xref: /freebsd/sys/netinet6/in6_fib.c (revision 62cfcf62f627e5093fb37026a6d8c98e4d2ef04c)
1 /*-
2  * Copyright (c) 2015
3  * 	Alexander V. Chernikov <melifaro@FreeBSD.org>
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 #include <sys/cdefs.h>
31 __FBSDID("$FreeBSD$");
32 
33 #include "opt_inet.h"
34 #include "opt_inet6.h"
35 #include "opt_route.h"
36 #include "opt_mpath.h"
37 
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/lock.h>
41 #include <sys/rmlock.h>
42 #include <sys/malloc.h>
43 #include <sys/mbuf.h>
44 #include <sys/socket.h>
45 #include <sys/sysctl.h>
46 #include <sys/kernel.h>
47 
48 #include <net/if.h>
49 #include <net/if_var.h>
50 #include <net/if_dl.h>
51 #include <net/route.h>
52 #include <net/route/route_var.h>
53 #include <net/route/nhop.h>
54 #include <net/route/shared.h>
55 #include <net/vnet.h>
56 
57 #ifdef RADIX_MPATH
58 #include <net/radix_mpath.h>
59 #endif
60 
61 #include <netinet/in.h>
62 #include <netinet/in_var.h>
63 #include <netinet/ip_mroute.h>
64 #include <netinet/ip6.h>
65 #include <netinet6/in6_fib.h>
66 #include <netinet6/in6_var.h>
67 #include <netinet6/nd6.h>
68 #include <netinet6/scope6_var.h>
69 
70 #include <net/if_types.h>
71 
72 #ifdef INET6
73 static void fib6_rte_to_nh_extended(const struct nhop_object *nh,
74     const struct in6_addr *dst, uint32_t flags, struct nhop6_extended *pnh6);
75 static void fib6_rte_to_nh_basic(const struct nhop_object *nh, const struct in6_addr *dst,
76     uint32_t flags, struct nhop6_basic *pnh6);
77 
78 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
79 
80 CHK_STRUCT_ROUTE_COMPAT(struct route_in6, ro_dst);
81 
82 
83 
84 static void
85 fib6_rte_to_nh_basic(const struct nhop_object *nh, const struct in6_addr *dst,
86     uint32_t flags, struct nhop6_basic *pnh6)
87 {
88 
89 	/* Do explicit nexthop zero unless we're copying it */
90 	memset(pnh6, 0, sizeof(*pnh6));
91 
92 	if ((flags & NHR_IFAIF) != 0)
93 		pnh6->nh_ifp = nh->nh_aifp;
94 	else
95 		pnh6->nh_ifp = nh->nh_ifp;
96 
97 	pnh6->nh_mtu = nh->nh_mtu;
98 	if (nh->nh_flags & NHF_GATEWAY) {
99 		/* Return address with embedded scope. */
100 		pnh6->nh_addr = nh->gw6_sa.sin6_addr;
101 	} else
102 		pnh6->nh_addr = *dst;
103 	/* Set flags */
104 	pnh6->nh_flags = nh->nh_flags;
105 }
106 
107 static void
108 fib6_rte_to_nh_extended(const struct nhop_object *nh, const struct in6_addr *dst,
109     uint32_t flags, struct nhop6_extended *pnh6)
110 {
111 
112 	/* Do explicit nexthop zero unless we're copying it */
113 	memset(pnh6, 0, sizeof(*pnh6));
114 
115 	if ((flags & NHR_IFAIF) != 0)
116 		pnh6->nh_ifp = nh->nh_aifp;
117 	else
118 		pnh6->nh_ifp = nh->nh_ifp;
119 
120 	pnh6->nh_mtu = nh->nh_mtu;
121 	if (nh->nh_flags & NHF_GATEWAY) {
122 		/* Return address with embedded scope. */
123 		pnh6->nh_addr = nh->gw6_sa.sin6_addr;
124 	} else
125 		pnh6->nh_addr = *dst;
126 	/* Set flags */
127 	pnh6->nh_flags = nh->nh_flags;
128 	pnh6->nh_ia = ifatoia6(nh->nh_ifa);
129 }
130 
131 /*
132  * Performs IPv6 route table lookup on @dst. Returns 0 on success.
133  * Stores basic nexthop info into provided @pnh6 structure.
134  * Note that
135  * - nh_ifp represents logical transmit interface (rt_ifp) by default
136  * - nh_ifp represents "address" interface if NHR_IFAIF flag is passed
137  * - mtu from logical transmit interface will be returned.
138  * - nh_ifp cannot be safely dereferenced
139  * - nh_ifp represents rt_ifp (e.g. if looking up address on
140  *   interface "ix0" pointer to "ix0" interface will be returned instead
141  *   of "lo0")
142  * - howewer mtu from "transmit" interface will be returned.
143  * - scope will be embedded in nh_addr
144  */
145 int
146 fib6_lookup_nh_basic(uint32_t fibnum, const struct in6_addr *dst, uint32_t scopeid,
147     uint32_t flags, uint32_t flowid, struct nhop6_basic *pnh6)
148 {
149 	RIB_RLOCK_TRACKER;
150 	struct rib_head *rh;
151 	struct radix_node *rn;
152 	struct sockaddr_in6 sin6;
153 	struct nhop_object *nh;
154 
155 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup_nh_basic: bad fibnum"));
156 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
157 	if (rh == NULL)
158 		return (ENOENT);
159 
160 	/* Prepare lookup key */
161 	memset(&sin6, 0, sizeof(sin6));
162 	sin6.sin6_addr = *dst;
163 	sin6.sin6_len = sizeof(struct sockaddr_in6);
164 	/* Assume scopeid is valid and embed it directly */
165 	if (IN6_IS_SCOPE_LINKLOCAL(dst))
166 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
167 
168 	RIB_RLOCK(rh);
169 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
170 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
171 		nh = RNTORT(rn)->rt_nhop;
172 		/* Ensure route & ifp is UP */
173 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
174 			fib6_rte_to_nh_basic(nh, &sin6.sin6_addr, flags, pnh6);
175 			RIB_RUNLOCK(rh);
176 			return (0);
177 		}
178 	}
179 	RIB_RUNLOCK(rh);
180 
181 	return (ENOENT);
182 }
183 
184 /*
185  * Performs IPv6 route table lookup on @dst. Returns 0 on success.
186  * Stores extended nexthop info into provided @pnh6 structure.
187  * Note that
188  * - nh_ifp cannot be safely dereferenced unless NHR_REF is specified.
189  * - in that case you need to call fib6_free_nh_ext()
190  * - nh_ifp represents logical transmit interface (rt_ifp) by default
191  * - nh_ifp represents "address" interface if NHR_IFAIF flag is passed
192  * - mtu from logical transmit interface will be returned.
193  * - scope will be embedded in nh_addr
194  */
195 int
196 fib6_lookup_nh_ext(uint32_t fibnum, const struct in6_addr *dst,uint32_t scopeid,
197     uint32_t flags, uint32_t flowid, struct nhop6_extended *pnh6)
198 {
199 	RIB_RLOCK_TRACKER;
200 	struct rib_head *rh;
201 	struct radix_node *rn;
202 	struct sockaddr_in6 sin6;
203 	struct rtentry *rte;
204 	struct nhop_object *nh;
205 
206 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup_nh_ext: bad fibnum"));
207 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
208 	if (rh == NULL)
209 		return (ENOENT);
210 
211 	/* Prepare lookup key */
212 	memset(&sin6, 0, sizeof(sin6));
213 	sin6.sin6_len = sizeof(struct sockaddr_in6);
214 	sin6.sin6_addr = *dst;
215 	/* Assume scopeid is valid and embed it directly */
216 	if (IN6_IS_SCOPE_LINKLOCAL(dst))
217 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
218 
219 	RIB_RLOCK(rh);
220 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
221 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
222 		rte = RNTORT(rn);
223 #ifdef RADIX_MPATH
224 		rte = rt_mpath_select(rte, flowid);
225 		if (rte == NULL) {
226 			RIB_RUNLOCK(rh);
227 			return (ENOENT);
228 		}
229 #endif
230 		nh = rte->rt_nhop;
231 		/* Ensure route & ifp is UP */
232 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
233 			fib6_rte_to_nh_extended(nh, &sin6.sin6_addr, flags,
234 			    pnh6);
235 			if ((flags & NHR_REF) != 0) {
236 				/* TODO: Do lwref on egress ifp's */
237 			}
238 			RIB_RUNLOCK(rh);
239 
240 			return (0);
241 		}
242 	}
243 	RIB_RUNLOCK(rh);
244 
245 	return (ENOENT);
246 }
247 
248 void
249 fib6_free_nh_ext(uint32_t fibnum, struct nhop6_extended *pnh6)
250 {
251 
252 }
253 
254 /*
255  * Looks up path in fib @fibnum specified by @dst.
256  * Assumes scope is deembedded and provided in @scopeid.
257  *
258  * Returns path nexthop on success. Nexthop is safe to use
259  *  within the current network epoch. If longer lifetime is required,
260  *  one needs to pass NHR_REF as a flag. This will return referenced
261  *  nexthop.
262  */
263 struct nhop_object *
264 fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6,
265     uint32_t scopeid, uint32_t flags, uint32_t flowid)
266 {
267 	RIB_RLOCK_TRACKER;
268 	struct rib_head *rh;
269 	struct radix_node *rn;
270 	struct rtentry *rt;
271 	struct nhop_object *nh;
272 	struct sockaddr_in6 sin6;
273 
274 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup: bad fibnum"));
275 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
276 	if (rh == NULL)
277 		return (NULL);
278 
279 	/* TODO: radix changes */
280 	//addr = *dst6;
281 	/* Prepare lookup key */
282 	memset(&sin6, 0, sizeof(sin6));
283 	sin6.sin6_len = sizeof(struct sockaddr_in6);
284 	sin6.sin6_addr = *dst6;
285 
286 	/* Assume scopeid is valid and embed it directly */
287 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
288 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
289 
290 	RIB_RLOCK(rh);
291 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
292 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
293 		rt = RNTORT(rn);
294 #ifdef RADIX_MPATH
295 		if (rt_mpath_next(rt) != NULL)
296 			rt = rt_mpath_selectrte(rt, flowid);
297 #endif
298 		nh = rt->rt_nhop;
299 		/* Ensure route & ifp is UP */
300 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
301 			if (flags & NHR_REF)
302 				nhop_ref_object(nh);
303 			RIB_RUNLOCK(rh);
304 			return (nh);
305 		}
306 	}
307 	RIB_RUNLOCK(rh);
308 
309 	RTSTAT_INC(rts_unreach);
310 	return (NULL);
311 }
312 
313 inline static int
314 check_urpf(const struct nhop_object *nh, uint32_t flags,
315     const struct ifnet *src_if)
316 {
317 
318 	if (src_if != NULL && nh->nh_aifp == src_if) {
319 		return (1);
320 	}
321 	if (src_if == NULL) {
322 		if ((flags & NHR_NODEFAULT) == 0)
323 			return (1);
324 		else if ((nh->nh_flags & NHF_DEFAULT) == 0)
325 			return (1);
326 	}
327 
328 	return (0);
329 }
330 
331 #ifdef RADIX_MPATH
332 inline static int
333 check_urpf_mpath(struct rtentry *rt, uint32_t flags,
334     const struct ifnet *src_if)
335 {
336 
337 	while (rt != NULL) {
338 		if (check_urpf(rt->rt_nhop, flags, src_if) != 0)
339 			return (1);
340 		rt = rt_mpath_next(rt);
341 	}
342 
343 	return (0);
344 }
345 #endif
346 
347 /*
348  * Performs reverse path forwarding lookup.
349  * If @src_if is non-zero, verifies that at least 1 path goes via
350  *   this interface.
351  * If @src_if is zero, verifies that route exist.
352  * if @flags contains NHR_NOTDEFAULT, do not consider default route.
353  *
354  * Returns 1 if route matching conditions is found, 0 otherwise.
355  */
356 int
357 fib6_check_urpf(uint32_t fibnum, const struct in6_addr *dst6,
358     uint32_t scopeid, uint32_t flags, const struct ifnet *src_if)
359 {
360 	RIB_RLOCK_TRACKER;
361 	struct rib_head *rh;
362 	struct radix_node *rn;
363 	struct rtentry *rt;
364 	struct sockaddr_in6 sin6;
365 	int ret;
366 
367 	KASSERT((fibnum < rt_numfibs), ("fib6_check_urpf: bad fibnum"));
368 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
369 	if (rh == NULL)
370 		return (0);
371 
372 	/* TODO: radix changes */
373 	/* Prepare lookup key */
374 	memset(&sin6, 0, sizeof(sin6));
375 	sin6.sin6_len = sizeof(struct sockaddr_in6);
376 	sin6.sin6_addr = *dst6;
377 
378 	/* Assume scopeid is valid and embed it directly */
379 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
380 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
381 
382 	RIB_RLOCK(rh);
383 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
384 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
385 		rt = RNTORT(rn);
386 #ifdef	RADIX_MPATH
387 		ret = check_urpf_mpath(rt, flags, src_if);
388 #else
389 		ret = check_urpf(rt->rt_nhop, flags, src_if);
390 #endif
391 		RIB_RUNLOCK(rh);
392 		return (ret);
393 	}
394 	RIB_RUNLOCK(rh);
395 
396 	return (0);
397 }
398 
399 struct nhop_object *
400 fib6_lookup_debugnet(uint32_t fibnum, const struct in6_addr *dst6,
401     uint32_t scopeid, uint32_t flags)
402 {
403 	struct rib_head *rh;
404 	struct radix_node *rn;
405 	struct rtentry *rt;
406 	struct nhop_object *nh;
407 	struct sockaddr_in6 sin6;
408 
409 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup: bad fibnum"));
410 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
411 	if (rh == NULL)
412 		return (NULL);
413 
414 	/* TODO: radix changes */
415 	//addr = *dst6;
416 	/* Prepare lookup key */
417 	memset(&sin6, 0, sizeof(sin6));
418 	sin6.sin6_len = sizeof(struct sockaddr_in6);
419 	sin6.sin6_addr = *dst6;
420 
421 	/* Assume scopeid is valid and embed it directly */
422 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
423 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
424 
425 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
426 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
427 		rt = RNTORT(rn);
428 		nh = rt->rt_nhop;
429 		/* Ensure route & ifp is UP */
430 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
431 			if (flags & NHR_REF)
432 				nhop_ref_object(nh);
433 			return (nh);
434 		}
435 	}
436 
437 	return (NULL);
438 }
439 
440 #endif
441 
442