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 * 3. 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_fib.h>
58 #include <netinet/in_systm.h>
59 #include <netinet/in_var.h>
60 #include <netinet/ip.h>
61 #include <netinet/in_pcb.h>
62 #include <netinet/ip_var.h>
63 #include <netinet/ip_options.h>
64 #include <netinet/ip_icmp.h>
65 #include <machine/in_cksum.h>
67 #include <sys/socketvar.h>
69 static VNET_DEFINE(int, ip_dosourceroute);
70 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute,
71 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_dosourceroute), 0,
72 "Enable forwarding source routed IP packets");
73 #define V_ip_dosourceroute VNET(ip_dosourceroute)
75 static VNET_DEFINE(int, ip_acceptsourceroute);
76 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
77 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_acceptsourceroute), 0,
78 "Enable accepting source routed IP packets");
79 #define V_ip_acceptsourceroute VNET(ip_acceptsourceroute)
81 VNET_DEFINE(int, ip_doopts) = 1; /* 0 = ignore, 1 = process, 2 = reject */
82 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_VNET | CTLFLAG_RW,
83 &VNET_NAME(ip_doopts), 0, "Enable IP options processing ([LS]SRR, RR, TS)");
85 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
88 * Do option processing on a datagram, possibly discarding it if bad options
89 * are encountered, or forwarding it if source-routed.
91 * The pass argument is used when operating in the IPSTEALTH mode to tell
92 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
93 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
94 * options should be processed only if the packet is for us.
96 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
100 ip_dooptions(struct mbuf *m, int pass)
102 struct ip *ip = mtod(m, struct ip *);
104 struct in_ifaddr *ia;
105 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
106 struct in_addr *sin, dst;
108 struct nhop4_extended nh_ext;
109 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
111 /* Ignore or reject packets with IP options. */
112 if (V_ip_doopts == 0)
114 else if (V_ip_doopts == 2) {
116 code = ICMP_UNREACH_FILTER_PROHIB;
121 cp = (u_char *)(ip + 1);
122 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
123 for (; cnt > 0; cnt -= optlen, cp += optlen) {
124 opt = cp[IPOPT_OPTVAL];
125 if (opt == IPOPT_EOL)
127 if (opt == IPOPT_NOP)
130 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
131 code = &cp[IPOPT_OLEN] - (u_char *)ip;
134 optlen = cp[IPOPT_OLEN];
135 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
136 code = &cp[IPOPT_OLEN] - (u_char *)ip;
146 * Source routing with record. Find interface with current
147 * destination address. If none on this machine then drop if
148 * strictly routed, or do nothing if loosely routed. Record
149 * interface address and bring up next address component. If
150 * strictly routed make sure next address is on directly
156 if (V_ipstealth && pass > 0)
159 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
160 code = &cp[IPOPT_OLEN] - (u_char *)ip;
163 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
164 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
167 ipaddr.sin_addr = ip->ip_dst;
168 if (ifa_ifwithaddr_check((struct sockaddr *)&ipaddr)
170 if (opt == IPOPT_SSRR) {
172 code = ICMP_UNREACH_SRCFAIL;
175 if (!V_ip_dosourceroute)
176 goto nosourcerouting;
178 * Loose routing, and not at next destination
179 * yet; nothing to do except forward.
183 off--; /* 0 origin */
184 if (off > optlen - (int)sizeof(struct in_addr)) {
186 * End of source route. Should be for us.
188 if (!V_ip_acceptsourceroute)
189 goto nosourcerouting;
190 save_rte(m, cp, ip->ip_src);
197 if (!V_ip_dosourceroute) {
198 if (V_ipforwarding) {
199 char srcbuf[INET_ADDRSTRLEN];
200 char dstbuf[INET_ADDRSTRLEN];
203 * Acting as a router, so generate
208 "attempted source route from %s "
210 inet_ntoa_r(ip->ip_src, srcbuf),
211 inet_ntoa_r(ip->ip_dst, dstbuf));
213 code = ICMP_UNREACH_SRCFAIL;
217 * Not acting as a router, so
223 IPSTAT_INC(ips_cantforward);
230 * locate outgoing interface
232 (void)memcpy(&ipaddr.sin_addr, cp + off,
233 sizeof(ipaddr.sin_addr));
236 code = ICMP_UNREACH_SRCFAIL;
238 if (opt == IPOPT_SSRR) {
239 #define INA struct in_ifaddr *
240 #define SA struct sockaddr *
241 ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr,
244 ia = (INA)ifa_ifwithnet((SA)&ipaddr, 0,
249 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
250 sizeof(struct in_addr));
251 ifa_free(&ia->ia_ifa);
253 /* XXX MRT 0 for routing */
254 if (fib4_lookup_nh_ext(M_GETFIB(m),
255 ipaddr.sin_addr, 0, 0, &nh_ext) != 0)
258 memcpy(cp + off, &nh_ext.nh_src,
259 sizeof(struct in_addr));
262 ip->ip_dst = ipaddr.sin_addr;
263 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
265 * Let ip_intr's mcast routing check handle mcast pkts
267 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
272 if (V_ipstealth && pass == 0)
275 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
276 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
279 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
280 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
284 * If no space remains, ignore.
286 off--; /* 0 origin */
287 if (off > optlen - (int)sizeof(struct in_addr))
289 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
290 sizeof(ipaddr.sin_addr));
292 * Locate outgoing interface; if we're the
293 * destination, use the incoming interface (should be
296 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) != NULL) {
297 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
298 sizeof(struct in_addr));
299 ifa_free(&ia->ia_ifa);
300 } else if (fib4_lookup_nh_ext(M_GETFIB(m),
301 ipaddr.sin_addr, 0, 0, &nh_ext) == 0) {
302 memcpy(cp + off, &nh_ext.nh_src,
303 sizeof(struct in_addr));
306 code = ICMP_UNREACH_HOST;
309 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
314 if (V_ipstealth && pass == 0)
317 code = cp - (u_char *)ip;
318 if (optlen < 4 || optlen > 40) {
319 code = &cp[IPOPT_OLEN] - (u_char *)ip;
322 if ((off = cp[IPOPT_OFFSET]) < 5) {
323 code = &cp[IPOPT_OLEN] - (u_char *)ip;
326 if (off > optlen - (int)sizeof(int32_t)) {
327 cp[IPOPT_OFFSET + 1] += (1 << 4);
328 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
329 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
334 off--; /* 0 origin */
335 sin = (struct in_addr *)(cp + off);
336 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
338 case IPOPT_TS_TSONLY:
341 case IPOPT_TS_TSANDADDR:
342 if (off + sizeof(uint32_t) +
343 sizeof(struct in_addr) > optlen) {
344 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
347 ipaddr.sin_addr = dst;
348 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
352 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
353 sizeof(struct in_addr));
354 ifa_free(&ia->ia_ifa);
355 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
356 off += sizeof(struct in_addr);
359 case IPOPT_TS_PRESPEC:
360 if (off + sizeof(uint32_t) +
361 sizeof(struct in_addr) > optlen) {
362 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
365 (void)memcpy(&ipaddr.sin_addr, sin,
366 sizeof(struct in_addr));
367 if (ifa_ifwithaddr_check((SA)&ipaddr) == 0)
369 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
370 off += sizeof(struct in_addr);
374 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
378 (void)memcpy(cp + off, &ntime, sizeof(uint32_t));
379 cp[IPOPT_OFFSET] += sizeof(uint32_t);
382 if (forward && V_ipforwarding) {
388 icmp_error(m, type, code, 0, 0);
389 IPSTAT_INC(ips_badoptions);
394 * Save incoming source route for use in replies, to be picked up later by
395 * ip_srcroute if the receiver is interested.
398 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
401 struct ipopt_tag *opts;
403 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
404 sizeof(struct ipopt_tag), M_NOWAIT);
408 olen = option[IPOPT_OLEN];
409 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
410 m_tag_free((struct m_tag *)opts);
413 bcopy(option, opts->ip_srcrt.srcopt, olen);
414 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
415 opts->ip_srcrt.dst = dst;
416 m_tag_prepend(m, (struct m_tag *)opts);
420 * Retrieve incoming source route for use in replies, in the same form used
421 * by setsockopt. The first hop is placed before the options, will be
425 ip_srcroute(struct mbuf *m0)
427 struct in_addr *p, *q;
429 struct ipopt_tag *opts;
431 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
435 if (opts->ip_nhops == 0)
437 m = m_get(M_NOWAIT, MT_DATA);
441 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
443 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
444 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
445 sizeof(struct in_addr) + OPTSIZ;
448 * First, save first hop for return route.
450 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
451 *(mtod(m, struct in_addr *)) = *p--;
454 * Copy option fields and padding (nop) to mbuf.
456 opts->ip_srcrt.nop = IPOPT_NOP;
457 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
458 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
459 &(opts->ip_srcrt.nop), OPTSIZ);
460 q = (struct in_addr *)(mtod(m, caddr_t) +
461 sizeof(struct in_addr) + OPTSIZ);
464 * Record return path as an IP source route, reversing the path
465 * (pointers are now aligned).
467 while (p >= opts->ip_srcrt.route) {
471 * Last hop goes to final destination.
473 *q = opts->ip_srcrt.dst;
474 m_tag_delete(m0, (struct m_tag *)opts);
479 * Strip out IP options, at higher level protocol in the kernel.
482 ip_stripoptions(struct mbuf *m)
484 struct ip *ip = mtod(m, struct ip *);
487 olen = (ip->ip_hl << 2) - sizeof(struct ip);
489 if (m->m_flags & M_PKTHDR)
490 m->m_pkthdr.len -= olen;
491 ip->ip_len = htons(ntohs(ip->ip_len) - olen);
492 ip->ip_hl = sizeof(struct ip) >> 2;
494 bcopy((char *)ip + sizeof(struct ip) + olen, (ip + 1),
495 (size_t )(m->m_len - sizeof(struct ip)));
499 * Insert IP options into preformed packet. Adjust IP destination as
500 * required for IP source routing, as indicated by a non-zero in_addr at the
501 * start of the options.
503 * XXX This routine assumes that the packet has no options in place.
506 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
508 struct ipoption *p = mtod(opt, struct ipoption *);
510 struct ip *ip = mtod(m, struct ip *);
513 optlen = opt->m_len - sizeof(p->ipopt_dst);
514 if (optlen + ntohs(ip->ip_len) > IP_MAXPACKET) {
516 return (m); /* XXX should fail */
518 if (p->ipopt_dst.s_addr)
519 ip->ip_dst = p->ipopt_dst;
520 if (!M_WRITABLE(m) || M_LEADINGSPACE(m) < optlen) {
521 n = m_gethdr(M_NOWAIT, MT_DATA);
527 n->m_pkthdr.rcvif = NULL;
528 n->m_pkthdr.len += optlen;
529 m->m_len -= sizeof(struct ip);
530 m->m_data += sizeof(struct ip);
533 m->m_len = optlen + sizeof(struct ip);
534 m->m_data += max_linkhdr;
535 bcopy(ip, mtod(m, void *), sizeof(struct ip));
539 m->m_pkthdr.len += optlen;
540 bcopy(ip, mtod(m, void *), sizeof(struct ip));
542 ip = mtod(m, struct ip *);
543 bcopy(p->ipopt_list, ip + 1, optlen);
544 *phlen = sizeof(struct ip) + optlen;
545 ip->ip_v = IPVERSION;
546 ip->ip_hl = *phlen >> 2;
547 ip->ip_len = htons(ntohs(ip->ip_len) + optlen);
552 * Copy options from ip to jp, omitting those not copied during
556 ip_optcopy(struct ip *ip, struct ip *jp)
559 int opt, optlen, cnt;
561 cp = (u_char *)(ip + 1);
562 dp = (u_char *)(jp + 1);
563 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
564 for (; cnt > 0; cnt -= optlen, cp += optlen) {
566 if (opt == IPOPT_EOL)
568 if (opt == IPOPT_NOP) {
569 /* Preserve for IP mcast tunnel's LSRR alignment. */
575 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
576 ("ip_optcopy: malformed ipv4 option"));
577 optlen = cp[IPOPT_OLEN];
578 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
579 ("ip_optcopy: malformed ipv4 option"));
581 /* Bogus lengths should have been caught by ip_dooptions. */
584 if (IPOPT_COPIED(opt)) {
585 bcopy(cp, dp, optlen);
589 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
595 * Set up IP options in pcb for insertion in output packets. Store in mbuf
596 * with pointer in pcbopt, adding pseudo-option with destination address if
600 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
604 struct mbuf **pcbopt;
607 INP_WLOCK_ASSERT(inp);
609 pcbopt = &inp->inp_options;
611 /* turn off any old options */
613 (void)m_free(*pcbopt);
615 if (m == NULL || m->m_len == 0) {
617 * Only turning off any previous options.
624 if (m->m_len % sizeof(int32_t))
627 * IP first-hop destination address will be stored before actual
628 * options; move other options back and clear it when none present.
630 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
633 m->m_len += sizeof(struct in_addr);
634 cp = mtod(m, u_char *) + sizeof(struct in_addr);
635 bcopy(mtod(m, void *), cp, (unsigned)cnt);
636 bzero(mtod(m, void *), sizeof(struct in_addr));
638 for (; cnt > 0; cnt -= optlen, cp += optlen) {
639 opt = cp[IPOPT_OPTVAL];
640 if (opt == IPOPT_EOL)
642 if (opt == IPOPT_NOP)
645 if (cnt < IPOPT_OLEN + sizeof(*cp))
647 optlen = cp[IPOPT_OLEN];
648 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
659 * User process specifies route as:
663 * D must be our final destination (but we can't
664 * check that since we may not have connected yet).
665 * A is first hop destination, which doesn't appear
666 * in actual IP option, but is stored before the
669 /* XXX-BZ PRIV_NETINET_SETHDROPTS? */
670 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
672 m->m_len -= sizeof(struct in_addr);
673 cnt -= sizeof(struct in_addr);
674 optlen -= sizeof(struct in_addr);
675 cp[IPOPT_OLEN] = optlen;
677 * Move first hop before start of options.
679 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
680 sizeof(struct in_addr));
682 * Then copy rest of options back
683 * to close up the deleted entry.
685 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
687 (unsigned)cnt - (IPOPT_MINOFF - 1));
691 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
702 * Check for the presence of the IP Router Alert option [RFC2113]
703 * in the header of an IPv4 datagram.
705 * This call is not intended for use from the forwarding path; it is here
706 * so that protocol domains may check for the presence of the option.
707 * Given how FreeBSD's IPv4 stack is currently structured, the Router Alert
708 * option does not have much relevance to the implementation, though this
709 * may change in future.
710 * Router alert options SHOULD be passed if running in IPSTEALTH mode and
711 * we are not the endpoint.
712 * Length checks on individual options should already have been performed
713 * by ip_dooptions() therefore they are folded under INVARIANTS here.
715 * Return zero if not present or options are invalid, non-zero if present.
718 ip_checkrouteralert(struct mbuf *m)
720 struct ip *ip = mtod(m, struct ip *);
722 int opt, optlen, cnt, found_ra;
725 cp = (u_char *)(ip + 1);
726 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
727 for (; cnt > 0; cnt -= optlen, cp += optlen) {
728 opt = cp[IPOPT_OPTVAL];
729 if (opt == IPOPT_EOL)
731 if (opt == IPOPT_NOP)
735 if (cnt < IPOPT_OLEN + sizeof(*cp))
738 optlen = cp[IPOPT_OLEN];
740 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
747 if (optlen != IPOPT_OFFSET + sizeof(uint16_t) ||
748 (*((uint16_t *)&cp[IPOPT_OFFSET]) != 0))