2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California.
4 * Copyright (c) 2005 Andre Oppermann, Internet Business Solutions AG.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 4. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 #include <sys/cdefs.h>
33 __FBSDID("$FreeBSD$");
35 #include "opt_ipstealth.h"
37 #include <sys/param.h>
38 #include <sys/systm.h>
40 #include <sys/domain.h>
41 #include <sys/protosw.h>
42 #include <sys/socket.h>
44 #include <sys/kernel.h>
45 #include <sys/syslog.h>
46 #include <sys/sysctl.h>
49 #include <net/if_types.h>
50 #include <net/if_var.h>
51 #include <net/if_dl.h>
52 #include <net/route.h>
53 #include <net/netisr.h>
56 #include <netinet/in.h>
57 #include <netinet/in_systm.h>
58 #include <netinet/in_var.h>
59 #include <netinet/ip.h>
60 #include <netinet/in_pcb.h>
61 #include <netinet/ip_var.h>
62 #include <netinet/ip_options.h>
63 #include <netinet/ip_icmp.h>
64 #include <machine/in_cksum.h>
66 #include <sys/socketvar.h>
68 static int ip_dosourceroute = 0;
69 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
70 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
72 static int ip_acceptsourceroute = 0;
73 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
74 CTLFLAG_RW, &ip_acceptsourceroute, 0,
75 "Enable accepting source routed IP packets");
77 int ip_doopts = 1; /* 0 = ignore, 1 = process, 2 = reject */
78 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_RW,
79 &ip_doopts, 0, "Enable IP options processing ([LS]SRR, RR, TS)");
81 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
84 * Do option processing on a datagram, possibly discarding it if bad options
85 * are encountered, or forwarding it if source-routed.
87 * The pass argument is used when operating in the IPSTEALTH mode to tell
88 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
89 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
90 * options should be processed only if the packet is for us.
92 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
96 ip_dooptions(struct mbuf *m, int pass)
98 struct ip *ip = mtod(m, struct ip *);
100 struct in_ifaddr *ia;
101 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
102 struct in_addr *sin, dst;
104 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
106 /* Ignore or reject packets with IP options. */
109 else if (ip_doopts == 2) {
111 code = ICMP_UNREACH_FILTER_PROHIB;
116 cp = (u_char *)(ip + 1);
117 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
118 for (; cnt > 0; cnt -= optlen, cp += optlen) {
119 opt = cp[IPOPT_OPTVAL];
120 if (opt == IPOPT_EOL)
122 if (opt == IPOPT_NOP)
125 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
126 code = &cp[IPOPT_OLEN] - (u_char *)ip;
129 optlen = cp[IPOPT_OLEN];
130 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
131 code = &cp[IPOPT_OLEN] - (u_char *)ip;
141 * Source routing with record. Find interface with current
142 * destination address. If none on this machine then drop if
143 * strictly routed, or do nothing if loosely routed. Record
144 * interface address and bring up next address component. If
145 * strictly routed make sure next address is on directly
151 if (V_ipstealth && pass > 0)
154 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
155 code = &cp[IPOPT_OLEN] - (u_char *)ip;
158 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
159 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
162 ipaddr.sin_addr = ip->ip_dst;
163 if (ifa_ifwithaddr_check((struct sockaddr *)&ipaddr)
165 if (opt == IPOPT_SSRR) {
167 code = ICMP_UNREACH_SRCFAIL;
170 if (!ip_dosourceroute)
171 goto nosourcerouting;
173 * Loose routing, and not at next destination
174 * yet; nothing to do except forward.
178 off--; /* 0 origin */
179 if (off > optlen - (int)sizeof(struct in_addr)) {
181 * End of source route. Should be for us.
183 if (!ip_acceptsourceroute)
184 goto nosourcerouting;
185 save_rte(m, cp, ip->ip_src);
192 if (!ip_dosourceroute) {
193 if (V_ipforwarding) {
194 char buf[16]; /* aaa.bbb.ccc.ddd\0 */
196 * Acting as a router, so generate
200 strcpy(buf, inet_ntoa(ip->ip_dst));
202 "attempted source route from %s to %s\n",
203 inet_ntoa(ip->ip_src), buf);
205 code = ICMP_UNREACH_SRCFAIL;
209 * Not acting as a router, so
215 IPSTAT_INC(ips_cantforward);
222 * locate outgoing interface
224 (void)memcpy(&ipaddr.sin_addr, cp + off,
225 sizeof(ipaddr.sin_addr));
227 if (opt == IPOPT_SSRR) {
228 #define INA struct in_ifaddr *
229 #define SA struct sockaddr *
230 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == NULL)
231 ia = (INA)ifa_ifwithnet((SA)&ipaddr, 0);
233 /* XXX MRT 0 for routing */
234 ia = ip_rtaddr(ipaddr.sin_addr, M_GETFIB(m));
237 code = ICMP_UNREACH_SRCFAIL;
240 ip->ip_dst = ipaddr.sin_addr;
241 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
242 sizeof(struct in_addr));
243 ifa_free(&ia->ia_ifa);
244 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
246 * Let ip_intr's mcast routing check handle mcast pkts
248 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
253 if (V_ipstealth && pass == 0)
256 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
257 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
260 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
261 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
265 * If no space remains, ignore.
267 off--; /* 0 origin */
268 if (off > optlen - (int)sizeof(struct in_addr))
270 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
271 sizeof(ipaddr.sin_addr));
273 * Locate outgoing interface; if we're the
274 * destination, use the incoming interface (should be
277 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
278 (ia = ip_rtaddr(ipaddr.sin_addr, M_GETFIB(m))) == NULL) {
280 code = ICMP_UNREACH_HOST;
283 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
284 sizeof(struct in_addr));
285 ifa_free(&ia->ia_ifa);
286 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
291 if (V_ipstealth && pass == 0)
294 code = cp - (u_char *)ip;
295 if (optlen < 4 || optlen > 40) {
296 code = &cp[IPOPT_OLEN] - (u_char *)ip;
299 if ((off = cp[IPOPT_OFFSET]) < 5) {
300 code = &cp[IPOPT_OLEN] - (u_char *)ip;
303 if (off > optlen - (int)sizeof(int32_t)) {
304 cp[IPOPT_OFFSET + 1] += (1 << 4);
305 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
306 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
311 off--; /* 0 origin */
312 sin = (struct in_addr *)(cp + off);
313 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
315 case IPOPT_TS_TSONLY:
318 case IPOPT_TS_TSANDADDR:
319 if (off + sizeof(uint32_t) +
320 sizeof(struct in_addr) > optlen) {
321 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
324 ipaddr.sin_addr = dst;
325 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
329 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
330 sizeof(struct in_addr));
331 ifa_free(&ia->ia_ifa);
332 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
333 off += sizeof(struct in_addr);
336 case IPOPT_TS_PRESPEC:
337 if (off + sizeof(uint32_t) +
338 sizeof(struct in_addr) > optlen) {
339 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
342 (void)memcpy(&ipaddr.sin_addr, sin,
343 sizeof(struct in_addr));
344 if (ifa_ifwithaddr_check((SA)&ipaddr) == 0)
346 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
347 off += sizeof(struct in_addr);
351 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
355 (void)memcpy(cp + off, &ntime, sizeof(uint32_t));
356 cp[IPOPT_OFFSET] += sizeof(uint32_t);
359 if (forward && V_ipforwarding) {
365 icmp_error(m, type, code, 0, 0);
366 IPSTAT_INC(ips_badoptions);
371 * Save incoming source route for use in replies, to be picked up later by
372 * ip_srcroute if the receiver is interested.
375 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
378 struct ipopt_tag *opts;
380 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
381 sizeof(struct ipopt_tag), M_NOWAIT);
385 olen = option[IPOPT_OLEN];
386 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
387 m_tag_free((struct m_tag *)opts);
390 bcopy(option, opts->ip_srcrt.srcopt, olen);
391 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
392 opts->ip_srcrt.dst = dst;
393 m_tag_prepend(m, (struct m_tag *)opts);
397 * Retrieve incoming source route for use in replies, in the same form used
398 * by setsockopt. The first hop is placed before the options, will be
402 ip_srcroute(struct mbuf *m0)
404 struct in_addr *p, *q;
406 struct ipopt_tag *opts;
408 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
412 if (opts->ip_nhops == 0)
414 m = m_get(M_NOWAIT, MT_DATA);
418 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
420 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
421 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
422 sizeof(struct in_addr) + OPTSIZ;
425 * First, save first hop for return route.
427 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
428 *(mtod(m, struct in_addr *)) = *p--;
431 * Copy option fields and padding (nop) to mbuf.
433 opts->ip_srcrt.nop = IPOPT_NOP;
434 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
435 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
436 &(opts->ip_srcrt.nop), OPTSIZ);
437 q = (struct in_addr *)(mtod(m, caddr_t) +
438 sizeof(struct in_addr) + OPTSIZ);
441 * Record return path as an IP source route, reversing the path
442 * (pointers are now aligned).
444 while (p >= opts->ip_srcrt.route) {
448 * Last hop goes to final destination.
450 *q = opts->ip_srcrt.dst;
451 m_tag_delete(m0, (struct m_tag *)opts);
456 * Strip out IP options, at higher level protocol in the kernel.
459 ip_stripoptions(struct mbuf *m)
461 struct ip *ip = mtod(m, struct ip *);
464 olen = (ip->ip_hl << 2) - sizeof(struct ip);
466 if (m->m_flags & M_PKTHDR)
467 m->m_pkthdr.len -= olen;
468 ip->ip_len = htons(ntohs(ip->ip_len) - olen);
469 ip->ip_hl = sizeof(struct ip) >> 2;
471 bcopy((char *)ip + sizeof(struct ip) + olen, (ip + 1),
472 (size_t )(m->m_len - sizeof(struct ip)));
476 * Insert IP options into preformed packet. Adjust IP destination as
477 * required for IP source routing, as indicated by a non-zero in_addr at the
478 * start of the options.
480 * XXX This routine assumes that the packet has no options in place.
483 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
485 struct ipoption *p = mtod(opt, struct ipoption *);
487 struct ip *ip = mtod(m, struct ip *);
490 optlen = opt->m_len - sizeof(p->ipopt_dst);
491 if (optlen + ntohs(ip->ip_len) > IP_MAXPACKET) {
493 return (m); /* XXX should fail */
495 if (p->ipopt_dst.s_addr)
496 ip->ip_dst = p->ipopt_dst;
497 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
498 n = m_gethdr(M_NOWAIT, MT_DATA);
504 n->m_pkthdr.rcvif = NULL;
505 n->m_pkthdr.len += optlen;
506 m->m_len -= sizeof(struct ip);
507 m->m_data += sizeof(struct ip);
510 m->m_len = optlen + sizeof(struct ip);
511 m->m_data += max_linkhdr;
512 bcopy(ip, mtod(m, void *), sizeof(struct ip));
516 m->m_pkthdr.len += optlen;
517 bcopy(ip, mtod(m, void *), sizeof(struct ip));
519 ip = mtod(m, struct ip *);
520 bcopy(p->ipopt_list, ip + 1, optlen);
521 *phlen = sizeof(struct ip) + optlen;
522 ip->ip_v = IPVERSION;
523 ip->ip_hl = *phlen >> 2;
524 ip->ip_len = htons(ntohs(ip->ip_len) + optlen);
529 * Copy options from ip to jp, omitting those not copied during
533 ip_optcopy(struct ip *ip, struct ip *jp)
536 int opt, optlen, cnt;
538 cp = (u_char *)(ip + 1);
539 dp = (u_char *)(jp + 1);
540 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
541 for (; cnt > 0; cnt -= optlen, cp += optlen) {
543 if (opt == IPOPT_EOL)
545 if (opt == IPOPT_NOP) {
546 /* Preserve for IP mcast tunnel's LSRR alignment. */
552 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
553 ("ip_optcopy: malformed ipv4 option"));
554 optlen = cp[IPOPT_OLEN];
555 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
556 ("ip_optcopy: malformed ipv4 option"));
558 /* Bogus lengths should have been caught by ip_dooptions. */
561 if (IPOPT_COPIED(opt)) {
562 bcopy(cp, dp, optlen);
566 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
572 * Set up IP options in pcb for insertion in output packets. Store in mbuf
573 * with pointer in pcbopt, adding pseudo-option with destination address if
577 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
581 struct mbuf **pcbopt;
584 INP_WLOCK_ASSERT(inp);
586 pcbopt = &inp->inp_options;
588 /* turn off any old options */
590 (void)m_free(*pcbopt);
592 if (m == NULL || m->m_len == 0) {
594 * Only turning off any previous options.
601 if (m->m_len % sizeof(int32_t))
604 * IP first-hop destination address will be stored before actual
605 * options; move other options back and clear it when none present.
607 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
610 m->m_len += sizeof(struct in_addr);
611 cp = mtod(m, u_char *) + sizeof(struct in_addr);
612 bcopy(mtod(m, void *), cp, (unsigned)cnt);
613 bzero(mtod(m, void *), sizeof(struct in_addr));
615 for (; cnt > 0; cnt -= optlen, cp += optlen) {
616 opt = cp[IPOPT_OPTVAL];
617 if (opt == IPOPT_EOL)
619 if (opt == IPOPT_NOP)
622 if (cnt < IPOPT_OLEN + sizeof(*cp))
624 optlen = cp[IPOPT_OLEN];
625 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
636 * User process specifies route as:
640 * D must be our final destination (but we can't
641 * check that since we may not have connected yet).
642 * A is first hop destination, which doesn't appear
643 * in actual IP option, but is stored before the
646 /* XXX-BZ PRIV_NETINET_SETHDROPTS? */
647 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
649 m->m_len -= sizeof(struct in_addr);
650 cnt -= sizeof(struct in_addr);
651 optlen -= sizeof(struct in_addr);
652 cp[IPOPT_OLEN] = optlen;
654 * Move first hop before start of options.
656 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
657 sizeof(struct in_addr));
659 * Then copy rest of options back
660 * to close up the deleted entry.
662 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
664 (unsigned)cnt - (IPOPT_MINOFF - 1));
668 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
679 * Check for the presence of the IP Router Alert option [RFC2113]
680 * in the header of an IPv4 datagram.
682 * This call is not intended for use from the forwarding path; it is here
683 * so that protocol domains may check for the presence of the option.
684 * Given how FreeBSD's IPv4 stack is currently structured, the Router Alert
685 * option does not have much relevance to the implementation, though this
686 * may change in future.
687 * Router alert options SHOULD be passed if running in IPSTEALTH mode and
688 * we are not the endpoint.
689 * Length checks on individual options should already have been peformed
690 * by ip_dooptions() therefore they are folded under INVARIANTS here.
692 * Return zero if not present or options are invalid, non-zero if present.
695 ip_checkrouteralert(struct mbuf *m)
697 struct ip *ip = mtod(m, struct ip *);
699 int opt, optlen, cnt, found_ra;
702 cp = (u_char *)(ip + 1);
703 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
704 for (; cnt > 0; cnt -= optlen, cp += optlen) {
705 opt = cp[IPOPT_OPTVAL];
706 if (opt == IPOPT_EOL)
708 if (opt == IPOPT_NOP)
712 if (cnt < IPOPT_OLEN + sizeof(*cp))
715 optlen = cp[IPOPT_OLEN];
717 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
724 if (optlen != IPOPT_OFFSET + sizeof(uint16_t) ||
725 (*((uint16_t *)&cp[IPOPT_OFFSET]) != 0))