xref: /freebsd/sys/netinet/in_fib.c (revision 8ddb146abcdf061be9f2c0db7e391697dafad85c)
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_route.h"
35 
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/lock.h>
39 #include <sys/rmlock.h>
40 #include <sys/malloc.h>
41 #include <sys/mbuf.h>
42 #include <sys/socket.h>
43 #include <sys/sysctl.h>
44 #include <sys/kernel.h>
45 
46 #include <net/if.h>
47 #include <net/if_var.h>
48 #include <net/if_dl.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/in_fib.h>
60 
61 #ifdef INET
62 
63 /* Verify struct route compatibility */
64 /* Assert 'struct route_in' is compatible with 'struct route' */
65 CHK_STRUCT_ROUTE_COMPAT(struct route_in, ro_dst4);
66 
67 #ifdef FIB_ALGO
68 VNET_DEFINE(struct fib_dp *, inet_dp);
69 #endif
70 
71 #ifdef ROUTE_MPATH
72 struct _hash_5tuple_ipv4 {
73 	struct in_addr src;
74 	struct in_addr dst;
75 	unsigned short src_port;
76 	unsigned short dst_port;
77 	char proto;
78 	char spare[3];
79 };
80 _Static_assert(sizeof(struct _hash_5tuple_ipv4) == 16,
81     "_hash_5tuple_ipv4 size is wrong");
82 
83 uint32_t
84 fib4_calc_software_hash(struct in_addr src, struct in_addr dst,
85     unsigned short src_port, unsigned short dst_port, char proto,
86     uint32_t *phashtype)
87 {
88 	struct _hash_5tuple_ipv4 data;
89 
90 	data.src = src;
91 	data.dst = dst;
92 	data.src_port = src_port;
93 	data.dst_port = dst_port;
94 	data.proto = proto;
95 	data.spare[0] = data.spare[1] = data.spare[2] = 0;
96 
97 	*phashtype = M_HASHTYPE_OPAQUE;
98 
99 	return (toeplitz_hash(MPATH_ENTROPY_KEY_LEN, mpath_entropy_key,
100 	  sizeof(data), (uint8_t *)&data));
101 }
102 #endif
103 
104 /*
105  * Looks up path in fib @fibnum specified by @dst.
106  * Returns path nexthop on success. Nexthop is safe to use
107  *  within the current network epoch. If longer lifetime is required,
108  *  one needs to pass NHR_REF as a flag. This will return referenced
109  *  nexthop.
110  */
111 #ifdef FIB_ALGO
112 struct nhop_object *
113 fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid,
114     uint32_t flags, uint32_t flowid)
115 {
116 	struct nhop_object *nh;
117 	struct fib_dp *dp = &V_inet_dp[fibnum];
118 	struct flm_lookup_key key = {.addr4 = dst };
119 
120 	nh = dp->f(dp->arg, key, scopeid);
121 	if (nh != NULL) {
122 		nh = nhop_select(nh, flowid);
123 		/* Ensure route & ifp is UP */
124 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
125 			if (flags & NHR_REF)
126 				nhop_ref_object(nh);
127 			return (nh);
128 		}
129 	}
130 	RTSTAT_INC(rts_unreach);
131 	return (NULL);
132 }
133 #else
134 struct nhop_object *
135 fib4_lookup(uint32_t fibnum, struct in_addr dst, uint32_t scopeid,
136     uint32_t flags, uint32_t flowid)
137 {
138 	RIB_RLOCK_TRACKER;
139 	struct rib_head *rh;
140 	struct radix_node *rn;
141 	struct nhop_object *nh;
142 
143 	KASSERT((fibnum < rt_numfibs), ("fib4_lookup: bad fibnum"));
144 	rh = rt_tables_get_rnh(fibnum, AF_INET);
145 	if (rh == NULL)
146 		return (NULL);
147 
148 	/* Prepare lookup key */
149 	struct sockaddr_in sin4 = {
150 		.sin_family = AF_INET,
151 		.sin_len = sizeof(struct sockaddr_in),
152 		.sin_addr = dst,
153 	};
154 
155 	nh = NULL;
156 	RIB_RLOCK(rh);
157 	rn = rh->rnh_matchaddr((void *)&sin4, &rh->head);
158 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
159 		nh = nhop_select((RNTORT(rn))->rt_nhop, flowid);
160 		/* Ensure route & ifp is UP */
161 		if (RT_LINK_IS_UP(nh->nh_ifp)) {
162 			if (flags & NHR_REF)
163 				nhop_ref_object(nh);
164 			RIB_RUNLOCK(rh);
165 			return (nh);
166 		}
167 	}
168 	RIB_RUNLOCK(rh);
169 
170 	RTSTAT_INC(rts_unreach);
171 	return (NULL);
172 }
173 #endif
174 
175 inline static int
176 check_urpf_nhop(const struct nhop_object *nh, uint32_t flags,
177     const struct ifnet *src_if)
178 {
179 
180 	if (src_if != NULL && nh->nh_aifp == src_if) {
181 		return (1);
182 	}
183 	if (src_if == NULL) {
184 		if ((flags & NHR_NODEFAULT) == 0)
185 			return (1);
186 		else if ((nh->nh_flags & NHF_DEFAULT) == 0)
187 			return (1);
188 	}
189 
190 	return (0);
191 }
192 
193 static int
194 check_urpf(struct nhop_object *nh, uint32_t flags,
195     const struct ifnet *src_if)
196 {
197 #ifdef ROUTE_MPATH
198 	if (NH_IS_NHGRP(nh)) {
199 		const struct weightened_nhop *wn;
200 		uint32_t num_nhops;
201 		wn = nhgrp_get_nhops((struct nhgrp_object *)nh, &num_nhops);
202 			for (int i = 0; i < num_nhops; i++) {
203 				if (check_urpf_nhop(wn[i].nh, flags, src_if) != 0)
204 				return (1);
205 		}
206 		return (0);
207 	} else
208 #endif
209 		return (check_urpf_nhop(nh, flags, src_if));
210 }
211 
212 #ifndef FIB_ALGO
213 static struct nhop_object *
214 lookup_nhop(uint32_t fibnum, struct in_addr dst, uint32_t scopeid)
215 {
216 	RIB_RLOCK_TRACKER;
217 	struct rib_head *rh;
218 	struct radix_node *rn;
219 	struct nhop_object *nh;
220 
221 	KASSERT((fibnum < rt_numfibs), ("fib4_check_urpf: bad fibnum"));
222 	rh = rt_tables_get_rnh(fibnum, AF_INET);
223 	if (rh == NULL)
224 		return (NULL);
225 
226 	/* Prepare lookup key */
227 	struct sockaddr_in sin4;
228 	memset(&sin4, 0, sizeof(sin4));
229 	sin4.sin_len = sizeof(struct sockaddr_in);
230 	sin4.sin_addr = dst;
231 
232 	nh = NULL;
233 	RIB_RLOCK(rh);
234 	rn = rh->rnh_matchaddr((void *)&sin4, &rh->head);
235 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0))
236 		nh = RNTORT(rn)->rt_nhop;
237 	RIB_RUNLOCK(rh);
238 
239 	return (nh);
240 }
241 #endif
242 
243 /*
244  * Performs reverse path forwarding lookup.
245  * If @src_if is non-zero, verifies that at least 1 path goes via
246  *   this interface.
247  * If @src_if is zero, verifies that route exist.
248  * if @flags contains NHR_NOTDEFAULT, do not consider default route.
249  *
250  * Returns 1 if route matching conditions is found, 0 otherwise.
251  */
252 int
253 fib4_check_urpf(uint32_t fibnum, struct in_addr dst, uint32_t scopeid,
254   uint32_t flags, const struct ifnet *src_if)
255 {
256 	struct nhop_object *nh;
257 #ifdef FIB_ALGO
258 	struct fib_dp *dp = &V_inet_dp[fibnum];
259 	struct flm_lookup_key key = {.addr4 = dst };
260 
261 	nh = dp->f(dp->arg, key, scopeid);
262 #else
263 	nh = lookup_nhop(fibnum, dst, scopeid);
264 #endif
265 	if (nh != NULL)
266 		return (check_urpf(nh, flags, src_if));
267 
268 	return (0);
269 }
270 
271 /*
272  * Function returning prefix match data along with the nexthop data.
273  * Intended to be used by the control plane code.
274  * Supported flags:
275  *  NHR_UNLOCKED: do not lock radix during lookup.
276  * Returns pointer to rtentry and raw nexthop in @rnd. Both rtentry
277  *  and nexthop are safe to use within current epoch. Note:
278  * Note: rnd_nhop can actually be the nexthop group.
279  */
280 struct rtentry *
281 fib4_lookup_rt(uint32_t fibnum, struct in_addr dst, uint32_t scopeid,
282     uint32_t flags, struct route_nhop_data *rnd)
283 {
284 	RIB_RLOCK_TRACKER;
285 	struct rib_head *rh;
286 	struct radix_node *rn;
287 	struct rtentry *rt;
288 
289 	KASSERT((fibnum < rt_numfibs), ("fib4_lookup_rt: bad fibnum"));
290 	rh = rt_tables_get_rnh(fibnum, AF_INET);
291 	if (rh == NULL)
292 		return (NULL);
293 
294 	/* Prepare lookup key */
295 	struct sockaddr_in sin4 = {
296 		.sin_family = AF_INET,
297 		.sin_len = sizeof(struct sockaddr_in),
298 		.sin_addr = dst,
299 	};
300 
301 	rt = NULL;
302 	if (!(flags & NHR_UNLOCKED))
303 		RIB_RLOCK(rh);
304 	rn = rh->rnh_matchaddr((void *)&sin4, &rh->head);
305 	if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
306 		rt = (struct rtentry *)rn;
307 		rnd->rnd_nhop = rt->rt_nhop;
308 		rnd->rnd_weight = rt->rt_weight;
309 	}
310 	if (!(flags & NHR_UNLOCKED))
311 		RIB_RUNLOCK(rh);
312 
313 	return (rt);
314 }
315 
316 struct nhop_object *
317 fib4_lookup_debugnet(uint32_t fibnum, struct in_addr dst, uint32_t scopeid,
318     uint32_t flags)
319 {
320 	struct rtentry *rt;
321 	struct route_nhop_data rnd;
322 
323 	rt = fib4_lookup_rt(fibnum, dst, scopeid, NHR_UNLOCKED, &rnd);
324 	if (rt != NULL) {
325 		struct nhop_object *nh = nhop_select(rnd.rnd_nhop, 0);
326 		/* Ensure route & ifp is UP */
327 		if (RT_LINK_IS_UP(nh->nh_ifp))
328 			return (nh);
329 	}
330 
331 	return (NULL);
332 }
333 
334 #endif
335