2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1988, 1993
5 * The Regents of the University of California.
6 * Copyright (c) 2005 Andre Oppermann, Internet Business Solutions AG.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. 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.
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
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
37 #include "opt_ipstealth.h"
39 #include <sys/param.h>
40 #include <sys/systm.h>
42 #include <sys/domain.h>
43 #include <sys/protosw.h>
44 #include <sys/socket.h>
46 #include <sys/kernel.h>
47 #include <sys/syslog.h>
48 #include <sys/sysctl.h>
51 #include <net/if_types.h>
52 #include <net/if_var.h>
53 #include <net/if_dl.h>
54 #include <net/route.h>
55 #include <net/route/nhop.h>
56 #include <net/netisr.h>
59 #include <netinet/in.h>
60 #include <netinet/in_fib.h>
61 #include <netinet/in_systm.h>
62 #include <netinet/in_var.h>
63 #include <netinet/ip.h>
64 #include <netinet/in_pcb.h>
65 #include <netinet/ip_var.h>
66 #include <netinet/ip_options.h>
67 #include <netinet/ip_icmp.h>
68 #include <machine/in_cksum.h>
70 #include <sys/socketvar.h>
72 VNET_DEFINE_STATIC(int, ip_dosourceroute);
73 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute,
74 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_dosourceroute), 0,
75 "Enable forwarding source routed IP packets");
76 #define V_ip_dosourceroute VNET(ip_dosourceroute)
78 VNET_DEFINE_STATIC(int, ip_acceptsourceroute);
79 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
80 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_acceptsourceroute), 0,
81 "Enable accepting source routed IP packets");
82 #define V_ip_acceptsourceroute VNET(ip_acceptsourceroute)
84 VNET_DEFINE(int, ip_doopts) = 1; /* 0 = ignore, 1 = process, 2 = reject */
85 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_VNET | CTLFLAG_RW,
86 &VNET_NAME(ip_doopts), 0, "Enable IP options processing ([LS]SRR, RR, TS)");
88 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
91 * Do option processing on a datagram, possibly discarding it if bad options
92 * are encountered, or forwarding it if source-routed.
94 * The pass argument is used when operating in the IPSTEALTH mode to tell
95 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
96 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
97 * options should be processed only if the packet is for us.
99 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
103 ip_dooptions(struct mbuf *m, int pass)
105 struct ip *ip = mtod(m, struct ip *);
107 struct in_ifaddr *ia;
108 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
109 struct in_addr *sin, dst;
111 struct nhop_object *nh;
112 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
116 /* Ignore or reject packets with IP options. */
117 if (V_ip_doopts == 0)
119 else if (V_ip_doopts == 2) {
121 code = ICMP_UNREACH_FILTER_PROHIB;
126 cp = (u_char *)(ip + 1);
127 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
128 for (; cnt > 0; cnt -= optlen, cp += optlen) {
129 opt = cp[IPOPT_OPTVAL];
130 if (opt == IPOPT_EOL)
132 if (opt == IPOPT_NOP)
135 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
136 code = &cp[IPOPT_OLEN] - (u_char *)ip;
139 optlen = cp[IPOPT_OLEN];
140 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
141 code = &cp[IPOPT_OLEN] - (u_char *)ip;
151 * Source routing with record. Find interface with current
152 * destination address. If none on this machine then drop if
153 * strictly routed, or do nothing if loosely routed. Record
154 * interface address and bring up next address component. If
155 * strictly routed make sure next address is on directly
161 if (V_ipstealth && pass > 0)
164 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
165 code = &cp[IPOPT_OLEN] - (u_char *)ip;
168 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
169 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
172 ipaddr.sin_addr = ip->ip_dst;
173 if (ifa_ifwithaddr_check((struct sockaddr *)&ipaddr)
175 if (opt == IPOPT_SSRR) {
177 code = ICMP_UNREACH_SRCFAIL;
180 if (!V_ip_dosourceroute)
181 goto nosourcerouting;
183 * Loose routing, and not at next destination
184 * yet; nothing to do except forward.
188 off--; /* 0 origin */
189 if (off > optlen - (int)sizeof(struct in_addr)) {
191 * End of source route. Should be for us.
193 if (!V_ip_acceptsourceroute)
194 goto nosourcerouting;
195 save_rte(m, cp, ip->ip_src);
202 if (!V_ip_dosourceroute) {
203 if (V_ipforwarding) {
204 char srcbuf[INET_ADDRSTRLEN];
205 char dstbuf[INET_ADDRSTRLEN];
208 * Acting as a router, so generate
213 "attempted source route from %s "
215 inet_ntoa_r(ip->ip_src, srcbuf),
216 inet_ntoa_r(ip->ip_dst, dstbuf));
218 code = ICMP_UNREACH_SRCFAIL;
222 * Not acting as a router, so
228 IPSTAT_INC(ips_cantforward);
235 * locate outgoing interface
237 (void)memcpy(&ipaddr.sin_addr, cp + off,
238 sizeof(ipaddr.sin_addr));
241 code = ICMP_UNREACH_SRCFAIL;
243 if (opt == IPOPT_SSRR) {
244 #define INA struct in_ifaddr *
245 #define SA struct sockaddr *
246 ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr,
249 ia = (INA)ifa_ifwithnet((SA)&ipaddr, 0,
254 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
255 sizeof(struct in_addr));
257 /* XXX MRT 0 for routing */
258 nh = fib4_lookup(M_GETFIB(m), ipaddr.sin_addr,
263 memcpy(cp + off, &(IA_SIN(nh->nh_ifa)->sin_addr),
264 sizeof(struct in_addr));
267 ip->ip_dst = ipaddr.sin_addr;
268 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
270 * Let ip_intr's mcast routing check handle mcast pkts
272 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
277 if (V_ipstealth && pass == 0)
280 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
281 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
284 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
285 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
289 * If no space remains, ignore.
291 off--; /* 0 origin */
292 if (off > optlen - (int)sizeof(struct in_addr))
294 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
295 sizeof(ipaddr.sin_addr));
297 * Locate outgoing interface; if we're the
298 * destination, use the incoming interface (should be
301 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) != NULL) {
302 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
303 sizeof(struct in_addr));
304 } else if ((nh = fib4_lookup(M_GETFIB(m),
305 ipaddr.sin_addr, 0, NHR_NONE, 0)) != NULL) {
306 memcpy(cp + off, &(IA_SIN(nh->nh_ifa)->sin_addr),
307 sizeof(struct in_addr));
310 code = ICMP_UNREACH_HOST;
313 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
318 if (V_ipstealth && pass == 0)
321 code = cp - (u_char *)ip;
322 if (optlen < 4 || optlen > 40) {
323 code = &cp[IPOPT_OLEN] - (u_char *)ip;
326 if ((off = cp[IPOPT_OFFSET]) < 5) {
327 code = &cp[IPOPT_OLEN] - (u_char *)ip;
330 if (off > optlen - (int)sizeof(int32_t)) {
331 cp[IPOPT_OFFSET + 1] += (1 << 4);
332 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
333 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
338 off--; /* 0 origin */
339 sin = (struct in_addr *)(cp + off);
340 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
342 case IPOPT_TS_TSONLY:
345 case IPOPT_TS_TSANDADDR:
346 if (off + sizeof(uint32_t) +
347 sizeof(struct in_addr) > optlen) {
348 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
351 ipaddr.sin_addr = dst;
352 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
356 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
357 sizeof(struct in_addr));
358 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
359 off += sizeof(struct in_addr);
362 case IPOPT_TS_PRESPEC:
363 if (off + sizeof(uint32_t) +
364 sizeof(struct in_addr) > optlen) {
365 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
368 (void)memcpy(&ipaddr.sin_addr, sin,
369 sizeof(struct in_addr));
370 if (ifa_ifwithaddr_check((SA)&ipaddr) == 0)
372 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
373 off += sizeof(struct in_addr);
377 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
381 (void)memcpy(cp + off, &ntime, sizeof(uint32_t));
382 cp[IPOPT_OFFSET] += sizeof(uint32_t);
385 if (forward && V_ipforwarding) {
391 icmp_error(m, type, code, 0, 0);
392 IPSTAT_INC(ips_badoptions);
397 * Save incoming source route for use in replies, to be picked up later by
398 * ip_srcroute if the receiver is interested.
401 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
404 struct ipopt_tag *opts;
406 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
407 sizeof(struct ipopt_tag), M_NOWAIT);
411 olen = option[IPOPT_OLEN];
412 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
413 m_tag_free((struct m_tag *)opts);
416 bcopy(option, opts->ip_srcrt.srcopt, olen);
417 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
418 opts->ip_srcrt.dst = dst;
419 m_tag_prepend(m, (struct m_tag *)opts);
423 * Retrieve incoming source route for use in replies, in the same form used
424 * by setsockopt. The first hop is placed before the options, will be
428 ip_srcroute(struct mbuf *m0)
430 struct in_addr *p, *q;
432 struct ipopt_tag *opts;
434 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
438 if (opts->ip_nhops == 0)
440 m = m_get(M_NOWAIT, MT_DATA);
444 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
446 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
447 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
448 sizeof(struct in_addr) + OPTSIZ;
451 * First, save first hop for return route.
453 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
454 *(mtod(m, struct in_addr *)) = *p--;
457 * Copy option fields and padding (nop) to mbuf.
459 opts->ip_srcrt.nop = IPOPT_NOP;
460 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
461 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
462 &(opts->ip_srcrt.nop), OPTSIZ);
463 q = (struct in_addr *)(mtod(m, caddr_t) +
464 sizeof(struct in_addr) + OPTSIZ);
467 * Record return path as an IP source route, reversing the path
468 * (pointers are now aligned).
470 while (p >= opts->ip_srcrt.route) {
474 * Last hop goes to final destination.
476 *q = opts->ip_srcrt.dst;
477 m_tag_delete(m0, (struct m_tag *)opts);
482 * Strip out IP options, at higher level protocol in the kernel.
485 ip_stripoptions(struct mbuf *m)
487 struct ip *ip = mtod(m, struct ip *);
490 olen = (ip->ip_hl << 2) - sizeof(struct ip);
492 if (m->m_flags & M_PKTHDR)
493 m->m_pkthdr.len -= olen;
494 ip->ip_len = htons(ntohs(ip->ip_len) - olen);
495 ip->ip_hl = sizeof(struct ip) >> 2;
497 bcopy((char *)ip + sizeof(struct ip) + olen, (ip + 1),
498 (size_t )(m->m_len - sizeof(struct ip)));
502 * Insert IP options into preformed packet. Adjust IP destination as
503 * required for IP source routing, as indicated by a non-zero in_addr at the
504 * start of the options.
506 * XXX This routine assumes that the packet has no options in place.
509 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
511 struct ipoption *p = mtod(opt, struct ipoption *);
513 struct ip *ip = mtod(m, struct ip *);
516 optlen = opt->m_len - sizeof(p->ipopt_dst);
517 if (optlen + ntohs(ip->ip_len) > IP_MAXPACKET) {
519 return (m); /* XXX should fail */
521 if (p->ipopt_dst.s_addr)
522 ip->ip_dst = p->ipopt_dst;
523 if (!M_WRITABLE(m) || M_LEADINGSPACE(m) < optlen) {
524 n = m_gethdr(M_NOWAIT, MT_DATA);
530 n->m_pkthdr.rcvif = NULL;
531 n->m_pkthdr.len += optlen;
532 m->m_len -= sizeof(struct ip);
533 m->m_data += sizeof(struct ip);
536 m->m_len = optlen + sizeof(struct ip);
537 m->m_data += max_linkhdr;
538 bcopy(ip, mtod(m, void *), sizeof(struct ip));
542 m->m_pkthdr.len += optlen;
543 bcopy(ip, mtod(m, void *), sizeof(struct ip));
545 ip = mtod(m, struct ip *);
546 bcopy(p->ipopt_list, ip + 1, optlen);
547 *phlen = sizeof(struct ip) + optlen;
548 ip->ip_v = IPVERSION;
549 ip->ip_hl = *phlen >> 2;
550 ip->ip_len = htons(ntohs(ip->ip_len) + optlen);
555 * Copy options from ip to jp, omitting those not copied during
559 ip_optcopy(struct ip *ip, struct ip *jp)
562 int opt, optlen, cnt;
564 cp = (u_char *)(ip + 1);
565 dp = (u_char *)(jp + 1);
566 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
567 for (; cnt > 0; cnt -= optlen, cp += optlen) {
569 if (opt == IPOPT_EOL)
571 if (opt == IPOPT_NOP) {
572 /* Preserve for IP mcast tunnel's LSRR alignment. */
578 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
579 ("ip_optcopy: malformed ipv4 option"));
580 optlen = cp[IPOPT_OLEN];
581 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
582 ("ip_optcopy: malformed ipv4 option"));
584 /* Bogus lengths should have been caught by ip_dooptions. */
587 if (IPOPT_COPIED(opt)) {
588 bcopy(cp, dp, optlen);
592 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
598 * Set up IP options in pcb for insertion in output packets. Store in mbuf
599 * with pointer in pcbopt, adding pseudo-option with destination address if
603 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
607 struct mbuf **pcbopt;
610 INP_WLOCK_ASSERT(inp);
612 pcbopt = &inp->inp_options;
614 /* turn off any old options */
616 (void)m_free(*pcbopt);
618 if (m == NULL || m->m_len == 0) {
620 * Only turning off any previous options.
627 if (m->m_len % sizeof(int32_t))
630 * IP first-hop destination address will be stored before actual
631 * options; move other options back and clear it when none present.
633 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
636 m->m_len += sizeof(struct in_addr);
637 cp = mtod(m, u_char *) + sizeof(struct in_addr);
638 bcopy(mtod(m, void *), cp, (unsigned)cnt);
639 bzero(mtod(m, void *), sizeof(struct in_addr));
641 for (; cnt > 0; cnt -= optlen, cp += optlen) {
642 opt = cp[IPOPT_OPTVAL];
643 if (opt == IPOPT_EOL)
645 if (opt == IPOPT_NOP)
648 if (cnt < IPOPT_OLEN + sizeof(*cp))
650 optlen = cp[IPOPT_OLEN];
651 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
662 * User process specifies route as:
666 * D must be our final destination (but we can't
667 * check that since we may not have connected yet).
668 * A is first hop destination, which doesn't appear
669 * in actual IP option, but is stored before the
672 /* XXX-BZ PRIV_NETINET_SETHDROPTS? */
673 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
675 m->m_len -= sizeof(struct in_addr);
676 cnt -= sizeof(struct in_addr);
677 optlen -= sizeof(struct in_addr);
678 cp[IPOPT_OLEN] = optlen;
680 * Move first hop before start of options.
682 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
683 sizeof(struct in_addr));
685 * Then copy rest of options back
686 * to close up the deleted entry.
688 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
690 (unsigned)cnt - (IPOPT_MINOFF - 1));
694 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
705 * Check for the presence of the IP Router Alert option [RFC2113]
706 * in the header of an IPv4 datagram.
708 * This call is not intended for use from the forwarding path; it is here
709 * so that protocol domains may check for the presence of the option.
710 * Given how FreeBSD's IPv4 stack is currently structured, the Router Alert
711 * option does not have much relevance to the implementation, though this
712 * may change in future.
713 * Router alert options SHOULD be passed if running in IPSTEALTH mode and
714 * we are not the endpoint.
715 * Length checks on individual options should already have been performed
716 * by ip_dooptions() therefore they are folded under INVARIANTS here.
718 * Return zero if not present or options are invalid, non-zero if present.
721 ip_checkrouteralert(struct mbuf *m)
723 struct ip *ip = mtod(m, struct ip *);
725 int opt, optlen, cnt, found_ra;
728 cp = (u_char *)(ip + 1);
729 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
730 for (; cnt > 0; cnt -= optlen, cp += optlen) {
731 opt = cp[IPOPT_OPTVAL];
732 if (opt == IPOPT_EOL)
734 if (opt == IPOPT_NOP)
738 if (cnt < IPOPT_OLEN + sizeof(*cp))
741 optlen = cp[IPOPT_OLEN];
743 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
750 if (optlen != IPOPT_OFFSET + sizeof(uint16_t) ||
751 (*((uint16_t *)&cp[IPOPT_OFFSET]) != 0))