1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1982, 1986, 1988, 1993
5 * The Regents of the University of California.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the University nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33 #include "opt_inet.h"
34 #include "opt_inet6.h"
35 #include "opt_ipsec.h"
36
37 #include <sys/param.h>
38 #include <sys/jail.h>
39 #include <sys/kernel.h>
40 #include <sys/eventhandler.h>
41 #include <sys/lock.h>
42 #include <sys/malloc.h>
43 #include <sys/mbuf.h>
44 #include <sys/hash.h>
45 #include <sys/priv.h>
46 #include <sys/proc.h>
47 #include <sys/protosw.h>
48 #include <sys/rwlock.h>
49 #include <sys/signalvar.h>
50 #include <sys/socket.h>
51 #include <sys/socketvar.h>
52 #include <sys/stdarg.h>
53 #include <sys/sx.h>
54 #include <sys/sysctl.h>
55 #include <sys/systm.h>
56
57 #include <vm/uma.h>
58
59 #include <net/if.h>
60 #include <net/if_var.h>
61 #include <net/route.h>
62 #include <net/route/route_ctl.h>
63 #include <net/vnet.h>
64
65 #include <netinet/in.h>
66 #include <netinet/in_systm.h>
67 #include <netinet/in_fib.h>
68 #include <netinet/in_pcb.h>
69 #include <netinet/in_var.h>
70 #include <netinet/if_ether.h>
71 #include <netinet/ip.h>
72 #include <netinet/ip_var.h>
73 #include <netinet/ip_mroute.h>
74 #include <netinet/ip_icmp.h>
75
76 #include <netipsec/ipsec_support.h>
77
78 #include <security/mac/mac_framework.h>
79
80 extern ipproto_input_t *ip_protox[];
81
82 VNET_DEFINE(int, ip_defttl) = IPDEFTTL;
83 SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_VNET | CTLFLAG_RW,
84 &VNET_NAME(ip_defttl), 0,
85 "Maximum TTL on IP packets");
86
87 VNET_DEFINE(struct inpcbinfo, ripcbinfo);
88 #define V_ripcbinfo VNET(ripcbinfo)
89
90 /*
91 * Control and data hooks for ipfw, dummynet, divert and so on.
92 * The data hooks are not used here but it is convenient
93 * to keep them all in one place.
94 */
95 VNET_DEFINE(ip_fw_ctl_ptr_t, ip_fw_ctl_ptr) = NULL;
96
97 int (*ip_dn_ctl_ptr)(struct sockopt *);
98 int (*ip_dn_io_ptr)(struct mbuf **, struct ip_fw_args *);
99 void (*ip_divert_ptr)(struct mbuf *, bool);
100 int (*ng_ipfw_input_p)(struct mbuf **, struct ip_fw_args *, bool);
101
102 #ifdef INET
103 /*
104 * Hooks for multicast routing. They all default to NULL, so leave them not
105 * initialized and rely on BSS being set to 0.
106 */
107
108 /*
109 * A per-VNET flag indicating whether multicast routing is enabled.
110 */
111 VNET_DEFINE(bool, ip_mrouting_enabled);
112
113 /*
114 * The various mrouter and rsvp functions.
115 */
116 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
117 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
118 void (*ip_mrouter_done)(struct socket *);
119 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
120 struct ip_moptions *);
121 int (*mrt_ioctl)(u_long, caddr_t, int);
122 int (*legal_vif_num)(int, int);
123 u_long (*ip_mcast_src)(int, int);
124
125 int (*rsvp_input_p)(struct mbuf **, int *, int);
126 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
127 void (*ip_rsvp_force_done)(struct socket *);
128 #endif /* INET */
129
130 #define V_rip_bind_all_fibs VNET(rip_bind_all_fibs)
131 VNET_DEFINE(int, rip_bind_all_fibs) = 1;
132 SYSCTL_INT(_net_inet_raw, OID_AUTO, bind_all_fibs, CTLFLAG_VNET | CTLFLAG_RDTUN,
133 &VNET_NAME(rip_bind_all_fibs), 0,
134 "Bound sockets receive traffic from all FIBs");
135
136 u_long rip_sendspace = 9216;
137 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
138 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
139
140 u_long rip_recvspace = 9216;
141 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
142 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
143
144 INPCBSTORAGE_DEFINE(ripcbstor, inpcb, "rawinp", "ripcb", "riphash");
145
146 static void
rip_init(void * arg __unused)147 rip_init(void *arg __unused)
148 {
149 #define INP_PCBHASH_RAW_SIZE 256
150 in_pcbinfo_init(&V_ripcbinfo, &ripcbstor, INP_PCBHASH_RAW_SIZE, 0, 0);
151 }
152 VNET_SYSINIT(rip_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rip_init, NULL);
153
154 #ifdef VIMAGE
155 static void
rip_destroy(void * unused __unused)156 rip_destroy(void *unused __unused)
157 {
158
159 in_pcbinfo_destroy(&V_ripcbinfo);
160 }
161 VNET_SYSUNINIT(raw_ip, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, rip_destroy, NULL);
162 #endif
163
164 #ifdef INET
165 static int
rip_append(struct inpcb * inp,struct ip * ip,struct mbuf * m,struct sockaddr_in * ripsrc)166 rip_append(struct inpcb *inp, struct ip *ip, struct mbuf *m,
167 struct sockaddr_in *ripsrc)
168 {
169 struct socket *so = inp->inp_socket;
170 struct mbuf *n, *opts = NULL;
171
172 INP_LOCK_ASSERT(inp);
173
174 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
175 /* check AH/ESP integrity. */
176 if (IPSEC_ENABLED(ipv4) && IPSEC_CHECK_POLICY(ipv4, m, inp) != 0)
177 return (0);
178 #endif /* IPSEC */
179 #ifdef MAC
180 if (mac_inpcb_check_deliver(inp, m) != 0)
181 return (0);
182 #endif
183 /* Check the minimum TTL for socket. */
184 if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl)
185 return (0);
186
187 if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
188 return (0);
189
190 if ((inp->inp_flags & INP_CONTROLOPTS) ||
191 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
192 ip_savecontrol(inp, &opts, ip, n);
193 SOCKBUF_LOCK(&so->so_rcv);
194 if (sbappendaddr_locked(&so->so_rcv,
195 (struct sockaddr *)ripsrc, n, opts) == 0) {
196 soroverflow_locked(so);
197 m_freem(n);
198 if (opts)
199 m_freem(opts);
200 return (0);
201 }
202 sorwakeup_locked(so);
203
204 return (1);
205 }
206
207 struct rip_inp_match_ctx {
208 struct ip *ip;
209 int proto;
210 };
211
212 static bool
rip_inp_match(const struct inpcb * inp,void * v)213 rip_inp_match(const struct inpcb *inp, void *v)
214 {
215 struct rip_inp_match_ctx *ctx = v;
216
217 if (inp->inp_ip_p && inp->inp_ip_p != ctx->proto)
218 return (false);
219 #ifdef INET6
220 /* XXX inp locking */
221 if ((inp->inp_vflag & INP_IPV4) == 0)
222 return (false);
223 #endif
224 if (!in_nullhost(inp->inp_laddr) &&
225 !in_hosteq(inp->inp_laddr, ctx->ip->ip_dst))
226 return (false);
227 if (!in_nullhost(inp->inp_faddr) &&
228 !in_hosteq(inp->inp_faddr, ctx->ip->ip_src))
229 return (false);
230 return (true);
231 }
232
233 /*
234 * Setup generic address and protocol structures for raw_input routine, then
235 * pass them along with mbuf chain.
236 */
237 int
rip_input(struct mbuf ** mp,int * offp,int proto)238 rip_input(struct mbuf **mp, int *offp, int proto)
239 {
240 struct rip_inp_match_ctx ctx = {
241 .ip = mtod(*mp, struct ip *),
242 .proto = proto,
243 };
244 struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
245 INPLOOKUP_RLOCKPCB, rip_inp_match, &ctx);
246 struct ifnet *ifp;
247 struct mbuf *m = *mp;
248 struct inpcb *inp;
249 struct sockaddr_in ripsrc;
250 int appended, fib;
251
252 M_ASSERTPKTHDR(m);
253
254 *mp = NULL;
255 appended = 0;
256
257 bzero(&ripsrc, sizeof(ripsrc));
258 ripsrc.sin_len = sizeof(ripsrc);
259 ripsrc.sin_family = AF_INET;
260 ripsrc.sin_addr = ctx.ip->ip_src;
261
262 fib = M_GETFIB(m);
263 ifp = m->m_pkthdr.rcvif;
264
265 inpi.mode = IN_ADDR_JHASH32(&ctx.ip->ip_src) & V_ripcbinfo.ipi_hashmask;
266 while ((inp = inp_next(&inpi)) != NULL) {
267 INP_RLOCK_ASSERT(inp);
268 if (jailed_without_vnet(inp->inp_cred) &&
269 prison_check_ip4(inp->inp_cred, &ctx.ip->ip_dst) != 0) {
270 /*
271 * XXX: If faddr was bound to multicast group,
272 * jailed raw socket will drop datagram.
273 */
274 continue;
275 }
276 if (V_rip_bind_all_fibs == 0 && fib != inp->inp_inc.inc_fibnum)
277 /*
278 * Sockets bound to a specific FIB can only receive
279 * packets from that FIB.
280 */
281 continue;
282 appended += rip_append(inp, ctx.ip, m, &ripsrc);
283 }
284
285 inpi.mode = INP_UNCONN_LIST;
286 MPASS(inpi.inp == NULL);
287 while ((inp = inp_next(&inpi)) != NULL) {
288 INP_RLOCK_ASSERT(inp);
289 if (jailed_without_vnet(inp->inp_cred) &&
290 !IN_MULTICAST(ntohl(ctx.ip->ip_dst.s_addr)) &&
291 prison_check_ip4(inp->inp_cred, &ctx.ip->ip_dst) != 0)
292 /*
293 * Allow raw socket in jail to receive multicast;
294 * assume process had PRIV_NETINET_RAW at attach,
295 * and fall through into normal filter path if so.
296 */
297 continue;
298 if (V_rip_bind_all_fibs == 0 && fib != inp->inp_inc.inc_fibnum)
299 continue;
300
301 /*
302 * If this raw socket has multicast state, and we
303 * have received a multicast, check if this socket
304 * should receive it, as multicast filtering is now
305 * the responsibility of the transport layer.
306 */
307 if (inp->inp_moptions != NULL &&
308 IN_MULTICAST(ntohl(ctx.ip->ip_dst.s_addr))) {
309 /*
310 * If the incoming datagram is for IGMP, allow it
311 * through unconditionally to the raw socket.
312 *
313 * In the case of IGMPv2, we may not have explicitly
314 * joined the group, and may have set IFF_ALLMULTI
315 * on the interface. imo_multi_filter() may discard
316 * control traffic we actually need to see.
317 *
318 * Userland multicast routing daemons should continue
319 * filter the control traffic appropriately.
320 */
321 int blocked;
322
323 blocked = MCAST_PASS;
324 if (proto != IPPROTO_IGMP) {
325 struct sockaddr_in group;
326
327 bzero(&group, sizeof(struct sockaddr_in));
328 group.sin_len = sizeof(struct sockaddr_in);
329 group.sin_family = AF_INET;
330 group.sin_addr = ctx.ip->ip_dst;
331
332 blocked = imo_multi_filter(inp->inp_moptions,
333 ifp,
334 (struct sockaddr *)&group,
335 (struct sockaddr *)&ripsrc);
336 }
337
338 if (blocked != MCAST_PASS) {
339 IPSTAT_INC(ips_notmember);
340 continue;
341 }
342 }
343 appended += rip_append(inp, ctx.ip, m, &ripsrc);
344 }
345 if (appended == 0 && ip_protox[ctx.ip->ip_p] == rip_input) {
346 IPSTAT_INC(ips_noproto);
347 IPSTAT_DEC(ips_delivered);
348 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL, 0, 0);
349 } else
350 m_freem(m);
351 return (IPPROTO_DONE);
352 }
353
354 /*
355 * Generate IP header and pass packet to ip_output. Tack on options user may
356 * have setup with control call.
357 */
358 static int
rip_send(struct socket * so,int pruflags,struct mbuf * m,struct sockaddr * nam,struct mbuf * control,struct thread * td)359 rip_send(struct socket *so, int pruflags, struct mbuf *m, struct sockaddr *nam,
360 struct mbuf *control, struct thread *td)
361 {
362 struct epoch_tracker et;
363 struct ip *ip;
364 struct inpcb *inp;
365 in_addr_t *dst;
366 int error, flags, cnt, hlen;
367 u_char opttype, optlen, *cp;
368
369 inp = sotoinpcb(so);
370 KASSERT(inp != NULL, ("rip_send: inp == NULL"));
371
372 if (control != NULL) {
373 m_freem(control);
374 control = NULL;
375 }
376
377 if (so->so_state & SS_ISCONNECTED) {
378 if (nam) {
379 error = EISCONN;
380 m_freem(m);
381 return (error);
382 }
383 dst = &inp->inp_faddr.s_addr;
384 } else {
385 if (nam == NULL)
386 error = ENOTCONN;
387 else if (nam->sa_family != AF_INET)
388 error = EAFNOSUPPORT;
389 else if (nam->sa_len != sizeof(struct sockaddr_in))
390 error = EINVAL;
391 else
392 error = 0;
393 if (error != 0) {
394 m_freem(m);
395 return (error);
396 }
397 dst = &((struct sockaddr_in *)nam)->sin_addr.s_addr;
398 }
399
400 flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
401 IP_ALLOWBROADCAST;
402
403 /*
404 * If the user handed us a complete IP packet, use it. Otherwise,
405 * allocate an mbuf for a header and fill it in.
406 */
407 if ((inp->inp_flags & INP_HDRINCL) == 0) {
408 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
409 m_freem(m);
410 return(EMSGSIZE);
411 }
412 M_PREPEND(m, sizeof(struct ip), M_NOWAIT);
413 if (m == NULL)
414 return(ENOBUFS);
415
416 INP_RLOCK(inp);
417 ip = mtod(m, struct ip *);
418 ip->ip_tos = inp->inp_ip_tos;
419 if (inp->inp_flags & INP_DONTFRAG)
420 ip->ip_off = htons(IP_DF);
421 else
422 ip->ip_off = htons(0);
423 ip->ip_p = inp->inp_ip_p;
424 ip->ip_len = htons(m->m_pkthdr.len);
425 ip->ip_src = inp->inp_laddr;
426 ip->ip_dst.s_addr = *dst;
427 if (V_fib_hash_outbound) {
428 uint32_t hash_type, hash_val;
429
430 hash_val = fib4_calc_software_hash(ip->ip_src,
431 ip->ip_dst, 0, 0, ip->ip_p, &hash_type);
432 m->m_pkthdr.flowid = hash_val;
433 M_HASHTYPE_SET(m, hash_type);
434 flags |= IP_NODEFAULTFLOWID;
435 }
436 if (jailed(inp->inp_cred)) {
437 /*
438 * prison_local_ip4() would be good enough but would
439 * let a source of INADDR_ANY pass, which we do not
440 * want to see from jails.
441 */
442 if (ip->ip_src.s_addr == INADDR_ANY) {
443 NET_EPOCH_ENTER(et);
444 error = in_pcbladdr(inp, &ip->ip_dst,
445 &ip->ip_src, inp->inp_cred);
446 NET_EPOCH_EXIT(et);
447 } else {
448 error = prison_local_ip4(inp->inp_cred,
449 &ip->ip_src);
450 }
451 if (error != 0) {
452 INP_RUNLOCK(inp);
453 m_freem(m);
454 return (error);
455 }
456 }
457 ip->ip_ttl = inp->inp_ip_ttl;
458 } else {
459 if (m->m_pkthdr.len > IP_MAXPACKET) {
460 m_freem(m);
461 return (EMSGSIZE);
462 }
463 if (m->m_pkthdr.len < sizeof(*ip)) {
464 m_freem(m);
465 return (EINVAL);
466 }
467 m = m_pullup(m, sizeof(*ip));
468 if (m == NULL)
469 return (ENOMEM);
470 ip = mtod(m, struct ip *);
471 hlen = ip->ip_hl << 2;
472 if (m->m_len < hlen) {
473 m = m_pullup(m, hlen);
474 if (m == NULL)
475 return (EINVAL);
476 ip = mtod(m, struct ip *);
477 }
478 if (V_fib_hash_outbound) {
479 uint32_t hash_type, hash_val;
480
481 hash_val = fib4_calc_software_hash(ip->ip_dst,
482 ip->ip_src, 0, 0, ip->ip_p, &hash_type);
483 m->m_pkthdr.flowid = hash_val;
484 M_HASHTYPE_SET(m, hash_type);
485 flags |= IP_NODEFAULTFLOWID;
486 }
487 INP_RLOCK(inp);
488 /*
489 * Don't allow both user specified and setsockopt options,
490 * and don't allow packet length sizes that will crash.
491 */
492 if ((hlen < sizeof (*ip))
493 || ((hlen > sizeof (*ip)) && inp->inp_options)
494 || (ntohs(ip->ip_len) != m->m_pkthdr.len)) {
495 INP_RUNLOCK(inp);
496 m_freem(m);
497 return (EINVAL);
498 }
499 error = prison_check_ip4(inp->inp_cred, &ip->ip_src);
500 if (error != 0) {
501 INP_RUNLOCK(inp);
502 m_freem(m);
503 return (error);
504 }
505 /*
506 * Don't allow IP options which do not have the required
507 * structure as specified in section 3.1 of RFC 791 on
508 * pages 15-23.
509 */
510 cp = (u_char *)(ip + 1);
511 cnt = hlen - sizeof (struct ip);
512 for (; cnt > 0; cnt -= optlen, cp += optlen) {
513 opttype = cp[IPOPT_OPTVAL];
514 if (opttype == IPOPT_EOL)
515 break;
516 if (opttype == IPOPT_NOP) {
517 optlen = 1;
518 continue;
519 }
520 if (cnt < IPOPT_OLEN + sizeof(u_char)) {
521 INP_RUNLOCK(inp);
522 m_freem(m);
523 return (EINVAL);
524 }
525 optlen = cp[IPOPT_OLEN];
526 if (optlen < IPOPT_OLEN + sizeof(u_char) ||
527 optlen > cnt) {
528 INP_RUNLOCK(inp);
529 m_freem(m);
530 return (EINVAL);
531 }
532 }
533 /*
534 * This doesn't allow application to specify ID of zero,
535 * but we got this limitation from the beginning of history.
536 */
537 if (ip->ip_id == 0)
538 ip_fillid(ip, V_ip_random_id);
539
540 /*
541 * XXX prevent ip_output from overwriting header fields.
542 */
543 flags |= IP_RAWOUTPUT;
544 IPSTAT_INC(ips_rawout);
545 }
546
547 if (inp->inp_flags & INP_ONESBCAST)
548 flags |= IP_SENDONES;
549
550 #ifdef MAC
551 mac_inpcb_create_mbuf(inp, m);
552 #endif
553
554 NET_EPOCH_ENTER(et);
555 error = ip_output(m, inp->inp_options, NULL, flags,
556 inp->inp_moptions, inp);
557 NET_EPOCH_EXIT(et);
558 INP_RUNLOCK(inp);
559 return (error);
560 }
561
562 /*
563 * Raw IP socket option processing.
564 *
565 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
566 * only be created by a privileged process, and as such, socket option
567 * operations to manage system properties on any raw socket were allowed to
568 * take place without explicit additional access control checks. However,
569 * raw sockets can now also be created in jail(), and therefore explicit
570 * checks are now required. Likewise, raw sockets can be used by a process
571 * after it gives up privilege, so some caution is required. For options
572 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
573 * performed in ip_ctloutput() and therefore no check occurs here.
574 * Unilaterally checking priv_check() here breaks normal IP socket option
575 * operations on raw sockets.
576 *
577 * When adding new socket options here, make sure to add access control
578 * checks here as necessary.
579 */
580 int
rip_ctloutput(struct socket * so,struct sockopt * sopt)581 rip_ctloutput(struct socket *so, struct sockopt *sopt)
582 {
583 struct inpcb *inp = sotoinpcb(so);
584 int error, optval;
585
586 if (sopt->sopt_level != IPPROTO_IP) {
587 if (sopt->sopt_dir == SOPT_SET &&
588 sopt->sopt_level == SOL_SOCKET &&
589 sopt->sopt_name == SO_SETFIB)
590 return (ip_ctloutput(so, sopt));
591 return (EINVAL);
592 }
593
594 error = 0;
595 switch (sopt->sopt_dir) {
596 case SOPT_GET:
597 switch (sopt->sopt_name) {
598 case IP_HDRINCL:
599 optval = inp->inp_flags & INP_HDRINCL;
600 error = sooptcopyout(sopt, &optval, sizeof optval);
601 break;
602
603 case IP_FW3: /* generic ipfw v.3 functions */
604 case IP_FW_ADD: /* ADD actually returns the body... */
605 case IP_FW_GET:
606 case IP_FW_TABLE_GETSIZE:
607 case IP_FW_TABLE_LIST:
608 case IP_FW_NAT_GET_CONFIG:
609 case IP_FW_NAT_GET_LOG:
610 if (V_ip_fw_ctl_ptr != NULL)
611 error = V_ip_fw_ctl_ptr(sopt);
612 else
613 error = ENOPROTOOPT;
614 break;
615
616 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
617 if (ip_dn_ctl_ptr != NULL)
618 error = ip_dn_ctl_ptr(sopt);
619 else
620 error = ENOPROTOOPT;
621 break ;
622
623 case MRT_INIT:
624 case MRT_DONE:
625 case MRT_ADD_VIF:
626 case MRT_DEL_VIF:
627 case MRT_ADD_MFC:
628 case MRT_DEL_MFC:
629 case MRT_VERSION:
630 case MRT_ASSERT:
631 case MRT_API_SUPPORT:
632 case MRT_API_CONFIG:
633 case MRT_ADD_BW_UPCALL:
634 case MRT_DEL_BW_UPCALL:
635 error = priv_check(curthread, PRIV_NETINET_MROUTE);
636 if (error != 0)
637 return (error);
638 if (inp->inp_ip_p != IPPROTO_IGMP)
639 return (EOPNOTSUPP);
640 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
641 EOPNOTSUPP;
642 break;
643
644 default:
645 error = ip_ctloutput(so, sopt);
646 break;
647 }
648 break;
649
650 case SOPT_SET:
651 switch (sopt->sopt_name) {
652 case IP_HDRINCL:
653 error = sooptcopyin(sopt, &optval, sizeof optval,
654 sizeof optval);
655 if (error)
656 break;
657 INP_WLOCK(inp);
658 if (optval)
659 inp->inp_flags |= INP_HDRINCL;
660 else
661 inp->inp_flags &= ~INP_HDRINCL;
662 INP_WUNLOCK(inp);
663 break;
664
665 case IP_FW3: /* generic ipfw v.3 functions */
666 case IP_FW_ADD:
667 case IP_FW_DEL:
668 case IP_FW_FLUSH:
669 case IP_FW_ZERO:
670 case IP_FW_RESETLOG:
671 case IP_FW_TABLE_ADD:
672 case IP_FW_TABLE_DEL:
673 case IP_FW_TABLE_FLUSH:
674 case IP_FW_NAT_CFG:
675 case IP_FW_NAT_DEL:
676 if (V_ip_fw_ctl_ptr != NULL)
677 error = V_ip_fw_ctl_ptr(sopt);
678 else
679 error = ENOPROTOOPT;
680 break;
681
682 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
683 if (ip_dn_ctl_ptr != NULL)
684 error = ip_dn_ctl_ptr(sopt);
685 else
686 error = ENOPROTOOPT ;
687 break ;
688
689 case IP_RSVP_ON:
690 error = priv_check(curthread, PRIV_NETINET_MROUTE);
691 if (error != 0)
692 return (error);
693 if (inp->inp_ip_p != IPPROTO_RSVP)
694 return (EOPNOTSUPP);
695 error = ip_rsvp_init(so);
696 break;
697
698 case IP_RSVP_OFF:
699 error = priv_check(curthread, PRIV_NETINET_MROUTE);
700 if (error != 0)
701 return (error);
702 error = ip_rsvp_done();
703 break;
704
705 case IP_RSVP_VIF_ON:
706 case IP_RSVP_VIF_OFF:
707 error = priv_check(curthread, PRIV_NETINET_MROUTE);
708 if (error != 0)
709 return (error);
710 if (inp->inp_ip_p != IPPROTO_RSVP)
711 return (EOPNOTSUPP);
712 error = ip_rsvp_vif ?
713 ip_rsvp_vif(so, sopt) : EINVAL;
714 break;
715
716 case MRT_INIT:
717 case MRT_DONE:
718 case MRT_ADD_VIF:
719 case MRT_DEL_VIF:
720 case MRT_ADD_MFC:
721 case MRT_DEL_MFC:
722 case MRT_VERSION:
723 case MRT_ASSERT:
724 case MRT_API_SUPPORT:
725 case MRT_API_CONFIG:
726 case MRT_ADD_BW_UPCALL:
727 case MRT_DEL_BW_UPCALL:
728 error = priv_check(curthread, PRIV_NETINET_MROUTE);
729 if (error != 0)
730 return (error);
731 if (inp->inp_ip_p != IPPROTO_IGMP)
732 return (EOPNOTSUPP);
733 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
734 EOPNOTSUPP;
735 break;
736
737 default:
738 error = ip_ctloutput(so, sopt);
739 break;
740 }
741 break;
742 }
743
744 return (error);
745 }
746
747 void
rip_ctlinput(struct icmp * icmp)748 rip_ctlinput(struct icmp *icmp)
749 {
750 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
751 if (IPSEC_ENABLED(ipv4))
752 IPSEC_CTLINPUT(ipv4, icmp);
753 #endif
754 }
755
756 static int
rip_attach(struct socket * so,int proto,struct thread * td)757 rip_attach(struct socket *so, int proto, struct thread *td)
758 {
759 struct inpcb *inp;
760 int error;
761
762 inp = sotoinpcb(so);
763 KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
764
765 error = priv_check(td, PRIV_NETINET_RAW);
766 if (error)
767 return (error);
768 if (proto >= IPPROTO_MAX || proto < 0)
769 return EPROTONOSUPPORT;
770 error = soreserve(so, rip_sendspace, rip_recvspace);
771 if (error)
772 return (error);
773 error = in_pcballoc(so, &V_ripcbinfo);
774 if (error)
775 return (error);
776 inp = (struct inpcb *)so->so_pcb;
777 inp->inp_ip_p = proto;
778 inp->inp_ip_ttl = V_ip_defttl;
779 INP_WUNLOCK(inp);
780 return (0);
781 }
782
783 static void
rip_detach(struct socket * so)784 rip_detach(struct socket *so)
785 {
786 struct inpcb *inp;
787
788 inp = sotoinpcb(so);
789 KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
790 KASSERT(inp->inp_faddr.s_addr == INADDR_ANY,
791 ("rip_detach: not closed"));
792
793 /* Disable mrouter first */
794 if (ip_mrouter_done != NULL)
795 ip_mrouter_done(so);
796
797 INP_WLOCK(inp);
798
799 if (ip_rsvp_force_done)
800 ip_rsvp_force_done(so);
801 if (so == V_ip_rsvpd)
802 ip_rsvp_done();
803 in_pcbfree(inp);
804 }
805
806 static void
rip_dodisconnect(struct inpcb * inp,bool disconnect_socket)807 rip_dodisconnect(struct inpcb *inp, bool disconnect_socket)
808 {
809
810 INP_WLOCK(inp);
811 inp->inp_faddr.s_addr = INADDR_ANY;
812 ripcb_disconnect(inp);
813 if (disconnect_socket) {
814 SOCK_LOCK(inp->inp_socket);
815 inp->inp_socket->so_state &= ~SS_ISCONNECTED;
816 SOCK_UNLOCK(inp->inp_socket);
817 }
818 INP_WUNLOCK(inp);
819 }
820
821 static void
rip_abort(struct socket * so)822 rip_abort(struct socket *so)
823 {
824 struct inpcb *inp;
825
826 inp = sotoinpcb(so);
827 KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
828
829 rip_dodisconnect(inp, true);
830 }
831
832 static void
rip_close(struct socket * so)833 rip_close(struct socket *so)
834 {
835 struct inpcb *inp;
836
837 inp = sotoinpcb(so);
838 KASSERT(inp != NULL, ("rip_close: inp == NULL"));
839
840 rip_dodisconnect(inp, true);
841 }
842
843 static int
rip_disconnect(struct socket * so)844 rip_disconnect(struct socket *so)
845 {
846 struct inpcb *inp;
847
848 if ((so->so_state & SS_ISCONNECTED) == 0)
849 return (ENOTCONN);
850
851 inp = sotoinpcb(so);
852 KASSERT(inp != NULL, ("rip_disconnect: inp == NULL"));
853
854 rip_dodisconnect(inp, true);
855 return (0);
856 }
857
858 static int
rip_bind(struct socket * so,struct sockaddr * nam,struct thread * td)859 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
860 {
861 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
862 struct inpcb *inp;
863 int error;
864
865 if (nam->sa_family != AF_INET)
866 return (EAFNOSUPPORT);
867 if (nam->sa_len != sizeof(*addr))
868 return (EINVAL);
869
870 error = prison_check_ip4(td->td_ucred, &addr->sin_addr);
871 if (error != 0)
872 return (error);
873
874 inp = sotoinpcb(so);
875 KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
876
877 if (CK_STAILQ_EMPTY(&V_ifnet) ||
878 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
879 (addr->sin_addr.s_addr &&
880 (inp->inp_flags & INP_BINDANY) == 0 &&
881 ifa_ifwithaddr_check((struct sockaddr *)addr) == 0))
882 return (EADDRNOTAVAIL);
883
884 INP_WLOCK(inp);
885 inp->inp_laddr = addr->sin_addr;
886 INP_WUNLOCK(inp);
887 return (0);
888 }
889
890 static int
rip_connect(struct socket * so,struct sockaddr * nam,struct thread * td)891 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
892 {
893 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
894 struct inpcb *inp;
895
896 if (nam->sa_len != sizeof(*addr))
897 return (EINVAL);
898 if (CK_STAILQ_EMPTY(&V_ifnet))
899 return (EADDRNOTAVAIL);
900 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
901 return (EAFNOSUPPORT);
902
903 inp = sotoinpcb(so);
904 KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
905
906 INP_WLOCK(inp);
907 if (inp->inp_faddr.s_addr != INADDR_ANY &&
908 addr->sin_addr.s_addr == INADDR_ANY)
909 rip_dodisconnect(inp, false);
910 if (addr->sin_addr.s_addr != INADDR_ANY) {
911 inp->inp_faddr = addr->sin_addr;
912 ripcb_connect(inp);
913 }
914 soisconnected(so);
915 INP_WUNLOCK(inp);
916 return (0);
917 }
918
919 static int
rip_shutdown(struct socket * so,enum shutdown_how how)920 rip_shutdown(struct socket *so, enum shutdown_how how)
921 {
922
923 SOCK_LOCK(so);
924 if (!(so->so_state & SS_ISCONNECTED)) {
925 SOCK_UNLOCK(so);
926 return (ENOTCONN);
927 }
928 SOCK_UNLOCK(so);
929
930 switch (how) {
931 case SHUT_RD:
932 sorflush(so);
933 break;
934 case SHUT_RDWR:
935 sorflush(so);
936 /* FALLTHROUGH */
937 case SHUT_WR:
938 socantsendmore(so);
939 }
940
941 return (0);
942 }
943 #endif /* INET */
944
945 static int
rip_pcblist(SYSCTL_HANDLER_ARGS)946 rip_pcblist(SYSCTL_HANDLER_ARGS)
947 {
948 struct inpcb_iterator inpi = INP_ALL_ITERATOR(&V_ripcbinfo,
949 INPLOOKUP_RLOCKPCB);
950 struct xinpgen xig;
951 struct inpcb *inp;
952 int error;
953
954 if (req->newptr != 0)
955 return (EPERM);
956
957 if (req->oldptr == 0) {
958 int n;
959
960 n = V_ripcbinfo.ipi_count;
961 n += imax(n / 8, 10);
962 req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb);
963 return (0);
964 }
965
966 if ((error = sysctl_wire_old_buffer(req, 0)) != 0)
967 return (error);
968
969 bzero(&xig, sizeof(xig));
970 xig.xig_len = sizeof xig;
971 xig.xig_count = V_ripcbinfo.ipi_count;
972 xig.xig_gen = V_ripcbinfo.ipi_gencnt;
973 xig.xig_sogen = so_gencnt;
974 error = SYSCTL_OUT(req, &xig, sizeof xig);
975 if (error)
976 return (error);
977
978 while ((inp = inp_next(&inpi)) != NULL) {
979 if (inp->inp_gencnt <= xig.xig_gen &&
980 cr_canseeinpcb(req->td->td_ucred, inp) == 0) {
981 struct xinpcb xi;
982
983 in_pcbtoxinpcb(inp, &xi);
984 error = SYSCTL_OUT(req, &xi, sizeof xi);
985 if (error) {
986 INP_RUNLOCK(inp);
987 break;
988 }
989 }
990 }
991
992 if (!error) {
993 /*
994 * Give the user an updated idea of our state. If the
995 * generation differs from what we told her before, she knows
996 * that something happened while we were processing this
997 * request, and it might be necessary to retry.
998 */
999 xig.xig_gen = V_ripcbinfo.ipi_gencnt;
1000 xig.xig_sogen = so_gencnt;
1001 xig.xig_count = V_ripcbinfo.ipi_count;
1002 error = SYSCTL_OUT(req, &xig, sizeof xig);
1003 }
1004
1005 return (error);
1006 }
1007
1008 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist,
1009 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0,
1010 rip_pcblist, "S,xinpcb",
1011 "List of active raw IP sockets");
1012
1013 #ifdef INET
1014 struct protosw rip_protosw = {
1015 .pr_type = SOCK_RAW,
1016 .pr_flags = PR_ATOMIC|PR_ADDR,
1017 .pr_ctloutput = rip_ctloutput,
1018 .pr_abort = rip_abort,
1019 .pr_attach = rip_attach,
1020 .pr_bind = rip_bind,
1021 .pr_connect = rip_connect,
1022 .pr_control = in_control,
1023 .pr_detach = rip_detach,
1024 .pr_disconnect = rip_disconnect,
1025 .pr_peeraddr = in_getpeeraddr,
1026 .pr_send = rip_send,
1027 .pr_shutdown = rip_shutdown,
1028 .pr_sockaddr = in_getsockaddr,
1029 .pr_sosetlabel = in_pcbsosetlabel,
1030 .pr_close = rip_close
1031 };
1032 #endif /* INET */
1033