udp_usrreq.c revision 1.44 1 /* $NetBSD: udp_usrreq.c,v 1.44 1998/01/05 10:32:16 thorpej Exp $ */
2
3 /*
4 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
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 8.6 (Berkeley) 5/23/95
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 iphlen += sizeof(struct udphdr);
203 m->m_len -= iphlen;
204 m->m_pkthdr.len -= iphlen;
205 m->m_data += iphlen;
206 /*
207 * Locate pcb(s) for datagram.
208 * (Algorithm copied from raw_intr().)
209 */
210 last = NULL;
211 for (inp = udbtable.inpt_queue.cqh_first;
212 inp != (struct inpcb *)&udbtable.inpt_queue;
213 inp = inp->inp_queue.cqe_next) {
214 if (inp->inp_lport != uh->uh_dport)
215 continue;
216 if (!in_nullhost(inp->inp_laddr)) {
217 if (!in_hosteq(inp->inp_laddr, ip->ip_dst))
218 continue;
219 }
220 if (!in_nullhost(inp->inp_faddr)) {
221 if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
222 inp->inp_fport != uh->uh_sport)
223 continue;
224 }
225
226 if (last != NULL) {
227 struct mbuf *n;
228
229 if ((n = m_copy(m, 0, M_COPYALL)) != NULL) {
230 if (last->inp_flags & INP_CONTROLOPTS
231 || last->inp_socket->so_options &
232 SO_TIMESTAMP) {
233 ip_savecontrol(last, &opts,
234 ip, n);
235 }
236 if (sbappendaddr(
237 &last->inp_socket->so_rcv,
238 sintosa(&udpsrc), n, opts) == 0) {
239 m_freem(n);
240 if (opts)
241 m_freem(opts);
242 } else
243 sorwakeup(last->inp_socket);
244 opts = 0;
245 }
246 }
247 last = inp;
248 /*
249 * Don't look for additional matches if this one does
250 * not have either the SO_REUSEPORT or SO_REUSEADDR
251 * socket options set. This heuristic avoids searching
252 * through all pcbs in the common case of a non-shared
253 * port. It * assumes that an application will never
254 * clear these options after setting them.
255 */
256 if ((last->inp_socket->so_options &
257 (SO_REUSEPORT|SO_REUSEADDR)) == 0)
258 break;
259 }
260
261 if (last == NULL) {
262 /*
263 * No matching pcb found; discard datagram.
264 * (No need to send an ICMP Port Unreachable
265 * for a broadcast or multicast datgram.)
266 */
267 udpstat.udps_noportbcast++;
268 goto bad;
269 }
270 if (last->inp_flags & INP_CONTROLOPTS ||
271 last->inp_socket->so_options & SO_TIMESTAMP)
272 ip_savecontrol(last, &opts, ip, m);
273 if (sbappendaddr(&last->inp_socket->so_rcv,
274 sintosa(&udpsrc), m, opts) == 0) {
275 udpstat.udps_fullsock++;
276 goto bad;
277 }
278 sorwakeup(last->inp_socket);
279 return;
280 }
281 /*
282 * Locate pcb for datagram.
283 */
284 inp = in_pcblookup_connect(&udbtable, ip->ip_src, uh->uh_sport,
285 ip->ip_dst, uh->uh_dport);
286 if (inp == 0) {
287 ++udpstat.udps_pcbhashmiss;
288 inp = in_pcblookup_bind(&udbtable, ip->ip_dst, uh->uh_dport);
289 if (inp == 0) {
290 udpstat.udps_noport++;
291 if (m->m_flags & (M_BCAST | M_MCAST)) {
292 udpstat.udps_noportbcast++;
293 goto bad;
294 }
295 *ip = save_ip;
296 #if NIPKDB > 0
297 if (checkipkdb(&ip->ip_src,
298 uh->uh_sport,
299 uh->uh_dport,
300 m,
301 iphlen + sizeof(struct udphdr),
302 len - sizeof(struct udphdr)))
303 /* It was a debugger connect packet, just drop it now */
304 goto bad;
305 #endif
306 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
307 return;
308 }
309 }
310
311 /*
312 * Construct sockaddr format source address.
313 * Stuff source address and datagram in user buffer.
314 */
315 udpsrc.sin_family = AF_INET;
316 udpsrc.sin_len = sizeof(struct sockaddr_in);
317 udpsrc.sin_addr = ip->ip_src;
318 udpsrc.sin_port = uh->uh_sport;
319 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero));
320
321 if (inp->inp_flags & INP_CONTROLOPTS ||
322 inp->inp_socket->so_options & SO_TIMESTAMP)
323 ip_savecontrol(inp, &opts, ip, m);
324 iphlen += sizeof(struct udphdr);
325 m->m_len -= iphlen;
326 m->m_pkthdr.len -= iphlen;
327 m->m_data += iphlen;
328 if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udpsrc), m,
329 opts) == 0) {
330 udpstat.udps_fullsock++;
331 goto bad;
332 }
333 sorwakeup(inp->inp_socket);
334 return;
335 bad:
336 m_freem(m);
337 if (opts)
338 m_freem(opts);
339 }
340
341 /*
342 * Notify a udp user of an asynchronous error;
343 * just wake up so that he can collect error status.
344 */
345 static void
346 udp_notify(inp, errno)
347 register struct inpcb *inp;
348 int errno;
349 {
350
351 inp->inp_socket->so_error = errno;
352 sorwakeup(inp->inp_socket);
353 sowwakeup(inp->inp_socket);
354 }
355
356 void *
357 udp_ctlinput(cmd, sa, v)
358 int cmd;
359 struct sockaddr *sa;
360 void *v;
361 {
362 register struct ip *ip = v;
363 register struct udphdr *uh;
364 extern int inetctlerrmap[];
365 void (*notify) __P((struct inpcb *, int)) = udp_notify;
366 int errno;
367
368 if ((unsigned)cmd >= PRC_NCMDS)
369 return NULL;
370 errno = inetctlerrmap[cmd];
371 if (PRC_IS_REDIRECT(cmd))
372 notify = in_rtchange, ip = 0;
373 else if (cmd == PRC_HOSTDEAD)
374 ip = 0;
375 else if (errno == 0)
376 return NULL;
377 if (ip) {
378 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
379 in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport,
380 ip->ip_src, uh->uh_sport, errno, notify);
381 } else
382 in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno,
383 notify);
384 return NULL;
385 }
386
387 int
388 #if __STDC__
389 udp_output(struct mbuf *m, ...)
390 #else
391 udp_output(m, va_alist)
392 struct mbuf *m;
393 va_dcl
394 #endif
395 {
396 register struct inpcb *inp;
397 register struct udpiphdr *ui;
398 register int len = m->m_pkthdr.len;
399 int error = 0;
400 va_list ap;
401
402 va_start(ap, m);
403 inp = va_arg(ap, struct inpcb *);
404 va_end(ap);
405
406 /*
407 * Calculate data length and get a mbuf
408 * for UDP and IP headers.
409 */
410 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
411 if (m == 0) {
412 error = ENOBUFS;
413 goto release;
414 }
415
416 /*
417 * Compute the packet length of the IP header, and
418 * punt if the length looks bogus.
419 */
420 if ((len + sizeof(struct udpiphdr)) > IP_MAXPACKET) {
421 error = EMSGSIZE;
422 goto release;
423 }
424
425 /*
426 * Fill in mbuf with extended UDP header
427 * and addresses and length put into network format.
428 */
429 ui = mtod(m, struct udpiphdr *);
430 bzero(ui->ui_x1, sizeof ui->ui_x1);
431 ui->ui_pr = IPPROTO_UDP;
432 ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr));
433 ui->ui_src = inp->inp_laddr;
434 ui->ui_dst = inp->inp_faddr;
435 ui->ui_sport = inp->inp_lport;
436 ui->ui_dport = inp->inp_fport;
437 ui->ui_ulen = ui->ui_len;
438
439 /*
440 * Stuff checksum and output datagram.
441 */
442 ui->ui_sum = 0;
443 if (udpcksum) {
444 if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0)
445 ui->ui_sum = 0xffff;
446 }
447 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
448 ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */
449 ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */
450 udpstat.udps_opackets++;
451 return (ip_output(m, inp->inp_options, &inp->inp_route,
452 inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST),
453 inp->inp_moptions));
454
455 release:
456 m_freem(m);
457 return (error);
458 }
459
460 int udp_sendspace = 9216; /* really max datagram size */
461 int udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in));
462 /* 40 1K datagrams */
463
464 /*ARGSUSED*/
465 int
466 udp_usrreq(so, req, m, nam, control, p)
467 struct socket *so;
468 int req;
469 struct mbuf *m, *nam, *control;
470 struct proc *p;
471 {
472 register struct inpcb *inp;
473 int s;
474 register int error = 0;
475
476 if (req == PRU_CONTROL)
477 return (in_control(so, (long)m, (caddr_t)nam,
478 (struct ifnet *)control, p));
479
480 s = splsoftnet();
481 inp = sotoinpcb(so);
482 #ifdef DIAGNOSTIC
483 if (req != PRU_SEND && req != PRU_SENDOOB && control)
484 panic("udp_usrreq: unexpected control mbuf");
485 #endif
486 if (inp == 0 && req != PRU_ATTACH) {
487 error = EINVAL;
488 goto release;
489 }
490
491 /*
492 * Note: need to block udp_input while changing
493 * the udp pcb queue and/or pcb addresses.
494 */
495 switch (req) {
496
497 case PRU_ATTACH:
498 if (inp != 0) {
499 error = EISCONN;
500 break;
501 }
502 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
503 error = soreserve(so, udp_sendspace, udp_recvspace);
504 if (error)
505 break;
506 }
507 error = in_pcballoc(so, &udbtable);
508 if (error)
509 break;
510 inp = sotoinpcb(so);
511 inp->inp_ip.ip_ttl = ip_defttl;
512 break;
513
514 case PRU_DETACH:
515 in_pcbdetach(inp);
516 break;
517
518 case PRU_BIND:
519 error = in_pcbbind(inp, nam, p);
520 break;
521
522 case PRU_LISTEN:
523 error = EOPNOTSUPP;
524 break;
525
526 case PRU_CONNECT:
527 error = in_pcbconnect(inp, nam);
528 if (error)
529 break;
530 soisconnected(so);
531 break;
532
533 case PRU_CONNECT2:
534 error = EOPNOTSUPP;
535 break;
536
537 case PRU_DISCONNECT:
538 /*soisdisconnected(so);*/
539 so->so_state &= ~SS_ISCONNECTED; /* XXX */
540 in_pcbdisconnect(inp);
541 inp->inp_laddr = zeroin_addr; /* XXX */
542 in_pcbstate(inp, INP_BOUND); /* XXX */
543 break;
544
545 case PRU_SHUTDOWN:
546 socantsendmore(so);
547 break;
548
549 case PRU_RCVD:
550 error = EOPNOTSUPP;
551 break;
552
553 case PRU_SEND:
554 if (control && control->m_len) {
555 m_freem(control);
556 m_freem(m);
557 error = EINVAL;
558 break;
559 }
560 {
561 struct in_addr laddr; /* XXX */
562
563 if (nam) {
564 laddr = inp->inp_laddr; /* XXX */
565 if ((so->so_state & SS_ISCONNECTED) != 0) {
566 error = EISCONN;
567 goto die;
568 }
569 error = in_pcbconnect(inp, nam);
570 if (error) {
571 die:
572 m_freem(m);
573 break;
574 }
575 } else {
576 if ((so->so_state & SS_ISCONNECTED) == 0) {
577 error = ENOTCONN;
578 goto die;
579 }
580 }
581 error = udp_output(m, inp);
582 if (nam) {
583 in_pcbdisconnect(inp);
584 inp->inp_laddr = laddr; /* XXX */
585 in_pcbstate(inp, INP_BOUND); /* XXX */
586 }
587 }
588 break;
589
590 case PRU_SENSE:
591 /*
592 * stat: don't bother with a blocksize.
593 */
594 splx(s);
595 return (0);
596
597 case PRU_RCVOOB:
598 error = EOPNOTSUPP;
599 break;
600
601 case PRU_SENDOOB:
602 m_freem(control);
603 m_freem(m);
604 error = EOPNOTSUPP;
605 break;
606
607 case PRU_SOCKADDR:
608 in_setsockaddr(inp, nam);
609 break;
610
611 case PRU_PEERADDR:
612 in_setpeeraddr(inp, nam);
613 break;
614
615 default:
616 panic("udp_usrreq");
617 }
618
619 release:
620 splx(s);
621 return (error);
622 }
623
624 /*
625 * Sysctl for udp variables.
626 */
627 int
628 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
629 int *name;
630 u_int namelen;
631 void *oldp;
632 size_t *oldlenp;
633 void *newp;
634 size_t newlen;
635 {
636 /* All sysctl names at this level are terminal. */
637 if (namelen != 1)
638 return (ENOTDIR);
639
640 switch (name[0]) {
641 case UDPCTL_CHECKSUM:
642 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum));
643 case UDPCTL_SENDSPACE:
644 return (sysctl_int(oldp, oldlenp, newp, newlen,
645 &udp_sendspace));
646 case UDPCTL_RECVSPACE:
647 return (sysctl_int(oldp, oldlenp, newp, newlen,
648 &udp_recvspace));
649 default:
650 return (ENOPROTOOPT);
651 }
652 /* NOTREACHED */
653 }
654