2 * Linux INET6 implementation
5 * Pedro Roque <roque@di.fc.ul.pt>
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version
10 * 2 of the License, or (at your option) any later version.
16 #include <linux/ipv6_route.h>
17 #include <linux/rtnetlink.h>
18 #include <linux/spinlock.h>
19 #include <linux/notifier.h>
22 #include <net/netlink.h>
23 #include <net/inetpeer.h>
24 #include <net/fib_notifier.h>
26 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
27 #define FIB6_TABLE_HASHSZ 256
29 #define FIB6_TABLE_HASHSZ 1
35 #define RT6_TRACE(x...) pr_debug(x)
37 #define RT6_TRACE(x...) do { ; } while (0)
51 u16 fc_type; /* only 8 bits are used */
52 u16 fc_delete_all_nh : 1,
56 struct in6_addr fc_dst;
57 struct in6_addr fc_src;
58 struct in6_addr fc_prefsrc;
59 struct in6_addr fc_gateway;
61 unsigned long fc_expires;
67 struct nl_info fc_nlinfo;
68 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 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
100 * routing information
105 struct in6_addr addr;
111 struct rt6_exception_bucket {
112 struct hlist_head chain;
116 struct rt6_exception {
117 struct hlist_node hlist;
118 struct rt6_info *rt6i;
123 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
124 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
125 #define FIB6_MAX_DEPTH 5
128 struct in6_addr nh_gw;
129 struct net_device *nh_dev;
130 struct lwtunnel_state *nh_lwtstate;
132 unsigned int nh_flags;
133 atomic_t nh_upper_bound;
138 struct fib6_table *fib6_table;
139 struct fib6_info __rcu *fib6_next;
140 struct fib6_node __rcu *fib6_node;
143 * siblings is a list of fib6_info that have the the same metric/weight,
144 * destination, but not the same gateway. nsiblings is just a cache
145 * to speed up lookup.
147 struct list_head fib6_siblings;
148 unsigned int fib6_nsiblings;
151 unsigned long expires;
152 struct dst_metrics *fib6_metrics;
153 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1]
155 struct rt6key fib6_dst;
157 struct rt6key fib6_src;
158 struct rt6key fib6_prefsrc;
160 struct rt6_info * __percpu *rt6i_pcpu;
161 struct rt6_exception_bucket __rcu *rt6i_exception_bucket;
163 #ifdef CONFIG_IPV6_ROUTER_PREF
164 unsigned long last_probe;
170 u8 exception_bucket_flushed:1,
177 struct fib6_nh fib6_nh;
182 struct dst_entry dst;
183 struct fib6_info __rcu *from;
185 struct rt6key rt6i_dst;
186 struct rt6key rt6i_src;
187 struct in6_addr rt6i_gateway;
188 struct inet6_dev *rt6i_idev;
191 struct list_head rt6i_uncached;
192 struct uncached_list *rt6i_uncached_list;
194 /* more non-fragment space at head required */
195 unsigned short rt6i_nfheader_len;
198 #define for_each_fib6_node_rt_rcu(fn) \
199 for (rt = rcu_dereference((fn)->leaf); rt; \
200 rt = rcu_dereference(rt->fib6_next))
202 #define for_each_fib6_walker_rt(w) \
203 for (rt = (w)->leaf; rt; \
204 rt = rcu_dereference_protected(rt->fib6_next, 1))
206 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst)
208 return ((struct rt6_info *)dst)->rt6i_idev;
211 static inline void fib6_clean_expires(struct fib6_info *f6i)
213 f6i->fib6_flags &= ~RTF_EXPIRES;
217 static inline void fib6_set_expires(struct fib6_info *f6i,
218 unsigned long expires)
220 f6i->expires = expires;
221 f6i->fib6_flags |= RTF_EXPIRES;
224 static inline bool fib6_check_expired(const struct fib6_info *f6i)
226 if (f6i->fib6_flags & RTF_EXPIRES)
227 return time_after(jiffies, f6i->expires);
231 /* Function to safely get fn->sernum for passed in rt
232 * and store result in passed in cookie.
233 * Return true if we can get cookie safely
234 * Return false if not
236 static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i,
239 struct fib6_node *fn;
242 fn = rcu_dereference(f6i->fib6_node);
245 *cookie = fn->fn_sernum;
246 /* pairs with smp_wmb() in fib6_update_sernum_upto_root() */
254 static inline u32 rt6_get_cookie(const struct rt6_info *rt)
256 struct fib6_info *from;
261 from = rcu_dereference(rt->from);
262 if (from && (rt->rt6i_flags & RTF_PCPU ||
263 unlikely(!list_empty(&rt->rt6i_uncached))))
264 fib6_get_cookie_safe(from, &cookie);
271 static inline void ip6_rt_put(struct rt6_info *rt)
273 /* dst_release() accepts a NULL parameter.
274 * We rely on dst being first structure in struct rt6_info
276 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0);
277 dst_release(&rt->dst);
280 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags);
281 void fib6_info_destroy_rcu(struct rcu_head *head);
283 static inline void fib6_info_hold(struct fib6_info *f6i)
285 atomic_inc(&f6i->fib6_ref);
288 static inline bool fib6_info_hold_safe(struct fib6_info *f6i)
290 return atomic_inc_not_zero(&f6i->fib6_ref);
293 static inline void fib6_info_release(struct fib6_info *f6i)
295 if (f6i && atomic_dec_and_test(&f6i->fib6_ref))
296 call_rcu(&f6i->rcu, fib6_info_destroy_rcu);
299 enum fib6_walk_state {
300 #ifdef CONFIG_IPV6_SUBTREES
311 struct fib6_node *root, *node;
312 struct fib6_info *leaf;
313 enum fib6_walk_state state;
316 int (*func)(struct fib6_walker *);
320 struct rt6_statistics {
321 __u32 fib_nodes; /* all fib6 nodes */
322 __u32 fib_route_nodes; /* intermediate nodes */
323 __u32 fib_rt_entries; /* rt entries in fib table */
324 __u32 fib_rt_cache; /* cached rt entries in exception table */
325 __u32 fib_discarded_routes; /* total number of routes delete */
327 /* The following stats are not protected by any lock */
328 atomic_t fib_rt_alloc; /* total number of routes alloced */
329 atomic_t fib_rt_uncache; /* rt entries in uncached list */
332 #define RTN_TL_ROOT 0x0001
333 #define RTN_ROOT 0x0002 /* tree root node */
334 #define RTN_RTINFO 0x0004 /* node with valid routing info */
337 * priority levels (or metrics)
343 struct hlist_node tb6_hlist;
346 struct fib6_node tb6_root;
347 struct inet_peer_base tb6_peers;
349 unsigned int fib_seq;
350 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
353 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
354 #define RT6_TABLE_MAIN RT_TABLE_MAIN
355 #define RT6_TABLE_DFLT RT6_TABLE_MAIN
356 #define RT6_TABLE_INFO RT6_TABLE_MAIN
357 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN
359 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
360 #define FIB6_TABLE_MIN 1
361 #define FIB6_TABLE_MAX RT_TABLE_MAX
362 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL
364 #define FIB6_TABLE_MIN RT_TABLE_MAIN
365 #define FIB6_TABLE_MAX FIB6_TABLE_MIN
366 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN
369 typedef struct rt6_info *(*pol_lookup_t)(struct net *,
372 const struct sk_buff *, int);
374 struct fib6_entry_notifier_info {
375 struct fib_notifier_info info; /* must be first */
376 struct fib6_info *rt;
383 struct fib6_table *fib6_get_table(struct net *net, u32 id);
384 struct fib6_table *fib6_new_table(struct net *net, u32 id);
385 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6,
386 const struct sk_buff *skb,
387 int flags, pol_lookup_t lookup);
389 /* called with rcu lock held; can return error pointer
390 * caller needs to select path
392 struct fib6_info *fib6_lookup(struct net *net, int oif, struct flowi6 *fl6,
395 /* called with rcu lock held; caller needs to select path */
396 struct fib6_info *fib6_table_lookup(struct net *net, struct fib6_table *table,
397 int oif, struct flowi6 *fl6, int strict);
399 struct fib6_info *fib6_multipath_select(const struct net *net,
400 struct fib6_info *match,
401 struct flowi6 *fl6, int oif,
402 const struct sk_buff *skb, int strict);
404 struct fib6_node *fib6_node_lookup(struct fib6_node *root,
405 const struct in6_addr *daddr,
406 const struct in6_addr *saddr);
408 struct fib6_node *fib6_locate(struct fib6_node *root,
409 const struct in6_addr *daddr, int dst_len,
410 const struct in6_addr *saddr, int src_len,
413 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg),
415 void fib6_clean_all_skip_notify(struct net *net,
416 int (*func)(struct fib6_info *, void *arg),
419 int fib6_add(struct fib6_node *root, struct fib6_info *rt,
420 struct nl_info *info, struct netlink_ext_ack *extack);
421 int fib6_del(struct fib6_info *rt, struct nl_info *info);
424 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr)
426 const struct fib6_info *from;
430 from = rcu_dereference(rt->from);
432 *addr = from->fib6_prefsrc.addr;
434 struct in6_addr in6_zero = {};
442 static inline struct net_device *fib6_info_nh_dev(const struct fib6_info *f6i)
444 return f6i->fib6_nh.nh_dev;
448 struct lwtunnel_state *fib6_info_nh_lwt(const struct fib6_info *f6i)
450 return f6i->fib6_nh.nh_lwtstate;
453 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info,
456 void fib6_run_gc(unsigned long expires, struct net *net, bool force);
458 void fib6_gc_cleanup(void);
462 struct ipv6_route_iter {
463 struct seq_net_private p;
464 struct fib6_walker w;
466 struct fib6_table *tbl;
470 extern const struct seq_operations ipv6_route_seq_ops;
472 int call_fib6_notifier(struct notifier_block *nb, struct net *net,
473 enum fib_event_type event_type,
474 struct fib_notifier_info *info);
475 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
476 struct fib_notifier_info *info);
478 int __net_init fib6_notifier_init(struct net *net);
479 void __net_exit fib6_notifier_exit(struct net *net);
481 unsigned int fib6_tables_seq_read(struct net *net);
482 int fib6_tables_dump(struct net *net, struct notifier_block *nb);
484 void fib6_update_sernum(struct net *net, struct fib6_info *rt);
485 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt);
487 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val);
488 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric)
490 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric));
493 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
494 int fib6_rules_init(void);
495 void fib6_rules_cleanup(void);
496 bool fib6_rule_default(const struct fib_rule *rule);
497 int fib6_rules_dump(struct net *net, struct notifier_block *nb);
498 unsigned int fib6_rules_seq_read(struct net *net);
500 static inline bool fib6_rules_early_flow_dissect(struct net *net,
503 struct flow_keys *flkeys)
505 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
507 if (!net->ipv6.fib6_rules_require_fldissect)
510 skb_flow_dissect_flow_keys(skb, flkeys, flag);
511 fl6->fl6_sport = flkeys->ports.src;
512 fl6->fl6_dport = flkeys->ports.dst;
513 fl6->flowi6_proto = flkeys->basic.ip_proto;
518 static inline int fib6_rules_init(void)
522 static inline void fib6_rules_cleanup(void)
526 static inline bool fib6_rule_default(const struct fib_rule *rule)
530 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb)
534 static inline unsigned int fib6_rules_seq_read(struct net *net)
538 static inline bool fib6_rules_early_flow_dissect(struct net *net,
541 struct flow_keys *flkeys)