raw_ip6.c revision 211501
1/*-
2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29
30/*-
31 * Copyright (c) 1982, 1986, 1988, 1993
32 *	The Regents of the University of California.
33 * All rights reserved.
34 *
35 * Redistribution and use in source and binary forms, with or without
36 * modification, are permitted provided that the following conditions
37 * are met:
38 * 1. Redistributions of source code must retain the above copyright
39 *    notice, this list of conditions and the following disclaimer.
40 * 2. Redistributions in binary form must reproduce the above copyright
41 *    notice, this list of conditions and the following disclaimer in the
42 *    documentation and/or other materials provided with the distribution.
43 * 4. Neither the name of the University nor the names of its contributors
44 *    may be used to endorse or promote products derived from this software
45 *    without specific prior written permission.
46 *
47 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
48 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
49 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
50 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
51 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
52 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
53 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
54 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
55 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
56 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
57 * SUCH DAMAGE.
58 *
59 *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
60 */
61
62#include <sys/cdefs.h>
63__FBSDID("$FreeBSD: head/sys/netinet6/raw_ip6.c 211501 2010-08-19 11:31:03Z anchie $");
64
65#include "opt_ipsec.h"
66#include "opt_inet6.h"
67
68#include <sys/param.h>
69#include <sys/errno.h>
70#include <sys/jail.h>
71#include <sys/lock.h>
72#include <sys/malloc.h>
73#include <sys/mbuf.h>
74#include <sys/priv.h>
75#include <sys/proc.h>
76#include <sys/protosw.h>
77#include <sys/signalvar.h>
78#include <sys/socket.h>
79#include <sys/socketvar.h>
80#include <sys/sx.h>
81#include <sys/syslog.h>
82
83#include <net/if.h>
84#include <net/if_types.h>
85#include <net/route.h>
86#include <net/vnet.h>
87
88#include <netinet/in.h>
89#include <netinet/in_var.h>
90#include <netinet/in_systm.h>
91#include <netinet/in_pcb.h>
92
93#include <netinet/icmp6.h>
94#include <netinet/ip6.h>
95#include <netinet/ip_var.h>
96#include <netinet6/ip6protosw.h>
97#include <netinet6/ip6_mroute.h>
98#include <netinet6/in6_pcb.h>
99#include <netinet6/ip6_var.h>
100#include <netinet6/nd6.h>
101#include <netinet6/raw_ip6.h>
102#include <netinet6/scope6_var.h>
103#include <netinet6/send.h>
104
105#ifdef IPSEC
106#include <netipsec/ipsec.h>
107#include <netipsec/ipsec6.h>
108#endif /* IPSEC */
109
110#include <machine/stdarg.h>
111
112#define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
113#define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
114
115/*
116 * Raw interface to IP6 protocol.
117 */
118
119VNET_DECLARE(struct inpcbhead, ripcb);
120VNET_DECLARE(struct inpcbinfo, ripcbinfo);
121#define	V_ripcb				VNET(ripcb)
122#define	V_ripcbinfo			VNET(ripcbinfo)
123
124extern u_long	rip_sendspace;
125extern u_long	rip_recvspace;
126
127VNET_DEFINE(struct rip6stat, rip6stat);
128
129/*
130 * Hooks for multicast routing. They all default to NULL, so leave them not
131 * initialized and rely on BSS being set to 0.
132 */
133
134/*
135 * The socket used to communicate with the multicast routing daemon.
136 */
137VNET_DEFINE(struct socket *, ip6_mrouter);
138
139/*
140 * The various mrouter functions.
141 */
142int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
143int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
144int (*ip6_mrouter_done)(void);
145int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
146int (*mrt6_ioctl)(u_long, caddr_t);
147
148/*
149 * Setup generic address and protocol structures for raw_input routine, then
150 * pass them along with mbuf chain.
151 */
152int
153rip6_input(struct mbuf **mp, int *offp, int proto)
154{
155	struct ifnet *ifp;
156	struct mbuf *m = *mp;
157	register struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
158	register struct inpcb *in6p;
159	struct inpcb *last = 0;
160	struct mbuf *opts = NULL;
161	struct sockaddr_in6 fromsa;
162
163	V_rip6stat.rip6s_ipackets++;
164
165	if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
166		/* XXX Send icmp6 host/port unreach? */
167		m_freem(m);
168		return (IPPROTO_DONE);
169	}
170
171	init_sin6(&fromsa, m); /* general init */
172
173	ifp = m->m_pkthdr.rcvif;
174
175	INP_INFO_RLOCK(&V_ripcbinfo);
176	LIST_FOREACH(in6p, &V_ripcb, inp_list) {
177		/* XXX inp locking */
178		if ((in6p->inp_vflag & INP_IPV6) == 0)
179			continue;
180		if (in6p->inp_ip_p &&
181		    in6p->inp_ip_p != proto)
182			continue;
183		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
184		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &ip6->ip6_dst))
185			continue;
186		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) &&
187		    !IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &ip6->ip6_src))
188			continue;
189		if (jailed_without_vnet(in6p->inp_cred)) {
190			/*
191			 * Allow raw socket in jail to receive multicast;
192			 * assume process had PRIV_NETINET_RAW at attach,
193			 * and fall through into normal filter path if so.
194			 */
195			if (!IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
196			    prison_check_ip6(in6p->inp_cred,
197			    &ip6->ip6_dst) != 0)
198				continue;
199		}
200		if (in6p->in6p_cksum != -1) {
201			V_rip6stat.rip6s_isum++;
202			if (in6_cksum(m, proto, *offp,
203			    m->m_pkthdr.len - *offp)) {
204				INP_RUNLOCK(in6p);
205				V_rip6stat.rip6s_badsum++;
206				continue;
207			}
208		}
209		INP_RLOCK(in6p);
210		/*
211		 * If this raw socket has multicast state, and we
212		 * have received a multicast, check if this socket
213		 * should receive it, as multicast filtering is now
214		 * the responsibility of the transport layer.
215		 */
216		if (in6p->in6p_moptions &&
217		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
218			/*
219			 * If the incoming datagram is for MLD, allow it
220			 * through unconditionally to the raw socket.
221			 *
222			 * Use the M_RTALERT_MLD flag to check for MLD
223			 * traffic without having to inspect the mbuf chain
224			 * more deeply, as all MLDv1/v2 host messages MUST
225			 * contain the Router Alert option.
226			 *
227			 * In the case of MLDv1, we may not have explicitly
228			 * joined the group, and may have set IFF_ALLMULTI
229			 * on the interface. im6o_mc_filter() may discard
230			 * control traffic we actually need to see.
231			 *
232			 * Userland multicast routing daemons should continue
233			 * filter the control traffic appropriately.
234			 */
235			int blocked;
236
237			blocked = MCAST_PASS;
238			if ((m->m_flags & M_RTALERT_MLD) == 0) {
239				struct sockaddr_in6 mcaddr;
240
241				bzero(&mcaddr, sizeof(struct sockaddr_in6));
242				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
243				mcaddr.sin6_family = AF_INET6;
244				mcaddr.sin6_addr = ip6->ip6_dst;
245
246				blocked = im6o_mc_filter(in6p->in6p_moptions,
247				    ifp,
248				    (struct sockaddr *)&mcaddr,
249				    (struct sockaddr *)&fromsa);
250			}
251			if (blocked != MCAST_PASS) {
252				IP6STAT_INC(ip6s_notmember);
253				INP_RUNLOCK(in6p);
254				continue;
255			}
256		}
257		if (last != NULL) {
258			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
259
260#ifdef IPSEC
261			/*
262			 * Check AH/ESP integrity.
263			 */
264			if (n && ipsec6_in_reject(n, last)) {
265				m_freem(n);
266				V_ipsec6stat.in_polvio++;
267				/* Do not inject data into pcb. */
268			} else
269#endif /* IPSEC */
270			if (n) {
271				if (last->inp_flags & INP_CONTROLOPTS ||
272				    last->inp_socket->so_options & SO_TIMESTAMP)
273					ip6_savecontrol(last, n, &opts);
274				/* strip intermediate headers */
275				m_adj(n, *offp);
276				if (sbappendaddr(&last->inp_socket->so_rcv,
277						(struct sockaddr *)&fromsa,
278						 n, opts) == 0) {
279					m_freem(n);
280					if (opts)
281						m_freem(opts);
282					V_rip6stat.rip6s_fullsock++;
283				} else
284					sorwakeup(last->inp_socket);
285				opts = NULL;
286			}
287			INP_RUNLOCK(last);
288		}
289		last = in6p;
290	}
291	INP_INFO_RUNLOCK(&V_ripcbinfo);
292#ifdef IPSEC
293	/*
294	 * Check AH/ESP integrity.
295	 */
296	if ((last != NULL) && ipsec6_in_reject(m, last)) {
297		m_freem(m);
298		V_ipsec6stat.in_polvio++;
299		V_ip6stat.ip6s_delivered--;
300		/* Do not inject data into pcb. */
301		INP_RUNLOCK(last);
302	} else
303#endif /* IPSEC */
304	if (last != NULL) {
305		if (last->inp_flags & INP_CONTROLOPTS ||
306		    last->inp_socket->so_options & SO_TIMESTAMP)
307			ip6_savecontrol(last, m, &opts);
308		/* Strip intermediate headers. */
309		m_adj(m, *offp);
310		if (sbappendaddr(&last->inp_socket->so_rcv,
311		    (struct sockaddr *)&fromsa, m, opts) == 0) {
312			m_freem(m);
313			if (opts)
314				m_freem(opts);
315			V_rip6stat.rip6s_fullsock++;
316		} else
317			sorwakeup(last->inp_socket);
318		INP_RUNLOCK(last);
319	} else {
320		V_rip6stat.rip6s_nosock++;
321		if (m->m_flags & M_MCAST)
322			V_rip6stat.rip6s_nosockmcast++;
323		if (proto == IPPROTO_NONE)
324			m_freem(m);
325		else {
326			char *prvnxtp = ip6_get_prevhdr(m, *offp); /* XXX */
327			icmp6_error(m, ICMP6_PARAM_PROB,
328			    ICMP6_PARAMPROB_NEXTHEADER,
329			    prvnxtp - mtod(m, char *));
330		}
331		V_ip6stat.ip6s_delivered--;
332	}
333	return (IPPROTO_DONE);
334}
335
336void
337rip6_ctlinput(int cmd, struct sockaddr *sa, void *d)
338{
339	struct ip6_hdr *ip6;
340	struct mbuf *m;
341	int off = 0;
342	struct ip6ctlparam *ip6cp = NULL;
343	const struct sockaddr_in6 *sa6_src = NULL;
344	void *cmdarg;
345	struct inpcb *(*notify)(struct inpcb *, int) = in6_rtchange;
346
347	if (sa->sa_family != AF_INET6 ||
348	    sa->sa_len != sizeof(struct sockaddr_in6))
349		return;
350
351	if ((unsigned)cmd >= PRC_NCMDS)
352		return;
353	if (PRC_IS_REDIRECT(cmd))
354		notify = in6_rtchange, d = NULL;
355	else if (cmd == PRC_HOSTDEAD)
356		d = NULL;
357	else if (inet6ctlerrmap[cmd] == 0)
358		return;
359
360	/*
361	 * If the parameter is from icmp6, decode it.
362	 */
363	if (d != NULL) {
364		ip6cp = (struct ip6ctlparam *)d;
365		m = ip6cp->ip6c_m;
366		ip6 = ip6cp->ip6c_ip6;
367		off = ip6cp->ip6c_off;
368		cmdarg = ip6cp->ip6c_cmdarg;
369		sa6_src = ip6cp->ip6c_src;
370	} else {
371		m = NULL;
372		ip6 = NULL;
373		cmdarg = NULL;
374		sa6_src = &sa6_any;
375	}
376
377	(void) in6_pcbnotify(&V_ripcbinfo, sa, 0,
378	    (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
379}
380
381/*
382 * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
383 * may have setup with control call.
384 */
385int
386#if __STDC__
387rip6_output(struct mbuf *m, ...)
388#else
389rip6_output(m, va_alist)
390	struct mbuf *m;
391	va_dcl
392#endif
393{
394	struct mbuf *control;
395	struct m_tag *mtag;
396	struct socket *so;
397	struct sockaddr_in6 *dstsock;
398	struct in6_addr *dst;
399	struct ip6_hdr *ip6;
400	struct inpcb *in6p;
401	u_int	plen = m->m_pkthdr.len;
402	int error = 0;
403	struct ip6_pktopts opt, *optp;
404	struct ifnet *oifp = NULL;
405	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
406	int scope_ambiguous = 0;
407	int use_defzone = 0;
408	struct in6_addr in6a;
409	va_list ap;
410
411	va_start(ap, m);
412	so = va_arg(ap, struct socket *);
413	dstsock = va_arg(ap, struct sockaddr_in6 *);
414	control = va_arg(ap, struct mbuf *);
415	va_end(ap);
416
417	in6p = sotoinpcb(so);
418	INP_WLOCK(in6p);
419
420	dst = &dstsock->sin6_addr;
421	if (control != NULL) {
422		if ((error = ip6_setpktopts(control, &opt,
423		    in6p->in6p_outputopts, so->so_cred,
424		    so->so_proto->pr_protocol)) != 0) {
425			goto bad;
426		}
427		optp = &opt;
428	} else
429		optp = in6p->in6p_outputopts;
430
431	/*
432	 * Check and convert scope zone ID into internal form.
433	 *
434	 * XXX: we may still need to determine the zone later.
435	 */
436	if (!(so->so_state & SS_ISCONNECTED)) {
437		if (!optp->ip6po_pktinfo || !optp->ip6po_pktinfo->ipi6_ifindex)
438			use_defzone = V_ip6_use_defzone;
439		if (dstsock->sin6_scope_id == 0 && !use_defzone)
440			scope_ambiguous = 1;
441		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
442			goto bad;
443	}
444
445	/*
446	 * For an ICMPv6 packet, we should know its type and code to update
447	 * statistics.
448	 */
449	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
450		struct icmp6_hdr *icmp6;
451		if (m->m_len < sizeof(struct icmp6_hdr) &&
452		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
453			error = ENOBUFS;
454			goto bad;
455		}
456		icmp6 = mtod(m, struct icmp6_hdr *);
457		type = icmp6->icmp6_type;
458		code = icmp6->icmp6_code;
459	}
460
461	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
462	if (m == NULL) {
463		error = ENOBUFS;
464		goto bad;
465	}
466	ip6 = mtod(m, struct ip6_hdr *);
467
468	/*
469	 * Source address selection.
470	 */
471	error = in6_selectsrc(dstsock, optp, in6p, NULL, so->so_cred,
472	    &oifp, &in6a);
473	if (error)
474		goto bad;
475	error = prison_check_ip6(in6p->inp_cred, &in6a);
476	if (error != 0)
477		goto bad;
478	ip6->ip6_src = in6a;
479
480	if (oifp && scope_ambiguous) {
481		/*
482		 * Application should provide a proper zone ID or the use of
483		 * default zone IDs should be enabled.  Unfortunately, some
484		 * applications do not behave as it should, so we need a
485		 * workaround.  Even if an appropriate ID is not determined
486		 * (when it's required), if we can determine the outgoing
487		 * interface. determine the zone ID based on the interface.
488		 */
489		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
490		if (error != 0)
491			goto bad;
492	}
493	ip6->ip6_dst = dstsock->sin6_addr;
494
495	/*
496	 * Fill in the rest of the IPv6 header fields.
497	 */
498	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
499	    (in6p->inp_flow & IPV6_FLOWINFO_MASK);
500	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
501	    (IPV6_VERSION & IPV6_VERSION_MASK);
502
503	/*
504	 * ip6_plen will be filled in ip6_output, so not fill it here.
505	 */
506	ip6->ip6_nxt = in6p->inp_ip_p;
507	ip6->ip6_hlim = in6_selecthlim(in6p, oifp);
508
509	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
510	    in6p->in6p_cksum != -1) {
511		struct mbuf *n;
512		int off;
513		u_int16_t *p;
514
515		/* Compute checksum. */
516		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
517			off = offsetof(struct icmp6_hdr, icmp6_cksum);
518		else
519			off = in6p->in6p_cksum;
520		if (plen < off + 1) {
521			error = EINVAL;
522			goto bad;
523		}
524		off += sizeof(struct ip6_hdr);
525
526		n = m;
527		while (n && n->m_len <= off) {
528			off -= n->m_len;
529			n = n->m_next;
530		}
531		if (!n)
532			goto bad;
533		p = (u_int16_t *)(mtod(n, caddr_t) + off);
534		*p = 0;
535		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
536	}
537
538	/*
539	 * Send RA/RS messages to user land for protection, before sending
540	 * them to rtadvd/rtsol.
541	 */
542	if ((send_sendso_input_hook != NULL) &&
543	    so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
544		switch (type) {
545		case ND_ROUTER_ADVERT:
546		case ND_ROUTER_SOLICIT:
547			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
548				sizeof(unsigned short), M_NOWAIT);
549			if (mtag == NULL)
550				goto bad;
551			m_tag_prepend(m, mtag);
552		}
553	}
554
555	error = ip6_output(m, optp, NULL, 0, in6p->in6p_moptions, &oifp, in6p);
556	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
557		if (oifp)
558			icmp6_ifoutstat_inc(oifp, type, code);
559		ICMP6STAT_INC(icp6s_outhist[type]);
560	} else
561		V_rip6stat.rip6s_opackets++;
562
563	goto freectl;
564
565 bad:
566	if (m)
567		m_freem(m);
568
569 freectl:
570	if (control != NULL) {
571		ip6_clearpktopts(&opt, -1);
572		m_freem(control);
573	}
574	INP_WUNLOCK(in6p);
575	return (error);
576}
577
578/*
579 * Raw IPv6 socket option processing.
580 */
581int
582rip6_ctloutput(struct socket *so, struct sockopt *sopt)
583{
584	int error;
585
586	if (sopt->sopt_level == IPPROTO_ICMPV6)
587		/*
588		 * XXX: is it better to call icmp6_ctloutput() directly
589		 * from protosw?
590		 */
591		return (icmp6_ctloutput(so, sopt));
592	else if (sopt->sopt_level != IPPROTO_IPV6)
593		return (EINVAL);
594
595	error = 0;
596
597	switch (sopt->sopt_dir) {
598	case SOPT_GET:
599		switch (sopt->sopt_name) {
600		case MRT6_INIT:
601		case MRT6_DONE:
602		case MRT6_ADD_MIF:
603		case MRT6_DEL_MIF:
604		case MRT6_ADD_MFC:
605		case MRT6_DEL_MFC:
606		case MRT6_PIM:
607			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
608			    EOPNOTSUPP;
609			break;
610		case IPV6_CHECKSUM:
611			error = ip6_raw_ctloutput(so, sopt);
612			break;
613		default:
614			error = ip6_ctloutput(so, sopt);
615			break;
616		}
617		break;
618
619	case SOPT_SET:
620		switch (sopt->sopt_name) {
621		case MRT6_INIT:
622		case MRT6_DONE:
623		case MRT6_ADD_MIF:
624		case MRT6_DEL_MIF:
625		case MRT6_ADD_MFC:
626		case MRT6_DEL_MFC:
627		case MRT6_PIM:
628			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
629			    EOPNOTSUPP;
630			break;
631		case IPV6_CHECKSUM:
632			error = ip6_raw_ctloutput(so, sopt);
633			break;
634		default:
635			error = ip6_ctloutput(so, sopt);
636			break;
637		}
638		break;
639	}
640
641	return (error);
642}
643
644static int
645rip6_attach(struct socket *so, int proto, struct thread *td)
646{
647	struct inpcb *inp;
648	struct icmp6_filter *filter;
649	int error;
650
651	inp = sotoinpcb(so);
652	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
653
654	error = priv_check(td, PRIV_NETINET_RAW);
655	if (error)
656		return (error);
657	error = soreserve(so, rip_sendspace, rip_recvspace);
658	if (error)
659		return (error);
660	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
661	if (filter == NULL)
662		return (ENOMEM);
663	INP_INFO_WLOCK(&V_ripcbinfo);
664	error = in_pcballoc(so, &V_ripcbinfo);
665	if (error) {
666		INP_INFO_WUNLOCK(&V_ripcbinfo);
667		free(filter, M_PCB);
668		return (error);
669	}
670	inp = (struct inpcb *)so->so_pcb;
671	INP_INFO_WUNLOCK(&V_ripcbinfo);
672	inp->inp_vflag |= INP_IPV6;
673	inp->inp_ip_p = (long)proto;
674	inp->in6p_hops = -1;	/* use kernel default */
675	inp->in6p_cksum = -1;
676	inp->in6p_icmp6filt = filter;
677	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
678	INP_WUNLOCK(inp);
679	return (0);
680}
681
682static void
683rip6_detach(struct socket *so)
684{
685	struct inpcb *inp;
686
687	inp = sotoinpcb(so);
688	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
689
690	if (so == V_ip6_mrouter && ip6_mrouter_done)
691		ip6_mrouter_done();
692	/* xxx: RSVP */
693	INP_INFO_WLOCK(&V_ripcbinfo);
694	INP_WLOCK(inp);
695	free(inp->in6p_icmp6filt, M_PCB);
696	in_pcbdetach(inp);
697	in_pcbfree(inp);
698	INP_INFO_WUNLOCK(&V_ripcbinfo);
699}
700
701/* XXXRW: This can't ever be called. */
702static void
703rip6_abort(struct socket *so)
704{
705	struct inpcb *inp;
706
707	inp = sotoinpcb(so);
708	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
709
710	soisdisconnected(so);
711}
712
713static void
714rip6_close(struct socket *so)
715{
716	struct inpcb *inp;
717
718	inp = sotoinpcb(so);
719	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
720
721	soisdisconnected(so);
722}
723
724static int
725rip6_disconnect(struct socket *so)
726{
727	struct inpcb *inp;
728
729	inp = sotoinpcb(so);
730	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
731
732	if ((so->so_state & SS_ISCONNECTED) == 0)
733		return (ENOTCONN);
734	inp->in6p_faddr = in6addr_any;
735	rip6_abort(so);
736	return (0);
737}
738
739static int
740rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
741{
742	struct inpcb *inp;
743	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
744	struct ifaddr *ifa = NULL;
745	int error = 0;
746
747	inp = sotoinpcb(so);
748	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
749
750	if (nam->sa_len != sizeof(*addr))
751		return (EINVAL);
752	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
753		return (error);
754	if (TAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
755		return (EADDRNOTAVAIL);
756	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
757		return (error);
758
759	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
760	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL)
761		return (EADDRNOTAVAIL);
762	if (ifa != NULL &&
763	    ((struct in6_ifaddr *)ifa)->ia6_flags &
764	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
765	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
766		ifa_free(ifa);
767		return (EADDRNOTAVAIL);
768	}
769	if (ifa != NULL)
770		ifa_free(ifa);
771	INP_INFO_WLOCK(&V_ripcbinfo);
772	INP_WLOCK(inp);
773	inp->in6p_laddr = addr->sin6_addr;
774	INP_WUNLOCK(inp);
775	INP_INFO_WUNLOCK(&V_ripcbinfo);
776	return (0);
777}
778
779static int
780rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
781{
782	struct inpcb *inp;
783	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
784	struct in6_addr in6a;
785	struct ifnet *ifp = NULL;
786	int error = 0, scope_ambiguous = 0;
787
788	inp = sotoinpcb(so);
789	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
790
791	if (nam->sa_len != sizeof(*addr))
792		return (EINVAL);
793	if (TAILQ_EMPTY(&V_ifnet))
794		return (EADDRNOTAVAIL);
795	if (addr->sin6_family != AF_INET6)
796		return (EAFNOSUPPORT);
797
798	/*
799	 * Application should provide a proper zone ID or the use of default
800	 * zone IDs should be enabled.  Unfortunately, some applications do
801	 * not behave as it should, so we need a workaround.  Even if an
802	 * appropriate ID is not determined, we'll see if we can determine
803	 * the outgoing interface.  If we can, determine the zone ID based on
804	 * the interface below.
805	 */
806	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
807		scope_ambiguous = 1;
808	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
809		return (error);
810
811	INP_INFO_WLOCK(&V_ripcbinfo);
812	INP_WLOCK(inp);
813	/* Source address selection. XXX: need pcblookup? */
814	error = in6_selectsrc(addr, inp->in6p_outputopts,
815	    inp, NULL, so->so_cred, &ifp, &in6a);
816	if (error) {
817		INP_WUNLOCK(inp);
818		INP_INFO_WUNLOCK(&V_ripcbinfo);
819		return (error);
820	}
821
822	/* XXX: see above */
823	if (ifp && scope_ambiguous &&
824	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
825		INP_WUNLOCK(inp);
826		INP_INFO_WUNLOCK(&V_ripcbinfo);
827		return (error);
828	}
829	inp->in6p_faddr = addr->sin6_addr;
830	inp->in6p_laddr = in6a;
831	soisconnected(so);
832	INP_WUNLOCK(inp);
833	INP_INFO_WUNLOCK(&V_ripcbinfo);
834	return (0);
835}
836
837static int
838rip6_shutdown(struct socket *so)
839{
840	struct inpcb *inp;
841
842	inp = sotoinpcb(so);
843	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
844
845	INP_WLOCK(inp);
846	socantsendmore(so);
847	INP_WUNLOCK(inp);
848	return (0);
849}
850
851static int
852rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
853    struct mbuf *control, struct thread *td)
854{
855	struct inpcb *inp;
856	struct sockaddr_in6 tmp;
857	struct sockaddr_in6 *dst;
858	int ret;
859
860	inp = sotoinpcb(so);
861	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
862
863	/* Always copy sockaddr to avoid overwrites. */
864	/* Unlocked read. */
865	if (so->so_state & SS_ISCONNECTED) {
866		if (nam) {
867			m_freem(m);
868			return (EISCONN);
869		}
870		/* XXX */
871		bzero(&tmp, sizeof(tmp));
872		tmp.sin6_family = AF_INET6;
873		tmp.sin6_len = sizeof(struct sockaddr_in6);
874		INP_RLOCK(inp);
875		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
876		    sizeof(struct in6_addr));
877		INP_RUNLOCK(inp);
878		dst = &tmp;
879	} else {
880		if (nam == NULL) {
881			m_freem(m);
882			return (ENOTCONN);
883		}
884		if (nam->sa_len != sizeof(struct sockaddr_in6)) {
885			m_freem(m);
886			return (EINVAL);
887		}
888		tmp = *(struct sockaddr_in6 *)nam;
889		dst = &tmp;
890
891		if (dst->sin6_family == AF_UNSPEC) {
892			/*
893			 * XXX: we allow this case for backward
894			 * compatibility to buggy applications that
895			 * rely on old (and wrong) kernel behavior.
896			 */
897			log(LOG_INFO, "rip6 SEND: address family is "
898			    "unspec. Assume AF_INET6\n");
899			dst->sin6_family = AF_INET6;
900		} else if (dst->sin6_family != AF_INET6) {
901			m_freem(m);
902			return(EAFNOSUPPORT);
903		}
904	}
905	ret = rip6_output(m, so, dst, control);
906	return (ret);
907}
908
909struct pr_usrreqs rip6_usrreqs = {
910	.pru_abort =		rip6_abort,
911	.pru_attach =		rip6_attach,
912	.pru_bind =		rip6_bind,
913	.pru_connect =		rip6_connect,
914	.pru_control =		in6_control,
915	.pru_detach =		rip6_detach,
916	.pru_disconnect =	rip6_disconnect,
917	.pru_peeraddr =		in6_getpeeraddr,
918	.pru_send =		rip6_send,
919	.pru_shutdown =		rip6_shutdown,
920	.pru_sockaddr =		in6_getsockaddr,
921	.pru_close =		rip6_close,
922};
923