xref: /openbmc/linux/include/net/ip_fib.h (revision 3557b3fd)
1 /*
2  * INET		An implementation of the TCP/IP protocol suite for the LINUX
3  *		operating system.  INET  is implemented using the  BSD Socket
4  *		interface as the means of communication with the user level.
5  *
6  *		Definitions for the Forwarding Information Base.
7  *
8  * Authors:	A.N.Kuznetsov, <kuznet@ms2.inr.ac.ru>
9  *
10  *		This program is free software; you can redistribute it and/or
11  *		modify it under the terms of the GNU General Public License
12  *		as published by the Free Software Foundation; either version
13  *		2 of the License, or (at your option) any later version.
14  */
15 
16 #ifndef _NET_IP_FIB_H
17 #define _NET_IP_FIB_H
18 
19 #include <net/flow.h>
20 #include <linux/seq_file.h>
21 #include <linux/rcupdate.h>
22 #include <net/fib_notifier.h>
23 #include <net/fib_rules.h>
24 #include <net/inetpeer.h>
25 #include <linux/percpu.h>
26 #include <linux/notifier.h>
27 #include <linux/refcount.h>
28 
29 struct fib_config {
30 	u8			fc_dst_len;
31 	u8			fc_tos;
32 	u8			fc_protocol;
33 	u8			fc_scope;
34 	u8			fc_type;
35 	u8			fc_gw_family;
36 	/* 2 bytes unused */
37 	u32			fc_table;
38 	__be32			fc_dst;
39 	union {
40 		__be32		fc_gw4;
41 		struct in6_addr	fc_gw6;
42 	};
43 	int			fc_oif;
44 	u32			fc_flags;
45 	u32			fc_priority;
46 	__be32			fc_prefsrc;
47 	struct nlattr		*fc_mx;
48 	struct rtnexthop	*fc_mp;
49 	int			fc_mx_len;
50 	int			fc_mp_len;
51 	u32			fc_flow;
52 	u32			fc_nlflags;
53 	struct nl_info		fc_nlinfo;
54 	struct nlattr		*fc_encap;
55 	u16			fc_encap_type;
56 };
57 
58 struct fib_info;
59 struct rtable;
60 
61 struct fib_nh_exception {
62 	struct fib_nh_exception __rcu	*fnhe_next;
63 	int				fnhe_genid;
64 	__be32				fnhe_daddr;
65 	u32				fnhe_pmtu;
66 	bool				fnhe_mtu_locked;
67 	__be32				fnhe_gw;
68 	unsigned long			fnhe_expires;
69 	struct rtable __rcu		*fnhe_rth_input;
70 	struct rtable __rcu		*fnhe_rth_output;
71 	unsigned long			fnhe_stamp;
72 	struct rcu_head			rcu;
73 };
74 
75 struct fnhe_hash_bucket {
76 	struct fib_nh_exception __rcu	*chain;
77 };
78 
79 #define FNHE_HASH_SHIFT		11
80 #define FNHE_HASH_SIZE		(1 << FNHE_HASH_SHIFT)
81 #define FNHE_RECLAIM_DEPTH	5
82 
83 struct fib_nh_common {
84 	struct net_device	*nhc_dev;
85 	int			nhc_oif;
86 	unsigned int		nhc_flags;
87 	struct lwtunnel_state	*nhc_lwtstate;
88 	unsigned char		nhc_scope;
89 	u8			nhc_family;
90 	u8			nhc_gw_family;
91 
92 	union {
93 		__be32          ipv4;
94 		struct in6_addr ipv6;
95 	} nhc_gw;
96 
97 	int			nhc_weight;
98 	atomic_t		nhc_upper_bound;
99 };
100 
101 struct fib_nh {
102 	struct fib_nh_common	nh_common;
103 	struct hlist_node	nh_hash;
104 	struct fib_info		*nh_parent;
105 #ifdef CONFIG_IP_ROUTE_CLASSID
106 	__u32			nh_tclassid;
107 #endif
108 	__be32			nh_saddr;
109 	int			nh_saddr_genid;
110 	struct rtable __rcu * __percpu *nh_pcpu_rth_output;
111 	struct rtable __rcu	*nh_rth_input;
112 	struct fnhe_hash_bucket	__rcu *nh_exceptions;
113 #define fib_nh_family		nh_common.nhc_family
114 #define fib_nh_dev		nh_common.nhc_dev
115 #define fib_nh_oif		nh_common.nhc_oif
116 #define fib_nh_flags		nh_common.nhc_flags
117 #define fib_nh_lws		nh_common.nhc_lwtstate
118 #define fib_nh_scope		nh_common.nhc_scope
119 #define fib_nh_gw_family	nh_common.nhc_gw_family
120 #define fib_nh_gw4		nh_common.nhc_gw.ipv4
121 #define fib_nh_gw6		nh_common.nhc_gw.ipv6
122 #define fib_nh_weight		nh_common.nhc_weight
123 #define fib_nh_upper_bound	nh_common.nhc_upper_bound
124 };
125 
126 /*
127  * This structure contains data shared by many of routes.
128  */
129 
130 struct fib_info {
131 	struct hlist_node	fib_hash;
132 	struct hlist_node	fib_lhash;
133 	struct net		*fib_net;
134 	int			fib_treeref;
135 	refcount_t		fib_clntref;
136 	unsigned int		fib_flags;
137 	unsigned char		fib_dead;
138 	unsigned char		fib_protocol;
139 	unsigned char		fib_scope;
140 	unsigned char		fib_type;
141 	__be32			fib_prefsrc;
142 	u32			fib_tb_id;
143 	u32			fib_priority;
144 	struct dst_metrics	*fib_metrics;
145 #define fib_mtu fib_metrics->metrics[RTAX_MTU-1]
146 #define fib_window fib_metrics->metrics[RTAX_WINDOW-1]
147 #define fib_rtt fib_metrics->metrics[RTAX_RTT-1]
148 #define fib_advmss fib_metrics->metrics[RTAX_ADVMSS-1]
149 	int			fib_nhs;
150 	bool			fib_nh_is_v6;
151 	struct rcu_head		rcu;
152 	struct fib_nh		fib_nh[0];
153 #define fib_dev		fib_nh[0].fib_nh_dev
154 };
155 
156 
157 #ifdef CONFIG_IP_MULTIPLE_TABLES
158 struct fib_rule;
159 #endif
160 
161 struct fib_table;
162 struct fib_result {
163 	__be32			prefix;
164 	unsigned char		prefixlen;
165 	unsigned char		nh_sel;
166 	unsigned char		type;
167 	unsigned char		scope;
168 	u32			tclassid;
169 	struct fib_nh_common	*nhc;
170 	struct fib_info		*fi;
171 	struct fib_table	*table;
172 	struct hlist_head	*fa_head;
173 };
174 
175 struct fib_result_nl {
176 	__be32		fl_addr;   /* To be looked up*/
177 	u32		fl_mark;
178 	unsigned char	fl_tos;
179 	unsigned char   fl_scope;
180 	unsigned char   tb_id_in;
181 
182 	unsigned char   tb_id;      /* Results */
183 	unsigned char	prefixlen;
184 	unsigned char	nh_sel;
185 	unsigned char	type;
186 	unsigned char	scope;
187 	int             err;
188 };
189 
190 static inline struct fib_nh_common *fib_info_nhc(struct fib_info *fi, int nhsel)
191 {
192 	return &fi->fib_nh[nhsel].nh_common;
193 }
194 
195 #ifdef CONFIG_IP_MULTIPLE_TABLES
196 #define FIB_TABLE_HASHSZ 256
197 #else
198 #define FIB_TABLE_HASHSZ 2
199 #endif
200 
201 __be32 fib_info_update_nh_saddr(struct net *net, struct fib_nh *nh);
202 __be32 fib_result_prefsrc(struct net *net, struct fib_result *res);
203 
204 #define FIB_RES_NHC(res)		((res).nhc)
205 #define FIB_RES_DEV(res)	(FIB_RES_NHC(res)->nhc_dev)
206 #define FIB_RES_OIF(res)	(FIB_RES_NHC(res)->nhc_oif)
207 
208 struct fib_entry_notifier_info {
209 	struct fib_notifier_info info; /* must be first */
210 	u32 dst;
211 	int dst_len;
212 	struct fib_info *fi;
213 	u8 tos;
214 	u8 type;
215 	u32 tb_id;
216 };
217 
218 struct fib_nh_notifier_info {
219 	struct fib_notifier_info info; /* must be first */
220 	struct fib_nh *fib_nh;
221 };
222 
223 int call_fib4_notifier(struct notifier_block *nb, struct net *net,
224 		       enum fib_event_type event_type,
225 		       struct fib_notifier_info *info);
226 int call_fib4_notifiers(struct net *net, enum fib_event_type event_type,
227 			struct fib_notifier_info *info);
228 
229 int __net_init fib4_notifier_init(struct net *net);
230 void __net_exit fib4_notifier_exit(struct net *net);
231 
232 void fib_notify(struct net *net, struct notifier_block *nb);
233 
234 struct fib_table {
235 	struct hlist_node	tb_hlist;
236 	u32			tb_id;
237 	int			tb_num_default;
238 	struct rcu_head		rcu;
239 	unsigned long 		*tb_data;
240 	unsigned long		__data[0];
241 };
242 
243 struct fib_dump_filter {
244 	u32			table_id;
245 	/* filter_set is an optimization that an entry is set */
246 	bool			filter_set;
247 	bool			dump_all_families;
248 	unsigned char		protocol;
249 	unsigned char		rt_type;
250 	unsigned int		flags;
251 	struct net_device	*dev;
252 };
253 
254 int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
255 		     struct fib_result *res, int fib_flags);
256 int fib_table_insert(struct net *, struct fib_table *, struct fib_config *,
257 		     struct netlink_ext_ack *extack);
258 int fib_table_delete(struct net *, struct fib_table *, struct fib_config *,
259 		     struct netlink_ext_ack *extack);
260 int fib_table_dump(struct fib_table *table, struct sk_buff *skb,
261 		   struct netlink_callback *cb, struct fib_dump_filter *filter);
262 int fib_table_flush(struct net *net, struct fib_table *table, bool flush_all);
263 struct fib_table *fib_trie_unmerge(struct fib_table *main_tb);
264 void fib_table_flush_external(struct fib_table *table);
265 void fib_free_table(struct fib_table *tb);
266 
267 #ifndef CONFIG_IP_MULTIPLE_TABLES
268 
269 #define TABLE_LOCAL_INDEX	(RT_TABLE_LOCAL & (FIB_TABLE_HASHSZ - 1))
270 #define TABLE_MAIN_INDEX	(RT_TABLE_MAIN  & (FIB_TABLE_HASHSZ - 1))
271 
272 static inline struct fib_table *fib_get_table(struct net *net, u32 id)
273 {
274 	struct hlist_node *tb_hlist;
275 	struct hlist_head *ptr;
276 
277 	ptr = id == RT_TABLE_LOCAL ?
278 		&net->ipv4.fib_table_hash[TABLE_LOCAL_INDEX] :
279 		&net->ipv4.fib_table_hash[TABLE_MAIN_INDEX];
280 
281 	tb_hlist = rcu_dereference_rtnl(hlist_first_rcu(ptr));
282 
283 	return hlist_entry(tb_hlist, struct fib_table, tb_hlist);
284 }
285 
286 static inline struct fib_table *fib_new_table(struct net *net, u32 id)
287 {
288 	return fib_get_table(net, id);
289 }
290 
291 static inline int fib_lookup(struct net *net, const struct flowi4 *flp,
292 			     struct fib_result *res, unsigned int flags)
293 {
294 	struct fib_table *tb;
295 	int err = -ENETUNREACH;
296 
297 	rcu_read_lock();
298 
299 	tb = fib_get_table(net, RT_TABLE_MAIN);
300 	if (tb)
301 		err = fib_table_lookup(tb, flp, res, flags | FIB_LOOKUP_NOREF);
302 
303 	if (err == -EAGAIN)
304 		err = -ENETUNREACH;
305 
306 	rcu_read_unlock();
307 
308 	return err;
309 }
310 
311 static inline bool fib4_rule_default(const struct fib_rule *rule)
312 {
313 	return true;
314 }
315 
316 static inline int fib4_rules_dump(struct net *net, struct notifier_block *nb)
317 {
318 	return 0;
319 }
320 
321 static inline unsigned int fib4_rules_seq_read(struct net *net)
322 {
323 	return 0;
324 }
325 
326 static inline bool fib4_rules_early_flow_dissect(struct net *net,
327 						 struct sk_buff *skb,
328 						 struct flowi4 *fl4,
329 						 struct flow_keys *flkeys)
330 {
331 	return false;
332 }
333 #else /* CONFIG_IP_MULTIPLE_TABLES */
334 int __net_init fib4_rules_init(struct net *net);
335 void __net_exit fib4_rules_exit(struct net *net);
336 
337 struct fib_table *fib_new_table(struct net *net, u32 id);
338 struct fib_table *fib_get_table(struct net *net, u32 id);
339 
340 int __fib_lookup(struct net *net, struct flowi4 *flp,
341 		 struct fib_result *res, unsigned int flags);
342 
343 static inline int fib_lookup(struct net *net, struct flowi4 *flp,
344 			     struct fib_result *res, unsigned int flags)
345 {
346 	struct fib_table *tb;
347 	int err = -ENETUNREACH;
348 
349 	flags |= FIB_LOOKUP_NOREF;
350 	if (net->ipv4.fib_has_custom_rules)
351 		return __fib_lookup(net, flp, res, flags);
352 
353 	rcu_read_lock();
354 
355 	res->tclassid = 0;
356 
357 	tb = rcu_dereference_rtnl(net->ipv4.fib_main);
358 	if (tb)
359 		err = fib_table_lookup(tb, flp, res, flags);
360 
361 	if (!err)
362 		goto out;
363 
364 	tb = rcu_dereference_rtnl(net->ipv4.fib_default);
365 	if (tb)
366 		err = fib_table_lookup(tb, flp, res, flags);
367 
368 out:
369 	if (err == -EAGAIN)
370 		err = -ENETUNREACH;
371 
372 	rcu_read_unlock();
373 
374 	return err;
375 }
376 
377 bool fib4_rule_default(const struct fib_rule *rule);
378 int fib4_rules_dump(struct net *net, struct notifier_block *nb);
379 unsigned int fib4_rules_seq_read(struct net *net);
380 
381 static inline bool fib4_rules_early_flow_dissect(struct net *net,
382 						 struct sk_buff *skb,
383 						 struct flowi4 *fl4,
384 						 struct flow_keys *flkeys)
385 {
386 	unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
387 
388 	if (!net->ipv4.fib_rules_require_fldissect)
389 		return false;
390 
391 	skb_flow_dissect_flow_keys(skb, flkeys, flag);
392 	fl4->fl4_sport = flkeys->ports.src;
393 	fl4->fl4_dport = flkeys->ports.dst;
394 	fl4->flowi4_proto = flkeys->basic.ip_proto;
395 
396 	return true;
397 }
398 
399 #endif /* CONFIG_IP_MULTIPLE_TABLES */
400 
401 /* Exported by fib_frontend.c */
402 extern const struct nla_policy rtm_ipv4_policy[];
403 void ip_fib_init(void);
404 int fib_gw_from_via(struct fib_config *cfg, struct nlattr *nla,
405 		    struct netlink_ext_ack *extack);
406 __be32 fib_compute_spec_dst(struct sk_buff *skb);
407 bool fib_info_nh_uses_dev(struct fib_info *fi, const struct net_device *dev);
408 int fib_validate_source(struct sk_buff *skb, __be32 src, __be32 dst,
409 			u8 tos, int oif, struct net_device *dev,
410 			struct in_device *idev, u32 *itag);
411 #ifdef CONFIG_IP_ROUTE_CLASSID
412 static inline int fib_num_tclassid_users(struct net *net)
413 {
414 	return net->ipv4.fib_num_tclassid_users;
415 }
416 #else
417 static inline int fib_num_tclassid_users(struct net *net)
418 {
419 	return 0;
420 }
421 #endif
422 int fib_unmerge(struct net *net);
423 
424 /* Exported by fib_semantics.c */
425 int ip_fib_check_default(__be32 gw, struct net_device *dev);
426 int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force);
427 int fib_sync_down_addr(struct net_device *dev, __be32 local);
428 int fib_sync_up(struct net_device *dev, unsigned int nh_flags);
429 void fib_sync_mtu(struct net_device *dev, u32 orig_mtu);
430 
431 #ifdef CONFIG_IP_ROUTE_MULTIPATH
432 int fib_multipath_hash(const struct net *net, const struct flowi4 *fl4,
433 		       const struct sk_buff *skb, struct flow_keys *flkeys);
434 #endif
435 void fib_select_multipath(struct fib_result *res, int hash);
436 void fib_select_path(struct net *net, struct fib_result *res,
437 		     struct flowi4 *fl4, const struct sk_buff *skb);
438 
439 int fib_nh_init(struct net *net, struct fib_nh *fib_nh,
440 		struct fib_config *cfg, int nh_weight,
441 		struct netlink_ext_ack *extack);
442 void fib_nh_release(struct net *net, struct fib_nh *fib_nh);
443 int fib_nh_common_init(struct fib_nh_common *nhc, struct nlattr *fc_encap,
444 		       u16 fc_encap_type, void *cfg, gfp_t gfp_flags,
445 		       struct netlink_ext_ack *extack);
446 void fib_nh_common_release(struct fib_nh_common *nhc);
447 
448 /* Exported by fib_trie.c */
449 void fib_trie_init(void);
450 struct fib_table *fib_trie_table(u32 id, struct fib_table *alias);
451 
452 static inline void fib_combine_itag(u32 *itag, const struct fib_result *res)
453 {
454 #ifdef CONFIG_IP_ROUTE_CLASSID
455 	struct fib_nh_common *nhc = res->nhc;
456 	struct fib_nh *nh = container_of(nhc, struct fib_nh, nh_common);
457 #ifdef CONFIG_IP_MULTIPLE_TABLES
458 	u32 rtag;
459 #endif
460 	*itag = nh->nh_tclassid << 16;
461 #ifdef CONFIG_IP_MULTIPLE_TABLES
462 	rtag = res->tclassid;
463 	if (*itag == 0)
464 		*itag = (rtag<<16);
465 	*itag |= (rtag>>16);
466 #endif
467 #endif
468 }
469 
470 void free_fib_info(struct fib_info *fi);
471 
472 static inline void fib_info_hold(struct fib_info *fi)
473 {
474 	refcount_inc(&fi->fib_clntref);
475 }
476 
477 static inline void fib_info_put(struct fib_info *fi)
478 {
479 	if (refcount_dec_and_test(&fi->fib_clntref))
480 		free_fib_info(fi);
481 }
482 
483 #ifdef CONFIG_PROC_FS
484 int __net_init fib_proc_init(struct net *net);
485 void __net_exit fib_proc_exit(struct net *net);
486 #else
487 static inline int fib_proc_init(struct net *net)
488 {
489 	return 0;
490 }
491 static inline void fib_proc_exit(struct net *net)
492 {
493 }
494 #endif
495 
496 u32 ip_mtu_from_fib_result(struct fib_result *res, __be32 daddr);
497 
498 int ip_valid_fib_dump_req(struct net *net, const struct nlmsghdr *nlh,
499 			  struct fib_dump_filter *filter,
500 			  struct netlink_callback *cb);
501 
502 int fib_nexthop_info(struct sk_buff *skb, const struct fib_nh_common *nh,
503 		     unsigned int *flags, bool skip_oif);
504 int fib_add_nexthop(struct sk_buff *skb, const struct fib_nh_common *nh,
505 		    int nh_weight);
506 #endif  /* _NET_FIB_H */
507