1 /* SPDX-License-Identifier: GPL-2.0-or-later */ 2 /* 3 * INET An implementation of the TCP/IP protocol suite for the LINUX 4 * operating system. INET is implemented using the BSD Socket 5 * interface as the means of communication with the user level. 6 * 7 * Definitions for the Forwarding Information Base. 8 * 9 * Authors: A.N.Kuznetsov, <kuznet@ms2.inr.ac.ru> 10 */ 11 12 #ifndef _NET_IP_FIB_H 13 #define _NET_IP_FIB_H 14 15 #include <net/flow.h> 16 #include <linux/seq_file.h> 17 #include <linux/rcupdate.h> 18 #include <net/fib_notifier.h> 19 #include <net/fib_rules.h> 20 #include <net/inetpeer.h> 21 #include <linux/percpu.h> 22 #include <linux/notifier.h> 23 #include <linux/refcount.h> 24 25 struct fib_config { 26 u8 fc_dst_len; 27 u8 fc_tos; 28 u8 fc_protocol; 29 u8 fc_scope; 30 u8 fc_type; 31 u8 fc_gw_family; 32 /* 2 bytes unused */ 33 u32 fc_table; 34 __be32 fc_dst; 35 union { 36 __be32 fc_gw4; 37 struct in6_addr fc_gw6; 38 }; 39 int fc_oif; 40 u32 fc_flags; 41 u32 fc_priority; 42 __be32 fc_prefsrc; 43 u32 fc_nh_id; 44 struct nlattr *fc_mx; 45 struct rtnexthop *fc_mp; 46 int fc_mx_len; 47 int fc_mp_len; 48 u32 fc_flow; 49 u32 fc_nlflags; 50 struct nl_info fc_nlinfo; 51 struct nlattr *fc_encap; 52 u16 fc_encap_type; 53 }; 54 55 struct fib_info; 56 struct rtable; 57 58 struct fib_nh_exception { 59 struct fib_nh_exception __rcu *fnhe_next; 60 int fnhe_genid; 61 __be32 fnhe_daddr; 62 u32 fnhe_pmtu; 63 bool fnhe_mtu_locked; 64 __be32 fnhe_gw; 65 unsigned long fnhe_expires; 66 struct rtable __rcu *fnhe_rth_input; 67 struct rtable __rcu *fnhe_rth_output; 68 unsigned long fnhe_stamp; 69 struct rcu_head rcu; 70 }; 71 72 struct fnhe_hash_bucket { 73 struct fib_nh_exception __rcu *chain; 74 }; 75 76 #define FNHE_HASH_SHIFT 11 77 #define FNHE_HASH_SIZE (1 << FNHE_HASH_SHIFT) 78 #define FNHE_RECLAIM_DEPTH 5 79 80 struct fib_nh_common { 81 struct net_device *nhc_dev; 82 netdevice_tracker nhc_dev_tracker; 83 int nhc_oif; 84 unsigned char nhc_scope; 85 u8 nhc_family; 86 u8 nhc_gw_family; 87 unsigned char nhc_flags; 88 struct lwtunnel_state *nhc_lwtstate; 89 90 union { 91 __be32 ipv4; 92 struct in6_addr ipv6; 93 } nhc_gw; 94 95 int nhc_weight; 96 atomic_t nhc_upper_bound; 97 98 /* v4 specific, but allows fib6_nh with v4 routes */ 99 struct rtable __rcu * __percpu *nhc_pcpu_rth_output; 100 struct rtable __rcu *nhc_rth_input; 101 struct fnhe_hash_bucket __rcu *nhc_exceptions; 102 }; 103 104 struct fib_nh { 105 struct fib_nh_common nh_common; 106 struct hlist_node nh_hash; 107 struct fib_info *nh_parent; 108 #ifdef CONFIG_IP_ROUTE_CLASSID 109 __u32 nh_tclassid; 110 #endif 111 __be32 nh_saddr; 112 int nh_saddr_genid; 113 #define fib_nh_family nh_common.nhc_family 114 #define fib_nh_dev nh_common.nhc_dev 115 #define fib_nh_dev_tracker nh_common.nhc_dev_tracker 116 #define fib_nh_oif nh_common.nhc_oif 117 #define fib_nh_flags nh_common.nhc_flags 118 #define fib_nh_lws nh_common.nhc_lwtstate 119 #define fib_nh_scope nh_common.nhc_scope 120 #define fib_nh_gw_family nh_common.nhc_gw_family 121 #define fib_nh_gw4 nh_common.nhc_gw.ipv4 122 #define fib_nh_gw6 nh_common.nhc_gw.ipv6 123 #define fib_nh_weight nh_common.nhc_weight 124 #define fib_nh_upper_bound nh_common.nhc_upper_bound 125 }; 126 127 /* 128 * This structure contains data shared by many of routes. 129 */ 130 131 struct nexthop; 132 133 struct fib_info { 134 struct hlist_node fib_hash; 135 struct hlist_node fib_lhash; 136 struct list_head nh_list; 137 struct net *fib_net; 138 refcount_t fib_treeref; 139 refcount_t fib_clntref; 140 unsigned int fib_flags; 141 unsigned char fib_dead; 142 unsigned char fib_protocol; 143 unsigned char fib_scope; 144 unsigned char fib_type; 145 __be32 fib_prefsrc; 146 u32 fib_tb_id; 147 u32 fib_priority; 148 struct dst_metrics *fib_metrics; 149 #define fib_mtu fib_metrics->metrics[RTAX_MTU-1] 150 #define fib_window fib_metrics->metrics[RTAX_WINDOW-1] 151 #define fib_rtt fib_metrics->metrics[RTAX_RTT-1] 152 #define fib_advmss fib_metrics->metrics[RTAX_ADVMSS-1] 153 int fib_nhs; 154 bool fib_nh_is_v6; 155 bool nh_updated; 156 struct nexthop *nh; 157 struct rcu_head rcu; 158 struct fib_nh fib_nh[]; 159 }; 160 161 162 #ifdef CONFIG_IP_MULTIPLE_TABLES 163 struct fib_rule; 164 #endif 165 166 struct fib_table; 167 struct fib_result { 168 __be32 prefix; 169 unsigned char prefixlen; 170 unsigned char nh_sel; 171 unsigned char type; 172 unsigned char scope; 173 u32 tclassid; 174 struct fib_nh_common *nhc; 175 struct fib_info *fi; 176 struct fib_table *table; 177 struct hlist_head *fa_head; 178 }; 179 180 struct fib_result_nl { 181 __be32 fl_addr; /* To be looked up*/ 182 u32 fl_mark; 183 unsigned char fl_tos; 184 unsigned char fl_scope; 185 unsigned char tb_id_in; 186 187 unsigned char tb_id; /* Results */ 188 unsigned char prefixlen; 189 unsigned char nh_sel; 190 unsigned char type; 191 unsigned char scope; 192 int err; 193 }; 194 195 #ifdef CONFIG_IP_MULTIPLE_TABLES 196 #define FIB_TABLE_HASHSZ 256 197 #else 198 #define FIB_TABLE_HASHSZ 2 199 #endif 200 201 __be32 fib_info_update_nhc_saddr(struct net *net, struct fib_nh_common *nhc, 202 unsigned char scope); 203 __be32 fib_result_prefsrc(struct net *net, struct fib_result *res); 204 205 #define FIB_RES_NHC(res) ((res).nhc) 206 #define FIB_RES_DEV(res) (FIB_RES_NHC(res)->nhc_dev) 207 #define FIB_RES_OIF(res) (FIB_RES_NHC(res)->nhc_oif) 208 209 struct fib_rt_info { 210 struct fib_info *fi; 211 u32 tb_id; 212 __be32 dst; 213 int dst_len; 214 u8 tos; 215 u8 type; 216 u8 offload:1, 217 trap:1, 218 offload_failed:1, 219 unused:5; 220 }; 221 222 struct fib_entry_notifier_info { 223 struct fib_notifier_info info; /* must be first */ 224 u32 dst; 225 int dst_len; 226 struct fib_info *fi; 227 u8 tos; 228 u8 type; 229 u32 tb_id; 230 }; 231 232 struct fib_nh_notifier_info { 233 struct fib_notifier_info info; /* must be first */ 234 struct fib_nh *fib_nh; 235 }; 236 237 int call_fib4_notifier(struct notifier_block *nb, 238 enum fib_event_type event_type, 239 struct fib_notifier_info *info); 240 int call_fib4_notifiers(struct net *net, enum fib_event_type event_type, 241 struct fib_notifier_info *info); 242 243 int __net_init fib4_notifier_init(struct net *net); 244 void __net_exit fib4_notifier_exit(struct net *net); 245 246 void fib_info_notify_update(struct net *net, struct nl_info *info); 247 int fib_notify(struct net *net, struct notifier_block *nb, 248 struct netlink_ext_ack *extack); 249 250 struct fib_table { 251 struct hlist_node tb_hlist; 252 u32 tb_id; 253 int tb_num_default; 254 struct rcu_head rcu; 255 unsigned long *tb_data; 256 unsigned long __data[]; 257 }; 258 259 struct fib_dump_filter { 260 u32 table_id; 261 /* filter_set is an optimization that an entry is set */ 262 bool filter_set; 263 bool dump_routes; 264 bool dump_exceptions; 265 unsigned char protocol; 266 unsigned char rt_type; 267 unsigned int flags; 268 struct net_device *dev; 269 }; 270 271 int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp, 272 struct fib_result *res, int fib_flags); 273 int fib_table_insert(struct net *, struct fib_table *, struct fib_config *, 274 struct netlink_ext_ack *extack); 275 int fib_table_delete(struct net *, struct fib_table *, struct fib_config *, 276 struct netlink_ext_ack *extack); 277 int fib_table_dump(struct fib_table *table, struct sk_buff *skb, 278 struct netlink_callback *cb, struct fib_dump_filter *filter); 279 int fib_table_flush(struct net *net, struct fib_table *table, bool flush_all); 280 struct fib_table *fib_trie_unmerge(struct fib_table *main_tb); 281 void fib_table_flush_external(struct fib_table *table); 282 void fib_free_table(struct fib_table *tb); 283 284 #ifndef CONFIG_IP_MULTIPLE_TABLES 285 286 #define TABLE_LOCAL_INDEX (RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1)) 287 #define TABLE_MAIN_INDEX (RT_TABLE_MAIN & (FIB_TABLE_HASHSZ - 1)) 288 289 static inline struct fib_table *fib_get_table(struct net *net, u32 id) 290 { 291 struct hlist_node *tb_hlist; 292 struct hlist_head *ptr; 293 294 ptr = id == RT_TABLE_LOCAL ? 295 &net->ipv4.fib_table_hash[TABLE_LOCAL_INDEX] : 296 &net->ipv4.fib_table_hash[TABLE_MAIN_INDEX]; 297 298 tb_hlist = rcu_dereference_rtnl(hlist_first_rcu(ptr)); 299 300 return hlist_entry(tb_hlist, struct fib_table, tb_hlist); 301 } 302 303 static inline struct fib_table *fib_new_table(struct net *net, u32 id) 304 { 305 return fib_get_table(net, id); 306 } 307 308 static inline int fib_lookup(struct net *net, const struct flowi4 *flp, 309 struct fib_result *res, unsigned int flags) 310 { 311 struct fib_table *tb; 312 int err = -ENETUNREACH; 313 314 rcu_read_lock(); 315 316 tb = fib_get_table(net, RT_TABLE_MAIN); 317 if (tb) 318 err = fib_table_lookup(tb, flp, res, flags | FIB_LOOKUP_NOREF); 319 320 if (err == -EAGAIN) 321 err = -ENETUNREACH; 322 323 rcu_read_unlock(); 324 325 return err; 326 } 327 328 static inline bool fib4_has_custom_rules(const struct net *net) 329 { 330 return false; 331 } 332 333 static inline bool fib4_rule_default(const struct fib_rule *rule) 334 { 335 return true; 336 } 337 338 static inline int fib4_rules_dump(struct net *net, struct notifier_block *nb, 339 struct netlink_ext_ack *extack) 340 { 341 return 0; 342 } 343 344 static inline unsigned int fib4_rules_seq_read(struct net *net) 345 { 346 return 0; 347 } 348 349 static inline bool fib4_rules_early_flow_dissect(struct net *net, 350 struct sk_buff *skb, 351 struct flowi4 *fl4, 352 struct flow_keys *flkeys) 353 { 354 return false; 355 } 356 #else /* CONFIG_IP_MULTIPLE_TABLES */ 357 int __net_init fib4_rules_init(struct net *net); 358 void __net_exit fib4_rules_exit(struct net *net); 359 360 struct fib_table *fib_new_table(struct net *net, u32 id); 361 struct fib_table *fib_get_table(struct net *net, u32 id); 362 363 int __fib_lookup(struct net *net, struct flowi4 *flp, 364 struct fib_result *res, unsigned int flags); 365 366 static inline int fib_lookup(struct net *net, struct flowi4 *flp, 367 struct fib_result *res, unsigned int flags) 368 { 369 struct fib_table *tb; 370 int err = -ENETUNREACH; 371 372 flags |= FIB_LOOKUP_NOREF; 373 if (net->ipv4.fib_has_custom_rules) 374 return __fib_lookup(net, flp, res, flags); 375 376 rcu_read_lock(); 377 378 res->tclassid = 0; 379 380 tb = rcu_dereference_rtnl(net->ipv4.fib_main); 381 if (tb) 382 err = fib_table_lookup(tb, flp, res, flags); 383 384 if (!err) 385 goto out; 386 387 tb = rcu_dereference_rtnl(net->ipv4.fib_default); 388 if (tb) 389 err = fib_table_lookup(tb, flp, res, flags); 390 391 out: 392 if (err == -EAGAIN) 393 err = -ENETUNREACH; 394 395 rcu_read_unlock(); 396 397 return err; 398 } 399 400 static inline bool fib4_has_custom_rules(const struct net *net) 401 { 402 return net->ipv4.fib_has_custom_rules; 403 } 404 405 bool fib4_rule_default(const struct fib_rule *rule); 406 int fib4_rules_dump(struct net *net, struct notifier_block *nb, 407 struct netlink_ext_ack *extack); 408 unsigned int fib4_rules_seq_read(struct net *net); 409 410 static inline bool fib4_rules_early_flow_dissect(struct net *net, 411 struct sk_buff *skb, 412 struct flowi4 *fl4, 413 struct flow_keys *flkeys) 414 { 415 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 416 417 if (!net->ipv4.fib_rules_require_fldissect) 418 return false; 419 420 skb_flow_dissect_flow_keys(skb, flkeys, flag); 421 fl4->fl4_sport = flkeys->ports.src; 422 fl4->fl4_dport = flkeys->ports.dst; 423 fl4->flowi4_proto = flkeys->basic.ip_proto; 424 425 return true; 426 } 427 428 #endif /* CONFIG_IP_MULTIPLE_TABLES */ 429 430 /* Exported by fib_frontend.c */ 431 extern const struct nla_policy rtm_ipv4_policy[]; 432 void ip_fib_init(void); 433 int fib_gw_from_via(struct fib_config *cfg, struct nlattr *nla, 434 struct netlink_ext_ack *extack); 435 __be32 fib_compute_spec_dst(struct sk_buff *skb); 436 bool fib_info_nh_uses_dev(struct fib_info *fi, const struct net_device *dev); 437 int fib_validate_source(struct sk_buff *skb, __be32 src, __be32 dst, 438 u8 tos, int oif, struct net_device *dev, 439 struct in_device *idev, u32 *itag); 440 #ifdef CONFIG_IP_ROUTE_CLASSID 441 static inline int fib_num_tclassid_users(struct net *net) 442 { 443 return atomic_read(&net->ipv4.fib_num_tclassid_users); 444 } 445 #else 446 static inline int fib_num_tclassid_users(struct net *net) 447 { 448 return 0; 449 } 450 #endif 451 int fib_unmerge(struct net *net); 452 453 static inline bool nhc_l3mdev_matches_dev(const struct fib_nh_common *nhc, 454 const struct net_device *dev) 455 { 456 if (nhc->nhc_dev == dev || 457 l3mdev_master_ifindex_rcu(nhc->nhc_dev) == dev->ifindex) 458 return true; 459 460 return false; 461 } 462 463 /* Exported by fib_semantics.c */ 464 int ip_fib_check_default(__be32 gw, struct net_device *dev); 465 int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force); 466 int fib_sync_down_addr(struct net_device *dev, __be32 local); 467 int fib_sync_up(struct net_device *dev, unsigned char nh_flags); 468 void fib_sync_mtu(struct net_device *dev, u32 orig_mtu); 469 void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig); 470 471 /* Fields used for sysctl_fib_multipath_hash_fields. 472 * Common to IPv4 and IPv6. 473 * 474 * Add new fields at the end. This is user API. 475 */ 476 #define FIB_MULTIPATH_HASH_FIELD_SRC_IP BIT(0) 477 #define FIB_MULTIPATH_HASH_FIELD_DST_IP BIT(1) 478 #define FIB_MULTIPATH_HASH_FIELD_IP_PROTO BIT(2) 479 #define FIB_MULTIPATH_HASH_FIELD_FLOWLABEL BIT(3) 480 #define FIB_MULTIPATH_HASH_FIELD_SRC_PORT BIT(4) 481 #define FIB_MULTIPATH_HASH_FIELD_DST_PORT BIT(5) 482 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP BIT(6) 483 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP BIT(7) 484 #define FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO BIT(8) 485 #define FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL BIT(9) 486 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT BIT(10) 487 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT BIT(11) 488 489 #define FIB_MULTIPATH_HASH_FIELD_OUTER_MASK \ 490 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 491 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 492 FIB_MULTIPATH_HASH_FIELD_IP_PROTO | \ 493 FIB_MULTIPATH_HASH_FIELD_FLOWLABEL | \ 494 FIB_MULTIPATH_HASH_FIELD_SRC_PORT | \ 495 FIB_MULTIPATH_HASH_FIELD_DST_PORT) 496 497 #define FIB_MULTIPATH_HASH_FIELD_INNER_MASK \ 498 (FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP | \ 499 FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP | \ 500 FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO | \ 501 FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL | \ 502 FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT | \ 503 FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT) 504 505 #define FIB_MULTIPATH_HASH_FIELD_ALL_MASK \ 506 (FIB_MULTIPATH_HASH_FIELD_OUTER_MASK | \ 507 FIB_MULTIPATH_HASH_FIELD_INNER_MASK) 508 509 #define FIB_MULTIPATH_HASH_FIELD_DEFAULT_MASK \ 510 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 511 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 512 FIB_MULTIPATH_HASH_FIELD_IP_PROTO) 513 514 #ifdef CONFIG_IP_ROUTE_MULTIPATH 515 int fib_multipath_hash(const struct net *net, const struct flowi4 *fl4, 516 const struct sk_buff *skb, struct flow_keys *flkeys); 517 #endif 518 int fib_check_nh(struct net *net, struct fib_nh *nh, u32 table, u8 scope, 519 struct netlink_ext_ack *extack); 520 void fib_select_multipath(struct fib_result *res, int hash); 521 void fib_select_path(struct net *net, struct fib_result *res, 522 struct flowi4 *fl4, const struct sk_buff *skb); 523 524 int fib_nh_init(struct net *net, struct fib_nh *fib_nh, 525 struct fib_config *cfg, int nh_weight, 526 struct netlink_ext_ack *extack); 527 void fib_nh_release(struct net *net, struct fib_nh *fib_nh); 528 int fib_nh_common_init(struct net *net, struct fib_nh_common *nhc, 529 struct nlattr *fc_encap, u16 fc_encap_type, 530 void *cfg, gfp_t gfp_flags, 531 struct netlink_ext_ack *extack); 532 void fib_nh_common_release(struct fib_nh_common *nhc); 533 534 /* Exported by fib_trie.c */ 535 void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri); 536 void fib_trie_init(void); 537 struct fib_table *fib_trie_table(u32 id, struct fib_table *alias); 538 bool fib_lookup_good_nhc(const struct fib_nh_common *nhc, int fib_flags, 539 const struct flowi4 *flp); 540 541 static inline void fib_combine_itag(u32 *itag, const struct fib_result *res) 542 { 543 #ifdef CONFIG_IP_ROUTE_CLASSID 544 struct fib_nh_common *nhc = res->nhc; 545 #ifdef CONFIG_IP_MULTIPLE_TABLES 546 u32 rtag; 547 #endif 548 if (nhc->nhc_family == AF_INET) { 549 struct fib_nh *nh; 550 551 nh = container_of(nhc, struct fib_nh, nh_common); 552 *itag = nh->nh_tclassid << 16; 553 } else { 554 *itag = 0; 555 } 556 557 #ifdef CONFIG_IP_MULTIPLE_TABLES 558 rtag = res->tclassid; 559 if (*itag == 0) 560 *itag = (rtag<<16); 561 *itag |= (rtag>>16); 562 #endif 563 #endif 564 } 565 566 void fib_flush(struct net *net); 567 void free_fib_info(struct fib_info *fi); 568 569 static inline void fib_info_hold(struct fib_info *fi) 570 { 571 refcount_inc(&fi->fib_clntref); 572 } 573 574 static inline void fib_info_put(struct fib_info *fi) 575 { 576 if (refcount_dec_and_test(&fi->fib_clntref)) 577 free_fib_info(fi); 578 } 579 580 #ifdef CONFIG_PROC_FS 581 int __net_init fib_proc_init(struct net *net); 582 void __net_exit fib_proc_exit(struct net *net); 583 #else 584 static inline int fib_proc_init(struct net *net) 585 { 586 return 0; 587 } 588 static inline void fib_proc_exit(struct net *net) 589 { 590 } 591 #endif 592 593 u32 ip_mtu_from_fib_result(struct fib_result *res, __be32 daddr); 594 595 int ip_valid_fib_dump_req(struct net *net, const struct nlmsghdr *nlh, 596 struct fib_dump_filter *filter, 597 struct netlink_callback *cb); 598 599 int fib_nexthop_info(struct sk_buff *skb, const struct fib_nh_common *nh, 600 u8 rt_family, unsigned char *flags, bool skip_oif); 601 int fib_add_nexthop(struct sk_buff *skb, const struct fib_nh_common *nh, 602 int nh_weight, u8 rt_family, u32 nh_tclassid); 603 #endif /* _NET_FIB_H */ 604