1/*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * Copyright (c) 2018 Andrey V. Elsukov <ae@FreeBSD.org>
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 *	$KAME: in6_gif.c,v 1.49 2001/05/14 14:02:17 itojun Exp $
33 */
34
35#include <sys/cdefs.h>
36#include "opt_inet.h"
37#include "opt_inet6.h"
38
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/jail.h>
42#include <sys/socket.h>
43#include <sys/sockio.h>
44#include <sys/mbuf.h>
45#include <sys/errno.h>
46#include <sys/kernel.h>
47#include <sys/syslog.h>
48#include <sys/sysctl.h>
49#include <sys/malloc.h>
50#include <sys/proc.h>
51
52#include <net/ethernet.h>
53#include <net/if.h>
54#include <net/if_var.h>
55#include <net/if_private.h>
56#include <net/route.h>
57#include <net/vnet.h>
58
59#include <netinet/in.h>
60#include <netinet/in_systm.h>
61#ifdef INET
62#include <netinet/ip.h>
63#include <netinet/ip_ecn.h>
64#endif
65#include <netinet/ip_encap.h>
66#include <netinet/ip6.h>
67#include <netinet6/ip6_var.h>
68#include <netinet6/in6_var.h>
69#include <netinet6/scope6_var.h>
70#include <netinet6/ip6_ecn.h>
71#include <netinet6/in6_fib.h>
72
73#include <net/if_gif.h>
74
75#define GIF_HLIM	30
76VNET_DEFINE_STATIC(int, ip6_gif_hlim) = GIF_HLIM;
77#define	V_ip6_gif_hlim			VNET(ip6_gif_hlim)
78
79SYSCTL_DECL(_net_inet6_ip6);
80SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim,
81    CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_gif_hlim), 0,
82    "Default hop limit for encapsulated packets");
83
84/*
85 * We keep interfaces in a hash table using src+dst as key.
86 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list.
87 */
88VNET_DEFINE_STATIC(struct gif_list *, ipv6_hashtbl) = NULL;
89VNET_DEFINE_STATIC(struct gif_list *, ipv6_srchashtbl) = NULL;
90VNET_DEFINE_STATIC(struct gif_list, ipv6_list) = CK_LIST_HEAD_INITIALIZER();
91#define	V_ipv6_hashtbl		VNET(ipv6_hashtbl)
92#define	V_ipv6_srchashtbl	VNET(ipv6_srchashtbl)
93#define	V_ipv6_list		VNET(ipv6_list)
94
95#define	GIF_HASH(src, dst)	(V_ipv6_hashtbl[\
96    in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)])
97#define	GIF_SRCHASH(src)	(V_ipv6_srchashtbl[\
98    fnv_32_buf((src), sizeof(*src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)])
99#define	GIF_HASH_SC(sc)		GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\
100    &(sc)->gif_ip6hdr->ip6_dst)
101static uint32_t
102in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst)
103{
104	uint32_t ret;
105
106	ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT);
107	return (fnv_32_buf(dst, sizeof(*dst), ret));
108}
109
110static int
111in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src,
112    const struct in6_addr *dst)
113{
114	struct gif_softc *tmp;
115
116	if (sc->gif_family == AF_INET6 &&
117	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) &&
118	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst))
119		return (EEXIST);
120
121	CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) {
122		if (tmp == sc)
123			continue;
124		if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) &&
125		    IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst))
126			return (EADDRNOTAVAIL);
127	}
128	return (0);
129}
130
131/*
132 * Check that ingress address belongs to local host.
133 */
134static void
135in6_gif_set_running(struct gif_softc *sc)
136{
137
138	if (in6_localip(&sc->gif_ip6hdr->ip6_src))
139		GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING;
140	else
141		GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING;
142}
143
144/*
145 * ifaddr_event handler.
146 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent
147 * source address spoofing.
148 */
149static void
150in6_gif_srcaddr(void *arg __unused, const struct sockaddr *sa, int event)
151{
152	const struct sockaddr_in6 *sin;
153	struct gif_softc *sc;
154
155	/* Check that VNET is ready */
156	if (V_ipv6_hashtbl == NULL)
157		return;
158
159	NET_EPOCH_ASSERT();
160	sin = (const struct sockaddr_in6 *)sa;
161	CK_LIST_FOREACH(sc, &GIF_SRCHASH(&sin->sin6_addr), srchash) {
162		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
163		    &sin->sin6_addr) == 0)
164			continue;
165		in6_gif_set_running(sc);
166	}
167}
168
169static void
170in6_gif_attach(struct gif_softc *sc)
171{
172
173	if (sc->gif_options & GIF_IGNORE_SOURCE)
174		CK_LIST_INSERT_HEAD(&V_ipv6_list, sc, chain);
175	else
176		CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain);
177
178	CK_LIST_INSERT_HEAD(&GIF_SRCHASH(&sc->gif_ip6hdr->ip6_src),
179	    sc, srchash);
180}
181
182int
183in6_gif_setopts(struct gif_softc *sc, u_int options)
184{
185
186	/* NOTE: we are protected with gif_ioctl_sx lock */
187	MPASS(sc->gif_family == AF_INET6);
188	MPASS(sc->gif_options != options);
189
190	if ((options & GIF_IGNORE_SOURCE) !=
191	    (sc->gif_options & GIF_IGNORE_SOURCE)) {
192		CK_LIST_REMOVE(sc, srchash);
193		CK_LIST_REMOVE(sc, chain);
194		sc->gif_options = options;
195		in6_gif_attach(sc);
196	}
197	return (0);
198}
199
200int
201in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data)
202{
203	struct in6_ifreq *ifr = (struct in6_ifreq *)data;
204	struct sockaddr_in6 *dst, *src;
205	struct ip6_hdr *ip6;
206	int error;
207
208	/* NOTE: we are protected with gif_ioctl_sx lock */
209	error = EINVAL;
210	switch (cmd) {
211	case SIOCSIFPHYADDR_IN6:
212		src = &((struct in6_aliasreq *)data)->ifra_addr;
213		dst = &((struct in6_aliasreq *)data)->ifra_dstaddr;
214
215		/* sanity checks */
216		if (src->sin6_family != dst->sin6_family ||
217		    src->sin6_family != AF_INET6 ||
218		    src->sin6_len != dst->sin6_len ||
219		    src->sin6_len != sizeof(*src))
220			break;
221		if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) ||
222		    IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) {
223			error = EADDRNOTAVAIL;
224			break;
225		}
226		/*
227		 * Check validity of the scope zone ID of the
228		 * addresses, and convert it into the kernel
229		 * internal form if necessary.
230		 */
231		if ((error = sa6_embedscope(src, 0)) != 0 ||
232		    (error = sa6_embedscope(dst, 0)) != 0)
233			break;
234
235		if (V_ipv6_hashtbl == NULL) {
236			V_ipv6_hashtbl = gif_hashinit();
237			V_ipv6_srchashtbl = gif_hashinit();
238		}
239		error = in6_gif_checkdup(sc, &src->sin6_addr,
240		    &dst->sin6_addr);
241		if (error == EADDRNOTAVAIL)
242			break;
243		if (error == EEXIST) {
244			/* Addresses are the same. Just return. */
245			error = 0;
246			break;
247		}
248		ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO);
249		ip6->ip6_src = src->sin6_addr;
250		ip6->ip6_dst = dst->sin6_addr;
251		ip6->ip6_vfc = IPV6_VERSION;
252		if (sc->gif_family != 0) {
253			/* Detach existing tunnel first */
254			CK_LIST_REMOVE(sc, srchash);
255			CK_LIST_REMOVE(sc, chain);
256			GIF_WAIT();
257			free(sc->gif_hdr, M_GIF);
258			/* XXX: should we notify about link state change? */
259		}
260		sc->gif_family = AF_INET6;
261		sc->gif_ip6hdr = ip6;
262		in6_gif_attach(sc);
263		in6_gif_set_running(sc);
264		break;
265	case SIOCGIFPSRCADDR_IN6:
266	case SIOCGIFPDSTADDR_IN6:
267		if (sc->gif_family != AF_INET6) {
268			error = EADDRNOTAVAIL;
269			break;
270		}
271		src = (struct sockaddr_in6 *)&ifr->ifr_addr;
272		memset(src, 0, sizeof(*src));
273		src->sin6_family = AF_INET6;
274		src->sin6_len = sizeof(*src);
275		src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ?
276		    sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst;
277		error = prison_if(curthread->td_ucred, (struct sockaddr *)src);
278		if (error == 0)
279			error = sa6_recoverscope(src);
280		if (error != 0)
281			memset(src, 0, sizeof(*src));
282		break;
283	}
284	return (error);
285}
286
287int
288in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn)
289{
290	struct gif_softc *sc = ifp->if_softc;
291	struct ip6_hdr *ip6;
292
293	/* prepend new IP header */
294	NET_EPOCH_ASSERT();
295	M_PREPEND(m, sizeof(struct ip6_hdr), M_NOWAIT);
296	if (m == NULL)
297		return (ENOBUFS);
298
299	ip6 = mtod(m, struct ip6_hdr *);
300	MPASS(sc->gif_family == AF_INET6);
301	memcpy(ip6, sc->gif_ip6hdr, sizeof(struct ip6_hdr));
302
303	ip6->ip6_flow  |= htonl((uint32_t)ecn << 20);
304	ip6->ip6_nxt	= proto;
305	ip6->ip6_hlim	= V_ip6_gif_hlim;
306	/*
307	 * force fragmentation to minimum MTU, to avoid path MTU discovery.
308	 * it is too painful to ask for resend of inner packet, to achieve
309	 * path MTU discovery for encapsulated packets.
310	 */
311	return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL));
312}
313
314static int
315in6_gif_input(struct mbuf *m, int off, int proto, void *arg)
316{
317	struct gif_softc *sc = arg;
318	struct ifnet *gifp;
319	struct ip6_hdr *ip6;
320	uint8_t ecn;
321
322	NET_EPOCH_ASSERT();
323	if (sc == NULL) {
324		m_freem(m);
325		IP6STAT_INC(ip6s_nogif);
326		return (IPPROTO_DONE);
327	}
328	gifp = GIF2IFP(sc);
329	if ((gifp->if_flags & IFF_UP) != 0) {
330		ip6 = mtod(m, struct ip6_hdr *);
331		ecn = IPV6_TRAFFIC_CLASS(ip6);
332		m_adj(m, off);
333		gif_input(m, gifp, proto, ecn);
334	} else {
335		m_freem(m);
336		IP6STAT_INC(ip6s_nogif);
337	}
338	return (IPPROTO_DONE);
339}
340
341static int
342in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg)
343{
344	const struct ip6_hdr *ip6;
345	struct gif_softc *sc;
346	int ret;
347
348	if (V_ipv6_hashtbl == NULL)
349		return (0);
350
351	NET_EPOCH_ASSERT();
352	/*
353	 * NOTE: it is safe to iterate without any locking here, because softc
354	 * can be reclaimed only when we are not within net_epoch_preempt
355	 * section, but ip_encap lookup+input are executed in epoch section.
356	 */
357	ip6 = mtod(m, const struct ip6_hdr *);
358	ret = 0;
359	CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) {
360		/*
361		 * This is an inbound packet, its ip6_dst is source address
362		 * in softc.
363		 */
364		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
365		    &ip6->ip6_dst) &&
366		    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst,
367		    &ip6->ip6_src)) {
368			ret = ENCAP_DRV_LOOKUP;
369			goto done;
370		}
371	}
372	/*
373	 * No exact match.
374	 * Check the list of interfaces with GIF_IGNORE_SOURCE flag.
375	 */
376	CK_LIST_FOREACH(sc, &V_ipv6_list, chain) {
377		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
378		    &ip6->ip6_dst)) {
379			ret = 128 + 8; /* src + proto */
380			goto done;
381		}
382	}
383	return (0);
384done:
385	if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0)
386		return (0);
387	/* ingress filters on outer source */
388	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) {
389		if (fib6_check_urpf(sc->gif_fibnum, &ip6->ip6_src,
390		    ntohs(in6_getscope(&ip6->ip6_src)), NHR_NONE,
391		    m->m_pkthdr.rcvif) == 0)
392			return (0);
393	}
394	*arg = sc;
395	return (ret);
396}
397
398static const struct srcaddrtab *ipv6_srcaddrtab;
399static struct {
400	const struct encap_config encap;
401	const struct encaptab *cookie;
402} ipv6_encap_cfg[] = {
403#ifdef INET
404	{
405		.encap = {
406			.proto = IPPROTO_IPV4,
407			.min_length = sizeof(struct ip6_hdr) +
408			    sizeof(struct ip),
409			.exact_match = ENCAP_DRV_LOOKUP,
410			.lookup = in6_gif_lookup,
411			.input = in6_gif_input
412		},
413	},
414#endif
415	{
416		.encap = {
417			.proto = IPPROTO_IPV6,
418			.min_length = 2 * sizeof(struct ip6_hdr),
419			.exact_match = ENCAP_DRV_LOOKUP,
420			.lookup = in6_gif_lookup,
421			.input = in6_gif_input
422		},
423	},
424	{
425		.encap = {
426			.proto = IPPROTO_ETHERIP,
427			.min_length = sizeof(struct ip6_hdr) +
428			    sizeof(struct etherip_header) +
429			    sizeof(struct ether_header),
430			.exact_match = ENCAP_DRV_LOOKUP,
431			.lookup = in6_gif_lookup,
432			.input = in6_gif_input
433		},
434	}
435};
436
437void
438in6_gif_init(void)
439{
440	int i;
441
442	if (!IS_DEFAULT_VNET(curvnet))
443		return;
444
445	ipv6_srcaddrtab = ip6_encap_register_srcaddr(in6_gif_srcaddr,
446	    NULL, M_WAITOK);
447	for (i = 0; i < nitems(ipv6_encap_cfg); i++)
448		ipv6_encap_cfg[i].cookie = ip6_encap_attach(
449		    &ipv6_encap_cfg[i].encap, NULL, M_WAITOK);
450}
451
452void
453in6_gif_uninit(void)
454{
455	int i;
456
457	if (IS_DEFAULT_VNET(curvnet)) {
458		for (i = 0; i < nitems(ipv6_encap_cfg); i++)
459			ip6_encap_detach(ipv6_encap_cfg[i].cookie);
460		ip6_encap_unregister_srcaddr(ipv6_srcaddrtab);
461	}
462	if (V_ipv6_hashtbl != NULL) {
463		gif_hashdestroy(V_ipv6_hashtbl);
464		V_ipv6_hashtbl = NULL;
465		GIF_WAIT();
466		gif_hashdestroy(V_ipv6_srchashtbl);
467	}
468}
469