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/ip_fib.h>
23 #include <net/netlink.h>
24 #include <net/inetpeer.h>
25 #include <net/fib_notifier.h>
27 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
28 #define FIB6_TABLE_HASHSZ 256
30 #define FIB6_TABLE_HASHSZ 1
36 #define RT6_TRACE(x...) pr_debug(x)
38 #define RT6_TRACE(x...) do { ; } while (0)
52 u16 fc_type; /* only 8 bits are used */
53 u16 fc_delete_all_nh : 1,
57 struct in6_addr fc_dst;
58 struct in6_addr fc_src;
59 struct in6_addr fc_prefsrc;
60 struct in6_addr fc_gateway;
62 unsigned long fc_expires;
68 struct nl_info fc_nlinfo;
69 struct nlattr *fc_encap;
74 struct fib6_node __rcu *parent;
75 struct fib6_node __rcu *left;
76 struct fib6_node __rcu *right;
77 #ifdef CONFIG_IPV6_SUBTREES
78 struct fib6_node __rcu *subtree;
80 struct fib6_info __rcu *leaf;
82 __u16 fn_bit; /* bit key */
85 struct fib6_info __rcu *rr_ptr;
94 #ifndef CONFIG_IPV6_SUBTREES
95 #define FIB6_SUBTREE(fn) NULL
97 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
101 * routing information
106 struct in6_addr addr;
112 struct rt6_exception_bucket {
113 struct hlist_head chain;
117 struct rt6_exception {
118 struct hlist_node hlist;
119 struct rt6_info *rt6i;
124 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
125 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
126 #define FIB6_MAX_DEPTH 5
129 struct fib_nh_common nh_common;
131 #ifdef CONFIG_IPV6_ROUTER_PREF
132 unsigned long last_probe;
137 struct fib6_table *fib6_table;
138 struct fib6_info __rcu *fib6_next;
139 struct fib6_node __rcu *fib6_node;
142 * siblings is a list of fib6_info that have the the same metric/weight,
143 * destination, but not the same gateway. nsiblings is just a cache
144 * to speed up lookup.
146 struct list_head fib6_siblings;
147 unsigned int fib6_nsiblings;
150 unsigned long expires;
151 struct dst_metrics *fib6_metrics;
152 #define fib6_pmtu fib6_metrics->metrics[RTAX_MTU-1]
154 struct rt6key fib6_dst;
156 struct rt6key fib6_src;
157 struct rt6key fib6_prefsrc;
159 struct rt6_info * __percpu *rt6i_pcpu;
160 struct rt6_exception_bucket __rcu *rt6i_exception_bucket;
165 u8 exception_bucket_flushed:1,
172 struct fib6_nh fib6_nh;
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);
265 if (from && (rt->rt6i_flags & RTF_PCPU ||
266 unlikely(!list_empty(&rt->rt6i_uncached))))
267 fib6_get_cookie_safe(from, &cookie);
274 static inline void ip6_rt_put(struct rt6_info *rt)
276 /* dst_release() accepts a NULL parameter.
277 * We rely on dst being first structure in struct rt6_info
279 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0);
280 dst_release(&rt->dst);
283 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags);
284 void fib6_info_destroy_rcu(struct rcu_head *head);
286 static inline void fib6_info_hold(struct fib6_info *f6i)
288 refcount_inc(&f6i->fib6_ref);
291 static inline bool fib6_info_hold_safe(struct fib6_info *f6i)
293 return refcount_inc_not_zero(&f6i->fib6_ref);
296 static inline void fib6_info_release(struct fib6_info *f6i)
298 if (f6i && refcount_dec_and_test(&f6i->fib6_ref))
299 call_rcu(&f6i->rcu, fib6_info_destroy_rcu);
302 enum fib6_walk_state {
303 #ifdef CONFIG_IPV6_SUBTREES
314 struct fib6_node *root, *node;
315 struct fib6_info *leaf;
316 enum fib6_walk_state state;
319 int (*func)(struct fib6_walker *);
323 struct rt6_statistics {
324 __u32 fib_nodes; /* all fib6 nodes */
325 __u32 fib_route_nodes; /* intermediate nodes */
326 __u32 fib_rt_entries; /* rt entries in fib table */
327 __u32 fib_rt_cache; /* cached rt entries in exception table */
328 __u32 fib_discarded_routes; /* total number of routes delete */
330 /* The following stats are not protected by any lock */
331 atomic_t fib_rt_alloc; /* total number of routes alloced */
332 atomic_t fib_rt_uncache; /* rt entries in uncached list */
335 #define RTN_TL_ROOT 0x0001
336 #define RTN_ROOT 0x0002 /* tree root node */
337 #define RTN_RTINFO 0x0004 /* node with valid routing info */
340 * priority levels (or metrics)
346 struct hlist_node tb6_hlist;
349 struct fib6_node tb6_root;
350 struct inet_peer_base tb6_peers;
352 unsigned int fib_seq;
353 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
356 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
357 #define RT6_TABLE_MAIN RT_TABLE_MAIN
358 #define RT6_TABLE_DFLT RT6_TABLE_MAIN
359 #define RT6_TABLE_INFO RT6_TABLE_MAIN
360 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN
362 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
363 #define FIB6_TABLE_MIN 1
364 #define FIB6_TABLE_MAX RT_TABLE_MAX
365 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL
367 #define FIB6_TABLE_MIN RT_TABLE_MAIN
368 #define FIB6_TABLE_MAX FIB6_TABLE_MIN
369 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN
372 typedef struct rt6_info *(*pol_lookup_t)(struct net *,
375 const struct sk_buff *, int);
377 struct fib6_entry_notifier_info {
378 struct fib_notifier_info info; /* must be first */
379 struct fib6_info *rt;
386 struct fib6_table *fib6_get_table(struct net *net, u32 id);
387 struct fib6_table *fib6_new_table(struct net *net, u32 id);
388 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6,
389 const struct sk_buff *skb,
390 int flags, pol_lookup_t lookup);
392 /* called with rcu lock held; can return error pointer
393 * caller needs to select path
395 int fib6_lookup(struct net *net, int oif, struct flowi6 *fl6,
396 struct fib6_result *res, int flags);
398 /* called with rcu lock held; caller needs to select path */
399 int fib6_table_lookup(struct net *net, struct fib6_table *table,
400 int oif, struct flowi6 *fl6, struct fib6_result *res,
403 void fib6_select_path(const struct net *net, struct fib6_result *res,
404 struct flowi6 *fl6, int oif, bool have_oif_match,
405 const struct sk_buff *skb, int strict);
406 struct fib6_node *fib6_node_lookup(struct fib6_node *root,
407 const struct in6_addr *daddr,
408 const struct in6_addr *saddr);
410 struct fib6_node *fib6_locate(struct fib6_node *root,
411 const struct in6_addr *daddr, int dst_len,
412 const struct in6_addr *saddr, int src_len,
415 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg),
417 void fib6_clean_all_skip_notify(struct net *net,
418 int (*func)(struct fib6_info *, void *arg),
421 int fib6_add(struct fib6_node *root, struct fib6_info *rt,
422 struct nl_info *info, struct netlink_ext_ack *extack);
423 int fib6_del(struct fib6_info *rt, struct nl_info *info);
426 void rt6_get_prefsrc(const struct rt6_info *rt, struct in6_addr *addr)
428 const struct fib6_info *from;
432 from = rcu_dereference(rt->from);
434 *addr = from->fib6_prefsrc.addr;
436 struct in6_addr in6_zero = {};
444 static inline struct net_device *fib6_info_nh_dev(const struct fib6_info *f6i)
446 return f6i->fib6_nh.fib_nh_dev;
449 int fib6_nh_init(struct net *net, struct fib6_nh *fib6_nh,
450 struct fib6_config *cfg, gfp_t gfp_flags,
451 struct netlink_ext_ack *extack);
452 void fib6_nh_release(struct fib6_nh *fib6_nh);
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);
488 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val);
489 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric)
491 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric));
494 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
495 int fib6_rules_init(void);
496 void fib6_rules_cleanup(void);
497 bool fib6_rule_default(const struct fib_rule *rule);
498 int fib6_rules_dump(struct net *net, struct notifier_block *nb);
499 unsigned int fib6_rules_seq_read(struct net *net);
501 static inline bool fib6_rules_early_flow_dissect(struct net *net,
504 struct flow_keys *flkeys)
506 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
508 if (!net->ipv6.fib6_rules_require_fldissect)
511 skb_flow_dissect_flow_keys(skb, flkeys, flag);
512 fl6->fl6_sport = flkeys->ports.src;
513 fl6->fl6_dport = flkeys->ports.dst;
514 fl6->flowi6_proto = flkeys->basic.ip_proto;
519 static inline int fib6_rules_init(void)
523 static inline void fib6_rules_cleanup(void)
527 static inline bool fib6_rule_default(const struct fib_rule *rule)
531 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb)
535 static inline unsigned int fib6_rules_seq_read(struct net *net)
539 static inline bool fib6_rules_early_flow_dissect(struct net *net,
542 struct flow_keys *flkeys)