]> git.saurik.com Git - apple/xnu.git/blame - bsd/netinet/udp_usrreq.c
xnu-1228.5.18.tar.gz
[apple/xnu.git] / bsd / netinet / udp_usrreq.c
CommitLineData
1c79356b 1/*
2d21ac55 2 * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
5d5c5d0d 3 *
2d21ac55 4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
1c79356b 5 *
2d21ac55
A
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
8f6c56a5 14 *
2d21ac55
A
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
8f6c56a5
A
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
2d21ac55
A
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
8f6c56a5 25 *
2d21ac55 26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
1c79356b
A
27 */
28/*
29 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
30 * The Regents of the University of California. All rights reserved.
31 *
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
34 * are met:
35 * 1. Redistributions of source code must retain the above copyright
36 * notice, this list of conditions and the following disclaimer.
37 * 2. Redistributions in binary form must reproduce the above copyright
38 * notice, this list of conditions and the following disclaimer in the
39 * documentation and/or other materials provided with the distribution.
40 * 3. All advertising materials mentioning features or use of this software
41 * must display the following acknowledgement:
42 * This product includes software developed by the University of
43 * California, Berkeley and its contributors.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
9bccf70c 61 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.13 2001/08/08 18:59:54 ghelmer Exp $
1c79356b
A
62 */
63
64#include <sys/param.h>
65#include <sys/systm.h>
66#include <sys/kernel.h>
67#include <sys/malloc.h>
68#include <sys/mbuf.h>
1c79356b 69#include <sys/domain.h>
1c79356b
A
70#include <sys/protosw.h>
71#include <sys/socket.h>
72#include <sys/socketvar.h>
73#include <sys/sysctl.h>
74#include <sys/syslog.h>
75
2d21ac55
A
76#include <kern/zalloc.h>
77
1c79356b 78#include <net/if.h>
91447636 79#include <net/if_types.h>
1c79356b
A
80#include <net/route.h>
81
82#include <netinet/in.h>
83#include <netinet/in_systm.h>
84#include <netinet/ip.h>
9bccf70c
A
85#if INET6
86#include <netinet/ip6.h>
87#endif
1c79356b
A
88#include <netinet/in_pcb.h>
89#include <netinet/in_var.h>
90#include <netinet/ip_var.h>
1c79356b 91#if INET6
1c79356b
A
92#include <netinet6/ip6_var.h>
93#endif
9bccf70c
A
94#include <netinet/ip_icmp.h>
95#include <netinet/icmp_var.h>
1c79356b
A
96#include <netinet/udp.h>
97#include <netinet/udp_var.h>
98#include <sys/kdebug.h>
99
100#if IPSEC
101#include <netinet6/ipsec.h>
2d21ac55 102#include <netinet6/esp.h>
9bccf70c 103extern int ipsec_bypass;
1c79356b
A
104#endif /*IPSEC*/
105
106
107#define DBG_LAYER_IN_BEG NETDBG_CODE(DBG_NETUDP, 0)
108#define DBG_LAYER_IN_END NETDBG_CODE(DBG_NETUDP, 2)
109#define DBG_LAYER_OUT_BEG NETDBG_CODE(DBG_NETUDP, 1)
110#define DBG_LAYER_OUT_END NETDBG_CODE(DBG_NETUDP, 3)
111#define DBG_FNC_UDP_INPUT NETDBG_CODE(DBG_NETUDP, (5 << 8))
112#define DBG_FNC_UDP_OUTPUT NETDBG_CODE(DBG_NETUDP, (6 << 8) | 1)
113
1c79356b
A
114/*
115 * UDP protocol implementation.
116 * Per RFC 768, August, 1980.
117 */
118#ifndef COMPAT_42
119static int udpcksum = 1;
120#else
121static int udpcksum = 0; /* XXX */
122#endif
123SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
124 &udpcksum, 0, "");
125
2d21ac55
A
126static u_int32_t udps_in_sw_cksum;
127SYSCTL_UINT(_net_inet_udp, OID_AUTO, in_sw_cksum, CTLFLAG_RD,
128 &udps_in_sw_cksum, 0,
129 "Number of received packets checksummed in software");
130
131static u_int64_t udps_in_sw_cksum_bytes;
132SYSCTL_QUAD(_net_inet_udp, OID_AUTO, in_sw_cksum_bytes, CTLFLAG_RD,
133 &udps_in_sw_cksum_bytes,
134 "Amount of received data checksummed in software");
135
136static u_int32_t udps_out_sw_cksum;
137SYSCTL_UINT(_net_inet_udp, OID_AUTO, out_sw_cksum, CTLFLAG_RD,
138 &udps_out_sw_cksum, 0,
139 "Number of transmitted packets checksummed in software");
140
141static u_int64_t udps_out_sw_cksum_bytes;
142SYSCTL_QUAD(_net_inet_udp, OID_AUTO, out_sw_cksum_bytes, CTLFLAG_RD,
143 &udps_out_sw_cksum_bytes,
144 "Amount of transmitted data checksummed in software");
145
9bccf70c 146int log_in_vain = 0;
1c79356b 147SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
9bccf70c
A
148 &log_in_vain, 0, "Log all incoming UDP packets");
149
150static int blackhole = 0;
151SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW,
152 &blackhole, 0, "Do not send port unreachables for refused connects");
1c79356b
A
153
154struct inpcbhead udb; /* from udp_var.h */
155#define udb6 udb /* for KAME src sync over BSD*'s */
156struct inpcbinfo udbinfo;
157
158#ifndef UDBHASHSIZE
159#define UDBHASHSIZE 16
160#endif
161
55e303ae
A
162extern int esp_udp_encap_port;
163extern u_long route_generation;
0b4e3aa0 164
2d21ac55 165extern void ipfwsyslog( int level, const char *format,...);
91447636
A
166
167extern int fw_verbose;
2d21ac55 168static int udp_gc_done = FALSE; /* Garbage collection performed last slowtimo */
91447636 169
2d21ac55 170#if IPFIREWALL
91447636
A
171#define log_in_vain_log( a ) { \
172 if ( (log_in_vain == 3 ) && (fw_verbose == 2)) { /* Apple logging, log to ipfw.log */ \
173 ipfwsyslog a ; \
174 } \
175 else log a ; \
176}
2d21ac55
A
177#else
178#define log_in_vain_log( a ) { log a; }
179#endif
91447636 180
1c79356b
A
181struct udpstat udpstat; /* from udp_var.h */
182SYSCTL_STRUCT(_net_inet_udp, UDPCTL_STATS, stats, CTLFLAG_RD,
9bccf70c 183 &udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)");
91447636
A
184SYSCTL_INT(_net_inet_udp, OID_AUTO, pcbcount, CTLFLAG_RD,
185 &udbinfo.ipi_count, 0, "Number of active PCBs");
1c79356b 186
1c79356b
A
187#if INET6
188struct udp_in6 {
189 struct sockaddr_in6 uin6_sin;
190 u_char uin6_init_done : 1;
1c79356b
A
191};
192struct udp_ip6 {
193 struct ip6_hdr uip6_ip6;
194 u_char uip6_init_done : 1;
2d21ac55 195};
91447636 196static void ip_2_ip6_hdr(struct ip6_hdr *ip6, struct ip *ip);
2d21ac55
A
197static void udp_append(struct inpcb *last, struct ip *ip,
198 struct mbuf *n, int off, struct sockaddr_in *pudp_in,
199 struct udp_in6 *pudp_in6, struct udp_ip6 *pudp_ip6);
200#else
201static void udp_append(struct inpcb *last, struct ip *ip,
202 struct mbuf *n, int off, struct sockaddr_in *pudp_in);
1c79356b
A
203#endif
204
91447636
A
205static int udp_detach(struct socket *so);
206static int udp_output(struct inpcb *, struct mbuf *, struct sockaddr *,
207 struct mbuf *, struct proc *);
208extern int ChkAddressOK( __uint32_t dstaddr, __uint32_t srcaddr );
1c79356b
A
209
210void
211udp_init()
212{
91447636
A
213 vm_size_t str_size;
214 struct inpcbinfo *pcbinfo;
215
1c79356b
A
216
217 LIST_INIT(&udb);
218 udbinfo.listhead = &udb;
219 udbinfo.hashbase = hashinit(UDBHASHSIZE, M_PCB, &udbinfo.hashmask);
220 udbinfo.porthashbase = hashinit(UDBHASHSIZE, M_PCB,
221 &udbinfo.porthashmask);
9bccf70c 222#ifdef __APPLE__
1c79356b 223 str_size = (vm_size_t) sizeof(struct inpcb);
9bccf70c 224 udbinfo.ipi_zone = (void *) zinit(str_size, 80000*str_size, 8192, "udpcb");
1c79356b 225
91447636
A
226 pcbinfo = &udbinfo;
227 /*
228 * allocate lock group attribute and group for udp pcb mutexes
229 */
230 pcbinfo->mtx_grp_attr = lck_grp_attr_alloc_init();
91447636
A
231
232 pcbinfo->mtx_grp = lck_grp_alloc_init("udppcb", pcbinfo->mtx_grp_attr);
233
234 pcbinfo->mtx_attr = lck_attr_alloc_init();
91447636
A
235
236 if ((pcbinfo->mtx = lck_rw_alloc_init(pcbinfo->mtx_grp, pcbinfo->mtx_attr)) == NULL)
237 return; /* pretty much dead if this fails... */
238
1c79356b 239 in_pcb_nat_init(&udbinfo, AF_INET, IPPROTO_UDP, SOCK_DGRAM);
9bccf70c
A
240#else
241 udbinfo.ipi_zone = zinit("udpcb", sizeof(struct inpcb), maxsockets,
242 ZONE_INTERRUPT, 0);
243#endif
1c79356b
A
244
245#if 0
9bccf70c 246 /* for pcb sharing testing only */
1c79356b
A
247 stat = in_pcb_new_share_client(&udbinfo, &fake_owner);
248 kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat);
249
250 laddr.s_addr = 0x11646464;
251 faddr.s_addr = 0x11646465;
252
253 lport = 1500;
254 in_pcb_grab_port(&udbinfo, 0, laddr, &lport, faddr, 1600, 0, fake_owner);
255 kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat);
256
257 stat = in_pcb_rem_share_client(&udbinfo, fake_owner);
258 kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat);
259
260 stat = in_pcb_new_share_client(&udbinfo, &fake_owner);
261 kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat);
262
263 laddr.s_addr = 0x11646464;
264 faddr.s_addr = 0x11646465;
265
266 lport = 1500;
267 stat = in_pcb_grab_port(&udbinfo, 0, laddr, &lport, faddr, 1600, 0, fake_owner);
268 kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat);
269#endif
270}
271
272void
273udp_input(m, iphlen)
274 register struct mbuf *m;
275 int iphlen;
276{
277 register struct ip *ip;
278 register struct udphdr *uh;
279 register struct inpcb *inp;
280 struct mbuf *opts = 0;
1c79356b
A
281 int len;
282 struct ip save_ip;
283 struct sockaddr *append_sa;
91447636 284 struct inpcbinfo *pcbinfo = &udbinfo;
2d21ac55
A
285 struct sockaddr_in udp_in = {
286 sizeof (udp_in), AF_INET, 0, { 0 }, { 0, 0, 0, 0, 0, 0, 0, 0 }
287 };
288#if INET6
289 struct udp_in6 udp_in6 = {
290 { sizeof (udp_in6.uin6_sin), AF_INET6, 0, 0,
291 IN6ADDR_ANY_INIT, 0 },
292 0
293 };
294 struct udp_ip6 udp_ip6;
295#endif /* INET6 */
1c79356b
A
296
297 udpstat.udps_ipackets++;
1c79356b
A
298
299 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_START, 0,0,0,0,0);
9bccf70c
A
300 if (m->m_pkthdr.csum_flags & CSUM_TCP_SUM16)
301 m->m_pkthdr.csum_flags = 0; /* invalidate hwcksum for UDP */
1c79356b
A
302
303 /*
304 * Strip IP options, if any; should skip this,
305 * make available to user, and use on returned packets,
306 * but we don't yet have a way to check the checksum
307 * with options still present.
308 */
309 if (iphlen > sizeof (struct ip)) {
310 ip_stripoptions(m, (struct mbuf *)0);
311 iphlen = sizeof(struct ip);
312 }
313
314 /*
315 * Get IP and UDP header together in first mbuf.
316 */
317 ip = mtod(m, struct ip *);
318 if (m->m_len < iphlen + sizeof(struct udphdr)) {
319 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
320 udpstat.udps_hdrops++;
321 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
322 return;
323 }
324 ip = mtod(m, struct ip *);
325 }
326 uh = (struct udphdr *)((caddr_t)ip + iphlen);
327
9bccf70c
A
328 /* destination port of 0 is illegal, based on RFC768. */
329 if (uh->uh_dport == 0)
330 goto bad;
331
1c79356b
A
332 KERNEL_DEBUG(DBG_LAYER_IN_BEG, uh->uh_dport, uh->uh_sport,
333 ip->ip_src.s_addr, ip->ip_dst.s_addr, uh->uh_ulen);
334
335 /*
336 * Make mbuf data length reflect UDP length.
337 * If not enough data to reflect UDP length, drop.
338 */
339 len = ntohs((u_short)uh->uh_ulen);
340 if (ip->ip_len != len) {
341 if (len > ip->ip_len || len < sizeof(struct udphdr)) {
342 udpstat.udps_badlen++;
343 goto bad;
344 }
345 m_adj(m, len - ip->ip_len);
346 /* ip->ip_len = len; */
347 }
348 /*
349 * Save a copy of the IP header in case we want restore it
350 * for sending an ICMP error message in response.
351 */
352 save_ip = *ip;
353
354 /*
355 * Checksum extended UDP header and data.
356 */
357 if (uh->uh_sum) {
ac5ea4a9
A
358 if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
359 if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
360 uh->uh_sum = m->m_pkthdr.csum_data;
361 else
362 goto doudpcksum;
363 uh->uh_sum ^= 0xffff;
364 } else {
365 char b[9];
0b4e3aa0 366doudpcksum:
ac5ea4a9
A
367 *(uint32_t*)&b[0] = *(uint32_t*)&((struct ipovly *)ip)->ih_x1[0];
368 *(uint32_t*)&b[4] = *(uint32_t*)&((struct ipovly *)ip)->ih_x1[4];
369 *(uint8_t*)&b[8] = *(uint8_t*)&((struct ipovly *)ip)->ih_x1[8];
370
0b4e3aa0
A
371 bzero(((struct ipovly *)ip)->ih_x1, 9);
372 ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
373 uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
ac5ea4a9
A
374
375 *(uint32_t*)&((struct ipovly *)ip)->ih_x1[0] = *(uint32_t*)&b[0];
376 *(uint32_t*)&((struct ipovly *)ip)->ih_x1[4] = *(uint32_t*)&b[4];
377 *(uint8_t*)&((struct ipovly *)ip)->ih_x1[8] = *(uint8_t*)&b[8];
2d21ac55 378 udp_in_cksum_stats(len);
0b4e3aa0 379 }
1c79356b
A
380 if (uh->uh_sum) {
381 udpstat.udps_badsum++;
382 m_freem(m);
383 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
384 return;
385 }
386 }
9bccf70c
A
387#ifndef __APPLE__
388 else
389 udpstat.udps_nosum++;
390#endif
1c79356b
A
391
392 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
393 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
2d21ac55
A
394
395 int reuse_sock = 0, mcast_delivered = 0;
396 struct mbuf *n = NULL;
397
91447636 398 lck_rw_lock_shared(pcbinfo->mtx);
1c79356b
A
399 /*
400 * Deliver a multicast or broadcast datagram to *all* sockets
401 * for which the local and remote addresses and ports match
402 * those of the incoming datagram. This allows more than
403 * one process to receive multi/broadcasts on the same port.
404 * (This really ought to be done for unicast datagrams as
405 * well, but that would cause problems with existing
406 * applications that open both address-specific sockets and
407 * a wildcard socket listening to the same port -- they would
408 * end up receiving duplicates of every unicast datagram.
409 * Those applications open the multiple sockets to overcome an
410 * inadequacy of the UDP socket interface, but for backwards
411 * compatibility we avoid the problem here rather than
412 * fixing the interface. Maybe 4.5BSD will remedy this?)
413 */
414
91447636 415
1c79356b
A
416 /*
417 * Construct sockaddr format source address.
418 */
419 udp_in.sin_port = uh->uh_sport;
420 udp_in.sin_addr = ip->ip_src;
421 /*
422 * Locate pcb(s) for datagram.
423 * (Algorithm copied from raw_intr().)
424 */
1c79356b
A
425#if INET6
426 udp_in6.uin6_init_done = udp_ip6.uip6_init_done = 0;
427#endif
428 LIST_FOREACH(inp, &udb, inp_list) {
9bccf70c
A
429#ifdef __APPLE__
430 /* Ignore nat/SharedIP dummy pcbs */
431 if (inp->inp_socket == &udbinfo.nat_dummy_socket)
432 continue;
433#endif
91447636 434 if (inp->inp_socket == NULL)
1c79356b 435 continue;
91447636 436 if (inp != sotoinpcb(inp->inp_socket))
2d21ac55 437 panic("udp_input: bad so back ptr inp=%p\n", inp);
91447636
A
438#if INET6
439 if ((inp->inp_vflag & INP_IPV4) == 0)
440 continue;
1c79356b 441#endif
2d21ac55 442
91447636
A
443 if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) == WNT_STOPUSING) {
444 continue;
445 }
446
447 udp_lock(inp->inp_socket, 1, 0);
448
449 if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
450 udp_unlock(inp->inp_socket, 1, 0);
1c79356b 451 continue;
91447636
A
452 }
453
454 if (inp->inp_lport != uh->uh_dport) {
455 udp_unlock(inp->inp_socket, 1, 0);
456 continue;
457 }
1c79356b
A
458 if (inp->inp_laddr.s_addr != INADDR_ANY) {
459 if (inp->inp_laddr.s_addr !=
91447636
A
460 ip->ip_dst.s_addr) {
461 udp_unlock(inp->inp_socket, 1, 0);
1c79356b 462 continue;
91447636 463 }
1c79356b
A
464 }
465 if (inp->inp_faddr.s_addr != INADDR_ANY) {
466 if (inp->inp_faddr.s_addr !=
467 ip->ip_src.s_addr ||
91447636
A
468 inp->inp_fport != uh->uh_sport) {
469 udp_unlock(inp->inp_socket, 1, 0);
1c79356b 470 continue;
91447636 471 }
1c79356b
A
472 }
473
2d21ac55
A
474 reuse_sock = inp->inp_socket->so_options& (SO_REUSEPORT|SO_REUSEADDR);
475 {
1c79356b 476#if IPSEC
91447636 477 int skipit = 0;
1c79356b 478 /* check AH/ESP integrity. */
91447636 479 if (ipsec_bypass == 0) {
2d21ac55
A
480 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
481 IPSEC_STAT_INCREMENT(ipsecstat.in_polvio);
91447636
A
482 /* do not inject data to pcb */
483 skipit = 1;
484 }
91447636 485 }
2d21ac55 486 if (skipit == 0)
1c79356b 487#endif /*IPSEC*/
2d21ac55
A
488 {
489 if (reuse_sock)
490 n = m_copy(m, 0, M_COPYALL);
491#if INET6
492 udp_append(inp, ip, m,
493 iphlen + sizeof(struct udphdr),
494 &udp_in, &udp_in6, &udp_ip6);
495#else
496 udp_append(inp, ip, m,
497 iphlen + sizeof(struct udphdr),
498 &udp_in);
499#endif /* INET6 */
500 mcast_delivered++;
1c79356b 501 }
2d21ac55 502 udp_unlock(inp->inp_socket, 1, 0);
1c79356b 503 }
1c79356b
A
504 /*
505 * Don't look for additional matches if this one does
506 * not have either the SO_REUSEPORT or SO_REUSEADDR
507 * socket options set. This heuristic avoids searching
508 * through all pcbs in the common case of a non-shared
2d21ac55 509 * port. It assumes that an application will never
1c79356b
A
510 * clear these options after setting them.
511 */
2d21ac55 512 if (reuse_sock == 0 || ((m = n) == NULL))
1c79356b
A
513 break;
514 }
91447636 515 lck_rw_done(pcbinfo->mtx);
1c79356b 516
2d21ac55 517 if (mcast_delivered == 0) {
1c79356b
A
518 /*
519 * No matching pcb found; discard datagram.
520 * (No need to send an ICMP Port Unreachable
521 * for a broadcast or multicast datgram.)
522 */
523 udpstat.udps_noportbcast++;
524 goto bad;
525 }
2d21ac55
A
526
527 if (reuse_sock != 0) /* free the extra copy of mbuf */
528 m_freem(m);
1c79356b
A
529 return;
530 }
55e303ae 531
91447636 532#if IPSEC
55e303ae
A
533 /*
534 * UDP to port 4500 with a payload where the first four bytes are
535 * not zero is a UDP encapsulated IPSec packet. Packets where
536 * the payload is one byte and that byte is 0xFF are NAT keepalive
537 * packets. Decapsulate the ESP packet and carry on with IPSec input
538 * or discard the NAT keep-alive.
539 */
540 if (ipsec_bypass == 0 && (esp_udp_encap_port & 0xFFFF) != 0 &&
541 uh->uh_dport == ntohs((u_short)esp_udp_encap_port)) {
542 int payload_len = len - sizeof(struct udphdr) > 4 ? 4 : len - sizeof(struct udphdr);
543 if (m->m_len < iphlen + sizeof(struct udphdr) + payload_len) {
544 if ((m = m_pullup(m, iphlen + sizeof(struct udphdr) + payload_len)) == 0) {
545 udpstat.udps_hdrops++;
546 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
547 return;
548 }
549 ip = mtod(m, struct ip *);
550 uh = (struct udphdr *)((caddr_t)ip + iphlen);
551 }
552 /* Check for NAT keepalive packet */
553 if (payload_len == 1 && *(u_int8_t*)((caddr_t)uh + sizeof(struct udphdr)) == 0xFF) {
554 m_freem(m);
555 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
556 return;
557 }
558 else if (payload_len == 4 && *(u_int32_t*)((caddr_t)uh + sizeof(struct udphdr)) != 0) {
559 /* UDP encapsulated IPSec packet to pass through NAT */
560 size_t stripsiz;
561
562 stripsiz = sizeof(struct udphdr);
563
564 ip = mtod(m, struct ip *);
565 ovbcopy((caddr_t)ip, (caddr_t)(((u_char *)ip) + stripsiz), iphlen);
566 m->m_data += stripsiz;
567 m->m_len -= stripsiz;
568 m->m_pkthdr.len -= stripsiz;
569 ip = mtod(m, struct ip *);
570 ip->ip_len = ip->ip_len - stripsiz;
571 ip->ip_p = IPPROTO_ESP;
572
573 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
574 esp4_input(m, iphlen);
575 return;
576 }
577 }
91447636 578#endif
55e303ae 579
1c79356b
A
580 /*
581 * Locate pcb for datagram.
582 */
583 inp = in_pcblookup_hash(&udbinfo, ip->ip_src, uh->uh_sport,
584 ip->ip_dst, uh->uh_dport, 1, m->m_pkthdr.rcvif);
585 if (inp == NULL) {
586 if (log_in_vain) {
91447636
A
587 char buf[MAX_IPv4_STR_LEN];
588 char buf2[MAX_IPv4_STR_LEN];
589
590 /* check src and dst address */
591 if (log_in_vain != 3)
592 log(LOG_INFO,
593 "Connection attempt to UDP %s:%d from %s:%d\n",
594 inet_ntop(AF_INET, &ip->ip_dst, buf, sizeof(buf)),
595 ntohs(uh->uh_dport),
596 inet_ntop(AF_INET, &ip->ip_src, buf2, sizeof(buf2)),
597 ntohs(uh->uh_sport));
598 else if (!(m->m_flags & (M_BCAST | M_MCAST)) &&
599 ip->ip_dst.s_addr != ip->ip_src.s_addr)
600 log_in_vain_log((LOG_INFO,
601 "Stealth Mode connection attempt to UDP %s:%d from %s:%d\n",
602 inet_ntop(AF_INET, &ip->ip_dst, buf, sizeof(buf)),
603 ntohs(uh->uh_dport),
604 inet_ntop(AF_INET, &ip->ip_src, buf2, sizeof(buf2)),
605 ntohs(uh->uh_sport)))
1c79356b
A
606 }
607 udpstat.udps_noport++;
608 if (m->m_flags & (M_BCAST | M_MCAST)) {
609 udpstat.udps_noportbcast++;
610 goto bad;
611 }
1c79356b 612#if ICMP_BANDLIM
9bccf70c 613 if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
1c79356b
A
614 goto bad;
615#endif
9bccf70c 616 if (blackhole)
91447636
A
617 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type != IFT_LOOP)
618 goto bad;
9bccf70c
A
619 *ip = save_ip;
620 ip->ip_len += iphlen;
1c79356b
A
621 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
622 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
623 return;
624 }
91447636
A
625 udp_lock(inp->inp_socket, 1, 0);
626
627 if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
628 udp_unlock(inp->inp_socket, 1, 0);
1c79356b
A
629 goto bad;
630 }
91447636
A
631#if IPSEC
632 if (ipsec_bypass == 0 && inp != NULL) {
91447636 633 if (ipsec4_in_reject_so(m, inp->inp_socket)) {
2d21ac55 634 IPSEC_STAT_INCREMENT(ipsecstat.in_polvio);
91447636
A
635 udp_unlock(inp->inp_socket, 1, 0);
636 goto bad;
637 }
91447636 638 }
1c79356b
A
639#endif /*IPSEC*/
640
641 /*
642 * Construct sockaddr format source address.
643 * Stuff source address and datagram in user buffer.
644 */
645 udp_in.sin_port = uh->uh_sport;
646 udp_in.sin_addr = ip->ip_src;
647 if (inp->inp_flags & INP_CONTROLOPTS
648 || inp->inp_socket->so_options & SO_TIMESTAMP) {
649#if INET6
650 if (inp->inp_vflag & INP_IPV6) {
651 int savedflags;
652
653 ip_2_ip6_hdr(&udp_ip6.uip6_ip6, ip);
654 savedflags = inp->inp_flags;
655 inp->inp_flags &= ~INP_UNMAPPABLEOPTS;
9bccf70c 656 ip6_savecontrol(inp, &opts, &udp_ip6.uip6_ip6, m);
1c79356b
A
657 inp->inp_flags = savedflags;
658 } else
659#endif
660 ip_savecontrol(inp, &opts, ip, m);
661 }
9bccf70c 662 m_adj(m, iphlen + sizeof(struct udphdr));
1c79356b
A
663
664 KERNEL_DEBUG(DBG_LAYER_IN_END, uh->uh_dport, uh->uh_sport,
665 save_ip.ip_src.s_addr, save_ip.ip_dst.s_addr, uh->uh_ulen);
666
667#if INET6
668 if (inp->inp_vflag & INP_IPV6) {
669 in6_sin_2_v4mapsin6(&udp_in, &udp_in6.uin6_sin);
2d21ac55 670 append_sa = (struct sockaddr *)&udp_in6.uin6_sin;
1c79356b
A
671 } else
672#endif
673 append_sa = (struct sockaddr *)&udp_in;
91447636 674 if (sbappendaddr(&inp->inp_socket->so_rcv, append_sa, m, opts, NULL) == 0) {
1c79356b 675 udpstat.udps_fullsock++;
1c79356b 676 }
91447636
A
677 else {
678 sorwakeup(inp->inp_socket);
679 }
680 udp_unlock(inp->inp_socket, 1, 0);
1c79356b
A
681 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
682 return;
683bad:
684 m_freem(m);
685 if (opts)
686 m_freem(opts);
687 KERNEL_DEBUG(DBG_FNC_UDP_INPUT | DBG_FUNC_END, 0,0,0,0,0);
9bccf70c 688 return;
1c79356b
A
689}
690
691#if INET6
692static void
693ip_2_ip6_hdr(ip6, ip)
694 struct ip6_hdr *ip6;
695 struct ip *ip;
696{
697 bzero(ip6, sizeof(*ip6));
698
699 ip6->ip6_vfc = IPV6_VERSION;
700 ip6->ip6_plen = ip->ip_len;
701 ip6->ip6_nxt = ip->ip_p;
702 ip6->ip6_hlim = ip->ip_ttl;
703 ip6->ip6_src.s6_addr32[2] = ip6->ip6_dst.s6_addr32[2] =
704 IPV6_ADDR_INT32_SMP;
705 ip6->ip6_src.s6_addr32[3] = ip->ip_src.s_addr;
706 ip6->ip6_dst.s6_addr32[3] = ip->ip_dst.s_addr;
707}
708#endif
709
710/*
711 * subroutine of udp_input(), mainly for source code readability.
1c79356b
A
712 */
713static void
2d21ac55
A
714#if INET6
715udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n, int off,
716 struct sockaddr_in *pudp_in, struct udp_in6 *pudp_in6,
717 struct udp_ip6 *pudp_ip6)
718#else
719udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n, int off,
720 struct sockaddr_in *pudp_in)
721#endif
1c79356b
A
722{
723 struct sockaddr *append_sa;
724 struct mbuf *opts = 0;
1c79356b 725
2d21ac55
A
726#if CONFIG_MACF_NET
727 if (mac_inpcb_check_deliver(last, n, AF_INET, SOCK_DGRAM) != 0) {
728 m_freem(n);
729 return;
730 }
731#endif
1c79356b
A
732 if (last->inp_flags & INP_CONTROLOPTS ||
733 last->inp_socket->so_options & SO_TIMESTAMP) {
734#if INET6
735 if (last->inp_vflag & INP_IPV6) {
736 int savedflags;
737
2d21ac55
A
738 if (pudp_ip6->uip6_init_done == 0) {
739 ip_2_ip6_hdr(&pudp_ip6->uip6_ip6, ip);
740 pudp_ip6->uip6_init_done = 1;
1c79356b
A
741 }
742 savedflags = last->inp_flags;
743 last->inp_flags &= ~INP_UNMAPPABLEOPTS;
2d21ac55 744 ip6_savecontrol(last, &opts, &pudp_ip6->uip6_ip6, n);
1c79356b
A
745 last->inp_flags = savedflags;
746 } else
747#endif
748 ip_savecontrol(last, &opts, ip, n);
749 }
750#if INET6
751 if (last->inp_vflag & INP_IPV6) {
2d21ac55
A
752 if (pudp_in6->uin6_init_done == 0) {
753 in6_sin_2_v4mapsin6(pudp_in, &pudp_in6->uin6_sin);
754 pudp_in6->uin6_init_done = 1;
1c79356b 755 }
2d21ac55 756 append_sa = (struct sockaddr *)&pudp_in6->uin6_sin;
1c79356b
A
757 } else
758#endif
2d21ac55 759 append_sa = (struct sockaddr *)pudp_in;
1c79356b 760 m_adj(n, off);
91447636 761 if (sbappendaddr(&last->inp_socket->so_rcv, append_sa, n, opts, NULL) == 0) {
1c79356b
A
762 udpstat.udps_fullsock++;
763 } else
764 sorwakeup(last->inp_socket);
765}
766
1c79356b
A
767/*
768 * Notify a udp user of an asynchronous error;
769 * just wake up so that he can collect error status.
770 */
771void
772udp_notify(inp, errno)
773 register struct inpcb *inp;
774 int errno;
775{
776 inp->inp_socket->so_error = errno;
777 sorwakeup(inp->inp_socket);
778 sowwakeup(inp->inp_socket);
779}
780
781void
782udp_ctlinput(cmd, sa, vip)
783 int cmd;
784 struct sockaddr *sa;
785 void *vip;
786{
9bccf70c
A
787 struct ip *ip = vip;
788 struct udphdr *uh;
91447636 789 void (*notify)(struct inpcb *, int) = udp_notify;
9bccf70c
A
790 struct in_addr faddr;
791 struct inpcb *inp;
9bccf70c
A
792
793 faddr = ((struct sockaddr_in *)sa)->sin_addr;
794 if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
795 return;
1c79356b 796
9bccf70c
A
797 if (PRC_IS_REDIRECT(cmd)) {
798 ip = 0;
799 notify = in_rtchange;
800 } else if (cmd == PRC_HOSTDEAD)
801 ip = 0;
802 else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0)
1c79356b
A
803 return;
804 if (ip) {
805 uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
9bccf70c
A
806 inp = in_pcblookup_hash(&udbinfo, faddr, uh->uh_dport,
807 ip->ip_src, uh->uh_sport, 0, NULL);
91447636
A
808 if (inp != NULL && inp->inp_socket != NULL) {
809 udp_lock(inp->inp_socket, 1, 0);
810 if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
811 udp_unlock(inp->inp_socket, 1, 0);
812 return;
813 }
9bccf70c 814 (*notify)(inp, inetctlerrmap[cmd]);
91447636
A
815 udp_unlock(inp->inp_socket, 1, 0);
816 }
1c79356b 817 } else
91447636 818 in_pcbnotifyall(&udbinfo, faddr, inetctlerrmap[cmd], notify);
1c79356b
A
819}
820
2d21ac55
A
821int
822udp_ctloutput(struct socket *so, struct sockopt *sopt)
823{
824 int error, optval;
825 struct inpcb *inp;
826
827 if (sopt->sopt_level != IPPROTO_UDP)
828 return (ip_ctloutput(so, sopt));
829
830 error = 0;
831 inp = sotoinpcb(so);
832
833 switch (sopt->sopt_dir) {
834 case SOPT_SET:
835 switch (sopt->sopt_name) {
836 case UDP_NOCKSUM:
837 /* This option is settable only for UDP over IPv4 */
838 if (!(inp->inp_vflag & INP_IPV4)) {
839 error = EINVAL;
840 break;
841 }
842
843 if ((error = sooptcopyin(sopt, &optval, sizeof (optval),
844 sizeof (optval))) != 0)
845 break;
846
847 if (optval != 0)
848 inp->inp_flags |= INP_UDP_NOCKSUM;
849 else
850 inp->inp_flags &= ~INP_UDP_NOCKSUM;
851 break;
852
853 default:
854 error = ENOPROTOOPT;
855 break;
856 }
857 break;
858
859 case SOPT_GET:
860 switch (sopt->sopt_name) {
861 case UDP_NOCKSUM:
862 optval = inp->inp_flags & INP_UDP_NOCKSUM;
863 break;
864
865 default:
866 error = ENOPROTOOPT;
867 break;
868 }
869 if (error == 0)
870 error = sooptcopyout(sopt, &optval, sizeof (optval));
871 break;
872 }
873 return (error);
874}
875
1c79356b
A
876static int
877udp_pcblist SYSCTL_HANDLER_ARGS
878{
2d21ac55 879#pragma unused(oidp, arg1, arg2)
91447636 880 int error, i, n;
1c79356b
A
881 struct inpcb *inp, **inp_list;
882 inp_gen_t gencnt;
883 struct xinpgen xig;
884
885 /*
886 * The process of preparing the TCB list is too time-consuming and
887 * resource-intensive to repeat twice on every request.
888 */
91447636
A
889 lck_rw_lock_exclusive(udbinfo.mtx);
890 if (req->oldptr == USER_ADDR_NULL) {
1c79356b
A
891 n = udbinfo.ipi_count;
892 req->oldidx = 2 * (sizeof xig)
893 + (n + n/8) * sizeof(struct xinpcb);
91447636 894 lck_rw_done(udbinfo.mtx);
1c79356b
A
895 return 0;
896 }
897
91447636
A
898 if (req->newptr != USER_ADDR_NULL) {
899 lck_rw_done(udbinfo.mtx);
1c79356b 900 return EPERM;
91447636 901 }
1c79356b
A
902
903 /*
904 * OK, now we're committed to doing something.
905 */
1c79356b
A
906 gencnt = udbinfo.ipi_gencnt;
907 n = udbinfo.ipi_count;
1c79356b 908
3a60a9f5 909 bzero(&xig, sizeof(xig));
1c79356b
A
910 xig.xig_len = sizeof xig;
911 xig.xig_count = n;
912 xig.xig_gen = gencnt;
913 xig.xig_sogen = so_gencnt;
914 error = SYSCTL_OUT(req, &xig, sizeof xig);
91447636
A
915 if (error) {
916 lck_rw_done(udbinfo.mtx);
1c79356b 917 return error;
91447636 918 }
9bccf70c
A
919 /*
920 * We are done if there is no pcb
921 */
91447636
A
922 if (n == 0) {
923 lck_rw_done(udbinfo.mtx);
9bccf70c 924 return 0;
91447636 925 }
1c79356b
A
926
927 inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
928 if (inp_list == 0) {
91447636 929 lck_rw_done(udbinfo.mtx);
1c79356b
A
930 return ENOMEM;
931 }
9bccf70c
A
932
933 for (inp = LIST_FIRST(udbinfo.listhead), i = 0; inp && i < n;
934 inp = LIST_NEXT(inp, inp_list)) {
91447636 935 if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD)
1c79356b
A
936 inp_list[i++] = inp;
937 }
1c79356b
A
938 n = i;
939
940 error = 0;
941 for (i = 0; i < n; i++) {
942 inp = inp_list[i];
91447636 943 if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
1c79356b 944 struct xinpcb xi;
3a60a9f5
A
945
946 bzero(&xi, sizeof(xi));
1c79356b
A
947 xi.xi_len = sizeof xi;
948 /* XXX should avoid extra copy */
91447636 949 inpcb_to_compat(inp, &xi.xi_inp);
1c79356b
A
950 if (inp->inp_socket)
951 sotoxsocket(inp->inp_socket, &xi.xi_socket);
952 error = SYSCTL_OUT(req, &xi, sizeof xi);
953 }
954 }
955 if (!error) {
956 /*
957 * Give the user an updated idea of our state.
958 * If the generation differs from what we told
959 * her before, she knows that something happened
960 * while we were processing this request, and it
961 * might be necessary to retry.
962 */
3a60a9f5
A
963 bzero(&xig, sizeof(xig));
964 xig.xig_len = sizeof xig;
1c79356b
A
965 xig.xig_gen = udbinfo.ipi_gencnt;
966 xig.xig_sogen = so_gencnt;
967 xig.xig_count = udbinfo.ipi_count;
1c79356b
A
968 error = SYSCTL_OUT(req, &xig, sizeof xig);
969 }
970 FREE(inp_list, M_TEMP);
91447636 971 lck_rw_done(udbinfo.mtx);
1c79356b
A
972 return error;
973}
974
975SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
976 udp_pcblist, "S,xinpcb", "List of active UDP sockets");
977
978
979
91447636
A
980static __inline__ u_int16_t
981get_socket_id(struct socket * s)
982{
983 u_int16_t val;
984
985 if (s == NULL) {
986 return (0);
987 }
988 val = (u_int16_t)(((u_int32_t)s) / sizeof(struct socket));
989 if (val == 0) {
990 val = 0xffff;
991 }
992 return (val);
993}
994
1c79356b
A
995static int
996udp_output(inp, m, addr, control, p)
997 register struct inpcb *inp;
9bccf70c 998 struct mbuf *m;
1c79356b
A
999 struct sockaddr *addr;
1000 struct mbuf *control;
1001 struct proc *p;
1002{
1003 register struct udpiphdr *ui;
1004 register int len = m->m_pkthdr.len;
2d21ac55 1005 struct sockaddr_in *sin;
91447636
A
1006 struct in_addr origladdr, laddr, faddr;
1007 u_short lport, fport;
1008 struct sockaddr_in *ifaddr;
1009 int error = 0, udp_dodisconnect = 0;
2d21ac55
A
1010 struct socket *so = inp->inp_socket;
1011 int soopts;
1012 struct mbuf *inpopts;
1013 struct ip_moptions *mopts;
1014 struct route ro;
1c79356b
A
1015
1016 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_START, 0,0,0,0,0);
1017
1018 if (control)
1019 m_freem(control); /* XXX */
1020
1021 KERNEL_DEBUG(DBG_LAYER_OUT_BEG, inp->inp_fport, inp->inp_lport,
1022 inp->inp_laddr.s_addr, inp->inp_faddr.s_addr,
1023 (htons((u_short)len + sizeof (struct udphdr))));
1024
1025 if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
1026 error = EMSGSIZE;
1027 goto release;
1028 }
1029
55e303ae
A
1030 /* If there was a routing change, discard cached route and check
1031 * that we have a valid source address.
1032 * Reacquire a new source address if INADDR_ANY was specified
1033 */
1034
91447636
A
1035#if 1
1036 lck_mtx_assert(inp->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
1037#endif
1038
55e303ae 1039 if (inp->inp_route.ro_rt && inp->inp_route.ro_rt->generation_id != route_generation) {
91447636 1040 if (ifa_foraddr(inp->inp_laddr.s_addr) == 0) { /* src address is gone */
55e303ae 1041 if (inp->inp_flags & INP_INADDR_ANY)
6601e61a 1042 inp->inp_laddr.s_addr = INADDR_ANY; /* new src will be set later */
55e303ae
A
1043 else {
1044 error = EADDRNOTAVAIL;
1045 goto release;
1046 }
1047 }
1048 rtfree(inp->inp_route.ro_rt);
1049 inp->inp_route.ro_rt = (struct rtentry *)0;
1050 }
1051
91447636
A
1052 origladdr= laddr = inp->inp_laddr;
1053 faddr = inp->inp_faddr;
1054 lport = inp->inp_lport;
1055 fport = inp->inp_fport;
1056
1c79356b 1057 if (addr) {
91447636
A
1058 sin = (struct sockaddr_in *)addr;
1059 if (faddr.s_addr != INADDR_ANY) {
1c79356b
A
1060 error = EISCONN;
1061 goto release;
1062 }
91447636
A
1063 if (lport == 0) {
1064 /*
1065 * In case we don't have a local port set, go through the full connect.
1066 * We don't have a local port yet (ie, we can't be looked up),
1067 * so it's not an issue if the input runs at the same time we do this.
1068 */
1069 error = in_pcbconnect(inp, addr, p);
1070 if (error) {
1071 goto release;
1072 }
1073 laddr = inp->inp_laddr;
1074 lport = inp->inp_lport;
1075 faddr = inp->inp_faddr;
1076 fport = inp->inp_fport;
1077 udp_dodisconnect = 1;
1078 }
1079 else {
1080 /* Fast path case
1081 * we have a full address and a local port.
1082 * use those info to build the packet without changing the pcb
1083 * and interfering with the input path. See 3851370
1084 */
1085 if (laddr.s_addr == INADDR_ANY) {
1086 if ((error = in_pcbladdr(inp, addr, &ifaddr)) != 0)
1087 goto release;
1088 laddr = ifaddr->sin_addr;
1089 inp->inp_flags |= INP_INADDR_ANY; /* from pcbconnect: remember we don't care about src addr.*/
1090 }
1091
1092 faddr = sin->sin_addr;
1093 fport = sin->sin_port;
1c79356b
A
1094 }
1095 } else {
91447636 1096 if (faddr.s_addr == INADDR_ANY) {
1c79356b
A
1097 error = ENOTCONN;
1098 goto release;
1099 }
1100 }
55e303ae 1101
2d21ac55
A
1102#if CONFIG_MACF_NET
1103 mac_mbuf_label_associate_inpcb(inp, m);
1104#endif
1105
55e303ae 1106
1c79356b
A
1107 /*
1108 * Calculate data length and get a mbuf
1109 * for UDP and IP headers.
1110 */
1111 M_PREPEND(m, sizeof(struct udpiphdr), M_DONTWAIT);
1112 if (m == 0) {
1113 error = ENOBUFS;
55e303ae 1114 goto abort;
1c79356b
A
1115 }
1116
1117 /*
1118 * Fill in mbuf with extended UDP header
1119 * and addresses and length put into network format.
1120 */
1121 ui = mtod(m, struct udpiphdr *);
9bccf70c 1122 bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */
1c79356b 1123 ui->ui_pr = IPPROTO_UDP;
91447636
A
1124 ui->ui_src = laddr;
1125 ui->ui_dst = faddr;
1126 ui->ui_sport = lport;
1127 ui->ui_dport = fport;
9bccf70c 1128 ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));
1c79356b
A
1129
1130 /*
9bccf70c 1131 * Set up checksum and output datagram.
1c79356b 1132 */
2d21ac55 1133 if (udpcksum && !(inp->inp_flags & INP_UDP_NOCKSUM)) {
9bccf70c
A
1134 ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
1135 htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
1136 m->m_pkthdr.csum_flags = CSUM_UDP;
1137 m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
1138 } else {
0b4e3aa0 1139 ui->ui_sum = 0;
9bccf70c 1140 }
1c79356b
A
1141 ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len;
1142 ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */
1143 ((struct ip *)ui)->ip_tos = inp->inp_ip_tos; /* XXX */
1144 udpstat.udps_opackets++;
1145
1146 KERNEL_DEBUG(DBG_LAYER_OUT_END, ui->ui_dport, ui->ui_sport,
1147 ui->ui_src.s_addr, ui->ui_dst.s_addr, ui->ui_ulen);
1148
1c79356b 1149#if IPSEC
9bccf70c
A
1150 if (ipsec_bypass == 0 && ipsec_setsocket(m, inp->inp_socket) != 0) {
1151 error = ENOBUFS;
55e303ae 1152 goto abort;
9bccf70c 1153 }
1c79356b 1154#endif /*IPSEC*/
91447636 1155 m->m_pkthdr.socket_id = get_socket_id(inp->inp_socket);
2d21ac55
A
1156
1157 inpopts = inp->inp_options;
1158 soopts = (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST));
1159 mopts = inp->inp_moptions;
1160
1161 /* We don't want to cache the route for non-connected UDP */
1162 if (udp_dodisconnect) {
1163 bcopy(&inp->inp_route, &ro, sizeof (ro));
1164 ro.ro_rt = NULL;
1165 }
1166
1167 socket_unlock(so, 0);
1168 /* XXX jgraessley please look at XXX */
1169 error = ip_output_list(m, 0, inpopts,
1170 udp_dodisconnect ? &ro : &inp->inp_route, soopts, mopts, NULL);
1171 socket_lock(so, 0);
1c79356b 1172
91447636 1173 if (udp_dodisconnect) {
2d21ac55
A
1174 /* Discard the cached route, if there is one */
1175 if (ro.ro_rt != NULL)
1176 rtfree(ro.ro_rt);
1c79356b 1177 in_pcbdisconnect(inp);
91447636 1178 inp->inp_laddr = origladdr; /* XXX rehash? */
1c79356b
A
1179 }
1180 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_END, error, 0,0,0,0);
1181 return (error);
1182
55e303ae 1183abort:
91447636 1184 if (udp_dodisconnect) {
55e303ae 1185 in_pcbdisconnect(inp);
91447636 1186 inp->inp_laddr = origladdr; /* XXX rehash? */
55e303ae
A
1187 }
1188
1c79356b
A
1189release:
1190 m_freem(m);
1191 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT | DBG_FUNC_END, error, 0,0,0,0);
1192 return (error);
1193}
1194
1195u_long udp_sendspace = 9216; /* really max datagram size */
1196 /* 40 1K datagrams */
1197SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
9bccf70c 1198 &udp_sendspace, 0, "Maximum outgoing UDP datagram size");
1c79356b 1199
9bccf70c 1200u_long udp_recvspace = 40 * (1024 +
1c79356b
A
1201#if INET6
1202 sizeof(struct sockaddr_in6)
9bccf70c 1203#else
1c79356b 1204 sizeof(struct sockaddr_in)
9bccf70c 1205#endif
1c79356b
A
1206 );
1207SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
9bccf70c 1208 &udp_recvspace, 0, "Maximum incoming UDP datagram size");
1c79356b
A
1209
1210static int
1211udp_abort(struct socket *so)
1212{
1213 struct inpcb *inp;
1c79356b
A
1214
1215 inp = sotoinpcb(so);
1216 if (inp == 0)
2d21ac55 1217 panic("udp_abort: so=%p null inp\n", so); /* ??? possible? panic instead? */
1c79356b 1218 soisdisconnected(so);
1c79356b 1219 in_pcbdetach(inp);
1c79356b
A
1220 return 0;
1221}
1222
1223static int
2d21ac55 1224udp_attach(struct socket *so, __unused int proto, struct proc *p)
1c79356b
A
1225{
1226 struct inpcb *inp;
91447636 1227 int error;
1c79356b 1228
9bccf70c
A
1229 inp = sotoinpcb(so);
1230 if (inp != 0)
2d21ac55 1231 panic ("udp_attach so=%p inp=%p\n", so, inp);
9bccf70c 1232
1c79356b 1233 error = in_pcballoc(so, &udbinfo, p);
1c79356b
A
1234 if (error)
1235 return error;
91447636
A
1236 error = soreserve(so, udp_sendspace, udp_recvspace);
1237 if (error)
1238 return error;
1c79356b
A
1239 inp = (struct inpcb *)so->so_pcb;
1240 inp->inp_vflag |= INP_IPV4;
1241 inp->inp_ip_ttl = ip_defttl;
1c79356b
A
1242 return 0;
1243}
1244
1245static int
1246udp_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
1247{
1248 struct inpcb *inp;
91447636 1249 int error;
1c79356b 1250
2d21ac55
A
1251 if (nam->sa_family != 0 && nam->sa_family != AF_INET
1252 && nam->sa_family != AF_INET6) {
1253 return EAFNOSUPPORT;
1254 }
1c79356b
A
1255 inp = sotoinpcb(so);
1256 if (inp == 0)
1257 return EINVAL;
1c79356b 1258 error = in_pcbbind(inp, nam, p);
1c79356b
A
1259 return error;
1260}
1261
1262static int
1263udp_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
1264{
1265 struct inpcb *inp;
91447636 1266 int error;
1c79356b
A
1267
1268 inp = sotoinpcb(so);
1269 if (inp == 0)
1270 return EINVAL;
1271 if (inp->inp_faddr.s_addr != INADDR_ANY)
1272 return EISCONN;
1c79356b 1273 error = in_pcbconnect(inp, nam, p);
91447636 1274 if (error == 0)
1c79356b
A
1275 soisconnected(so);
1276 return error;
1277}
1278
1279static int
1280udp_detach(struct socket *so)
1281{
1282 struct inpcb *inp;
1c79356b
A
1283
1284 inp = sotoinpcb(so);
1285 if (inp == 0)
2d21ac55 1286 panic("udp_detach: so=%p null inp\n", so); /* ??? possible? panic instead? */
1c79356b 1287 in_pcbdetach(inp);
91447636 1288 inp->inp_state = INPCB_STATE_DEAD;
1c79356b
A
1289 return 0;
1290}
1291
1292static int
1293udp_disconnect(struct socket *so)
1294{
1295 struct inpcb *inp;
1c79356b
A
1296
1297 inp = sotoinpcb(so);
1298 if (inp == 0)
1299 return EINVAL;
1300 if (inp->inp_faddr.s_addr == INADDR_ANY)
1301 return ENOTCONN;
1302
1c79356b
A
1303 in_pcbdisconnect(inp);
1304 inp->inp_laddr.s_addr = INADDR_ANY;
1c79356b
A
1305 so->so_state &= ~SS_ISCONNECTED; /* XXX */
1306 return 0;
1307}
1308
1309static int
2d21ac55 1310udp_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr *addr,
1c79356b
A
1311 struct mbuf *control, struct proc *p)
1312{
1313 struct inpcb *inp;
1314
1315 inp = sotoinpcb(so);
1316 if (inp == 0) {
1317 m_freem(m);
1318 return EINVAL;
1319 }
1320 return udp_output(inp, m, addr, control, p);
1321}
1322
1323int
1324udp_shutdown(struct socket *so)
1325{
1326 struct inpcb *inp;
1327
1328 inp = sotoinpcb(so);
1329 if (inp == 0)
1330 return EINVAL;
1331 socantsendmore(so);
1332 return 0;
1333}
1334
1335struct pr_usrreqs udp_usrreqs = {
1336 udp_abort, pru_accept_notsupp, udp_attach, udp_bind, udp_connect,
1337 pru_connect2_notsupp, in_control, udp_detach, udp_disconnect,
1338 pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
1339 pru_rcvoob_notsupp, udp_send, pru_sense_null, udp_shutdown,
91447636 1340 in_setsockaddr, sosend, soreceive, pru_sopoll_notsupp
1c79356b
A
1341};
1342
91447636
A
1343
1344int
2d21ac55 1345udp_lock(struct socket *so, int refcount, int debug)
91447636
A
1346{
1347 int lr_saved;
0c530ab8
A
1348 if (debug == 0)
1349 lr_saved = (unsigned int) __builtin_return_address(0);
91447636 1350 else lr_saved = debug;
91447636
A
1351
1352 if (so->so_pcb) {
1353 lck_mtx_assert(((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_NOTOWNED);
1354 lck_mtx_lock(((struct inpcb *)so->so_pcb)->inpcb_mtx);
1355 }
0c530ab8 1356 else
2d21ac55 1357 panic("udp_lock: so=%p NO PCB! lr=%x\n", so, lr_saved);
91447636
A
1358
1359 if (refcount)
1360 so->so_usecount++;
1361
2d21ac55 1362 so->lock_lr[so->next_lock_lr] = (u_int32_t)lr_saved;
0c530ab8 1363 so->next_lock_lr = (so->next_lock_lr+1) % SO_LCKDBG_MAX;
91447636
A
1364 return (0);
1365}
1366
1367int
2d21ac55 1368udp_unlock(struct socket *so, int refcount, int debug)
91447636
A
1369{
1370 int lr_saved;
0c530ab8
A
1371
1372 if (debug == 0)
1373 lr_saved = (unsigned int) __builtin_return_address(0);
91447636 1374 else lr_saved = debug;
0c530ab8 1375
91447636
A
1376 if (refcount) {
1377 so->so_usecount--;
1378#if 0
2d21ac55
A
1379 {
1380 struct inpcb *inp = sotoinpcb(so);
1381 struct inpcbinfo *pcbinfo = &udbinfo;
1382
1383 if (so->so_usecount == 0 && (inp->inp_wantcnt == WNT_STOPUSING)) {
1384
1385 if (lck_rw_try_lock_exclusive(pcbinfo->mtx)) {
1386 in_pcbdispose(inp);
1387 lck_rw_done(pcbinfo->mtx);
1388 return(0);
1389 }
91447636
A
1390 }
1391 }
1392#endif
1393 }
0c530ab8 1394 if (so->so_pcb == NULL)
2d21ac55 1395 panic("udp_unlock: so=%p NO PCB! lr=%x\n", so, lr_saved);
91447636
A
1396 else {
1397 lck_mtx_assert(((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
2d21ac55 1398 so->unlock_lr[so->next_unlock_lr] = (u_int32_t)lr_saved;
0c530ab8 1399 so->next_unlock_lr = (so->next_unlock_lr+1) % SO_LCKDBG_MAX;
91447636
A
1400 lck_mtx_unlock(((struct inpcb *)so->so_pcb)->inpcb_mtx);
1401 }
1402
1403
91447636
A
1404 return (0);
1405}
1406
1407lck_mtx_t *
2d21ac55 1408udp_getlock(struct socket *so, __unused int locktype)
91447636
A
1409{
1410 struct inpcb *inp = sotoinpcb(so);
1411
1412
1413 if (so->so_pcb)
1414 return(inp->inpcb_mtx);
1415 else {
2d21ac55 1416 panic("udp_getlock: so=%p NULL so_pcb\n", so);
91447636
A
1417 return (so->so_proto->pr_domain->dom_mtx);
1418 }
1419}
1420
1421void
1422udp_slowtimo()
1423{
1424 struct inpcb *inp, *inpnxt;
1425 struct socket *so;
1426 struct inpcbinfo *pcbinfo = &udbinfo;
1427
2d21ac55
A
1428 if (lck_rw_try_lock_exclusive(pcbinfo->mtx) == FALSE) {
1429 if (udp_gc_done == TRUE) {
1430 udp_gc_done = FALSE;
1431 return; /* couldn't get the lock, better lock next time */
1432 }
1433 lck_rw_lock_exclusive(pcbinfo->mtx);
1434 }
1435
1436 udp_gc_done = TRUE;
91447636
A
1437
1438 for (inp = udb.lh_first; inp != NULL; inp = inpnxt) {
1439 inpnxt = inp->inp_list.le_next;
1440
1441 /* Ignore nat/SharedIP dummy pcbs */
1442 if (inp->inp_socket == &udbinfo.nat_dummy_socket)
1443 continue;
1444
1445 if (inp->inp_wantcnt != WNT_STOPUSING)
1446 continue;
1447
1448 so = inp->inp_socket;
1449 if (!lck_mtx_try_lock(inp->inpcb_mtx)) /* skip if busy, no hurry for cleanup... */
1450 continue;
1451
1452 if (so->so_usecount == 0)
1453 in_pcbdispose(inp);
1454 else
1455 lck_mtx_unlock(inp->inpcb_mtx);
1456 }
1457 lck_rw_done(pcbinfo->mtx);
1458}
1459
1460int
1461ChkAddressOK( __uint32_t dstaddr, __uint32_t srcaddr )
1462{
1463 if ( dstaddr == srcaddr ){
1464 return 0;
1465 }
1466 return 1;
1467}
1468
2d21ac55
A
1469void
1470udp_in_cksum_stats(u_int32_t len)
1471{
1472 udps_in_sw_cksum++;
1473 udps_in_sw_cksum_bytes += len;
1474}
1475
1476void
1477udp_out_cksum_stats(u_int32_t len)
1478{
1479 udps_out_sw_cksum++;
1480 udps_out_sw_cksum_bytes += len;
1481}