bsd/netinet/udp_usrreq.c

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * The contents of this file constitute Original Code as defined in and
   7  * are subject to the Apple Public Source License Version 1.1 (the
   8  * "License").  You may not use this file except in compliance with the
   9  * License.  Please obtain a copy of the License at
  10  * http://www.apple.com/publicsource and read it before using this file.
  11  *
  12  * This Original Code and all software distributed under the License are
  13  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  14  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  15  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
  17  * License for the specific language governing rights and limitations
  18  * under the License.
  19  *
  20  * @APPLE_LICENSE_HEADER_END@
  21  */
  22 /*
  23  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
  24  *      The Regents of the University of California.  All rights reserved.
  25  *
  26  * Redistribution and use in source and binary forms, with or without
  27  * modification, are permitted provided that the following conditions
  28  * are met:
  29  * 1. Redistributions of source code must retain the above copyright
  30  *    notice, this list of conditions and the following disclaimer.
  31  * 2. Redistributions in binary form must reproduce the above copyright
  32  *    notice, this list of conditions and the following disclaimer in the
  33  *    documentation and/or other materials provided with the distribution.
  34  * 3. All advertising materials mentioning features or use of this software
  35  *    must display the following acknowledgement:
  36  *      This product includes software developed by the University of
  37  *      California, Berkeley and its contributors.
  38  * 4. Neither the name of the University nor the names of its contributors
  39  *    may be used to endorse or promote products derived from this software
  40  *    without specific prior written permission.
  41  *
  42  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  43  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  44  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  45  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  46  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  47  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  48  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  49  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  50  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  51  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  52  * SUCH DAMAGE.
  53  *
  54  *      @(#)udp_usrreq.c        8.6 (Berkeley) 5/23/95
  55  */
  56
  57 #include <sys/param.h>
  58 #include <sys/systm.h>
  59 #include <sys/kernel.h>
  60 #include <sys/malloc.h>
  61 #include <sys/mbuf.h>
  62 #if INET6
  63 #include <sys/domain.h>
  64 #endif
  65 #include <sys/protosw.h>
  66 #include <sys/socket.h>
  67 #include <sys/socketvar.h>
  68 #include <sys/sysctl.h>
  69 #include <sys/syslog.h>
  70
  71 #if ISFB31
  72 #include <vm/vm_zone.h>
  73 #endif
  74
  75
  76 #include <net/if.h>
  77 #include <net/route.h>
  78
  79 #include <netinet/in.h>
  80 #include <netinet/in_systm.h>
  81 #include <netinet/ip.h>
  82 #include <netinet/in_pcb.h>
  83 #include <netinet/in_var.h>
  84 #include <netinet/ip_var.h>
  85 #include <netinet/ip_icmp.h>
  86 #include <netinet/icmp_var.h>
  87 #if INET6
  88 #include <netinet/ip6.h>
  89 #include <netinet6/ip6_var.h>
  90 #endif
  91 #include <netinet/udp.h>
  92 #include <netinet/udp_var.h>
  93 #include <sys/kdebug.h>
  94
  95 #if IPSEC
  96 #include <netinet6/ipsec.h>
  97 #endif /*IPSEC*/
  98
  99
 100 #define DBG_LAYER_IN_BEG        NETDBG_CODE(DBG_NETUDP, 0)
 101 #define DBG_LAYER_IN_END        NETDBG_CODE(DBG_NETUDP, 2)
 102 #define DBG_LAYER_OUT_BEG       NETDBG_CODE(DBG_NETUDP, 1)
 103 #define DBG_LAYER_OUT_END       NETDBG_CODE(DBG_NETUDP, 3)
 104 #define DBG_FNC_UDP_INPUT       NETDBG_CODE(DBG_NETUDP, (5 << 8))
 105 #define DBG_FNC_UDP_OUTPUT      NETDBG_CODE(DBG_NETUDP, (6 << 8) | 1)
 106
 107
 108 #ifndef offsetof               /* XXX */
 109 #define        offsetof(type, member)  ((size_t)(&((type *)0)->member))
 110 #endif
 111
 112 #define __STDC__ 1
 113 /*
 114  * UDP protocol implementation.
 115  * Per RFC 768, August, 1980.
 116  */
 117 #ifndef COMPAT_42
 118 static int      udpcksum = 1;
 119 #else
 120 static int      udpcksum = 0;           /* XXX */
 121 #endif
 122 SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
 123                 &udpcksum, 0, "");
 124
 125 int log_in_vain;
 126 SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
 127         &log_in_vain, 0, "");
 128
 129 struct  inpcbhead udb;          /* from udp_var.h */
 130 #define udb6    udb  /* for KAME src sync over BSD*'s */
 131 struct  inpcbinfo udbinfo;
 132
 133 #ifndef UDBHASHSIZE
 134 #define UDBHASHSIZE 16
 135 #endif
 136
 137 extern  int apple_hwcksum_rx;
 138
 139 struct  udpstat udpstat;        /* from udp_var.h */
 140 SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RD,
 141         &udpstat, udpstat, "");
 142
 143 static struct   sockaddr_in udp_in = { sizeof(udp_in), AF_INET };
 144 #if INET6
 145 struct udp_in6 {
 146         struct sockaddr_in6     uin6_sin;
 147         u_char                  uin6_init_done : 1;
 148 } udp_in6 = {
 149         { sizeof(udp_in6.uin6_sin), AF_INET6 },
 150         0
 151 };
 152 struct udp_ip6 {
 153         struct ip6_hdr          uip6_ip6;
 154         u_char                  uip6_init_done : 1;
 155 } udp_ip6;
 156 #endif /* INET6 */
 157
 158 static void udp_append __P((struct inpcb *last, struct ip *ip,
 159                             struct mbuf *n, int off));
 160 #if INET6
 161 static void ip_2_ip6_hdr __P((struct ip6_hdr *ip6, struct ip *ip));
 162 #endif
 163
 164 static int udp_detach __P((struct socket *so));
 165 static int udp_output __P((struct inpcb *, struct mbuf *, struct sockaddr *,
 166                            struct mbuf *, struct proc *));
 167
 168 void
 169 udp_init()
 170 {
 171         vm_size_t       str_size;
 172         int             stat;
 173         u_char          fake_owner;
 174         struct in_addr  laddr;
 175         struct in_addr  faddr;
 176         u_short         lport;
 177
 178         LIST_INIT(&udb);
 179         udbinfo.listhead = &udb;
 180         udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
 181         udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
 182                                         &udbinfo.porthashmask);
 183 #if ISFB31
 184         udbinfo.ipi_zone = zinit("udpcb", sizeof(struct inpcb), maxsockets,
 185                                  ZONE_INTERRUPT, 0);
 186 #else
 187         str_size = (vm_size_t) sizeof(struct inpcb);
 188         udbinfo.ipi_zone = (void *) zinit(str_size, 80000*str_size, 8192, "inpcb_zone");
 189 #endif
 190
 191         udbinfo.last_pcb = 0;
 192         in_pcb_nat_init(&udbinfo, AF_INET, IPPROTO_UDP, SOCK_DGRAM);
 193
 194 #if 0
 195         stat = in_pcb_new_share_client(&udbinfo, &fake_owner);
 196         kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat);
 197
 198         laddr.s_addr = 0x11646464;
 199         faddr.s_addr = 0x11646465;
 200
 201         lport = 1500;
 202         in_pcb_grab_port(&udbinfo, 0, laddr, &lport, faddr, 1600, 0, fake_owner);
 203         kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat);
 204
 205         stat = in_pcb_rem_share_client(&udbinfo, fake_owner);
 206         kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat);
 207
 208         stat = in_pcb_new_share_client(&udbinfo, &fake_owner);
 209         kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat);
 210
 211         laddr.s_addr = 0x11646464;
 212         faddr.s_addr = 0x11646465;
 213
 214         lport = 1500;
 215         stat = in_pcb_grab_port(&udbinfo, 0, laddr, &lport, faddr, 1600, 0, fake_owner);
 216         kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat);
 217 #endif
 218 }
 219
 220 void
 221 udp_input(m, iphlen)
 222         register struct mbuf *m;
 223         int iphlen;
 224 {
 225         register struct ip *ip;
 226         register struct udphdr *uh;
 227         register struct inpcb *inp;
 228         struct mbuf *opts = 0;
 229 #if INET6
 230         struct ip6_recvpktopts opts6;
 231 #endif
 232         int len;
 233         struct ip save_ip;
 234         struct sockaddr *append_sa;
 235
 236         udpstat.udps_ipackets++;
 237 #if INET6
 238         bzero(&opts6, sizeof(opts6));
 239 #endif
 240
 241
 242         KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_START, 0,0,0,0,0);
 243
 244         /*
 245          * Strip IP options, if any; should skip this,
 246          * make available to user, and use on returned packets,
 247          * but we don't yet have a way to check the checksum
 248          * with options still present.
 249          */
 250         if (iphlen > sizeof (struct ip)) {
 251                 ip_stripoptions(m, (struct mbuf *)0);
 252                 iphlen = sizeof(struct ip);
 253                 if (m->m_pkthdr.csum_flags & CSUM_TCP_SUM16)
 254                         m->m_pkthdr.csum_flags = 0; /* invalidate hwcksum */
 255         }
 256
 257         /*
 258          * Get IP and UDP header together in first mbuf.
 259          */
 260         ip = mtod(m, struct ip *);
 261         if (m->m_len < iphlen + sizeof(struct udphdr)) {
 262                 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
 263                         udpstat.udps_hdrops++;
 264                         KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
 265                         return;
 266                 }
 267                 ip = mtod(m, struct ip *);
 268         }
 269         uh = (struct udphdr *)((caddr_t)ip + iphlen);
 270
 271         KERNEL_DEBUG(DBG_LAYER_IN_BEG, uh->uh_dport, uh->uh_sport,
 272                      ip->ip_src.s_addr, ip->ip_dst.s_addr, uh->uh_ulen);
 273
 274         /*
 275          * Make mbuf data length reflect UDP length.
 276          * If not enough data to reflect UDP length, drop.
 277          */
 278         len = ntohs((u_short)uh->uh_ulen);
 279         if (ip->ip_len != len) {
 280                 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
 281                         udpstat.udps_badlen++;
 282                         goto bad;
 283                 }
 284                 m_adj(m, len - ip->ip_len);
 285                 /* ip->ip_len = len; */
 286         }
 287         /*
 288          * Save a copy of the IP header in case we want restore it
 289          * for sending an ICMP error message in response.
 290          */
 291         save_ip = *ip;
 292
 293         /*
 294          * Checksum extended UDP header and data.
 295          */
 296         if (uh->uh_sum) {
 297                if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
 298                         if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
 299                                 uh->uh_sum = m->m_pkthdr.csum_data;
 300                         else {
 301                                 if (apple_hwcksum_rx && (m->m_pkthdr.csum_flags & CSUM_TCP_SUM16)) {
 302                                         bzero(((struct ipovly *)ip)->ih_x1, 9);
 303                                         ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
 304                                         uh->uh_sum = in_addword(in_cksum(m, sizeof(struct ip)),
 305                                                  m->m_pkthdr.csum_data & 0xFFFF);
 306                                 }
 307                                 else  {
 308                                    goto doudpcksum;
 309                                 }
 310                         }
 311                       uh->uh_sum ^= 0xffff;
 312                 } else {
 313 doudpcksum:
 314                         bzero(((struct ipovly *)ip)->ih_x1, 9);
 315                         ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
 316                         uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
 317                 }
 318                 if (uh->uh_sum) {
 319                         udpstat.udps_badsum++;
 320                         m_freem(m);
 321                         KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
 322                         return;
 323                 }
 324         }
 325
 326         if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
 327             in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
 328                 struct inpcb *last;
 329                 /*
 330                  * Deliver a multicast or broadcast datagram to *all* sockets
 331                  * for which the local and remote addresses and ports match
 332                  * those of the incoming datagram.  This allows more than
 333                  * one process to receive multi/broadcasts on the same port.
 334                  * (This really ought to be done for unicast datagrams as
 335                  * well, but that would cause problems with existing
 336                  * applications that open both address-specific sockets and
 337                  * a wildcard socket listening to the same port -- they would
 338                  * end up receiving duplicates of every unicast datagram.
 339                  * Those applications open the multiple sockets to overcome an
 340                  * inadequacy of the UDP socket interface, but for backwards
 341                  * compatibility we avoid the problem here rather than
 342                  * fixing the interface.  Maybe 4.5BSD will remedy this?)
 343                  */
 344
 345                 /*
 346                  * Construct sockaddr format source address.
 347                  */
 348                 udp_in.sin_port = uh->uh_sport;
 349                 udp_in.sin_addr = ip->ip_src;
 350                 /*
 351                  * Locate pcb(s) for datagram.
 352                  * (Algorithm copied from raw_intr().)
 353                  */
 354                 last = NULL;
 355 #if INET6
 356                 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
 357 #endif
 358                 LIST_FOREACH(inp, &udb, inp_list) {
 359 #if INET6
 360                         if ((inp->inp_vflag & INP_IPV4) == 0)
 361                                 continue;
 362 #endif
 363                         if (inp->inp_lport != uh->uh_dport)
 364                                 continue;
 365                         if (inp->inp_laddr.s_addr != INADDR_ANY) {
 366                                 if (inp->inp_laddr.s_addr !=
 367                                     ip->ip_dst.s_addr)
 368                                         continue;
 369                         }
 370                         if (inp->inp_faddr.s_addr != INADDR_ANY) {
 371                                 if (inp->inp_faddr.s_addr !=
 372                                     ip->ip_src.s_addr ||
 373                                     inp->inp_fport != uh->uh_sport)
 374                                         continue;
 375                         }
 376
 377                         if (last != NULL) {
 378                                 struct mbuf *n;
 379
 380 #if IPSEC
 381                                 /* check AH/ESP integrity. */
 382                                 if (ipsec4_in_reject_so(m, last->inp_socket)) {
 383                                         ipsecstat.in_polvio++;
 384                                         /* do not inject data to pcb */
 385                                 } else
 386 #endif /*IPSEC*/
 387                                 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) {
 388                                         udp_append(last, ip, n, iphlen +
 389                                                 sizeof (struct udphdr));
 390                                 }
 391                         }
 392                         last = inp;
 393                         /*
 394                          * Don't look for additional matches if this one does
 395                          * not have either the SO_REUSEPORT or SO_REUSEADDR
 396                          * socket options set.  This heuristic avoids searching
 397                          * through all pcbs in the common case of a non-shared
 398                          * port.  It * assumes that an application will never
 399                          * clear these options after setting them.
 400                          */
 401                         if ((last->inp_socket->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
 402                                 break;
 403                 }
 404
 405                 if (last == NULL) {
 406                         /*
 407                          * No matching pcb found; discard datagram.
 408                          * (No need to send an ICMP Port Unreachable
 409                          * for a broadcast or multicast datgram.)
 410                          */
 411                         udpstat.udps_noportbcast++;
 412                         goto bad;
 413                 }
 414 #if IPSEC
 415                 else
 416                 /* check AH/ESP integrity. */
 417                 if (m && ipsec4_in_reject_so(m, last->inp_socket)) {
 418                         ipsecstat.in_polvio++;
 419                         goto bad;
 420                 }
 421 #endif /*IPSEC*/
 422                 udp_append(last, ip, m, iphlen + sizeof (struct udphdr));
 423                 return;
 424         }
 425         /*
 426          * Locate pcb for datagram.
 427          */
 428         inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
 429             ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
 430         if (inp == NULL) {
 431                 if (log_in_vain) {
 432                         char buf[4*sizeof "123"];
 433
 434                         strcpy(buf, inet_ntoa(ip->ip_dst));
 435                         log(LOG_INFO,
 436                             "Connection attempt to UDP %s:%d from %s:%d\n",
 437                             buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
 438                             ntohs(uh->uh_sport));
 439                 }
 440                 udpstat.udps_noport++;
 441                 if (m->m_flags & (M_BCAST | M_MCAST)) {
 442                         udpstat.udps_noportbcast++;
 443                         goto bad;
 444                 }
 445                 *ip = save_ip;
 446 #if ICMP_BANDLIM
 447                 if (badport_bandlim(0) < 0)
 448                         goto bad;
 449 #endif
 450                 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
 451                 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
 452                 return;
 453         }
 454 #if IPSEC
 455         if (inp != NULL && ipsec4_in_reject_so(m, inp->inp_socket)) {
 456                 ipsecstat.in_polvio++;
 457                 goto bad;
 458         }
 459 #endif /*IPSEC*/
 460
 461         /*
 462          * Construct sockaddr format source address.
 463          * Stuff source address and datagram in user buffer.
 464          */
 465         udp_in.sin_port = uh->uh_sport;
 466         udp_in.sin_addr = ip->ip_src;
 467         if (inp->inp_flags & INP_CONTROLOPTS
 468             || inp->inp_socket->so_options & SO_TIMESTAMP) {
 469 #if INET6
 470                 if (inp->inp_vflag & INP_IPV6) {
 471                         int savedflags;
 472
 473                         ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
 474                         savedflags = inp->inp_flags;
 475                         inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
 476                         ip6_savecontrol(inp, &udp_ip6.uip6_ip6, m,
 477                                         &opts6, NULL);
 478
 479                         inp->inp_flags = savedflags;
 480                 } else
 481 #endif
 482                 ip_savecontrol(inp, &opts, ip, m);
 483         }
 484         m_adj(m, iphlen + sizeof(struct udphdr));
 485
 486         KERNEL_DEBUG(DBG_LAYER_IN_END, uh->uh_dport, uh->uh_sport,
 487                      save_ip.ip_src.s_addr, save_ip.ip_dst.s_addr, uh->uh_ulen);
 488
 489 #if INET6
 490         if (inp->inp_vflag & INP_IPV6) {
 491                 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
 492                 append_sa = (struct sockaddr *)&udp_in6;
 493                 opts = opts6.head;
 494         } else
 495 #endif
 496         append_sa = (struct sockaddr *)&udp_in;
 497         if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) {
 498                 udpstat.udps_fullsock++;
 499                 goto bad;
 500         }
 501         sorwakeup(inp->inp_socket);
 502         KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
 503         return;
 504 bad:
 505         m_freem(m);
 506         if (opts)
 507                 m_freem(opts);
 508         KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
 509 }
 510
 511 #if INET6
 512 static void
 513 ip_2_ip6_hdr(ip6, ip)
 514         struct ip6_hdr *ip6;
 515         struct ip *ip;
 516 {
 517         bzero(ip6, sizeof(*ip6));
 518
 519         ip6->ip6_vfc = IPV6_VERSION;
 520         ip6->ip6_plen = ip->ip_len;
 521         ip6->ip6_nxt = ip->ip_p;
 522         ip6->ip6_hlim = ip->ip_ttl;
 523         ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
 524                 IPV6_ADDR_INT32_SMP;
 525         ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
 526         ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
 527 }
 528 #endif
 529
 530 /*
 531  * subroutine of udp_input(), mainly for source code readability.
 532  * caller must properly init udp_ip6 and udp_in6 beforehand.
 533  */
 534 static void
 535 udp_append(last, ip, n, off)
 536         struct inpcb *last;
 537         struct ip *ip;
 538         struct mbuf *n;
 539 {
 540         struct sockaddr *append_sa;
 541         struct mbuf *opts = 0;
 542 #if INET6
 543         struct ip6_recvpktopts opts6;
 544         bzero(&opts6, sizeof(opts6));
 545 #endif
 546
 547
 548         if (last->inp_flags & INP_CONTROLOPTS ||
 549             last->inp_socket->so_options & SO_TIMESTAMP) {
 550 #if INET6
 551                 if (last->inp_vflag & INP_IPV6) {
 552                         int savedflags;
 553
 554                         if (udp_ip6.uip6_init_done == 0) {
 555                                 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
 556                                 udp_ip6.uip6_init_done = 1;
 557                         }
 558                         savedflags = last->inp_flags;
 559                         last->inp_flags &= ~INP_UNMAPPABLEOPTS;
 560                         ip6_savecontrol(last, &udp_ip6.uip6_ip6, n,
 561                                         &opts6, NULL);
 562                         last->inp_flags = savedflags;
 563                 } else
 564 #endif
 565                 ip_savecontrol(last, &opts, ip, n);
 566         }
 567 #if INET6
 568         if (last->inp_vflag & INP_IPV6) {
 569                 if (udp_in6.uin6_init_done == 0) {
 570                         in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
 571                         udp_in6.uin6_init_done = 1;
 572                 }
 573                 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
 574                 opts = opts6.head;
 575         } else
 576 #endif
 577         append_sa = (struct sockaddr *)&udp_in;
 578         m_adj(n, off);
 579
 580         if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) {
 581                 m_freem(n);
 582                 if (opts)
 583                         m_freem(opts);
 584                 udpstat.udps_fullsock++;
 585         } else
 586                 sorwakeup(last->inp_socket);
 587 }
 588
 589
 590
 591 /*
 592  * Notify a udp user of an asynchronous error;
 593  * just wake up so that he can collect error status.
 594  */
 595 void
 596 udp_notify(inp, errno)
 597         register struct inpcb *inp;
 598         int errno;
 599 {
 600         inp->inp_socket->so_error = errno;
 601         sorwakeup(inp->inp_socket);
 602         sowwakeup(inp->inp_socket);
 603 }
 604
 605 void
 606 udp_ctlinput(cmd, sa, vip)
 607         int cmd;
 608         struct sockaddr *sa;
 609         void *vip;
 610 {
 611         register struct ip *ip = vip;
 612         register struct udphdr *uh;
 613
 614         if (!PRC_IS_REDIRECT(cmd) &&
 615             ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0))
 616                 return;
 617         if (ip) {
 618                 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
 619                 in_pcbnotify(&udb, sa, uh->uh_dport, ip->ip_src, uh->uh_sport,
 620                         cmd, udp_notify);
 621         } else
 622                 in_pcbnotify(&udb, sa, 0, zeroin_addr, 0, cmd, udp_notify);
 623 }
 624
 625
 626 static int
 627 udp_pcblist SYSCTL_HANDLER_ARGS
 628 {
 629         int error, i, n, s;
 630         struct inpcb *inp, **inp_list;
 631         inp_gen_t gencnt;
 632         struct xinpgen xig;
 633
 634         /*
 635          * The process of preparing the TCB list is too time-consuming and
 636          * resource-intensive to repeat twice on every request.
 637          */
 638         if (req->oldptr == 0) {
 639                 n = udbinfo.ipi_count;
 640                 req->oldidx = 2 * (sizeof xig)
 641                         + (n + n/8) * sizeof(struct xinpcb);
 642                 return 0;
 643         }
 644
 645         if (req->newptr != 0)
 646                 return EPERM;
 647
 648         /*
 649          * OK, now we're committed to doing something.
 650          */
 651         s = splnet();
 652         gencnt = udbinfo.ipi_gencnt;
 653         n = udbinfo.ipi_count;
 654         splx(s);
 655
 656         xig.xig_len = sizeof xig;
 657         xig.xig_count = n;
 658         xig.xig_gen = gencnt;
 659         xig.xig_sogen = so_gencnt;
 660         error = SYSCTL_OUT(req, &xig, sizeof xig);
 661         if (error)
 662                 return error;
 663         /*
 664          * We are done if there is no pcb
 665          */
 666         if (n == 0)
 667             return 0;
 668
 669         inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
 670         if (inp_list == 0) {
 671                 return ENOMEM;
 672         }
 673         s = splnet();
 674         for (inp = udbinfo.listhead->lh_first, i = 0; inp && i < n;
 675              inp = inp->inp_list.le_next) {
 676                 if (inp->inp_gencnt <= gencnt)
 677                         inp_list[i++] = inp;
 678         }
 679         splx(s);
 680         n = i;
 681
 682         error = 0;
 683         for (i = 0; i < n; i++) {
 684                 inp = inp_list[i];
 685                 if (inp->inp_gencnt <= gencnt) {
 686                         struct xinpcb xi;
 687                         xi.xi_len = sizeof xi;
 688                         /* XXX should avoid extra copy */
 689                         bcopy(inp, &xi.xi_inp, sizeof *inp);
 690                         if (inp->inp_socket)
 691                                 sotoxsocket(inp->inp_socket, &xi.xi_socket);
 692                         error = SYSCTL_OUT(req, &xi, sizeof xi);
 693                 }
 694         }
 695         if (!error) {
 696                 /*
 697                  * Give the user an updated idea of our state.
 698                  * If the generation differs from what we told
 699                  * her before, she knows that something happened
 700                  * while we were processing this request, and it
 701                  * might be necessary to retry.
 702                  */
 703                 s = splnet();
 704                 xig.xig_gen = udbinfo.ipi_gencnt;
 705                 xig.xig_sogen = so_gencnt;
 706                 xig.xig_count = udbinfo.ipi_count;
 707                 splx(s);
 708                 error = SYSCTL_OUT(req, &xig, sizeof xig);
 709         }
 710         FREE(inp_list, M_TEMP);
 711         return error;
 712 }
 713
 714 SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
 715             udp_pcblist, "S,xinpcb", "List of active UDP sockets");
 716
 717
 718
 719 static int
 720 udp_output(inp, m, addr, control, p)
 721         register struct inpcb *inp;
 722         register struct mbuf *m;
 723         struct sockaddr *addr;
 724         struct mbuf *control;
 725         struct proc *p;
 726 {
 727         register struct udpiphdr *ui;
 728         register int len = m->m_pkthdr.len;
 729         struct in_addr laddr;
 730         int s = 0, error = 0;
 731
 732         KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
 733
 734         if (control)
 735                 m_freem(control);               /* XXX */
 736
 737         KERNEL_DEBUG(DBG_LAYER_OUT_BEG, inp->inp_fport, inp->inp_lport,
 738                      inp->inp_laddr.s_addr, inp->inp_faddr.s_addr,
 739                      (htons((u_short)len + sizeof (struct udphdr))));
 740
 741         if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
 742                 error = EMSGSIZE;
 743                 goto release;
 744         }
 745
 746         if (addr) {
 747                 laddr = inp->inp_laddr;
 748                 if (inp->inp_faddr.s_addr != INADDR_ANY) {
 749                         error = EISCONN;
 750                         goto release;
 751                 }
 752                 /*
 753                  * Must block input while temporarily connected.
 754                  */
 755                 s = splnet();
 756                 error = in_pcbconnect(inp, addr, p);
 757                 if (error) {
 758                         splx(s);
 759                         goto release;
 760                 }
 761         } else {
 762                 if (inp->inp_faddr.s_addr == INADDR_ANY) {
 763                         error = ENOTCONN;
 764                         goto release;
 765                 }
 766         }
 767         /*
 768          * Calculate data length and get a mbuf
 769          * for UDP and IP headers.
 770          */
 771         M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
 772         if (m == 0) {
 773                 error = ENOBUFS;
 774                 if (addr)
 775                         splx(s);
 776                 goto release;
 777         }
 778
 779         /*
 780          * Fill in mbuf with extended UDP header
 781          * and addresses and length put into network format.
 782          */
 783         ui = mtod(m, struct udpiphdr *);
 784         bzero(ui->ui_x1, sizeof(ui->ui_x1));
 785         ui->ui_pr = IPPROTO_UDP;
 786         ui->ui_len = htons((u_short)len + sizeof (struct udphdr));
 787         ui->ui_src = inp->inp_laddr;
 788         ui->ui_dst = inp->inp_faddr;
 789         ui->ui_sport = inp->inp_lport;
 790         ui->ui_dport = inp->inp_fport;
 791         ui->ui_ulen = ui->ui_len;
 792
 793         /*
 794          * Stuff checksum and output datagram.
 795          */
 796         if (udpcksum) {
 797                 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
 798                     htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
 799                 m->m_pkthdr.csum_flags = CSUM_UDP;
 800                 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
 801         }
 802         else
 803                 ui->ui_sum = 0;
 804         ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
 805         ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl;    /* XXX */
 806         ((struct ip *)ui)->ip_tos = inp->inp_ip_tos;    /* XXX */
 807         udpstat.udps_opackets++;
 808
 809         KERNEL_DEBUG(DBG_LAYER_OUT_END, ui->ui_dport, ui->ui_sport,
 810                      ui->ui_src.s_addr, ui->ui_dst.s_addr, ui->ui_ulen);
 811
 812
 813 #if IPSEC
 814         ipsec_setsocket(m, inp->inp_socket);
 815 #endif /*IPSEC*/
 816
 817         error = ip_output(m, inp->inp_options, &inp->inp_route,
 818             inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST),
 819             inp->inp_moptions);
 820
 821         if (addr) {
 822                 in_pcbdisconnect(inp);
 823                 inp->inp_laddr = laddr; /* XXX rehash? */
 824                 splx(s);
 825         }
 826         KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_END, error, 0,0,0,0);
 827         return (error);
 828
 829 release:
 830         m_freem(m);
 831         KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_END, error, 0,0,0,0);
 832         return (error);
 833 }
 834
 835 u_long  udp_sendspace = 9216;           /* really max datagram size */
 836                                         /* 40 1K datagrams */
 837 SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
 838         &udp_sendspace, 0, "");
 839
 840
 841 u_long  udp_recvspace = 40 * (1024 + /* 40 1K datagrams */
 842 #if INET6
 843                                       sizeof(struct sockaddr_in6)
 844 #else /* INET6 */
 845                                       sizeof(struct sockaddr_in)
 846 #endif /* INET6 */
 847                                       );
 848 SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
 849         &udp_recvspace, 0, "");
 850
 851 static int
 852 udp_abort(struct socket *so)
 853 {
 854         struct inpcb *inp;
 855         int s;
 856
 857         inp = sotoinpcb(so);
 858         if (inp == 0)
 859                 return EINVAL;  /* ??? possible? panic instead? */
 860         soisdisconnected(so);
 861         s = splnet();
 862         in_pcbdetach(inp);
 863         splx(s);
 864         return 0;
 865 }
 866
 867 static int
 868 udp_attach(struct socket *so, int proto, struct proc *p)
 869 {
 870         struct inpcb *inp;
 871         int error; long s;
 872
 873         if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
 874                 error = soreserve(so, udp_sendspace, udp_recvspace);
 875                 if (error)
 876                         return error;
 877         }
 878         s = splnet();
 879         error = in_pcballoc(so, &udbinfo, p);
 880         splx(s);
 881         if (error)
 882                 return error;
 883         error = soreserve(so, udp_sendspace, udp_recvspace);
 884         if (error)
 885                 return error;
 886         inp = (struct inpcb *)so->so_pcb;
 887         inp->inp_vflag |= INP_IPV4;
 888         inp->inp_ip_ttl = ip_defttl;
 889 #if IPSEC
 890         error = ipsec_init_policy(so, &inp->inp_sp);
 891         if (error != 0) {
 892                 in_pcbdetach(inp);
 893                 return error;
 894         }
 895 #endif /*IPSEC*/
 896         return 0;
 897 }
 898
 899 static int
 900 udp_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 901 {
 902         struct inpcb *inp;
 903         int s, error;
 904
 905         inp = sotoinpcb(so);
 906         if (inp == 0)
 907                 return EINVAL;
 908         s = splnet();
 909         error = in_pcbbind(inp, nam, p);
 910         splx(s);
 911         return error;
 912 }
 913
 914 static int
 915 udp_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
 916 {
 917         struct inpcb *inp;
 918         int s, error;
 919
 920         inp = sotoinpcb(so);
 921         if (inp == 0)
 922                 return EINVAL;
 923         if (inp->inp_faddr.s_addr != INADDR_ANY)
 924                 return EISCONN;
 925         s = splnet();
 926         error = in_pcbconnect(inp, nam, p);
 927         splx(s);
 928         if (error == 0)
 929                 soisconnected(so);
 930         return error;
 931 }
 932
 933 static int
 934 udp_detach(struct socket *so)
 935 {
 936         struct inpcb *inp;
 937         int s;
 938
 939         inp = sotoinpcb(so);
 940         if (inp == 0)
 941                 return EINVAL;
 942         s = splnet();
 943         in_pcbdetach(inp);
 944         splx(s);
 945         return 0;
 946 }
 947
 948 static int
 949 udp_disconnect(struct socket *so)
 950 {
 951         struct inpcb *inp;
 952         int s;
 953
 954         inp = sotoinpcb(so);
 955         if (inp == 0)
 956                 return EINVAL;
 957         if (inp->inp_faddr.s_addr == INADDR_ANY)
 958                 return ENOTCONN;
 959
 960         s = splnet();
 961         in_pcbdisconnect(inp);
 962         inp->inp_laddr.s_addr = INADDR_ANY;
 963         splx(s);
 964         so->so_state &= ~SS_ISCONNECTED;                /* XXX */
 965         return 0;
 966 }
 967
 968 static int
 969 udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr,
 970             struct mbuf *control, struct proc *p)
 971 {
 972         struct inpcb *inp;
 973
 974         inp = sotoinpcb(so);
 975         if (inp == 0) {
 976                 m_freem(m);
 977                 return EINVAL;
 978         }
 979         return udp_output(inp, m, addr, control, p);
 980 }
 981
 982 int
 983 udp_shutdown(struct socket *so)
 984 {
 985         struct inpcb *inp;
 986
 987         inp = sotoinpcb(so);
 988         if (inp == 0)
 989                 return EINVAL;
 990         socantsendmore(so);
 991         return 0;
 992 }
 993
 994 struct pr_usrreqs udp_usrreqs = {
 995         udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect,
 996         pru_connect2_notsupp, in_control, udp_detach, udp_disconnect,
 997         pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
 998         pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown,
 999         in_setsockaddr, sosend, soreceive, sopoll
1000 };
1001