| 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 1.3 2003/07/26 21:00:04 rob 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 | |
| 49 | #include <net/if.h> |
| 50 | #include <net/if_types.h> |
| 51 | #include <net/route.h> |
| 52 | |
| 53 | #define _IP_VHL |
| 54 | #include <netinet/in.h> |
| 55 | #include <netinet/in_systm.h> |
| 56 | #include <netinet/in_var.h> |
| 57 | #include <netinet/ip.h> |
| 58 | #include <netinet/ip_icmp.h> |
| 59 | #include <netinet/ip_var.h> |
| 60 | #include <netinet/icmp_var.h> |
| 61 | |
| 62 | #ifdef IPSEC |
| 63 | #include <netinet6/ipsec.h> |
| 64 | #include <netkey/key.h> |
| 65 | #endif |
| 66 | |
| 67 | #ifdef FAST_IPSEC |
| 68 | #include <netipsec/ipsec.h> |
| 69 | #include <netipsec/key.h> |
| 70 | #define IPSEC |
| 71 | #endif |
| 72 | |
| 73 | /* |
| 74 | * ICMP routines: error generation, receive packet processing, and |
| 75 | * routines to turnaround packets back to the originator, and |
| 76 | * host table maintenance routines. |
| 77 | */ |
| 78 | |
| 79 | static struct icmpstat icmpstat; |
| 80 | SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW, |
| 81 | &icmpstat, icmpstat, ""); |
| 82 | |
| 83 | static int icmpmaskrepl = 0; |
| 84 | SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW, |
| 85 | &icmpmaskrepl, 0, ""); |
| 86 | |
| 87 | static int drop_redirect = 0; |
| 88 | SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW, |
| 89 | &drop_redirect, 0, ""); |
| 90 | |
| 91 | static int log_redirect = 0; |
| 92 | SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW, |
| 93 | &log_redirect, 0, ""); |
| 94 | |
| 95 | #ifdef ICMP_BANDLIM |
| 96 | |
| 97 | /* |
| 98 | * ICMP error-response bandwidth limiting sysctl. If not enabled, sysctl |
| 99 | * variable content is -1 and read-only. |
| 100 | */ |
| 101 | |
| 102 | static int icmplim = 200; |
| 103 | SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW, |
| 104 | &icmplim, 0, ""); |
| 105 | #else |
| 106 | |
| 107 | static int icmplim = -1; |
| 108 | SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RD, |
| 109 | &icmplim, 0, ""); |
| 110 | |
| 111 | #endif |
| 112 | |
| 113 | static int icmplim_output = 1; |
| 114 | SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW, |
| 115 | &icmplim_output, 0, ""); |
| 116 | |
| 117 | /* |
| 118 | * ICMP broadcast echo sysctl |
| 119 | */ |
| 120 | |
| 121 | static int icmpbmcastecho = 0; |
| 122 | SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW, |
| 123 | &icmpbmcastecho, 0, ""); |
| 124 | |
| 125 | |
| 126 | #ifdef ICMPPRINTFS |
| 127 | int icmpprintfs = 0; |
| 128 | #endif |
| 129 | |
| 130 | static void icmp_reflect __P((struct mbuf *)); |
| 131 | static void icmp_send __P((struct mbuf *, struct mbuf *, struct route *)); |
| 132 | static int ip_next_mtu __P((int, int)); |
| 133 | |
| 134 | extern struct protosw inetsw[]; |
| 135 | |
| 136 | /* |
| 137 | * Generate an error packet of type error |
| 138 | * in response to bad packet ip. |
| 139 | */ |
| 140 | void |
| 141 | icmp_error(n, type, code, dest, destifp) |
| 142 | struct mbuf *n; |
| 143 | int type, code; |
| 144 | n_long dest; |
| 145 | struct ifnet *destifp; |
| 146 | { |
| 147 | struct ip *oip = mtod(n, struct ip *), *nip; |
| 148 | unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2; |
| 149 | struct icmp *icp; |
| 150 | struct mbuf *m; |
| 151 | unsigned icmplen; |
| 152 | |
| 153 | #ifdef ICMPPRINTFS |
| 154 | if (icmpprintfs) |
| 155 | printf("icmp_error(%p, %x, %d)\n", oip, type, code); |
| 156 | #endif |
| 157 | if (type != ICMP_REDIRECT) |
| 158 | icmpstat.icps_error++; |
| 159 | /* |
| 160 | * Don't send error if not the first fragment of message. |
| 161 | * Don't error if the old packet protocol was ICMP |
| 162 | * error message, only known informational types. |
| 163 | */ |
| 164 | if (oip->ip_off &~ (IP_MF|IP_DF)) |
| 165 | goto freeit; |
| 166 | if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && |
| 167 | n->m_len >= oiplen + ICMP_MINLEN && |
| 168 | !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) { |
| 169 | icmpstat.icps_oldicmp++; |
| 170 | goto freeit; |
| 171 | } |
| 172 | /* Don't send error in response to a multicast or broadcast packet */ |
| 173 | if (n->m_flags & (M_BCAST|M_MCAST)) |
| 174 | goto freeit; |
| 175 | /* |
| 176 | * First, formulate icmp message |
| 177 | */ |
| 178 | m = m_gethdr(M_DONTWAIT, MT_HEADER); |
| 179 | if (m == NULL) |
| 180 | goto freeit; |
| 181 | icmplen = min(oiplen + 8, oip->ip_len); |
| 182 | if (icmplen < sizeof(struct ip)) |
| 183 | panic("icmp_error: bad length"); |
| 184 | m->m_len = icmplen + ICMP_MINLEN; |
| 185 | MH_ALIGN(m, m->m_len); |
| 186 | icp = mtod(m, struct icmp *); |
| 187 | if ((u_int)type > ICMP_MAXTYPE) |
| 188 | panic("icmp_error"); |
| 189 | icmpstat.icps_outhist[type]++; |
| 190 | icp->icmp_type = type; |
| 191 | if (type == ICMP_REDIRECT) |
| 192 | icp->icmp_gwaddr.s_addr = dest; |
| 193 | else { |
| 194 | icp->icmp_void = 0; |
| 195 | /* |
| 196 | * The following assignments assume an overlay with the |
| 197 | * zeroed icmp_void field. |
| 198 | */ |
| 199 | if (type == ICMP_PARAMPROB) { |
| 200 | icp->icmp_pptr = code; |
| 201 | code = 0; |
| 202 | } else if (type == ICMP_UNREACH && |
| 203 | code == ICMP_UNREACH_NEEDFRAG && destifp) { |
| 204 | icp->icmp_nextmtu = htons(destifp->if_mtu); |
| 205 | } |
| 206 | } |
| 207 | |
| 208 | icp->icmp_code = code; |
| 209 | m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip); |
| 210 | nip = &icp->icmp_ip; |
| 211 | |
| 212 | /* |
| 213 | * Convert fields to network representation. |
| 214 | */ |
| 215 | HTONS(nip->ip_len); |
| 216 | HTONS(nip->ip_off); |
| 217 | |
| 218 | /* |
| 219 | * Now, copy old ip header (without options) |
| 220 | * in front of icmp message. |
| 221 | */ |
| 222 | if (m->m_data - sizeof(struct ip) < m->m_pktdat) |
| 223 | panic("icmp len"); |
| 224 | m->m_data -= sizeof(struct ip); |
| 225 | m->m_len += sizeof(struct ip); |
| 226 | m->m_pkthdr.len = m->m_len; |
| 227 | m->m_pkthdr.rcvif = n->m_pkthdr.rcvif; |
| 228 | nip = mtod(m, struct ip *); |
| 229 | bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip)); |
| 230 | nip->ip_len = m->m_len; |
| 231 | nip->ip_vhl = IP_VHL_BORING; |
| 232 | nip->ip_p = IPPROTO_ICMP; |
| 233 | nip->ip_tos = 0; |
| 234 | icmp_reflect(m); |
| 235 | |
| 236 | freeit: |
| 237 | m_freem(n); |
| 238 | } |
| 239 | |
| 240 | static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET }; |
| 241 | static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET }; |
| 242 | static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET }; |
| 243 | |
| 244 | /* |
| 245 | * Process a received ICMP message. |
| 246 | */ |
| 247 | void |
| 248 | icmp_input(m, off, proto) |
| 249 | struct mbuf *m; |
| 250 | int off, proto; |
| 251 | { |
| 252 | int hlen = off; |
| 253 | struct icmp *icp; |
| 254 | struct ip *ip = mtod(m, struct ip *); |
| 255 | int icmplen = ip->ip_len; |
| 256 | int i; |
| 257 | struct in_ifaddr *ia; |
| 258 | void (*ctlfunc) __P((int, struct sockaddr *, void *)); |
| 259 | int code; |
| 260 | |
| 261 | /* |
| 262 | * Locate icmp structure in mbuf, and check |
| 263 | * that not corrupted and of at least minimum length. |
| 264 | */ |
| 265 | #ifdef ICMPPRINTFS |
| 266 | if (icmpprintfs) { |
| 267 | char buf[4 * sizeof "123"]; |
| 268 | strcpy(buf, inet_ntoa(ip->ip_src)); |
| 269 | printf("icmp_input from %s to %s, len %d\n", |
| 270 | buf, inet_ntoa(ip->ip_dst), icmplen); |
| 271 | } |
| 272 | #endif |
| 273 | if (icmplen < ICMP_MINLEN) { |
| 274 | icmpstat.icps_tooshort++; |
| 275 | goto freeit; |
| 276 | } |
| 277 | i = hlen + min(icmplen, ICMP_ADVLENMIN); |
| 278 | if (m->m_len < i && (m = m_pullup(m, i)) == 0) { |
| 279 | icmpstat.icps_tooshort++; |
| 280 | return; |
| 281 | } |
| 282 | ip = mtod(m, struct ip *); |
| 283 | m->m_len -= hlen; |
| 284 | m->m_data += hlen; |
| 285 | icp = mtod(m, struct icmp *); |
| 286 | if (in_cksum(m, icmplen)) { |
| 287 | icmpstat.icps_checksum++; |
| 288 | goto freeit; |
| 289 | } |
| 290 | m->m_len += hlen; |
| 291 | m->m_data -= hlen; |
| 292 | |
| 293 | if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) { |
| 294 | /* |
| 295 | * Deliver very specific ICMP type only. |
| 296 | */ |
| 297 | switch (icp->icmp_type) { |
| 298 | case ICMP_UNREACH: |
| 299 | case ICMP_TIMXCEED: |
| 300 | break; |
| 301 | default: |
| 302 | goto freeit; |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | #ifdef ICMPPRINTFS |
| 307 | if (icmpprintfs) |
| 308 | printf("icmp_input, type %d code %d\n", icp->icmp_type, |
| 309 | icp->icmp_code); |
| 310 | #endif |
| 311 | |
| 312 | /* |
| 313 | * Message type specific processing. |
| 314 | */ |
| 315 | if (icp->icmp_type > ICMP_MAXTYPE) |
| 316 | goto raw; |
| 317 | icmpstat.icps_inhist[icp->icmp_type]++; |
| 318 | code = icp->icmp_code; |
| 319 | switch (icp->icmp_type) { |
| 320 | |
| 321 | case ICMP_UNREACH: |
| 322 | switch (code) { |
| 323 | case ICMP_UNREACH_NET: |
| 324 | case ICMP_UNREACH_HOST: |
| 325 | case ICMP_UNREACH_SRCFAIL: |
| 326 | case ICMP_UNREACH_NET_UNKNOWN: |
| 327 | case ICMP_UNREACH_HOST_UNKNOWN: |
| 328 | case ICMP_UNREACH_ISOLATED: |
| 329 | case ICMP_UNREACH_TOSNET: |
| 330 | case ICMP_UNREACH_TOSHOST: |
| 331 | case ICMP_UNREACH_HOST_PRECEDENCE: |
| 332 | case ICMP_UNREACH_PRECEDENCE_CUTOFF: |
| 333 | code = PRC_UNREACH_NET; |
| 334 | break; |
| 335 | |
| 336 | case ICMP_UNREACH_NEEDFRAG: |
| 337 | code = PRC_MSGSIZE; |
| 338 | break; |
| 339 | |
| 340 | /* |
| 341 | * RFC 1122, Sections 3.2.2.1 and 4.2.3.9. |
| 342 | * Treat subcodes 2,3 as immediate RST |
| 343 | */ |
| 344 | case ICMP_UNREACH_PROTOCOL: |
| 345 | case ICMP_UNREACH_PORT: |
| 346 | code = PRC_UNREACH_PORT; |
| 347 | break; |
| 348 | |
| 349 | case ICMP_UNREACH_NET_PROHIB: |
| 350 | case ICMP_UNREACH_HOST_PROHIB: |
| 351 | case ICMP_UNREACH_FILTER_PROHIB: |
| 352 | code = PRC_UNREACH_ADMIN_PROHIB; |
| 353 | break; |
| 354 | |
| 355 | default: |
| 356 | goto badcode; |
| 357 | } |
| 358 | goto deliver; |
| 359 | |
| 360 | case ICMP_TIMXCEED: |
| 361 | if (code > 1) |
| 362 | goto badcode; |
| 363 | code += PRC_TIMXCEED_INTRANS; |
| 364 | goto deliver; |
| 365 | |
| 366 | case ICMP_PARAMPROB: |
| 367 | if (code > 1) |
| 368 | goto badcode; |
| 369 | code = PRC_PARAMPROB; |
| 370 | goto deliver; |
| 371 | |
| 372 | case ICMP_SOURCEQUENCH: |
| 373 | if (code) |
| 374 | goto badcode; |
| 375 | code = PRC_QUENCH; |
| 376 | deliver: |
| 377 | /* |
| 378 | * Problem with datagram; advise higher level routines. |
| 379 | */ |
| 380 | if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || |
| 381 | IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { |
| 382 | icmpstat.icps_badlen++; |
| 383 | goto freeit; |
| 384 | } |
| 385 | NTOHS(icp->icmp_ip.ip_len); |
| 386 | /* Discard ICMP's in response to multicast packets */ |
| 387 | if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr))) |
| 388 | goto badcode; |
| 389 | #ifdef ICMPPRINTFS |
| 390 | if (icmpprintfs) |
| 391 | printf("deliver to protocol %d\n", icp->icmp_ip.ip_p); |
| 392 | #endif |
| 393 | icmpsrc.sin_addr = icp->icmp_ip.ip_dst; |
| 394 | #if 1 |
| 395 | /* |
| 396 | * MTU discovery: |
| 397 | * If we got a needfrag and there is a host route to the |
| 398 | * original destination, and the MTU is not locked, then |
| 399 | * set the MTU in the route to the suggested new value |
| 400 | * (if given) and then notify as usual. The ULPs will |
| 401 | * notice that the MTU has changed and adapt accordingly. |
| 402 | * If no new MTU was suggested, then we guess a new one |
| 403 | * less than the current value. If the new MTU is |
| 404 | * unreasonably small (arbitrarily set at 296), then |
| 405 | * we reset the MTU to the interface value and enable the |
| 406 | * lock bit, indicating that we are no longer doing MTU |
| 407 | * discovery. |
| 408 | */ |
| 409 | if (code == PRC_MSGSIZE) { |
| 410 | struct rtentry *rt; |
| 411 | int mtu; |
| 412 | |
| 413 | rt = rtalloc1((struct sockaddr *)&icmpsrc, 0, |
| 414 | RTF_CLONING | RTF_PRCLONING); |
| 415 | if (rt && (rt->rt_flags & RTF_HOST) |
| 416 | && !(rt->rt_rmx.rmx_locks & RTV_MTU)) { |
| 417 | mtu = ntohs(icp->icmp_nextmtu); |
| 418 | if (!mtu) |
| 419 | mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu, |
| 420 | 1); |
| 421 | #ifdef DEBUG_MTUDISC |
| 422 | printf("MTU for %s reduced to %d\n", |
| 423 | inet_ntoa(icmpsrc.sin_addr), mtu); |
| 424 | #endif |
| 425 | if (mtu < 296) { |
| 426 | /* rt->rt_rmx.rmx_mtu = |
| 427 | rt->rt_ifp->if_mtu; */ |
| 428 | rt->rt_rmx.rmx_locks |= RTV_MTU; |
| 429 | } else if (rt->rt_rmx.rmx_mtu > mtu) { |
| 430 | rt->rt_rmx.rmx_mtu = mtu; |
| 431 | } |
| 432 | } |
| 433 | if (rt) |
| 434 | RTFREE(rt); |
| 435 | } |
| 436 | |
| 437 | #endif |
| 438 | /* |
| 439 | * XXX if the packet contains [IPv4 AH TCP], we can't make a |
| 440 | * notification to TCP layer. |
| 441 | */ |
| 442 | ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput; |
| 443 | if (ctlfunc) |
| 444 | (*ctlfunc)(code, (struct sockaddr *)&icmpsrc, |
| 445 | (void *)&icp->icmp_ip); |
| 446 | break; |
| 447 | |
| 448 | badcode: |
| 449 | icmpstat.icps_badcode++; |
| 450 | break; |
| 451 | |
| 452 | case ICMP_ECHO: |
| 453 | if (!icmpbmcastecho |
| 454 | && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { |
| 455 | icmpstat.icps_bmcastecho++; |
| 456 | break; |
| 457 | } |
| 458 | icp->icmp_type = ICMP_ECHOREPLY; |
| 459 | #ifdef ICMP_BANDLIM |
| 460 | if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0) |
| 461 | goto freeit; |
| 462 | else |
| 463 | #endif |
| 464 | goto reflect; |
| 465 | |
| 466 | case ICMP_TSTAMP: |
| 467 | if (!icmpbmcastecho |
| 468 | && (m->m_flags & (M_MCAST | M_BCAST)) != 0) { |
| 469 | icmpstat.icps_bmcasttstamp++; |
| 470 | break; |
| 471 | } |
| 472 | if (icmplen < ICMP_TSLEN) { |
| 473 | icmpstat.icps_badlen++; |
| 474 | break; |
| 475 | } |
| 476 | icp->icmp_type = ICMP_TSTAMPREPLY; |
| 477 | icp->icmp_rtime = iptime(); |
| 478 | icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ |
| 479 | #ifdef ICMP_BANDLIM |
| 480 | if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0) |
| 481 | goto freeit; |
| 482 | else |
| 483 | #endif |
| 484 | goto reflect; |
| 485 | |
| 486 | case ICMP_MASKREQ: |
| 487 | if (icmpmaskrepl == 0) |
| 488 | break; |
| 489 | /* |
| 490 | * We are not able to respond with all ones broadcast |
| 491 | * unless we receive it over a point-to-point interface. |
| 492 | */ |
| 493 | if (icmplen < ICMP_MASKLEN) |
| 494 | break; |
| 495 | switch (ip->ip_dst.s_addr) { |
| 496 | |
| 497 | case INADDR_BROADCAST: |
| 498 | case INADDR_ANY: |
| 499 | icmpdst.sin_addr = ip->ip_src; |
| 500 | break; |
| 501 | |
| 502 | default: |
| 503 | icmpdst.sin_addr = ip->ip_dst; |
| 504 | } |
| 505 | ia = (struct in_ifaddr *)ifaof_ifpforaddr( |
| 506 | (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif); |
| 507 | if (ia == 0) |
| 508 | break; |
| 509 | if (ia->ia_ifp == 0) |
| 510 | break; |
| 511 | icp->icmp_type = ICMP_MASKREPLY; |
| 512 | icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; |
| 513 | if (ip->ip_src.s_addr == 0) { |
| 514 | if (ia->ia_ifp->if_flags & IFF_BROADCAST) |
| 515 | ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr; |
| 516 | else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) |
| 517 | ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr; |
| 518 | } |
| 519 | reflect: |
| 520 | ip->ip_len += hlen; /* since ip_input deducts this */ |
| 521 | icmpstat.icps_reflect++; |
| 522 | icmpstat.icps_outhist[icp->icmp_type]++; |
| 523 | icmp_reflect(m); |
| 524 | return; |
| 525 | |
| 526 | case ICMP_REDIRECT: |
| 527 | if (log_redirect) { |
| 528 | u_long src, dst, gw; |
| 529 | |
| 530 | src = ntohl(ip->ip_src.s_addr); |
| 531 | dst = ntohl(icp->icmp_ip.ip_dst.s_addr); |
| 532 | gw = ntohl(icp->icmp_gwaddr.s_addr); |
| 533 | printf("icmp redirect from %d.%d.%d.%d: " |
| 534 | "%d.%d.%d.%d => %d.%d.%d.%d\n", |
| 535 | (int)(src >> 24), (int)((src >> 16) & 0xff), |
| 536 | (int)((src >> 8) & 0xff), (int)(src & 0xff), |
| 537 | (int)(dst >> 24), (int)((dst >> 16) & 0xff), |
| 538 | (int)((dst >> 8) & 0xff), (int)(dst & 0xff), |
| 539 | (int)(gw >> 24), (int)((gw >> 16) & 0xff), |
| 540 | (int)((gw >> 8) & 0xff), (int)(gw & 0xff)); |
| 541 | } |
| 542 | if (drop_redirect) |
| 543 | break; |
| 544 | if (code > 3) |
| 545 | goto badcode; |
| 546 | if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || |
| 547 | IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) { |
| 548 | icmpstat.icps_badlen++; |
| 549 | break; |
| 550 | } |
| 551 | /* |
| 552 | * Short circuit routing redirects to force |
| 553 | * immediate change in the kernel's routing |
| 554 | * tables. The message is also handed to anyone |
| 555 | * listening on a raw socket (e.g. the routing |
| 556 | * daemon for use in updating its tables). |
| 557 | */ |
| 558 | icmpgw.sin_addr = ip->ip_src; |
| 559 | icmpdst.sin_addr = icp->icmp_gwaddr; |
| 560 | #ifdef ICMPPRINTFS |
| 561 | if (icmpprintfs) { |
| 562 | char buf[4 * sizeof "123"]; |
| 563 | strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst)); |
| 564 | |
| 565 | printf("redirect dst %s to %s\n", |
| 566 | buf, inet_ntoa(icp->icmp_gwaddr)); |
| 567 | } |
| 568 | #endif |
| 569 | icmpsrc.sin_addr = icp->icmp_ip.ip_dst; |
| 570 | rtredirect((struct sockaddr *)&icmpsrc, |
| 571 | (struct sockaddr *)&icmpdst, |
| 572 | (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST, |
| 573 | (struct sockaddr *)&icmpgw, (struct rtentry **)0); |
| 574 | pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc); |
| 575 | #ifdef IPSEC |
| 576 | key_sa_routechange((struct sockaddr *)&icmpsrc); |
| 577 | #endif |
| 578 | break; |
| 579 | |
| 580 | /* |
| 581 | * No kernel processing for the following; |
| 582 | * just fall through to send to raw listener. |
| 583 | */ |
| 584 | case ICMP_ECHOREPLY: |
| 585 | case ICMP_ROUTERADVERT: |
| 586 | case ICMP_ROUTERSOLICIT: |
| 587 | case ICMP_TSTAMPREPLY: |
| 588 | case ICMP_IREQREPLY: |
| 589 | case ICMP_MASKREPLY: |
| 590 | default: |
| 591 | break; |
| 592 | } |
| 593 | |
| 594 | raw: |
| 595 | rip_input(m, off, proto); |
| 596 | return; |
| 597 | |
| 598 | freeit: |
| 599 | m_freem(m); |
| 600 | } |
| 601 | |
| 602 | /* |
| 603 | * Reflect the ip packet back to the source |
| 604 | */ |
| 605 | static void |
| 606 | icmp_reflect(m) |
| 607 | struct mbuf *m; |
| 608 | { |
| 609 | struct ip *ip = mtod(m, struct ip *); |
| 610 | struct ifaddr *ifa; |
| 611 | struct in_ifaddr *ia; |
| 612 | struct in_addr t; |
| 613 | struct mbuf *opts = 0; |
| 614 | int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); |
| 615 | struct route *ro = NULL, rt; |
| 616 | |
| 617 | if (!in_canforward(ip->ip_src) && |
| 618 | ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) != |
| 619 | (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { |
| 620 | m_freem(m); /* Bad return address */ |
| 621 | icmpstat.icps_badaddr++; |
| 622 | goto done; /* Ip_output() will check for broadcast */ |
| 623 | } |
| 624 | t = ip->ip_dst; |
| 625 | ip->ip_dst = ip->ip_src; |
| 626 | ro = &rt; |
| 627 | bzero(ro, sizeof(*ro)); |
| 628 | /* |
| 629 | * If the incoming packet was addressed directly to us, |
| 630 | * use dst as the src for the reply. Otherwise (broadcast |
| 631 | * or anonymous), use the address which corresponds |
| 632 | * to the incoming interface. |
| 633 | */ |
| 634 | LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash) |
| 635 | if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr) |
| 636 | goto match; |
| 637 | if (m->m_pkthdr.rcvif != NULL && |
| 638 | m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { |
| 639 | TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) { |
| 640 | if (ifa->ifa_addr->sa_family != AF_INET) |
| 641 | continue; |
| 642 | ia = ifatoia(ifa); |
| 643 | if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr == |
| 644 | t.s_addr) |
| 645 | goto match; |
| 646 | } |
| 647 | } |
| 648 | ia = ip_rtaddr(ip->ip_dst, ro); |
| 649 | /* We need a route to do anything useful. */ |
| 650 | if (ia == NULL) { |
| 651 | m_freem(m); |
| 652 | icmpstat.icps_noroute++; |
| 653 | goto done; |
| 654 | } |
| 655 | match: |
| 656 | t = IA_SIN(ia)->sin_addr; |
| 657 | ip->ip_src = t; |
| 658 | ip->ip_ttl = ip_defttl; |
| 659 | |
| 660 | if (optlen > 0) { |
| 661 | u_char *cp; |
| 662 | int opt, cnt; |
| 663 | u_int len; |
| 664 | |
| 665 | /* |
| 666 | * Retrieve any source routing from the incoming packet; |
| 667 | * add on any record-route or timestamp options. |
| 668 | */ |
| 669 | cp = (u_char *) (ip + 1); |
| 670 | if ((opts = ip_srcroute()) == 0 && |
| 671 | (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) { |
| 672 | opts->m_len = sizeof(struct in_addr); |
| 673 | mtod(opts, struct in_addr *)->s_addr = 0; |
| 674 | } |
| 675 | if (opts) { |
| 676 | #ifdef ICMPPRINTFS |
| 677 | if (icmpprintfs) |
| 678 | printf("icmp_reflect optlen %d rt %d => ", |
| 679 | optlen, opts->m_len); |
| 680 | #endif |
| 681 | for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { |
| 682 | opt = cp[IPOPT_OPTVAL]; |
| 683 | if (opt == IPOPT_EOL) |
| 684 | break; |
| 685 | if (opt == IPOPT_NOP) |
| 686 | len = 1; |
| 687 | else { |
| 688 | if (cnt < IPOPT_OLEN + sizeof(*cp)) |
| 689 | break; |
| 690 | len = cp[IPOPT_OLEN]; |
| 691 | if (len < IPOPT_OLEN + sizeof(*cp) || |
| 692 | len > cnt) |
| 693 | break; |
| 694 | } |
| 695 | /* |
| 696 | * Should check for overflow, but it "can't happen" |
| 697 | */ |
| 698 | if (opt == IPOPT_RR || opt == IPOPT_TS || |
| 699 | opt == IPOPT_SECURITY) { |
| 700 | bcopy((caddr_t)cp, |
| 701 | mtod(opts, caddr_t) + opts->m_len, len); |
| 702 | opts->m_len += len; |
| 703 | } |
| 704 | } |
| 705 | /* Terminate & pad, if necessary */ |
| 706 | cnt = opts->m_len % 4; |
| 707 | if (cnt) { |
| 708 | for (; cnt < 4; cnt++) { |
| 709 | *(mtod(opts, caddr_t) + opts->m_len) = |
| 710 | IPOPT_EOL; |
| 711 | opts->m_len++; |
| 712 | } |
| 713 | } |
| 714 | #ifdef ICMPPRINTFS |
| 715 | if (icmpprintfs) |
| 716 | printf("%d\n", opts->m_len); |
| 717 | #endif |
| 718 | } |
| 719 | /* |
| 720 | * Now strip out original options by copying rest of first |
| 721 | * mbuf's data back, and adjust the IP length. |
| 722 | */ |
| 723 | ip->ip_len -= optlen; |
| 724 | ip->ip_vhl = IP_VHL_BORING; |
| 725 | m->m_len -= optlen; |
| 726 | if (m->m_flags & M_PKTHDR) |
| 727 | m->m_pkthdr.len -= optlen; |
| 728 | optlen += sizeof(struct ip); |
| 729 | bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1), |
| 730 | (unsigned)(m->m_len - sizeof(struct ip))); |
| 731 | } |
| 732 | m->m_flags &= ~(M_BCAST|M_MCAST); |
| 733 | icmp_send(m, opts, ro); |
| 734 | done: |
| 735 | if (opts) |
| 736 | (void)m_free(opts); |
| 737 | if (ro && ro->ro_rt) |
| 738 | RTFREE(ro->ro_rt); |
| 739 | } |
| 740 | |
| 741 | /* |
| 742 | * Send an icmp packet back to the ip level, |
| 743 | * after supplying a checksum. |
| 744 | */ |
| 745 | static void |
| 746 | icmp_send(m, opts, rt) |
| 747 | struct mbuf *m; |
| 748 | struct mbuf *opts; |
| 749 | struct route *rt; |
| 750 | { |
| 751 | struct ip *ip = mtod(m, struct ip *); |
| 752 | int hlen; |
| 753 | struct icmp *icp; |
| 754 | |
| 755 | hlen = IP_VHL_HL(ip->ip_vhl) << 2; |
| 756 | m->m_data += hlen; |
| 757 | m->m_len -= hlen; |
| 758 | icp = mtod(m, struct icmp *); |
| 759 | icp->icmp_cksum = 0; |
| 760 | icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen); |
| 761 | m->m_data -= hlen; |
| 762 | m->m_len += hlen; |
| 763 | m->m_pkthdr.rcvif = (struct ifnet *)0; |
| 764 | #ifdef ICMPPRINTFS |
| 765 | if (icmpprintfs) { |
| 766 | char buf[4 * sizeof "123"]; |
| 767 | strcpy(buf, inet_ntoa(ip->ip_dst)); |
| 768 | printf("icmp_send dst %s src %s\n", |
| 769 | buf, inet_ntoa(ip->ip_src)); |
| 770 | } |
| 771 | #endif |
| 772 | (void) ip_output(m, opts, rt, 0, NULL, NULL); |
| 773 | } |
| 774 | |
| 775 | n_time |
| 776 | iptime() |
| 777 | { |
| 778 | struct timeval atv; |
| 779 | u_long t; |
| 780 | |
| 781 | getmicrotime(&atv); |
| 782 | t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; |
| 783 | return (htonl(t)); |
| 784 | } |
| 785 | |
| 786 | #if 1 |
| 787 | /* |
| 788 | * Return the next larger or smaller MTU plateau (table from RFC 1191) |
| 789 | * given current value MTU. If DIR is less than zero, a larger plateau |
| 790 | * is returned; otherwise, a smaller value is returned. |
| 791 | */ |
| 792 | static int |
| 793 | ip_next_mtu(mtu, dir) |
| 794 | int mtu; |
| 795 | int dir; |
| 796 | { |
| 797 | static int mtutab[] = { |
| 798 | 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296, |
| 799 | 68, 0 |
| 800 | }; |
| 801 | int i; |
| 802 | |
| 803 | for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) { |
| 804 | if (mtu >= mtutab[i]) |
| 805 | break; |
| 806 | } |
| 807 | |
| 808 | if (dir < 0) { |
| 809 | if (i == 0) { |
| 810 | return 0; |
| 811 | } else { |
| 812 | return mtutab[i - 1]; |
| 813 | } |
| 814 | } else { |
| 815 | if (mtutab[i] == 0) { |
| 816 | return 0; |
| 817 | } else if(mtu > mtutab[i]) { |
| 818 | return mtutab[i]; |
| 819 | } else { |
| 820 | return mtutab[i + 1]; |
| 821 | } |
| 822 | } |
| 823 | } |
| 824 | #endif |
| 825 | |
| 826 | #ifdef ICMP_BANDLIM |
| 827 | |
| 828 | /* |
| 829 | * badport_bandlim() - check for ICMP bandwidth limit |
| 830 | * |
| 831 | * Return 0 if it is ok to send an ICMP error response, -1 if we have |
| 832 | * hit our bandwidth limit and it is not ok. |
| 833 | * |
| 834 | * If icmplim is <= 0, the feature is disabled and 0 is returned. |
| 835 | * |
| 836 | * For now we separate the TCP and UDP subsystems w/ different 'which' |
| 837 | * values. We may eventually remove this separation (and simplify the |
| 838 | * code further). |
| 839 | * |
| 840 | * Note that the printing of the error message is delayed so we can |
| 841 | * properly print the icmp error rate that the system was trying to do |
| 842 | * (i.e. 22000/100 pps, etc...). This can cause long delays in printing |
| 843 | * the 'final' error, but it doesn't make sense to solve the printing |
| 844 | * delay with more complex code. |
| 845 | */ |
| 846 | |
| 847 | int |
| 848 | badport_bandlim(int which) |
| 849 | { |
| 850 | static int lticks[BANDLIM_MAX + 1]; |
| 851 | static int lpackets[BANDLIM_MAX + 1]; |
| 852 | int dticks; |
| 853 | const char *bandlimittype[] = { |
| 854 | "Limiting icmp unreach response", |
| 855 | "Limiting icmp ping response", |
| 856 | "Limiting icmp tstamp response", |
| 857 | "Limiting closed port RST response", |
| 858 | "Limiting open port RST response" |
| 859 | }; |
| 860 | |
| 861 | /* |
| 862 | * Return ok status if feature disabled or argument out of |
| 863 | * ranage. |
| 864 | */ |
| 865 | |
| 866 | if (icmplim <= 0 || which > BANDLIM_MAX || which < 0) |
| 867 | return(0); |
| 868 | dticks = ticks - lticks[which]; |
| 869 | |
| 870 | /* |
| 871 | * reset stats when cumulative dt exceeds one second. |
| 872 | */ |
| 873 | |
| 874 | if ((unsigned int)dticks > hz) { |
| 875 | if (lpackets[which] > icmplim && icmplim_output) { |
| 876 | printf("%s from %d to %d packets per second\n", |
| 877 | bandlimittype[which], |
| 878 | lpackets[which], |
| 879 | icmplim |
| 880 | ); |
| 881 | } |
| 882 | lticks[which] = ticks; |
| 883 | lpackets[which] = 0; |
| 884 | } |
| 885 | |
| 886 | /* |
| 887 | * bump packet count |
| 888 | */ |
| 889 | |
| 890 | if (++lpackets[which] > icmplim) { |
| 891 | return(-1); |
| 892 | } |
| 893 | return(0); |
| 894 | } |
| 895 | |
| 896 | #endif |
| 897 | |
| 898 | |