udp_usrreq.c revision 1.36 1 /* $NetBSD: udp_usrreq.c,v 1.36 1996/09/16 17:45:19 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, 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_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 (sbappendaddr(&last->so_rcv,
229 sintosa(&udpsrc), n,
230 (struct mbuf *)0) == 0) {
231 m_freem(n);
232 udpstat.udps_fullsock++;
233 } else
234 sorwakeup(last);
235 }
236 }
237 last = inp->inp_socket;
238 /*
239 * Don't look for additional matches if this one does
240 * not have either the SO_REUSEPORT or SO_REUSEADDR
241 * socket options set. This heuristic avoids searching
242 * through all pcbs in the common case of a non-shared
243 * port. It * assumes that an application will never
244 * clear these options after setting them.
245 */
246 if ((last->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0)
247 break;
248 }
249
250 if (last == NULL) {
251 /*
252 * No matching pcb found; discard datagram.
253 * (No need to send an ICMP Port Unreachable
254 * for a broadcast or multicast datgram.)
255 */
256 udpstat.udps_noportbcast++;
257 goto bad;
258 }
259 if (sbappendaddr(&last->so_rcv, sintosa(&udpsrc), m,
260 (struct mbuf *)0) == 0) {
261 udpstat.udps_fullsock++;
262 goto bad;
263 }
264 sorwakeup(last);
265 return;
266 }
267 /*
268 * Locate pcb for datagram.
269 */
270 inp = in_pcblookup_connect(&udbtable, ip->ip_src, uh->uh_sport,
271 ip->ip_dst, uh->uh_dport);
272 if (inp == 0) {
273 ++udpstat.udps_pcbhashmiss;
274 inp = in_pcblookup_bind(&udbtable, ip->ip_dst, uh->uh_dport);
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_addr = ip->ip_src;
295 udpsrc.sin_port = uh->uh_sport;
296 bzero((caddr_t)udpsrc.sin_zero, sizeof(udpsrc.sin_zero));
297
298 if (inp->inp_flags & INP_CONTROLOPTS) {
299 struct mbuf **mp = &opts;
300
301 if (inp->inp_flags & INP_RECVDSTADDR) {
302 *mp = udp_saveopt((caddr_t) &ip->ip_dst,
303 sizeof(struct in_addr), IP_RECVDSTADDR);
304 if (*mp)
305 mp = &(*mp)->m_next;
306 }
307 #ifdef notyet
308 /* options were tossed above */
309 if (inp->inp_flags & INP_RECVOPTS) {
310 *mp = udp_saveopt((caddr_t) opts_deleted_above,
311 sizeof(struct in_addr), IP_RECVOPTS);
312 if (*mp)
313 mp = &(*mp)->m_next;
314 }
315 /* ip_srcroute doesn't do what we want here, need to fix */
316 if (inp->inp_flags & INP_RECVRETOPTS) {
317 *mp = udp_saveopt((caddr_t) ip_srcroute(),
318 sizeof(struct in_addr), IP_RECVRETOPTS);
319 if (*mp)
320 mp = &(*mp)->m_next;
321 }
322 #endif
323 }
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 * Create a "control" mbuf containing the specified data
343 * with the specified type for presentation with a datagram.
344 */
345 struct mbuf *
346 udp_saveopt(p, size, type)
347 caddr_t p;
348 register int size;
349 int type;
350 {
351 register struct cmsghdr *cp;
352 struct mbuf *m;
353
354 if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL)
355 return ((struct mbuf *) NULL);
356 cp = (struct cmsghdr *) mtod(m, struct cmsghdr *);
357 bcopy(p, CMSG_DATA(cp), size);
358 size += sizeof(*cp);
359 m->m_len = size;
360 cp->cmsg_len = size;
361 cp->cmsg_level = IPPROTO_IP;
362 cp->cmsg_type = type;
363 return (m);
364 }
365
366 /*
367 * Notify a udp user of an asynchronous error;
368 * just wake up so that he can collect error status.
369 */
370 static void
371 udp_notify(inp, errno)
372 register struct inpcb *inp;
373 int errno;
374 {
375
376 inp->inp_socket->so_error = errno;
377 sorwakeup(inp->inp_socket);
378 sowwakeup(inp->inp_socket);
379 }
380
381 void *
382 udp_ctlinput(cmd, sa, v)
383 int cmd;
384 struct sockaddr *sa;
385 void *v;
386 {
387 register struct ip *ip = v;
388 register struct udphdr *uh;
389 extern int inetctlerrmap[];
390 void (*notify) __P((struct inpcb *, int)) = udp_notify;
391 int errno;
392
393 if ((unsigned)cmd >= PRC_NCMDS)
394 return NULL;
395 errno = inetctlerrmap[cmd];
396 if (PRC_IS_REDIRECT(cmd))
397 notify = in_rtchange, ip = 0;
398 else if (cmd == PRC_HOSTDEAD)
399 ip = 0;
400 else if (errno == 0)
401 return NULL;
402 if (ip) {
403 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
404 in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport,
405 ip->ip_src, uh->uh_sport, errno, notify);
406 } else
407 in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno,
408 notify);
409 return NULL;
410 }
411
412 int
413 #if __STDC__
414 udp_output(struct mbuf *m, ...)
415 #else
416 udp_output(m, va_alist)
417 struct mbuf *m;
418 va_dcl
419 #endif
420 {
421 register struct inpcb *inp;
422 register struct udpiphdr *ui;
423 register int len = m->m_pkthdr.len;
424 int error = 0;
425 va_list ap;
426
427 va_start(ap, m);
428 inp = va_arg(ap, struct inpcb *);
429 va_end(ap);
430
431 /*
432 * Calculate data length and get a mbuf
433 * for UDP and IP headers.
434 */
435 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
436 if (m == 0) {
437 error = ENOBUFS;
438 goto release;
439 }
440
441 /*
442 * Fill in mbuf with extended UDP header
443 * and addresses and length put into network format.
444 */
445 ui = mtod(m, struct udpiphdr *);
446 bzero(ui->ui_x1, sizeof ui->ui_x1);
447 ui->ui_pr = IPPROTO_UDP;
448 ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr));
449 ui->ui_src = inp->inp_laddr;
450 ui->ui_dst = inp->inp_faddr;
451 ui->ui_sport = inp->inp_lport;
452 ui->ui_dport = inp->inp_fport;
453 ui->ui_ulen = ui->ui_len;
454
455 /*
456 * Stuff checksum and output datagram.
457 */
458 ui->ui_sum = 0;
459 if (udpcksum) {
460 if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0)
461 ui->ui_sum = 0xffff;
462 }
463 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
464 ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */
465 ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */
466 udpstat.udps_opackets++;
467 return (ip_output(m, inp->inp_options, &inp->inp_route,
468 inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST),
469 inp->inp_moptions));
470
471 release:
472 m_freem(m);
473 return (error);
474 }
475
476 u_long udp_sendspace = 9216; /* really max datagram size */
477 u_long udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in));
478 /* 40 1K datagrams */
479
480 /*ARGSUSED*/
481 int
482 udp_usrreq(so, req, m, nam, control, p)
483 struct socket *so;
484 int req;
485 struct mbuf *m, *nam, *control;
486 struct proc *p;
487 {
488 register struct inpcb *inp;
489 int s;
490 register int error = 0;
491
492 if (req == PRU_CONTROL)
493 return (in_control(so, (long)m, (caddr_t)nam,
494 (struct ifnet *)control, p));
495
496 s = splsoftnet();
497 inp = sotoinpcb(so);
498 #ifdef DIAGNOSTIC
499 if (req != PRU_SEND && req != PRU_SENDOOB && control)
500 panic("udp_usrreq: unexpected control mbuf");
501 #endif
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 = zeroin_addr; /* XXX */
558 in_pcbstate(inp, INP_BOUND); /* XXX */
559 break;
560
561 case PRU_SHUTDOWN:
562 socantsendmore(so);
563 break;
564
565 case PRU_RCVD:
566 error = EOPNOTSUPP;
567 break;
568
569 case PRU_SEND:
570 if (control && control->m_len) {
571 m_freem(control);
572 m_freem(m);
573 error = EINVAL;
574 break;
575 }
576 {
577 struct in_addr laddr; /* XXX */
578
579 if (nam) {
580 laddr = inp->inp_laddr; /* XXX */
581 if ((so->so_state & SS_ISCONNECTED) != 0) {
582 error = EISCONN;
583 goto die;
584 }
585 error = in_pcbconnect(inp, nam);
586 if (error) {
587 die:
588 m_freem(m);
589 break;
590 }
591 } else {
592 if ((so->so_state & SS_ISCONNECTED) == 0) {
593 error = ENOTCONN;
594 goto die;
595 }
596 }
597 error = udp_output(m, inp);
598 if (nam) {
599 in_pcbdisconnect(inp);
600 inp->inp_laddr = laddr; /* XXX */
601 in_pcbstate(inp, INP_BOUND); /* XXX */
602 }
603 }
604 break;
605
606 case PRU_SENSE:
607 /*
608 * stat: don't bother with a blocksize.
609 */
610 splx(s);
611 return (0);
612
613 case PRU_RCVOOB:
614 error = EOPNOTSUPP;
615 break;
616
617 case PRU_SENDOOB:
618 m_freem(control);
619 m_freem(m);
620 error = EOPNOTSUPP;
621 break;
622
623 case PRU_SOCKADDR:
624 in_setsockaddr(inp, nam);
625 break;
626
627 case PRU_PEERADDR:
628 in_setpeeraddr(inp, nam);
629 break;
630
631 default:
632 panic("udp_usrreq");
633 }
634
635 release:
636 splx(s);
637 return (error);
638 }
639
640 /*
641 * Sysctl for udp variables.
642 */
643 int
644 udp_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
645 int *name;
646 u_int namelen;
647 void *oldp;
648 size_t *oldlenp;
649 void *newp;
650 size_t newlen;
651 {
652 /* All sysctl names at this level are terminal. */
653 if (namelen != 1)
654 return (ENOTDIR);
655
656 switch (name[0]) {
657 case UDPCTL_CHECKSUM:
658 return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum));
659 default:
660 return (ENOPROTOOPT);
661 }
662 /* NOTREACHED */
663 }
664