1/*
2 * Copyright (c) 2000-2013 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28
29/*	$FreeBSD: src/sys/net/if_stf.c,v 1.1.2.6 2001/07/24 19:10:18 brooks Exp $	*/
30/*	$KAME: if_stf.c,v 1.62 2001/06/07 22:32:16 itojun Exp $	*/
31
32/*
33 * Copyright (C) 2000 WIDE Project.
34 * All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 *    notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 *    notice, this list of conditions and the following disclaimer in the
43 *    documentation and/or other materials provided with the distribution.
44 * 3. Neither the name of the project nor the names of its contributors
45 *    may be used to endorse or promote products derived from this software
46 *    without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 */
60/*
61 * NOTICE: This file was modified by SPARTA, Inc. in 2006 to introduce
62 * support for mandatory and extensible security protections.  This notice
63 * is included in support of clause 2.2 (b) of the Apple Public License,
64 * Version 2.0.
65 */
66
67/*
68 * 6to4 interface, based on RFC3056.
69 *
70 * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
71 * There is no address mapping defined from IPv6 multicast address to IPv4
72 * address.  Therefore, we do not have IFF_MULTICAST on the interface.
73 *
74 * Due to the lack of address mapping for link-local addresses, we cannot
75 * throw packets toward link-local addresses (fe80::x).  Also, we cannot throw
76 * packets to link-local multicast addresses (ff02::x).
77 *
78 * Here are interesting symptoms due to the lack of link-local address:
79 *
80 * Unicast routing exchange:
81 * - RIPng: Impossible.  Uses link-local multicast packet toward ff02::9,
82 *   and link-local addresses as nexthop.
83 * - OSPFv6: Impossible.  OSPFv6 assumes that there's link-local address
84 *   assigned to the link, and makes use of them.  Also, HELLO packets use
85 *   link-local multicast addresses (ff02::5 and ff02::6).
86 * - BGP4+: Maybe.  You can only use global address as nexthop, and global
87 *   address as TCP endpoint address.
88 *
89 * Multicast routing protocols:
90 * - PIM: Hello packet cannot be used to discover adjacent PIM routers.
91 *   Adjacent PIM routers must be configured manually (is it really spec-wise
92 *   correct thing to do?).
93 *
94 * ICMPv6:
95 * - Redirects cannot be used due to the lack of link-local address.
96 *
97 * stf interface does not have, and will not need, a link-local address.
98 * It seems to have no real benefit and does not help the above symptoms much.
99 * Even if we assign link-locals to interface, we cannot really
100 * use link-local unicast/multicast on top of 6to4 cloud (since there's no
101 * encapsulation defined for link-local address), and the above analysis does
102 * not change.  RFC3056 does not mandate the assignment of link-local address
103 * either.
104 *
105 * 6to4 interface has security issues.  Refer to
106 * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
107 * for details.  The code tries to filter out some of malicious packets.
108 * Note that there is no way to be 100% secure.
109 */
110
111#include <sys/param.h>
112#include <sys/systm.h>
113#include <sys/socket.h>
114#include <sys/sockio.h>
115#include <sys/mbuf.h>
116#include <sys/errno.h>
117#include <sys/protosw.h>
118#include <sys/kernel.h>
119#include <sys/syslog.h>
120
121#include <sys/malloc.h>
122
123#include <kern/locks.h>
124
125#include <net/if.h>
126#include <net/route.h>
127#include <net/if_types.h>
128
129#include <netinet/in.h>
130#include <netinet/in_systm.h>
131#include <netinet/ip.h>
132#include <netinet/ip_var.h>
133#include <netinet/in_var.h>
134
135#include <netinet/ip6.h>
136#include <netinet6/ip6_var.h>
137#include <netinet6/in6_var.h>
138#include <netinet/ip_ecn.h>
139
140#include <netinet/ip_encap.h>
141#include <net/kpi_interface.h>
142#include <net/kpi_protocol.h>
143
144
145#include <net/net_osdep.h>
146
147#include <net/bpf.h>
148
149#if CONFIG_MACF_NET
150#include <security/mac_framework.h>
151#endif
152
153#define GET_V4(x) ((const struct in_addr *)(const void *)(&(x)->s6_addr16[1]))
154
155static lck_grp_t *stf_mtx_grp;
156
157struct stf_softc {
158	ifnet_t				sc_if;	   /* common area */
159	u_int32_t				sc_protocol_family; /* dlil protocol attached */
160	union {
161		struct route  __sc_ro4;
162		struct route_in6 __sc_ro6; /* just for safety */
163	} __sc_ro46;
164#define sc_ro	__sc_ro46.__sc_ro4
165	decl_lck_mtx_data(, sc_ro_mtx);
166	const struct encaptab *encap_cookie;
167	bpf_tap_mode		tap_mode;
168	bpf_packet_func		tap_callback;
169};
170
171void stfattach (void);
172
173static int ip_stf_ttl = 40;
174static int stf_init_done;
175
176static void in_stf_input(struct mbuf *, int);
177static void stfinit(void);
178
179static struct protosw in_stf_protosw =
180{
181	.pr_type =		SOCK_RAW,
182	.pr_protocol =		IPPROTO_IPV6,
183	.pr_flags =		PR_ATOMIC|PR_ADDR,
184	.pr_input =		in_stf_input,
185	.pr_ctloutput =		rip_ctloutput,
186	.pr_usrreqs =		&rip_usrreqs,
187	.pr_unlock =		rip_unlock,
188};
189
190static int stf_encapcheck(const struct mbuf *, int, int, void *);
191static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
192int stf_pre_output(struct ifnet *, protocol_family_t, struct mbuf **,
193	const struct sockaddr *, void *, char *, char *);
194static int stf_checkaddr4(struct stf_softc *, const struct in_addr *,
195	struct ifnet *);
196static int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
197	struct ifnet *);
198static void stf_rtrequest(int, struct rtentry *, struct sockaddr *);
199static errno_t stf_ioctl(ifnet_t ifp, u_long cmd, void *data);
200static errno_t stf_output(ifnet_t ifp, mbuf_t m);
201
202static void
203stfinit(void)
204{
205	if (!stf_init_done) {
206		stf_mtx_grp = lck_grp_alloc_init("stf", LCK_GRP_ATTR_NULL);
207		stf_init_done = 1;
208	}
209}
210
211/*
212 * gif_input is the input handler for IP and IPv6 attached to gif
213 */
214static errno_t
215stf_media_input(
216	__unused ifnet_t	ifp,
217	protocol_family_t	protocol_family,
218	mbuf_t				m,
219	__unused char		*frame_header)
220{
221	if (proto_input(protocol_family, m) != 0)
222		m_freem(m);
223
224	return (0);
225}
226
227
228
229static errno_t
230stf_add_proto(
231	ifnet_t									ifp,
232	protocol_family_t						protocol_family,
233	__unused const struct ifnet_demux_desc	*demux_array,
234	__unused u_int32_t						demux_count)
235{
236	/* Only one protocol may be attached at a time */
237	struct stf_softc* stf = ifnet_softc(ifp);
238	if (stf->sc_protocol_family == 0)
239		stf->sc_protocol_family = protocol_family;
240	else {
241		printf("stf_add_proto: stf already has a proto\n");
242		return EBUSY;
243	}
244
245	return 0;
246}
247
248static errno_t
249stf_del_proto(
250	ifnet_t				ifp,
251	protocol_family_t	protocol_family)
252{
253	if (((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family == protocol_family)
254		((struct stf_softc*)ifnet_softc(ifp))->sc_protocol_family = 0;
255
256	return 0;
257}
258
259static errno_t
260stf_attach_inet6(
261	ifnet_t				ifp,
262	protocol_family_t	protocol_family)
263{
264    struct ifnet_attach_proto_param	reg;
265    errno_t							stat;
266
267    if (protocol_family != PF_INET6)
268    	return EPROTONOSUPPORT;
269
270	bzero(&reg, sizeof(reg));
271    reg.input = stf_media_input;
272    reg.pre_output = stf_pre_output;
273
274    stat = ifnet_attach_protocol(ifp, protocol_family, &reg);
275    if (stat && stat != EEXIST) {
276        printf("stf_attach_proto_family can't attach interface fam=%d\n",
277        	   protocol_family);
278    }
279
280    return stat;
281}
282
283static errno_t
284stf_demux(
285	ifnet_t					ifp,
286	__unused mbuf_t			m,
287	__unused char			*frame_ptr,
288	protocol_family_t		*protocol_family)
289{
290	struct stf_softc* stf = ifnet_softc(ifp);
291	*protocol_family = stf->sc_protocol_family;
292	return 0;
293}
294
295static errno_t
296stf_set_bpf_tap(
297	ifnet_t			ifp,
298	bpf_tap_mode	mode,
299	bpf_packet_func	callback)
300{
301	struct stf_softc	*sc = ifnet_softc(ifp);
302
303	sc->tap_mode = mode;
304	sc->tap_callback = callback;
305
306	return 0;
307}
308
309void
310stfattach(void)
311{
312	struct stf_softc *sc;
313	int error;
314	const struct encaptab *p;
315	struct ifnet_init_params	stf_init;
316
317	stfinit();
318
319	error = proto_register_plumber(PF_INET6, APPLE_IF_FAM_STF,
320								   stf_attach_inet6, NULL);
321	if (error != 0)
322		printf("proto_register_plumber failed for AF_INET6 error=%d\n", error);
323
324	sc = _MALLOC(sizeof(struct stf_softc), M_DEVBUF, M_WAITOK);
325	if (sc == 0) {
326		printf("stf softc attach failed\n" );
327		return;
328	}
329
330	bzero(sc, sizeof(*sc));
331
332	p = encap_attach_func(AF_INET, IPPROTO_IPV6, stf_encapcheck,
333	    &in_stf_protosw, sc);
334	if (p == NULL) {
335		printf("sftattach encap_attach_func failed\n");
336		FREE(sc, M_DEVBUF);
337		return;
338	}
339	sc->encap_cookie = p;
340	lck_mtx_init(&sc->sc_ro_mtx, stf_mtx_grp, LCK_ATTR_NULL);
341
342	bzero(&stf_init, sizeof(stf_init));
343	stf_init.name = "stf";
344	stf_init.unit = 0;
345	stf_init.type = IFT_STF;
346	stf_init.family = IFNET_FAMILY_STF;
347	stf_init.output = stf_output;
348	stf_init.demux = stf_demux;
349	stf_init.add_proto = stf_add_proto;
350	stf_init.del_proto = stf_del_proto;
351	stf_init.softc = sc;
352	stf_init.ioctl = stf_ioctl;
353	stf_init.set_bpf_tap = stf_set_bpf_tap;
354
355	error = ifnet_allocate(&stf_init, &sc->sc_if);
356	if (error != 0) {
357		printf("stfattach, ifnet_allocate failed - %d\n", error);
358		encap_detach(sc->encap_cookie);
359		lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
360		FREE(sc, M_DEVBUF);
361		return;
362	}
363	ifnet_set_mtu(sc->sc_if, IPV6_MMTU);
364	ifnet_set_flags(sc->sc_if, 0, 0xffff); /* clear all flags */
365#if 0
366	/* turn off ingress filter */
367	ifnet_set_flags(sc->sc_if, IFF_LINK2, IFF_LINK2);
368#endif
369
370#if CONFIG_MACF_NET
371	mac_ifnet_label_init(&sc->sc_if);
372#endif
373
374	error = ifnet_attach(sc->sc_if, NULL);
375	if (error != 0) {
376		printf("stfattach: ifnet_attach returned error=%d\n", error);
377		encap_detach(sc->encap_cookie);
378		ifnet_release(sc->sc_if);
379		lck_mtx_destroy(&sc->sc_ro_mtx, stf_mtx_grp);
380		FREE(sc, M_DEVBUF);
381		return;
382	}
383
384	bpfattach(sc->sc_if, DLT_NULL, sizeof(u_int));
385
386	return;
387}
388
389static int
390stf_encapcheck(
391	const struct mbuf *m,
392	__unused int off,
393	int proto,
394	void *arg)
395{
396	struct ip ip;
397	struct in6_ifaddr *ia6;
398	struct stf_softc *sc;
399	struct in_addr a, b;
400
401	sc = (struct stf_softc *)arg;
402	if (sc == NULL)
403		return 0;
404
405	if ((ifnet_flags(sc->sc_if) & IFF_UP) == 0)
406		return 0;
407
408	/* IFF_LINK0 means "no decapsulation" */
409	if ((ifnet_flags(sc->sc_if) & IFF_LINK0) != 0)
410		return 0;
411
412	if (proto != IPPROTO_IPV6)
413		return 0;
414
415	mbuf_copydata((struct mbuf *)(size_t)m, 0, sizeof(ip), &ip);
416
417	if (ip.ip_v != 4)
418		return 0;
419
420	ia6 = stf_getsrcifa6(sc->sc_if);
421	if (ia6 == NULL)
422		return 0;
423
424	/*
425	 * check if IPv4 dst matches the IPv4 address derived from the
426	 * local 6to4 address.
427	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
428	 */
429	IFA_LOCK(&ia6->ia_ifa);
430	if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
431	    sizeof(ip.ip_dst)) != 0) {
432		IFA_UNLOCK(&ia6->ia_ifa);
433		IFA_REMREF(&ia6->ia_ifa);
434		return 0;
435	}
436	/*
437	 * check if IPv4 src matches the IPv4 address derived from the
438	 * local 6to4 address masked by prefixmask.
439	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
440	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
441	 */
442	bzero(&a, sizeof(a));
443	a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
444	a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
445	b = ip.ip_src;
446	b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
447	if (a.s_addr != b.s_addr) {
448		IFA_UNLOCK(&ia6->ia_ifa);
449		IFA_REMREF(&ia6->ia_ifa);
450		return 0;
451	}
452	/* stf interface makes single side match only */
453	IFA_UNLOCK(&ia6->ia_ifa);
454	IFA_REMREF(&ia6->ia_ifa);
455	return 32;
456}
457
458static struct in6_ifaddr *
459stf_getsrcifa6(struct ifnet *ifp)
460{
461	struct ifaddr *ia;
462	struct in_ifaddr *ia4;
463	struct sockaddr_in6 *sin6;
464	struct in_addr in;
465
466	ifnet_lock_shared(ifp);
467	for (ia = ifp->if_addrlist.tqh_first; ia; ia = ia->ifa_list.tqe_next) {
468		IFA_LOCK(ia);
469		if (ia->ifa_addr == NULL) {
470			IFA_UNLOCK(ia);
471			continue;
472		}
473		if (ia->ifa_addr->sa_family != AF_INET6) {
474			IFA_UNLOCK(ia);
475			continue;
476		}
477		sin6 = (struct sockaddr_in6 *)(void *)ia->ifa_addr;
478		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
479			IFA_UNLOCK(ia);
480			continue;
481		}
482		bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
483		IFA_UNLOCK(ia);
484		lck_rw_lock_shared(in_ifaddr_rwlock);
485		for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
486		     ia4;
487		     ia4 = TAILQ_NEXT(ia4, ia_link))
488		{
489			IFA_LOCK(&ia4->ia_ifa);
490			if (ia4->ia_addr.sin_addr.s_addr == in.s_addr) {
491				IFA_UNLOCK(&ia4->ia_ifa);
492				break;
493			}
494			IFA_UNLOCK(&ia4->ia_ifa);
495		}
496		lck_rw_done(in_ifaddr_rwlock);
497		if (ia4 == NULL)
498			continue;
499
500		IFA_ADDREF(ia);		/* for caller */
501		ifnet_lock_done(ifp);
502		return ((struct in6_ifaddr *)ia);
503	}
504	ifnet_lock_done(ifp);
505
506	return (NULL);
507}
508
509int
510stf_pre_output(
511	struct ifnet	*ifp,
512	__unused protocol_family_t  protocol_family,
513	struct mbuf	**m0,
514	const struct sockaddr	*dst,
515	__unused void *route,
516	__unused char *desk_linkaddr,
517	__unused char *frame_type)
518{
519	struct mbuf *m = *m0;
520	struct stf_softc *sc;
521	const struct sockaddr_in6 *dst6;
522	const struct in_addr *in4;
523	u_int8_t tos;
524	struct ip *ip;
525	struct ip6_hdr *ip6;
526	struct in6_ifaddr *ia6;
527	struct sockaddr_in 	*dst4;
528	struct ip_out_args ipoa =
529	    { IFSCOPE_NONE, { 0 }, IPOAF_SELECT_SRCIF, 0 };
530	errno_t				result = 0;
531
532	sc = ifnet_softc(ifp);
533	dst6 = (const struct sockaddr_in6 *)(const void *)dst;
534
535	/* just in case */
536	if ((ifnet_flags(ifp) & IFF_UP) == 0) {
537		printf("stf: IFF_DOWN\n");
538		return ENETDOWN;
539	}
540
541	/*
542	 * If we don't have an ip4 address that match my inner ip6 address,
543	 * we shouldn't generate output.  Without this check, we'll end up
544	 * using wrong IPv4 source.
545	 */
546	ia6 = stf_getsrcifa6(ifp);
547	if (ia6 == NULL) {
548		return ENETDOWN;
549	}
550
551	if (mbuf_len(m) < sizeof(*ip6)) {
552		m = m_pullup(m, sizeof(*ip6));
553		if (!m) {
554			*m0 = NULL; /* makes sure this won't be double freed */
555			IFA_REMREF(&ia6->ia_ifa);
556			return ENOBUFS;
557		}
558	}
559	ip6 = mtod(m, struct ip6_hdr *);
560	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
561
562	/*
563	 * Pickup the right outer dst addr from the list of candidates.
564	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
565	 */
566	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
567		in4 = GET_V4(&ip6->ip6_dst);
568	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
569		in4 = GET_V4(&dst6->sin6_addr);
570	else {
571		IFA_REMREF(&ia6->ia_ifa);
572		return ENETUNREACH;
573	}
574
575	if (ifp->if_bpf) {
576		/* We need to prepend the address family as a four byte field. */
577		u_int32_t af = AF_INET6;
578
579		bpf_tap_out(ifp, 0, m, &af, sizeof(af));
580	}
581
582	M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
583	if (m && mbuf_len(m) < sizeof(struct ip))
584		m = m_pullup(m, sizeof(struct ip));
585	if (m == NULL) {
586		*m0 = NULL;
587		IFA_REMREF(&ia6->ia_ifa);
588		return ENOBUFS;
589	}
590	ip = mtod(m, struct ip *);
591
592	bzero(ip, sizeof(*ip));
593
594	IFA_LOCK_SPIN(&ia6->ia_ifa);
595	bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
596	    &ip->ip_src, sizeof(ip->ip_src));
597	IFA_UNLOCK(&ia6->ia_ifa);
598	bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
599	ip->ip_p = IPPROTO_IPV6;
600	ip->ip_ttl = ip_stf_ttl;
601	ip->ip_len = m->m_pkthdr.len;	/*host order*/
602	if (ifp->if_flags & IFF_LINK1)
603		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
604	else
605		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
606
607	lck_mtx_lock(&sc->sc_ro_mtx);
608	dst4 = (struct sockaddr_in *)(void *)&sc->sc_ro.ro_dst;
609	if (ROUTE_UNUSABLE(&sc->sc_ro) || dst4->sin_family != AF_INET ||
610	    bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
611		ROUTE_RELEASE(&sc->sc_ro);
612		/* cache route doesn't match: always the case during the first use */
613		dst4->sin_family = AF_INET;
614		dst4->sin_len = sizeof(struct sockaddr_in);
615		bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
616	}
617
618	result = ip_output(m, NULL, &sc->sc_ro, IP_OUTARGS, NULL, &ipoa);
619	lck_mtx_unlock(&sc->sc_ro_mtx);
620
621	/* Assumption: ip_output will free mbuf on errors */
622	/* All the output processing is done here, don't let stf_output be called */
623	if (result == 0)
624		result = EJUSTRETURN;
625	*m0 = NULL;
626	IFA_REMREF(&ia6->ia_ifa);
627	return result;
628}
629static errno_t
630stf_output(
631	__unused ifnet_t	ifp,
632	__unused mbuf_t	m)
633{
634	/* All processing is done in stf_pre_output
635	 * this shouldn't be called as the pre_output returns "EJUSTRETURN"
636	 */
637	return 0;
638}
639
640static int
641stf_checkaddr4(
642	struct stf_softc *sc,
643	const struct in_addr *in,
644	struct ifnet *inifp)	/* incoming interface */
645{
646	struct in_ifaddr *ia4;
647
648	/*
649	 * reject packets with the following address:
650	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
651	 */
652	if (IN_MULTICAST(ntohl(in->s_addr)))
653		return -1;
654	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
655	case 0: case 127: case 255:
656		return -1;
657	}
658
659	/*
660	 * reject packets with broadcast
661	 */
662	lck_rw_lock_shared(in_ifaddr_rwlock);
663	for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
664	     ia4;
665	     ia4 = TAILQ_NEXT(ia4, ia_link))
666	{
667		IFA_LOCK(&ia4->ia_ifa);
668		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0) {
669			IFA_UNLOCK(&ia4->ia_ifa);
670			continue;
671		}
672		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
673			IFA_UNLOCK(&ia4->ia_ifa);
674			lck_rw_done(in_ifaddr_rwlock);
675			return -1;
676		}
677		IFA_UNLOCK(&ia4->ia_ifa);
678	}
679	lck_rw_done(in_ifaddr_rwlock);
680
681	/*
682	 * perform ingress filter
683	 */
684	if (sc && (ifnet_flags(sc->sc_if) & IFF_LINK2) == 0 && inifp) {
685		struct sockaddr_in sin;
686		struct rtentry *rt;
687
688		bzero(&sin, sizeof(sin));
689		sin.sin_family = AF_INET;
690		sin.sin_len = sizeof(struct sockaddr_in);
691		sin.sin_addr = *in;
692		rt = rtalloc1((struct sockaddr *)&sin, 0, 0);
693		if (rt != NULL)
694			RT_LOCK(rt);
695		if (rt == NULL || rt->rt_ifp != inifp) {
696#if 1
697			log(LOG_WARNING, "%s: packet from 0x%x dropped "
698			    "due to ingress filter\n", if_name(sc->sc_if),
699			    (u_int32_t)ntohl(sin.sin_addr.s_addr));
700#endif
701			if (rt != NULL) {
702				RT_UNLOCK(rt);
703				rtfree(rt);
704			}
705			return -1;
706		}
707		RT_UNLOCK(rt);
708		rtfree(rt);
709	}
710
711	return 0;
712}
713
714static int
715stf_checkaddr6(
716	struct stf_softc *sc,
717	struct in6_addr *in6,
718	struct ifnet *inifp)	/* incoming interface */
719{
720	/*
721	 * check 6to4 addresses
722	 */
723	if (IN6_IS_ADDR_6TO4(in6))
724		return stf_checkaddr4(sc, GET_V4(in6), inifp);
725
726	/*
727	 * reject anything that look suspicious.  the test is implemented
728	 * in ip6_input too, but we check here as well to
729	 * (1) reject bad packets earlier, and
730	 * (2) to be safe against future ip6_input change.
731	 */
732	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
733		return -1;
734
735	return 0;
736}
737
738static void
739in_stf_input(
740	struct mbuf *m,
741	int off)
742{
743	struct stf_softc *sc;
744	struct ip *ip;
745	struct ip6_hdr ip6;
746	u_int8_t otos, itos;
747	int proto;
748	struct ifnet *ifp;
749	struct ifnet_stat_increment_param	stats;
750
751	ip = mtod(m, struct ip *);
752	proto = ip->ip_p;
753
754	if (proto != IPPROTO_IPV6) {
755		m_freem(m);
756		return;
757	}
758
759	ip = mtod(m, struct ip *);
760
761	sc = (struct stf_softc *)encap_getarg(m);
762
763	if (sc == NULL || (ifnet_flags(sc->sc_if) & IFF_UP) == 0) {
764		m_freem(m);
765		return;
766	}
767
768	ifp = sc->sc_if;
769
770#if MAC_LABEL
771	mac_mbuf_label_associate_ifnet(ifp, m);
772#endif
773
774	/*
775	 * perform sanity check against outer src/dst.
776	 * for source, perform ingress filter as well.
777	 */
778	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
779	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
780		m_freem(m);
781		return;
782	}
783
784	otos = ip->ip_tos;
785	mbuf_copydata(m, off, sizeof(ip6), &ip6);
786
787	/*
788	 * perform sanity check against inner src/dst.
789	 * for source, perform ingress filter as well.
790	 */
791	if (stf_checkaddr6(sc, &ip6.ip6_dst, NULL) < 0 ||
792	    stf_checkaddr6(sc, &ip6.ip6_src, m->m_pkthdr.rcvif) < 0) {
793		m_freem(m);
794		return;
795	}
796
797	itos = (ntohl(ip6.ip6_flow) >> 20) & 0xff;
798	if ((ifnet_flags(ifp) & IFF_LINK1) != 0)
799		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
800	else
801		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
802	ip6.ip6_flow &= ~htonl(0xff << 20);
803	ip6.ip6_flow |= htonl((u_int32_t)itos << 20);
804
805	m->m_pkthdr.rcvif = ifp;
806	mbuf_pkthdr_setheader(m, mbuf_data(m));
807	mbuf_adj(m, off);
808
809	if (ifp->if_bpf) {
810		/* We need to prepend the address family as a four byte field. */
811		u_int32_t af = AF_INET6;
812		bpf_tap_in(ifp, 0, m, &af, sizeof(af));
813	}
814
815	/*
816	 * Put the packet to the network layer input queue according to the
817	 * specified address family.
818	 * See net/if_gif.c for possible issues with packet processing
819	 * reorder due to extra queueing.
820	 */
821	bzero(&stats, sizeof(stats));
822	stats.packets_in = 1;
823	stats.bytes_in = mbuf_pkthdr_len(m);
824	mbuf_pkthdr_setrcvif(m, ifp);
825	ifnet_input(ifp, m, &stats);
826
827	return;
828}
829
830static void
831stf_rtrequest(
832	__unused int cmd,
833	struct rtentry *rt,
834	__unused struct sockaddr *sa)
835{
836	if (rt != NULL) {
837		RT_LOCK_ASSERT_HELD(rt);
838		rt->rt_rmx.rmx_mtu = IPV6_MMTU;
839	}
840}
841
842static errno_t
843stf_ioctl(
844	ifnet_t		ifp,
845	u_long		cmd,
846	void		*data)
847{
848	struct ifaddr *ifa;
849	struct ifreq *ifr;
850	struct sockaddr_in6 *sin6;
851	int error;
852
853	error = 0;
854	switch (cmd) {
855	case SIOCSIFADDR:
856		ifa = (struct ifaddr *)data;
857		if (ifa == NULL) {
858			error = EAFNOSUPPORT;
859			break;
860		}
861		IFA_LOCK(ifa);
862		if (ifa->ifa_addr->sa_family != AF_INET6) {
863			IFA_UNLOCK(ifa);
864			error = EAFNOSUPPORT;
865			break;
866		}
867		sin6 = (struct sockaddr_in6 *)(void *)ifa->ifa_addr;
868		if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
869                        if ( !(ifnet_flags( ifp ) & IFF_UP) ) {
870                                /* do this only if the interface is not already up */
871				ifa->ifa_rtrequest = stf_rtrequest;
872				IFA_UNLOCK(ifa);
873				ifnet_set_flags(ifp, IFF_UP, IFF_UP);
874			} else {
875				IFA_UNLOCK(ifa);
876			}
877		} else {
878			IFA_UNLOCK(ifa);
879			error = EINVAL;
880		}
881		IFA_LOCK_ASSERT_NOTHELD(ifa);
882		break;
883
884	case SIOCADDMULTI:
885	case SIOCDELMULTI:
886		ifr = (struct ifreq *)data;
887		if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
888			;
889		else
890			error = EAFNOSUPPORT;
891		break;
892
893	default:
894		error = EOPNOTSUPP;
895		break;
896	}
897
898	return error;
899}
900