Switched back to using pthread_create versus lwp_create.
[dragonfly/vkernel-mp.git] / sys / netinet / udp_usrreq.c
blobc62adfa1b242d97bcec5bf988b6e373836f56b6e
1 /*
2 * Copyright (c) 2004 Jeffrey M. Hsu. All rights reserved.
3 * Copyright (c) 2004 The DragonFly Project. All rights reserved.
5 * This code is derived from software contributed to The DragonFly Project
6 * by Jeffrey M. Hsu.
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.
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.
35 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64 * SUCH DAMAGE.
66 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
67 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.18 2003/01/24 05:11:34 sam Exp $
68 * $DragonFly: src/sys/netinet/udp_usrreq.c,v 1.42 2007/04/22 01:13:14 dillon Exp $
71 #include "opt_ipsec.h"
72 #include "opt_inet6.h"
74 #include <sys/param.h>
75 #include <sys/systm.h>
76 #include <sys/kernel.h>
77 #include <sys/malloc.h>
78 #include <sys/mbuf.h>
79 #include <sys/domain.h>
80 #include <sys/proc.h>
81 #include <sys/protosw.h>
82 #include <sys/socket.h>
83 #include <sys/socketvar.h>
84 #include <sys/sysctl.h>
85 #include <sys/syslog.h>
86 #include <sys/thread2.h>
87 #include <sys/in_cksum.h>
89 #include <machine/stdarg.h>
91 #include <vm/vm_zone.h>
93 #include <net/if.h>
94 #include <net/route.h>
96 #include <netinet/in.h>
97 #include <netinet/in_systm.h>
98 #include <netinet/ip.h>
99 #ifdef INET6
100 #include <netinet/ip6.h>
101 #endif
102 #include <netinet/in_pcb.h>
103 #include <netinet/in_var.h>
104 #include <netinet/ip_var.h>
105 #ifdef INET6
106 #include <netinet6/ip6_var.h>
107 #endif
108 #include <netinet/ip_icmp.h>
109 #include <netinet/icmp_var.h>
110 #include <netinet/udp.h>
111 #include <netinet/udp_var.h>
113 #ifdef FAST_IPSEC
114 #include <netproto/ipsec/ipsec.h>
115 #endif
117 #ifdef IPSEC
118 #include <netinet6/ipsec.h>
119 #endif
122 * UDP protocol implementation.
123 * Per RFC 768, August, 1980.
125 #ifndef COMPAT_42
126 static int udpcksum = 1;
127 #else
128 static int udpcksum = 0; /* XXX */
129 #endif
130 SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
131 &udpcksum, 0, "");
133 int log_in_vain = 0;
134 SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
135 &log_in_vain, 0, "Log all incoming UDP packets");
137 static int blackhole = 0;
138 SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW,
139 &blackhole, 0, "Do not send port unreachables for refused connects");
141 static int strict_mcast_mship = 1;
142 SYSCTL_INT(_net_inet_udp, OID_AUTO, strict_mcast_mship, CTLFLAG_RW,
143 &strict_mcast_mship, 0, "Only send multicast to member sockets");
145 struct inpcbinfo udbinfo;
147 #ifndef UDBHASHSIZE
148 #define UDBHASHSIZE 16
149 #endif
151 struct udpstat udpstat; /* from udp_var.h */
152 SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW,
153 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)");
155 static struct sockaddr_in udp_in = { sizeof udp_in, AF_INET };
156 #ifdef INET6
157 struct udp_in6 {
158 struct sockaddr_in6 uin6_sin;
159 u_char uin6_init_done : 1;
160 } udp_in6 = {
161 { sizeof udp_in6.uin6_sin, AF_INET6 },
164 struct udp_ip6 {
165 struct ip6_hdr uip6_ip6;
166 u_char uip6_init_done : 1;
167 } udp_ip6;
168 #endif /* INET6 */
170 static void udp_append (struct inpcb *last, struct ip *ip,
171 struct mbuf *n, int off);
172 #ifdef INET6
173 static void ip_2_ip6_hdr (struct ip6_hdr *ip6, struct ip *ip);
174 #endif
176 static int udp_detach (struct socket *so);
177 static int udp_output (struct inpcb *, struct mbuf *, struct sockaddr *,
178 struct mbuf *, struct thread *);
180 void
181 udp_init(void)
183 in_pcbinfo_init(&udbinfo);
184 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
185 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
186 &udbinfo.porthashmask);
187 udbinfo.wildcardhashbase = hashinit(UDBHASHSIZE, M_PCB,
188 &udbinfo.wildcardhashmask);
189 udbinfo.ipi_zone = zinit("udpcb", sizeof(struct inpcb), maxsockets,
190 ZONE_INTERRUPT, 0);
191 udp_thread_init();
195 * Check multicast packets to make sure they are only sent to sockets with
196 * multicast memberships for the packet's destination address and arrival
197 * interface. Multicast packets to multicast-unaware sockets are also
198 * disallowed.
200 * Returns 0 if the packet is acceptable, -1 if it is not.
202 static __inline int
203 check_multicast_membership(struct ip *ip, struct inpcb *inp, struct mbuf *m)
205 int mshipno;
206 struct ip_moptions *mopt;
208 if (strict_mcast_mship == 0 ||
209 !IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
210 return (0);
212 mopt = inp->inp_moptions;
213 if (mopt == NULL)
214 return (-1);
215 for (mshipno = 0; mshipno <= mopt->imo_num_memberships; ++mshipno) {
216 struct in_multi *maddr = mopt->imo_membership[mshipno];
218 if (ip->ip_dst.s_addr == maddr->inm_addr.s_addr &&
219 m->m_pkthdr.rcvif == maddr->inm_ifp) {
220 return (0);
223 return (-1);
226 void
227 udp_input(struct mbuf *m, ...)
229 int iphlen;
230 struct ip *ip;
231 struct udphdr *uh;
232 struct inpcb *inp;
233 struct mbuf *opts = NULL;
234 int len, off, proto;
235 struct ip save_ip;
236 struct sockaddr *append_sa;
237 __va_list ap;
239 __va_start(ap, m);
240 off = __va_arg(ap, int);
241 proto = __va_arg(ap, int);
242 __va_end(ap);
244 iphlen = off;
245 udpstat.udps_ipackets++;
248 * Strip IP options, if any; should skip this,
249 * make available to user, and use on returned packets,
250 * but we don't yet have a way to check the checksum
251 * with options still present.
253 if (iphlen > sizeof(struct ip)) {
254 ip_stripoptions(m);
255 iphlen = sizeof(struct ip);
259 * IP and UDP headers are together in first mbuf.
260 * Already checked and pulled up in ip_demux().
262 KASSERT(m->m_len >= iphlen + sizeof(struct udphdr),
263 ("UDP header not in one mbuf"));
265 ip = mtod(m, struct ip *);
266 uh = (struct udphdr *)((caddr_t)ip + iphlen);
268 /* destination port of 0 is illegal, based on RFC768. */
269 if (uh->uh_dport == 0)
270 goto bad;
273 * Make mbuf data length reflect UDP length.
274 * If not enough data to reflect UDP length, drop.
276 len = ntohs((u_short)uh->uh_ulen);
277 if (ip->ip_len != len) {
278 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
279 udpstat.udps_badlen++;
280 goto bad;
282 m_adj(m, len - ip->ip_len);
283 /* ip->ip_len = len; */
286 * Save a copy of the IP header in case we want restore it
287 * for sending an ICMP error message in response.
289 save_ip = *ip;
292 * Checksum extended UDP header and data.
294 if (uh->uh_sum) {
295 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
296 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
297 uh->uh_sum = m->m_pkthdr.csum_data;
298 else
299 uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
300 ip->ip_dst.s_addr, htonl((u_short)len +
301 m->m_pkthdr.csum_data + IPPROTO_UDP));
302 uh->uh_sum ^= 0xffff;
303 } else {
304 char b[9];
306 bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
307 bzero(((struct ipovly *)ip)->ih_x1, 9);
308 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
309 uh->uh_sum = in_cksum(m, len + sizeof(struct ip));
310 bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
312 if (uh->uh_sum) {
313 udpstat.udps_badsum++;
314 m_freem(m);
315 return;
317 } else
318 udpstat.udps_nosum++;
320 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
321 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
322 struct inpcb *last;
325 * Deliver a multicast or broadcast datagram to *all* sockets
326 * for which the local and remote addresses and ports match
327 * those of the incoming datagram. This allows more than
328 * one process to receive multi/broadcasts on the same port.
329 * (This really ought to be done for unicast datagrams as
330 * well, but that would cause problems with existing
331 * applications that open both address-specific sockets and
332 * a wildcard socket listening to the same port -- they would
333 * end up receiving duplicates of every unicast datagram.
334 * Those applications open the multiple sockets to overcome an
335 * inadequacy of the UDP socket interface, but for backwards
336 * compatibility we avoid the problem here rather than
337 * fixing the interface. Maybe 4.5BSD will remedy this?)
341 * Construct sockaddr format source address.
343 udp_in.sin_port = uh->uh_sport;
344 udp_in.sin_addr = ip->ip_src;
346 * Locate pcb(s) for datagram.
347 * (Algorithm copied from raw_intr().)
349 last = NULL;
350 #ifdef INET6
351 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
352 #endif
353 LIST_FOREACH(inp, &udbinfo.pcblisthead, inp_list) {
354 if (inp->inp_flags & INP_PLACEMARKER)
355 continue;
356 #ifdef INET6
357 if (!(inp->inp_vflag & INP_IPV4))
358 continue;
359 #endif
360 if (inp->inp_lport != uh->uh_dport)
361 continue;
362 if (inp->inp_laddr.s_addr != INADDR_ANY) {
363 if (inp->inp_laddr.s_addr !=
364 ip->ip_dst.s_addr)
365 continue;
367 if (inp->inp_faddr.s_addr != INADDR_ANY) {
368 if (inp->inp_faddr.s_addr !=
369 ip->ip_src.s_addr ||
370 inp->inp_fport != uh->uh_sport)
371 continue;
374 if (check_multicast_membership(ip, inp, m) < 0)
375 continue;
377 if (last != NULL) {
378 struct mbuf *n;
380 #ifdef 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 #ifdef FAST_IPSEC
388 /* check AH/ESP integrity. */
389 if (ipsec4_in_reject(m, last))
391 else
392 #endif /*FAST_IPSEC*/
393 if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL)
394 udp_append(last, ip, n,
395 iphlen +
396 sizeof(struct udphdr));
398 last = inp;
400 * Don't look for additional matches if this one does
401 * not have either the SO_REUSEPORT or SO_REUSEADDR
402 * socket options set. This heuristic avoids searching
403 * through all pcbs in the common case of a non-shared
404 * port. It * assumes that an application will never
405 * clear these options after setting them.
407 if (!(last->inp_socket->so_options &
408 (SO_REUSEPORT | SO_REUSEADDR)))
409 break;
412 if (last == NULL) {
414 * No matching pcb found; discard datagram.
415 * (No need to send an ICMP Port Unreachable
416 * for a broadcast or multicast datgram.)
418 udpstat.udps_noportbcast++;
419 goto bad;
421 #ifdef IPSEC
422 /* check AH/ESP integrity. */
423 if (ipsec4_in_reject_so(m, last->inp_socket)) {
424 ipsecstat.in_polvio++;
425 goto bad;
427 #endif /*IPSEC*/
428 #ifdef FAST_IPSEC
429 /* check AH/ESP integrity. */
430 if (ipsec4_in_reject(m, last))
431 goto bad;
432 #endif /*FAST_IPSEC*/
433 udp_append(last, ip, m, iphlen + sizeof(struct udphdr));
434 return;
437 * Locate pcb for datagram.
439 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
440 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
441 if (inp == NULL) {
442 if (log_in_vain) {
443 char buf[sizeof "aaa.bbb.ccc.ddd"];
445 strcpy(buf, inet_ntoa(ip->ip_dst));
446 log(LOG_INFO,
447 "Connection attempt to UDP %s:%d from %s:%d\n",
448 buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
449 ntohs(uh->uh_sport));
451 udpstat.udps_noport++;
452 if (m->m_flags & (M_BCAST | M_MCAST)) {
453 udpstat.udps_noportbcast++;
454 goto bad;
456 if (blackhole)
457 goto bad;
458 #ifdef ICMP_BANDLIM
459 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
460 goto bad;
461 #endif
462 *ip = save_ip;
463 ip->ip_len += iphlen;
464 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
465 return;
467 #ifdef IPSEC
468 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
469 ipsecstat.in_polvio++;
470 goto bad;
472 #endif /*IPSEC*/
473 #ifdef FAST_IPSEC
474 if (ipsec4_in_reject(m, inp))
475 goto bad;
476 #endif /*FAST_IPSEC*/
478 * Check the minimum TTL for socket.
480 if (ip->ip_ttl < inp->inp_ip_minttl)
481 goto bad;
484 * Construct sockaddr format source address.
485 * Stuff source address and datagram in user buffer.
487 udp_in.sin_port = uh->uh_sport;
488 udp_in.sin_addr = ip->ip_src;
489 if ((inp->inp_flags & INP_CONTROLOPTS) ||
490 (inp->inp_socket->so_options & SO_TIMESTAMP)) {
491 #ifdef INET6
492 if (inp->inp_vflag & INP_IPV6) {
493 int savedflags;
495 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
496 savedflags = inp->inp_flags;
497 inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
498 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m);
499 inp->inp_flags = savedflags;
500 } else
501 #endif
502 ip_savecontrol(inp, &opts, ip, m);
504 m_adj(m, iphlen + sizeof(struct udphdr));
505 #ifdef INET6
506 if (inp->inp_vflag & INP_IPV6) {
507 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
508 append_sa = (struct sockaddr *)&udp_in6;
509 } else
510 #endif
511 append_sa = (struct sockaddr *)&udp_in;
512 if (ssb_appendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) {
513 udpstat.udps_fullsock++;
514 goto bad;
516 sorwakeup(inp->inp_socket);
517 return;
518 bad:
519 m_freem(m);
520 if (opts)
521 m_freem(opts);
522 return;
525 #ifdef INET6
526 static void
527 ip_2_ip6_hdr(struct ip6_hdr *ip6, struct ip *ip)
529 bzero(ip6, sizeof *ip6);
531 ip6->ip6_vfc = IPV6_VERSION;
532 ip6->ip6_plen = ip->ip_len;
533 ip6->ip6_nxt = ip->ip_p;
534 ip6->ip6_hlim = ip->ip_ttl;
535 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
536 IPV6_ADDR_INT32_SMP;
537 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
538 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
540 #endif
543 * subroutine of udp_input(), mainly for source code readability.
544 * caller must properly init udp_ip6 and udp_in6 beforehand.
546 static void
547 udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n, int off)
549 struct sockaddr *append_sa;
550 struct mbuf *opts = NULL;
552 if (last->inp_flags & INP_CONTROLOPTS ||
553 last->inp_socket->so_options & SO_TIMESTAMP) {
554 #ifdef INET6
555 if (last->inp_vflag & INP_IPV6) {
556 int savedflags;
558 if (udp_ip6.uip6_init_done == 0) {
559 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
560 udp_ip6.uip6_init_done = 1;
562 savedflags = last->inp_flags;
563 last->inp_flags &= ~INP_UNMAPPABLEOPTS;
564 ip6_savecontrol(last, &opts, &udp_ip6.uip6_ip6, n);
565 last->inp_flags = savedflags;
566 } else
567 #endif
568 ip_savecontrol(last, &opts, ip, n);
570 #ifdef INET6
571 if (last->inp_vflag & INP_IPV6) {
572 if (udp_in6.uin6_init_done == 0) {
573 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
574 udp_in6.uin6_init_done = 1;
576 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
577 } else
578 #endif
579 append_sa = (struct sockaddr *)&udp_in;
580 m_adj(n, off);
581 if (ssb_appendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) {
582 m_freem(n);
583 if (opts)
584 m_freem(opts);
585 udpstat.udps_fullsock++;
586 } else
587 sorwakeup(last->inp_socket);
591 * Notify a udp user of an asynchronous error;
592 * just wake up so that he can collect error status.
594 void
595 udp_notify(struct inpcb *inp, int error)
597 inp->inp_socket->so_error = error;
598 sorwakeup(inp->inp_socket);
599 sowwakeup(inp->inp_socket);
602 void
603 udp_ctlinput(int cmd, struct sockaddr *sa, void *vip)
605 struct ip *ip = vip;
606 struct udphdr *uh;
607 void (*notify) (struct inpcb *, int) = udp_notify;
608 struct in_addr faddr;
609 struct inpcb *inp;
611 faddr = ((struct sockaddr_in *)sa)->sin_addr;
612 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
613 return;
615 if (PRC_IS_REDIRECT(cmd)) {
616 ip = NULL;
617 notify = in_rtchange;
618 } else if (cmd == PRC_HOSTDEAD)
619 ip = NULL;
620 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0)
621 return;
622 if (ip) {
623 crit_enter();
624 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
625 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport,
626 ip->ip_src, uh->uh_sport, 0, NULL);
627 if (inp != NULL && inp->inp_socket != NULL)
628 (*notify)(inp, inetctlerrmap[cmd]);
629 crit_exit();
630 } else
631 in_pcbnotifyall(&udbinfo.pcblisthead, faddr, inetctlerrmap[cmd],
632 notify);
635 SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, &udbinfo, 0,
636 in_pcblist_global, "S,xinpcb", "List of active UDP sockets");
638 static int
639 udp_getcred(SYSCTL_HANDLER_ARGS)
641 struct sockaddr_in addrs[2];
642 struct inpcb *inp;
643 int error;
645 error = suser(req->td);
646 if (error)
647 return (error);
648 error = SYSCTL_IN(req, addrs, sizeof addrs);
649 if (error)
650 return (error);
651 crit_enter();
652 inp = in_pcblookup_hash(&udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
653 addrs[0].sin_addr, addrs[0].sin_port, 1, NULL);
654 if (inp == NULL || inp->inp_socket == NULL) {
655 error = ENOENT;
656 goto out;
658 error = SYSCTL_OUT(req, inp->inp_socket->so_cred, sizeof(struct ucred));
659 out:
660 crit_exit();
661 return (error);
664 SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW,
665 0, 0, udp_getcred, "S,ucred", "Get the ucred of a UDP connection");
667 static int
668 udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *dstaddr,
669 struct mbuf *control, struct thread *td)
671 struct udpiphdr *ui;
672 int len = m->m_pkthdr.len;
673 struct sockaddr_in *sin; /* really is initialized before use */
674 int error = 0;
676 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
677 error = EMSGSIZE;
678 goto release;
681 if (inp->inp_lport == 0) { /* unbound socket */
682 error = in_pcbbind(inp, (struct sockaddr *)NULL, td);
683 if (error)
684 goto release;
685 in_pcbinswildcardhash(inp);
688 if (dstaddr != NULL) { /* destination address specified */
689 if (inp->inp_faddr.s_addr != INADDR_ANY) {
690 /* already connected */
691 error = EISCONN;
692 goto release;
694 sin = (struct sockaddr_in *)dstaddr;
695 if (!prison_remote_ip(td, (struct sockaddr *)&sin)) {
696 error = EAFNOSUPPORT; /* IPv6 only jail */
697 goto release;
699 } else {
700 if (inp->inp_faddr.s_addr == INADDR_ANY) {
701 /* no destination specified and not already connected */
702 error = ENOTCONN;
703 goto release;
705 sin = NULL;
709 * Calculate data length and get a mbuf
710 * for UDP and IP headers.
712 M_PREPEND(m, sizeof(struct udpiphdr), MB_DONTWAIT);
713 if (m == NULL) {
714 error = ENOBUFS;
715 goto release;
719 * Fill in mbuf with extended UDP header
720 * and addresses and length put into network format.
722 ui = mtod(m, struct udpiphdr *);
723 bzero(ui->ui_x1, sizeof ui->ui_x1); /* XXX still needed? */
724 ui->ui_pr = IPPROTO_UDP;
727 * Set destination address.
729 if (dstaddr != NULL) { /* use specified destination */
730 ui->ui_dst = sin->sin_addr;
731 ui->ui_dport = sin->sin_port;
732 } else { /* use connected destination */
733 ui->ui_dst = inp->inp_faddr;
734 ui->ui_dport = inp->inp_fport;
738 * Set source address.
740 if (inp->inp_laddr.s_addr == INADDR_ANY) {
741 struct sockaddr_in *if_sin;
743 if (dstaddr == NULL) {
745 * connect() had (or should have) failed because
746 * the interface had no IP address, but the
747 * application proceeded to call send() anyways.
749 error = ENOTCONN;
750 goto release;
753 /* Look up outgoing interface. */
754 if ((error = in_pcbladdr(inp, dstaddr, &if_sin, td)))
755 goto release;
756 ui->ui_src = if_sin->sin_addr; /* use address of interface */
757 } else {
758 ui->ui_src = inp->inp_laddr; /* use non-null bound address */
760 ui->ui_sport = inp->inp_lport;
761 KASSERT(inp->inp_lport != 0, ("inp lport should have been bound"));
763 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));
766 * Set up checksum and output datagram.
768 if (udpcksum) {
769 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
770 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
771 m->m_pkthdr.csum_flags = CSUM_UDP;
772 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
773 } else {
774 ui->ui_sum = 0;
776 ((struct ip *)ui)->ip_len = sizeof(struct udpiphdr) + len;
777 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */
778 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */
779 udpstat.udps_opackets++;
781 error = ip_output(m, inp->inp_options, &inp->inp_route,
782 (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST)),
783 inp->inp_moptions, inp);
785 return (error);
787 release:
788 m_freem(m);
789 return (error);
792 u_long udp_sendspace = 9216; /* really max datagram size */
793 /* 40 1K datagrams */
794 SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
795 &udp_sendspace, 0, "Maximum outgoing UDP datagram size");
797 u_long udp_recvspace = 40 * (1024 +
798 #ifdef INET6
799 sizeof(struct sockaddr_in6)
800 #else
801 sizeof(struct sockaddr_in)
802 #endif
804 SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
805 &udp_recvspace, 0, "Maximum incoming UDP datagram size");
807 static int
808 udp_abort(struct socket *so)
810 struct inpcb *inp;
812 inp = so->so_pcb;
813 if (inp == NULL)
814 return EINVAL; /* ??? possible? panic instead? */
815 soisdisconnected(so);
816 crit_enter();
817 in_pcbdetach(inp);
818 crit_exit();
819 return 0;
822 static int
823 udp_attach(struct socket *so, int proto, struct pru_attach_info *ai)
825 struct inpcb *inp;
826 int error;
828 inp = so->so_pcb;
829 if (inp != NULL)
830 return EINVAL;
832 error = soreserve(so, udp_sendspace, udp_recvspace, ai->sb_rlimit);
833 if (error)
834 return error;
835 crit_enter();
836 error = in_pcballoc(so, &udbinfo);
837 crit_exit();
838 if (error)
839 return error;
841 inp = (struct inpcb *)so->so_pcb;
842 inp->inp_vflag |= INP_IPV4;
843 inp->inp_ip_ttl = ip_defttl;
844 return 0;
847 static int
848 udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
850 struct sockaddr_in *sin = (struct sockaddr_in *)nam;
851 struct inpcb *inp;
852 int error;
854 inp = so->so_pcb;
855 if (inp == NULL)
856 return EINVAL;
857 crit_enter();
858 error = in_pcbbind(inp, nam, td);
859 crit_exit();
860 if (error == 0) {
861 if (sin->sin_addr.s_addr != INADDR_ANY)
862 inp->inp_flags |= INP_WASBOUND_NOTANY;
863 in_pcbinswildcardhash(inp);
865 return error;
868 static int
869 udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
871 struct inpcb *inp;
872 int error;
874 inp = so->so_pcb;
875 if (inp == NULL)
876 return EINVAL;
877 if (inp->inp_faddr.s_addr != INADDR_ANY)
878 return EISCONN;
879 error = 0;
880 crit_enter();
881 if (td->td_proc && td->td_proc->p_ucred->cr_prison != NULL &&
882 inp->inp_laddr.s_addr == INADDR_ANY) {
883 error = in_pcbbind(inp, NULL, td);
885 if (error == 0) {
886 if (!prison_remote_ip(td, nam))
887 return(EAFNOSUPPORT); /* IPv6 only jail */
888 if (inp->inp_flags & INP_WILDCARD)
889 in_pcbremwildcardhash(inp);
890 error = in_pcbconnect(inp, nam, td);
892 crit_exit();
893 if (error == 0)
894 soisconnected(so);
895 else if (error == EAFNOSUPPORT) { /* connection dissolved */
897 * Follow traditional BSD behavior and retain
898 * the local port binding. But, fix the old misbehavior
899 * of overwriting any previously bound local address.
901 if (!(inp->inp_flags & INP_WASBOUND_NOTANY))
902 inp->inp_laddr.s_addr = INADDR_ANY;
903 in_pcbinswildcardhash(inp);
905 return error;
908 static int
909 udp_detach(struct socket *so)
911 struct inpcb *inp;
913 inp = so->so_pcb;
914 if (inp == NULL)
915 return EINVAL;
916 crit_enter();
917 in_pcbdetach(inp);
918 crit_exit();
919 return 0;
922 static int
923 udp_disconnect(struct socket *so)
925 struct inpcb *inp;
927 inp = so->so_pcb;
928 if (inp == NULL)
929 return EINVAL;
930 if (inp->inp_faddr.s_addr == INADDR_ANY)
931 return ENOTCONN;
933 crit_enter();
934 in_pcbdisconnect(inp);
935 crit_exit();
936 so->so_state &= ~SS_ISCONNECTED; /* XXX */
937 return 0;
940 static int
941 udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr,
942 struct mbuf *control, struct thread *td)
944 struct inpcb *inp;
946 inp = so->so_pcb;
947 if (inp == NULL) {
948 m_freem(m);
949 return EINVAL;
951 return udp_output(inp, m, addr, control, td);
955 udp_shutdown(struct socket *so)
957 struct inpcb *inp;
959 inp = so->so_pcb;
960 if (inp == NULL)
961 return EINVAL;
962 socantsendmore(so);
963 return 0;
966 struct pr_usrreqs udp_usrreqs = {
967 .pru_abort = udp_abort,
968 .pru_accept = pru_accept_notsupp,
969 .pru_attach = udp_attach,
970 .pru_bind = udp_bind,
971 .pru_connect = udp_connect,
972 .pru_connect2 = pru_connect2_notsupp,
973 .pru_control = in_control,
974 .pru_detach = udp_detach,
975 .pru_disconnect = udp_disconnect,
976 .pru_listen = pru_listen_notsupp,
977 .pru_peeraddr = in_setpeeraddr,
978 .pru_rcvd = pru_rcvd_notsupp,
979 .pru_rcvoob = pru_rcvoob_notsupp,
980 .pru_send = udp_send,
981 .pru_sense = pru_sense_null,
982 .pru_shutdown = udp_shutdown,
983 .pru_sockaddr = in_setsockaddr,
984 .pru_sosend = sosendudp,
985 .pru_soreceive = soreceive,
986 .pru_sopoll = sopoll