1 /* SPDX-License-Identifier: GPL-2.0-or-later */
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
12 #include <linux/ipv6_route.h>
13 #include <linux/rtnetlink.h>
14 #include <linux/spinlock.h>
15 #include <linux/notifier.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>
24 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
25 #define FIB6_TABLE_HASHSZ 256
27 #define FIB6_TABLE_HASHSZ 1
33 #define RT6_TRACE(x...) pr_debug(x)
35 #define RT6_TRACE(x...) do { ; } while (0)
49 u16 fc_type
; /* only 8 bits are used */
50 u16 fc_delete_all_nh
: 1,
55 struct in6_addr fc_dst
;
56 struct in6_addr fc_src
;
57 struct in6_addr fc_prefsrc
;
58 struct in6_addr fc_gateway
;
60 unsigned long fc_expires
;
66 struct nl_info fc_nlinfo
;
67 struct nlattr
*fc_encap
;
73 struct fib6_node __rcu
*parent
;
74 struct fib6_node __rcu
*left
;
75 struct fib6_node __rcu
*right
;
76 #ifdef CONFIG_IPV6_SUBTREES
77 struct fib6_node __rcu
*subtree
;
79 struct fib6_info __rcu
*leaf
;
81 __u16 fn_bit
; /* bit key */
84 struct fib6_info __rcu
*rr_ptr
;
93 #ifndef CONFIG_IPV6_SUBTREES
94 #define FIB6_SUBTREE(fn) NULL
96 static inline bool fib6_routes_require_src(const struct net
*net
)
101 static inline void fib6_routes_require_src_inc(struct net
*net
) {}
102 static inline void fib6_routes_require_src_dec(struct net
*net
) {}
106 static inline bool fib6_routes_require_src(const struct net
*net
)
108 return net
->ipv6
.fib6_routes_require_src
> 0;
111 static inline void fib6_routes_require_src_inc(struct net
*net
)
113 net
->ipv6
.fib6_routes_require_src
++;
116 static inline void fib6_routes_require_src_dec(struct net
*net
)
118 net
->ipv6
.fib6_routes_require_src
--;
121 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
125 * routing information
130 struct in6_addr addr
;
136 struct rt6_exception_bucket
{
137 struct hlist_head chain
;
141 struct rt6_exception
{
142 struct hlist_node hlist
;
143 struct rt6_info
*rt6i
;
148 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
149 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
150 #define FIB6_MAX_DEPTH 5
153 struct fib_nh_common nh_common
;
155 #ifdef CONFIG_IPV6_ROUTER_PREF
156 unsigned long last_probe
;
159 struct rt6_info
* __percpu
*rt6i_pcpu
;
160 struct rt6_exception_bucket __rcu
*rt6i_exception_bucket
;
164 struct fib6_table
*fib6_table
;
165 struct fib6_info __rcu
*fib6_next
;
166 struct fib6_node __rcu
*fib6_node
;
169 * siblings is a list of fib6_info that have the same metric/weight,
170 * destination, but not the same gateway. nsiblings is just a cache
171 * to speed up lookup.
174 struct list_head fib6_siblings
;
175 struct list_head nh_list
;
177 unsigned int fib6_nsiblings
;
180 unsigned long expires
;
181 struct dst_metrics
*fib6_metrics
;
182 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1]
184 struct rt6key fib6_dst
;
186 struct rt6key fib6_src
;
187 struct rt6key fib6_prefsrc
;
202 struct fib6_nh fib6_nh
[];
206 struct dst_entry dst
;
207 struct fib6_info __rcu
*from
;
210 struct rt6key rt6i_dst
;
211 struct rt6key rt6i_src
;
212 struct in6_addr rt6i_gateway
;
213 struct inet6_dev
*rt6i_idev
;
216 struct list_head rt6i_uncached
;
217 struct uncached_list
*rt6i_uncached_list
;
219 /* more non-fragment space at head required */
220 unsigned short rt6i_nfheader_len
;
225 struct fib6_info
*f6i
;
228 struct rt6_info
*rt6
;
231 #define for_each_fib6_node_rt_rcu(fn) \
232 for (rt = rcu_dereference((fn)->leaf); rt; \
233 rt = rcu_dereference(rt->fib6_next))
235 #define for_each_fib6_walker_rt(w) \
236 for (rt = (w)->leaf; rt; \
237 rt = rcu_dereference_protected(rt->fib6_next, 1))
239 static inline struct inet6_dev
*ip6_dst_idev(struct dst_entry
*dst
)
241 return ((struct rt6_info
*)dst
)->rt6i_idev
;
244 static inline bool fib6_requires_src(const struct fib6_info
*rt
)
246 return rt
->fib6_src
.plen
> 0;
249 static inline void fib6_clean_expires(struct fib6_info
*f6i
)
251 f6i
->fib6_flags
&= ~RTF_EXPIRES
;
255 static inline void fib6_set_expires(struct fib6_info
*f6i
,
256 unsigned long expires
)
258 f6i
->expires
= expires
;
259 f6i
->fib6_flags
|= RTF_EXPIRES
;
262 static inline bool fib6_check_expired(const struct fib6_info
*f6i
)
264 if (f6i
->fib6_flags
& RTF_EXPIRES
)
265 return time_after(jiffies
, f6i
->expires
);
269 /* Function to safely get fn->sernum for passed in rt
270 * and store result in passed in cookie.
271 * Return true if we can get cookie safely
272 * Return false if not
274 static inline bool fib6_get_cookie_safe(const struct fib6_info
*f6i
,
277 struct fib6_node
*fn
;
280 fn
= rcu_dereference(f6i
->fib6_node
);
283 *cookie
= fn
->fn_sernum
;
284 /* pairs with smp_wmb() in fib6_update_sernum_upto_root() */
292 static inline u32
rt6_get_cookie(const struct rt6_info
*rt
)
294 struct fib6_info
*from
;
302 from
= rcu_dereference(rt
->from
);
304 fib6_get_cookie_safe(from
, &cookie
);
311 static inline void ip6_rt_put(struct rt6_info
*rt
)
313 /* dst_release() accepts a NULL parameter.
314 * We rely on dst being first structure in struct rt6_info
316 BUILD_BUG_ON(offsetof(struct rt6_info
, dst
) != 0);
317 dst_release(&rt
->dst
);
320 struct fib6_info
*fib6_info_alloc(gfp_t gfp_flags
, bool with_fib6_nh
);
321 void fib6_info_destroy_rcu(struct rcu_head
*head
);
323 static inline void fib6_info_hold(struct fib6_info
*f6i
)
325 refcount_inc(&f6i
->fib6_ref
);
328 static inline bool fib6_info_hold_safe(struct fib6_info
*f6i
)
330 return refcount_inc_not_zero(&f6i
->fib6_ref
);
333 static inline void fib6_info_release(struct fib6_info
*f6i
)
335 if (f6i
&& refcount_dec_and_test(&f6i
->fib6_ref
))
336 call_rcu(&f6i
->rcu
, fib6_info_destroy_rcu
);
339 static inline void fib6_info_hw_flags_set(struct fib6_info
*f6i
, bool offload
,
342 f6i
->offload
= offload
;
346 enum fib6_walk_state
{
347 #ifdef CONFIG_IPV6_SUBTREES
358 struct fib6_node
*root
, *node
;
359 struct fib6_info
*leaf
;
360 enum fib6_walk_state state
;
363 unsigned int skip_in_node
;
364 int (*func
)(struct fib6_walker
*);
368 struct rt6_statistics
{
369 __u32 fib_nodes
; /* all fib6 nodes */
370 __u32 fib_route_nodes
; /* intermediate nodes */
371 __u32 fib_rt_entries
; /* rt entries in fib table */
372 __u32 fib_rt_cache
; /* cached rt entries in exception table */
373 __u32 fib_discarded_routes
; /* total number of routes delete */
375 /* The following stats are not protected by any lock */
376 atomic_t fib_rt_alloc
; /* total number of routes alloced */
377 atomic_t fib_rt_uncache
; /* rt entries in uncached list */
380 #define RTN_TL_ROOT 0x0001
381 #define RTN_ROOT 0x0002 /* tree root node */
382 #define RTN_RTINFO 0x0004 /* node with valid routing info */
385 * priority levels (or metrics)
391 struct hlist_node tb6_hlist
;
394 struct fib6_node tb6_root
;
395 struct inet_peer_base tb6_peers
;
397 unsigned int fib_seq
;
398 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
401 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
402 #define RT6_TABLE_MAIN RT_TABLE_MAIN
403 #define RT6_TABLE_DFLT RT6_TABLE_MAIN
404 #define RT6_TABLE_INFO RT6_TABLE_MAIN
405 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN
407 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
408 #define FIB6_TABLE_MIN 1
409 #define FIB6_TABLE_MAX RT_TABLE_MAX
410 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL
412 #define FIB6_TABLE_MIN RT_TABLE_MAIN
413 #define FIB6_TABLE_MAX FIB6_TABLE_MIN
414 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN
417 typedef struct rt6_info
*(*pol_lookup_t
)(struct net
*,
420 const struct sk_buff
*, int);
422 struct fib6_entry_notifier_info
{
423 struct fib_notifier_info info
; /* must be first */
424 struct fib6_info
*rt
;
425 unsigned int nsiblings
;
432 struct fib6_table
*fib6_get_table(struct net
*net
, u32 id
);
433 struct fib6_table
*fib6_new_table(struct net
*net
, u32 id
);
434 struct dst_entry
*fib6_rule_lookup(struct net
*net
, struct flowi6
*fl6
,
435 const struct sk_buff
*skb
,
436 int flags
, pol_lookup_t lookup
);
438 /* called with rcu lock held; can return error pointer
439 * caller needs to select path
441 int fib6_lookup(struct net
*net
, int oif
, struct flowi6
*fl6
,
442 struct fib6_result
*res
, int flags
);
444 /* called with rcu lock held; caller needs to select path */
445 int fib6_table_lookup(struct net
*net
, struct fib6_table
*table
,
446 int oif
, struct flowi6
*fl6
, struct fib6_result
*res
,
449 void fib6_select_path(const struct net
*net
, struct fib6_result
*res
,
450 struct flowi6
*fl6
, int oif
, bool have_oif_match
,
451 const struct sk_buff
*skb
, int strict
);
452 struct fib6_node
*fib6_node_lookup(struct fib6_node
*root
,
453 const struct in6_addr
*daddr
,
454 const struct in6_addr
*saddr
);
456 struct fib6_node
*fib6_locate(struct fib6_node
*root
,
457 const struct in6_addr
*daddr
, int dst_len
,
458 const struct in6_addr
*saddr
, int src_len
,
461 void fib6_clean_all(struct net
*net
, int (*func
)(struct fib6_info
*, void *arg
),
463 void fib6_clean_all_skip_notify(struct net
*net
,
464 int (*func
)(struct fib6_info
*, void *arg
),
467 int fib6_add(struct fib6_node
*root
, struct fib6_info
*rt
,
468 struct nl_info
*info
, struct netlink_ext_ack
*extack
);
469 int fib6_del(struct fib6_info
*rt
, struct nl_info
*info
);
472 void rt6_get_prefsrc(const struct rt6_info
*rt
, struct in6_addr
*addr
)
474 const struct fib6_info
*from
;
478 from
= rcu_dereference(rt
->from
);
480 *addr
= from
->fib6_prefsrc
.addr
;
482 struct in6_addr in6_zero
= {};
490 int fib6_nh_init(struct net
*net
, struct fib6_nh
*fib6_nh
,
491 struct fib6_config
*cfg
, gfp_t gfp_flags
,
492 struct netlink_ext_ack
*extack
);
493 void fib6_nh_release(struct fib6_nh
*fib6_nh
);
495 int call_fib6_entry_notifiers(struct net
*net
,
496 enum fib_event_type event_type
,
497 struct fib6_info
*rt
,
498 struct netlink_ext_ack
*extack
);
499 int call_fib6_multipath_entry_notifiers(struct net
*net
,
500 enum fib_event_type event_type
,
501 struct fib6_info
*rt
,
502 unsigned int nsiblings
,
503 struct netlink_ext_ack
*extack
);
504 int call_fib6_entry_notifiers_replace(struct net
*net
, struct fib6_info
*rt
);
505 void fib6_rt_update(struct net
*net
, struct fib6_info
*rt
,
506 struct nl_info
*info
);
507 void inet6_rt_notify(int event
, struct fib6_info
*rt
, struct nl_info
*info
,
510 void fib6_run_gc(unsigned long expires
, struct net
*net
, bool force
);
512 void fib6_gc_cleanup(void);
516 struct ipv6_route_iter
{
517 struct seq_net_private p
;
518 struct fib6_walker w
;
520 struct fib6_table
*tbl
;
524 extern const struct seq_operations ipv6_route_seq_ops
;
526 int call_fib6_notifier(struct notifier_block
*nb
,
527 enum fib_event_type event_type
,
528 struct fib_notifier_info
*info
);
529 int call_fib6_notifiers(struct net
*net
, enum fib_event_type event_type
,
530 struct fib_notifier_info
*info
);
532 int __net_init
fib6_notifier_init(struct net
*net
);
533 void __net_exit
fib6_notifier_exit(struct net
*net
);
535 unsigned int fib6_tables_seq_read(struct net
*net
);
536 int fib6_tables_dump(struct net
*net
, struct notifier_block
*nb
,
537 struct netlink_ext_ack
*extack
);
539 void fib6_update_sernum(struct net
*net
, struct fib6_info
*rt
);
540 void fib6_update_sernum_upto_root(struct net
*net
, struct fib6_info
*rt
);
541 void fib6_update_sernum_stub(struct net
*net
, struct fib6_info
*f6i
);
543 void fib6_metric_set(struct fib6_info
*f6i
, int metric
, u32 val
);
544 static inline bool fib6_metric_locked(struct fib6_info
*f6i
, int metric
)
546 return !!(f6i
->fib6_metrics
->metrics
[RTAX_LOCK
- 1] & (1 << metric
));
549 #if IS_BUILTIN(CONFIG_IPV6) && defined(CONFIG_BPF_SYSCALL)
550 struct bpf_iter__ipv6_route
{
551 __bpf_md_ptr(struct bpf_iter_meta
*, meta
);
552 __bpf_md_ptr(struct fib6_info
*, rt
);
556 INDIRECT_CALLABLE_DECLARE(struct rt6_info
*ip6_pol_route_output(struct net
*net
,
557 struct fib6_table
*table
,
559 const struct sk_buff
*skb
,
561 INDIRECT_CALLABLE_DECLARE(struct rt6_info
*ip6_pol_route_input(struct net
*net
,
562 struct fib6_table
*table
,
564 const struct sk_buff
*skb
,
566 INDIRECT_CALLABLE_DECLARE(struct rt6_info
*__ip6_route_redirect(struct net
*net
,
567 struct fib6_table
*table
,
569 const struct sk_buff
*skb
,
571 INDIRECT_CALLABLE_DECLARE(struct rt6_info
*ip6_pol_route_lookup(struct net
*net
,
572 struct fib6_table
*table
,
574 const struct sk_buff
*skb
,
576 static inline struct rt6_info
*pol_lookup_func(pol_lookup_t lookup
,
578 struct fib6_table
*table
,
580 const struct sk_buff
*skb
,
583 return INDIRECT_CALL_4(lookup
,
584 ip6_pol_route_output
,
586 ip6_pol_route_lookup
,
587 __ip6_route_redirect
,
588 net
, table
, fl6
, skb
, flags
);
591 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
592 static inline bool fib6_has_custom_rules(const struct net
*net
)
594 return net
->ipv6
.fib6_has_custom_rules
;
597 int fib6_rules_init(void);
598 void fib6_rules_cleanup(void);
599 bool fib6_rule_default(const struct fib_rule
*rule
);
600 int fib6_rules_dump(struct net
*net
, struct notifier_block
*nb
,
601 struct netlink_ext_ack
*extack
);
602 unsigned int fib6_rules_seq_read(struct net
*net
);
604 static inline bool fib6_rules_early_flow_dissect(struct net
*net
,
607 struct flow_keys
*flkeys
)
609 unsigned int flag
= FLOW_DISSECTOR_F_STOP_AT_ENCAP
;
611 if (!net
->ipv6
.fib6_rules_require_fldissect
)
614 skb_flow_dissect_flow_keys(skb
, flkeys
, flag
);
615 fl6
->fl6_sport
= flkeys
->ports
.src
;
616 fl6
->fl6_dport
= flkeys
->ports
.dst
;
617 fl6
->flowi6_proto
= flkeys
->basic
.ip_proto
;
622 static inline bool fib6_has_custom_rules(const struct net
*net
)
626 static inline int fib6_rules_init(void)
630 static inline void fib6_rules_cleanup(void)
634 static inline bool fib6_rule_default(const struct fib_rule
*rule
)
638 static inline int fib6_rules_dump(struct net
*net
, struct notifier_block
*nb
,
639 struct netlink_ext_ack
*extack
)
643 static inline unsigned int fib6_rules_seq_read(struct net
*net
)
647 static inline bool fib6_rules_early_flow_dissect(struct net
*net
,
650 struct flow_keys
*flkeys
)