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"
38 #include <sys/param.h>
39 #include <sys/systm.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>
48 #include <sys/vimage.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/netisr.h>
57 #include <netinet/in.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 #include <security/mac/mac_framework.h>
71 static int ip_dosourceroute = 0;
72 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
73 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
75 static int ip_acceptsourceroute = 0;
76 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
77 CTLFLAG_RW, &ip_acceptsourceroute, 0,
78 "Enable accepting source routed IP packets");
80 int ip_doopts = 1; /* 0 = ignore, 1 = process, 2 = reject */
81 SYSCTL_INT(_net_inet_ip, OID_AUTO, process_options, CTLFLAG_RW,
82 &ip_doopts, 0, "Enable IP options processing ([LS]SRR, RR, TS)");
84 static void save_rte(struct mbuf *m, u_char *, struct in_addr);
87 * Do option processing on a datagram, possibly discarding it if bad options
88 * are encountered, or forwarding it if source-routed.
90 * The pass argument is used when operating in the IPSTEALTH mode to tell
91 * what options to process: [LS]SRR (pass 0) or the others (pass 1). The
92 * reason for as many as two passes is that when doing IPSTEALTH, non-routing
93 * options should be processed only if the packet is for us.
95 * Returns 1 if packet has been forwarded/freed, 0 if the packet should be
99 ip_dooptions(struct mbuf *m, int pass)
101 struct ip *ip = mtod(m, struct ip *);
103 struct in_ifaddr *ia;
104 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
105 struct in_addr *sin, dst;
107 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
109 /* Ignore or reject packets with IP options. */
112 else if (ip_doopts == 2) {
114 code = ICMP_UNREACH_FILTER_PROHIB;
119 cp = (u_char *)(ip + 1);
120 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
121 for (; cnt > 0; cnt -= optlen, cp += optlen) {
122 opt = cp[IPOPT_OPTVAL];
123 if (opt == IPOPT_EOL)
125 if (opt == IPOPT_NOP)
128 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
129 code = &cp[IPOPT_OLEN] - (u_char *)ip;
132 optlen = cp[IPOPT_OLEN];
133 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
134 code = &cp[IPOPT_OLEN] - (u_char *)ip;
144 * Source routing with record. Find interface with current
145 * destination address. If none on this machine then drop if
146 * strictly routed, or do nothing if loosely routed. Record
147 * interface address and bring up next address component. If
148 * strictly routed make sure next address is on directly
154 if (V_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) {
197 if (V_ipforwarding) {
198 char buf[16]; /* aaa.bbb.ccc.ddd\0 */
200 * Acting as a router, so generate
204 strcpy(buf, inet_ntoa(ip->ip_dst));
206 "attempted source route from %s to %s\n",
207 inet_ntoa(ip->ip_src), buf);
209 code = ICMP_UNREACH_SRCFAIL;
213 * Not acting as a router, so
219 V_ipstat.ips_cantforward++;
226 * locate outgoing interface
228 (void)memcpy(&ipaddr.sin_addr, cp + off,
229 sizeof(ipaddr.sin_addr));
231 if (opt == IPOPT_SSRR) {
232 #define INA struct in_ifaddr *
233 #define SA struct sockaddr *
234 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr)) == NULL)
235 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
237 /* XXX MRT 0 for routing */
238 ia = ip_rtaddr(ipaddr.sin_addr, M_GETFIB(m));
241 code = ICMP_UNREACH_SRCFAIL;
244 ip->ip_dst = ipaddr.sin_addr;
245 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
246 sizeof(struct in_addr));
247 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
249 * Let ip_intr's mcast routing check handle mcast pkts
251 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
256 if (V_ipstealth && pass == 0)
259 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
260 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
263 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
264 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
268 * If no space remains, ignore.
270 off--; /* 0 origin */
271 if (off > optlen - (int)sizeof(struct in_addr))
273 (void)memcpy(&ipaddr.sin_addr, &ip->ip_dst,
274 sizeof(ipaddr.sin_addr));
276 * Locate outgoing interface; if we're the
277 * destination, use the incoming interface (should be
280 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
281 (ia = ip_rtaddr(ipaddr.sin_addr, M_GETFIB(m))) == NULL) {
283 code = ICMP_UNREACH_HOST;
286 (void)memcpy(cp + off, &(IA_SIN(ia)->sin_addr),
287 sizeof(struct in_addr));
288 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
293 if (V_ipstealth && pass == 0)
296 code = cp - (u_char *)ip;
297 if (optlen < 4 || optlen > 40) {
298 code = &cp[IPOPT_OLEN] - (u_char *)ip;
301 if ((off = cp[IPOPT_OFFSET]) < 5) {
302 code = &cp[IPOPT_OLEN] - (u_char *)ip;
305 if (off > optlen - (int)sizeof(int32_t)) {
306 cp[IPOPT_OFFSET + 1] += (1 << 4);
307 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
308 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
313 off--; /* 0 origin */
314 sin = (struct in_addr *)(cp + off);
315 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
317 case IPOPT_TS_TSONLY:
320 case IPOPT_TS_TSANDADDR:
321 if (off + sizeof(n_time) +
322 sizeof(struct in_addr) > optlen) {
323 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
326 ipaddr.sin_addr = dst;
327 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
331 (void)memcpy(sin, &IA_SIN(ia)->sin_addr,
332 sizeof(struct in_addr));
333 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
334 off += sizeof(struct in_addr);
337 case IPOPT_TS_PRESPEC:
338 if (off + sizeof(n_time) +
339 sizeof(struct in_addr) > optlen) {
340 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
343 (void)memcpy(&ipaddr.sin_addr, sin,
344 sizeof(struct in_addr));
345 if (ifa_ifwithaddr((SA)&ipaddr) == NULL)
347 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
348 off += sizeof(struct in_addr);
352 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
356 (void)memcpy(cp + off, &ntime, sizeof(n_time));
357 cp[IPOPT_OFFSET] += sizeof(n_time);
360 if (forward && V_ipforwarding) {
366 icmp_error(m, type, code, 0, 0);
367 V_ipstat.ips_badoptions++;
372 * Save incoming source route for use in replies, to be picked up later by
373 * ip_srcroute if the receiver is interested.
376 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
379 struct ipopt_tag *opts;
381 opts = (struct ipopt_tag *)m_tag_get(PACKET_TAG_IPOPTIONS,
382 sizeof(struct ipopt_tag), M_NOWAIT);
386 olen = option[IPOPT_OLEN];
387 if (olen > sizeof(opts->ip_srcrt) - (1 + sizeof(dst))) {
388 m_tag_free((struct m_tag *)opts);
391 bcopy(option, opts->ip_srcrt.srcopt, olen);
392 opts->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
393 opts->ip_srcrt.dst = dst;
394 m_tag_prepend(m, (struct m_tag *)opts);
398 * Retrieve incoming source route for use in replies, in the same form used
399 * by setsockopt. The first hop is placed before the options, will be
403 ip_srcroute(struct mbuf *m0)
405 struct in_addr *p, *q;
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, reversing the path
443 * (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 level protocol in the kernel. Second
458 * argument is buffer to which options will be moved, and return value is
461 * XXX should be deleted; last arg currently ignored.
464 ip_stripoptions(struct mbuf *m, struct mbuf *mopt)
467 struct ip *ip = mtod(m, struct ip *);
471 olen = (ip->ip_hl << 2) - sizeof (struct ip);
472 opts = (caddr_t)(ip + 1);
473 i = m->m_len - (sizeof (struct ip) + olen);
474 bcopy(opts + olen, opts, (unsigned)i);
476 if (m->m_flags & M_PKTHDR)
477 m->m_pkthdr.len -= olen;
478 ip->ip_v = IPVERSION;
479 ip->ip_hl = sizeof(struct ip) >> 2;
483 * Insert IP options into preformed packet. Adjust IP destination as
484 * required for IP source routing, as indicated by a non-zero in_addr at the
485 * start of the options.
487 * XXX This routine assumes that the packet has no options in place.
490 ip_insertoptions(struct mbuf *m, struct mbuf *opt, int *phlen)
492 struct ipoption *p = mtod(opt, struct ipoption *);
494 struct ip *ip = mtod(m, struct ip *);
497 optlen = opt->m_len - sizeof(p->ipopt_dst);
498 if (optlen + ip->ip_len > IP_MAXPACKET) {
500 return (m); /* XXX should fail */
502 if (p->ipopt_dst.s_addr)
503 ip->ip_dst = p->ipopt_dst;
504 if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
505 MGETHDR(n, M_DONTWAIT, MT_DATA);
511 n->m_pkthdr.rcvif = NULL;
512 n->m_pkthdr.len += optlen;
513 m->m_len -= sizeof(struct ip);
514 m->m_data += sizeof(struct ip);
517 m->m_len = optlen + sizeof(struct ip);
518 m->m_data += max_linkhdr;
519 bcopy(ip, mtod(m, void *), sizeof(struct ip));
523 m->m_pkthdr.len += optlen;
524 bcopy(ip, mtod(m, void *), sizeof(struct ip));
526 ip = mtod(m, struct ip *);
527 bcopy(p->ipopt_list, ip + 1, optlen);
528 *phlen = sizeof(struct ip) + optlen;
529 ip->ip_v = IPVERSION;
530 ip->ip_hl = *phlen >> 2;
531 ip->ip_len += optlen;
536 * Copy options from ip to jp, omitting those not copied during
540 ip_optcopy(struct ip *ip, struct ip *jp)
543 int opt, optlen, cnt;
545 cp = (u_char *)(ip + 1);
546 dp = (u_char *)(jp + 1);
547 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
548 for (; cnt > 0; cnt -= optlen, cp += optlen) {
550 if (opt == IPOPT_EOL)
552 if (opt == IPOPT_NOP) {
553 /* Preserve for IP mcast tunnel's LSRR alignment. */
559 KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
560 ("ip_optcopy: malformed ipv4 option"));
561 optlen = cp[IPOPT_OLEN];
562 KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
563 ("ip_optcopy: malformed ipv4 option"));
565 /* Bogus lengths should have been caught by ip_dooptions. */
568 if (IPOPT_COPIED(opt)) {
569 bcopy(cp, dp, optlen);
573 for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
579 * Set up IP options in pcb for insertion in output packets. Store in mbuf
580 * with pointer in pcbopt, adding pseudo-option with destination address if
584 ip_pcbopts(struct inpcb *inp, int optname, struct mbuf *m)
588 struct mbuf **pcbopt;
591 INP_WLOCK_ASSERT(inp);
593 pcbopt = &inp->inp_options;
595 /* turn off any old options */
597 (void)m_free(*pcbopt);
599 if (m == NULL || m->m_len == 0) {
601 * Only turning off any previous options.
608 if (m->m_len % sizeof(int32_t))
611 * IP first-hop destination address will be stored before actual
612 * options; move other options back and clear it when none present.
614 if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
617 m->m_len += sizeof(struct in_addr);
618 cp = mtod(m, u_char *) + sizeof(struct in_addr);
619 bcopy(mtod(m, void *), cp, (unsigned)cnt);
620 bzero(mtod(m, void *), sizeof(struct in_addr));
622 for (; cnt > 0; cnt -= optlen, cp += optlen) {
623 opt = cp[IPOPT_OPTVAL];
624 if (opt == IPOPT_EOL)
626 if (opt == IPOPT_NOP)
629 if (cnt < IPOPT_OLEN + sizeof(*cp))
631 optlen = cp[IPOPT_OLEN];
632 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
643 * User process specifies route as:
647 * D must be our final destination (but we can't
648 * check that since we may not have connected yet).
649 * A is first hop destination, which doesn't appear
650 * in actual IP option, but is stored before the
653 /* XXX-BZ PRIV_NETINET_SETHDROPTS? */
654 if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
656 m->m_len -= sizeof(struct in_addr);
657 cnt -= sizeof(struct in_addr);
658 optlen -= sizeof(struct in_addr);
659 cp[IPOPT_OLEN] = optlen;
661 * Move first hop before start of options.
663 bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
664 sizeof(struct in_addr));
666 * Then copy rest of options back
667 * to close up the deleted entry.
669 bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
671 (unsigned)cnt - (IPOPT_MINOFF - 1));
675 if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))