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();
209 pcbinfo
->mtx_grp
= lck_grp_alloc_init("udppcb", pcbinfo
->mtx_grp_attr
);
211 pcbinfo
->mtx_attr
= lck_attr_alloc_init();
213 if ((pcbinfo
->mtx
= lck_rw_alloc_init(pcbinfo
->mtx_grp
, pcbinfo
->mtx_attr
)) == NULL
)
214 return; /* pretty much dead if this fails... */
216 in_pcb_nat_init(&udbinfo
, AF_INET
, IPPROTO_UDP
, SOCK_DGRAM
);
218 udbinfo
.ipi_zone
= zinit("udpcb", sizeof(struct inpcb
), maxsockets
,
223 /* for pcb sharing testing only */
224 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
225 kprintf("udp_init in_pcb_new_share_client - stat = %d\n", stat
);
227 laddr
.s_addr
= 0x11646464;
228 faddr
.s_addr
= 0x11646465;
231 in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
232 kprintf("udp_init in_pcb_grab_port - stat = %d\n", stat
);
234 stat
= in_pcb_rem_share_client(&udbinfo
, fake_owner
);
235 kprintf("udp_init in_pcb_rem_share_client - stat = %d\n", stat
);
237 stat
= in_pcb_new_share_client(&udbinfo
, &fake_owner
);
238 kprintf("udp_init in_pcb_new_share_client(2) - stat = %d\n", stat
);
240 laddr
.s_addr
= 0x11646464;
241 faddr
.s_addr
= 0x11646465;
244 stat
= in_pcb_grab_port(&udbinfo
, 0, laddr
, &lport
, faddr
, 1600, 0, fake_owner
);
245 kprintf("udp_init in_pcb_grab_port(2) - stat = %d\n", stat
);
251 register struct mbuf
*m
;
254 register struct ip
*ip
;
255 register struct udphdr
*uh
;
256 register struct inpcb
*inp
;
257 struct mbuf
*opts
= 0;
260 struct sockaddr
*append_sa
;
261 struct inpcbinfo
*pcbinfo
= &udbinfo
;
263 udpstat
.udps_ipackets
++;
265 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_START
, 0,0,0,0,0);
266 if (m
->m_pkthdr
.csum_flags
& CSUM_TCP_SUM16
)
267 m
->m_pkthdr
.csum_flags
= 0; /* invalidate hwcksum for UDP */
270 * Strip IP options, if any; should skip this,
271 * make available to user, and use on returned packets,
272 * but we don't yet have a way to check the checksum
273 * with options still present.
275 if (iphlen
> sizeof (struct ip
)) {
276 ip_stripoptions(m
, (struct mbuf
*)0);
277 iphlen
= sizeof(struct ip
);
281 * Get IP and UDP header together in first mbuf.
283 ip
= mtod(m
, struct ip
*);
284 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
)) {
285 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
))) == 0) {
286 udpstat
.udps_hdrops
++;
287 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
290 ip
= mtod(m
, struct ip
*);
292 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
294 /* destination port of 0 is illegal, based on RFC768. */
295 if (uh
->uh_dport
== 0)
298 KERNEL_DEBUG(DBG_LAYER_IN_BEG
, uh
->uh_dport
, uh
->uh_sport
,
299 ip
->ip_src
.s_addr
, ip
->ip_dst
.s_addr
, uh
->uh_ulen
);
302 * Make mbuf data length reflect UDP length.
303 * If not enough data to reflect UDP length, drop.
305 len
= ntohs((u_short
)uh
->uh_ulen
);
306 if (ip
->ip_len
!= len
) {
307 if (len
> ip
->ip_len
|| len
< sizeof(struct udphdr
)) {
308 udpstat
.udps_badlen
++;
311 m_adj(m
, len
- ip
->ip_len
);
312 /* ip->ip_len = len; */
315 * Save a copy of the IP header in case we want restore it
316 * for sending an ICMP error message in response.
321 * Checksum extended UDP header and data.
324 if (m
->m_pkthdr
.csum_flags
& CSUM_DATA_VALID
) {
325 if (m
->m_pkthdr
.csum_flags
& CSUM_PSEUDO_HDR
)
326 uh
->uh_sum
= m
->m_pkthdr
.csum_data
;
329 uh
->uh_sum
^= 0xffff;
333 *(uint32_t*)&b
[0] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0];
334 *(uint32_t*)&b
[4] = *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4];
335 *(uint8_t*)&b
[8] = *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8];
337 bzero(((struct ipovly
*)ip
)->ih_x1
, 9);
338 ((struct ipovly
*)ip
)->ih_len
= uh
->uh_ulen
;
339 uh
->uh_sum
= in_cksum(m
, len
+ sizeof (struct ip
));
341 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[0] = *(uint32_t*)&b
[0];
342 *(uint32_t*)&((struct ipovly
*)ip
)->ih_x1
[4] = *(uint32_t*)&b
[4];
343 *(uint8_t*)&((struct ipovly
*)ip
)->ih_x1
[8] = *(uint8_t*)&b
[8];
346 udpstat
.udps_badsum
++;
348 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
354 udpstat
.udps_nosum
++;
357 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
)) ||
358 in_broadcast(ip
->ip_dst
, m
->m_pkthdr
.rcvif
)) {
360 lck_rw_lock_shared(pcbinfo
->mtx
);
362 * Deliver a multicast or broadcast datagram to *all* sockets
363 * for which the local and remote addresses and ports match
364 * those of the incoming datagram. This allows more than
365 * one process to receive multi/broadcasts on the same port.
366 * (This really ought to be done for unicast datagrams as
367 * well, but that would cause problems with existing
368 * applications that open both address-specific sockets and
369 * a wildcard socket listening to the same port -- they would
370 * end up receiving duplicates of every unicast datagram.
371 * Those applications open the multiple sockets to overcome an
372 * inadequacy of the UDP socket interface, but for backwards
373 * compatibility we avoid the problem here rather than
374 * fixing the interface. Maybe 4.5BSD will remedy this?)
379 * Construct sockaddr format source address.
381 udp_in
.sin_port
= uh
->uh_sport
;
382 udp_in
.sin_addr
= ip
->ip_src
;
384 * Locate pcb(s) for datagram.
385 * (Algorithm copied from raw_intr().)
389 udp_in6
.uin6_init_done
= udp_ip6
.uip6_init_done
= 0;
391 LIST_FOREACH(inp
, &udb
, inp_list
) {
393 /* Ignore nat/SharedIP dummy pcbs */
394 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
397 if (inp
->inp_socket
== NULL
)
399 if (inp
!= sotoinpcb(inp
->inp_socket
))
400 panic("udp_input: bad so back ptr inp=%x\n", inp
);
402 if ((inp
->inp_vflag
& INP_IPV4
) == 0)
405 if (in_pcb_checkstate(inp
, WNT_ACQUIRE
, 0) == WNT_STOPUSING
) {
409 udp_lock(inp
->inp_socket
, 1, 0);
411 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
412 udp_unlock(inp
->inp_socket
, 1, 0);
416 if (inp
->inp_lport
!= uh
->uh_dport
) {
417 udp_unlock(inp
->inp_socket
, 1, 0);
420 if (inp
->inp_laddr
.s_addr
!= INADDR_ANY
) {
421 if (inp
->inp_laddr
.s_addr
!=
423 udp_unlock(inp
->inp_socket
, 1, 0);
427 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
) {
428 if (inp
->inp_faddr
.s_addr
!=
430 inp
->inp_fport
!= uh
->uh_sport
) {
431 udp_unlock(inp
->inp_socket
, 1, 0);
440 /* check AH/ESP integrity. */
441 if (ipsec_bypass
== 0) {
442 lck_mtx_lock(sadb_mutex
);
443 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
444 ipsecstat
.in_polvio
++;
445 /* do not inject data to pcb */
448 lck_mtx_unlock(sadb_mutex
);
452 if ((n
= m_copy(m
, 0, M_COPYALL
)) != NULL
) {
453 udp_append(last
, ip
, n
,
455 sizeof(struct udphdr
));
457 udp_unlock(last
->inp_socket
, 1, 0);
461 * Don't look for additional matches if this one does
462 * not have either the SO_REUSEPORT or SO_REUSEADDR
463 * socket options set. This heuristic avoids searching
464 * through all pcbs in the common case of a non-shared
465 * port. It * assumes that an application will never
466 * clear these options after setting them.
468 if ((last
->inp_socket
->so_options
&(SO_REUSEPORT
|SO_REUSEADDR
)) == 0)
471 lck_rw_done(pcbinfo
->mtx
);
475 * No matching pcb found; discard datagram.
476 * (No need to send an ICMP Port Unreachable
477 * for a broadcast or multicast datgram.)
479 udpstat
.udps_noportbcast
++;
483 /* check AH/ESP integrity. */
484 if (ipsec_bypass
== 0 && m
) {
485 lck_mtx_lock(sadb_mutex
);
486 if (ipsec4_in_reject_so(m
, last
->inp_socket
)) {
487 ipsecstat
.in_polvio
++;
488 lck_mtx_unlock(sadb_mutex
);
489 udp_unlock(last
->inp_socket
, 1, 0);
492 lck_mtx_unlock(sadb_mutex
);
495 udp_append(last
, ip
, m
, iphlen
+ sizeof(struct udphdr
));
496 udp_unlock(last
->inp_socket
, 1, 0);
502 * UDP to port 4500 with a payload where the first four bytes are
503 * not zero is a UDP encapsulated IPSec packet. Packets where
504 * the payload is one byte and that byte is 0xFF are NAT keepalive
505 * packets. Decapsulate the ESP packet and carry on with IPSec input
506 * or discard the NAT keep-alive.
508 if (ipsec_bypass
== 0 && (esp_udp_encap_port
& 0xFFFF) != 0 &&
509 uh
->uh_dport
== ntohs((u_short
)esp_udp_encap_port
)) {
510 int payload_len
= len
- sizeof(struct udphdr
) > 4 ? 4 : len
- sizeof(struct udphdr
);
511 if (m
->m_len
< iphlen
+ sizeof(struct udphdr
) + payload_len
) {
512 if ((m
= m_pullup(m
, iphlen
+ sizeof(struct udphdr
) + payload_len
)) == 0) {
513 udpstat
.udps_hdrops
++;
514 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
517 ip
= mtod(m
, struct ip
*);
518 uh
= (struct udphdr
*)((caddr_t
)ip
+ iphlen
);
520 /* Check for NAT keepalive packet */
521 if (payload_len
== 1 && *(u_int8_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) == 0xFF) {
523 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
526 else if (payload_len
== 4 && *(u_int32_t
*)((caddr_t
)uh
+ sizeof(struct udphdr
)) != 0) {
527 /* UDP encapsulated IPSec packet to pass through NAT */
530 stripsiz
= sizeof(struct udphdr
);
532 ip
= mtod(m
, struct ip
*);
533 ovbcopy((caddr_t
)ip
, (caddr_t
)(((u_char
*)ip
) + stripsiz
), iphlen
);
534 m
->m_data
+= stripsiz
;
535 m
->m_len
-= stripsiz
;
536 m
->m_pkthdr
.len
-= stripsiz
;
537 ip
= mtod(m
, struct ip
*);
538 ip
->ip_len
= ip
->ip_len
- stripsiz
;
539 ip
->ip_p
= IPPROTO_ESP
;
541 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
542 esp4_input(m
, iphlen
);
549 * Locate pcb for datagram.
551 inp
= in_pcblookup_hash(&udbinfo
, ip
->ip_src
, uh
->uh_sport
,
552 ip
->ip_dst
, uh
->uh_dport
, 1, m
->m_pkthdr
.rcvif
);
555 char buf
[MAX_IPv4_STR_LEN
];
556 char buf2
[MAX_IPv4_STR_LEN
];
558 /* check src and dst address */
559 if (log_in_vain
!= 3)
561 "Connection attempt to UDP %s:%d from %s:%d\n",
562 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
564 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
565 ntohs(uh
->uh_sport
));
566 else if (!(m
->m_flags
& (M_BCAST
| M_MCAST
)) &&
567 ip
->ip_dst
.s_addr
!= ip
->ip_src
.s_addr
)
568 log_in_vain_log((LOG_INFO
,
569 "Stealth Mode connection attempt to UDP %s:%d from %s:%d\n",
570 inet_ntop(AF_INET
, &ip
->ip_dst
, buf
, sizeof(buf
)),
572 inet_ntop(AF_INET
, &ip
->ip_src
, buf2
, sizeof(buf2
)),
573 ntohs(uh
->uh_sport
)))
575 udpstat
.udps_noport
++;
576 if (m
->m_flags
& (M_BCAST
| M_MCAST
)) {
577 udpstat
.udps_noportbcast
++;
581 if (badport_bandlim(BANDLIM_ICMP_UNREACH
) < 0)
585 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
!= IFT_LOOP
)
588 ip
->ip_len
+= iphlen
;
589 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_PORT
, 0, 0);
590 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
593 udp_lock(inp
->inp_socket
, 1, 0);
595 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
596 udp_unlock(inp
->inp_socket
, 1, 0);
600 if (ipsec_bypass
== 0 && inp
!= NULL
) {
601 lck_mtx_lock(sadb_mutex
);
602 if (ipsec4_in_reject_so(m
, inp
->inp_socket
)) {
603 ipsecstat
.in_polvio
++;
604 lck_mtx_unlock(sadb_mutex
);
605 udp_unlock(inp
->inp_socket
, 1, 0);
608 lck_mtx_unlock(sadb_mutex
);
613 * Construct sockaddr format source address.
614 * Stuff source address and datagram in user buffer.
616 udp_in
.sin_port
= uh
->uh_sport
;
617 udp_in
.sin_addr
= ip
->ip_src
;
618 if (inp
->inp_flags
& INP_CONTROLOPTS
619 || inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
621 if (inp
->inp_vflag
& INP_IPV6
) {
624 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
625 savedflags
= inp
->inp_flags
;
626 inp
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
627 ip6_savecontrol(inp
, &opts
, &udp_ip6
.uip6_ip6
, m
);
628 inp
->inp_flags
= savedflags
;
631 ip_savecontrol(inp
, &opts
, ip
, m
);
633 m_adj(m
, iphlen
+ sizeof(struct udphdr
));
635 KERNEL_DEBUG(DBG_LAYER_IN_END
, uh
->uh_dport
, uh
->uh_sport
,
636 save_ip
.ip_src
.s_addr
, save_ip
.ip_dst
.s_addr
, uh
->uh_ulen
);
639 if (inp
->inp_vflag
& INP_IPV6
) {
640 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
641 append_sa
= (struct sockaddr
*)&udp_in6
;
644 append_sa
= (struct sockaddr
*)&udp_in
;
645 if (sbappendaddr(&inp
->inp_socket
->so_rcv
, append_sa
, m
, opts
, NULL
) == 0) {
646 udpstat
.udps_fullsock
++;
649 sorwakeup(inp
->inp_socket
);
651 udp_unlock(inp
->inp_socket
, 1, 0);
652 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
658 KERNEL_DEBUG(DBG_FNC_UDP_INPUT
| DBG_FUNC_END
, 0,0,0,0,0);
664 ip_2_ip6_hdr(ip6
, ip
)
668 bzero(ip6
, sizeof(*ip6
));
670 ip6
->ip6_vfc
= IPV6_VERSION
;
671 ip6
->ip6_plen
= ip
->ip_len
;
672 ip6
->ip6_nxt
= ip
->ip_p
;
673 ip6
->ip6_hlim
= ip
->ip_ttl
;
674 ip6
->ip6_src
.s6_addr32
[2] = ip6
->ip6_dst
.s6_addr32
[2] =
676 ip6
->ip6_src
.s6_addr32
[3] = ip
->ip_src
.s_addr
;
677 ip6
->ip6_dst
.s6_addr32
[3] = ip
->ip_dst
.s_addr
;
682 * subroutine of udp_input(), mainly for source code readability.
683 * caller must properly init udp_ip6 and udp_in6 beforehand.
686 udp_append(last
, ip
, n
, off
)
692 struct sockaddr
*append_sa
;
693 struct mbuf
*opts
= 0;
695 if (last
->inp_flags
& INP_CONTROLOPTS
||
696 last
->inp_socket
->so_options
& SO_TIMESTAMP
) {
698 if (last
->inp_vflag
& INP_IPV6
) {
701 if (udp_ip6
.uip6_init_done
== 0) {
702 ip_2_ip6_hdr(&udp_ip6
.uip6_ip6
, ip
);
703 udp_ip6
.uip6_init_done
= 1;
705 savedflags
= last
->inp_flags
;
706 last
->inp_flags
&= ~INP_UNMAPPABLEOPTS
;
707 ip6_savecontrol(last
, &opts
, &udp_ip6
.uip6_ip6
, n
);
708 last
->inp_flags
= savedflags
;
711 ip_savecontrol(last
, &opts
, ip
, n
);
714 if (last
->inp_vflag
& INP_IPV6
) {
715 if (udp_in6
.uin6_init_done
== 0) {
716 in6_sin_2_v4mapsin6(&udp_in
, &udp_in6
.uin6_sin
);
717 udp_in6
.uin6_init_done
= 1;
719 append_sa
= (struct sockaddr
*)&udp_in6
.uin6_sin
;
722 append_sa
= (struct sockaddr
*)&udp_in
;
724 if (sbappendaddr(&last
->inp_socket
->so_rcv
, append_sa
, n
, opts
, NULL
) == 0) {
725 udpstat
.udps_fullsock
++;
727 sorwakeup(last
->inp_socket
);
731 * Notify a udp user of an asynchronous error;
732 * just wake up so that he can collect error status.
735 udp_notify(inp
, errno
)
736 register struct inpcb
*inp
;
739 inp
->inp_socket
->so_error
= errno
;
740 sorwakeup(inp
->inp_socket
);
741 sowwakeup(inp
->inp_socket
);
745 udp_ctlinput(cmd
, sa
, vip
)
752 void (*notify
)(struct inpcb
*, int) = udp_notify
;
753 struct in_addr faddr
;
756 faddr
= ((struct sockaddr_in
*)sa
)->sin_addr
;
757 if (sa
->sa_family
!= AF_INET
|| faddr
.s_addr
== INADDR_ANY
)
760 if (PRC_IS_REDIRECT(cmd
)) {
762 notify
= in_rtchange
;
763 } else if (cmd
== PRC_HOSTDEAD
)
765 else if ((unsigned)cmd
>= PRC_NCMDS
|| inetctlerrmap
[cmd
] == 0)
768 uh
= (struct udphdr
*)((caddr_t
)ip
+ (ip
->ip_hl
<< 2));
769 inp
= in_pcblookup_hash(&udbinfo
, faddr
, uh
->uh_dport
,
770 ip
->ip_src
, uh
->uh_sport
, 0, NULL
);
771 if (inp
!= NULL
&& inp
->inp_socket
!= NULL
) {
772 udp_lock(inp
->inp_socket
, 1, 0);
773 if (in_pcb_checkstate(inp
, WNT_RELEASE
, 1) == WNT_STOPUSING
) {
774 udp_unlock(inp
->inp_socket
, 1, 0);
777 (*notify
)(inp
, inetctlerrmap
[cmd
]);
778 udp_unlock(inp
->inp_socket
, 1, 0);
781 in_pcbnotifyall(&udbinfo
, faddr
, inetctlerrmap
[cmd
], notify
);
785 udp_pcblist SYSCTL_HANDLER_ARGS
788 struct inpcb
*inp
, **inp_list
;
793 * The process of preparing the TCB list is too time-consuming and
794 * resource-intensive to repeat twice on every request.
796 lck_rw_lock_exclusive(udbinfo
.mtx
);
797 if (req
->oldptr
== USER_ADDR_NULL
) {
798 n
= udbinfo
.ipi_count
;
799 req
->oldidx
= 2 * (sizeof xig
)
800 + (n
+ n
/8) * sizeof(struct xinpcb
);
801 lck_rw_done(udbinfo
.mtx
);
805 if (req
->newptr
!= USER_ADDR_NULL
) {
806 lck_rw_done(udbinfo
.mtx
);
811 * OK, now we're committed to doing something.
813 gencnt
= udbinfo
.ipi_gencnt
;
814 n
= udbinfo
.ipi_count
;
816 bzero(&xig
, sizeof(xig
));
817 xig
.xig_len
= sizeof xig
;
819 xig
.xig_gen
= gencnt
;
820 xig
.xig_sogen
= so_gencnt
;
821 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
823 lck_rw_done(udbinfo
.mtx
);
827 * We are done if there is no pcb
830 lck_rw_done(udbinfo
.mtx
);
834 inp_list
= _MALLOC(n
* sizeof *inp_list
, M_TEMP
, M_WAITOK
);
836 lck_rw_done(udbinfo
.mtx
);
840 for (inp
= LIST_FIRST(udbinfo
.listhead
), i
= 0; inp
&& i
< n
;
841 inp
= LIST_NEXT(inp
, inp_list
)) {
842 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
)
848 for (i
= 0; i
< n
; i
++) {
850 if (inp
->inp_gencnt
<= gencnt
&& inp
->inp_state
!= INPCB_STATE_DEAD
) {
853 bzero(&xi
, sizeof(xi
));
854 xi
.xi_len
= sizeof xi
;
855 /* XXX should avoid extra copy */
856 inpcb_to_compat(inp
, &xi
.xi_inp
);
858 sotoxsocket(inp
->inp_socket
, &xi
.xi_socket
);
859 error
= SYSCTL_OUT(req
, &xi
, sizeof xi
);
864 * Give the user an updated idea of our state.
865 * If the generation differs from what we told
866 * her before, she knows that something happened
867 * while we were processing this request, and it
868 * might be necessary to retry.
870 bzero(&xig
, sizeof(xig
));
871 xig
.xig_len
= sizeof xig
;
872 xig
.xig_gen
= udbinfo
.ipi_gencnt
;
873 xig
.xig_sogen
= so_gencnt
;
874 xig
.xig_count
= udbinfo
.ipi_count
;
875 error
= SYSCTL_OUT(req
, &xig
, sizeof xig
);
877 FREE(inp_list
, M_TEMP
);
878 lck_rw_done(udbinfo
.mtx
);
882 SYSCTL_PROC(_net_inet_udp
, UDPCTL_PCBLIST
, pcblist
, CTLFLAG_RD
, 0, 0,
883 udp_pcblist
, "S,xinpcb", "List of active UDP sockets");
887 static __inline__ u_int16_t
888 get_socket_id(struct socket
* s
)
895 val
= (u_int16_t
)(((u_int32_t
)s
) / sizeof(struct socket
));
903 udp_output(inp
, m
, addr
, control
, p
)
904 register struct inpcb
*inp
;
906 struct sockaddr
*addr
;
907 struct mbuf
*control
;
910 register struct udpiphdr
*ui
;
911 register int len
= m
->m_pkthdr
.len
;
912 struct sockaddr_in
*sin
, src
;
913 struct in_addr origladdr
, laddr
, faddr
;
914 u_short lport
, fport
;
915 struct sockaddr_in
*ifaddr
;
916 int error
= 0, udp_dodisconnect
= 0;
919 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_START
, 0,0,0,0,0);
922 m_freem(control
); /* XXX */
924 KERNEL_DEBUG(DBG_LAYER_OUT_BEG
, inp
->inp_fport
, inp
->inp_lport
,
925 inp
->inp_laddr
.s_addr
, inp
->inp_faddr
.s_addr
,
926 (htons((u_short
)len
+ sizeof (struct udphdr
))));
928 if (len
+ sizeof(struct udpiphdr
) > IP_MAXPACKET
) {
933 /* If there was a routing change, discard cached route and check
934 * that we have a valid source address.
935 * Reacquire a new source address if INADDR_ANY was specified
939 lck_mtx_assert(inp
->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
942 if (inp
->inp_route
.ro_rt
&& inp
->inp_route
.ro_rt
->generation_id
!= route_generation
) {
943 if (ifa_foraddr(inp
->inp_laddr
.s_addr
) == 0) { /* src address is gone */
944 if (inp
->inp_flags
& INP_INADDR_ANY
)
945 inp
->inp_faddr
.s_addr
= INADDR_ANY
; /* new src will be set later */
947 error
= EADDRNOTAVAIL
;
951 rtfree(inp
->inp_route
.ro_rt
);
952 inp
->inp_route
.ro_rt
= (struct rtentry
*)0;
955 origladdr
= laddr
= inp
->inp_laddr
;
956 faddr
= inp
->inp_faddr
;
957 lport
= inp
->inp_lport
;
958 fport
= inp
->inp_fport
;
961 sin
= (struct sockaddr_in
*)addr
;
962 if (faddr
.s_addr
!= INADDR_ANY
) {
968 * In case we don't have a local port set, go through the full connect.
969 * We don't have a local port yet (ie, we can't be looked up),
970 * so it's not an issue if the input runs at the same time we do this.
972 error
= in_pcbconnect(inp
, addr
, p
);
976 laddr
= inp
->inp_laddr
;
977 lport
= inp
->inp_lport
;
978 faddr
= inp
->inp_faddr
;
979 fport
= inp
->inp_fport
;
980 udp_dodisconnect
= 1;
984 * we have a full address and a local port.
985 * use those info to build the packet without changing the pcb
986 * and interfering with the input path. See 3851370
988 if (laddr
.s_addr
== INADDR_ANY
) {
989 if ((error
= in_pcbladdr(inp
, addr
, &ifaddr
)) != 0)
991 laddr
= ifaddr
->sin_addr
;
992 inp
->inp_flags
|= INP_INADDR_ANY
; /* from pcbconnect: remember we don't care about src addr.*/
995 faddr
= sin
->sin_addr
;
996 fport
= sin
->sin_port
;
999 if (faddr
.s_addr
== INADDR_ANY
) {
1007 * Calculate data length and get a mbuf
1008 * for UDP and IP headers.
1010 M_PREPEND(m
, sizeof(struct udpiphdr
), M_DONTWAIT
);
1017 * Fill in mbuf with extended UDP header
1018 * and addresses and length put into network format.
1020 ui
= mtod(m
, struct udpiphdr
*);
1021 bzero(ui
->ui_x1
, sizeof(ui
->ui_x1
)); /* XXX still needed? */
1022 ui
->ui_pr
= IPPROTO_UDP
;
1025 ui
->ui_sport
= lport
;
1026 ui
->ui_dport
= fport
;
1027 ui
->ui_ulen
= htons((u_short
)len
+ sizeof(struct udphdr
));
1030 * Set up checksum and output datagram.
1033 ui
->ui_sum
= in_pseudo(ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
,
1034 htons((u_short
)len
+ sizeof(struct udphdr
) + IPPROTO_UDP
));
1035 m
->m_pkthdr
.csum_flags
= CSUM_UDP
;
1036 m
->m_pkthdr
.csum_data
= offsetof(struct udphdr
, uh_sum
);
1040 ((struct ip
*)ui
)->ip_len
= sizeof (struct udpiphdr
) + len
;
1041 ((struct ip
*)ui
)->ip_ttl
= inp
->inp_ip_ttl
; /* XXX */
1042 ((struct ip
*)ui
)->ip_tos
= inp
->inp_ip_tos
; /* XXX */
1043 udpstat
.udps_opackets
++;
1045 KERNEL_DEBUG(DBG_LAYER_OUT_END
, ui
->ui_dport
, ui
->ui_sport
,
1046 ui
->ui_src
.s_addr
, ui
->ui_dst
.s_addr
, ui
->ui_ulen
);
1049 if (ipsec_bypass
== 0 && ipsec_setsocket(m
, inp
->inp_socket
) != 0) {
1054 m
->m_pkthdr
.socket_id
= get_socket_id(inp
->inp_socket
);
1055 error
= ip_output_list(m
, 0, inp
->inp_options
, &inp
->inp_route
,
1056 (inp
->inp_socket
->so_options
& (SO_DONTROUTE
| SO_BROADCAST
)),
1059 if (udp_dodisconnect
) {
1060 in_pcbdisconnect(inp
);
1061 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1063 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1067 if (udp_dodisconnect
) {
1068 in_pcbdisconnect(inp
);
1069 inp
->inp_laddr
= origladdr
; /* XXX rehash? */
1074 KERNEL_DEBUG(DBG_FNC_UDP_OUTPUT
| DBG_FUNC_END
, error
, 0,0,0,0);
1078 u_long udp_sendspace
= 9216; /* really max datagram size */
1079 /* 40 1K datagrams */
1080 SYSCTL_INT(_net_inet_udp
, UDPCTL_MAXDGRAM
, maxdgram
, CTLFLAG_RW
,
1081 &udp_sendspace
, 0, "Maximum outgoing UDP datagram size");
1083 u_long udp_recvspace
= 40 * (1024 +
1085 sizeof(struct sockaddr_in6
)
1087 sizeof(struct sockaddr_in
)
1090 SYSCTL_INT(_net_inet_udp
, UDPCTL_RECVSPACE
, recvspace
, CTLFLAG_RW
,
1091 &udp_recvspace
, 0, "Maximum incoming UDP datagram size");
1094 udp_abort(struct socket
*so
)
1098 inp
= sotoinpcb(so
);
1100 panic("udp_abort: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1101 soisdisconnected(so
);
1107 udp_attach(struct socket
*so
, int proto
, struct proc
*p
)
1112 inp
= sotoinpcb(so
);
1114 panic ("udp_attach so=%x inp=%x\n", so
, inp
);
1116 error
= in_pcballoc(so
, &udbinfo
, p
);
1119 error
= soreserve(so
, udp_sendspace
, udp_recvspace
);
1122 inp
= (struct inpcb
*)so
->so_pcb
;
1123 inp
->inp_vflag
|= INP_IPV4
;
1124 inp
->inp_ip_ttl
= ip_defttl
;
1129 udp_bind(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1134 inp
= sotoinpcb(so
);
1137 error
= in_pcbbind(inp
, nam
, p
);
1142 udp_connect(struct socket
*so
, struct sockaddr
*nam
, struct proc
*p
)
1147 inp
= sotoinpcb(so
);
1150 if (inp
->inp_faddr
.s_addr
!= INADDR_ANY
)
1152 error
= in_pcbconnect(inp
, nam
, p
);
1159 udp_detach(struct socket
*so
)
1163 inp
= sotoinpcb(so
);
1165 panic("udp_detach: so=%x null inp\n", so
); /* ??? possible? panic instead? */
1167 inp
->inp_state
= INPCB_STATE_DEAD
;
1172 udp_disconnect(struct socket
*so
)
1176 inp
= sotoinpcb(so
);
1179 if (inp
->inp_faddr
.s_addr
== INADDR_ANY
)
1182 in_pcbdisconnect(inp
);
1183 inp
->inp_laddr
.s_addr
= INADDR_ANY
;
1184 so
->so_state
&= ~SS_ISCONNECTED
; /* XXX */
1189 udp_send(struct socket
*so
, int flags
, struct mbuf
*m
, struct sockaddr
*addr
,
1190 struct mbuf
*control
, struct proc
*p
)
1194 inp
= sotoinpcb(so
);
1199 return udp_output(inp
, m
, addr
, control
, p
);
1203 udp_shutdown(struct socket
*so
)
1207 inp
= sotoinpcb(so
);
1214 struct pr_usrreqs udp_usrreqs
= {
1215 udp_abort
, pru_accept_notsupp
, udp_attach
, udp_bind
, udp_connect
,
1216 pru_connect2_notsupp
, in_control
, udp_detach
, udp_disconnect
,
1217 pru_listen_notsupp
, in_setpeeraddr
, pru_rcvd_notsupp
,
1218 pru_rcvoob_notsupp
, udp_send
, pru_sense_null
, udp_shutdown
,
1219 in_setsockaddr
, sosend
, soreceive
, pru_sopoll_notsupp
1224 udp_lock(so
, refcount
, debug
)
1226 int refcount
, debug
;
1230 lr_saved
= (unsigned int) __builtin_return_address(0);
1231 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
);
1243 so
->lock_lr
[so
->next_lock_lr
] = (void *)lr_saved
;
1244 so
->next_lock_lr
= (so
->next_lock_lr
+1) % SO_LCKDBG_MAX
;
1249 udp_unlock(so
, refcount
, debug
)
1255 struct inpcb
*inp
= sotoinpcb(so
);
1256 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1259 lr_saved
= (unsigned int) __builtin_return_address(0);
1260 else lr_saved
= debug
;
1265 if (so
->so_usecount
== 0 && (inp
->inp_wantcnt
== WNT_STOPUSING
)) {
1266 if (lck_rw_try_lock_exclusive(pcbinfo
->mtx
)) {
1268 lck_rw_done(pcbinfo
->mtx
);
1274 if (so
->so_pcb
== NULL
)
1275 panic("udp_unlock: so=%x NO PCB! lr=%x\n", so
, lr_saved
);
1277 lck_mtx_assert(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
, LCK_MTX_ASSERT_OWNED
);
1278 so
->unlock_lr
[so
->next_unlock_lr
] = (void *)lr_saved
;
1279 so
->next_unlock_lr
= (so
->next_unlock_lr
+1) % SO_LCKDBG_MAX
;
1280 lck_mtx_unlock(((struct inpcb
*)so
->so_pcb
)->inpcb_mtx
);
1288 udp_getlock(so
, locktype
)
1292 struct inpcb
*inp
= sotoinpcb(so
);
1296 return(inp
->inpcb_mtx
);
1298 panic("udp_getlock: so=%x NULL so_pcb\n", so
);
1299 return (so
->so_proto
->pr_domain
->dom_mtx
);
1306 struct inpcb
*inp
, *inpnxt
;
1308 struct inpcbinfo
*pcbinfo
= &udbinfo
;
1310 lck_rw_lock_exclusive(pcbinfo
->mtx
);
1312 for (inp
= udb
.lh_first
; inp
!= NULL
; inp
= inpnxt
) {
1313 inpnxt
= inp
->inp_list
.le_next
;
1315 /* Ignore nat/SharedIP dummy pcbs */
1316 if (inp
->inp_socket
== &udbinfo
.nat_dummy_socket
)
1319 if (inp
->inp_wantcnt
!= WNT_STOPUSING
)
1322 so
= inp
->inp_socket
;
1323 if (!lck_mtx_try_lock(inp
->inpcb_mtx
)) /* skip if busy, no hurry for cleanup... */
1326 if (so
->so_usecount
== 0)
1329 lck_mtx_unlock(inp
->inpcb_mtx
);
1331 lck_rw_done(pcbinfo
->mtx
);
1335 ChkAddressOK( __uint32_t dstaddr
, __uint32_t srcaddr
)
1337 if ( dstaddr
== srcaddr
){