xref: /freebsd/sys/netinet6/raw_ip6.c (revision 0bb9c2b665d90934ae16eee152b6b4f4f4be0dd5)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1982, 1986, 1988, 1993
34  *	The Regents of the University of California.
35  * All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. Neither the name of the University nor the names of its contributors
46  *    may be used to endorse or promote products derived from this software
47  *    without specific prior written permission.
48  *
49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59  * SUCH DAMAGE.
60  */
61 
62 #include "opt_ipsec.h"
63 #include "opt_inet6.h"
64 
65 #include <sys/param.h>
66 #include <sys/errno.h>
67 #include <sys/jail.h>
68 #include <sys/kernel.h>
69 #include <sys/lock.h>
70 #include <sys/malloc.h>
71 #include <sys/mbuf.h>
72 #include <sys/priv.h>
73 #include <sys/proc.h>
74 #include <sys/protosw.h>
75 #include <sys/signalvar.h>
76 #include <sys/socket.h>
77 #include <sys/socketvar.h>
78 #include <sys/stdarg.h>
79 #include <sys/sx.h>
80 #include <sys/syslog.h>
81 
82 #include <net/if.h>
83 #include <net/if_var.h>
84 #include <net/if_private.h>
85 #include <net/if_types.h>
86 #include <net/route.h>
87 #include <net/vnet.h>
88 
89 #include <netinet/in.h>
90 #include <netinet/in_var.h>
91 #include <netinet/in_systm.h>
92 #include <netinet/in_pcb.h>
93 
94 #include <netinet/icmp6.h>
95 #include <netinet/ip6.h>
96 #include <netinet/ip_var.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/in6_fib.h>
103 #include <netinet6/scope6_var.h>
104 #include <netinet6/send.h>
105 
106 #include <netipsec/ipsec_support.h>
107 
108 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
109 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
110 
111 /*
112  * Raw interface to IP6 protocol.
113  */
114 
115 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
116 #define	V_ripcbinfo			VNET(ripcbinfo)
117 
118 VNET_DECLARE(int, rip_bind_all_fibs);
119 #define	V_rip_bind_all_fibs	VNET(rip_bind_all_fibs)
120 
121 extern u_long	rip_sendspace;
122 extern u_long	rip_recvspace;
123 
124 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
125 VNET_PCPUSTAT_SYSINIT(rip6stat);
126 
127 #ifdef VIMAGE
128 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
129 #endif /* VIMAGE */
130 
131 /*
132  * Hooks for multicast routing. They all default to NULL, so leave them not
133  * initialized and rely on BSS being set to 0.
134  */
135 
136 /*
137  * The socket used to communicate with the multicast routing daemon.
138  */
139 VNET_DEFINE(bool, ip6_mrouting_enabled);
140 
141 /*
142  * The various mrouter functions.
143  */
144 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
145 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
146 void (*ip6_mrouter_done)(struct socket *);
147 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
148 int (*mrt6_ioctl)(u_long, caddr_t, int);
149 
150 struct rip6_inp_match_ctx {
151 	struct ip6_hdr *ip6;
152 	int proto;
153 };
154 
155 static bool
rip6_inp_match(const struct inpcb * inp,void * v)156 rip6_inp_match(const struct inpcb *inp, void *v)
157 {
158 	struct rip6_inp_match_ctx *c = v;
159 	struct ip6_hdr *ip6 = c->ip6;
160 	int proto = c->proto;
161 
162 	/* XXX inp locking */
163 	if ((inp->inp_vflag & INP_IPV6) == 0)
164 		return (false);
165 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
166 		return (false);
167 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
168 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
169 		return (false);
170 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
171 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
172 		return (false);
173 
174 	return (true);
175 }
176 
177 /*
178  * Setup generic address and protocol structures for raw_input routine, then
179  * pass them along with mbuf chain.
180  */
181 int
rip6_input(struct mbuf ** mp,int * offp,int proto)182 rip6_input(struct mbuf **mp, int *offp, int proto)
183 {
184 	struct ifnet *ifp;
185 	struct mbuf *n, *m = *mp;
186 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
187 	struct inpcb *inp;
188 	struct mbuf *opts = NULL;
189 	struct sockaddr_in6 fromsa;
190 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
191 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
192 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
193 	int delivered = 0, fib;
194 
195 	M_ASSERTPKTHDR(m);
196 	NET_EPOCH_ASSERT();
197 
198 	RIP6STAT_INC(rip6s_ipackets);
199 
200 	init_sin6(&fromsa, m, 0); /* general init */
201 
202 	fib = M_GETFIB(m);
203 	ifp = m->m_pkthdr.rcvif;
204 
205 	while ((inp = inp_next(&inpi)) != NULL) {
206 		INP_RLOCK_ASSERT(inp);
207 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
208 		/*
209 		 * Check AH/ESP integrity.
210 		 */
211 		if (IPSEC_ENABLED(ipv6) &&
212 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
213 			/* Do not inject data into pcb. */
214 			continue;
215 		}
216 #endif /* IPSEC */
217 		if (jailed_without_vnet(inp->inp_cred) &&
218 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
219 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
220 			/*
221 			 * Allow raw socket in jail to receive multicast;
222 			 * assume process had PRIV_NETINET_RAW at attach,
223 			 * and fall through into normal filter path if so.
224 			 */
225 			continue;
226 		if (V_rip_bind_all_fibs == 0 && fib != inp->inp_inc.inc_fibnum)
227 			/*
228 			 * Sockets bound to a specific FIB can only receive
229 			 * packets from that FIB.
230 			 */
231 			continue;
232 		if (inp->in6p_cksum != -1) {
233 			RIP6STAT_INC(rip6s_isum);
234 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
235 			    in6_cksum(m, proto, *offp,
236 			    m->m_pkthdr.len - *offp)) {
237 				RIP6STAT_INC(rip6s_badsum);
238 				/*
239 				 * Drop the received message, don't send an
240 				 * ICMP6 message. Set proto to IPPROTO_NONE
241 				 * to achieve that.
242 				 */
243 				INP_RUNLOCK(inp);
244 				proto = IPPROTO_NONE;
245 				break;
246 			}
247 		}
248 		/*
249 		 * If this raw socket has multicast state, and we
250 		 * have received a multicast, check if this socket
251 		 * should receive it, as multicast filtering is now
252 		 * the responsibility of the transport layer.
253 		 */
254 		if (inp->in6p_moptions &&
255 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
256 			/*
257 			 * If the incoming datagram is for MLD, allow it
258 			 * through unconditionally to the raw socket.
259 			 *
260 			 * Use the M_RTALERT_MLD flag to check for MLD
261 			 * traffic without having to inspect the mbuf chain
262 			 * more deeply, as all MLDv1/v2 host messages MUST
263 			 * contain the Router Alert option.
264 			 *
265 			 * In the case of MLDv1, we may not have explicitly
266 			 * joined the group, and may have set IFF_ALLMULTI
267 			 * on the interface. im6o_mc_filter() may discard
268 			 * control traffic we actually need to see.
269 			 *
270 			 * Userland multicast routing daemons should continue
271 			 * filter the control traffic appropriately.
272 			 */
273 			int blocked;
274 
275 			blocked = MCAST_PASS;
276 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
277 				struct sockaddr_in6 mcaddr;
278 
279 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
280 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
281 				mcaddr.sin6_family = AF_INET6;
282 				mcaddr.sin6_addr = ip6->ip6_dst;
283 
284 				blocked = im6o_mc_filter(inp->in6p_moptions,
285 				    ifp,
286 				    (struct sockaddr *)&mcaddr,
287 				    (struct sockaddr *)&fromsa);
288 			}
289 			if (blocked != MCAST_PASS) {
290 				IP6STAT_INC(ip6s_notmember);
291 				continue;
292 			}
293 		}
294 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
295 			continue;
296 		if (inp->inp_flags & INP_CONTROLOPTS ||
297 		    inp->inp_socket->so_options & SO_TIMESTAMP)
298 			ip6_savecontrol(inp, n, &opts);
299 		/* strip intermediate headers */
300 		m_adj(n, *offp);
301 		if (sbappendaddr(&inp->inp_socket->so_rcv,
302 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
303 			soroverflow(inp->inp_socket);
304 			m_freem(n);
305 			if (opts)
306 				m_freem(opts);
307 			RIP6STAT_INC(rip6s_fullsock);
308 		} else {
309 			sorwakeup(inp->inp_socket);
310 			delivered++;
311 		}
312 		opts = NULL;
313 	}
314 	if (delivered == 0) {
315 		RIP6STAT_INC(rip6s_nosock);
316 		if (m->m_flags & M_MCAST)
317 			RIP6STAT_INC(rip6s_nosockmcast);
318 		if (proto == IPPROTO_NONE)
319 			m_freem(m);
320 		else
321 			icmp6_error(m, ICMP6_PARAM_PROB,
322 			    ICMP6_PARAMPROB_NEXTHEADER,
323 			    ip6_get_prevhdr(m, *offp));
324 		IP6STAT_DEC(ip6s_delivered);
325 	} else
326 		m_freem(m);
327 	return (IPPROTO_DONE);
328 }
329 
330 void
rip6_ctlinput(struct ip6ctlparam * ip6cp)331 rip6_ctlinput(struct ip6ctlparam *ip6cp)
332 {
333 	int errno;
334 
335 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
336 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
337 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
338 		    in6_rtchange);
339 }
340 
341 /*
342  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
343  * may have setup with control call.
344  */
345 static int
rip6_send(struct socket * so,int flags,struct mbuf * m,struct sockaddr * nam,struct mbuf * control,struct thread * td)346 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
347     struct mbuf *control, struct thread *td)
348 {
349 	struct epoch_tracker et;
350 	struct inpcb *inp;
351 	struct sockaddr_in6 tmp, *dstsock;
352 	struct m_tag *mtag;
353 	struct ip6_hdr *ip6;
354 	u_int	plen = m->m_pkthdr.len;
355 	struct ip6_pktopts opt, *optp;
356 	struct ifnet *oifp = NULL;
357 	int error;
358 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
359 	int scope_ambiguous = 0;
360 	int use_defzone = 0;
361 	int hlim = 0;
362 	struct in6_addr in6a;
363 	uint32_t hash_type, hash_val;
364 
365 	inp = sotoinpcb(so);
366 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
367 
368 	/* Always copy sockaddr to avoid overwrites. */
369 	/* Unlocked read. */
370 	if (so->so_state & SS_ISCONNECTED) {
371 		if (nam) {
372 			error = EISCONN;
373 			goto release;
374 		}
375 		tmp = (struct sockaddr_in6 ){
376 			.sin6_family = AF_INET6,
377 			.sin6_len = sizeof(struct sockaddr_in6),
378 		};
379 		INP_RLOCK(inp);
380 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
381 		    sizeof(struct in6_addr));
382 		INP_RUNLOCK(inp);
383 		dstsock = &tmp;
384 	} else {
385 		if (nam == NULL)
386 			error = ENOTCONN;
387 		else if (nam->sa_family != AF_INET6)
388 			error = EAFNOSUPPORT;
389 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
390 			error = EINVAL;
391 		else
392 			error = 0;
393 		if (error != 0)
394 			goto release;
395 		dstsock = (struct sockaddr_in6 *)nam;
396 		if (dstsock->sin6_family != AF_INET6) {
397 			error = EAFNOSUPPORT;
398 			goto release;
399 		}
400 	}
401 
402 	INP_WLOCK(inp);
403 
404 	if (control != NULL) {
405 		NET_EPOCH_ENTER(et);
406 		error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
407 		    so->so_cred, inp->inp_ip_p);
408 		NET_EPOCH_EXIT(et);
409 
410 		if (error != 0) {
411 			goto bad;
412 		}
413 		optp = &opt;
414 	} else
415 		optp = inp->in6p_outputopts;
416 
417 	/*
418 	 * Check and convert scope zone ID into internal form.
419 	 *
420 	 * XXX: we may still need to determine the zone later.
421 	 */
422 	if (!(so->so_state & SS_ISCONNECTED)) {
423 		if (!optp || !optp->ip6po_pktinfo ||
424 		    !optp->ip6po_pktinfo->ipi6_ifindex)
425 			use_defzone = V_ip6_use_defzone;
426 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
427 			scope_ambiguous = 1;
428 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
429 			goto bad;
430 	}
431 
432 	/*
433 	 * For an ICMPv6 packet, we should know its type and code to update
434 	 * statistics.
435 	 */
436 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
437 		struct icmp6_hdr *icmp6;
438 		if (m->m_len < sizeof(struct icmp6_hdr) &&
439 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
440 			error = ENOBUFS;
441 			goto bad;
442 		}
443 		icmp6 = mtod(m, struct icmp6_hdr *);
444 		type = icmp6->icmp6_type;
445 		code = icmp6->icmp6_code;
446 	}
447 
448 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
449 	if (m == NULL) {
450 		error = ENOBUFS;
451 		goto bad;
452 	}
453 	ip6 = mtod(m, struct ip6_hdr *);
454 
455 	if (V_fib_hash_outbound) {
456 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
457 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
458 		inp->inp_flowid = hash_val;
459 		inp->inp_flowtype = hash_type;
460 	}
461 	/*
462 	 * Source address selection.
463 	 */
464 	NET_EPOCH_ENTER(et);
465 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
466 	    scope_ambiguous, &in6a, &hlim);
467 	NET_EPOCH_EXIT(et);
468 
469 	if (error)
470 		goto bad;
471 	error = prison_check_ip6(inp->inp_cred, &in6a);
472 	if (error != 0)
473 		goto bad;
474 	ip6->ip6_src = in6a;
475 
476 	ip6->ip6_dst = dstsock->sin6_addr;
477 
478 	/*
479 	 * Fill in the rest of the IPv6 header fields.
480 	 */
481 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
482 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
483 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
484 	    (IPV6_VERSION & IPV6_VERSION_MASK);
485 
486 	/*
487 	 * ip6_plen will be filled in ip6_output, so not fill it here.
488 	 */
489 	ip6->ip6_nxt = inp->inp_ip_p;
490 	ip6->ip6_hlim = hlim;
491 
492 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
493 		struct mbuf *n;
494 		int off;
495 		u_int16_t *p;
496 
497 		/* Compute checksum. */
498 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
499 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
500 		else
501 			off = inp->in6p_cksum;
502 		if (plen < off + 2) {
503 			error = EINVAL;
504 			goto bad;
505 		}
506 		off += sizeof(struct ip6_hdr);
507 
508 		n = m;
509 		while (n && n->m_len <= off) {
510 			off -= n->m_len;
511 			n = n->m_next;
512 		}
513 		if (!n)
514 			goto bad;
515 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
516 		*p = 0;
517 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
518 	}
519 
520 	/*
521 	 * Send RA/RS messages to user land for protection, before sending
522 	 * them to rtadvd/rtsol.
523 	 */
524 	if ((send_sendso_input_hook != NULL) &&
525 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
526 		switch (type) {
527 		case ND_ROUTER_ADVERT:
528 		case ND_ROUTER_SOLICIT:
529 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
530 				sizeof(unsigned short), M_NOWAIT);
531 			if (mtag == NULL)
532 				goto bad;
533 			m_tag_prepend(m, mtag);
534 		}
535 	}
536 
537 	NET_EPOCH_ENTER(et);
538 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
539 	NET_EPOCH_EXIT(et);
540 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
541 		if (oifp)
542 			icmp6_ifoutstat_inc(oifp, type, code);
543 		ICMP6STAT_INC2(icp6s_outhist, type);
544 	} else
545 		RIP6STAT_INC(rip6s_opackets);
546 
547 	goto freectl;
548 
549  bad:
550 	if (m)
551 		m_freem(m);
552 
553  freectl:
554 	if (control != NULL) {
555 		ip6_clearpktopts(&opt, -1);
556 		m_freem(control);
557 	}
558 	INP_WUNLOCK(inp);
559 	return (error);
560 
561 release:
562 	if (control != NULL)
563 		m_freem(control);
564 	m_freem(m);
565 	return (error);
566 }
567 
568 /*
569  * Raw IPv6 socket option processing.
570  */
571 int
rip6_ctloutput(struct socket * so,struct sockopt * sopt)572 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
573 {
574 	struct inpcb *inp = sotoinpcb(so);
575 	int error;
576 
577 	if (sopt->sopt_level == IPPROTO_ICMPV6)
578 		/*
579 		 * XXX: is it better to call icmp6_ctloutput() directly
580 		 * from protosw?
581 		 */
582 		return (icmp6_ctloutput(so, sopt));
583 	else if (sopt->sopt_level != IPPROTO_IPV6) {
584 		if (sopt->sopt_dir == SOPT_SET &&
585 		    sopt->sopt_level == SOL_SOCKET &&
586 		    sopt->sopt_name == SO_SETFIB)
587 			return (ip6_ctloutput(so, sopt));
588 		return (EINVAL);
589 	}
590 
591 	error = 0;
592 
593 	switch (sopt->sopt_dir) {
594 	case SOPT_GET:
595 		switch (sopt->sopt_name) {
596 		case MRT6_INIT:
597 		case MRT6_DONE:
598 		case MRT6_ADD_MIF:
599 		case MRT6_DEL_MIF:
600 		case MRT6_ADD_MFC:
601 		case MRT6_DEL_MFC:
602 		case MRT6_PIM:
603 			error = priv_check(curthread, PRIV_NETINET_MROUTE);
604 			if (error != 0)
605 				return (error);
606 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
607 				return (EOPNOTSUPP);
608 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
609 			    EOPNOTSUPP;
610 			break;
611 		case IPV6_CHECKSUM:
612 			error = ip6_raw_ctloutput(so, sopt);
613 			break;
614 		default:
615 			error = ip6_ctloutput(so, sopt);
616 			break;
617 		}
618 		break;
619 
620 	case SOPT_SET:
621 		switch (sopt->sopt_name) {
622 		case MRT6_INIT:
623 		case MRT6_DONE:
624 		case MRT6_ADD_MIF:
625 		case MRT6_DEL_MIF:
626 		case MRT6_ADD_MFC:
627 		case MRT6_DEL_MFC:
628 		case MRT6_PIM:
629 			error = priv_check(curthread, PRIV_NETINET_MROUTE);
630 			if (error != 0)
631 				return (error);
632 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
633 				return (EOPNOTSUPP);
634 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
635 			    EOPNOTSUPP;
636 			break;
637 		case IPV6_CHECKSUM:
638 			error = ip6_raw_ctloutput(so, sopt);
639 			break;
640 		default:
641 			error = ip6_ctloutput(so, sopt);
642 			break;
643 		}
644 		break;
645 	}
646 
647 	return (error);
648 }
649 
650 static int
rip6_attach(struct socket * so,int proto,struct thread * td)651 rip6_attach(struct socket *so, int proto, struct thread *td)
652 {
653 	struct inpcb *inp;
654 	struct icmp6_filter *filter;
655 	int error;
656 
657 	inp = sotoinpcb(so);
658 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
659 
660 	error = priv_check(td, PRIV_NETINET_RAW);
661 	if (error)
662 		return (error);
663 	if (proto >= IPPROTO_MAX || proto < 0)
664 		return (EPROTONOSUPPORT);
665 	error = soreserve(so, rip_sendspace, rip_recvspace);
666 	if (error)
667 		return (error);
668 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
669 	if (filter == NULL)
670 		return (ENOMEM);
671 	error = in_pcballoc(so, &V_ripcbinfo);
672 	if (error) {
673 		free(filter, M_PCB);
674 		return (error);
675 	}
676 	inp = (struct inpcb *)so->so_pcb;
677 	inp->inp_ip_p = proto;
678 	inp->in6p_cksum = -1;
679 	inp->in6p_icmp6filt = filter;
680 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
681 	INP_WUNLOCK(inp);
682 	return (0);
683 }
684 
685 static void
rip6_detach(struct socket * so)686 rip6_detach(struct socket *so)
687 {
688 	struct inpcb *inp;
689 
690 	inp = sotoinpcb(so);
691 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
692 
693 	if (ip6_mrouter_done != NULL)
694 		ip6_mrouter_done(so);
695 	/* xxx: RSVP */
696 	INP_WLOCK(inp);
697 	free(inp->in6p_icmp6filt, M_PCB);
698 	in_pcbfree(inp);
699 }
700 
701 /* XXXRW: This can't ever be called. */
702 static void
rip6_abort(struct socket * so)703 rip6_abort(struct socket *so)
704 {
705 	struct inpcb *inp __diagused;
706 
707 	inp = sotoinpcb(so);
708 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
709 
710 	soisdisconnected(so);
711 }
712 
713 static void
rip6_close(struct socket * so)714 rip6_close(struct socket *so)
715 {
716 	struct inpcb *inp __diagused;
717 
718 	inp = sotoinpcb(so);
719 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
720 
721 	soisdisconnected(so);
722 }
723 
724 static int
rip6_disconnect(struct socket * so)725 rip6_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 
739 static int
rip6_bind(struct socket * so,struct sockaddr * nam,struct thread * td)740 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
741 {
742 	struct epoch_tracker et;
743 	struct inpcb *inp;
744 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
745 	struct ifaddr *ifa = NULL;
746 	int error = 0;
747 
748 	inp = sotoinpcb(so);
749 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
750 
751 	if (nam->sa_family != AF_INET6)
752 		return (EAFNOSUPPORT);
753 	if (nam->sa_len != sizeof(*addr))
754 		return (EINVAL);
755 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
756 		return (error);
757 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
758 		return (EADDRNOTAVAIL);
759 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
760 		return (error);
761 
762 	NET_EPOCH_ENTER(et);
763 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
764 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
765 		NET_EPOCH_EXIT(et);
766 		return (EADDRNOTAVAIL);
767 	}
768 	if (ifa != NULL &&
769 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
770 	    (IN6_IFF_NOTREADY|IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
771 		NET_EPOCH_EXIT(et);
772 		return (EADDRNOTAVAIL);
773 	}
774 	NET_EPOCH_EXIT(et);
775 	INP_WLOCK(inp);
776 	inp->in6p_laddr = addr->sin6_addr;
777 	INP_WUNLOCK(inp);
778 	return (0);
779 }
780 
781 static int
rip6_connect(struct socket * so,struct sockaddr * nam,struct thread * td)782 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
783 {
784 	struct inpcb *inp;
785 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
786 	struct in6_addr in6a;
787 	struct epoch_tracker et;
788 	int error = 0, scope_ambiguous = 0;
789 
790 	inp = sotoinpcb(so);
791 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
792 
793 	if (nam->sa_len != sizeof(*addr))
794 		return (EINVAL);
795 	if (CK_STAILQ_EMPTY(&V_ifnet))
796 		return (EADDRNOTAVAIL);
797 	if (addr->sin6_family != AF_INET6)
798 		return (EAFNOSUPPORT);
799 
800 	/*
801 	 * Application should provide a proper zone ID or the use of default
802 	 * zone IDs should be enabled.  Unfortunately, some applications do
803 	 * not behave as it should, so we need a workaround.  Even if an
804 	 * appropriate ID is not determined, we'll see if we can determine
805 	 * the outgoing interface.  If we can, determine the zone ID based on
806 	 * the interface below.
807 	 */
808 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
809 		scope_ambiguous = 1;
810 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
811 		return (error);
812 
813 	INP_WLOCK(inp);
814 	/* Source address selection. XXX: need pcblookup? */
815 	NET_EPOCH_ENTER(et);
816 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
817 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
818 	NET_EPOCH_EXIT(et);
819 	if (error) {
820 		INP_WUNLOCK(inp);
821 		return (error);
822 	}
823 
824 	inp->in6p_faddr = addr->sin6_addr;
825 	inp->in6p_laddr = in6a;
826 	soisconnected(so);
827 	INP_WUNLOCK(inp);
828 	return (0);
829 }
830 
831 static int
rip6_shutdown(struct socket * so,enum shutdown_how how)832 rip6_shutdown(struct socket *so, enum shutdown_how how)
833 {
834 
835 	SOCK_LOCK(so);
836 	if (!(so->so_state & SS_ISCONNECTED)) {
837 		SOCK_UNLOCK(so);
838 		return (ENOTCONN);
839 	}
840 	SOCK_UNLOCK(so);
841 
842 	switch (how) {
843 	case SHUT_RD:
844 		sorflush(so);
845 		break;
846 	case SHUT_RDWR:
847 		sorflush(so);
848 		/* FALLTHROUGH */
849 	case SHUT_WR:
850 		socantsendmore(so);
851 	}
852 
853 	return (0);
854 }
855 
856 struct protosw rip6_protosw = {
857 	.pr_type =		SOCK_RAW,
858 	.pr_flags =		PR_ATOMIC|PR_ADDR,
859 	.pr_ctloutput =		rip6_ctloutput,
860 	.pr_abort =		rip6_abort,
861 	.pr_attach =		rip6_attach,
862 	.pr_bind =		rip6_bind,
863 	.pr_connect =		rip6_connect,
864 	.pr_control =		in6_control,
865 	.pr_detach =		rip6_detach,
866 	.pr_disconnect =	rip6_disconnect,
867 	.pr_peeraddr =		in6_getpeeraddr,
868 	.pr_send =		rip6_send,
869 	.pr_shutdown =		rip6_shutdown,
870 	.pr_sockaddr =		in6_getsockaddr,
871 	.pr_close =		rip6_close
872 };
873