2 * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
4 * @APPLE_LICENSE_HEADER_START@
6 * The contents of this file constitute Original Code as defined in and
7 * are subject to the Apple Public Source License Version 1.1 (the
8 * "License"). You may not use this file except in compliance with the
9 * License. Please obtain a copy of the License at
10 * http://www.apple.com/publicsource and read it before using this file.
12 * This Original Code and all software distributed under the License are
13 * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
14 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
15 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. Please see the
17 * License for the specific language governing rights and limitations
20 * @APPLE_LICENSE_HEADER_END@
23 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
24 * The Regents of the University of California. All rights reserved.
26 * Redistribution and use in source and binary forms, with or without
27 * modification, are permitted provided that the following conditions
29 * 1. Redistributions of source code must retain the above copyright
30 * notice, this list of conditions and the following disclaimer.
31 * 2. Redistributions in binary form must reproduce the above copyright
32 * notice, this list of conditions and the following disclaimer in the
33 * documentation and/or other materials provided with the distribution.
34 * 3. All advertising materials mentioning features or use of this software
35 * must display the following acknowledgement:
36 * This product includes software developed by the University of
37 * California, Berkeley and its contributors.
38 * 4. Neither the name of the University nor the names of its contributors
39 * may be used to endorse or promote products derived from this software
40 * without specific prior written permission.
42 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
43 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
44 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
45 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
46 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
47 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
48 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
49 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
50 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
51 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
54 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
55 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.13 2001/08/08 18:59:54 ghelmer Exp $
58 #include <sys/param.h>
59 #include <sys/systm.h>
60 #include <sys/kernel.h>
61 #include <sys/malloc.h>
63 #include <sys/domain.h>
64 #include <sys/protosw.h>
65 #include <sys/socket.h>
66 #include <sys/socketvar.h>
67 #include <sys/sysctl.h>
68 #include <sys/syslog.h>
71 #include <net/if_types.h>
72 #include <net/route.h>
74 #include <netinet/in.h>
75 #include <netinet/in_systm.h>
76 #include <netinet/ip.h>
78 #include <netinet/ip6.h>
80 #include <netinet/in_pcb.h>
81 #include <netinet/in_var.h>
82 #include <netinet/ip_var.h>
84 #include <netinet6/ip6_var.h>
86 #include <netinet/ip_icmp.h>
87 #include <netinet/icmp_var.h>
88 #include <netinet/udp.h>
89 #include <netinet/udp_var.h>
90 #include <sys/kdebug.h>
93 #include <netinet6/ipsec.h>
94 extern int ipsec_bypass
;
95 extern lck_mtx_t
*sadb_mutex
;
99 #define DBG_LAYER_IN_BEG NETDBG_CODE(DBG_NETUDP, 0)
100 #define DBG_LAYER_IN_END NETDBG_CODE(DBG_NETUDP, 2)
101 #define DBG_LAYER_OUT_BEG NETDBG_CODE(DBG_NETUDP, 1)
102 #define DBG_LAYER_OUT_END NETDBG_CODE(DBG_NETUDP, 3)
103 #define DBG_FNC_UDP_INPUT NETDBG_CODE(DBG_NETUDP, (5 << 8))
104 #define DBG_FNC_UDP_OUTPUT NETDBG_CODE(DBG_NETUDP, (6 << 8) | 1)
107 * UDP protocol implementation.
108 * Per RFC 768, August, 1980.
111 static int udpcksum
= 1;
113 static int udpcksum
= 0; /* XXX */
115 SYSCTL_INT(_net_inet_udp
, UDPCTL_CHECKSUM
, checksum
, CTLFLAG_RW
,
119 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, log_in_vain
, CTLFLAG_RW
,
120 &log_in_vain
, 0, "Log all incoming UDP packets");
122 static int blackhole
= 0;
123 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, blackhole
, CTLFLAG_RW
,
124 &blackhole
, 0, "Do not send port unreachables for refused connects");
126 struct inpcbhead udb
; /* from udp_var.h */
127 #define udb6 udb /* for KAME src sync over BSD*'s */
128 struct inpcbinfo udbinfo
;
131 #define UDBHASHSIZE 16
134 extern int apple_hwcksum_rx
;
135 extern int esp_udp_encap_port
;
136 extern u_long route_generation
;
138 extern void ipfwsyslog( int level
, char *format
,...);
140 extern int fw_verbose
;
142 #define log_in_vain_log( a ) { \
143 if ( (log_in_vain == 3 ) && (fw_verbose == 2)) { /* Apple logging, log to ipfw.log */ \
149 struct udpstat udpstat
; /* from udp_var.h */
150 SYSCTL_STRUCT(_net_inet_udp
, UDPCTL_STATS
, stats
, CTLFLAG_RD
,
151 &udpstat
, udpstat
, "UDP statistics (struct udpstat, netinet/udp_var.h)");
152 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, pcbcount
, CTLFLAG_RD
,
153 &udbinfo
.ipi_count
, 0, "Number of active PCBs");
155 static struct sockaddr_in udp_in
= { sizeof(udp_in
), AF_INET
};
158 struct sockaddr_in6 uin6_sin
;
159 u_char uin6_init_done
: 1;
161 { sizeof(udp_in6
.uin6_sin
), AF_INET6
},
165 struct ip6_hdr uip6_ip6
;
166 u_char uip6_init_done
: 1;
170 static void udp_append(struct inpcb
*last
, struct ip
*ip
,
171 struct mbuf
*n
, int off
);
173 static void ip_2_ip6_hdr(struct ip6_hdr
*ip6
, struct ip
*ip
);
176 static int udp_detach(struct socket
*so
);
177 static int udp_output(struct inpcb
*, struct mbuf
*, struct sockaddr
*,
178 struct mbuf
*, struct proc
*);
179 extern int ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
);
185 struct inpcbinfo
*pcbinfo
;
189 udbinfo
.listhead
= &udb
;
190 udbinfo
.hashbase
= hashinit(UDBHASHSIZE
, M_PCB
, &udbinfo
.hashmask
);
191 udbinfo
.porthashbase
= hashinit(UDBHASHSIZE
, M_PCB
,
192 &udbinfo
.porthashmask
);
194 str_size
= (vm_size_t
) sizeof(struct inpcb
);
195 udbinfo
.ipi_zone
= (void *) zinit(str_size
, 80000*str_size
, 8192, "udpcb");
199 * allocate lock group attribute and group for udp pcb mutexes
201 pcbinfo
->mtx_grp_attr
= lck_grp_attr_alloc_init();
203 pcbinfo
->mtx_grp
= lck_grp_alloc_init("udppcb", pcbinfo
->mtx_grp_attr
);
205 pcbinfo
->mtx_attr
= lck_attr_alloc_init();
207 if ((pcbinfo
->mtx
= lck_rw_alloc_init(pcbinfo
->mtx_grp
, pcbinfo
->mtx_attr
)) == NULL
)
208 return; /* pretty much dead if this fails... */
210 in_pcb_nat_init(&udbinfo
, AF_INET
, IPPROTO_UDP
, SOCK_DGRAM
);
212 udbinfo
.ipi_zone
= zinit("udpcb", sizeof(struct inpcb
), maxsockets
,
217 /* for pcb sharing testing only */
218 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
219 kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat
);
221 laddr
.s_addr
= 0x11646464;
222 faddr
.s_addr
= 0x11646465;
225 in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
226 kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat
);
228 stat
= in_pcb_rem_share_client(&udbinfo
, fake_owner
);
229 kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat
);
231 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
232 kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat
);
234 laddr
.s_addr
= 0x11646464;
235 faddr
.s_addr
= 0x11646465;
238 stat
= in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
239 kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat
);
245 register struct mbuf
*m
;
248 register struct ip
*ip
;
249 register struct udphdr
*uh
;
250 register struct inpcb
*inp
;
251 struct mbuf
*opts
= 0;
254 struct sockaddr
*append_sa
;
255 struct inpcbinfo
*pcbinfo
= &udbinfo
;
257 udpstat
.udps_ipackets
++;
259 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_START
, 0,0,0,0,0);
260 if (m
->m_pkthdr
.csum_flags
& CSUM_TCP_SUM16
)
261 m
->m_pkthdr
.csum_flags
= 0; /* invalidate hwcksum for UDP */
264 * Strip IP options, if any; should skip this,
265 * make available to user, and use on returned packets,
266 * but we don't yet have a way to check the checksum
267 * with options still present.
269 if (iphlen
> sizeof (struct ip
)) {
270 ip_stripoptions(m
, (struct mbuf
*)0);
271 iphlen
= sizeof(struct ip
);
275 * Get IP and UDP header together in first mbuf.
277 ip
= mtod(m
, struct ip
*);
278 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
)) {
279 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
))) == 0) {
280 udpstat
.udps_hdrops
++;
281 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
284 ip
= mtod(m
, struct ip
*);
286 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
288 /* destination port of 0 is illegal, based on RFC768. */
289 if (uh
->uh_dport
== 0)
292 KERNEL_DEBUG(DBG_LAYER_IN_BEG
, uh
->uh_dport
, uh
->uh_sport
,
293 ip
->ip_src
.s_addr
, ip
->ip_dst
.s_addr
, uh
->uh_ulen
);
296 * Make mbuf data length reflect UDP length.
297 * If not enough data to reflect UDP length, drop.
299 len
= ntohs((u_short
)uh
->uh_ulen
);
300 if (ip
->ip_len
!= len
) {
301 if (len
> ip
->ip_len
|| len
< sizeof(struct udphdr
)) {
302 udpstat
.udps_badlen
++;
305 m_adj(m
, len
- ip
->ip_len
);
306 /* ip->ip_len = len; */
309 * Save a copy of the IP header in case we want restore it
310 * for sending an ICMP error message in response.
315 * Checksum extended UDP header and data.
318 if (m
->m_pkthdr
.csum_flags
& CSUM_DATA_VALID
) {
319 if (m
->m_pkthdr
.csum_flags
& CSUM_PSEUDO_HDR
)
320 uh
->uh_sum
= m
->m_pkthdr
.csum_data
;
323 uh
->uh_sum
^= 0xffff;
327 *(uint32_t*)&b
[0] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0];
328 *(uint32_t*)&b
[4] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4];
329 *(uint8_t*)&b
[8] = *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8];
331 bzero(((struct ipovly
*)ip
)->ih_x1
, 9);
332 ((struct ipovly
*)ip
)->ih_len
= uh
->uh_ulen
;
333 uh
->uh_sum
= in_cksum(m
, len
+ sizeof (struct ip
));
335 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0] = *(uint32_t*)&b
[0];
336 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4] = *(uint32_t*)&b
[4];
337 *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8] = *(uint8_t*)&b
[8];
340 udpstat
.udps_badsum
++;
342 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
348 udpstat
.udps_nosum
++;
351 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
)) ||
352 in_broadcast(ip
->ip_dst
, m
->m_pkthdr
.rcvif
)) {
354 lck_rw_lock_shared(pcbinfo
->mtx
);
356 * Deliver a multicast or broadcast datagram to *all* sockets
357 * for which the local and remote addresses and ports match
358 * those of the incoming datagram. This allows more than
359 * one process to receive multi/broadcasts on the same port.
360 * (This really ought to be done for unicast datagrams as
361 * well, but that would cause problems with existing
362 * applications that open both address-specific sockets and
363 * a wildcard socket listening to the same port -- they would
364 * end up receiving duplicates of every unicast datagram.
365 * Those applications open the multiple sockets to overcome an
366 * inadequacy of the UDP socket interface, but for backwards
367 * compatibility we avoid the problem here rather than
368 * fixing the interface. Maybe 4.5BSD will remedy this?)
373 * Construct sockaddr format source address.
375 udp_in
.sin_port
= uh
->uh_sport
;
376 udp_in
.sin_addr
= ip
->ip_src
;
378 * Locate pcb(s) for datagram.
379 * (Algorithm copied from raw_intr().)
383 udp_in6
.uin6_init_done
= udp_ip6
.uip6_init_done
= 0;
385 LIST_FOREACH(inp
, &udb
, inp_list
) {
387 /* Ignore nat/SharedIP dummy pcbs */
388 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
391 if (inp
->inp_socket
== NULL
)
393 if (inp
!= sotoinpcb(inp
->inp_socket
))
394 panic("udp_input: bad so back ptr inp=%x\n", inp
);
396 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
399 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) == WNT_STOPUSING
) {
403 udp_lock(inp
->inp_socket
, 1, 0);
405 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
406 udp_unlock(inp
->inp_socket
, 1, 0);
410 if (inp
->inp_lport
!= uh
->uh_dport
) {
411 udp_unlock(inp
->inp_socket
, 1, 0);
414 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
) {
415 if (inp
->inp_laddr
.s_addr
!=
417 udp_unlock(inp
->inp_socket
, 1, 0);
421 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
) {
422 if (inp
->inp_faddr
.s_addr
!=
424 inp
->inp_fport
!= uh
->uh_sport
) {
425 udp_unlock(inp
->inp_socket
, 1, 0);
434 /* check AH/ESP integrity. */
435 if (ipsec_bypass
== 0) {
436 lck_mtx_lock(sadb_mutex
);
437 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
438 ipsecstat
.in_polvio
++;
439 /* do not inject data to pcb */
442 lck_mtx_unlock(sadb_mutex
);
446 if ((n
= m_copy(m
, 0, M_COPYALL
)) != NULL
) {
447 udp_append(last
, ip
, n
,
449 sizeof(struct udphdr
));
451 udp_unlock(last
->inp_socket
, 1, 0);
455 * Don't look for additional matches if this one does
456 * not have either the SO_REUSEPORT or SO_REUSEADDR
457 * socket options set. This heuristic avoids searching
458 * through all pcbs in the common case of a non-shared
459 * port. It * assumes that an application will never
460 * clear these options after setting them.
462 if ((last
->inp_socket
->so_options
&(SO_REUSEPORT
|SO_REUSEADDR
)) == 0)
465 lck_rw_done(pcbinfo
->mtx
);
469 * No matching pcb found; discard datagram.
470 * (No need to send an ICMP Port Unreachable
471 * for a broadcast or multicast datgram.)
473 udpstat
.udps_noportbcast
++;
477 /* check AH/ESP integrity. */
478 if (ipsec_bypass
== 0 && m
) {
479 lck_mtx_lock(sadb_mutex
);
480 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
481 ipsecstat
.in_polvio
++;
482 lck_mtx_unlock(sadb_mutex
);
483 udp_unlock(last
->inp_socket
, 1, 0);
486 lck_mtx_unlock(sadb_mutex
);
489 udp_append(last
, ip
, m
, iphlen
+ sizeof(struct udphdr
));
490 udp_unlock(last
->inp_socket
, 1, 0);
496 * UDP to port 4500 with a payload where the first four bytes are
497 * not zero is a UDP encapsulated IPSec packet. Packets where
498 * the payload is one byte and that byte is 0xFF are NAT keepalive
499 * packets. Decapsulate the ESP packet and carry on with IPSec input
500 * or discard the NAT keep-alive.
502 if (ipsec_bypass
== 0 && (esp_udp_encap_port
& 0xFFFF) != 0 &&
503 uh
->uh_dport
== ntohs((u_short
)esp_udp_encap_port
)) {
504 int payload_len
= len
- sizeof(struct udphdr
) > 4 ? 4 : len
- sizeof(struct udphdr
);
505 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
) + payload_len
) {
506 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
) + payload_len
)) == 0) {
507 udpstat
.udps_hdrops
++;
508 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
511 ip
= mtod(m
, struct ip
*);
512 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
514 /* Check for NAT keepalive packet */
515 if (payload_len
== 1 && *(u_int8_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) == 0xFF) {
517 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
520 else if (payload_len
== 4 && *(u_int32_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) != 0) {
521 /* UDP encapsulated IPSec packet to pass through NAT */
524 stripsiz
= sizeof(struct udphdr
);
526 ip
= mtod(m
, struct ip
*);
527 ovbcopy((caddr_t
)ip
, (caddr_t
)(((u_char
*)ip
) + stripsiz
), iphlen
);
528 m
->m_data
+= stripsiz
;
529 m
->m_len
-= stripsiz
;
530 m
->m_pkthdr
.len
-= stripsiz
;
531 ip
= mtod(m
, struct ip
*);
532 ip
->ip_len
= ip
->ip_len
- stripsiz
;
533 ip
->ip_p
= IPPROTO_ESP
;
535 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
536 esp4_input(m
, iphlen
);
543 * Locate pcb for datagram.
545 inp
= in_pcblookup_hash(&udbinfo
, ip
->ip_src
, uh
->uh_sport
,
546 ip
->ip_dst
, uh
->uh_dport
, 1, m
->m_pkthdr
.rcvif
);
549 char buf
[MAX_IPv4_STR_LEN
];
550 char buf2
[MAX_IPv4_STR_LEN
];
552 /* check src and dst address */
553 if (log_in_vain
!= 3)
555 "Connection attempt to UDP %s:%d from %s:%d\n",
556 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
558 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
559 ntohs(uh
->uh_sport
));
560 else if (!(m
->m_flags
& (M_BCAST
| M_MCAST
)) &&
561 ip
->ip_dst
.s_addr
!= ip
->ip_src
.s_addr
)
562 log_in_vain_log((LOG_INFO
,
563 "Stealth Mode connection attempt to UDP %s:%d from %s:%d\n",
564 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
566 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
567 ntohs(uh
->uh_sport
)))
569 udpstat
.udps_noport
++;
570 if (m
->m_flags
& (M_BCAST
| M_MCAST
)) {
571 udpstat
.udps_noportbcast
++;
575 if (badport_bandlim(BANDLIM_ICMP_UNREACH
) < 0)
579 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
!= IFT_LOOP
)
582 ip
->ip_len
+= iphlen
;
583 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_PORT
, 0, 0);
584 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
587 udp_lock(inp
->inp_socket
, 1, 0);
589 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
590 udp_unlock(inp
->inp_socket
, 1, 0);
594 if (ipsec_bypass
== 0 && inp
!= NULL
) {
595 lck_mtx_lock(sadb_mutex
);
596 if (ipsec4_in_reject_so(m
, inp
->inp_socket
)) {
597 ipsecstat
.in_polvio
++;
598 lck_mtx_unlock(sadb_mutex
);
599 udp_unlock(inp
->inp_socket
, 1, 0);
602 lck_mtx_unlock(sadb_mutex
);
607 * Construct sockaddr format source address.
608 * Stuff source address and datagram in user buffer.
610 udp_in
.sin_port
= uh
->uh_sport
;
611 udp_in
.sin_addr
= ip
->ip_src
;
612 if (inp
->inp_flags
& INP_CONTROLOPTS
613 || inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
615 if (inp
->inp_vflag
& INP_IPV6
) {
618 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
619 savedflags
= inp
->inp_flags
;
620 inp
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
621 ip6_savecontrol(inp
, &opts
, &udp_ip6
.uip6_ip6
, m
);
622 inp
->inp_flags
= savedflags
;
625 ip_savecontrol(inp
, &opts
, ip
, m
);
627 m_adj(m
, iphlen
+ sizeof(struct udphdr
));
629 KERNEL_DEBUG(DBG_LAYER_IN_END
, uh
->uh_dport
, uh
->uh_sport
,
630 save_ip
.ip_src
.s_addr
, save_ip
.ip_dst
.s_addr
, uh
->uh_ulen
);
633 if (inp
->inp_vflag
& INP_IPV6
) {
634 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
635 append_sa
= (struct sockaddr
*)&udp_in6
;
638 append_sa
= (struct sockaddr
*)&udp_in
;
639 if (sbappendaddr(&inp
->inp_socket
->so_rcv
, append_sa
, m
, opts
, NULL
) == 0) {
640 udpstat
.udps_fullsock
++;
643 sorwakeup(inp
->inp_socket
);
645 udp_unlock(inp
->inp_socket
, 1, 0);
646 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
652 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
658 ip_2_ip6_hdr(ip6
, ip
)
662 bzero(ip6
, sizeof(*ip6
));
664 ip6
->ip6_vfc
= IPV6_VERSION
;
665 ip6
->ip6_plen
= ip
->ip_len
;
666 ip6
->ip6_nxt
= ip
->ip_p
;
667 ip6
->ip6_hlim
= ip
->ip_ttl
;
668 ip6
->ip6_src
.s6_addr32
[2] = ip6
->ip6_dst
.s6_addr32
[2] =
670 ip6
->ip6_src
.s6_addr32
[3] = ip
->ip_src
.s_addr
;
671 ip6
->ip6_dst
.s6_addr32
[3] = ip
->ip_dst
.s_addr
;
676 * subroutine of udp_input(), mainly for source code readability.
677 * caller must properly init udp_ip6 and udp_in6 beforehand.
680 udp_append(last
, ip
, n
, off
)
686 struct sockaddr
*append_sa
;
687 struct mbuf
*opts
= 0;
689 if (last
->inp_flags
& INP_CONTROLOPTS
||
690 last
->inp_socket
->so_options
& SO_TIMESTAMP
) {
692 if (last
->inp_vflag
& INP_IPV6
) {
695 if (udp_ip6
.uip6_init_done
== 0) {
696 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
697 udp_ip6
.uip6_init_done
= 1;
699 savedflags
= last
->inp_flags
;
700 last
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
701 ip6_savecontrol(last
, &opts
, &udp_ip6
.uip6_ip6
, n
);
702 last
->inp_flags
= savedflags
;
705 ip_savecontrol(last
, &opts
, ip
, n
);
708 if (last
->inp_vflag
& INP_IPV6
) {
709 if (udp_in6
.uin6_init_done
== 0) {
710 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
711 udp_in6
.uin6_init_done
= 1;
713 append_sa
= (struct sockaddr
*)&udp_in6
.uin6_sin
;
716 append_sa
= (struct sockaddr
*)&udp_in
;
718 if (sbappendaddr(&last
->inp_socket
->so_rcv
, append_sa
, n
, opts
, NULL
) == 0) {
719 udpstat
.udps_fullsock
++;
721 sorwakeup(last
->inp_socket
);
725 * Notify a udp user of an asynchronous error;
726 * just wake up so that he can collect error status.
729 udp_notify(inp
, errno
)
730 register struct inpcb
*inp
;
733 inp
->inp_socket
->so_error
= errno
;
734 sorwakeup(inp
->inp_socket
);
735 sowwakeup(inp
->inp_socket
);
739 udp_ctlinput(cmd
, sa
, vip
)
746 void (*notify
)(struct inpcb
*, int) = udp_notify
;
747 struct in_addr faddr
;
750 faddr
= ((struct sockaddr_in
*)sa
)->sin_addr
;
751 if (sa
->sa_family
!= AF_INET
|| faddr
.s_addr
== INADDR_ANY
)
754 if (PRC_IS_REDIRECT(cmd
)) {
756 notify
= in_rtchange
;
757 } else if (cmd
== PRC_HOSTDEAD
)
759 else if ((unsigned)cmd
>= PRC_NCMDS
|| inetctlerrmap
[cmd
] == 0)
762 uh
= (struct udphdr
*)((caddr_t
)ip
+ (ip
->ip_hl
<< 2));
763 inp
= in_pcblookup_hash(&udbinfo
, faddr
, uh
->uh_dport
,
764 ip
->ip_src
, uh
->uh_sport
, 0, NULL
);
765 if (inp
!= NULL
&& inp
->inp_socket
!= NULL
) {
766 udp_lock(inp
->inp_socket
, 1, 0);
767 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
768 udp_unlock(inp
->inp_socket
, 1, 0);
771 (*notify
)(inp
, inetctlerrmap
[cmd
]);
772 udp_unlock(inp
->inp_socket
, 1, 0);
775 in_pcbnotifyall(&udbinfo
, faddr
, inetctlerrmap
[cmd
], notify
);
779 udp_pcblist SYSCTL_HANDLER_ARGS
782 struct inpcb
*inp
, **inp_list
;
787 * The process of preparing the TCB list is too time-consuming and
788 * resource-intensive to repeat twice on every request.
790 lck_rw_lock_exclusive(udbinfo
.mtx
);
791 if (req
->oldptr
== USER_ADDR_NULL
) {
792 n
= udbinfo
.ipi_count
;
793 req
->oldidx
= 2 * (sizeof xig
)
794 + (n
+ n
/8) * sizeof(struct xinpcb
);
795 lck_rw_done(udbinfo
.mtx
);
799 if (req
->newptr
!= USER_ADDR_NULL
) {
800 lck_rw_done(udbinfo
.mtx
);
805 * OK, now we're committed to doing something.
807 gencnt
= udbinfo
.ipi_gencnt
;
808 n
= udbinfo
.ipi_count
;
810 bzero(&xig
, sizeof(xig
));
811 xig
.xig_len
= sizeof xig
;
813 xig
.xig_gen
= gencnt
;
814 xig
.xig_sogen
= so_gencnt
;
815 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
817 lck_rw_done(udbinfo
.mtx
);
821 * We are done if there is no pcb
824 lck_rw_done(udbinfo
.mtx
);
828 inp_list
= _MALLOC(n
* sizeof *inp_list
, M_TEMP
, M_WAITOK
);
830 lck_rw_done(udbinfo
.mtx
);
834 for (inp
= LIST_FIRST(udbinfo
.listhead
), i
= 0; inp
&& i
< n
;
835 inp
= LIST_NEXT(inp
, inp_list
)) {
836 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
)
842 for (i
= 0; i
< n
; i
++) {
844 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
) {
847 bzero(&xi
, sizeof(xi
));
848 xi
.xi_len
= sizeof xi
;
849 /* XXX should avoid extra copy */
850 inpcb_to_compat(inp
, &xi
.xi_inp
);
852 sotoxsocket(inp
->inp_socket
, &xi
.xi_socket
);
853 error
= SYSCTL_OUT(req
, &xi
, sizeof xi
);
858 * Give the user an updated idea of our state.
859 * If the generation differs from what we told
860 * her before, she knows that something happened
861 * while we were processing this request, and it
862 * might be necessary to retry.
864 bzero(&xig
, sizeof(xig
));
865 xig
.xig_len
= sizeof xig
;
866 xig
.xig_gen
= udbinfo
.ipi_gencnt
;
867 xig
.xig_sogen
= so_gencnt
;
868 xig
.xig_count
= udbinfo
.ipi_count
;
869 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
871 FREE(inp_list
, M_TEMP
);
872 lck_rw_done(udbinfo
.mtx
);
876 SYSCTL_PROC(_net_inet_udp
, UDPCTL_PCBLIST
, pcblist
, CTLFLAG_RD
, 0, 0,
877 udp_pcblist
, "S,xinpcb", "List of active UDP sockets");
881 static __inline__ u_int16_t
882 get_socket_id(struct socket
* s
)
889 val
= (u_int16_t
)(((u_int32_t
)s
) / sizeof(struct socket
));
897 udp_output(inp
, m
, addr
, control
, p
)
898 register struct inpcb
*inp
;
900 struct sockaddr
*addr
;
901 struct mbuf
*control
;
904 register struct udpiphdr
*ui
;
905 register int len
= m
->m_pkthdr
.len
;
906 struct sockaddr_in
*sin
, src
;
907 struct in_addr origladdr
, laddr
, faddr
;
908 u_short lport
, fport
;
909 struct sockaddr_in
*ifaddr
;
910 int error
= 0, udp_dodisconnect
= 0;
913 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_START
, 0,0,0,0,0);
916 m_freem(control
); /* XXX */
918 KERNEL_DEBUG(DBG_LAYER_OUT_BEG
, inp
->inp_fport
, inp
->inp_lport
,
919 inp
->inp_laddr
.s_addr
, inp
->inp_faddr
.s_addr
,
920 (htons((u_short
)len
+ sizeof (struct udphdr
))));
922 if (len
+ sizeof(struct udpiphdr
) > IP_MAXPACKET
) {
927 /* If there was a routing change, discard cached route and check
928 * that we have a valid source address.
929 * Reacquire a new source address if INADDR_ANY was specified
933 lck_mtx_assert(inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
936 if (inp
->inp_route
.ro_rt
&& inp
->inp_route
.ro_rt
->generation_id
!= route_generation
) {
937 if (ifa_foraddr(inp
->inp_laddr
.s_addr
) == 0) { /* src address is gone */
938 if (inp
->inp_flags
& INP_INADDR_ANY
)
939 inp
->inp_faddr
.s_addr
= INADDR_ANY
; /* new src will be set later */
941 error
= EADDRNOTAVAIL
;
945 rtfree(inp
->inp_route
.ro_rt
);
946 inp
->inp_route
.ro_rt
= (struct rtentry
*)0;
949 origladdr
= laddr
= inp
->inp_laddr
;
950 faddr
= inp
->inp_faddr
;
951 lport
= inp
->inp_lport
;
952 fport
= inp
->inp_fport
;
955 sin
= (struct sockaddr_in
*)addr
;
956 if (faddr
.s_addr
!= INADDR_ANY
) {
962 * In case we don't have a local port set, go through the full connect.
963 * We don't have a local port yet (ie, we can't be looked up),
964 * so it's not an issue if the input runs at the same time we do this.
966 error
= in_pcbconnect(inp
, addr
, p
);
970 laddr
= inp
->inp_laddr
;
971 lport
= inp
->inp_lport
;
972 faddr
= inp
->inp_faddr
;
973 fport
= inp
->inp_fport
;
974 udp_dodisconnect
= 1;
978 * we have a full address and a local port.
979 * use those info to build the packet without changing the pcb
980 * and interfering with the input path. See 3851370
982 if (laddr
.s_addr
== INADDR_ANY
) {
983 if ((error
= in_pcbladdr(inp
, addr
, &ifaddr
)) != 0)
985 laddr
= ifaddr
->sin_addr
;
986 inp
->inp_flags
|= INP_INADDR_ANY
; /* from pcbconnect: remember we don't care about src addr.*/
989 faddr
= sin
->sin_addr
;
990 fport
= sin
->sin_port
;
993 if (faddr
.s_addr
== INADDR_ANY
) {
1001 * Calculate data length and get a mbuf
1002 * for UDP and IP headers.
1004 M_PREPEND(m
, sizeof(struct udpiphdr
), M_DONTWAIT
);
1011 * Fill in mbuf with extended UDP header
1012 * and addresses and length put into network format.
1014 ui
= mtod(m
, struct udpiphdr
*);
1015 bzero(ui
->ui_x1
, sizeof(ui
->ui_x1
)); /* XXX still needed? */
1016 ui
->ui_pr
= IPPROTO_UDP
;
1019 ui
->ui_sport
= lport
;
1020 ui
->ui_dport
= fport
;
1021 ui
->ui_ulen
= htons((u_short
)len
+ sizeof(struct udphdr
));
1024 * Set up checksum and output datagram.
1027 ui
->ui_sum
= in_pseudo(ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
,
1028 htons((u_short
)len
+ sizeof(struct udphdr
) + IPPROTO_UDP
));
1029 m
->m_pkthdr
.csum_flags
= CSUM_UDP
;
1030 m
->m_pkthdr
.csum_data
= offsetof(struct udphdr
, uh_sum
);
1034 ((struct ip
*)ui
)->ip_len
= sizeof (struct udpiphdr
) + len
;
1035 ((struct ip
*)ui
)->ip_ttl
= inp
->inp_ip_ttl
; /* XXX */
1036 ((struct ip
*)ui
)->ip_tos
= inp
->inp_ip_tos
; /* XXX */
1037 udpstat
.udps_opackets
++;
1039 KERNEL_DEBUG(DBG_LAYER_OUT_END
, ui
->ui_dport
, ui
->ui_sport
,
1040 ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
, ui
->ui_ulen
);
1043 if (ipsec_bypass
== 0 && ipsec_setsocket(m
, inp
->inp_socket
) != 0) {
1048 m
->m_pkthdr
.socket_id
= get_socket_id(inp
->inp_socket
);
1049 error
= ip_output_list(m
, 0, inp
->inp_options
, &inp
->inp_route
,
1050 (inp
->inp_socket
->so_options
& (SO_DONTROUTE
| SO_BROADCAST
)),
1053 if (udp_dodisconnect
) {
1054 in_pcbdisconnect(inp
);
1055 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1057 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1061 if (udp_dodisconnect
) {
1062 in_pcbdisconnect(inp
);
1063 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1068 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1072 u_long udp_sendspace
= 9216; /* really max datagram size */
1073 /* 40 1K datagrams */
1074 SYSCTL_INT(_net_inet_udp
, UDPCTL_MAXDGRAM
, maxdgram
, CTLFLAG_RW
,
1075 &udp_sendspace
, 0, "Maximum outgoing UDP datagram size");
1077 u_long udp_recvspace
= 40 * (1024 +
1079 sizeof(struct sockaddr_in6
)
1081 sizeof(struct sockaddr_in
)
1084 SYSCTL_INT(_net_inet_udp
, UDPCTL_RECVSPACE
, recvspace
, CTLFLAG_RW
,
1085 &udp_recvspace
, 0, "Maximum incoming UDP datagram size");
1088 udp_abort(struct socket
*so
)
1092 inp
= sotoinpcb(so
);
1094 panic("udp_abort: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1095 soisdisconnected(so
);
1101 udp_attach(struct socket
*so
, int proto
, struct proc
*p
)
1106 inp
= sotoinpcb(so
);
1108 panic ("udp_attach so=%x inp=%x\n", so
, inp
);
1110 error
= in_pcballoc(so
, &udbinfo
, p
);
1113 error
= soreserve(so
, udp_sendspace
, udp_recvspace
);
1116 inp
= (struct inpcb
*)so
->so_pcb
;
1117 inp
->inp_vflag
|= INP_IPV4
;
1118 inp
->inp_ip_ttl
= ip_defttl
;
1123 udp_bind(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1128 inp
= sotoinpcb(so
);
1131 error
= in_pcbbind(inp
, nam
, p
);
1136 udp_connect(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1141 inp
= sotoinpcb(so
);
1144 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
)
1146 error
= in_pcbconnect(inp
, nam
, p
);
1153 udp_detach(struct socket
*so
)
1157 inp
= sotoinpcb(so
);
1159 panic("udp_detach: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1161 inp
->inp_state
= INPCB_STATE_DEAD
;
1166 udp_disconnect(struct socket
*so
)
1170 inp
= sotoinpcb(so
);
1173 if (inp
->inp_faddr
.s_addr
== INADDR_ANY
)
1176 in_pcbdisconnect(inp
);
1177 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
1178 so
->so_state
&= ~SS_ISCONNECTED
; /* XXX */
1183 udp_send(struct socket
*so
, int flags
, struct mbuf
*m
, struct sockaddr
*addr
,
1184 struct mbuf
*control
, struct proc
*p
)
1188 inp
= sotoinpcb(so
);
1193 return udp_output(inp
, m
, addr
, control
, p
);
1197 udp_shutdown(struct socket
*so
)
1201 inp
= sotoinpcb(so
);
1208 struct pr_usrreqs udp_usrreqs
= {
1209 udp_abort
, pru_accept_notsupp
, udp_attach
, udp_bind
, udp_connect
,
1210 pru_connect2_notsupp
, in_control
, udp_detach
, udp_disconnect
,
1211 pru_listen_notsupp
, in_setpeeraddr
, pru_rcvd_notsupp
,
1212 pru_rcvoob_notsupp
, udp_send
, pru_sense_null
, udp_shutdown
,
1213 in_setsockaddr
, sosend
, soreceive
, pru_sopoll_notsupp
1218 udp_lock(so
, refcount
, debug
)
1220 int refcount
, debug
;
1224 lr_saved
= (unsigned int) __builtin_return_address(0);
1225 else lr_saved
= debug
;
1228 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_NOTOWNED
);
1229 lck_mtx_lock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1232 panic("udp_lock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1237 so
->lock_lr
[so
->next_lock_lr
] = (void *)lr_saved
;
1238 so
->next_lock_lr
= (so
->next_lock_lr
+1) % SO_LCKDBG_MAX
;
1243 udp_unlock(so
, refcount
, debug
)
1249 struct inpcb
*inp
= sotoinpcb(so
);
1250 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1253 lr_saved
= (unsigned int) __builtin_return_address(0);
1254 else lr_saved
= debug
;
1259 if (so
->so_usecount
== 0 && (inp
->inp_wantcnt
== WNT_STOPUSING
)) {
1260 if (lck_rw_try_lock_exclusive(pcbinfo
->mtx
)) {
1262 lck_rw_done(pcbinfo
->mtx
);
1268 if (so
->so_pcb
== NULL
)
1269 panic("udp_unlock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1271 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1272 so
->unlock_lr
[so
->next_unlock_lr
] = (void *)lr_saved
;
1273 so
->next_unlock_lr
= (so
->next_unlock_lr
+1) % SO_LCKDBG_MAX
;
1274 lck_mtx_unlock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1282 udp_getlock(so
, locktype
)
1286 struct inpcb
*inp
= sotoinpcb(so
);
1290 return(inp
->inpcb_mtx
);
1292 panic("udp_getlock: so=%x NULL so_pcb\n", so
);
1293 return (so
->so_proto
->pr_domain
->dom_mtx
);
1300 struct inpcb
*inp
, *inpnxt
;
1302 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1304 lck_rw_lock_exclusive(pcbinfo
->mtx
);
1306 for (inp
= udb
.lh_first
; inp
!= NULL
; inp
= inpnxt
) {
1307 inpnxt
= inp
->inp_list
.le_next
;
1309 /* Ignore nat/SharedIP dummy pcbs */
1310 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
1313 if (inp
->inp_wantcnt
!= WNT_STOPUSING
)
1316 so
= inp
->inp_socket
;
1317 if (!lck_mtx_try_lock(inp
->inpcb_mtx
)) /* skip if busy, no hurry for cleanup... */
1320 if (so
->so_usecount
== 0)
1323 lck_mtx_unlock(inp
->inpcb_mtx
);
1325 lck_rw_done(pcbinfo
->mtx
);
1329 ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
)
1331 if ( dstaddr
== srcaddr
){