1 /* $NetBSD: udp_usrreq.c,v 1.47 1999/01/19 23:03:22 mycroft Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 36 */ 37 #include "ipkdb.h" 38 39 #include <sys/param.h> 40 #include <sys/malloc.h> 41 #include <sys/mbuf.h> 42 #include <sys/protosw.h> 43 #include <sys/socket.h> 44 #include <sys/socketvar.h> 45 #include <sys/errno.h> 46 #include <sys/stat.h> 47 #include <sys/systm.h> 48 #include <sys/proc.h> 49 50 #include <vm/vm.h> 51 #include <sys/sysctl.h> 52 53 #include <net/if.h> 54 #include <net/route.h> 55 56 #include <netinet/in.h> 57 #include <netinet/in_systm.h> 58 #include <netinet/in_var.h> 59 #include <netinet/ip.h> 60 #include <netinet/in_pcb.h> 61 #include <netinet/ip_var.h> 62 #include <netinet/ip_icmp.h> 63 #include <netinet/udp.h> 64 #include <netinet/udp_var.h> 65 66 #include <machine/stdarg.h> 67 68 /* 69 * UDP protocol implementation. 70 * Per RFC 768, August, 1980. 71 */ 72 #ifndef COMPAT_42 73 int udpcksum = 1; 74 #else 75 int udpcksum = 0; /* XXX */ 76 #endif 77 78 static void udp_notify __P((struct inpcb *, int)); 79 80 #ifndef UDBHASHSIZE 81 #define UDBHASHSIZE 128 82 #endif 83 int udbhashsize = UDBHASHSIZE; 84 85 void 86 udp_init() 87 { 88 89 in_pcbinit(&udbtable, udbhashsize, udbhashsize); 90 } 91 92 void 93 #if __STDC__ 94 udp_input(struct mbuf *m, ...) 95 #else 96 udp_input(m, va_alist) 97 struct mbuf *m; 98 va_dcl 99 #endif 100 { 101 register struct ip *ip; 102 register struct udphdr *uh; 103 register struct inpcb *inp; 104 struct mbuf *opts = 0; 105 int len; 106 struct ip save_ip; 107 int iphlen; 108 va_list ap; 109 struct sockaddr_in udpsrc; 110 111 va_start(ap, m); 112 iphlen = va_arg(ap, int); 113 va_end(ap); 114 115 udpstat.udps_ipackets++; 116 117 /* 118 * Strip IP options, if any; should skip this, 119 * make available to user, and use on returned packets, 120 * but we don't yet have a way to check the checksum 121 * with options still present. 122 */ 123 if (iphlen > sizeof (struct ip)) { 124 ip_stripoptions(m, (struct mbuf *)0); 125 iphlen = sizeof(struct ip); 126 } 127 128 /* 129 * Get IP and UDP header together in first mbuf. 130 */ 131 ip = mtod(m, struct ip *); 132 if (m->m_len < iphlen + sizeof(struct udphdr)) { 133 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) { 134 udpstat.udps_hdrops++; 135 return; 136 } 137 ip = mtod(m, struct ip *); 138 } 139 uh = (struct udphdr *)((caddr_t)ip + iphlen); 140 141 /* 142 * Make mbuf data length reflect UDP length. 143 * If not enough data to reflect UDP length, drop. 144 */ 145 len = ntohs((u_int16_t)uh->uh_ulen); 146 if (ip->ip_len != iphlen + len) { 147 if (ip->ip_len < iphlen + len) { 148 udpstat.udps_badlen++; 149 goto bad; 150 } 151 m_adj(m, iphlen + len - ip->ip_len); 152 } 153 /* 154 * Save a copy of the IP header in case we want restore it 155 * for sending an ICMP error message in response. 156 */ 157 save_ip = *ip; 158 159 /* 160 * Checksum extended UDP header and data. 161 */ 162 if (uh->uh_sum) { 163 bzero(((struct ipovly *)ip)->ih_x1, 164 sizeof ((struct ipovly *)ip)->ih_x1); 165 ((struct ipovly *)ip)->ih_len = uh->uh_ulen; 166 if (in_cksum(m, len + sizeof (struct ip)) != 0) { 167 udpstat.udps_badsum++; 168 m_freem(m); 169 return; 170 } 171 } 172 173 if (IN_MULTICAST(ip->ip_dst.s_addr) || 174 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) { 175 struct inpcb *last; 176 /* 177 * Deliver a multicast or broadcast datagram to *all* sockets 178 * for which the local and remote addresses and ports match 179 * those of the incoming datagram. This allows more than 180 * one process to receive multi/broadcasts on the same port. 181 * (This really ought to be done for unicast datagrams as 182 * well, but that would cause problems with existing 183 * applications that open both address-specific sockets and 184 * a wildcard socket listening to the same port -- they would 185 * end up receiving duplicates of every unicast datagram. 186 * Those applications open the multiple sockets to overcome an 187 * inadequacy of the UDP socket interface, but for backwards 188 * compatibility we avoid the problem here rather than 189 * fixing the interface. Maybe 4.5BSD will remedy this?) 190 */ 191 192 /* 193 * Construct sockaddr format source address. 194 */ 195 udpsrc.sin_family = AF_INET; 196 udpsrc.sin_len = sizeof(struct sockaddr_in); 197 udpsrc.sin_addr = ip->ip_src; 198 udpsrc.sin_port = uh->uh_sport; 199 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero)); 200 201 iphlen += sizeof(struct udphdr); 202 m->m_len -= iphlen; 203 m->m_pkthdr.len -= iphlen; 204 m->m_data += iphlen; 205 /* 206 * Locate pcb(s) for datagram. 207 * (Algorithm copied from raw_intr().) 208 */ 209 last = NULL; 210 for (inp = udbtable.inpt_queue.cqh_first; 211 inp != (struct inpcb *)&udbtable.inpt_queue; 212 inp = inp->inp_queue.cqe_next) { 213 if (inp->inp_lport != uh->uh_dport) 214 continue; 215 if (!in_nullhost(inp->inp_laddr)) { 216 if (!in_hosteq(inp->inp_laddr, ip->ip_dst)) 217 continue; 218 } 219 if (!in_nullhost(inp->inp_faddr)) { 220 if (!in_hosteq(inp->inp_faddr, ip->ip_src) || 221 inp->inp_fport != uh->uh_sport) 222 continue; 223 } 224 225 if (last != NULL) { 226 struct mbuf *n; 227 228 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) { 229 if (last->inp_flags & INP_CONTROLOPTS 230 || last->inp_socket->so_options & 231 SO_TIMESTAMP) { 232 ip_savecontrol(last, &opts, 233 ip, n); 234 } 235 if (sbappendaddr( 236 &last->inp_socket->so_rcv, 237 sintosa(&udpsrc), n, opts) == 0) { 238 m_freem(n); 239 if (opts) 240 m_freem(opts); 241 } else 242 sorwakeup(last->inp_socket); 243 opts = 0; 244 } 245 } 246 last = inp; 247 /* 248 * Don't look for additional matches if this one does 249 * not have either the SO_REUSEPORT or SO_REUSEADDR 250 * socket options set. This heuristic avoids searching 251 * through all pcbs in the common case of a non-shared 252 * port. It * assumes that an application will never 253 * clear these options after setting them. 254 */ 255 if ((last->inp_socket->so_options & 256 (SO_REUSEPORT|SO_REUSEADDR)) == 0) 257 break; 258 } 259 260 if (last == NULL) { 261 /* 262 * No matching pcb found; discard datagram. 263 * (No need to send an ICMP Port Unreachable 264 * for a broadcast or multicast datgram.) 265 */ 266 udpstat.udps_noportbcast++; 267 goto bad; 268 } 269 if (last->inp_flags & INP_CONTROLOPTS || 270 last->inp_socket->so_options & SO_TIMESTAMP) 271 ip_savecontrol(last, &opts, ip, m); 272 if (sbappendaddr(&last->inp_socket->so_rcv, 273 sintosa(&udpsrc), m, opts) == 0) { 274 udpstat.udps_fullsock++; 275 goto bad; 276 } 277 sorwakeup(last->inp_socket); 278 return; 279 } 280 /* 281 * Locate pcb for datagram. 282 */ 283 inp = in_pcblookup_connect(&udbtable, ip->ip_src, uh->uh_sport, 284 ip->ip_dst, uh->uh_dport); 285 if (inp == 0) { 286 ++udpstat.udps_pcbhashmiss; 287 inp = in_pcblookup_bind(&udbtable, ip->ip_dst, uh->uh_dport); 288 if (inp == 0) { 289 udpstat.udps_noport++; 290 if (m->m_flags & (M_BCAST | M_MCAST)) { 291 udpstat.udps_noportbcast++; 292 goto bad; 293 } 294 *ip = save_ip; 295 #if NIPKDB > 0 296 if (checkipkdb(&ip->ip_src, 297 uh->uh_sport, 298 uh->uh_dport, 299 m, 300 iphlen + sizeof(struct udphdr), 301 len - sizeof(struct udphdr))) 302 /* It was a debugger connect packet, just drop it now */ 303 goto bad; 304 #endif 305 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); 306 return; 307 } 308 } 309 310 /* 311 * Construct sockaddr format source address. 312 * Stuff source address and datagram in user buffer. 313 */ 314 udpsrc.sin_family = AF_INET; 315 udpsrc.sin_len = sizeof(struct sockaddr_in); 316 udpsrc.sin_addr = ip->ip_src; 317 udpsrc.sin_port = uh->uh_sport; 318 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero)); 319 320 if (inp->inp_flags & INP_CONTROLOPTS || 321 inp->inp_socket->so_options & SO_TIMESTAMP) 322 ip_savecontrol(inp, &opts, ip, m); 323 iphlen += sizeof(struct udphdr); 324 m->m_len -= iphlen; 325 m->m_pkthdr.len -= iphlen; 326 m->m_data += iphlen; 327 if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udpsrc), m, 328 opts) == 0) { 329 udpstat.udps_fullsock++; 330 goto bad; 331 } 332 sorwakeup(inp->inp_socket); 333 return; 334 bad: 335 m_freem(m); 336 if (opts) 337 m_freem(opts); 338 } 339 340 /* 341 * Notify a udp user of an asynchronous error; 342 * just wake up so that he can collect error status. 343 */ 344 static void 345 udp_notify(inp, errno) 346 register struct inpcb *inp; 347 int errno; 348 { 349 350 inp->inp_socket->so_error = errno; 351 sorwakeup(inp->inp_socket); 352 sowwakeup(inp->inp_socket); 353 } 354 355 void * 356 udp_ctlinput(cmd, sa, v) 357 int cmd; 358 struct sockaddr *sa; 359 void *v; 360 { 361 register struct ip *ip = v; 362 register struct udphdr *uh; 363 extern int inetctlerrmap[]; 364 void (*notify) __P((struct inpcb *, int)) = udp_notify; 365 int errno; 366 367 if ((unsigned)cmd >= PRC_NCMDS) 368 return NULL; 369 errno = inetctlerrmap[cmd]; 370 if (PRC_IS_REDIRECT(cmd)) 371 notify = in_rtchange, ip = 0; 372 else if (cmd == PRC_HOSTDEAD) 373 ip = 0; 374 else if (errno == 0) 375 return NULL; 376 if (ip) { 377 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 378 in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport, 379 ip->ip_src, uh->uh_sport, errno, notify); 380 } else 381 in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno, 382 notify); 383 return NULL; 384 } 385 386 int 387 #if __STDC__ 388 udp_output(struct mbuf *m, ...) 389 #else 390 udp_output(m, va_alist) 391 struct mbuf *m; 392 va_dcl 393 #endif 394 { 395 register struct inpcb *inp; 396 register struct udpiphdr *ui; 397 register int len = m->m_pkthdr.len; 398 int error = 0; 399 va_list ap; 400 401 va_start(ap, m); 402 inp = va_arg(ap, struct inpcb *); 403 va_end(ap); 404 405 /* 406 * Calculate data length and get a mbuf 407 * for UDP and IP headers. 408 */ 409 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT); 410 if (m == 0) { 411 error = ENOBUFS; 412 goto release; 413 } 414 415 /* 416 * Compute the packet length of the IP header, and 417 * punt if the length looks bogus. 418 */ 419 if ((len + sizeof(struct udpiphdr)) > IP_MAXPACKET) { 420 error = EMSGSIZE; 421 goto release; 422 } 423 424 /* 425 * Fill in mbuf with extended UDP header 426 * and addresses and length put into network format. 427 */ 428 ui = mtod(m, struct udpiphdr *); 429 bzero(ui->ui_x1, sizeof ui->ui_x1); 430 ui->ui_pr = IPPROTO_UDP; 431 ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr)); 432 ui->ui_src = inp->inp_laddr; 433 ui->ui_dst = inp->inp_faddr; 434 ui->ui_sport = inp->inp_lport; 435 ui->ui_dport = inp->inp_fport; 436 ui->ui_ulen = ui->ui_len; 437 438 /* 439 * Stuff checksum and output datagram. 440 */ 441 ui->ui_sum = 0; 442 if (udpcksum) { 443 if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0) 444 ui->ui_sum = 0xffff; 445 } 446 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len; 447 ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */ 448 ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */ 449 udpstat.udps_opackets++; 450 return (ip_output(m, inp->inp_options, &inp->inp_route, 451 inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST), 452 inp->inp_moptions)); 453 454 release: 455 m_freem(m); 456 return (error); 457 } 458 459 int udp_sendspace = 9216; /* really max datagram size */ 460 int udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in)); 461 /* 40 1K datagrams */ 462 463 /*ARGSUSED*/ 464 int 465 udp_usrreq(so, req, m, nam, control, p) 466 struct socket *so; 467 int req; 468 struct mbuf *m, *nam, *control; 469 struct proc *p; 470 { 471 register struct inpcb *inp; 472 int s; 473 register int error = 0; 474 475 if (req == PRU_CONTROL) 476 return (in_control(so, (long)m, (caddr_t)nam, 477 (struct ifnet *)control, p)); 478 479 s = splsoftnet(); 480 inp = sotoinpcb(so); 481 #ifdef DIAGNOSTIC 482 if (req != PRU_SEND && req != PRU_SENDOOB && control) 483 panic("udp_usrreq: unexpected control mbuf"); 484 #endif 485 if (inp == 0 && req != PRU_ATTACH) { 486 error = EINVAL; 487 goto release; 488 } 489 490 /* 491 * Note: need to block udp_input while changing 492 * the udp pcb queue and/or pcb addresses. 493 */ 494 switch (req) { 495 496 case PRU_ATTACH: 497 if (inp != 0) { 498 error = EISCONN; 499 break; 500 } 501 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 502 error = soreserve(so, udp_sendspace, udp_recvspace); 503 if (error) 504 break; 505 } 506 error = in_pcballoc(so, &udbtable); 507 if (error) 508 break; 509 inp = sotoinpcb(so); 510 inp->inp_ip.ip_ttl = ip_defttl; 511 break; 512 513 case PRU_DETACH: 514 in_pcbdetach(inp); 515 break; 516 517 case PRU_BIND: 518 error = in_pcbbind(inp, nam, p); 519 break; 520 521 case PRU_LISTEN: 522 error = EOPNOTSUPP; 523 break; 524 525 case PRU_CONNECT: 526 error = in_pcbconnect(inp, nam); 527 if (error) 528 break; 529 soisconnected(so); 530 break; 531 532 case PRU_CONNECT2: 533 error = EOPNOTSUPP; 534 break; 535 536 case PRU_DISCONNECT: 537 /*soisdisconnected(so);*/ 538 so->so_state &= ~SS_ISCONNECTED; /* XXX */ 539 in_pcbdisconnect(inp); 540 inp->inp_laddr = zeroin_addr; /* XXX */ 541 in_pcbstate(inp, INP_BOUND); /* XXX */ 542 break; 543 544 case PRU_SHUTDOWN: 545 socantsendmore(so); 546 break; 547 548 case PRU_RCVD: 549 error = EOPNOTSUPP; 550 break; 551 552 case PRU_SEND: 553 if (control && control->m_len) { 554 m_freem(control); 555 m_freem(m); 556 error = EINVAL; 557 break; 558 } 559 { 560 struct in_addr laddr; /* XXX */ 561 562 if (nam) { 563 laddr = inp->inp_laddr; /* XXX */ 564 if ((so->so_state & SS_ISCONNECTED) != 0) { 565 error = EISCONN; 566 goto die; 567 } 568 error = in_pcbconnect(inp, nam); 569 if (error) { 570 die: 571 m_freem(m); 572 break; 573 } 574 } else { 575 if ((so->so_state & SS_ISCONNECTED) == 0) { 576 error = ENOTCONN; 577 goto die; 578 } 579 } 580 error = udp_output(m, inp); 581 if (nam) { 582 in_pcbdisconnect(inp); 583 inp->inp_laddr = laddr; /* XXX */ 584 in_pcbstate(inp, INP_BOUND); /* XXX */ 585 } 586 } 587 break; 588 589 case PRU_SENSE: 590 /* 591 * stat: don't bother with a blocksize. 592 */ 593 splx(s); 594 return (0); 595 596 case PRU_RCVOOB: 597 error = EOPNOTSUPP; 598 break; 599 600 case PRU_SENDOOB: 601 m_freem(control); 602 m_freem(m); 603 error = EOPNOTSUPP; 604 break; 605 606 case PRU_SOCKADDR: 607 in_setsockaddr(inp, nam); 608 break; 609 610 case PRU_PEERADDR: 611 in_setpeeraddr(inp, nam); 612 break; 613 614 default: 615 panic("udp_usrreq"); 616 } 617 618 release: 619 splx(s); 620 return (error); 621 } 622 623 /* 624 * Sysctl for udp variables. 625 */ 626 int 627 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 628 int *name; 629 u_int namelen; 630 void *oldp; 631 size_t *oldlenp; 632 void *newp; 633 size_t newlen; 634 { 635 /* All sysctl names at this level are terminal. */ 636 if (namelen != 1) 637 return (ENOTDIR); 638 639 switch (name[0]) { 640 case UDPCTL_CHECKSUM: 641 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum)); 642 case UDPCTL_SENDSPACE: 643 return (sysctl_int(oldp, oldlenp, newp, newlen, 644 &udp_sendspace)); 645 case UDPCTL_RECVSPACE: 646 return (sysctl_int(oldp, oldlenp, newp, newlen, 647 &udp_recvspace)); 648 default: 649 return (ENOPROTOOPT); 650 } 651 /* NOTREACHED */ 652 } 653