1 /* SPDX-License-Identifier: GPL-2.0-or-later */
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.
7 * Definitions for the Forwarding Information Base.
9 * Authors: A.N.Kuznetsov, <kuznet@ms2.inr.ac.ru>
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>
26 #include <linux/in_route.h>
40 struct in6_addr fc_gw6
;
48 struct rtnexthop
*fc_mp
;
53 struct nl_info fc_nlinfo
;
54 struct nlattr
*fc_encap
;
61 struct fib_nh_exception
{
62 struct fib_nh_exception __rcu
*fnhe_next
;
68 unsigned long fnhe_expires
;
69 struct rtable __rcu
*fnhe_rth_input
;
70 struct rtable __rcu
*fnhe_rth_output
;
71 unsigned long fnhe_stamp
;
75 struct fnhe_hash_bucket
{
76 struct fib_nh_exception __rcu
*chain
;
79 #define FNHE_HASH_SHIFT 11
80 #define FNHE_HASH_SIZE (1 << FNHE_HASH_SHIFT)
81 #define FNHE_RECLAIM_DEPTH 5
83 struct fib_nh_common
{
84 struct net_device
*nhc_dev
;
85 netdevice_tracker nhc_dev_tracker
;
87 unsigned char nhc_scope
;
90 unsigned char nhc_flags
;
91 struct lwtunnel_state
*nhc_lwtstate
;
99 atomic_t nhc_upper_bound
;
101 /* v4 specific, but allows fib6_nh with v4 routes */
102 struct rtable __rcu
* __percpu
*nhc_pcpu_rth_output
;
103 struct rtable __rcu
*nhc_rth_input
;
104 struct fnhe_hash_bucket __rcu
*nhc_exceptions
;
108 struct fib_nh_common nh_common
;
109 struct hlist_node nh_hash
;
110 struct fib_info
*nh_parent
;
111 #ifdef CONFIG_IP_ROUTE_CLASSID
116 #define fib_nh_family nh_common.nhc_family
117 #define fib_nh_dev nh_common.nhc_dev
118 #define fib_nh_dev_tracker nh_common.nhc_dev_tracker
119 #define fib_nh_oif nh_common.nhc_oif
120 #define fib_nh_flags nh_common.nhc_flags
121 #define fib_nh_lws nh_common.nhc_lwtstate
122 #define fib_nh_scope nh_common.nhc_scope
123 #define fib_nh_gw_family nh_common.nhc_gw_family
124 #define fib_nh_gw4 nh_common.nhc_gw.ipv4
125 #define fib_nh_gw6 nh_common.nhc_gw.ipv6
126 #define fib_nh_weight nh_common.nhc_weight
127 #define fib_nh_upper_bound nh_common.nhc_upper_bound
131 * This structure contains data shared by many of routes.
137 struct hlist_node fib_hash
;
138 struct hlist_node fib_lhash
;
139 struct list_head nh_list
;
141 refcount_t fib_treeref
;
142 refcount_t fib_clntref
;
143 unsigned int fib_flags
;
144 unsigned char fib_dead
;
145 unsigned char fib_protocol
;
146 unsigned char fib_scope
;
147 unsigned char fib_type
;
151 struct dst_metrics
*fib_metrics
;
152 #define fib_mtu fib_metrics->metrics[RTAX_MTU-1]
153 #define fib_window fib_metrics->metrics[RTAX_WINDOW-1]
154 #define fib_rtt fib_metrics->metrics[RTAX_RTT-1]
155 #define fib_advmss fib_metrics->metrics[RTAX_ADVMSS-1]
162 struct fib_nh fib_nh
[] __counted_by(fib_nhs
);
166 #ifdef CONFIG_IP_MULTIPLE_TABLES
173 unsigned char prefixlen
;
174 unsigned char nh_sel
;
179 struct fib_nh_common
*nhc
;
181 struct fib_table
*table
;
182 struct hlist_head
*fa_head
;
185 struct fib_result_nl
{
186 __be32 fl_addr
; /* To be looked up*/
188 unsigned char fl_tos
;
189 unsigned char fl_scope
;
190 unsigned char tb_id_in
;
192 unsigned char tb_id
; /* Results */
193 unsigned char prefixlen
;
194 unsigned char nh_sel
;
200 #ifdef CONFIG_IP_MULTIPLE_TABLES
201 #define FIB_TABLE_HASHSZ 256
203 #define FIB_TABLE_HASHSZ 2
206 __be32
fib_info_update_nhc_saddr(struct net
*net
, struct fib_nh_common
*nhc
,
207 unsigned char scope
);
208 __be32
fib_result_prefsrc(struct net
*net
, struct fib_result
*res
);
210 #define FIB_RES_NHC(res) ((res).nhc)
211 #define FIB_RES_DEV(res) (FIB_RES_NHC(res)->nhc_dev)
212 #define FIB_RES_OIF(res) (FIB_RES_NHC(res)->nhc_oif)
227 struct fib_entry_notifier_info
{
228 struct fib_notifier_info info
; /* must be first */
237 struct fib_nh_notifier_info
{
238 struct fib_notifier_info info
; /* must be first */
239 struct fib_nh
*fib_nh
;
242 int call_fib4_notifier(struct notifier_block
*nb
,
243 enum fib_event_type event_type
,
244 struct fib_notifier_info
*info
);
245 int call_fib4_notifiers(struct net
*net
, enum fib_event_type event_type
,
246 struct fib_notifier_info
*info
);
248 int __net_init
fib4_notifier_init(struct net
*net
);
249 void __net_exit
fib4_notifier_exit(struct net
*net
);
251 void fib_info_notify_update(struct net
*net
, struct nl_info
*info
);
252 int fib_notify(struct net
*net
, struct notifier_block
*nb
,
253 struct netlink_ext_ack
*extack
);
256 struct hlist_node tb_hlist
;
260 unsigned long *tb_data
;
261 unsigned long __data
[];
264 struct fib_dump_filter
{
266 /* filter_set is an optimization that an entry is set */
269 bool dump_exceptions
;
271 unsigned char protocol
;
272 unsigned char rt_type
;
274 struct net_device
*dev
;
277 int fib_table_lookup(struct fib_table
*tb
, const struct flowi4
*flp
,
278 struct fib_result
*res
, int fib_flags
);
279 int fib_table_insert(struct net
*, struct fib_table
*, struct fib_config
*,
280 struct netlink_ext_ack
*extack
);
281 int fib_table_delete(struct net
*, struct fib_table
*, struct fib_config
*,
282 struct netlink_ext_ack
*extack
);
283 int fib_table_dump(struct fib_table
*table
, struct sk_buff
*skb
,
284 struct netlink_callback
*cb
, struct fib_dump_filter
*filter
);
285 int fib_table_flush(struct net
*net
, struct fib_table
*table
, bool flush_all
);
286 struct fib_table
*fib_trie_unmerge(struct fib_table
*main_tb
);
287 void fib_table_flush_external(struct fib_table
*table
);
288 void fib_free_table(struct fib_table
*tb
);
290 #ifndef CONFIG_IP_MULTIPLE_TABLES
292 #define TABLE_LOCAL_INDEX (RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1))
293 #define TABLE_MAIN_INDEX (RT_TABLE_MAIN & (FIB_TABLE_HASHSZ - 1))
295 static inline struct fib_table
*fib_get_table(struct net
*net
, u32 id
)
297 struct hlist_node
*tb_hlist
;
298 struct hlist_head
*ptr
;
300 ptr
= id
== RT_TABLE_LOCAL
?
301 &net
->ipv4
.fib_table_hash
[TABLE_LOCAL_INDEX
] :
302 &net
->ipv4
.fib_table_hash
[TABLE_MAIN_INDEX
];
304 tb_hlist
= rcu_dereference_rtnl(hlist_first_rcu(ptr
));
306 return hlist_entry(tb_hlist
, struct fib_table
, tb_hlist
);
309 static inline struct fib_table
*fib_new_table(struct net
*net
, u32 id
)
311 return fib_get_table(net
, id
);
314 static inline int fib_lookup(struct net
*net
, const struct flowi4
*flp
,
315 struct fib_result
*res
, unsigned int flags
)
317 struct fib_table
*tb
;
318 int err
= -ENETUNREACH
;
322 tb
= fib_get_table(net
, RT_TABLE_MAIN
);
324 err
= fib_table_lookup(tb
, flp
, res
, flags
| FIB_LOOKUP_NOREF
);
334 static inline bool fib4_has_custom_rules(const struct net
*net
)
339 static inline bool fib4_rule_default(const struct fib_rule
*rule
)
344 static inline int fib4_rules_dump(struct net
*net
, struct notifier_block
*nb
,
345 struct netlink_ext_ack
*extack
)
350 static inline unsigned int fib4_rules_seq_read(const struct net
*net
)
355 static inline bool fib4_rules_early_flow_dissect(struct net
*net
,
358 struct flow_keys
*flkeys
)
362 #else /* CONFIG_IP_MULTIPLE_TABLES */
363 int __net_init
fib4_rules_init(struct net
*net
);
364 void __net_exit
fib4_rules_exit(struct net
*net
);
366 struct fib_table
*fib_new_table(struct net
*net
, u32 id
);
367 struct fib_table
*fib_get_table(struct net
*net
, u32 id
);
369 int __fib_lookup(struct net
*net
, struct flowi4
*flp
,
370 struct fib_result
*res
, unsigned int flags
);
372 static inline int fib_lookup(struct net
*net
, struct flowi4
*flp
,
373 struct fib_result
*res
, unsigned int flags
)
375 struct fib_table
*tb
;
376 int err
= -ENETUNREACH
;
378 flags
|= FIB_LOOKUP_NOREF
;
379 if (net
->ipv4
.fib_has_custom_rules
)
380 return __fib_lookup(net
, flp
, res
, flags
);
386 tb
= rcu_dereference_rtnl(net
->ipv4
.fib_main
);
388 err
= fib_table_lookup(tb
, flp
, res
, flags
);
393 tb
= rcu_dereference_rtnl(net
->ipv4
.fib_default
);
395 err
= fib_table_lookup(tb
, flp
, res
, flags
);
406 static inline bool fib4_has_custom_rules(const struct net
*net
)
408 return net
->ipv4
.fib_has_custom_rules
;
411 bool fib4_rule_default(const struct fib_rule
*rule
);
412 int fib4_rules_dump(struct net
*net
, struct notifier_block
*nb
,
413 struct netlink_ext_ack
*extack
);
414 unsigned int fib4_rules_seq_read(const struct net
*net
);
416 static inline bool fib4_rules_early_flow_dissect(struct net
*net
,
419 struct flow_keys
*flkeys
)
421 unsigned int flag
= FLOW_DISSECTOR_F_STOP_AT_ENCAP
;
423 if (!net
->ipv4
.fib_rules_require_fldissect
)
426 memset(flkeys
, 0, sizeof(*flkeys
));
427 __skb_flow_dissect(net
, skb
, &flow_keys_dissector
,
428 flkeys
, NULL
, 0, 0, 0, flag
);
430 fl4
->fl4_sport
= flkeys
->ports
.src
;
431 fl4
->fl4_dport
= flkeys
->ports
.dst
;
432 fl4
->flowi4_proto
= flkeys
->basic
.ip_proto
;
437 #endif /* CONFIG_IP_MULTIPLE_TABLES */
439 static inline bool fib_dscp_masked_match(dscp_t dscp
, const struct flowi4
*fl4
)
441 return dscp
== inet_dsfield_to_dscp(RT_TOS(fl4
->flowi4_tos
));
444 /* Exported by fib_frontend.c */
445 extern const struct nla_policy rtm_ipv4_policy
[];
446 void ip_fib_init(void);
447 int fib_gw_from_via(struct fib_config
*cfg
, struct nlattr
*nla
,
448 struct netlink_ext_ack
*extack
);
449 __be32
fib_compute_spec_dst(struct sk_buff
*skb
);
450 bool fib_info_nh_uses_dev(struct fib_info
*fi
, const struct net_device
*dev
);
451 int fib_validate_source(struct sk_buff
*skb
, __be32 src
, __be32 dst
,
452 dscp_t dscp
, int oif
, struct net_device
*dev
,
453 struct in_device
*idev
, u32
*itag
);
455 static inline enum skb_drop_reason
456 fib_validate_source_reason(struct sk_buff
*skb
, __be32 src
, __be32 dst
,
457 dscp_t dscp
, int oif
, struct net_device
*dev
,
458 struct in_device
*idev
, u32
*itag
)
460 int err
= fib_validate_source(skb
, src
, dst
, dscp
, oif
, dev
, idev
,
464 return SKB_NOT_DROPPED_YET
;
467 #ifdef CONFIG_IP_ROUTE_CLASSID
468 static inline int fib_num_tclassid_users(struct net
*net
)
470 return atomic_read(&net
->ipv4
.fib_num_tclassid_users
);
473 static inline int fib_num_tclassid_users(struct net
*net
)
478 int fib_unmerge(struct net
*net
);
480 static inline bool nhc_l3mdev_matches_dev(const struct fib_nh_common
*nhc
,
481 const struct net_device
*dev
)
483 if (nhc
->nhc_dev
== dev
||
484 l3mdev_master_ifindex_rcu(nhc
->nhc_dev
) == dev
->ifindex
)
490 /* Exported by fib_semantics.c */
491 int ip_fib_check_default(__be32 gw
, struct net_device
*dev
);
492 int fib_sync_down_dev(struct net_device
*dev
, unsigned long event
, bool force
);
493 int fib_sync_down_addr(struct net_device
*dev
, __be32 local
);
494 int fib_sync_up(struct net_device
*dev
, unsigned char nh_flags
);
495 void fib_sync_mtu(struct net_device
*dev
, u32 orig_mtu
);
496 void fib_nhc_update_mtu(struct fib_nh_common
*nhc
, u32
new, u32 orig
);
498 /* Fields used for sysctl_fib_multipath_hash_fields.
499 * Common to IPv4 and IPv6.
501 * Add new fields at the end. This is user API.
503 #define FIB_MULTIPATH_HASH_FIELD_SRC_IP BIT(0)
504 #define FIB_MULTIPATH_HASH_FIELD_DST_IP BIT(1)
505 #define FIB_MULTIPATH_HASH_FIELD_IP_PROTO BIT(2)
506 #define FIB_MULTIPATH_HASH_FIELD_FLOWLABEL BIT(3)
507 #define FIB_MULTIPATH_HASH_FIELD_SRC_PORT BIT(4)
508 #define FIB_MULTIPATH_HASH_FIELD_DST_PORT BIT(5)
509 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP BIT(6)
510 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP BIT(7)
511 #define FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO BIT(8)
512 #define FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL BIT(9)
513 #define FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT BIT(10)
514 #define FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT BIT(11)
516 #define FIB_MULTIPATH_HASH_FIELD_OUTER_MASK \
517 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \
518 FIB_MULTIPATH_HASH_FIELD_DST_IP | \
519 FIB_MULTIPATH_HASH_FIELD_IP_PROTO | \
520 FIB_MULTIPATH_HASH_FIELD_FLOWLABEL | \
521 FIB_MULTIPATH_HASH_FIELD_SRC_PORT | \
522 FIB_MULTIPATH_HASH_FIELD_DST_PORT)
524 #define FIB_MULTIPATH_HASH_FIELD_INNER_MASK \
525 (FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP | \
526 FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP | \
527 FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO | \
528 FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL | \
529 FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT | \
530 FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT)
532 #define FIB_MULTIPATH_HASH_FIELD_ALL_MASK \
533 (FIB_MULTIPATH_HASH_FIELD_OUTER_MASK | \
534 FIB_MULTIPATH_HASH_FIELD_INNER_MASK)
536 #define FIB_MULTIPATH_HASH_FIELD_DEFAULT_MASK \
537 (FIB_MULTIPATH_HASH_FIELD_SRC_IP | \
538 FIB_MULTIPATH_HASH_FIELD_DST_IP | \
539 FIB_MULTIPATH_HASH_FIELD_IP_PROTO)
541 #ifdef CONFIG_IP_ROUTE_MULTIPATH
542 int fib_multipath_hash(const struct net
*net
, const struct flowi4
*fl4
,
543 const struct sk_buff
*skb
, struct flow_keys
*flkeys
);
546 fib_multipath_hash_construct_key(siphash_key_t
*key
, u32 mp_seed
)
548 u64 mp_seed_64
= mp_seed
;
550 key
->key
[0] = (mp_seed_64
<< 32) | mp_seed_64
;
551 key
->key
[1] = key
->key
[0];
554 static inline u32
fib_multipath_hash_from_keys(const struct net
*net
,
555 struct flow_keys
*keys
)
557 siphash_aligned_key_t hash_key
;
560 mp_seed
= READ_ONCE(net
->ipv4
.sysctl_fib_multipath_hash_seed
).mp_seed
;
561 fib_multipath_hash_construct_key(&hash_key
, mp_seed
);
563 return flow_hash_from_keys_seed(keys
, &hash_key
);
566 static inline u32
fib_multipath_hash_from_keys(const struct net
*net
,
567 struct flow_keys
*keys
)
569 return flow_hash_from_keys(keys
);
573 int fib_check_nh(struct net
*net
, struct fib_nh
*nh
, u32 table
, u8 scope
,
574 struct netlink_ext_ack
*extack
);
575 void fib_select_multipath(struct fib_result
*res
, int hash
);
576 void fib_select_path(struct net
*net
, struct fib_result
*res
,
577 struct flowi4
*fl4
, const struct sk_buff
*skb
);
579 int fib_nh_init(struct net
*net
, struct fib_nh
*fib_nh
,
580 struct fib_config
*cfg
, int nh_weight
,
581 struct netlink_ext_ack
*extack
);
582 void fib_nh_release(struct net
*net
, struct fib_nh
*fib_nh
);
583 int fib_nh_common_init(struct net
*net
, struct fib_nh_common
*nhc
,
584 struct nlattr
*fc_encap
, u16 fc_encap_type
,
585 void *cfg
, gfp_t gfp_flags
,
586 struct netlink_ext_ack
*extack
);
587 void fib_nh_common_release(struct fib_nh_common
*nhc
);
589 /* Exported by fib_trie.c */
590 void fib_alias_hw_flags_set(struct net
*net
, const struct fib_rt_info
*fri
);
591 void fib_trie_init(void);
592 struct fib_table
*fib_trie_table(u32 id
, struct fib_table
*alias
);
593 bool fib_lookup_good_nhc(const struct fib_nh_common
*nhc
, int fib_flags
,
594 const struct flowi4
*flp
);
596 static inline void fib_combine_itag(u32
*itag
, const struct fib_result
*res
)
598 #ifdef CONFIG_IP_ROUTE_CLASSID
599 struct fib_nh_common
*nhc
= res
->nhc
;
600 #ifdef CONFIG_IP_MULTIPLE_TABLES
603 if (nhc
->nhc_family
== AF_INET
) {
606 nh
= container_of(nhc
, struct fib_nh
, nh_common
);
607 *itag
= nh
->nh_tclassid
<< 16;
612 #ifdef CONFIG_IP_MULTIPLE_TABLES
613 rtag
= res
->tclassid
;
621 void fib_flush(struct net
*net
);
622 void free_fib_info(struct fib_info
*fi
);
624 static inline void fib_info_hold(struct fib_info
*fi
)
626 refcount_inc(&fi
->fib_clntref
);
629 static inline void fib_info_put(struct fib_info
*fi
)
631 if (refcount_dec_and_test(&fi
->fib_clntref
))
635 #ifdef CONFIG_PROC_FS
636 int __net_init
fib_proc_init(struct net
*net
);
637 void __net_exit
fib_proc_exit(struct net
*net
);
639 static inline int fib_proc_init(struct net
*net
)
643 static inline void fib_proc_exit(struct net
*net
)
648 u32
ip_mtu_from_fib_result(struct fib_result
*res
, __be32 daddr
);
650 int ip_valid_fib_dump_req(struct net
*net
, const struct nlmsghdr
*nlh
,
651 struct fib_dump_filter
*filter
,
652 struct netlink_callback
*cb
);
654 int fib_nexthop_info(struct sk_buff
*skb
, const struct fib_nh_common
*nh
,
655 u8 rt_family
, unsigned char *flags
, bool skip_oif
);
656 int fib_add_nexthop(struct sk_buff
*skb
, const struct fib_nh_common
*nh
,
657 int nh_weight
, u8 rt_family
, u32 nh_tclassid
);
658 #endif /* _NET_FIB_H */