udp_usrreq.c revision 1.31 1 /* $NetBSD: udp_usrreq.c,v 1.31 1996/05/23 16:22:32 mycroft Exp $ */
2
3 /*
4 * Copyright (c) 1982, 1986, 1988, 1990, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)udp_usrreq.c 8.4 (Berkeley) 1/21/94
36 */
37
38 #include <sys/param.h>
39 #include <sys/malloc.h>
40 #include <sys/mbuf.h>
41 #include <sys/protosw.h>
42 #include <sys/socket.h>
43 #include <sys/socketvar.h>
44 #include <sys/errno.h>
45 #include <sys/stat.h>
46 #include <sys/systm.h>
47 #include <sys/proc.h>
48
49 #include <vm/vm.h>
50 #include <sys/sysctl.h>
51
52 #include <net/if.h>
53 #include <net/route.h>
54
55 #include <netinet/in.h>
56 #include <netinet/in_systm.h>
57 #include <netinet/in_var.h>
58 #include <netinet/ip.h>
59 #include <netinet/in_pcb.h>
60 #include <netinet/ip_var.h>
61 #include <netinet/ip_icmp.h>
62 #include <netinet/udp.h>
63 #include <netinet/udp_var.h>
64
65 #include <machine/stdarg.h>
66
67 /*
68 * UDP protocol implementation.
69 * Per RFC 768, August, 1980.
70 */
71 #ifndef COMPAT_42
72 int udpcksum = 1;
73 #else
74 int udpcksum = 0; /* XXX */
75 #endif
76
77 struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET };
78
79 static void udp_notify __P((struct inpcb *, int));
80 static struct mbuf *udp_saveopt __P((caddr_t, int, int));
81
82 #ifndef UDBHASHSIZE
83 #define UDBHASHSIZE 128
84 #endif
85 int udbhashsize = UDBHASHSIZE;
86
87 void
88 udp_init()
89 {
90
91 in_pcbinit(&udbtable, udbhashsize);
92 }
93
94 void
95 #if __STDC__
96 udp_input(struct mbuf *m, ...)
97 #else
98 udp_input(m, va_alist)
99 struct mbuf *m;
100 va_dcl
101 #endif
102 {
103 register struct ip *ip;
104 register struct udphdr *uh;
105 register struct inpcb *inp;
106 struct mbuf *opts = 0;
107 int len;
108 struct ip save_ip;
109 int iphlen;
110 va_list ap;
111
112 va_start(ap, m);
113 iphlen = va_arg(ap, int);
114 va_end(ap);
115
116 udpstat.udps_ipackets++;
117
118 /*
119 * Strip IP options, if any; should skip this,
120 * make available to user, and use on returned packets,
121 * but we don't yet have a way to check the checksum
122 * with options still present.
123 */
124 if (iphlen > sizeof (struct ip)) {
125 ip_stripoptions(m, (struct mbuf *)0);
126 iphlen = sizeof(struct ip);
127 }
128
129 /*
130 * Get IP and UDP header together in first mbuf.
131 */
132 ip = mtod(m, struct ip *);
133 if (m->m_len < iphlen + sizeof(struct udphdr)) {
134 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
135 udpstat.udps_hdrops++;
136 return;
137 }
138 ip = mtod(m, struct ip *);
139 }
140 uh = (struct udphdr *)((caddr_t)ip + iphlen);
141
142 /*
143 * Make mbuf data length reflect UDP length.
144 * If not enough data to reflect UDP length, drop.
145 */
146 len = ntohs((u_int16_t)uh->uh_ulen);
147 if (ip->ip_len != len) {
148 if (len > ip->ip_len) {
149 udpstat.udps_badlen++;
150 goto bad;
151 }
152 m_adj(m, len - ip->ip_len);
153 /* ip->ip_len = len; */
154 }
155 /*
156 * Save a copy of the IP header in case we want restore it
157 * for sending an ICMP error message in response.
158 */
159 save_ip = *ip;
160
161 /*
162 * Checksum extended UDP header and data.
163 */
164 if (uh->uh_sum) {
165 bzero(((struct ipovly *)ip)->ih_x1,
166 sizeof ((struct ipovly *)ip)->ih_x1);
167 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
168 if ((uh->uh_sum = in_cksum(m, len + sizeof (struct ip))) != 0) {
169 udpstat.udps_badsum++;
170 m_freem(m);
171 return;
172 }
173 }
174
175 if (IN_MULTICAST(ip->ip_dst.s_addr) ||
176 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
177 struct socket *last;
178 /*
179 * Deliver a multicast or broadcast datagram to *all* sockets
180 * for which the local and remote addresses and ports match
181 * those of the incoming datagram. This allows more than
182 * one process to receive multi/broadcasts on the same port.
183 * (This really ought to be done for unicast datagrams as
184 * well, but that would cause problems with existing
185 * applications that open both address-specific sockets and
186 * a wildcard socket listening to the same port -- they would
187 * end up receiving duplicates of every unicast datagram.
188 * Those applications open the multiple sockets to overcome an
189 * inadequacy of the UDP socket interface, but for backwards
190 * compatibility we avoid the problem here rather than
191 * fixing the interface. Maybe 4.5BSD will remedy this?)
192 */
193
194 /*
195 * Construct sockaddr format source address.
196 */
197 udp_in.sin_port = uh->uh_sport;
198 udp_in.sin_addr = ip->ip_src;
199 m->m_len -= sizeof (struct udpiphdr);
200 m->m_data += sizeof (struct udpiphdr);
201 /*
202 * Locate pcb(s) for datagram.
203 * (Algorithm copied from raw_intr().)
204 */
205 last = NULL;
206 for (inp = udbtable.inpt_queue.cqh_first;
207 inp != (struct inpcb *)&udbtable.inpt_queue;
208 inp = inp->inp_queue.cqe_next) {
209 if (inp->inp_lport != uh->uh_dport)
210 continue;
211 if (inp->inp_laddr.s_addr != INADDR_ANY) {
212 if (inp->inp_laddr.s_addr !=
213 ip->ip_dst.s_addr)
214 continue;
215 }
216 if (inp->inp_faddr.s_addr != INADDR_ANY) {
217 if (inp->inp_faddr.s_addr !=
218 ip->ip_src.s_addr ||
219 inp->inp_fport != uh->uh_sport)
220 continue;
221 }
222
223 if (last != NULL) {
224 struct mbuf *n;
225
226 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) {
227 if (sbappendaddr(&last->so_rcv,
228 sintosa(&udp_in), n,
229 (struct mbuf *)0) == 0) {
230 m_freem(n);
231 udpstat.udps_fullsock++;
232 } else
233 sorwakeup(last);
234 }
235 }
236 last = inp->inp_socket;
237 /*
238 * Don't look for additional matches if this one does
239 * not have either the SO_REUSEPORT or SO_REUSEADDR
240 * socket options set. This heuristic avoids searching
241 * through all pcbs in the common case of a non-shared
242 * port. It * assumes that an application will never
243 * clear these options after setting them.
244 */
245 if ((last->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
246 break;
247 }
248
249 if (last == NULL) {
250 /*
251 * No matching pcb found; discard datagram.
252 * (No need to send an ICMP Port Unreachable
253 * for a broadcast or multicast datgram.)
254 */
255 udpstat.udps_noportbcast++;
256 goto bad;
257 }
258 if (sbappendaddr(&last->so_rcv, sintosa(&udp_in), m,
259 (struct mbuf *)0) == 0) {
260 udpstat.udps_fullsock++;
261 goto bad;
262 }
263 sorwakeup(last);
264 return;
265 }
266 /*
267 * Locate pcb for datagram.
268 */
269 inp = in_pcbhashlookup(&udbtable, ip->ip_src, uh->uh_sport,
270 ip->ip_dst, uh->uh_dport);
271 if (inp == 0) {
272 ++udpstat.udps_pcbhashmiss;
273 inp = in_pcblookup(&udbtable, ip->ip_src, uh->uh_sport,
274 ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD);
275 if (inp == 0) {
276 udpstat.udps_noport++;
277 if (m->m_flags & (M_BCAST | M_MCAST)) {
278 udpstat.udps_noportbcast++;
279 goto bad;
280 }
281 *ip = save_ip;
282 ip->ip_len += iphlen;
283 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
284 return;
285 }
286 }
287
288 /*
289 * Construct sockaddr format source address.
290 * Stuff source address and datagram in user buffer.
291 */
292 udp_in.sin_port = uh->uh_sport;
293 udp_in.sin_addr = ip->ip_src;
294 if (inp->inp_flags & INP_CONTROLOPTS) {
295 struct mbuf **mp = &opts;
296
297 if (inp->inp_flags & INP_RECVDSTADDR) {
298 *mp = udp_saveopt((caddr_t) &ip->ip_dst,
299 sizeof(struct in_addr), IP_RECVDSTADDR);
300 if (*mp)
301 mp = &(*mp)->m_next;
302 }
303 #ifdef notyet
304 /* options were tossed above */
305 if (inp->inp_flags & INP_RECVOPTS) {
306 *mp = udp_saveopt((caddr_t) opts_deleted_above,
307 sizeof(struct in_addr), IP_RECVOPTS);
308 if (*mp)
309 mp = &(*mp)->m_next;
310 }
311 /* ip_srcroute doesn't do what we want here, need to fix */
312 if (inp->inp_flags & INP_RECVRETOPTS) {
313 *mp = udp_saveopt((caddr_t) ip_srcroute(),
314 sizeof(struct in_addr), IP_RECVRETOPTS);
315 if (*mp)
316 mp = &(*mp)->m_next;
317 }
318 #endif
319 }
320 iphlen += sizeof(struct udphdr);
321 m->m_len -= iphlen;
322 m->m_pkthdr.len -= iphlen;
323 m->m_data += iphlen;
324 if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udp_in), m,
325 opts) == 0) {
326 udpstat.udps_fullsock++;
327 goto bad;
328 }
329 sorwakeup(inp->inp_socket);
330 return;
331 bad:
332 m_freem(m);
333 if (opts)
334 m_freem(opts);
335 }
336
337 /*
338 * Create a "control" mbuf containing the specified data
339 * with the specified type for presentation with a datagram.
340 */
341 struct mbuf *
342 udp_saveopt(p, size, type)
343 caddr_t p;
344 register int size;
345 int type;
346 {
347 register struct cmsghdr *cp;
348 struct mbuf *m;
349
350 if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL)
351 return ((struct mbuf *) NULL);
352 cp = (struct cmsghdr *) mtod(m, struct cmsghdr *);
353 bcopy(p, CMSG_DATA(cp), size);
354 size += sizeof(*cp);
355 m->m_len = size;
356 cp->cmsg_len = size;
357 cp->cmsg_level = IPPROTO_IP;
358 cp->cmsg_type = type;
359 return (m);
360 }
361
362 /*
363 * Notify a udp user of an asynchronous error;
364 * just wake up so that he can collect error status.
365 */
366 static void
367 udp_notify(inp, errno)
368 register struct inpcb *inp;
369 int errno;
370 {
371 inp->inp_socket->so_error = errno;
372 sorwakeup(inp->inp_socket);
373 sowwakeup(inp->inp_socket);
374 }
375
376 void *
377 udp_ctlinput(cmd, sa, v)
378 int cmd;
379 struct sockaddr *sa;
380 void *v;
381 {
382 register struct ip *ip = v;
383 register struct udphdr *uh;
384 extern int inetctlerrmap[];
385 void (*notify) __P((struct inpcb *, int)) = udp_notify;
386 int errno;
387
388 if ((unsigned)cmd >= PRC_NCMDS)
389 return NULL;
390 errno = inetctlerrmap[cmd];
391 if (PRC_IS_REDIRECT(cmd))
392 notify = in_rtchange, ip = 0;
393 else if (cmd == PRC_HOSTDEAD)
394 ip = 0;
395 else if (errno == 0)
396 return NULL;
397 if (ip) {
398 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
399 in_pcbnotify(&udbtable, sa, uh->uh_dport, ip->ip_src,
400 uh->uh_sport, errno, notify);
401 } else
402 in_pcbnotifyall(&udbtable, sa, errno, notify);
403 return NULL;
404 }
405
406 int
407 #if __STDC__
408 udp_output(struct mbuf *m, ...)
409 #else
410 udp_output(m, va_alist)
411 struct mbuf *m;
412 va_dcl
413 #endif
414 {
415 struct mbuf *control;
416 register struct inpcb *inp;
417 register struct udpiphdr *ui;
418 register int len = m->m_pkthdr.len;
419 int error = 0;
420 va_list ap;
421
422 va_start(ap, m);
423 control = va_arg(ap, struct mbuf *);
424 inp = va_arg(ap, struct inpcb *);
425 va_end(ap);
426
427 if (control)
428 m_freem(control); /* XXX */
429
430 /*
431 * Calculate data length and get a mbuf
432 * for UDP and IP headers.
433 */
434 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
435 if (m == 0) {
436 error = ENOBUFS;
437 goto release;
438 }
439
440 /*
441 * Fill in mbuf with extended UDP header
442 * and addresses and length put into network format.
443 */
444 ui = mtod(m, struct udpiphdr *);
445 bzero(ui->ui_x1, sizeof ui->ui_x1);
446 ui->ui_pr = IPPROTO_UDP;
447 ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr));
448 ui->ui_src = inp->inp_laddr;
449 ui->ui_dst = inp->inp_faddr;
450 ui->ui_sport = inp->inp_lport;
451 ui->ui_dport = inp->inp_fport;
452 ui->ui_ulen = ui->ui_len;
453
454 /*
455 * Stuff checksum and output datagram.
456 */
457 ui->ui_sum = 0;
458 if (udpcksum) {
459 if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0)
460 ui->ui_sum = 0xffff;
461 }
462 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
463 ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */
464 ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */
465 udpstat.udps_opackets++;
466 return (ip_output(m, inp->inp_options, &inp->inp_route,
467 inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST),
468 inp->inp_moptions));
469
470 release:
471 m_freem(m);
472 return (error);
473 }
474
475 u_long udp_sendspace = 9216; /* really max datagram size */
476 u_long udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in));
477 /* 40 1K datagrams */
478
479 /*ARGSUSED*/
480 int
481 udp_usrreq(so, req, m, nam, control, p)
482 struct socket *so;
483 int req;
484 struct mbuf *m, *nam, *control;
485 struct proc *p;
486 {
487 register struct inpcb *inp;
488 int s;
489 register int error = 0;
490
491 if (req == PRU_CONTROL)
492 return (in_control(so, (long)m, (caddr_t)nam,
493 (struct ifnet *)control, p));
494
495 s = splsoftnet();
496 inp = sotoinpcb(so);
497 if (control && control->m_len) {
498 m_freem(control);
499 error = EINVAL;
500 goto release;
501 }
502 if (inp == 0 && req != PRU_ATTACH) {
503 error = EINVAL;
504 goto release;
505 }
506
507 /*
508 * Note: need to block udp_input while changing
509 * the udp pcb queue and/or pcb addresses.
510 */
511 switch (req) {
512
513 case PRU_ATTACH:
514 if (inp != 0) {
515 error = EISCONN;
516 break;
517 }
518 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
519 error = soreserve(so, udp_sendspace, udp_recvspace);
520 if (error)
521 break;
522 }
523 error = in_pcballoc(so, &udbtable);
524 if (error)
525 break;
526 inp = sotoinpcb(so);
527 inp->inp_ip.ip_ttl = ip_defttl;
528 break;
529
530 case PRU_DETACH:
531 in_pcbdetach(inp);
532 break;
533
534 case PRU_BIND:
535 error = in_pcbbind(inp, nam, p);
536 break;
537
538 case PRU_LISTEN:
539 error = EOPNOTSUPP;
540 break;
541
542 case PRU_CONNECT:
543 error = in_pcbconnect(inp, nam);
544 if (error)
545 break;
546 soisconnected(so);
547 break;
548
549 case PRU_CONNECT2:
550 error = EOPNOTSUPP;
551 break;
552
553 case PRU_DISCONNECT:
554 /*soisdisconnected(so);*/
555 so->so_state &= ~SS_ISCONNECTED; /* XXX */
556 in_pcbdisconnect(inp);
557 inp->inp_laddr.s_addr = INADDR_ANY; /* XXX */
558 break;
559
560 case PRU_SHUTDOWN:
561 socantsendmore(so);
562 break;
563
564 case PRU_RCVD:
565 error = EOPNOTSUPP;
566 break;
567
568 case PRU_SEND:
569 {
570 struct in_addr laddr;
571
572 if (nam) {
573 laddr = inp->inp_laddr;
574 if ((so->so_state & SS_ISCONNECTED) != 0) {
575 error = EISCONN;
576 break;
577 }
578 error = in_pcbconnect(inp, nam);
579 if (error)
580 break;
581 } else {
582 if ((so->so_state & SS_ISCONNECTED) == 0) {
583 error = ENOTCONN;
584 break;
585 }
586 }
587 error = udp_output(m, control, inp);
588 if (nam) {
589 in_pcbdisconnect(inp);
590 inp->inp_laddr = laddr;
591 }
592 }
593 break;
594
595 case PRU_SENSE:
596 /*
597 * stat: don't bother with a blocksize.
598 */
599 splx(s);
600 return (0);
601
602 case PRU_RCVOOB:
603 error = EOPNOTSUPP;
604 break;
605
606 case PRU_SENDOOB:
607 m_freem(m);
608 error = EOPNOTSUPP;
609 break;
610
611 case PRU_SOCKADDR:
612 in_setsockaddr(inp, nam);
613 break;
614
615 case PRU_PEERADDR:
616 in_setpeeraddr(inp, nam);
617 break;
618
619 default:
620 panic("udp_usrreq");
621 }
622
623 release:
624 splx(s);
625 return (error);
626 }
627
628 /*
629 * Sysctl for udp variables.
630 */
631 int
632 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
633 int *name;
634 u_int namelen;
635 void *oldp;
636 size_t *oldlenp;
637 void *newp;
638 size_t newlen;
639 {
640 /* All sysctl names at this level are terminal. */
641 if (namelen != 1)
642 return (ENOTDIR);
643
644 switch (name[0]) {
645 case UDPCTL_CHECKSUM:
646 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum));
647 default:
648 return (ENOPROTOOPT);
649 }
650 /* NOTREACHED */
651 }
652