE_MIN 369 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN 370 #endif 371 372 typedef struct rt6_info *(*pol_lookup_t)(struct net *, 373 struct fib6_table *, 374 struct flowi6 *, 375 const struct sk_buff *, int); 376 377 struct fib6_entry_notifier_info { 378 struct fib_notifier_info info; /* must be first */ 379 struct fib6_info *rt; 380 }; 381 382 /* 383 * exported functions 384 */ 385 386 struct fib6_table *fib6_get_table(struct net *net, u32 id); 387 struct fib6_table *fib6_new_table(struct net *net, u32 id); 388 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6, 389 const struct sk_buff *skb, 390 int flags, pol_lookup_t lookup); 391 392 /* called with rcu lock held; can return error pointer 393 * caller needs to select path 394 */ 395 int fib6_lookup(struct net *net, int oif, struct flowi6 *fl6, 396 struct fib6_result *res, int flags); 397 398 /* called with rcu lock held; caller needs to select path */ 399 int fib6_table_lookup(struct net *net, struct fib6_table *table, 400 int oif, struct flowi6 *fl6, struct fib6_result *res, 401 int strict); 402 403 void fib6_select_path(const struct net *net, struct fib6_result *res, 404 struct flowi6 *fl6, int oif, bool have_oif_match, 405 const struct sk_buff *skb, int strict); 406 struct fib6_node *fib6_node_lookup(struct fib6_node *root, 407 const struct in6_addr *daddr, 408 const struct in6_addr *saddr); 409 410 struct fib6_node *fib6_locate(struct fib6_node *root, 411 const struct in6_addr *daddr, int dst_len, 412 const struct in6_addr *saddr, int src_len, 413 bool exact_match); 414 415 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg), 416 void *arg); 417 void fib6_clean_all_skip_notify(struct net *net, 418 int (*func)(struct fib6_info *, void *arg), 419 void *arg); 420 421 int fib6_add(struct fib6_node *root, struct fib6_info *rt, 422 struct nl_info *info, struct netlink_ext_ack *extack); 423 int fib6_del(struct fib6_info *rt, struct nl_info *info); 424 425 static inline 426 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr) 427 { 428 const struct fib6_info *from; 429 430 rcu_read_lock(); 431 432 from = rcu_dereference(rt->from); 433 if (from) { 434 *addr = from->fib6_prefsrc.addr; 435 } else { 436 struct in6_addr in6_zero = {}; 437 438 *addr = in6_zero; 439 } 440 441 rcu_read_unlock(); 442 } 443 444 static inline struct net_device *fib6_info_nh_dev(const struct fib6_info *f6i) 445 { 446 return f6i->fib6_nh.fib_nh_dev; 447 } 448 449 int fib6_nh_init(struct net *net, struct fib6_nh *fib6_nh, 450 struct fib6_config *cfg, gfp_t gfp_flags, 451 struct netlink_ext_ack *extack); 452 void fib6_nh_release(struct fib6_nh *fib6_nh); 453 454 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info, 455 unsigned int flags); 456 457 void fib6_run_gc(unsigned long expires, struct net *net, bool force); 458 459 void fib6_gc_cleanup(void); 460 461 int fib6_init(void); 462 463 struct ipv6_route_iter { 464 struct seq_net_private p; 465 struct fib6_walker w; 466 loff_t skip; 467 struct fib6_table *tbl; 468 int sernum; 469 }; 470 471 extern const struct seq_operations ipv6_route_seq_ops; 472 473 int call_fib6_notifier(struct notifier_block *nb, struct net *net, 474 enum fib_event_type event_type, 475 struct fib_notifier_info *info); 476 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type, 477 struct fib_notifier_info *info); 478 479 int __net_init fib6_notifier_init(struct net *net); 480 void __net_exit fib6_notifier_exit(struct net *net); 481 482 unsigned int fib6_tables_seq_read(struct net *net); 483 int fib6_tables_dump(struct net *net, struct notifier_block *nb); 484 485 void fib6_update_sernum(struct net *net, struct fib6_info *rt); 486 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt); 487 488 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val); 489 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric) 490 { 491 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric)); 492 } 493 494 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 495 int fib6_rules_init(void); 496 void fib6_rules_cleanup(void); 497 bool fib6_rule_default(const struct fib_rule *rule); 498 int fib6_rules_dump(struct net *net, struct notifier_block *nb); 499 unsigned int fib6_rules_seq_read(struct net *net); 500 501 static inline bool fib6_rules_early_flow_dissect(struct net *net, 502 struct sk_buff *skb, 503 struct flowi6 *fl6, 504 struct flow_keys *flkeys) 505 { 506 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 507 508 if (!net->ipv6.fib6_rules_require_fldissect) 509 return false; 510 511 skb_flow_dissect_flow_keys(skb, flkeys, flag); 512 fl6->fl6_sport = flkeys->ports.src; 513 fl6->fl6_dport = flkeys->ports.dst; 514 fl6->flowi6_proto = flkeys->basic.ip_proto; 515 516 return true; 517 } 518 #else 519 static inline int fib6_rules_init(void) 520 { 521 return 0; 522 } 523 static inline void fib6_rules_cleanup(void) 524 { 525 return ; 526 } 527 static inline bool fib6_rule_default(const struct fib_rule *rule) 528 { 529 return true; 530 } 531 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb) 532 { 533 return 0; 534 } 535 static inline unsigned int fib6_rules_seq_read(struct net *net) 536 { 537 return 0; 538 } 539 static inline bool fib6_rules_early_flow_dissect(struct net *net, 540 struct sk_buff *skb, 541 struct flowi6 *fl6, 542 struct flow_keys *flkeys) 543 { 544 return false; 545 } 546 #endif 547 #endif 548