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>
23 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
24 #define FIB6_TABLE_HASHSZ 256
26 #define FIB6_TABLE_HASHSZ 1
32 #define RT6_TRACE(x...) pr_debug(x)
34 #define RT6_TRACE(x...) do { ; } while (0)
48 u16 fc_type; /* only 8 bits are used */
49 u16 fc_delete_all_nh : 1,
53 struct in6_addr fc_dst;
54 struct in6_addr fc_src;
55 struct in6_addr fc_prefsrc;
56 struct in6_addr fc_gateway;
58 unsigned long fc_expires;
64 struct nl_info fc_nlinfo;
65 struct nlattr *fc_encap;
70 struct fib6_node __rcu *parent;
71 struct fib6_node __rcu *left;
72 struct fib6_node __rcu *right;
73 #ifdef CONFIG_IPV6_SUBTREES
74 struct fib6_node __rcu *subtree;
76 struct fib6_info __rcu *leaf;
78 __u16 fn_bit; /* bit key */
81 struct fib6_info __rcu *rr_ptr;
90 #ifndef CONFIG_IPV6_SUBTREES
91 #define FIB6_SUBTREE(fn) NULL
93 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
102 struct in6_addr addr;
108 struct rt6_exception_bucket {
109 struct hlist_head chain;
113 struct rt6_exception {
114 struct hlist_node hlist;
115 struct rt6_info *rt6i;
120 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
121 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
122 #define FIB6_MAX_DEPTH 5
125 struct fib_nh_common nh_common;
127 #ifdef CONFIG_IPV6_ROUTER_PREF
128 unsigned long last_probe;
131 struct rt6_info * __percpu *rt6i_pcpu;
132 struct rt6_exception_bucket __rcu *rt6i_exception_bucket;
136 struct fib6_table *fib6_table;
137 struct fib6_info __rcu *fib6_next;
138 struct fib6_node __rcu *fib6_node;
141 * siblings is a list of fib6_info that have the the same metric/weight,
142 * destination, but not the same gateway. nsiblings is just a cache
143 * to speed up lookup.
146 struct list_head fib6_siblings;
147 struct list_head nh_list;
149 unsigned int fib6_nsiblings;
152 unsigned long expires;
153 struct dst_metrics *fib6_metrics;
154 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1]
156 struct rt6key fib6_dst;
158 struct rt6key fib6_src;
159 struct rt6key fib6_prefsrc;
173 struct fib6_nh fib6_nh[0];
177 struct dst_entry dst;
178 struct fib6_info __rcu *from;
180 struct rt6key rt6i_dst;
181 struct rt6key rt6i_src;
182 struct in6_addr rt6i_gateway;
183 struct inet6_dev *rt6i_idev;
186 struct list_head rt6i_uncached;
187 struct uncached_list *rt6i_uncached_list;
189 /* more non-fragment space at head required */
190 unsigned short rt6i_nfheader_len;
195 struct fib6_info *f6i;
198 struct rt6_info *rt6;
201 #define for_each_fib6_node_rt_rcu(fn) \
202 for (rt = rcu_dereference((fn)->leaf); rt; \
203 rt = rcu_dereference(rt->fib6_next))
205 #define for_each_fib6_walker_rt(w) \
206 for (rt = (w)->leaf; rt; \
207 rt = rcu_dereference_protected(rt->fib6_next, 1))
209 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst)
211 return ((struct rt6_info *)dst)->rt6i_idev;
214 static inline void fib6_clean_expires(struct fib6_info *f6i)
216 f6i->fib6_flags &= ~RTF_EXPIRES;
220 static inline void fib6_set_expires(struct fib6_info *f6i,
221 unsigned long expires)
223 f6i->expires = expires;
224 f6i->fib6_flags |= RTF_EXPIRES;
227 static inline bool fib6_check_expired(const struct fib6_info *f6i)
229 if (f6i->fib6_flags & RTF_EXPIRES)
230 return time_after(jiffies, f6i->expires);
234 /* Function to safely get fn->sernum for passed in rt
235 * and store result in passed in cookie.
236 * Return true if we can get cookie safely
237 * Return false if not
239 static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i,
242 struct fib6_node *fn;
245 fn = rcu_dereference(f6i->fib6_node);
248 *cookie = fn->fn_sernum;
249 /* pairs with smp_wmb() in fib6_update_sernum_upto_root() */
257 static inline u32 rt6_get_cookie(const struct rt6_info *rt)
259 struct fib6_info *from;
264 from = rcu_dereference(rt->from);
266 fib6_get_cookie_safe(from, &cookie);
273 static inline void ip6_rt_put(struct rt6_info *rt)
275 /* dst_release() accepts a NULL parameter.
276 * We rely on dst being first structure in struct rt6_info
278 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0);
279 dst_release(&rt->dst);
282 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags, bool with_fib6_nh);
283 void fib6_info_destroy_rcu(struct rcu_head *head);
285 static inline void fib6_info_hold(struct fib6_info *f6i)
287 refcount_inc(&f6i->fib6_ref);
290 static inline bool fib6_info_hold_safe(struct fib6_info *f6i)
292 return refcount_inc_not_zero(&f6i->fib6_ref);
295 static inline void fib6_info_release(struct fib6_info *f6i)
297 if (f6i && refcount_dec_and_test(&f6i->fib6_ref))
298 call_rcu(&f6i->rcu, fib6_info_destroy_rcu);
301 enum fib6_walk_state {
302 #ifdef CONFIG_IPV6_SUBTREES
313 struct fib6_node *root, *node;
314 struct fib6_info *leaf;
315 enum fib6_walk_state state;
318 int (*func)(struct fib6_walker *);
322 struct rt6_statistics {
323 __u32 fib_nodes; /* all fib6 nodes */
324 __u32 fib_route_nodes; /* intermediate nodes */
325 __u32 fib_rt_entries; /* rt entries in fib table */
326 __u32 fib_rt_cache; /* cached rt entries in exception table */
327 __u32 fib_discarded_routes; /* total number of routes delete */
329 /* The following stats are not protected by any lock */
330 atomic_t fib_rt_alloc; /* total number of routes alloced */
331 atomic_t fib_rt_uncache; /* rt entries in uncached list */
334 #define RTN_TL_ROOT 0x0001
335 #define RTN_ROOT 0x0002 /* tree root node */
336 #define RTN_RTINFO 0x0004 /* node with valid routing info */
339 * priority levels (or metrics)
345 struct hlist_node tb6_hlist;
348 struct fib6_node tb6_root;
349 struct inet_peer_base tb6_peers;
351 unsigned int fib_seq;
352 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
355 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
356 #define RT6_TABLE_MAIN RT_TABLE_MAIN
357 #define RT6_TABLE_DFLT RT6_TABLE_MAIN
358 #define RT6_TABLE_INFO RT6_TABLE_MAIN
359 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN
361 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
362 #define FIB6_TABLE_MIN 1
363 #define FIB6_TABLE_MAX RT_TABLE_MAX
364 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL
366 #define FIB6_TABLE_MIN RT_TABLE_MAIN
367 #define FIB6_TABLE_MAX FIB6_TABLE_MIN
368 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN
371 typedef struct rt6_info *(*pol_lookup_t)(struct net *,
374 const struct sk_buff *, int);
376 struct fib6_entry_notifier_info {
377 struct fib_notifier_info info; /* must be first */
378 struct fib6_info *rt;
385 struct fib6_table *fib6_get_table(struct net *net, u32 id);
386 struct fib6_table *fib6_new_table(struct net *net, u32 id);
387 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6,
388 const struct sk_buff *skb,
389 int flags, pol_lookup_t lookup);
391 /* called with rcu lock held; can return error pointer
392 * caller needs to select path
394 int fib6_lookup(struct net *net, int oif, struct flowi6 *fl6,
395 struct fib6_result *res, int flags);
397 /* called with rcu lock held; caller needs to select path */
398 int fib6_table_lookup(struct net *net, struct fib6_table *table,
399 int oif, struct flowi6 *fl6, struct fib6_result *res,
402 void fib6_select_path(const struct net *net, struct fib6_result *res,
403 struct flowi6 *fl6, int oif, bool have_oif_match,
404 const struct sk_buff *skb, int strict);
405 struct fib6_node *fib6_node_lookup(struct fib6_node *root,
406 const struct in6_addr *daddr,
407 const struct in6_addr *saddr);
409 struct fib6_node *fib6_locate(struct fib6_node *root,
410 const struct in6_addr *daddr, int dst_len,
411 const struct in6_addr *saddr, int src_len,
414 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg),
416 void fib6_clean_all_skip_notify(struct net *net,
417 int (*func)(struct fib6_info *, void *arg),
420 int fib6_add(struct fib6_node *root, struct fib6_info *rt,
421 struct nl_info *info, struct netlink_ext_ack *extack);
422 int fib6_del(struct fib6_info *rt, struct nl_info *info);
425 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr)
427 const struct fib6_info *from;
431 from = rcu_dereference(rt->from);
433 *addr = from->fib6_prefsrc.addr;
435 struct in6_addr in6_zero = {};
443 int fib6_nh_init(struct net *net, struct fib6_nh *fib6_nh,
444 struct fib6_config *cfg, gfp_t gfp_flags,
445 struct netlink_ext_ack *extack);
446 void fib6_nh_release(struct fib6_nh *fib6_nh);
448 int call_fib6_entry_notifiers(struct net *net,
449 enum fib_event_type event_type,
450 struct fib6_info *rt,
451 struct netlink_ext_ack *extack);
452 void fib6_rt_update(struct net *net, struct fib6_info *rt,
453 struct nl_info *info);
454 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info,
457 void fib6_run_gc(unsigned long expires, struct net *net, bool force);
459 void fib6_gc_cleanup(void);
463 struct ipv6_route_iter {
464 struct seq_net_private p;
465 struct fib6_walker w;
467 struct fib6_table *tbl;
471 extern const struct seq_operations ipv6_route_seq_ops;
473 int call_fib6_notifier(struct notifier_block *nb, struct net *net,
474 enum fib_event_type event_type,
475 struct fib_notifier_info *info);
476 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
477 struct fib_notifier_info *info);
479 int __net_init fib6_notifier_init(struct net *net);
480 void __net_exit fib6_notifier_exit(struct net *net);
482 unsigned int fib6_tables_seq_read(struct net *net);
483 int fib6_tables_dump(struct net *net, struct notifier_block *nb);
485 void fib6_update_sernum(struct net *net, struct fib6_info *rt);
486 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt);
487 void fib6_update_sernum_stub(struct net *net, struct fib6_info *f6i);
489 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val);
490 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric)
492 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric));
495 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
496 int fib6_rules_init(void);
497 void fib6_rules_cleanup(void);
498 bool fib6_rule_default(const struct fib_rule *rule);
499 int fib6_rules_dump(struct net *net, struct notifier_block *nb);
500 unsigned int fib6_rules_seq_read(struct net *net);
502 static inline bool fib6_rules_early_flow_dissect(struct net *net,
505 struct flow_keys *flkeys)
507 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
509 if (!net->ipv6.fib6_rules_require_fldissect)
512 skb_flow_dissect_flow_keys(skb, flkeys, flag);
513 fl6->fl6_sport = flkeys->ports.src;
514 fl6->fl6_dport = flkeys->ports.dst;
515 fl6->flowi6_proto = flkeys->basic.ip_proto;
520 static inline int fib6_rules_init(void)
524 static inline void fib6_rules_cleanup(void)
528 static inline bool fib6_rule_default(const struct fib_rule *rule)
532 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb)
536 static inline unsigned int fib6_rules_seq_read(struct net *net)
540 static inline bool fib6_rules_early_flow_dissect(struct net *net,
543 struct flow_keys *flkeys)