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 struct nexthop *nh; 158 struct rcu_head rcu; 159 struct fib_nh fib_nh[]; 160 }; 161 162 163 #ifdef CONFIG_IP_MULTIPLE_TABLES 164 struct fib_rule; 165 #endif 166 167 struct fib_table; 168 struct fib_result { 169 __be32 prefix; 170 unsigned char prefixlen; 171 unsigned char nh_sel; 172 unsigned char type; 173 unsigned char scope; 174 u32 tclassid; 175 struct fib_nh_common *nhc; 176 struct fib_info *fi; 177 struct fib_table *table; 178 struct hlist_head *fa_head; 179 }; 180 181 struct fib_result_nl { 182 __be32 fl_addr; /* To be looked up*/ 183 u32 fl_mark; 184 unsigned char fl_tos; 185 unsigned char fl_scope; 186 unsigned char tb_id_in; 187 188 unsigned char tb_id; /* Results */ 189 unsigned char prefixlen; 190 unsigned char nh_sel; 191 unsigned char type; 192 unsigned char scope; 193 int err; 194 }; 195 196 #ifdef CONFIG_IP_MULTIPLE_TABLES 197 #define FIB_TABLE_HASHSZ 256 198 #else 199 #define FIB_TABLE_HASHSZ 2 200 #endif 201 202 __be32 fib_info_update_nhc_saddr(struct net *net, struct fib_nh_common *nhc, 203 unsigned char scope); 204 __be32 fib_result_prefsrc(struct net *net, struct fib_result *res); 205 206 #define FIB_RES_NHC(res) ((res).nhc) 207 #define FIB_RES_DEV(res) (FIB_RES_NHC(res)->nhc_dev) 208 #define FIB_RES_OIF(res) (FIB_RES_NHC(res)->nhc_oif) 209 210 struct fib_rt_info { 211 struct fib_info *fi; 212 u32 tb_id; 213 __be32 dst; 214 int dst_len; 215 dscp_t dscp; 216 u8 type; 217 u8 offload:1, 218 trap:1, 219 offload_failed:1, 220 unused:5; 221 }; 222 223 struct fib_entry_notifier_info { 224 struct fib_notifier_info info; /* must be first */ 225 u32 dst; 226 int dst_len; 227 struct fib_info *fi; 228 dscp_t dscp; 229 u8 type; 230 u32 tb_id; 231 }; 232 233 struct fib_nh_notifier_info { 234 struct fib_notifier_info info; /* must be first */ 235 struct fib_nh *fib_nh; 236 }; 237 238 int call_fib4_notifier(struct notifier_block *nb, 239 enum fib_event_type event_type, 240 struct fib_notifier_info *info); 241 int call_fib4_notifiers(struct net *net, enum fib_event_type event_type, 242 struct fib_notifier_info *info); 243 244 int __net_init fib4_notifier_init(struct net *net); 245 void __net_exit fib4_notifier_exit(struct net *net); 246 247 void fib_info_notify_update(struct net *net, struct nl_info *info); 248 int fib_notify(struct net *net, struct notifier_block *nb, 249 struct netlink_ext_ack *extack); 250 251 struct fib_table { 252 struct hlist_node tb_hlist; 253 u32 tb_id; 254 int tb_num_default; 255 struct rcu_head rcu; 256 unsigned long *tb_data; 257 unsigned long __data[]; 258 }; 259 260 struct fib_dump_filter { 261 u32 table_id; 262 /* filter_set is an optimization that an entry is set */ 263 bool filter_set; 264 bool dump_routes; 265 bool dump_exceptions; 266 unsigned char protocol; 267 unsigned char rt_type; 268 unsigned int flags; 269 struct net_device *dev; 270 }; 271 272 int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp, 273 struct fib_result *res, int fib_flags); 274 int fib_table_insert(struct net *, struct fib_table *, struct fib_config *, 275 struct netlink_ext_ack *extack); 276 int fib_table_delete(struct net *, struct fib_table *, struct fib_config *, 277 struct netlink_ext_ack *extack); 278 int fib_table_dump(struct fib_table *table, struct sk_buff *skb, 279 struct netlink_callback *cb, struct fib_dump_filter *filter); 280 int fib_table_flush(struct net *net, struct fib_table *table, bool flush_all); 281 struct fib_table *fib_trie_unmerge(struct fib_table *main_tb); 282 void fib_table_flush_external(struct fib_table *table); 283 void fib_free_table(struct fib_table *tb); 284 285 #ifndef CONFIG_IP_MULTIPLE_TABLES 286 287 #define TABLE_LOCAL_INDEX (RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1)) 288 #define TABLE_MAIN_INDEX (RT_TABLE_MAIN & (FIB_TABLE_HASHSZ - 1)) 289 290 static inline struct fib_table *fib_get_table(struct net *net, u32 id) 291 { 292 struct hlist_node *tb_hlist; 293 struct hlist_head *ptr; 294 295 ptr = id == RT_TABLE_LOCAL ? 296 &net->ipv4.fib_table_hash[TABLE_LOCAL_INDEX] : 297 &net->ipv4.fib_table_hash[TABLE_MAIN_INDEX]; 298 299 tb_hlist = rcu_dereference_rtnl(hlist_first_rcu(ptr)); 300 301 return hlist_entry(tb_hlist, struct fib_table, tb_hlist); 302 } 303 304 static inline struct fib_table *fib_new_table(struct net *net, u32 id) 305 { 306 return fib_get_table(net, id); 307 } 308 309 static inline int fib_lookup(struct net *net, const struct flowi4 *flp, 310 struct fib_result *res, unsigned int flags) 311 { 312 struct fib_table *tb; 313 int err = -ENETUNREACH; 314 315 rcu_read_lock(); 316 317 tb = fib_get_table(net, RT_TABLE_MAIN); 318 if (tb) 319 err = fib_table_lookup(tb, flp, res, flags | FIB_LOOKUP_NOREF); 320 321 if (err == -EAGAIN) 322 err = -ENETUNREACH; 323 324 rcu_read_unlock(); 325 326 return err; 327 } 328 329 static inline bool fib4_has_custom_rules(const struct net *net) 330 { 331 return false; 332 } 333 334 static inline bool fib4_rule_default(const struct fib_rule *rule) 335 { 336 return true; 337 } 338 339 static inline int fib4_rules_dump(struct net *net, struct notifier_block *nb, 340 struct netlink_ext_ack *extack) 341 { 342 return 0; 343 } 344 345 static inline unsigned int fib4_rules_seq_read(struct net *net) 346 { 347 return 0; 348 } 349 350 static inline bool fib4_rules_early_flow_dissect(struct net *net, 351 struct sk_buff *skb, 352 struct flowi4 *fl4, 353 struct flow_keys *flkeys) 354 { 355 return false; 356 } 357 #else /* CONFIG_IP_MULTIPLE_TABLES */ 358 int __net_init fib4_rules_init(struct net *net); 359 void __net_exit fib4_rules_exit(struct net *net); 360 361 struct fib_table *fib_new_table(struct net *net, u32 id); 362 struct fib_table *fib_get_table(struct net *net, u32 id); 363 364 int __fib_lookup(struct net *net, struct flowi4 *flp, 365 struct fib_result *res, unsigned int flags); 366 367 static inline int fib_lookup(struct net *net, struct flowi4 *flp, 368 struct fib_result *res, unsigned int flags) 369 { 370 struct fib_table *tb; 371 int err = -ENETUNREACH; 372 373 flags |= FIB_LOOKUP_NOREF; 374 if (net->ipv4.fib_has_custom_rules) 375 return __fib_lookup(net, flp, res, flags); 376 377 rcu_read_lock(); 378 379 res->tclassid = 0; 380 381 tb = rcu_dereference_rtnl(net->ipv4.fib_main); 382 if (tb) 383 err = fib_table_lookup(tb, flp, res, flags); 384 385 if (!err) 386 goto out; 387 388 tb = rcu_dereference_rtnl(net->ipv4.fib_default); 389 if (tb) 390 err = fib_table_lookup(tb, flp, res, flags); 391 392 out: 393 if (err == -EAGAIN) 394 err = -ENETUNREACH; 395 396 rcu_read_unlock(); 397 398 return err; 399 } 400 401 static inline bool fib4_has_custom_rules(const struct net *net) 402 { 403 return net->ipv4.fib_has_custom_rules; 404 } 405 406 bool fib4_rule_default(const struct fib_rule *rule); 407 int fib4_rules_dump(struct net *net, struct notifier_block *nb, 408 struct netlink_ext_ack *extack); 409 unsigned int fib4_rules_seq_read(struct net *net); 410 411 static inline bool fib4_rules_early_flow_dissect(struct net *net, 412 struct sk_buff *skb, 413 struct flowi4 *fl4, 414 struct flow_keys *flkeys) 415 { 416 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP; 417 418 if (!net->ipv4.fib_rules_require_fldissect) 419 return false; 420 421 skb_flow_dissect_flow_keys(skb, flkeys, flag); 422 fl4->fl4_sport = flkeys->ports.src; 423 fl4->fl4_dport = flkeys->ports.dst; 424 fl4->flowi4_proto = flkeys->basic.ip_proto; 425 426 return true; 427 } 428 429 #endif /* CONFIG_IP_MULTIPLE_TABLES */ 430 431 /* Exported by fib_frontend.c */ 432 extern const struct nla_policy rtm_ipv4_policy[]; 433 void ip_fib_init(void); 434 int fib_gw_from_via(struct fib_config *cfg, struct nlattr *nla, 435 struct netlink_ext_ack *extack); 436 __be32 fib_compute_spec_dst(struct sk_buff *skb); 437 bool fib_info_nh_uses_dev(struct fib_info *fi, const struct net_device *dev); 438 int fib_validate_source(struct sk_buff *skb, __be32 src, __be32 dst, 439 u8 tos, int oif, struct net_device *dev, 440 struct in_device *idev, u32 *itag); 441 #ifdef CONFIG_IP_ROUTE_CLASSID 442 static inline int fib_num_tclassid_users(struct net *net) 443 { 444 return atomic_read(&net->ipv4.fib_num_tclassid_users); 445 } 446 #else 447 static inline int fib_num_tclassid_users(struct net *net) 448 { 449 return 0; 450 } 451 #endif 452 int fib_unmerge(struct net *net); 453 454 static inline bool nhc_l3mdev_matches_dev(const struct fib_nh_common *nhc, 455 const struct net_device *dev) 456 { 457 if (nhc->nhc_dev == dev || 458 l3mdev_master_ifindex_rcu(nhc->nhc_dev) == dev->ifindex) 459 return true; 460 461 return false; 462 } 463 464 /* Exported by fib_semantics.c */ 465 int ip_fib_check_default(__be32 gw, struct net_device *dev); 466 int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force); 467 int fib_sync_down_addr(struct net_device *dev, __be32 local); 468 int fib_sync_up(struct net_device *dev, unsigned char nh_flags); 469 void fib_sync_mtu(struct net_device *dev, u32 orig_mtu); 470 void fib_nhc_update_mtu(struct fib_nh_common *nhc, u32 new, u32 orig); 471 472 /* Fields used for sysctl_fib_multipath_hash_fields. 473 * Common to IPv4 and IPv6. 474 * 475 * Add new fields at the end. This is user API. 476 */ 477 #define FIB_MULTIPATH_HASH_FIELD_SRC_IP BIT(0) 478 #define FIB_MULTIPATH_HASH_FIELD_DST_IP BIT(1) 479 #define FIB_MULTIPATH_HASH_FIELD_IP_PROTO BIT(2) 480 #define FIB_MULTIPATH_HASH_FIELD_FLOWLABEL BIT(3) 481 #define FIB_MULTIPATH_HASH_FIELD_SRC_PORT BIT(4) 482 #define FIB_MULTIPATH_HASH_FIELD_DST_PORT BIT(5) 483 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP BIT(6) 484 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP BIT(7) 485 #define FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO BIT(8) 486 #define FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL BIT(9) 487 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT BIT(10) 488 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT BIT(11) 489 490 #define FIB_MULTIPATH_HASH_FIELD_OUTER_MASK \ 491 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 492 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 493 FIB_MULTIPATH_HASH_FIELD_IP_PROTO | \ 494 FIB_MULTIPATH_HASH_FIELD_FLOWLABEL | \ 495 FIB_MULTIPATH_HASH_FIELD_SRC_PORT | \ 496 FIB_MULTIPATH_HASH_FIELD_DST_PORT) 497 498 #define FIB_MULTIPATH_HASH_FIELD_INNER_MASK \ 499 (FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP | \ 500 FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP | \ 501 FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO | \ 502 FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL | \ 503 FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT | \ 504 FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT) 505 506 #define FIB_MULTIPATH_HASH_FIELD_ALL_MASK \ 507 (FIB_MULTIPATH_HASH_FIELD_OUTER_MASK | \ 508 FIB_MULTIPATH_HASH_FIELD_INNER_MASK) 509 510 #define FIB_MULTIPATH_HASH_FIELD_DEFAULT_MASK \ 511 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \ 512 FIB_MULTIPATH_HASH_FIELD_DST_IP | \ 513 FIB_MULTIPATH_HASH_FIELD_IP_PROTO) 514 515 #ifdef CONFIG_IP_ROUTE_MULTIPATH 516 int fib_multipath_hash(const struct net *net, const struct flowi4 *fl4, 517 const struct sk_buff *skb, struct flow_keys *flkeys); 518 #endif 519 int fib_check_nh(struct net *net, struct fib_nh *nh, u32 table, u8 scope, 520 struct netlink_ext_ack *extack); 521 void fib_select_multipath(struct fib_result *res, int hash); 522 void fib_select_path(struct net *net, struct fib_result *res, 523 struct flowi4 *fl4, const struct sk_buff *skb); 524 525 int fib_nh_init(struct net *net, struct fib_nh *fib_nh, 526 struct fib_config *cfg, int nh_weight, 527 struct netlink_ext_ack *extack); 528 void fib_nh_release(struct net *net, struct fib_nh *fib_nh); 529 int fib_nh_common_init(struct net *net, struct fib_nh_common *nhc, 530 struct nlattr *fc_encap, u16 fc_encap_type, 531 void *cfg, gfp_t gfp_flags, 532 struct netlink_ext_ack *extack); 533 void fib_nh_common_release(struct fib_nh_common *nhc); 534 535 /* Exported by fib_trie.c */ 536 void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri); 537 void fib_trie_init(void); 538 struct fib_table *fib_trie_table(u32 id, struct fib_table *alias); 539 bool fib_lookup_good_nhc(const struct fib_nh_common *nhc, int fib_flags, 540 const struct flowi4 *flp); 541 542 static inline void fib_combine_itag(u32 *itag, const struct fib_result *res) 543 { 544 #ifdef CONFIG_IP_ROUTE_CLASSID 545 struct fib_nh_common *nhc = res->nhc; 546 #ifdef CONFIG_IP_MULTIPLE_TABLES 547 u32 rtag; 548 #endif 549 if (nhc->nhc_family == AF_INET) { 550 struct fib_nh *nh; 551 552 nh = container_of(nhc, struct fib_nh, nh_common); 553 *itag = nh->nh_tclassid << 16; 554 } else { 555 *itag = 0; 556 } 557 558 #ifdef CONFIG_IP_MULTIPLE_TABLES 559 rtag = res->tclassid; 560 if (*itag == 0) 561 *itag = (rtag<<16); 562 *itag |= (rtag>>16); 563 #endif 564 #endif 565 } 566 567 void fib_flush(struct net *net); 568 void free_fib_info(struct fib_info *fi); 569 570 static inline void fib_info_hold(struct fib_info *fi) 571 { 572 refcount_inc(&fi->fib_clntref); 573 } 574 575 static inline void fib_info_put(struct fib_info *fi) 576 { 577 if (refcount_dec_and_test(&fi->fib_clntref)) 578 free_fib_info(fi); 579 } 580 581 #ifdef CONFIG_PROC_FS 582 int __net_init fib_proc_init(struct net *net); 583 void __net_exit fib_proc_exit(struct net *net); 584 #else 585 static inline int fib_proc_init(struct net *net) 586 { 587 return 0; 588 } 589 static inline void fib_proc_exit(struct net *net) 590 { 591 } 592 #endif 593 594 u32 ip_mtu_from_fib_result(struct fib_result *res, __be32 daddr); 595 596 int ip_valid_fib_dump_req(struct net *net, const struct nlmsghdr *nlh, 597 struct fib_dump_filter *filter, 598 struct netlink_callback *cb); 599 600 int fib_nexthop_info(struct sk_buff *skb, const struct fib_nh_common *nh, 601 u8 rt_family, unsigned char *flags, bool skip_oif); 602 int fib_add_nexthop(struct sk_buff *skb, const struct fib_nh_common *nh, 603 int nh_weight, u8 rt_family, u32 nh_tclassid); 604 #endif /* _NET_FIB_H */ 605