2 * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
4 * @APPLE_OSREFERENCE_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. 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.
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
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
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
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.
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
29 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
30 * The Regents of the University of California. All rights reserved.
32 * Redistribution and use in source and binary forms, with or without
33 * modification, are permitted provided that the following conditions
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.
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
60 * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
61 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.13 2001/08/08 18:59:54 ghelmer Exp $
64 #include <sys/param.h>
65 #include <sys/systm.h>
66 #include <sys/kernel.h>
67 #include <sys/malloc.h>
69 #include <sys/domain.h>
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>
77 #include <net/if_types.h>
78 #include <net/route.h>
80 #include <netinet/in.h>
81 #include <netinet/in_systm.h>
82 #include <netinet/ip.h>
84 #include <netinet/ip6.h>
86 #include <netinet/in_pcb.h>
87 #include <netinet/in_var.h>
88 #include <netinet/ip_var.h>
90 #include <netinet6/ip6_var.h>
92 #include <netinet/ip_icmp.h>
93 #include <netinet/icmp_var.h>
94 #include <netinet/udp.h>
95 #include <netinet/udp_var.h>
96 #include <sys/kdebug.h>
99 #include <netinet6/ipsec.h>
100 extern int ipsec_bypass
;
101 extern lck_mtx_t
*sadb_mutex
;
105 #define DBG_LAYER_IN_BEG NETDBG_CODE(DBG_NETUDP, 0)
106 #define DBG_LAYER_IN_END NETDBG_CODE(DBG_NETUDP, 2)
107 #define DBG_LAYER_OUT_BEG NETDBG_CODE(DBG_NETUDP, 1)
108 #define DBG_LAYER_OUT_END NETDBG_CODE(DBG_NETUDP, 3)
109 #define DBG_FNC_UDP_INPUT NETDBG_CODE(DBG_NETUDP, (5 << 8))
110 #define DBG_FNC_UDP_OUTPUT NETDBG_CODE(DBG_NETUDP, (6 << 8) | 1)
113 * UDP protocol implementation.
114 * Per RFC 768, August, 1980.
117 static int udpcksum
= 1;
119 static int udpcksum
= 0; /* XXX */
121 SYSCTL_INT(_net_inet_udp
, UDPCTL_CHECKSUM
, checksum
, CTLFLAG_RW
,
125 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, log_in_vain
, CTLFLAG_RW
,
126 &log_in_vain
, 0, "Log all incoming UDP packets");
128 static int blackhole
= 0;
129 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, blackhole
, CTLFLAG_RW
,
130 &blackhole
, 0, "Do not send port unreachables for refused connects");
132 struct inpcbhead udb
; /* from udp_var.h */
133 #define udb6 udb /* for KAME src sync over BSD*'s */
134 struct inpcbinfo udbinfo
;
137 #define UDBHASHSIZE 16
140 extern int apple_hwcksum_rx
;
141 extern int esp_udp_encap_port
;
142 extern u_long route_generation
;
144 extern void ipfwsyslog( int level
, char *format
,...);
146 extern int fw_verbose
;
148 #define log_in_vain_log( a ) { \
149 if ( (log_in_vain == 3 ) && (fw_verbose == 2)) { /* Apple logging, log to ipfw.log */ \
155 struct udpstat udpstat
; /* from udp_var.h */
156 SYSCTL_STRUCT(_net_inet_udp
, UDPCTL_STATS
, stats
, CTLFLAG_RD
,
157 &udpstat
, udpstat
, "UDP statistics (struct udpstat, netinet/udp_var.h)");
158 SYSCTL_INT(_net_inet_udp
, OID_AUTO
, pcbcount
, CTLFLAG_RD
,
159 &udbinfo
.ipi_count
, 0, "Number of active PCBs");
161 static struct sockaddr_in udp_in
= { sizeof(udp_in
), AF_INET
};
164 struct sockaddr_in6 uin6_sin
;
165 u_char uin6_init_done
: 1;
167 { sizeof(udp_in6
.uin6_sin
), AF_INET6
},
171 struct ip6_hdr uip6_ip6
;
172 u_char uip6_init_done
: 1;
176 static void udp_append(struct inpcb
*last
, struct ip
*ip
,
177 struct mbuf
*n
, int off
);
179 static void ip_2_ip6_hdr(struct ip6_hdr
*ip6
, struct ip
*ip
);
182 static int udp_detach(struct socket
*so
);
183 static int udp_output(struct inpcb
*, struct mbuf
*, struct sockaddr
*,
184 struct mbuf
*, struct proc
*);
185 extern int ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
);
191 struct inpcbinfo
*pcbinfo
;
195 udbinfo
.listhead
= &udb
;
196 udbinfo
.hashbase
= hashinit(UDBHASHSIZE
, M_PCB
, &udbinfo
.hashmask
);
197 udbinfo
.porthashbase
= hashinit(UDBHASHSIZE
, M_PCB
,
198 &udbinfo
.porthashmask
);
200 str_size
= (vm_size_t
) sizeof(struct inpcb
);
201 udbinfo
.ipi_zone
= (void *) zinit(str_size
, 80000*str_size
, 8192, "udpcb");
205 * allocate lock group attribute and group for udp pcb mutexes
207 pcbinfo
->mtx_grp_attr
= lck_grp_attr_alloc_init();
208 lck_grp_attr_setdefault(pcbinfo
->mtx_grp_attr
);
210 pcbinfo
->mtx_grp
= lck_grp_alloc_init("udppcb", pcbinfo
->mtx_grp_attr
);
212 pcbinfo
->mtx_attr
= lck_attr_alloc_init();
213 lck_attr_setdefault(pcbinfo
->mtx_attr
);
215 if ((pcbinfo
->mtx
= lck_rw_alloc_init(pcbinfo
->mtx_grp
, pcbinfo
->mtx_attr
)) == NULL
)
216 return; /* pretty much dead if this fails... */
218 in_pcb_nat_init(&udbinfo
, AF_INET
, IPPROTO_UDP
, SOCK_DGRAM
);
220 udbinfo
.ipi_zone
= zinit("udpcb", sizeof(struct inpcb
), maxsockets
,
225 /* for pcb sharing testing only */
226 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
227 kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat
);
229 laddr
.s_addr
= 0x11646464;
230 faddr
.s_addr
= 0x11646465;
233 in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
234 kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat
);
236 stat
= in_pcb_rem_share_client(&udbinfo
, fake_owner
);
237 kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat
);
239 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
240 kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat
);
242 laddr
.s_addr
= 0x11646464;
243 faddr
.s_addr
= 0x11646465;
246 stat
= in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
247 kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat
);
253 register struct mbuf
*m
;
256 register struct ip
*ip
;
257 register struct udphdr
*uh
;
258 register struct inpcb
*inp
;
259 struct mbuf
*opts
= 0;
262 struct sockaddr
*append_sa
;
263 struct inpcbinfo
*pcbinfo
= &udbinfo
;
265 udpstat
.udps_ipackets
++;
267 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_START
, 0,0,0,0,0);
268 if (m
->m_pkthdr
.csum_flags
& CSUM_TCP_SUM16
)
269 m
->m_pkthdr
.csum_flags
= 0; /* invalidate hwcksum for UDP */
272 * Strip IP options, if any; should skip this,
273 * make available to user, and use on returned packets,
274 * but we don't yet have a way to check the checksum
275 * with options still present.
277 if (iphlen
> sizeof (struct ip
)) {
278 ip_stripoptions(m
, (struct mbuf
*)0);
279 iphlen
= sizeof(struct ip
);
283 * Get IP and UDP header together in first mbuf.
285 ip
= mtod(m
, struct ip
*);
286 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
)) {
287 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
))) == 0) {
288 udpstat
.udps_hdrops
++;
289 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
292 ip
= mtod(m
, struct ip
*);
294 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
296 /* destination port of 0 is illegal, based on RFC768. */
297 if (uh
->uh_dport
== 0)
300 KERNEL_DEBUG(DBG_LAYER_IN_BEG
, uh
->uh_dport
, uh
->uh_sport
,
301 ip
->ip_src
.s_addr
, ip
->ip_dst
.s_addr
, uh
->uh_ulen
);
304 * Make mbuf data length reflect UDP length.
305 * If not enough data to reflect UDP length, drop.
307 len
= ntohs((u_short
)uh
->uh_ulen
);
308 if (ip
->ip_len
!= len
) {
309 if (len
> ip
->ip_len
|| len
< sizeof(struct udphdr
)) {
310 udpstat
.udps_badlen
++;
313 m_adj(m
, len
- ip
->ip_len
);
314 /* ip->ip_len = len; */
317 * Save a copy of the IP header in case we want restore it
318 * for sending an ICMP error message in response.
323 * Checksum extended UDP header and data.
326 if (m
->m_pkthdr
.csum_flags
& CSUM_DATA_VALID
) {
327 if (m
->m_pkthdr
.csum_flags
& CSUM_PSEUDO_HDR
)
328 uh
->uh_sum
= m
->m_pkthdr
.csum_data
;
331 uh
->uh_sum
^= 0xffff;
335 *(uint32_t*)&b
[0] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0];
336 *(uint32_t*)&b
[4] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4];
337 *(uint8_t*)&b
[8] = *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8];
339 bzero(((struct ipovly
*)ip
)->ih_x1
, 9);
340 ((struct ipovly
*)ip
)->ih_len
= uh
->uh_ulen
;
341 uh
->uh_sum
= in_cksum(m
, len
+ sizeof (struct ip
));
343 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0] = *(uint32_t*)&b
[0];
344 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4] = *(uint32_t*)&b
[4];
345 *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8] = *(uint8_t*)&b
[8];
348 udpstat
.udps_badsum
++;
350 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
356 udpstat
.udps_nosum
++;
359 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
)) ||
360 in_broadcast(ip
->ip_dst
, m
->m_pkthdr
.rcvif
)) {
362 lck_rw_lock_shared(pcbinfo
->mtx
);
364 * Deliver a multicast or broadcast datagram to *all* sockets
365 * for which the local and remote addresses and ports match
366 * those of the incoming datagram. This allows more than
367 * one process to receive multi/broadcasts on the same port.
368 * (This really ought to be done for unicast datagrams as
369 * well, but that would cause problems with existing
370 * applications that open both address-specific sockets and
371 * a wildcard socket listening to the same port -- they would
372 * end up receiving duplicates of every unicast datagram.
373 * Those applications open the multiple sockets to overcome an
374 * inadequacy of the UDP socket interface, but for backwards
375 * compatibility we avoid the problem here rather than
376 * fixing the interface. Maybe 4.5BSD will remedy this?)
381 * Construct sockaddr format source address.
383 udp_in
.sin_port
= uh
->uh_sport
;
384 udp_in
.sin_addr
= ip
->ip_src
;
386 * Locate pcb(s) for datagram.
387 * (Algorithm copied from raw_intr().)
391 udp_in6
.uin6_init_done
= udp_ip6
.uip6_init_done
= 0;
393 LIST_FOREACH(inp
, &udb
, inp_list
) {
395 /* Ignore nat/SharedIP dummy pcbs */
396 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
399 if (inp
->inp_socket
== NULL
)
401 if (inp
!= sotoinpcb(inp
->inp_socket
))
402 panic("udp_input: bad so back ptr inp=%x\n", inp
);
404 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
407 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) == WNT_STOPUSING
) {
411 udp_lock(inp
->inp_socket
, 1, 0);
413 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
414 udp_unlock(inp
->inp_socket
, 1, 0);
418 if (inp
->inp_lport
!= uh
->uh_dport
) {
419 udp_unlock(inp
->inp_socket
, 1, 0);
422 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
) {
423 if (inp
->inp_laddr
.s_addr
!=
425 udp_unlock(inp
->inp_socket
, 1, 0);
429 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
) {
430 if (inp
->inp_faddr
.s_addr
!=
432 inp
->inp_fport
!= uh
->uh_sport
) {
433 udp_unlock(inp
->inp_socket
, 1, 0);
442 /* check AH/ESP integrity. */
443 if (ipsec_bypass
== 0) {
444 lck_mtx_lock(sadb_mutex
);
445 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
446 ipsecstat
.in_polvio
++;
447 /* do not inject data to pcb */
450 lck_mtx_unlock(sadb_mutex
);
454 if ((n
= m_copy(m
, 0, M_COPYALL
)) != NULL
) {
455 udp_append(last
, ip
, n
,
457 sizeof(struct udphdr
));
459 udp_unlock(last
->inp_socket
, 1, 0);
463 * Don't look for additional matches if this one does
464 * not have either the SO_REUSEPORT or SO_REUSEADDR
465 * socket options set. This heuristic avoids searching
466 * through all pcbs in the common case of a non-shared
467 * port. It * assumes that an application will never
468 * clear these options after setting them.
470 if ((last
->inp_socket
->so_options
&(SO_REUSEPORT
|SO_REUSEADDR
)) == 0)
473 lck_rw_done(pcbinfo
->mtx
);
477 * No matching pcb found; discard datagram.
478 * (No need to send an ICMP Port Unreachable
479 * for a broadcast or multicast datgram.)
481 udpstat
.udps_noportbcast
++;
485 /* check AH/ESP integrity. */
486 if (ipsec_bypass
== 0 && m
) {
487 lck_mtx_lock(sadb_mutex
);
488 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
489 ipsecstat
.in_polvio
++;
490 lck_mtx_unlock(sadb_mutex
);
491 udp_unlock(last
->inp_socket
, 1, 0);
494 lck_mtx_unlock(sadb_mutex
);
497 udp_append(last
, ip
, m
, iphlen
+ sizeof(struct udphdr
));
498 udp_unlock(last
->inp_socket
, 1, 0);
504 * UDP to port 4500 with a payload where the first four bytes are
505 * not zero is a UDP encapsulated IPSec packet. Packets where
506 * the payload is one byte and that byte is 0xFF are NAT keepalive
507 * packets. Decapsulate the ESP packet and carry on with IPSec input
508 * or discard the NAT keep-alive.
510 if (ipsec_bypass
== 0 && (esp_udp_encap_port
& 0xFFFF) != 0 &&
511 uh
->uh_dport
== ntohs((u_short
)esp_udp_encap_port
)) {
512 int payload_len
= len
- sizeof(struct udphdr
) > 4 ? 4 : len
- sizeof(struct udphdr
);
513 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
) + payload_len
) {
514 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
) + payload_len
)) == 0) {
515 udpstat
.udps_hdrops
++;
516 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
519 ip
= mtod(m
, struct ip
*);
520 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
522 /* Check for NAT keepalive packet */
523 if (payload_len
== 1 && *(u_int8_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) == 0xFF) {
525 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
528 else if (payload_len
== 4 && *(u_int32_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) != 0) {
529 /* UDP encapsulated IPSec packet to pass through NAT */
532 stripsiz
= sizeof(struct udphdr
);
534 ip
= mtod(m
, struct ip
*);
535 ovbcopy((caddr_t
)ip
, (caddr_t
)(((u_char
*)ip
) + stripsiz
), iphlen
);
536 m
->m_data
+= stripsiz
;
537 m
->m_len
-= stripsiz
;
538 m
->m_pkthdr
.len
-= stripsiz
;
539 ip
= mtod(m
, struct ip
*);
540 ip
->ip_len
= ip
->ip_len
- stripsiz
;
541 ip
->ip_p
= IPPROTO_ESP
;
543 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
544 esp4_input(m
, iphlen
);
551 * Locate pcb for datagram.
553 inp
= in_pcblookup_hash(&udbinfo
, ip
->ip_src
, uh
->uh_sport
,
554 ip
->ip_dst
, uh
->uh_dport
, 1, m
->m_pkthdr
.rcvif
);
557 char buf
[MAX_IPv4_STR_LEN
];
558 char buf2
[MAX_IPv4_STR_LEN
];
560 /* check src and dst address */
561 if (log_in_vain
!= 3)
563 "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
));
568 else if (!(m
->m_flags
& (M_BCAST
| M_MCAST
)) &&
569 ip
->ip_dst
.s_addr
!= ip
->ip_src
.s_addr
)
570 log_in_vain_log((LOG_INFO
,
571 "Stealth Mode connection attempt to UDP %s:%d from %s:%d\n",
572 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
574 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
575 ntohs(uh
->uh_sport
)))
577 udpstat
.udps_noport
++;
578 if (m
->m_flags
& (M_BCAST
| M_MCAST
)) {
579 udpstat
.udps_noportbcast
++;
583 if (badport_bandlim(BANDLIM_ICMP_UNREACH
) < 0)
587 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
!= IFT_LOOP
)
590 ip
->ip_len
+= iphlen
;
591 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_PORT
, 0, 0);
592 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
595 udp_lock(inp
->inp_socket
, 1, 0);
597 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
598 udp_unlock(inp
->inp_socket
, 1, 0);
602 if (ipsec_bypass
== 0 && inp
!= NULL
) {
603 lck_mtx_lock(sadb_mutex
);
604 if (ipsec4_in_reject_so(m
, inp
->inp_socket
)) {
605 ipsecstat
.in_polvio
++;
606 lck_mtx_unlock(sadb_mutex
);
607 udp_unlock(inp
->inp_socket
, 1, 0);
610 lck_mtx_unlock(sadb_mutex
);
615 * Construct sockaddr format source address.
616 * Stuff source address and datagram in user buffer.
618 udp_in
.sin_port
= uh
->uh_sport
;
619 udp_in
.sin_addr
= ip
->ip_src
;
620 if (inp
->inp_flags
& INP_CONTROLOPTS
621 || inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
623 if (inp
->inp_vflag
& INP_IPV6
) {
626 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
627 savedflags
= inp
->inp_flags
;
628 inp
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
629 ip6_savecontrol(inp
, &opts
, &udp_ip6
.uip6_ip6
, m
);
630 inp
->inp_flags
= savedflags
;
633 ip_savecontrol(inp
, &opts
, ip
, m
);
635 m_adj(m
, iphlen
+ sizeof(struct udphdr
));
637 KERNEL_DEBUG(DBG_LAYER_IN_END
, uh
->uh_dport
, uh
->uh_sport
,
638 save_ip
.ip_src
.s_addr
, save_ip
.ip_dst
.s_addr
, uh
->uh_ulen
);
641 if (inp
->inp_vflag
& INP_IPV6
) {
642 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
643 append_sa
= (struct sockaddr
*)&udp_in6
;
646 append_sa
= (struct sockaddr
*)&udp_in
;
647 if (sbappendaddr(&inp
->inp_socket
->so_rcv
, append_sa
, m
, opts
, NULL
) == 0) {
648 udpstat
.udps_fullsock
++;
651 sorwakeup(inp
->inp_socket
);
653 udp_unlock(inp
->inp_socket
, 1, 0);
654 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
660 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
666 ip_2_ip6_hdr(ip6
, ip
)
670 bzero(ip6
, sizeof(*ip6
));
672 ip6
->ip6_vfc
= IPV6_VERSION
;
673 ip6
->ip6_plen
= ip
->ip_len
;
674 ip6
->ip6_nxt
= ip
->ip_p
;
675 ip6
->ip6_hlim
= ip
->ip_ttl
;
676 ip6
->ip6_src
.s6_addr32
[2] = ip6
->ip6_dst
.s6_addr32
[2] =
678 ip6
->ip6_src
.s6_addr32
[3] = ip
->ip_src
.s_addr
;
679 ip6
->ip6_dst
.s6_addr32
[3] = ip
->ip_dst
.s_addr
;
684 * subroutine of udp_input(), mainly for source code readability.
685 * caller must properly init udp_ip6 and udp_in6 beforehand.
688 udp_append(last
, ip
, n
, off
)
694 struct sockaddr
*append_sa
;
695 struct mbuf
*opts
= 0;
697 if (last
->inp_flags
& INP_CONTROLOPTS
||
698 last
->inp_socket
->so_options
& SO_TIMESTAMP
) {
700 if (last
->inp_vflag
& INP_IPV6
) {
703 if (udp_ip6
.uip6_init_done
== 0) {
704 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
705 udp_ip6
.uip6_init_done
= 1;
707 savedflags
= last
->inp_flags
;
708 last
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
709 ip6_savecontrol(last
, &opts
, &udp_ip6
.uip6_ip6
, n
);
710 last
->inp_flags
= savedflags
;
713 ip_savecontrol(last
, &opts
, ip
, n
);
716 if (last
->inp_vflag
& INP_IPV6
) {
717 if (udp_in6
.uin6_init_done
== 0) {
718 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
719 udp_in6
.uin6_init_done
= 1;
721 append_sa
= (struct sockaddr
*)&udp_in6
.uin6_sin
;
724 append_sa
= (struct sockaddr
*)&udp_in
;
726 if (sbappendaddr(&last
->inp_socket
->so_rcv
, append_sa
, n
, opts
, NULL
) == 0) {
727 udpstat
.udps_fullsock
++;
729 sorwakeup(last
->inp_socket
);
733 * Notify a udp user of an asynchronous error;
734 * just wake up so that he can collect error status.
737 udp_notify(inp
, errno
)
738 register struct inpcb
*inp
;
741 inp
->inp_socket
->so_error
= errno
;
742 sorwakeup(inp
->inp_socket
);
743 sowwakeup(inp
->inp_socket
);
747 udp_ctlinput(cmd
, sa
, vip
)
754 void (*notify
)(struct inpcb
*, int) = udp_notify
;
755 struct in_addr faddr
;
758 faddr
= ((struct sockaddr_in
*)sa
)->sin_addr
;
759 if (sa
->sa_family
!= AF_INET
|| faddr
.s_addr
== INADDR_ANY
)
762 if (PRC_IS_REDIRECT(cmd
)) {
764 notify
= in_rtchange
;
765 } else if (cmd
== PRC_HOSTDEAD
)
767 else if ((unsigned)cmd
>= PRC_NCMDS
|| inetctlerrmap
[cmd
] == 0)
770 uh
= (struct udphdr
*)((caddr_t
)ip
+ (ip
->ip_hl
<< 2));
771 inp
= in_pcblookup_hash(&udbinfo
, faddr
, uh
->uh_dport
,
772 ip
->ip_src
, uh
->uh_sport
, 0, NULL
);
773 if (inp
!= NULL
&& inp
->inp_socket
!= NULL
) {
774 udp_lock(inp
->inp_socket
, 1, 0);
775 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
776 udp_unlock(inp
->inp_socket
, 1, 0);
779 (*notify
)(inp
, inetctlerrmap
[cmd
]);
780 udp_unlock(inp
->inp_socket
, 1, 0);
783 in_pcbnotifyall(&udbinfo
, faddr
, inetctlerrmap
[cmd
], notify
);
787 udp_pcblist SYSCTL_HANDLER_ARGS
790 struct inpcb
*inp
, **inp_list
;
795 * The process of preparing the TCB list is too time-consuming and
796 * resource-intensive to repeat twice on every request.
798 lck_rw_lock_exclusive(udbinfo
.mtx
);
799 if (req
->oldptr
== USER_ADDR_NULL
) {
800 n
= udbinfo
.ipi_count
;
801 req
->oldidx
= 2 * (sizeof xig
)
802 + (n
+ n
/8) * sizeof(struct xinpcb
);
803 lck_rw_done(udbinfo
.mtx
);
807 if (req
->newptr
!= USER_ADDR_NULL
) {
808 lck_rw_done(udbinfo
.mtx
);
813 * OK, now we're committed to doing something.
815 gencnt
= udbinfo
.ipi_gencnt
;
816 n
= udbinfo
.ipi_count
;
818 bzero(&xig
, sizeof(xig
));
819 xig
.xig_len
= sizeof xig
;
821 xig
.xig_gen
= gencnt
;
822 xig
.xig_sogen
= so_gencnt
;
823 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
825 lck_rw_done(udbinfo
.mtx
);
829 * We are done if there is no pcb
832 lck_rw_done(udbinfo
.mtx
);
836 inp_list
= _MALLOC(n
* sizeof *inp_list
, M_TEMP
, M_WAITOK
);
838 lck_rw_done(udbinfo
.mtx
);
842 for (inp
= LIST_FIRST(udbinfo
.listhead
), i
= 0; inp
&& i
< n
;
843 inp
= LIST_NEXT(inp
, inp_list
)) {
844 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
)
850 for (i
= 0; i
< n
; i
++) {
852 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
) {
855 bzero(&xi
, sizeof(xi
));
856 xi
.xi_len
= sizeof xi
;
857 /* XXX should avoid extra copy */
858 inpcb_to_compat(inp
, &xi
.xi_inp
);
860 sotoxsocket(inp
->inp_socket
, &xi
.xi_socket
);
861 error
= SYSCTL_OUT(req
, &xi
, sizeof xi
);
866 * Give the user an updated idea of our state.
867 * If the generation differs from what we told
868 * her before, she knows that something happened
869 * while we were processing this request, and it
870 * might be necessary to retry.
872 bzero(&xig
, sizeof(xig
));
873 xig
.xig_len
= sizeof xig
;
874 xig
.xig_gen
= udbinfo
.ipi_gencnt
;
875 xig
.xig_sogen
= so_gencnt
;
876 xig
.xig_count
= udbinfo
.ipi_count
;
877 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
879 FREE(inp_list
, M_TEMP
);
880 lck_rw_done(udbinfo
.mtx
);
884 SYSCTL_PROC(_net_inet_udp
, UDPCTL_PCBLIST
, pcblist
, CTLFLAG_RD
, 0, 0,
885 udp_pcblist
, "S,xinpcb", "List of active UDP sockets");
889 static __inline__ u_int16_t
890 get_socket_id(struct socket
* s
)
897 val
= (u_int16_t
)(((u_int32_t
)s
) / sizeof(struct socket
));
905 udp_output(inp
, m
, addr
, control
, p
)
906 register struct inpcb
*inp
;
908 struct sockaddr
*addr
;
909 struct mbuf
*control
;
912 register struct udpiphdr
*ui
;
913 register int len
= m
->m_pkthdr
.len
;
914 struct sockaddr_in
*sin
, src
;
915 struct in_addr origladdr
, laddr
, faddr
;
916 u_short lport
, fport
;
917 struct sockaddr_in
*ifaddr
;
918 int error
= 0, udp_dodisconnect
= 0;
921 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_START
, 0,0,0,0,0);
924 m_freem(control
); /* XXX */
926 KERNEL_DEBUG(DBG_LAYER_OUT_BEG
, inp
->inp_fport
, inp
->inp_lport
,
927 inp
->inp_laddr
.s_addr
, inp
->inp_faddr
.s_addr
,
928 (htons((u_short
)len
+ sizeof (struct udphdr
))));
930 if (len
+ sizeof(struct udpiphdr
) > IP_MAXPACKET
) {
935 /* If there was a routing change, discard cached route and check
936 * that we have a valid source address.
937 * Reacquire a new source address if INADDR_ANY was specified
941 lck_mtx_assert(inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
944 if (inp
->inp_route
.ro_rt
&& inp
->inp_route
.ro_rt
->generation_id
!= route_generation
) {
945 if (ifa_foraddr(inp
->inp_laddr
.s_addr
) == 0) { /* src address is gone */
946 if (inp
->inp_flags
& INP_INADDR_ANY
)
947 inp
->inp_faddr
.s_addr
= INADDR_ANY
; /* new src will be set later */
949 error
= EADDRNOTAVAIL
;
953 rtfree(inp
->inp_route
.ro_rt
);
954 inp
->inp_route
.ro_rt
= (struct rtentry
*)0;
957 origladdr
= laddr
= inp
->inp_laddr
;
958 faddr
= inp
->inp_faddr
;
959 lport
= inp
->inp_lport
;
960 fport
= inp
->inp_fport
;
963 sin
= (struct sockaddr_in
*)addr
;
964 if (faddr
.s_addr
!= INADDR_ANY
) {
970 * In case we don't have a local port set, go through the full connect.
971 * We don't have a local port yet (ie, we can't be looked up),
972 * so it's not an issue if the input runs at the same time we do this.
974 error
= in_pcbconnect(inp
, addr
, p
);
978 laddr
= inp
->inp_laddr
;
979 lport
= inp
->inp_lport
;
980 faddr
= inp
->inp_faddr
;
981 fport
= inp
->inp_fport
;
982 udp_dodisconnect
= 1;
986 * we have a full address and a local port.
987 * use those info to build the packet without changing the pcb
988 * and interfering with the input path. See 3851370
990 if (laddr
.s_addr
== INADDR_ANY
) {
991 if ((error
= in_pcbladdr(inp
, addr
, &ifaddr
)) != 0)
993 laddr
= ifaddr
->sin_addr
;
994 inp
->inp_flags
|= INP_INADDR_ANY
; /* from pcbconnect: remember we don't care about src addr.*/
997 faddr
= sin
->sin_addr
;
998 fport
= sin
->sin_port
;
1001 if (faddr
.s_addr
== INADDR_ANY
) {
1009 * Calculate data length and get a mbuf
1010 * for UDP and IP headers.
1012 M_PREPEND(m
, sizeof(struct udpiphdr
), M_DONTWAIT
);
1019 * Fill in mbuf with extended UDP header
1020 * and addresses and length put into network format.
1022 ui
= mtod(m
, struct udpiphdr
*);
1023 bzero(ui
->ui_x1
, sizeof(ui
->ui_x1
)); /* XXX still needed? */
1024 ui
->ui_pr
= IPPROTO_UDP
;
1027 ui
->ui_sport
= lport
;
1028 ui
->ui_dport
= fport
;
1029 ui
->ui_ulen
= htons((u_short
)len
+ sizeof(struct udphdr
));
1032 * Set up checksum and output datagram.
1035 ui
->ui_sum
= in_pseudo(ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
,
1036 htons((u_short
)len
+ sizeof(struct udphdr
) + IPPROTO_UDP
));
1037 m
->m_pkthdr
.csum_flags
= CSUM_UDP
;
1038 m
->m_pkthdr
.csum_data
= offsetof(struct udphdr
, uh_sum
);
1042 ((struct ip
*)ui
)->ip_len
= sizeof (struct udpiphdr
) + len
;
1043 ((struct ip
*)ui
)->ip_ttl
= inp
->inp_ip_ttl
; /* XXX */
1044 ((struct ip
*)ui
)->ip_tos
= inp
->inp_ip_tos
; /* XXX */
1045 udpstat
.udps_opackets
++;
1047 KERNEL_DEBUG(DBG_LAYER_OUT_END
, ui
->ui_dport
, ui
->ui_sport
,
1048 ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
, ui
->ui_ulen
);
1051 if (ipsec_bypass
== 0 && ipsec_setsocket(m
, inp
->inp_socket
) != 0) {
1056 m
->m_pkthdr
.socket_id
= get_socket_id(inp
->inp_socket
);
1057 error
= ip_output_list(m
, 0, inp
->inp_options
, &inp
->inp_route
,
1058 (inp
->inp_socket
->so_options
& (SO_DONTROUTE
| SO_BROADCAST
)),
1061 if (udp_dodisconnect
) {
1062 in_pcbdisconnect(inp
);
1063 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1065 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1069 if (udp_dodisconnect
) {
1070 in_pcbdisconnect(inp
);
1071 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1076 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1080 u_long udp_sendspace
= 9216; /* really max datagram size */
1081 /* 40 1K datagrams */
1082 SYSCTL_INT(_net_inet_udp
, UDPCTL_MAXDGRAM
, maxdgram
, CTLFLAG_RW
,
1083 &udp_sendspace
, 0, "Maximum outgoing UDP datagram size");
1085 u_long udp_recvspace
= 40 * (1024 +
1087 sizeof(struct sockaddr_in6
)
1089 sizeof(struct sockaddr_in
)
1092 SYSCTL_INT(_net_inet_udp
, UDPCTL_RECVSPACE
, recvspace
, CTLFLAG_RW
,
1093 &udp_recvspace
, 0, "Maximum incoming UDP datagram size");
1096 udp_abort(struct socket
*so
)
1100 inp
= sotoinpcb(so
);
1102 panic("udp_abort: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1103 soisdisconnected(so
);
1109 udp_attach(struct socket
*so
, int proto
, struct proc
*p
)
1114 inp
= sotoinpcb(so
);
1116 panic ("udp_attach so=%x inp=%x\n", so
, inp
);
1118 error
= in_pcballoc(so
, &udbinfo
, p
);
1121 error
= soreserve(so
, udp_sendspace
, udp_recvspace
);
1124 inp
= (struct inpcb
*)so
->so_pcb
;
1125 inp
->inp_vflag
|= INP_IPV4
;
1126 inp
->inp_ip_ttl
= ip_defttl
;
1131 udp_bind(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1136 inp
= sotoinpcb(so
);
1139 error
= in_pcbbind(inp
, nam
, p
);
1144 udp_connect(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1149 inp
= sotoinpcb(so
);
1152 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
)
1154 error
= in_pcbconnect(inp
, nam
, p
);
1161 udp_detach(struct socket
*so
)
1165 inp
= sotoinpcb(so
);
1167 panic("udp_detach: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1169 inp
->inp_state
= INPCB_STATE_DEAD
;
1174 udp_disconnect(struct socket
*so
)
1178 inp
= sotoinpcb(so
);
1181 if (inp
->inp_faddr
.s_addr
== INADDR_ANY
)
1184 in_pcbdisconnect(inp
);
1185 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
1186 so
->so_state
&= ~SS_ISCONNECTED
; /* XXX */
1191 udp_send(struct socket
*so
, int flags
, struct mbuf
*m
, struct sockaddr
*addr
,
1192 struct mbuf
*control
, struct proc
*p
)
1196 inp
= sotoinpcb(so
);
1201 return udp_output(inp
, m
, addr
, control
, p
);
1205 udp_shutdown(struct socket
*so
)
1209 inp
= sotoinpcb(so
);
1216 struct pr_usrreqs udp_usrreqs
= {
1217 udp_abort
, pru_accept_notsupp
, udp_attach
, udp_bind
, udp_connect
,
1218 pru_connect2_notsupp
, in_control
, udp_detach
, udp_disconnect
,
1219 pru_listen_notsupp
, in_setpeeraddr
, pru_rcvd_notsupp
,
1220 pru_rcvoob_notsupp
, udp_send
, pru_sense_null
, udp_shutdown
,
1221 in_setsockaddr
, sosend
, soreceive
, pru_sopoll_notsupp
1226 udp_lock(so
, refcount
, debug
)
1228 int refcount
, debug
;
1233 __asm__
volatile("mflr %0" : "=r" (lr_saved
));
1235 else lr_saved
= debug
;
1239 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_NOTOWNED
);
1240 lck_mtx_lock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1243 panic("udp_lock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1244 lck_mtx_assert(so
->so_proto
->pr_domain
->dom_mtx
, LCK_MTX_ASSERT_NOTOWNED
);
1245 lck_mtx_lock(so
->so_proto
->pr_domain
->dom_mtx
);
1251 so
->reserved3
= lr_saved
;
1256 udp_unlock(so
, refcount
, debug
)
1262 struct inpcb
*inp
= sotoinpcb(so
);
1263 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1266 __asm__
volatile("mflr %0" : "=r" (lr_saved
));
1268 else lr_saved
= debug
;
1273 if (so
->so_usecount
== 0 && (inp
->inp_wantcnt
== WNT_STOPUSING
)) {
1274 if (lck_rw_try_lock_exclusive(pcbinfo
->mtx
)) {
1276 lck_rw_done(pcbinfo
->mtx
);
1282 if (so
->so_pcb
== NULL
) {
1283 panic("udp_unlock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1284 lck_mtx_assert(so
->so_proto
->pr_domain
->dom_mtx
, LCK_MTX_ASSERT_OWNED
);
1285 lck_mtx_unlock(so
->so_proto
->pr_domain
->dom_mtx
);
1288 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1289 lck_mtx_unlock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1293 so
->reserved4
= lr_saved
;
1298 udp_getlock(so
, locktype
)
1302 struct inpcb
*inp
= sotoinpcb(so
);
1306 return(inp
->inpcb_mtx
);
1308 panic("udp_getlock: so=%x NULL so_pcb\n", so
);
1309 return (so
->so_proto
->pr_domain
->dom_mtx
);
1316 struct inpcb
*inp
, *inpnxt
;
1318 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1320 lck_rw_lock_exclusive(pcbinfo
->mtx
);
1322 for (inp
= udb
.lh_first
; inp
!= NULL
; inp
= inpnxt
) {
1323 inpnxt
= inp
->inp_list
.le_next
;
1325 /* Ignore nat/SharedIP dummy pcbs */
1326 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
1329 if (inp
->inp_wantcnt
!= WNT_STOPUSING
)
1332 so
= inp
->inp_socket
;
1333 if (!lck_mtx_try_lock(inp
->inpcb_mtx
)) /* skip if busy, no hurry for cleanup... */
1336 if (so
->so_usecount
== 0)
1339 lck_mtx_unlock(inp
->inpcb_mtx
);
1341 lck_rw_done(pcbinfo
->mtx
);
1345 ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
)
1347 if ( dstaddr
== srcaddr
){