udp_usrreq.c revision 1.42 1 /* $NetBSD: udp_usrreq.c,v 1.42 1997/07/28 22:19:53 thorpej 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 #include "ipkdb.h"
38
39 #include <sys/param.h>
40 #include <sys/malloc.h>
41 #include <sys/mbuf.h>
42 #include <sys/protosw.h>
43 #include <sys/socket.h>
44 #include <sys/socketvar.h>
45 #include <sys/errno.h>
46 #include <sys/stat.h>
47 #include <sys/systm.h>
48 #include <sys/proc.h>
49
50 #include <vm/vm.h>
51 #include <sys/sysctl.h>
52
53 #include <net/if.h>
54 #include <net/route.h>
55
56 #include <netinet/in.h>
57 #include <netinet/in_systm.h>
58 #include <netinet/in_var.h>
59 #include <netinet/ip.h>
60 #include <netinet/in_pcb.h>
61 #include <netinet/ip_var.h>
62 #include <netinet/ip_icmp.h>
63 #include <netinet/udp.h>
64 #include <netinet/udp_var.h>
65
66 #include <machine/stdarg.h>
67
68 /*
69 * UDP protocol implementation.
70 * Per RFC 768, August, 1980.
71 */
72 #ifndef COMPAT_42
73 int udpcksum = 1;
74 #else
75 int udpcksum = 0; /* XXX */
76 #endif
77
78 static void udp_notify __P((struct inpcb *, 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, 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 inpcb *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_addr = ip->ip_src;
199 udpsrc.sin_port = uh->uh_sport;
200 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero));
201
202 m->m_len -= sizeof (struct udpiphdr);
203 m->m_data += sizeof (struct udpiphdr);
204 /*
205 * Locate pcb(s) for datagram.
206 * (Algorithm copied from raw_intr().)
207 */
208 last = NULL;
209 for (inp = udbtable.inpt_queue.cqh_first;
210 inp != (struct inpcb *)&udbtable.inpt_queue;
211 inp = inp->inp_queue.cqe_next) {
212 if (inp->inp_lport != uh->uh_dport)
213 continue;
214 if (!in_nullhost(inp->inp_laddr)) {
215 if (!in_hosteq(inp->inp_laddr, ip->ip_dst))
216 continue;
217 }
218 if (!in_nullhost(inp->inp_faddr)) {
219 if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
220 inp->inp_fport != uh->uh_sport)
221 continue;
222 }
223
224 if (last != NULL) {
225 struct mbuf *n;
226
227 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) {
228 if (last->inp_flags & INP_CONTROLOPTS
229 || last->inp_socket->so_options &
230 SO_TIMESTAMP) {
231 ip_savecontrol(last, &opts,
232 ip, n);
233 }
234 if (sbappendaddr(
235 &last->inp_socket->so_rcv,
236 sintosa(&udpsrc), n, opts) == 0) {
237 m_freem(n);
238 if (opts)
239 m_freem(opts);
240 } else
241 sorwakeup(last->inp_socket);
242 opts = 0;
243 }
244 }
245 last = inp;
246 /*
247 * Don't look for additional matches if this one does
248 * not have either the SO_REUSEPORT or SO_REUSEADDR
249 * socket options set. This heuristic avoids searching
250 * through all pcbs in the common case of a non-shared
251 * port. It * assumes that an application will never
252 * clear these options after setting them.
253 */
254 if ((last->inp_socket->so_options &
255 (SO_REUSEPORT|SO_REUSEADDR)) == 0)
256 break;
257 }
258
259 if (last == NULL) {
260 /*
261 * No matching pcb found; discard datagram.
262 * (No need to send an ICMP Port Unreachable
263 * for a broadcast or multicast datgram.)
264 */
265 udpstat.udps_noportbcast++;
266 goto bad;
267 }
268 if (last->inp_flags & INP_CONTROLOPTS ||
269 last->inp_socket->so_options & SO_TIMESTAMP)
270 ip_savecontrol(last, &opts, ip, m);
271 if (sbappendaddr(&last->inp_socket->so_rcv,
272 sintosa(&udpsrc), m, opts) == 0) {
273 udpstat.udps_fullsock++;
274 goto bad;
275 }
276 sorwakeup(last->inp_socket);
277 return;
278 }
279 /*
280 * Locate pcb for datagram.
281 */
282 inp = in_pcblookup_connect(&udbtable, ip->ip_src, uh->uh_sport,
283 ip->ip_dst, uh->uh_dport);
284 if (inp == 0) {
285 ++udpstat.udps_pcbhashmiss;
286 inp = in_pcblookup_bind(&udbtable, ip->ip_dst, uh->uh_dport);
287 if (inp == 0) {
288 udpstat.udps_noport++;
289 if (m->m_flags & (M_BCAST | M_MCAST)) {
290 udpstat.udps_noportbcast++;
291 goto bad;
292 }
293 *ip = save_ip;
294 #if NIPKDB > 0
295 if (checkipkdb(&ip->ip_src,
296 uh->uh_sport,
297 uh->uh_dport,
298 m,
299 iphlen + sizeof(struct udphdr),
300 len - sizeof(struct udphdr)))
301 /* It was a debugger connect packet, just drop it now */
302 goto bad;
303 #endif
304 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
305 return;
306 }
307 }
308
309 /*
310 * Construct sockaddr format source address.
311 * Stuff source address and datagram in user buffer.
312 */
313 udpsrc.sin_family = AF_INET;
314 udpsrc.sin_len = sizeof(struct sockaddr_in);
315 udpsrc.sin_addr = ip->ip_src;
316 udpsrc.sin_port = uh->uh_sport;
317 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero));
318
319 if (inp->inp_flags & INP_CONTROLOPTS ||
320 inp->inp_socket->so_options & SO_TIMESTAMP)
321 ip_savecontrol(inp, &opts, ip, m);
322 iphlen += sizeof(struct udphdr);
323 m->m_len -= iphlen;
324 m->m_pkthdr.len -= iphlen;
325 m->m_data += iphlen;
326 if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udpsrc), m,
327 opts) == 0) {
328 udpstat.udps_fullsock++;
329 goto bad;
330 }
331 sorwakeup(inp->inp_socket);
332 return;
333 bad:
334 m_freem(m);
335 if (opts)
336 m_freem(opts);
337 }
338
339 /*
340 * Notify a udp user of an asynchronous error;
341 * just wake up so that he can collect error status.
342 */
343 static void
344 udp_notify(inp, errno)
345 register struct inpcb *inp;
346 int errno;
347 {
348
349 inp->inp_socket->so_error = errno;
350 sorwakeup(inp->inp_socket);
351 sowwakeup(inp->inp_socket);
352 }
353
354 void *
355 udp_ctlinput(cmd, sa, v)
356 int cmd;
357 struct sockaddr *sa;
358 void *v;
359 {
360 register struct ip *ip = v;
361 register struct udphdr *uh;
362 extern int inetctlerrmap[];
363 void (*notify) __P((struct inpcb *, int)) = udp_notify;
364 int errno;
365
366 if ((unsigned)cmd >= PRC_NCMDS)
367 return NULL;
368 errno = inetctlerrmap[cmd];
369 if (PRC_IS_REDIRECT(cmd))
370 notify = in_rtchange, ip = 0;
371 else if (cmd == PRC_HOSTDEAD)
372 ip = 0;
373 else if (errno == 0)
374 return NULL;
375 if (ip) {
376 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
377 in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport,
378 ip->ip_src, uh->uh_sport, errno, notify);
379 } else
380 in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno,
381 notify);
382 return NULL;
383 }
384
385 int
386 #if __STDC__
387 udp_output(struct mbuf *m, ...)
388 #else
389 udp_output(m, va_alist)
390 struct mbuf *m;
391 va_dcl
392 #endif
393 {
394 register struct inpcb *inp;
395 register struct udpiphdr *ui;
396 register int len = m->m_pkthdr.len;
397 int error = 0;
398 va_list ap;
399
400 va_start(ap, m);
401 inp = va_arg(ap, struct inpcb *);
402 va_end(ap);
403
404 /*
405 * Calculate data length and get a mbuf
406 * for UDP and IP headers.
407 */
408 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
409 if (m == 0) {
410 error = ENOBUFS;
411 goto release;
412 }
413
414 /*
415 * Compute the packet length of the IP header, and
416 * punt if the length looks bogus.
417 */
418 if ((len + sizeof(struct udpiphdr)) > IP_MAXPACKET) {
419 error = EMSGSIZE;
420 goto release;
421 }
422
423 /*
424 * Fill in mbuf with extended UDP header
425 * and addresses and length put into network format.
426 */
427 ui = mtod(m, struct udpiphdr *);
428 bzero(ui->ui_x1, sizeof ui->ui_x1);
429 ui->ui_pr = IPPROTO_UDP;
430 ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr));
431 ui->ui_src = inp->inp_laddr;
432 ui->ui_dst = inp->inp_faddr;
433 ui->ui_sport = inp->inp_lport;
434 ui->ui_dport = inp->inp_fport;
435 ui->ui_ulen = ui->ui_len;
436
437 /*
438 * Stuff checksum and output datagram.
439 */
440 ui->ui_sum = 0;
441 if (udpcksum) {
442 if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0)
443 ui->ui_sum = 0xffff;
444 }
445 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
446 ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */
447 ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */
448 udpstat.udps_opackets++;
449 return (ip_output(m, inp->inp_options, &inp->inp_route,
450 inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST),
451 inp->inp_moptions));
452
453 release:
454 m_freem(m);
455 return (error);
456 }
457
458 int udp_sendspace = 9216; /* really max datagram size */
459 int udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in));
460 /* 40 1K datagrams */
461
462 /*ARGSUSED*/
463 int
464 udp_usrreq(so, req, m, nam, control, p)
465 struct socket *so;
466 int req;
467 struct mbuf *m, *nam, *control;
468 struct proc *p;
469 {
470 register struct inpcb *inp;
471 int s;
472 register int error = 0;
473
474 if (req == PRU_CONTROL)
475 return (in_control(so, (long)m, (caddr_t)nam,
476 (struct ifnet *)control, p));
477
478 s = splsoftnet();
479 inp = sotoinpcb(so);
480 #ifdef DIAGNOSTIC
481 if (req != PRU_SEND && req != PRU_SENDOOB && control)
482 panic("udp_usrreq: unexpected control mbuf");
483 #endif
484 if (inp == 0 && req != PRU_ATTACH) {
485 error = EINVAL;
486 goto release;
487 }
488
489 /*
490 * Note: need to block udp_input while changing
491 * the udp pcb queue and/or pcb addresses.
492 */
493 switch (req) {
494
495 case PRU_ATTACH:
496 if (inp != 0) {
497 error = EISCONN;
498 break;
499 }
500 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
501 error = soreserve(so, udp_sendspace, udp_recvspace);
502 if (error)
503 break;
504 }
505 error = in_pcballoc(so, &udbtable);
506 if (error)
507 break;
508 inp = sotoinpcb(so);
509 inp->inp_ip.ip_ttl = ip_defttl;
510 break;
511
512 case PRU_DETACH:
513 in_pcbdetach(inp);
514 break;
515
516 case PRU_BIND:
517 error = in_pcbbind(inp, nam, p);
518 break;
519
520 case PRU_LISTEN:
521 error = EOPNOTSUPP;
522 break;
523
524 case PRU_CONNECT:
525 error = in_pcbconnect(inp, nam);
526 if (error)
527 break;
528 soisconnected(so);
529 break;
530
531 case PRU_CONNECT2:
532 error = EOPNOTSUPP;
533 break;
534
535 case PRU_DISCONNECT:
536 /*soisdisconnected(so);*/
537 so->so_state &= ~SS_ISCONNECTED; /* XXX */
538 in_pcbdisconnect(inp);
539 inp->inp_laddr = zeroin_addr; /* XXX */
540 in_pcbstate(inp, INP_BOUND); /* XXX */
541 break;
542
543 case PRU_SHUTDOWN:
544 socantsendmore(so);
545 break;
546
547 case PRU_RCVD:
548 error = EOPNOTSUPP;
549 break;
550
551 case PRU_SEND:
552 if (control && control->m_len) {
553 m_freem(control);
554 m_freem(m);
555 error = EINVAL;
556 break;
557 }
558 {
559 struct in_addr laddr; /* XXX */
560
561 if (nam) {
562 laddr = inp->inp_laddr; /* XXX */
563 if ((so->so_state & SS_ISCONNECTED) != 0) {
564 error = EISCONN;
565 goto die;
566 }
567 error = in_pcbconnect(inp, nam);
568 if (error) {
569 die:
570 m_freem(m);
571 break;
572 }
573 } else {
574 if ((so->so_state & SS_ISCONNECTED) == 0) {
575 error = ENOTCONN;
576 goto die;
577 }
578 }
579 error = udp_output(m, inp);
580 if (nam) {
581 in_pcbdisconnect(inp);
582 inp->inp_laddr = laddr; /* XXX */
583 in_pcbstate(inp, INP_BOUND); /* XXX */
584 }
585 }
586 break;
587
588 case PRU_SENSE:
589 /*
590 * stat: don't bother with a blocksize.
591 */
592 splx(s);
593 return (0);
594
595 case PRU_RCVOOB:
596 error = EOPNOTSUPP;
597 break;
598
599 case PRU_SENDOOB:
600 m_freem(control);
601 m_freem(m);
602 error = EOPNOTSUPP;
603 break;
604
605 case PRU_SOCKADDR:
606 in_setsockaddr(inp, nam);
607 break;
608
609 case PRU_PEERADDR:
610 in_setpeeraddr(inp, nam);
611 break;
612
613 default:
614 panic("udp_usrreq");
615 }
616
617 release:
618 splx(s);
619 return (error);
620 }
621
622 /*
623 * Sysctl for udp variables.
624 */
625 int
626 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
627 int *name;
628 u_int namelen;
629 void *oldp;
630 size_t *oldlenp;
631 void *newp;
632 size_t newlen;
633 {
634 /* All sysctl names at this level are terminal. */
635 if (namelen != 1)
636 return (ENOTDIR);
637
638 switch (name[0]) {
639 case UDPCTL_CHECKSUM:
640 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum));
641 case UDPCTL_SENDSPACE:
642 return (sysctl_int(oldp, oldlenp, newp, newlen,
643 &udp_sendspace));
644 case UDPCTL_RECVSPACE:
645 return (sysctl_int(oldp, oldlenp, newp, newlen,
646 &udp_recvspace));
647 default:
648 return (ENOPROTOOPT);
649 }
650 /* NOTREACHED */
651 }
652