xref: /openbsd/sys/netinet6/raw_ip6.c (revision ea7e0400)
1 /*	$OpenBSD: raw_ip6.c,v 1.186 2024/11/08 10:24:13 bluhm Exp $	*/
2 /*	$KAME: raw_ip6.c,v 1.69 2001/03/04 15:55:44 itojun Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * Copyright (c) 1982, 1986, 1988, 1993
35  *	The Regents of the University of California.  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  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
62  */
63 
64 #include "pf.h"
65 
66 #include <sys/param.h>
67 #include <sys/malloc.h>
68 #include <sys/mbuf.h>
69 #include <sys/socket.h>
70 #include <sys/protosw.h>
71 #include <sys/socketvar.h>
72 #include <sys/errno.h>
73 #include <sys/systm.h>
74 #include <sys/sysctl.h>
75 
76 #include <net/if.h>
77 #include <net/if_var.h>
78 #include <net/route.h>
79 
80 #include <netinet/in.h>
81 #include <netinet6/in6_var.h>
82 #include <netinet/ip6.h>
83 #include <netinet6/ip6_var.h>
84 #ifdef MROUTING
85 #include <netinet6/ip6_mroute.h>
86 #endif
87 #include <netinet/icmp6.h>
88 #include <netinet/ip.h>
89 #include <netinet/in_pcb.h>
90 #include <netinet6/nd6.h>
91 #include <netinet6/ip6protosw.h>
92 #include <netinet6/raw_ip6.h>
93 
94 #if NPF > 0
95 #include <net/pfvar.h>
96 #endif
97 
98 #include <sys/stdarg.h>
99 
100 /*
101  * Raw interface to IP6 protocol.
102  */
103 
104 struct	inpcbtable rawin6pcbtable;
105 
106 struct cpumem *rip6counters;
107 
108 const struct pr_usrreqs rip6_usrreqs = {
109 	.pru_attach	= rip6_attach,
110 	.pru_detach	= rip6_detach,
111 	.pru_bind	= rip6_bind,
112 	.pru_connect	= rip6_connect,
113 	.pru_disconnect	= rip6_disconnect,
114 	.pru_shutdown	= rip6_shutdown,
115 	.pru_send	= rip6_send,
116 	.pru_control	= in6_control,
117 	.pru_sockaddr	= in6_sockaddr,
118 	.pru_peeraddr	= in6_peeraddr,
119 };
120 
121 void	rip6_sbappend(struct inpcb *, struct mbuf *, struct ip6_hdr *, int,
122 	    struct sockaddr_in6 *);
123 
124 /*
125  * Initialize raw connection block queue.
126  */
127 void
rip6_init(void)128 rip6_init(void)
129 {
130 	in_pcbinit(&rawin6pcbtable, 1);
131 	rip6counters = counters_alloc(rip6s_ncounters);
132 }
133 
134 int
rip6_input(struct mbuf ** mp,int * offp,int proto,int af)135 rip6_input(struct mbuf **mp, int *offp, int proto, int af)
136 {
137 	struct mbuf *m = *mp;
138 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
139 	struct inpcb_iterator iter = { .inp_table = NULL };
140 	struct inpcb *inp, *last;
141 	struct in6_addr *key;
142 	struct sockaddr_in6 rip6src;
143 	uint8_t type;
144 
145 	KASSERT(af == AF_INET6);
146 
147 	if (proto == IPPROTO_ICMPV6) {
148 		struct icmp6_hdr *icmp6;
149 
150 		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, *offp,
151 		    sizeof(*icmp6));
152 		if (icmp6 == NULL)
153 			return IPPROTO_DONE;
154 		type = icmp6->icmp6_type;
155 	} else
156 		rip6stat_inc(rip6s_ipackets);
157 
158 	memset(&rip6src, 0, sizeof(rip6src));
159 	rip6src.sin6_family = AF_INET6;
160 	rip6src.sin6_len = sizeof(rip6src);
161 	/* KAME hack: recover scopeid */
162 	in6_recoverscope(&rip6src, &ip6->ip6_src);
163 
164 	key = &ip6->ip6_dst;
165 #if NPF > 0
166 	if (m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
167 		struct pf_divert *divert;
168 
169 		divert = pf_find_divert(m);
170 		KASSERT(divert != NULL);
171 		switch (divert->type) {
172 		case PF_DIVERT_TO:
173 			key = &divert->addr.v6;
174 			break;
175 		case PF_DIVERT_REPLY:
176 			break;
177 		default:
178 			panic("%s: unknown divert type %d, mbuf %p, divert %p",
179 			    __func__, divert->type, m, divert);
180 		}
181 	}
182 #endif
183 	mtx_enter(&rawin6pcbtable.inpt_mtx);
184 	last = inp = NULL;
185 	while ((inp = in_pcb_iterator(&rawin6pcbtable, inp, &iter)) != NULL) {
186 		KASSERT(ISSET(inp->inp_flags, INP_IPV6));
187 
188 		/*
189 		 * Packet must not be inserted after disconnected wakeup
190 		 * call.  To avoid race, check again when holding receive
191 		 * buffer mutex.
192 		 */
193 		if (ISSET(READ_ONCE(inp->inp_socket->so_rcv.sb_state),
194 		    SS_CANTRCVMORE))
195 			continue;
196 		if (rtable_l2(inp->inp_rtableid) !=
197 		    rtable_l2(m->m_pkthdr.ph_rtableid))
198 			continue;
199 
200 		if ((inp->inp_ipv6.ip6_nxt || proto == IPPROTO_ICMPV6) &&
201 		    inp->inp_ipv6.ip6_nxt != proto)
202 			continue;
203 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) &&
204 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, key))
205 			continue;
206 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_faddr6) &&
207 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_faddr6, &ip6->ip6_src))
208 			continue;
209 		if (proto == IPPROTO_ICMPV6 && inp->inp_icmp6filt) {
210 			if (ICMP6_FILTER_WILLBLOCK(type, inp->inp_icmp6filt))
211 				continue;
212 		}
213 		if (proto != IPPROTO_ICMPV6 && inp->inp_cksum6 != -1) {
214 			rip6stat_inc(rip6s_isum);
215 			/*
216 			 * Although in6_cksum() does not need the position of
217 			 * the checksum field for verification, enforce that it
218 			 * is located within the packet.  Userland has given
219 			 * a checksum offset, a packet too short for that is
220 			 * invalid.  Avoid overflow with user supplied offset.
221 			 */
222 			if (m->m_pkthdr.len < *offp + 2 ||
223 			    m->m_pkthdr.len - *offp - 2 < inp->inp_cksum6 ||
224 			    in6_cksum(m, proto, *offp,
225 			    m->m_pkthdr.len - *offp)) {
226 				rip6stat_inc(rip6s_badsum);
227 				continue;
228 			}
229 		}
230 
231 		if (last != NULL) {
232 			struct mbuf *n;
233 
234 			mtx_leave(&rawin6pcbtable.inpt_mtx);
235 
236 			n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
237 			if (n != NULL)
238 				rip6_sbappend(last, n, ip6, *offp, &rip6src);
239 			in_pcbunref(last);
240 
241 			mtx_enter(&rawin6pcbtable.inpt_mtx);
242 		}
243 		last = in_pcbref(inp);
244 	}
245 	mtx_leave(&rawin6pcbtable.inpt_mtx);
246 
247 	if (last == NULL) {
248 		struct counters_ref ref;
249 		uint64_t *counters;
250 
251 		if (proto != IPPROTO_ICMPV6) {
252 			rip6stat_inc(rip6s_nosock);
253 			if (m->m_flags & M_MCAST)
254 				rip6stat_inc(rip6s_nosockmcast);
255 		}
256 		if (proto == IPPROTO_NONE || proto == IPPROTO_ICMPV6) {
257 			m_freem(m);
258 		} else {
259 			int prvnxt = ip6_get_prevhdr(m, *offp);
260 
261 			icmp6_error(m, ICMP6_PARAM_PROB,
262 			    ICMP6_PARAMPROB_NEXTHEADER, prvnxt);
263 		}
264 		counters = counters_enter(&ref, ip6counters);
265 		counters[ip6s_delivered]--;
266 		counters_leave(&ref, ip6counters);
267 
268 		return IPPROTO_DONE;
269 	}
270 
271 	rip6_sbappend(last, m, ip6, *offp, &rip6src);
272 	in_pcbunref(last);
273 
274 	return IPPROTO_DONE;
275 }
276 
277 void
rip6_sbappend(struct inpcb * inp,struct mbuf * m,struct ip6_hdr * ip6,int hlen,struct sockaddr_in6 * rip6src)278 rip6_sbappend(struct inpcb *inp, struct mbuf *m, struct ip6_hdr *ip6, int hlen,
279     struct sockaddr_in6 *rip6src)
280 {
281 	struct socket *so = inp->inp_socket;
282 	struct mbuf *opts = NULL;
283 	int ret = 0;
284 
285 	if (inp->inp_flags & IN6P_CONTROLOPTS)
286 		ip6_savecontrol(inp, m, &opts);
287 	/* strip intermediate headers */
288 	m_adj(m, hlen);
289 
290 	mtx_enter(&so->so_rcv.sb_mtx);
291 	if (!ISSET(inp->inp_socket->so_rcv.sb_state, SS_CANTRCVMORE))
292 		ret = sbappendaddr(so, &so->so_rcv, sin6tosa(rip6src), m, opts);
293 	mtx_leave(&so->so_rcv.sb_mtx);
294 
295 	if (ret == 0) {
296 		m_freem(m);
297 		m_freem(opts);
298 		rip6stat_inc(rip6s_fullsock);
299 	} else
300 		sorwakeup(so);
301 }
302 
303 void
rip6_ctlinput(int cmd,struct sockaddr * sa,u_int rdomain,void * d)304 rip6_ctlinput(int cmd, struct sockaddr *sa, u_int rdomain, void *d)
305 {
306 	struct ip6_hdr *ip6;
307 	struct ip6ctlparam *ip6cp = NULL;
308 	struct sockaddr_in6 *sa6 = satosin6(sa);
309 	const struct sockaddr_in6 *sa6_src = NULL;
310 	void *cmdarg;
311 	void (*notify)(struct inpcb *, int) = in_rtchange;
312 	int nxt;
313 
314 	if (sa->sa_family != AF_INET6 ||
315 	    sa->sa_len != sizeof(struct sockaddr_in6))
316 		return;
317 
318 	if ((unsigned)cmd >= PRC_NCMDS)
319 		return;
320 	if (PRC_IS_REDIRECT(cmd))
321 		notify = in_rtchange, d = NULL;
322 	else if (cmd == PRC_HOSTDEAD)
323 		d = NULL;
324 	else if (cmd == PRC_MSGSIZE)
325 		; /* special code is present, see below */
326 	else if (inet6ctlerrmap[cmd] == 0)
327 		return;
328 
329 	/* if the parameter is from icmp6, decode it. */
330 	if (d != NULL) {
331 		ip6cp = (struct ip6ctlparam *)d;
332 		ip6 = ip6cp->ip6c_ip6;
333 		cmdarg = ip6cp->ip6c_cmdarg;
334 		sa6_src = ip6cp->ip6c_src;
335 		nxt = ip6cp->ip6c_nxt;
336 	} else {
337 		ip6 = NULL;
338 		cmdarg = NULL;
339 		sa6_src = &sa6_any;
340 		nxt = -1;
341 	}
342 
343 	if (ip6 && cmd == PRC_MSGSIZE) {
344 		int valid = 0;
345 		struct inpcb *inp;
346 
347 		/*
348 		 * Check to see if we have a valid raw IPv6 socket
349 		 * corresponding to the address in the ICMPv6 message
350 		 * payload, and the protocol (ip6_nxt) meets the socket.
351 		 * XXX chase extension headers, or pass final nxt value
352 		 * from icmp6_notify_error()
353 		 */
354 		inp = in6_pcblookup(&rawin6pcbtable, &sa6->sin6_addr, 0,
355 		    &sa6_src->sin6_addr, 0, rdomain);
356 
357 		if (inp && inp->inp_ipv6.ip6_nxt &&
358 		    inp->inp_ipv6.ip6_nxt == nxt)
359 			valid = 1;
360 
361 		/*
362 		 * Depending on the value of "valid" and routing table
363 		 * size (mtudisc_{hi,lo}wat), we will:
364 		 * - recalculate the new MTU and create the
365 		 *   corresponding routing entry, or
366 		 * - ignore the MTU change notification.
367 		 */
368 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
369 		in_pcbunref(inp);
370 
371 		/*
372 		 * regardless of if we called icmp6_mtudisc_update(),
373 		 * we need to call in6_pcbnotify(), to notify path
374 		 * MTU change to the userland (2292bis-02), because
375 		 * some unconnected sockets may share the same
376 		 * destination and want to know the path MTU.
377 		 */
378 	}
379 
380 	in6_pcbnotify(&rawin6pcbtable, sa6, 0,
381 	    sa6_src, 0, rdomain, cmd, cmdarg, notify);
382 }
383 
384 /*
385  * Generate IPv6 header and pass packet to ip6_output.
386  * Tack on options user may have setup with control call.
387  */
388 int
rip6_output(struct mbuf * m,struct socket * so,struct sockaddr * dstaddr,struct mbuf * control)389 rip6_output(struct mbuf *m, struct socket *so, struct sockaddr *dstaddr,
390     struct mbuf *control)
391 {
392 	struct in6_addr *dst;
393 	struct ip6_hdr *ip6;
394 	struct inpcb *inp;
395 	u_int	plen = m->m_pkthdr.len;
396 	int error = 0;
397 	struct ip6_pktopts opt, *optp = NULL;
398 	int type;		/* for ICMPv6 output statistics only */
399 	int priv = 0;
400 	int flags;
401 
402 	inp = sotoinpcb(so);
403 
404 	priv = 0;
405 	if ((so->so_state & SS_PRIV) != 0)
406 		priv = 1;
407 	if (control) {
408 		if ((error = ip6_setpktopts(control, &opt,
409 		    inp->inp_outputopts6,
410 		    priv, so->so_proto->pr_protocol)) != 0)
411 			goto bad;
412 		optp = &opt;
413 	} else
414 		optp = inp->inp_outputopts6;
415 
416 	if (dstaddr->sa_family != AF_INET6) {
417 		error = EAFNOSUPPORT;
418 		goto bad;
419 	}
420 	dst = &satosin6(dstaddr)->sin6_addr;
421 	if (IN6_IS_ADDR_V4MAPPED(dst)) {
422 		error = EADDRNOTAVAIL;
423 		goto bad;
424 	}
425 
426 	/*
427 	 * For an ICMPv6 packet, we should know its type and code
428 	 * to update statistics.
429 	 */
430 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
431 		struct icmp6_hdr *icmp6;
432 		if (m->m_len < sizeof(struct icmp6_hdr) &&
433 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
434 			error = ENOBUFS;
435 			goto bad;
436 		}
437 		icmp6 = mtod(m, struct icmp6_hdr *);
438 		type = icmp6->icmp6_type;
439 	}
440 
441 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
442 	if (!m) {
443 		error = ENOBUFS;
444 		goto bad;
445 	}
446 	ip6 = mtod(m, struct ip6_hdr *);
447 
448 	/*
449 	 * Next header might not be ICMP6 but use its pseudo header anyway.
450 	 */
451 	ip6->ip6_dst = *dst;
452 
453 	/* KAME hack: embed scopeid */
454 	if (in6_embedscope(&ip6->ip6_dst, satosin6(dstaddr),
455 	    optp, inp->inp_moptions6) != 0) {
456 		error = EINVAL;
457 		goto bad;
458 	}
459 
460 	/*
461 	 * Source address selection.
462 	 */
463 	{
464 		const struct in6_addr *in6a;
465 
466 		error = in6_pcbselsrc(&in6a, satosin6(dstaddr), inp, optp);
467 		if (error)
468 			goto bad;
469 
470 		ip6->ip6_src = *in6a;
471 	}
472 
473 	ip6->ip6_flow = inp->inp_flowinfo & IPV6_FLOWINFO_MASK;
474 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
475 	ip6->ip6_vfc  |= IPV6_VERSION;
476 #if 0				/* ip6_plen will be filled in ip6_output. */
477 	ip6->ip6_plen  = htons((u_short)plen);
478 #endif
479 	ip6->ip6_nxt   = inp->inp_ipv6.ip6_nxt;
480 	ip6->ip6_hlim = in6_selecthlim(inp);
481 
482 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
483 	    inp->inp_cksum6 != -1) {
484 		struct mbuf *n;
485 		int off;
486 		u_int16_t *sump;
487 		int sumoff;
488 
489 		/* compute checksum */
490 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
491 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
492 		else
493 			off = inp->inp_cksum6;
494 		if (plen < 2 || plen - 2 < off) {
495 			error = EINVAL;
496 			goto bad;
497 		}
498 		off += sizeof(struct ip6_hdr);
499 
500 		n = m_pulldown(m, off, sizeof(*sump), &sumoff);
501 		if (n == NULL) {
502 			m = NULL;
503 			error = ENOBUFS;
504 			goto bad;
505 		}
506 		sump = (u_int16_t *)(mtod(n, caddr_t) + sumoff);
507 		*sump = 0;
508 		*sump = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
509 	}
510 
511 	flags = 0;
512 	if (inp->inp_flags & IN6P_MINMTU)
513 		flags |= IPV6_MINMTU;
514 
515 	/* force routing table */
516 	m->m_pkthdr.ph_rtableid = inp->inp_rtableid;
517 
518 #if NPF > 0
519 	if (inp->inp_socket->so_state & SS_ISCONNECTED &&
520 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
521 		pf_mbuf_link_inpcb(m, inp);
522 #endif
523 
524 	error = ip6_output(m, optp, &inp->inp_route, flags,
525 	    inp->inp_moptions6, &inp->inp_seclevel);
526 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
527 		icmp6stat_inc(icp6s_outhist + type);
528 	} else
529 		rip6stat_inc(rip6s_opackets);
530 
531 	goto freectl;
532 
533  bad:
534 	m_freem(m);
535 
536  freectl:
537 	if (control) {
538 		ip6_clearpktopts(&opt, -1);
539 		m_freem(control);
540 	}
541 	return (error);
542 }
543 
544 /*
545  * Raw IPv6 socket option processing.
546  */
547 int
rip6_ctloutput(int op,struct socket * so,int level,int optname,struct mbuf * m)548 rip6_ctloutput(int op, struct socket *so, int level, int optname,
549     struct mbuf *m)
550 {
551 #ifdef MROUTING
552 	int error;
553 #endif
554 
555 	switch (level) {
556 	case IPPROTO_IPV6:
557 		switch (optname) {
558 #ifdef MROUTING
559 		case MRT6_INIT:
560 		case MRT6_DONE:
561 		case MRT6_ADD_MIF:
562 		case MRT6_DEL_MIF:
563 		case MRT6_ADD_MFC:
564 		case MRT6_DEL_MFC:
565 			if (op == PRCO_SETOPT) {
566 				error = ip6_mrouter_set(optname, so, m);
567 			} else if (op == PRCO_GETOPT)
568 				error = ip6_mrouter_get(optname, so, m);
569 			else
570 				error = EINVAL;
571 			return (error);
572 #endif
573 		case IPV6_CHECKSUM:
574 			return (ip6_raw_ctloutput(op, so, level, optname, m));
575 		default:
576 			return (ip6_ctloutput(op, so, level, optname, m));
577 		}
578 
579 	case IPPROTO_ICMPV6:
580 		/*
581 		 * XXX: is it better to call icmp6_ctloutput() directly
582 		 * from protosw?
583 		 */
584 		return (icmp6_ctloutput(op, so, level, optname, m));
585 
586 	default:
587 		return EINVAL;
588 	}
589 }
590 
591 extern	u_long rip6_sendspace;
592 extern	u_long rip6_recvspace;
593 
594 int
rip6_attach(struct socket * so,int proto,int wait)595 rip6_attach(struct socket *so, int proto, int wait)
596 {
597 	struct inpcb *inp;
598 	int error;
599 
600 	if (so->so_pcb)
601 		panic("%s", __func__);
602 	if ((so->so_state & SS_PRIV) == 0)
603 		return (EACCES);
604 	if (proto < 0 || proto >= IPPROTO_MAX)
605 		return EPROTONOSUPPORT;
606 
607 	if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)))
608 		return error;
609 	NET_ASSERT_LOCKED();
610 	if ((error = in_pcballoc(so, &rawin6pcbtable, wait)))
611 		return error;
612 
613 	inp = sotoinpcb(so);
614 	inp->inp_ipv6.ip6_nxt = proto;
615 	inp->inp_cksum6 = -1;
616 
617 	inp->inp_icmp6filt = malloc(sizeof(struct icmp6_filter), M_PCB,
618 	    wait == M_WAIT ? M_WAITOK : M_NOWAIT);
619 	if (inp->inp_icmp6filt == NULL) {
620 		in_pcbdetach(inp);
621 		return ENOMEM;
622 	}
623 	ICMP6_FILTER_SETPASSALL(inp->inp_icmp6filt);
624 	return 0;
625 }
626 
627 int
rip6_detach(struct socket * so)628 rip6_detach(struct socket *so)
629 {
630 	struct inpcb *inp = sotoinpcb(so);
631 
632 	soassertlocked(so);
633 
634 	if (inp == NULL)
635 		panic("%s", __func__);
636 #ifdef MROUTING
637 	if (so == ip6_mrouter[inp->inp_rtableid])
638 		ip6_mrouter_done(so);
639 #endif
640 	free(inp->inp_icmp6filt, M_PCB, sizeof(struct icmp6_filter));
641 	inp->inp_icmp6filt = NULL;
642 
643 	in_pcbdetach(inp);
644 
645 	return (0);
646 }
647 
648 int
rip6_bind(struct socket * so,struct mbuf * nam,struct proc * p)649 rip6_bind(struct socket *so, struct mbuf *nam, struct proc *p)
650 {
651 	struct inpcb *inp = sotoinpcb(so);
652 	struct sockaddr_in6 *addr;
653 	int error;
654 
655 	soassertlocked(so);
656 
657 	if ((error = in6_nam2sin6(nam, &addr)))
658 		return (error);
659 
660 	/*
661 	 * Make sure to not enter in_pcblookup_local(), local ports
662 	 * are non-sensical for raw sockets.
663 	 */
664 	addr->sin6_port = 0;
665 
666 	if ((error = in6_pcbaddrisavail(inp, addr, 0, p)))
667 		return (error);
668 
669 	mtx_enter(&rawin6pcbtable.inpt_mtx);
670 	inp->inp_laddr6 = addr->sin6_addr;
671 	mtx_leave(&rawin6pcbtable.inpt_mtx);
672 
673 	return (0);
674 }
675 
676 int
rip6_connect(struct socket * so,struct mbuf * nam)677 rip6_connect(struct socket *so, struct mbuf *nam)
678 {
679 	struct inpcb *inp = sotoinpcb(so);
680 	struct sockaddr_in6 *addr;
681 	const struct in6_addr *in6a;
682 	int error;
683 
684 	soassertlocked(so);
685 
686 	if ((error = in6_nam2sin6(nam, &addr)))
687 		return (error);
688 
689 	/* Source address selection. XXX: need pcblookup? */
690 	error = in6_pcbselsrc(&in6a, addr, inp, inp->inp_outputopts6);
691 	if (error)
692 		return (error);
693 
694 	mtx_enter(&rawin6pcbtable.inpt_mtx);
695 	inp->inp_laddr6 = *in6a;
696 	inp->inp_faddr6 = addr->sin6_addr;
697 	mtx_leave(&rawin6pcbtable.inpt_mtx);
698 	soisconnected(so);
699 
700 	return (0);
701 }
702 
703 int
rip6_disconnect(struct socket * so)704 rip6_disconnect(struct socket *so)
705 {
706 	struct inpcb *inp = sotoinpcb(so);
707 
708 	soassertlocked(so);
709 
710 	if ((so->so_state & SS_ISCONNECTED) == 0)
711 		return (ENOTCONN);
712 
713 	soisdisconnected(so);
714 	mtx_enter(&rawin6pcbtable.inpt_mtx);
715 	inp->inp_faddr6 = in6addr_any;
716 	mtx_leave(&rawin6pcbtable.inpt_mtx);
717 
718 	return (0);
719 }
720 
721 int
rip6_shutdown(struct socket * so)722 rip6_shutdown(struct socket *so)
723 {
724 	/*
725 	 * Mark the connection as being incapable of further input.
726 	 */
727 	soassertlocked(so);
728 	socantsendmore(so);
729 	return (0);
730 }
731 
732 int
rip6_send(struct socket * so,struct mbuf * m,struct mbuf * nam,struct mbuf * control)733 rip6_send(struct socket *so, struct mbuf *m, struct mbuf *nam,
734 	struct mbuf *control)
735 {
736 	struct inpcb *inp = sotoinpcb(so);
737 	struct sockaddr_in6 dst;
738 	int error;
739 
740 	soassertlocked(so);
741 
742 	/*
743 	 * Ship a packet out. The appropriate raw output
744 	 * routine handles any messaging necessary.
745 	 */
746 
747 	/* always copy sockaddr to avoid overwrites */
748 	memset(&dst, 0, sizeof(dst));
749 	dst.sin6_family = AF_INET6;
750 	dst.sin6_len = sizeof(dst);
751 	if (so->so_state & SS_ISCONNECTED) {
752 		if (nam) {
753 			error = EISCONN;
754 			goto out;
755 		}
756 		dst.sin6_addr = inp->inp_faddr6;
757 	} else {
758 		struct sockaddr_in6 *addr6;
759 
760 		if (nam == NULL) {
761 			error = ENOTCONN;
762 			goto out;
763 		}
764 		if ((error = in6_nam2sin6(nam, &addr6)))
765 			goto out;
766 		dst.sin6_addr = addr6->sin6_addr;
767 		dst.sin6_scope_id = addr6->sin6_scope_id;
768 	}
769 	error = rip6_output(m, so, sin6tosa(&dst), control);
770 	control = NULL;
771 	m = NULL;
772 
773 out:
774 	m_freem(control);
775 	m_freem(m);
776 
777 	return (error);
778 }
779 
780 int
rip6_sysctl_rip6stat(void * oldp,size_t * oldplen,void * newp)781 rip6_sysctl_rip6stat(void *oldp, size_t *oldplen, void *newp)
782 {
783 	struct rip6stat rip6stat;
784 
785 	CTASSERT(sizeof(rip6stat) == rip6s_ncounters * sizeof(uint64_t));
786 	counters_read(rip6counters, (uint64_t *)&rip6stat, rip6s_ncounters,
787 	    NULL);
788 
789 	return (sysctl_rdstruct(oldp, oldplen, newp,
790 	    &rip6stat, sizeof(rip6stat)));
791 }
792 
793 int
rip6_sysctl(int * name,u_int namelen,void * oldp,size_t * oldlenp,void * newp,size_t newlen)794 rip6_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
795     void *newp, size_t newlen)
796 {
797 	/* All sysctl names at this level are terminal. */
798 	if (namelen != 1)
799 		return ENOTDIR;
800 
801 	switch (name[0]) {
802 	case RIPV6CTL_STATS:
803 		return (rip6_sysctl_rip6stat(oldp, oldlenp, newp));
804 	default:
805 		return (EOPNOTSUPP);
806 	}
807 	/* NOTREACHED */
808 }
809