Once we distribute socket protocol processing requests to different
[dragonfly.git] / sys / netinet / udp_usrreq.c
CommitLineData
984263bc
MD
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: src/sys/netinet/udp_usrreq.c,v 1.64.2.18 2003/01/24 05:11:34 sam Exp $
e4700d00 35 * $DragonFly: src/sys/netinet/udp_usrreq.c,v 1.12 2004/03/05 16:57:15 hsu Exp $
984263bc
MD
36 */
37
38#include "opt_ipsec.h"
39#include "opt_inet6.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/kernel.h>
44#include <sys/malloc.h>
45#include <sys/mbuf.h>
46#include <sys/domain.h>
47#include <sys/proc.h>
48#include <sys/protosw.h>
49#include <sys/socket.h>
50#include <sys/socketvar.h>
51#include <sys/sysctl.h>
52#include <sys/syslog.h>
3f9db7f8 53#include <sys/in_cksum.h>
984263bc
MD
54
55#include <vm/vm_zone.h>
56
57#include <net/if.h>
58#include <net/route.h>
59
60#include <netinet/in.h>
61#include <netinet/in_systm.h>
62#include <netinet/ip.h>
63#ifdef INET6
64#include <netinet/ip6.h>
65#endif
66#include <netinet/in_pcb.h>
67#include <netinet/in_var.h>
68#include <netinet/ip_var.h>
69#ifdef INET6
70#include <netinet6/ip6_var.h>
71#endif
72#include <netinet/ip_icmp.h>
73#include <netinet/icmp_var.h>
74#include <netinet/udp.h>
75#include <netinet/udp_var.h>
76
77#ifdef FAST_IPSEC
78#include <netipsec/ipsec.h>
79#endif /*FAST_IPSEC*/
80
81#ifdef IPSEC
82#include <netinet6/ipsec.h>
83#endif /*IPSEC*/
84
984263bc
MD
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
56ae102f
MD
105static int strict_mcast_mship = 1;
106SYSCTL_INT(_net_inet_udp, OID_AUTO, strict_mcast_mship, CTLFLAG_RW,
107 &strict_mcast_mship, 0, "Only send multicast to member sockets");
108
984263bc
MD
109struct inpcbhead udb; /* from udp_var.h */
110#define udb6 udb /* for KAME src sync over BSD*'s */
111struct inpcbinfo udbinfo;
112
113#ifndef UDBHASHSIZE
114#define UDBHASHSIZE 16
115#endif
116
117struct udpstat udpstat; /* from udp_var.h */
118SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RW,
119 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)");
120
121static struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET };
122#ifdef INET6
123struct udp_in6 {
124 struct sockaddr_in6 uin6_sin;
125 u_char uin6_init_done : 1;
126} udp_in6 = {
127 { sizeof(udp_in6.uin6_sin), AF_INET6 },
128 0
129};
130struct udp_ip6 {
131 struct ip6_hdr uip6_ip6;
132 u_char uip6_init_done : 1;
133} udp_ip6;
134#endif /* INET6 */
135
42a7fc75
RG
136static void udp_append (struct inpcb *last, struct ip *ip,
137 struct mbuf *n, int off);
984263bc 138#ifdef INET6
42a7fc75 139static void ip_2_ip6_hdr (struct ip6_hdr *ip6, struct ip *ip);
984263bc
MD
140#endif
141
42a7fc75
RG
142static int udp_detach (struct socket *so);
143static int udp_output (struct inpcb *, struct mbuf *, struct sockaddr *,
144 struct mbuf *, struct thread *);
984263bc
MD
145
146void
147udp_init()
148{
149 LIST_INIT(&udb);
150 udbinfo.listhead = &udb;
151 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
152 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
153 &udbinfo.porthashmask);
13d8907a
JH
154 udbinfo.bindhashbase = hashinit(UDBHASHSIZE, M_PCB,
155 &udbinfo.bindhashmask);
984263bc
MD
156 udbinfo.ipi_zone = zinit("udpcb", sizeof(struct inpcb), maxsockets,
157 ZONE_INTERRUPT, 0);
bf82f9b7 158 udp_thread_init();
984263bc
MD
159}
160
56ae102f
MD
161/*
162 * Check multicast packets to make sure they are only sent to sockets with
163 * multicast memberships for the packet's destination address and arrival
164 * interface. Multicast packets to multicast-unaware sockets are also
165 * disallowed.
166 *
167 * Returns 0 if the packet is acceptable, -1 if it is not.
168 */
169static __inline
170int
171check_multicast_membership(struct ip *ip, struct inpcb *inp, struct mbuf *m)
172{
173 int mshipno;
174 struct ip_moptions *mopt;
175
176 if (strict_mcast_mship == 0 ||
177 !IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
178 return(0);
179 }
180 mopt = inp->inp_moptions;
181 if (mopt == NULL)
182 return(-1);
183 for (mshipno = 0; mshipno <= mopt->imo_num_memberships; ++mshipno) {
184 if (ip->ip_dst.s_addr == mopt->imo_membership[mshipno]->inm_addr.s_addr &&
185 m->m_pkthdr.rcvif == mopt->imo_membership[mshipno]->inm_ifp) {
186 return(0);
187 }
188 }
189 return(-1);
190}
191
984263bc
MD
192void
193udp_input(m, off, proto)
2256ba69 194 struct mbuf *m;
984263bc
MD
195 int off, proto;
196{
197 int iphlen = off;
2256ba69
RG
198 struct ip *ip;
199 struct udphdr *uh;
200 struct inpcb *inp;
984263bc
MD
201 struct mbuf *opts = 0;
202 int len;
203 struct ip save_ip;
204 struct sockaddr *append_sa;
205
206 udpstat.udps_ipackets++;
207
208 /*
209 * Strip IP options, if any; should skip this,
210 * make available to user, and use on returned packets,
211 * but we don't yet have a way to check the checksum
212 * with options still present.
213 */
214 if (iphlen > sizeof (struct ip)) {
215 ip_stripoptions(m, (struct mbuf *)0);
216 iphlen = sizeof(struct ip);
217 }
218
219 /*
220 * Get IP and UDP header together in first mbuf.
221 */
222 ip = mtod(m, struct ip *);
223 if (m->m_len < iphlen + sizeof(struct udphdr)) {
224 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
225 udpstat.udps_hdrops++;
226 return;
227 }
228 ip = mtod(m, struct ip *);
229 }
230 uh = (struct udphdr *)((caddr_t)ip + iphlen);
231
232 /* destination port of 0 is illegal, based on RFC768. */
233 if (uh->uh_dport == 0)
234 goto bad;
235
236 /*
237 * Make mbuf data length reflect UDP length.
238 * If not enough data to reflect UDP length, drop.
239 */
240 len = ntohs((u_short)uh->uh_ulen);
241 if (ip->ip_len != len) {
242 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
243 udpstat.udps_badlen++;
244 goto bad;
245 }
246 m_adj(m, len - ip->ip_len);
247 /* ip->ip_len = len; */
248 }
249 /*
250 * Save a copy of the IP header in case we want restore it
251 * for sending an ICMP error message in response.
252 */
253 save_ip = *ip;
254
255 /*
256 * Checksum extended UDP header and data.
257 */
258 if (uh->uh_sum) {
259 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
260 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
261 uh->uh_sum = m->m_pkthdr.csum_data;
262 else
263 uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
264 ip->ip_dst.s_addr, htonl((u_short)len +
265 m->m_pkthdr.csum_data + IPPROTO_UDP));
266 uh->uh_sum ^= 0xffff;
267 } else {
268 char b[9];
269 bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
270 bzero(((struct ipovly *)ip)->ih_x1, 9);
271 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
272 uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
273 bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
274 }
275 if (uh->uh_sum) {
276 udpstat.udps_badsum++;
277 m_freem(m);
278 return;
279 }
280 } else
281 udpstat.udps_nosum++;
282
283 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
284 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
285 struct inpcb *last;
286 /*
287 * Deliver a multicast or broadcast datagram to *all* sockets
288 * for which the local and remote addresses and ports match
289 * those of the incoming datagram. This allows more than
290 * one process to receive multi/broadcasts on the same port.
291 * (This really ought to be done for unicast datagrams as
292 * well, but that would cause problems with existing
293 * applications that open both address-specific sockets and
294 * a wildcard socket listening to the same port -- they would
295 * end up receiving duplicates of every unicast datagram.
296 * Those applications open the multiple sockets to overcome an
297 * inadequacy of the UDP socket interface, but for backwards
298 * compatibility we avoid the problem here rather than
299 * fixing the interface. Maybe 4.5BSD will remedy this?)
300 */
301
302 /*
303 * Construct sockaddr format source address.
304 */
305 udp_in.sin_port = uh->uh_sport;
306 udp_in.sin_addr = ip->ip_src;
307 /*
308 * Locate pcb(s) for datagram.
309 * (Algorithm copied from raw_intr().)
310 */
311 last = NULL;
312#ifdef INET6
313 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
314#endif
315 LIST_FOREACH(inp, &udb, inp_list) {
316#ifdef INET6
317 if ((inp->inp_vflag & INP_IPV4) == 0)
318 continue;
319#endif
320 if (inp->inp_lport != uh->uh_dport)
321 continue;
322 if (inp->inp_laddr.s_addr != INADDR_ANY) {
323 if (inp->inp_laddr.s_addr !=
324 ip->ip_dst.s_addr)
325 continue;
326 }
327 if (inp->inp_faddr.s_addr != INADDR_ANY) {
328 if (inp->inp_faddr.s_addr !=
329 ip->ip_src.s_addr ||
330 inp->inp_fport != uh->uh_sport)
331 continue;
332 }
333
56ae102f
MD
334 if (check_multicast_membership(ip, inp, m) < 0)
335 continue;
336
984263bc
MD
337 if (last != NULL) {
338 struct mbuf *n;
339
340#ifdef IPSEC
341 /* check AH/ESP integrity. */
342 if (ipsec4_in_reject_so(m, last->inp_socket))
343 ipsecstat.in_polvio++;
344 /* do not inject data to pcb */
345 else
346#endif /*IPSEC*/
347#ifdef FAST_IPSEC
348 /* check AH/ESP integrity. */
349 if (ipsec4_in_reject(m, last))
350 ;
351 else
352#endif /*FAST_IPSEC*/
353 if ((n = m_copy(m, 0, M_COPYALL)) != NULL)
354 udp_append(last, ip, n,
355 iphlen +
356 sizeof(struct udphdr));
357 }
358 last = inp;
359 /*
360 * Don't look for additional matches if this one does
361 * not have either the SO_REUSEPORT or SO_REUSEADDR
362 * socket options set. This heuristic avoids searching
363 * through all pcbs in the common case of a non-shared
364 * port. It * assumes that an application will never
365 * clear these options after setting them.
366 */
367 if ((last->inp_socket->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
368 break;
369 }
370
371 if (last == NULL) {
372 /*
373 * No matching pcb found; discard datagram.
374 * (No need to send an ICMP Port Unreachable
375 * for a broadcast or multicast datgram.)
376 */
377 udpstat.udps_noportbcast++;
378 goto bad;
379 }
380#ifdef IPSEC
381 /* check AH/ESP integrity. */
382 if (ipsec4_in_reject_so(m, last->inp_socket)) {
383 ipsecstat.in_polvio++;
384 goto bad;
385 }
386#endif /*IPSEC*/
387#ifdef FAST_IPSEC
388 /* check AH/ESP integrity. */
389 if (ipsec4_in_reject(m, last))
390 goto bad;
391#endif /*FAST_IPSEC*/
392 udp_append(last, ip, m, iphlen + sizeof(struct udphdr));
393 return;
394 }
395 /*
396 * Locate pcb for datagram.
397 */
398 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
399 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
400 if (inp == NULL) {
401 if (log_in_vain) {
402 char buf[4*sizeof "123"];
403
404 strcpy(buf, inet_ntoa(ip->ip_dst));
405 log(LOG_INFO,
406 "Connection attempt to UDP %s:%d from %s:%d\n",
407 buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
408 ntohs(uh->uh_sport));
409 }
410 udpstat.udps_noport++;
411 if (m->m_flags & (M_BCAST | M_MCAST)) {
412 udpstat.udps_noportbcast++;
413 goto bad;
414 }
415 if (blackhole)
416 goto bad;
417#ifdef ICMP_BANDLIM
418 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
419 goto bad;
420#endif
421 *ip = save_ip;
422 ip->ip_len += iphlen;
423 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
424 return;
425 }
426#ifdef IPSEC
427 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
428 ipsecstat.in_polvio++;
429 goto bad;
430 }
431#endif /*IPSEC*/
432#ifdef FAST_IPSEC
433 if (ipsec4_in_reject(m, inp))
434 goto bad;
435#endif /*FAST_IPSEC*/
436
437 /*
438 * Construct sockaddr format source address.
439 * Stuff source address and datagram in user buffer.
440 */
441 udp_in.sin_port = uh->uh_sport;
442 udp_in.sin_addr = ip->ip_src;
443 if (inp->inp_flags & INP_CONTROLOPTS
444 || inp->inp_socket->so_options & SO_TIMESTAMP) {
445#ifdef INET6
446 if (inp->inp_vflag & INP_IPV6) {
447 int savedflags;
448
449 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
450 savedflags = inp->inp_flags;
451 inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
452 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m);
453 inp->inp_flags = savedflags;
454 } else
455#endif
456 ip_savecontrol(inp, &opts, ip, m);
457 }
458 m_adj(m, iphlen + sizeof(struct udphdr));
459#ifdef INET6
460 if (inp->inp_vflag & INP_IPV6) {
461 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
462 append_sa = (struct sockaddr *)&udp_in6;
463 } else
464#endif
465 append_sa = (struct sockaddr *)&udp_in;
466 if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts) == 0) {
467 udpstat.udps_fullsock++;
468 goto bad;
469 }
470 sorwakeup(inp->inp_socket);
471 return;
472bad:
473 m_freem(m);
474 if (opts)
475 m_freem(opts);
476 return;
477}
478
479#ifdef INET6
480static void
481ip_2_ip6_hdr(ip6, ip)
482 struct ip6_hdr *ip6;
483 struct ip *ip;
484{
485 bzero(ip6, sizeof(*ip6));
486
487 ip6->ip6_vfc = IPV6_VERSION;
488 ip6->ip6_plen = ip->ip_len;
489 ip6->ip6_nxt = ip->ip_p;
490 ip6->ip6_hlim = ip->ip_ttl;
491 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
492 IPV6_ADDR_INT32_SMP;
493 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
494 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
495}
496#endif
497
498/*
499 * subroutine of udp_input(), mainly for source code readability.
500 * caller must properly init udp_ip6 and udp_in6 beforehand.
501 */
502static void
503udp_append(last, ip, n, off)
504 struct inpcb *last;
505 struct ip *ip;
506 struct mbuf *n;
507 int off;
508{
509 struct sockaddr *append_sa;
510 struct mbuf *opts = 0;
511
512 if (last->inp_flags & INP_CONTROLOPTS ||
513 last->inp_socket->so_options & SO_TIMESTAMP) {
514#ifdef INET6
515 if (last->inp_vflag & INP_IPV6) {
516 int savedflags;
517
518 if (udp_ip6.uip6_init_done == 0) {
519 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
520 udp_ip6.uip6_init_done = 1;
521 }
522 savedflags = last->inp_flags;
523 last->inp_flags &= ~INP_UNMAPPABLEOPTS;
524 ip6_savecontrol(last, &opts, &udp_ip6.uip6_ip6, n);
525 last->inp_flags = savedflags;
526 } else
527#endif
528 ip_savecontrol(last, &opts, ip, n);
529 }
530#ifdef INET6
531 if (last->inp_vflag & INP_IPV6) {
532 if (udp_in6.uin6_init_done == 0) {
533 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
534 udp_in6.uin6_init_done = 1;
535 }
536 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
537 } else
538#endif
539 append_sa = (struct sockaddr *)&udp_in;
540 m_adj(n, off);
541 if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts) == 0) {
542 m_freem(n);
543 if (opts)
544 m_freem(opts);
545 udpstat.udps_fullsock++;
546 } else
547 sorwakeup(last->inp_socket);
548}
549
550/*
551 * Notify a udp user of an asynchronous error;
552 * just wake up so that he can collect error status.
553 */
554void
555udp_notify(inp, errno)
2256ba69 556 struct inpcb *inp;
984263bc
MD
557 int errno;
558{
559 inp->inp_socket->so_error = errno;
560 sorwakeup(inp->inp_socket);
561 sowwakeup(inp->inp_socket);
562}
563
564void
565udp_ctlinput(cmd, sa, vip)
566 int cmd;
567 struct sockaddr *sa;
568 void *vip;
569{
570 struct ip *ip = vip;
571 struct udphdr *uh;
42a7fc75 572 void (*notify) (struct inpcb *, int) = udp_notify;
984263bc
MD
573 struct in_addr faddr;
574 struct inpcb *inp;
575 int s;
576
577 faddr = ((struct sockaddr_in *)sa)->sin_addr;
578 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
579 return;
580
581 if (PRC_IS_REDIRECT(cmd)) {
582 ip = 0;
583 notify = in_rtchange;
584 } else if (cmd == PRC_HOSTDEAD)
585 ip = 0;
586 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0)
587 return;
588 if (ip) {
589 s = splnet();
590 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
591 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport,
592 ip->ip_src, uh->uh_sport, 0, NULL);
593 if (inp != NULL && inp->inp_socket != NULL)
594 (*notify)(inp, inetctlerrmap[cmd]);
595 splx(s);
596 } else
597 in_pcbnotifyall(&udb, faddr, inetctlerrmap[cmd], notify);
598}
599
600static int
601udp_pcblist(SYSCTL_HANDLER_ARGS)
602{
603 int error, i, n, s;
604 struct inpcb *inp, **inp_list;
605 inp_gen_t gencnt;
606 struct xinpgen xig;
607
608 /*
609 * The process of preparing the TCB list is too time-consuming and
610 * resource-intensive to repeat twice on every request.
611 */
612 if (req->oldptr == 0) {
613 n = udbinfo.ipi_count;
614 req->oldidx = 2 * (sizeof xig)
615 + (n + n/8) * sizeof(struct xinpcb);
616 return 0;
617 }
618
619 if (req->newptr != 0)
620 return EPERM;
621
622 /*
623 * OK, now we're committed to doing something.
624 */
625 s = splnet();
626 gencnt = udbinfo.ipi_gencnt;
627 n = udbinfo.ipi_count;
628 splx(s);
629
630 xig.xig_len = sizeof xig;
631 xig.xig_count = n;
632 xig.xig_gen = gencnt;
633 xig.xig_sogen = so_gencnt;
634 error = SYSCTL_OUT(req, &xig, sizeof xig);
635 if (error)
636 return error;
637
638 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
639 if (inp_list == 0)
640 return ENOMEM;
641
642 s = splnet();
643 for (inp = LIST_FIRST(udbinfo.listhead), i = 0; inp && i < n;
644 inp = LIST_NEXT(inp, inp_list)) {
dadab5e9 645 if (inp->inp_gencnt <= gencnt && !prison_xinpcb(req->td, inp))
984263bc
MD
646 inp_list[i++] = inp;
647 }
648 splx(s);
649 n = i;
650
651 error = 0;
652 for (i = 0; i < n; i++) {
653 inp = inp_list[i];
654 if (inp->inp_gencnt <= gencnt) {
655 struct xinpcb xi;
656 xi.xi_len = sizeof xi;
657 /* XXX should avoid extra copy */
658 bcopy(inp, &xi.xi_inp, sizeof *inp);
659 if (inp->inp_socket)
660 sotoxsocket(inp->inp_socket, &xi.xi_socket);
661 error = SYSCTL_OUT(req, &xi, sizeof xi);
662 }
663 }
664 if (!error) {
665 /*
666 * Give the user an updated idea of our state.
667 * If the generation differs from what we told
668 * her before, she knows that something happened
669 * while we were processing this request, and it
670 * might be necessary to retry.
671 */
672 s = splnet();
673 xig.xig_gen = udbinfo.ipi_gencnt;
674 xig.xig_sogen = so_gencnt;
675 xig.xig_count = udbinfo.ipi_count;
676 splx(s);
677 error = SYSCTL_OUT(req, &xig, sizeof xig);
678 }
679 free(inp_list, M_TEMP);
680 return error;
681}
682
683SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
684 udp_pcblist, "S,xinpcb", "List of active UDP sockets");
685
686static int
687udp_getcred(SYSCTL_HANDLER_ARGS)
688{
689 struct sockaddr_in addrs[2];
690 struct inpcb *inp;
691 int error, s;
692
dadab5e9 693 error = suser(req->td);
984263bc
MD
694 if (error)
695 return (error);
696 error = SYSCTL_IN(req, addrs, sizeof(addrs));
697 if (error)
698 return (error);
699 s = splnet();
700 inp = in_pcblookup_hash(&udbinfo, addrs[1].sin_addr, addrs[1].sin_port,
701 addrs[0].sin_addr, addrs[0].sin_port, 1, NULL);
702 if (inp == NULL || inp->inp_socket == NULL) {
703 error = ENOENT;
704 goto out;
705 }
706 error = SYSCTL_OUT(req, inp->inp_socket->so_cred, sizeof(struct ucred));
707out:
708 splx(s);
709 return (error);
710}
711
712SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW,
713 0, 0, udp_getcred, "S,ucred", "Get the ucred of a UDP connection");
714
715static int
dadab5e9
MD
716udp_output(inp, m, addr, control, td)
717 struct inpcb *inp;
984263bc
MD
718 struct mbuf *m;
719 struct sockaddr *addr;
720 struct mbuf *control;
dadab5e9 721 struct thread *td;
984263bc 722{
dadab5e9
MD
723 struct udpiphdr *ui;
724 int len = m->m_pkthdr.len;
984263bc
MD
725 struct in_addr laddr;
726 struct sockaddr_in *sin;
727 int s = 0, error = 0;
728
729 if (control)
730 m_freem(control); /* XXX */
731
732 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
733 error = EMSGSIZE;
734 goto release;
735 }
736
737 if (addr) {
738 sin = (struct sockaddr_in *)addr;
dadab5e9 739 prison_remote_ip(td, 0, &sin->sin_addr.s_addr);
984263bc
MD
740 laddr = inp->inp_laddr;
741 if (inp->inp_faddr.s_addr != INADDR_ANY) {
742 error = EISCONN;
743 goto release;
744 }
745 /*
746 * Must block input while temporarily connected.
747 */
748 s = splnet();
dadab5e9 749 error = in_pcbconnect(inp, addr, td);
984263bc
MD
750 if (error) {
751 splx(s);
752 goto release;
753 }
754 } else {
755 if (inp->inp_faddr.s_addr == INADDR_ANY) {
756 error = ENOTCONN;
757 goto release;
758 }
759 }
760 /*
761 * Calculate data length and get a mbuf
762 * for UDP and IP headers.
763 */
764 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
765 if (m == 0) {
766 error = ENOBUFS;
767 if (addr)
768 splx(s);
769 goto release;
770 }
771
772 /*
773 * Fill in mbuf with extended UDP header
774 * and addresses and length put into network format.
775 */
776 ui = mtod(m, struct udpiphdr *);
777 bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */
778 ui->ui_pr = IPPROTO_UDP;
779 ui->ui_src = inp->inp_laddr;
780 ui->ui_dst = inp->inp_faddr;
781 ui->ui_sport = inp->inp_lport;
782 ui->ui_dport = inp->inp_fport;
783 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));
784
785 /*
786 * Set up checksum and output datagram.
787 */
788 if (udpcksum) {
789 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
790 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
791 m->m_pkthdr.csum_flags = CSUM_UDP;
792 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
793 } else {
794 ui->ui_sum = 0;
795 }
796 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
797 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */
798 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */
799 udpstat.udps_opackets++;
800
801 error = ip_output(m, inp->inp_options, &inp->inp_route,
802 (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST)),
803 inp->inp_moptions, inp);
804
805 if (addr) {
806 in_pcbdisconnect(inp);
688bf2e6
JH
807 inp->inp_laddr = laddr;
808 in_pcbinsbindhash(inp);
984263bc
MD
809 splx(s);
810 }
811 return (error);
812
813release:
814 m_freem(m);
815 return (error);
816}
817
818u_long udp_sendspace = 9216; /* really max datagram size */
819 /* 40 1K datagrams */
820SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
821 &udp_sendspace, 0, "Maximum outgoing UDP datagram size");
822
823u_long udp_recvspace = 40 * (1024 +
824#ifdef INET6
825 sizeof(struct sockaddr_in6)
826#else
827 sizeof(struct sockaddr_in)
828#endif
829 );
830SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
831 &udp_recvspace, 0, "Maximum incoming UDP datagram size");
832
833static int
834udp_abort(struct socket *so)
835{
836 struct inpcb *inp;
837 int s;
838
839 inp = sotoinpcb(so);
840 if (inp == 0)
841 return EINVAL; /* ??? possible? panic instead? */
842 soisdisconnected(so);
843 s = splnet();
844 in_pcbdetach(inp);
845 splx(s);
846 return 0;
847}
848
849static int
e4700d00 850udp_attach(struct socket *so, int proto, struct pru_attach_info *ai)
984263bc
MD
851{
852 struct inpcb *inp;
853 int s, error;
854
855 inp = sotoinpcb(so);
856 if (inp != 0)
857 return EINVAL;
858
e4700d00 859 error = soreserve(so, udp_sendspace, udp_recvspace, ai->sb_rlimit);
984263bc
MD
860 if (error)
861 return error;
862 s = splnet();
e4700d00 863 error = in_pcballoc(so, &udbinfo);
984263bc
MD
864 splx(s);
865 if (error)
866 return error;
867
868 inp = (struct inpcb *)so->so_pcb;
869 inp->inp_vflag |= INP_IPV4;
870 inp->inp_ip_ttl = ip_defttl;
871 return 0;
872}
873
874static int
dadab5e9 875udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
984263bc
MD
876{
877 struct inpcb *inp;
878 int s, error;
879
880 inp = sotoinpcb(so);
881 if (inp == 0)
882 return EINVAL;
883 s = splnet();
dadab5e9 884 error = in_pcbbind(inp, nam, td);
984263bc
MD
885 splx(s);
886 return error;
887}
888
889static int
dadab5e9 890udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
984263bc
MD
891{
892 struct inpcb *inp;
893 int s, error;
894 struct sockaddr_in *sin;
895
896 inp = sotoinpcb(so);
897 if (inp == 0)
898 return EINVAL;
899 if (inp->inp_faddr.s_addr != INADDR_ANY)
900 return EISCONN;
901 error = 0;
902 s = splnet();
dadab5e9
MD
903 if (inp->inp_laddr.s_addr == INADDR_ANY &&
904 td->td_proc &&
905 td->td_proc->p_ucred->cr_prison != NULL) {
906 error = in_pcbbind(inp, NULL, td);
907 }
984263bc
MD
908 if (error == 0) {
909 sin = (struct sockaddr_in *)nam;
dadab5e9
MD
910 prison_remote_ip(td, 0, &sin->sin_addr.s_addr);
911 error = in_pcbconnect(inp, nam, td);
984263bc
MD
912 }
913 splx(s);
914 if (error == 0)
915 soisconnected(so);
916 return error;
917}
918
919static int
920udp_detach(struct socket *so)
921{
922 struct inpcb *inp;
923 int s;
924
925 inp = sotoinpcb(so);
926 if (inp == 0)
927 return EINVAL;
928 s = splnet();
929 in_pcbdetach(inp);
930 splx(s);
931 return 0;
932}
933
934static int
935udp_disconnect(struct socket *so)
936{
937 struct inpcb *inp;
938 int s;
939
940 inp = sotoinpcb(so);
941 if (inp == 0)
942 return EINVAL;
943 if (inp->inp_faddr.s_addr == INADDR_ANY)
944 return ENOTCONN;
945
946 s = splnet();
947 in_pcbdisconnect(inp);
948 inp->inp_laddr.s_addr = INADDR_ANY;
688bf2e6 949 in_pcbinsbindhash(inp);
984263bc
MD
950 splx(s);
951 so->so_state &= ~SS_ISCONNECTED; /* XXX */
952 return 0;
953}
954
955static int
956udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr,
dadab5e9 957 struct mbuf *control, struct thread *td)
984263bc
MD
958{
959 struct inpcb *inp;
960
961 inp = sotoinpcb(so);
962 if (inp == 0) {
963 m_freem(m);
964 return EINVAL;
965 }
dadab5e9 966 return udp_output(inp, m, addr, control, td);
984263bc
MD
967}
968
969int
970udp_shutdown(struct socket *so)
971{
972 struct inpcb *inp;
973
974 inp = sotoinpcb(so);
975 if (inp == 0)
976 return EINVAL;
977 socantsendmore(so);
978 return 0;
979}
980
981struct pr_usrreqs udp_usrreqs = {
982 udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect,
983 pru_connect2_notsupp, in_control, udp_detach, udp_disconnect,
984 pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
985 pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown,
986 in_setsockaddr, sosend, soreceive, sopoll
987};
988