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/inet_dscp.h> 21 #include <net/inetpeer.h> 22 #include <linux/percpu.h> 23 #include <linux/notifier.h> 24 #include <linux/refcount.h> 25 26 struct fib_config { 27 u8 fc_dst_len; 28 dscp_t fc_dscp; 29 u8 fc_protocol; 30 u8 fc_scope; 31 u8 fc_type; 32 u8 fc_gw_family; 33 /* 2 bytes unused */ 34 u32 fc_table; 35 __be32 fc_dst; 36 union { 37 __be32 fc_gw4; 38 struct in6_addr fc_gw6; 39 }; 40 int fc_oif; 41 u32 fc_flags; 42 u32 fc_priority; 43 __be32 fc_prefsrc; 44 u32 fc_nh_id; 45 struct nlattr *fc_mx; 46 struct rtnexthop *fc_mp; 47 int fc_mx_len; 48 int fc_mp_len; 49 u32 fc_flow; 50 u32 fc_nlflags; 51 struct nl_info fc_nlinfo; 52 struct nlattr *fc_encap; 53 u16 fc_encap_type; 54 }; 55 56 struct fib_info; 57 struct rtable; 58 59 struct fib_nh_exception { 60 struct fib_nh_exception __rcu *fnhe_next; 61 int fnhe_genid; 62 __be32 fnhe_daddr; 63 u32 fnhe_pmtu; 64 bool fnhe_mtu_locked; 65 __be32 fnhe_gw; 66 unsigned long fnhe_expires; 67 struct rtable __rcu *fnhe_rth_input; 68 struct rtable __rcu *fnhe_rth_output; 69 unsigned long fnhe_stamp; 70 struct rcu_head rcu; 71 }; 72 73 struct fnhe_hash_bucket { 74 struct fib_nh_exception __rcu *chain; 75 }; 76 77 #define FNHE_HASH_SHIFT 11 78 #define FNHE_HASH_SIZE (1 << FNHE_HASH_SHIFT) 79 #define FNHE_RECLAIM_DEPTH 5 80 81 struct fib_nh_common { 82 struct net_device *nhc_dev; 83 netdevice_tracker nhc_dev_tracker; 84 int nhc_oif; 85 unsigned char nhc_scope; 86 u8 nhc_family; 87 u8 nhc_gw_family; 88 unsigned char nhc_flags; 89 struct lwtunnel_state *nhc_lwtstate; 90 91 union { 92 __be32 ipv4; 93 struct in6_addr ipv6; 94 } nhc_gw; 95 96 int nhc_weight; 97 atomic_t nhc_upper_bound; 98 99 /* v4 specific, but allows fib6_nh with v4 routes */ 100 struct rtable __rcu * __percpu *nhc_pcpu_rth_output; 101 struct rtable __rcu *nhc_rth_input; 102 struct fnhe_hash_bucket __rcu *nhc_exceptions; 103 }; 104 105 struct fib_nh { 106 struct fib_nh_common nh_common; 107 struct hlist_node nh_hash; 108 struct fib_info *nh_parent; 109 #ifdef CONFIG_IP_ROUTE_CLASSID 110 __u32 nh_tclassid; 111 #endif 112 __be32 nh_saddr; 113 int nh_saddr_genid; 114 #define fib_nh_family nh_common.nhc_family 115 #define fib_nh_dev nh_common.nhc_dev 116 #define fib_nh_dev_tracker nh_common.nhc_dev_tracker 117 #define fib_nh_oif nh_common.nhc_oif 118 #define fib_nh_flags nh_common.nhc_flags 119 #define fib_nh_lws nh_common.nhc_lwtstate 120 #define fib_nh_scope nh_common.nhc_scope 121 #define fib_nh_gw_family nh_common.nhc_gw_family 122 #define fib_nh_gw4 nh_common.nhc_gw.ipv4 123 #define fib_nh_gw6 nh_common.nhc_gw.ipv6 124 #define fib_nh_weight nh_common.nhc_weight 125 #define fib_nh_upper_bound nh_common.nhc_upper_bound 126 }; 127 128 /* 129 * This structure contains data shared by many of routes. 130 */ 131 132 struct nexthop; 133 134 struct fib_info { 135 struct hlist_node fib_hash; 136 struct hlist_node fib_lhash; 137 struct list_head nh_list; 138 struct net *fib_net; 139 refcount_t fib_treeref; 140 refcount_t fib_clntref; 141 unsigned int fib_flags; 142 unsigned char fib_dead; 143 unsigned char fib_protocol; 144 unsigned char fib_scope; 145 unsigned char fib_type; 146 __be32 fib_prefsrc; 147 u32 fib_tb_id; 148 u32 fib_priority; 149 struct dst_metrics *fib_metrics; 150 #define fib_mtu fib_metrics->metrics[RTAX_MTU-1] 151 #define fib_window fib_metrics->metrics[RTAX_WINDOW-1] 152 #define fib_rtt fib_metrics->metrics[RTAX_RTT-1] 153 #define fib_advmss fib_metrics->metrics[RTAX_ADVMSS-1] 154 int fib_nhs; 155 bool fib_nh_is_v6; 156 bool nh_updated; 157 bool pfsrc_removed; 158 struct nexthop *nh; 159 struct rcu_head rcu; 160 struct fib_nh fib_nh[]; 161 }; 162 163 164 #ifdef CONFIG_IP_MULTIPLE_TABLES 165 struct fib_rule; 166 #endif 167 168 struct fib_table; 169 struct fib_result { 170 __be32 prefix; 171 unsigned char prefixlen; 172 unsigned char nh_sel; 173 unsigned char type; 174 unsigned char scope; 175 u32 tclassid; 176 dscp_t dscp; 177 struct fib_nh_common *nhc; 178 struct fib_info *fi; 179 struct fib_table *table; 180 struct hlist_head *fa_head; 181 }; 182 183 struct fib_result_nl { 184 __be32 fl_addr; /* To be looked up*/ 185 u32 fl_mark; 186 unsigned char fl_tos; 187 unsigned char fl_scope; 188 unsigned char tb_id_in; 189 190 unsigned char tb_id; /* Results */ 191 unsigned char prefixlen; 192 unsigned char nh_sel; 193 unsigned char type; 194 unsigned char scope; 195 int err; 196 }; 197 198 #ifdef CONFIG_IP_MULTIPLE_TABLES 199 #define FIB_TABLE_HASHSZ 256 200 #else 201 #define FIB_TABLE_HASHSZ 2 202 #endif 203 204 __be32 fib_info_update_nhc_saddr(struct net *net, struct fib_nh_common *nhc, 205 unsigned char scope); 206 __be32 fib_result_prefsrc(struct net *net, struct fib_result *res); 207 208 #define FIB_RES_NHC(res) ((res).nhc) 209 #define FIB_RES_DEV(res) (FIB_RES_NHC(res)->nhc_dev) 210 #define FIB_RES_OIF(res) (FIB_RES_NHC(res)->nhc_oif) 211 212 struct fib_rt_info { 213 struct fib_info *fi; 214 u32 tb_id; 215 __be32 dst; 216 int dst_len; 217 dscp_t dscp; 218 u8 type; 219 u8 offload:1, 220 trap:1, 221 offload_failed:1, 222 unused:5; 223 }; 224 225 struct fib_entry_notifier_info { 226 struct fib_notifier_info info; /* must be first */ 227 u32 dst; 228 int dst_len; 229 struct fib_info *fi; 230 dscp_t dscp; 231 u8 type; 232 u32 tb_id; 233 }; 234 235 struct fib_nh_notifier_info { 236 struct fib_notifier_info info; /* must be first */ 237 struct fib_nh *fib_nh; 238 }; 239 240 int call_fib4_notifier(struct notifier_block *nb, 241 enum fib_event_type event_type, 242 struct fib_notifier_info *info); 243 int call_fib4_notifiers(struct net *net, enum fib_event_type event_type, 244 struct fib_notifier_info *info); 245 246 int __net_init fib4_notifier_init(struct net *net); 247 void __net_exit fib4_notifier_exit(struct net *net); 248 249 void fib_info_notify_update(struct net *net, struct nl_info *info); 250 int fib_notify(struct net *net, struct notifier_block *nb, 251 struct netlink_ext_ack *extack); 252 253 struct fib_table { 254 struct hlist_node tb_hlist; 255 u32 tb_id; 256 int tb_num_default; 257 struct rcu_head rcu; 258 unsigned long *tb_data; 259 unsigned long __data[]; 260 }; 261 262 struct fib_dump_filter { 263 u32 table_id; 264 /* filter_set is an optimization that an entry is set */ 265 bool filter_set; 266 bool dump_routes; 267 bool dump_exceptions; 268 unsigned char protocol; 269 unsigned char rt_type; 270 unsigned int flags; 271 struct net_device *dev; 272 }; 273 274 int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp, 275 struct fib_result *res, int fib_flags); 276 int fib_table_insert(struct net *, struct fib_table *, struct fib_config *, 277 struct netlink_ext_ack *extack); 278 int fib_table_delete(struct net *, struct fib_table *, struct fib_config *, 279 struct netlink_ext_ack *extack); 280 int fib_table_dump(struct fib_table *table, struct sk_buff *skb, 281 struct netlink_callback *cb, struct fib_dump_filter *filter); 282 int fib_table_flush(struct net *net, struct fib_table *table, bool flush_all); 283 struct fib_table *fib_trie_unmerge(struct fib_table *main_tb); 284 void fib_table_flush_external(struct fib_table *table); 285 void fib_free_table(struct fib_table *tb); 286 287 #ifndef CONFIG_IP_MULTIPLE_TABLES 288 289 #define TABLE_LOCAL_INDEX (RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1)) 290 #define TABLE_MAIN_INDEX (RT_TABLE_MAIN & (FIB_TABLE_HASHSZ - 1)) 291 292 static inline struct fib_table *fib_get_table(struct net *net, u32 id) 293 { 294 struct hlist_node *tb_hlist; 295 struct hlist_head *ptr; 296 297 ptr = id == RT_TABLE_LOCAL ? 298 &net->ipv4.fib_table_hash[TABLE_LOCAL_INDEX] : 299 &net->ipv4.fib_table_hash[TABLE_MAIN_INDEX]; 300 301 tb_hlist = rcu_dereference_rtnl(hlist_first_rcu(ptr)); 302 303 return hlist_entry(tb_hlist, struct fib_table, tb_hlist); 304 } 305 306 static inline struct fib_table *fib_new_table(struct net *net, u32 id) 307 { 308 return fib_get_table(net, id); 309 } 310 311 static inline int fib_lookup(struct net *net, const struct flowi4 *flp, 312 struct fib_result *res, unsigned int flags) 313 { 314 struct fib_table *tb; 315 int err = -ENETUNREACH; 316 317 rcu_read_lock(); 318 319 tb = fib_get_table(net, RT_TABLE_MAIN); 320 if (tb) 321 err = fib_table_lookup(tb, flp, res, flags | FIB_LOOKUP_NOREF); 322 323 if (err == -EAGAIN) 324 err = -ENETUNREACH; 325 326 rcu_read_unlock(); 327 328 return err; 329 } 330 331 static inline bool fib4_has_custom_rules(const struct net *net) 332 { 333 return false; 334 } 335 336 static inline bool fib4_rule_default(const struct fib_rule *rule) 337 { 338 return true; 339 } 340 341 static inline int fib4_rules_dump(struct net *net, struct notifier_block *nb, 342 struct netlink_ext_ack *extack) 343 { 344 return 0; 345 } 346 347 static inline unsigned int fib4_rules_seq_read(struct net *net) 348 { 349 return 0; 350 } 351 352 static inline bool fib4_rules_early_flow_dissect(struct net *net, 353 struct sk_buff *skb, 354 struct flowi4 *fl4, 355 struct flow_keys *flkeys) 356 { 357 return false; 358 } 359 #else /* CONFIG_IP_MULTIPLE_TABLES */ 360 int __net_init fib4_rules_init(struct net *net); 361 void __net_exit fib4_rules_exit(struct net *net); 362 363 struct fib_table *fib_new_table(struct net *net, u32 id); 364 struct fib_table *fib_get_table(struct net *net, u32 id); 365 366 int __fib_lookup(struct net *net, struct flowi4 *flp, 367 struct fib_result *res, unsigned int flags); 368 369 static inline int fib_lookup(struct net *net, struct flowi4 *flp, 370 struct fib_result *res, unsigned int flags) 371 { 372 struct fib_table *tb; 373 int err = -ENETUNREACH; 374 375 flags |= FIB_LOOKUP_NOREF; 376 if (net->ipv4.fib_has_custom_rules) 377 return __fib_lookup(net, flp, res, flags); 378 379 rcu_read_lock(); 380 381 res->tclassid = 0; 382 383 tb = rcu_dereference_rtnl(net->ipv4.fib_main); 384 if (tb) 385 err = fib_table_lookup(tb, flp, res, flags); 386 387 if (!err) 388 goto out; 389 390 tb = rcu_dereference_rtnl(net->ipv4.fib_default); 391 if (tb) 392 err = fib_table_lookup(tb, flp, res, flags); 393 394 out: 395 if (err == -EAGAIN) 396 err = -ENETUNREACH; 397 398 rcu_read_unlock(); 399 400 return err; 401 } 402 403 static inline bool fib4_has_custom_rules(const struct net *net) 404 { 405 return net->ipv4.fib_has_custom_rules; 406 } 407 408 bool fib4_rule_default(const struct fib_rule *rule); 409 int fib4_rules_dump(struct net *net, struct notifier_block *nb, 410 struct netlink_ext_ack *extack); 411 unsigned int fib4_rules_seq_read(struct net *net); 412 413 static inline bool fib4_rules_early_flow_dissect(struct net *net, 414 struct sk_buff *skb, 415 struct flowi4 *fl4, 416 struct flow_keys *flkeys) 417 { 418 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 419 420 if (!net->ipv4.fib_rules_require_fldissect) 421 return false; 422 423 memset(flkeys, 0, sizeof(*flkeys)); 424 __skb_flow_dissect(net, skb, &flow_keys_dissector, 425 flkeys, NULL, 0, 0, 0, flag); 426 427 fl4->fl4_sport = flkeys->ports.src; 428 fl4->fl4_dport = flkeys->ports.dst; 429 fl4->flowi4_proto = flkeys->basic.ip_proto; 430 431 return true; 432 } 433 434 #endif /* CONFIG_IP_MULTIPLE_TABLES */ 435 436 /* Exported by fib_frontend.c */ 437 extern const struct nla_policy rtm_ipv4_policy[]; 438 void ip_fib_init(void); 439 int fib_gw_from_via(struct fib_config *cfg, struct nlattr *nla, 440 struct netlink_ext_ack *extack); 441 __be32 fib_compute_spec_dst(struct sk_buff *skb); 442 bool fib_info_nh_uses_dev(struct fib_info *fi, const struct net_device *dev); 443 int fib_validate_source(struct sk_buff *skb, __be32 src, __be32 dst, 444 u8 tos, int oif, struct net_device *dev, 445 struct in_device *idev, u32 *itag); 446 #ifdef CONFIG_IP_ROUTE_CLASSID 447 static inline int fib_num_tclassid_users(struct net *net) 448 { 449 return atomic_read(&net->ipv4.fib_num_tclassid_users); 450 } 451 #else 452 static inline int fib_num_tclassid_users(struct net *net) 453 { 454 return 0; 455 } 456 #endif 457 int fib_unmerge(struct net *net); 458 459 static inline bool nhc_l3mdev_matches_dev(const struct fib_nh_common *nhc, 460 const struct net_device *dev) 461 { 462 if (nhc->nhc_dev == dev || 463 l3mdev_master_ifindex_rcu(nhc->nhc_dev) == dev->ifindex) 464 return true; 465 466 return false; 467 } 468 469 /* Exported by fib_semantics.c */ 470 int ip_fib_check_default(__be32 gw, struct net_device *dev); 471 int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force); 472 int fib_sync_down_addr(struct net_device *dev, __be32 local); 473 int fib_sync_up(struct net_device *dev, unsigned char nh_flags); 474 void fib_sync_mtu(struct net_device *dev, u32 orig_mtu); 475 void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig); 476 477 /* Fields used for sysctl_fib_multipath_hash_fields. 478 * Common to IPv4 and IPv6. 479 * 480 * Add new fields at the end. This is user API. 481 */ 482 #define FIB_MULTIPATH_HASH_FIELD_SRC_IP BIT(0) 483 #define FIB_MULTIPATH_HASH_FIELD_DST_IP BIT(1) 484 #define FIB_MULTIPATH_HASH_FIELD_IP_PROTO BIT(2) 485 #define FIB_MULTIPATH_HASH_FIELD_FLOWLABEL BIT(3) 486 #define FIB_MULTIPATH_HASH_FIELD_SRC_PORT BIT(4) 487 #define FIB_MULTIPATH_HASH_FIELD_DST_PORT BIT(5) 488 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP BIT(6) 489 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP BIT(7) 490 #define FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO BIT(8) 491 #define FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL BIT(9) 492 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT BIT(10) 493 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT BIT(11) 494 495 #define FIB_MULTIPATH_HASH_FIELD_OUTER_MASK \ 496 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 497 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 498 FIB_MULTIPATH_HASH_FIELD_IP_PROTO | \ 499 FIB_MULTIPATH_HASH_FIELD_FLOWLABEL | \ 500 FIB_MULTIPATH_HASH_FIELD_SRC_PORT | \ 501 FIB_MULTIPATH_HASH_FIELD_DST_PORT) 502 503 #define FIB_MULTIPATH_HASH_FIELD_INNER_MASK \ 504 (FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP | \ 505 FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP | \ 506 FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO | \ 507 FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL | \ 508 FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT | \ 509 FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT) 510 511 #define FIB_MULTIPATH_HASH_FIELD_ALL_MASK \ 512 (FIB_MULTIPATH_HASH_FIELD_OUTER_MASK | \ 513 FIB_MULTIPATH_HASH_FIELD_INNER_MASK) 514 515 #define FIB_MULTIPATH_HASH_FIELD_DEFAULT_MASK \ 516 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 517 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 518 FIB_MULTIPATH_HASH_FIELD_IP_PROTO) 519 520 #ifdef CONFIG_IP_ROUTE_MULTIPATH 521 int fib_multipath_hash(const struct net *net, const struct flowi4 *fl4, 522 const struct sk_buff *skb, struct flow_keys *flkeys); 523 #endif 524 int fib_check_nh(struct net *net, struct fib_nh *nh, u32 table, u8 scope, 525 struct netlink_ext_ack *extack); 526 void fib_select_multipath(struct fib_result *res, int hash); 527 void fib_select_path(struct net *net, struct fib_result *res, 528 struct flowi4 *fl4, const struct sk_buff *skb); 529 530 int fib_nh_init(struct net *net, struct fib_nh *fib_nh, 531 struct fib_config *cfg, int nh_weight, 532 struct netlink_ext_ack *extack); 533 void fib_nh_release(struct net *net, struct fib_nh *fib_nh); 534 int fib_nh_common_init(struct net *net, struct fib_nh_common *nhc, 535 struct nlattr *fc_encap, u16 fc_encap_type, 536 void *cfg, gfp_t gfp_flags, 537 struct netlink_ext_ack *extack); 538 void fib_nh_common_release(struct fib_nh_common *nhc); 539 540 /* Exported by fib_trie.c */ 541 void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri); 542 void fib_trie_init(void); 543 struct fib_table *fib_trie_table(u32 id, struct fib_table *alias); 544 bool fib_lookup_good_nhc(const struct fib_nh_common *nhc, int fib_flags, 545 const struct flowi4 *flp); 546 547 static inline void fib_combine_itag(u32 *itag, const struct fib_result *res) 548 { 549 #ifdef CONFIG_IP_ROUTE_CLASSID 550 struct fib_nh_common *nhc = res->nhc; 551 #ifdef CONFIG_IP_MULTIPLE_TABLES 552 u32 rtag; 553 #endif 554 if (nhc->nhc_family == AF_INET) { 555 struct fib_nh *nh; 556 557 nh = container_of(nhc, struct fib_nh, nh_common); 558 *itag = nh->nh_tclassid << 16; 559 } else { 560 *itag = 0; 561 } 562 563 #ifdef CONFIG_IP_MULTIPLE_TABLES 564 rtag = res->tclassid; 565 if (*itag == 0) 566 *itag = (rtag<<16); 567 *itag |= (rtag>>16); 568 #endif 569 #endif 570 } 571 572 void fib_flush(struct net *net); 573 void free_fib_info(struct fib_info *fi); 574 575 static inline void fib_info_hold(struct fib_info *fi) 576 { 577 refcount_inc(&fi->fib_clntref); 578 } 579 580 static inline void fib_info_put(struct fib_info *fi) 581 { 582 if (refcount_dec_and_test(&fi->fib_clntref)) 583 free_fib_info(fi); 584 } 585 586 #ifdef CONFIG_PROC_FS 587 int __net_init fib_proc_init(struct net *net); 588 void __net_exit fib_proc_exit(struct net *net); 589 #else 590 static inline int fib_proc_init(struct net *net) 591 { 592 return 0; 593 } 594 static inline void fib_proc_exit(struct net *net) 595 { 596 } 597 #endif 598 599 u32 ip_mtu_from_fib_result(struct fib_result *res, __be32 daddr); 600 601 int ip_valid_fib_dump_req(struct net *net, const struct nlmsghdr *nlh, 602 struct fib_dump_filter *filter, 603 struct netlink_callback *cb); 604 605 int fib_nexthop_info(struct sk_buff *skb, const struct fib_nh_common *nh, 606 u8 rt_family, unsigned char *flags, bool skip_oif); 607 int fib_add_nexthop(struct sk_buff *skb, const struct fib_nh_common *nh, 608 int nh_weight, u8 rt_family, u32 nh_tclassid); 609 #endif /* _NET_FIB_H */ 610