xref: /freebsd/sys/netinet/raw_ip.c (revision 5129159789cc9d7bc514e4546b88e3427695002d)
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  * $FreeBSD$
35  */
36 
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/kernel.h>
40 #include <sys/malloc.h>
41 #include <sys/mbuf.h>
42 #include <sys/proc.h>
43 #include <sys/protosw.h>
44 #include <sys/socket.h>
45 #include <sys/socketvar.h>
46 #include <sys/sysctl.h>
47 
48 #include <vm/vm_zone.h>
49 
50 #include <net/if.h>
51 #include <net/route.h>
52 
53 #define _IP_VHL
54 #include <netinet/in.h>
55 #include <netinet/in_systm.h>
56 #include <netinet/ip.h>
57 #include <netinet/in_pcb.h>
58 #include <netinet/in_var.h>
59 #include <netinet/ip_var.h>
60 #include <netinet/ip_mroute.h>
61 
62 #include <netinet/ip_fw.h>
63 
64 #include "opt_ipdn.h"
65 #ifdef DUMMYNET
66 #include <netinet/ip_dummynet.h>
67 #endif
68 
69 struct	inpcbhead ripcb;
70 struct	inpcbinfo ripcbinfo;
71 
72 /*
73  * Nominal space allocated to a raw ip socket.
74  */
75 #define	RIPSNDQ		8192
76 #define	RIPRCVQ		8192
77 
78 /*
79  * Raw interface to IP protocol.
80  */
81 
82 /*
83  * Initialize raw connection block q.
84  */
85 void
86 rip_init()
87 {
88 	LIST_INIT(&ripcb);
89 	ripcbinfo.listhead = &ripcb;
90 	/*
91 	 * XXX We don't use the hash list for raw IP, but it's easier
92 	 * to allocate a one entry hash list than it is to check all
93 	 * over the place for hashbase == NULL.
94 	 */
95 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
96 	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
97 	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
98 				   maxsockets, ZONE_INTERRUPT, 0);
99 }
100 
101 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
102 /*
103  * Setup generic address and protocol structures
104  * for raw_input routine, then pass them along with
105  * mbuf chain.
106  */
107 void
108 rip_input(m, iphlen)
109 	struct mbuf *m;
110 	int iphlen;
111 {
112 	register struct ip *ip = mtod(m, struct ip *);
113 	register struct inpcb *inp;
114 	struct inpcb *last = 0;
115 	struct mbuf *opts = 0;
116 
117 	ripsrc.sin_addr = ip->ip_src;
118 	for (inp = ripcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
119 		if (inp->inp_ip_p && inp->inp_ip_p != ip->ip_p)
120 			continue;
121 		if (inp->inp_laddr.s_addr &&
122                   inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
123 			continue;
124 		if (inp->inp_faddr.s_addr &&
125                   inp->inp_faddr.s_addr != ip->ip_src.s_addr)
126 			continue;
127 		if (last) {
128 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
129 			if (n) {
130 				if (last->inp_flags & INP_CONTROLOPTS ||
131 				    last->inp_socket->so_options & SO_TIMESTAMP)
132 				    ip_savecontrol(last, &opts, ip, n);
133 				if (sbappendaddr(&last->inp_socket->so_rcv,
134 				    (struct sockaddr *)&ripsrc, n,
135 				    opts) == 0) {
136 					/* should notify about lost packet */
137 					m_freem(n);
138 					if (opts)
139 					    m_freem(opts);
140 				} else
141 					sorwakeup(last->inp_socket);
142 				opts = 0;
143 			}
144 		}
145 		last = inp;
146 	}
147 	if (last) {
148 		if (last->inp_flags & INP_CONTROLOPTS ||
149 		    last->inp_socket->so_options & SO_TIMESTAMP)
150 			ip_savecontrol(last, &opts, ip, m);
151 		if (sbappendaddr(&last->inp_socket->so_rcv,
152 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
153 			m_freem(m);
154 			if (opts)
155 			    m_freem(opts);
156 		} else
157 			sorwakeup(last->inp_socket);
158 	} else {
159 		m_freem(m);
160               ipstat.ips_noproto++;
161               ipstat.ips_delivered--;
162       }
163 }
164 
165 /*
166  * Generate IP header and pass packet to ip_output.
167  * Tack on options user may have setup with control call.
168  */
169 int
170 rip_output(m, so, dst)
171 	struct mbuf *m;
172 	struct socket *so;
173 	u_long dst;
174 {
175 	register struct ip *ip;
176 	register struct inpcb *inp = sotoinpcb(so);
177 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
178 
179 	/*
180 	 * If the user handed us a complete IP packet, use it.
181 	 * Otherwise, allocate an mbuf for a header and fill it in.
182 	 */
183 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
184 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
185 			m_freem(m);
186 			return(EMSGSIZE);
187 		}
188 		M_PREPEND(m, sizeof(struct ip), M_WAIT);
189 		ip = mtod(m, struct ip *);
190 		ip->ip_tos = 0;
191 		ip->ip_off = 0;
192 		ip->ip_p = inp->inp_ip_p;
193 		ip->ip_len = m->m_pkthdr.len;
194 		ip->ip_src = inp->inp_laddr;
195 		ip->ip_dst.s_addr = dst;
196 		ip->ip_ttl = MAXTTL;
197 	} else {
198 		if (m->m_pkthdr.len > IP_MAXPACKET) {
199 			m_freem(m);
200 			return(EMSGSIZE);
201 		}
202 		ip = mtod(m, struct ip *);
203 		/* don't allow both user specified and setsockopt options,
204 		   and don't allow packet length sizes that will crash */
205 		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
206 		     && inp->inp_options)
207 		    || (ip->ip_len > m->m_pkthdr.len)
208 		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
209 			m_freem(m);
210 			return EINVAL;
211 		}
212 		if (ip->ip_id == 0)
213 			ip->ip_id = htons(ip_id++);
214 		/* XXX prevent ip_output from overwriting header fields */
215 		flags |= IP_RAWOUTPUT;
216 		ipstat.ips_rawout++;
217 	}
218 	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
219 			  inp->inp_moptions));
220 }
221 
222 /*
223  * Raw IP socket option processing.
224  */
225 int
226 rip_ctloutput(so, sopt)
227 	struct socket *so;
228 	struct sockopt *sopt;
229 {
230 	struct	inpcb *inp = sotoinpcb(so);
231 	int	error, optval;
232 
233 	if (sopt->sopt_level != IPPROTO_IP)
234 		return (EINVAL);
235 
236 	error = 0;
237 
238 	switch (sopt->sopt_dir) {
239 	case SOPT_GET:
240 		switch (sopt->sopt_name) {
241 		case IP_HDRINCL:
242 			optval = inp->inp_flags & INP_HDRINCL;
243 			error = sooptcopyout(sopt, &optval, sizeof optval);
244 			break;
245 
246 		case IP_FW_GET:
247 			if (ip_fw_ctl_ptr == 0)
248 				error = ENOPROTOOPT;
249 			else
250 				error = ip_fw_ctl_ptr(sopt);
251 			break;
252 
253 #ifdef DUMMYNET
254 		case IP_DUMMYNET_GET:
255 			if (ip_dn_ctl_ptr == NULL)
256 				error = ENOPROTOOPT ;
257 			else
258 				error = ip_dn_ctl_ptr(sopt);
259 			break ;
260 #endif /* DUMMYNET */
261 
262 		case MRT_INIT:
263 		case MRT_DONE:
264 		case MRT_ADD_VIF:
265 		case MRT_DEL_VIF:
266 		case MRT_ADD_MFC:
267 		case MRT_DEL_MFC:
268 		case MRT_VERSION:
269 		case MRT_ASSERT:
270 			error = ip_mrouter_get(so, sopt);
271 			break;
272 
273 		default:
274 			error = ip_ctloutput(so, sopt);
275 			break;
276 		}
277 		break;
278 
279 	case SOPT_SET:
280 		switch (sopt->sopt_name) {
281 		case IP_HDRINCL:
282 			error = sooptcopyin(sopt, &optval, sizeof optval,
283 					    sizeof optval);
284 			if (error)
285 				break;
286 			if (optval)
287 				inp->inp_flags |= INP_HDRINCL;
288 			else
289 				inp->inp_flags &= ~INP_HDRINCL;
290 			break;
291 
292 		case IP_FW_ADD:
293 		case IP_FW_DEL:
294 		case IP_FW_FLUSH:
295 		case IP_FW_ZERO:
296 		case IP_FW_RESETLOG:
297 			if (ip_fw_ctl_ptr == 0)
298 				error = ENOPROTOOPT;
299 			else
300 				error = ip_fw_ctl_ptr(sopt);
301 			break;
302 
303 #ifdef DUMMYNET
304 		case IP_DUMMYNET_CONFIGURE:
305 		case IP_DUMMYNET_DEL:
306 		case IP_DUMMYNET_FLUSH:
307 			if (ip_dn_ctl_ptr == NULL)
308 				error = ENOPROTOOPT ;
309 			else
310 				error = ip_dn_ctl_ptr(sopt);
311 			break ;
312 #endif
313 
314 		case IP_RSVP_ON:
315 			error = ip_rsvp_init(so);
316 			break;
317 
318 		case IP_RSVP_OFF:
319 			error = ip_rsvp_done();
320 			break;
321 
322 			/* XXX - should be combined */
323 		case IP_RSVP_VIF_ON:
324 			error = ip_rsvp_vif_init(so, sopt);
325 			break;
326 
327 		case IP_RSVP_VIF_OFF:
328 			error = ip_rsvp_vif_done(so, sopt);
329 			break;
330 
331 		case MRT_INIT:
332 		case MRT_DONE:
333 		case MRT_ADD_VIF:
334 		case MRT_DEL_VIF:
335 		case MRT_ADD_MFC:
336 		case MRT_DEL_MFC:
337 		case MRT_VERSION:
338 		case MRT_ASSERT:
339 			error = ip_mrouter_set(so, sopt);
340 			break;
341 
342 		default:
343 			error = ip_ctloutput(so, sopt);
344 			break;
345 		}
346 		break;
347 	}
348 
349 	return (error);
350 }
351 
352 /*
353  * This function exists solely to receive the PRC_IFDOWN messages which
354  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
355  * and calls in_ifadown() to remove all routes corresponding to that address.
356  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
357  * interface routes.
358  */
359 void
360 rip_ctlinput(cmd, sa, vip)
361 	int cmd;
362 	struct sockaddr *sa;
363 	void *vip;
364 {
365 	struct in_ifaddr *ia;
366 	struct ifnet *ifp;
367 	int err;
368 	int flags;
369 
370 	switch (cmd) {
371 	case PRC_IFDOWN:
372 		for (ia = in_ifaddrhead.tqh_first; ia;
373 		     ia = ia->ia_link.tqe_next) {
374 			if (ia->ia_ifa.ifa_addr == sa
375 			    && (ia->ia_flags & IFA_ROUTE)) {
376 				/*
377 				 * in_ifscrub kills the interface route.
378 				 */
379 				in_ifscrub(ia->ia_ifp, ia);
380 				/*
381 				 * in_ifadown gets rid of all the rest of
382 				 * the routes.  This is not quite the right
383 				 * thing to do, but at least if we are running
384 				 * a routing process they will come back.
385 				 */
386 				in_ifadown(&ia->ia_ifa);
387 				break;
388 			}
389 		}
390 		break;
391 
392 	case PRC_IFUP:
393 		for (ia = in_ifaddrhead.tqh_first; ia;
394 		     ia = ia->ia_link.tqe_next) {
395 			if (ia->ia_ifa.ifa_addr == sa)
396 				break;
397 		}
398 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
399 			return;
400 		flags = RTF_UP;
401 		ifp = ia->ia_ifa.ifa_ifp;
402 
403 		if ((ifp->if_flags & IFF_LOOPBACK)
404 		    || (ifp->if_flags & IFF_POINTOPOINT))
405 			flags |= RTF_HOST;
406 
407 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
408 		if (err == 0)
409 			ia->ia_flags |= IFA_ROUTE;
410 		break;
411 	}
412 }
413 
414 u_long	rip_sendspace = RIPSNDQ;
415 u_long	rip_recvspace = RIPRCVQ;
416 
417 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
418     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
419 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
420     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
421 
422 static int
423 rip_attach(struct socket *so, int proto, struct proc *p)
424 {
425 	struct inpcb *inp;
426 	int error, s;
427 
428 	inp = sotoinpcb(so);
429 	if (inp)
430 		panic("rip_attach");
431 	if (p && (error = suser(p)) != 0)
432 		return error;
433 
434 	s = splnet();
435 	error = in_pcballoc(so, &ripcbinfo, p);
436 	splx(s);
437 	if (error)
438 		return error;
439 	error = soreserve(so, rip_sendspace, rip_recvspace);
440 	if (error)
441 		return error;
442 	inp = (struct inpcb *)so->so_pcb;
443 	inp->inp_ip_p = proto;
444 	return 0;
445 }
446 
447 static int
448 rip_detach(struct socket *so)
449 {
450 	struct inpcb *inp;
451 
452 	inp = sotoinpcb(so);
453 	if (inp == 0)
454 		panic("rip_detach");
455 	if (so == ip_mrouter)
456 		ip_mrouter_done();
457 	ip_rsvp_force_done(so);
458 	if (so == ip_rsvpd)
459 		ip_rsvp_done();
460 	in_pcbdetach(inp);
461 	return 0;
462 }
463 
464 static int
465 rip_abort(struct socket *so)
466 {
467 	soisdisconnected(so);
468 	return rip_detach(so);
469 }
470 
471 static int
472 rip_disconnect(struct socket *so)
473 {
474 	if ((so->so_state & SS_ISCONNECTED) == 0)
475 		return ENOTCONN;
476 	return rip_abort(so);
477 }
478 
479 static int
480 rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
481 {
482 	struct inpcb *inp = sotoinpcb(so);
483 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
484 
485 	if (nam->sa_len != sizeof(*addr))
486 		return EINVAL;
487 
488 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
489 				    (addr->sin_family != AF_IMPLINK)) ||
490 	    (addr->sin_addr.s_addr &&
491 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
492 		return EADDRNOTAVAIL;
493 	inp->inp_laddr = addr->sin_addr;
494 	return 0;
495 }
496 
497 static int
498 rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
499 {
500 	struct inpcb *inp = sotoinpcb(so);
501 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
502 
503 	if (nam->sa_len != sizeof(*addr))
504 		return EINVAL;
505 	if (TAILQ_EMPTY(&ifnet))
506 		return EADDRNOTAVAIL;
507 	if ((addr->sin_family != AF_INET) &&
508 	    (addr->sin_family != AF_IMPLINK))
509 		return EAFNOSUPPORT;
510 	inp->inp_faddr = addr->sin_addr;
511 	soisconnected(so);
512 	return 0;
513 }
514 
515 static int
516 rip_shutdown(struct socket *so)
517 {
518 	socantsendmore(so);
519 	return 0;
520 }
521 
522 static int
523 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
524 	 struct mbuf *control, struct proc *p)
525 {
526 	struct inpcb *inp = sotoinpcb(so);
527 	register u_long dst;
528 
529 	if (so->so_state & SS_ISCONNECTED) {
530 		if (nam) {
531 			m_freem(m);
532 			return EISCONN;
533 		}
534 		dst = inp->inp_faddr.s_addr;
535 	} else {
536 		if (nam == NULL) {
537 			m_freem(m);
538 			return ENOTCONN;
539 		}
540 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
541 	}
542 	return rip_output(m, so, dst);
543 }
544 
545 static int
546 rip_pcblist SYSCTL_HANDLER_ARGS
547 {
548 	int error, i, n, s;
549 	struct inpcb *inp, **inp_list;
550 	inp_gen_t gencnt;
551 	struct xinpgen xig;
552 
553 	/*
554 	 * The process of preparing the TCB list is too time-consuming and
555 	 * resource-intensive to repeat twice on every request.
556 	 */
557 	if (req->oldptr == 0) {
558 		n = ripcbinfo.ipi_count;
559 		req->oldidx = 2 * (sizeof xig)
560 			+ (n + n/8) * sizeof(struct xinpcb);
561 		return 0;
562 	}
563 
564 	if (req->newptr != 0)
565 		return EPERM;
566 
567 	/*
568 	 * OK, now we're committed to doing something.
569 	 */
570 	s = splnet();
571 	gencnt = ripcbinfo.ipi_gencnt;
572 	n = ripcbinfo.ipi_count;
573 	splx(s);
574 
575 	xig.xig_len = sizeof xig;
576 	xig.xig_count = n;
577 	xig.xig_gen = gencnt;
578 	xig.xig_sogen = so_gencnt;
579 	error = SYSCTL_OUT(req, &xig, sizeof xig);
580 	if (error)
581 		return error;
582 
583 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
584 	if (inp_list == 0)
585 		return ENOMEM;
586 
587 	s = splnet();
588 	for (inp = ripcbinfo.listhead->lh_first, i = 0; inp && i < n;
589 	     inp = inp->inp_list.le_next) {
590 		if (inp->inp_gencnt <= gencnt)
591 			inp_list[i++] = inp;
592 	}
593 	splx(s);
594 	n = i;
595 
596 	error = 0;
597 	for (i = 0; i < n; i++) {
598 		inp = inp_list[i];
599 		if (inp->inp_gencnt <= gencnt) {
600 			struct xinpcb xi;
601 			xi.xi_len = sizeof xi;
602 			/* XXX should avoid extra copy */
603 			bcopy(inp, &xi.xi_inp, sizeof *inp);
604 			if (inp->inp_socket)
605 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
606 			error = SYSCTL_OUT(req, &xi, sizeof xi);
607 		}
608 	}
609 	if (!error) {
610 		/*
611 		 * Give the user an updated idea of our state.
612 		 * If the generation differs from what we told
613 		 * her before, she knows that something happened
614 		 * while we were processing this request, and it
615 		 * might be necessary to retry.
616 		 */
617 		s = splnet();
618 		xig.xig_gen = ripcbinfo.ipi_gencnt;
619 		xig.xig_sogen = so_gencnt;
620 		xig.xig_count = ripcbinfo.ipi_count;
621 		splx(s);
622 		error = SYSCTL_OUT(req, &xig, sizeof xig);
623 	}
624 	free(inp_list, M_TEMP);
625 	return error;
626 }
627 
628 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
629 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
630 
631 struct pr_usrreqs rip_usrreqs = {
632 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
633 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
634 	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
635 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
636 	in_setsockaddr, sosend, soreceive, sopoll
637 };
638