1 /* 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. 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. 20 * 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 31 * SUCH DAMAGE. 32 * 33 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94 34 * $FreeBSD: src/sys/netinet/ip_icmp.c,v 1.39.2.19 2003/01/24 05:11:34 sam Exp $ 35 * $DragonFly: src/sys/netinet/ip_icmp.c,v 1.14 2004/12/21 02:54:15 hsu Exp $ 36 */ 37 38 #include "opt_ipsec.h" 39 40 #include <sys/param.h> 41 #include <sys/systm.h> 42 #include <sys/mbuf.h> 43 #include <sys/protosw.h> 44 #include <sys/socket.h> 45 #include <sys/time.h> 46 #include <sys/kernel.h> 47 #include <sys/sysctl.h> 48 #include <sys/in_cksum.h> 49 50 #include <machine/stdarg.h> 51 52 #include <net/if.h> 53 #include <net/if_types.h> 54 #include <net/route.h> 55 56 #define _IP_VHL 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/ip_icmp.h> 62 #include <netinet/ip_var.h> 63 #include <netinet/icmp_var.h> 64 65 #ifdef IPSEC 66 #include <netinet6/ipsec.h> 67 #include <netproto/key/key.h> 68 #endif 69 70 #ifdef FAST_IPSEC 71 #include <netproto/ipsec/ipsec.h> 72 #include <netproto/ipsec/key.h> 73 #define IPSEC 74 #endif 75 76 /* 77 * ICMP routines: error generation, receive packet processing, and 78 * routines to turnaround packets back to the originator, and 79 * host table maintenance routines. 80 */ 81 82 struct icmpstat icmpstat; 83 SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW, 84 &icmpstat, icmpstat, ""); 85 86 static int icmpmaskrepl = 0; 87 SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW, 88 &icmpmaskrepl, 0, ""); 89 90 static int drop_redirect = 0; 91 SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW, 92 &drop_redirect, 0, ""); 93 94 static int log_redirect = 0; 95 SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW, 96 &log_redirect, 0, ""); 97 98 #ifdef ICMP_BANDLIM 99 100 /* 101 * ICMP error-response bandwidth limiting sysctl. If not enabled, sysctl 102 * variable content is -1 and read-only. 103 */ 104 105 static int icmplim = 200; 106 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW, 107 &icmplim, 0, ""); 108 #else 109 110 static int icmplim = -1; 111 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RD, 112 &icmplim, 0, ""); 113 114 #endif 115 116 static int icmplim_output = 1; 117 SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW, 118 &icmplim_output, 0, ""); 119 120 /* 121 * ICMP broadcast echo sysctl 122 */ 123 124 static int icmpbmcastecho = 0; 125 SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW, 126 &icmpbmcastecho, 0, ""); 127 128 129 #ifdef ICMPPRINTFS 130 int icmpprintfs = 0; 131 #endif 132 133 static void icmp_reflect (struct mbuf *); 134 static void icmp_send (struct mbuf *, struct mbuf *, struct route *); 135 static int ip_next_mtu (int, int); 136 137 extern struct protosw inetsw[]; 138 139 /* 140 * Generate an error packet of type error 141 * in response to bad packet ip. 142 */ 143 void 144 icmp_error(n, type, code, dest, destifp) 145 struct mbuf *n; 146 int type, code; 147 n_long dest; 148 struct ifnet *destifp; 149 { 150 struct ip *oip = mtod(n, struct ip *), *nip; 151 unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2; 152 struct icmp *icp; 153 struct mbuf *m; 154 unsigned icmplen; 155 156 #ifdef ICMPPRINTFS 157 if (icmpprintfs) 158 printf("icmp_error(%p, %x, %d)\n", oip, type, code); 159 #endif 160 if (type != ICMP_REDIRECT) 161 icmpstat.icps_error++; 162 /* 163 * Don't send error if the original packet was encrypted. 164 * Don't send error if not the first fragment of message. 165 * Don't error if the old packet protocol was ICMP 166 * error message, only known informational types. 167 */ 168 if (n->m_flags & M_DECRYPTED) 169 goto freeit; 170 if (oip->ip_off &~ (IP_MF|IP_DF)) 171 goto freeit; 172 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && 173 n->m_len >= oiplen + ICMP_MINLEN && 174 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) { 175 icmpstat.icps_oldicmp++; 176 goto freeit; 177 } 178 /* Don't send error in response to a multicast or broadcast packet */ 179 if (n->m_flags & (M_BCAST|M_MCAST)) 180 goto freeit; 181 /* 182 * First, formulate icmp message 183 */ 184 m = m_gethdr(MB_DONTWAIT, MT_HEADER); 185 if (m == NULL) 186 goto freeit; 187 icmplen = min(oiplen + 8, oip->ip_len); 188 if (icmplen < sizeof(struct ip)) 189 panic("icmp_error: bad length"); 190 m->m_len = icmplen + ICMP_MINLEN; 191 MH_ALIGN(m, m->m_len); 192 icp = mtod(m, struct icmp *); 193 if ((u_int)type > ICMP_MAXTYPE) 194 panic("icmp_error"); 195 icmpstat.icps_outhist[type]++; 196 icp->icmp_type = type; 197 if (type == ICMP_REDIRECT) 198 icp->icmp_gwaddr.s_addr = dest; 199 else { 200 icp->icmp_void = 0; 201 /* 202 * The following assignments assume an overlay with the 203 * zeroed icmp_void field. 204 */ 205 if (type == ICMP_PARAMPROB) { 206 icp->icmp_pptr = code; 207 code = 0; 208 } else if (type == ICMP_UNREACH && 209 code == ICMP_UNREACH_NEEDFRAG && destifp) { 210 icp->icmp_nextmtu = htons(destifp->if_mtu); 211 } 212 } 213 214 icp->icmp_code = code; 215 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip); 216 nip = &icp->icmp_ip; 217 218 /* 219 * Convert fields to network representation. 220 */ 221 nip->ip_len = htons(nip->ip_len); 222 nip->ip_off = htons(nip->ip_off); 223 224 /* 225 * Now, copy old ip header (without options) 226 * in front of icmp message. 227 */ 228 if (m->m_data - sizeof(struct ip) < m->m_pktdat) 229 panic("icmp len"); 230 m->m_data -= sizeof(struct ip); 231 m->m_len += sizeof(struct ip); 232 m->m_pkthdr.len = m->m_len; 233 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif; 234 nip = mtod(m, struct ip *); 235 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip)); 236 nip->ip_len = m->m_len; 237 nip->ip_vhl = IP_VHL_BORING; 238 nip->ip_p = IPPROTO_ICMP; 239 nip->ip_tos = 0; 240 m->m_pkthdr.pf_flags |= n->m_pkthdr.pf_flags & PF_MBUF_GENERATED; 241 icmp_reflect(m); 242 243 freeit: 244 m_freem(n); 245 } 246 247 static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET }; 248 static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET }; 249 static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET }; 250 251 /* 252 * Process a received ICMP message. 253 */ 254 void 255 icmp_input(struct mbuf *m, ...) 256 { 257 int hlen; 258 struct icmp *icp; 259 struct ip *ip = mtod(m, struct ip *); 260 int icmplen = ip->ip_len; 261 int i; 262 struct in_ifaddr *ia; 263 void (*ctlfunc) (int, struct sockaddr *, void *); 264 int code, off, proto; 265 __va_list ap; 266 267 __va_start(ap, m); 268 off = __va_arg(ap, int); 269 proto = __va_arg(ap, int); 270 __va_end(ap); 271 272 hlen = off; 273 274 /* 275 * Locate icmp structure in mbuf, and check 276 * that not corrupted and of at least minimum length. 277 */ 278 #ifdef ICMPPRINTFS 279 if (icmpprintfs) { 280 char buf[4 * sizeof "123"]; 281 strcpy(buf, inet_ntoa(ip->ip_src)); 282 printf("icmp_input from %s to %s, len %d\n", 283 buf, inet_ntoa(ip->ip_dst), icmplen); 284 } 285 #endif 286 if (icmplen < ICMP_MINLEN) { 287 icmpstat.icps_tooshort++; 288 goto freeit; 289 } 290 i = hlen + min(icmplen, ICMP_ADVLENMIN); 291 if (m->m_len < i && (m = m_pullup(m, i)) == 0) { 292 icmpstat.icps_tooshort++; 293 return; 294 } 295 ip = mtod(m, struct ip *); 296 m->m_len -= hlen; 297 m->m_data += hlen; 298 icp = mtod(m, struct icmp *); 299 if (in_cksum(m, icmplen)) { 300 icmpstat.icps_checksum++; 301 goto freeit; 302 } 303 m->m_len += hlen; 304 m->m_data -= hlen; 305 306 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) { 307 /* 308 * Deliver very specific ICMP type only. 309 */ 310 switch (icp->icmp_type) { 311 case ICMP_UNREACH: 312 case ICMP_TIMXCEED: 313 break; 314 default: 315 goto freeit; 316 } 317 } 318 319 #ifdef ICMPPRINTFS 320 if (icmpprintfs) 321 printf("icmp_input, type %d code %d\n", icp->icmp_type, 322 icp->icmp_code); 323 #endif 324 325 /* 326 * Message type specific processing. 327 */ 328 if (icp->icmp_type > ICMP_MAXTYPE) 329 goto raw; 330 icmpstat.icps_inhist[icp->icmp_type]++; 331 code = icp->icmp_code; 332 switch (icp->icmp_type) { 333 334 case ICMP_UNREACH: 335 switch (code) { 336 case ICMP_UNREACH_NET: 337 case ICMP_UNREACH_HOST: 338 case ICMP_UNREACH_SRCFAIL: 339 case ICMP_UNREACH_NET_UNKNOWN: 340 case ICMP_UNREACH_HOST_UNKNOWN: 341 case ICMP_UNREACH_ISOLATED: 342 case ICMP_UNREACH_TOSNET: 343 case ICMP_UNREACH_TOSHOST: 344 case ICMP_UNREACH_HOST_PRECEDENCE: 345 case ICMP_UNREACH_PRECEDENCE_CUTOFF: 346 code = PRC_UNREACH_NET; 347 break; 348 349 case ICMP_UNREACH_NEEDFRAG: 350 code = PRC_MSGSIZE; 351 break; 352 353 /* 354 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9. 355 * Treat subcodes 2,3 as immediate RST 356 */ 357 case ICMP_UNREACH_PROTOCOL: 358 case ICMP_UNREACH_PORT: 359 code = PRC_UNREACH_PORT; 360 break; 361 362 case ICMP_UNREACH_NET_PROHIB: 363 case ICMP_UNREACH_HOST_PROHIB: 364 case ICMP_UNREACH_FILTER_PROHIB: 365 code = PRC_UNREACH_ADMIN_PROHIB; 366 break; 367 368 default: 369 goto badcode; 370 } 371 goto deliver; 372 373 case ICMP_TIMXCEED: 374 if (code > 1) 375 goto badcode; 376 code += PRC_TIMXCEED_INTRANS; 377 goto deliver; 378 379 case ICMP_PARAMPROB: 380 if (code > 1) 381 goto badcode; 382 code = PRC_PARAMPROB; 383 goto deliver; 384 385 case ICMP_SOURCEQUENCH: 386 if (code) 387 goto badcode; 388 code = PRC_QUENCH; 389 deliver: 390 /* 391 * Problem with datagram; advise higher level routines. 392 */ 393 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 394 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { 395 icmpstat.icps_badlen++; 396 goto freeit; 397 } 398 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len); 399 /* Discard ICMP's in response to multicast packets */ 400 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr))) 401 goto badcode; 402 #ifdef ICMPPRINTFS 403 if (icmpprintfs) 404 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p); 405 #endif 406 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 407 #if 1 408 /* 409 * MTU discovery: 410 * If we got a needfrag and there is a host route to the 411 * original destination, and the MTU is not locked, then 412 * set the MTU in the route to the suggested new value 413 * (if given) and then notify as usual. The ULPs will 414 * notice that the MTU has changed and adapt accordingly. 415 * If no new MTU was suggested, then we guess a new one 416 * less than the current value. If the new MTU is 417 * unreasonably small (arbitrarily set at 296), then 418 * we reset the MTU to the interface value and enable the 419 * lock bit, indicating that we are no longer doing MTU 420 * discovery. 421 */ 422 if (code == PRC_MSGSIZE) { 423 struct rtentry *rt; 424 int mtu; 425 426 rt = rtlookup((struct sockaddr *)&icmpsrc, 0, 427 RTF_CLONING | RTF_PRCLONING); 428 if (rt && (rt->rt_flags & RTF_HOST) && 429 !(rt->rt_rmx.rmx_locks & RTV_MTU)) { 430 mtu = ntohs(icp->icmp_nextmtu); 431 if (!mtu) 432 mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu, 433 1); 434 #ifdef DEBUG_MTUDISC 435 printf("MTU for %s reduced to %d\n", 436 inet_ntoa(icmpsrc.sin_addr), mtu); 437 #endif 438 if (mtu < 296) { 439 /* rt->rt_rmx.rmx_mtu = 440 rt->rt_ifp->if_mtu; */ 441 rt->rt_rmx.rmx_locks |= RTV_MTU; 442 } else if (rt->rt_rmx.rmx_mtu > mtu) { 443 rt->rt_rmx.rmx_mtu = mtu; 444 } 445 } 446 if (rt) 447 --rt->rt_refcnt; 448 } 449 450 #endif 451 /* 452 * XXX if the packet contains [IPv4 AH TCP], we can't make a 453 * notification to TCP layer. 454 */ 455 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput; 456 if (ctlfunc) 457 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc, 458 (void *)&icp->icmp_ip); 459 break; 460 461 badcode: 462 icmpstat.icps_badcode++; 463 break; 464 465 case ICMP_ECHO: 466 if (!icmpbmcastecho 467 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 468 icmpstat.icps_bmcastecho++; 469 break; 470 } 471 icp->icmp_type = ICMP_ECHOREPLY; 472 #ifdef ICMP_BANDLIM 473 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0) 474 goto freeit; 475 else 476 #endif 477 goto reflect; 478 479 case ICMP_TSTAMP: 480 if (!icmpbmcastecho 481 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 482 icmpstat.icps_bmcasttstamp++; 483 break; 484 } 485 if (icmplen < ICMP_TSLEN) { 486 icmpstat.icps_badlen++; 487 break; 488 } 489 icp->icmp_type = ICMP_TSTAMPREPLY; 490 icp->icmp_rtime = iptime(); 491 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ 492 #ifdef ICMP_BANDLIM 493 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0) 494 goto freeit; 495 else 496 #endif 497 goto reflect; 498 499 case ICMP_MASKREQ: 500 if (icmpmaskrepl == 0) 501 break; 502 /* 503 * We are not able to respond with all ones broadcast 504 * unless we receive it over a point-to-point interface. 505 */ 506 if (icmplen < ICMP_MASKLEN) 507 break; 508 switch (ip->ip_dst.s_addr) { 509 510 case INADDR_BROADCAST: 511 case INADDR_ANY: 512 icmpdst.sin_addr = ip->ip_src; 513 break; 514 515 default: 516 icmpdst.sin_addr = ip->ip_dst; 517 } 518 ia = (struct in_ifaddr *)ifaof_ifpforaddr( 519 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif); 520 if (ia == 0) 521 break; 522 if (ia->ia_ifp == 0) 523 break; 524 icp->icmp_type = ICMP_MASKREPLY; 525 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; 526 if (ip->ip_src.s_addr == 0) { 527 if (ia->ia_ifp->if_flags & IFF_BROADCAST) 528 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr; 529 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) 530 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr; 531 } 532 reflect: 533 ip->ip_len += hlen; /* since ip_input deducts this */ 534 icmpstat.icps_reflect++; 535 icmpstat.icps_outhist[icp->icmp_type]++; 536 icmp_reflect(m); 537 return; 538 539 case ICMP_REDIRECT: 540 if (log_redirect) { 541 u_long src, dst, gw; 542 543 src = ntohl(ip->ip_src.s_addr); 544 dst = ntohl(icp->icmp_ip.ip_dst.s_addr); 545 gw = ntohl(icp->icmp_gwaddr.s_addr); 546 printf("icmp redirect from %d.%d.%d.%d: " 547 "%d.%d.%d.%d => %d.%d.%d.%d\n", 548 (int)(src >> 24), (int)((src >> 16) & 0xff), 549 (int)((src >> 8) & 0xff), (int)(src & 0xff), 550 (int)(dst >> 24), (int)((dst >> 16) & 0xff), 551 (int)((dst >> 8) & 0xff), (int)(dst & 0xff), 552 (int)(gw >> 24), (int)((gw >> 16) & 0xff), 553 (int)((gw >> 8) & 0xff), (int)(gw & 0xff)); 554 } 555 if (drop_redirect) 556 break; 557 if (code > 3) 558 goto badcode; 559 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 560 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { 561 icmpstat.icps_badlen++; 562 break; 563 } 564 /* 565 * Short circuit routing redirects to force 566 * immediate change in the kernel's routing 567 * tables. The message is also handed to anyone 568 * listening on a raw socket (e.g. the routing 569 * daemon for use in updating its tables). 570 */ 571 icmpgw.sin_addr = ip->ip_src; 572 icmpdst.sin_addr = icp->icmp_gwaddr; 573 #ifdef ICMPPRINTFS 574 if (icmpprintfs) { 575 char buf[4 * sizeof "123"]; 576 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst)); 577 578 printf("redirect dst %s to %s\n", 579 buf, inet_ntoa(icp->icmp_gwaddr)); 580 } 581 #endif 582 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 583 rtredirect((struct sockaddr *)&icmpsrc, 584 (struct sockaddr *)&icmpdst, 585 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST, 586 (struct sockaddr *)&icmpgw, (struct rtentry **)0); 587 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc); 588 #ifdef IPSEC 589 key_sa_routechange((struct sockaddr *)&icmpsrc); 590 #endif 591 break; 592 593 /* 594 * No kernel processing for the following; 595 * just fall through to send to raw listener. 596 */ 597 case ICMP_ECHOREPLY: 598 case ICMP_ROUTERADVERT: 599 case ICMP_ROUTERSOLICIT: 600 case ICMP_TSTAMPREPLY: 601 case ICMP_IREQREPLY: 602 case ICMP_MASKREPLY: 603 default: 604 break; 605 } 606 607 raw: 608 rip_input(m, off, proto); 609 return; 610 611 freeit: 612 m_freem(m); 613 } 614 615 /* 616 * Reflect the ip packet back to the source 617 */ 618 static void 619 icmp_reflect(m) 620 struct mbuf *m; 621 { 622 struct ip *ip = mtod(m, struct ip *); 623 struct ifaddr *ifa; 624 struct in_ifaddr *ia; 625 struct in_addr t; 626 struct mbuf *opts = 0; 627 int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); 628 struct route *ro = NULL, rt; 629 630 if (!in_canforward(ip->ip_src) && 631 ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) != 632 (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { 633 m_freem(m); /* Bad return address */ 634 icmpstat.icps_badaddr++; 635 goto done; /* Ip_output() will check for broadcast */ 636 } 637 t = ip->ip_dst; 638 ip->ip_dst = ip->ip_src; 639 ro = &rt; 640 bzero(ro, sizeof(*ro)); 641 /* 642 * If the incoming packet was addressed directly to us, 643 * use dst as the src for the reply. Otherwise (broadcast 644 * or anonymous), use the address which corresponds 645 * to the incoming interface. 646 */ 647 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash) 648 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr) 649 goto match; 650 if (m->m_pkthdr.rcvif != NULL && 651 m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 652 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) { 653 if (ifa->ifa_addr->sa_family != AF_INET) 654 continue; 655 ia = ifatoia(ifa); 656 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 657 t.s_addr) 658 goto match; 659 } 660 } 661 ia = ip_rtaddr(ip->ip_dst, ro); 662 /* We need a route to do anything useful. */ 663 if (ia == NULL) { 664 m_freem(m); 665 icmpstat.icps_noroute++; 666 goto done; 667 } 668 match: 669 t = IA_SIN(ia)->sin_addr; 670 ip->ip_src = t; 671 ip->ip_ttl = ip_defttl; 672 673 if (optlen > 0) { 674 u_char *cp; 675 int opt, cnt; 676 u_int len; 677 678 /* 679 * Retrieve any source routing from the incoming packet; 680 * add on any record-route or timestamp options. 681 */ 682 cp = (u_char *) (ip + 1); 683 if ((opts = ip_srcroute()) == 0 && 684 (opts = m_gethdr(MB_DONTWAIT, MT_HEADER))) { 685 opts->m_len = sizeof(struct in_addr); 686 mtod(opts, struct in_addr *)->s_addr = 0; 687 } 688 if (opts) { 689 #ifdef ICMPPRINTFS 690 if (icmpprintfs) 691 printf("icmp_reflect optlen %d rt %d => ", 692 optlen, opts->m_len); 693 #endif 694 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { 695 opt = cp[IPOPT_OPTVAL]; 696 if (opt == IPOPT_EOL) 697 break; 698 if (opt == IPOPT_NOP) 699 len = 1; 700 else { 701 if (cnt < IPOPT_OLEN + sizeof(*cp)) 702 break; 703 len = cp[IPOPT_OLEN]; 704 if (len < IPOPT_OLEN + sizeof(*cp) || 705 len > cnt) 706 break; 707 } 708 /* 709 * Should check for overflow, but it 710 * "can't happen". 711 */ 712 if (opt == IPOPT_RR || opt == IPOPT_TS || 713 opt == IPOPT_SECURITY) { 714 bcopy((caddr_t)cp, 715 mtod(opts, caddr_t) + opts->m_len, 716 len); 717 opts->m_len += len; 718 } 719 } 720 /* Terminate & pad, if necessary */ 721 cnt = opts->m_len % 4; 722 if (cnt) { 723 for (; cnt < 4; cnt++) { 724 *(mtod(opts, caddr_t) + opts->m_len) = 725 IPOPT_EOL; 726 opts->m_len++; 727 } 728 } 729 #ifdef ICMPPRINTFS 730 if (icmpprintfs) 731 printf("%d\n", opts->m_len); 732 #endif 733 } 734 /* 735 * Now strip out original options by copying rest of first 736 * mbuf's data back, and adjust the IP length. 737 */ 738 ip->ip_len -= optlen; 739 ip->ip_vhl = IP_VHL_BORING; 740 m->m_len -= optlen; 741 if (m->m_flags & M_PKTHDR) 742 m->m_pkthdr.len -= optlen; 743 optlen += sizeof(struct ip); 744 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1), 745 (unsigned)(m->m_len - sizeof(struct ip))); 746 } 747 m->m_pkthdr.pf_flags &= PF_MBUF_GENERATED; 748 m->m_flags &= ~(M_BCAST|M_MCAST); 749 icmp_send(m, opts, ro); 750 done: 751 if (opts) 752 m_free(opts); 753 if (ro && ro->ro_rt) 754 RTFREE(ro->ro_rt); 755 } 756 757 /* 758 * Send an icmp packet back to the ip level, 759 * after supplying a checksum. 760 */ 761 static void 762 icmp_send(m, opts, rt) 763 struct mbuf *m; 764 struct mbuf *opts; 765 struct route *rt; 766 { 767 struct ip *ip = mtod(m, struct ip *); 768 int hlen; 769 struct icmp *icp; 770 771 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 772 m->m_data += hlen; 773 m->m_len -= hlen; 774 icp = mtod(m, struct icmp *); 775 icp->icmp_cksum = 0; 776 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen); 777 m->m_data -= hlen; 778 m->m_len += hlen; 779 m->m_pkthdr.rcvif = (struct ifnet *)0; 780 #ifdef ICMPPRINTFS 781 if (icmpprintfs) { 782 char buf[4 * sizeof "123"]; 783 strcpy(buf, inet_ntoa(ip->ip_dst)); 784 printf("icmp_send dst %s src %s\n", 785 buf, inet_ntoa(ip->ip_src)); 786 } 787 #endif 788 ip_output(m, opts, rt, 0, NULL, NULL); 789 } 790 791 n_time 792 iptime() 793 { 794 struct timeval atv; 795 u_long t; 796 797 getmicrotime(&atv); 798 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; 799 return (htonl(t)); 800 } 801 802 #if 1 803 /* 804 * Return the next larger or smaller MTU plateau (table from RFC 1191) 805 * given current value MTU. If DIR is less than zero, a larger plateau 806 * is returned; otherwise, a smaller value is returned. 807 */ 808 static int 809 ip_next_mtu(mtu, dir) 810 int mtu; 811 int dir; 812 { 813 static int mtutab[] = { 814 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296, 815 68, 0 816 }; 817 int i; 818 819 for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) { 820 if (mtu >= mtutab[i]) 821 break; 822 } 823 824 if (dir < 0) { 825 if (i == 0) { 826 return 0; 827 } else { 828 return mtutab[i - 1]; 829 } 830 } else { 831 if (mtutab[i] == 0) { 832 return 0; 833 } else if(mtu > mtutab[i]) { 834 return mtutab[i]; 835 } else { 836 return mtutab[i + 1]; 837 } 838 } 839 } 840 #endif 841 842 #ifdef ICMP_BANDLIM 843 /* 844 * badport_bandlim() - check for ICMP bandwidth limit 845 * 846 * Return 0 if it is ok to send an ICMP error response, -1 if we have 847 * hit our bandwidth limit and it is not ok. 848 * 849 * If icmplim is <= 0, the feature is disabled and 0 is returned. 850 * 851 * For now we separate the TCP and UDP subsystems w/ different 'which' 852 * values. We may eventually remove this separation (and simplify the 853 * code further). 854 * 855 * Note that the printing of the error message is delayed so we can 856 * properly print the icmp error rate that the system was trying to do 857 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing 858 * the 'final' error, but it doesn't make sense to solve the printing 859 * delay with more complex code. 860 */ 861 int 862 badport_bandlim(int which) 863 { 864 static int lticks[BANDLIM_MAX + 1]; 865 static int lpackets[BANDLIM_MAX + 1]; 866 int dticks; 867 const char *bandlimittype[] = { 868 "Limiting icmp unreach response", 869 "Limiting icmp ping response", 870 "Limiting icmp tstamp response", 871 "Limiting closed port RST response", 872 "Limiting open port RST response" 873 }; 874 875 /* 876 * Return ok status if feature disabled or argument out of 877 * ranage. 878 */ 879 880 if (icmplim <= 0 || which > BANDLIM_MAX || which < 0) 881 return(0); 882 dticks = ticks - lticks[which]; 883 884 /* 885 * reset stats when cumulative dt exceeds one second. 886 */ 887 888 if ((unsigned int)dticks > hz) { 889 if (lpackets[which] > icmplim && icmplim_output) { 890 printf("%s from %d to %d packets per second\n", 891 bandlimittype[which], 892 lpackets[which], 893 icmplim 894 ); 895 } 896 lticks[which] = ticks; 897 lpackets[which] = 0; 898 } 899 900 /* 901 * bump packet count 902 */ 903 904 if (++lpackets[which] > icmplim) { 905 return(-1); 906 } 907 return(0); 908 } 909 #endif 910