xref: /dragonfly/sys/netinet/raw_ip.c (revision ef3ac1d1)
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. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
30  * $FreeBSD: src/sys/netinet/raw_ip.c,v 1.64.2.16 2003/08/24 08:24:38 hsu Exp $
31  */
32 
33 #include "opt_inet6.h"
34 #include "opt_ipsec.h"
35 #include "opt_carp.h"
36 
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/kernel.h>
40 #include <sys/jail.h>
41 #include <sys/malloc.h>
42 #include <sys/mbuf.h>
43 #include <sys/proc.h>
44 #include <sys/priv.h>
45 #include <sys/protosw.h>
46 #include <sys/socket.h>
47 #include <sys/socketvar.h>
48 #include <sys/sysctl.h>
49 
50 #include <sys/thread2.h>
51 #include <sys/socketvar2.h>
52 #include <sys/msgport2.h>
53 
54 #include <machine/stdarg.h>
55 
56 #include <net/if.h>
57 #ifdef CARP
58 #include <net/if_types.h>
59 #endif
60 #include <net/route.h>
61 
62 #define _IP_VHL
63 #include <netinet/in.h>
64 #include <netinet/in_systm.h>
65 #include <netinet/ip.h>
66 #include <netinet/in_pcb.h>
67 #include <netinet/in_var.h>
68 #include <netinet/ip_var.h>
69 
70 #include <net/ip_mroute/ip_mroute.h>
71 #include <net/ipfw/ip_fw.h>
72 #include <net/dummynet/ip_dummynet.h>
73 
74 #ifdef FAST_IPSEC
75 #include <netproto/ipsec/ipsec.h>
76 #endif /*FAST_IPSEC*/
77 
78 #ifdef IPSEC
79 #include <netinet6/ipsec.h>
80 #endif /*IPSEC*/
81 
82 struct	inpcbinfo ripcbinfo;
83 struct	inpcbportinfo ripcbportinfo;
84 
85 /* control hooks for ipfw and dummynet */
86 ip_fw_ctl_t *ip_fw_ctl_ptr;
87 ip_dn_ctl_t *ip_dn_ctl_ptr;
88 
89 /*
90  * hooks for multicast routing. They all default to NULL,
91  * so leave them not initialized and rely on BSS being set to 0.
92  */
93 
94 /* The socket used to communicate with the multicast routing daemon.  */
95 struct socket  *ip_mrouter;
96 
97 /* The various mrouter and rsvp functions */
98 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
99 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
100 int (*ip_mrouter_done)(void);
101 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
102 		struct ip_moptions *);
103 int (*mrt_ioctl)(int, caddr_t);
104 int (*legal_vif_num)(int);
105 u_long (*ip_mcast_src)(int);
106 
107 int (*rsvp_input_p)(struct mbuf **, int *, int);
108 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
109 void (*ip_rsvp_force_done)(struct socket *);
110 
111 /*
112  * Nominal space allocated to a raw ip socket.
113  */
114 #define	RIPSNDQ		8192
115 #define	RIPRCVQ		8192
116 
117 /*
118  * Raw interface to IP protocol.
119  */
120 
121 /*
122  * Initialize raw connection block queue.
123  */
124 void
125 rip_init(void)
126 {
127 	in_pcbinfo_init(&ripcbinfo, 0, FALSE);
128 	in_pcbportinfo_init(&ripcbportinfo, 1, FALSE, 0);
129 	/*
130 	 * XXX We don't use the hash list for raw IP, but it's easier
131 	 * to allocate a one entry hash list than it is to check all
132 	 * over the place for hashbase == NULL.
133 	 */
134 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
135 	ripcbinfo.portinfo = &ripcbportinfo;
136 	ripcbinfo.wildcardhashbase = hashinit(1, M_PCB,
137 					      &ripcbinfo.wildcardhashmask);
138 	ripcbinfo.ipi_size = sizeof(struct inpcb);
139 }
140 
141 /*
142  * Setup generic address and protocol structures
143  * for raw_input routine, then pass them along with
144  * mbuf chain.
145  */
146 int
147 rip_input(struct mbuf **mp, int *offp, int proto)
148 {
149 	struct sockaddr_in ripsrc = { sizeof ripsrc, AF_INET };
150 	struct mbuf *m = *mp;
151 	struct ip *ip = mtod(m, struct ip *);
152 	struct inpcb *inp;
153 	struct inpcb *last = NULL;
154 	struct mbuf *opts = NULL;
155 
156 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
157 	    ("not in netisr0"));
158 
159 	*mp = NULL;
160 
161 	ripsrc.sin_addr = ip->ip_src;
162 	LIST_FOREACH(inp, &ripcbinfo.pcblisthead, inp_list) {
163 		if (inp->inp_flags & INP_PLACEMARKER)
164 			continue;
165 #ifdef INET6
166 		if ((inp->inp_vflag & INP_IPV4) == 0)
167 			continue;
168 #endif
169 		if (inp->inp_ip_p && inp->inp_ip_p != proto)
170 			continue;
171 		if (inp->inp_laddr.s_addr != INADDR_ANY &&
172 		    inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
173 			continue;
174 		if (inp->inp_faddr.s_addr != INADDR_ANY &&
175 		    inp->inp_faddr.s_addr != ip->ip_src.s_addr)
176 			continue;
177 		if (last) {
178 			struct mbuf *n = m_copypacket(m, MB_DONTWAIT);
179 
180 #ifdef IPSEC
181 			/* check AH/ESP integrity. */
182 			if (n && ipsec4_in_reject_so(n, last->inp_socket)) {
183 				m_freem(n);
184 				ipsecstat.in_polvio++;
185 				/* do not inject data to pcb */
186 			} else
187 #endif /*IPSEC*/
188 #ifdef FAST_IPSEC
189 			/* check AH/ESP integrity. */
190 			if (ipsec4_in_reject(n, last)) {
191 				m_freem(n);
192 				/* do not inject data to pcb */
193 			} else
194 #endif /*FAST_IPSEC*/
195 			if (n) {
196 				lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
197 				if (last->inp_flags & INP_CONTROLOPTS ||
198 				    last->inp_socket->so_options & SO_TIMESTAMP)
199 				    ip_savecontrol(last, &opts, ip, n);
200 				if (ssb_appendaddr(&last->inp_socket->so_rcv,
201 					    (struct sockaddr *)&ripsrc, n,
202 					    opts) == 0) {
203 					/* should notify about lost packet */
204 					m_freem(n);
205 					if (opts)
206 					    m_freem(opts);
207 				} else {
208 					sorwakeup(last->inp_socket);
209 				}
210 				lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
211 				opts = NULL;
212 			}
213 		}
214 		last = inp;
215 	}
216 #ifdef IPSEC
217 	/* check AH/ESP integrity. */
218 	if (last && ipsec4_in_reject_so(m, last->inp_socket)) {
219 		m_freem(m);
220 		ipsecstat.in_polvio++;
221 		ipstat.ips_delivered--;
222 		/* do not inject data to pcb */
223 	} else
224 #endif /*IPSEC*/
225 #ifdef FAST_IPSEC
226 	/* check AH/ESP integrity. */
227 	if (last && ipsec4_in_reject(m, last)) {
228 		m_freem(m);
229 		ipstat.ips_delivered--;
230 		/* do not inject data to pcb */
231 	} else
232 #endif /*FAST_IPSEC*/
233 	/* Check the minimum TTL for socket. */
234 	if (last && ip->ip_ttl < last->inp_ip_minttl) {
235 		m_freem(opts);
236 		ipstat.ips_delivered--;
237 	} else if (last) {
238 		if (last->inp_flags & INP_CONTROLOPTS ||
239 		    last->inp_socket->so_options & SO_TIMESTAMP)
240 			ip_savecontrol(last, &opts, ip, m);
241 		lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
242 		if (ssb_appendaddr(&last->inp_socket->so_rcv,
243 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
244 			m_freem(m);
245 			if (opts)
246 			    m_freem(opts);
247 		} else {
248 			sorwakeup(last->inp_socket);
249 		}
250 		lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
251 	} else {
252 		m_freem(m);
253 		ipstat.ips_noproto++;
254 		ipstat.ips_delivered--;
255 	}
256 	return(IPPROTO_DONE);
257 }
258 
259 /*
260  * Generate IP header and pass packet to ip_output.
261  * Tack on options user may have setup with control call.
262  */
263 int
264 rip_output(struct mbuf *m, struct socket *so, ...)
265 {
266 	struct ip *ip;
267 	struct inpcb *inp = so->so_pcb;
268 	__va_list ap;
269 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
270 	u_long dst;
271 
272 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
273 	    ("not in netisr0"));
274 
275 	__va_start(ap, so);
276 	dst = __va_arg(ap, u_long);
277 	__va_end(ap);
278 
279 	/*
280 	 * If the user handed us a complete IP packet, use it.
281 	 * Otherwise, allocate an mbuf for a header and fill it in.
282 	 */
283 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
284 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
285 			m_freem(m);
286 			return(EMSGSIZE);
287 		}
288 		M_PREPEND(m, sizeof(struct ip), MB_WAIT);
289 		if (m == NULL)
290 			return(ENOBUFS);
291 		ip = mtod(m, struct ip *);
292 		ip->ip_tos = inp->inp_ip_tos;
293 		ip->ip_off = 0;
294 		ip->ip_p = inp->inp_ip_p;
295 		ip->ip_len = m->m_pkthdr.len;
296 		ip->ip_src = inp->inp_laddr;
297 		ip->ip_dst.s_addr = dst;
298 		ip->ip_ttl = inp->inp_ip_ttl;
299 	} else {
300 		int hlen;
301 
302 		if (m->m_pkthdr.len > IP_MAXPACKET) {
303 			m_freem(m);
304 			return(EMSGSIZE);
305 		}
306 		if (m->m_len < sizeof(struct ip)) {
307 			m = m_pullup(m, sizeof(struct ip));
308 			if (m == NULL)
309 				return ENOBUFS;
310 		}
311 		ip = mtod(m, struct ip *);
312 		hlen = IP_VHL_HL(ip->ip_vhl) << 2;
313 
314 		/* Don't allow header length less than the minimum. */
315 		if (hlen < sizeof(struct ip)) {
316 			m_freem(m);
317 			return EINVAL;
318 		}
319 
320 		/*
321 		 * Don't allow both user specified and setsockopt options.
322 		 * Don't allow packet length sizes that will crash.
323 		 */
324 		if ((hlen != sizeof(struct ip) && inp->inp_options) ||
325 		    ip->ip_len > m->m_pkthdr.len || ip->ip_len < hlen) {
326 			m_freem(m);
327 			return EINVAL;
328 		}
329 		if (ip->ip_id == 0)
330 			ip->ip_id = ip_newid();
331 
332 		/* Prevent ip_output from overwriting header fields */
333 		flags |= IP_RAWOUTPUT;
334 		ipstat.ips_rawout++;
335 	}
336 
337 	return ip_output(m, inp->inp_options, &inp->inp_route, flags,
338 			 inp->inp_moptions, inp);
339 }
340 
341 /*
342  * Raw IP socket option processing.
343  */
344 void
345 rip_ctloutput(netmsg_t msg)
346 {
347 	struct socket *so = msg->base.nm_so;
348 	struct sockopt *sopt = msg->ctloutput.nm_sopt;
349 	struct	inpcb *inp = so->so_pcb;
350 	int	error, optval;
351 
352 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
353 	    ("not in netisr0"));
354 
355 	error = 0;
356 
357 	/* Get socket's owner cpuid hint */
358 	if (sopt->sopt_level == SOL_SOCKET &&
359 	    sopt->sopt_dir == SOPT_GET &&
360 	    sopt->sopt_name == SO_CPUHINT) {
361 		optval = mycpuid;
362 		soopt_from_kbuf(sopt, &optval, sizeof(optval));
363 		goto done;
364 	}
365 
366 	if (sopt->sopt_level != IPPROTO_IP) {
367 		error = EINVAL;
368 		goto done;
369 	}
370 
371 	switch (sopt->sopt_dir) {
372 	case SOPT_GET:
373 		switch (sopt->sopt_name) {
374 		case IP_HDRINCL:
375 			optval = inp->inp_flags & INP_HDRINCL;
376 			soopt_from_kbuf(sopt, &optval, sizeof optval);
377 			break;
378 
379 		case IP_FW_ADD: /* ADD actually returns the body... */
380 		case IP_FW_GET:
381 			if (IPFW_LOADED)
382 				error = ip_fw_sockopt(sopt);
383 			else
384 				error = ENOPROTOOPT;
385 			break;
386 
387 		case IP_DUMMYNET_GET:
388 			error = ip_dn_sockopt(sopt);
389 			break ;
390 
391 		case MRT_INIT:
392 		case MRT_DONE:
393 		case MRT_ADD_VIF:
394 		case MRT_DEL_VIF:
395 		case MRT_ADD_MFC:
396 		case MRT_DEL_MFC:
397 		case MRT_VERSION:
398 		case MRT_ASSERT:
399 		case MRT_API_SUPPORT:
400 		case MRT_API_CONFIG:
401 		case MRT_ADD_BW_UPCALL:
402 		case MRT_DEL_BW_UPCALL:
403 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
404 				EOPNOTSUPP;
405 			break;
406 
407 		default:
408 			ip_ctloutput(msg);
409 			/* msg invalid now */
410 			return;
411 		}
412 		break;
413 
414 	case SOPT_SET:
415 		switch (sopt->sopt_name) {
416 		case IP_HDRINCL:
417 			error = soopt_to_kbuf(sopt, &optval, sizeof optval,
418 					      sizeof optval);
419 			if (error)
420 				break;
421 			if (optval)
422 				inp->inp_flags |= INP_HDRINCL;
423 			else
424 				inp->inp_flags &= ~INP_HDRINCL;
425 			break;
426 
427 		case IP_FW_ADD:
428 		case IP_FW_DEL:
429 		case IP_FW_FLUSH:
430 		case IP_FW_ZERO:
431 		case IP_FW_RESETLOG:
432 			if (IPFW_LOADED)
433 				error = ip_fw_ctl_ptr(sopt);
434 			else
435 				error = ENOPROTOOPT;
436 			break;
437 
438 		case IP_DUMMYNET_CONFIGURE:
439 		case IP_DUMMYNET_DEL:
440 		case IP_DUMMYNET_FLUSH:
441 			error = ip_dn_sockopt(sopt);
442 			break ;
443 
444 		case IP_RSVP_ON:
445 			error = ip_rsvp_init(so);
446 			break;
447 
448 		case IP_RSVP_OFF:
449 			error = ip_rsvp_done();
450 			break;
451 
452 		case IP_RSVP_VIF_ON:
453 		case IP_RSVP_VIF_OFF:
454 			error = ip_rsvp_vif ?
455 				ip_rsvp_vif(so, sopt) : EINVAL;
456 			break;
457 
458 		case MRT_INIT:
459 		case MRT_DONE:
460 		case MRT_ADD_VIF:
461 		case MRT_DEL_VIF:
462 		case MRT_ADD_MFC:
463 		case MRT_DEL_MFC:
464 		case MRT_VERSION:
465 		case MRT_ASSERT:
466 		case MRT_API_SUPPORT:
467 		case MRT_API_CONFIG:
468 		case MRT_ADD_BW_UPCALL:
469 		case MRT_DEL_BW_UPCALL:
470 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
471 					EOPNOTSUPP;
472 			break;
473 
474 		default:
475 			ip_ctloutput(msg);
476 			/* msg invalid now */
477 			return;
478 		}
479 		break;
480 	}
481 done:
482 	lwkt_replymsg(&msg->lmsg, error);
483 }
484 
485 /*
486  * This function exists solely to receive the PRC_IFDOWN messages which
487  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
488  * and calls in_ifadown() to remove all routes corresponding to that address.
489  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
490  * interface routes.
491  */
492 void
493 rip_ctlinput(netmsg_t msg)
494 {
495 	int cmd = msg->ctlinput.nm_cmd;
496 	struct sockaddr *sa = msg->ctlinput.nm_arg;
497 	struct in_ifaddr *ia;
498 	struct in_ifaddr_container *iac;
499 	struct ifnet *ifp;
500 	int err;
501 	int flags;
502 
503 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
504 	    ("not in netisr0"));
505 
506 	switch (cmd) {
507 	case PRC_IFDOWN:
508 		TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) {
509 			ia = iac->ia;
510 
511 			if (ia->ia_ifa.ifa_addr == sa &&
512 			    (ia->ia_flags & IFA_ROUTE)) {
513 				/*
514 				 * in_ifscrub kills the interface route.
515 				 */
516 				in_ifscrub(ia->ia_ifp, ia);
517 				/*
518 				 * in_ifadown gets rid of all the rest of
519 				 * the routes.  This is not quite the right
520 				 * thing to do, but at least if we are running
521 				 * a routing process they will come back.
522 				 */
523 				in_ifadown(&ia->ia_ifa, 0);
524 				break;
525 			}
526 		}
527 		break;
528 
529 	case PRC_IFUP:
530 		ia = NULL;
531 		TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) {
532 			if (iac->ia->ia_ifa.ifa_addr == sa) {
533 				ia = iac->ia;
534 				break;
535 			}
536 		}
537 		if (ia == NULL || (ia->ia_flags & IFA_ROUTE))
538 			goto done;
539 		flags = RTF_UP;
540 		ifp = ia->ia_ifa.ifa_ifp;
541 
542 #ifdef CARP
543 		/*
544 		 * Don't add prefix routes for CARP interfaces.
545 		 * Prefix routes creation is handled by CARP
546 		 * interfaces themselves.
547 		 */
548 		if (ifp->if_type == IFT_CARP)
549 			goto done;
550 #endif
551 
552 		if ((ifp->if_flags & IFF_LOOPBACK) ||
553 		    (ifp->if_flags & IFF_POINTOPOINT))
554 			flags |= RTF_HOST;
555 
556 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
557 		if (err == 0)
558 			ia->ia_flags |= IFA_ROUTE;
559 		break;
560 	}
561 done:
562 	lwkt_replymsg(&msg->lmsg, 0);
563 }
564 
565 u_long	rip_sendspace = RIPSNDQ;
566 u_long	rip_recvspace = RIPRCVQ;
567 
568 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
569     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
570 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
571     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
572 
573 static void
574 rip_attach(netmsg_t msg)
575 {
576 	struct socket *so = msg->base.nm_so;
577 	int proto = msg->attach.nm_proto;
578 	struct pru_attach_info *ai = msg->attach.nm_ai;
579 	struct inpcb *inp;
580 	int error;
581 
582 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
583 	    ("not in netisr0"));
584 
585 	inp = so->so_pcb;
586 	if (inp)
587 		panic("rip_attach");
588 	error = priv_check_cred(ai->p_ucred, PRIV_NETINET_RAW, NULL_CRED_OKAY);
589 	if (error)
590 		goto done;
591 
592 	error = soreserve(so, rip_sendspace, rip_recvspace, ai->sb_rlimit);
593 	if (error)
594 		goto done;
595 
596 	error = in_pcballoc(so, &ripcbinfo);
597 	if (error == 0) {
598 		inp = (struct inpcb *)so->so_pcb;
599 		inp->inp_vflag |= INP_IPV4;
600 		inp->inp_ip_p = proto;
601 		inp->inp_ip_ttl = ip_defttl;
602 	}
603 	error = 0;
604 done:
605 	lwkt_replymsg(&msg->lmsg, error);
606 }
607 
608 static void
609 rip_detach(netmsg_t msg)
610 {
611 	struct socket *so = msg->base.nm_so;
612 	struct inpcb *inp;
613 
614 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
615 	    ("not in netisr0"));
616 
617 	inp = so->so_pcb;
618 	if (inp == NULL)
619 		panic("rip_detach");
620 	if (so == ip_mrouter && ip_mrouter_done)
621 		ip_mrouter_done();
622 	if (ip_rsvp_force_done)
623 		ip_rsvp_force_done(so);
624 	if (so == ip_rsvpd)
625 		ip_rsvp_done();
626 	in_pcbdetach(inp);
627 	lwkt_replymsg(&msg->lmsg, 0);
628 }
629 
630 static void
631 rip_abort(netmsg_t msg)
632 {
633 	/*
634 	 * Raw socket does not support listen(2),
635 	 * so this should never be called.
636 	 */
637 	panic("rip_abort is called");
638 }
639 
640 static void
641 rip_disconnect(netmsg_t msg)
642 {
643 	struct socket *so = msg->base.nm_so;
644 	int error;
645 
646 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
647 	    ("not in netisr0"));
648 
649 	if (so->so_state & SS_ISCONNECTED) {
650 		soisdisconnected(so);
651 		error = 0;
652 	} else {
653 		error = ENOTCONN;
654 	}
655 	lwkt_replymsg(&msg->lmsg, error);
656 }
657 
658 static void
659 rip_bind(netmsg_t msg)
660 {
661 	struct socket *so = msg->base.nm_so;
662 	struct sockaddr *nam = msg->bind.nm_nam;
663 	struct inpcb *inp = so->so_pcb;
664 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
665 	int error;
666 
667 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
668 	    ("not in netisr0"));
669 
670 	if (nam->sa_len == sizeof(*addr)) {
671 		if (TAILQ_EMPTY(&ifnet) ||
672 		    ((addr->sin_family != AF_INET) &&
673 		     (addr->sin_family != AF_IMPLINK)) ||
674 		    (addr->sin_addr.s_addr != INADDR_ANY &&
675 		     ifa_ifwithaddr((struct sockaddr *)addr) == 0)) {
676 			error = EADDRNOTAVAIL;
677 		} else {
678 			inp->inp_laddr = addr->sin_addr;
679 			error = 0;
680 		}
681 	} else {
682 		error = EINVAL;
683 	}
684 	lwkt_replymsg(&msg->lmsg, error);
685 }
686 
687 static void
688 rip_connect(netmsg_t msg)
689 {
690 	struct socket *so = msg->base.nm_so;
691 	struct sockaddr *nam = msg->connect.nm_nam;
692 	struct inpcb *inp = so->so_pcb;
693 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
694 	int error;
695 
696 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
697 	    ("not in netisr0"));
698 
699 	if (nam->sa_len != sizeof(*addr)) {
700 		error = EINVAL;
701 	} else if (TAILQ_EMPTY(&ifnet)) {
702 		error = EADDRNOTAVAIL;
703 	} else {
704 		if ((addr->sin_family != AF_INET) &&
705 		    (addr->sin_family != AF_IMPLINK)) {
706 			error = EAFNOSUPPORT;
707 		} else {
708 			inp->inp_faddr = addr->sin_addr;
709 			soisconnected(so);
710 			error = 0;
711 		}
712 	}
713 	lwkt_replymsg(&msg->lmsg, error);
714 }
715 
716 static void
717 rip_shutdown(netmsg_t msg)
718 {
719 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
720 	    ("not in netisr0"));
721 
722 	socantsendmore(msg->base.nm_so);
723 	lwkt_replymsg(&msg->lmsg, 0);
724 }
725 
726 static void
727 rip_send(netmsg_t msg)
728 {
729 	struct socket *so = msg->base.nm_so;
730 	struct mbuf *m = msg->send.nm_m;
731 	/*struct mbuf *control = msg->send.nm_control;*/
732 	struct sockaddr *nam = msg->send.nm_addr;
733 	/*int flags = msg->send.nm_flags;*/
734 	struct inpcb *inp = so->so_pcb;
735 	u_long dst;
736 	int error;
737 
738 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
739 	    ("not in netisr0"));
740 
741 	if (so->so_state & SS_ISCONNECTED) {
742 		if (nam) {
743 			m_freem(m);
744 			error = EISCONN;
745 		} else {
746 			dst = inp->inp_faddr.s_addr;
747 			error = rip_output(m, so, dst);
748 		}
749 	} else {
750 		if (nam == NULL) {
751 			m_freem(m);
752 			error = ENOTCONN;
753 		} else {
754 			dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
755 			error = rip_output(m, so, dst);
756 		}
757 	}
758 	lwkt_replymsg(&msg->lmsg, error);
759 }
760 
761 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, &ripcbinfo, 1,
762 	    in_pcblist_global, "S,xinpcb", "List of active raw IP sockets");
763 
764 struct pr_usrreqs rip_usrreqs = {
765 	.pru_abort = rip_abort,
766 	.pru_accept = pr_generic_notsupp,
767 	.pru_attach = rip_attach,
768 	.pru_bind = rip_bind,
769 	.pru_connect = rip_connect,
770 	.pru_connect2 = pr_generic_notsupp,
771 	.pru_control = in_control_dispatch,
772 	.pru_detach = rip_detach,
773 	.pru_disconnect = rip_disconnect,
774 	.pru_listen = pr_generic_notsupp,
775 	.pru_peeraddr = in_setpeeraddr_dispatch,
776 	.pru_rcvd = pr_generic_notsupp,
777 	.pru_rcvoob = pr_generic_notsupp,
778 	.pru_send = rip_send,
779 	.pru_sense = pru_sense_null,
780 	.pru_shutdown = rip_shutdown,
781 	.pru_sockaddr = in_setsockaddr_dispatch,
782 	.pru_sosend = sosend,
783 	.pru_soreceive = soreceive
784 };
785