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__FBSDID("$FreeBSD$");
37
38#include "opt_inet.h"
39#include "opt_inet6.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/jail.h>
44#include <sys/socket.h>
45#include <sys/sockio.h>
46#include <sys/mbuf.h>
47#include <sys/errno.h>
48#include <sys/kernel.h>
49#include <sys/syslog.h>
50#include <sys/sysctl.h>
51#include <sys/malloc.h>
52#include <sys/proc.h>
53
54#include <net/ethernet.h>
55#include <net/if.h>
56#include <net/if_var.h>
57#include <net/route.h>
58#include <net/vnet.h>
59
60#include <netinet/in.h>
61#include <netinet/in_systm.h>
62#ifdef INET
63#include <netinet/ip.h>
64#include <netinet/ip_ecn.h>
65#endif
66#include <netinet/ip_encap.h>
67#include <netinet/ip6.h>
68#include <netinet6/ip6_var.h>
69#include <netinet6/in6_var.h>
70#include <netinet6/scope6_var.h>
71#include <netinet6/ip6_ecn.h>
72#include <netinet6/in6_fib.h>
73
74#include <net/if_gif.h>
75
76#define GIF_HLIM	30
77VNET_DEFINE_STATIC(int, ip6_gif_hlim) = GIF_HLIM;
78#define	V_ip6_gif_hlim			VNET(ip6_gif_hlim)
79
80SYSCTL_DECL(_net_inet6_ip6);
81SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim,
82    CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_gif_hlim), 0,
83    "Default hop limit for encapsulated packets");
84
85/*
86 * We keep interfaces in a hash table using src+dst as key.
87 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list.
88 */
89VNET_DEFINE_STATIC(struct gif_list *, ipv6_hashtbl) = NULL;
90VNET_DEFINE_STATIC(struct gif_list *, ipv6_srchashtbl) = NULL;
91VNET_DEFINE_STATIC(struct gif_list, ipv6_list) = CK_LIST_HEAD_INITIALIZER();
92#define	V_ipv6_hashtbl		VNET(ipv6_hashtbl)
93#define	V_ipv6_srchashtbl	VNET(ipv6_srchashtbl)
94#define	V_ipv6_list		VNET(ipv6_list)
95
96#define	GIF_HASH(src, dst)	(V_ipv6_hashtbl[\
97    in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)])
98#define	GIF_SRCHASH(src)	(V_ipv6_srchashtbl[\
99    fnv_32_buf((src), sizeof(*src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)])
100#define	GIF_HASH_SC(sc)		GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\
101    &(sc)->gif_ip6hdr->ip6_dst)
102static uint32_t
103in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst)
104{
105	uint32_t ret;
106
107	ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT);
108	return (fnv_32_buf(dst, sizeof(*dst), ret));
109}
110
111static int
112in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src,
113    const struct in6_addr *dst)
114{
115	struct gif_softc *tmp;
116
117	if (sc->gif_family == AF_INET6 &&
118	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) &&
119	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst))
120		return (EEXIST);
121
122	CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) {
123		if (tmp == sc)
124			continue;
125		if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) &&
126		    IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst))
127			return (EADDRNOTAVAIL);
128	}
129	return (0);
130}
131
132/*
133 * Check that ingress address belongs to local host.
134 */
135static void
136in6_gif_set_running(struct gif_softc *sc)
137{
138
139	if (in6_localip(&sc->gif_ip6hdr->ip6_src))
140		GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING;
141	else
142		GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING;
143}
144
145/*
146 * ifaddr_event handler.
147 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent
148 * source address spoofing.
149 */
150static void
151in6_gif_srcaddr(void *arg __unused, const struct sockaddr *sa, int event)
152{
153	const struct sockaddr_in6 *sin;
154	struct gif_softc *sc;
155
156	/* Check that VNET is ready */
157	if (V_ipv6_hashtbl == NULL)
158		return;
159
160	NET_EPOCH_ASSERT();
161	sin = (const struct sockaddr_in6 *)sa;
162	CK_LIST_FOREACH(sc, &GIF_SRCHASH(&sin->sin6_addr), srchash) {
163		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
164		    &sin->sin6_addr) == 0)
165			continue;
166		in6_gif_set_running(sc);
167	}
168}
169
170static void
171in6_gif_attach(struct gif_softc *sc)
172{
173
174	if (sc->gif_options & GIF_IGNORE_SOURCE)
175		CK_LIST_INSERT_HEAD(&V_ipv6_list, sc, chain);
176	else
177		CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain);
178
179	CK_LIST_INSERT_HEAD(&GIF_SRCHASH(&sc->gif_ip6hdr->ip6_src),
180	    sc, srchash);
181}
182
183int
184in6_gif_setopts(struct gif_softc *sc, u_int options)
185{
186
187	/* NOTE: we are protected with gif_ioctl_sx lock */
188	MPASS(sc->gif_family == AF_INET6);
189	MPASS(sc->gif_options != options);
190
191	if ((options & GIF_IGNORE_SOURCE) !=
192	    (sc->gif_options & GIF_IGNORE_SOURCE)) {
193		CK_LIST_REMOVE(sc, srchash);
194		CK_LIST_REMOVE(sc, chain);
195		sc->gif_options = options;
196		in6_gif_attach(sc);
197	}
198	return (0);
199}
200
201int
202in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data)
203{
204	struct in6_ifreq *ifr = (struct in6_ifreq *)data;
205	struct sockaddr_in6 *dst, *src;
206	struct ip6_hdr *ip6;
207	int error;
208
209	/* NOTE: we are protected with gif_ioctl_sx lock */
210	error = EINVAL;
211	switch (cmd) {
212	case SIOCSIFPHYADDR_IN6:
213		src = &((struct in6_aliasreq *)data)->ifra_addr;
214		dst = &((struct in6_aliasreq *)data)->ifra_dstaddr;
215
216		/* sanity checks */
217		if (src->sin6_family != dst->sin6_family ||
218		    src->sin6_family != AF_INET6 ||
219		    src->sin6_len != dst->sin6_len ||
220		    src->sin6_len != sizeof(*src))
221			break;
222		if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) ||
223		    IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) {
224			error = EADDRNOTAVAIL;
225			break;
226		}
227		/*
228		 * Check validity of the scope zone ID of the
229		 * addresses, and convert it into the kernel
230		 * internal form if necessary.
231		 */
232		if ((error = sa6_embedscope(src, 0)) != 0 ||
233		    (error = sa6_embedscope(dst, 0)) != 0)
234			break;
235
236		if (V_ipv6_hashtbl == NULL) {
237			V_ipv6_hashtbl = gif_hashinit();
238			V_ipv6_srchashtbl = gif_hashinit();
239		}
240		error = in6_gif_checkdup(sc, &src->sin6_addr,
241		    &dst->sin6_addr);
242		if (error == EADDRNOTAVAIL)
243			break;
244		if (error == EEXIST) {
245			/* Addresses are the same. Just return. */
246			error = 0;
247			break;
248		}
249		ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO);
250		ip6->ip6_src = src->sin6_addr;
251		ip6->ip6_dst = dst->sin6_addr;
252		ip6->ip6_vfc = IPV6_VERSION;
253		if (sc->gif_family != 0) {
254			/* Detach existing tunnel first */
255			CK_LIST_REMOVE(sc, srchash);
256			CK_LIST_REMOVE(sc, chain);
257			GIF_WAIT();
258			free(sc->gif_hdr, M_GIF);
259			/* XXX: should we notify about link state change? */
260		}
261		sc->gif_family = AF_INET6;
262		sc->gif_ip6hdr = ip6;
263		in6_gif_attach(sc);
264		in6_gif_set_running(sc);
265		break;
266	case SIOCGIFPSRCADDR_IN6:
267	case SIOCGIFPDSTADDR_IN6:
268		if (sc->gif_family != AF_INET6) {
269			error = EADDRNOTAVAIL;
270			break;
271		}
272		src = (struct sockaddr_in6 *)&ifr->ifr_addr;
273		memset(src, 0, sizeof(*src));
274		src->sin6_family = AF_INET6;
275		src->sin6_len = sizeof(*src);
276		src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ?
277		    sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst;
278		error = prison_if(curthread->td_ucred, (struct sockaddr *)src);
279		if (error == 0)
280			error = sa6_recoverscope(src);
281		if (error != 0)
282			memset(src, 0, sizeof(*src));
283		break;
284	}
285	return (error);
286}
287
288int
289in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn)
290{
291	struct gif_softc *sc = ifp->if_softc;
292	struct ip6_hdr *ip6;
293	int len;
294
295	/* prepend new IP header */
296	NET_EPOCH_ASSERT();
297	len = sizeof(struct ip6_hdr);
298#ifndef __NO_STRICT_ALIGNMENT
299	if (proto == IPPROTO_ETHERIP)
300		len += ETHERIP_ALIGN;
301#endif
302	M_PREPEND(m, len, M_NOWAIT);
303	if (m == NULL)
304		return (ENOBUFS);
305#ifndef __NO_STRICT_ALIGNMENT
306	if (proto == IPPROTO_ETHERIP) {
307		len = mtod(m, vm_offset_t) & 3;
308		KASSERT(len == 0 || len == ETHERIP_ALIGN,
309		    ("in6_gif_output: unexpected misalignment"));
310		m->m_data += len;
311		m->m_len -= ETHERIP_ALIGN;
312	}
313#endif
314
315	ip6 = mtod(m, struct ip6_hdr *);
316	MPASS(sc->gif_family == AF_INET6);
317	bcopy(sc->gif_ip6hdr, ip6, sizeof(struct ip6_hdr));
318
319	ip6->ip6_flow  |= htonl((uint32_t)ecn << 20);
320	ip6->ip6_nxt	= proto;
321	ip6->ip6_hlim	= V_ip6_gif_hlim;
322	/*
323	 * force fragmentation to minimum MTU, to avoid path MTU discovery.
324	 * it is too painful to ask for resend of inner packet, to achieve
325	 * path MTU discovery for encapsulated packets.
326	 */
327	return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL));
328}
329
330static int
331in6_gif_input(struct mbuf *m, int off, int proto, void *arg)
332{
333	struct gif_softc *sc = arg;
334	struct ifnet *gifp;
335	struct ip6_hdr *ip6;
336	uint8_t ecn;
337
338	NET_EPOCH_ASSERT();
339	if (sc == NULL) {
340		m_freem(m);
341		IP6STAT_INC(ip6s_nogif);
342		return (IPPROTO_DONE);
343	}
344	gifp = GIF2IFP(sc);
345	if ((gifp->if_flags & IFF_UP) != 0) {
346		ip6 = mtod(m, struct ip6_hdr *);
347		ecn = IPV6_TRAFFIC_CLASS(ip6);
348		m_adj(m, off);
349		gif_input(m, gifp, proto, ecn);
350	} else {
351		m_freem(m);
352		IP6STAT_INC(ip6s_nogif);
353	}
354	return (IPPROTO_DONE);
355}
356
357static int
358in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg)
359{
360	const struct ip6_hdr *ip6;
361	struct gif_softc *sc;
362	int ret;
363
364	if (V_ipv6_hashtbl == NULL)
365		return (0);
366
367	NET_EPOCH_ASSERT();
368	/*
369	 * NOTE: it is safe to iterate without any locking here, because softc
370	 * can be reclaimed only when we are not within net_epoch_preempt
371	 * section, but ip_encap lookup+input are executed in epoch section.
372	 */
373	ip6 = mtod(m, const struct ip6_hdr *);
374	ret = 0;
375	CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) {
376		/*
377		 * This is an inbound packet, its ip6_dst is source address
378		 * in softc.
379		 */
380		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
381		    &ip6->ip6_dst) &&
382		    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst,
383		    &ip6->ip6_src)) {
384			ret = ENCAP_DRV_LOOKUP;
385			goto done;
386		}
387	}
388	/*
389	 * No exact match.
390	 * Check the list of interfaces with GIF_IGNORE_SOURCE flag.
391	 */
392	CK_LIST_FOREACH(sc, &V_ipv6_list, chain) {
393		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
394		    &ip6->ip6_dst)) {
395			ret = 128 + 8; /* src + proto */
396			goto done;
397		}
398	}
399	return (0);
400done:
401	if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0)
402		return (0);
403	/* ingress filters on outer source */
404	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) {
405		if (fib6_check_urpf(sc->gif_fibnum, &ip6->ip6_src,
406		    ntohs(in6_getscope(&ip6->ip6_src)), NHR_NONE,
407		    m->m_pkthdr.rcvif) == 0)
408			return (0);
409	}
410	*arg = sc;
411	return (ret);
412}
413
414static const struct srcaddrtab *ipv6_srcaddrtab;
415static struct {
416	const struct encap_config encap;
417	const struct encaptab *cookie;
418} ipv6_encap_cfg[] = {
419#ifdef INET
420	{
421		.encap = {
422			.proto = IPPROTO_IPV4,
423			.min_length = sizeof(struct ip6_hdr) +
424			    sizeof(struct ip),
425			.exact_match = ENCAP_DRV_LOOKUP,
426			.lookup = in6_gif_lookup,
427			.input = in6_gif_input
428		},
429	},
430#endif
431	{
432		.encap = {
433			.proto = IPPROTO_IPV6,
434			.min_length = 2 * sizeof(struct ip6_hdr),
435			.exact_match = ENCAP_DRV_LOOKUP,
436			.lookup = in6_gif_lookup,
437			.input = in6_gif_input
438		},
439	},
440	{
441		.encap = {
442			.proto = IPPROTO_ETHERIP,
443			.min_length = sizeof(struct ip6_hdr) +
444			    sizeof(struct etherip_header) +
445			    sizeof(struct ether_header),
446			.exact_match = ENCAP_DRV_LOOKUP,
447			.lookup = in6_gif_lookup,
448			.input = in6_gif_input
449		},
450	}
451};
452
453void
454in6_gif_init(void)
455{
456	int i;
457
458	if (!IS_DEFAULT_VNET(curvnet))
459		return;
460
461	ipv6_srcaddrtab = ip6_encap_register_srcaddr(in6_gif_srcaddr,
462	    NULL, M_WAITOK);
463	for (i = 0; i < nitems(ipv6_encap_cfg); i++)
464		ipv6_encap_cfg[i].cookie = ip6_encap_attach(
465		    &ipv6_encap_cfg[i].encap, NULL, M_WAITOK);
466}
467
468void
469in6_gif_uninit(void)
470{
471	int i;
472
473	if (IS_DEFAULT_VNET(curvnet)) {
474		for (i = 0; i < nitems(ipv6_encap_cfg); i++)
475			ip6_encap_detach(ipv6_encap_cfg[i].cookie);
476		ip6_encap_unregister_srcaddr(ipv6_srcaddrtab);
477	}
478	if (V_ipv6_hashtbl != NULL) {
479		gif_hashdestroy(V_ipv6_hashtbl);
480		V_ipv6_hashtbl = NULL;
481		GIF_WAIT();
482		gif_hashdestroy(V_ipv6_srchashtbl);
483	}
484}
485