xref: /freebsd/sys/netinet6/raw_ip6.c (revision e453e498cbb88570a3ff7b3679de65c88707da95)
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 #include "opt_route.h"
65 
66 #include <sys/param.h>
67 #include <sys/errno.h>
68 #include <sys/jail.h>
69 #include <sys/kernel.h>
70 #include <sys/lock.h>
71 #include <sys/malloc.h>
72 #include <sys/mbuf.h>
73 #include <sys/priv.h>
74 #include <sys/proc.h>
75 #include <sys/protosw.h>
76 #include <sys/signalvar.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/stdarg.h>
80 #include <sys/sx.h>
81 #include <sys/syslog.h>
82 
83 #include <net/if.h>
84 #include <net/if_var.h>
85 #include <net/if_private.h>
86 #include <net/if_types.h>
87 #include <net/route.h>
88 #include <net/vnet.h>
89 
90 #include <netinet/in.h>
91 #include <netinet/in_var.h>
92 #include <netinet/in_systm.h>
93 #include <netinet/in_pcb.h>
94 
95 #include <netinet/icmp6.h>
96 #include <netinet/ip6.h>
97 #include <netinet/ip_var.h>
98 #include <netinet6/ip6_mroute.h>
99 #include <netinet6/in6_pcb.h>
100 #include <netinet6/ip6_var.h>
101 #include <netinet6/nd6.h>
102 #include <netinet6/raw_ip6.h>
103 #include <netinet6/in6_fib.h>
104 #include <netinet6/scope6_var.h>
105 #include <netinet6/send.h>
106 
107 #include <netipsec/ipsec_support.h>
108 
109 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
110 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
111 
112 /*
113  * Raw interface to IP6 protocol.
114  */
115 
116 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
117 #define	V_ripcbinfo			VNET(ripcbinfo)
118 
119 VNET_DECLARE(int, rip_bind_all_fibs);
120 #define	V_rip_bind_all_fibs	VNET(rip_bind_all_fibs)
121 
122 extern u_long	rip_sendspace;
123 extern u_long	rip_recvspace;
124 
125 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
126 VNET_PCPUSTAT_SYSINIT(rip6stat);
127 
128 #ifdef VIMAGE
129 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
130 #endif /* VIMAGE */
131 
132 /*
133  * Hooks for multicast routing. They all default to NULL, so leave them not
134  * initialized and rely on BSS being set to 0.
135  */
136 
137 /*
138  * The socket used to communicate with the multicast routing daemon.
139  */
140 VNET_DEFINE(struct socket *, ip6_mrouter);
141 
142 /*
143  * The various mrouter functions.
144  */
145 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
146 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
147 int (*ip6_mrouter_done)(void);
148 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
149 int (*mrt6_ioctl)(u_long, caddr_t);
150 
151 struct rip6_inp_match_ctx {
152 	struct ip6_hdr *ip6;
153 	int proto;
154 };
155 
156 static bool
rip6_inp_match(const struct inpcb * inp,void * v)157 rip6_inp_match(const struct inpcb *inp, void *v)
158 {
159 	struct rip6_inp_match_ctx *c = v;
160 	struct ip6_hdr *ip6 = c->ip6;
161 	int proto = c->proto;
162 
163 	/* XXX inp locking */
164 	if ((inp->inp_vflag & INP_IPV6) == 0)
165 		return (false);
166 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
167 		return (false);
168 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
169 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
170 		return (false);
171 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
172 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
173 		return (false);
174 
175 	return (true);
176 }
177 
178 /*
179  * Setup generic address and protocol structures for raw_input routine, then
180  * pass them along with mbuf chain.
181  */
182 int
rip6_input(struct mbuf ** mp,int * offp,int proto)183 rip6_input(struct mbuf **mp, int *offp, int proto)
184 {
185 	struct ifnet *ifp;
186 	struct mbuf *n, *m = *mp;
187 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
188 	struct inpcb *inp;
189 	struct mbuf *opts = NULL;
190 	struct sockaddr_in6 fromsa;
191 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
192 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
193 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
194 	int delivered = 0, fib;
195 
196 	M_ASSERTPKTHDR(m);
197 	NET_EPOCH_ASSERT();
198 
199 	RIP6STAT_INC(rip6s_ipackets);
200 
201 	init_sin6(&fromsa, m, 0); /* general init */
202 
203 	fib = M_GETFIB(m);
204 	ifp = m->m_pkthdr.rcvif;
205 
206 	while ((inp = inp_next(&inpi)) != NULL) {
207 		INP_RLOCK_ASSERT(inp);
208 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
209 		/*
210 		 * Check AH/ESP integrity.
211 		 */
212 		if (IPSEC_ENABLED(ipv6) &&
213 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
214 			/* Do not inject data into pcb. */
215 			continue;
216 		}
217 #endif /* IPSEC */
218 		if (jailed_without_vnet(inp->inp_cred) &&
219 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
220 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
221 			/*
222 			 * Allow raw socket in jail to receive multicast;
223 			 * assume process had PRIV_NETINET_RAW at attach,
224 			 * and fall through into normal filter path if so.
225 			 */
226 			continue;
227 		if (V_rip_bind_all_fibs == 0 && fib != inp->inp_inc.inc_fibnum)
228 			/*
229 			 * Sockets bound to a specific FIB can only receive
230 			 * packets from that FIB.
231 			 */
232 			continue;
233 		if (inp->in6p_cksum != -1) {
234 			RIP6STAT_INC(rip6s_isum);
235 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
236 			    in6_cksum(m, proto, *offp,
237 			    m->m_pkthdr.len - *offp)) {
238 				RIP6STAT_INC(rip6s_badsum);
239 				/*
240 				 * Drop the received message, don't send an
241 				 * ICMP6 message. Set proto to IPPROTO_NONE
242 				 * to achieve that.
243 				 */
244 				INP_RUNLOCK(inp);
245 				proto = IPPROTO_NONE;
246 				break;
247 			}
248 		}
249 		/*
250 		 * If this raw socket has multicast state, and we
251 		 * have received a multicast, check if this socket
252 		 * should receive it, as multicast filtering is now
253 		 * the responsibility of the transport layer.
254 		 */
255 		if (inp->in6p_moptions &&
256 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
257 			/*
258 			 * If the incoming datagram is for MLD, allow it
259 			 * through unconditionally to the raw socket.
260 			 *
261 			 * Use the M_RTALERT_MLD flag to check for MLD
262 			 * traffic without having to inspect the mbuf chain
263 			 * more deeply, as all MLDv1/v2 host messages MUST
264 			 * contain the Router Alert option.
265 			 *
266 			 * In the case of MLDv1, we may not have explicitly
267 			 * joined the group, and may have set IFF_ALLMULTI
268 			 * on the interface. im6o_mc_filter() may discard
269 			 * control traffic we actually need to see.
270 			 *
271 			 * Userland multicast routing daemons should continue
272 			 * filter the control traffic appropriately.
273 			 */
274 			int blocked;
275 
276 			blocked = MCAST_PASS;
277 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
278 				struct sockaddr_in6 mcaddr;
279 
280 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
281 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
282 				mcaddr.sin6_family = AF_INET6;
283 				mcaddr.sin6_addr = ip6->ip6_dst;
284 
285 				blocked = im6o_mc_filter(inp->in6p_moptions,
286 				    ifp,
287 				    (struct sockaddr *)&mcaddr,
288 				    (struct sockaddr *)&fromsa);
289 			}
290 			if (blocked != MCAST_PASS) {
291 				IP6STAT_INC(ip6s_notmember);
292 				continue;
293 			}
294 		}
295 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
296 			continue;
297 		if (inp->inp_flags & INP_CONTROLOPTS ||
298 		    inp->inp_socket->so_options & SO_TIMESTAMP)
299 			ip6_savecontrol(inp, n, &opts);
300 		/* strip intermediate headers */
301 		m_adj(n, *offp);
302 		if (sbappendaddr(&inp->inp_socket->so_rcv,
303 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
304 			soroverflow(inp->inp_socket);
305 			m_freem(n);
306 			if (opts)
307 				m_freem(opts);
308 			RIP6STAT_INC(rip6s_fullsock);
309 		} else {
310 			sorwakeup(inp->inp_socket);
311 			delivered++;
312 		}
313 		opts = NULL;
314 	}
315 	if (delivered == 0) {
316 		RIP6STAT_INC(rip6s_nosock);
317 		if (m->m_flags & M_MCAST)
318 			RIP6STAT_INC(rip6s_nosockmcast);
319 		if (proto == IPPROTO_NONE)
320 			m_freem(m);
321 		else
322 			icmp6_error(m, ICMP6_PARAM_PROB,
323 			    ICMP6_PARAMPROB_NEXTHEADER,
324 			    ip6_get_prevhdr(m, *offp));
325 		IP6STAT_DEC(ip6s_delivered);
326 	} else
327 		m_freem(m);
328 	return (IPPROTO_DONE);
329 }
330 
331 void
rip6_ctlinput(struct ip6ctlparam * ip6cp)332 rip6_ctlinput(struct ip6ctlparam *ip6cp)
333 {
334 	int errno;
335 
336 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
337 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
338 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
339 		    in6_rtchange);
340 }
341 
342 /*
343  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
344  * may have setup with control call.
345  */
346 static int
rip6_send(struct socket * so,int flags,struct mbuf * m,struct sockaddr * nam,struct mbuf * control,struct thread * td)347 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
348     struct mbuf *control, struct thread *td)
349 {
350 	struct epoch_tracker et;
351 	struct inpcb *inp;
352 	struct sockaddr_in6 tmp, *dstsock;
353 	struct m_tag *mtag;
354 	struct ip6_hdr *ip6;
355 	u_int	plen = m->m_pkthdr.len;
356 	struct ip6_pktopts opt, *optp;
357 	struct ifnet *oifp = NULL;
358 	int error;
359 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
360 	int scope_ambiguous = 0;
361 	int use_defzone = 0;
362 	int hlim = 0;
363 	struct in6_addr in6a;
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 #ifdef ROUTE_MPATH
456 	if (CALC_FLOWID_OUTBOUND) {
457 		uint32_t hash_type, hash_val;
458 
459 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
460 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
461 		inp->inp_flowid = hash_val;
462 		inp->inp_flowtype = hash_type;
463 	}
464 #endif
465 	/*
466 	 * Source address selection.
467 	 */
468 	NET_EPOCH_ENTER(et);
469 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
470 	    scope_ambiguous, &in6a, &hlim);
471 	NET_EPOCH_EXIT(et);
472 
473 	if (error)
474 		goto bad;
475 	error = prison_check_ip6(inp->inp_cred, &in6a);
476 	if (error != 0)
477 		goto bad;
478 	ip6->ip6_src = in6a;
479 
480 	ip6->ip6_dst = dstsock->sin6_addr;
481 
482 	/*
483 	 * Fill in the rest of the IPv6 header fields.
484 	 */
485 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
486 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
487 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
488 	    (IPV6_VERSION & IPV6_VERSION_MASK);
489 
490 	/*
491 	 * ip6_plen will be filled in ip6_output, so not fill it here.
492 	 */
493 	ip6->ip6_nxt = inp->inp_ip_p;
494 	ip6->ip6_hlim = hlim;
495 
496 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
497 		struct mbuf *n;
498 		int off;
499 		u_int16_t *p;
500 
501 		/* Compute checksum. */
502 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
503 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
504 		else
505 			off = inp->in6p_cksum;
506 		if (plen < off + 2) {
507 			error = EINVAL;
508 			goto bad;
509 		}
510 		off += sizeof(struct ip6_hdr);
511 
512 		n = m;
513 		while (n && n->m_len <= off) {
514 			off -= n->m_len;
515 			n = n->m_next;
516 		}
517 		if (!n)
518 			goto bad;
519 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
520 		*p = 0;
521 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
522 	}
523 
524 	/*
525 	 * Send RA/RS messages to user land for protection, before sending
526 	 * them to rtadvd/rtsol.
527 	 */
528 	if ((send_sendso_input_hook != NULL) &&
529 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
530 		switch (type) {
531 		case ND_ROUTER_ADVERT:
532 		case ND_ROUTER_SOLICIT:
533 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
534 				sizeof(unsigned short), M_NOWAIT);
535 			if (mtag == NULL)
536 				goto bad;
537 			m_tag_prepend(m, mtag);
538 		}
539 	}
540 
541 	NET_EPOCH_ENTER(et);
542 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
543 	NET_EPOCH_EXIT(et);
544 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
545 		if (oifp)
546 			icmp6_ifoutstat_inc(oifp, type, code);
547 		ICMP6STAT_INC2(icp6s_outhist, type);
548 	} else
549 		RIP6STAT_INC(rip6s_opackets);
550 
551 	goto freectl;
552 
553  bad:
554 	if (m)
555 		m_freem(m);
556 
557  freectl:
558 	if (control != NULL) {
559 		ip6_clearpktopts(&opt, -1);
560 		m_freem(control);
561 	}
562 	INP_WUNLOCK(inp);
563 	return (error);
564 
565 release:
566 	if (control != NULL)
567 		m_freem(control);
568 	m_freem(m);
569 	return (error);
570 }
571 
572 /*
573  * Raw IPv6 socket option processing.
574  */
575 int
rip6_ctloutput(struct socket * so,struct sockopt * sopt)576 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
577 {
578 	struct inpcb *inp = sotoinpcb(so);
579 	int error;
580 
581 	if (sopt->sopt_level == IPPROTO_ICMPV6)
582 		/*
583 		 * XXX: is it better to call icmp6_ctloutput() directly
584 		 * from protosw?
585 		 */
586 		return (icmp6_ctloutput(so, sopt));
587 	else if (sopt->sopt_level != IPPROTO_IPV6) {
588 		if (sopt->sopt_dir == SOPT_SET &&
589 		    sopt->sopt_level == SOL_SOCKET &&
590 		    sopt->sopt_name == SO_SETFIB)
591 			return (ip6_ctloutput(so, sopt));
592 		return (EINVAL);
593 	}
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 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
608 				return (EOPNOTSUPP);
609 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
610 			    EOPNOTSUPP;
611 			break;
612 		case IPV6_CHECKSUM:
613 			error = ip6_raw_ctloutput(so, sopt);
614 			break;
615 		default:
616 			error = ip6_ctloutput(so, sopt);
617 			break;
618 		}
619 		break;
620 
621 	case SOPT_SET:
622 		switch (sopt->sopt_name) {
623 		case MRT6_INIT:
624 		case MRT6_DONE:
625 		case MRT6_ADD_MIF:
626 		case MRT6_DEL_MIF:
627 		case MRT6_ADD_MFC:
628 		case MRT6_DEL_MFC:
629 		case MRT6_PIM:
630 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
631 				return (EOPNOTSUPP);
632 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
633 			    EOPNOTSUPP;
634 			break;
635 		case IPV6_CHECKSUM:
636 			error = ip6_raw_ctloutput(so, sopt);
637 			break;
638 		default:
639 			error = ip6_ctloutput(so, sopt);
640 			break;
641 		}
642 		break;
643 	}
644 
645 	return (error);
646 }
647 
648 static int
rip6_attach(struct socket * so,int proto,struct thread * td)649 rip6_attach(struct socket *so, int proto, struct thread *td)
650 {
651 	struct inpcb *inp;
652 	struct icmp6_filter *filter;
653 	int error;
654 
655 	inp = sotoinpcb(so);
656 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
657 
658 	error = priv_check(td, PRIV_NETINET_RAW);
659 	if (error)
660 		return (error);
661 	if (proto >= IPPROTO_MAX || proto < 0)
662 		return (EPROTONOSUPPORT);
663 	error = soreserve(so, rip_sendspace, rip_recvspace);
664 	if (error)
665 		return (error);
666 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
667 	if (filter == NULL)
668 		return (ENOMEM);
669 	error = in_pcballoc(so, &V_ripcbinfo);
670 	if (error) {
671 		free(filter, M_PCB);
672 		return (error);
673 	}
674 	inp = (struct inpcb *)so->so_pcb;
675 	inp->inp_ip_p = proto;
676 	inp->in6p_cksum = -1;
677 	inp->in6p_icmp6filt = filter;
678 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
679 	INP_WUNLOCK(inp);
680 	return (0);
681 }
682 
683 static void
rip6_detach(struct socket * so)684 rip6_detach(struct socket *so)
685 {
686 	struct inpcb *inp;
687 
688 	inp = sotoinpcb(so);
689 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
690 
691 	if (so == V_ip6_mrouter && ip6_mrouter_done)
692 		ip6_mrouter_done();
693 	/* xxx: RSVP */
694 	INP_WLOCK(inp);
695 	free(inp->in6p_icmp6filt, M_PCB);
696 	in_pcbfree(inp);
697 }
698 
699 /* XXXRW: This can't ever be called. */
700 static void
rip6_abort(struct socket * so)701 rip6_abort(struct socket *so)
702 {
703 	struct inpcb *inp __diagused;
704 
705 	inp = sotoinpcb(so);
706 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
707 
708 	soisdisconnected(so);
709 }
710 
711 static void
rip6_close(struct socket * so)712 rip6_close(struct socket *so)
713 {
714 	struct inpcb *inp __diagused;
715 
716 	inp = sotoinpcb(so);
717 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
718 
719 	soisdisconnected(so);
720 }
721 
722 static int
rip6_disconnect(struct socket * so)723 rip6_disconnect(struct socket *so)
724 {
725 	struct inpcb *inp;
726 
727 	inp = sotoinpcb(so);
728 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
729 
730 	if ((so->so_state & SS_ISCONNECTED) == 0)
731 		return (ENOTCONN);
732 	inp->in6p_faddr = in6addr_any;
733 	rip6_abort(so);
734 	return (0);
735 }
736 
737 static int
rip6_bind(struct socket * so,struct sockaddr * nam,struct thread * td)738 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
739 {
740 	struct epoch_tracker et;
741 	struct inpcb *inp;
742 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
743 	struct ifaddr *ifa = NULL;
744 	int error = 0;
745 
746 	inp = sotoinpcb(so);
747 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
748 
749 	if (nam->sa_family != AF_INET6)
750 		return (EAFNOSUPPORT);
751 	if (nam->sa_len != sizeof(*addr))
752 		return (EINVAL);
753 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
754 		return (error);
755 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
756 		return (EADDRNOTAVAIL);
757 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
758 		return (error);
759 
760 	NET_EPOCH_ENTER(et);
761 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
762 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
763 		NET_EPOCH_EXIT(et);
764 		return (EADDRNOTAVAIL);
765 	}
766 	if (ifa != NULL &&
767 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
768 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
769 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
770 		NET_EPOCH_EXIT(et);
771 		return (EADDRNOTAVAIL);
772 	}
773 	NET_EPOCH_EXIT(et);
774 	INP_WLOCK(inp);
775 	inp->in6p_laddr = addr->sin6_addr;
776 	INP_WUNLOCK(inp);
777 	return (0);
778 }
779 
780 static int
rip6_connect(struct socket * so,struct sockaddr * nam,struct thread * td)781 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
782 {
783 	struct inpcb *inp;
784 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
785 	struct in6_addr in6a;
786 	struct epoch_tracker et;
787 	int error = 0, scope_ambiguous = 0;
788 
789 	inp = sotoinpcb(so);
790 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
791 
792 	if (nam->sa_len != sizeof(*addr))
793 		return (EINVAL);
794 	if (CK_STAILQ_EMPTY(&V_ifnet))
795 		return (EADDRNOTAVAIL);
796 	if (addr->sin6_family != AF_INET6)
797 		return (EAFNOSUPPORT);
798 
799 	/*
800 	 * Application should provide a proper zone ID or the use of default
801 	 * zone IDs should be enabled.  Unfortunately, some applications do
802 	 * not behave as it should, so we need a workaround.  Even if an
803 	 * appropriate ID is not determined, we'll see if we can determine
804 	 * the outgoing interface.  If we can, determine the zone ID based on
805 	 * the interface below.
806 	 */
807 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
808 		scope_ambiguous = 1;
809 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
810 		return (error);
811 
812 	INP_WLOCK(inp);
813 	/* Source address selection. XXX: need pcblookup? */
814 	NET_EPOCH_ENTER(et);
815 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
816 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
817 	NET_EPOCH_EXIT(et);
818 	if (error) {
819 		INP_WUNLOCK(inp);
820 		return (error);
821 	}
822 
823 	inp->in6p_faddr = addr->sin6_addr;
824 	inp->in6p_laddr = in6a;
825 	soisconnected(so);
826 	INP_WUNLOCK(inp);
827 	return (0);
828 }
829 
830 static int
rip6_shutdown(struct socket * so,enum shutdown_how how)831 rip6_shutdown(struct socket *so, enum shutdown_how how)
832 {
833 
834 	SOCK_LOCK(so);
835 	if (!(so->so_state & SS_ISCONNECTED)) {
836 		SOCK_UNLOCK(so);
837 		return (ENOTCONN);
838 	}
839 	SOCK_UNLOCK(so);
840 
841 	switch (how) {
842 	case SHUT_RD:
843 		sorflush(so);
844 		break;
845 	case SHUT_RDWR:
846 		sorflush(so);
847 		/* FALLTHROUGH */
848 	case SHUT_WR:
849 		socantsendmore(so);
850 	}
851 
852 	return (0);
853 }
854 
855 struct protosw rip6_protosw = {
856 	.pr_type =		SOCK_RAW,
857 	.pr_flags =		PR_ATOMIC|PR_ADDR,
858 	.pr_ctloutput =		rip6_ctloutput,
859 	.pr_abort =		rip6_abort,
860 	.pr_attach =		rip6_attach,
861 	.pr_bind =		rip6_bind,
862 	.pr_connect =		rip6_connect,
863 	.pr_control =		in6_control,
864 	.pr_detach =		rip6_detach,
865 	.pr_disconnect =	rip6_disconnect,
866 	.pr_peeraddr =		in6_getpeeraddr,
867 	.pr_send =		rip6_send,
868 	.pr_shutdown =		rip6_shutdown,
869 	.pr_sockaddr =		in6_getsockaddr,
870 	.pr_close =		rip6_close
871 };
872