162587Sitojun/*	$FreeBSD: stable/11/sys/net/if_stf.c 340951 2018-11-26 11:17:12Z eugen $	*/
295023Ssuz/*	$KAME: if_stf.c,v 1.73 2001/12/03 11:08:30 keiichi Exp $	*/
362587Sitojun
4139823Simp/*-
562587Sitojun * Copyright (C) 2000 WIDE Project.
662587Sitojun * All rights reserved.
762587Sitojun *
862587Sitojun * Redistribution and use in source and binary forms, with or without
962587Sitojun * modification, are permitted provided that the following conditions
1062587Sitojun * are met:
1162587Sitojun * 1. Redistributions of source code must retain the above copyright
1262587Sitojun *    notice, this list of conditions and the following disclaimer.
1362587Sitojun * 2. Redistributions in binary form must reproduce the above copyright
1462587Sitojun *    notice, this list of conditions and the following disclaimer in the
1562587Sitojun *    documentation and/or other materials provided with the distribution.
1662587Sitojun * 3. Neither the name of the project nor the names of its contributors
1762587Sitojun *    may be used to endorse or promote products derived from this software
1862587Sitojun *    without specific prior written permission.
1962587Sitojun *
2062587Sitojun * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
2162587Sitojun * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
2262587Sitojun * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
2362587Sitojun * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
2462587Sitojun * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
2562587Sitojun * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
2662587Sitojun * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
2762587Sitojun * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
2862587Sitojun * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
2962587Sitojun * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
3062587Sitojun * SUCH DAMAGE.
3162587Sitojun */
3262587Sitojun
3362587Sitojun/*
3478064Sume * 6to4 interface, based on RFC3056.
3562587Sitojun *
3662587Sitojun * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
3762587Sitojun * There is no address mapping defined from IPv6 multicast address to IPv4
3862587Sitojun * address.  Therefore, we do not have IFF_MULTICAST on the interface.
3962587Sitojun *
4062587Sitojun * Due to the lack of address mapping for link-local addresses, we cannot
4162587Sitojun * throw packets toward link-local addresses (fe80::x).  Also, we cannot throw
4262587Sitojun * packets to link-local multicast addresses (ff02::x).
4362587Sitojun *
4462587Sitojun * Here are interesting symptoms due to the lack of link-local address:
4562587Sitojun *
4662587Sitojun * Unicast routing exchange:
4762587Sitojun * - RIPng: Impossible.  Uses link-local multicast packet toward ff02::9,
4862587Sitojun *   and link-local addresses as nexthop.
4962587Sitojun * - OSPFv6: Impossible.  OSPFv6 assumes that there's link-local address
5062587Sitojun *   assigned to the link, and makes use of them.  Also, HELLO packets use
5162587Sitojun *   link-local multicast addresses (ff02::5 and ff02::6).
5262587Sitojun * - BGP4+: Maybe.  You can only use global address as nexthop, and global
5362587Sitojun *   address as TCP endpoint address.
5462587Sitojun *
5562587Sitojun * Multicast routing protocols:
5662587Sitojun * - PIM: Hello packet cannot be used to discover adjacent PIM routers.
5762587Sitojun *   Adjacent PIM routers must be configured manually (is it really spec-wise
5862587Sitojun *   correct thing to do?).
5962587Sitojun *
6062587Sitojun * ICMPv6:
6162587Sitojun * - Redirects cannot be used due to the lack of link-local address.
6262587Sitojun *
6378064Sume * stf interface does not have, and will not need, a link-local address.
6478064Sume * It seems to have no real benefit and does not help the above symptoms much.
6578064Sume * Even if we assign link-locals to interface, we cannot really
6678064Sume * use link-local unicast/multicast on top of 6to4 cloud (since there's no
6778064Sume * encapsulation defined for link-local address), and the above analysis does
6878064Sume * not change.  RFC3056 does not mandate the assignment of link-local address
6978064Sume * either.
7062587Sitojun *
7162587Sitojun * 6to4 interface has security issues.  Refer to
7262587Sitojun * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
7362587Sitojun * for details.  The code tries to filter out some of malicious packets.
7462587Sitojun * Note that there is no way to be 100% secure.
7562587Sitojun */
7662587Sitojun
7762587Sitojun#include <sys/param.h>
7862587Sitojun#include <sys/systm.h>
7962587Sitojun#include <sys/socket.h>
8062587Sitojun#include <sys/sockio.h>
8162587Sitojun#include <sys/mbuf.h>
8262587Sitojun#include <sys/errno.h>
8383655Sbrooks#include <sys/kernel.h>
84286001Sae#include <sys/lock.h>
85129880Sphk#include <sys/module.h>
8662587Sitojun#include <sys/protosw.h>
87178888Sjulian#include <sys/proc.h>
8883655Sbrooks#include <sys/queue.h>
89286001Sae#include <sys/rmlock.h>
90183351Sdwmalone#include <sys/sysctl.h>
9162587Sitojun#include <machine/cpu.h>
9262587Sitojun
9362587Sitojun#include <sys/malloc.h>
9462587Sitojun
9562587Sitojun#include <net/if.h>
96257176Sglebius#include <net/if_var.h>
97130933Sbrooks#include <net/if_clone.h>
9862587Sitojun#include <net/route.h>
9962587Sitojun#include <net/netisr.h>
10062587Sitojun#include <net/if_types.h>
101196019Srwatson#include <net/vnet.h>
10262587Sitojun
10362587Sitojun#include <netinet/in.h>
104292331Smelifaro#include <netinet/in_fib.h>
10562587Sitojun#include <netinet/in_systm.h>
10662587Sitojun#include <netinet/ip.h>
10762587Sitojun#include <netinet/ip_var.h>
10862587Sitojun#include <netinet/in_var.h>
10962587Sitojun
11062587Sitojun#include <netinet/ip6.h>
11162587Sitojun#include <netinet6/ip6_var.h>
11262587Sitojun#include <netinet6/in6_var.h>
11362587Sitojun#include <netinet/ip_ecn.h>
11462587Sitojun
11562587Sitojun#include <netinet/ip_encap.h>
11662587Sitojun
11762587Sitojun#include <machine/stdarg.h>
11862587Sitojun
11962587Sitojun#include <net/bpf.h>
12062587Sitojun
121163606Srwatson#include <security/mac/mac_framework.h>
122163606Srwatson
123183351SdwmaloneSYSCTL_DECL(_net_link);
124227309Sedstatic SYSCTL_NODE(_net_link, IFT_STF, stf, CTLFLAG_RW, 0, "6to4 Interface");
125183351Sdwmalone
126244750Saestatic int stf_permit_rfc1918 = 0;
127267992ShselaskySYSCTL_INT(_net_link_stf, OID_AUTO, permit_rfc1918, CTLFLAG_RWTUN,
128244750Sae    &stf_permit_rfc1918, 0, "Permit the use of private IPv4 addresses");
129244750Sae
130130933Sbrooks#define STFUNIT		0
13183655Sbrooks
13262587Sitojun#define IN6_IS_ADDR_6TO4(x)	(ntohs((x)->s6_addr16[0]) == 0x2002)
13362587Sitojun
134109322Ssuz/*
135109322Ssuz * XXX: Return a pointer with 16-bit aligned.  Don't cast it to
136109322Ssuz * struct in_addr *; use bcopy() instead.
137109322Ssuz */
138249925Sglebius#define GET_V4(x)	(&(x)->s6_addr16[1])
139109322Ssuz
14062587Sitojunstruct stf_softc {
141147256Sbrooks	struct ifnet	*sc_ifp;
142178888Sjulian	u_int	sc_fibnum;
14362587Sitojun	const struct encaptab *encap_cookie;
14462587Sitojun};
145147256Sbrooks#define STF2IFP(sc)	((sc)->sc_ifp)
14662587Sitojun
147241610Sglebiusstatic const char stfname[] = "stf";
148241610Sglebius
149241610Sglebiusstatic MALLOC_DEFINE(M_STF, stfname, "6to4 Tunnel Interface");
150126709Srwatsonstatic const int ip_stf_ttl = 40;
15162587Sitojun
15279106Sbrooksextern  struct domain inetdomain;
153276152Saestatic int in_stf_input(struct mbuf **, int *, int);
154276152Saestatic struct protosw in_stf_protosw = {
155152242Sru	.pr_type =		SOCK_RAW,
156152242Sru	.pr_domain =		&inetdomain,
157152242Sru	.pr_protocol =		IPPROTO_IPV6,
158152242Sru	.pr_flags =		PR_ATOMIC|PR_ADDR,
159152242Sru	.pr_input =		in_stf_input,
160270008Skevlo	.pr_output =		rip_output,
161152242Sru	.pr_ctloutput =		rip_ctloutput,
162152242Sru	.pr_usrreqs =		&rip_usrreqs
16379106Sbrooks};
16462587Sitojun
165130933Sbrooksstatic char *stfnames[] = {"stf0", "stf", "6to4", NULL};
166130933Sbrooks
16792725Salfredstatic int stfmodevent(module_t, int, void *);
16892725Salfredstatic int stf_encapcheck(const struct mbuf *, int, int, void *);
169273220Smelifarostatic int stf_getsrcifa6(struct ifnet *, struct in6_addr *, struct in6_addr *);
170249925Sglebiusstatic int stf_output(struct ifnet *, struct mbuf *, const struct sockaddr *,
171191148Skmacy	struct route *);
172103475Sumestatic int isrfc1918addr(struct in_addr *);
17392725Salfredstatic int stf_checkaddr4(struct stf_softc *, struct in_addr *,
17492725Salfred	struct ifnet *);
17592725Salfredstatic int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
17692725Salfred	struct ifnet *);
17792725Salfredstatic int stf_ioctl(struct ifnet *, u_long, caddr_t);
17862587Sitojun
179130933Sbrooksstatic int stf_clone_match(struct if_clone *, const char *);
180160195Ssamstatic int stf_clone_create(struct if_clone *, char *, size_t, caddr_t);
181130933Sbrooksstatic int stf_clone_destroy(struct if_clone *, struct ifnet *);
182241610Sglebiusstatic struct if_clone *stf_cloner;
18383655Sbrooks
184130933Sbrooksstatic int
185130933Sbrooksstf_clone_match(struct if_clone *ifc, const char *name)
186130933Sbrooks{
187130933Sbrooks	int i;
18883655Sbrooks
189130933Sbrooks	for(i = 0; stfnames[i] != NULL; i++) {
190130933Sbrooks		if (strcmp(stfnames[i], name) == 0)
191130933Sbrooks			return (1);
192130933Sbrooks	}
193130933Sbrooks
194130933Sbrooks	return (0);
195130933Sbrooks}
196130933Sbrooks
197128209Sbrooksstatic int
198160195Ssamstf_clone_create(struct if_clone *ifc, char *name, size_t len, caddr_t params)
19983655Sbrooks{
200337855Sloos	char *dp;
201337855Sloos	int err, unit, wildcard;
20283655Sbrooks	struct stf_softc *sc;
203128417Sbrooks	struct ifnet *ifp;
20483655Sbrooks
205337855Sloos	err = ifc_name2unit(name, &unit);
206337855Sloos	if (err != 0)
207337855Sloos		return (err);
208337855Sloos	wildcard = (unit < 0);
209337855Sloos
210130933Sbrooks	/*
211130933Sbrooks	 * We can only have one unit, but since unit allocation is
212130933Sbrooks	 * already locked, we use it to keep from allocating extra
213130933Sbrooks	 * interfaces.
214130933Sbrooks	 */
215130933Sbrooks	unit = STFUNIT;
216130933Sbrooks	err = ifc_alloc_unit(ifc, &unit);
217130933Sbrooks	if (err != 0)
218130933Sbrooks		return (err);
219130933Sbrooks
220111119Simp	sc = malloc(sizeof(struct stf_softc), M_STF, M_WAITOK | M_ZERO);
221147346Sbrooks	ifp = STF2IFP(sc) = if_alloc(IFT_STF);
222147256Sbrooks	if (ifp == NULL) {
223147256Sbrooks		free(sc, M_STF);
224147256Sbrooks		ifc_free_unit(ifc, unit);
225147256Sbrooks		return (ENOSPC);
226147256Sbrooks	}
227147346Sbrooks	ifp->if_softc = sc;
228178888Sjulian	sc->sc_fibnum = curthread->td_proc->p_fibnum;
229147346Sbrooks
230130933Sbrooks	/*
231130933Sbrooks	 * Set the name manually rather then using if_initname because
232130933Sbrooks	 * we don't conform to the default naming convention for interfaces.
233337855Sloos	 * In the wildcard case, we need to update the name.
234130933Sbrooks	 */
235337855Sloos	if (wildcard) {
236337855Sloos		for (dp = name; *dp != '\0'; dp++);
237337855Sloos		if (snprintf(dp, len - (dp-name), "%d", unit) >
238337855Sloos		    len - (dp-name) - 1) {
239337855Sloos			/*
240337855Sloos			 * This can only be a programmer error and
241337855Sloos			 * there's no straightforward way to recover if
242337855Sloos			 * it happens.
243337855Sloos			 */
244337855Sloos			panic("if_clone_create(): interface name too long");
245337855Sloos		}
246337855Sloos	}
247130933Sbrooks	strlcpy(ifp->if_xname, name, IFNAMSIZ);
248241610Sglebius	ifp->if_dname = stfname;
249130933Sbrooks	ifp->if_dunit = IF_DUNIT_NONE;
25083655Sbrooks
25183655Sbrooks	sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
25283655Sbrooks	    stf_encapcheck, &in_stf_protosw, sc);
25383655Sbrooks	if (sc->encap_cookie == NULL) {
254128417Sbrooks		if_printf(ifp, "attach failed\n");
25583655Sbrooks		free(sc, M_STF);
256130933Sbrooks		ifc_free_unit(ifc, unit);
25783655Sbrooks		return (ENOMEM);
25883655Sbrooks	}
25983655Sbrooks
260128417Sbrooks	ifp->if_mtu    = IPV6_MMTU;
261128417Sbrooks	ifp->if_ioctl  = stf_ioctl;
262128417Sbrooks	ifp->if_output = stf_output;
263207554Ssobomax	ifp->if_snd.ifq_maxlen = ifqmaxlen;
264128417Sbrooks	if_attach(ifp);
265147611Sdwmalone	bpfattach(ifp, DLT_NULL, sizeof(u_int32_t));
26683655Sbrooks	return (0);
26783655Sbrooks}
26883655Sbrooks
269130933Sbrooksstatic int
270130933Sbrooksstf_clone_destroy(struct if_clone *ifc, struct ifnet *ifp)
27183655Sbrooks{
272147256Sbrooks	struct stf_softc *sc = ifp->if_softc;
273151266Sthompsa	int err;
27483655Sbrooks
275151266Sthompsa	err = encap_detach(sc->encap_cookie);
276151266Sthompsa	KASSERT(err == 0, ("Unexpected error detaching encap_cookie"));
277151266Sthompsa	bpfdetach(ifp);
278151266Sthompsa	if_detach(ifp);
279151266Sthompsa	if_free(ifp);
280151266Sthompsa
281151266Sthompsa	free(sc, M_STF);
282130933Sbrooks	ifc_free_unit(ifc, STFUNIT);
283130933Sbrooks
284130933Sbrooks	return (0);
28583655Sbrooks}
28683655Sbrooks
28779106Sbrooksstatic int
288286114Saestfmodevent(module_t mod, int type, void *data)
28962587Sitojun{
29062587Sitojun
29179106Sbrooks	switch (type) {
29279106Sbrooks	case MOD_LOAD:
293241610Sglebius		stf_cloner = if_clone_advanced(stfname, 0, stf_clone_match,
294241610Sglebius		    stf_clone_create, stf_clone_destroy);
29579106Sbrooks		break;
29679106Sbrooks	case MOD_UNLOAD:
297241610Sglebius		if_clone_detach(stf_cloner);
29879106Sbrooks		break;
299132199Sphk	default:
300132199Sphk		return (EOPNOTSUPP);
30162587Sitojun	}
30279106Sbrooks
30379106Sbrooks	return (0);
30462587Sitojun}
30562587Sitojun
30679106Sbrooksstatic moduledata_t stf_mod = {
30779106Sbrooks	"if_stf",
30879106Sbrooks	stfmodevent,
309241394Skevlo	0
31079106Sbrooks};
31162587Sitojun
31279106SbrooksDECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
31379106Sbrooks
31462587Sitojunstatic int
315286114Saestf_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
31662587Sitojun{
31762587Sitojun	struct ip ip;
31862587Sitojun	struct stf_softc *sc;
319108710Sfenner	struct in_addr a, b, mask;
320273220Smelifaro	struct in6_addr addr6, mask6;
32162587Sitojun
32262587Sitojun	sc = (struct stf_softc *)arg;
32362587Sitojun	if (sc == NULL)
32462587Sitojun		return 0;
32562587Sitojun
326147256Sbrooks	if ((STF2IFP(sc)->if_flags & IFF_UP) == 0)
32762587Sitojun		return 0;
32862587Sitojun
32978064Sume	/* IFF_LINK0 means "no decapsulation" */
330147256Sbrooks	if ((STF2IFP(sc)->if_flags & IFF_LINK0) != 0)
33178064Sume		return 0;
33278064Sume
33362587Sitojun	if (proto != IPPROTO_IPV6)
33462587Sitojun		return 0;
33562587Sitojun
33662587Sitojun	/* LINTED const cast */
33791452Speter	m_copydata((struct mbuf *)(uintptr_t)m, 0, sizeof(ip), (caddr_t)&ip);
33862587Sitojun
33962587Sitojun	if (ip.ip_v != 4)
34062587Sitojun		return 0;
34162587Sitojun
342273220Smelifaro	if (stf_getsrcifa6(STF2IFP(sc), &addr6, &mask6) != 0)
343273220Smelifaro		return (0);
34462587Sitojun
34562587Sitojun	/*
34662587Sitojun	 * check if IPv4 dst matches the IPv4 address derived from the
34762587Sitojun	 * local 6to4 address.
34862587Sitojun	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
34962587Sitojun	 */
350273220Smelifaro	if (bcmp(GET_V4(&addr6), &ip.ip_dst, sizeof(ip.ip_dst)) != 0)
35162587Sitojun		return 0;
35262587Sitojun
35362587Sitojun	/*
35462587Sitojun	 * check if IPv4 src matches the IPv4 address derived from the
35562587Sitojun	 * local 6to4 address masked by prefixmask.
35662587Sitojun	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
35762587Sitojun	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
35862587Sitojun	 */
35962587Sitojun	bzero(&a, sizeof(a));
360273220Smelifaro	bcopy(GET_V4(&addr6), &a, sizeof(a));
361273220Smelifaro	bcopy(GET_V4(&mask6), &mask, sizeof(mask));
362108710Sfenner	a.s_addr &= mask.s_addr;
36362587Sitojun	b = ip.ip_src;
364108710Sfenner	b.s_addr &= mask.s_addr;
36562587Sitojun	if (a.s_addr != b.s_addr)
36662587Sitojun		return 0;
36762587Sitojun
36862587Sitojun	/* stf interface makes single side match only */
36962587Sitojun	return 32;
37062587Sitojun}
37162587Sitojun
372273220Smelifarostatic int
373273220Smelifarostf_getsrcifa6(struct ifnet *ifp, struct in6_addr *addr, struct in6_addr *mask)
37462587Sitojun{
375340951Seugen	struct rm_priotracker in_ifa_tracker;
37662587Sitojun	struct ifaddr *ia;
37762587Sitojun	struct in_ifaddr *ia4;
378273220Smelifaro	struct in6_ifaddr *ia6;
37962587Sitojun	struct sockaddr_in6 *sin6;
38062587Sitojun	struct in_addr in;
38162587Sitojun
382195022Srwatson	if_addr_rlock(ifp);
383191339Srwatson	TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_link) {
38462587Sitojun		if (ia->ifa_addr->sa_family != AF_INET6)
38562587Sitojun			continue;
38662587Sitojun		sin6 = (struct sockaddr_in6 *)ia->ifa_addr;
38762587Sitojun		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
38862587Sitojun			continue;
38962587Sitojun
39062587Sitojun		bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
391340951Seugen		IN_IFADDR_RLOCK(&in_ifa_tracker);
39284104Sjlemon		LIST_FOREACH(ia4, INADDR_HASH(in.s_addr), ia_hash)
39362587Sitojun			if (ia4->ia_addr.sin_addr.s_addr == in.s_addr)
39462587Sitojun				break;
395340951Seugen		IN_IFADDR_RUNLOCK(&in_ifa_tracker);
39662587Sitojun		if (ia4 == NULL)
39762587Sitojun			continue;
39862587Sitojun
399273220Smelifaro		ia6 = (struct in6_ifaddr *)ia;
400273220Smelifaro
401273220Smelifaro		*addr = sin6->sin6_addr;
402273220Smelifaro		*mask = ia6->ia_prefixmask.sin6_addr;
403195022Srwatson		if_addr_runlock(ifp);
404273220Smelifaro		return (0);
40562587Sitojun	}
406195022Srwatson	if_addr_runlock(ifp);
40762587Sitojun
408273220Smelifaro	return (ENOENT);
40962587Sitojun}
41062587Sitojun
41162587Sitojunstatic int
412249925Sglebiusstf_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
413286114Sae    struct route *ro)
41462587Sitojun{
41562587Sitojun	struct stf_softc *sc;
416249925Sglebius	const struct sockaddr_in6 *dst6;
417109322Ssuz	struct in_addr in4;
418249925Sglebius	const void *ptr;
41962587Sitojun	u_int8_t tos;
42062587Sitojun	struct ip *ip;
42162587Sitojun	struct ip6_hdr *ip6;
422273220Smelifaro	struct in6_addr addr6, mask6;
423105580Srwatson	int error;
42462587Sitojun
425183351Sdwmalone#ifdef MAC
426172930Srwatson	error = mac_ifnet_check_transmit(ifp, m);
427105580Srwatson	if (error) {
428105580Srwatson		m_freem(m);
429105580Srwatson		return (error);
430105580Srwatson	}
431105580Srwatson#endif
432105580Srwatson
433147256Sbrooks	sc = ifp->if_softc;
434249925Sglebius	dst6 = (const struct sockaddr_in6 *)dst;
43562587Sitojun
43662587Sitojun	/* just in case */
43762587Sitojun	if ((ifp->if_flags & IFF_UP) == 0) {
43862587Sitojun		m_freem(m);
439271867Sglebius		if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
44062587Sitojun		return ENETDOWN;
44162587Sitojun	}
44262587Sitojun
44362587Sitojun	/*
44462587Sitojun	 * If we don't have an ip4 address that match my inner ip6 address,
44562587Sitojun	 * we shouldn't generate output.  Without this check, we'll end up
44662587Sitojun	 * using wrong IPv4 source.
44762587Sitojun	 */
448273220Smelifaro	if (stf_getsrcifa6(ifp, &addr6, &mask6) != 0) {
44962587Sitojun		m_freem(m);
450271867Sglebius		if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
45162587Sitojun		return ENETDOWN;
45262587Sitojun	}
45362587Sitojun
45462587Sitojun	if (m->m_len < sizeof(*ip6)) {
45562587Sitojun		m = m_pullup(m, sizeof(*ip6));
456103487Sume		if (!m) {
457271867Sglebius			if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
45862587Sitojun			return ENOBUFS;
459103487Sume		}
46062587Sitojun	}
46162587Sitojun	ip6 = mtod(m, struct ip6_hdr *);
46262587Sitojun	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
46362587Sitojun
46478064Sume	/*
46578064Sume	 * Pickup the right outer dst addr from the list of candidates.
46678064Sume	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
46778064Sume	 */
468109322Ssuz	ptr = NULL;
46978064Sume	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
470109322Ssuz		ptr = GET_V4(&ip6->ip6_dst);
47178064Sume	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
472109322Ssuz		ptr = GET_V4(&dst6->sin6_addr);
47378064Sume	else {
47478064Sume		m_freem(m);
475271867Sglebius		if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
47678064Sume		return ENETUNREACH;
47778064Sume	}
478109322Ssuz	bcopy(ptr, &in4, sizeof(in4));
47978064Sume
480159180Scsjp	if (bpf_peers_present(ifp->if_bpf)) {
48178701Sume		/*
48278701Sume		 * We need to prepend the address family as
48378701Sume		 * a four byte field.  Cons up a dummy header
48478701Sume		 * to pacify bpf.  This is safe because bpf
48578701Sume		 * will only read from the mbuf (i.e., it won't
48678701Sume		 * try to free it or keep a pointer a to it).
48778701Sume		 */
488249925Sglebius		u_int af = AF_INET6;
489123922Ssam		bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m);
49078701Sume	}
49178701Sume
492243882Sglebius	M_PREPEND(m, sizeof(struct ip), M_NOWAIT);
493103487Sume	if (m == NULL) {
494271867Sglebius		if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
49562587Sitojun		return ENOBUFS;
496103487Sume	}
49762587Sitojun	ip = mtod(m, struct ip *);
49862587Sitojun
49962587Sitojun	bzero(ip, sizeof(*ip));
50062587Sitojun
501273220Smelifaro	bcopy(GET_V4(&addr6), &ip->ip_src, sizeof(ip->ip_src));
502109322Ssuz	bcopy(&in4, &ip->ip_dst, sizeof(ip->ip_dst));
50362587Sitojun	ip->ip_p = IPPROTO_IPV6;
50479106Sbrooks	ip->ip_ttl = ip_stf_ttl;
505241913Sglebius	ip->ip_len = htons(m->m_pkthdr.len);
50662587Sitojun	if (ifp->if_flags & IFF_LINK1)
50762587Sitojun		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
50878064Sume	else
50978064Sume		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
51062587Sitojun
511178888Sjulian	M_SETFIB(m, sc->sc_fibnum);
512271867Sglebius	if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
513273220Smelifaro	error = ip_output(m, NULL, NULL, 0, NULL, NULL);
514183351Sdwmalone
515183351Sdwmalone	return error;
51662587Sitojun}
51762587Sitojun
51862587Sitojunstatic int
519286114Saeisrfc1918addr(struct in_addr *in)
520103475Sume{
521103475Sume	/*
522103475Sume	 * returns 1 if private address range:
523103475Sume	 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
524103475Sume	 */
525244750Sae	if (stf_permit_rfc1918 == 0 && (
526244750Sae	    (ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
527109322Ssuz	    (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
528244750Sae	    (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168))
529103475Sume		return 1;
530103475Sume
531103475Sume	return 0;
532103475Sume}
533103475Sume
534103475Sumestatic int
535286114Saestf_checkaddr4(struct stf_softc *sc, struct in_addr *in, struct ifnet *inifp)
53662587Sitojun{
537286001Sae	struct rm_priotracker in_ifa_tracker;
53862587Sitojun	struct in_ifaddr *ia4;
53962587Sitojun
54062587Sitojun	/*
54162587Sitojun	 * reject packets with the following address:
54262587Sitojun	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
54362587Sitojun	 */
544109322Ssuz	if (IN_MULTICAST(ntohl(in->s_addr)))
54562587Sitojun		return -1;
546109322Ssuz	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
547109322Ssuz	case 0: case 127: case 255:
54862587Sitojun		return -1;
549109322Ssuz	}
55062587Sitojun
55162587Sitojun	/*
552103475Sume	 * reject packets with private address range.
553103475Sume	 * (requirement from RFC3056 section 2 1st paragraph)
554103475Sume	 */
555103475Sume	if (isrfc1918addr(in))
556103475Sume		return -1;
557103475Sume
558103475Sume	/*
55962587Sitojun	 * reject packets with broadcast
56062587Sitojun	 */
561286001Sae	IN_IFADDR_RLOCK(&in_ifa_tracker);
562239357Sjhb	TAILQ_FOREACH(ia4, &V_in_ifaddrhead, ia_link) {
56362587Sitojun		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
56462587Sitojun			continue;
565194951Srwatson		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr) {
566286001Sae			IN_IFADDR_RUNLOCK(&in_ifa_tracker);
56762587Sitojun			return -1;
568194951Srwatson		}
56962587Sitojun	}
570286001Sae	IN_IFADDR_RUNLOCK(&in_ifa_tracker);
57162587Sitojun
57262587Sitojun	/*
57362587Sitojun	 * perform ingress filter
57462587Sitojun	 */
575147256Sbrooks	if (sc && (STF2IFP(sc)->if_flags & IFF_LINK2) == 0 && inifp) {
576292331Smelifaro		struct nhop4_basic nh4;
57762587Sitojun
578292331Smelifaro		if (fib4_lookup_nh_basic(sc->sc_fibnum, *in, 0, 0, &nh4) != 0)
579292331Smelifaro			return (-1);
580292331Smelifaro
581292331Smelifaro		if (nh4.nh_ifp != inifp)
582292331Smelifaro			return (-1);
58362587Sitojun	}
58462587Sitojun
58562587Sitojun	return 0;
58662587Sitojun}
58762587Sitojun
58862587Sitojunstatic int
589286114Saestf_checkaddr6(struct stf_softc *sc, struct in6_addr *in6, struct ifnet *inifp)
59062587Sitojun{
59162587Sitojun	/*
59262587Sitojun	 * check 6to4 addresses
59362587Sitojun	 */
594109322Ssuz	if (IN6_IS_ADDR_6TO4(in6)) {
595109322Ssuz		struct in_addr in4;
596109322Ssuz		bcopy(GET_V4(in6), &in4, sizeof(in4));
597109322Ssuz		return stf_checkaddr4(sc, &in4, inifp);
598109322Ssuz	}
59962587Sitojun
60062587Sitojun	/*
60162587Sitojun	 * reject anything that look suspicious.  the test is implemented
60262587Sitojun	 * in ip6_input too, but we check here as well to
60362587Sitojun	 * (1) reject bad packets earlier, and
60462587Sitojun	 * (2) to be safe against future ip6_input change.
60562587Sitojun	 */
60662587Sitojun	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
60762587Sitojun		return -1;
60862587Sitojun
60962587Sitojun	return 0;
61062587Sitojun}
61162587Sitojun
612276152Saestatic int
613269699Skevloin_stf_input(struct mbuf **mp, int *offp, int proto)
61462587Sitojun{
61562587Sitojun	struct stf_softc *sc;
61662587Sitojun	struct ip *ip;
61762587Sitojun	struct ip6_hdr *ip6;
618269699Skevlo	struct mbuf *m;
61962587Sitojun	u_int8_t otos, itos;
62062587Sitojun	struct ifnet *ifp;
621269699Skevlo	int off;
62262587Sitojun
623269699Skevlo	m = *mp;
624269699Skevlo	off = *offp;
62562587Sitojun
62662587Sitojun	if (proto != IPPROTO_IPV6) {
62762587Sitojun		m_freem(m);
628269699Skevlo		return (IPPROTO_DONE);
62962587Sitojun	}
63062587Sitojun
63162587Sitojun	ip = mtod(m, struct ip *);
63262587Sitojun
63362587Sitojun	sc = (struct stf_softc *)encap_getarg(m);
63462587Sitojun
635147256Sbrooks	if (sc == NULL || (STF2IFP(sc)->if_flags & IFF_UP) == 0) {
63662587Sitojun		m_freem(m);
637269699Skevlo		return (IPPROTO_DONE);
63862587Sitojun	}
63962587Sitojun
640147256Sbrooks	ifp = STF2IFP(sc);
64162587Sitojun
642105580Srwatson#ifdef MAC
643172930Srwatson	mac_ifnet_create_mbuf(ifp, m);
644105580Srwatson#endif
645105580Srwatson
64662587Sitojun	/*
64762587Sitojun	 * perform sanity check against outer src/dst.
64862587Sitojun	 * for source, perform ingress filter as well.
64962587Sitojun	 */
65078064Sume	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
65178064Sume	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
65262587Sitojun		m_freem(m);
653269699Skevlo		return (IPPROTO_DONE);
65462587Sitojun	}
65562587Sitojun
65662587Sitojun	otos = ip->ip_tos;
65762587Sitojun	m_adj(m, off);
65862587Sitojun
65962587Sitojun	if (m->m_len < sizeof(*ip6)) {
66062587Sitojun		m = m_pullup(m, sizeof(*ip6));
66162587Sitojun		if (!m)
662269699Skevlo			return (IPPROTO_DONE);
66362587Sitojun	}
66462587Sitojun	ip6 = mtod(m, struct ip6_hdr *);
66562587Sitojun
66662587Sitojun	/*
66762587Sitojun	 * perform sanity check against inner src/dst.
66862587Sitojun	 * for source, perform ingress filter as well.
66962587Sitojun	 */
67078064Sume	if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
67178064Sume	    stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
67262587Sitojun		m_freem(m);
673269699Skevlo		return (IPPROTO_DONE);
67462587Sitojun	}
67562587Sitojun
67662587Sitojun	itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
67762587Sitojun	if ((ifp->if_flags & IFF_LINK1) != 0)
67862587Sitojun		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
67978064Sume	else
68078064Sume		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
68162587Sitojun	ip6->ip6_flow &= ~htonl(0xff << 20);
68262587Sitojun	ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
68362587Sitojun
68462587Sitojun	m->m_pkthdr.rcvif = ifp;
68562587Sitojun
686159180Scsjp	if (bpf_peers_present(ifp->if_bpf)) {
68762587Sitojun		/*
68862587Sitojun		 * We need to prepend the address family as
68962587Sitojun		 * a four byte field.  Cons up a dummy header
69062587Sitojun		 * to pacify bpf.  This is safe because bpf
69162587Sitojun		 * will only read from the mbuf (i.e., it won't
69262587Sitojun		 * try to free it or keep a pointer a to it).
69362587Sitojun		 */
694123922Ssam		u_int32_t af = AF_INET6;
695140042Sume		bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m);
69662587Sitojun	}
69762587Sitojun
69862587Sitojun	/*
69962587Sitojun	 * Put the packet to the network layer input queue according to the
70062587Sitojun	 * specified address family.
70162587Sitojun	 * See net/if_gif.c for possible issues with packet processing
70262587Sitojun	 * reorder due to extra queueing.
70362587Sitojun	 */
704271867Sglebius	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
705271867Sglebius	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
706223741Sbz	M_SETFIB(m, ifp->if_fib);
707111888Sjlemon	netisr_dispatch(NETISR_IPV6, m);
708269699Skevlo	return (IPPROTO_DONE);
70962587Sitojun}
71062587Sitojun
71162587Sitojunstatic int
712286114Saestf_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
71362587Sitojun{
71462587Sitojun	struct ifaddr *ifa;
71562587Sitojun	struct ifreq *ifr;
71662587Sitojun	struct sockaddr_in6 *sin6;
717109322Ssuz	struct in_addr addr;
718238492Smelifaro	int error, mtu;
71962587Sitojun
72062587Sitojun	error = 0;
72162587Sitojun	switch (cmd) {
72262587Sitojun	case SIOCSIFADDR:
72362587Sitojun		ifa = (struct ifaddr *)data;
72462587Sitojun		if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
72562587Sitojun			error = EAFNOSUPPORT;
72662587Sitojun			break;
72762587Sitojun		}
72862587Sitojun		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
729109322Ssuz		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
73062587Sitojun			error = EINVAL;
731109322Ssuz			break;
732109322Ssuz		}
733109322Ssuz		bcopy(GET_V4(&sin6->sin6_addr), &addr, sizeof(addr));
734109322Ssuz		if (isrfc1918addr(&addr)) {
735109322Ssuz			error = EINVAL;
736109322Ssuz			break;
737109322Ssuz		}
738109322Ssuz
739109322Ssuz		ifp->if_flags |= IFF_UP;
74062587Sitojun		break;
74162587Sitojun
74262587Sitojun	case SIOCADDMULTI:
74362587Sitojun	case SIOCDELMULTI:
74462587Sitojun		ifr = (struct ifreq *)data;
74562587Sitojun		if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
74662587Sitojun			;
74762587Sitojun		else
74862587Sitojun			error = EAFNOSUPPORT;
74962587Sitojun		break;
75062587Sitojun
751238492Smelifaro	case SIOCGIFMTU:
752238492Smelifaro		break;
753238492Smelifaro
754238492Smelifaro	case SIOCSIFMTU:
755238492Smelifaro		ifr = (struct ifreq *)data;
756238492Smelifaro		mtu = ifr->ifr_mtu;
757238492Smelifaro		/* RFC 4213 3.2 ideal world MTU */
758238492Smelifaro		if (mtu < IPV6_MINMTU || mtu > IF_MAXMTU - 20)
759238492Smelifaro			return (EINVAL);
760238492Smelifaro		ifp->if_mtu = mtu;
761238492Smelifaro		break;
762238492Smelifaro
76362587Sitojun	default:
76462587Sitojun		error = EINVAL;
76562587Sitojun		break;
76662587Sitojun	}
76762587Sitojun
76862587Sitojun	return error;
76962587Sitojun}
770