xref: /dragonfly/sys/netinet/raw_ip.c (revision 1d1731fa)
1 /*
2  * Copyright (c) 1982, 1986, 1988, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
34  * $FreeBSD: src/sys/netinet/raw_ip.c,v 1.64.2.16 2003/08/24 08:24:38 hsu Exp $
35  * $DragonFly: src/sys/netinet/raw_ip.c,v 1.8 2003/09/03 13:19:02 hmp Exp $
36  */
37 
38 #include "opt_inet6.h"
39 #include "opt_ipsec.h"
40 #include "opt_random_ip_id.h"
41 
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/kernel.h>
45 #include <sys/malloc.h>
46 #include <sys/mbuf.h>
47 #include <sys/proc.h>
48 #include <sys/protosw.h>
49 #include <sys/socket.h>
50 #include <sys/socketvar.h>
51 #include <sys/sysctl.h>
52 
53 #include <vm/vm_zone.h>
54 
55 #include <net/if.h>
56 #include <net/route.h>
57 
58 #define _IP_VHL
59 #include <netinet/in.h>
60 #include <netinet/in_systm.h>
61 #include <netinet/ip.h>
62 #include <netinet/in_pcb.h>
63 #include <netinet/in_var.h>
64 #include <netinet/ip_var.h>
65 
66 #include <net/ip_mroute/ip_mroute.h>
67 #include <net/ipfw/ip_fw.h>
68 #include <net/dummynet/ip_dummynet.h>
69 
70 #ifdef FAST_IPSEC
71 #include <netipsec/ipsec.h>
72 #endif /*FAST_IPSEC*/
73 
74 #ifdef IPSEC
75 #include <netinet6/ipsec.h>
76 #endif /*IPSEC*/
77 
78 struct	inpcbhead ripcb;
79 struct	inpcbinfo ripcbinfo;
80 
81 /* control hooks for ipfw and dummynet */
82 ip_fw_ctl_t *ip_fw_ctl_ptr;
83 ip_dn_ctl_t *ip_dn_ctl_ptr;
84 
85 /*
86  * hooks for multicast routing. They all default to NULL,
87  * so leave them not initialized and rely on BSS being set to 0.
88  */
89 
90 /* The socket used to communicate with the multicast routing daemon.  */
91 struct socket  *ip_mrouter;
92 
93 /* The various mrouter and rsvp functions */
94 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
95 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
96 int (*ip_mrouter_done)(void);
97 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
98 		struct ip_moptions *);
99 int (*mrt_ioctl)(int, caddr_t);
100 int (*legal_vif_num)(int);
101 u_long (*ip_mcast_src)(int);
102 
103 void (*rsvp_input_p)(struct mbuf *m, int off, int proto);
104 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
105 void (*ip_rsvp_force_done)(struct socket *);
106 
107 /*
108  * Nominal space allocated to a raw ip socket.
109  */
110 #define	RIPSNDQ		8192
111 #define	RIPRCVQ		8192
112 
113 /*
114  * Raw interface to IP protocol.
115  */
116 
117 /*
118  * Initialize raw connection block queue.
119  */
120 void
121 rip_init(void)
122 {
123 	LIST_INIT(&ripcb);
124 	ripcbinfo.listhead = &ripcb;
125 	/*
126 	 * XXX We don't use the hash list for raw IP, but it's easier
127 	 * to allocate a one entry hash list than it is to check all
128 	 * over the place for hashbase == NULL.
129 	 */
130 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
131 	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
132 	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
133 				   maxsockets, ZONE_INTERRUPT, 0);
134 }
135 
136 /*
137  * XXX ripsrc is modified in rip_input, so we must be fix this
138  * when we want to make this code smp-friendly.
139  */
140 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
141 
142 /*
143  * Setup generic address and protocol structures
144  * for raw_input routine, then pass them along with
145  * mbuf chain.
146  */
147 void
148 rip_input(struct mbuf *m, int off, int proto)
149 {
150 	struct ip *ip = mtod(m, struct ip *);
151 	struct inpcb *inp;
152 	struct inpcb *last = NULL;
153 	struct mbuf *opts = NULL;
154 
155 	ripsrc.sin_addr = ip->ip_src;
156 	LIST_FOREACH(inp, &ripcb, inp_list) {
157 #ifdef INET6
158 		if ((inp->inp_vflag & INP_IPV4) == 0)
159 			continue;
160 #endif
161 		if (inp->inp_ip_p && inp->inp_ip_p != proto)
162 			continue;
163 		if (inp->inp_laddr.s_addr != INADDR_ANY &&
164 		    inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
165 			continue;
166 		if (inp->inp_faddr.s_addr != INADDR_ANY &&
167 		    inp->inp_faddr.s_addr != ip->ip_src.s_addr)
168 			continue;
169 		if (last) {
170 			struct mbuf *n = m_copypacket(m, M_DONTWAIT);
171 
172 #ifdef IPSEC
173 			/* check AH/ESP integrity. */
174 			if (n && ipsec4_in_reject_so(n, last->inp_socket)) {
175 				m_freem(n);
176 				ipsecstat.in_polvio++;
177 				/* do not inject data to pcb */
178 			} else
179 #endif /*IPSEC*/
180 #ifdef FAST_IPSEC
181 			/* check AH/ESP integrity. */
182 			if (ipsec4_in_reject(n, last)) {
183 				m_freem(n);
184 				/* do not inject data to pcb */
185 			} else
186 #endif /*FAST_IPSEC*/
187 			if (n) {
188 				if (last->inp_flags & INP_CONTROLOPTS ||
189 				    last->inp_socket->so_options & SO_TIMESTAMP)
190 				    ip_savecontrol(last, &opts, ip, n);
191 				if (sbappendaddr(&last->inp_socket->so_rcv,
192 				    (struct sockaddr *)&ripsrc, n,
193 				    opts) == 0) {
194 					/* should notify about lost packet */
195 					m_freem(n);
196 					if (opts)
197 					    m_freem(opts);
198 				} else
199 					sorwakeup(last->inp_socket);
200 				opts = 0;
201 			}
202 		}
203 		last = inp;
204 	}
205 #ifdef IPSEC
206 	/* check AH/ESP integrity. */
207 	if (last && ipsec4_in_reject_so(m, last->inp_socket)) {
208 		m_freem(m);
209 		ipsecstat.in_polvio++;
210 		ipstat.ips_delivered--;
211 		/* do not inject data to pcb */
212 	} else
213 #endif /*IPSEC*/
214 #ifdef FAST_IPSEC
215 	/* check AH/ESP integrity. */
216 	if (last && ipsec4_in_reject(m, last)) {
217 		m_freem(m);
218 		ipstat.ips_delivered--;
219 		/* do not inject data to pcb */
220 	} else
221 #endif /*FAST_IPSEC*/
222 	if (last) {
223 		if (last->inp_flags & INP_CONTROLOPTS ||
224 		    last->inp_socket->so_options & SO_TIMESTAMP)
225 			ip_savecontrol(last, &opts, ip, m);
226 		if (sbappendaddr(&last->inp_socket->so_rcv,
227 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
228 			m_freem(m);
229 			if (opts)
230 			    m_freem(opts);
231 		} else
232 			sorwakeup(last->inp_socket);
233 	} else {
234 		m_freem(m);
235 		ipstat.ips_noproto++;
236 		ipstat.ips_delivered--;
237 	}
238 }
239 
240 /*
241  * Generate IP header and pass packet to ip_output.
242  * Tack on options user may have setup with control call.
243  */
244 int
245 rip_output(struct mbuf *m, struct socket *so, u_long dst)
246 {
247 	struct ip *ip;
248 	struct inpcb *inp = sotoinpcb(so);
249 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
250 
251 	/*
252 	 * If the user handed us a complete IP packet, use it.
253 	 * Otherwise, allocate an mbuf for a header and fill it in.
254 	 */
255 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
256 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
257 			m_freem(m);
258 			return(EMSGSIZE);
259 		}
260 		M_PREPEND(m, sizeof(struct ip), M_WAIT);
261 		if (m == NULL)
262 			return(ENOBUFS);
263 		ip = mtod(m, struct ip *);
264 		ip->ip_tos = inp->inp_ip_tos;
265 		ip->ip_off = 0;
266 		ip->ip_p = inp->inp_ip_p;
267 		ip->ip_len = m->m_pkthdr.len;
268 		ip->ip_src = inp->inp_laddr;
269 		ip->ip_dst.s_addr = dst;
270 		ip->ip_ttl = inp->inp_ip_ttl;
271 	} else {
272 		if (m->m_pkthdr.len > IP_MAXPACKET) {
273 			m_freem(m);
274 			return(EMSGSIZE);
275 		}
276 		ip = mtod(m, struct ip *);
277 		/* don't allow both user specified and setsockopt options,
278 		   and don't allow packet length sizes that will crash */
279 		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
280 		     && inp->inp_options)
281 		    || (ip->ip_len > m->m_pkthdr.len)
282 		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
283 			m_freem(m);
284 			return EINVAL;
285 		}
286 		if (ip->ip_id == 0)
287 #ifdef RANDOM_IP_ID
288 			ip->ip_id = ip_randomid();
289 #else
290 			ip->ip_id = htons(ip_id++);
291 #endif
292 		/* XXX prevent ip_output from overwriting header fields */
293 		flags |= IP_RAWOUTPUT;
294 		ipstat.ips_rawout++;
295 	}
296 
297 	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
298 			  inp->inp_moptions, inp));
299 }
300 
301 /*
302  * Raw IP socket option processing.
303  */
304 int
305 rip_ctloutput(struct socket *so, struct sockopt *sopt)
306 {
307 	struct	inpcb *inp = sotoinpcb(so);
308 	int	error, optval;
309 
310 	if (sopt->sopt_level != IPPROTO_IP)
311 		return (EINVAL);
312 
313 	error = 0;
314 
315 	switch (sopt->sopt_dir) {
316 	case SOPT_GET:
317 		switch (sopt->sopt_name) {
318 		case IP_HDRINCL:
319 			optval = inp->inp_flags & INP_HDRINCL;
320 			error = sooptcopyout(sopt, &optval, sizeof optval);
321 			break;
322 
323 		case IP_FW_ADD: /* ADD actually returns the body... */
324 		case IP_FW_GET:
325 			if (IPFW_LOADED)
326 				error = ip_fw_ctl_ptr(sopt);
327 			else
328 				error = ENOPROTOOPT;
329 			break;
330 
331 		case IP_DUMMYNET_GET:
332 			if (DUMMYNET_LOADED)
333 				error = ip_dn_ctl_ptr(sopt);
334 			else
335 				error = ENOPROTOOPT;
336 			break ;
337 
338 		case MRT_INIT:
339 		case MRT_DONE:
340 		case MRT_ADD_VIF:
341 		case MRT_DEL_VIF:
342 		case MRT_ADD_MFC:
343 		case MRT_DEL_MFC:
344 		case MRT_VERSION:
345 		case MRT_ASSERT:
346 		case MRT_API_SUPPORT:
347 		case MRT_API_CONFIG:
348 		case MRT_ADD_BW_UPCALL:
349 		case MRT_DEL_BW_UPCALL:
350 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
351 				EOPNOTSUPP;
352 			break;
353 
354 		default:
355 			error = ip_ctloutput(so, sopt);
356 			break;
357 		}
358 		break;
359 
360 	case SOPT_SET:
361 		switch (sopt->sopt_name) {
362 		case IP_HDRINCL:
363 			error = sooptcopyin(sopt, &optval, sizeof optval,
364 					    sizeof optval);
365 			if (error)
366 				break;
367 			if (optval)
368 				inp->inp_flags |= INP_HDRINCL;
369 			else
370 				inp->inp_flags &= ~INP_HDRINCL;
371 			break;
372 
373 		case IP_FW_ADD:
374 		case IP_FW_DEL:
375 		case IP_FW_FLUSH:
376 		case IP_FW_ZERO:
377 		case IP_FW_RESETLOG:
378 			if (IPFW_LOADED)
379 				error = ip_fw_ctl_ptr(sopt);
380 			else
381 				error = ENOPROTOOPT;
382 			break;
383 
384 		case IP_DUMMYNET_CONFIGURE:
385 		case IP_DUMMYNET_DEL:
386 		case IP_DUMMYNET_FLUSH:
387 			if (DUMMYNET_LOADED)
388 				error = ip_dn_ctl_ptr(sopt);
389 			else
390 				error = ENOPROTOOPT ;
391 			break ;
392 
393 		case IP_RSVP_ON:
394 			error = ip_rsvp_init(so);
395 			break;
396 
397 		case IP_RSVP_OFF:
398 			error = ip_rsvp_done();
399 			break;
400 
401 		case IP_RSVP_VIF_ON:
402 		case IP_RSVP_VIF_OFF:
403 			error = ip_rsvp_vif ?
404 				ip_rsvp_vif(so, sopt) : EINVAL;
405 			break;
406 
407 		case MRT_INIT:
408 		case MRT_DONE:
409 		case MRT_ADD_VIF:
410 		case MRT_DEL_VIF:
411 		case MRT_ADD_MFC:
412 		case MRT_DEL_MFC:
413 		case MRT_VERSION:
414 		case MRT_ASSERT:
415 		case MRT_API_SUPPORT:
416 		case MRT_API_CONFIG:
417 		case MRT_ADD_BW_UPCALL:
418 		case MRT_DEL_BW_UPCALL:
419 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
420 					EOPNOTSUPP;
421 			break;
422 
423 		default:
424 			error = ip_ctloutput(so, sopt);
425 			break;
426 		}
427 		break;
428 	}
429 
430 	return (error);
431 }
432 
433 /*
434  * This function exists solely to receive the PRC_IFDOWN messages which
435  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
436  * and calls in_ifadown() to remove all routes corresponding to that address.
437  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
438  * interface routes.
439  */
440 void
441 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
442 {
443 	struct in_ifaddr *ia;
444 	struct ifnet *ifp;
445 	int err;
446 	int flags;
447 
448 	switch (cmd) {
449 	case PRC_IFDOWN:
450 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
451 			if (ia->ia_ifa.ifa_addr == sa
452 			    && (ia->ia_flags & IFA_ROUTE)) {
453 				/*
454 				 * in_ifscrub kills the interface route.
455 				 */
456 				in_ifscrub(ia->ia_ifp, ia);
457 				/*
458 				 * in_ifadown gets rid of all the rest of
459 				 * the routes.  This is not quite the right
460 				 * thing to do, but at least if we are running
461 				 * a routing process they will come back.
462 				 */
463 				in_ifadown(&ia->ia_ifa, 0);
464 				break;
465 			}
466 		}
467 		break;
468 
469 	case PRC_IFUP:
470 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
471 			if (ia->ia_ifa.ifa_addr == sa)
472 				break;
473 		}
474 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
475 			return;
476 		flags = RTF_UP;
477 		ifp = ia->ia_ifa.ifa_ifp;
478 
479 		if ((ifp->if_flags & IFF_LOOPBACK)
480 		    || (ifp->if_flags & IFF_POINTOPOINT))
481 			flags |= RTF_HOST;
482 
483 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
484 		if (err == 0)
485 			ia->ia_flags |= IFA_ROUTE;
486 		break;
487 	}
488 }
489 
490 u_long	rip_sendspace = RIPSNDQ;
491 u_long	rip_recvspace = RIPRCVQ;
492 
493 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
494     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
495 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
496     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
497 
498 static int
499 rip_attach(struct socket *so, int proto, struct thread *td)
500 {
501 	struct inpcb *inp;
502 	int error, s;
503 
504 	inp = sotoinpcb(so);
505 	if (inp)
506 		panic("rip_attach");
507 	if ((error = suser(td)) != 0)
508 		return error;
509 
510 	error = soreserve(so, rip_sendspace, rip_recvspace);
511 	if (error)
512 		return error;
513 	s = splnet();
514 	error = in_pcballoc(so, &ripcbinfo, td);
515 	splx(s);
516 	if (error)
517 		return error;
518 	inp = (struct inpcb *)so->so_pcb;
519 	inp->inp_vflag |= INP_IPV4;
520 	inp->inp_ip_p = proto;
521 	inp->inp_ip_ttl = ip_defttl;
522 	return 0;
523 }
524 
525 static int
526 rip_detach(struct socket *so)
527 {
528 	struct inpcb *inp;
529 
530 	inp = sotoinpcb(so);
531 	if (inp == 0)
532 		panic("rip_detach");
533 	if (so == ip_mrouter && ip_mrouter_done)
534 		ip_mrouter_done();
535 	if (ip_rsvp_force_done)
536 		ip_rsvp_force_done(so);
537 	if (so == ip_rsvpd)
538 		ip_rsvp_done();
539 	in_pcbdetach(inp);
540 	return 0;
541 }
542 
543 static int
544 rip_abort(struct socket *so)
545 {
546 	soisdisconnected(so);
547 	if (so->so_state & SS_NOFDREF)
548 		return rip_detach(so);
549 	return 0;
550 }
551 
552 static int
553 rip_disconnect(struct socket *so)
554 {
555 	if ((so->so_state & SS_ISCONNECTED) == 0)
556 		return ENOTCONN;
557 	return rip_abort(so);
558 }
559 
560 static int
561 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
562 {
563 	struct inpcb *inp = sotoinpcb(so);
564 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
565 
566 	if (nam->sa_len != sizeof(*addr))
567 		return EINVAL;
568 
569 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
570 				    (addr->sin_family != AF_IMPLINK)) ||
571 	    (addr->sin_addr.s_addr != INADDR_ANY &&
572 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
573 		return EADDRNOTAVAIL;
574 	inp->inp_laddr = addr->sin_addr;
575 	return 0;
576 }
577 
578 static int
579 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
580 {
581 	struct inpcb *inp = sotoinpcb(so);
582 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
583 
584 	if (nam->sa_len != sizeof(*addr))
585 		return EINVAL;
586 	if (TAILQ_EMPTY(&ifnet))
587 		return EADDRNOTAVAIL;
588 	if ((addr->sin_family != AF_INET) &&
589 	    (addr->sin_family != AF_IMPLINK))
590 		return EAFNOSUPPORT;
591 	inp->inp_faddr = addr->sin_addr;
592 	soisconnected(so);
593 	return 0;
594 }
595 
596 static int
597 rip_shutdown(struct socket *so)
598 {
599 	socantsendmore(so);
600 	return 0;
601 }
602 
603 static int
604 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
605 	 struct mbuf *control, struct thread *td)
606 {
607 	struct inpcb *inp = sotoinpcb(so);
608 	u_long dst;
609 
610 	if (so->so_state & SS_ISCONNECTED) {
611 		if (nam) {
612 			m_freem(m);
613 			return EISCONN;
614 		}
615 		dst = inp->inp_faddr.s_addr;
616 	} else {
617 		if (nam == NULL) {
618 			m_freem(m);
619 			return ENOTCONN;
620 		}
621 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
622 	}
623 	return rip_output(m, so, dst);
624 }
625 
626 static int
627 rip_pcblist(SYSCTL_HANDLER_ARGS)
628 {
629 	int error, i, n, s;
630 	struct inpcb *inp, **inp_list;
631 	inp_gen_t gencnt;
632 	struct xinpgen xig;
633 
634 	/*
635 	 * The process of preparing the TCB list is too time-consuming and
636 	 * resource-intensive to repeat twice on every request.
637 	 */
638 	if (req->oldptr == 0) {
639 		n = ripcbinfo.ipi_count;
640 		req->oldidx = 2 * (sizeof xig)
641 			+ (n + n/8) * sizeof(struct xinpcb);
642 		return 0;
643 	}
644 
645 	if (req->newptr != 0)
646 		return EPERM;
647 
648 	/*
649 	 * OK, now we're committed to doing something.
650 	 */
651 	s = splnet();
652 	gencnt = ripcbinfo.ipi_gencnt;
653 	n = ripcbinfo.ipi_count;
654 	splx(s);
655 
656 	xig.xig_len = sizeof xig;
657 	xig.xig_count = n;
658 	xig.xig_gen = gencnt;
659 	xig.xig_sogen = so_gencnt;
660 	error = SYSCTL_OUT(req, &xig, sizeof xig);
661 	if (error)
662 		return error;
663 
664 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
665 	if (inp_list == 0)
666 		return ENOMEM;
667 
668 	s = splnet();
669 	for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
670 	     inp = LIST_NEXT(inp, inp_list)) {
671 		if (inp->inp_gencnt <= gencnt)
672 			inp_list[i++] = inp;
673 	}
674 	splx(s);
675 	n = i;
676 
677 	error = 0;
678 	for (i = 0; i < n; i++) {
679 		inp = inp_list[i];
680 		if (inp->inp_gencnt <= gencnt) {
681 			struct xinpcb xi;
682 			xi.xi_len = sizeof xi;
683 			/* XXX should avoid extra copy */
684 			bcopy(inp, &xi.xi_inp, sizeof *inp);
685 			if (inp->inp_socket)
686 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
687 			error = SYSCTL_OUT(req, &xi, sizeof xi);
688 		}
689 	}
690 	if (!error) {
691 		/*
692 		 * Give the user an updated idea of our state.
693 		 * If the generation differs from what we told
694 		 * her before, she knows that something happened
695 		 * while we were processing this request, and it
696 		 * might be necessary to retry.
697 		 */
698 		s = splnet();
699 		xig.xig_gen = ripcbinfo.ipi_gencnt;
700 		xig.xig_sogen = so_gencnt;
701 		xig.xig_count = ripcbinfo.ipi_count;
702 		splx(s);
703 		error = SYSCTL_OUT(req, &xig, sizeof xig);
704 	}
705 	free(inp_list, M_TEMP);
706 	return error;
707 }
708 
709 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
710 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
711 
712 struct pr_usrreqs rip_usrreqs = {
713 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
714 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
715 	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
716 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
717 	in_setsockaddr, sosend, soreceive, sopoll
718 };
719