2 * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
4 * @APPLE_LICENSE_HEADER_START@
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. Please obtain a copy of the License at
10 * http://www.opensource.apple.com/apsl/ and read it before using this
13 * The Original Code and all software distributed under the License are
14 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
15 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
16 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
18 * Please see the License for the specific language governing rights and
19 * limitations under the License.
21 * @APPLE_LICENSE_HEADER_END@
24 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
25 * The Regents of the University of California. All rights reserved.
27 * Redistribution and use in source and binary forms, with or without
28 * modification, are permitted provided that the following conditions
30 * 1. Redistributions of source code must retain the above copyright
31 * notice, this list of conditions and the following disclaimer.
32 * 2. Redistributions in binary form must reproduce the above copyright
33 * notice, this list of conditions and the following disclaimer in the
34 * documentation and/or other materials provided with the distribution.
35 * 3. All advertising materials mentioning features or use of this software
36 * must display the following acknowledgement:
37 * This product includes software developed by the University of
38 * California, Berkeley and its contributors.
39 * 4. Neither the name of the University nor the names of its contributors
40 * may be used to endorse or promote products derived from this software
41 * without specific prior written permission.
43 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
44 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
45 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
46 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
47 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
48 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
49 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
50 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
51 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
52 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
55 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
56 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.13 2001/08/08 18:59:54 ghelmer Exp $
59 #include <sys/param.h>
60 #include <sys/systm.h>
61 #include <sys/kernel.h>
62 #include <sys/malloc.h>
64 #include <sys/domain.h>
65 #include <sys/protosw.h>
66 #include <sys/socket.h>
67 #include <sys/socketvar.h>
68 #include <sys/sysctl.h>
69 #include <sys/syslog.h>
72 #include <net/if_types.h>
73 #include <net/route.h>
75 #include <netinet/in.h>
76 #include <netinet/in_systm.h>
77 #include <netinet/ip.h>
79 #include <netinet/ip6.h>
81 #include <netinet/in_pcb.h>
82 #include <netinet/in_var.h>
83 #include <netinet/ip_var.h>
85 #include <netinet6/ip6_var.h>
87 #include <netinet/ip_icmp.h>
88 #include <netinet/icmp_var.h>
89 #include <netinet/udp.h>
90 #include <netinet/udp_var.h>
91 #include <sys/kdebug.h>
94 #include <netinet6/ipsec.h>
95 extern int ipsec_bypass
;
96 extern lck_mtx_t
*sadb_mutex
;
100 #define DBG_LAYER_IN_BEG NETDBG_CODE(DBG_NETUDP, 0)
101 #define DBG_LAYER_IN_END NETDBG_CODE(DBG_NETUDP, 2)
102 #define DBG_LAYER_OUT_BEG NETDBG_CODE(DBG_NETUDP, 1)
103 #define DBG_LAYER_OUT_END NETDBG_CODE(DBG_NETUDP, 3)
104 #define DBG_FNC_UDP_INPUT NETDBG_CODE(DBG_NETUDP, (5 << 8))
105 #define DBG_FNC_UDP_OUTPUT NETDBG_CODE(DBG_NETUDP, (6 << 8) | 1)
108 * UDP protocol implementation.
109 * Per RFC 768, August, 1980.
112 static int udpcksum
= 1;
114 static int udpcksum
= 0; /* XXX */
116 SYSCTL_INT(_net_inet_udp
, UDPCTL_CHECKSUM
, checksum
, CTLFLAG_RW
,
120 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, log_in_vain
, CTLFLAG_RW
,
121 &log_in_vain
, 0, "Log all incoming UDP packets");
123 static int blackhole
= 0;
124 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, blackhole
, CTLFLAG_RW
,
125 &blackhole
, 0, "Do not send port unreachables for refused connects");
127 struct inpcbhead udb
; /* from udp_var.h */
128 #define udb6 udb /* for KAME src sync over BSD*'s */
129 struct inpcbinfo udbinfo
;
132 #define UDBHASHSIZE 16
135 extern int apple_hwcksum_rx
;
136 extern int esp_udp_encap_port
;
137 extern u_long route_generation
;
139 extern void ipfwsyslog( int level
, char *format
,...);
141 extern int fw_verbose
;
143 #define log_in_vain_log( a ) { \
144 if ( (log_in_vain == 3 ) && (fw_verbose == 2)) { /* Apple logging, log to ipfw.log */ \
150 struct udpstat udpstat
; /* from udp_var.h */
151 SYSCTL_STRUCT(_net_inet_udp
, UDPCTL_STATS
, stats
, CTLFLAG_RD
,
152 &udpstat
, udpstat
, "UDP statistics (struct udpstat, netinet/udp_var.h)");
153 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, pcbcount
, CTLFLAG_RD
,
154 &udbinfo
.ipi_count
, 0, "Number of active PCBs");
156 static struct sockaddr_in udp_in
= { sizeof(udp_in
), AF_INET
};
159 struct sockaddr_in6 uin6_sin
;
160 u_char uin6_init_done
: 1;
162 { sizeof(udp_in6
.uin6_sin
), AF_INET6
},
166 struct ip6_hdr uip6_ip6
;
167 u_char uip6_init_done
: 1;
171 static void udp_append(struct inpcb
*last
, struct ip
*ip
,
172 struct mbuf
*n
, int off
);
174 static void ip_2_ip6_hdr(struct ip6_hdr
*ip6
, struct ip
*ip
);
177 static int udp_detach(struct socket
*so
);
178 static int udp_output(struct inpcb
*, struct mbuf
*, struct sockaddr
*,
179 struct mbuf
*, struct proc
*);
180 extern int ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
);
186 struct inpcbinfo
*pcbinfo
;
190 udbinfo
.listhead
= &udb
;
191 udbinfo
.hashbase
= hashinit(UDBHASHSIZE
, M_PCB
, &udbinfo
.hashmask
);
192 udbinfo
.porthashbase
= hashinit(UDBHASHSIZE
, M_PCB
,
193 &udbinfo
.porthashmask
);
195 str_size
= (vm_size_t
) sizeof(struct inpcb
);
196 udbinfo
.ipi_zone
= (void *) zinit(str_size
, 80000*str_size
, 8192, "udpcb");
200 * allocate lock group attribute and group for udp pcb mutexes
202 pcbinfo
->mtx_grp_attr
= lck_grp_attr_alloc_init();
203 lck_grp_attr_setdefault(pcbinfo
->mtx_grp_attr
);
205 pcbinfo
->mtx_grp
= lck_grp_alloc_init("udppcb", pcbinfo
->mtx_grp_attr
);
207 pcbinfo
->mtx_attr
= lck_attr_alloc_init();
208 lck_attr_setdefault(pcbinfo
->mtx_attr
);
210 if ((pcbinfo
->mtx
= lck_rw_alloc_init(pcbinfo
->mtx_grp
, pcbinfo
->mtx_attr
)) == NULL
)
211 return; /* pretty much dead if this fails... */
213 in_pcb_nat_init(&udbinfo
, AF_INET
, IPPROTO_UDP
, SOCK_DGRAM
);
215 udbinfo
.ipi_zone
= zinit("udpcb", sizeof(struct inpcb
), maxsockets
,
220 /* for pcb sharing testing only */
221 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
222 kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat
);
224 laddr
.s_addr
= 0x11646464;
225 faddr
.s_addr
= 0x11646465;
228 in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
229 kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat
);
231 stat
= in_pcb_rem_share_client(&udbinfo
, fake_owner
);
232 kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat
);
234 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
235 kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat
);
237 laddr
.s_addr
= 0x11646464;
238 faddr
.s_addr
= 0x11646465;
241 stat
= in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
242 kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat
);
248 register struct mbuf
*m
;
251 register struct ip
*ip
;
252 register struct udphdr
*uh
;
253 register struct inpcb
*inp
;
254 struct mbuf
*opts
= 0;
257 struct sockaddr
*append_sa
;
258 struct inpcbinfo
*pcbinfo
= &udbinfo
;
260 udpstat
.udps_ipackets
++;
262 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_START
, 0,0,0,0,0);
263 if (m
->m_pkthdr
.csum_flags
& CSUM_TCP_SUM16
)
264 m
->m_pkthdr
.csum_flags
= 0; /* invalidate hwcksum for UDP */
267 * Strip IP options, if any; should skip this,
268 * make available to user, and use on returned packets,
269 * but we don't yet have a way to check the checksum
270 * with options still present.
272 if (iphlen
> sizeof (struct ip
)) {
273 ip_stripoptions(m
, (struct mbuf
*)0);
274 iphlen
= sizeof(struct ip
);
278 * Get IP and UDP header together in first mbuf.
280 ip
= mtod(m
, struct ip
*);
281 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
)) {
282 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
))) == 0) {
283 udpstat
.udps_hdrops
++;
284 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
287 ip
= mtod(m
, struct ip
*);
289 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
291 /* destination port of 0 is illegal, based on RFC768. */
292 if (uh
->uh_dport
== 0)
295 KERNEL_DEBUG(DBG_LAYER_IN_BEG
, uh
->uh_dport
, uh
->uh_sport
,
296 ip
->ip_src
.s_addr
, ip
->ip_dst
.s_addr
, uh
->uh_ulen
);
299 * Make mbuf data length reflect UDP length.
300 * If not enough data to reflect UDP length, drop.
302 len
= ntohs((u_short
)uh
->uh_ulen
);
303 if (ip
->ip_len
!= len
) {
304 if (len
> ip
->ip_len
|| len
< sizeof(struct udphdr
)) {
305 udpstat
.udps_badlen
++;
308 m_adj(m
, len
- ip
->ip_len
);
309 /* ip->ip_len = len; */
312 * Save a copy of the IP header in case we want restore it
313 * for sending an ICMP error message in response.
318 * Checksum extended UDP header and data.
321 if (m
->m_pkthdr
.csum_flags
& CSUM_DATA_VALID
) {
322 if (m
->m_pkthdr
.csum_flags
& CSUM_PSEUDO_HDR
)
323 uh
->uh_sum
= m
->m_pkthdr
.csum_data
;
326 uh
->uh_sum
^= 0xffff;
330 *(uint32_t*)&b
[0] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0];
331 *(uint32_t*)&b
[4] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4];
332 *(uint8_t*)&b
[8] = *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8];
334 bzero(((struct ipovly
*)ip
)->ih_x1
, 9);
335 ((struct ipovly
*)ip
)->ih_len
= uh
->uh_ulen
;
336 uh
->uh_sum
= in_cksum(m
, len
+ sizeof (struct ip
));
338 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0] = *(uint32_t*)&b
[0];
339 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4] = *(uint32_t*)&b
[4];
340 *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8] = *(uint8_t*)&b
[8];
343 udpstat
.udps_badsum
++;
345 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
351 udpstat
.udps_nosum
++;
354 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
)) ||
355 in_broadcast(ip
->ip_dst
, m
->m_pkthdr
.rcvif
)) {
357 lck_rw_lock_shared(pcbinfo
->mtx
);
359 * Deliver a multicast or broadcast datagram to *all* sockets
360 * for which the local and remote addresses and ports match
361 * those of the incoming datagram. This allows more than
362 * one process to receive multi/broadcasts on the same port.
363 * (This really ought to be done for unicast datagrams as
364 * well, but that would cause problems with existing
365 * applications that open both address-specific sockets and
366 * a wildcard socket listening to the same port -- they would
367 * end up receiving duplicates of every unicast datagram.
368 * Those applications open the multiple sockets to overcome an
369 * inadequacy of the UDP socket interface, but for backwards
370 * compatibility we avoid the problem here rather than
371 * fixing the interface. Maybe 4.5BSD will remedy this?)
376 * Construct sockaddr format source address.
378 udp_in
.sin_port
= uh
->uh_sport
;
379 udp_in
.sin_addr
= ip
->ip_src
;
381 * Locate pcb(s) for datagram.
382 * (Algorithm copied from raw_intr().)
386 udp_in6
.uin6_init_done
= udp_ip6
.uip6_init_done
= 0;
388 LIST_FOREACH(inp
, &udb
, inp_list
) {
390 /* Ignore nat/SharedIP dummy pcbs */
391 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
394 if (inp
->inp_socket
== NULL
)
396 if (inp
!= sotoinpcb(inp
->inp_socket
))
397 panic("udp_input: bad so back ptr inp=%x\n", inp
);
399 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
402 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) == WNT_STOPUSING
) {
406 udp_lock(inp
->inp_socket
, 1, 0);
408 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
409 udp_unlock(inp
->inp_socket
, 1, 0);
413 if (inp
->inp_lport
!= uh
->uh_dport
) {
414 udp_unlock(inp
->inp_socket
, 1, 0);
417 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
) {
418 if (inp
->inp_laddr
.s_addr
!=
420 udp_unlock(inp
->inp_socket
, 1, 0);
424 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
) {
425 if (inp
->inp_faddr
.s_addr
!=
427 inp
->inp_fport
!= uh
->uh_sport
) {
428 udp_unlock(inp
->inp_socket
, 1, 0);
437 /* check AH/ESP integrity. */
438 if (ipsec_bypass
== 0) {
439 lck_mtx_lock(sadb_mutex
);
440 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
441 ipsecstat
.in_polvio
++;
442 /* do not inject data to pcb */
445 lck_mtx_unlock(sadb_mutex
);
449 if ((n
= m_copy(m
, 0, M_COPYALL
)) != NULL
) {
450 udp_append(last
, ip
, n
,
452 sizeof(struct udphdr
));
454 udp_unlock(last
->inp_socket
, 1, 0);
458 * Don't look for additional matches if this one does
459 * not have either the SO_REUSEPORT or SO_REUSEADDR
460 * socket options set. This heuristic avoids searching
461 * through all pcbs in the common case of a non-shared
462 * port. It * assumes that an application will never
463 * clear these options after setting them.
465 if ((last
->inp_socket
->so_options
&(SO_REUSEPORT
|SO_REUSEADDR
)) == 0)
468 lck_rw_done(pcbinfo
->mtx
);
472 * No matching pcb found; discard datagram.
473 * (No need to send an ICMP Port Unreachable
474 * for a broadcast or multicast datgram.)
476 udpstat
.udps_noportbcast
++;
480 /* check AH/ESP integrity. */
481 if (ipsec_bypass
== 0 && m
) {
482 lck_mtx_lock(sadb_mutex
);
483 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
484 ipsecstat
.in_polvio
++;
485 lck_mtx_unlock(sadb_mutex
);
486 udp_unlock(last
->inp_socket
, 1, 0);
489 lck_mtx_unlock(sadb_mutex
);
492 udp_append(last
, ip
, m
, iphlen
+ sizeof(struct udphdr
));
493 udp_unlock(last
->inp_socket
, 1, 0);
499 * UDP to port 4500 with a payload where the first four bytes are
500 * not zero is a UDP encapsulated IPSec packet. Packets where
501 * the payload is one byte and that byte is 0xFF are NAT keepalive
502 * packets. Decapsulate the ESP packet and carry on with IPSec input
503 * or discard the NAT keep-alive.
505 if (ipsec_bypass
== 0 && (esp_udp_encap_port
& 0xFFFF) != 0 &&
506 uh
->uh_dport
== ntohs((u_short
)esp_udp_encap_port
)) {
507 int payload_len
= len
- sizeof(struct udphdr
) > 4 ? 4 : len
- sizeof(struct udphdr
);
508 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
) + payload_len
) {
509 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
) + payload_len
)) == 0) {
510 udpstat
.udps_hdrops
++;
511 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
514 ip
= mtod(m
, struct ip
*);
515 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
517 /* Check for NAT keepalive packet */
518 if (payload_len
== 1 && *(u_int8_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) == 0xFF) {
520 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
523 else if (payload_len
== 4 && *(u_int32_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) != 0) {
524 /* UDP encapsulated IPSec packet to pass through NAT */
527 stripsiz
= sizeof(struct udphdr
);
529 ip
= mtod(m
, struct ip
*);
530 ovbcopy((caddr_t
)ip
, (caddr_t
)(((u_char
*)ip
) + stripsiz
), iphlen
);
531 m
->m_data
+= stripsiz
;
532 m
->m_len
-= stripsiz
;
533 m
->m_pkthdr
.len
-= stripsiz
;
534 ip
= mtod(m
, struct ip
*);
535 ip
->ip_len
= ip
->ip_len
- stripsiz
;
536 ip
->ip_p
= IPPROTO_ESP
;
538 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
539 esp4_input(m
, iphlen
);
546 * Locate pcb for datagram.
548 inp
= in_pcblookup_hash(&udbinfo
, ip
->ip_src
, uh
->uh_sport
,
549 ip
->ip_dst
, uh
->uh_dport
, 1, m
->m_pkthdr
.rcvif
);
552 char buf
[MAX_IPv4_STR_LEN
];
553 char buf2
[MAX_IPv4_STR_LEN
];
555 /* check src and dst address */
556 if (log_in_vain
!= 3)
558 "Connection attempt to UDP %s:%d from %s:%d\n",
559 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
561 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
562 ntohs(uh
->uh_sport
));
563 else if (!(m
->m_flags
& (M_BCAST
| M_MCAST
)) &&
564 ip
->ip_dst
.s_addr
!= ip
->ip_src
.s_addr
)
565 log_in_vain_log((LOG_INFO
,
566 "Stealth Mode connection attempt to UDP %s:%d from %s:%d\n",
567 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
569 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
570 ntohs(uh
->uh_sport
)))
572 udpstat
.udps_noport
++;
573 if (m
->m_flags
& (M_BCAST
| M_MCAST
)) {
574 udpstat
.udps_noportbcast
++;
578 if (badport_bandlim(BANDLIM_ICMP_UNREACH
) < 0)
582 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
!= IFT_LOOP
)
585 ip
->ip_len
+= iphlen
;
586 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_PORT
, 0, 0);
587 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
590 udp_lock(inp
->inp_socket
, 1, 0);
592 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
593 udp_unlock(inp
->inp_socket
, 1, 0);
597 if (ipsec_bypass
== 0 && inp
!= NULL
) {
598 lck_mtx_lock(sadb_mutex
);
599 if (ipsec4_in_reject_so(m
, inp
->inp_socket
)) {
600 ipsecstat
.in_polvio
++;
601 lck_mtx_unlock(sadb_mutex
);
602 udp_unlock(inp
->inp_socket
, 1, 0);
605 lck_mtx_unlock(sadb_mutex
);
610 * Construct sockaddr format source address.
611 * Stuff source address and datagram in user buffer.
613 udp_in
.sin_port
= uh
->uh_sport
;
614 udp_in
.sin_addr
= ip
->ip_src
;
615 if (inp
->inp_flags
& INP_CONTROLOPTS
616 || inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
618 if (inp
->inp_vflag
& INP_IPV6
) {
621 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
622 savedflags
= inp
->inp_flags
;
623 inp
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
624 ip6_savecontrol(inp
, &opts
, &udp_ip6
.uip6_ip6
, m
);
625 inp
->inp_flags
= savedflags
;
628 ip_savecontrol(inp
, &opts
, ip
, m
);
630 m_adj(m
, iphlen
+ sizeof(struct udphdr
));
632 KERNEL_DEBUG(DBG_LAYER_IN_END
, uh
->uh_dport
, uh
->uh_sport
,
633 save_ip
.ip_src
.s_addr
, save_ip
.ip_dst
.s_addr
, uh
->uh_ulen
);
636 if (inp
->inp_vflag
& INP_IPV6
) {
637 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
638 append_sa
= (struct sockaddr
*)&udp_in6
;
641 append_sa
= (struct sockaddr
*)&udp_in
;
642 if (sbappendaddr(&inp
->inp_socket
->so_rcv
, append_sa
, m
, opts
, NULL
) == 0) {
643 udpstat
.udps_fullsock
++;
646 sorwakeup(inp
->inp_socket
);
648 udp_unlock(inp
->inp_socket
, 1, 0);
649 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
655 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
661 ip_2_ip6_hdr(ip6
, ip
)
665 bzero(ip6
, sizeof(*ip6
));
667 ip6
->ip6_vfc
= IPV6_VERSION
;
668 ip6
->ip6_plen
= ip
->ip_len
;
669 ip6
->ip6_nxt
= ip
->ip_p
;
670 ip6
->ip6_hlim
= ip
->ip_ttl
;
671 ip6
->ip6_src
.s6_addr32
[2] = ip6
->ip6_dst
.s6_addr32
[2] =
673 ip6
->ip6_src
.s6_addr32
[3] = ip
->ip_src
.s_addr
;
674 ip6
->ip6_dst
.s6_addr32
[3] = ip
->ip_dst
.s_addr
;
679 * subroutine of udp_input(), mainly for source code readability.
680 * caller must properly init udp_ip6 and udp_in6 beforehand.
683 udp_append(last
, ip
, n
, off
)
689 struct sockaddr
*append_sa
;
690 struct mbuf
*opts
= 0;
692 if (last
->inp_flags
& INP_CONTROLOPTS
||
693 last
->inp_socket
->so_options
& SO_TIMESTAMP
) {
695 if (last
->inp_vflag
& INP_IPV6
) {
698 if (udp_ip6
.uip6_init_done
== 0) {
699 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
700 udp_ip6
.uip6_init_done
= 1;
702 savedflags
= last
->inp_flags
;
703 last
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
704 ip6_savecontrol(last
, &opts
, &udp_ip6
.uip6_ip6
, n
);
705 last
->inp_flags
= savedflags
;
708 ip_savecontrol(last
, &opts
, ip
, n
);
711 if (last
->inp_vflag
& INP_IPV6
) {
712 if (udp_in6
.uin6_init_done
== 0) {
713 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
714 udp_in6
.uin6_init_done
= 1;
716 append_sa
= (struct sockaddr
*)&udp_in6
.uin6_sin
;
719 append_sa
= (struct sockaddr
*)&udp_in
;
721 if (sbappendaddr(&last
->inp_socket
->so_rcv
, append_sa
, n
, opts
, NULL
) == 0) {
722 udpstat
.udps_fullsock
++;
724 sorwakeup(last
->inp_socket
);
728 * Notify a udp user of an asynchronous error;
729 * just wake up so that he can collect error status.
732 udp_notify(inp
, errno
)
733 register struct inpcb
*inp
;
736 inp
->inp_socket
->so_error
= errno
;
737 sorwakeup(inp
->inp_socket
);
738 sowwakeup(inp
->inp_socket
);
742 udp_ctlinput(cmd
, sa
, vip
)
749 void (*notify
)(struct inpcb
*, int) = udp_notify
;
750 struct in_addr faddr
;
753 faddr
= ((struct sockaddr_in
*)sa
)->sin_addr
;
754 if (sa
->sa_family
!= AF_INET
|| faddr
.s_addr
== INADDR_ANY
)
757 if (PRC_IS_REDIRECT(cmd
)) {
759 notify
= in_rtchange
;
760 } else if (cmd
== PRC_HOSTDEAD
)
762 else if ((unsigned)cmd
>= PRC_NCMDS
|| inetctlerrmap
[cmd
] == 0)
765 uh
= (struct udphdr
*)((caddr_t
)ip
+ (ip
->ip_hl
<< 2));
766 inp
= in_pcblookup_hash(&udbinfo
, faddr
, uh
->uh_dport
,
767 ip
->ip_src
, uh
->uh_sport
, 0, NULL
);
768 if (inp
!= NULL
&& inp
->inp_socket
!= NULL
) {
769 udp_lock(inp
->inp_socket
, 1, 0);
770 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
771 udp_unlock(inp
->inp_socket
, 1, 0);
774 (*notify
)(inp
, inetctlerrmap
[cmd
]);
775 udp_unlock(inp
->inp_socket
, 1, 0);
778 in_pcbnotifyall(&udbinfo
, faddr
, inetctlerrmap
[cmd
], notify
);
782 udp_pcblist SYSCTL_HANDLER_ARGS
785 struct inpcb
*inp
, **inp_list
;
790 * The process of preparing the TCB list is too time-consuming and
791 * resource-intensive to repeat twice on every request.
793 lck_rw_lock_exclusive(udbinfo
.mtx
);
794 if (req
->oldptr
== USER_ADDR_NULL
) {
795 n
= udbinfo
.ipi_count
;
796 req
->oldidx
= 2 * (sizeof xig
)
797 + (n
+ n
/8) * sizeof(struct xinpcb
);
798 lck_rw_done(udbinfo
.mtx
);
802 if (req
->newptr
!= USER_ADDR_NULL
) {
803 lck_rw_done(udbinfo
.mtx
);
808 * OK, now we're committed to doing something.
810 gencnt
= udbinfo
.ipi_gencnt
;
811 n
= udbinfo
.ipi_count
;
813 bzero(&xig
, sizeof(xig
));
814 xig
.xig_len
= sizeof xig
;
816 xig
.xig_gen
= gencnt
;
817 xig
.xig_sogen
= so_gencnt
;
818 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
820 lck_rw_done(udbinfo
.mtx
);
824 * We are done if there is no pcb
827 lck_rw_done(udbinfo
.mtx
);
831 inp_list
= _MALLOC(n
* sizeof *inp_list
, M_TEMP
, M_WAITOK
);
833 lck_rw_done(udbinfo
.mtx
);
837 for (inp
= LIST_FIRST(udbinfo
.listhead
), i
= 0; inp
&& i
< n
;
838 inp
= LIST_NEXT(inp
, inp_list
)) {
839 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
)
845 for (i
= 0; i
< n
; i
++) {
847 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
) {
850 bzero(&xi
, sizeof(xi
));
851 xi
.xi_len
= sizeof xi
;
852 /* XXX should avoid extra copy */
853 inpcb_to_compat(inp
, &xi
.xi_inp
);
855 sotoxsocket(inp
->inp_socket
, &xi
.xi_socket
);
856 error
= SYSCTL_OUT(req
, &xi
, sizeof xi
);
861 * Give the user an updated idea of our state.
862 * If the generation differs from what we told
863 * her before, she knows that something happened
864 * while we were processing this request, and it
865 * might be necessary to retry.
867 bzero(&xig
, sizeof(xig
));
868 xig
.xig_len
= sizeof xig
;
869 xig
.xig_gen
= udbinfo
.ipi_gencnt
;
870 xig
.xig_sogen
= so_gencnt
;
871 xig
.xig_count
= udbinfo
.ipi_count
;
872 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
874 FREE(inp_list
, M_TEMP
);
875 lck_rw_done(udbinfo
.mtx
);
879 SYSCTL_PROC(_net_inet_udp
, UDPCTL_PCBLIST
, pcblist
, CTLFLAG_RD
, 0, 0,
880 udp_pcblist
, "S,xinpcb", "List of active UDP sockets");
884 static __inline__ u_int16_t
885 get_socket_id(struct socket
* s
)
892 val
= (u_int16_t
)(((u_int32_t
)s
) / sizeof(struct socket
));
900 udp_output(inp
, m
, addr
, control
, p
)
901 register struct inpcb
*inp
;
903 struct sockaddr
*addr
;
904 struct mbuf
*control
;
907 register struct udpiphdr
*ui
;
908 register int len
= m
->m_pkthdr
.len
;
909 struct sockaddr_in
*sin
, src
;
910 struct in_addr origladdr
, laddr
, faddr
;
911 u_short lport
, fport
;
912 struct sockaddr_in
*ifaddr
;
913 int error
= 0, udp_dodisconnect
= 0;
916 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_START
, 0,0,0,0,0);
919 m_freem(control
); /* XXX */
921 KERNEL_DEBUG(DBG_LAYER_OUT_BEG
, inp
->inp_fport
, inp
->inp_lport
,
922 inp
->inp_laddr
.s_addr
, inp
->inp_faddr
.s_addr
,
923 (htons((u_short
)len
+ sizeof (struct udphdr
))));
925 if (len
+ sizeof(struct udpiphdr
) > IP_MAXPACKET
) {
930 /* If there was a routing change, discard cached route and check
931 * that we have a valid source address.
932 * Reacquire a new source address if INADDR_ANY was specified
936 lck_mtx_assert(inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
939 if (inp
->inp_route
.ro_rt
&& inp
->inp_route
.ro_rt
->generation_id
!= route_generation
) {
940 if (ifa_foraddr(inp
->inp_laddr
.s_addr
) == 0) { /* src address is gone */
941 if (inp
->inp_flags
& INP_INADDR_ANY
)
942 inp
->inp_faddr
.s_addr
= INADDR_ANY
; /* new src will be set later */
944 error
= EADDRNOTAVAIL
;
948 rtfree(inp
->inp_route
.ro_rt
);
949 inp
->inp_route
.ro_rt
= (struct rtentry
*)0;
952 origladdr
= laddr
= inp
->inp_laddr
;
953 faddr
= inp
->inp_faddr
;
954 lport
= inp
->inp_lport
;
955 fport
= inp
->inp_fport
;
958 sin
= (struct sockaddr_in
*)addr
;
959 if (faddr
.s_addr
!= INADDR_ANY
) {
965 * In case we don't have a local port set, go through the full connect.
966 * We don't have a local port yet (ie, we can't be looked up),
967 * so it's not an issue if the input runs at the same time we do this.
969 error
= in_pcbconnect(inp
, addr
, p
);
973 laddr
= inp
->inp_laddr
;
974 lport
= inp
->inp_lport
;
975 faddr
= inp
->inp_faddr
;
976 fport
= inp
->inp_fport
;
977 udp_dodisconnect
= 1;
981 * we have a full address and a local port.
982 * use those info to build the packet without changing the pcb
983 * and interfering with the input path. See 3851370
985 if (laddr
.s_addr
== INADDR_ANY
) {
986 if ((error
= in_pcbladdr(inp
, addr
, &ifaddr
)) != 0)
988 laddr
= ifaddr
->sin_addr
;
989 inp
->inp_flags
|= INP_INADDR_ANY
; /* from pcbconnect: remember we don't care about src addr.*/
992 faddr
= sin
->sin_addr
;
993 fport
= sin
->sin_port
;
996 if (faddr
.s_addr
== INADDR_ANY
) {
1004 * Calculate data length and get a mbuf
1005 * for UDP and IP headers.
1007 M_PREPEND(m
, sizeof(struct udpiphdr
), M_DONTWAIT
);
1014 * Fill in mbuf with extended UDP header
1015 * and addresses and length put into network format.
1017 ui
= mtod(m
, struct udpiphdr
*);
1018 bzero(ui
->ui_x1
, sizeof(ui
->ui_x1
)); /* XXX still needed? */
1019 ui
->ui_pr
= IPPROTO_UDP
;
1022 ui
->ui_sport
= lport
;
1023 ui
->ui_dport
= fport
;
1024 ui
->ui_ulen
= htons((u_short
)len
+ sizeof(struct udphdr
));
1027 * Set up checksum and output datagram.
1030 ui
->ui_sum
= in_pseudo(ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
,
1031 htons((u_short
)len
+ sizeof(struct udphdr
) + IPPROTO_UDP
));
1032 m
->m_pkthdr
.csum_flags
= CSUM_UDP
;
1033 m
->m_pkthdr
.csum_data
= offsetof(struct udphdr
, uh_sum
);
1037 ((struct ip
*)ui
)->ip_len
= sizeof (struct udpiphdr
) + len
;
1038 ((struct ip
*)ui
)->ip_ttl
= inp
->inp_ip_ttl
; /* XXX */
1039 ((struct ip
*)ui
)->ip_tos
= inp
->inp_ip_tos
; /* XXX */
1040 udpstat
.udps_opackets
++;
1042 KERNEL_DEBUG(DBG_LAYER_OUT_END
, ui
->ui_dport
, ui
->ui_sport
,
1043 ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
, ui
->ui_ulen
);
1046 if (ipsec_bypass
== 0 && ipsec_setsocket(m
, inp
->inp_socket
) != 0) {
1051 m
->m_pkthdr
.socket_id
= get_socket_id(inp
->inp_socket
);
1052 error
= ip_output_list(m
, 0, inp
->inp_options
, &inp
->inp_route
,
1053 (inp
->inp_socket
->so_options
& (SO_DONTROUTE
| SO_BROADCAST
)),
1056 if (udp_dodisconnect
) {
1057 in_pcbdisconnect(inp
);
1058 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1060 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1064 if (udp_dodisconnect
) {
1065 in_pcbdisconnect(inp
);
1066 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1071 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1075 u_long udp_sendspace
= 9216; /* really max datagram size */
1076 /* 40 1K datagrams */
1077 SYSCTL_INT(_net_inet_udp
, UDPCTL_MAXDGRAM
, maxdgram
, CTLFLAG_RW
,
1078 &udp_sendspace
, 0, "Maximum outgoing UDP datagram size");
1080 u_long udp_recvspace
= 40 * (1024 +
1082 sizeof(struct sockaddr_in6
)
1084 sizeof(struct sockaddr_in
)
1087 SYSCTL_INT(_net_inet_udp
, UDPCTL_RECVSPACE
, recvspace
, CTLFLAG_RW
,
1088 &udp_recvspace
, 0, "Maximum incoming UDP datagram size");
1091 udp_abort(struct socket
*so
)
1095 inp
= sotoinpcb(so
);
1097 panic("udp_abort: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1098 soisdisconnected(so
);
1104 udp_attach(struct socket
*so
, int proto
, struct proc
*p
)
1109 inp
= sotoinpcb(so
);
1111 panic ("udp_attach so=%x inp=%x\n", so
, inp
);
1113 error
= in_pcballoc(so
, &udbinfo
, p
);
1116 error
= soreserve(so
, udp_sendspace
, udp_recvspace
);
1119 inp
= (struct inpcb
*)so
->so_pcb
;
1120 inp
->inp_vflag
|= INP_IPV4
;
1121 inp
->inp_ip_ttl
= ip_defttl
;
1126 udp_bind(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1131 inp
= sotoinpcb(so
);
1134 error
= in_pcbbind(inp
, nam
, p
);
1139 udp_connect(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1144 inp
= sotoinpcb(so
);
1147 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
)
1149 error
= in_pcbconnect(inp
, nam
, p
);
1156 udp_detach(struct socket
*so
)
1160 inp
= sotoinpcb(so
);
1162 panic("udp_detach: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1164 inp
->inp_state
= INPCB_STATE_DEAD
;
1169 udp_disconnect(struct socket
*so
)
1173 inp
= sotoinpcb(so
);
1176 if (inp
->inp_faddr
.s_addr
== INADDR_ANY
)
1179 in_pcbdisconnect(inp
);
1180 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
1181 so
->so_state
&= ~SS_ISCONNECTED
; /* XXX */
1186 udp_send(struct socket
*so
, int flags
, struct mbuf
*m
, struct sockaddr
*addr
,
1187 struct mbuf
*control
, struct proc
*p
)
1191 inp
= sotoinpcb(so
);
1196 return udp_output(inp
, m
, addr
, control
, p
);
1200 udp_shutdown(struct socket
*so
)
1204 inp
= sotoinpcb(so
);
1211 struct pr_usrreqs udp_usrreqs
= {
1212 udp_abort
, pru_accept_notsupp
, udp_attach
, udp_bind
, udp_connect
,
1213 pru_connect2_notsupp
, in_control
, udp_detach
, udp_disconnect
,
1214 pru_listen_notsupp
, in_setpeeraddr
, pru_rcvd_notsupp
,
1215 pru_rcvoob_notsupp
, udp_send
, pru_sense_null
, udp_shutdown
,
1216 in_setsockaddr
, sosend
, soreceive
, pru_sopoll_notsupp
1221 udp_lock(so
, refcount
, debug
)
1223 int refcount
, debug
;
1228 __asm__
volatile("mflr %0" : "=r" (lr_saved
));
1230 else lr_saved
= debug
;
1234 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_NOTOWNED
);
1235 lck_mtx_lock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1238 panic("udp_lock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1239 lck_mtx_assert(so
->so_proto
->pr_domain
->dom_mtx
, LCK_MTX_ASSERT_NOTOWNED
);
1240 lck_mtx_lock(so
->so_proto
->pr_domain
->dom_mtx
);
1246 so
->reserved3
= lr_saved
;
1251 udp_unlock(so
, refcount
, debug
)
1257 struct inpcb
*inp
= sotoinpcb(so
);
1258 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1261 __asm__
volatile("mflr %0" : "=r" (lr_saved
));
1263 else lr_saved
= debug
;
1268 if (so
->so_usecount
== 0 && (inp
->inp_wantcnt
== WNT_STOPUSING
)) {
1269 if (lck_rw_try_lock_exclusive(pcbinfo
->mtx
)) {
1271 lck_rw_done(pcbinfo
->mtx
);
1277 if (so
->so_pcb
== NULL
) {
1278 panic("udp_unlock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1279 lck_mtx_assert(so
->so_proto
->pr_domain
->dom_mtx
, LCK_MTX_ASSERT_OWNED
);
1280 lck_mtx_unlock(so
->so_proto
->pr_domain
->dom_mtx
);
1283 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1284 lck_mtx_unlock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1288 so
->reserved4
= lr_saved
;
1293 udp_getlock(so
, locktype
)
1297 struct inpcb
*inp
= sotoinpcb(so
);
1301 return(inp
->inpcb_mtx
);
1303 panic("udp_getlock: so=%x NULL so_pcb\n", so
);
1304 return (so
->so_proto
->pr_domain
->dom_mtx
);
1311 struct inpcb
*inp
, *inpnxt
;
1313 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1315 lck_rw_lock_exclusive(pcbinfo
->mtx
);
1317 for (inp
= udb
.lh_first
; inp
!= NULL
; inp
= inpnxt
) {
1318 inpnxt
= inp
->inp_list
.le_next
;
1320 /* Ignore nat/SharedIP dummy pcbs */
1321 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
1324 if (inp
->inp_wantcnt
!= WNT_STOPUSING
)
1327 so
= inp
->inp_socket
;
1328 if (!lck_mtx_try_lock(inp
->inpcb_mtx
)) /* skip if busy, no hurry for cleanup... */
1331 if (so
->so_usecount
== 0)
1334 lck_mtx_unlock(inp
->inpcb_mtx
);
1336 lck_rw_done(pcbinfo
->mtx
);
1340 ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
)
1342 if ( dstaddr
== srcaddr
){