1 /* 2 * Copyright (c) 2004 Jeffrey M. Hsu. All rights reserved. 3 * Copyright (c) 2004 The DragonFly Project. All rights reserved. 4 * 5 * This code is derived from software contributed to The DragonFly Project 6 * by Jeffrey M. Hsu. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of The DragonFly Project nor the names of its 17 * contributors may be used to endorse or promote products derived 18 * from this software without specific, prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 /* 35 * Copyright (c) 2004 Jeffrey M. Hsu. All rights reserved. 36 * 37 * License terms: all terms for the DragonFly license above plus the following: 38 * 39 * 4. All advertising materials mentioning features or use of this software 40 * must display the following acknowledgement: 41 * 42 * This product includes software developed by Jeffrey M. Hsu 43 * for the DragonFly Project. 44 * 45 * This requirement may be waived with permission from Jeffrey Hsu. 46 * This requirement will sunset and may be removed on July 8 2005, 47 * after which the standard DragonFly license (as shown above) will 48 * apply. 49 */ 50 51 /* 52 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 53 * The Regents of the University of California. All rights reserved. 54 * 55 * Redistribution and use in source and binary forms, with or without 56 * modification, are permitted provided that the following conditions 57 * are met: 58 * 1. Redistributions of source code must retain the above copyright 59 * notice, this list of conditions and the following disclaimer. 60 * 2. Redistributions in binary form must reproduce the above copyright 61 * notice, this list of conditions and the following disclaimer in the 62 * documentation and/or other materials provided with the distribution. 63 * 3. All advertising materials mentioning features or use of this software 64 * must display the following acknowledgement: 65 * This product includes software developed by the University of 66 * California, Berkeley and its contributors. 67 * 4. Neither the name of the University nor the names of its contributors 68 * may be used to endorse or promote products derived from this software 69 * without specific prior written permission. 70 * 71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 81 * SUCH DAMAGE. 82 * 83 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 84 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.18 2003/01/24 05:11:34 sam Exp $ 85 * $DragonFly: src/sys/netinet/udp_usrreq.c,v 1.27 2004/10/15 22:59:10 hsu Exp $ 86 */ 87 88 #include "opt_ipsec.h" 89 #include "opt_inet6.h" 90 91 #include <sys/param.h> 92 #include <sys/systm.h> 93 #include <sys/kernel.h> 94 #include <sys/malloc.h> 95 #include <sys/mbuf.h> 96 #include <sys/domain.h> 97 #include <sys/proc.h> 98 #include <sys/protosw.h> 99 #include <sys/socket.h> 100 #include <sys/socketvar.h> 101 #include <sys/sysctl.h> 102 #include <sys/syslog.h> 103 #include <sys/in_cksum.h> 104 105 #include <machine/stdarg.h> 106 107 #include <vm/vm_zone.h> 108 109 #include <net/if.h> 110 #include <net/route.h> 111 112 #include <netinet/in.h> 113 #include <netinet/in_systm.h> 114 #include <netinet/ip.h> 115 #ifdef INET6 116 #include <netinet/ip6.h> 117 #endif 118 #include <netinet/in_pcb.h> 119 #include <netinet/in_var.h> 120 #include <netinet/ip_var.h> 121 #ifdef INET6 122 #include <netinet6/ip6_var.h> 123 #endif 124 #include <netinet/ip_icmp.h> 125 #include <netinet/icmp_var.h> 126 #include <netinet/udp.h> 127 #include <netinet/udp_var.h> 128 129 #ifdef FAST_IPSEC 130 #include <netproto/ipsec/ipsec.h> 131 #endif /*FAST_IPSEC*/ 132 133 #ifdef IPSEC 134 #include <netinet6/ipsec.h> 135 #endif /*IPSEC*/ 136 137 /* 138 * UDP protocol implementation. 139 * Per RFC 768, August, 1980. 140 */ 141 #ifndef COMPAT_42 142 static int udpcksum = 1; 143 #else 144 static int udpcksum = 0; /* XXX */ 145 #endif 146 SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW, 147 &udpcksum, 0, ""); 148 149 int log_in_vain = 0; 150 SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW, 151 &log_in_vain, 0, "Log all incoming UDP packets"); 152 153 static int blackhole = 0; 154 SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW, 155 &blackhole, 0, "Do not send port unreachables for refused connects"); 156 157 static int strict_mcast_mship = 1; 158 SYSCTL_INT(_net_inet_udp, OID_AUTO, strict_mcast_mship, CTLFLAG_RW, 159 &strict_mcast_mship, 0, "Only send multicast to member sockets"); 160 161 struct inpcbinfo udbinfo; 162 163 #ifndef UDBHASHSIZE 164 #define UDBHASHSIZE 16 165 #endif 166 167 struct udpstat udpstat; /* from udp_var.h */ 168 SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW, 169 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)"); 170 171 static struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET }; 172 #ifdef INET6 173 struct udp_in6 { 174 struct sockaddr_in6 uin6_sin; 175 u_char uin6_init_done : 1; 176 } udp_in6 = { 177 { sizeof(udp_in6.uin6_sin), AF_INET6 }, 178 0 179 }; 180 struct udp_ip6 { 181 struct ip6_hdr uip6_ip6; 182 u_char uip6_init_done : 1; 183 } udp_ip6; 184 #endif /* INET6 */ 185 186 static void udp_append (struct inpcb *last, struct ip *ip, 187 struct mbuf *n, int off); 188 #ifdef INET6 189 static void ip_2_ip6_hdr (struct ip6_hdr *ip6, struct ip *ip); 190 #endif 191 192 static int udp_detach (struct socket *so); 193 static int udp_output (struct inpcb *, struct mbuf *, struct sockaddr *, 194 struct mbuf *, struct thread *); 195 196 void 197 udp_init() 198 { 199 in_pcbinfo_init(&udbinfo); 200 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask); 201 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB, 202 &udbinfo.porthashmask); 203 udbinfo.wildcardhashbase = hashinit(UDBHASHSIZE, M_PCB, 204 &udbinfo.wildcardhashmask); 205 udbinfo.ipi_zone = zinit("udpcb", sizeof(struct inpcb), maxsockets, 206 ZONE_INTERRUPT, 0); 207 udp_thread_init(); 208 } 209 210 /* 211 * Check multicast packets to make sure they are only sent to sockets with 212 * multicast memberships for the packet's destination address and arrival 213 * interface. Multicast packets to multicast-unaware sockets are also 214 * disallowed. 215 * 216 * Returns 0 if the packet is acceptable, -1 if it is not. 217 */ 218 static __inline 219 int 220 check_multicast_membership(struct ip *ip, struct inpcb *inp, struct mbuf *m) 221 { 222 int mshipno; 223 struct ip_moptions *mopt; 224 225 if (strict_mcast_mship == 0 || 226 !IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 227 return(0); 228 } 229 mopt = inp->inp_moptions; 230 if (mopt == NULL) 231 return(-1); 232 for (mshipno = 0; mshipno <= mopt->imo_num_memberships; ++mshipno) { 233 if (ip->ip_dst.s_addr == mopt->imo_membership[mshipno]->inm_addr.s_addr && 234 m->m_pkthdr.rcvif == mopt->imo_membership[mshipno]->inm_ifp) { 235 return(0); 236 } 237 } 238 return(-1); 239 } 240 241 void 242 udp_input(struct mbuf *m, ...) 243 { 244 int iphlen; 245 struct ip *ip; 246 struct udphdr *uh; 247 struct inpcb *inp; 248 struct mbuf *opts = 0; 249 int len, off, proto; 250 struct ip save_ip; 251 struct sockaddr *append_sa; 252 __va_list ap; 253 254 __va_start(ap, m); 255 off = __va_arg(ap, int); 256 proto = __va_arg(ap, int); 257 __va_end(ap); 258 259 iphlen = off; 260 udpstat.udps_ipackets++; 261 262 /* 263 * Strip IP options, if any; should skip this, 264 * make available to user, and use on returned packets, 265 * but we don't yet have a way to check the checksum 266 * with options still present. 267 */ 268 if (iphlen > sizeof (struct ip)) { 269 ip_stripoptions(m); 270 iphlen = sizeof(struct ip); 271 } 272 273 /* 274 * IP and UDP headers are together in first mbuf. 275 * Already checked and pulled up in ip_demux(). 276 */ 277 KASSERT(m->m_len >= iphlen + sizeof(struct udphdr), 278 ("UDP header not in one mbuf")); 279 280 ip = mtod(m, struct ip *); 281 uh = (struct udphdr *)((caddr_t)ip + iphlen); 282 283 /* destination port of 0 is illegal, based on RFC768. */ 284 if (uh->uh_dport == 0) 285 goto bad; 286 287 /* 288 * Make mbuf data length reflect UDP length. 289 * If not enough data to reflect UDP length, drop. 290 */ 291 len = ntohs((u_short)uh->uh_ulen); 292 if (ip->ip_len != len) { 293 if (len > ip->ip_len || len < sizeof(struct udphdr)) { 294 udpstat.udps_badlen++; 295 goto bad; 296 } 297 m_adj(m, len - ip->ip_len); 298 /* ip->ip_len = len; */ 299 } 300 /* 301 * Save a copy of the IP header in case we want restore it 302 * for sending an ICMP error message in response. 303 */ 304 save_ip = *ip; 305 306 /* 307 * Checksum extended UDP header and data. 308 */ 309 if (uh->uh_sum) { 310 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { 311 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) 312 uh->uh_sum = m->m_pkthdr.csum_data; 313 else 314 uh->uh_sum = in_pseudo(ip->ip_src.s_addr, 315 ip->ip_dst.s_addr, htonl((u_short)len + 316 m->m_pkthdr.csum_data + IPPROTO_UDP)); 317 uh->uh_sum ^= 0xffff; 318 } else { 319 char b[9]; 320 bcopy(((struct ipovly *)ip)->ih_x1, b, 9); 321 bzero(((struct ipovly *)ip)->ih_x1, 9); 322 ((struct ipovly *)ip)->ih_len = uh->uh_ulen; 323 uh->uh_sum = in_cksum(m, len + sizeof (struct ip)); 324 bcopy(b, ((struct ipovly *)ip)->ih_x1, 9); 325 } 326 if (uh->uh_sum) { 327 udpstat.udps_badsum++; 328 m_freem(m); 329 return; 330 } 331 } else 332 udpstat.udps_nosum++; 333 334 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || 335 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) { 336 struct inpcb *last; 337 /* 338 * Deliver a multicast or broadcast datagram to *all* sockets 339 * for which the local and remote addresses and ports match 340 * those of the incoming datagram. This allows more than 341 * one process to receive multi/broadcasts on the same port. 342 * (This really ought to be done for unicast datagrams as 343 * well, but that would cause problems with existing 344 * applications that open both address-specific sockets and 345 * a wildcard socket listening to the same port -- they would 346 * end up receiving duplicates of every unicast datagram. 347 * Those applications open the multiple sockets to overcome an 348 * inadequacy of the UDP socket interface, but for backwards 349 * compatibility we avoid the problem here rather than 350 * fixing the interface. Maybe 4.5BSD will remedy this?) 351 */ 352 353 /* 354 * Construct sockaddr format source address. 355 */ 356 udp_in.sin_port = uh->uh_sport; 357 udp_in.sin_addr = ip->ip_src; 358 /* 359 * Locate pcb(s) for datagram. 360 * (Algorithm copied from raw_intr().) 361 */ 362 last = NULL; 363 #ifdef INET6 364 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0; 365 #endif 366 LIST_FOREACH(inp, &udbinfo.pcblisthead, inp_list) { 367 if (inp->inp_flags & INP_PLACEMARKER) 368 continue; 369 #ifdef INET6 370 if ((inp->inp_vflag & INP_IPV4) == 0) 371 continue; 372 #endif 373 if (inp->inp_lport != uh->uh_dport) 374 continue; 375 if (inp->inp_laddr.s_addr != INADDR_ANY) { 376 if (inp->inp_laddr.s_addr != 377 ip->ip_dst.s_addr) 378 continue; 379 } 380 if (inp->inp_faddr.s_addr != INADDR_ANY) { 381 if (inp->inp_faddr.s_addr != 382 ip->ip_src.s_addr || 383 inp->inp_fport != uh->uh_sport) 384 continue; 385 } 386 387 if (check_multicast_membership(ip, inp, m) < 0) 388 continue; 389 390 if (last != NULL) { 391 struct mbuf *n; 392 393 #ifdef IPSEC 394 /* check AH/ESP integrity. */ 395 if (ipsec4_in_reject_so(m, last->inp_socket)) 396 ipsecstat.in_polvio++; 397 /* do not inject data to pcb */ 398 else 399 #endif /*IPSEC*/ 400 #ifdef FAST_IPSEC 401 /* check AH/ESP integrity. */ 402 if (ipsec4_in_reject(m, last)) 403 ; 404 else 405 #endif /*FAST_IPSEC*/ 406 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) 407 udp_append(last, ip, n, 408 iphlen + 409 sizeof(struct udphdr)); 410 } 411 last = inp; 412 /* 413 * Don't look for additional matches if this one does 414 * not have either the SO_REUSEPORT or SO_REUSEADDR 415 * socket options set. This heuristic avoids searching 416 * through all pcbs in the common case of a non-shared 417 * port. It * assumes that an application will never 418 * clear these options after setting them. 419 */ 420 if ((last->inp_socket->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0) 421 break; 422 } 423 424 if (last == NULL) { 425 /* 426 * No matching pcb found; discard datagram. 427 * (No need to send an ICMP Port Unreachable 428 * for a broadcast or multicast datgram.) 429 */ 430 udpstat.udps_noportbcast++; 431 goto bad; 432 } 433 #ifdef IPSEC 434 /* check AH/ESP integrity. */ 435 if (ipsec4_in_reject_so(m, last->inp_socket)) { 436 ipsecstat.in_polvio++; 437 goto bad; 438 } 439 #endif /*IPSEC*/ 440 #ifdef FAST_IPSEC 441 /* check AH/ESP integrity. */ 442 if (ipsec4_in_reject(m, last)) 443 goto bad; 444 #endif /*FAST_IPSEC*/ 445 udp_append(last, ip, m, iphlen + sizeof(struct udphdr)); 446 return; 447 } 448 /* 449 * Locate pcb for datagram. 450 */ 451 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport, 452 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif); 453 if (inp == NULL) { 454 if (log_in_vain) { 455 char buf[4*sizeof "123"]; 456 457 strcpy(buf, inet_ntoa(ip->ip_dst)); 458 log(LOG_INFO, 459 "Connection attempt to UDP %s:%d from %s:%d\n", 460 buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src), 461 ntohs(uh->uh_sport)); 462 } 463 udpstat.udps_noport++; 464 if (m->m_flags & (M_BCAST | M_MCAST)) { 465 udpstat.udps_noportbcast++; 466 goto bad; 467 } 468 if (blackhole) 469 goto bad; 470 #ifdef ICMP_BANDLIM 471 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0) 472 goto bad; 473 #endif 474 *ip = save_ip; 475 ip->ip_len += iphlen; 476 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); 477 return; 478 } 479 #ifdef IPSEC 480 if (ipsec4_in_reject_so(m, inp->inp_socket)) { 481 ipsecstat.in_polvio++; 482 goto bad; 483 } 484 #endif /*IPSEC*/ 485 #ifdef FAST_IPSEC 486 if (ipsec4_in_reject(m, inp)) 487 goto bad; 488 #endif /*FAST_IPSEC*/ 489 490 /* 491 * Construct sockaddr format source address. 492 * Stuff source address and datagram in user buffer. 493 */ 494 udp_in.sin_port = uh->uh_sport; 495 udp_in.sin_addr = ip->ip_src; 496 if (inp->inp_flags & INP_CONTROLOPTS 497 || inp->inp_socket->so_options & SO_TIMESTAMP) { 498 #ifdef INET6 499 if (inp->inp_vflag & INP_IPV6) { 500 int savedflags; 501 502 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip); 503 savedflags = inp->inp_flags; 504 inp->inp_flags &= ~INP_UNMAPPABLEOPTS; 505 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m); 506 inp->inp_flags = savedflags; 507 } else 508 #endif 509 ip_savecontrol(inp, &opts, ip, m); 510 } 511 m_adj(m, iphlen + sizeof(struct udphdr)); 512 #ifdef INET6 513 if (inp->inp_vflag & INP_IPV6) { 514 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin); 515 append_sa = (struct sockaddr *)&udp_in6; 516 } else 517 #endif 518 append_sa = (struct sockaddr *)&udp_in; 519 if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) { 520 udpstat.udps_fullsock++; 521 goto bad; 522 } 523 sorwakeup(inp->inp_socket); 524 return; 525 bad: 526 m_freem(m); 527 if (opts) 528 m_freem(opts); 529 return; 530 } 531 532 #ifdef INET6 533 static void 534 ip_2_ip6_hdr(ip6, ip) 535 struct ip6_hdr *ip6; 536 struct ip *ip; 537 { 538 bzero(ip6, sizeof(*ip6)); 539 540 ip6->ip6_vfc = IPV6_VERSION; 541 ip6->ip6_plen = ip->ip_len; 542 ip6->ip6_nxt = ip->ip_p; 543 ip6->ip6_hlim = ip->ip_ttl; 544 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] = 545 IPV6_ADDR_INT32_SMP; 546 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr; 547 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr; 548 } 549 #endif 550 551 /* 552 * subroutine of udp_input(), mainly for source code readability. 553 * caller must properly init udp_ip6 and udp_in6 beforehand. 554 */ 555 static void 556 udp_append(last, ip, n, off) 557 struct inpcb *last; 558 struct ip *ip; 559 struct mbuf *n; 560 int off; 561 { 562 struct sockaddr *append_sa; 563 struct mbuf *opts = 0; 564 565 if (last->inp_flags & INP_CONTROLOPTS || 566 last->inp_socket->so_options & SO_TIMESTAMP) { 567 #ifdef INET6 568 if (last->inp_vflag & INP_IPV6) { 569 int savedflags; 570 571 if (udp_ip6.uip6_init_done == 0) { 572 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip); 573 udp_ip6.uip6_init_done = 1; 574 } 575 savedflags = last->inp_flags; 576 last->inp_flags &= ~INP_UNMAPPABLEOPTS; 577 ip6_savecontrol(last, &opts, &udp_ip6.uip6_ip6, n); 578 last->inp_flags = savedflags; 579 } else 580 #endif 581 ip_savecontrol(last, &opts, ip, n); 582 } 583 #ifdef INET6 584 if (last->inp_vflag & INP_IPV6) { 585 if (udp_in6.uin6_init_done == 0) { 586 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin); 587 udp_in6.uin6_init_done = 1; 588 } 589 append_sa = (struct sockaddr *)&udp_in6.uin6_sin; 590 } else 591 #endif 592 append_sa = (struct sockaddr *)&udp_in; 593 m_adj(n, off); 594 if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) { 595 m_freem(n); 596 if (opts) 597 m_freem(opts); 598 udpstat.udps_fullsock++; 599 } else 600 sorwakeup(last->inp_socket); 601 } 602 603 /* 604 * Notify a udp user of an asynchronous error; 605 * just wake up so that he can collect error status. 606 */ 607 void 608 udp_notify(inp, errno) 609 struct inpcb *inp; 610 int errno; 611 { 612 inp->inp_socket->so_error = errno; 613 sorwakeup(inp->inp_socket); 614 sowwakeup(inp->inp_socket); 615 } 616 617 void 618 udp_ctlinput(cmd, sa, vip) 619 int cmd; 620 struct sockaddr *sa; 621 void *vip; 622 { 623 struct ip *ip = vip; 624 struct udphdr *uh; 625 void (*notify) (struct inpcb *, int) = udp_notify; 626 struct in_addr faddr; 627 struct inpcb *inp; 628 int s; 629 630 faddr = ((struct sockaddr_in *)sa)->sin_addr; 631 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) 632 return; 633 634 if (PRC_IS_REDIRECT(cmd)) { 635 ip = 0; 636 notify = in_rtchange; 637 } else if (cmd == PRC_HOSTDEAD) 638 ip = 0; 639 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) 640 return; 641 if (ip) { 642 s = splnet(); 643 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 644 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport, 645 ip->ip_src, uh->uh_sport, 0, NULL); 646 if (inp != NULL && inp->inp_socket != NULL) 647 (*notify)(inp, inetctlerrmap[cmd]); 648 splx(s); 649 } else 650 in_pcbnotifyall(&udbinfo.pcblisthead, faddr, inetctlerrmap[cmd], 651 notify); 652 } 653 654 static int 655 udp_pcblist(SYSCTL_HANDLER_ARGS) 656 { 657 int error, i, n; 658 struct inpcb *inp; 659 struct inpcb *marker; 660 inp_gen_t gencnt; 661 struct xinpgen xig; 662 struct xinpcb xi; 663 664 /* 665 * The process of preparing the TCB list is too time-consuming and 666 * resource-intensive to repeat twice on every request. 667 */ 668 if (req->oldptr == 0) { 669 n = udbinfo.ipi_count; 670 req->oldidx = 2 * (sizeof xig) 671 + (n + n/8) * sizeof(struct xinpcb); 672 return 0; 673 } 674 675 if (req->newptr != 0) 676 return EPERM; 677 678 /* 679 * OK, now we're committed to doing something. 680 */ 681 gencnt = udbinfo.ipi_gencnt; 682 n = udbinfo.ipi_count; 683 684 xig.xig_len = sizeof xig; 685 xig.xig_count = n; 686 xig.xig_gen = gencnt; 687 xig.xig_sogen = so_gencnt; 688 xig.xig_cpu = 0; 689 error = SYSCTL_OUT(req, &xig, sizeof xig); 690 if (error) 691 return error; 692 693 marker = malloc(sizeof(struct inpcb), M_TEMP, M_WAITOK|M_ZERO); 694 marker->inp_flags |= INP_PLACEMARKER; 695 696 LIST_INSERT_HEAD(&udbinfo.pcblisthead, marker, inp_list); 697 i = 0; 698 while ((inp = LIST_NEXT(marker, inp_list)) != NULL && i < n) { 699 LIST_REMOVE(marker, inp_list); 700 LIST_INSERT_AFTER(inp, marker, inp_list); 701 if (inp->inp_flags & INP_PLACEMARKER) 702 continue; 703 if (inp->inp_gencnt > gencnt) 704 continue; 705 if (prison_xinpcb(req->td, inp)) 706 continue; 707 xi.xi_len = sizeof xi; 708 bcopy(inp, &xi.xi_inp, sizeof *inp); 709 if (inp->inp_socket) 710 sotoxsocket(inp->inp_socket, &xi.xi_socket); 711 if ((error = SYSCTL_OUT(req, &xi, sizeof xi)) != 0) 712 break; 713 ++i; 714 } 715 LIST_REMOVE(marker, inp_list); 716 if (error == 0 && i < n) { 717 bzero(&xi, sizeof(xi)); 718 xi.xi_len = sizeof(xi); 719 while (i < n) { 720 if ((error = SYSCTL_OUT(req, &xi, sizeof(xi))) != 0) 721 break; 722 ++i; 723 } 724 } 725 if (error == 0) { 726 /* 727 * Give the user an updated idea of our state. 728 * If the generation differs from what we told 729 * her before, she knows that something happened 730 * while we were processing this request, and it 731 * might be necessary to retry. 732 */ 733 xig.xig_gen = udbinfo.ipi_gencnt; 734 xig.xig_sogen = so_gencnt; 735 xig.xig_count = udbinfo.ipi_count; 736 error = SYSCTL_OUT(req, &xig, sizeof xig); 737 } 738 free(marker, M_TEMP); 739 return error; 740 } 741 742 SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0, 743 udp_pcblist, "S,xinpcb", "List of active UDP sockets"); 744 745 static int 746 udp_getcred(SYSCTL_HANDLER_ARGS) 747 { 748 struct sockaddr_in addrs[2]; 749 struct inpcb *inp; 750 int error, s; 751 752 error = suser(req->td); 753 if (error) 754 return (error); 755 error = SYSCTL_IN(req, addrs, sizeof(addrs)); 756 if (error) 757 return (error); 758 s = splnet(); 759 inp = in_pcblookup_hash(&udbinfo, addrs[1].sin_addr, addrs[1].sin_port, 760 addrs[0].sin_addr, addrs[0].sin_port, 1, NULL); 761 if (inp == NULL || inp->inp_socket == NULL) { 762 error = ENOENT; 763 goto out; 764 } 765 error = SYSCTL_OUT(req, inp->inp_socket->so_cred, sizeof(struct ucred)); 766 out: 767 splx(s); 768 return (error); 769 } 770 771 SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW, 772 0, 0, udp_getcred, "S,ucred", "Get the ucred of a UDP connection"); 773 774 static int 775 udp_output(inp, m, dstaddr, control, td) 776 struct inpcb *inp; 777 struct mbuf *m; 778 struct sockaddr *dstaddr; 779 struct mbuf *control; 780 struct thread *td; 781 { 782 struct udpiphdr *ui; 783 int len = m->m_pkthdr.len; 784 struct sockaddr_in *sin; /* really is initialized before use */ 785 int error = 0; 786 787 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) { 788 error = EMSGSIZE; 789 goto release; 790 } 791 792 if (inp->inp_lport == 0) { /* unbound socket */ 793 error = in_pcbbind(inp, (struct sockaddr *)NULL, td); 794 if (error) 795 goto release; 796 in_pcbinswildcardhash(inp); 797 } 798 799 if (dstaddr != NULL) { /* destination address specified */ 800 if (inp->inp_faddr.s_addr != INADDR_ANY) { 801 /* already connected */ 802 error = EISCONN; 803 goto release; 804 } 805 sin = (struct sockaddr_in *)dstaddr; 806 prison_remote_ip(td, 0, &sin->sin_addr.s_addr); 807 } else { 808 if (inp->inp_faddr.s_addr == INADDR_ANY) { 809 /* no destination specified and not already connected */ 810 error = ENOTCONN; 811 goto release; 812 } 813 sin = NULL; 814 } 815 816 /* 817 * Calculate data length and get a mbuf 818 * for UDP and IP headers. 819 */ 820 M_PREPEND(m, sizeof(struct udpiphdr), MB_DONTWAIT); 821 if (m == 0) { 822 error = ENOBUFS; 823 goto release; 824 } 825 826 /* 827 * Fill in mbuf with extended UDP header 828 * and addresses and length put into network format. 829 */ 830 ui = mtod(m, struct udpiphdr *); 831 bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */ 832 ui->ui_pr = IPPROTO_UDP; 833 834 /* 835 * Set destination address. 836 */ 837 if (dstaddr != NULL) { /* use specified destination */ 838 ui->ui_dst = sin->sin_addr; 839 ui->ui_dport = sin->sin_port; 840 } else { /* use connected destination */ 841 ui->ui_dst = inp->inp_faddr; 842 ui->ui_dport = inp->inp_fport; 843 } 844 845 /* 846 * Set source address. 847 */ 848 if (inp->inp_laddr.s_addr == INADDR_ANY) { 849 struct sockaddr_in *if_sin; 850 851 KASSERT(dstaddr != NULL, 852 ("connected UDP socket without local addr: " 853 "lport %d, faddr %x, fport %d", 854 inp->inp_lport, inp->inp_faddr.s_addr, inp->inp_fport)); 855 856 /* Look up outgoing interface. */ 857 if ((error = in_pcbladdr(inp, dstaddr, &if_sin))) 858 goto release; 859 ui->ui_src = if_sin->sin_addr; /* use address of interface */ 860 } else { 861 ui->ui_src = inp->inp_laddr; /* use non-null bound address */ 862 } 863 ui->ui_sport = inp->inp_lport; 864 KASSERT(inp->inp_lport != 0, ("inp lport should have been bound")); 865 866 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr)); 867 868 /* 869 * Set up checksum and output datagram. 870 */ 871 if (udpcksum) { 872 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr, 873 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP)); 874 m->m_pkthdr.csum_flags = CSUM_UDP; 875 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum); 876 } else { 877 ui->ui_sum = 0; 878 } 879 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len; 880 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */ 881 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */ 882 udpstat.udps_opackets++; 883 884 error = ip_output(m, inp->inp_options, &inp->inp_route, 885 (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST)), 886 inp->inp_moptions, inp); 887 888 return (error); 889 890 release: 891 m_freem(m); 892 return (error); 893 } 894 895 u_long udp_sendspace = 9216; /* really max datagram size */ 896 /* 40 1K datagrams */ 897 SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW, 898 &udp_sendspace, 0, "Maximum outgoing UDP datagram size"); 899 900 u_long udp_recvspace = 40 * (1024 + 901 #ifdef INET6 902 sizeof(struct sockaddr_in6) 903 #else 904 sizeof(struct sockaddr_in) 905 #endif 906 ); 907 SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW, 908 &udp_recvspace, 0, "Maximum incoming UDP datagram size"); 909 910 static int 911 udp_abort(struct socket *so) 912 { 913 struct inpcb *inp; 914 int s; 915 916 inp = sotoinpcb(so); 917 if (inp == 0) 918 return EINVAL; /* ??? possible? panic instead? */ 919 soisdisconnected(so); 920 s = splnet(); 921 in_pcbdetach(inp); 922 splx(s); 923 return 0; 924 } 925 926 static int 927 udp_attach(struct socket *so, int proto, struct pru_attach_info *ai) 928 { 929 struct inpcb *inp; 930 int s, error; 931 932 inp = sotoinpcb(so); 933 if (inp != 0) 934 return EINVAL; 935 936 error = soreserve(so, udp_sendspace, udp_recvspace, ai->sb_rlimit); 937 if (error) 938 return error; 939 s = splnet(); 940 error = in_pcballoc(so, &udbinfo); 941 splx(s); 942 if (error) 943 return error; 944 945 inp = (struct inpcb *)so->so_pcb; 946 inp->inp_vflag |= INP_IPV4; 947 inp->inp_ip_ttl = ip_defttl; 948 return 0; 949 } 950 951 static int 952 udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td) 953 { 954 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 955 struct inpcb *inp; 956 int s, error; 957 958 inp = sotoinpcb(so); 959 if (inp == 0) 960 return EINVAL; 961 s = splnet(); 962 error = in_pcbbind(inp, nam, td); 963 splx(s); 964 if (error == 0) { 965 if (sin->sin_addr.s_addr != INADDR_ANY) 966 inp->inp_flags |= INP_WASBOUND_NOTANY; 967 in_pcbinswildcardhash(inp); 968 } 969 return error; 970 } 971 972 static int 973 udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td) 974 { 975 struct inpcb *inp; 976 int s, error; 977 struct sockaddr_in *sin; 978 979 inp = sotoinpcb(so); 980 if (inp == 0) 981 return EINVAL; 982 if (inp->inp_faddr.s_addr != INADDR_ANY) 983 return EISCONN; 984 error = 0; 985 s = splnet(); 986 if (td->td_proc && td->td_proc->p_ucred->cr_prison != NULL && 987 inp->inp_laddr.s_addr == INADDR_ANY) { 988 error = in_pcbbind(inp, NULL, td); 989 } 990 if (error == 0) { 991 sin = (struct sockaddr_in *)nam; 992 prison_remote_ip(td, 0, &sin->sin_addr.s_addr); 993 if (inp->inp_flags & INP_WILDCARD) 994 in_pcbremwildcardhash(inp); 995 error = in_pcbconnect(inp, nam, td); 996 } 997 splx(s); 998 if (error == 0) 999 soisconnected(so); 1000 else if (error == EAFNOSUPPORT) { /* connection dissolved */ 1001 /* 1002 * Follow traditional BSD behavior and retain 1003 * the local port binding. But, fix the old misbehavior 1004 * of overwriting any previously bound local address. 1005 */ 1006 if (!(inp->inp_flags & INP_WASBOUND_NOTANY)) 1007 inp->inp_laddr.s_addr = INADDR_ANY; 1008 in_pcbinswildcardhash(inp); 1009 } 1010 return error; 1011 } 1012 1013 static int 1014 udp_detach(struct socket *so) 1015 { 1016 struct inpcb *inp; 1017 int s; 1018 1019 inp = sotoinpcb(so); 1020 if (inp == 0) 1021 return EINVAL; 1022 s = splnet(); 1023 in_pcbdetach(inp); 1024 splx(s); 1025 return 0; 1026 } 1027 1028 static int 1029 udp_disconnect(struct socket *so) 1030 { 1031 struct inpcb *inp; 1032 int s; 1033 1034 inp = sotoinpcb(so); 1035 if (inp == 0) 1036 return EINVAL; 1037 if (inp->inp_faddr.s_addr == INADDR_ANY) 1038 return ENOTCONN; 1039 1040 s = splnet(); 1041 in_pcbdisconnect(inp); 1042 splx(s); 1043 so->so_state &= ~SS_ISCONNECTED; /* XXX */ 1044 return 0; 1045 } 1046 1047 static int 1048 udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, 1049 struct mbuf *control, struct thread *td) 1050 { 1051 struct inpcb *inp; 1052 1053 inp = sotoinpcb(so); 1054 if (inp == 0) { 1055 m_freem(m); 1056 return EINVAL; 1057 } 1058 return udp_output(inp, m, addr, control, td); 1059 } 1060 1061 int 1062 udp_shutdown(struct socket *so) 1063 { 1064 struct inpcb *inp; 1065 1066 inp = sotoinpcb(so); 1067 if (inp == 0) 1068 return EINVAL; 1069 socantsendmore(so); 1070 return 0; 1071 } 1072 1073 struct pr_usrreqs udp_usrreqs = { 1074 udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect, 1075 pru_connect2_notsupp, in_control, udp_detach, udp_disconnect, 1076 pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp, 1077 pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown, 1078 in_setsockaddr, sosendudp, soreceive, sopoll 1079 }; 1080 1081