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,
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;
72 struct fib6_node __rcu *parent;
73 struct fib6_node __rcu *left;
74 struct fib6_node __rcu *right;
75 #ifdef CONFIG_IPV6_SUBTREES
76 struct fib6_node __rcu *subtree;
78 struct fib6_info __rcu *leaf;
80 __u16 fn_bit; /* bit key */
83 struct fib6_info __rcu *rr_ptr;
92 #ifndef CONFIG_IPV6_SUBTREES
93 #define FIB6_SUBTREE(fn) NULL
95 #define FIB6_SUBTREE(fn) (rcu_dereference_protected((fn)->subtree, 1))
104 struct in6_addr addr;
110 struct rt6_exception_bucket {
111 struct hlist_head chain;
115 struct rt6_exception {
116 struct hlist_node hlist;
117 struct rt6_info *rt6i;
122 #define FIB6_EXCEPTION_BUCKET_SIZE_SHIFT 10
123 #define FIB6_EXCEPTION_BUCKET_SIZE (1 << FIB6_EXCEPTION_BUCKET_SIZE_SHIFT)
124 #define FIB6_MAX_DEPTH 5
127 struct in6_addr nh_gw;
128 struct net_device *nh_dev;
129 struct lwtunnel_state *nh_lwtstate;
131 unsigned int nh_flags;
132 atomic_t nh_upper_bound;
137 struct fib6_table *fib6_table;
138 struct fib6_info __rcu *rt6_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;
176 struct dst_entry dst;
177 struct fib6_info *from;
179 struct rt6key rt6i_dst;
180 struct rt6key rt6i_src;
181 struct in6_addr rt6i_gateway;
182 struct inet6_dev *rt6i_idev;
184 struct rt6key rt6i_prefsrc;
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;
193 #define for_each_fib6_node_rt_rcu(fn) \
194 for (rt = rcu_dereference((fn)->leaf); rt; \
195 rt = rcu_dereference(rt->rt6_next))
197 #define for_each_fib6_walker_rt(w) \
198 for (rt = (w)->leaf; rt; \
199 rt = rcu_dereference_protected(rt->rt6_next, 1))
201 static inline struct inet6_dev *ip6_dst_idev(struct dst_entry *dst)
203 return ((struct rt6_info *)dst)->rt6i_idev;
206 static inline void fib6_clean_expires(struct fib6_info *f6i)
208 f6i->fib6_flags &= ~RTF_EXPIRES;
212 static inline void fib6_set_expires(struct fib6_info *f6i,
213 unsigned long expires)
215 f6i->expires = expires;
216 f6i->fib6_flags |= RTF_EXPIRES;
219 static inline bool fib6_check_expired(const struct fib6_info *f6i)
221 if (f6i->fib6_flags & RTF_EXPIRES)
222 return time_after(jiffies, f6i->expires);
226 /* Function to safely get fn->sernum for passed in rt
227 * and store result in passed in cookie.
228 * Return true if we can get cookie safely
229 * Return false if not
231 static inline bool rt6_get_cookie_safe(const struct fib6_info *f6i,
234 struct fib6_node *fn;
238 fn = rcu_dereference(f6i->fib6_node);
241 *cookie = fn->fn_sernum;
242 /* pairs with smp_wmb() in fib6_update_sernum_upto_root() */
251 static inline u32 rt6_get_cookie(const struct rt6_info *rt)
255 if (rt->rt6i_flags & RTF_PCPU ||
256 (unlikely(!list_empty(&rt->rt6i_uncached)) && rt->from))
257 rt6_get_cookie_safe(rt->from, &cookie);
262 static inline void ip6_rt_put(struct rt6_info *rt)
264 /* dst_release() accepts a NULL parameter.
265 * We rely on dst being first structure in struct rt6_info
267 BUILD_BUG_ON(offsetof(struct rt6_info, dst) != 0);
268 dst_release(&rt->dst);
271 struct fib6_info *fib6_info_alloc(gfp_t gfp_flags);
272 void fib6_info_destroy(struct fib6_info *f6i);
274 static inline void fib6_info_hold(struct fib6_info *f6i)
276 atomic_inc(&f6i->fib6_ref);
279 static inline void fib6_info_release(struct fib6_info *f6i)
281 if (f6i && atomic_dec_and_test(&f6i->fib6_ref))
282 fib6_info_destroy(f6i);
285 enum fib6_walk_state {
286 #ifdef CONFIG_IPV6_SUBTREES
297 struct fib6_node *root, *node;
298 struct fib6_info *leaf;
299 enum fib6_walk_state state;
302 int (*func)(struct fib6_walker *);
306 struct rt6_statistics {
307 __u32 fib_nodes; /* all fib6 nodes */
308 __u32 fib_route_nodes; /* intermediate nodes */
309 __u32 fib_rt_entries; /* rt entries in fib table */
310 __u32 fib_rt_cache; /* cached rt entries in exception table */
311 __u32 fib_discarded_routes; /* total number of routes delete */
313 /* The following stats are not protected by any lock */
314 atomic_t fib_rt_alloc; /* total number of routes alloced */
315 atomic_t fib_rt_uncache; /* rt entries in uncached list */
318 #define RTN_TL_ROOT 0x0001
319 #define RTN_ROOT 0x0002 /* tree root node */
320 #define RTN_RTINFO 0x0004 /* node with valid routing info */
323 * priority levels (or metrics)
329 struct hlist_node tb6_hlist;
332 struct fib6_node tb6_root;
333 struct inet_peer_base tb6_peers;
335 unsigned int fib_seq;
336 #define RT6_TABLE_HAS_DFLT_ROUTER BIT(0)
339 #define RT6_TABLE_UNSPEC RT_TABLE_UNSPEC
340 #define RT6_TABLE_MAIN RT_TABLE_MAIN
341 #define RT6_TABLE_DFLT RT6_TABLE_MAIN
342 #define RT6_TABLE_INFO RT6_TABLE_MAIN
343 #define RT6_TABLE_PREFIX RT6_TABLE_MAIN
345 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
346 #define FIB6_TABLE_MIN 1
347 #define FIB6_TABLE_MAX RT_TABLE_MAX
348 #define RT6_TABLE_LOCAL RT_TABLE_LOCAL
350 #define FIB6_TABLE_MIN RT_TABLE_MAIN
351 #define FIB6_TABLE_MAX FIB6_TABLE_MIN
352 #define RT6_TABLE_LOCAL RT6_TABLE_MAIN
355 typedef struct rt6_info *(*pol_lookup_t)(struct net *,
358 const struct sk_buff *, int);
360 struct fib6_entry_notifier_info {
361 struct fib_notifier_info info; /* must be first */
362 struct fib6_info *rt;
369 struct fib6_table *fib6_get_table(struct net *net, u32 id);
370 struct fib6_table *fib6_new_table(struct net *net, u32 id);
371 struct dst_entry *fib6_rule_lookup(struct net *net, struct flowi6 *fl6,
372 const struct sk_buff *skb,
373 int flags, pol_lookup_t lookup);
375 struct fib6_node *fib6_lookup(struct fib6_node *root,
376 const struct in6_addr *daddr,
377 const struct in6_addr *saddr);
379 struct fib6_node *fib6_locate(struct fib6_node *root,
380 const struct in6_addr *daddr, int dst_len,
381 const struct in6_addr *saddr, int src_len,
384 void fib6_clean_all(struct net *net, int (*func)(struct fib6_info *, void *arg),
387 int fib6_add(struct fib6_node *root, struct fib6_info *rt,
388 struct nl_info *info, struct netlink_ext_ack *extack);
389 int fib6_del(struct fib6_info *rt, struct nl_info *info);
391 static inline struct net_device *fib6_info_nh_dev(const struct fib6_info *f6i)
393 return f6i->fib6_nh.nh_dev;
396 void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info,
399 void fib6_run_gc(unsigned long expires, struct net *net, bool force);
401 void fib6_gc_cleanup(void);
405 int ipv6_route_open(struct inode *inode, struct file *file);
407 int call_fib6_notifier(struct notifier_block *nb, struct net *net,
408 enum fib_event_type event_type,
409 struct fib_notifier_info *info);
410 int call_fib6_notifiers(struct net *net, enum fib_event_type event_type,
411 struct fib_notifier_info *info);
413 int __net_init fib6_notifier_init(struct net *net);
414 void __net_exit fib6_notifier_exit(struct net *net);
416 unsigned int fib6_tables_seq_read(struct net *net);
417 int fib6_tables_dump(struct net *net, struct notifier_block *nb);
419 void fib6_update_sernum(struct net *net, struct fib6_info *rt);
420 void fib6_update_sernum_upto_root(struct net *net, struct fib6_info *rt);
422 void fib6_metric_set(struct fib6_info *f6i, int metric, u32 val);
423 static inline bool fib6_metric_locked(struct fib6_info *f6i, int metric)
425 return !!(f6i->fib6_metrics->metrics[RTAX_LOCK - 1] & (1 << metric));
428 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
429 int fib6_rules_init(void);
430 void fib6_rules_cleanup(void);
431 bool fib6_rule_default(const struct fib_rule *rule);
432 int fib6_rules_dump(struct net *net, struct notifier_block *nb);
433 unsigned int fib6_rules_seq_read(struct net *net);
435 static inline bool fib6_rules_early_flow_dissect(struct net *net,
438 struct flow_keys *flkeys)
440 unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
442 if (!net->ipv6.fib6_rules_require_fldissect)
445 skb_flow_dissect_flow_keys(skb, flkeys, flag);
446 fl6->fl6_sport = flkeys->ports.src;
447 fl6->fl6_dport = flkeys->ports.dst;
448 fl6->flowi6_proto = flkeys->basic.ip_proto;
453 static inline int fib6_rules_init(void)
457 static inline void fib6_rules_cleanup(void)
461 static inline bool fib6_rule_default(const struct fib_rule *rule)
465 static inline int fib6_rules_dump(struct net *net, struct notifier_block *nb)
469 static inline unsigned int fib6_rules_seq_read(struct net *net)
473 static inline bool fib6_rules_early_flow_dissect(struct net *net,
476 struct flow_keys *flkeys)