in6_gif.c revision a6605d2938cce3ab84122eb8f7e12469b7d3356f
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_list) = CK_LIST_HEAD_INITIALIZER();
91#define	V_ipv6_hashtbl		VNET(ipv6_hashtbl)
92#define	V_ipv6_list		VNET(ipv6_list)
93
94#define	GIF_HASH(src, dst)	(V_ipv6_hashtbl[\
95    in6_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)])
96#define	GIF_HASH_SC(sc)		GIF_HASH(&(sc)->gif_ip6hdr->ip6_src,\
97    &(sc)->gif_ip6hdr->ip6_dst)
98static uint32_t
99in6_gif_hashval(const struct in6_addr *src, const struct in6_addr *dst)
100{
101	uint32_t ret;
102
103	ret = fnv_32_buf(src, sizeof(*src), FNV1_32_INIT);
104	return (fnv_32_buf(dst, sizeof(*dst), ret));
105}
106
107static int
108in6_gif_checkdup(const struct gif_softc *sc, const struct in6_addr *src,
109    const struct in6_addr *dst)
110{
111	struct gif_softc *tmp;
112
113	if (sc->gif_family == AF_INET6 &&
114	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src, src) &&
115	    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst, dst))
116		return (EEXIST);
117
118	CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) {
119		if (tmp == sc)
120			continue;
121		if (IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_src, src) &&
122		    IN6_ARE_ADDR_EQUAL(&tmp->gif_ip6hdr->ip6_dst, dst))
123			return (EADDRNOTAVAIL);
124	}
125	return (0);
126}
127
128static void
129in6_gif_attach(struct gif_softc *sc)
130{
131
132	if (sc->gif_options & GIF_IGNORE_SOURCE)
133		CK_LIST_INSERT_HEAD(&V_ipv6_list, sc, chain);
134	else
135		CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain);
136}
137
138int
139in6_gif_setopts(struct gif_softc *sc, u_int options)
140{
141
142	/* NOTE: we are protected with gif_ioctl_sx lock */
143	MPASS(sc->gif_family == AF_INET6);
144	MPASS(sc->gif_options != options);
145
146	if ((options & GIF_IGNORE_SOURCE) !=
147	    (sc->gif_options & GIF_IGNORE_SOURCE)) {
148		CK_LIST_REMOVE(sc, chain);
149		sc->gif_options = options;
150		in6_gif_attach(sc);
151	}
152	return (0);
153}
154
155int
156in6_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data)
157{
158	struct in6_ifreq *ifr = (struct in6_ifreq *)data;
159	struct sockaddr_in6 *dst, *src;
160	struct ip6_hdr *ip6;
161	int error;
162
163	/* NOTE: we are protected with gif_ioctl_sx lock */
164	error = EINVAL;
165	switch (cmd) {
166	case SIOCSIFPHYADDR_IN6:
167		src = &((struct in6_aliasreq *)data)->ifra_addr;
168		dst = &((struct in6_aliasreq *)data)->ifra_dstaddr;
169
170		/* sanity checks */
171		if (src->sin6_family != dst->sin6_family ||
172		    src->sin6_family != AF_INET6 ||
173		    src->sin6_len != dst->sin6_len ||
174		    src->sin6_len != sizeof(*src))
175			break;
176		if (IN6_IS_ADDR_UNSPECIFIED(&src->sin6_addr) ||
177		    IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr)) {
178			error = EADDRNOTAVAIL;
179			break;
180		}
181		/*
182		 * Check validity of the scope zone ID of the
183		 * addresses, and convert it into the kernel
184		 * internal form if necessary.
185		 */
186		if ((error = sa6_embedscope(src, 0)) != 0 ||
187		    (error = sa6_embedscope(dst, 0)) != 0)
188			break;
189
190		if (V_ipv6_hashtbl == NULL)
191			V_ipv6_hashtbl = gif_hashinit();
192		error = in6_gif_checkdup(sc, &src->sin6_addr,
193		    &dst->sin6_addr);
194		if (error == EADDRNOTAVAIL)
195			break;
196		if (error == EEXIST) {
197			/* Addresses are the same. Just return. */
198			error = 0;
199			break;
200		}
201		ip6 = malloc(sizeof(*ip6), M_GIF, M_WAITOK | M_ZERO);
202		ip6->ip6_src = src->sin6_addr;
203		ip6->ip6_dst = dst->sin6_addr;
204		if (sc->gif_family != 0) {
205			/* Detach existing tunnel first */
206			CK_LIST_REMOVE(sc, chain);
207			GIF_WAIT();
208			free(sc->gif_hdr, M_GIF);
209			/* XXX: should we notify about link state change? */
210		}
211		sc->gif_family = AF_INET6;
212		sc->gif_ip6hdr = ip6;
213		in6_gif_attach(sc);
214		break;
215	case SIOCGIFPSRCADDR_IN6:
216	case SIOCGIFPDSTADDR_IN6:
217		if (sc->gif_family != AF_INET6) {
218			error = EADDRNOTAVAIL;
219			break;
220		}
221		src = (struct sockaddr_in6 *)&ifr->ifr_addr;
222		memset(src, 0, sizeof(*src));
223		src->sin6_family = AF_INET6;
224		src->sin6_len = sizeof(*src);
225		src->sin6_addr = (cmd == SIOCGIFPSRCADDR_IN6) ?
226		    sc->gif_ip6hdr->ip6_src: sc->gif_ip6hdr->ip6_dst;
227		error = prison_if(curthread->td_ucred, (struct sockaddr *)src);
228		if (error == 0)
229			error = sa6_recoverscope(src);
230		if (error != 0)
231			memset(src, 0, sizeof(*src));
232		break;
233	}
234	return (error);
235}
236
237int
238in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn)
239{
240	struct gif_softc *sc = ifp->if_softc;
241	struct ip6_hdr *ip6;
242	int len;
243
244	/* prepend new IP header */
245	MPASS(in_epoch(net_epoch_preempt));
246	len = sizeof(struct ip6_hdr);
247#ifndef __NO_STRICT_ALIGNMENT
248	if (proto == IPPROTO_ETHERIP)
249		len += ETHERIP_ALIGN;
250#endif
251	M_PREPEND(m, len, M_NOWAIT);
252	if (m == NULL)
253		return (ENOBUFS);
254#ifndef __NO_STRICT_ALIGNMENT
255	if (proto == IPPROTO_ETHERIP) {
256		len = mtod(m, vm_offset_t) & 3;
257		KASSERT(len == 0 || len == ETHERIP_ALIGN,
258		    ("in6_gif_output: unexpected misalignment"));
259		m->m_data += len;
260		m->m_len -= ETHERIP_ALIGN;
261	}
262#endif
263
264	ip6 = mtod(m, struct ip6_hdr *);
265	MPASS(sc->gif_family == AF_INET6);
266	bcopy(sc->gif_ip6hdr, ip6, sizeof(struct ip6_hdr));
267
268	ip6->ip6_flow  |= htonl((uint32_t)ecn << 20);
269	ip6->ip6_nxt	= proto;
270	ip6->ip6_hlim	= V_ip6_gif_hlim;
271	/*
272	 * force fragmentation to minimum MTU, to avoid path MTU discovery.
273	 * it is too painful to ask for resend of inner packet, to achieve
274	 * path MTU discovery for encapsulated packets.
275	 */
276	return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL));
277}
278
279static int
280in6_gif_input(struct mbuf *m, int off, int proto, void *arg)
281{
282	struct gif_softc *sc = arg;
283	struct ifnet *gifp;
284	struct ip6_hdr *ip6;
285	uint8_t ecn;
286
287	MPASS(in_epoch(net_epoch_preempt));
288	if (sc == NULL) {
289		m_freem(m);
290		IP6STAT_INC(ip6s_nogif);
291		return (IPPROTO_DONE);
292	}
293	gifp = GIF2IFP(sc);
294	if ((gifp->if_flags & IFF_UP) != 0) {
295		ip6 = mtod(m, struct ip6_hdr *);
296		ecn = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
297		m_adj(m, off);
298		gif_input(m, gifp, proto, ecn);
299	} else {
300		m_freem(m);
301		IP6STAT_INC(ip6s_nogif);
302	}
303	return (IPPROTO_DONE);
304}
305
306static int
307in6_gif_lookup(const struct mbuf *m, int off, int proto, void **arg)
308{
309	const struct ip6_hdr *ip6;
310	struct gif_softc *sc;
311	int ret;
312
313	if (V_ipv6_hashtbl == NULL)
314		return (0);
315
316	MPASS(in_epoch(net_epoch_preempt));
317	/*
318	 * NOTE: it is safe to iterate without any locking here, because softc
319	 * can be reclaimed only when we are not within net_epoch_preempt
320	 * section, but ip_encap lookup+input are executed in epoch section.
321	 */
322	ip6 = mtod(m, const struct ip6_hdr *);
323	ret = 0;
324	CK_LIST_FOREACH(sc, &GIF_HASH(&ip6->ip6_dst, &ip6->ip6_src), chain) {
325		/*
326		 * This is an inbound packet, its ip6_dst is source address
327		 * in softc.
328		 */
329		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
330		    &ip6->ip6_dst) &&
331		    IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_dst,
332		    &ip6->ip6_src)) {
333			ret = ENCAP_DRV_LOOKUP;
334			goto done;
335		}
336	}
337	/*
338	 * No exact match.
339	 * Check the list of interfaces with GIF_IGNORE_SOURCE flag.
340	 */
341	CK_LIST_FOREACH(sc, &V_ipv6_list, chain) {
342		if (IN6_ARE_ADDR_EQUAL(&sc->gif_ip6hdr->ip6_src,
343		    &ip6->ip6_dst)) {
344			ret = 128 + 8; /* src + proto */
345			goto done;
346		}
347	}
348	return (0);
349done:
350	if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0)
351		return (0);
352	/* ingress filters on outer source */
353	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) {
354		struct nhop6_basic nh6;
355
356		if (fib6_lookup_nh_basic(sc->gif_fibnum, &ip6->ip6_src,
357		    ntohs(in6_getscope(&ip6->ip6_src)), 0, 0, &nh6) != 0)
358			return (0);
359
360		if (nh6.nh_ifp != m->m_pkthdr.rcvif)
361			return (0);
362	}
363	*arg = sc;
364	return (ret);
365}
366
367static struct {
368	const struct encap_config encap;
369	const struct encaptab *cookie;
370} ipv6_encap_cfg[] = {
371#ifdef INET
372	{
373		.encap = {
374			.proto = IPPROTO_IPV4,
375			.min_length = sizeof(struct ip6_hdr) +
376			    sizeof(struct ip),
377			.exact_match = ENCAP_DRV_LOOKUP,
378			.lookup = in6_gif_lookup,
379			.input = in6_gif_input
380		},
381	},
382#endif
383	{
384		.encap = {
385			.proto = IPPROTO_IPV6,
386			.min_length = 2 * sizeof(struct ip6_hdr),
387			.exact_match = ENCAP_DRV_LOOKUP,
388			.lookup = in6_gif_lookup,
389			.input = in6_gif_input
390		},
391	},
392	{
393		.encap = {
394			.proto = IPPROTO_ETHERIP,
395			.min_length = sizeof(struct ip6_hdr) +
396			    sizeof(struct etherip_header) +
397			    sizeof(struct ether_header),
398			.exact_match = ENCAP_DRV_LOOKUP,
399			.lookup = in6_gif_lookup,
400			.input = in6_gif_input
401		},
402	}
403};
404
405void
406in6_gif_init(void)
407{
408	int i;
409
410	if (!IS_DEFAULT_VNET(curvnet))
411		return;
412	for (i = 0; i < nitems(ipv6_encap_cfg); i++)
413		ipv6_encap_cfg[i].cookie = ip6_encap_attach(
414		    &ipv6_encap_cfg[i].encap, NULL, M_WAITOK);
415}
416
417void
418in6_gif_uninit(void)
419{
420	int i;
421
422	if (IS_DEFAULT_VNET(curvnet)) {
423		for (i = 0; i < nitems(ipv6_encap_cfg); i++)
424			ip6_encap_detach(ipv6_encap_cfg[i].cookie);
425	}
426	if (V_ipv6_hashtbl != NULL)
427		gif_hashdestroy(V_ipv6_hashtbl);
428}
429