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 /* more non-fragment space at head required */ 221 unsigned short rt6i_nfheader_len; 222 }; 223 224 struct fib6_result { 225 struct fib6_nh *nh; 226 struct fib6_info *f6i; 227 u32 fib6_flags; 228 u8 fib6_type; 229 struct rt6_info *rt6; 230 }; 231 232 #define for_each_fib6_node_rt_rcu(fn) \ 233 for (rt = rcu_dereference((fn)->leaf); rt; \ 234 rt = rcu_dereference(rt->fib6_next)) 235 236 #define for_each_fib6_walker_rt(w) \ 237 for (rt = (w)->leaf; rt; \ 238 rt = rcu_dereference_protected(rt->fib6_next, 1)) 239 240 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst) 241 { 242 return ((struct rt6_info *)dst)->rt6i_idev; 243 } 244 245 static inline bool fib6_requires_src(const struct fib6_info *rt) 246 { 247 return rt->fib6_src.plen > 0; 248 } 249 250 static inline void fib6_clean_expires(struct fib6_info *f6i) 251 { 252 f6i->fib6_flags &= ~RTF_EXPIRES; 253 f6i->expires = 0; 254 } 255 256 static inline void fib6_set_expires(struct fib6_info *f6i, 257 unsigned long expires) 258 { 259 f6i->expires = expires; 260 f6i->fib6_flags |= RTF_EXPIRES; 261 } 262 263 static inline bool fib6_check_expired(const struct fib6_info *f6i) 264 { 265 if (f6i->fib6_flags & RTF_EXPIRES) 266 return time_after(jiffies, f6i->expires); 267 return false; 268 } 269 270 /* Function to safely get fn->fn_sernum for passed in rt 271 * and store result in passed in cookie. 272 * Return true if we can get cookie safely 273 * Return false if not 274 */ 275 static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i, 276 u32 *cookie) 277 { 278 struct fib6_node *fn; 279 bool status = false; 280 281 fn = rcu_dereference(f6i->fib6_node); 282 283 if (fn) { 284 *cookie = READ_ONCE(fn->fn_sernum); 285 /* pairs with smp_wmb() in __fib6_update_sernum_upto_root() */ 286 smp_rmb(); 287 status = true; 288 } 289 290 return status; 291 } 292 293 static inline u32 rt6_get_cookie(const struct rt6_info *rt) 294 { 295 struct fib6_info *from; 296 u32 cookie = 0; 297 298 if (rt->sernum) 299 return rt->sernum; 300 301 rcu_read_lock(); 302 303 from = rcu_dereference(rt->from); 304 if (from) 305 fib6_get_cookie_safe(from, &cookie); 306 307 rcu_read_unlock(); 308 309 return cookie; 310 } 311 312 static inline void ip6_rt_put(struct rt6_info *rt) 313 { 314 /* dst_release() accepts a NULL parameter. 315 * We rely on dst being first structure in struct rt6_info 316 */ 317 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0); 318 dst_release(&rt->dst); 319 } 320 321 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags, bool with_fib6_nh); 322 void fib6_info_destroy_rcu(struct rcu_head *head); 323 324 static inline void fib6_info_hold(struct fib6_info *f6i) 325 { 326 refcount_inc(&f6i->fib6_ref); 327 } 328 329 static inline bool fib6_info_hold_safe(struct fib6_info *f6i) 330 { 331 return refcount_inc_not_zero(&f6i->fib6_ref); 332 } 333 334 static inline void fib6_info_release(struct fib6_info *f6i) 335 { 336 if (f6i && refcount_dec_and_test(&f6i->fib6_ref)) 337 call_rcu(&f6i->rcu, fib6_info_destroy_rcu); 338 } 339 340 enum fib6_walk_state { 341 #ifdef CONFIG_IPV6_SUBTREES 342 FWS_S, 343 #endif 344 FWS_L, 345 FWS_R, 346 FWS_C, 347 FWS_U 348 }; 349 350 struct fib6_walker { 351 struct list_head lh; 352 struct fib6_node *root, *node; 353 struct fib6_info *leaf; 354 enum fib6_walk_state state; 355 unsigned int skip; 356 unsigned int count; 357 unsigned int skip_in_node; 358 int (*func)(struct fib6_walker *); 359 void *args; 360 }; 361 362 struct rt6_statistics { 363 __u32 fib_nodes; /* all fib6 nodes */ 364 __u32 fib_route_nodes; /* intermediate nodes */ 365 __u32 fib_rt_entries; /* rt entries in fib table */ 366 __u32 fib_rt_cache; /* cached rt entries in exception table */ 367 __u32 fib_discarded_routes; /* total number of routes delete */ 368 369 /* The following stat is not protected by any lock */ 370 atomic_t fib_rt_alloc; /* total number of routes alloced */ 371 }; 372 373 #define RTN_TL_ROOT 0x0001 374 #define RTN_ROOT 0x0002 /* tree root node */ 375 #define RTN_RTINFO 0x0004 /* node with valid routing info */ 376 377 /* 378 * priority levels (or metrics) 379 * 380 */ 381 382 383 struct fib6_table { 384 struct hlist_node tb6_hlist; 385 u32 tb6_id; 386 spinlock_t tb6_lock; 387 struct fib6_node tb6_root; 388 struct inet_peer_base tb6_peers; 389 unsigned int flags; 390 unsigned int fib_seq; 391 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0) 392 }; 393 394 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC 395 #define RT6_TABLE_MAIN RT_TABLE_MAIN 396 #define RT6_TABLE_DFLT RT6_TABLE_MAIN 397 #define RT6_TABLE_INFO RT6_TABLE_MAIN 398 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN 399 400 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 401 #define FIB6_TABLE_MIN 1 402 #define FIB6_TABLE_MAX RT_TABLE_MAX 403 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL 404 #else 405 #define FIB6_TABLE_MIN RT_TABLE_MAIN 406 #define FIB6_TABLE_MAX FIB6_TABLE_MIN 407 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN 408 #endif 409 410 typedef struct rt6_info *(*pol_lookup_t)(struct net *, 411 struct fib6_table *, 412 struct flowi6 *, 413 const struct sk_buff *, int); 414 415 struct fib6_entry_notifier_info { 416 struct fib_notifier_info info; /* must be first */ 417 struct fib6_info *rt; 418 unsigned int nsiblings; 419 }; 420 421 /* 422 * exported functions 423 */ 424 425 struct fib6_table *fib6_get_table(struct net *net, u32 id); 426 struct fib6_table *fib6_new_table(struct net *net, u32 id); 427 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6, 428 const struct sk_buff *skb, 429 int flags, pol_lookup_t lookup); 430 431 /* called with rcu lock held; can return error pointer 432 * caller needs to select path 433 */ 434 int fib6_lookup(struct net *net, int oif, struct flowi6 *fl6, 435 struct fib6_result *res, int flags); 436 437 /* called with rcu lock held; caller needs to select path */ 438 int fib6_table_lookup(struct net *net, struct fib6_table *table, 439 int oif, struct flowi6 *fl6, struct fib6_result *res, 440 int strict); 441 442 void fib6_select_path(const struct net *net, struct fib6_result *res, 443 struct flowi6 *fl6, int oif, bool have_oif_match, 444 const struct sk_buff *skb, int strict); 445 struct fib6_node *fib6_node_lookup(struct fib6_node *root, 446 const struct in6_addr *daddr, 447 const struct in6_addr *saddr); 448 449 struct fib6_node *fib6_locate(struct fib6_node *root, 450 const struct in6_addr *daddr, int dst_len, 451 const struct in6_addr *saddr, int src_len, 452 bool exact_match); 453 454 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg), 455 void *arg); 456 void fib6_clean_all_skip_notify(struct net *net, 457 int (*func)(struct fib6_info *, void *arg), 458 void *arg); 459 460 int fib6_add(struct fib6_node *root, struct fib6_info *rt, 461 struct nl_info *info, struct netlink_ext_ack *extack); 462 int fib6_del(struct fib6_info *rt, struct nl_info *info); 463 464 static inline 465 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr) 466 { 467 const struct fib6_info *from; 468 469 rcu_read_lock(); 470 471 from = rcu_dereference(rt->from); 472 if (from) 473 *addr = from->fib6_prefsrc.addr; 474 else 475 *addr = in6addr_any; 476 477 rcu_read_unlock(); 478 } 479 480 int fib6_nh_init(struct net *net, struct fib6_nh *fib6_nh, 481 struct fib6_config *cfg, gfp_t gfp_flags, 482 struct netlink_ext_ack *extack); 483 void fib6_nh_release(struct fib6_nh *fib6_nh); 484 void fib6_nh_release_dsts(struct fib6_nh *fib6_nh); 485 486 int call_fib6_entry_notifiers(struct net *net, 487 enum fib_event_type event_type, 488 struct fib6_info *rt, 489 struct netlink_ext_ack *extack); 490 int call_fib6_multipath_entry_notifiers(struct net *net, 491 enum fib_event_type event_type, 492 struct fib6_info *rt, 493 unsigned int nsiblings, 494 struct netlink_ext_ack *extack); 495 int call_fib6_entry_notifiers_replace(struct net *net, struct fib6_info *rt); 496 void fib6_rt_update(struct net *net, struct fib6_info *rt, 497 struct nl_info *info); 498 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info, 499 unsigned int flags); 500 501 void fib6_run_gc(unsigned long expires, struct net *net, bool force); 502 503 void fib6_gc_cleanup(void); 504 505 int fib6_init(void); 506 507 struct ipv6_route_iter { 508 struct seq_net_private p; 509 struct fib6_walker w; 510 loff_t skip; 511 struct fib6_table *tbl; 512 int sernum; 513 }; 514 515 extern const struct seq_operations ipv6_route_seq_ops; 516 517 int call_fib6_notifier(struct notifier_block *nb, 518 enum fib_event_type event_type, 519 struct fib_notifier_info *info); 520 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type, 521 struct fib_notifier_info *info); 522 523 int __net_init fib6_notifier_init(struct net *net); 524 void __net_exit fib6_notifier_exit(struct net *net); 525 526 unsigned int fib6_tables_seq_read(struct net *net); 527 int fib6_tables_dump(struct net *net, struct notifier_block *nb, 528 struct netlink_ext_ack *extack); 529 530 void fib6_update_sernum(struct net *net, struct fib6_info *rt); 531 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt); 532 void fib6_update_sernum_stub(struct net *net, struct fib6_info *f6i); 533 534 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val); 535 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric) 536 { 537 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric)); 538 } 539 void fib6_info_hw_flags_set(struct net *net, struct fib6_info *f6i, 540 bool offload, bool trap, bool offload_failed); 541 542 #if IS_BUILTIN(CONFIG_IPV6) && defined(CONFIG_BPF_SYSCALL) 543 struct bpf_iter__ipv6_route { 544 __bpf_md_ptr(struct bpf_iter_meta *, meta); 545 __bpf_md_ptr(struct fib6_info *, rt); 546 }; 547 #endif 548 549 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_output(struct net *net, 550 struct fib6_table *table, 551 struct flowi6 *fl6, 552 const struct sk_buff *skb, 553 int flags)); 554 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_input(struct net *net, 555 struct fib6_table *table, 556 struct flowi6 *fl6, 557 const struct sk_buff *skb, 558 int flags)); 559 INDIRECT_CALLABLE_DECLARE(struct rt6_info *__ip6_route_redirect(struct net *net, 560 struct fib6_table *table, 561 struct flowi6 *fl6, 562 const struct sk_buff *skb, 563 int flags)); 564 INDIRECT_CALLABLE_DECLARE(struct rt6_info *ip6_pol_route_lookup(struct net *net, 565 struct fib6_table *table, 566 struct flowi6 *fl6, 567 const struct sk_buff *skb, 568 int flags)); 569 static inline struct rt6_info *pol_lookup_func(pol_lookup_t lookup, 570 struct net *net, 571 struct fib6_table *table, 572 struct flowi6 *fl6, 573 const struct sk_buff *skb, 574 int flags) 575 { 576 return INDIRECT_CALL_4(lookup, 577 ip6_pol_route_output, 578 ip6_pol_route_input, 579 ip6_pol_route_lookup, 580 __ip6_route_redirect, 581 net, table, fl6, skb, flags); 582 } 583 584 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 585 static inline bool fib6_has_custom_rules(const struct net *net) 586 { 587 return net->ipv6.fib6_has_custom_rules; 588 } 589 590 int fib6_rules_init(void); 591 void fib6_rules_cleanup(void); 592 bool fib6_rule_default(const struct fib_rule *rule); 593 int fib6_rules_dump(struct net *net, struct notifier_block *nb, 594 struct netlink_ext_ack *extack); 595 unsigned int fib6_rules_seq_read(struct net *net); 596 597 static inline bool fib6_rules_early_flow_dissect(struct net *net, 598 struct sk_buff *skb, 599 struct flowi6 *fl6, 600 struct flow_keys *flkeys) 601 { 602 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 603 604 if (!net->ipv6.fib6_rules_require_fldissect) 605 return false; 606 607 memset(flkeys, 0, sizeof(*flkeys)); 608 __skb_flow_dissect(net, skb, &flow_keys_dissector, 609 flkeys, NULL, 0, 0, 0, flag); 610 611 fl6->fl6_sport = flkeys->ports.src; 612 fl6->fl6_dport = flkeys->ports.dst; 613 fl6->flowi6_proto = flkeys->basic.ip_proto; 614 615 return true; 616 } 617 #else 618 static inline bool fib6_has_custom_rules(const struct net *net) 619 { 620 return false; 621 } 622 static inline int fib6_rules_init(void) 623 { 624 return 0; 625 } 626 static inline void fib6_rules_cleanup(void) 627 { 628 return ; 629 } 630 static inline bool fib6_rule_default(const struct fib_rule *rule) 631 { 632 return true; 633 } 634 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb, 635 struct netlink_ext_ack *extack) 636 { 637 return 0; 638 } 639 static inline unsigned int fib6_rules_seq_read(struct net *net) 640 { 641 return 0; 642 } 643 static inline bool fib6_rules_early_flow_dissect(struct net *net, 644 struct sk_buff *skb, 645 struct flowi6 *fl6, 646 struct flow_keys *flkeys) 647 { 648 return false; 649 } 650 #endif 651 #endif 652