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 #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/route/nhop.h>
54 #include <net/netisr.h>
57 #include <netinet/in.h>
58 #include <netinet/in_fib.h>
59 #include <netinet/in_systm.h>
60 #include <netinet/in_var.h>
61 #include <netinet/ip.h>
62 #include <netinet/in_pcb.h>
63 #include <netinet/ip_var.h>
64 #include <netinet/ip_options.h>
65 #include <netinet/ip_icmp.h>
66 #include <machine/in_cksum.h>
68 #include <sys/socketvar.h>
70 VNET_DEFINE_STATIC(int, ip_dosourceroute);
71 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute,
72 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_dosourceroute), 0,
73 "Enable forwarding source routed IP packets");
74 #define V_ip_dosourceroute VNET(ip_dosourceroute)
76 VNET_DEFINE_STATIC(int, ip_acceptsourceroute);
77 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
78 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_acceptsourceroute), 0,
79 "Enable accepting source routed IP packets");
80 #define V_ip_acceptsourceroute VNET(ip_acceptsourceroute)
82 VNET_DEFINE(int, ip_doopts) = 1; /* 0 = ignore, 1 = process, 2 = reject */
83 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_VNET | CTLFLAG_RW,
84 &VNET_NAME(ip_doopts), 0, "Enable IP options processing ([LS]SRR, RR, TS)");
86 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
89 * Do option processing on a datagram, possibly discarding it if bad options
90 * are encountered, or forwarding it if source-routed.
92 * The pass argument is used when operating in the IPSTEALTH mode to tell
93 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
94 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
95 * options should be processed only if the packet is for us.
97 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
101 ip_dooptions(struct mbuf *m, int pass)
103 struct ip *ip = mtod(m, struct ip *);
105 struct in_ifaddr *ia;
106 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
107 struct in_addr *sin, dst;
109 struct nhop_object *nh;
110 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
114 /* Ignore or reject packets with IP options. */
115 if (V_ip_doopts == 0)
117 else if (V_ip_doopts == 2) {
119 code = ICMP_UNREACH_FILTER_PROHIB;
124 cp = (u_char *)(ip + 1);
125 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
126 for (; cnt > 0; cnt -= optlen, cp += optlen) {
127 opt = cp[IPOPT_OPTVAL];
128 if (opt == IPOPT_EOL)
130 if (opt == IPOPT_NOP)
133 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
134 code = &cp[IPOPT_OLEN] - (u_char *)ip;
137 optlen = cp[IPOPT_OLEN];
138 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
139 code = &cp[IPOPT_OLEN] - (u_char *)ip;
148 * Source routing with record. Find interface with current
149 * destination address. If none on this machine then drop if
150 * strictly routed, or do nothing if loosely routed. Record
151 * interface address and bring up next address component. If
152 * strictly routed make sure next address is on directly
158 if (V_ipstealth && pass > 0)
161 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
162 code = &cp[IPOPT_OLEN] - (u_char *)ip;
165 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
166 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
169 ipaddr.sin_addr = ip->ip_dst;
170 if (ifa_ifwithaddr_check((struct sockaddr *)&ipaddr)
172 if (opt == IPOPT_SSRR) {
174 code = ICMP_UNREACH_SRCFAIL;
177 if (!V_ip_dosourceroute)
178 goto nosourcerouting;
180 * Loose routing, and not at next destination
181 * yet; nothing to do except forward.
185 off--; /* 0 origin */
186 if (off > optlen - (int)sizeof(struct in_addr)) {
188 * End of source route. Should be for us.
190 if (!V_ip_acceptsourceroute)
191 goto nosourcerouting;
192 save_rte(m, cp, ip->ip_src);
199 if (!V_ip_dosourceroute) {
200 if (V_ipforwarding) {
201 char srcbuf[INET_ADDRSTRLEN];
202 char dstbuf[INET_ADDRSTRLEN];
205 * Acting as a router, so generate
210 "attempted source route from %s "
212 inet_ntoa_r(ip->ip_src, srcbuf),
213 inet_ntoa_r(ip->ip_dst, dstbuf));
215 code = ICMP_UNREACH_SRCFAIL;
219 * Not acting as a router, so
225 IPSTAT_INC(ips_cantforward);
232 * locate outgoing interface
234 (void)memcpy(&ipaddr.sin_addr, cp + off,
235 sizeof(ipaddr.sin_addr));
238 code = ICMP_UNREACH_SRCFAIL;
240 if (opt == IPOPT_SSRR) {
241 #define INA struct in_ifaddr *
242 #define SA struct sockaddr *
243 ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr,
246 ia = (INA)ifa_ifwithnet((SA)&ipaddr, 0,
251 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
252 sizeof(struct in_addr));
254 /* XXX MRT 0 for routing */
255 nh = fib4_lookup(M_GETFIB(m), ipaddr.sin_addr,
260 memcpy(cp + off, &(IA_SIN(nh->nh_ifa)->sin_addr),
261 sizeof(struct in_addr));
264 ip->ip_dst = ipaddr.sin_addr;
265 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
267 * Let ip_intr's mcast routing check handle mcast pkts
269 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
274 if (V_ipstealth && pass == 0)
277 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
278 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
281 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
282 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
286 * If no space remains, ignore.
288 off--; /* 0 origin */
289 if (off > optlen - (int)sizeof(struct in_addr))
291 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
292 sizeof(ipaddr.sin_addr));
294 * Locate outgoing interface; if we're the
295 * destination, use the incoming interface (should be
298 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) != NULL) {
299 memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
300 sizeof(struct in_addr));
301 } else if ((nh = fib4_lookup(M_GETFIB(m),
302 ipaddr.sin_addr, 0, NHR_NONE, 0)) != NULL) {
303 memcpy(cp + off, &(IA_SIN(nh->nh_ifa)->sin_addr),
304 sizeof(struct in_addr));
307 code = ICMP_UNREACH_HOST;
310 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
315 if (V_ipstealth && pass == 0)
318 code = cp - (u_char *)ip;
319 if (optlen < 4 || optlen > 40) {
320 code = &cp[IPOPT_OLEN] - (u_char *)ip;
323 if ((off = cp[IPOPT_OFFSET]) < 5) {
324 code = &cp[IPOPT_OLEN] - (u_char *)ip;
327 if (off > optlen - (int)sizeof(int32_t)) {
328 cp[IPOPT_OFFSET + 1] += (1 << 4);
329 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
330 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
335 off--; /* 0 origin */
336 sin = (struct in_addr *)(cp + off);
337 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 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
355 off += sizeof(struct in_addr);
358 case IPOPT_TS_PRESPEC:
359 if (off + sizeof(uint32_t) +
360 sizeof(struct in_addr) > optlen) {
361 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
364 (void)memcpy(&ipaddr.sin_addr, sin,
365 sizeof(struct in_addr));
366 if (ifa_ifwithaddr_check((SA)&ipaddr) == 0)
368 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
369 off += sizeof(struct in_addr);
373 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
377 (void)memcpy(cp + off, &ntime, sizeof(uint32_t));
378 cp[IPOPT_OFFSET] += sizeof(uint32_t);
381 if (forward && V_ipforwarding) {
387 icmp_error(m, type, code, 0, 0);
388 IPSTAT_INC(ips_badoptions);
393 * Save incoming source route for use in replies, to be picked up later by
394 * ip_srcroute if the receiver is interested.
397 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
400 struct ipopt_tag *opts;
402 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
403 sizeof(struct ipopt_tag), M_NOWAIT);
407 olen = option[IPOPT_OLEN];
408 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
409 m_tag_free((struct m_tag *)opts);
412 bcopy(option, opts->ip_srcrt.srcopt, olen);
413 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
414 opts->ip_srcrt.dst = dst;
415 m_tag_prepend(m, (struct m_tag *)opts);
419 * Retrieve incoming source route for use in replies, in the same form used
420 * by setsockopt. The first hop is placed before the options, will be
424 ip_srcroute(struct mbuf *m0)
426 struct in_addr *p, *q;
428 struct ipopt_tag *opts;
430 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
434 if (opts->ip_nhops == 0)
436 m = m_get(M_NOWAIT, MT_DATA);
440 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
442 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
443 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
444 sizeof(struct in_addr) + OPTSIZ;
447 * First, save first hop for return route.
449 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
450 *(mtod(m, struct in_addr *)) = *p--;
453 * Copy option fields and padding (nop) to mbuf.
455 opts->ip_srcrt.nop = IPOPT_NOP;
456 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
457 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
458 &(opts->ip_srcrt.nop), OPTSIZ);
459 q = (struct in_addr *)(mtod(m, caddr_t) +
460 sizeof(struct in_addr) + OPTSIZ);
463 * Record return path as an IP source route, reversing the path
464 * (pointers are now aligned).
466 while (p >= opts->ip_srcrt.route) {
470 * Last hop goes to final destination.
472 *q = opts->ip_srcrt.dst;
473 m_tag_delete(m0, (struct m_tag *)opts);
478 * Strip out IP options, at higher level protocol in the kernel.
481 ip_stripoptions(struct mbuf *m)
483 struct ip *ip = mtod(m, struct ip *);
486 olen = (ip->ip_hl << 2) - sizeof(struct ip);
488 if (m->m_flags & M_PKTHDR)
489 m->m_pkthdr.len -= olen;
490 ip->ip_len = htons(ntohs(ip->ip_len) - olen);
491 ip->ip_hl = sizeof(struct ip) >> 2;
493 bcopy((char *)ip + sizeof(struct ip) + olen, (ip + 1),
494 (size_t )(m->m_len - sizeof(struct ip)));
498 * Insert IP options into preformed packet. Adjust IP destination as
499 * required for IP source routing, as indicated by a non-zero in_addr at the
500 * start of the options.
502 * XXX This routine assumes that the packet has no options in place.
505 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
507 struct ipoption *p = mtod(opt, struct ipoption *);
509 struct ip *ip = mtod(m, struct ip *);
512 optlen = opt->m_len - sizeof(p->ipopt_dst);
513 if (optlen + ntohs(ip->ip_len) > IP_MAXPACKET) {
515 return (m); /* XXX should fail */
517 if (p->ipopt_dst.s_addr)
518 ip->ip_dst = p->ipopt_dst;
519 if (!M_WRITABLE(m) || M_LEADINGSPACE(m) < optlen) {
520 n = m_gethdr(M_NOWAIT, MT_DATA);
526 n->m_pkthdr.rcvif = NULL;
527 n->m_pkthdr.len += optlen;
528 m->m_len -= sizeof(struct ip);
529 m->m_data += sizeof(struct ip);
532 m->m_len = optlen + sizeof(struct ip);
533 m->m_data += max_linkhdr;
534 bcopy(ip, mtod(m, void *), sizeof(struct ip));
538 m->m_pkthdr.len += optlen;
539 bcopy(ip, mtod(m, void *), sizeof(struct ip));
541 ip = mtod(m, struct ip *);
542 bcopy(p->ipopt_list, ip + 1, optlen);
543 *phlen = sizeof(struct ip) + optlen;
544 ip->ip_v = IPVERSION;
545 ip->ip_hl = *phlen >> 2;
546 ip->ip_len = htons(ntohs(ip->ip_len) + optlen);
551 * Copy options from ip to jp, omitting those not copied during
555 ip_optcopy(struct ip *ip, struct ip *jp)
558 int opt, optlen, cnt;
560 cp = (u_char *)(ip + 1);
561 dp = (u_char *)(jp + 1);
562 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
563 for (; cnt > 0; cnt -= optlen, cp += optlen) {
565 if (opt == IPOPT_EOL)
567 if (opt == IPOPT_NOP) {
568 /* Preserve for IP mcast tunnel's LSRR alignment. */
574 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
575 ("ip_optcopy: malformed ipv4 option"));
576 optlen = cp[IPOPT_OLEN];
577 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
578 ("ip_optcopy: malformed ipv4 option"));
580 /* Bogus lengths should have been caught by ip_dooptions. */
583 if (IPOPT_COPIED(opt)) {
584 bcopy(cp, dp, optlen);
588 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
594 * Set up IP options in pcb for insertion in output packets. Store in mbuf
595 * with pointer in pcbopt, adding pseudo-option with destination address if
599 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
603 struct mbuf **pcbopt;
606 INP_WLOCK_ASSERT(inp);
608 pcbopt = &inp->inp_options;
610 /* turn off any old options */
612 (void)m_free(*pcbopt);
614 if (m == NULL || m->m_len == 0) {
616 * Only turning off any previous options.
623 if (m->m_len % sizeof(int32_t))
626 * IP first-hop destination address will be stored before actual
627 * options; move other options back and clear it when none present.
629 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
632 m->m_len += sizeof(struct in_addr);
633 cp = mtod(m, u_char *) + sizeof(struct in_addr);
634 bcopy(mtod(m, void *), cp, (unsigned)cnt);
635 bzero(mtod(m, void *), sizeof(struct in_addr));
637 for (; cnt > 0; cnt -= optlen, cp += optlen) {
638 opt = cp[IPOPT_OPTVAL];
639 if (opt == IPOPT_EOL)
641 if (opt == IPOPT_NOP)
644 if (cnt < IPOPT_OLEN + sizeof(*cp))
646 optlen = cp[IPOPT_OLEN];
647 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
657 * User process specifies route as:
661 * D must be our final destination (but we can't
662 * check that since we may not have connected yet).
663 * A is first hop destination, which doesn't appear
664 * in actual IP option, but is stored before the
667 /* XXX-BZ PRIV_NETINET_SETHDROPTS? */
668 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
670 m->m_len -= sizeof(struct in_addr);
671 cnt -= sizeof(struct in_addr);
672 optlen -= sizeof(struct in_addr);
673 cp[IPOPT_OLEN] = optlen;
675 * Move first hop before start of options.
677 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
678 sizeof(struct in_addr));
680 * Then copy rest of options back
681 * to close up the deleted entry.
683 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
685 (unsigned)cnt - (IPOPT_MINOFF - 1));
689 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
700 * Check for the presence of the IP Router Alert option [RFC2113]
701 * in the header of an IPv4 datagram.
703 * This call is not intended for use from the forwarding path; it is here
704 * so that protocol domains may check for the presence of the option.
705 * Given how FreeBSD's IPv4 stack is currently structured, the Router Alert
706 * option does not have much relevance to the implementation, though this
707 * may change in future.
708 * Router alert options SHOULD be passed if running in IPSTEALTH mode and
709 * we are not the endpoint.
710 * Length checks on individual options should already have been performed
711 * by ip_dooptions() therefore they are folded under INVARIANTS here.
713 * Return zero if not present or options are invalid, non-zero if present.
716 ip_checkrouteralert(struct mbuf *m)
718 struct ip *ip = mtod(m, struct ip *);
720 int opt, optlen, cnt, found_ra;
723 cp = (u_char *)(ip + 1);
724 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
725 for (; cnt > 0; cnt -= optlen, cp += optlen) {
726 opt = cp[IPOPT_OPTVAL];
727 if (opt == IPOPT_EOL)
729 if (opt == IPOPT_NOP)
733 if (cnt < IPOPT_OLEN + sizeof(*cp))
736 optlen = cp[IPOPT_OLEN];
738 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
745 if (optlen != IPOPT_OFFSET + sizeof(uint16_t) ||
746 (*((uint16_t *)&cp[IPOPT_OFFSET]) != 0))