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. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94 30 * $FreeBSD: src/sys/netinet/ip_icmp.c,v 1.39.2.19 2003/01/24 05:11:34 sam Exp $ 31 */ 32 33 #include "opt_ipsec.h" 34 35 #include <sys/param.h> 36 #include <sys/systm.h> 37 #include <sys/mbuf.h> 38 #include <sys/protosw.h> 39 #include <sys/socket.h> 40 #include <sys/socketops.h> 41 #include <sys/time.h> 42 #include <sys/kernel.h> 43 #include <sys/sysctl.h> 44 #include <sys/in_cksum.h> 45 46 #include <machine/stdarg.h> 47 48 #include <net/if.h> 49 #include <net/if_types.h> 50 #include <net/route.h> 51 52 #define _IP_VHL 53 #include <netinet/in.h> 54 #include <netinet/in_systm.h> 55 #include <netinet/in_var.h> 56 #include <netinet/ip.h> 57 #include <netinet/ip_icmp.h> 58 #include <netinet/ip_var.h> 59 #include <netinet/icmp_var.h> 60 61 #ifdef IPSEC 62 #include <netinet6/ipsec.h> 63 #include <netproto/key/key.h> 64 #endif 65 66 #ifdef FAST_IPSEC 67 #include <netproto/ipsec/ipsec.h> 68 #include <netproto/ipsec/key.h> 69 #define IPSEC 70 #endif 71 72 /* 73 * ICMP routines: error generation, receive packet processing, and 74 * routines to turnaround packets back to the originator, and 75 * host table maintenance routines. 76 */ 77 78 struct icmpstat icmpstat; 79 SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW, 80 &icmpstat, icmpstat, "ICMP statistics"); 81 82 static int icmpmaskrepl = 0; 83 SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW, 84 &icmpmaskrepl, 0, "Allow replies to netmask requests"); 85 86 static int drop_redirect = 0; 87 SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW, 88 &drop_redirect, 0, "Ignore ICMP redirects"); 89 90 static int log_redirect = 0; 91 SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW, 92 &log_redirect, 0, "Enable output about ICMP redirects"); 93 94 static int discard_sourcequench = 1; 95 SYSCTL_INT(_net_inet_icmp, OID_AUTO, discard_sourcequench, CTLFLAG_RW, 96 &discard_sourcequench, 0, "Discard ICMP Source Quench"); 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, "ICMP bandwidth limit"); 108 #else 109 110 static int icmplim = -1; 111 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RD, 112 &icmplim, 0, "ICMP bandwidth limit"); 113 114 #endif 115 116 static int icmplim_output = 0; 117 SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW, 118 &icmplim_output, 0, "Enable output about ICMP bandwidth limits"); 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 static char icmp_reply_src[IFNAMSIZ]; 129 SYSCTL_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW, 130 icmp_reply_src, IFNAMSIZ, "icmp reply source for non-local packets."); 131 132 static int icmp_rfi; 133 SYSCTL_INT(_net_inet_icmp, OID_AUTO, reply_from_interface, CTLFLAG_RW, 134 &icmp_rfi, 0, "ICMP reply from incoming interface for " 135 "non-local packets"); 136 137 #ifdef ICMPPRINTFS 138 int icmpprintfs = 0; 139 #endif 140 141 static void icmp_reflect (struct mbuf *); 142 static void icmp_send (struct mbuf *, struct mbuf *, struct route *); 143 144 extern struct protosw inetsw[]; 145 146 /* 147 * Generate an error packet of type error 148 * in response to bad packet ip. 149 */ 150 void 151 icmp_error(struct mbuf *n, int type, int code, n_long dest, int destmtu) 152 { 153 struct ip *oip = mtod(n, struct ip *), *nip; 154 unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2; 155 struct icmp *icp; 156 struct mbuf *m; 157 unsigned icmplen; 158 159 #ifdef ICMPPRINTFS 160 if (icmpprintfs) 161 kprintf("icmp_error(%p, %d, %d)\n", oip, type, code); 162 #endif 163 if (type != ICMP_REDIRECT) 164 icmpstat.icps_error++; 165 /* 166 * Don't send error if the original packet was encrypted. 167 * Don't send error if not the first fragment of message. 168 * Don't error if the old packet protocol was ICMP 169 * error message, only known informational types. 170 */ 171 if (n->m_flags & M_DECRYPTED) 172 goto freeit; 173 if (oip->ip_off &~ (IP_MF|IP_DF)) 174 goto freeit; 175 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && 176 n->m_len >= oiplen + ICMP_MINLEN && 177 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) { 178 icmpstat.icps_oldicmp++; 179 goto freeit; 180 } 181 /* Don't send error in response to a multicast or broadcast packet */ 182 if (n->m_flags & (M_BCAST|M_MCAST)) 183 goto freeit; 184 /* 185 * First, formulate icmp message 186 */ 187 m = m_gethdr(MB_DONTWAIT, MT_HEADER); 188 if (m == NULL) 189 goto freeit; 190 icmplen = min(oiplen + 8, oip->ip_len); 191 if (icmplen < sizeof(struct ip)) 192 panic("icmp_error: bad length"); 193 m->m_len = icmplen + ICMP_MINLEN; 194 MH_ALIGN(m, m->m_len); 195 icp = mtod(m, struct icmp *); 196 if ((u_int)type > ICMP_MAXTYPE) 197 panic("icmp_error"); 198 icmpstat.icps_outhist[type]++; 199 icp->icmp_type = type; 200 if (type == ICMP_REDIRECT) 201 icp->icmp_gwaddr.s_addr = dest; 202 else { 203 icp->icmp_void = 0; 204 /* 205 * The following assignments assume an overlay with the 206 * zeroed icmp_void field. 207 */ 208 if (type == ICMP_PARAMPROB) { 209 icp->icmp_pptr = code; 210 code = 0; 211 } else if (type == ICMP_UNREACH && 212 code == ICMP_UNREACH_NEEDFRAG && destmtu) { 213 icp->icmp_nextmtu = htons(destmtu); 214 } 215 } 216 217 icp->icmp_code = code; 218 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip); 219 nip = &icp->icmp_ip; 220 221 /* 222 * Convert fields to network representation. 223 */ 224 nip->ip_len = htons(nip->ip_len); 225 nip->ip_off = htons(nip->ip_off); 226 227 /* 228 * Now, copy old ip header (without options) 229 * in front of icmp message. 230 */ 231 if (m->m_data - sizeof(struct ip) < m->m_pktdat) 232 panic("icmp len"); 233 m->m_data -= sizeof(struct ip); 234 m->m_len += sizeof(struct ip); 235 m->m_pkthdr.len = m->m_len; 236 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif; 237 nip = mtod(m, struct ip *); 238 bcopy(oip, nip, sizeof(struct ip)); 239 nip->ip_len = m->m_len; 240 nip->ip_vhl = IP_VHL_BORING; 241 nip->ip_p = IPPROTO_ICMP; 242 nip->ip_tos = 0; 243 m->m_pkthdr.fw_flags |= n->m_pkthdr.fw_flags & FW_MBUF_GENERATED; 244 icmp_reflect(m); 245 246 freeit: 247 m_freem(n); 248 } 249 250 /* 251 * Process a received ICMP message. 252 */ 253 int 254 icmp_input(struct mbuf **mp, int *offp, int proto) 255 { 256 struct sockaddr_in icmpsrc = { sizeof(struct sockaddr_in), AF_INET }; 257 struct sockaddr_in icmpdst = { sizeof(struct sockaddr_in), AF_INET }; 258 struct sockaddr_in icmpgw = { sizeof(struct sockaddr_in), AF_INET }; 259 struct icmp *icp; 260 struct in_ifaddr *ia; 261 struct mbuf *m = *mp; 262 struct ip *ip = mtod(m, struct ip *); 263 int icmplen = ip->ip_len; 264 int i, hlen; 265 int code; 266 267 *mp = NULL; 268 hlen = *offp; 269 270 /* 271 * Locate icmp structure in mbuf, and check 272 * that not corrupted and of at least minimum length. 273 */ 274 #ifdef ICMPPRINTFS 275 if (icmpprintfs) { 276 char buf[sizeof "aaa.bbb.ccc.ddd"]; 277 278 strcpy(buf, inet_ntoa(ip->ip_src)); 279 kprintf("icmp_input from %s to %s, len %d\n", 280 buf, inet_ntoa(ip->ip_dst), icmplen); 281 } 282 #endif 283 if (icmplen < ICMP_MINLEN) { 284 icmpstat.icps_tooshort++; 285 goto freeit; 286 } 287 i = hlen + min(icmplen, ICMP_ADVLENMIN); 288 if (m->m_len < i && (m = m_pullup(m, i)) == NULL) { 289 icmpstat.icps_tooshort++; 290 return(IPPROTO_DONE); 291 } 292 ip = mtod(m, struct ip *); 293 m->m_len -= hlen; 294 m->m_data += hlen; 295 icp = mtod(m, struct icmp *); 296 if (in_cksum(m, icmplen)) { 297 icmpstat.icps_checksum++; 298 goto freeit; 299 } 300 m->m_len += hlen; 301 m->m_data -= hlen; 302 303 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) { 304 /* 305 * Deliver very specific ICMP type only. 306 */ 307 switch (icp->icmp_type) { 308 case ICMP_UNREACH: 309 case ICMP_TIMXCEED: 310 break; 311 default: 312 goto freeit; 313 } 314 } 315 316 #ifdef ICMPPRINTFS 317 if (icmpprintfs) 318 kprintf("icmp_input, type %d code %d\n", icp->icmp_type, 319 icp->icmp_code); 320 #endif 321 322 /* 323 * Message type specific processing. 324 */ 325 if (icp->icmp_type > ICMP_MAXTYPE) 326 goto raw; 327 icmpstat.icps_inhist[icp->icmp_type]++; 328 code = icp->icmp_code; 329 switch (icp->icmp_type) { 330 331 case ICMP_UNREACH: 332 switch (code) { 333 case ICMP_UNREACH_NET: 334 case ICMP_UNREACH_HOST: 335 case ICMP_UNREACH_SRCFAIL: 336 case ICMP_UNREACH_NET_UNKNOWN: 337 case ICMP_UNREACH_HOST_UNKNOWN: 338 case ICMP_UNREACH_ISOLATED: 339 case ICMP_UNREACH_TOSNET: 340 case ICMP_UNREACH_TOSHOST: 341 case ICMP_UNREACH_HOST_PRECEDENCE: 342 case ICMP_UNREACH_PRECEDENCE_CUTOFF: 343 code = PRC_UNREACH_NET; 344 break; 345 346 case ICMP_UNREACH_NEEDFRAG: 347 code = PRC_MSGSIZE; 348 break; 349 350 /* 351 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9. 352 * Treat subcodes 2,3 as immediate RST 353 */ 354 case ICMP_UNREACH_PROTOCOL: 355 case ICMP_UNREACH_PORT: 356 code = PRC_UNREACH_PORT; 357 break; 358 359 case ICMP_UNREACH_NET_PROHIB: 360 case ICMP_UNREACH_HOST_PROHIB: 361 case ICMP_UNREACH_FILTER_PROHIB: 362 code = PRC_UNREACH_ADMIN_PROHIB; 363 break; 364 365 default: 366 goto badcode; 367 } 368 goto deliver; 369 370 case ICMP_TIMXCEED: 371 if (code > 1) 372 goto badcode; 373 code += PRC_TIMXCEED_INTRANS; 374 goto deliver; 375 376 case ICMP_PARAMPROB: 377 if (code > 1) 378 goto badcode; 379 code = PRC_PARAMPROB; 380 goto deliver; 381 382 case ICMP_SOURCEQUENCH: 383 if (code) 384 goto badcode; 385 if (discard_sourcequench) 386 break; 387 code = PRC_QUENCH; 388 deliver: 389 /* 390 * Problem with datagram; advise higher level routines. 391 */ 392 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 393 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { 394 icmpstat.icps_badlen++; 395 goto freeit; 396 } 397 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len); 398 /* Discard ICMP's in response to multicast packets */ 399 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr))) 400 goto badcode; 401 #ifdef ICMPPRINTFS 402 if (icmpprintfs) 403 kprintf("deliver to protocol %d\n", icp->icmp_ip.ip_p); 404 #endif 405 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 406 #if 1 407 /* 408 * MTU discovery: 409 * If we got a needfrag and there is a host route to the 410 * original destination, and the MTU is not locked, then 411 * set the MTU in the route to the suggested new value 412 * (if given) and then notify as usual. The ULPs will 413 * notice that the MTU has changed and adapt accordingly. 414 * If no new MTU was suggested, then we guess a new one 415 * less than the current value. If the new MTU is 416 * unreasonably small (arbitrarily set at 296), then 417 * we reset the MTU to the interface value and enable the 418 * lock bit, indicating that we are no longer doing MTU 419 * discovery. 420 */ 421 if (code == PRC_MSGSIZE) { 422 struct rtentry *rt; 423 int mtu; 424 425 rt = rtpurelookup((struct sockaddr *)&icmpsrc); 426 if (rt != NULL && (rt->rt_flags & RTF_HOST) && 427 !(rt->rt_rmx.rmx_locks & RTV_MTU)) { 428 mtu = ntohs(icp->icmp_nextmtu); 429 if (!mtu) 430 mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu, 431 1); 432 #ifdef DEBUG_MTUDISC 433 kprintf("MTU for %s reduced to %d\n", 434 inet_ntoa(icmpsrc.sin_addr), mtu); 435 #endif 436 if (mtu < 296) { 437 /* rt->rt_rmx.rmx_mtu = 438 rt->rt_ifp->if_mtu; */ 439 rt->rt_rmx.rmx_locks |= RTV_MTU; 440 } else if (rt->rt_rmx.rmx_mtu > mtu) { 441 rt->rt_rmx.rmx_mtu = mtu; 442 } 443 } 444 if (rt != NULL) 445 --rt->rt_refcnt; 446 } 447 #endif 448 /* 449 * XXX if the packet contains [IPv4 AH TCP], we can't make a 450 * notification to TCP layer. 451 */ 452 so_pru_ctlinput( 453 &inetsw[ip_protox[icp->icmp_ip.ip_p]], 454 code, (struct sockaddr *)&icmpsrc, &icp->icmp_ip); 455 break; 456 457 badcode: 458 icmpstat.icps_badcode++; 459 break; 460 461 case ICMP_ECHO: 462 if (!icmpbmcastecho 463 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 464 icmpstat.icps_bmcastecho++; 465 break; 466 } 467 icp->icmp_type = ICMP_ECHOREPLY; 468 #ifdef ICMP_BANDLIM 469 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0) 470 goto freeit; 471 else 472 #endif 473 goto reflect; 474 475 case ICMP_TSTAMP: 476 if (!icmpbmcastecho 477 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 478 icmpstat.icps_bmcasttstamp++; 479 break; 480 } 481 if (icmplen < ICMP_TSLEN) { 482 icmpstat.icps_badlen++; 483 break; 484 } 485 icp->icmp_type = ICMP_TSTAMPREPLY; 486 icp->icmp_rtime = iptime(); 487 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ 488 #ifdef ICMP_BANDLIM 489 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0) 490 goto freeit; 491 else 492 #endif 493 goto reflect; 494 495 case ICMP_MASKREQ: 496 if (icmpmaskrepl == 0) 497 break; 498 /* 499 * We are not able to respond with all ones broadcast 500 * unless we receive it over a point-to-point interface. 501 */ 502 if (icmplen < ICMP_MASKLEN) 503 break; 504 switch (ip->ip_dst.s_addr) { 505 506 case INADDR_BROADCAST: 507 case INADDR_ANY: 508 icmpdst.sin_addr = ip->ip_src; 509 break; 510 511 default: 512 icmpdst.sin_addr = ip->ip_dst; 513 } 514 ia = (struct in_ifaddr *)ifaof_ifpforaddr( 515 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif); 516 if (ia == NULL) 517 break; 518 if (ia->ia_ifp == 0) 519 break; 520 icp->icmp_type = ICMP_MASKREPLY; 521 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; 522 if (ip->ip_src.s_addr == 0) { 523 if (ia->ia_ifp->if_flags & IFF_BROADCAST) 524 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr; 525 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) 526 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr; 527 } 528 reflect: 529 ip->ip_len += hlen; /* since ip_input deducts this */ 530 icmpstat.icps_reflect++; 531 icmpstat.icps_outhist[icp->icmp_type]++; 532 icmp_reflect(m); 533 return(IPPROTO_DONE); 534 535 case ICMP_REDIRECT: 536 if (log_redirect) { 537 u_long src, dst, gw; 538 539 src = ntohl(ip->ip_src.s_addr); 540 dst = ntohl(icp->icmp_ip.ip_dst.s_addr); 541 gw = ntohl(icp->icmp_gwaddr.s_addr); 542 kprintf("icmp redirect from %d.%d.%d.%d: " 543 "%d.%d.%d.%d => %d.%d.%d.%d\n", 544 (int)(src >> 24), (int)((src >> 16) & 0xff), 545 (int)((src >> 8) & 0xff), (int)(src & 0xff), 546 (int)(dst >> 24), (int)((dst >> 16) & 0xff), 547 (int)((dst >> 8) & 0xff), (int)(dst & 0xff), 548 (int)(gw >> 24), (int)((gw >> 16) & 0xff), 549 (int)((gw >> 8) & 0xff), (int)(gw & 0xff)); 550 } 551 if (drop_redirect) 552 break; 553 if (code > 3) 554 goto badcode; 555 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 556 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { 557 icmpstat.icps_badlen++; 558 break; 559 } 560 /* 561 * Short circuit routing redirects to force 562 * immediate change in the kernel's routing 563 * tables. The message is also handed to anyone 564 * listening on a raw socket (e.g. the routing 565 * daemon for use in updating its tables). 566 */ 567 icmpgw.sin_addr = ip->ip_src; 568 icmpdst.sin_addr = icp->icmp_gwaddr; 569 #ifdef ICMPPRINTFS 570 if (icmpprintfs) { 571 char buf[sizeof "aaa.bbb.ccc.ddd"]; 572 573 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst)); 574 kprintf("redirect dst %s to %s\n", 575 buf, inet_ntoa(icp->icmp_gwaddr)); 576 } 577 #endif 578 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 579 rtredirect((struct sockaddr *)&icmpsrc, 580 (struct sockaddr *)&icmpdst, 581 NULL, RTF_GATEWAY | RTF_HOST, 582 (struct sockaddr *)&icmpgw); 583 kpfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc); 584 #ifdef IPSEC 585 key_sa_routechange((struct sockaddr *)&icmpsrc); 586 #endif 587 break; 588 589 /* 590 * No kernel processing for the following; 591 * just fall through to send to raw listener. 592 */ 593 case ICMP_ECHOREPLY: 594 case ICMP_ROUTERADVERT: 595 case ICMP_ROUTERSOLICIT: 596 case ICMP_TSTAMPREPLY: 597 case ICMP_IREQREPLY: 598 case ICMP_MASKREPLY: 599 default: 600 break; 601 } 602 603 raw: 604 *mp = m; 605 rip_input(mp, offp, proto); 606 return(IPPROTO_DONE); 607 608 freeit: 609 m_freem(m); 610 return(IPPROTO_DONE); 611 } 612 613 /* 614 * Reflect the ip packet back to the source 615 */ 616 static void 617 icmp_reflect(struct mbuf *m) 618 { 619 struct ip *ip = mtod(m, struct ip *); 620 struct in_ifaddr *ia; 621 struct in_ifaddr_container *iac; 622 struct ifaddr_container *ifac; 623 struct ifnet *ifp; 624 struct in_addr t; 625 struct mbuf *opts = NULL; 626 int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); 627 struct route *ro = NULL, rt; 628 629 if (!in_canforward(ip->ip_src) && 630 ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) != 631 (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { 632 m_freem(m); /* Bad return address */ 633 icmpstat.icps_badaddr++; 634 goto done; /* Ip_output() will check for broadcast */ 635 } 636 t = ip->ip_dst; 637 ip->ip_dst = ip->ip_src; 638 639 ro = &rt; 640 bzero(ro, sizeof *ro); 641 642 /* 643 * If the incoming packet was addressed directly to us, 644 * use dst as the src for the reply. Otherwise (broadcast 645 * or anonymous), use the address which corresponds 646 * to the incoming interface. 647 */ 648 ia = NULL; 649 LIST_FOREACH(iac, INADDR_HASH(t.s_addr), ia_hash) { 650 if (t.s_addr == IA_SIN(iac->ia)->sin_addr.s_addr) { 651 ia = iac->ia; 652 goto match; 653 } 654 } 655 ifp = m->m_pkthdr.rcvif; 656 if (ifp != NULL && (ifp->if_flags & IFF_BROADCAST)) { 657 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) { 658 struct ifaddr *ifa = ifac->ifa; 659 660 if (ifa->ifa_addr->sa_family != AF_INET) 661 continue; 662 ia = ifatoia(ifa); 663 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == 664 t.s_addr) 665 goto match; 666 } 667 } 668 /* 669 * If the packet was transiting through us, use the address of 670 * the interface the packet came through in. If that interface 671 * doesn't have a suitable IP address, the normal selection 672 * criteria apply. 673 */ 674 if (icmp_rfi && ifp != NULL) { 675 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) { 676 struct ifaddr *ifa = ifac->ifa; 677 678 if (ifa->ifa_addr->sa_family != AF_INET) 679 continue; 680 ia = ifatoia(ifa); 681 goto match; 682 } 683 } 684 /* 685 * If the incoming packet was not addressed directly to us, use 686 * designated interface for icmp replies specified by sysctl 687 * net.inet.icmp.reply_src (default not set). Otherwise continue 688 * with normal source selection. 689 */ 690 if (icmp_reply_src[0] != '\0' && (ifp = ifunit(icmp_reply_src))) { 691 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) { 692 struct ifaddr *ifa = ifac->ifa; 693 694 if (ifa->ifa_addr->sa_family != AF_INET) 695 continue; 696 ia = ifatoia(ifa); 697 goto match; 698 } 699 } 700 /* 701 * If the packet was transiting through us, use the address of 702 * the interface that is the closest to the packet source. 703 * When we don't have a route back to the packet source, stop here 704 * and drop the packet. 705 */ 706 ia = ip_rtaddr(ip->ip_dst, ro); 707 if (ia == NULL) { 708 m_freem(m); 709 icmpstat.icps_noroute++; 710 goto done; 711 } 712 match: 713 t = IA_SIN(ia)->sin_addr; 714 ip->ip_src = t; 715 ip->ip_ttl = ip_defttl; 716 717 if (optlen > 0) { 718 u_char *cp; 719 int opt, cnt; 720 u_int len; 721 722 /* 723 * Retrieve any source routing from the incoming packet; 724 * add on any record-route or timestamp options. 725 */ 726 cp = (u_char *) (ip + 1); 727 if ((opts = ip_srcroute(m)) == NULL && 728 (opts = m_gethdr(MB_DONTWAIT, MT_HEADER))) { 729 opts->m_len = sizeof(struct in_addr); 730 mtod(opts, struct in_addr *)->s_addr = 0; 731 } 732 if (opts) { 733 #ifdef ICMPPRINTFS 734 if (icmpprintfs) 735 kprintf("icmp_reflect optlen %d rt %d => ", 736 optlen, opts->m_len); 737 #endif 738 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { 739 opt = cp[IPOPT_OPTVAL]; 740 if (opt == IPOPT_EOL) 741 break; 742 if (opt == IPOPT_NOP) 743 len = 1; 744 else { 745 if (cnt < IPOPT_OLEN + sizeof *cp) 746 break; 747 len = cp[IPOPT_OLEN]; 748 if (len < IPOPT_OLEN + sizeof *cp || 749 len > cnt) 750 break; 751 } 752 /* 753 * Should check for overflow, but it 754 * "can't happen". 755 */ 756 if (opt == IPOPT_RR || opt == IPOPT_TS || 757 opt == IPOPT_SECURITY) { 758 bcopy(cp, 759 mtod(opts, caddr_t) + opts->m_len, 760 len); 761 opts->m_len += len; 762 } 763 } 764 /* Terminate & pad, if necessary */ 765 cnt = opts->m_len % 4; 766 if (cnt) { 767 for (; cnt < 4; cnt++) { 768 *(mtod(opts, caddr_t) + opts->m_len) = 769 IPOPT_EOL; 770 opts->m_len++; 771 } 772 } 773 #ifdef ICMPPRINTFS 774 if (icmpprintfs) 775 kprintf("%d\n", opts->m_len); 776 #endif 777 } 778 /* 779 * Now strip out original options by copying rest of first 780 * mbuf's data back, and adjust the IP length. 781 */ 782 ip->ip_len -= optlen; 783 ip->ip_vhl = IP_VHL_BORING; 784 m->m_len -= optlen; 785 if (m->m_flags & M_PKTHDR) 786 m->m_pkthdr.len -= optlen; 787 optlen += sizeof(struct ip); 788 bcopy((caddr_t)ip + optlen, ip + 1, 789 m->m_len - sizeof(struct ip)); 790 } 791 m->m_pkthdr.fw_flags &= FW_MBUF_GENERATED; 792 m->m_flags &= ~(M_BCAST|M_MCAST); 793 icmp_send(m, opts, ro); 794 done: 795 if (opts) 796 m_free(opts); 797 if (ro && ro->ro_rt) 798 RTFREE(ro->ro_rt); 799 } 800 801 /* 802 * Send an icmp packet back to the ip level, 803 * after supplying a checksum. 804 */ 805 static void 806 icmp_send(struct mbuf *m, struct mbuf *opts, struct route *rt) 807 { 808 struct ip *ip = mtod(m, struct ip *); 809 struct icmp *icp; 810 int hlen; 811 812 hlen = IP_VHL_HL(ip->ip_vhl) << 2; 813 m->m_data += hlen; 814 m->m_len -= hlen; 815 icp = mtod(m, struct icmp *); 816 icp->icmp_cksum = 0; 817 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen); 818 m->m_data -= hlen; 819 m->m_len += hlen; 820 m->m_pkthdr.rcvif = NULL; 821 #ifdef ICMPPRINTFS 822 if (icmpprintfs) { 823 char buf[sizeof "aaa.bbb.ccc.ddd"]; 824 825 strcpy(buf, inet_ntoa(ip->ip_dst)); 826 kprintf("icmp_send dst %s src %s\n", buf, inet_ntoa(ip->ip_src)); 827 } 828 #endif 829 ip_output(m, opts, rt, 0, NULL, NULL); 830 } 831 832 n_time 833 iptime(void) 834 { 835 struct timeval atv; 836 u_long t; 837 838 getmicrotime(&atv); 839 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; 840 return (htonl(t)); 841 } 842 843 #if 1 844 /* 845 * Return the next larger or smaller MTU plateau (table from RFC 1191) 846 * given current value MTU. If DIR is less than zero, a larger plateau 847 * is returned; otherwise, a smaller value is returned. 848 */ 849 int 850 ip_next_mtu(int mtu, int dir) 851 { 852 static int mtutab[] = { 853 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296, 854 68, 0 855 }; 856 int i; 857 858 for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) { 859 if (mtu >= mtutab[i]) 860 break; 861 } 862 863 if (dir < 0) { 864 if (i == 0) { 865 return 0; 866 } else { 867 return mtutab[i - 1]; 868 } 869 } else { 870 if (mtutab[i] == 0) { 871 return 0; 872 } else if(mtu > mtutab[i]) { 873 return mtutab[i]; 874 } else { 875 return mtutab[i + 1]; 876 } 877 } 878 } 879 #endif 880 881 #ifdef ICMP_BANDLIM 882 /* 883 * badport_bandlim() - check for ICMP bandwidth limit 884 * 885 * Return 0 if it is ok to send an ICMP error response, -1 if we have 886 * hit our bandwidth limit and it is not ok. 887 * 888 * If icmplim is <= 0, the feature is disabled and 0 is returned. 889 * 890 * For now we separate the TCP and UDP subsystems w/ different 'which' 891 * values. We may eventually remove this separation (and simplify the 892 * code further). 893 * 894 * Note that the printing of the error message is delayed so we can 895 * properly print the icmp error rate that the system was trying to do 896 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing 897 * the 'final' error, but it doesn't make sense to solve the printing 898 * delay with more complex code. 899 */ 900 int 901 badport_bandlim(int which) 902 { 903 static int lticks[BANDLIM_MAX + 1]; 904 static int lpackets[BANDLIM_MAX + 1]; 905 int dticks; 906 const char *bandlimittype[] = { 907 "Limiting icmp unreach response", 908 "Limiting icmp ping response", 909 "Limiting icmp tstamp response", 910 "Limiting closed port RST response", 911 "Limiting open port RST response" 912 }; 913 914 /* 915 * Return ok status if feature disabled or argument out of 916 * ranage. 917 */ 918 919 if (icmplim <= 0 || which > BANDLIM_MAX || which < 0) 920 return(0); 921 dticks = ticks - lticks[which]; 922 923 /* 924 * reset stats when cumulative dt exceeds one second. 925 */ 926 927 if ((unsigned int)dticks > hz) { 928 if (lpackets[which] > icmplim && icmplim_output) { 929 kprintf("%s from %d to %d packets per second\n", 930 bandlimittype[which], 931 lpackets[which], 932 icmplim 933 ); 934 } 935 lticks[which] = ticks; 936 lpackets[which] = 0; 937 } 938 939 /* 940 * bump packet count 941 */ 942 943 if (++lpackets[which] > icmplim) { 944 return(-1); 945 } 946 return(0); 947 } 948 #endif 949