udp_usrreq.c revision 1.34 1 /* $NetBSD: udp_usrreq.c,v 1.34 1996/09/09 14:51:23 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 static void udp_notify __P((struct inpcb *, int));
78 static struct mbuf *udp_saveopt __P((caddr_t, int, int));
79
80 #ifndef UDBHASHSIZE
81 #define UDBHASHSIZE 128
82 #endif
83 int udbhashsize = UDBHASHSIZE;
84
85 void
86 udp_init()
87 {
88
89 in_pcbinit(&udbtable, udbhashsize);
90 }
91
92 void
93 #if __STDC__
94 udp_input(struct mbuf *m, ...)
95 #else
96 udp_input(m, va_alist)
97 struct mbuf *m;
98 va_dcl
99 #endif
100 {
101 register struct ip *ip;
102 register struct udphdr *uh;
103 register struct inpcb *inp;
104 struct mbuf *opts = 0;
105 int len;
106 struct ip save_ip;
107 int iphlen;
108 va_list ap;
109 struct sockaddr_in udpsrc;
110
111 va_start(ap, m);
112 iphlen = va_arg(ap, int);
113 va_end(ap);
114
115 udpstat.udps_ipackets++;
116
117 /*
118 * Strip IP options, if any; should skip this,
119 * make available to user, and use on returned packets,
120 * but we don't yet have a way to check the checksum
121 * with options still present.
122 */
123 if (iphlen > sizeof (struct ip)) {
124 ip_stripoptions(m, (struct mbuf *)0);
125 iphlen = sizeof(struct ip);
126 }
127
128 /*
129 * Get IP and UDP header together in first mbuf.
130 */
131 ip = mtod(m, struct ip *);
132 if (m->m_len < iphlen + sizeof(struct udphdr)) {
133 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
134 udpstat.udps_hdrops++;
135 return;
136 }
137 ip = mtod(m, struct ip *);
138 }
139 uh = (struct udphdr *)((caddr_t)ip + iphlen);
140
141 /*
142 * Make mbuf data length reflect UDP length.
143 * If not enough data to reflect UDP length, drop.
144 */
145 len = ntohs((u_int16_t)uh->uh_ulen);
146 if (ip->ip_len != len) {
147 if (len > ip->ip_len) {
148 udpstat.udps_badlen++;
149 goto bad;
150 }
151 m_adj(m, len - ip->ip_len);
152 /* ip->ip_len = len; */
153 }
154 /*
155 * Save a copy of the IP header in case we want restore it
156 * for sending an ICMP error message in response.
157 */
158 save_ip = *ip;
159
160 /*
161 * Checksum extended UDP header and data.
162 */
163 if (uh->uh_sum) {
164 bzero(((struct ipovly *)ip)->ih_x1,
165 sizeof ((struct ipovly *)ip)->ih_x1);
166 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
167 if ((uh->uh_sum = in_cksum(m, len + sizeof (struct ip))) != 0) {
168 udpstat.udps_badsum++;
169 m_freem(m);
170 return;
171 }
172 }
173
174 if (IN_MULTICAST(ip->ip_dst.s_addr) ||
175 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
176 struct socket *last;
177 /*
178 * Deliver a multicast or broadcast datagram to *all* sockets
179 * for which the local and remote addresses and ports match
180 * those of the incoming datagram. This allows more than
181 * one process to receive multi/broadcasts on the same port.
182 * (This really ought to be done for unicast datagrams as
183 * well, but that would cause problems with existing
184 * applications that open both address-specific sockets and
185 * a wildcard socket listening to the same port -- they would
186 * end up receiving duplicates of every unicast datagram.
187 * Those applications open the multiple sockets to overcome an
188 * inadequacy of the UDP socket interface, but for backwards
189 * compatibility we avoid the problem here rather than
190 * fixing the interface. Maybe 4.5BSD will remedy this?)
191 */
192
193 /*
194 * Construct sockaddr format source address.
195 */
196 udpsrc.sin_family = AF_INET;
197 udpsrc.sin_len = sizeof(struct sockaddr_in);
198 udpsrc.sin_port = uh->uh_sport;
199 udpsrc.sin_addr = ip->ip_src;
200
201 m->m_len -= sizeof (struct udpiphdr);
202 m->m_data += sizeof (struct udpiphdr);
203 /*
204 * Locate pcb(s) for datagram.
205 * (Algorithm copied from raw_intr().)
206 */
207 last = NULL;
208 for (inp = udbtable.inpt_queue.cqh_first;
209 inp != (struct inpcb *)&udbtable.inpt_queue;
210 inp = inp->inp_queue.cqe_next) {
211 if (inp->inp_lport != uh->uh_dport)
212 continue;
213 if (!in_nullhost(inp->inp_laddr)) {
214 if (!in_hosteq(inp->inp_laddr, ip->ip_dst))
215 continue;
216 }
217 if (!in_nullhost(inp->inp_faddr)) {
218 if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
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(&udpsrc), 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(&udpsrc), 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 udpsrc.sin_family = AF_INET;
293 udpsrc.sin_len = sizeof(struct sockaddr_in);
294 udpsrc.sin_port = uh->uh_sport;
295 udpsrc.sin_addr = ip->ip_src;
296
297 if (inp->inp_flags & INP_CONTROLOPTS) {
298 struct mbuf **mp = &opts;
299
300 if (inp->inp_flags & INP_RECVDSTADDR) {
301 *mp = udp_saveopt((caddr_t) &ip->ip_dst,
302 sizeof(struct in_addr), IP_RECVDSTADDR);
303 if (*mp)
304 mp = &(*mp)->m_next;
305 }
306 #ifdef notyet
307 /* options were tossed above */
308 if (inp->inp_flags & INP_RECVOPTS) {
309 *mp = udp_saveopt((caddr_t) opts_deleted_above,
310 sizeof(struct in_addr), IP_RECVOPTS);
311 if (*mp)
312 mp = &(*mp)->m_next;
313 }
314 /* ip_srcroute doesn't do what we want here, need to fix */
315 if (inp->inp_flags & INP_RECVRETOPTS) {
316 *mp = udp_saveopt((caddr_t) ip_srcroute(),
317 sizeof(struct in_addr), IP_RECVRETOPTS);
318 if (*mp)
319 mp = &(*mp)->m_next;
320 }
321 #endif
322 }
323 iphlen += sizeof(struct udphdr);
324 m->m_len -= iphlen;
325 m->m_pkthdr.len -= iphlen;
326 m->m_data += iphlen;
327 if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udpsrc), m,
328 opts) == 0) {
329 udpstat.udps_fullsock++;
330 goto bad;
331 }
332 sorwakeup(inp->inp_socket);
333 return;
334 bad:
335 m_freem(m);
336 if (opts)
337 m_freem(opts);
338 }
339
340 /*
341 * Create a "control" mbuf containing the specified data
342 * with the specified type for presentation with a datagram.
343 */
344 struct mbuf *
345 udp_saveopt(p, size, type)
346 caddr_t p;
347 register int size;
348 int type;
349 {
350 register struct cmsghdr *cp;
351 struct mbuf *m;
352
353 if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL)
354 return ((struct mbuf *) NULL);
355 cp = (struct cmsghdr *) mtod(m, struct cmsghdr *);
356 bcopy(p, CMSG_DATA(cp), size);
357 size += sizeof(*cp);
358 m->m_len = size;
359 cp->cmsg_len = size;
360 cp->cmsg_level = IPPROTO_IP;
361 cp->cmsg_type = type;
362 return (m);
363 }
364
365 /*
366 * Notify a udp user of an asynchronous error;
367 * just wake up so that he can collect error status.
368 */
369 static void
370 udp_notify(inp, errno)
371 register struct inpcb *inp;
372 int errno;
373 {
374
375 inp->inp_socket->so_error = errno;
376 sorwakeup(inp->inp_socket);
377 sowwakeup(inp->inp_socket);
378 }
379
380 void *
381 udp_ctlinput(cmd, sa, v)
382 int cmd;
383 struct sockaddr *sa;
384 void *v;
385 {
386 register struct ip *ip = v;
387 register struct udphdr *uh;
388 extern int inetctlerrmap[];
389 void (*notify) __P((struct inpcb *, int)) = udp_notify;
390 int errno;
391
392 if ((unsigned)cmd >= PRC_NCMDS)
393 return NULL;
394 errno = inetctlerrmap[cmd];
395 if (PRC_IS_REDIRECT(cmd))
396 notify = in_rtchange, ip = 0;
397 else if (cmd == PRC_HOSTDEAD)
398 ip = 0;
399 else if (errno == 0)
400 return NULL;
401 if (ip) {
402 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
403 in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport,
404 ip->ip_src, uh->uh_sport, errno, notify);
405 } else
406 in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno,
407 notify);
408 return NULL;
409 }
410
411 int
412 #if __STDC__
413 udp_output(struct mbuf *m, ...)
414 #else
415 udp_output(m, va_alist)
416 struct mbuf *m;
417 va_dcl
418 #endif
419 {
420 register struct inpcb *inp;
421 register struct udpiphdr *ui;
422 register int len = m->m_pkthdr.len;
423 int error = 0;
424 va_list ap;
425
426 va_start(ap, m);
427 inp = va_arg(ap, struct inpcb *);
428 va_end(ap);
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 #ifdef DIAGNOSTIC
498 if (req != PRU_SEND && req != PRU_SENDOOB && control)
499 panic("udp_usrreq: unexpected control mbuf");
500 #endif
501 if (inp == 0 && req != PRU_ATTACH) {
502 error = EINVAL;
503 goto release;
504 }
505
506 /*
507 * Note: need to block udp_input while changing
508 * the udp pcb queue and/or pcb addresses.
509 */
510 switch (req) {
511
512 case PRU_ATTACH:
513 if (inp != 0) {
514 error = EISCONN;
515 break;
516 }
517 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
518 error = soreserve(so, udp_sendspace, udp_recvspace);
519 if (error)
520 break;
521 }
522 error = in_pcballoc(so, &udbtable);
523 if (error)
524 break;
525 inp = sotoinpcb(so);
526 inp->inp_ip.ip_ttl = ip_defttl;
527 break;
528
529 case PRU_DETACH:
530 in_pcbdetach(inp);
531 break;
532
533 case PRU_BIND:
534 error = in_pcbbind(inp, nam, p);
535 break;
536
537 case PRU_LISTEN:
538 error = EOPNOTSUPP;
539 break;
540
541 case PRU_CONNECT:
542 error = in_pcbconnect(inp, nam);
543 if (error)
544 break;
545 soisconnected(so);
546 break;
547
548 case PRU_CONNECT2:
549 error = EOPNOTSUPP;
550 break;
551
552 case PRU_DISCONNECT:
553 /*soisdisconnected(so);*/
554 so->so_state &= ~SS_ISCONNECTED; /* XXX */
555 in_pcbdisconnect(inp);
556 inp->inp_laddr = zeroin_addr; /* XXX */
557 break;
558
559 case PRU_SHUTDOWN:
560 socantsendmore(so);
561 break;
562
563 case PRU_RCVD:
564 error = EOPNOTSUPP;
565 break;
566
567 case PRU_SEND:
568 if (control && control->m_len) {
569 m_freem(control);
570 m_freem(m);
571 error = EINVAL;
572 break;
573 }
574 {
575 struct in_addr laddr;
576
577 if (nam) {
578 laddr = inp->inp_laddr;
579 if ((so->so_state & SS_ISCONNECTED) != 0) {
580 error = EISCONN;
581 goto die;
582 }
583 error = in_pcbconnect(inp, nam);
584 if (error) {
585 die:
586 m_freem(m);
587 break;
588 }
589 } else {
590 if ((so->so_state & SS_ISCONNECTED) == 0) {
591 error = ENOTCONN;
592 goto die;
593 }
594 }
595 error = udp_output(m, inp);
596 if (nam) {
597 in_pcbdisconnect(inp);
598 inp->inp_laddr = laddr;
599 }
600 }
601 break;
602
603 case PRU_SENSE:
604 /*
605 * stat: don't bother with a blocksize.
606 */
607 splx(s);
608 return (0);
609
610 case PRU_RCVOOB:
611 error = EOPNOTSUPP;
612 break;
613
614 case PRU_SENDOOB:
615 m_freem(control);
616 m_freem(m);
617 error = EOPNOTSUPP;
618 break;
619
620 case PRU_SOCKADDR:
621 in_setsockaddr(inp, nam);
622 break;
623
624 case PRU_PEERADDR:
625 in_setpeeraddr(inp, nam);
626 break;
627
628 default:
629 panic("udp_usrreq");
630 }
631
632 release:
633 splx(s);
634 return (error);
635 }
636
637 /*
638 * Sysctl for udp variables.
639 */
640 int
641 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
642 int *name;
643 u_int namelen;
644 void *oldp;
645 size_t *oldlenp;
646 void *newp;
647 size_t newlen;
648 {
649 /* All sysctl names at this level are terminal. */
650 if (namelen != 1)
651 return (ENOTDIR);
652
653 switch (name[0]) {
654 case UDPCTL_CHECKSUM:
655 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum));
656 default:
657 return (ENOPROTOOPT);
658 }
659 /* NOTREACHED */
660 }
661