xref: /freebsd/sys/netinet6/in6_fib.c (revision 685dc743)
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 #include "opt_inet.h"
32 #include "opt_inet6.h"
33 #include "opt_route.h"
34 
35 #include <sys/param.h>
36 #include <sys/systm.h>
37 #include <sys/lock.h>
38 #include <sys/rmlock.h>
39 #include <sys/malloc.h>
40 #include <sys/mbuf.h>
41 #include <sys/socket.h>
42 #include <sys/sysctl.h>
43 #include <sys/kernel.h>
44 
45 #include <net/if.h>
46 #include <net/if_var.h>
47 #include <net/if_dl.h>
48 #include <net/if_private.h>
49 #include <net/route.h>
50 #include <net/route/route_ctl.h>
51 #include <net/route/route_var.h>
52 #include <net/route/fib_algo.h>
53 #include <net/route/nhop.h>
54 #include <net/toeplitz.h>
55 #include <net/vnet.h>
56 
57 #include <netinet/in.h>
58 #include <netinet/in_var.h>
59 #include <netinet/ip_mroute.h>
60 #include <netinet/ip6.h>
61 #include <netinet6/in6_fib.h>
62 #include <netinet6/in6_var.h>
63 #include <netinet6/nd6.h>
64 #include <netinet6/scope6_var.h>
65 
66 #include <net/if_types.h>
67 
68 #ifdef INET6
69 
70 CHK_STRUCT_ROUTE_COMPAT(struct route_in6, ro_dst);
71 
72 #ifdef FIB_ALGO
73 VNET_DEFINE(struct fib_dp *, inet6_dp);
74 #endif
75 
76 #ifdef ROUTE_MPATH
77 struct _hash_5tuple_ipv6 {
78 	struct in6_addr src;
79 	struct in6_addr dst;
80 	unsigned short src_port;
81 	unsigned short dst_port;
82 	char proto;
83 	char spare[3];
84 };
85 _Static_assert(sizeof(struct _hash_5tuple_ipv6) == 40,
86     "_hash_5tuple_ipv6 size is wrong");
87 
88 uint32_t
fib6_calc_software_hash(const struct in6_addr * src,const struct in6_addr * dst,unsigned short src_port,unsigned short dst_port,char proto,uint32_t * phashtype)89 fib6_calc_software_hash(const struct in6_addr *src, const struct in6_addr *dst,
90     unsigned short src_port, unsigned short dst_port, char proto,
91     uint32_t *phashtype)
92 {
93 	struct _hash_5tuple_ipv6 data;
94 
95 	data.src = *src;
96 	data.dst = *dst;
97 	data.src_port = src_port;
98 	data.dst_port = dst_port;
99 	data.proto = proto;
100 	data.spare[0] = data.spare[1] = data.spare[2] = 0;
101 
102 	*phashtype = M_HASHTYPE_OPAQUE_HASH;
103 
104 	return (toeplitz_hash(MPATH_ENTROPY_KEY_LEN, mpath_entropy_key,
105 	  sizeof(data), (uint8_t *)&data));
106 }
107 #endif
108 
109 /*
110  * Looks up path in fib @fibnum specified by @dst.
111  * Assumes scope is deembedded and provided in @scopeid.
112  *
113  * Returns path nexthop on success. Nexthop is safe to use
114  *  within the current network epoch. If longer lifetime is required,
115  *  one needs to pass NHR_REF as a flag. This will return referenced
116  *  nexthop.
117  */
118 #ifdef FIB_ALGO
119 struct nhop_object *
fib6_lookup(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid,uint32_t flags,uint32_t flowid)120 fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6,
121     uint32_t scopeid, uint32_t flags, uint32_t flowid)
122 {
123 	struct nhop_object *nh;
124 	struct fib_dp *dp = &V_inet6_dp[fibnum];
125 	struct flm_lookup_key key = {.addr6 = dst6 };
126 
127 	nh = dp->f(dp->arg, key, scopeid);
128 	if (nh != NULL) {
129 		nh = nhop_select(nh, flowid);
130 		/* Ensure route & ifp is UP */
131 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
132 			if (flags & NHR_REF)
133 				nhop_ref_object(nh);
134 			return (nh);
135 		}
136 	}
137 	RTSTAT_INC(rts_unreach);
138 	return (NULL);
139 }
140 #else
141 struct nhop_object *
fib6_lookup(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid,uint32_t flags,uint32_t flowid)142 fib6_lookup(uint32_t fibnum, const struct in6_addr *dst6,
143     uint32_t scopeid, uint32_t flags, uint32_t flowid)
144 {
145 	RIB_RLOCK_TRACKER;
146 	struct rib_head *rh;
147 	struct radix_node *rn;
148 	struct nhop_object *nh;
149 
150 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup: bad fibnum"));
151 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
152 	if (rh == NULL)
153 		return (NULL);
154 
155 	struct sockaddr_in6 sin6 = {
156 		.sin6_len = sizeof(struct sockaddr_in6),
157 		.sin6_addr = *dst6,
158 	};
159 
160 	/* Assume scopeid is valid and embed it directly */
161 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
162 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
163 
164 	RIB_RLOCK(rh);
165 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
166 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
167 		nh = nhop_select((RNTORT(rn))->rt_nhop, flowid);
168 		/* Ensure route & ifp is UP */
169 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
170 			if (flags & NHR_REF)
171 				nhop_ref_object(nh);
172 			RIB_RUNLOCK(rh);
173 			return (nh);
174 		}
175 	}
176 	RIB_RUNLOCK(rh);
177 
178 	RTSTAT_INC(rts_unreach);
179 	return (NULL);
180 }
181 #endif
182 
183 inline static int
check_urpf_nhop(const struct nhop_object * nh,uint32_t flags,const struct ifnet * src_if)184 check_urpf_nhop(const struct nhop_object *nh, uint32_t flags,
185     const struct ifnet *src_if)
186 {
187 
188 	if (src_if != NULL && nh->nh_aifp == src_if) {
189 		return (1);
190 	}
191 	if (src_if == NULL) {
192 		if ((flags & NHR_NODEFAULT) == 0)
193 			return (1);
194 		else if ((nh->nh_flags & NHF_DEFAULT) == 0)
195 			return (1);
196 	}
197 
198 	return (0);
199 }
200 
201 static int
check_urpf(struct nhop_object * nh,uint32_t flags,const struct ifnet * src_if)202 check_urpf(struct nhop_object *nh, uint32_t flags,
203     const struct ifnet *src_if)
204 {
205 #ifdef ROUTE_MPATH
206 	if (NH_IS_NHGRP(nh)) {
207 		const struct weightened_nhop *wn;
208 		uint32_t num_nhops;
209 		wn = nhgrp_get_nhops((struct nhgrp_object *)nh, &num_nhops);
210 		for (int i = 0; i < num_nhops; i++) {
211 			if (check_urpf_nhop(wn[i].nh, flags, src_if) != 0)
212 				return (1);
213 		}
214 		return (0);
215 	} else
216 #endif
217 		return (check_urpf_nhop(nh, flags, src_if));
218 }
219 
220 #ifndef FIB_ALGO
221 static struct nhop_object *
lookup_nhop(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid)222 lookup_nhop(uint32_t fibnum, const struct in6_addr *dst6,
223     uint32_t scopeid)
224 {
225 	RIB_RLOCK_TRACKER;
226 	struct rib_head *rh;
227 	struct radix_node *rn;
228 	struct nhop_object *nh;
229 
230 	KASSERT((fibnum < rt_numfibs), ("fib6_check_urpf: bad fibnum"));
231 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
232 	if (rh == NULL)
233 		return (NULL);
234 
235 	/* Prepare lookup key */
236 	struct sockaddr_in6 sin6 = {
237 		.sin6_len = sizeof(struct sockaddr_in6),
238 		.sin6_addr = *dst6,
239 	};
240 
241 	/* Assume scopeid is valid and embed it directly */
242 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
243 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
244 
245 	nh = NULL;
246 	RIB_RLOCK(rh);
247 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
248 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0))
249 		nh = RNTORT(rn)->rt_nhop;
250 	RIB_RUNLOCK(rh);
251 
252 	return (nh);
253 }
254 #endif
255 
256 /*
257  * Performs reverse path forwarding lookup.
258  * If @src_if is non-zero, verifies that at least 1 path goes via
259  *   this interface.
260  * If @src_if is zero, verifies that route exist.
261  * if @flags contains NHR_NOTDEFAULT, do not consider default route.
262  *
263  * Returns 1 if route matching conditions is found, 0 otherwise.
264  */
265 int
fib6_check_urpf(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid,uint32_t flags,const struct ifnet * src_if)266 fib6_check_urpf(uint32_t fibnum, const struct in6_addr *dst6,
267     uint32_t scopeid, uint32_t flags, const struct ifnet *src_if)
268 {
269 	struct nhop_object *nh;
270 #ifdef FIB_ALGO
271 	struct fib_dp *dp = &V_inet6_dp[fibnum];
272 	struct flm_lookup_key key = {.addr6 = dst6 };
273 
274 	nh = dp->f(dp->arg, key, scopeid);
275 #else
276 	nh = lookup_nhop(fibnum, dst6, scopeid);
277 #endif
278 	if (nh != NULL)
279 		return (check_urpf(nh, flags, src_if));
280 	return (0);
281 }
282 
283 /*
284  * Function returning prefix match data along with the nexthop data.
285  * Intended to be used by the control plane code.
286  * Supported flags:
287  *  NHR_UNLOCKED: do not lock radix during lookup.
288  * Returns pointer to rtentry and raw nexthop in @rnd. Both rtentry
289  *  and nexthop are safe to use within current epoch. Note:
290  * Note: rnd_nhop can actually be the nexthop group.
291  */
292 struct rtentry *
fib6_lookup_rt(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid,uint32_t flags,struct route_nhop_data * rnd)293 fib6_lookup_rt(uint32_t fibnum, const struct in6_addr *dst6,
294     uint32_t scopeid, uint32_t flags, struct route_nhop_data *rnd)
295 {
296 	RIB_RLOCK_TRACKER;
297 	struct rib_head *rh;
298 	struct radix_node *rn;
299 	struct rtentry *rt;
300 
301 	KASSERT((fibnum < rt_numfibs), ("fib6_lookup: bad fibnum"));
302 	rh = rt_tables_get_rnh(fibnum, AF_INET6);
303 	if (rh == NULL)
304 		return (NULL);
305 
306 	struct sockaddr_in6 sin6 = {
307 		.sin6_len = sizeof(struct sockaddr_in6),
308 		.sin6_addr = *dst6,
309 	};
310 
311 	/* Assume scopeid is valid and embed it directly */
312 	if (IN6_IS_SCOPE_LINKLOCAL(dst6))
313 		sin6.sin6_addr.s6_addr16[1] = htons(scopeid & 0xffff);
314 
315 	rt = NULL;
316 	if (!(flags & NHR_UNLOCKED))
317 		RIB_RLOCK(rh);
318 	rn = rh->rnh_matchaddr((void *)&sin6, &rh->head);
319 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
320 		rt = (struct rtentry *)rn;
321 		rnd->rnd_nhop = rt->rt_nhop;
322 		rnd->rnd_weight = rt->rt_weight;
323 	}
324 	if (!(flags & NHR_UNLOCKED))
325 		RIB_RUNLOCK(rh);
326 
327 	return (rt);
328 }
329 
330 struct nhop_object *
fib6_lookup_debugnet(uint32_t fibnum,const struct in6_addr * dst6,uint32_t scopeid,uint32_t flags)331 fib6_lookup_debugnet(uint32_t fibnum, const struct in6_addr *dst6,
332     uint32_t scopeid, uint32_t flags)
333 {
334 	struct rtentry *rt;
335 	struct route_nhop_data rnd;
336 
337 	rt = fib6_lookup_rt(fibnum, dst6, scopeid, NHR_UNLOCKED, &rnd);
338 	if (rt != NULL) {
339 		struct nhop_object *nh = nhop_select(rnd.rnd_nhop, 0);
340 		/* Ensure route & ifp is UP */
341 		if (RT_LINK_IS_UP(nh->nh_ifp))
342 			return (nh);
343 	}
344 
345 	return (NULL);
346 }
347 
348 #endif
349