xref: /freebsd/sys/netinet6/in6_fib.c (revision 0fbdc3726a50a2e5f8530f08da858d3ff175d960)
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_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 #define RNTORT(p)	((struct rtentry *)(p))
78 
79 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
80 
81 CHK_STRUCT_ROUTE_COMPAT(struct route_in6, ro_dst);
82 
83 
84 
85 static void
86 fib6_rte_to_nh_basic(const struct nhop_object *nh, const struct in6_addr *dst,
87     uint32_t flags, struct nhop6_basic *pnh6)
88 {
89 
90 	/* Do explicit nexthop zero unless we're copying it */
91 	memset(pnh6, 0, sizeof(*pnh6));
92 
93 	if ((flags & NHR_IFAIF) != 0)
94 		pnh6->nh_ifp = nh->nh_aifp;
95 	else
96 		pnh6->nh_ifp = nh->nh_ifp;
97 
98 	pnh6->nh_mtu = nh->nh_mtu;
99 	if (nh->nh_flags & NHF_GATEWAY) {
100 		/* Return address with embedded scope. */
101 		pnh6->nh_addr = nh->gw6_sa.sin6_addr;
102 	} else
103 		pnh6->nh_addr = *dst;
104 	/* Set flags */
105 	pnh6->nh_flags = nh->nh_flags;
106 }
107 
108 static void
109 fib6_rte_to_nh_extended(const struct nhop_object *nh, const struct in6_addr *dst,
110     uint32_t flags, struct nhop6_extended *pnh6)
111 {
112 
113 	/* Do explicit nexthop zero unless we're copying it */
114 	memset(pnh6, 0, sizeof(*pnh6));
115 
116 	if ((flags & NHR_IFAIF) != 0)
117 		pnh6->nh_ifp = nh->nh_aifp;
118 	else
119 		pnh6->nh_ifp = nh->nh_ifp;
120 
121 	pnh6->nh_mtu = nh->nh_mtu;
122 	if (nh->nh_flags & NHF_GATEWAY) {
123 		/* Return address with embedded scope. */
124 		pnh6->nh_addr = nh->gw6_sa.sin6_addr;
125 	} else
126 		pnh6->nh_addr = *dst;
127 	/* Set flags */
128 	pnh6->nh_flags = nh->nh_flags;
129 	pnh6->nh_ia = ifatoia6(nh->nh_ifa);
130 }
131 
132 /*
133  * Performs IPv6 route table lookup on @dst. Returns 0 on success.
134  * Stores basic nexthop info into provided @pnh6 structure.
135  * Note that
136  * - nh_ifp represents logical transmit interface (rt_ifp) by default
137  * - nh_ifp represents "address" interface if NHR_IFAIF flag is passed
138  * - mtu from logical transmit interface will be returned.
139  * - nh_ifp cannot be safely dereferenced
140  * - nh_ifp represents rt_ifp (e.g. if looking up address on
141  *   interface "ix0" pointer to "ix0" interface will be returned instead
142  *   of "lo0")
143  * - howewer mtu from "transmit" interface will be returned.
144  * - scope will be embedded in nh_addr
145  */
146 int
147 fib6_lookup_nh_basic(uint32_t fibnum, const struct in6_addr *dst, uint32_t scopeid,
148     uint32_t flags, uint32_t flowid, struct nhop6_basic *pnh6)
149 {
150 	RIB_RLOCK_TRACKER;
151 	struct rib_head *rh;
152 	struct radix_node *rn;
153 	struct sockaddr_in6 sin6;
154 	struct nhop_object *nh;
155 
156 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup_nh_basic: bad fibnum"));
157 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
158 	if (rh == NULL)
159 		return (ENOENT);
160 
161 	/* Prepare lookup key */
162 	memset(&sin6, 0, sizeof(sin6));
163 	sin6.sin6_addr = *dst;
164 	sin6.sin6_len = sizeof(struct sockaddr_in6);
165 	/* Assume scopeid is valid and embed it directly */
166 	if (IN6_IS_SCOPE_LINKLOCAL(dst))
167 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
168 
169 	RIB_RLOCK(rh);
170 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
171 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
172 		nh = RNTORT(rn)->rt_nhop;
173 		/* Ensure route & ifp is UP */
174 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
175 			fib6_rte_to_nh_basic(nh, &sin6.sin6_addr, flags, pnh6);
176 			RIB_RUNLOCK(rh);
177 			return (0);
178 		}
179 	}
180 	RIB_RUNLOCK(rh);
181 
182 	return (ENOENT);
183 }
184 
185 /*
186  * Performs IPv6 route table lookup on @dst. Returns 0 on success.
187  * Stores extended nexthop info into provided @pnh6 structure.
188  * Note that
189  * - nh_ifp cannot be safely dereferenced unless NHR_REF is specified.
190  * - in that case you need to call fib6_free_nh_ext()
191  * - nh_ifp represents logical transmit interface (rt_ifp) by default
192  * - nh_ifp represents "address" interface if NHR_IFAIF flag is passed
193  * - mtu from logical transmit interface will be returned.
194  * - scope will be embedded in nh_addr
195  */
196 int
197 fib6_lookup_nh_ext(uint32_t fibnum, const struct in6_addr *dst,uint32_t scopeid,
198     uint32_t flags, uint32_t flowid, struct nhop6_extended *pnh6)
199 {
200 	RIB_RLOCK_TRACKER;
201 	struct rib_head *rh;
202 	struct radix_node *rn;
203 	struct sockaddr_in6 sin6;
204 	struct rtentry *rte;
205 	struct nhop_object *nh;
206 
207 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup_nh_ext: bad fibnum"));
208 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
209 	if (rh == NULL)
210 		return (ENOENT);
211 
212 	/* Prepare lookup key */
213 	memset(&sin6, 0, sizeof(sin6));
214 	sin6.sin6_len = sizeof(struct sockaddr_in6);
215 	sin6.sin6_addr = *dst;
216 	/* Assume scopeid is valid and embed it directly */
217 	if (IN6_IS_SCOPE_LINKLOCAL(dst))
218 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
219 
220 	RIB_RLOCK(rh);
221 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
222 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
223 		rte = RNTORT(rn);
224 #ifdef RADIX_MPATH
225 		rte = rt_mpath_select(rte, flowid);
226 		if (rte == NULL) {
227 			RIB_RUNLOCK(rh);
228 			return (ENOENT);
229 		}
230 #endif
231 		nh = rte->rt_nhop;
232 		/* Ensure route & ifp is UP */
233 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
234 			fib6_rte_to_nh_extended(nh, &sin6.sin6_addr, flags,
235 			    pnh6);
236 			if ((flags & NHR_REF) != 0) {
237 				/* TODO: Do lwref on egress ifp's */
238 			}
239 			RIB_RUNLOCK(rh);
240 
241 			return (0);
242 		}
243 	}
244 	RIB_RUNLOCK(rh);
245 
246 	return (ENOENT);
247 }
248 
249 void
250 fib6_free_nh_ext(uint32_t fibnum, struct nhop6_extended *pnh6)
251 {
252 
253 }
254 
255 /*
256  * Looks up path in fib @fibnum specified by @dst.
257  * Assumes scope is deembedded and provided in @scopeid.
258  *
259  * Returns path nexthop on success. Nexthop is safe to use
260  *  within the current network epoch. If longer lifetime is required,
261  *  one needs to pass NHR_REF as a flag. This will return referenced
262  *  nexthop.
263  */
264 struct nhop_object *
265 fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6,
266     uint32_t scopeid, uint32_t flags, uint32_t flowid)
267 {
268 	RIB_RLOCK_TRACKER;
269 	struct rib_head *rh;
270 	struct radix_node *rn;
271 	struct rtentry *rt;
272 	struct nhop_object *nh;
273 	struct sockaddr_in6 sin6;
274 
275 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup: bad fibnum"));
276 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
277 	if (rh == NULL)
278 		return (NULL);
279 
280 	/* TODO: radix changes */
281 	//addr = *dst6;
282 	/* Prepare lookup key */
283 	memset(&sin6, 0, sizeof(sin6));
284 	sin6.sin6_len = sizeof(struct sockaddr_in6);
285 	sin6.sin6_addr = *dst6;
286 
287 	/* Assume scopeid is valid and embed it directly */
288 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
289 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
290 
291 	RIB_RLOCK(rh);
292 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
293 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
294 		rt = RNTORT(rn);
295 #ifdef RADIX_MPATH
296 		if (rt_mpath_next(rt) != NULL)
297 			rt = rt_mpath_selectrte(rt, flowid);
298 #endif
299 		nh = rt->rt_nhop;
300 		/* Ensure route & ifp is UP */
301 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
302 			if (flags & NHR_REF)
303 				nhop_ref_object(nh);
304 			RIB_RUNLOCK(rh);
305 			return (nh);
306 		}
307 	}
308 	RIB_RUNLOCK(rh);
309 
310 	RTSTAT_INC(rts_unreach);
311 	return (NULL);
312 }
313 
314 inline static int
315 check_urpf(const struct nhop_object *nh, uint32_t flags,
316     const struct ifnet *src_if)
317 {
318 
319 	if (src_if != NULL && nh->nh_aifp == src_if) {
320 		return (1);
321 	}
322 	if (src_if == NULL) {
323 		if ((flags & NHR_NODEFAULT) == 0)
324 			return (1);
325 		else if ((nh->nh_flags & NHF_DEFAULT) == 0)
326 			return (1);
327 	}
328 
329 	return (0);
330 }
331 
332 #ifdef RADIX_MPATH
333 inline static int
334 check_urpf_mpath(struct rtentry *rt, uint32_t flags,
335     const struct ifnet *src_if)
336 {
337 
338 	while (rt != NULL) {
339 		if (check_urpf(rt->rt_nhop, flags, src_if) != 0)
340 			return (1);
341 		rt = rt_mpath_next(rt);
342 	}
343 
344 	return (0);
345 }
346 #endif
347 
348 /*
349  * Performs reverse path forwarding lookup.
350  * If @src_if is non-zero, verifies that at least 1 path goes via
351  *   this interface.
352  * If @src_if is zero, verifies that route exist.
353  * if @flags contains NHR_NOTDEFAULT, do not consider default route.
354  *
355  * Returns 1 if route matching conditions is found, 0 otherwise.
356  */
357 int
358 fib6_check_urpf(uint32_t fibnum, const struct in6_addr *dst6,
359     uint32_t scopeid, uint32_t flags, const struct ifnet *src_if)
360 {
361 	RIB_RLOCK_TRACKER;
362 	struct rib_head *rh;
363 	struct radix_node *rn;
364 	struct rtentry *rt;
365 	struct sockaddr_in6 sin6;
366 	int ret;
367 
368 	KASSERT((fibnum < rt_numfibs), ("fib6_check_urpf: bad fibnum"));
369 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
370 	if (rh == NULL)
371 		return (0);
372 
373 	/* TODO: radix changes */
374 	/* Prepare lookup key */
375 	memset(&sin6, 0, sizeof(sin6));
376 	sin6.sin6_len = sizeof(struct sockaddr_in6);
377 	sin6.sin6_addr = *dst6;
378 
379 	/* Assume scopeid is valid and embed it directly */
380 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
381 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
382 
383 	RIB_RLOCK(rh);
384 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
385 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
386 		rt = RNTORT(rn);
387 #ifdef	RADIX_MPATH
388 		ret = check_urpf_mpath(rt, flags, src_if);
389 #else
390 		ret = check_urpf(rt->rt_nhop, flags, src_if);
391 #endif
392 		RIB_RUNLOCK(rh);
393 		return (ret);
394 	}
395 	RIB_RUNLOCK(rh);
396 
397 	return (0);
398 }
399 
400 #endif
401 
402