1 | /* |
2 | * Copyright (c) 2000-2013 Apple Inc. All rights reserved. |
3 | * |
4 | * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ |
5 | * |
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. |
14 | * |
15 | * Please obtain a copy of the License at |
16 | * http://www.opensource.apple.com/apsl/ and read it before using this file. |
17 | * |
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. |
25 | * |
26 | * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ |
27 | */ |
28 | /* |
29 | * Copyright (c) 1982, 1986, 1993 |
30 | * The Regents of the University of California. All rights reserved. |
31 | * |
32 | * Redistribution and use in source and binary forms, with or without |
33 | * modification, are permitted provided that the following conditions |
34 | * are met: |
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. |
47 | * |
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 |
58 | * SUCH DAMAGE. |
59 | * |
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 $ |
62 | */ |
63 | /* |
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, |
67 | * Version 2.0. |
68 | */ |
69 | |
70 | /* |
71 | * Loopback interface driver for protocol testing and timing. |
72 | */ |
73 | #include "loop.h" |
74 | #if NLOOP > 0 |
75 | |
76 | #if NLOOP != 1 |
77 | #error "More than one loopback interface is not supported." |
78 | #endif |
79 | |
80 | #include <sys/param.h> |
81 | #include <sys/systm.h> |
82 | #include <sys/kernel.h> |
83 | #include <sys/mbuf.h> |
84 | #include <sys/socket.h> |
85 | #include <sys/sockio.h> |
86 | #include <sys/mcache.h> |
87 | #include <sys/sysctl.h> |
88 | |
89 | #include <net/if.h> |
90 | #include <net/if_types.h> |
91 | #include <net/route.h> |
92 | #include <net/bpf.h> |
93 | #include <sys/malloc.h> |
94 | |
95 | #if INET |
96 | #include <netinet/in.h> |
97 | #include <netinet/in_var.h> |
98 | #endif |
99 | |
100 | #if INET6 |
101 | #if !INET |
102 | #include <netinet/in.h> |
103 | #endif |
104 | #include <netinet6/in6_var.h> |
105 | #include <netinet/ip6.h> |
106 | #endif |
107 | |
108 | #include <net/dlil.h> |
109 | #include <net/kpi_protocol.h> |
110 | |
111 | #if CONFIG_MACF_NET |
112 | #include <security/mac_framework.h> |
113 | #endif |
114 | |
115 | #include <pexpert/pexpert.h> |
116 | |
117 | #define LOMTU 16384 |
118 | #define LOSNDQ_MAXLEN 256 |
119 | |
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)); \ |
125 | } \ |
126 | } |
127 | |
128 | #define LO_BPF_TAP_OUT_MULTI(_m) { \ |
129 | if (lo_statics[0].bpf_callback != NULL) { \ |
130 | struct mbuf *_n; \ |
131 | for (_n = _m; _n != NULL; _n = _n->m_nextpkt) \ |
132 | LO_BPF_TAP_OUT(_n); \ |
133 | } \ |
134 | } |
135 | |
136 | struct lo_statics_str { |
137 | int bpf_mode; |
138 | bpf_packet_func bpf_callback; |
139 | }; |
140 | |
141 | static struct lo_statics_str lo_statics[NLOOP]; |
142 | static int lo_txstart = 0; |
143 | |
144 | struct ifnet *lo_ifp = NULL; |
145 | |
146 | struct { |
147 | protocol_family_t ; |
148 | }; |
149 | |
150 | /* Local forward declerations */ |
151 | void loopattach(void); |
152 | static errno_t lo_demux(struct ifnet *, struct mbuf *, char *, |
153 | protocol_family_t *); |
154 | static errno_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; |
174 | |
175 | SYSCTL_DECL(_net_link); |
176 | |
177 | SYSCTL_NODE(_net_link, OID_AUTO, loopback, CTLFLAG_RW | CTLFLAG_LOCKED, 0, |
178 | "loopback interface" ); |
179 | |
180 | static u_int32_t lo_dequeue_max = LOSNDQ_MAXLEN; |
181 | SYSCTL_PROC(_net_link_loopback, OID_AUTO, max_dequeue, |
182 | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &lo_dequeue_max, LOSNDQ_MAXLEN, |
183 | sysctl_dequeue_max, "I" , "Maximum number of packets dequeued at a time" ); |
184 | |
185 | static u_int32_t lo_sched_model = IFNET_SCHED_MODEL_NORMAL; |
186 | SYSCTL_PROC(_net_link_loopback, OID_AUTO, sched_model, |
187 | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &lo_sched_model, |
188 | IFNET_SCHED_MODEL_NORMAL, sysctl_sched_model, "I" , "Scheduling model" ); |
189 | |
190 | static u_int32_t lo_dequeue_sc = MBUF_SC_BE; |
191 | static int lo_dequeue_scidx = MBUF_SCIDX(MBUF_SC_BE); |
192 | SYSCTL_PROC(_net_link_loopback, OID_AUTO, dequeue_sc, |
193 | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &lo_dequeue_scidx, |
194 | MBUF_SC_BE, sysctl_dequeue_scidx, "I" , "Dequeue a specific SC index" ); |
195 | |
196 | static errno_t |
197 | lo_demux(struct ifnet *ifp, struct mbuf *m, char *, |
198 | protocol_family_t *protocol_family) |
199 | { |
200 | #pragma unused(ifp, m) |
201 | struct loopback_header * = |
202 | (struct loopback_header *)(void *)frame_header; |
203 | |
204 | *protocol_family = header->protocol; |
205 | |
206 | return (0); |
207 | } |
208 | |
209 | static errno_t |
210 | lo_framer(struct ifnet *ifp, struct mbuf **m, const struct sockaddr *dest, |
211 | const char *dest_linkaddr, const char *frame_type, |
212 | u_int32_t *prepend_len, u_int32_t *postpend_len) |
213 | { |
214 | #pragma unused(ifp, dest, dest_linkaddr) |
215 | struct loopback_header *; |
216 | |
217 | M_PREPEND(*m, sizeof (struct loopback_header), M_WAITOK, 1); |
218 | if (*m == NULL) { |
219 | /* Tell caller not to try to free passed-in mbuf */ |
220 | return (EJUSTRETURN); |
221 | } |
222 | |
223 | if (prepend_len != NULL) |
224 | *prepend_len = sizeof (struct loopback_header); |
225 | if (postpend_len != NULL) |
226 | *postpend_len = 0; |
227 | |
228 | header = mtod(*m, struct loopback_header *); |
229 | bcopy(frame_type, &header->protocol, sizeof (u_int32_t)); |
230 | return (0); |
231 | } |
232 | |
233 | static errno_t |
234 | lo_add_proto(struct ifnet *interface, protocol_family_t protocol_family, |
235 | const struct ifnet_demux_desc *demux_array, u_int32_t demux_count) |
236 | { |
237 | #pragma unused(interface, protocol_family, demux_array, demux_count) |
238 | return (0); |
239 | } |
240 | |
241 | static errno_t |
242 | lo_del_proto(struct ifnet *ifp, protocol_family_t protocol) |
243 | { |
244 | #pragma unused(ifp, protocol) |
245 | return (0); |
246 | } |
247 | |
248 | static void |
249 | lo_tx_compl(struct ifnet *ifp, struct mbuf *m) |
250 | { |
251 | errno_t error; |
252 | |
253 | if ((ifp->if_xflags & IFXF_TIMESTAMP_ENABLED) != 0) { |
254 | boolean_t requested; |
255 | |
256 | error = mbuf_get_timestamp_requested(m, &requested); |
257 | if (requested) { |
258 | struct timespec now; |
259 | u_int64_t ts; |
260 | |
261 | nanouptime(&now); |
262 | net_timernsec(&now, &ts); |
263 | |
264 | error = mbuf_set_timestamp(m, ts, TRUE); |
265 | if (error != 0) |
266 | printf("%s: mbuf_set_timestamp() failed %d\n" , |
267 | __func__, error); |
268 | } |
269 | } |
270 | error = mbuf_set_status(m, KERN_SUCCESS); |
271 | if (error != 0) |
272 | printf("%s: mbuf_set_status() failed %d\n" , |
273 | __func__, error); |
274 | |
275 | ifnet_tx_compl(ifp, m); |
276 | } |
277 | |
278 | /* |
279 | * Output callback. |
280 | * |
281 | * This routine is called only when lo_txstart is disabled. |
282 | */ |
283 | static int |
284 | lo_output(struct ifnet *ifp, struct mbuf *m_list) |
285 | { |
286 | struct mbuf *m, *m_tail = NULL; |
287 | struct ifnet_stat_increment_param s; |
288 | u_int32_t cnt = 0, len = 0; |
289 | |
290 | bzero(&s, sizeof(s)); |
291 | |
292 | for (m = m_list; m; m = m->m_nextpkt) { |
293 | VERIFY(m->m_flags & M_PKTHDR); |
294 | cnt++; |
295 | len += m->m_pkthdr.len; |
296 | |
297 | /* |
298 | * Don't overwrite the rcvif field if it is in use. |
299 | * This is used to match multicast packets, sent looping |
300 | * back, with the appropriate group record on input. |
301 | */ |
302 | if (m->m_pkthdr.rcvif == NULL) |
303 | m->m_pkthdr.rcvif = ifp; |
304 | |
305 | m->m_pkthdr.pkt_flags |= PKTF_LOOP; |
306 | m->m_pkthdr.pkt_hdr = mtod(m, char *); |
307 | |
308 | /* loopback checksums are always OK */ |
309 | m->m_pkthdr.csum_data = 0xffff; |
310 | m->m_pkthdr.csum_flags = |
311 | CSUM_DATA_VALID | CSUM_PSEUDO_HDR | |
312 | CSUM_IP_CHECKED | CSUM_IP_VALID; |
313 | |
314 | m_adj(m, sizeof (struct loopback_header)); |
315 | |
316 | LO_BPF_TAP_OUT(m); |
317 | if (m->m_nextpkt == NULL) { |
318 | m_tail = m; |
319 | } |
320 | lo_tx_compl(ifp, m); |
321 | } |
322 | |
323 | s.packets_in = cnt; |
324 | s.packets_out = cnt; |
325 | s.bytes_in = len; |
326 | s.bytes_out = len; |
327 | |
328 | return (ifnet_input_extended(ifp, m_list, m_tail, &s)); |
329 | } |
330 | |
331 | /* |
332 | * Pre-enqueue callback. |
333 | * |
334 | * This routine is called only when lo_txstart is enabled. |
335 | */ |
336 | static errno_t |
337 | lo_pre_enqueue(struct ifnet *ifp, struct mbuf *m0) |
338 | { |
339 | struct mbuf *m = m0, *n; |
340 | int error = 0; |
341 | |
342 | while (m != NULL) { |
343 | VERIFY(m->m_flags & M_PKTHDR); |
344 | |
345 | n = m->m_nextpkt; |
346 | m->m_nextpkt = NULL; |
347 | |
348 | /* |
349 | * Don't overwrite the rcvif field if it is in use. |
350 | * This is used to match multicast packets, sent looping |
351 | * back, with the appropriate group record on input. |
352 | */ |
353 | if (m->m_pkthdr.rcvif == NULL) |
354 | m->m_pkthdr.rcvif = ifp; |
355 | |
356 | m->m_pkthdr.pkt_flags |= PKTF_LOOP; |
357 | m->m_pkthdr.pkt_hdr = mtod(m, char *); |
358 | |
359 | /* loopback checksums are always OK */ |
360 | m->m_pkthdr.csum_data = 0xffff; |
361 | m->m_pkthdr.csum_flags = |
362 | CSUM_DATA_VALID | CSUM_PSEUDO_HDR | |
363 | CSUM_IP_CHECKED | CSUM_IP_VALID; |
364 | |
365 | m_adj(m, sizeof (struct loopback_header)); |
366 | |
367 | /* |
368 | * Let the callee free it in case of error, |
369 | * and perform any necessary accounting. |
370 | */ |
371 | (void) ifnet_enqueue(ifp, m); |
372 | |
373 | m = n; |
374 | } |
375 | |
376 | return (error); |
377 | } |
378 | |
379 | /* |
380 | * Start output callback. |
381 | * |
382 | * This routine is invoked by the start worker thread; because we never call |
383 | * it directly, there is no need do deploy any serialization mechanism other |
384 | * than what's already used by the worker thread, i.e. this is already single |
385 | * threaded. |
386 | * |
387 | * This routine is called only when lo_txstart is enabled. |
388 | */ |
389 | static void |
390 | lo_start(struct ifnet *ifp) |
391 | { |
392 | struct ifnet_stat_increment_param s; |
393 | |
394 | bzero(&s, sizeof (s)); |
395 | |
396 | for (;;) { |
397 | struct mbuf *m = NULL, *m_tail = NULL; |
398 | u_int32_t cnt, len = 0; |
399 | |
400 | if (lo_sched_model == IFNET_SCHED_MODEL_NORMAL) { |
401 | if (ifnet_dequeue_multi(ifp, lo_dequeue_max, &m, |
402 | &m_tail, &cnt, &len) != 0) |
403 | break; |
404 | } else { |
405 | if (ifnet_dequeue_service_class_multi(ifp, |
406 | lo_dequeue_sc, lo_dequeue_max, &m, |
407 | &m_tail, &cnt, &len) != 0) |
408 | break; |
409 | } |
410 | |
411 | LO_BPF_TAP_OUT_MULTI(m); |
412 | lo_tx_compl(ifp, m); |
413 | |
414 | /* stats are required for extended variant */ |
415 | s.packets_in = cnt; |
416 | s.packets_out = cnt; |
417 | s.bytes_in = len; |
418 | s.bytes_out = len; |
419 | |
420 | (void) ifnet_input_extended(ifp, m, m_tail, &s); |
421 | } |
422 | } |
423 | |
424 | /* |
425 | * This is a common pre-output route used by INET and INET6. This could |
426 | * (should?) be split into separate pre-output routines for each protocol. |
427 | */ |
428 | static errno_t |
429 | lo_pre_output(struct ifnet *ifp, protocol_family_t protocol_family, |
430 | struct mbuf **m, const struct sockaddr *dst, void *route, char *frame_type, |
431 | char *dst_addr) |
432 | { |
433 | #pragma unused(ifp, dst, dst_addr) |
434 | struct rtentry *rt = route; |
435 | |
436 | VERIFY((*m)->m_flags & M_PKTHDR); |
437 | |
438 | (*m)->m_flags |= M_LOOP; |
439 | |
440 | if (rt != NULL) { |
441 | u_int32_t rt_flags = rt->rt_flags; |
442 | if (rt_flags & (RTF_REJECT | RTF_BLACKHOLE)) { |
443 | if (rt_flags & RTF_BLACKHOLE) { |
444 | m_freem(*m); |
445 | return (EJUSTRETURN); |
446 | } else { |
447 | return ((rt_flags & RTF_HOST) ? |
448 | EHOSTUNREACH : ENETUNREACH); |
449 | } |
450 | } |
451 | } |
452 | |
453 | bcopy(&protocol_family, frame_type, sizeof (protocol_family)); |
454 | |
455 | return (0); |
456 | } |
457 | |
458 | /* |
459 | * lo_input - This should work for all attached protocols that use the |
460 | * ifq/schednetisr input mechanism. |
461 | */ |
462 | static errno_t |
463 | lo_input(struct ifnet *ifp, protocol_family_t protocol_family, struct mbuf *m) |
464 | { |
465 | #pragma unused(ifp, protocol_family) |
466 | |
467 | if ((ifp->if_xflags & IFXF_TIMESTAMP_ENABLED) != 0) { |
468 | errno_t error; |
469 | struct timespec now; |
470 | u_int64_t ts; |
471 | |
472 | nanouptime(&now); |
473 | net_timernsec(&now, &ts); |
474 | |
475 | error = mbuf_set_timestamp(m, ts, TRUE); |
476 | if (error != 0) |
477 | printf("%s: mbuf_set_timestamp() failed %d\n" , |
478 | __func__, error); |
479 | } |
480 | |
481 | if (proto_input(protocol_family, m) != 0) |
482 | m_freem(m); |
483 | return (0); |
484 | } |
485 | |
486 | /* ARGSUSED */ |
487 | static void |
488 | lo_rtrequest(int cmd, struct rtentry *rt, struct sockaddr *sa) |
489 | { |
490 | #pragma unused(cmd, sa) |
491 | if (rt != NULL) { |
492 | RT_LOCK_ASSERT_HELD(rt); |
493 | rt->rt_rmx.rmx_mtu = rt->rt_ifp->if_mtu; /* for ISO */ |
494 | /* |
495 | * For optimal performance, the send and receive buffers |
496 | * should be at least twice the MTU plus a little more for |
497 | * overhead. |
498 | */ |
499 | rt->rt_rmx.rmx_recvpipe = rt->rt_rmx.rmx_sendpipe = 3 * LOMTU; |
500 | } |
501 | } |
502 | |
503 | /* |
504 | * Process an ioctl request. |
505 | */ |
506 | static errno_t |
507 | lo_ioctl(struct ifnet *ifp, u_long cmd, void *data) |
508 | { |
509 | int error = 0; |
510 | |
511 | switch (cmd) { |
512 | |
513 | case SIOCSIFADDR: { /* struct ifaddr pointer */ |
514 | struct ifaddr *ifa = data; |
515 | |
516 | ifnet_set_flags(ifp, IFF_UP|IFF_RUNNING, IFF_UP|IFF_RUNNING); |
517 | IFA_LOCK_SPIN(ifa); |
518 | ifa->ifa_rtrequest = lo_rtrequest; |
519 | IFA_UNLOCK(ifa); |
520 | /* |
521 | * Everything else is done at a higher level. |
522 | */ |
523 | break; |
524 | } |
525 | |
526 | case SIOCADDMULTI: /* struct ifreq */ |
527 | case SIOCDELMULTI: { /* struct ifreq */ |
528 | struct ifreq *ifr = data; |
529 | |
530 | if (ifr == NULL) { |
531 | error = EAFNOSUPPORT; /* XXX */ |
532 | break; |
533 | } |
534 | switch (ifr->ifr_addr.sa_family) { |
535 | |
536 | #if INET |
537 | case AF_INET: |
538 | break; |
539 | #endif |
540 | #if INET6 |
541 | case AF_INET6: |
542 | break; |
543 | #endif |
544 | |
545 | default: |
546 | error = EAFNOSUPPORT; |
547 | break; |
548 | } |
549 | break; |
550 | } |
551 | |
552 | case SIOCSIFMTU: { /* struct ifreq */ |
553 | struct ifreq *ifr = data; |
554 | |
555 | bcopy(&ifr->ifr_mtu, &ifp->if_mtu, sizeof (int)); |
556 | break; |
557 | } |
558 | |
559 | case SIOCSIFFLAGS: /* struct ifreq */ |
560 | case SIOCSIFTIMESTAMPENABLE: |
561 | case SIOCSIFTIMESTAMPDISABLE: |
562 | break; |
563 | |
564 | default: |
565 | error = EOPNOTSUPP; |
566 | break; |
567 | } |
568 | return (error); |
569 | } |
570 | #endif /* NLOOP > 0 */ |
571 | |
572 | |
573 | static errno_t |
574 | lo_attach_proto(struct ifnet *ifp, protocol_family_t protocol_family) |
575 | { |
576 | struct ifnet_attach_proto_param_v2 proto; |
577 | errno_t result = 0; |
578 | |
579 | bzero(&proto, sizeof (proto)); |
580 | proto.input = lo_input; |
581 | proto.pre_output = lo_pre_output; |
582 | |
583 | result = ifnet_attach_protocol_v2(ifp, protocol_family, &proto); |
584 | |
585 | if (result && result != EEXIST) { |
586 | printf("lo_attach_proto: ifnet_attach_protocol for %u " |
587 | "returned=%d\n" , protocol_family, result); |
588 | } |
589 | |
590 | return (result); |
591 | } |
592 | |
593 | static void |
594 | lo_reg_if_mods(void) |
595 | { |
596 | int error; |
597 | |
598 | /* Register protocol registration functions */ |
599 | if ((error = proto_register_plumber(PF_INET, |
600 | APPLE_IF_FAM_LOOPBACK, lo_attach_proto, NULL)) != 0) |
601 | printf("proto_register_plumber failed for AF_INET " |
602 | "error=%d\n" , error); |
603 | |
604 | if ((error = proto_register_plumber(PF_INET6, |
605 | APPLE_IF_FAM_LOOPBACK, lo_attach_proto, NULL)) != 0) |
606 | printf("proto_register_plumber failed for AF_INET6 " |
607 | "error=%d\n" , error); |
608 | } |
609 | |
610 | static errno_t |
611 | lo_set_bpf_tap(struct ifnet *ifp, bpf_tap_mode mode, |
612 | bpf_packet_func bpf_callback) |
613 | { |
614 | VERIFY(ifp == lo_ifp); |
615 | |
616 | lo_statics[0].bpf_mode = mode; |
617 | |
618 | switch (mode) { |
619 | case BPF_TAP_DISABLE: |
620 | case BPF_TAP_INPUT: |
621 | lo_statics[0].bpf_callback = NULL; |
622 | break; |
623 | |
624 | case BPF_TAP_OUTPUT: |
625 | case BPF_TAP_INPUT_OUTPUT: |
626 | lo_statics[0].bpf_callback = bpf_callback; |
627 | break; |
628 | } |
629 | |
630 | return (0); |
631 | } |
632 | |
633 | /* ARGSUSED */ |
634 | void |
635 | loopattach(void) |
636 | { |
637 | struct ifnet_init_eparams lo_init; |
638 | errno_t result = 0; |
639 | |
640 | PE_parse_boot_argn("lo_txstart" , &lo_txstart, sizeof (lo_txstart)); |
641 | |
642 | lo_reg_if_mods(); |
643 | |
644 | lo_statics[0].bpf_callback = NULL; |
645 | lo_statics[0].bpf_mode = BPF_TAP_DISABLE; |
646 | |
647 | bzero(&lo_init, sizeof (lo_init)); |
648 | lo_init.ver = IFNET_INIT_CURRENT_VERSION; |
649 | lo_init.len = sizeof (lo_init); |
650 | lo_init.sndq_maxlen = LOSNDQ_MAXLEN; |
651 | if (lo_txstart) { |
652 | lo_init.flags = 0; |
653 | lo_init.pre_enqueue = lo_pre_enqueue; |
654 | lo_init.start = lo_start; |
655 | lo_init.output_sched_model = lo_sched_model; |
656 | } else { |
657 | lo_init.flags = IFNET_INIT_LEGACY; |
658 | lo_init.output = lo_output; |
659 | } |
660 | lo_init.flags |= IFNET_INIT_NX_NOAUTO; |
661 | lo_init.name = "lo" ; |
662 | lo_init.unit = 0; |
663 | lo_init.family = IFNET_FAMILY_LOOPBACK; |
664 | lo_init.type = IFT_LOOP; |
665 | lo_init.demux = lo_demux; |
666 | lo_init.add_proto = lo_add_proto; |
667 | lo_init.del_proto = lo_del_proto; |
668 | lo_init.framer_extended = lo_framer; |
669 | lo_init.softc = &lo_statics[0]; |
670 | lo_init.ioctl = lo_ioctl; |
671 | lo_init.set_bpf_tap = lo_set_bpf_tap; |
672 | |
673 | result = ifnet_allocate_extended(&lo_init, &lo_ifp); |
674 | if (result != 0) { |
675 | panic("%s: couldn't allocate loopback ifnet (%d)\n" , |
676 | __func__, result); |
677 | /* NOTREACHED */ |
678 | } |
679 | |
680 | ifnet_set_mtu(lo_ifp, LOMTU); |
681 | ifnet_set_flags(lo_ifp, IFF_LOOPBACK | IFF_MULTICAST, |
682 | IFF_LOOPBACK | IFF_MULTICAST); |
683 | ifnet_set_offload(lo_ifp, |
684 | IFNET_CSUM_IP | IFNET_CSUM_TCP | IFNET_CSUM_UDP | |
685 | IFNET_CSUM_TCPIPV6 | IFNET_CSUM_UDPIPV6 | IFNET_IPV6_FRAGMENT | |
686 | IFNET_CSUM_FRAGMENT | IFNET_IP_FRAGMENT | IFNET_MULTIPAGES | |
687 | IFNET_TX_STATUS | IFNET_SW_TIMESTAMP); |
688 | ifnet_set_hdrlen(lo_ifp, sizeof (struct loopback_header)); |
689 | ifnet_set_eflags(lo_ifp, IFEF_SENDLIST, IFEF_SENDLIST); |
690 | |
691 | #if CONFIG_MACF_NET |
692 | mac_ifnet_label_init(ifp); |
693 | #endif |
694 | |
695 | result = ifnet_attach(lo_ifp, NULL); |
696 | if (result != 0) { |
697 | panic("%s: couldn't attach loopback ifnet (%d)\n" , |
698 | __func__, result); |
699 | /* NOTREACHED */ |
700 | } |
701 | /* |
702 | * Disable ECN on loopback as ECN serves no purpose and otherwise |
703 | * TCP connections are subject to heuristics like SYN retransmits on RST |
704 | */ |
705 | lo_ifp->if_eflags &= ~IFEF_ECN_ENABLE; |
706 | lo_ifp->if_eflags |= IFEF_ECN_DISABLE; |
707 | |
708 | bpfattach(lo_ifp, DLT_NULL, sizeof (u_int32_t)); |
709 | } |
710 | |
711 | static int |
712 | sysctl_dequeue_max SYSCTL_HANDLER_ARGS |
713 | { |
714 | #pragma unused(arg1, arg2) |
715 | u_int32_t i; |
716 | int err; |
717 | |
718 | i = lo_dequeue_max; |
719 | |
720 | err = sysctl_handle_int(oidp, &i, 0, req); |
721 | if (err != 0 || req->newptr == USER_ADDR_NULL) |
722 | return (err); |
723 | |
724 | if (i < 1) |
725 | i = 1; |
726 | else if (i > LOSNDQ_MAXLEN) |
727 | i = LOSNDQ_MAXLEN; |
728 | |
729 | lo_dequeue_max = i; |
730 | |
731 | return (err); |
732 | } |
733 | |
734 | static int |
735 | sysctl_sched_model SYSCTL_HANDLER_ARGS |
736 | { |
737 | #pragma unused(arg1, arg2) |
738 | u_int32_t i; |
739 | int err; |
740 | |
741 | i = lo_sched_model; |
742 | |
743 | err = sysctl_handle_int(oidp, &i, 0, req); |
744 | if (err != 0 || req->newptr == USER_ADDR_NULL) |
745 | return (err); |
746 | |
747 | switch (i) { |
748 | case IFNET_SCHED_MODEL_NORMAL: |
749 | case IFNET_SCHED_MODEL_DRIVER_MANAGED: |
750 | case IFNET_SCHED_MODEL_FQ_CODEL: |
751 | break; |
752 | |
753 | default: |
754 | err = EINVAL; |
755 | break; |
756 | } |
757 | |
758 | if (err == 0 && (err = ifnet_set_output_sched_model(lo_ifp, i)) == 0) |
759 | lo_sched_model = i; |
760 | |
761 | return (err); |
762 | } |
763 | |
764 | static int |
765 | sysctl_dequeue_scidx SYSCTL_HANDLER_ARGS |
766 | { |
767 | #pragma unused(arg1, arg2) |
768 | u_int32_t i; |
769 | int err; |
770 | |
771 | i = lo_dequeue_scidx; |
772 | |
773 | err = sysctl_handle_int(oidp, &i, 0, req); |
774 | if (err != 0 || req->newptr == USER_ADDR_NULL) |
775 | return (err); |
776 | |
777 | if (!MBUF_VALID_SCIDX(i)) |
778 | return (EINVAL); |
779 | |
780 | if (lo_sched_model != IFNET_SCHED_MODEL_DRIVER_MANAGED) |
781 | return (ENODEV); |
782 | |
783 | lo_dequeue_sc = m_service_class_from_idx(i); |
784 | lo_dequeue_scidx = MBUF_SCIDX(lo_dequeue_sc); |
785 | |
786 | return (err); |
787 | } |
788 | |