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 * 4. 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: head/sys/netinet/raw_ip.c 157927 2006-04-21 09:25:40Z ps $
31 */
32
33#include "opt_inet6.h"
34#include "opt_ipsec.h"
35#include "opt_mac.h"
36
37#include <sys/param.h>
38#include <sys/jail.h>
39#include <sys/kernel.h>
40#include <sys/lock.h>
41#include <sys/mac.h>
42#include <sys/malloc.h>
43#include <sys/mbuf.h>
44#include <sys/proc.h>
45#include <sys/protosw.h>
46#include <sys/signalvar.h>
47#include <sys/socket.h>
48#include <sys/socketvar.h>
49#include <sys/sx.h>
50#include <sys/sysctl.h>
51#include <sys/systm.h>
52
53#include <vm/uma.h>
54
55#include <net/if.h>
56#include <net/route.h>
57
58#include <netinet/in.h>
59#include <netinet/in_systm.h>
60#include <netinet/in_pcb.h>
61#include <netinet/in_var.h>
62#include <netinet/ip.h>
63#include <netinet/ip_var.h>
64#include <netinet/ip_mroute.h>
65
66#include <netinet/ip_fw.h>
67#include <netinet/ip_dummynet.h>
68
69#ifdef FAST_IPSEC
70#include <netipsec/ipsec.h>
71#endif /*FAST_IPSEC*/
72
73#ifdef IPSEC
74#include <netinet6/ipsec.h>
75#endif /*IPSEC*/
76
77struct inpcbhead ripcb;
78struct inpcbinfo ripcbinfo;
79
80/* control hooks for ipfw and dummynet */
81ip_fw_ctl_t *ip_fw_ctl_ptr = NULL;
82ip_dn_ctl_t *ip_dn_ctl_ptr = NULL;
83
84/*
85 * hooks for multicast routing. They all default to NULL,
86 * so leave them not initialized and rely on BSS being set to 0.
87 */
88
89/* The socket used to communicate with the multicast routing daemon. */
90struct socket *ip_mrouter;
91
92/* The various mrouter and rsvp functions */
93int (*ip_mrouter_set)(struct socket *, struct sockopt *);
94int (*ip_mrouter_get)(struct socket *, struct sockopt *);
95int (*ip_mrouter_done)(void);
96int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
97 struct ip_moptions *);
98int (*mrt_ioctl)(int, caddr_t);
99int (*legal_vif_num)(int);
100u_long (*ip_mcast_src)(int);
101
102void (*rsvp_input_p)(struct mbuf *m, int off);
103int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
104void (*ip_rsvp_force_done)(struct socket *);
105
106/*
107 * Nominal space allocated to a raw ip socket.
108 */
109#define RIPSNDQ 8192
110#define RIPRCVQ 8192
111
112/*
113 * Raw interface to IP protocol.
114 */
115
116/*
117 * Initialize raw connection block q.
118 */
119static void
120rip_zone_change(void *tag)
121{
122
123 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
124}
125
126void
127rip_init()
128{
129 INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
130 LIST_INIT(&ripcb);
131 ripcbinfo.listhead = &ripcb;
132 /*
133 * XXX We don't use the hash list for raw IP, but it's easier
134 * to allocate a one entry hash list than it is to check all
135 * over the place for hashbase == NULL.
136 */
137 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
138 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
139 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb),
140 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
141 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
142 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change,
143 NULL, EVENTHANDLER_PRI_ANY);
144}
145
146static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
147
148static int
149raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n)
150{
151 int policyfail = 0;
152
153 INP_LOCK_ASSERT(last);
154
155#if defined(IPSEC) || defined(FAST_IPSEC)
156 /* check AH/ESP integrity. */
157 if (ipsec4_in_reject(n, last)) {
158 policyfail = 1;
159#ifdef IPSEC
160 ipsecstat.in_polvio++;
161#endif /*IPSEC*/
162 /* do not inject data to pcb */
163 }
164#endif /*IPSEC || FAST_IPSEC*/
165#ifdef MAC
166 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0)
167 policyfail = 1;
168#endif
169 /* Check the minimum TTL for socket. */
170 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
171 policyfail = 1;
172 if (!policyfail) {
173 struct mbuf *opts = NULL;
174 struct socket *so;
175
176 so = last->inp_socket;
177 if ((last->inp_flags & INP_CONTROLOPTS) ||
178 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
179 ip_savecontrol(last, &opts, ip, n);
180 SOCKBUF_LOCK(&so->so_rcv);
181 if (sbappendaddr_locked(&so->so_rcv,
182 (struct sockaddr *)&ripsrc, n, opts) == 0) {
183 /* should notify about lost packet */
184 m_freem(n);
185 if (opts)
186 m_freem(opts);
187 SOCKBUF_UNLOCK(&so->so_rcv);
188 } else
189 sorwakeup_locked(so);
190 } else
191 m_freem(n);
192 return policyfail;
193}
194
195/*
196 * Setup generic address and protocol structures
197 * for raw_input routine, then pass them along with
198 * mbuf chain.
199 */
200void
201rip_input(struct mbuf *m, int off)
202{
203 struct ip *ip = mtod(m, struct ip *);
204 int proto = ip->ip_p;
205 struct inpcb *inp, *last;
206
207 INP_INFO_RLOCK(&ripcbinfo);
208 ripsrc.sin_addr = ip->ip_src;
209 last = NULL;
210 LIST_FOREACH(inp, &ripcb, inp_list) {
211 INP_LOCK(inp);
212 if (inp->inp_ip_p && inp->inp_ip_p != proto) {
213 docontinue:
214 INP_UNLOCK(inp);
215 continue;
216 }
217#ifdef INET6
218 if ((inp->inp_vflag & INP_IPV4) == 0)
219 goto docontinue;
220#endif
221 if (inp->inp_laddr.s_addr &&
222 inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
223 goto docontinue;
224 if (inp->inp_faddr.s_addr &&
225 inp->inp_faddr.s_addr != ip->ip_src.s_addr)
226 goto docontinue;
227 if (jailed(inp->inp_socket->so_cred))
228 if (htonl(prison_getip(inp->inp_socket->so_cred)) !=
229 ip->ip_dst.s_addr)
230 goto docontinue;
231 if (last) {
232 struct mbuf *n;
233
234 n = m_copy(m, 0, (int)M_COPYALL);
235 if (n != NULL)
236 (void) raw_append(last, ip, n);
237 /* XXX count dropped packet */
238 INP_UNLOCK(last);
239 }
240 last = inp;
241 }
242 if (last != NULL) {
243 if (raw_append(last, ip, m) != 0)
244 ipstat.ips_delivered--;
245 INP_UNLOCK(last);
246 } else {
247 m_freem(m);
248 ipstat.ips_noproto++;
249 ipstat.ips_delivered--;
250 }
251 INP_INFO_RUNLOCK(&ripcbinfo);
252}
253
254/*
255 * Generate IP header and pass packet to ip_output.
256 * Tack on options user may have setup with control call.
257 */
258int
259rip_output(struct mbuf *m, struct socket *so, u_long dst)
260{
261 struct ip *ip;
262 int error;
263 struct inpcb *inp = sotoinpcb(so);
264 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
265 IP_ALLOWBROADCAST;
266
267 /*
268 * If the user handed us a complete IP packet, use it.
269 * Otherwise, allocate an mbuf for a header and fill it in.
270 */
271 if ((inp->inp_flags & INP_HDRINCL) == 0) {
272 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
273 m_freem(m);
274 return(EMSGSIZE);
275 }
276 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
277 if (m == NULL)
278 return(ENOBUFS);
279
280 INP_LOCK(inp);
281 ip = mtod(m, struct ip *);
282 ip->ip_tos = inp->inp_ip_tos;
283 if (inp->inp_flags & INP_DONTFRAG)
284 ip->ip_off = IP_DF;
285 else
286 ip->ip_off = 0;
287 ip->ip_p = inp->inp_ip_p;
288 ip->ip_len = m->m_pkthdr.len;
289 if (jailed(inp->inp_socket->so_cred))
290 ip->ip_src.s_addr =
291 htonl(prison_getip(inp->inp_socket->so_cred));
292 else
293 ip->ip_src = inp->inp_laddr;
294 ip->ip_dst.s_addr = dst;
295 ip->ip_ttl = inp->inp_ip_ttl;
296 } else {
297 if (m->m_pkthdr.len > IP_MAXPACKET) {
298 m_freem(m);
299 return(EMSGSIZE);
300 }
301 INP_LOCK(inp);
302 ip = mtod(m, struct ip *);
303 if (jailed(inp->inp_socket->so_cred)) {
304 if (ip->ip_src.s_addr !=
305 htonl(prison_getip(inp->inp_socket->so_cred))) {
306 INP_UNLOCK(inp);
307 m_freem(m);
308 return (EPERM);
309 }
310 }
311 /* don't allow both user specified and setsockopt options,
312 and don't allow packet length sizes that will crash */
313 if (((ip->ip_hl != (sizeof (*ip) >> 2))
314 && inp->inp_options)
315 || (ip->ip_len > m->m_pkthdr.len)
316 || (ip->ip_len < (ip->ip_hl << 2))) {
317 INP_UNLOCK(inp);
318 m_freem(m);
319 return EINVAL;
320 }
321 if (ip->ip_id == 0)
322 ip->ip_id = ip_newid();
323 /* XXX prevent ip_output from overwriting header fields */
324 flags |= IP_RAWOUTPUT;
325 ipstat.ips_rawout++;
326 }
327
328 if (inp->inp_vflag & INP_ONESBCAST)
329 flags |= IP_SENDONES;
330
331#ifdef MAC
332 mac_create_mbuf_from_inpcb(inp, m);
333#endif
334
335 error = ip_output(m, inp->inp_options, NULL, flags,
336 inp->inp_moptions, inp);
337 INP_UNLOCK(inp);
338 return error;
339}
340
341/*
342 * Raw IP socket option processing.
343 *
344 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
345 * only be created by a privileged process, and as such, socket option
346 * operations to manage system properties on any raw socket were allowed to
347 * take place without explicit additional access control checks. However,
348 * raw sockets can now also be created in jail(), and therefore explicit
349 * checks are now required. Likewise, raw sockets can be used by a process
350 * after it gives up privilege, so some caution is required. For options
351 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
352 * performed in ip_ctloutput() and therefore no check occurs here.
353 * Unilaterally checking suser() here breaks normal IP socket option
354 * operations on raw sockets.
355 *
356 * When adding new socket options here, make sure to add access control
357 * checks here as necessary.
358 */
359int
360rip_ctloutput(struct socket *so, struct sockopt *sopt)
361{
362 struct inpcb *inp = sotoinpcb(so);
363 int error, optval;
364
365 if (sopt->sopt_level != IPPROTO_IP)
366 return (EINVAL);
367
368 error = 0;
369 switch (sopt->sopt_dir) {
370 case SOPT_GET:
371 switch (sopt->sopt_name) {
372 case IP_HDRINCL:
373 optval = inp->inp_flags & INP_HDRINCL;
374 error = sooptcopyout(sopt, &optval, sizeof optval);
375 break;
376
377 case IP_FW_ADD: /* ADD actually returns the body... */
378 case IP_FW_GET:
379 case IP_FW_TABLE_GETSIZE:
380 case IP_FW_TABLE_LIST:
381 error = suser(curthread);
382 if (error != 0)
383 return (error);
384 if (ip_fw_ctl_ptr != NULL)
385 error = ip_fw_ctl_ptr(sopt);
386 else
387 error = ENOPROTOOPT;
388 break;
389
390 case IP_DUMMYNET_GET:
391 error = suser(curthread);
392 if (error != 0)
393 return (error);
394 if (ip_dn_ctl_ptr != NULL)
395 error = ip_dn_ctl_ptr(sopt);
396 else
397 error = ENOPROTOOPT;
398 break ;
399
400 case MRT_INIT:
401 case MRT_DONE:
402 case MRT_ADD_VIF:
403 case MRT_DEL_VIF:
404 case MRT_ADD_MFC:
405 case MRT_DEL_MFC:
406 case MRT_VERSION:
407 case MRT_ASSERT:
408 case MRT_API_SUPPORT:
409 case MRT_API_CONFIG:
410 case MRT_ADD_BW_UPCALL:
411 case MRT_DEL_BW_UPCALL:
412 error = suser(curthread);
413 if (error != 0)
414 return (error);
415 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
416 EOPNOTSUPP;
417 break;
418
419 default:
420 error = ip_ctloutput(so, sopt);
421 break;
422 }
423 break;
424
425 case SOPT_SET:
426 switch (sopt->sopt_name) {
427 case IP_HDRINCL:
428 error = sooptcopyin(sopt, &optval, sizeof optval,
429 sizeof optval);
430 if (error)
431 break;
432 if (optval)
433 inp->inp_flags |= INP_HDRINCL;
434 else
435 inp->inp_flags &= ~INP_HDRINCL;
436 break;
437
438 case IP_FW_ADD:
439 case IP_FW_DEL:
440 case IP_FW_FLUSH:
441 case IP_FW_ZERO:
442 case IP_FW_RESETLOG:
443 case IP_FW_TABLE_ADD:
444 case IP_FW_TABLE_DEL:
445 case IP_FW_TABLE_FLUSH:
446 error = suser(curthread);
447 if (error != 0)
448 return (error);
449 if (ip_fw_ctl_ptr != NULL)
450 error = ip_fw_ctl_ptr(sopt);
451 else
452 error = ENOPROTOOPT;
453 break;
454
455 case IP_DUMMYNET_CONFIGURE:
456 case IP_DUMMYNET_DEL:
457 case IP_DUMMYNET_FLUSH:
458 error = suser(curthread);
459 if (error != 0)
460 return (error);
461 if (ip_dn_ctl_ptr != NULL)
462 error = ip_dn_ctl_ptr(sopt);
463 else
464 error = ENOPROTOOPT ;
465 break ;
466
467 case IP_RSVP_ON:
468 error = suser(curthread);
469 if (error != 0)
470 return (error);
471 error = ip_rsvp_init(so);
472 break;
473
474 case IP_RSVP_OFF:
475 error = suser(curthread);
476 if (error != 0)
477 return (error);
478 error = ip_rsvp_done();
479 break;
480
481 case IP_RSVP_VIF_ON:
482 case IP_RSVP_VIF_OFF:
483 error = suser(curthread);
484 if (error != 0)
485 return (error);
486 error = ip_rsvp_vif ?
487 ip_rsvp_vif(so, sopt) : EINVAL;
488 break;
489
490 case MRT_INIT:
491 case MRT_DONE:
492 case MRT_ADD_VIF:
493 case MRT_DEL_VIF:
494 case MRT_ADD_MFC:
495 case MRT_DEL_MFC:
496 case MRT_VERSION:
497 case MRT_ASSERT:
498 case MRT_API_SUPPORT:
499 case MRT_API_CONFIG:
500 case MRT_ADD_BW_UPCALL:
501 case MRT_DEL_BW_UPCALL:
502 error = suser(curthread);
503 if (error != 0)
504 return (error);
505 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
506 EOPNOTSUPP;
507 break;
508
509 default:
510 error = ip_ctloutput(so, sopt);
511 break;
512 }
513 break;
514 }
515
516 return (error);
517}
518
519/*
520 * This function exists solely to receive the PRC_IFDOWN messages which
521 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa,
522 * and calls in_ifadown() to remove all routes corresponding to that address.
523 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
524 * interface routes.
525 */
526void
527rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
528{
529 struct in_ifaddr *ia;
530 struct ifnet *ifp;
531 int err;
532 int flags;
533
534 switch (cmd) {
535 case PRC_IFDOWN:
536 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
537 if (ia->ia_ifa.ifa_addr == sa
538 && (ia->ia_flags & IFA_ROUTE)) {
539 /*
540 * in_ifscrub kills the interface route.
541 */
542 in_ifscrub(ia->ia_ifp, ia);
543 /*
544 * in_ifadown gets rid of all the rest of
545 * the routes. This is not quite the right
546 * thing to do, but at least if we are running
547 * a routing process they will come back.
548 */
549 in_ifadown(&ia->ia_ifa, 0);
550 break;
551 }
552 }
553 break;
554
555 case PRC_IFUP:
556 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
557 if (ia->ia_ifa.ifa_addr == sa)
558 break;
559 }
560 if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
561 return;
562 flags = RTF_UP;
563 ifp = ia->ia_ifa.ifa_ifp;
564
565 if ((ifp->if_flags & IFF_LOOPBACK)
566 || (ifp->if_flags & IFF_POINTOPOINT))
567 flags |= RTF_HOST;
568
569 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
570 if (err == 0)
571 ia->ia_flags |= IFA_ROUTE;
572 break;
573 }
574}
575
576u_long rip_sendspace = RIPSNDQ;
577u_long rip_recvspace = RIPRCVQ;
578
579SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
580 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
581SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
582 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
583
584static int
585rip_attach(struct socket *so, int proto, struct thread *td)
586{
587 struct inpcb *inp;
588 int error;
589
590 inp = sotoinpcb(so);
591 KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
592 if (jailed(td->td_ucred) && !jail_allow_raw_sockets)
593 return (EPERM);
594 if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0)
595 return error;
596 if (proto >= IPPROTO_MAX || proto < 0)
597 return EPROTONOSUPPORT;
598 error = soreserve(so, rip_sendspace, rip_recvspace);
599 if (error)
600 return error;
601 INP_INFO_WLOCK(&ripcbinfo);
602 error = in_pcballoc(so, &ripcbinfo, "rawinp");
603 if (error) {
604 INP_INFO_WUNLOCK(&ripcbinfo);
605 return error;
606 }
607 inp = (struct inpcb *)so->so_pcb;
608 INP_LOCK(inp);
609 INP_INFO_WUNLOCK(&ripcbinfo);
610 inp->inp_vflag |= INP_IPV4;
611 inp->inp_ip_p = proto;
612 inp->inp_ip_ttl = ip_defttl;
613 INP_UNLOCK(inp);
614 return 0;
615}
616
617static void
618rip_pcbdetach(struct socket *so, struct inpcb *inp)
619{
620
621 INP_INFO_WLOCK_ASSERT(&ripcbinfo);
622 INP_LOCK_ASSERT(inp);
623
624 if (so == ip_mrouter && ip_mrouter_done)
625 ip_mrouter_done();
626 if (ip_rsvp_force_done)
627 ip_rsvp_force_done(so);
628 if (so == ip_rsvpd)
629 ip_rsvp_done();
630 in_pcbdetach(inp);
631 in_pcbfree(inp);
632}
633
634static void
635rip_detach(struct socket *so)
636{
637 struct inpcb *inp;
638
639 inp = sotoinpcb(so);
640 KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
641 INP_INFO_WLOCK(&ripcbinfo);
642 INP_LOCK(inp);
643 rip_pcbdetach(so, inp);
644 INP_INFO_WUNLOCK(&ripcbinfo);
645}
646
647static void
648rip_abort(struct socket *so)
649{
650 struct inpcb *inp;
651
652 inp = sotoinpcb(so);
653 KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
654 INP_INFO_WLOCK(&ripcbinfo);
655 INP_LOCK(inp);
656 soisdisconnected(so);
657 rip_pcbdetach(so, inp);
658 INP_INFO_WUNLOCK(&ripcbinfo);
659}
660
661static int
662rip_disconnect(struct socket *so)
663{
664 if ((so->so_state & SS_ISCONNECTED) == 0)
665 return ENOTCONN;
666 rip_abort(so);
667 return (0);
668}
669
670static int
671rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
672{
673 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
674 struct inpcb *inp;
675
676 if (nam->sa_len != sizeof(*addr))
677 return EINVAL;
678
679 if (jailed(td->td_ucred)) {
680 if (addr->sin_addr.s_addr == INADDR_ANY)
681 addr->sin_addr.s_addr =
682 htonl(prison_getip(td->td_ucred));
683 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr)
684 return (EADDRNOTAVAIL);
685 }
686
687 if (TAILQ_EMPTY(&ifnet) ||
688 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
689 (addr->sin_addr.s_addr &&
690 ifa_ifwithaddr((struct sockaddr *)addr) == 0))
691 return EADDRNOTAVAIL;
692
693 inp = sotoinpcb(so);
694 KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
695 INP_INFO_WLOCK(&ripcbinfo);
696 INP_LOCK(inp);
697 inp->inp_laddr = addr->sin_addr;
698 INP_UNLOCK(inp);
699 INP_INFO_WUNLOCK(&ripcbinfo);
700 return 0;
701}
702
703static int
704rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
705{
706 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
707 struct inpcb *inp;
708
709 if (nam->sa_len != sizeof(*addr))
710 return EINVAL;
711 if (TAILQ_EMPTY(&ifnet))
712 return EADDRNOTAVAIL;
713 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
714 return EAFNOSUPPORT;
715
716 inp = sotoinpcb(so);
717 KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
718 INP_INFO_WLOCK(&ripcbinfo);
719 INP_LOCK(inp);
720 inp->inp_faddr = addr->sin_addr;
721 soisconnected(so);
722 INP_UNLOCK(inp);
723 INP_INFO_WUNLOCK(&ripcbinfo);
724 return 0;
725}
726
727static int
728rip_shutdown(struct socket *so)
729{
730 struct inpcb *inp;
731
732 inp = sotoinpcb(so);
733 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL"));
734 INP_LOCK(inp);
735 socantsendmore(so);
736 INP_UNLOCK(inp);
737 return 0;
738}
739
740static int
741rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
742 struct mbuf *control, struct thread *td)
743{
744 struct inpcb *inp;
745 u_long dst;
746
747 inp = sotoinpcb(so);
748 KASSERT(inp != NULL, ("rip_send: inp == NULL"));
749 /*
750 * Note: 'dst' reads below are unlocked.
751 */
752 if (so->so_state & SS_ISCONNECTED) {
753 if (nam) {
754 m_freem(m);
755 return EISCONN;
756 }
757 dst = inp->inp_faddr.s_addr; /* Unlocked read. */
758 } else {
759 if (nam == NULL) {
760 m_freem(m);
761 return ENOTCONN;
762 }
763 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
764 }
765 return rip_output(m, so, dst);
766}
767
768static int
769rip_pcblist(SYSCTL_HANDLER_ARGS)
770{
771 int error, i, n;
772 struct inpcb *inp, **inp_list;
773 inp_gen_t gencnt;
774 struct xinpgen xig;
775
776 /*
777 * The process of preparing the TCB list is too time-consuming and
778 * resource-intensive to repeat twice on every request.
779 */
780 if (req->oldptr == 0) {
781 n = ripcbinfo.ipi_count;
782 req->oldidx = 2 * (sizeof xig)
783 + (n + n/8) * sizeof(struct xinpcb);
784 return 0;
785 }
786
787 if (req->newptr != 0)
788 return EPERM;
789
790 /*
791 * OK, now we're committed to doing something.
792 */
793 INP_INFO_RLOCK(&ripcbinfo);
794 gencnt = ripcbinfo.ipi_gencnt;
795 n = ripcbinfo.ipi_count;
796 INP_INFO_RUNLOCK(&ripcbinfo);
797
798 xig.xig_len = sizeof xig;
799 xig.xig_count = n;
800 xig.xig_gen = gencnt;
801 xig.xig_sogen = so_gencnt;
802 error = SYSCTL_OUT(req, &xig, sizeof xig);
803 if (error)
804 return error;
805
806 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
807 if (inp_list == 0)
808 return ENOMEM;
809
810 INP_INFO_RLOCK(&ripcbinfo);
811 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
812 inp = LIST_NEXT(inp, inp_list)) {
813 INP_LOCK(inp);
814 if (inp->inp_gencnt <= gencnt &&
815 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) {
816 /* XXX held references? */
817 inp_list[i++] = inp;
818 }
819 INP_UNLOCK(inp);
820 }
821 INP_INFO_RUNLOCK(&ripcbinfo);
822 n = i;
823
824 error = 0;
825 for (i = 0; i < n; i++) {
826 inp = inp_list[i];
827 if (inp->inp_gencnt <= gencnt) {
828 struct xinpcb xi;
829 bzero(&xi, sizeof(xi));
830 xi.xi_len = sizeof xi;
831 /* XXX should avoid extra copy */
832 bcopy(inp, &xi.xi_inp, sizeof *inp);
833 if (inp->inp_socket)
834 sotoxsocket(inp->inp_socket, &xi.xi_socket);
835 error = SYSCTL_OUT(req, &xi, sizeof xi);
836 }
837 }
838 if (!error) {
839 /*
840 * Give the user an updated idea of our state.
841 * If the generation differs from what we told
842 * her before, she knows that something happened
843 * while we were processing this request, and it
844 * might be necessary to retry.
845 */
846 INP_INFO_RLOCK(&ripcbinfo);
847 xig.xig_gen = ripcbinfo.ipi_gencnt;
848 xig.xig_sogen = so_gencnt;
849 xig.xig_count = ripcbinfo.ipi_count;
850 INP_INFO_RUNLOCK(&ripcbinfo);
851 error = SYSCTL_OUT(req, &xig, sizeof xig);
852 }
853 free(inp_list, M_TEMP);
854 return error;
855}
856
857/*
858 * This is the wrapper function for in_setsockaddr. We just pass down
859 * the pcbinfo for in_setpeeraddr to lock.
860 */
861static int
862rip_sockaddr(struct socket *so, struct sockaddr **nam)
863{
864 return (in_setsockaddr(so, nam, &ripcbinfo));
865}
866
867/*
868 * This is the wrapper function for in_setpeeraddr. We just pass down
869 * the pcbinfo for in_setpeeraddr to lock.
870 */
871static int
872rip_peeraddr(struct socket *so, struct sockaddr **nam)
873{
874 return (in_setpeeraddr(so, nam, &ripcbinfo));
875}
876
877
878SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
879 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
880
881struct pr_usrreqs rip_usrreqs = {
882 .pru_abort = rip_abort,
883 .pru_attach = rip_attach,
884 .pru_bind = rip_bind,
885 .pru_connect = rip_connect,
886 .pru_control = in_control,
887 .pru_detach = rip_detach,
888 .pru_disconnect = rip_disconnect,
889 .pru_peeraddr = rip_peeraddr,
890 .pru_send = rip_send,
891 .pru_shutdown = rip_shutdown,
892 .pru_sockaddr = rip_sockaddr,
893 .pru_sosetlabel = in_pcbsosetlabel
894};
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 * 4. 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: head/sys/netinet/raw_ip.c 157927 2006-04-21 09:25:40Z ps $
31 */
32
33#include "opt_inet6.h"
34#include "opt_ipsec.h"
35#include "opt_mac.h"
36
37#include <sys/param.h>
38#include <sys/jail.h>
39#include <sys/kernel.h>
40#include <sys/lock.h>
41#include <sys/mac.h>
42#include <sys/malloc.h>
43#include <sys/mbuf.h>
44#include <sys/proc.h>
45#include <sys/protosw.h>
46#include <sys/signalvar.h>
47#include <sys/socket.h>
48#include <sys/socketvar.h>
49#include <sys/sx.h>
50#include <sys/sysctl.h>
51#include <sys/systm.h>
52
53#include <vm/uma.h>
54
55#include <net/if.h>
56#include <net/route.h>
57
58#include <netinet/in.h>
59#include <netinet/in_systm.h>
60#include <netinet/in_pcb.h>
61#include <netinet/in_var.h>
62#include <netinet/ip.h>
63#include <netinet/ip_var.h>
64#include <netinet/ip_mroute.h>
65
66#include <netinet/ip_fw.h>
67#include <netinet/ip_dummynet.h>
68
69#ifdef FAST_IPSEC
70#include <netipsec/ipsec.h>
71#endif /*FAST_IPSEC*/
72
73#ifdef IPSEC
74#include <netinet6/ipsec.h>
75#endif /*IPSEC*/
76
77struct inpcbhead ripcb;
78struct inpcbinfo ripcbinfo;
79
80/* control hooks for ipfw and dummynet */
81ip_fw_ctl_t *ip_fw_ctl_ptr = NULL;
82ip_dn_ctl_t *ip_dn_ctl_ptr = NULL;
83
84/*
85 * hooks for multicast routing. They all default to NULL,
86 * so leave them not initialized and rely on BSS being set to 0.
87 */
88
89/* The socket used to communicate with the multicast routing daemon. */
90struct socket *ip_mrouter;
91
92/* The various mrouter and rsvp functions */
93int (*ip_mrouter_set)(struct socket *, struct sockopt *);
94int (*ip_mrouter_get)(struct socket *, struct sockopt *);
95int (*ip_mrouter_done)(void);
96int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
97 struct ip_moptions *);
98int (*mrt_ioctl)(int, caddr_t);
99int (*legal_vif_num)(int);
100u_long (*ip_mcast_src)(int);
101
102void (*rsvp_input_p)(struct mbuf *m, int off);
103int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
104void (*ip_rsvp_force_done)(struct socket *);
105
106/*
107 * Nominal space allocated to a raw ip socket.
108 */
109#define RIPSNDQ 8192
110#define RIPRCVQ 8192
111
112/*
113 * Raw interface to IP protocol.
114 */
115
116/*
117 * Initialize raw connection block q.
118 */
119static void
120rip_zone_change(void *tag)
121{
122
123 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
124}
125
126void
127rip_init()
128{
129 INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
130 LIST_INIT(&ripcb);
131 ripcbinfo.listhead = &ripcb;
132 /*
133 * XXX We don't use the hash list for raw IP, but it's easier
134 * to allocate a one entry hash list than it is to check all
135 * over the place for hashbase == NULL.
136 */
137 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
138 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
139 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb),
140 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
141 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
142 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change,
143 NULL, EVENTHANDLER_PRI_ANY);
144}
145
146static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
147
148static int
149raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n)
150{
151 int policyfail = 0;
152
153 INP_LOCK_ASSERT(last);
154
155#if defined(IPSEC) || defined(FAST_IPSEC)
156 /* check AH/ESP integrity. */
157 if (ipsec4_in_reject(n, last)) {
158 policyfail = 1;
159#ifdef IPSEC
160 ipsecstat.in_polvio++;
161#endif /*IPSEC*/
162 /* do not inject data to pcb */
163 }
164#endif /*IPSEC || FAST_IPSEC*/
165#ifdef MAC
166 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0)
167 policyfail = 1;
168#endif
169 /* Check the minimum TTL for socket. */
170 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
171 policyfail = 1;
172 if (!policyfail) {
173 struct mbuf *opts = NULL;
174 struct socket *so;
175
176 so = last->inp_socket;
177 if ((last->inp_flags & INP_CONTROLOPTS) ||
178 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
179 ip_savecontrol(last, &opts, ip, n);
180 SOCKBUF_LOCK(&so->so_rcv);
181 if (sbappendaddr_locked(&so->so_rcv,
182 (struct sockaddr *)&ripsrc, n, opts) == 0) {
183 /* should notify about lost packet */
184 m_freem(n);
185 if (opts)
186 m_freem(opts);
187 SOCKBUF_UNLOCK(&so->so_rcv);
188 } else
189 sorwakeup_locked(so);
190 } else
191 m_freem(n);
192 return policyfail;
193}
194
195/*
196 * Setup generic address and protocol structures
197 * for raw_input routine, then pass them along with
198 * mbuf chain.
199 */
200void
201rip_input(struct mbuf *m, int off)
202{
203 struct ip *ip = mtod(m, struct ip *);
204 int proto = ip->ip_p;
205 struct inpcb *inp, *last;
206
207 INP_INFO_RLOCK(&ripcbinfo);
208 ripsrc.sin_addr = ip->ip_src;
209 last = NULL;
210 LIST_FOREACH(inp, &ripcb, inp_list) {
211 INP_LOCK(inp);
212 if (inp->inp_ip_p && inp->inp_ip_p != proto) {
213 docontinue:
214 INP_UNLOCK(inp);
215 continue;
216 }
217#ifdef INET6
218 if ((inp->inp_vflag & INP_IPV4) == 0)
219 goto docontinue;
220#endif
221 if (inp->inp_laddr.s_addr &&
222 inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
223 goto docontinue;
224 if (inp->inp_faddr.s_addr &&
225 inp->inp_faddr.s_addr != ip->ip_src.s_addr)
226 goto docontinue;
227 if (jailed(inp->inp_socket->so_cred))
228 if (htonl(prison_getip(inp->inp_socket->so_cred)) !=
229 ip->ip_dst.s_addr)
230 goto docontinue;
231 if (last) {
232 struct mbuf *n;
233
234 n = m_copy(m, 0, (int)M_COPYALL);
235 if (n != NULL)
236 (void) raw_append(last, ip, n);
237 /* XXX count dropped packet */
238 INP_UNLOCK(last);
239 }
240 last = inp;
241 }
242 if (last != NULL) {
243 if (raw_append(last, ip, m) != 0)
244 ipstat.ips_delivered--;
245 INP_UNLOCK(last);
246 } else {
247 m_freem(m);
248 ipstat.ips_noproto++;
249 ipstat.ips_delivered--;
250 }
251 INP_INFO_RUNLOCK(&ripcbinfo);
252}
253
254/*
255 * Generate IP header and pass packet to ip_output.
256 * Tack on options user may have setup with control call.
257 */
258int
259rip_output(struct mbuf *m, struct socket *so, u_long dst)
260{
261 struct ip *ip;
262 int error;
263 struct inpcb *inp = sotoinpcb(so);
264 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
265 IP_ALLOWBROADCAST;
266
267 /*
268 * If the user handed us a complete IP packet, use it.
269 * Otherwise, allocate an mbuf for a header and fill it in.
270 */
271 if ((inp->inp_flags & INP_HDRINCL) == 0) {
272 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
273 m_freem(m);
274 return(EMSGSIZE);
275 }
276 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
277 if (m == NULL)
278 return(ENOBUFS);
279
280 INP_LOCK(inp);
281 ip = mtod(m, struct ip *);
282 ip->ip_tos = inp->inp_ip_tos;
283 if (inp->inp_flags & INP_DONTFRAG)
284 ip->ip_off = IP_DF;
285 else
286 ip->ip_off = 0;
287 ip->ip_p = inp->inp_ip_p;
288 ip->ip_len = m->m_pkthdr.len;
289 if (jailed(inp->inp_socket->so_cred))
290 ip->ip_src.s_addr =
291 htonl(prison_getip(inp->inp_socket->so_cred));
292 else
293 ip->ip_src = inp->inp_laddr;
294 ip->ip_dst.s_addr = dst;
295 ip->ip_ttl = inp->inp_ip_ttl;
296 } else {
297 if (m->m_pkthdr.len > IP_MAXPACKET) {
298 m_freem(m);
299 return(EMSGSIZE);
300 }
301 INP_LOCK(inp);
302 ip = mtod(m, struct ip *);
303 if (jailed(inp->inp_socket->so_cred)) {
304 if (ip->ip_src.s_addr !=
305 htonl(prison_getip(inp->inp_socket->so_cred))) {
306 INP_UNLOCK(inp);
307 m_freem(m);
308 return (EPERM);
309 }
310 }
311 /* don't allow both user specified and setsockopt options,
312 and don't allow packet length sizes that will crash */
313 if (((ip->ip_hl != (sizeof (*ip) >> 2))
314 && inp->inp_options)
315 || (ip->ip_len > m->m_pkthdr.len)
316 || (ip->ip_len < (ip->ip_hl << 2))) {
317 INP_UNLOCK(inp);
318 m_freem(m);
319 return EINVAL;
320 }
321 if (ip->ip_id == 0)
322 ip->ip_id = ip_newid();
323 /* XXX prevent ip_output from overwriting header fields */
324 flags |= IP_RAWOUTPUT;
325 ipstat.ips_rawout++;
326 }
327
328 if (inp->inp_vflag & INP_ONESBCAST)
329 flags |= IP_SENDONES;
330
331#ifdef MAC
332 mac_create_mbuf_from_inpcb(inp, m);
333#endif
334
335 error = ip_output(m, inp->inp_options, NULL, flags,
336 inp->inp_moptions, inp);
337 INP_UNLOCK(inp);
338 return error;
339}
340
341/*
342 * Raw IP socket option processing.
343 *
344 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
345 * only be created by a privileged process, and as such, socket option
346 * operations to manage system properties on any raw socket were allowed to
347 * take place without explicit additional access control checks. However,
348 * raw sockets can now also be created in jail(), and therefore explicit
349 * checks are now required. Likewise, raw sockets can be used by a process
350 * after it gives up privilege, so some caution is required. For options
351 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
352 * performed in ip_ctloutput() and therefore no check occurs here.
353 * Unilaterally checking suser() here breaks normal IP socket option
354 * operations on raw sockets.
355 *
356 * When adding new socket options here, make sure to add access control
357 * checks here as necessary.
358 */
359int
360rip_ctloutput(struct socket *so, struct sockopt *sopt)
361{
362 struct inpcb *inp = sotoinpcb(so);
363 int error, optval;
364
365 if (sopt->sopt_level != IPPROTO_IP)
366 return (EINVAL);
367
368 error = 0;
369 switch (sopt->sopt_dir) {
370 case SOPT_GET:
371 switch (sopt->sopt_name) {
372 case IP_HDRINCL:
373 optval = inp->inp_flags & INP_HDRINCL;
374 error = sooptcopyout(sopt, &optval, sizeof optval);
375 break;
376
377 case IP_FW_ADD: /* ADD actually returns the body... */
378 case IP_FW_GET:
379 case IP_FW_TABLE_GETSIZE:
380 case IP_FW_TABLE_LIST:
381 error = suser(curthread);
382 if (error != 0)
383 return (error);
384 if (ip_fw_ctl_ptr != NULL)
385 error = ip_fw_ctl_ptr(sopt);
386 else
387 error = ENOPROTOOPT;
388 break;
389
390 case IP_DUMMYNET_GET:
391 error = suser(curthread);
392 if (error != 0)
393 return (error);
394 if (ip_dn_ctl_ptr != NULL)
395 error = ip_dn_ctl_ptr(sopt);
396 else
397 error = ENOPROTOOPT;
398 break ;
399
400 case MRT_INIT:
401 case MRT_DONE:
402 case MRT_ADD_VIF:
403 case MRT_DEL_VIF:
404 case MRT_ADD_MFC:
405 case MRT_DEL_MFC:
406 case MRT_VERSION:
407 case MRT_ASSERT:
408 case MRT_API_SUPPORT:
409 case MRT_API_CONFIG:
410 case MRT_ADD_BW_UPCALL:
411 case MRT_DEL_BW_UPCALL:
412 error = suser(curthread);
413 if (error != 0)
414 return (error);
415 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
416 EOPNOTSUPP;
417 break;
418
419 default:
420 error = ip_ctloutput(so, sopt);
421 break;
422 }
423 break;
424
425 case SOPT_SET:
426 switch (sopt->sopt_name) {
427 case IP_HDRINCL:
428 error = sooptcopyin(sopt, &optval, sizeof optval,
429 sizeof optval);
430 if (error)
431 break;
432 if (optval)
433 inp->inp_flags |= INP_HDRINCL;
434 else
435 inp->inp_flags &= ~INP_HDRINCL;
436 break;
437
438 case IP_FW_ADD:
439 case IP_FW_DEL:
440 case IP_FW_FLUSH:
441 case IP_FW_ZERO:
442 case IP_FW_RESETLOG:
443 case IP_FW_TABLE_ADD:
444 case IP_FW_TABLE_DEL:
445 case IP_FW_TABLE_FLUSH:
446 error = suser(curthread);
447 if (error != 0)
448 return (error);
449 if (ip_fw_ctl_ptr != NULL)
450 error = ip_fw_ctl_ptr(sopt);
451 else
452 error = ENOPROTOOPT;
453 break;
454
455 case IP_DUMMYNET_CONFIGURE:
456 case IP_DUMMYNET_DEL:
457 case IP_DUMMYNET_FLUSH:
458 error = suser(curthread);
459 if (error != 0)
460 return (error);
461 if (ip_dn_ctl_ptr != NULL)
462 error = ip_dn_ctl_ptr(sopt);
463 else
464 error = ENOPROTOOPT ;
465 break ;
466
467 case IP_RSVP_ON:
468 error = suser(curthread);
469 if (error != 0)
470 return (error);
471 error = ip_rsvp_init(so);
472 break;
473
474 case IP_RSVP_OFF:
475 error = suser(curthread);
476 if (error != 0)
477 return (error);
478 error = ip_rsvp_done();
479 break;
480
481 case IP_RSVP_VIF_ON:
482 case IP_RSVP_VIF_OFF:
483 error = suser(curthread);
484 if (error != 0)
485 return (error);
486 error = ip_rsvp_vif ?
487 ip_rsvp_vif(so, sopt) : EINVAL;
488 break;
489
490 case MRT_INIT:
491 case MRT_DONE:
492 case MRT_ADD_VIF:
493 case MRT_DEL_VIF:
494 case MRT_ADD_MFC:
495 case MRT_DEL_MFC:
496 case MRT_VERSION:
497 case MRT_ASSERT:
498 case MRT_API_SUPPORT:
499 case MRT_API_CONFIG:
500 case MRT_ADD_BW_UPCALL:
501 case MRT_DEL_BW_UPCALL:
502 error = suser(curthread);
503 if (error != 0)
504 return (error);
505 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
506 EOPNOTSUPP;
507 break;
508
509 default:
510 error = ip_ctloutput(so, sopt);
511 break;
512 }
513 break;
514 }
515
516 return (error);
517}
518
519/*
520 * This function exists solely to receive the PRC_IFDOWN messages which
521 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa,
522 * and calls in_ifadown() to remove all routes corresponding to that address.
523 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
524 * interface routes.
525 */
526void
527rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
528{
529 struct in_ifaddr *ia;
530 struct ifnet *ifp;
531 int err;
532 int flags;
533
534 switch (cmd) {
535 case PRC_IFDOWN:
536 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
537 if (ia->ia_ifa.ifa_addr == sa
538 && (ia->ia_flags & IFA_ROUTE)) {
539 /*
540 * in_ifscrub kills the interface route.
541 */
542 in_ifscrub(ia->ia_ifp, ia);
543 /*
544 * in_ifadown gets rid of all the rest of
545 * the routes. This is not quite the right
546 * thing to do, but at least if we are running
547 * a routing process they will come back.
548 */
549 in_ifadown(&ia->ia_ifa, 0);
550 break;
551 }
552 }
553 break;
554
555 case PRC_IFUP:
556 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
557 if (ia->ia_ifa.ifa_addr == sa)
558 break;
559 }
560 if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
561 return;
562 flags = RTF_UP;
563 ifp = ia->ia_ifa.ifa_ifp;
564
565 if ((ifp->if_flags & IFF_LOOPBACK)
566 || (ifp->if_flags & IFF_POINTOPOINT))
567 flags |= RTF_HOST;
568
569 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
570 if (err == 0)
571 ia->ia_flags |= IFA_ROUTE;
572 break;
573 }
574}
575
576u_long rip_sendspace = RIPSNDQ;
577u_long rip_recvspace = RIPRCVQ;
578
579SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
580 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
581SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
582 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
583
584static int
585rip_attach(struct socket *so, int proto, struct thread *td)
586{
587 struct inpcb *inp;
588 int error;
589
590 inp = sotoinpcb(so);
591 KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
592 if (jailed(td->td_ucred) && !jail_allow_raw_sockets)
593 return (EPERM);
594 if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0)
595 return error;
596 if (proto >= IPPROTO_MAX || proto < 0)
597 return EPROTONOSUPPORT;
598 error = soreserve(so, rip_sendspace, rip_recvspace);
599 if (error)
600 return error;
601 INP_INFO_WLOCK(&ripcbinfo);
602 error = in_pcballoc(so, &ripcbinfo, "rawinp");
603 if (error) {
604 INP_INFO_WUNLOCK(&ripcbinfo);
605 return error;
606 }
607 inp = (struct inpcb *)so->so_pcb;
608 INP_LOCK(inp);
609 INP_INFO_WUNLOCK(&ripcbinfo);
610 inp->inp_vflag |= INP_IPV4;
611 inp->inp_ip_p = proto;
612 inp->inp_ip_ttl = ip_defttl;
613 INP_UNLOCK(inp);
614 return 0;
615}
616
617static void
618rip_pcbdetach(struct socket *so, struct inpcb *inp)
619{
620
621 INP_INFO_WLOCK_ASSERT(&ripcbinfo);
622 INP_LOCK_ASSERT(inp);
623
624 if (so == ip_mrouter && ip_mrouter_done)
625 ip_mrouter_done();
626 if (ip_rsvp_force_done)
627 ip_rsvp_force_done(so);
628 if (so == ip_rsvpd)
629 ip_rsvp_done();
630 in_pcbdetach(inp);
631 in_pcbfree(inp);
632}
633
634static void
635rip_detach(struct socket *so)
636{
637 struct inpcb *inp;
638
639 inp = sotoinpcb(so);
640 KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
641 INP_INFO_WLOCK(&ripcbinfo);
642 INP_LOCK(inp);
643 rip_pcbdetach(so, inp);
644 INP_INFO_WUNLOCK(&ripcbinfo);
645}
646
647static void
648rip_abort(struct socket *so)
649{
650 struct inpcb *inp;
651
652 inp = sotoinpcb(so);
653 KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
654 INP_INFO_WLOCK(&ripcbinfo);
655 INP_LOCK(inp);
656 soisdisconnected(so);
657 rip_pcbdetach(so, inp);
658 INP_INFO_WUNLOCK(&ripcbinfo);
659}
660
661static int
662rip_disconnect(struct socket *so)
663{
664 if ((so->so_state & SS_ISCONNECTED) == 0)
665 return ENOTCONN;
666 rip_abort(so);
667 return (0);
668}
669
670static int
671rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
672{
673 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
674 struct inpcb *inp;
675
676 if (nam->sa_len != sizeof(*addr))
677 return EINVAL;
678
679 if (jailed(td->td_ucred)) {
680 if (addr->sin_addr.s_addr == INADDR_ANY)
681 addr->sin_addr.s_addr =
682 htonl(prison_getip(td->td_ucred));
683 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr)
684 return (EADDRNOTAVAIL);
685 }
686
687 if (TAILQ_EMPTY(&ifnet) ||
688 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
689 (addr->sin_addr.s_addr &&
690 ifa_ifwithaddr((struct sockaddr *)addr) == 0))
691 return EADDRNOTAVAIL;
692
693 inp = sotoinpcb(so);
694 KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
695 INP_INFO_WLOCK(&ripcbinfo);
696 INP_LOCK(inp);
697 inp->inp_laddr = addr->sin_addr;
698 INP_UNLOCK(inp);
699 INP_INFO_WUNLOCK(&ripcbinfo);
700 return 0;
701}
702
703static int
704rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
705{
706 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
707 struct inpcb *inp;
708
709 if (nam->sa_len != sizeof(*addr))
710 return EINVAL;
711 if (TAILQ_EMPTY(&ifnet))
712 return EADDRNOTAVAIL;
713 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
714 return EAFNOSUPPORT;
715
716 inp = sotoinpcb(so);
717 KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
718 INP_INFO_WLOCK(&ripcbinfo);
719 INP_LOCK(inp);
720 inp->inp_faddr = addr->sin_addr;
721 soisconnected(so);
722 INP_UNLOCK(inp);
723 INP_INFO_WUNLOCK(&ripcbinfo);
724 return 0;
725}
726
727static int
728rip_shutdown(struct socket *so)
729{
730 struct inpcb *inp;
731
732 inp = sotoinpcb(so);
733 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL"));
734 INP_LOCK(inp);
735 socantsendmore(so);
736 INP_UNLOCK(inp);
737 return 0;
738}
739
740static int
741rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
742 struct mbuf *control, struct thread *td)
743{
744 struct inpcb *inp;
745 u_long dst;
746
747 inp = sotoinpcb(so);
748 KASSERT(inp != NULL, ("rip_send: inp == NULL"));
749 /*
750 * Note: 'dst' reads below are unlocked.
751 */
752 if (so->so_state & SS_ISCONNECTED) {
753 if (nam) {
754 m_freem(m);
755 return EISCONN;
756 }
757 dst = inp->inp_faddr.s_addr; /* Unlocked read. */
758 } else {
759 if (nam == NULL) {
760 m_freem(m);
761 return ENOTCONN;
762 }
763 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
764 }
765 return rip_output(m, so, dst);
766}
767
768static int
769rip_pcblist(SYSCTL_HANDLER_ARGS)
770{
771 int error, i, n;
772 struct inpcb *inp, **inp_list;
773 inp_gen_t gencnt;
774 struct xinpgen xig;
775
776 /*
777 * The process of preparing the TCB list is too time-consuming and
778 * resource-intensive to repeat twice on every request.
779 */
780 if (req->oldptr == 0) {
781 n = ripcbinfo.ipi_count;
782 req->oldidx = 2 * (sizeof xig)
783 + (n + n/8) * sizeof(struct xinpcb);
784 return 0;
785 }
786
787 if (req->newptr != 0)
788 return EPERM;
789
790 /*
791 * OK, now we're committed to doing something.
792 */
793 INP_INFO_RLOCK(&ripcbinfo);
794 gencnt = ripcbinfo.ipi_gencnt;
795 n = ripcbinfo.ipi_count;
796 INP_INFO_RUNLOCK(&ripcbinfo);
797
798 xig.xig_len = sizeof xig;
799 xig.xig_count = n;
800 xig.xig_gen = gencnt;
801 xig.xig_sogen = so_gencnt;
802 error = SYSCTL_OUT(req, &xig, sizeof xig);
803 if (error)
804 return error;
805
806 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
807 if (inp_list == 0)
808 return ENOMEM;
809
810 INP_INFO_RLOCK(&ripcbinfo);
811 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
812 inp = LIST_NEXT(inp, inp_list)) {
813 INP_LOCK(inp);
814 if (inp->inp_gencnt <= gencnt &&
815 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) {
816 /* XXX held references? */
817 inp_list[i++] = inp;
818 }
819 INP_UNLOCK(inp);
820 }
821 INP_INFO_RUNLOCK(&ripcbinfo);
822 n = i;
823
824 error = 0;
825 for (i = 0; i < n; i++) {
826 inp = inp_list[i];
827 if (inp->inp_gencnt <= gencnt) {
828 struct xinpcb xi;
829 bzero(&xi, sizeof(xi));
830 xi.xi_len = sizeof xi;
831 /* XXX should avoid extra copy */
832 bcopy(inp, &xi.xi_inp, sizeof *inp);
833 if (inp->inp_socket)
834 sotoxsocket(inp->inp_socket, &xi.xi_socket);
835 error = SYSCTL_OUT(req, &xi, sizeof xi);
836 }
837 }
838 if (!error) {
839 /*
840 * Give the user an updated idea of our state.
841 * If the generation differs from what we told
842 * her before, she knows that something happened
843 * while we were processing this request, and it
844 * might be necessary to retry.
845 */
846 INP_INFO_RLOCK(&ripcbinfo);
847 xig.xig_gen = ripcbinfo.ipi_gencnt;
848 xig.xig_sogen = so_gencnt;
849 xig.xig_count = ripcbinfo.ipi_count;
850 INP_INFO_RUNLOCK(&ripcbinfo);
851 error = SYSCTL_OUT(req, &xig, sizeof xig);
852 }
853 free(inp_list, M_TEMP);
854 return error;
855}
856
857/*
858 * This is the wrapper function for in_setsockaddr. We just pass down
859 * the pcbinfo for in_setpeeraddr to lock.
860 */
861static int
862rip_sockaddr(struct socket *so, struct sockaddr **nam)
863{
864 return (in_setsockaddr(so, nam, &ripcbinfo));
865}
866
867/*
868 * This is the wrapper function for in_setpeeraddr. We just pass down
869 * the pcbinfo for in_setpeeraddr to lock.
870 */
871static int
872rip_peeraddr(struct socket *so, struct sockaddr **nam)
873{
874 return (in_setpeeraddr(so, nam, &ripcbinfo));
875}
876
877
878SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
879 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
880
881struct pr_usrreqs rip_usrreqs = {
882 .pru_abort = rip_abort,
883 .pru_attach = rip_attach,
884 .pru_bind = rip_bind,
885 .pru_connect = rip_connect,
886 .pru_control = in_control,
887 .pru_detach = rip_detach,
888 .pru_disconnect = rip_disconnect,
889 .pru_peeraddr = rip_peeraddr,
890 .pru_send = rip_send,
891 .pru_shutdown = rip_shutdown,
892 .pru_sockaddr = rip_sockaddr,
893 .pru_sosetlabel = in_pcbsosetlabel
894};