xref: /netbsd-src/sys/netinet6/raw_ip6.c (revision c42dbd0ed2e61fe6eda8590caa852ccf34719964)
1 /*	$NetBSD: raw_ip6.c,v 1.184 2024/02/24 21:41:13 mlelstv Exp $	*/
2 /*	$KAME: raw_ip6.c,v 1.82 2001/07/23 18:57:56 jinmei 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 <sys/cdefs.h>
65 __KERNEL_RCSID(0, "$NetBSD: raw_ip6.c,v 1.184 2024/02/24 21:41:13 mlelstv Exp $");
66 
67 #ifdef _KERNEL_OPT
68 #include "opt_ipsec.h"
69 #include "opt_net_mpsafe.h"
70 #endif
71 
72 #include <sys/param.h>
73 #include <sys/sysctl.h>
74 #include <sys/mbuf.h>
75 #include <sys/socket.h>
76 #include <sys/protosw.h>
77 #include <sys/socketvar.h>
78 #include <sys/systm.h>
79 #include <sys/proc.h>
80 #include <sys/kauth.h>
81 #include <sys/kmem.h>
82 
83 #include <net/if.h>
84 #include <net/if_types.h>
85 #include <net/net_stats.h>
86 
87 #include <netinet/in.h>
88 #include <netinet/in_var.h>
89 #include <netinet/ip6.h>
90 #include <netinet6/ip6_var.h>
91 #include <netinet6/ip6_private.h>
92 #include <netinet6/ip6_mroute.h>
93 #include <netinet/icmp6.h>
94 #include <netinet6/icmp6_private.h>
95 #include <netinet6/in6_pcb.h>
96 #include <netinet6/ip6protosw.h>
97 #include <netinet6/scope6_var.h>
98 #include <netinet6/raw_ip6.h>
99 
100 #ifdef IPSEC
101 #include <netipsec/ipsec.h>
102 #include <netipsec/ipsec6.h>
103 #endif
104 
105 #include "faith.h"
106 #if defined(NFAITH) && 0 < NFAITH
107 #include <net/if_faith.h>
108 #endif
109 
110 extern struct inpcbtable rawcbtable;
111 struct	inpcbtable raw6cbtable;
112 #define ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
113 
114 /*
115  * Raw interface to IP6 protocol.
116  */
117 
118 static percpu_t *rip6stat_percpu;
119 
120 #define	RIP6_STATINC(x)		_NET_STATINC(rip6stat_percpu, x)
121 
122 static void sysctl_net_inet6_raw6_setup(struct sysctllog **);
123 
124 /*
125  * Initialize raw connection block queue.
126  */
127 void
128 rip6_init(void)
129 {
130 
131 	sysctl_net_inet6_raw6_setup(NULL);
132 	in6pcb_init(&raw6cbtable, 1, 1);
133 
134 	rip6stat_percpu = percpu_alloc(sizeof(uint64_t) * RIP6_NSTATS);
135 }
136 
137 static void
138 rip6_sbappendaddr(struct inpcb *last, struct ip6_hdr *ip6,
139     const struct sockaddr *sa, int hlen, struct mbuf *n)
140 {
141 	struct mbuf *opts = NULL;
142 
143 	if (last->inp_flags & IN6P_CONTROLOPTS ||
144 	    SOOPT_TIMESTAMP(last->inp_socket->so_options))
145 		ip6_savecontrol(last, &opts, ip6, n);
146 
147 	m_adj(n, hlen);
148 
149 	if (sbappendaddr(&last->inp_socket->so_rcv, sa, n, opts) == 0) {
150 		soroverflow(last->inp_socket);
151 		m_freem(n);
152 		if (opts)
153 			m_freem(opts);
154 		RIP6_STATINC(RIP6_STAT_FULLSOCK);
155 	} else {
156 		sorwakeup(last->inp_socket);
157 	}
158 }
159 
160 /*
161  * Setup generic address and protocol structures
162  * for raw_input routine, then pass them along with
163  * mbuf chain.
164  */
165 int
166 rip6_input(struct mbuf **mp, int *offp, int proto)
167 {
168 	struct mbuf *m = *mp;
169 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
170 	struct inpcb *inp;
171 	struct inpcb *last = NULL;
172 	struct sockaddr_in6 rip6src;
173 	struct mbuf *n;
174 
175 	RIP6_STATINC(RIP6_STAT_IPACKETS);
176 
177 #if defined(NFAITH) && 0 < NFAITH
178 	if (faithprefix(&ip6->ip6_dst)) {
179 		/* send icmp6 host unreach? */
180 		m_freem(m);
181 		return IPPROTO_DONE;
182 	}
183 #endif
184 
185 	sockaddr_in6_init(&rip6src, &ip6->ip6_src, 0, 0, 0);
186 	if (sa6_recoverscope(&rip6src) != 0) {
187 		/* XXX: should be impossible. */
188 		m_freem(m);
189 		return IPPROTO_DONE;
190 	}
191 
192 	TAILQ_FOREACH(inp, &raw6cbtable.inpt_queue, inp_queue) {
193 		if (inp->inp_af != AF_INET6)
194 			continue;
195 		if (in6p_ip6(inp).ip6_nxt &&
196 		    in6p_ip6(inp).ip6_nxt != proto)
197 			continue;
198 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p_laddr(inp)) &&
199 		    !IN6_ARE_ADDR_EQUAL(&in6p_laddr(inp), &ip6->ip6_dst))
200 			continue;
201 		if (!IN6_IS_ADDR_UNSPECIFIED(&in6p_faddr(inp)) &&
202 		    !IN6_ARE_ADDR_EQUAL(&in6p_faddr(inp), &ip6->ip6_src))
203 			continue;
204 		if (in6p_cksum(inp) != -1) {
205 			RIP6_STATINC(RIP6_STAT_ISUM);
206 			/*
207 			 * Although in6_cksum() does not need the position of
208 			 * the checksum field for verification, enforce that it
209 			 * is located within the packet.  Userland has given
210 			 * a checksum offset, a packet too short for that is
211 			 * invalid.  Avoid overflow with user supplied offset.
212 			 */
213 			if (m->m_pkthdr.len < *offp + 2 ||
214 			    m->m_pkthdr.len - *offp - 2 < in6p_cksum(inp) ||
215 			    in6_cksum(m, proto, *offp,
216 			    m->m_pkthdr.len - *offp)) {
217 				RIP6_STATINC(RIP6_STAT_BADSUM);
218 				continue;
219 			}
220 		}
221 
222 		if (last == NULL) {
223 			;
224 		}
225 #ifdef IPSEC
226 		else if (ipsec_used && ipsec_in_reject(m, last)) {
227 			/* do not inject data into pcb */
228 		}
229 #endif
230 		else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) {
231 			rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src),
232 			    *offp, n);
233 		}
234 
235 		last = inp;
236 	}
237 
238 #ifdef IPSEC
239 	if (ipsec_used && last && ipsec_in_reject(m, last)) {
240 		m_freem(m);
241 		IP6_STATDEC(IP6_STAT_DELIVERED);
242 		/* do not inject data into pcb */
243 	} else
244 #endif
245 	if (last != NULL) {
246 		rip6_sbappendaddr(last, ip6, sin6tosa(&rip6src), *offp, m);
247 	} else {
248 		RIP6_STATINC(RIP6_STAT_NOSOCK);
249 		if (m->m_flags & M_MCAST)
250 			RIP6_STATINC(RIP6_STAT_NOSOCKMCAST);
251 		if (proto == IPPROTO_NONE)
252 			m_freem(m);
253 		else {
254 			int s;
255 			struct ifnet *rcvif = m_get_rcvif(m, &s);
256 			const int prvnxt = ip6_get_prevhdr(m, *offp);
257 			in6_ifstat_inc(rcvif, ifs6_in_protounknown);
258 			m_put_rcvif(rcvif, &s);
259 			icmp6_error(m, ICMP6_PARAM_PROB,
260 			    ICMP6_PARAMPROB_NEXTHEADER,
261 			    prvnxt);
262 		}
263 		IP6_STATDEC(IP6_STAT_DELIVERED);
264 	}
265 	return IPPROTO_DONE;
266 }
267 
268 void *
269 rip6_ctlinput(int cmd, const struct sockaddr *sa, void *d)
270 {
271 	struct ip6_hdr *ip6;
272 	struct ip6ctlparam *ip6cp = NULL;
273 	const struct sockaddr_in6 *sa6_src = NULL;
274 	void *cmdarg;
275 	void (*notify)(struct inpcb *, int) = in6pcb_rtchange;
276 	int nxt;
277 
278 	if (sa->sa_family != AF_INET6 ||
279 	    sa->sa_len != sizeof(struct sockaddr_in6))
280 		return NULL;
281 
282 	if ((unsigned)cmd >= PRC_NCMDS)
283 		return NULL;
284 	if (PRC_IS_REDIRECT(cmd))
285 		notify = in6pcb_rtchange, d = NULL;
286 	else if (cmd == PRC_HOSTDEAD)
287 		d = NULL;
288 	else if (cmd == PRC_MSGSIZE)
289 		; /* special code is present, see below */
290 	else if (inet6ctlerrmap[cmd] == 0)
291 		return NULL;
292 
293 	/* if the parameter is from icmp6, decode it. */
294 	if (d != NULL) {
295 		ip6cp = (struct ip6ctlparam *)d;
296 		ip6 = ip6cp->ip6c_ip6;
297 		cmdarg = ip6cp->ip6c_cmdarg;
298 		sa6_src = ip6cp->ip6c_src;
299 		nxt = ip6cp->ip6c_nxt;
300 	} else {
301 		ip6 = NULL;
302 		cmdarg = NULL;
303 		sa6_src = &sa6_any;
304 		nxt = -1;
305 	}
306 
307 	if (ip6 && cmd == PRC_MSGSIZE) {
308 		const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa;
309 		int valid = 0;
310 		struct inpcb *inp;
311 
312 		/*
313 		 * Check to see if we have a valid raw IPv6 socket
314 		 * corresponding to the address in the ICMPv6 message
315 		 * payload, and the protocol (ip6_nxt) meets the socket.
316 		 * XXX chase extension headers, or pass final nxt value
317 		 * from icmp6_notify_error()
318 		 */
319 		inp = NULL;
320 		inp = in6pcb_lookup(&raw6cbtable, &sa6->sin6_addr, 0,
321 					     (const struct in6_addr *)&sa6_src->sin6_addr, 0, 0, 0);
322 #if 0
323 		if (!inp) {
324 			/*
325 			 * As the use of sendto(2) is fairly popular,
326 			 * we may want to allow non-connected pcb too.
327 			 * But it could be too weak against attacks...
328 			 * We should at least check if the local
329 			 * address (= s) is really ours.
330 			 */
331 			inp = in6pcb_lookup_bound(&raw6cbtable,
332 			    &sa6->sin6_addr, 0, 0);
333 		}
334 #endif
335 
336 		if (inp && in6p_ip6(inp).ip6_nxt &&
337 		    in6p_ip6(inp).ip6_nxt == nxt)
338 			valid++;
339 
340 		/*
341 		 * Depending on the value of "valid" and routing table
342 		 * size (mtudisc_{hi,lo}wat), we will:
343 		 * - recalculate the new MTU and create the
344 		 *   corresponding routing entry, or
345 		 * - ignore the MTU change notification.
346 		 */
347 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
348 
349 		/*
350 		 * regardless of if we called icmp6_mtudisc_update(),
351 		 * we need to call in6pcb_notify(), to notify path MTU
352 		 * change to the userland (RFC3542), because some
353 		 * unconnected sockets may share the same destination
354 		 * and want to know the path MTU.
355 		 */
356 	}
357 
358 	(void) in6pcb_notify(&raw6cbtable, sa, 0,
359 	    sin6tocsa(sa6_src), 0, cmd, cmdarg, notify);
360 	return NULL;
361 }
362 
363 /*
364  * Generate IPv6 header and pass packet to ip6_output.
365  * Tack on options user may have setup with control call.
366  */
367 int
368 rip6_output(struct mbuf *m, struct socket * const so,
369     struct sockaddr_in6 * const dstsock, struct mbuf * const control)
370 {
371 	struct in6_addr *dst;
372 	struct ip6_hdr *ip6;
373 	struct inpcb *inp;
374 	u_int	plen = m->m_pkthdr.len;
375 	int error = 0;
376 	struct ip6_pktopts opt, *optp = NULL;
377 	struct ifnet *oifp = NULL;
378 	int type, code;		/* for ICMPv6 output statistics only */
379 	int scope_ambiguous = 0;
380 	int bound = curlwp_bind();
381 	struct psref psref;
382 
383 	inp = sotoinpcb(so);
384 
385 	dst = &dstsock->sin6_addr;
386 	if (control) {
387 		if ((error = ip6_setpktopts(control, &opt,
388 		    in6p_outputopts(inp),
389 		    kauth_cred_get(), so->so_proto->pr_protocol)) != 0) {
390 			goto bad;
391 		}
392 		optp = &opt;
393 	} else
394 		optp = in6p_outputopts(inp);
395 
396 	/*
397 	 * Check and convert scope zone ID into internal form.
398 	 * XXX: we may still need to determine the zone later.
399 	 */
400 	if (!(so->so_state & SS_ISCONNECTED)) {
401 		if (dstsock->sin6_scope_id == 0 && !ip6_use_defzone)
402 			scope_ambiguous = 1;
403 		if ((error = sa6_embedscope(dstsock, ip6_use_defzone)) != 0)
404 			goto bad;
405 	}
406 
407 	/*
408 	 * For an ICMPv6 packet, we should know its type and code
409 	 * to update statistics.
410 	 */
411 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
412 		struct icmp6_hdr *icmp6;
413 		if (m->m_len < sizeof(struct icmp6_hdr) &&
414 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
415 			error = ENOBUFS;
416 			goto bad;
417 		}
418 		icmp6 = mtod(m, struct icmp6_hdr *);
419 		type = icmp6->icmp6_type;
420 		code = icmp6->icmp6_code;
421 	} else {
422 		type = 0;
423 		code = 0;
424 	}
425 
426 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
427 	if (!m) {
428 		error = ENOBUFS;
429 		goto bad;
430 	}
431 	ip6 = mtod(m, struct ip6_hdr *);
432 
433 	/*
434 	 * Next header might not be ICMP6 but use its pseudo header anyway.
435 	 */
436 	ip6->ip6_dst = *dst;
437 
438 	/*
439 	 * Source address selection.
440 	 */
441 	error = in6_selectsrc(dstsock, optp, in6p_moptions(inp),
442 	    &inp->inp_route, &in6p_laddr(inp), &oifp, &psref, &ip6->ip6_src);
443 	if (error != 0)
444 		goto bad;
445 
446 	if (oifp && scope_ambiguous) {
447 		/*
448 		 * Application should provide a proper zone ID or the use of
449 		 * default zone IDs should be enabled.  Unfortunately, some
450 		 * applications do not behave as it should, so we need a
451 		 * workaround.  Even if an appropriate ID is not determined
452 		 * (when it's required), if we can determine the outgoing
453 		 * interface. determine the zone ID based on the interface.
454 		 */
455 		error = in6_setscope(&dstsock->sin6_addr, oifp, NULL);
456 		if (error != 0)
457 			goto bad;
458 	}
459 	ip6->ip6_dst = dstsock->sin6_addr;
460 
461 	/* fill in the rest of the IPv6 header fields */
462 	ip6->ip6_flow = in6p_flowinfo(inp) & IPV6_FLOWINFO_MASK;
463 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
464 	ip6->ip6_vfc  |= IPV6_VERSION;
465 	/* ip6_plen will be filled in ip6_output, so not fill it here. */
466 	ip6->ip6_nxt   = in6p_ip6(inp).ip6_nxt;
467 	ip6->ip6_hlim = in6pcb_selecthlim(inp, oifp);
468 
469 	if_put(oifp, &psref);
470 	oifp = NULL;
471 
472 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
473 	    in6p_cksum(inp) != -1) {
474 		const uint8_t nxt = ip6->ip6_nxt;
475 		int off;
476 		u_int16_t sum;
477 
478 		/* compute checksum */
479 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
480 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
481 		else
482 			off = in6p_cksum(inp);
483 		if (plen < 2 || plen - 2 < off) {
484 			error = EINVAL;
485 			goto bad;
486 		}
487 		off += sizeof(struct ip6_hdr);
488 
489 		sum = 0;
490 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
491 		    M_DONTWAIT);
492 		if (m == NULL) {
493 			error = ENOBUFS;
494 			goto bad;
495 		}
496 		sum = in6_cksum(m, nxt, sizeof(*ip6), plen);
497 		m = m_copyback_cow(m, off, sizeof(sum), (void *)&sum,
498 		    M_DONTWAIT);
499 		if (m == NULL) {
500 			error = ENOBUFS;
501 			goto bad;
502 		}
503 	}
504 
505 	{
506 		struct ifnet *ret_oifp = NULL;
507 
508 		error = ip6_output(m, optp, &inp->inp_route, 0,
509 		    in6p_moptions(inp), inp, &ret_oifp);
510 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
511 			if (ret_oifp)
512 				icmp6_ifoutstat_inc(ret_oifp, type, code);
513 			ICMP6_STATINC(ICMP6_STAT_OUTHIST + type);
514 		} else
515 			RIP6_STATINC(RIP6_STAT_OPACKETS);
516 	}
517 
518 	goto freectl;
519 
520  bad:
521 	if (m)
522 		m_freem(m);
523 
524  freectl:
525 	if (control) {
526 		ip6_clearpktopts(&opt, -1);
527 		m_freem(control);
528 	}
529 	if_put(oifp, &psref);
530 	curlwp_bindx(bound);
531 	return error;
532 }
533 
534 /*
535  * Raw IPv6 socket option processing.
536  */
537 int
538 rip6_ctloutput(int op, struct socket *so, struct sockopt *sopt)
539 {
540 	int error = 0;
541 
542 	if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == SO_NOHEADER) {
543 		int optval;
544 
545 		/* need to fiddle w/ opt(IPPROTO_IPV6, IPV6_CHECKSUM)? */
546 		if (op == PRCO_GETOPT) {
547 			optval = 1;
548 			error = sockopt_set(sopt, &optval, sizeof(optval));
549 		} else if (op == PRCO_SETOPT) {
550 			error = sockopt_getint(sopt, &optval);
551 			if (error)
552 				goto out;
553 			if (optval == 0)
554 				error = EINVAL;
555 		}
556 
557 		goto out;
558 	} else if (sopt->sopt_level != IPPROTO_IPV6)
559 		return ip6_ctloutput(op, so, sopt);
560 
561 	switch (sopt->sopt_name) {
562 	case MRT6_INIT:
563 	case MRT6_DONE:
564 	case MRT6_ADD_MIF:
565 	case MRT6_DEL_MIF:
566 	case MRT6_ADD_MFC:
567 	case MRT6_DEL_MFC:
568 	case MRT6_PIM:
569 		if (op == PRCO_SETOPT)
570 			error = ip6_mrouter_set(so, sopt);
571 		else if (op == PRCO_GETOPT)
572 			error = ip6_mrouter_get(so, sopt);
573 		else
574 			error = EINVAL;
575 		break;
576 	case IPV6_CHECKSUM:
577 		return ip6_raw_ctloutput(op, so, sopt);
578 	default:
579 		return ip6_ctloutput(op, so, sopt);
580 	}
581  out:
582 	return error;
583 }
584 
585 extern	u_long rip6_sendspace;
586 extern	u_long rip6_recvspace;
587 
588 int
589 rip6_attach(struct socket *so, int proto)
590 {
591 	struct inpcb *inp;
592 	int s, error;
593 
594 	KASSERT(sotoinpcb(so) == NULL);
595 	sosetlock(so);
596 
597 	error = kauth_authorize_network(kauth_cred_get(),
598 	    KAUTH_NETWORK_SOCKET, KAUTH_REQ_NETWORK_SOCKET_RAWSOCK,
599 	    KAUTH_ARG(AF_INET6),
600 	    KAUTH_ARG(SOCK_RAW),
601 	    KAUTH_ARG(so->so_proto->pr_protocol));
602 	if (error) {
603 		return error;
604 	}
605 	s = splsoftnet();
606 	error = soreserve(so, rip6_sendspace, rip6_recvspace);
607 	if (error) {
608 		splx(s);
609 		return error;
610 	}
611 	if ((error = inpcb_create(so, &raw6cbtable)) != 0) {
612 		splx(s);
613 		return error;
614 	}
615 	splx(s);
616 	inp = sotoinpcb(so);
617 	in6p_ip6(inp).ip6_nxt = proto;
618 	in6p_cksum(inp) = -1;
619 
620 	in6p_icmp6filt(inp) = kmem_alloc(sizeof(struct icmp6_filter), KM_SLEEP);
621 	ICMP6_FILTER_SETPASSALL(in6p_icmp6filt(inp));
622 	KASSERT(solocked(so));
623 	return error;
624 }
625 
626 static void
627 rip6_detach(struct socket *so)
628 {
629 	struct inpcb *inp = sotoinpcb(so);
630 
631 	KASSERT(solocked(so));
632 	KASSERT(inp != NULL);
633 
634 	if (so == ip6_mrouter) {
635 		ip6_mrouter_done();
636 	}
637 	/* xxx: RSVP */
638 	if (in6p_icmp6filt(inp) != NULL) {
639 		kmem_free(in6p_icmp6filt(inp), sizeof(struct icmp6_filter));
640 		in6p_icmp6filt(inp) = NULL;
641 	}
642 	inpcb_destroy(inp);
643 }
644 
645 static int
646 rip6_accept(struct socket *so, struct sockaddr *nam)
647 {
648 	KASSERT(solocked(so));
649 
650 	return EOPNOTSUPP;
651 }
652 
653 static int
654 rip6_bind(struct socket *so, struct sockaddr *nam, struct lwp *l)
655 {
656 	struct inpcb *inp = sotoinpcb(so);
657 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
658 	struct ifaddr *ifa = NULL;
659 	int error = 0;
660 	int s;
661 
662 	KASSERT(solocked(so));
663 	KASSERT(inp != NULL);
664 	KASSERT(nam != NULL);
665 
666 	if (addr->sin6_len != sizeof(*addr))
667 		return EINVAL;
668 	if (IFNET_READER_EMPTY() || addr->sin6_family != AF_INET6)
669 		return EADDRNOTAVAIL;
670 
671 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
672 		return error;
673 
674 	/*
675 	 * we don't support mapped address here, it would confuse
676 	 * users so reject it
677 	 */
678 	if (IN6_IS_ADDR_V4MAPPED(&addr->sin6_addr))
679 		return EADDRNOTAVAIL;
680 	s = pserialize_read_enter();
681 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
682 	    (ifa = ifa_ifwithaddr(sin6tosa(addr))) == NULL) {
683 		error = EADDRNOTAVAIL;
684 		goto out;
685 	}
686 	if (ifa && (ifatoia6(ifa))->ia6_flags &
687 	    (IN6_IFF_ANYCAST | IN6_IFF_DUPLICATED)) {
688 		error = EADDRNOTAVAIL;
689 		goto out;
690 	}
691 
692 	in6p_laddr(inp) = addr->sin6_addr;
693 	error = 0;
694 out:
695 	pserialize_read_exit(s);
696 	return error;
697 }
698 
699 static int
700 rip6_listen(struct socket *so, struct lwp *l)
701 {
702 	KASSERT(solocked(so));
703 
704 	return EOPNOTSUPP;
705 }
706 
707 static int
708 rip6_connect(struct socket *so, struct sockaddr *nam, struct lwp *l)
709 {
710 	struct inpcb *inp = sotoinpcb(so);
711 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
712 	struct in6_addr in6a;
713 	struct ifnet *ifp = NULL;
714 	int scope_ambiguous = 0;
715 	int error = 0;
716 	struct psref psref;
717 	int bound;
718 
719 	KASSERT(solocked(so));
720 	KASSERT(inp != NULL);
721 	KASSERT(nam != NULL);
722 
723 	if (IFNET_READER_EMPTY())
724 		return EADDRNOTAVAIL;
725 	if (addr->sin6_family != AF_INET6)
726 		return EAFNOSUPPORT;
727 	if (addr->sin6_len != sizeof(*addr))
728 		return EINVAL;
729 
730 	/*
731 	 * Application should provide a proper zone ID or the use of
732 	 * default zone IDs should be enabled.  Unfortunately, some
733 	 * applications do not behave as it should, so we need a
734 	 * workaround.  Even if an appropriate ID is not determined,
735 	 * we'll see if we can determine the outgoing interface.  If we
736 	 * can, determine the zone ID based on the interface below.
737 	 */
738 	if (addr->sin6_scope_id == 0 && !ip6_use_defzone)
739 		scope_ambiguous = 1;
740 	if ((error = sa6_embedscope(addr, ip6_use_defzone)) != 0)
741 		return error;
742 
743 	bound = curlwp_bind();
744 	/* Source address selection. XXX: need pcblookup? */
745 	error = in6_selectsrc(addr, in6p_outputopts(inp),
746 	    in6p_moptions(inp), &inp->inp_route,
747 	    &in6p_laddr(inp), &ifp, &psref, &in6a);
748 	if (error != 0)
749 		goto out;
750 	/* XXX: see above */
751 	if (ifp && scope_ambiguous &&
752 	    (error = in6_setscope(&addr->sin6_addr, ifp, NULL)) != 0) {
753 		goto out;
754 	}
755 	in6p_laddr(inp) = in6a;
756 	in6p_faddr(inp) = addr->sin6_addr;
757 	soisconnected(so);
758 out:
759 	if_put(ifp, &psref);
760 	curlwp_bindx(bound);
761 	return error;
762 }
763 
764 static int
765 rip6_connect2(struct socket *so, struct socket *so2)
766 {
767 	KASSERT(solocked(so));
768 
769 	return EOPNOTSUPP;
770 }
771 
772 static int
773 rip6_disconnect(struct socket *so)
774 {
775 	struct inpcb *inp = sotoinpcb(so);
776 
777 	KASSERT(solocked(so));
778 	KASSERT(inp != NULL);
779 
780 	if ((so->so_state & SS_ISCONNECTED) == 0)
781 		return ENOTCONN;
782 
783 	in6p_faddr(inp) = in6addr_any;
784 	so->so_state &= ~SS_ISCONNECTED;	/* XXX */
785 	return 0;
786 }
787 
788 static int
789 rip6_shutdown(struct socket *so)
790 {
791 	KASSERT(solocked(so));
792 
793 	/*
794 	 * Mark the connection as being incapable of further input.
795 	 */
796 	socantsendmore(so);
797 	return 0;
798 }
799 
800 static int
801 rip6_abort(struct socket *so)
802 {
803 	KASSERT(solocked(so));
804 
805 	soisdisconnected(so);
806 	rip6_detach(so);
807 	return 0;
808 }
809 
810 static int
811 rip6_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
812 {
813 	return in6_control(so, cmd, nam, ifp);
814 }
815 
816 static int
817 rip6_stat(struct socket *so, struct stat *ub)
818 {
819 	KASSERT(solocked(so));
820 
821 	/* stat: don't bother with a blocksize */
822 	return 0;
823 }
824 
825 static int
826 rip6_peeraddr(struct socket *so, struct sockaddr *nam)
827 {
828 	KASSERT(solocked(so));
829 	KASSERT(sotoinpcb(so) != NULL);
830 	KASSERT(nam != NULL);
831 
832 	in6pcb_fetch_peeraddr(sotoinpcb(so), (struct sockaddr_in6 *)nam);
833 	return 0;
834 }
835 
836 static int
837 rip6_sockaddr(struct socket *so, struct sockaddr *nam)
838 {
839 	KASSERT(solocked(so));
840 	KASSERT(sotoinpcb(so) != NULL);
841 	KASSERT(nam != NULL);
842 
843 	in6pcb_fetch_sockaddr(sotoinpcb(so), (struct sockaddr_in6 *)nam);
844 	return 0;
845 }
846 
847 static int
848 rip6_rcvd(struct socket *so, int flags, struct lwp *l)
849 {
850 	KASSERT(solocked(so));
851 
852 	return EOPNOTSUPP;
853 }
854 
855 static int
856 rip6_recvoob(struct socket *so, struct mbuf *m, int flags)
857 {
858 	KASSERT(solocked(so));
859 
860 	return EOPNOTSUPP;
861 }
862 
863 static int
864 rip6_send(struct socket *so, struct mbuf *m, struct sockaddr *nam,
865     struct mbuf *control, struct lwp *l)
866 {
867 	struct inpcb *inp = sotoinpcb(so);
868 	struct sockaddr_in6 tmp;
869 	struct sockaddr_in6 *dst;
870 	int error = 0;
871 
872 	KASSERT(solocked(so));
873 	KASSERT(inp != NULL);
874 	KASSERT(m != NULL);
875 
876 	/*
877 	 * Ship a packet out. The appropriate raw output
878 	 * routine handles any messaging necessary.
879 	 */
880 
881 	/* always copy sockaddr to avoid overwrites */
882 	if (so->so_state & SS_ISCONNECTED) {
883 		if (nam) {
884 			error = EISCONN;
885 			goto release;
886 		}
887 		/* XXX */
888 		sockaddr_in6_init(&tmp, &in6p_faddr(inp), 0, 0, 0);
889 		dst = &tmp;
890 	} else {
891 		if (nam == NULL) {
892 			error = ENOTCONN;
893 			goto release;
894 		}
895 		tmp = *(struct sockaddr_in6 *)nam;
896 		dst = &tmp;
897 
898 		if (dst->sin6_family != AF_INET6) {
899 			error = EAFNOSUPPORT;
900 			goto release;
901 		}
902 		if (dst->sin6_len != sizeof(*dst)) {
903 			error = EINVAL;
904 			goto release;
905 		}
906 	}
907 	error = rip6_output(m, so, dst, control);
908 	m = NULL;
909 
910 release:
911 	if (m)
912 		m_freem(m);
913 
914 	return error;
915 }
916 
917 static int
918 rip6_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
919 {
920 	KASSERT(solocked(so));
921 
922 	m_freem(m);
923 	m_freem(control);
924 
925 	return EOPNOTSUPP;
926 }
927 
928 static int
929 rip6_purgeif(struct socket *so, struct ifnet *ifp)
930 {
931 
932 	mutex_enter(softnet_lock);
933 	in6pcb_purgeif0(&raw6cbtable, ifp);
934 #ifdef NET_MPSAFE
935 	mutex_exit(softnet_lock);
936 #endif
937 	in6_purgeif(ifp);
938 #ifdef NET_MPSAFE
939 	mutex_enter(softnet_lock);
940 #endif
941 	in6pcb_purgeif(&raw6cbtable, ifp);
942 	mutex_exit(softnet_lock);
943 
944 	return 0;
945 }
946 
947 static int
948 sysctl_net_inet6_raw6_stats(SYSCTLFN_ARGS)
949 {
950 
951 	return (NETSTAT_SYSCTL(rip6stat_percpu, RIP6_NSTATS));
952 }
953 
954 static void
955 sysctl_net_inet6_raw6_setup(struct sysctllog **clog)
956 {
957 
958 	sysctl_createv(clog, 0, NULL, NULL,
959 		       CTLFLAG_PERMANENT,
960 		       CTLTYPE_NODE, "inet6", NULL,
961 		       NULL, 0, NULL, 0,
962 		       CTL_NET, PF_INET6, CTL_EOL);
963 	sysctl_createv(clog, 0, NULL, NULL,
964 		       CTLFLAG_PERMANENT,
965 		       CTLTYPE_NODE, "raw6",
966 		       SYSCTL_DESCR("Raw IPv6 settings"),
967 		       NULL, 0, NULL, 0,
968 		       CTL_NET, PF_INET6, IPPROTO_RAW, CTL_EOL);
969 
970 	sysctl_createv(clog, 0, NULL, NULL,
971 		       CTLFLAG_PERMANENT,
972 		       CTLTYPE_STRUCT, "pcblist",
973 		       SYSCTL_DESCR("Raw IPv6 control block list"),
974 		       sysctl_inpcblist, 0, &raw6cbtable, 0,
975 		       CTL_NET, PF_INET6, IPPROTO_RAW,
976 		       CTL_CREATE, CTL_EOL);
977 	sysctl_createv(clog, 0, NULL, NULL,
978 		       CTLFLAG_PERMANENT,
979 		       CTLTYPE_STRUCT, "stats",
980 		       SYSCTL_DESCR("Raw IPv6 statistics"),
981 		       sysctl_net_inet6_raw6_stats, 0, NULL, 0,
982 		       CTL_NET, PF_INET6, IPPROTO_RAW, RAW6CTL_STATS,
983 		       CTL_EOL);
984 }
985 
986 PR_WRAP_USRREQS(rip6)
987 #define	rip6_attach		rip6_attach_wrapper
988 #define	rip6_detach		rip6_detach_wrapper
989 #define	rip6_accept		rip6_accept_wrapper
990 #define	rip6_bind		rip6_bind_wrapper
991 #define	rip6_listen		rip6_listen_wrapper
992 #define	rip6_connect		rip6_connect_wrapper
993 #define	rip6_connect2		rip6_connect2_wrapper
994 #define	rip6_disconnect		rip6_disconnect_wrapper
995 #define	rip6_shutdown		rip6_shutdown_wrapper
996 #define	rip6_abort		rip6_abort_wrapper
997 #define	rip6_ioctl		rip6_ioctl_wrapper
998 #define	rip6_stat		rip6_stat_wrapper
999 #define	rip6_peeraddr		rip6_peeraddr_wrapper
1000 #define	rip6_sockaddr		rip6_sockaddr_wrapper
1001 #define	rip6_rcvd		rip6_rcvd_wrapper
1002 #define	rip6_recvoob		rip6_recvoob_wrapper
1003 #define	rip6_send		rip6_send_wrapper
1004 #define	rip6_sendoob		rip6_sendoob_wrapper
1005 #define	rip6_purgeif		rip6_purgeif_wrapper
1006 
1007 const struct pr_usrreqs rip6_usrreqs = {
1008 	.pr_attach	= rip6_attach,
1009 	.pr_detach	= rip6_detach,
1010 	.pr_accept	= rip6_accept,
1011 	.pr_bind	= rip6_bind,
1012 	.pr_listen	= rip6_listen,
1013 	.pr_connect	= rip6_connect,
1014 	.pr_connect2	= rip6_connect2,
1015 	.pr_disconnect	= rip6_disconnect,
1016 	.pr_shutdown	= rip6_shutdown,
1017 	.pr_abort	= rip6_abort,
1018 	.pr_ioctl	= rip6_ioctl,
1019 	.pr_stat	= rip6_stat,
1020 	.pr_peeraddr	= rip6_peeraddr,
1021 	.pr_sockaddr	= rip6_sockaddr,
1022 	.pr_rcvd	= rip6_rcvd,
1023 	.pr_recvoob	= rip6_recvoob,
1024 	.pr_send	= rip6_send,
1025 	.pr_sendoob	= rip6_sendoob,
1026 	.pr_purgeif	= rip6_purgeif,
1027 };
1028