raw_ip.c revision 80406 
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: head/sys/netinet/raw_ip.c 80406 2001-07-26 19:19:49Z ume $
35 */
36
37#include "opt_inet6.h"
38#include "opt_ipsec.h"
39#include "opt_random_ip_id.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/kernel.h>
44#include <sys/malloc.h>
45#include <sys/mbuf.h>
46#include <sys/protosw.h>
47#include <sys/socket.h>
48#include <sys/socketvar.h>
49#include <sys/sysctl.h>
50
51#include <vm/vm_zone.h>
52
53#include <net/if.h>
54#include <net/route.h>
55
56#define _IP_VHL
57#include <netinet/in.h>
58#include <netinet/in_systm.h>
59#include <netinet/ip.h>
60#include <netinet/in_pcb.h>
61#include <netinet/in_var.h>
62#include <netinet/ip_var.h>
63#include <netinet/ip_mroute.h>
64
65#include <netinet/ip_fw.h>
66
67#ifdef IPSEC
68#include <netinet6/ipsec.h>
69#endif /*IPSEC*/
70
71#include "opt_ipdn.h"
72#ifdef DUMMYNET
73#include <netinet/ip_dummynet.h>
74#endif
75
76struct	inpcbhead ripcb;
77struct	inpcbinfo ripcbinfo;
78
79/*
80 * Nominal space allocated to a raw ip socket.
81 */
82#define	RIPSNDQ		8192
83#define	RIPRCVQ		8192
84
85/*
86 * Raw interface to IP protocol.
87 */
88
89/*
90 * Initialize raw connection block q.
91 */
92void
93rip_init()
94{
95	LIST_INIT(&ripcb);
96	ripcbinfo.listhead = &ripcb;
97	/*
98	 * XXX We don't use the hash list for raw IP, but it's easier
99	 * to allocate a one entry hash list than it is to check all
100	 * over the place for hashbase == NULL.
101	 */
102	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
103	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
104	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
105				   maxsockets, ZONE_INTERRUPT, 0);
106}
107
108static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
109/*
110 * Setup generic address and protocol structures
111 * for raw_input routine, then pass them along with
112 * mbuf chain.
113 */
114void
115rip_input(m, off, proto)
116	struct mbuf *m;
117	int off, proto;
118{
119	register struct ip *ip = mtod(m, struct ip *);
120	register struct inpcb *inp;
121	struct inpcb *last = 0;
122	struct mbuf *opts = 0;
123
124	ripsrc.sin_addr = ip->ip_src;
125	LIST_FOREACH(inp, &ripcb, inp_list) {
126#ifdef INET6
127		if ((inp->inp_vflag & INP_IPV4) == 0)
128			continue;
129#endif
130		if (inp->inp_ip_p && inp->inp_ip_p != proto)
131			continue;
132		if (inp->inp_laddr.s_addr &&
133                  inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
134			continue;
135		if (inp->inp_faddr.s_addr &&
136                  inp->inp_faddr.s_addr != ip->ip_src.s_addr)
137			continue;
138		if (last) {
139			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
140
141#ifdef IPSEC
142			/* check AH/ESP integrity. */
143			if (n && ipsec4_in_reject_so(n, last->inp_socket)) {
144				m_freem(n);
145				ipsecstat.in_polvio++;
146				/* do not inject data to pcb */
147			} else
148#endif /*IPSEC*/
149			if (n) {
150				if (last->inp_flags & INP_CONTROLOPTS ||
151				    last->inp_socket->so_options & SO_TIMESTAMP)
152				    ip_savecontrol(last, &opts, ip, n);
153				if (sbappendaddr(&last->inp_socket->so_rcv,
154				    (struct sockaddr *)&ripsrc, n,
155				    opts) == 0) {
156					/* should notify about lost packet */
157					m_freem(n);
158					if (opts)
159					    m_freem(opts);
160				} else
161					sorwakeup(last->inp_socket);
162				opts = 0;
163			}
164		}
165		last = inp;
166	}
167#ifdef IPSEC
168	/* check AH/ESP integrity. */
169	if (last && ipsec4_in_reject_so(m, last->inp_socket)) {
170		m_freem(m);
171		ipsecstat.in_polvio++;
172		ipstat.ips_delivered--;
173		/* do not inject data to pcb */
174	} else
175#endif /*IPSEC*/
176	if (last) {
177		if (last->inp_flags & INP_CONTROLOPTS ||
178		    last->inp_socket->so_options & SO_TIMESTAMP)
179			ip_savecontrol(last, &opts, ip, m);
180		if (sbappendaddr(&last->inp_socket->so_rcv,
181		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
182			m_freem(m);
183			if (opts)
184			    m_freem(opts);
185		} else
186			sorwakeup(last->inp_socket);
187	} else {
188		m_freem(m);
189		ipstat.ips_noproto++;
190		ipstat.ips_delivered--;
191	}
192}
193
194/*
195 * Generate IP header and pass packet to ip_output.
196 * Tack on options user may have setup with control call.
197 */
198int
199rip_output(m, so, dst)
200	struct mbuf *m;
201	struct socket *so;
202	u_long dst;
203{
204	register struct ip *ip;
205	register struct inpcb *inp = sotoinpcb(so);
206	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
207
208	/*
209	 * If the user handed us a complete IP packet, use it.
210	 * Otherwise, allocate an mbuf for a header and fill it in.
211	 */
212	if ((inp->inp_flags & INP_HDRINCL) == 0) {
213		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
214			m_freem(m);
215			return(EMSGSIZE);
216		}
217		M_PREPEND(m, sizeof(struct ip), M_TRYWAIT);
218		ip = mtod(m, struct ip *);
219		ip->ip_tos = inp->inp_ip_tos;
220		ip->ip_off = 0;
221		ip->ip_p = inp->inp_ip_p;
222		ip->ip_len = m->m_pkthdr.len;
223		ip->ip_src = inp->inp_laddr;
224		ip->ip_dst.s_addr = dst;
225		ip->ip_ttl = inp->inp_ip_ttl;
226	} else {
227		if (m->m_pkthdr.len > IP_MAXPACKET) {
228			m_freem(m);
229			return(EMSGSIZE);
230		}
231		ip = mtod(m, struct ip *);
232		/* don't allow both user specified and setsockopt options,
233		   and don't allow packet length sizes that will crash */
234		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
235		     && inp->inp_options)
236		    || (ip->ip_len > m->m_pkthdr.len)
237		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
238			m_freem(m);
239			return EINVAL;
240		}
241		if (ip->ip_id == 0)
242#ifdef RANDOM_IP_ID
243			ip->ip_id = ip_randomid();
244#else
245			ip->ip_id = htons(ip_id++);
246#endif
247		/* XXX prevent ip_output from overwriting header fields */
248		flags |= IP_RAWOUTPUT;
249		ipstat.ips_rawout++;
250	}
251
252#ifdef IPSEC
253	if (ipsec_setsocket(m, so) != 0) {
254		m_freem(m);
255		return ENOBUFS;
256	}
257#endif /*IPSEC*/
258
259	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
260			  inp->inp_moptions));
261}
262
263/*
264 * Raw IP socket option processing.
265 */
266int
267rip_ctloutput(so, sopt)
268	struct socket *so;
269	struct sockopt *sopt;
270{
271	struct	inpcb *inp = sotoinpcb(so);
272	int	error, optval;
273
274	if (sopt->sopt_level != IPPROTO_IP)
275		return (EINVAL);
276
277	error = 0;
278
279	switch (sopt->sopt_dir) {
280	case SOPT_GET:
281		switch (sopt->sopt_name) {
282		case IP_HDRINCL:
283			optval = inp->inp_flags & INP_HDRINCL;
284			error = sooptcopyout(sopt, &optval, sizeof optval);
285			break;
286
287		case IP_FW_ADD:
288		case IP_FW_GET:
289			if (ip_fw_ctl_ptr == 0)
290				error = ENOPROTOOPT;
291			else
292				error = ip_fw_ctl_ptr(sopt);
293			break;
294
295#ifdef DUMMYNET
296		case IP_DUMMYNET_GET:
297			if (ip_dn_ctl_ptr == NULL)
298				error = ENOPROTOOPT ;
299			else
300				error = ip_dn_ctl_ptr(sopt);
301			break ;
302#endif /* DUMMYNET */
303
304		case MRT_INIT:
305		case MRT_DONE:
306		case MRT_ADD_VIF:
307		case MRT_DEL_VIF:
308		case MRT_ADD_MFC:
309		case MRT_DEL_MFC:
310		case MRT_VERSION:
311		case MRT_ASSERT:
312			error = ip_mrouter_get(so, sopt);
313			break;
314
315		default:
316			error = ip_ctloutput(so, sopt);
317			break;
318		}
319		break;
320
321	case SOPT_SET:
322		switch (sopt->sopt_name) {
323		case IP_HDRINCL:
324			error = sooptcopyin(sopt, &optval, sizeof optval,
325					    sizeof optval);
326			if (error)
327				break;
328			if (optval)
329				inp->inp_flags |= INP_HDRINCL;
330			else
331				inp->inp_flags &= ~INP_HDRINCL;
332			break;
333
334		case IP_FW_ADD:
335		case IP_FW_DEL:
336		case IP_FW_FLUSH:
337		case IP_FW_ZERO:
338		case IP_FW_RESETLOG:
339			if (ip_fw_ctl_ptr == 0)
340				error = ENOPROTOOPT;
341			else
342				error = ip_fw_ctl_ptr(sopt);
343			break;
344
345#ifdef DUMMYNET
346		case IP_DUMMYNET_CONFIGURE:
347		case IP_DUMMYNET_DEL:
348		case IP_DUMMYNET_FLUSH:
349			if (ip_dn_ctl_ptr == NULL)
350				error = ENOPROTOOPT ;
351			else
352				error = ip_dn_ctl_ptr(sopt);
353			break ;
354#endif
355
356		case IP_RSVP_ON:
357			error = ip_rsvp_init(so);
358			break;
359
360		case IP_RSVP_OFF:
361			error = ip_rsvp_done();
362			break;
363
364			/* XXX - should be combined */
365		case IP_RSVP_VIF_ON:
366			error = ip_rsvp_vif_init(so, sopt);
367			break;
368
369		case IP_RSVP_VIF_OFF:
370			error = ip_rsvp_vif_done(so, sopt);
371			break;
372
373		case MRT_INIT:
374		case MRT_DONE:
375		case MRT_ADD_VIF:
376		case MRT_DEL_VIF:
377		case MRT_ADD_MFC:
378		case MRT_DEL_MFC:
379		case MRT_VERSION:
380		case MRT_ASSERT:
381			error = ip_mrouter_set(so, sopt);
382			break;
383
384		default:
385			error = ip_ctloutput(so, sopt);
386			break;
387		}
388		break;
389	}
390
391	return (error);
392}
393
394/*
395 * This function exists solely to receive the PRC_IFDOWN messages which
396 * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
397 * and calls in_ifadown() to remove all routes corresponding to that address.
398 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
399 * interface routes.
400 */
401void
402rip_ctlinput(cmd, sa, vip)
403	int cmd;
404	struct sockaddr *sa;
405	void *vip;
406{
407	struct in_ifaddr *ia;
408	struct ifnet *ifp;
409	int err;
410	int flags;
411
412	switch (cmd) {
413	case PRC_IFDOWN:
414		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
415			if (ia->ia_ifa.ifa_addr == sa
416			    && (ia->ia_flags & IFA_ROUTE)) {
417				/*
418				 * in_ifscrub kills the interface route.
419				 */
420				in_ifscrub(ia->ia_ifp, ia);
421				/*
422				 * in_ifadown gets rid of all the rest of
423				 * the routes.  This is not quite the right
424				 * thing to do, but at least if we are running
425				 * a routing process they will come back.
426				 */
427				in_ifadown(&ia->ia_ifa, 0);
428				break;
429			}
430		}
431		break;
432
433	case PRC_IFUP:
434		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
435			if (ia->ia_ifa.ifa_addr == sa)
436				break;
437		}
438		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
439			return;
440		flags = RTF_UP;
441		ifp = ia->ia_ifa.ifa_ifp;
442
443		if ((ifp->if_flags & IFF_LOOPBACK)
444		    || (ifp->if_flags & IFF_POINTOPOINT))
445			flags |= RTF_HOST;
446
447		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
448		if (err == 0)
449			ia->ia_flags |= IFA_ROUTE;
450		break;
451	}
452}
453
454u_long	rip_sendspace = RIPSNDQ;
455u_long	rip_recvspace = RIPRCVQ;
456
457SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
458    &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
459SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
460    &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
461
462static int
463rip_attach(struct socket *so, int proto, struct proc *p)
464{
465	struct inpcb *inp;
466	int error, s;
467
468	inp = sotoinpcb(so);
469	if (inp)
470		panic("rip_attach");
471	if (p && (error = suser(p)) != 0)
472		return error;
473
474	error = soreserve(so, rip_sendspace, rip_recvspace);
475	if (error)
476		return error;
477	s = splnet();
478	error = in_pcballoc(so, &ripcbinfo, p);
479	splx(s);
480	if (error)
481		return error;
482	inp = (struct inpcb *)so->so_pcb;
483	inp->inp_vflag |= INP_IPV4;
484	inp->inp_ip_p = proto;
485	inp->inp_ip_ttl = ip_defttl;
486	return 0;
487}
488
489static int
490rip_detach(struct socket *so)
491{
492	struct inpcb *inp;
493
494	inp = sotoinpcb(so);
495	if (inp == 0)
496		panic("rip_detach");
497	if (so == ip_mrouter)
498		ip_mrouter_done();
499	ip_rsvp_force_done(so);
500	if (so == ip_rsvpd)
501		ip_rsvp_done();
502	in_pcbdetach(inp);
503	return 0;
504}
505
506static int
507rip_abort(struct socket *so)
508{
509	soisdisconnected(so);
510	return rip_detach(so);
511}
512
513static int
514rip_disconnect(struct socket *so)
515{
516	if ((so->so_state & SS_ISCONNECTED) == 0)
517		return ENOTCONN;
518	return rip_abort(so);
519}
520
521static int
522rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
523{
524	struct inpcb *inp = sotoinpcb(so);
525	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
526
527	if (nam->sa_len != sizeof(*addr))
528		return EINVAL;
529
530	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
531				    (addr->sin_family != AF_IMPLINK)) ||
532	    (addr->sin_addr.s_addr &&
533	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
534		return EADDRNOTAVAIL;
535	inp->inp_laddr = addr->sin_addr;
536	return 0;
537}
538
539static int
540rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
541{
542	struct inpcb *inp = sotoinpcb(so);
543	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
544
545	if (nam->sa_len != sizeof(*addr))
546		return EINVAL;
547	if (TAILQ_EMPTY(&ifnet))
548		return EADDRNOTAVAIL;
549	if ((addr->sin_family != AF_INET) &&
550	    (addr->sin_family != AF_IMPLINK))
551		return EAFNOSUPPORT;
552	inp->inp_faddr = addr->sin_addr;
553	soisconnected(so);
554	return 0;
555}
556
557static int
558rip_shutdown(struct socket *so)
559{
560	socantsendmore(so);
561	return 0;
562}
563
564static int
565rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
566	 struct mbuf *control, struct proc *p)
567{
568	struct inpcb *inp = sotoinpcb(so);
569	register u_long dst;
570
571	if (so->so_state & SS_ISCONNECTED) {
572		if (nam) {
573			m_freem(m);
574			return EISCONN;
575		}
576		dst = inp->inp_faddr.s_addr;
577	} else {
578		if (nam == NULL) {
579			m_freem(m);
580			return ENOTCONN;
581		}
582		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
583	}
584	return rip_output(m, so, dst);
585}
586
587static int
588rip_pcblist(SYSCTL_HANDLER_ARGS)
589{
590	int error, i, n, s;
591	struct inpcb *inp, **inp_list;
592	inp_gen_t gencnt;
593	struct xinpgen xig;
594
595	/*
596	 * The process of preparing the TCB list is too time-consuming and
597	 * resource-intensive to repeat twice on every request.
598	 */
599	if (req->oldptr == 0) {
600		n = ripcbinfo.ipi_count;
601		req->oldidx = 2 * (sizeof xig)
602			+ (n + n/8) * sizeof(struct xinpcb);
603		return 0;
604	}
605
606	if (req->newptr != 0)
607		return EPERM;
608
609	/*
610	 * OK, now we're committed to doing something.
611	 */
612	s = splnet();
613	gencnt = ripcbinfo.ipi_gencnt;
614	n = ripcbinfo.ipi_count;
615	splx(s);
616
617	xig.xig_len = sizeof xig;
618	xig.xig_count = n;
619	xig.xig_gen = gencnt;
620	xig.xig_sogen = so_gencnt;
621	error = SYSCTL_OUT(req, &xig, sizeof xig);
622	if (error)
623		return error;
624
625	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
626	if (inp_list == 0)
627		return ENOMEM;
628
629	s = splnet();
630	for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
631	     inp = LIST_NEXT(inp, inp_list)) {
632		if (inp->inp_gencnt <= gencnt)
633			inp_list[i++] = inp;
634	}
635	splx(s);
636	n = i;
637
638	error = 0;
639	for (i = 0; i < n; i++) {
640		inp = inp_list[i];
641		if (inp->inp_gencnt <= gencnt) {
642			struct xinpcb xi;
643			xi.xi_len = sizeof xi;
644			/* XXX should avoid extra copy */
645			bcopy(inp, &xi.xi_inp, sizeof *inp);
646			if (inp->inp_socket)
647				sotoxsocket(inp->inp_socket, &xi.xi_socket);
648			error = SYSCTL_OUT(req, &xi, sizeof xi);
649		}
650	}
651	if (!error) {
652		/*
653		 * Give the user an updated idea of our state.
654		 * If the generation differs from what we told
655		 * her before, she knows that something happened
656		 * while we were processing this request, and it
657		 * might be necessary to retry.
658		 */
659		s = splnet();
660		xig.xig_gen = ripcbinfo.ipi_gencnt;
661		xig.xig_sogen = so_gencnt;
662		xig.xig_count = ripcbinfo.ipi_count;
663		splx(s);
664		error = SYSCTL_OUT(req, &xig, sizeof xig);
665	}
666	free(inp_list, M_TEMP);
667	return error;
668}
669
670SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
671	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
672
673struct pr_usrreqs rip_usrreqs = {
674	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
675	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
676	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
677	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
678	in_setsockaddr, sosend, soreceive, sopoll
679};
680