2 * Copyright (c) 2000-2013 Apple 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, 1993
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 * @(#)if_loop.c 8.1 (Berkeley) 6/10/93
61 * $FreeBSD: src/sys/net/if_loop.c,v 1.47.2.5 2001/07/03 11:01:41 ume Exp $
64 * NOTICE: This file was modified by SPARTA, Inc. in 2006 to introduce
65 * support for mandatory and extensible security protections. This notice
66 * is included in support of clause 2.2 (b) of the Apple Public License,
71 * Loopback interface driver for protocol testing and timing.
77 #error "More than one loopback interface is not supported."
80 #include <sys/param.h>
81 #include <sys/systm.h>
82 #include <sys/kernel.h>
84 #include <sys/socket.h>
85 #include <sys/sockio.h>
86 #include <sys/mcache.h>
87 #include <sys/sysctl.h>
90 #include <net/if_types.h>
91 #include <net/route.h>
93 #include <sys/malloc.h>
96 #include <netinet/in.h>
97 #include <netinet/in_var.h>
102 #include <netinet/in.h>
104 #include <netinet6/in6_var.h>
105 #include <netinet/ip6.h>
108 #include <net/dlil.h>
109 #include <net/kpi_protocol.h>
112 #include <security/mac_framework.h>
115 #include <pexpert/pexpert.h>
118 #define LOSNDQ_MAXLEN 256
120 #define LO_BPF_TAP_OUT(_m) { \
121 if (lo_statics[0].bpf_callback != NULL) { \
122 bpf_tap_out(lo_ifp, DLT_NULL, _m, \
123 &((struct loopback_header *)_m->m_pkthdr.pkt_hdr)-> \
124 protocol, sizeof (u_int32_t)); \
128 #define LO_BPF_TAP_OUT_MULTI(_m) { \
129 if (lo_statics[0].bpf_callback != NULL) { \
131 for (_n = _m; _n != NULL; _n = _n->m_nextpkt) \
132 LO_BPF_TAP_OUT(_n); \
136 struct lo_statics_str
{
138 bpf_packet_func bpf_callback
;
141 static struct lo_statics_str lo_statics
[NLOOP
];
142 static int lo_txstart
= 0;
144 struct ifnet
*lo_ifp
= NULL
;
146 struct loopback_header
{
147 protocol_family_t protocol
;
150 /* Local forward declerations */
151 void loopattach(void);
152 static errno_t
lo_demux(struct ifnet
*, struct mbuf
*, char *,
153 protocol_family_t
*);
155 lo_framer(struct ifnet
*, struct mbuf
**, const struct sockaddr
*,
156 const char *, const char *, u_int32_t
*, u_int32_t
*);
157 static errno_t
lo_add_proto(struct ifnet
*, protocol_family_t
,
158 const struct ifnet_demux_desc
*, u_int32_t
);
159 static errno_t
lo_del_proto(struct ifnet
*, protocol_family_t
);
160 static int lo_output(struct ifnet
*, struct mbuf
*);
161 static errno_t
lo_pre_enqueue(struct ifnet
*, struct mbuf
*);
162 static void lo_start(struct ifnet
*);
163 static errno_t
lo_pre_output(struct ifnet
*, protocol_family_t
, struct mbuf
**,
164 const struct sockaddr
*, void *, char *, char *);
165 static errno_t
lo_input(struct ifnet
*, protocol_family_t
, struct mbuf
*);
166 static void lo_rtrequest(int, struct rtentry
*, struct sockaddr
*);
167 static errno_t
lo_ioctl(struct ifnet
*, u_long
, void *);
168 static errno_t
lo_attach_proto(struct ifnet
*, protocol_family_t
);
169 static void lo_reg_if_mods(void);
170 static errno_t
lo_set_bpf_tap(struct ifnet
*, bpf_tap_mode
, bpf_packet_func
);
171 static int sysctl_dequeue_max SYSCTL_HANDLER_ARGS
;
172 static int sysctl_sched_model SYSCTL_HANDLER_ARGS
;
173 static int sysctl_dequeue_scidx SYSCTL_HANDLER_ARGS
;
175 SYSCTL_DECL(_net_link
);
177 SYSCTL_NODE(_net_link
, OID_AUTO
, loopback
, CTLFLAG_RW
| CTLFLAG_LOCKED
, 0,
178 "loopback interface");
180 #define LO_BW_SLEEP 10
181 static u_int32_t lo_bw_sleep_usec
= LO_BW_SLEEP
;
182 SYSCTL_UINT(_net_link_loopback
, OID_AUTO
, bw_sleep_usec
,
183 CTLFLAG_RW
| CTLFLAG_LOCKED
, &lo_bw_sleep_usec
, LO_BW_SLEEP
, "");
185 static u_int32_t lo_bw_measure
= 0;
186 SYSCTL_UINT(_net_link_loopback
, OID_AUTO
, bw_measure
,
187 CTLFLAG_RW
| CTLFLAG_LOCKED
, &lo_bw_measure
, 0, "");
189 static u_int32_t lo_dequeue_max
= LOSNDQ_MAXLEN
;
190 SYSCTL_PROC(_net_link_loopback
, OID_AUTO
, max_dequeue
,
191 CTLTYPE_INT
| CTLFLAG_RW
| CTLFLAG_LOCKED
, &lo_dequeue_max
, LOSNDQ_MAXLEN
,
192 sysctl_dequeue_max
, "I", "Maximum number of packets dequeued at a time");
194 static u_int32_t lo_sched_model
= IFNET_SCHED_MODEL_NORMAL
;
195 SYSCTL_PROC(_net_link_loopback
, OID_AUTO
, sched_model
,
196 CTLTYPE_INT
| CTLFLAG_RW
| CTLFLAG_LOCKED
, &lo_sched_model
,
197 IFNET_SCHED_MODEL_NORMAL
, sysctl_sched_model
, "I", "Scheduling model");
199 static u_int32_t lo_dequeue_sc
= MBUF_SC_BE
;
200 static int lo_dequeue_scidx
= MBUF_SCIDX(MBUF_SC_BE
);
201 SYSCTL_PROC(_net_link_loopback
, OID_AUTO
, dequeue_sc
,
202 CTLTYPE_INT
| CTLFLAG_RW
| CTLFLAG_LOCKED
, &lo_dequeue_scidx
,
203 MBUF_SC_BE
, sysctl_dequeue_scidx
, "I", "Dequeue a specific SC index");
206 lo_demux(struct ifnet
*ifp
, struct mbuf
*m
, char *frame_header
,
207 protocol_family_t
*protocol_family
)
209 #pragma unused(ifp, m)
210 struct loopback_header
*header
=
211 (struct loopback_header
*)(void *)frame_header
;
213 *protocol_family
= header
->protocol
;
219 lo_framer(struct ifnet
*ifp
, struct mbuf
**m
, const struct sockaddr
*dest
,
220 const char *dest_linkaddr
, const char *frame_type
,
221 u_int32_t
*prepend_len
, u_int32_t
*postpend_len
)
223 #pragma unused(ifp, dest, dest_linkaddr)
224 struct loopback_header
*header
;
226 M_PREPEND(*m
, sizeof (struct loopback_header
), M_WAITOK
);
228 /* Tell caller not to try to free passed-in mbuf */
229 return (EJUSTRETURN
);
232 if (prepend_len
!= NULL
)
233 *prepend_len
= sizeof (struct loopback_header
);
234 if (postpend_len
!= NULL
)
237 header
= mtod(*m
, struct loopback_header
*);
238 bcopy(frame_type
, &header
->protocol
, sizeof (u_int32_t
));
243 lo_add_proto(struct ifnet
*interface
, protocol_family_t protocol_family
,
244 const struct ifnet_demux_desc
*demux_array
, u_int32_t demux_count
)
246 #pragma unused(interface, protocol_family, demux_array, demux_count)
251 lo_del_proto(struct ifnet
*ifp
, protocol_family_t protocol
)
253 #pragma unused(ifp, protocol)
260 * This routine is called only when lo_txstart is disabled.
263 lo_output(struct ifnet
*ifp
, struct mbuf
*m_list
)
265 struct mbuf
*m
, *m_tail
= NULL
;
266 struct ifnet_stat_increment_param s
;
267 u_int32_t cnt
= 0, len
= 0;
269 bzero(&s
, sizeof(s
));
271 for (m
= m_list
; m
; m
= m
->m_nextpkt
) {
272 VERIFY(m
->m_flags
& M_PKTHDR
);
274 len
+= m
->m_pkthdr
.len
;
277 * Don't overwrite the rcvif field if it is in use.
278 * This is used to match multicast packets, sent looping
279 * back, with the appropriate group record on input.
281 if (m
->m_pkthdr
.rcvif
== NULL
)
282 m
->m_pkthdr
.rcvif
= ifp
;
284 m
->m_pkthdr
.pkt_flags
|= PKTF_LOOP
;
285 m
->m_pkthdr
.pkt_hdr
= mtod(m
, char *);
287 /* loopback checksums are always OK */
288 m
->m_pkthdr
.csum_data
= 0xffff;
289 m
->m_pkthdr
.csum_flags
=
290 CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
|
291 CSUM_IP_CHECKED
| CSUM_IP_VALID
;
293 m_adj(m
, sizeof (struct loopback_header
));
296 if (m
->m_nextpkt
== NULL
) {
306 return (ifnet_input_extended(ifp
, m_list
, m_tail
, &s
));
310 * Pre-enqueue callback.
312 * This routine is called only when lo_txstart is enabled.
315 lo_pre_enqueue(struct ifnet
*ifp
, struct mbuf
*m0
)
317 struct mbuf
*m
= m0
, *n
;
321 VERIFY(m
->m_flags
& M_PKTHDR
);
327 * Don't overwrite the rcvif field if it is in use.
328 * This is used to match multicast packets, sent looping
329 * back, with the appropriate group record on input.
331 if (m
->m_pkthdr
.rcvif
== NULL
)
332 m
->m_pkthdr
.rcvif
= ifp
;
334 m
->m_pkthdr
.pkt_flags
|= PKTF_LOOP
;
335 m
->m_pkthdr
.pkt_hdr
= mtod(m
, char *);
337 /* loopback checksums are always OK */
338 m
->m_pkthdr
.csum_data
= 0xffff;
339 m
->m_pkthdr
.csum_flags
=
340 CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
|
341 CSUM_IP_CHECKED
| CSUM_IP_VALID
;
343 m_adj(m
, sizeof (struct loopback_header
));
346 * Let the callee free it in case of error,
347 * and perform any necessary accounting.
349 (void) ifnet_enqueue(ifp
, m
);
358 * Start output callback.
360 * This routine is invoked by the start worker thread; because we never call
361 * it directly, there is no need do deploy any serialization mechanism other
362 * than what's already used by the worker thread, i.e. this is already single
365 * This routine is called only when lo_txstart is enabled.
368 lo_start(struct ifnet
*ifp
)
370 struct ifnet_stat_increment_param s
;
372 bzero(&s
, sizeof (s
));
375 struct mbuf
*m
= NULL
, *m_tail
= NULL
;
376 u_int32_t cnt
, len
= 0;
380 if (lo_sched_model
== IFNET_SCHED_MODEL_NORMAL
) {
381 if (ifnet_dequeue_multi(ifp
, lo_dequeue_max
, &m
,
382 &m_tail
, &cnt
, &len
) != 0)
385 if (ifnet_dequeue_service_class_multi(ifp
,
386 lo_dequeue_sc
, lo_dequeue_max
, &m
,
387 &m_tail
, &cnt
, &len
) != 0)
391 LO_BPF_TAP_OUT_MULTI(m
);
394 if (cnt
>= if_bw_measure_size
)
395 ifnet_transmit_burst_start(ifp
, m
);
396 if (lo_bw_sleep_usec
> 0) {
397 bzero(&ts
, sizeof(ts
));
398 ts
.tv_nsec
= (lo_bw_sleep_usec
<< 10) * cnt
;
400 /* Add msleep with timeout */
401 (void) msleep(&sleep_chan
, NULL
,
402 PSOCK
, "lo_start", &ts
);
404 if (cnt
>= if_bw_measure_size
)
405 ifnet_transmit_burst_end(ifp
, m_tail
);
408 /* stats are required for extended variant */
414 (void) ifnet_input_extended(ifp
, m
, m_tail
, &s
);
419 * This is a common pre-output route used by INET and INET6. This could
420 * (should?) be split into separate pre-output routines for each protocol.
423 lo_pre_output(struct ifnet
*ifp
, protocol_family_t protocol_family
,
424 struct mbuf
**m
, const struct sockaddr
*dst
, void *route
, char *frame_type
,
427 #pragma unused(ifp, dst, dst_addr)
428 struct rtentry
*rt
= route
;
430 VERIFY((*m
)->m_flags
& M_PKTHDR
);
432 (*m
)->m_flags
|= M_LOOP
;
435 u_int32_t rt_flags
= rt
->rt_flags
;
436 if (rt_flags
& (RTF_REJECT
| RTF_BLACKHOLE
)) {
437 if (rt_flags
& RTF_BLACKHOLE
) {
439 return (EJUSTRETURN
);
441 return ((rt_flags
& RTF_HOST
) ?
442 EHOSTUNREACH
: ENETUNREACH
);
447 bcopy(&protocol_family
, frame_type
, sizeof (protocol_family
));
453 * lo_input - This should work for all attached protocols that use the
454 * ifq/schednetisr input mechanism.
457 lo_input(struct ifnet
*ifp
, protocol_family_t protocol_family
, struct mbuf
*m
)
459 #pragma unused(ifp, protocol_family)
460 if (proto_input(protocol_family
, m
) != 0)
467 lo_rtrequest(int cmd
, struct rtentry
*rt
, struct sockaddr
*sa
)
469 #pragma unused(cmd, sa)
471 RT_LOCK_ASSERT_HELD(rt
);
472 rt
->rt_rmx
.rmx_mtu
= rt
->rt_ifp
->if_mtu
; /* for ISO */
474 * For optimal performance, the send and receive buffers
475 * should be at least twice the MTU plus a little more for
478 rt
->rt_rmx
.rmx_recvpipe
= rt
->rt_rmx
.rmx_sendpipe
= 3 * LOMTU
;
483 * Process an ioctl request.
486 lo_ioctl(struct ifnet
*ifp
, u_long cmd
, void *data
)
492 case SIOCSIFADDR
: { /* struct ifaddr pointer */
493 struct ifaddr
*ifa
= data
;
495 ifnet_set_flags(ifp
, IFF_UP
|IFF_RUNNING
, IFF_UP
|IFF_RUNNING
);
497 ifa
->ifa_rtrequest
= lo_rtrequest
;
500 * Everything else is done at a higher level.
505 case SIOCADDMULTI
: /* struct ifreq */
506 case SIOCDELMULTI
: { /* struct ifreq */
507 struct ifreq
*ifr
= data
;
510 error
= EAFNOSUPPORT
; /* XXX */
513 switch (ifr
->ifr_addr
.sa_family
) {
525 error
= EAFNOSUPPORT
;
531 case SIOCSIFMTU
: { /* struct ifreq */
532 struct ifreq
*ifr
= data
;
534 bcopy(&ifr
->ifr_mtu
, &ifp
->if_mtu
, sizeof (int));
538 case SIOCSIFFLAGS
: /* struct ifreq */
547 #endif /* NLOOP > 0 */
551 lo_attach_proto(struct ifnet
*ifp
, protocol_family_t protocol_family
)
553 struct ifnet_attach_proto_param_v2 proto
;
556 bzero(&proto
, sizeof (proto
));
557 proto
.input
= lo_input
;
558 proto
.pre_output
= lo_pre_output
;
560 result
= ifnet_attach_protocol_v2(ifp
, protocol_family
, &proto
);
562 if (result
&& result
!= EEXIST
) {
563 printf("lo_attach_proto: ifnet_attach_protocol for %u "
564 "returned=%d\n", protocol_family
, result
);
575 /* Register protocol registration functions */
576 if ((error
= proto_register_plumber(PF_INET
,
577 APPLE_IF_FAM_LOOPBACK
, lo_attach_proto
, NULL
)) != 0)
578 printf("proto_register_plumber failed for AF_INET "
579 "error=%d\n", error
);
581 if ((error
= proto_register_plumber(PF_INET6
,
582 APPLE_IF_FAM_LOOPBACK
, lo_attach_proto
, NULL
)) != 0)
583 printf("proto_register_plumber failed for AF_INET6 "
584 "error=%d\n", error
);
588 lo_set_bpf_tap(struct ifnet
*ifp
, bpf_tap_mode mode
,
589 bpf_packet_func bpf_callback
)
591 VERIFY(ifp
== lo_ifp
);
593 lo_statics
[0].bpf_mode
= mode
;
596 case BPF_TAP_DISABLE
:
598 lo_statics
[0].bpf_callback
= NULL
;
602 case BPF_TAP_INPUT_OUTPUT
:
603 lo_statics
[0].bpf_callback
= bpf_callback
;
614 struct ifnet_init_eparams lo_init
;
617 PE_parse_boot_argn("lo_txstart", &lo_txstart
, sizeof (lo_txstart
));
621 lo_statics
[0].bpf_callback
= NULL
;
622 lo_statics
[0].bpf_mode
= BPF_TAP_DISABLE
;
624 bzero(&lo_init
, sizeof (lo_init
));
625 lo_init
.ver
= IFNET_INIT_CURRENT_VERSION
;
626 lo_init
.len
= sizeof (lo_init
);
627 lo_init
.sndq_maxlen
= LOSNDQ_MAXLEN
;
630 lo_init
.pre_enqueue
= lo_pre_enqueue
;
631 lo_init
.start
= lo_start
;
632 lo_init
.output_sched_model
= lo_sched_model
;
634 lo_init
.flags
= IFNET_INIT_LEGACY
;
635 lo_init
.output
= lo_output
;
639 lo_init
.family
= IFNET_FAMILY_LOOPBACK
;
640 lo_init
.type
= IFT_LOOP
;
641 lo_init
.demux
= lo_demux
;
642 lo_init
.add_proto
= lo_add_proto
;
643 lo_init
.del_proto
= lo_del_proto
;
644 lo_init
.framer_extended
= lo_framer
;
645 lo_init
.softc
= &lo_statics
[0];
646 lo_init
.ioctl
= lo_ioctl
;
647 lo_init
.set_bpf_tap
= lo_set_bpf_tap
;
649 result
= ifnet_allocate_extended(&lo_init
, &lo_ifp
);
651 panic("%s: couldn't allocate loopback ifnet (%d)\n",
656 ifnet_set_mtu(lo_ifp
, LOMTU
);
657 ifnet_set_flags(lo_ifp
, IFF_LOOPBACK
| IFF_MULTICAST
,
658 IFF_LOOPBACK
| IFF_MULTICAST
);
659 ifnet_set_offload(lo_ifp
,
660 IFNET_CSUM_IP
| IFNET_CSUM_TCP
| IFNET_CSUM_UDP
|
661 IFNET_CSUM_TCPIPV6
| IFNET_CSUM_UDPIPV6
| IFNET_IPV6_FRAGMENT
|
662 IFNET_CSUM_FRAGMENT
| IFNET_IP_FRAGMENT
| IFNET_MULTIPAGES
);
663 ifnet_set_hdrlen(lo_ifp
, sizeof (struct loopback_header
));
664 ifnet_set_eflags(lo_ifp
, IFEF_SENDLIST
, IFEF_SENDLIST
);
667 mac_ifnet_label_init(ifp
);
670 result
= ifnet_attach(lo_ifp
, NULL
);
672 panic("%s: couldn't attach loopback ifnet (%d)\n",
676 bpfattach(lo_ifp
, DLT_NULL
, sizeof (u_int32_t
));
680 sysctl_dequeue_max SYSCTL_HANDLER_ARGS
682 #pragma unused(arg1, arg2)
688 err
= sysctl_handle_int(oidp
, &i
, 0, req
);
689 if (err
!= 0 || req
->newptr
== USER_ADDR_NULL
)
694 else if (i
> LOSNDQ_MAXLEN
)
703 sysctl_sched_model SYSCTL_HANDLER_ARGS
705 #pragma unused(arg1, arg2)
711 err
= sysctl_handle_int(oidp
, &i
, 0, req
);
712 if (err
!= 0 || req
->newptr
== USER_ADDR_NULL
)
716 case IFNET_SCHED_MODEL_NORMAL
:
717 case IFNET_SCHED_MODEL_DRIVER_MANAGED
:
725 if (err
== 0 && (err
= ifnet_set_output_sched_model(lo_ifp
, i
)) == 0)
732 sysctl_dequeue_scidx SYSCTL_HANDLER_ARGS
734 #pragma unused(arg1, arg2)
738 i
= lo_dequeue_scidx
;
740 err
= sysctl_handle_int(oidp
, &i
, 0, req
);
741 if (err
!= 0 || req
->newptr
== USER_ADDR_NULL
)
744 if (!MBUF_VALID_SCIDX(i
))
747 if (lo_sched_model
!= IFNET_SCHED_MODEL_DRIVER_MANAGED
)
750 lo_dequeue_sc
= m_service_class_from_idx(i
);
751 lo_dequeue_scidx
= MBUF_SCIDX(lo_dequeue_sc
);