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: in_gif.c,v 1.54 2001/05/14 14:02:16 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/sysctl.h>
50#include <sys/malloc.h>
51#include <sys/proc.h>
52
53#include <net/ethernet.h>
54#include <net/if.h>
55#include <net/if_var.h>
56#include <net/route.h>
57#include <net/vnet.h>
58
59#include <netinet/in.h>
60#include <netinet/in_systm.h>
61#include <netinet/ip.h>
62#include <netinet/ip_var.h>
63#include <netinet/in_var.h>
64#include <netinet/ip_encap.h>
65#include <netinet/ip_ecn.h>
66#include <netinet/in_fib.h>
67
68#ifdef INET6
69#include <netinet/ip6.h>
70#endif
71
72#include <net/if_gif.h>
73
74#define GIF_TTL		30
75VNET_DEFINE_STATIC(int, ip_gif_ttl) = GIF_TTL;
76#define	V_ip_gif_ttl		VNET(ip_gif_ttl)
77SYSCTL_INT(_net_inet_ip, IPCTL_GIF_TTL, gifttl, CTLFLAG_VNET | CTLFLAG_RW,
78    &VNET_NAME(ip_gif_ttl), 0, "Default TTL value for encapsulated packets");
79
80/*
81 * We keep interfaces in a hash table using src+dst as key.
82 * Interfaces with GIF_IGNORE_SOURCE flag are linked into plain list.
83 */
84VNET_DEFINE_STATIC(struct gif_list *, ipv4_hashtbl) = NULL;
85VNET_DEFINE_STATIC(struct gif_list *, ipv4_srchashtbl) = NULL;
86VNET_DEFINE_STATIC(struct gif_list, ipv4_list) = CK_LIST_HEAD_INITIALIZER();
87#define	V_ipv4_hashtbl		VNET(ipv4_hashtbl)
88#define	V_ipv4_srchashtbl	VNET(ipv4_srchashtbl)
89#define	V_ipv4_list		VNET(ipv4_list)
90
91#define	GIF_HASH(src, dst)	(V_ipv4_hashtbl[\
92    in_gif_hashval((src), (dst)) & (GIF_HASH_SIZE - 1)])
93#define	GIF_SRCHASH(src)	(V_ipv4_srchashtbl[\
94    fnv_32_buf(&(src), sizeof(src), FNV1_32_INIT) & (GIF_HASH_SIZE - 1)])
95#define	GIF_HASH_SC(sc)		GIF_HASH((sc)->gif_iphdr->ip_src.s_addr,\
96    (sc)->gif_iphdr->ip_dst.s_addr)
97static uint32_t
98in_gif_hashval(in_addr_t src, in_addr_t dst)
99{
100	uint32_t ret;
101
102	ret = fnv_32_buf(&src, sizeof(src), FNV1_32_INIT);
103	return (fnv_32_buf(&dst, sizeof(dst), ret));
104}
105
106static int
107in_gif_checkdup(const struct gif_softc *sc, in_addr_t src, in_addr_t dst)
108{
109	struct gif_softc *tmp;
110
111	if (sc->gif_family == AF_INET &&
112	    sc->gif_iphdr->ip_src.s_addr == src &&
113	    sc->gif_iphdr->ip_dst.s_addr == dst)
114		return (EEXIST);
115
116	CK_LIST_FOREACH(tmp, &GIF_HASH(src, dst), chain) {
117		if (tmp == sc)
118			continue;
119		if (tmp->gif_iphdr->ip_src.s_addr == src &&
120		    tmp->gif_iphdr->ip_dst.s_addr == dst)
121			return (EADDRNOTAVAIL);
122	}
123	return (0);
124}
125
126/*
127 * Check that ingress address belongs to local host.
128 */
129static void
130in_gif_set_running(struct gif_softc *sc)
131{
132
133	if (in_localip(sc->gif_iphdr->ip_src))
134		GIF2IFP(sc)->if_drv_flags |= IFF_DRV_RUNNING;
135	else
136		GIF2IFP(sc)->if_drv_flags &= ~IFF_DRV_RUNNING;
137}
138
139/*
140 * ifaddr_event handler.
141 * Clear IFF_DRV_RUNNING flag when ingress address disappears to prevent
142 * source address spoofing.
143 */
144static void
145in_gif_srcaddr(void *arg __unused, const struct sockaddr *sa,
146    int event __unused)
147{
148	const struct sockaddr_in *sin;
149	struct gif_softc *sc;
150
151	/* Check that VNET is ready */
152	if (V_ipv4_hashtbl == NULL)
153		return;
154
155	NET_EPOCH_ASSERT();
156	sin = (const struct sockaddr_in *)sa;
157	CK_LIST_FOREACH(sc, &GIF_SRCHASH(sin->sin_addr.s_addr), srchash) {
158		if (sc->gif_iphdr->ip_src.s_addr != sin->sin_addr.s_addr)
159			continue;
160		in_gif_set_running(sc);
161	}
162}
163
164static void
165in_gif_attach(struct gif_softc *sc)
166{
167
168	if (sc->gif_options & GIF_IGNORE_SOURCE)
169		CK_LIST_INSERT_HEAD(&V_ipv4_list, sc, chain);
170	else
171		CK_LIST_INSERT_HEAD(&GIF_HASH_SC(sc), sc, chain);
172
173	CK_LIST_INSERT_HEAD(&GIF_SRCHASH(sc->gif_iphdr->ip_src.s_addr),
174	    sc, srchash);
175}
176
177int
178in_gif_setopts(struct gif_softc *sc, u_int options)
179{
180
181	/* NOTE: we are protected with gif_ioctl_sx lock */
182	MPASS(sc->gif_family == AF_INET);
183	MPASS(sc->gif_options != options);
184
185	if ((options & GIF_IGNORE_SOURCE) !=
186	    (sc->gif_options & GIF_IGNORE_SOURCE)) {
187		CK_LIST_REMOVE(sc, srchash);
188		CK_LIST_REMOVE(sc, chain);
189		sc->gif_options = options;
190		in_gif_attach(sc);
191	}
192	return (0);
193}
194
195int
196in_gif_ioctl(struct gif_softc *sc, u_long cmd, caddr_t data)
197{
198	struct ifreq *ifr = (struct ifreq *)data;
199	struct sockaddr_in *dst, *src;
200	struct ip *ip;
201	int error;
202
203	/* NOTE: we are protected with gif_ioctl_sx lock */
204	error = EINVAL;
205	switch (cmd) {
206	case SIOCSIFPHYADDR:
207		src = &((struct in_aliasreq *)data)->ifra_addr;
208		dst = &((struct in_aliasreq *)data)->ifra_dstaddr;
209
210		/* sanity checks */
211		if (src->sin_family != dst->sin_family ||
212		    src->sin_family != AF_INET ||
213		    src->sin_len != dst->sin_len ||
214		    src->sin_len != sizeof(*src))
215			break;
216		if (src->sin_addr.s_addr == INADDR_ANY ||
217		    dst->sin_addr.s_addr == INADDR_ANY) {
218			error = EADDRNOTAVAIL;
219			break;
220		}
221		if (V_ipv4_hashtbl == NULL) {
222			V_ipv4_hashtbl = gif_hashinit();
223			V_ipv4_srchashtbl = gif_hashinit();
224		}
225		error = in_gif_checkdup(sc, src->sin_addr.s_addr,
226		    dst->sin_addr.s_addr);
227		if (error == EADDRNOTAVAIL)
228			break;
229		if (error == EEXIST) {
230			/* Addresses are the same. Just return. */
231			error = 0;
232			break;
233		}
234		ip = malloc(sizeof(*ip), M_GIF, M_WAITOK | M_ZERO);
235		ip->ip_src.s_addr = src->sin_addr.s_addr;
236		ip->ip_dst.s_addr = dst->sin_addr.s_addr;
237		if (sc->gif_family != 0) {
238			/* Detach existing tunnel first */
239			CK_LIST_REMOVE(sc, srchash);
240			CK_LIST_REMOVE(sc, chain);
241			GIF_WAIT();
242			free(sc->gif_hdr, M_GIF);
243			/* XXX: should we notify about link state change? */
244		}
245		sc->gif_family = AF_INET;
246		sc->gif_iphdr = ip;
247		in_gif_attach(sc);
248		in_gif_set_running(sc);
249		break;
250	case SIOCGIFPSRCADDR:
251	case SIOCGIFPDSTADDR:
252		if (sc->gif_family != AF_INET) {
253			error = EADDRNOTAVAIL;
254			break;
255		}
256		src = (struct sockaddr_in *)&ifr->ifr_addr;
257		memset(src, 0, sizeof(*src));
258		src->sin_family = AF_INET;
259		src->sin_len = sizeof(*src);
260		src->sin_addr = (cmd == SIOCGIFPSRCADDR) ?
261		    sc->gif_iphdr->ip_src: sc->gif_iphdr->ip_dst;
262		error = prison_if(curthread->td_ucred, (struct sockaddr *)src);
263		if (error != 0)
264			memset(src, 0, sizeof(*src));
265		break;
266	}
267	return (error);
268}
269
270int
271in_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn)
272{
273	struct gif_softc *sc = ifp->if_softc;
274	struct ip *ip;
275	int len;
276
277	/* prepend new IP header */
278	NET_EPOCH_ASSERT();
279	len = sizeof(struct ip);
280#ifndef __NO_STRICT_ALIGNMENT
281	if (proto == IPPROTO_ETHERIP)
282		len += ETHERIP_ALIGN;
283#endif
284	M_PREPEND(m, len, M_NOWAIT);
285	if (m == NULL)
286		return (ENOBUFS);
287#ifndef __NO_STRICT_ALIGNMENT
288	if (proto == IPPROTO_ETHERIP) {
289		len = mtod(m, vm_offset_t) & 3;
290		KASSERT(len == 0 || len == ETHERIP_ALIGN,
291		    ("in_gif_output: unexpected misalignment"));
292		m->m_data += len;
293		m->m_len -= ETHERIP_ALIGN;
294	}
295#endif
296	ip = mtod(m, struct ip *);
297
298	MPASS(sc->gif_family == AF_INET);
299	bcopy(sc->gif_iphdr, ip, sizeof(struct ip));
300	ip->ip_p = proto;
301	/* version will be set in ip_output() */
302	ip->ip_ttl = V_ip_gif_ttl;
303	ip->ip_len = htons(m->m_pkthdr.len);
304	ip->ip_tos = ecn;
305
306	return (ip_output(m, NULL, NULL, 0, NULL, NULL));
307}
308
309static int
310in_gif_input(struct mbuf *m, int off, int proto, void *arg)
311{
312	struct gif_softc *sc = arg;
313	struct ifnet *gifp;
314	struct ip *ip;
315	uint8_t ecn;
316
317	NET_EPOCH_ASSERT();
318	if (sc == NULL) {
319		m_freem(m);
320		KMOD_IPSTAT_INC(ips_nogif);
321		return (IPPROTO_DONE);
322	}
323	gifp = GIF2IFP(sc);
324	if ((gifp->if_flags & IFF_UP) != 0) {
325		ip = mtod(m, struct ip *);
326		ecn = ip->ip_tos;
327		m_adj(m, off);
328		gif_input(m, gifp, proto, ecn);
329	} else {
330		m_freem(m);
331		KMOD_IPSTAT_INC(ips_nogif);
332	}
333	return (IPPROTO_DONE);
334}
335
336static int
337in_gif_lookup(const struct mbuf *m, int off, int proto, void **arg)
338{
339	const struct ip *ip;
340	struct gif_softc *sc;
341	int ret;
342
343	if (V_ipv4_hashtbl == NULL)
344		return (0);
345
346	NET_EPOCH_ASSERT();
347	ip = mtod(m, const struct ip *);
348	/*
349	 * NOTE: it is safe to iterate without any locking here, because softc
350	 * can be reclaimed only when we are not within net_epoch_preempt
351	 * section, but ip_encap lookup+input are executed in epoch section.
352	 */
353	ret = 0;
354	CK_LIST_FOREACH(sc, &GIF_HASH(ip->ip_dst.s_addr,
355	    ip->ip_src.s_addr), chain) {
356		/*
357		 * This is an inbound packet, its ip_dst is source address
358		 * in softc.
359		 */
360		if (sc->gif_iphdr->ip_src.s_addr == ip->ip_dst.s_addr &&
361		    sc->gif_iphdr->ip_dst.s_addr == ip->ip_src.s_addr) {
362			ret = ENCAP_DRV_LOOKUP;
363			goto done;
364		}
365	}
366	/*
367	 * No exact match.
368	 * Check the list of interfaces with GIF_IGNORE_SOURCE flag.
369	 */
370	CK_LIST_FOREACH(sc, &V_ipv4_list, chain) {
371		if (sc->gif_iphdr->ip_src.s_addr == ip->ip_dst.s_addr) {
372			ret = 32 + 8; /* src + proto */
373			goto done;
374		}
375	}
376	return (0);
377done:
378	if ((GIF2IFP(sc)->if_flags & IFF_UP) == 0)
379		return (0);
380	/* ingress filters on outer source */
381	if ((GIF2IFP(sc)->if_flags & IFF_LINK2) == 0) {
382		if (fib4_check_urpf(sc->gif_fibnum, ip->ip_src, 0, NHR_NONE,
383					m->m_pkthdr.rcvif) == 0)
384			return (0);
385	}
386	*arg = sc;
387	return (ret);
388}
389
390static const struct srcaddrtab *ipv4_srcaddrtab;
391static struct {
392	const struct encap_config encap;
393	const struct encaptab *cookie;
394} ipv4_encap_cfg[] = {
395	{
396		.encap = {
397			.proto = IPPROTO_IPV4,
398			.min_length = 2 * sizeof(struct ip),
399			.exact_match = ENCAP_DRV_LOOKUP,
400			.lookup = in_gif_lookup,
401			.input = in_gif_input
402		},
403	},
404#ifdef INET6
405	{
406		.encap = {
407			.proto = IPPROTO_IPV6,
408			.min_length = sizeof(struct ip) +
409			    sizeof(struct ip6_hdr),
410			.exact_match = ENCAP_DRV_LOOKUP,
411			.lookup = in_gif_lookup,
412			.input = in_gif_input
413		},
414	},
415#endif
416	{
417		.encap = {
418			.proto = IPPROTO_ETHERIP,
419			.min_length = sizeof(struct ip) +
420			    sizeof(struct etherip_header) +
421			    sizeof(struct ether_header),
422			.exact_match = ENCAP_DRV_LOOKUP,
423			.lookup = in_gif_lookup,
424			.input = in_gif_input
425		},
426	}
427};
428
429void
430in_gif_init(void)
431{
432	int i;
433
434	if (!IS_DEFAULT_VNET(curvnet))
435		return;
436
437	ipv4_srcaddrtab = ip_encap_register_srcaddr(in_gif_srcaddr,
438	    NULL, M_WAITOK);
439	for (i = 0; i < nitems(ipv4_encap_cfg); i++)
440		ipv4_encap_cfg[i].cookie = ip_encap_attach(
441		    &ipv4_encap_cfg[i].encap, NULL, M_WAITOK);
442}
443
444void
445in_gif_uninit(void)
446{
447	int i;
448
449	if (IS_DEFAULT_VNET(curvnet)) {
450		for (i = 0; i < nitems(ipv4_encap_cfg); i++)
451			ip_encap_detach(ipv4_encap_cfg[i].cookie);
452		ip_encap_unregister_srcaddr(ipv4_srcaddrtab);
453	}
454	if (V_ipv4_hashtbl != NULL) {
455		gif_hashdestroy(V_ipv4_hashtbl);
456		V_ipv4_hashtbl = NULL;
457		GIF_WAIT();
458		gif_hashdestroy(V_ipv4_srchashtbl);
459	}
460}
461