1/*
2 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
34 * $FreeBSD: head/sys/netinet/udp_usrreq.c 95759 2002-04-30 01:54:54Z tanimura $
35 */
36
37#include "opt_ipsec.h"
38#include "opt_inet6.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/domain.h>
43#include <sys/jail.h>
44#include <sys/kernel.h>
45#include <sys/lock.h>
46#include <sys/malloc.h>
47#include <sys/mbuf.h>
48#include <sys/proc.h>
49#include <sys/protosw.h>
50#include <sys/signalvar.h>
51#include <sys/socket.h>
52#include <sys/socketvar.h>
53#include <sys/sx.h>
54#include <sys/sysctl.h>
55#include <sys/syslog.h>
56
57#include <vm/uma.h>
58
59#include <net/if.h>
60#include <net/route.h>
61
62#include <netinet/in.h>
63#include <netinet/in_systm.h>
64#include <netinet/in_pcb.h>
65#include <netinet/in_var.h>
66#include <netinet/ip.h>
67#ifdef INET6
68#include <netinet/ip6.h>
69#endif
70#include <netinet/ip_icmp.h>
71#include <netinet/icmp_var.h>
72#include <netinet/ip_var.h>
73#ifdef INET6
74#include <netinet6/ip6_var.h>
75#endif
76#include <netinet/udp.h>
77#include <netinet/udp_var.h>
78
79#ifdef IPSEC
80#include <netinet6/ipsec.h>
81#endif /*IPSEC*/
82
83#include <machine/in_cksum.h>
84
85/*
86 * UDP protocol implementation.
87 * Per RFC 768, August, 1980.
88 */
89#ifndef COMPAT_42
90static int udpcksum = 1;
91#else
92static int udpcksum = 0; /* XXX */
93#endif
94SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
95 &udpcksum, 0, "");
96
97int log_in_vain = 0;
98SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
99 &log_in_vain, 0, "Log all incoming UDP packets");
100
101static int blackhole = 0;
102SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW,
103 &blackhole, 0, "Do not send port unreachables for refused connects");
104
105struct inpcbhead udb; /* from udp_var.h */
106#define udb6 udb /* for KAME src sync over BSD*'s */
107struct inpcbinfo udbinfo;
108
109#ifndef UDBHASHSIZE
110#define UDBHASHSIZE 16
111#endif
112
113struct udpstat udpstat; /* from udp_var.h */
114SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW,
115 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)");
116
117static struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET };
118#ifdef INET6
119struct udp_in6 {
120 struct sockaddr_in6 uin6_sin;
121 u_char uin6_init_done : 1;
122} udp_in6 = {
123 { sizeof(udp_in6.uin6_sin), AF_INET6 },
124 0
125};
126struct udp_ip6 {
127 struct ip6_hdr uip6_ip6;
128 u_char uip6_init_done : 1;
129} udp_ip6;
130#endif /* INET6 */
131
132static void udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n,
133 int off);
134#ifdef INET6
135static void ip_2_ip6_hdr(struct ip6_hdr *ip6, struct ip *ip);
136#endif
137
138static int udp_detach(struct socket *so);
139static int udp_output(struct inpcb *, struct mbuf *, struct sockaddr *,
140 struct mbuf *, struct thread *);
141
142void
143udp_init()
144{
145 LIST_INIT(&udb);
146 udbinfo.listhead = &udb;
147 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
148 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
149 &udbinfo.porthashmask);
150 udbinfo.ipi_zone = uma_zcreate("udpcb", sizeof(struct inpcb), NULL,
151 NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
152 uma_zone_set_max(udbinfo.ipi_zone, maxsockets);
153}
154
155void
156udp_input(m, off)
157 register struct mbuf *m;
158 int off;
159{
160 int iphlen = off;
161 register struct ip *ip;
162 register struct udphdr *uh;
163 register struct inpcb *inp;
164 struct mbuf *opts = 0;
165 int len;
166 struct ip save_ip;
167 struct sockaddr *append_sa;
168
169 udpstat.udps_ipackets++;
170
171 /*
172 * Strip IP options, if any; should skip this,
173 * make available to user, and use on returned packets,
174 * but we don't yet have a way to check the checksum
175 * with options still present.
176 */
177 if (iphlen > sizeof (struct ip)) {
178 ip_stripoptions(m, (struct mbuf *)0);
179 iphlen = sizeof(struct ip);
180 }
181
182 /*
183 * Get IP and UDP header together in first mbuf.
184 */
185 ip = mtod(m, struct ip *);
186 if (m->m_len < iphlen + sizeof(struct udphdr)) {
187 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
188 udpstat.udps_hdrops++;
189 return;
190 }
191 ip = mtod(m, struct ip *);
192 }
193 uh = (struct udphdr *)((caddr_t)ip + iphlen);
194
195 /* destination port of 0 is illegal, based on RFC768. */
196 if (uh->uh_dport == 0)
197 goto bad;
198
199 /*
200 * Make mbuf data length reflect UDP length.
201 * If not enough data to reflect UDP length, drop.
202 */
203 len = ntohs((u_short)uh->uh_ulen);
204 if (ip->ip_len != len) {
205 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
206 udpstat.udps_badlen++;
207 goto bad;
208 }
209 m_adj(m, len - ip->ip_len);
210 /* ip->ip_len = len; */
211 }
212 /*
213 * Save a copy of the IP header in case we want restore it
214 * for sending an ICMP error message in response.
215 */
216 if (!blackhole)
217 save_ip = *ip;
218
219 /*
220 * Checksum extended UDP header and data.
221 */
222 if (uh->uh_sum) {
223 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
224 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
225 uh->uh_sum = m->m_pkthdr.csum_data;
226 else
227 uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
228 ip->ip_dst.s_addr, htonl((u_short)len +
229 m->m_pkthdr.csum_data + IPPROTO_UDP));
230 uh->uh_sum ^= 0xffff;
231 } else {
232 char b[9];
233 bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
234 bzero(((struct ipovly *)ip)->ih_x1, 9);
235 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
236 uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
237 bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
238 }
239 if (uh->uh_sum) {
240 udpstat.udps_badsum++;
241 m_freem(m);
242 return;
243 }
244 } else
245 udpstat.udps_nosum++;
246
247 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
248 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
249 struct inpcb *last;
250 /*
251 * Deliver a multicast or broadcast datagram to *all* sockets
252 * for which the local and remote addresses and ports match
253 * those of the incoming datagram. This allows more than
254 * one process to receive multi/broadcasts on the same port.
255 * (This really ought to be done for unicast datagrams as
256 * well, but that would cause problems with existing
257 * applications that open both address-specific sockets and
258 * a wildcard socket listening to the same port -- they would
259 * end up receiving duplicates of every unicast datagram.
260 * Those applications open the multiple sockets to overcome an
261 * inadequacy of the UDP socket interface, but for backwards
262 * compatibility we avoid the problem here rather than
263 * fixing the interface. Maybe 4.5BSD will remedy this?)
264 */
265
266 /*
267 * Construct sockaddr format source address.
268 */
269 udp_in.sin_port = uh->uh_sport;
270 udp_in.sin_addr = ip->ip_src;
271 /*
272 * Locate pcb(s) for datagram.
273 * (Algorithm copied from raw_intr().)
274 */
275 last = NULL;
276#ifdef INET6
277 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
278#endif
279 LIST_FOREACH(inp, &udb, inp_list) {
280#ifdef INET6
281 if ((inp->inp_vflag & INP_IPV4) == 0)
282 continue;
283#endif
284 if (inp->inp_lport != uh->uh_dport)
285 continue;
286 if (inp->inp_laddr.s_addr != INADDR_ANY) {
287 if (inp->inp_laddr.s_addr !=
288 ip->ip_dst.s_addr)
289 continue;
290 }
291 if (inp->inp_faddr.s_addr != INADDR_ANY) {
292 if (inp->inp_faddr.s_addr !=
293 ip->ip_src.s_addr ||
294 inp->inp_fport != uh->uh_sport)
295 continue;
296 }
297
298 if (last != NULL) {
299 struct mbuf *n;
300
301#ifdef IPSEC
302 /* check AH/ESP integrity. */
303 if (ipsec4_in_reject_so(m, last->inp_socket))
304 ipsecstat.in_polvio++;
305 /* do not inject data to pcb */
306 else
307#endif /*IPSEC*/
308 if ((n = m_copy(m, 0, M_COPYALL)) != NULL)
309 udp_append(last, ip, n,
310 iphlen +
311 sizeof(struct udphdr));
312 }
313 last = inp;
314 /*
315 * Don't look for additional matches if this one does
316 * not have either the SO_REUSEPORT or SO_REUSEADDR
317 * socket options set. This heuristic avoids searching
318 * through all pcbs in the common case of a non-shared
319 * port. It * assumes that an application will never
320 * clear these options after setting them.
321 */
322 if ((last->inp_socket->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
323 break;
324 }
325
326 if (last == NULL) {
327 /*
328 * No matching pcb found; discard datagram.
329 * (No need to send an ICMP Port Unreachable
330 * for a broadcast or multicast datgram.)
331 */
332 udpstat.udps_noportbcast++;
333 goto bad;
334 }
335#ifdef IPSEC
336 /* check AH/ESP integrity. */
337 if (ipsec4_in_reject_so(m, last->inp_socket)) {
338 ipsecstat.in_polvio++;
339 goto bad;
340 }
341#endif /*IPSEC*/
342 udp_append(last, ip, m, iphlen + sizeof(struct udphdr));
343 return;
344 }
345 /*
346 * Locate pcb for datagram.
347 */
348 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
349 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
350 if (inp == NULL) {
351 if (log_in_vain) {
352 char buf[4*sizeof "123"];
353
354 strcpy(buf, inet_ntoa(ip->ip_dst));
355 log(LOG_INFO,
356 "Connection attempt to UDP %s:%d from %s:%d\n",
357 buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
358 ntohs(uh->uh_sport));
359 }
360 udpstat.udps_noport++;
361 if (m->m_flags & (M_BCAST | M_MCAST)) {
362 udpstat.udps_noportbcast++;
363 goto bad;
364 }
365 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
366 goto bad;
367 if (blackhole)
368 goto bad;
369 *ip = save_ip;
370 ip->ip_len += iphlen;
371 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
372 return;
373 }
374#ifdef IPSEC
375 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
376 ipsecstat.in_polvio++;
377 goto bad;
378 }
379#endif /*IPSEC*/
380
381 /*
382 * Construct sockaddr format source address.
383 * Stuff source address and datagram in user buffer.
384 */
385 udp_in.sin_port = uh->uh_sport;
386 udp_in.sin_addr = ip->ip_src;
387 if (inp->inp_flags & INP_CONTROLOPTS
388 || inp->inp_socket->so_options & SO_TIMESTAMP) {
389#ifdef INET6
390 if (inp->inp_vflag & INP_IPV6) {
391 int savedflags;
392
393 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
394 savedflags = inp->inp_flags;
395 inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
396 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m);
397 inp->inp_flags = savedflags;
398 } else
399#endif
400 ip_savecontrol(inp, &opts, ip, m);
401 }
402 m_adj(m, iphlen + sizeof(struct udphdr));
403#ifdef INET6
404 if (inp->inp_vflag & INP_IPV6) {
405 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
406 append_sa = (struct sockaddr *)&udp_in6;
407 } else
408#endif
409 append_sa = (struct sockaddr *)&udp_in;
410 if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) {
411 udpstat.udps_fullsock++;
412 goto bad;
413 }
414 sorwakeup(inp->inp_socket);
415 return;
416bad:
417 m_freem(m);
418 if (opts)
419 m_freem(opts);
420 return;
421}
422
423#ifdef INET6
424static void
425ip_2_ip6_hdr(ip6, ip)
426 struct ip6_hdr *ip6;
427 struct ip *ip;
428{
429 bzero(ip6, sizeof(*ip6));
430
431 ip6->ip6_vfc = IPV6_VERSION;
432 ip6->ip6_plen = ip->ip_len;
433 ip6->ip6_nxt = ip->ip_p;
434 ip6->ip6_hlim = ip->ip_ttl;
435 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
436 IPV6_ADDR_INT32_SMP;
437 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
438 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
439}
440#endif
441
442/*
443 * subroutine of udp_input(), mainly for source code readability.
444 * caller must properly init udp_ip6 and udp_in6 beforehand.
445 */
446static void
447udp_append(last, ip, n, off)
448 struct inpcb *last;
449 struct ip *ip;
450 struct mbuf *n;
451 int off;
452{
453 struct sockaddr *append_sa;
454 struct mbuf *opts = 0;
455
456 if (last->inp_flags & INP_CONTROLOPTS ||
457 last->inp_socket->so_options & SO_TIMESTAMP) {
458#ifdef INET6
459 if (last->inp_vflag & INP_IPV6) {
460 int savedflags;
461
462 if (udp_ip6.uip6_init_done == 0) {
463 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
464 udp_ip6.uip6_init_done = 1;
465 }
466 savedflags = last->inp_flags;
467 last->inp_flags &= ~INP_UNMAPPABLEOPTS;
468 ip6_savecontrol(last, &opts, &udp_ip6.uip6_ip6, n);
469 last->inp_flags = savedflags;
470 } else
471#endif
472 ip_savecontrol(last, &opts, ip, n);
473 }
474#ifdef INET6
475 if (last->inp_vflag & INP_IPV6) {
476 if (udp_in6.uin6_init_done == 0) {
477 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
478 udp_in6.uin6_init_done = 1;
479 }
480 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
481 } else
482#endif
483 append_sa = (struct sockaddr *)&udp_in;
484 m_adj(n, off);
485 if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) {
486 m_freem(n);
487 if (opts)
488 m_freem(opts);
489 udpstat.udps_fullsock++;
490 } else
491 sorwakeup(last->inp_socket);
492}
493
494/*
495 * Notify a udp user of an asynchronous error;
496 * just wake up so that he can collect error status.
497 */
498void
499udp_notify(inp, errno)
500 register struct inpcb *inp;
501 int errno;
502{
503 inp->inp_socket->so_error = errno;
504 sorwakeup(inp->inp_socket);
505 sowwakeup(inp->inp_socket);
506}
507
508void
509udp_ctlinput(cmd, sa, vip)
510 int cmd;
511 struct sockaddr *sa;
512 void *vip;
513{
514 struct ip *ip = vip;
515 struct udphdr *uh;
516 void (*notify)(struct inpcb *, int) = udp_notify;
517 struct in_addr faddr;
518 struct inpcb *inp;
519 int s;
520
521 faddr = ((struct sockaddr_in *)sa)->sin_addr;
522 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
523 return;
524
525 if (PRC_IS_REDIRECT(cmd)) {
526 ip = 0;
527 notify = in_rtchange;
528 } else if (cmd == PRC_HOSTDEAD)
529 ip = 0;
530 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0)
531 return;
532 if (ip) {
533 s = splnet();
534 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
535 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport,
536 ip->ip_src, uh->uh_sport, 0, NULL);
537 if (inp != NULL && inp->inp_socket != NULL)
538 (*notify)(inp, inetctlerrmap[cmd]);
539 splx(s);
540 } else
541 in_pcbnotifyall(&udb, faddr, inetctlerrmap[cmd], notify);
542}
543
544static int
545udp_pcblist(SYSCTL_HANDLER_ARGS)
546{
547 int error, i, n, s;
548 struct inpcb *inp, **inp_list;
549 inp_gen_t gencnt;
550 struct xinpgen xig;
551
552 /*
553 * The process of preparing the TCB list is too time-consuming and
554 * resource-intensive to repeat twice on every request.
555 */
556 if (req->oldptr == 0) {
557 n = udbinfo.ipi_count;
558 req->oldidx = 2 * (sizeof xig)
559 + (n + n/8) * sizeof(struct xinpcb);
560 return 0;
561 }
562
563 if (req->newptr != 0)
564 return EPERM;
565
566 /*
567 * OK, now we're committed to doing something.
568 */
569 s = splnet();
570 gencnt = udbinfo.ipi_gencnt;
571 n = udbinfo.ipi_count;
572 splx(s);
573
574 xig.xig_len = sizeof xig;
575 xig.xig_count = n;
576 xig.xig_gen = gencnt;
577 xig.xig_sogen = so_gencnt;
578 error = SYSCTL_OUT(req, &xig, sizeof xig);
579 if (error)
580 return error;
581
582 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
583 if (inp_list == 0)
584 return ENOMEM;
585
586 s = splnet();
587 for (inp = LIST_FIRST(udbinfo.listhead), i = 0; inp && i < n;
588 inp = LIST_NEXT(inp, inp_list)) {
589 if (inp->inp_gencnt <= gencnt) {
590 if (cr_canseesocket(req->td->td_ucred,
591 inp->inp_socket))
592 continue;
593 inp_list[i++] = inp;
594 }
595 }
596 splx(s);
597 n = i;
598
599 error = 0;
600 for (i = 0; i < n; i++) {
601 inp = inp_list[i];
602 if (inp->inp_gencnt <= gencnt) {
603 struct xinpcb xi;
604 xi.xi_len = sizeof xi;
605 /* XXX should avoid extra copy */
606 bcopy(inp, &xi.xi_inp, sizeof *inp);
607 if (inp->inp_socket)
608 sotoxsocket(inp->inp_socket, &xi.xi_socket);
609 error = SYSCTL_OUT(req, &xi, sizeof xi);
610 }
611 }
612 if (!error) {
613 /*
614 * Give the user an updated idea of our state.
615 * If the generation differs from what we told
616 * her before, she knows that something happened
617 * while we were processing this request, and it
618 * might be necessary to retry.
619 */
620 s = splnet();
621 xig.xig_gen = udbinfo.ipi_gencnt;
622 xig.xig_sogen = so_gencnt;
623 xig.xig_count = udbinfo.ipi_count;
624 splx(s);
625 error = SYSCTL_OUT(req, &xig, sizeof xig);
626 }
627 free(inp_list, M_TEMP);
628 return error;
629}
630
631SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
632 udp_pcblist, "S,xinpcb", "List of active UDP sockets");
633
634static int
635udp_getcred(SYSCTL_HANDLER_ARGS)
636{
637 struct xucred xuc;
638 struct sockaddr_in addrs[2];
639 struct inpcb *inp;
640 int error, s;
641
642 error = suser_cred(req->td->td_ucred, PRISON_ROOT);
643 if (error)
644 return (error);
645 error = SYSCTL_IN(req, addrs, sizeof(addrs));
646 if (error)
647 return (error);
648 s = splnet();
649 inp = in_pcblookup_hash(&udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
650 addrs[0].sin_addr, addrs[0].sin_port, 1, NULL);
651 if (inp == NULL || inp->inp_socket == NULL) {
652 error = ENOENT;
653 goto out;
654 }
655 error = cr_canseesocket(req->td->td_ucred, inp->inp_socket);
656 if (error)
657 goto out;
658 cru2x(inp->inp_socket->so_cred, &xuc);
659 error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
660out:
661 splx(s);
662 return (error);
663}
664
665SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred,
666 CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0,
667 udp_getcred, "S,xucred", "Get the xucred of a UDP connection");
668
669static int
670udp_output(inp, m, addr, control, td)
671 register struct inpcb *inp;
672 struct mbuf *m;
673 struct sockaddr *addr;
674 struct mbuf *control;
675 struct thread *td;
676{
677 register struct udpiphdr *ui;
678 register int len = m->m_pkthdr.len;
679 struct in_addr laddr;
680 struct sockaddr_in *sin;
681 int s = 0, error = 0;
682
683 if (control)
684 m_freem(control); /* XXX */
685
686 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
687 error = EMSGSIZE;
688 goto release;
689 }
690
691 if (addr) {
692 sin = (struct sockaddr_in *)addr;
693 if (td && jailed(td->td_ucred))
694 prison_remote_ip(td->td_ucred, 0, &sin->sin_addr.s_addr);
695 laddr = inp->inp_laddr;
696 if (inp->inp_faddr.s_addr != INADDR_ANY) {
697 error = EISCONN;
698 goto release;
699 }
700 /*
701 * Must block input while temporarily connected.
702 */
703 s = splnet();
704 error = in_pcbconnect(inp, addr, td);
705 if (error) {
706 splx(s);
707 goto release;
708 }
709 } else {
710 if (inp->inp_faddr.s_addr == INADDR_ANY) {
711 error = ENOTCONN;
712 goto release;
713 }
714 }
715 /*
716 * Calculate data length and get a mbuf
717 * for UDP and IP headers.
718 */
719 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
720 if (m == 0) {
721 error = ENOBUFS;
722 if (addr)
723 splx(s);
724 goto release;
725 }
726
727 /*
728 * Fill in mbuf with extended UDP header
729 * and addresses and length put into network format.
730 */
731 ui = mtod(m, struct udpiphdr *);
732 bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */
733 ui->ui_pr = IPPROTO_UDP;
734 ui->ui_src = inp->inp_laddr;
735 ui->ui_dst = inp->inp_faddr;
736 ui->ui_sport = inp->inp_lport;
737 ui->ui_dport = inp->inp_fport;
738 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));
739
740 /*
741 * Set up checksum and output datagram.
742 */
743 if (udpcksum) {
744 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
745 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
746 m->m_pkthdr.csum_flags = CSUM_UDP;
747 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
748 } else {
749 ui->ui_sum = 0;
750 }
751 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
752 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */
753 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */
754 udpstat.udps_opackets++;
755
756#ifdef IPSEC
757 if (ipsec_setsocket(m, inp->inp_socket) != 0) {
758 error = ENOBUFS;
759 goto release;
760 }
761#endif /*IPSEC*/
762 error = ip_output(m, inp->inp_options, &inp->inp_route,
763 (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST)),
764 inp->inp_moptions);
765
766 if (addr) {
767 in_pcbdisconnect(inp);
768 inp->inp_laddr = laddr; /* XXX rehash? */
769 splx(s);
770 }
771 return (error);
772
773release:
774 m_freem(m);
775 return (error);
776}
777
778u_long udp_sendspace = 9216; /* really max datagram size */
779 /* 40 1K datagrams */
780SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
781 &udp_sendspace, 0, "Maximum outgoing UDP datagram size");
782
783u_long udp_recvspace = 40 * (1024 +
784#ifdef INET6
785 sizeof(struct sockaddr_in6)
786#else
787 sizeof(struct sockaddr_in)
788#endif
789 );
790SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
791 &udp_recvspace, 0, "Maximum incoming UDP datagram size");
792
793static int
794udp_abort(struct socket *so)
795{
796 struct inpcb *inp;
797 int s;
798
799 inp = sotoinpcb(so);
800 if (inp == 0)
801 return EINVAL; /* ??? possible? panic instead? */
802 soisdisconnected(so);
803 s = splnet();
804 in_pcbdetach(inp);
805 splx(s);
806 return 0;
807}
808
809static int
810udp_attach(struct socket *so, int proto, struct thread *td)
811{
812 struct inpcb *inp;
813 int s, error;
814
815 inp = sotoinpcb(so);
816 if (inp != 0)
817 return EINVAL;
818
819 error = soreserve(so, udp_sendspace, udp_recvspace);
820 if (error)
821 return error;
822 s = splnet();
823 error = in_pcballoc(so, &udbinfo, td);
824 splx(s);
825 if (error)
826 return error;
827
828 inp = (struct inpcb *)so->so_pcb;
829 inp->inp_vflag |= INP_IPV4;
830 inp->inp_ip_ttl = ip_defttl;
831 return 0;
832}
833
834static int
835udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
836{
837 struct inpcb *inp;
838 int s, error;
839
840 inp = sotoinpcb(so);
841 if (inp == 0)
842 return EINVAL;
843 s = splnet();
844 error = in_pcbbind(inp, nam, td);
845 splx(s);
846 return error;
847}
848
849static int
850udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
851{
852 struct inpcb *inp;
853 int s, error;
854 struct sockaddr_in *sin;
855
856 inp = sotoinpcb(so);
857 if (inp == 0)
858 return EINVAL;
859 if (inp->inp_faddr.s_addr != INADDR_ANY)
860 return EISCONN;
861 s = splnet();
862 sin = (struct sockaddr_in *)nam;
863 if (td && jailed(td->td_ucred))
864 prison_remote_ip(td->td_ucred, 0, &sin->sin_addr.s_addr);
865 error = in_pcbconnect(inp, nam, td);
866 splx(s);
867 if (error == 0)
868 soisconnected(so);
869 return error;
870}
871
872static int
873udp_detach(struct socket *so)
874{
875 struct inpcb *inp;
876 int s;
877
878 inp = sotoinpcb(so);
879 if (inp == 0)
880 return EINVAL;
881 s = splnet();
882 in_pcbdetach(inp);
883 splx(s);
884 return 0;
885}
886
887static int
888udp_disconnect(struct socket *so)
889{
890 struct inpcb *inp;
891 int s;
892
893 inp = sotoinpcb(so);
894 if (inp == 0)
895 return EINVAL;
896 if (inp->inp_faddr.s_addr == INADDR_ANY)
897 return ENOTCONN;
898
899 s = splnet();
900 in_pcbdisconnect(inp);
901 inp->inp_laddr.s_addr = INADDR_ANY;
902 splx(s);
903 so->so_state &= ~SS_ISCONNECTED; /* XXX */
904 return 0;
905}
906
907static int
908udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr,
909 struct mbuf *control, struct thread *td)
910{
911 struct inpcb *inp;
912
913 inp = sotoinpcb(so);
914 if (inp == 0) {
915 m_freem(m);
916 return EINVAL;
917 }
918 return udp_output(inp, m, addr, control, td);
919}
920
921int
922udp_shutdown(struct socket *so)
923{
924 struct inpcb *inp;
925
926 inp = sotoinpcb(so);
927 if (inp == 0)
928 return EINVAL;
929 socantsendmore(so);
930 return 0;
931}
932
933struct pr_usrreqs udp_usrreqs = {
934 udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect,
935 pru_connect2_notsupp, in_control, udp_detach, udp_disconnect,
936 pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
937 pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown,
938 in_setsockaddr, sosend, soreceive, sopoll
939};
2 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
34 * $FreeBSD: head/sys/netinet/udp_usrreq.c 95759 2002-04-30 01:54:54Z tanimura $
35 */
36
37#include "opt_ipsec.h"
38#include "opt_inet6.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/domain.h>
43#include <sys/jail.h>
44#include <sys/kernel.h>
45#include <sys/lock.h>
46#include <sys/malloc.h>
47#include <sys/mbuf.h>
48#include <sys/proc.h>
49#include <sys/protosw.h>
50#include <sys/signalvar.h>
51#include <sys/socket.h>
52#include <sys/socketvar.h>
53#include <sys/sx.h>
54#include <sys/sysctl.h>
55#include <sys/syslog.h>
56
57#include <vm/uma.h>
58
59#include <net/if.h>
60#include <net/route.h>
61
62#include <netinet/in.h>
63#include <netinet/in_systm.h>
64#include <netinet/in_pcb.h>
65#include <netinet/in_var.h>
66#include <netinet/ip.h>
67#ifdef INET6
68#include <netinet/ip6.h>
69#endif
70#include <netinet/ip_icmp.h>
71#include <netinet/icmp_var.h>
72#include <netinet/ip_var.h>
73#ifdef INET6
74#include <netinet6/ip6_var.h>
75#endif
76#include <netinet/udp.h>
77#include <netinet/udp_var.h>
78
79#ifdef IPSEC
80#include <netinet6/ipsec.h>
81#endif /*IPSEC*/
82
83#include <machine/in_cksum.h>
84
85/*
86 * UDP protocol implementation.
87 * Per RFC 768, August, 1980.
88 */
89#ifndef COMPAT_42
90static int udpcksum = 1;
91#else
92static int udpcksum = 0; /* XXX */
93#endif
94SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
95 &udpcksum, 0, "");
96
97int log_in_vain = 0;
98SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
99 &log_in_vain, 0, "Log all incoming UDP packets");
100
101static int blackhole = 0;
102SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW,
103 &blackhole, 0, "Do not send port unreachables for refused connects");
104
105struct inpcbhead udb; /* from udp_var.h */
106#define udb6 udb /* for KAME src sync over BSD*'s */
107struct inpcbinfo udbinfo;
108
109#ifndef UDBHASHSIZE
110#define UDBHASHSIZE 16
111#endif
112
113struct udpstat udpstat; /* from udp_var.h */
114SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW,
115 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)");
116
117static struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET };
118#ifdef INET6
119struct udp_in6 {
120 struct sockaddr_in6 uin6_sin;
121 u_char uin6_init_done : 1;
122} udp_in6 = {
123 { sizeof(udp_in6.uin6_sin), AF_INET6 },
124 0
125};
126struct udp_ip6 {
127 struct ip6_hdr uip6_ip6;
128 u_char uip6_init_done : 1;
129} udp_ip6;
130#endif /* INET6 */
131
132static void udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n,
133 int off);
134#ifdef INET6
135static void ip_2_ip6_hdr(struct ip6_hdr *ip6, struct ip *ip);
136#endif
137
138static int udp_detach(struct socket *so);
139static int udp_output(struct inpcb *, struct mbuf *, struct sockaddr *,
140 struct mbuf *, struct thread *);
141
142void
143udp_init()
144{
145 LIST_INIT(&udb);
146 udbinfo.listhead = &udb;
147 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
148 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
149 &udbinfo.porthashmask);
150 udbinfo.ipi_zone = uma_zcreate("udpcb", sizeof(struct inpcb), NULL,
151 NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
152 uma_zone_set_max(udbinfo.ipi_zone, maxsockets);
153}
154
155void
156udp_input(m, off)
157 register struct mbuf *m;
158 int off;
159{
160 int iphlen = off;
161 register struct ip *ip;
162 register struct udphdr *uh;
163 register struct inpcb *inp;
164 struct mbuf *opts = 0;
165 int len;
166 struct ip save_ip;
167 struct sockaddr *append_sa;
168
169 udpstat.udps_ipackets++;
170
171 /*
172 * Strip IP options, if any; should skip this,
173 * make available to user, and use on returned packets,
174 * but we don't yet have a way to check the checksum
175 * with options still present.
176 */
177 if (iphlen > sizeof (struct ip)) {
178 ip_stripoptions(m, (struct mbuf *)0);
179 iphlen = sizeof(struct ip);
180 }
181
182 /*
183 * Get IP and UDP header together in first mbuf.
184 */
185 ip = mtod(m, struct ip *);
186 if (m->m_len < iphlen + sizeof(struct udphdr)) {
187 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
188 udpstat.udps_hdrops++;
189 return;
190 }
191 ip = mtod(m, struct ip *);
192 }
193 uh = (struct udphdr *)((caddr_t)ip + iphlen);
194
195 /* destination port of 0 is illegal, based on RFC768. */
196 if (uh->uh_dport == 0)
197 goto bad;
198
199 /*
200 * Make mbuf data length reflect UDP length.
201 * If not enough data to reflect UDP length, drop.
202 */
203 len = ntohs((u_short)uh->uh_ulen);
204 if (ip->ip_len != len) {
205 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
206 udpstat.udps_badlen++;
207 goto bad;
208 }
209 m_adj(m, len - ip->ip_len);
210 /* ip->ip_len = len; */
211 }
212 /*
213 * Save a copy of the IP header in case we want restore it
214 * for sending an ICMP error message in response.
215 */
216 if (!blackhole)
217 save_ip = *ip;
218
219 /*
220 * Checksum extended UDP header and data.
221 */
222 if (uh->uh_sum) {
223 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
224 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
225 uh->uh_sum = m->m_pkthdr.csum_data;
226 else
227 uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
228 ip->ip_dst.s_addr, htonl((u_short)len +
229 m->m_pkthdr.csum_data + IPPROTO_UDP));
230 uh->uh_sum ^= 0xffff;
231 } else {
232 char b[9];
233 bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
234 bzero(((struct ipovly *)ip)->ih_x1, 9);
235 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
236 uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
237 bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
238 }
239 if (uh->uh_sum) {
240 udpstat.udps_badsum++;
241 m_freem(m);
242 return;
243 }
244 } else
245 udpstat.udps_nosum++;
246
247 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
248 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
249 struct inpcb *last;
250 /*
251 * Deliver a multicast or broadcast datagram to *all* sockets
252 * for which the local and remote addresses and ports match
253 * those of the incoming datagram. This allows more than
254 * one process to receive multi/broadcasts on the same port.
255 * (This really ought to be done for unicast datagrams as
256 * well, but that would cause problems with existing
257 * applications that open both address-specific sockets and
258 * a wildcard socket listening to the same port -- they would
259 * end up receiving duplicates of every unicast datagram.
260 * Those applications open the multiple sockets to overcome an
261 * inadequacy of the UDP socket interface, but for backwards
262 * compatibility we avoid the problem here rather than
263 * fixing the interface. Maybe 4.5BSD will remedy this?)
264 */
265
266 /*
267 * Construct sockaddr format source address.
268 */
269 udp_in.sin_port = uh->uh_sport;
270 udp_in.sin_addr = ip->ip_src;
271 /*
272 * Locate pcb(s) for datagram.
273 * (Algorithm copied from raw_intr().)
274 */
275 last = NULL;
276#ifdef INET6
277 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
278#endif
279 LIST_FOREACH(inp, &udb, inp_list) {
280#ifdef INET6
281 if ((inp->inp_vflag & INP_IPV4) == 0)
282 continue;
283#endif
284 if (inp->inp_lport != uh->uh_dport)
285 continue;
286 if (inp->inp_laddr.s_addr != INADDR_ANY) {
287 if (inp->inp_laddr.s_addr !=
288 ip->ip_dst.s_addr)
289 continue;
290 }
291 if (inp->inp_faddr.s_addr != INADDR_ANY) {
292 if (inp->inp_faddr.s_addr !=
293 ip->ip_src.s_addr ||
294 inp->inp_fport != uh->uh_sport)
295 continue;
296 }
297
298 if (last != NULL) {
299 struct mbuf *n;
300
301#ifdef IPSEC
302 /* check AH/ESP integrity. */
303 if (ipsec4_in_reject_so(m, last->inp_socket))
304 ipsecstat.in_polvio++;
305 /* do not inject data to pcb */
306 else
307#endif /*IPSEC*/
308 if ((n = m_copy(m, 0, M_COPYALL)) != NULL)
309 udp_append(last, ip, n,
310 iphlen +
311 sizeof(struct udphdr));
312 }
313 last = inp;
314 /*
315 * Don't look for additional matches if this one does
316 * not have either the SO_REUSEPORT or SO_REUSEADDR
317 * socket options set. This heuristic avoids searching
318 * through all pcbs in the common case of a non-shared
319 * port. It * assumes that an application will never
320 * clear these options after setting them.
321 */
322 if ((last->inp_socket->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
323 break;
324 }
325
326 if (last == NULL) {
327 /*
328 * No matching pcb found; discard datagram.
329 * (No need to send an ICMP Port Unreachable
330 * for a broadcast or multicast datgram.)
331 */
332 udpstat.udps_noportbcast++;
333 goto bad;
334 }
335#ifdef IPSEC
336 /* check AH/ESP integrity. */
337 if (ipsec4_in_reject_so(m, last->inp_socket)) {
338 ipsecstat.in_polvio++;
339 goto bad;
340 }
341#endif /*IPSEC*/
342 udp_append(last, ip, m, iphlen + sizeof(struct udphdr));
343 return;
344 }
345 /*
346 * Locate pcb for datagram.
347 */
348 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
349 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
350 if (inp == NULL) {
351 if (log_in_vain) {
352 char buf[4*sizeof "123"];
353
354 strcpy(buf, inet_ntoa(ip->ip_dst));
355 log(LOG_INFO,
356 "Connection attempt to UDP %s:%d from %s:%d\n",
357 buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
358 ntohs(uh->uh_sport));
359 }
360 udpstat.udps_noport++;
361 if (m->m_flags & (M_BCAST | M_MCAST)) {
362 udpstat.udps_noportbcast++;
363 goto bad;
364 }
365 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
366 goto bad;
367 if (blackhole)
368 goto bad;
369 *ip = save_ip;
370 ip->ip_len += iphlen;
371 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
372 return;
373 }
374#ifdef IPSEC
375 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
376 ipsecstat.in_polvio++;
377 goto bad;
378 }
379#endif /*IPSEC*/
380
381 /*
382 * Construct sockaddr format source address.
383 * Stuff source address and datagram in user buffer.
384 */
385 udp_in.sin_port = uh->uh_sport;
386 udp_in.sin_addr = ip->ip_src;
387 if (inp->inp_flags & INP_CONTROLOPTS
388 || inp->inp_socket->so_options & SO_TIMESTAMP) {
389#ifdef INET6
390 if (inp->inp_vflag & INP_IPV6) {
391 int savedflags;
392
393 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
394 savedflags = inp->inp_flags;
395 inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
396 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m);
397 inp->inp_flags = savedflags;
398 } else
399#endif
400 ip_savecontrol(inp, &opts, ip, m);
401 }
402 m_adj(m, iphlen + sizeof(struct udphdr));
403#ifdef INET6
404 if (inp->inp_vflag & INP_IPV6) {
405 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
406 append_sa = (struct sockaddr *)&udp_in6;
407 } else
408#endif
409 append_sa = (struct sockaddr *)&udp_in;
410 if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) {
411 udpstat.udps_fullsock++;
412 goto bad;
413 }
414 sorwakeup(inp->inp_socket);
415 return;
416bad:
417 m_freem(m);
418 if (opts)
419 m_freem(opts);
420 return;
421}
422
423#ifdef INET6
424static void
425ip_2_ip6_hdr(ip6, ip)
426 struct ip6_hdr *ip6;
427 struct ip *ip;
428{
429 bzero(ip6, sizeof(*ip6));
430
431 ip6->ip6_vfc = IPV6_VERSION;
432 ip6->ip6_plen = ip->ip_len;
433 ip6->ip6_nxt = ip->ip_p;
434 ip6->ip6_hlim = ip->ip_ttl;
435 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
436 IPV6_ADDR_INT32_SMP;
437 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
438 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
439}
440#endif
441
442/*
443 * subroutine of udp_input(), mainly for source code readability.
444 * caller must properly init udp_ip6 and udp_in6 beforehand.
445 */
446static void
447udp_append(last, ip, n, off)
448 struct inpcb *last;
449 struct ip *ip;
450 struct mbuf *n;
451 int off;
452{
453 struct sockaddr *append_sa;
454 struct mbuf *opts = 0;
455
456 if (last->inp_flags & INP_CONTROLOPTS ||
457 last->inp_socket->so_options & SO_TIMESTAMP) {
458#ifdef INET6
459 if (last->inp_vflag & INP_IPV6) {
460 int savedflags;
461
462 if (udp_ip6.uip6_init_done == 0) {
463 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
464 udp_ip6.uip6_init_done = 1;
465 }
466 savedflags = last->inp_flags;
467 last->inp_flags &= ~INP_UNMAPPABLEOPTS;
468 ip6_savecontrol(last, &opts, &udp_ip6.uip6_ip6, n);
469 last->inp_flags = savedflags;
470 } else
471#endif
472 ip_savecontrol(last, &opts, ip, n);
473 }
474#ifdef INET6
475 if (last->inp_vflag & INP_IPV6) {
476 if (udp_in6.uin6_init_done == 0) {
477 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
478 udp_in6.uin6_init_done = 1;
479 }
480 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
481 } else
482#endif
483 append_sa = (struct sockaddr *)&udp_in;
484 m_adj(n, off);
485 if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) {
486 m_freem(n);
487 if (opts)
488 m_freem(opts);
489 udpstat.udps_fullsock++;
490 } else
491 sorwakeup(last->inp_socket);
492}
493
494/*
495 * Notify a udp user of an asynchronous error;
496 * just wake up so that he can collect error status.
497 */
498void
499udp_notify(inp, errno)
500 register struct inpcb *inp;
501 int errno;
502{
503 inp->inp_socket->so_error = errno;
504 sorwakeup(inp->inp_socket);
505 sowwakeup(inp->inp_socket);
506}
507
508void
509udp_ctlinput(cmd, sa, vip)
510 int cmd;
511 struct sockaddr *sa;
512 void *vip;
513{
514 struct ip *ip = vip;
515 struct udphdr *uh;
516 void (*notify)(struct inpcb *, int) = udp_notify;
517 struct in_addr faddr;
518 struct inpcb *inp;
519 int s;
520
521 faddr = ((struct sockaddr_in *)sa)->sin_addr;
522 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
523 return;
524
525 if (PRC_IS_REDIRECT(cmd)) {
526 ip = 0;
527 notify = in_rtchange;
528 } else if (cmd == PRC_HOSTDEAD)
529 ip = 0;
530 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0)
531 return;
532 if (ip) {
533 s = splnet();
534 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
535 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport,
536 ip->ip_src, uh->uh_sport, 0, NULL);
537 if (inp != NULL && inp->inp_socket != NULL)
538 (*notify)(inp, inetctlerrmap[cmd]);
539 splx(s);
540 } else
541 in_pcbnotifyall(&udb, faddr, inetctlerrmap[cmd], notify);
542}
543
544static int
545udp_pcblist(SYSCTL_HANDLER_ARGS)
546{
547 int error, i, n, s;
548 struct inpcb *inp, **inp_list;
549 inp_gen_t gencnt;
550 struct xinpgen xig;
551
552 /*
553 * The process of preparing the TCB list is too time-consuming and
554 * resource-intensive to repeat twice on every request.
555 */
556 if (req->oldptr == 0) {
557 n = udbinfo.ipi_count;
558 req->oldidx = 2 * (sizeof xig)
559 + (n + n/8) * sizeof(struct xinpcb);
560 return 0;
561 }
562
563 if (req->newptr != 0)
564 return EPERM;
565
566 /*
567 * OK, now we're committed to doing something.
568 */
569 s = splnet();
570 gencnt = udbinfo.ipi_gencnt;
571 n = udbinfo.ipi_count;
572 splx(s);
573
574 xig.xig_len = sizeof xig;
575 xig.xig_count = n;
576 xig.xig_gen = gencnt;
577 xig.xig_sogen = so_gencnt;
578 error = SYSCTL_OUT(req, &xig, sizeof xig);
579 if (error)
580 return error;
581
582 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
583 if (inp_list == 0)
584 return ENOMEM;
585
586 s = splnet();
587 for (inp = LIST_FIRST(udbinfo.listhead), i = 0; inp && i < n;
588 inp = LIST_NEXT(inp, inp_list)) {
589 if (inp->inp_gencnt <= gencnt) {
590 if (cr_canseesocket(req->td->td_ucred,
591 inp->inp_socket))
592 continue;
593 inp_list[i++] = inp;
594 }
595 }
596 splx(s);
597 n = i;
598
599 error = 0;
600 for (i = 0; i < n; i++) {
601 inp = inp_list[i];
602 if (inp->inp_gencnt <= gencnt) {
603 struct xinpcb xi;
604 xi.xi_len = sizeof xi;
605 /* XXX should avoid extra copy */
606 bcopy(inp, &xi.xi_inp, sizeof *inp);
607 if (inp->inp_socket)
608 sotoxsocket(inp->inp_socket, &xi.xi_socket);
609 error = SYSCTL_OUT(req, &xi, sizeof xi);
610 }
611 }
612 if (!error) {
613 /*
614 * Give the user an updated idea of our state.
615 * If the generation differs from what we told
616 * her before, she knows that something happened
617 * while we were processing this request, and it
618 * might be necessary to retry.
619 */
620 s = splnet();
621 xig.xig_gen = udbinfo.ipi_gencnt;
622 xig.xig_sogen = so_gencnt;
623 xig.xig_count = udbinfo.ipi_count;
624 splx(s);
625 error = SYSCTL_OUT(req, &xig, sizeof xig);
626 }
627 free(inp_list, M_TEMP);
628 return error;
629}
630
631SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
632 udp_pcblist, "S,xinpcb", "List of active UDP sockets");
633
634static int
635udp_getcred(SYSCTL_HANDLER_ARGS)
636{
637 struct xucred xuc;
638 struct sockaddr_in addrs[2];
639 struct inpcb *inp;
640 int error, s;
641
642 error = suser_cred(req->td->td_ucred, PRISON_ROOT);
643 if (error)
644 return (error);
645 error = SYSCTL_IN(req, addrs, sizeof(addrs));
646 if (error)
647 return (error);
648 s = splnet();
649 inp = in_pcblookup_hash(&udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
650 addrs[0].sin_addr, addrs[0].sin_port, 1, NULL);
651 if (inp == NULL || inp->inp_socket == NULL) {
652 error = ENOENT;
653 goto out;
654 }
655 error = cr_canseesocket(req->td->td_ucred, inp->inp_socket);
656 if (error)
657 goto out;
658 cru2x(inp->inp_socket->so_cred, &xuc);
659 error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
660out:
661 splx(s);
662 return (error);
663}
664
665SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred,
666 CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0,
667 udp_getcred, "S,xucred", "Get the xucred of a UDP connection");
668
669static int
670udp_output(inp, m, addr, control, td)
671 register struct inpcb *inp;
672 struct mbuf *m;
673 struct sockaddr *addr;
674 struct mbuf *control;
675 struct thread *td;
676{
677 register struct udpiphdr *ui;
678 register int len = m->m_pkthdr.len;
679 struct in_addr laddr;
680 struct sockaddr_in *sin;
681 int s = 0, error = 0;
682
683 if (control)
684 m_freem(control); /* XXX */
685
686 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
687 error = EMSGSIZE;
688 goto release;
689 }
690
691 if (addr) {
692 sin = (struct sockaddr_in *)addr;
693 if (td && jailed(td->td_ucred))
694 prison_remote_ip(td->td_ucred, 0, &sin->sin_addr.s_addr);
695 laddr = inp->inp_laddr;
696 if (inp->inp_faddr.s_addr != INADDR_ANY) {
697 error = EISCONN;
698 goto release;
699 }
700 /*
701 * Must block input while temporarily connected.
702 */
703 s = splnet();
704 error = in_pcbconnect(inp, addr, td);
705 if (error) {
706 splx(s);
707 goto release;
708 }
709 } else {
710 if (inp->inp_faddr.s_addr == INADDR_ANY) {
711 error = ENOTCONN;
712 goto release;
713 }
714 }
715 /*
716 * Calculate data length and get a mbuf
717 * for UDP and IP headers.
718 */
719 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
720 if (m == 0) {
721 error = ENOBUFS;
722 if (addr)
723 splx(s);
724 goto release;
725 }
726
727 /*
728 * Fill in mbuf with extended UDP header
729 * and addresses and length put into network format.
730 */
731 ui = mtod(m, struct udpiphdr *);
732 bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */
733 ui->ui_pr = IPPROTO_UDP;
734 ui->ui_src = inp->inp_laddr;
735 ui->ui_dst = inp->inp_faddr;
736 ui->ui_sport = inp->inp_lport;
737 ui->ui_dport = inp->inp_fport;
738 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));
739
740 /*
741 * Set up checksum and output datagram.
742 */
743 if (udpcksum) {
744 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
745 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
746 m->m_pkthdr.csum_flags = CSUM_UDP;
747 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
748 } else {
749 ui->ui_sum = 0;
750 }
751 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
752 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */
753 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */
754 udpstat.udps_opackets++;
755
756#ifdef IPSEC
757 if (ipsec_setsocket(m, inp->inp_socket) != 0) {
758 error = ENOBUFS;
759 goto release;
760 }
761#endif /*IPSEC*/
762 error = ip_output(m, inp->inp_options, &inp->inp_route,
763 (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST)),
764 inp->inp_moptions);
765
766 if (addr) {
767 in_pcbdisconnect(inp);
768 inp->inp_laddr = laddr; /* XXX rehash? */
769 splx(s);
770 }
771 return (error);
772
773release:
774 m_freem(m);
775 return (error);
776}
777
778u_long udp_sendspace = 9216; /* really max datagram size */
779 /* 40 1K datagrams */
780SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
781 &udp_sendspace, 0, "Maximum outgoing UDP datagram size");
782
783u_long udp_recvspace = 40 * (1024 +
784#ifdef INET6
785 sizeof(struct sockaddr_in6)
786#else
787 sizeof(struct sockaddr_in)
788#endif
789 );
790SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
791 &udp_recvspace, 0, "Maximum incoming UDP datagram size");
792
793static int
794udp_abort(struct socket *so)
795{
796 struct inpcb *inp;
797 int s;
798
799 inp = sotoinpcb(so);
800 if (inp == 0)
801 return EINVAL; /* ??? possible? panic instead? */
802 soisdisconnected(so);
803 s = splnet();
804 in_pcbdetach(inp);
805 splx(s);
806 return 0;
807}
808
809static int
810udp_attach(struct socket *so, int proto, struct thread *td)
811{
812 struct inpcb *inp;
813 int s, error;
814
815 inp = sotoinpcb(so);
816 if (inp != 0)
817 return EINVAL;
818
819 error = soreserve(so, udp_sendspace, udp_recvspace);
820 if (error)
821 return error;
822 s = splnet();
823 error = in_pcballoc(so, &udbinfo, td);
824 splx(s);
825 if (error)
826 return error;
827
828 inp = (struct inpcb *)so->so_pcb;
829 inp->inp_vflag |= INP_IPV4;
830 inp->inp_ip_ttl = ip_defttl;
831 return 0;
832}
833
834static int
835udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
836{
837 struct inpcb *inp;
838 int s, error;
839
840 inp = sotoinpcb(so);
841 if (inp == 0)
842 return EINVAL;
843 s = splnet();
844 error = in_pcbbind(inp, nam, td);
845 splx(s);
846 return error;
847}
848
849static int
850udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
851{
852 struct inpcb *inp;
853 int s, error;
854 struct sockaddr_in *sin;
855
856 inp = sotoinpcb(so);
857 if (inp == 0)
858 return EINVAL;
859 if (inp->inp_faddr.s_addr != INADDR_ANY)
860 return EISCONN;
861 s = splnet();
862 sin = (struct sockaddr_in *)nam;
863 if (td && jailed(td->td_ucred))
864 prison_remote_ip(td->td_ucred, 0, &sin->sin_addr.s_addr);
865 error = in_pcbconnect(inp, nam, td);
866 splx(s);
867 if (error == 0)
868 soisconnected(so);
869 return error;
870}
871
872static int
873udp_detach(struct socket *so)
874{
875 struct inpcb *inp;
876 int s;
877
878 inp = sotoinpcb(so);
879 if (inp == 0)
880 return EINVAL;
881 s = splnet();
882 in_pcbdetach(inp);
883 splx(s);
884 return 0;
885}
886
887static int
888udp_disconnect(struct socket *so)
889{
890 struct inpcb *inp;
891 int s;
892
893 inp = sotoinpcb(so);
894 if (inp == 0)
895 return EINVAL;
896 if (inp->inp_faddr.s_addr == INADDR_ANY)
897 return ENOTCONN;
898
899 s = splnet();
900 in_pcbdisconnect(inp);
901 inp->inp_laddr.s_addr = INADDR_ANY;
902 splx(s);
903 so->so_state &= ~SS_ISCONNECTED; /* XXX */
904 return 0;
905}
906
907static int
908udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr,
909 struct mbuf *control, struct thread *td)
910{
911 struct inpcb *inp;
912
913 inp = sotoinpcb(so);
914 if (inp == 0) {
915 m_freem(m);
916 return EINVAL;
917 }
918 return udp_output(inp, m, addr, control, td);
919}
920
921int
922udp_shutdown(struct socket *so)
923{
924 struct inpcb *inp;
925
926 inp = sotoinpcb(so);
927 if (inp == 0)
928 return EINVAL;
929 socantsendmore(so);
930 return 0;
931}
932
933struct pr_usrreqs udp_usrreqs = {
934 udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect,
935 pru_connect2_notsupp, in_control, udp_detach, udp_disconnect,
936 pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
937 pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown,
938 in_setsockaddr, sosend, soreceive, sopoll
939};