1 /* SPDX-License-Identifier: GPL-2.0-or-later */ 2 /* 3 * Linux INET6 implementation 4 * 5 * Authors: 6 * Pedro Roque <roque@di.fc.ul.pt> 7 */ 8 9 #ifndef _IP6_FIB_H 10 #define _IP6_FIB_H 11 12 #include <linux/ipv6_route.h> 13 #include <linux/rtnetlink.h> 14 #include <linux/spinlock.h> 15 #include <linux/notifier.h> 16 #include <net/dst.h> 17 #include <net/flow.h> 18 #include <net/ip_fib.h> 19 #include <net/netlink.h> 20 #include <net/inetpeer.h> 21 #include <net/fib_notifier.h> 22 #include <linux/indirect_call_wrapper.h> 23 #include <uapi/linux/bpf.h> 24 25 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 26 #define FIB6_TABLE_HASHSZ 256 27 #else 28 #define FIB6_TABLE_HASHSZ 1 29 #endif 30 31 #define RT6_DEBUG 2 32 33 #if RT6_DEBUG >= 3 34 #define RT6_TRACE(x...) pr_debug(x) 35 #else 36 #define RT6_TRACE(x...) do { ; } while (0) 37 #endif 38 39 struct rt6_info; 40 struct fib6_info; 41 42 struct fib6_config { 43 u32 fc_table; 44 u32 fc_metric; 45 int fc_dst_len; 46 int fc_src_len; 47 int fc_ifindex; 48 u32 fc_flags; 49 u32 fc_protocol; 50 u16 fc_type; /* only 8 bits are used */ 51 u16 fc_delete_all_nh : 1, 52 fc_ignore_dev_down:1, 53 __unused : 14; 54 u32 fc_nh_id; 55 56 struct in6_addr fc_dst; 57 struct in6_addr fc_src; 58 struct in6_addr fc_prefsrc; 59 struct in6_addr fc_gateway; 60 61 unsigned long fc_expires; 62 struct nlattr *fc_mx; 63 int fc_mx_len; 64 int fc_mp_len; 65 struct nlattr *fc_mp; 66 67 struct nl_info fc_nlinfo; 68 struct nlattr *fc_encap; 69 u16 fc_encap_type; 70 bool fc_is_fdb; 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 97 static inline bool fib6_routes_require_src(const struct net *net) 98 { 99 return false; 100 } 101 102 static inline void fib6_routes_require_src_inc(struct net *net) {} 103 static inline void fib6_routes_require_src_dec(struct net *net) {} 104 105 #else 106 107 static inline bool fib6_routes_require_src(const struct net *net) 108 { 109 return net->ipv6.fib6_routes_require_src > 0; 110 } 111 112 static inline void fib6_routes_require_src_inc(struct net *net) 113 { 114 net->ipv6.fib6_routes_require_src++; 115 } 116 117 static inline void fib6_routes_require_src_dec(struct net *net) 118 { 119 net->ipv6.fib6_routes_require_src--; 120 } 121 122 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1)) 123 #endif 124 125 /* 126 * routing information 127 * 128 */ 129 130 struct rt6key { 131 struct in6_addr addr; 132 int plen; 133 }; 134 135 struct fib6_table; 136 137 struct rt6_exception_bucket { 138 struct hlist_head chain; 139 int depth; 140 }; 141 142 struct rt6_exception { 143 struct hlist_node hlist; 144 struct rt6_info *rt6i; 145 unsigned long stamp; 146 struct rcu_head rcu; 147 }; 148 149 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10 150 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT) 151 #define FIB6_MAX_DEPTH 5 152 153 struct fib6_nh { 154 struct fib_nh_common nh_common; 155 156 #ifdef CONFIG_IPV6_ROUTER_PREF 157 unsigned long last_probe; 158 #endif 159 160 struct rt6_info * __percpu *rt6i_pcpu; 161 struct rt6_exception_bucket __rcu *rt6i_exception_bucket; 162 }; 163 164 struct fib6_info { 165 struct fib6_table *fib6_table; 166 struct fib6_info __rcu *fib6_next; 167 struct fib6_node __rcu *fib6_node; 168 169 /* Multipath routes: 170 * siblings is a list of fib6_info that have the same metric/weight, 171 * destination, but not the same gateway. nsiblings is just a cache 172 * to speed up lookup. 173 */ 174 union { 175 struct list_head fib6_siblings; 176 struct list_head nh_list; 177 }; 178 unsigned int fib6_nsiblings; 179 180 refcount_t fib6_ref; 181 unsigned long expires; 182 struct dst_metrics *fib6_metrics; 183 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1] 184 185 struct rt6key fib6_dst; 186 u32 fib6_flags; 187 struct rt6key fib6_src; 188 struct rt6key fib6_prefsrc; 189 190 u32 fib6_metric; 191 u8 fib6_protocol; 192 u8 fib6_type; 193 194 u8 offload; 195 u8 trap; 196 u8 offload_failed; 197 198 u8 should_flush:1, 199 dst_nocount:1, 200 dst_nopolicy:1, 201 fib6_destroying:1, 202 unused:4; 203 204 struct rcu_head rcu; 205 struct nexthop *nh; 206 struct fib6_nh fib6_nh[]; 207 }; 208 209 struct rt6_info { 210 struct dst_entry dst; 211 struct fib6_info __rcu *from; 212 int sernum; 213 214 struct rt6key rt6i_dst; 215 struct rt6key rt6i_src; 216 struct in6_addr rt6i_gateway; 217 struct inet6_dev *rt6i_idev; 218 u32 rt6i_flags; 219 220 struct list_head rt6i_uncached; 221 struct uncached_list *rt6i_uncached_list; 222 223 /* more non-fragment space at head required */ 224 unsigned short rt6i_nfheader_len; 225 }; 226 227 struct fib6_result { 228 struct fib6_nh *nh; 229 struct fib6_info *f6i; 230 u32 fib6_flags; 231 u8 fib6_type; 232 struct rt6_info *rt6; 233 }; 234 235 #define for_each_fib6_node_rt_rcu(fn) \ 236 for (rt = rcu_dereference((fn)->leaf); rt; \ 237 rt = rcu_dereference(rt->fib6_next)) 238 239 #define for_each_fib6_walker_rt(w) \ 240 for (rt = (w)->leaf; rt; \ 241 rt = rcu_dereference_protected(rt->fib6_next, 1)) 242 243 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst) 244 { 245 return ((struct rt6_info *)dst)->rt6i_idev; 246 } 247 248 static inline bool fib6_requires_src(const struct fib6_info *rt) 249 { 250 return rt->fib6_src.plen > 0; 251 } 252 253 static inline void fib6_clean_expires(struct fib6_info *f6i) 254 { 255 f6i->fib6_flags &= ~RTF_EXPIRES; 256 f6i->expires = 0; 257 } 258 259 static inline void fib6_set_expires(struct fib6_info *f6i, 260 unsigned long expires) 261 { 262 f6i->expires = expires; 263 f6i->fib6_flags |= RTF_EXPIRES; 264 } 265 266 static inline bool fib6_check_expired(const struct fib6_info *f6i) 267 { 268 if (f6i->fib6_flags & RTF_EXPIRES) 269 return time_after(jiffies, f6i->expires); 270 return false; 271 } 272 273 /* Function to safely get fn->fn_sernum for passed in rt 274 * and store result in passed in cookie. 275 * Return true if we can get cookie safely 276 * Return false if not 277 */ 278 static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i, 279 u32 *cookie) 280 { 281 struct fib6_node *fn; 282 bool status = false; 283 284 fn = rcu_dereference(f6i->fib6_node); 285 286 if (fn) { 287 *cookie = READ_ONCE(fn->fn_sernum); 288 /* pairs with smp_wmb() in __fib6_update_sernum_upto_root() */ 289 smp_rmb(); 290 status = true; 291 } 292 293 return status; 294 } 295 296 static inline u32 rt6_get_cookie(const struct rt6_info *rt) 297 { 298 struct fib6_info *from; 299 u32 cookie = 0; 300 301 if (rt->sernum) 302 return rt->sernum; 303 304 rcu_read_lock(); 305 306 from = rcu_dereference(rt->from); 307 if (from) 308 fib6_get_cookie_safe(from, &cookie); 309 310 rcu_read_unlock(); 311 312 return cookie; 313 } 314 315 static inline void ip6_rt_put(struct rt6_info *rt) 316 { 317 /* dst_release() accepts a NULL parameter. 318 * We rely on dst being first structure in struct rt6_info 319 */ 320 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0); 321 dst_release(&rt->dst); 322 } 323 324 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags, bool with_fib6_nh); 325 void fib6_info_destroy_rcu(struct rcu_head *head); 326 327 static inline void fib6_info_hold(struct fib6_info *f6i) 328 { 329 refcount_inc(&f6i->fib6_ref); 330 } 331 332 static inline bool fib6_info_hold_safe(struct fib6_info *f6i) 333 { 334 return refcount_inc_not_zero(&f6i->fib6_ref); 335 } 336 337 static inline void fib6_info_release(struct fib6_info *f6i) 338 { 339 if (f6i && refcount_dec_and_test(&f6i->fib6_ref)) 340 call_rcu(&f6i->rcu, fib6_info_destroy_rcu); 341 } 342 343 enum fib6_walk_state { 344 #ifdef CONFIG_IPV6_SUBTREES 345 FWS_S, 346 #endif 347 FWS_L, 348 FWS_R, 349 FWS_C, 350 FWS_U 351 }; 352 353 struct fib6_walker { 354 struct list_head lh; 355 struct fib6_node *root, *node; 356 struct fib6_info *leaf; 357 enum fib6_walk_state state; 358 unsigned int skip; 359 unsigned int count; 360 unsigned int skip_in_node; 361 int (*func)(struct fib6_walker *); 362 void *args; 363 }; 364 365 struct rt6_statistics { 366 __u32 fib_nodes; /* all fib6 nodes */ 367 __u32 fib_route_nodes; /* intermediate nodes */ 368 __u32 fib_rt_entries; /* rt entries in fib table */ 369 __u32 fib_rt_cache; /* cached rt entries in exception table */ 370 __u32 fib_discarded_routes; /* total number of routes delete */ 371 372 /* The following stats are not protected by any lock */ 373 atomic_t fib_rt_alloc; /* total number of routes alloced */ 374 atomic_t fib_rt_uncache; /* rt entries in uncached list */ 375 }; 376 377 #define RTN_TL_ROOT 0x0001 378 #define RTN_ROOT 0x0002 /* tree root node */ 379 #define RTN_RTINFO 0x0004 /* node with valid routing info */ 380 381 /* 382 * priority levels (or metrics) 383 * 384 */ 385 386 387 struct fib6_table { 388 struct hlist_node tb6_hlist; 389 u32 tb6_id; 390 spinlock_t tb6_lock; 391 struct fib6_node tb6_root; 392 struct inet_peer_base tb6_peers; 393 unsigned int flags; 394 unsigned int fib_seq; 395 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0) 396 }; 397 398 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC 399 #define RT6_TABLE_MAIN RT_TABLE_MAIN 400 #define RT6_TABLE_DFLT RT6_TABLE_MAIN 401 #define RT6_TABLE_INFO RT6_TABLE_MAIN 402 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN 403 404 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 405 #define FIB6_TABLE_MIN 1 406 #define FIB6_TABLE_MAX RT_TABLE_MAX 407 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL 408 #else 409 #define FIB6_TABLE_MIN RT_TABLE_MAIN 410 #define FIB6_TABLE_MAX FIB6_TABLE_MIN 411 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN 412 #endif 413 414 typedef struct rt6_info *(*pol_lookup_t)(struct net *, 415 struct fib6_table *, 416 struct flowi6 *, 417 const struct sk_buff *, int); 418 419 struct fib6_entry_notifier_info { 420 struct fib_notifier_info info; /* must be first */ 421 struct fib6_info *rt; 422 unsigned int nsiblings; 423 }; 424 425 /* 426 * exported functions 427 */ 428 429 struct fib6_table *fib6_get_table(struct net *net, u32 id); 430 struct fib6_table *fib6_new_table(struct net *net, u32 id); 431 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6, 432 const struct sk_buff *skb, 433 int flags, pol_lookup_t lookup); 434 435 /* called with rcu lock held; can return error pointer 436 * caller needs to select path 437 */ 438 int fib6_lookup(struct net *net, int oif, struct flowi6 *fl6, 439 struct fib6_result *res, int flags); 440 441 /* called with rcu lock held; caller needs to select path */ 442 int fib6_table_lookup(struct net *net, struct fib6_table *table, 443 int oif, struct flowi6 *fl6, struct fib6_result *res, 444 int strict); 445 446 void fib6_select_path(const struct net *net, struct fib6_result *res, 447 struct flowi6 *fl6, int oif, bool have_oif_match, 448 const struct sk_buff *skb, int strict); 449 struct fib6_node *fib6_node_lookup(struct fib6_node *root, 450 const struct in6_addr *daddr, 451 const struct in6_addr *saddr); 452 453 struct fib6_node *fib6_locate(struct fib6_node *root, 454 const struct in6_addr *daddr, int dst_len, 455 const struct in6_addr *saddr, int src_len, 456 bool exact_match); 457 458 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg), 459 void *arg); 460 void fib6_clean_all_skip_notify(struct net *net, 461 int (*func)(struct fib6_info *, void *arg), 462 void *arg); 463 464 int fib6_add(struct fib6_node *root, struct fib6_info *rt, 465 struct nl_info *info, struct netlink_ext_ack *extack); 466 int fib6_del(struct fib6_info *rt, struct nl_info *info); 467 468 static inline 469 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr) 470 { 471 const struct fib6_info *from; 472 473 rcu_read_lock(); 474 475 from = rcu_dereference(rt->from); 476 if (from) { 477 *addr = from->fib6_prefsrc.addr; 478 } else { 479 struct in6_addr in6_zero = {}; 480 481 *addr = in6_zero; 482 } 483 484 rcu_read_unlock(); 485 } 486 487 int fib6_nh_init(struct net *net, struct fib6_nh *fib6_nh, 488 struct fib6_config *cfg, gfp_t gfp_flags, 489 struct netlink_ext_ack *extack); 490 void fib6_nh_release(struct fib6_nh *fib6_nh); 491 void fib6_nh_release_dsts(struct fib6_nh *fib6_nh); 492 493 int call_fib6_entry_notifiers(struct net *net, 494 enum fib_event_type event_type, 495 struct fib6_info *rt, 496 struct netlink_ext_ack *extack); 497 int call_fib6_multipath_entry_notifiers(struct net *net, 498 enum fib_event_type event_type, 499 struct fib6_info *rt, 500 unsigned int nsiblings, 501 struct netlink_ext_ack *extack); 502 int call_fib6_entry_notifiers_replace(struct net *net, struct fib6_info *rt); 503 void fib6_rt_update(struct net *net, struct fib6_info *rt, 504 struct nl_info *info); 505 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info, 506 unsigned int flags); 507 508 void fib6_run_gc(unsigned long expires, struct net *net, bool force); 509 510 void fib6_gc_cleanup(void); 511 512 int fib6_init(void); 513 514 struct ipv6_route_iter { 515 struct seq_net_private p; 516 struct fib6_walker w; 517 loff_t skip; 518 struct fib6_table *tbl; 519 int sernum; 520 }; 521 522 extern const struct seq_operations ipv6_route_seq_ops; 523 524 int call_fib6_notifier(struct notifier_block *nb, 525 enum fib_event_type event_type, 526 struct fib_notifier_info *info); 527 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type, 528 struct fib_notifier_info *info); 529 530 int __net_init fib6_notifier_init(struct net *net); 531 void __net_exit fib6_notifier_exit(struct net *net); 532 533 unsigned int fib6_tables_seq_read(struct net *net); 534 int fib6_tables_dump(struct net *net, struct notifier_block *nb, 535 struct netlink_ext_ack *extack); 536 537 void fib6_update_sernum(struct net *net, struct fib6_info *rt); 538 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt); 539 void fib6_update_sernum_stub(struct net *net, struct fib6_info *f6i); 540 541 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val); 542 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric) 543 { 544 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric)); 545 } 546 void fib6_info_hw_flags_set(struct net *net, struct fib6_info *f6i, 547 bool offload, bool trap, bool offload_failed); 548 549 #if IS_BUILTIN(CONFIG_IPV6) && defined(CONFIG_BPF_SYSCALL) 550 struct bpf_iter__ipv6_route { 551 __bpf_md_ptr(struct bpf_iter_meta *, meta); 552 __bpf_md_ptr(struct fib6_info *, rt); 553 }; 554 #endif 555 556 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_output(struct net *net, 557 struct fib6_table *table, 558 struct flowi6 *fl6, 559 const struct sk_buff *skb, 560 int flags)); 561 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_input(struct net *net, 562 struct fib6_table *table, 563 struct flowi6 *fl6, 564 const struct sk_buff *skb, 565 int flags)); 566 INDIRECT_CALLABLE_DECLARE(struct rt6_info *__ip6_route_redirect(struct net *net, 567 struct fib6_table *table, 568 struct flowi6 *fl6, 569 const struct sk_buff *skb, 570 int flags)); 571 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_lookup(struct net *net, 572 struct fib6_table *table, 573 struct flowi6 *fl6, 574 const struct sk_buff *skb, 575 int flags)); 576 static inline struct rt6_info *pol_lookup_func(pol_lookup_t lookup, 577 struct net *net, 578 struct fib6_table *table, 579 struct flowi6 *fl6, 580 const struct sk_buff *skb, 581 int flags) 582 { 583 return INDIRECT_CALL_4(lookup, 584 ip6_pol_route_output, 585 ip6_pol_route_input, 586 ip6_pol_route_lookup, 587 __ip6_route_redirect, 588 net, table, fl6, skb, flags); 589 } 590 591 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 592 static inline bool fib6_has_custom_rules(const struct net *net) 593 { 594 return net->ipv6.fib6_has_custom_rules; 595 } 596 597 int fib6_rules_init(void); 598 void fib6_rules_cleanup(void); 599 bool fib6_rule_default(const struct fib_rule *rule); 600 int fib6_rules_dump(struct net *net, struct notifier_block *nb, 601 struct netlink_ext_ack *extack); 602 unsigned int fib6_rules_seq_read(struct net *net); 603 604 static inline bool fib6_rules_early_flow_dissect(struct net *net, 605 struct sk_buff *skb, 606 struct flowi6 *fl6, 607 struct flow_keys *flkeys) 608 { 609 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 610 611 if (!net->ipv6.fib6_rules_require_fldissect) 612 return false; 613 614 skb_flow_dissect_flow_keys(skb, flkeys, flag); 615 fl6->fl6_sport = flkeys->ports.src; 616 fl6->fl6_dport = flkeys->ports.dst; 617 fl6->flowi6_proto = flkeys->basic.ip_proto; 618 619 return true; 620 } 621 #else 622 static inline bool fib6_has_custom_rules(const struct net *net) 623 { 624 return false; 625 } 626 static inline int fib6_rules_init(void) 627 { 628 return 0; 629 } 630 static inline void fib6_rules_cleanup(void) 631 { 632 return ; 633 } 634 static inline bool fib6_rule_default(const struct fib_rule *rule) 635 { 636 return true; 637 } 638 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb, 639 struct netlink_ext_ack *extack) 640 { 641 return 0; 642 } 643 static inline unsigned int fib6_rules_seq_read(struct net *net) 644 { 645 return 0; 646 } 647 static inline bool fib6_rules_early_flow_dissect(struct net *net, 648 struct sk_buff *skb, 649 struct flowi6 *fl6, 650 struct flow_keys *flkeys) 651 { 652 return false; 653 } 654 #endif 655 #endif 656