2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Andre Oppermann, Internet Business Solutions AG. All right reserved.
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
34 #include "opt_ipstealth.h"
37 #include <sys/param.h>
38 #include <sys/systm.h>
40 /* #include <sys/malloc.h> */
41 #include <sys/domain.h>
42 #include <sys/protosw.h>
43 #include <sys/socket.h>
45 #include <sys/kernel.h>
46 #include <sys/syslog.h>
47 #include <sys/sysctl.h>
50 #include <net/if_types.h>
51 #include <net/if_var.h>
52 #include <net/if_dl.h>
53 #include <net/route.h>
54 #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 #include <security/mac/mac_framework.h>
70 static int ip_dosourceroute = 0;
71 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
72 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
74 static int ip_acceptsourceroute = 0;
75 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
76 CTLFLAG_RW, &ip_acceptsourceroute, 0,
77 "Enable accepting source routed IP packets");
79 int ip_doopts = 1; /* 0 = ignore, 1 = process, 2 = reject */
80 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_RW,
81 &ip_doopts, 0, "Enable IP options processing ([LS]SRR, RR, TS)");
83 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
86 * Do option processing on a datagram,
87 * possibly discarding it if bad options are encountered,
88 * or forwarding it if source-routed.
89 * The pass argument is used when operating in the IPSTEALTH
90 * mode to tell what options to process:
91 * [LS]SRR (pass 0) or the others (pass 1).
92 * The reason for as many as two passes is that when doing IPSTEALTH,
93 * non-routing options should be processed only if the packet is for us.
94 * Returns 1 if packet has been forwarded/freed,
95 * 0 if the packet should be processed further.
98 ip_dooptions(struct mbuf *m, int pass)
100 struct ip *ip = mtod(m, struct ip *);
102 struct in_ifaddr *ia;
103 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
104 struct in_addr *sin, dst;
106 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
108 /* ignore or reject packets with IP options */
111 else if (ip_doopts == 2) {
113 code = ICMP_UNREACH_FILTER_PROHIB;
118 cp = (u_char *)(ip + 1);
119 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
120 for (; cnt > 0; cnt -= optlen, cp += optlen) {
121 opt = cp[IPOPT_OPTVAL];
122 if (opt == IPOPT_EOL)
124 if (opt == IPOPT_NOP)
127 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
128 code = &cp[IPOPT_OLEN] - (u_char *)ip;
131 optlen = cp[IPOPT_OLEN];
132 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
133 code = &cp[IPOPT_OLEN] - (u_char *)ip;
143 * Source routing with record.
144 * Find interface with current destination address.
145 * If none on this machine then drop if strictly routed,
146 * or do nothing if loosely routed.
147 * Record interface address and bring up next address
148 * component. If strictly routed make sure next
149 * address is on directly accessible net.
154 if (ipstealth && pass > 0)
157 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
158 code = &cp[IPOPT_OLEN] - (u_char *)ip;
161 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
162 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
165 ipaddr.sin_addr = ip->ip_dst;
166 ia = (struct in_ifaddr *)
167 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
169 if (opt == IPOPT_SSRR) {
171 code = ICMP_UNREACH_SRCFAIL;
174 if (!ip_dosourceroute)
175 goto nosourcerouting;
177 * Loose routing, and not at next destination
178 * yet; nothing to do except forward.
182 off--; /* 0 origin */
183 if (off > optlen - (int)sizeof(struct in_addr)) {
185 * End of source route. Should be for us.
187 if (!ip_acceptsourceroute)
188 goto nosourcerouting;
189 save_rte(m, cp, ip->ip_src);
196 if (!ip_dosourceroute) {
198 char buf[16]; /* aaa.bbb.ccc.ddd\0 */
200 * Acting as a router, so generate ICMP
203 strcpy(buf, inet_ntoa(ip->ip_dst));
205 "attempted source route from %s to %s\n",
206 inet_ntoa(ip->ip_src), buf);
208 code = ICMP_UNREACH_SRCFAIL;
212 * Not acting as a router, so silently drop.
217 ipstat.ips_cantforward++;
224 * locate outgoing interface
226 (void)memcpy(&ipaddr.sin_addr, cp + off,
227 sizeof(ipaddr.sin_addr));
229 if (opt == IPOPT_SSRR) {
230 #define INA struct in_ifaddr *
231 #define SA struct sockaddr *
232 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == NULL)
233 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
235 ia = ip_rtaddr(ipaddr.sin_addr);
238 code = ICMP_UNREACH_SRCFAIL;
241 ip->ip_dst = ipaddr.sin_addr;
242 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
243 sizeof(struct in_addr));
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 (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 destination,
274 * use the incoming interface (should be same).
276 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
277 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) {
279 code = ICMP_UNREACH_HOST;
282 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
283 sizeof(struct in_addr));
284 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
289 if (ipstealth && pass == 0)
292 code = cp - (u_char *)ip;
293 if (optlen < 4 || optlen > 40) {
294 code = &cp[IPOPT_OLEN] - (u_char *)ip;
297 if ((off = cp[IPOPT_OFFSET]) < 5) {
298 code = &cp[IPOPT_OLEN] - (u_char *)ip;
301 if (off > optlen - (int)sizeof(int32_t)) {
302 cp[IPOPT_OFFSET + 1] += (1 << 4);
303 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
304 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
309 off--; /* 0 origin */
310 sin = (struct in_addr *)(cp + off);
311 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
313 case IPOPT_TS_TSONLY:
316 case IPOPT_TS_TSANDADDR:
317 if (off + sizeof(n_time) +
318 sizeof(struct in_addr) > optlen) {
319 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
322 ipaddr.sin_addr = dst;
323 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
327 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
328 sizeof(struct in_addr));
329 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
330 off += sizeof(struct in_addr);
333 case IPOPT_TS_PRESPEC:
334 if (off + sizeof(n_time) +
335 sizeof(struct in_addr) > optlen) {
336 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
339 (void)memcpy(&ipaddr.sin_addr, sin,
340 sizeof(struct in_addr));
341 if (ifa_ifwithaddr((SA)&ipaddr) == NULL)
343 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
344 off += sizeof(struct in_addr);
348 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
352 (void)memcpy(cp + off, &ntime, sizeof(n_time));
353 cp[IPOPT_OFFSET] += sizeof(n_time);
356 if (forward && ipforwarding) {
362 icmp_error(m, type, code, 0, 0);
363 ipstat.ips_badoptions++;
368 * Save incoming source route for use in replies,
369 * to be picked up later by ip_srcroute if the receiver is interested.
372 save_rte(m, option, 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,
398 * in the same form used by setsockopt.
399 * The first hop is placed before the options, will be removed later.
405 register struct in_addr *p, *q;
406 register struct mbuf *m;
407 struct ipopt_tag *opts;
409 opts = (struct ipopt_tag *)m_tag_find(m0, PACKET_TAG_IPOPTIONS, NULL);
413 if (opts->ip_nhops == 0)
415 m = m_get(M_DONTWAIT, MT_DATA);
419 #define OPTSIZ (sizeof(opts->ip_srcrt.nop) + sizeof(opts->ip_srcrt.srcopt))
421 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
422 m->m_len = opts->ip_nhops * sizeof(struct in_addr) +
423 sizeof(struct in_addr) + OPTSIZ;
426 * First save first hop for return route
428 p = &(opts->ip_srcrt.route[opts->ip_nhops - 1]);
429 *(mtod(m, struct in_addr *)) = *p--;
432 * Copy option fields and padding (nop) to mbuf.
434 opts->ip_srcrt.nop = IPOPT_NOP;
435 opts->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
436 (void)memcpy(mtod(m, caddr_t) + sizeof(struct in_addr),
437 &(opts->ip_srcrt.nop), OPTSIZ);
438 q = (struct in_addr *)(mtod(m, caddr_t) +
439 sizeof(struct in_addr) + OPTSIZ);
442 * Record return path as an IP source route,
443 * reversing the path (pointers are now aligned).
445 while (p >= opts->ip_srcrt.route) {
449 * Last hop goes to final destination.
451 *q = opts->ip_srcrt.dst;
452 m_tag_delete(m0, (struct m_tag *)opts);
457 * Strip out IP options, at higher
458 * level protocol in the kernel.
459 * Second argument is buffer to which options
460 * will be moved, and return value is their length.
461 * XXX should be deleted; last arg currently ignored.
464 ip_stripoptions(m, mopt)
465 register struct mbuf *m;
469 struct ip *ip = mtod(m, struct ip *);
470 register caddr_t opts;
473 olen = (ip->ip_hl << 2) - sizeof (struct ip);
474 opts = (caddr_t)(ip + 1);
475 i = m->m_len - (sizeof (struct ip) + olen);
476 bcopy(opts + olen, opts, (unsigned)i);
478 if (m->m_flags & M_PKTHDR)
479 m->m_pkthdr.len -= olen;
480 ip->ip_v = IPVERSION;
481 ip->ip_hl = sizeof(struct ip) >> 2;
485 * Insert IP options into preformed packet.
486 * Adjust IP destination as required for IP source routing,
487 * as indicated by a non-zero in_addr at the start of the options.
489 * XXX This routine assumes that the packet has no options in place.
492 ip_insertoptions(m, opt, phlen)
493 register struct mbuf *m;
497 register struct ipoption *p = mtod(opt, struct ipoption *);
499 register struct ip *ip = mtod(m, struct ip *);
502 optlen = opt->m_len - sizeof(p->ipopt_dst);
503 if (optlen + ip->ip_len > IP_MAXPACKET) {
505 return (m); /* XXX should fail */
507 if (p->ipopt_dst.s_addr)
508 ip->ip_dst = p->ipopt_dst;
509 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
510 MGETHDR(n, M_DONTWAIT, MT_DATA);
516 n->m_pkthdr.rcvif = NULL;
520 n->m_pkthdr.len += optlen;
521 m->m_len -= sizeof(struct ip);
522 m->m_data += sizeof(struct ip);
525 m->m_len = optlen + sizeof(struct ip);
526 m->m_data += max_linkhdr;
527 bcopy(ip, mtod(m, void *), sizeof(struct ip));
531 m->m_pkthdr.len += optlen;
532 bcopy(ip, mtod(m, void *), sizeof(struct ip));
534 ip = mtod(m, struct ip *);
535 bcopy(p->ipopt_list, ip + 1, optlen);
536 *phlen = sizeof(struct ip) + optlen;
537 ip->ip_v = IPVERSION;
538 ip->ip_hl = *phlen >> 2;
539 ip->ip_len += optlen;
544 * Copy options from ip to jp,
545 * omitting those not copied during fragmentation.
551 register u_char *cp, *dp;
552 int opt, optlen, cnt;
554 cp = (u_char *)(ip + 1);
555 dp = (u_char *)(jp + 1);
556 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
557 for (; cnt > 0; cnt -= optlen, cp += optlen) {
559 if (opt == IPOPT_EOL)
561 if (opt == IPOPT_NOP) {
562 /* Preserve for IP mcast tunnel's LSRR alignment. */
568 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
569 ("ip_optcopy: malformed ipv4 option"));
570 optlen = cp[IPOPT_OLEN];
571 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
572 ("ip_optcopy: malformed ipv4 option"));
574 /* bogus lengths should have been caught by ip_dooptions */
577 if (IPOPT_COPIED(opt)) {
578 bcopy(cp, dp, optlen);
582 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
588 * Set up IP options in pcb for insertion in output packets.
589 * Store in mbuf with pointer in pcbopt, adding pseudo-option
590 * with destination address if source routed.
593 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
595 register int cnt, optlen;
597 struct mbuf **pcbopt;
600 INP_LOCK_ASSERT(inp);
602 pcbopt = &inp->inp_options;
604 /* turn off any old options */
606 (void)m_free(*pcbopt);
608 if (m == NULL || m->m_len == 0) {
610 * Only turning off any previous options.
617 if (m->m_len % sizeof(int32_t))
620 * IP first-hop destination address will be stored before
621 * actual options; move other options back
622 * and clear it when none present.
624 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
627 m->m_len += sizeof(struct in_addr);
628 cp = mtod(m, u_char *) + sizeof(struct in_addr);
629 bcopy(mtod(m, void *), cp, (unsigned)cnt);
630 bzero(mtod(m, void *), sizeof(struct in_addr));
632 for (; cnt > 0; cnt -= optlen, cp += optlen) {
633 opt = cp[IPOPT_OPTVAL];
634 if (opt == IPOPT_EOL)
636 if (opt == IPOPT_NOP)
639 if (cnt < IPOPT_OLEN + sizeof(*cp))
641 optlen = cp[IPOPT_OLEN];
642 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
653 * user process specifies route as:
655 * D must be our final destination (but we can't
656 * check that since we may not have connected yet).
657 * A is first hop destination, which doesn't appear in
658 * actual IP option, but is stored before the options.
660 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
662 m->m_len -= sizeof(struct in_addr);
663 cnt -= sizeof(struct in_addr);
664 optlen -= sizeof(struct in_addr);
665 cp[IPOPT_OLEN] = optlen;
667 * Move first hop before start of options.
669 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
670 sizeof(struct in_addr));
672 * Then copy rest of options back
673 * to close up the deleted entry.
675 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
677 (unsigned)cnt - (IPOPT_MINOFF - 1));
681 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))