]> git.saurik.com Git - apple/xnu.git/blob - bsd/netinet/tcp_usrreq.c
projects / apple / xnu.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
xnu-3789.60.24.tar.gz
[apple/xnu.git] / bsd / netinet / tcp_usrreq.c
1 /*
2 * Copyright (c) 2000-2016 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, 1988, 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 * From: @(#)tcp_usrreq.c 8.2 (Berkeley) 1/3/94
61 * $FreeBSD: src/sys/netinet/tcp_usrreq.c,v 1.51.2.9 2001/08/22 00:59:12 silby Exp $
62 */
63
64
65 #include <sys/param.h>
66 #include <sys/systm.h>
67 #include <sys/kernel.h>
68 #include <sys/sysctl.h>
69 #include <sys/mbuf.h>
70 #if INET6
71 #include <sys/domain.h>
72 #endif /* INET6 */
73 #include <sys/kasl.h>
74 #include <sys/socket.h>
75 #include <sys/socketvar.h>
76 #include <sys/protosw.h>
77 #include <sys/syslog.h>
78
79 #include <net/if.h>
80 #include <net/route.h>
81 #include <net/ntstat.h>
82 #include <net/content_filter.h>
83
84 #include <netinet/in.h>
85 #include <netinet/in_systm.h>
86 #if INET6
87 #include <netinet/ip6.h>
88 #endif
89 #include <netinet/in_pcb.h>
90 #if INET6
91 #include <netinet6/in6_pcb.h>
92 #endif
93 #include <netinet/in_var.h>
94 #include <netinet/ip_var.h>
95 #if INET6
96 #include <netinet6/ip6_var.h>
97 #endif
98 #include <netinet/tcp.h>
99 #include <netinet/tcp_fsm.h>
100 #include <netinet/tcp_seq.h>
101 #include <netinet/tcp_timer.h>
102 #include <netinet/tcp_var.h>
103 #include <netinet/tcpip.h>
104 #include <mach/sdt.h>
105 #if TCPDEBUG
106 #include <netinet/tcp_debug.h>
107 #endif
108 #if MPTCP
109 #include <netinet/mptcp_var.h>
110 #endif /* MPTCP */
111
112 #if IPSEC
113 #include <netinet6/ipsec.h>
114 #endif /*IPSEC*/
115
116 #if FLOW_DIVERT
117 #include <netinet/flow_divert.h>
118 #endif /* FLOW_DIVERT */
119
120 void tcp_fill_info(struct tcpcb *, struct tcp_info *);
121 errno_t tcp_fill_info_for_info_tuple(struct info_tuple *, struct tcp_info *);
122
123 int tcp_sysctl_info(struct sysctl_oid *, void *, int , struct sysctl_req *);
124 static void tcp_connection_fill_info(struct tcpcb *tp,
125 struct tcp_connection_info *tci);
126
127 /*
128 * TCP protocol interface to socket abstraction.
129 */
130 extern char *tcpstates[]; /* XXX ??? */
131
132 static int tcp_attach(struct socket *, struct proc *);
133 static int tcp_connect(struct tcpcb *, struct sockaddr *, struct proc *);
134 #if INET6
135 static int tcp6_connect(struct tcpcb *, struct sockaddr *, struct proc *);
136 static int tcp6_usr_connect(struct socket *, struct sockaddr *,
137 struct proc *);
138 #endif /* INET6 */
139 static struct tcpcb *tcp_disconnect(struct tcpcb *);
140 static struct tcpcb *tcp_usrclosed(struct tcpcb *);
141 extern void tcp_sbrcv_trim(struct tcpcb *tp, struct sockbuf *sb);
142
143 #if TCPDEBUG
144 #define TCPDEBUG0 int ostate = 0
145 #define TCPDEBUG1() ostate = tp ? tp->t_state : 0
146 #define TCPDEBUG2(req) if (tp && (so->so_options & SO_DEBUG)) \
147 tcp_trace(TA_USER, ostate, tp, 0, 0, req)
148 #else
149 #define TCPDEBUG0
150 #define TCPDEBUG1()
151 #define TCPDEBUG2(req)
152 #endif
153
154 SYSCTL_PROC(_net_inet_tcp, OID_AUTO, info,
155 CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_KERN,
156 0 , 0, tcp_sysctl_info, "S", "TCP info per tuple");
157
158 /*
159 * TCP attaches to socket via pru_attach(), reserving space,
160 * and an internet control block.
161 *
162 * Returns: 0 Success
163 * EISCONN
164 * tcp_attach:ENOBUFS
165 * tcp_attach:ENOMEM
166 * tcp_attach:??? [IPSEC specific]
167 */
168 static int
169 tcp_usr_attach(struct socket *so, __unused int proto, struct proc *p)
170 {
171 int error;
172 struct inpcb *inp = sotoinpcb(so);
173 struct tcpcb *tp = 0;
174 TCPDEBUG0;
175
176 TCPDEBUG1();
177 if (inp) {
178 error = EISCONN;
179 goto out;
180 }
181
182 error = tcp_attach(so, p);
183 if (error)
184 goto out;
185
186 if ((so->so_options & SO_LINGER) && so->so_linger == 0)
187 so->so_linger = TCP_LINGERTIME * hz;
188 tp = sototcpcb(so);
189 out:
190 TCPDEBUG2(PRU_ATTACH);
191 return error;
192 }
193
194 /*
195 * pru_detach() detaches the TCP protocol from the socket.
196 * If the protocol state is non-embryonic, then can't
197 * do this directly: have to initiate a pru_disconnect(),
198 * which may finish later; embryonic TCB's can just
199 * be discarded here.
200 */
201 static int
202 tcp_usr_detach(struct socket *so)
203 {
204 int error = 0;
205 struct inpcb *inp = sotoinpcb(so);
206 struct tcpcb *tp;
207 TCPDEBUG0;
208
209 if (inp == 0 || (inp->inp_state == INPCB_STATE_DEAD)) {
210 return EINVAL; /* XXX */
211 }
212 lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
213 tp = intotcpcb(inp);
214 /* In case we got disconnected from the peer */
215 if (tp == NULL)
216 goto out;
217 TCPDEBUG1();
218
219 calculate_tcp_clock();
220
221 tp = tcp_disconnect(tp);
222 out:
223 TCPDEBUG2(PRU_DETACH);
224 return error;
225 }
226
227 #if NECP
228 #define COMMON_START() TCPDEBUG0; \
229 do { \
230 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD) \
231 return (EINVAL); \
232 if (necp_socket_should_use_flow_divert(inp)) \
233 return (EPROTOTYPE); \
234 tp = intotcpcb(inp); \
235 TCPDEBUG1(); \
236 calculate_tcp_clock(); \
237 } while (0)
238 #else /* NECP */
239 #define COMMON_START() TCPDEBUG0; \
240 do { \
241 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD) \
242 return (EINVAL); \
243 tp = intotcpcb(inp); \
244 TCPDEBUG1(); \
245 calculate_tcp_clock(); \
246 } while (0)
247 #endif /* !NECP */
248
249 #define COMMON_END(req) out: TCPDEBUG2(req); return error; goto out
250
251
252 /*
253 * Give the socket an address.
254 *
255 * Returns: 0 Success
256 * EINVAL Invalid argument [COMMON_START]
257 * EAFNOSUPPORT Address family not supported
258 * in_pcbbind:EADDRNOTAVAIL Address not available.
259 * in_pcbbind:EINVAL Invalid argument
260 * in_pcbbind:EAFNOSUPPORT Address family not supported [notdef]
261 * in_pcbbind:EACCES Permission denied
262 * in_pcbbind:EADDRINUSE Address in use
263 * in_pcbbind:EAGAIN Resource unavailable, try again
264 * in_pcbbind:EPERM Operation not permitted
265 */
266 static int
267 tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
268 {
269 int error = 0;
270 struct inpcb *inp = sotoinpcb(so);
271 struct tcpcb *tp;
272 struct sockaddr_in *sinp;
273
274 COMMON_START();
275
276 if (nam->sa_family != 0 && nam->sa_family != AF_INET) {
277 error = EAFNOSUPPORT;
278 goto out;
279 }
280
281 /*
282 * Must check for multicast addresses and disallow binding
283 * to them.
284 */
285 sinp = (struct sockaddr_in *)(void *)nam;
286 if (sinp->sin_family == AF_INET &&
287 IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
288 error = EAFNOSUPPORT;
289 goto out;
290 }
291 error = in_pcbbind(inp, nam, p);
292 if (error)
293 goto out;
294 COMMON_END(PRU_BIND);
295
296 }
297
298 #if INET6
299 static int
300 tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
301 {
302 int error = 0;
303 struct inpcb *inp = sotoinpcb(so);
304 struct tcpcb *tp;
305 struct sockaddr_in6 *sin6p;
306
307 COMMON_START();
308
309 if (nam->sa_family != 0 && nam->sa_family != AF_INET6) {
310 error = EAFNOSUPPORT;
311 goto out;
312 }
313
314 /*
315 * Must check for multicast addresses and disallow binding
316 * to them.
317 */
318 sin6p = (struct sockaddr_in6 *)(void *)nam;
319 if (sin6p->sin6_family == AF_INET6 &&
320 IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
321 error = EAFNOSUPPORT;
322 goto out;
323 }
324 inp->inp_vflag &= ~INP_IPV4;
325 inp->inp_vflag |= INP_IPV6;
326 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) {
327 if (IN6_IS_ADDR_UNSPECIFIED(&sin6p->sin6_addr))
328 inp->inp_vflag |= INP_IPV4;
329 else if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
330 struct sockaddr_in sin;
331
332 in6_sin6_2_sin(&sin, sin6p);
333 inp->inp_vflag |= INP_IPV4;
334 inp->inp_vflag &= ~INP_IPV6;
335 error = in_pcbbind(inp, (struct sockaddr *)&sin, p);
336 goto out;
337 }
338 }
339 error = in6_pcbbind(inp, nam, p);
340 if (error)
341 goto out;
342 COMMON_END(PRU_BIND);
343 }
344 #endif /* INET6 */
345
346 /*
347 * Prepare to accept connections.
348 *
349 * Returns: 0 Success
350 * EINVAL [COMMON_START]
351 * in_pcbbind:EADDRNOTAVAIL Address not available.
352 * in_pcbbind:EINVAL Invalid argument
353 * in_pcbbind:EAFNOSUPPORT Address family not supported [notdef]
354 * in_pcbbind:EACCES Permission denied
355 * in_pcbbind:EADDRINUSE Address in use
356 * in_pcbbind:EAGAIN Resource unavailable, try again
357 * in_pcbbind:EPERM Operation not permitted
358 */
359 static int
360 tcp_usr_listen(struct socket *so, struct proc *p)
361 {
362 int error = 0;
363 struct inpcb *inp = sotoinpcb(so);
364 struct tcpcb *tp;
365
366 COMMON_START();
367 if (inp->inp_lport == 0)
368 error = in_pcbbind(inp, NULL, p);
369 if (error == 0)
370 tp->t_state = TCPS_LISTEN;
371 COMMON_END(PRU_LISTEN);
372 }
373
374 #if INET6
375 static int
376 tcp6_usr_listen(struct socket *so, struct proc *p)
377 {
378 int error = 0;
379 struct inpcb *inp = sotoinpcb(so);
380 struct tcpcb *tp;
381
382 COMMON_START();
383 if (inp->inp_lport == 0) {
384 inp->inp_vflag &= ~INP_IPV4;
385 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0)
386 inp->inp_vflag |= INP_IPV4;
387 error = in6_pcbbind(inp, NULL, p);
388 }
389 if (error == 0)
390 tp->t_state = TCPS_LISTEN;
391 COMMON_END(PRU_LISTEN);
392 }
393 #endif /* INET6 */
394
395 static int
396 tcp_connect_complete(struct socket *so)
397 {
398 struct tcpcb *tp = sototcpcb(so);
399 int error = 0;
400
401 /* TFO delays the tcp_output until later, when the app calls write() */
402 if (so->so_flags1 & SOF1_PRECONNECT_DATA) {
403 if (!necp_socket_is_allowed_to_send_recv(sotoinpcb(so), NULL, NULL))
404 return (EHOSTUNREACH);
405
406 /* Initialize enough state so that we can actually send data */
407 tcp_mss(tp, -1, IFSCOPE_NONE);
408 tp->snd_wnd = tp->t_maxseg;
409 } else {
410 error = tcp_output(tp);
411 }
412
413 return (error);
414 }
415
416 /*
417 * Initiate connection to peer.
418 * Create a template for use in transmissions on this connection.
419 * Enter SYN_SENT state, and mark socket as connecting.
420 * Start keep-alive timer, and seed output sequence space.
421 * Send initial segment on connection.
422 */
423 static int
424 tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
425 {
426 int error = 0;
427 struct inpcb *inp = sotoinpcb(so);
428 struct tcpcb *tp;
429 struct sockaddr_in *sinp;
430
431 TCPDEBUG0;
432 if (inp == NULL) {
433 return EINVAL;
434 } else if (inp->inp_state == INPCB_STATE_DEAD) {
435 if (so->so_error) {
436 error = so->so_error;
437 so->so_error = 0;
438 return error;
439 } else
440 return EINVAL;
441 }
442 #if NECP
443 #if FLOW_DIVERT
444 else if (necp_socket_should_use_flow_divert(inp)) {
445 uint32_t fd_ctl_unit = necp_socket_get_flow_divert_control_unit(inp);
446 if (fd_ctl_unit > 0) {
447 error = flow_divert_pcb_init(so, fd_ctl_unit);
448 if (error == 0) {
449 error = flow_divert_connect_out(so, nam, p);
450 }
451 } else {
452 error = ENETDOWN;
453 }
454
455 return error;
456 }
457 #endif /* FLOW_DIVERT */
458 #if CONTENT_FILTER
459 error = cfil_sock_attach(so);
460 if (error != 0)
461 return error;
462 #endif /* CONTENT_FILTER */
463 #endif /* NECP */
464 tp = intotcpcb(inp);
465 TCPDEBUG1();
466
467 calculate_tcp_clock();
468
469 if (nam->sa_family != 0 && nam->sa_family != AF_INET) {
470 error = EAFNOSUPPORT;
471 goto out;
472 }
473 /*
474 * Must disallow TCP ``connections'' to multicast addresses.
475 */
476 sinp = (struct sockaddr_in *)(void *)nam;
477 if (sinp->sin_family == AF_INET
478 && IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
479 error = EAFNOSUPPORT;
480 goto out;
481 }
482
483 if ((error = tcp_connect(tp, nam, p)) != 0)
484 goto out;
485
486 error = tcp_connect_complete(so);
487
488 COMMON_END(PRU_CONNECT);
489 }
490
491 static int
492 tcp_usr_connectx_common(struct socket *so, int af,
493 struct sockaddr *src, struct sockaddr *dst,
494 struct proc *p, uint32_t ifscope, sae_associd_t aid, sae_connid_t *pcid,
495 uint32_t flags, void *arg, uint32_t arglen, struct uio *auio,
496 user_ssize_t *bytes_written)
497 {
498 #pragma unused(aid)
499 #if !MPTCP
500 #pragma unused(flags, arg, arglen)
501 #endif /* !MPTCP */
502 struct inpcb *inp = sotoinpcb(so);
503 int error;
504 user_ssize_t datalen = 0;
505
506 if (inp == NULL)
507 return (EINVAL);
508
509 VERIFY(dst != NULL);
510
511 #if NECP
512 inp_update_necp_policy(inp, src, dst, ifscope);
513 #endif /* NECP */
514
515 if ((so->so_flags1 & SOF1_DATA_IDEMPOTENT) &&
516 (tcp_fastopen & TCP_FASTOPEN_CLIENT))
517 sototcpcb(so)->t_flagsext |= TF_FASTOPEN;
518
519 /*
520 * We get here for 2 cases:
521 *
522 * a. From MPTCP, to connect a subflow. There is no need to
523 * bind the socket to the source address and/or interface,
524 * since everything has been taken care of by MPTCP. We
525 * simply check whether or not this is for the initial
526 * MPTCP connection attempt, or to join an existing one.
527 *
528 * b. From the socket layer, to connect a TCP. Perform the
529 * bind to source address and/or interface as necessary.
530 */
531 #if MPTCP
532 if (flags & CONNREQF_MPTCP) {
533 struct mptsub_connreq *mpcr = arg;
534
535 /* Check to make sure this came down from MPTCP */
536 if (arg == NULL || arglen != sizeof (*mpcr))
537 return (EOPNOTSUPP);
538
539 switch (mpcr->mpcr_type) {
540 case MPTSUB_CONNREQ_MP_ENABLE:
541 break;
542 case MPTSUB_CONNREQ_MP_ADD:
543 break;
544 default:
545 return (EOPNOTSUPP);
546 }
547 } else
548 #endif /* MPTCP */
549 {
550 /* bind socket to the specified interface, if requested */
551 if (ifscope != IFSCOPE_NONE &&
552 (error = inp_bindif(inp, ifscope, NULL)) != 0)
553 return (error);
554
555 /* if source address and/or port is specified, bind to it */
556 if (src != NULL) {
557 error = sobindlock(so, src, 0); /* already locked */
558 if (error != 0)
559 return (error);
560 }
561 }
562
563 switch (af) {
564 case AF_INET:
565 error = tcp_usr_connect(so, dst, p);
566 break;
567 #if INET6
568 case AF_INET6:
569 error = tcp6_usr_connect(so, dst, p);
570 break;
571 #endif /* INET6 */
572 default:
573 VERIFY(0);
574 /* NOTREACHED */
575 }
576
577 if (error != 0)
578 return (error);
579
580 /* if there is data, copy it */
581 if (auio != NULL) {
582 socket_unlock(so, 0);
583
584 VERIFY(bytes_written != NULL);
585
586 datalen = uio_resid(auio);
587 error = so->so_proto->pr_usrreqs->pru_sosend(so, NULL,
588 (uio_t)auio, NULL, NULL, 0);
589 socket_lock(so, 0);
590
591 if (error == 0 || error == EWOULDBLOCK)
592 *bytes_written = datalen - uio_resid(auio);
593
594 /*
595 * sosend returns EWOULDBLOCK if it's a non-blocking
596 * socket or a timeout occured (this allows to return
597 * the amount of queued data through sendit()).
598 *
599 * However, connectx() returns EINPROGRESS in case of a
600 * blocking socket. So we change the return value here.
601 */
602 if (error == EWOULDBLOCK)
603 error = EINPROGRESS;
604 }
605
606 if (error == 0 && pcid != NULL)
607 *pcid = 1; /* there is only one connection in regular TCP */
608
609 return (error);
610 }
611
612 static int
613 tcp_usr_connectx(struct socket *so, struct sockaddr *src,
614 struct sockaddr *dst, struct proc *p, uint32_t ifscope,
615 sae_associd_t aid, sae_connid_t *pcid, uint32_t flags, void *arg,
616 uint32_t arglen, struct uio *uio, user_ssize_t *bytes_written)
617 {
618 return (tcp_usr_connectx_common(so, AF_INET, src, dst, p, ifscope, aid,
619 pcid, flags, arg, arglen, uio, bytes_written));
620 }
621
622 #if INET6
623 static int
624 tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
625 {
626 int error = 0;
627 struct inpcb *inp = sotoinpcb(so);
628 struct tcpcb *tp;
629 struct sockaddr_in6 *sin6p;
630
631 TCPDEBUG0;
632 if (inp == NULL) {
633 return EINVAL;
634 } else if (inp->inp_state == INPCB_STATE_DEAD) {
635 if (so->so_error) {
636 error = so->so_error;
637 so->so_error = 0;
638 return error;
639 } else
640 return EINVAL;
641 }
642 #if NECP
643 #if FLOW_DIVERT
644 else if (necp_socket_should_use_flow_divert(inp)) {
645 uint32_t fd_ctl_unit = necp_socket_get_flow_divert_control_unit(inp);
646 if (fd_ctl_unit > 0) {
647 error = flow_divert_pcb_init(so, fd_ctl_unit);
648 if (error == 0) {
649 error = flow_divert_connect_out(so, nam, p);
650 }
651 } else {
652 error = ENETDOWN;
653 }
654
655 return error;
656 }
657 #endif /* FLOW_DIVERT */
658 #if CONTENT_FILTER
659 error = cfil_sock_attach(so);
660 if (error != 0)
661 return error;
662 #endif /* CONTENT_FILTER */
663 #endif /* NECP */
664
665 tp = intotcpcb(inp);
666 TCPDEBUG1();
667
668 calculate_tcp_clock();
669
670 if (nam->sa_family != 0 && nam->sa_family != AF_INET6) {
671 error = EAFNOSUPPORT;
672 goto out;
673 }
674
675 /*
676 * Must disallow TCP ``connections'' to multicast addresses.
677 */
678 sin6p = (struct sockaddr_in6 *)(void *)nam;
679 if (sin6p->sin6_family == AF_INET6
680 && IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
681 error = EAFNOSUPPORT;
682 goto out;
683 }
684
685 if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
686 struct sockaddr_in sin;
687
688 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)
689 return (EINVAL);
690
691 in6_sin6_2_sin(&sin, sin6p);
692 inp->inp_vflag |= INP_IPV4;
693 inp->inp_vflag &= ~INP_IPV6;
694 if ((error = tcp_connect(tp, (struct sockaddr *)&sin, p)) != 0)
695 goto out;
696
697 error = tcp_connect_complete(so);
698 goto out;
699 }
700 inp->inp_vflag &= ~INP_IPV4;
701 inp->inp_vflag |= INP_IPV6;
702 if ((error = tcp6_connect(tp, nam, p)) != 0)
703 goto out;
704
705 error = tcp_connect_complete(so);
706 COMMON_END(PRU_CONNECT);
707 }
708
709 static int
710 tcp6_usr_connectx(struct socket *so, struct sockaddr*src,
711 struct sockaddr *dst, struct proc *p, uint32_t ifscope,
712 sae_associd_t aid, sae_connid_t *pcid, uint32_t flags, void *arg,
713 uint32_t arglen, struct uio *uio, user_ssize_t *bytes_written)
714 {
715 return (tcp_usr_connectx_common(so, AF_INET6, src, dst, p, ifscope, aid,
716 pcid, flags, arg, arglen, uio, bytes_written));
717 }
718 #endif /* INET6 */
719
720 /*
721 * Initiate disconnect from peer.
722 * If connection never passed embryonic stage, just drop;
723 * else if don't need to let data drain, then can just drop anyways,
724 * else have to begin TCP shutdown process: mark socket disconnecting,
725 * drain unread data, state switch to reflect user close, and
726 * send segment (e.g. FIN) to peer. Socket will be really disconnected
727 * when peer sends FIN and acks ours.
728 *
729 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB.
730 */
731 static int
732 tcp_usr_disconnect(struct socket *so)
733 {
734 int error = 0;
735 struct inpcb *inp = sotoinpcb(so);
736 struct tcpcb *tp;
737
738 lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx,
739 LCK_MTX_ASSERT_OWNED);
740 COMMON_START();
741 /* In case we got disconnected from the peer */
742 if (tp == NULL)
743 goto out;
744 tp = tcp_disconnect(tp);
745 COMMON_END(PRU_DISCONNECT);
746 }
747
748 /*
749 * User-protocol pru_disconnectx callback.
750 */
751 static int
752 tcp_usr_disconnectx(struct socket *so, sae_associd_t aid, sae_connid_t cid)
753 {
754 #pragma unused(cid)
755 if (aid != SAE_ASSOCID_ANY && aid != SAE_ASSOCID_ALL)
756 return (EINVAL);
757
758 return (tcp_usr_disconnect(so));
759 }
760
761 /*
762 * Accept a connection. Essentially all the work is
763 * done at higher levels; just return the address
764 * of the peer, storing through addr.
765 */
766 static int
767 tcp_usr_accept(struct socket *so, struct sockaddr **nam)
768 {
769 int error = 0;
770 struct inpcb *inp = sotoinpcb(so);
771 struct tcpcb *tp = NULL;
772 TCPDEBUG0;
773
774 in_getpeeraddr(so, nam);
775
776 if (so->so_state & SS_ISDISCONNECTED) {
777 error = ECONNABORTED;
778 goto out;
779 }
780 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD)
781 return (EINVAL);
782 #if NECP
783 else if (necp_socket_should_use_flow_divert(inp))
784 return (EPROTOTYPE);
785 #if CONTENT_FILTER
786 error = cfil_sock_attach(so);
787 if (error != 0)
788 return (error);
789 #endif /* CONTENT_FILTER */
790 #endif /* NECP */
791
792 tp = intotcpcb(inp);
793 TCPDEBUG1();
794
795 calculate_tcp_clock();
796
797 COMMON_END(PRU_ACCEPT);
798 }
799
800 #if INET6
801 static int
802 tcp6_usr_accept(struct socket *so, struct sockaddr **nam)
803 {
804 int error = 0;
805 struct inpcb *inp = sotoinpcb(so);
806 struct tcpcb *tp = NULL;
807 TCPDEBUG0;
808
809 if (so->so_state & SS_ISDISCONNECTED) {
810 error = ECONNABORTED;
811 goto out;
812 }
813 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD)
814 return (EINVAL);
815 #if NECP
816 else if (necp_socket_should_use_flow_divert(inp))
817 return (EPROTOTYPE);
818 #if CONTENT_FILTER
819 error = cfil_sock_attach(so);
820 if (error != 0)
821 return (error);
822 #endif /* CONTENT_FILTER */
823 #endif /* NECP */
824
825 tp = intotcpcb(inp);
826 TCPDEBUG1();
827
828 calculate_tcp_clock();
829
830 in6_mapped_peeraddr(so, nam);
831 COMMON_END(PRU_ACCEPT);
832 }
833 #endif /* INET6 */
834
835 /*
836 * Mark the connection as being incapable of further output.
837 *
838 * Returns: 0 Success
839 * EINVAL [COMMON_START]
840 * tcp_output:EADDRNOTAVAIL
841 * tcp_output:ENOBUFS
842 * tcp_output:EMSGSIZE
843 * tcp_output:EHOSTUNREACH
844 * tcp_output:ENETUNREACH
845 * tcp_output:ENETDOWN
846 * tcp_output:ENOMEM
847 * tcp_output:EACCES
848 * tcp_output:EMSGSIZE
849 * tcp_output:ENOBUFS
850 * tcp_output:??? [ignorable: mostly IPSEC/firewall/DLIL]
851 */
852 static int
853 tcp_usr_shutdown(struct socket *so)
854 {
855 int error = 0;
856 struct inpcb *inp = sotoinpcb(so);
857 struct tcpcb *tp;
858
859 TCPDEBUG0;
860 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD)
861 return (EINVAL);
862
863 socantsendmore(so);
864
865 /*
866 * In case we got disconnected from the peer, or if this is
867 * a socket that is to be flow-diverted (but not yet).
868 */
869 tp = intotcpcb(inp);
870 TCPDEBUG1();
871
872 if (tp == NULL
873 #if NECP
874 || (necp_socket_should_use_flow_divert(inp))
875 #endif /* NECP */
876 ) {
877 if (tp != NULL)
878 error = EPROTOTYPE;
879 goto out;
880 }
881
882 calculate_tcp_clock();
883
884 tp = tcp_usrclosed(tp);
885 #if MPTCP
886 /* A reset has been sent but socket exists, do not send FIN */
887 if ((so->so_flags & SOF_MP_SUBFLOW) &&
888 (tp) && (tp->t_mpflags & TMPF_RESET)) {
889 goto out;
890 }
891 #endif
892 #if CONTENT_FILTER
893 /* Don't send a FIN yet */
894 if (tp && !(so->so_state & SS_ISDISCONNECTED) &&
895 cfil_sock_data_pending(&so->so_snd))
896 goto out;
897 #endif /* CONTENT_FILTER */
898 if (tp)
899 error = tcp_output(tp);
900 COMMON_END(PRU_SHUTDOWN);
901 }
902
903 /*
904 * After a receive, possibly send window update to peer.
905 */
906 static int
907 tcp_usr_rcvd(struct socket *so, __unused int flags)
908 {
909 int error = 0;
910 struct inpcb *inp = sotoinpcb(so);
911 struct tcpcb *tp;
912
913 COMMON_START();
914 /* In case we got disconnected from the peer */
915 if (tp == NULL)
916 goto out;
917 tcp_sbrcv_trim(tp, &so->so_rcv);
918
919 /*
920 * This tcp_output is solely there to trigger window-updates.
921 * However, we really do not want these window-updates while we
922 * are still in SYN_SENT or SYN_RECEIVED.
923 */
924 if (TCPS_HAVEESTABLISHED(tp->t_state))
925 tcp_output(tp);
926
927 #if CONTENT_FILTER
928 cfil_sock_buf_update(&so->so_rcv);
929 #endif /* CONTENT_FILTER */
930
931 COMMON_END(PRU_RCVD);
932 }
933
934 /*
935 * Do a send by putting data in output queue and updating urgent
936 * marker if URG set. Possibly send more data. Unlike the other
937 * pru_*() routines, the mbuf chains are our responsibility. We
938 * must either enqueue them or free them. The other pru_* routines
939 * generally are caller-frees.
940 *
941 * Returns: 0 Success
942 * ECONNRESET
943 * EINVAL
944 * ENOBUFS
945 * tcp_connect:EADDRINUSE Address in use
946 * tcp_connect:EADDRNOTAVAIL Address not available.
947 * tcp_connect:EINVAL Invalid argument
948 * tcp_connect:EAFNOSUPPORT Address family not supported [notdef]
949 * tcp_connect:EACCES Permission denied
950 * tcp_connect:EAGAIN Resource unavailable, try again
951 * tcp_connect:EPERM Operation not permitted
952 * tcp_output:EADDRNOTAVAIL
953 * tcp_output:ENOBUFS
954 * tcp_output:EMSGSIZE
955 * tcp_output:EHOSTUNREACH
956 * tcp_output:ENETUNREACH
957 * tcp_output:ENETDOWN
958 * tcp_output:ENOMEM
959 * tcp_output:EACCES
960 * tcp_output:EMSGSIZE
961 * tcp_output:ENOBUFS
962 * tcp_output:??? [ignorable: mostly IPSEC/firewall/DLIL]
963 * tcp6_connect:??? [IPV6 only]
964 */
965 static int
966 tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
967 struct sockaddr *nam, struct mbuf *control, struct proc *p)
968 {
969 int error = 0;
970 struct inpcb *inp = sotoinpcb(so);
971 struct tcpcb *tp;
972 uint32_t msgpri = MSG_PRI_DEFAULT;
973 #if INET6
974 int isipv6;
975 #endif
976 TCPDEBUG0;
977
978 if (inp == NULL || inp->inp_state == INPCB_STATE_DEAD
979 #if NECP
980 || (necp_socket_should_use_flow_divert(inp))
981 #endif /* NECP */
982 ) {
983 /*
984 * OOPS! we lost a race, the TCP session got reset after
985 * we checked SS_CANTSENDMORE, eg: while doing uiomove or a
986 * network interrupt in the non-splnet() section of sosend().
987 */
988 if (m != NULL)
989 m_freem(m);
990 if (control != NULL) {
991 m_freem(control);
992 control = NULL;
993 }
994
995 if (inp == NULL)
996 error = ECONNRESET; /* XXX EPIPE? */
997 else
998 error = EPROTOTYPE;
999 tp = NULL;
1000 TCPDEBUG1();
1001 goto out;
1002 }
1003 #if INET6
1004 isipv6 = nam && nam->sa_family == AF_INET6;
1005 #endif /* INET6 */
1006 tp = intotcpcb(inp);
1007 TCPDEBUG1();
1008
1009 calculate_tcp_clock();
1010
1011 if (control != NULL) {
1012 if (so->so_flags & SOF_ENABLE_MSGS) {
1013 /* Get the msg priority from control mbufs */
1014 error = tcp_get_msg_priority(control, &msgpri);
1015 if (error) {
1016 m_freem(control);
1017 if (m != NULL)
1018 m_freem(m);
1019 control = NULL;
1020 m = NULL;
1021 goto out;
1022 }
1023 m_freem(control);
1024 control = NULL;
1025 } else if (control->m_len) {
1026 /*
1027 * if not unordered, TCP should not have
1028 * control mbufs
1029 */
1030 m_freem(control);
1031 if (m != NULL)
1032 m_freem(m);
1033 control = NULL;
1034 m = NULL;
1035 error = EINVAL;
1036 goto out;
1037 }
1038 }
1039
1040 if (so->so_flags & SOF_ENABLE_MSGS) {
1041 VERIFY(m->m_flags & M_PKTHDR);
1042 m->m_pkthdr.msg_pri = msgpri;
1043 }
1044
1045 /* MPTCP sublow socket buffers must not be compressed */
1046 VERIFY(!(so->so_flags & SOF_MP_SUBFLOW) ||
1047 (so->so_snd.sb_flags & SB_NOCOMPRESS));
1048
1049 if(!(flags & PRUS_OOB) || (so->so_flags1 & SOF1_PRECONNECT_DATA)) {
1050 /* Call msg send if message delivery is enabled */
1051 if (so->so_flags & SOF_ENABLE_MSGS)
1052 sbappendmsg_snd(&so->so_snd, m);
1053 else
1054 sbappendstream(&so->so_snd, m);
1055
1056 if (nam && tp->t_state < TCPS_SYN_SENT) {
1057
1058 /*
1059 * Do implied connect if not yet connected,
1060 * initialize window to default value, and
1061 * initialize maxseg/maxopd using peer's cached
1062 * MSS.
1063 */
1064 #if INET6
1065 if (isipv6)
1066 error = tcp6_connect(tp, nam, p);
1067 else
1068 #endif /* INET6 */
1069 error = tcp_connect(tp, nam, p);
1070 if (error)
1071 goto out;
1072 tp->snd_wnd = TTCP_CLIENT_SND_WND;
1073 tcp_mss(tp, -1, IFSCOPE_NONE);
1074 }
1075
1076 if (flags & PRUS_EOF) {
1077 /*
1078 * Close the send side of the connection after
1079 * the data is sent.
1080 */
1081 socantsendmore(so);
1082 tp = tcp_usrclosed(tp);
1083 }
1084 if (tp != NULL) {
1085 if (flags & PRUS_MORETOCOME)
1086 tp->t_flags |= TF_MORETOCOME;
1087 error = tcp_output(tp);
1088 if (flags & PRUS_MORETOCOME)
1089 tp->t_flags &= ~TF_MORETOCOME;
1090 }
1091 } else {
1092 if (sbspace(&so->so_snd) == 0) {
1093 /* if no space is left in sockbuf,
1094 * do not try to squeeze in OOB traffic */
1095 m_freem(m);
1096 error = ENOBUFS;
1097 goto out;
1098 }
1099 /*
1100 * According to RFC961 (Assigned Protocols),
1101 * the urgent pointer points to the last octet
1102 * of urgent data. We continue, however,
1103 * to consider it to indicate the first octet
1104 * of data past the urgent section.
1105 * Otherwise, snd_up should be one lower.
1106 */
1107 sbappendstream(&so->so_snd, m);
1108 if (nam && tp->t_state < TCPS_SYN_SENT) {
1109 /*
1110 * Do implied connect if not yet connected,
1111 * initialize window to default value, and
1112 * initialize maxseg/maxopd using peer's cached
1113 * MSS.
1114 */
1115 #if INET6
1116 if (isipv6)
1117 error = tcp6_connect(tp, nam, p);
1118 else
1119 #endif /* INET6 */
1120 error = tcp_connect(tp, nam, p);
1121 if (error)
1122 goto out;
1123 tp->snd_wnd = TTCP_CLIENT_SND_WND;
1124 tcp_mss(tp, -1, IFSCOPE_NONE);
1125 }
1126 tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
1127 tp->t_flagsext |= TF_FORCE;
1128 error = tcp_output(tp);
1129 tp->t_flagsext &= ~TF_FORCE;
1130 }
1131
1132
1133 /*
1134 * We wait for the socket to successfully connect before returning.
1135 * This allows us to signal a timeout to the application.
1136 */
1137 if (so->so_state & SS_ISCONNECTING) {
1138 if (so->so_state & SS_NBIO)
1139 error = EWOULDBLOCK;
1140 else
1141 error = sbwait(&so->so_snd);
1142 }
1143
1144 COMMON_END((flags & PRUS_OOB) ? PRU_SENDOOB :
1145 ((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND));
1146 }
1147
1148 /*
1149 * Abort the TCP.
1150 */
1151 static int
1152 tcp_usr_abort(struct socket *so)
1153 {
1154 int error = 0;
1155 struct inpcb *inp = sotoinpcb(so);
1156 struct tcpcb *tp;
1157
1158 COMMON_START();
1159 /* In case we got disconnected from the peer */
1160 if (tp == NULL)
1161 goto out;
1162 tp = tcp_drop(tp, ECONNABORTED);
1163 VERIFY(so->so_usecount > 0);
1164 so->so_usecount--;
1165 COMMON_END(PRU_ABORT);
1166 }
1167
1168 /*
1169 * Receive out-of-band data.
1170 *
1171 * Returns: 0 Success
1172 * EINVAL [COMMON_START]
1173 * EINVAL
1174 * EWOULDBLOCK
1175 */
1176 static int
1177 tcp_usr_rcvoob(struct socket *so, struct mbuf *m, int flags)
1178 {
1179 int error = 0;
1180 struct inpcb *inp = sotoinpcb(so);
1181 struct tcpcb *tp;
1182
1183 COMMON_START();
1184 if ((so->so_oobmark == 0 &&
1185 (so->so_state & SS_RCVATMARK) == 0) ||
1186 so->so_options & SO_OOBINLINE ||
1187 tp->t_oobflags & TCPOOB_HADDATA) {
1188 error = EINVAL;
1189 goto out;
1190 }
1191 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) {
1192 error = EWOULDBLOCK;
1193 goto out;
1194 }
1195 m->m_len = 1;
1196 *mtod(m, caddr_t) = tp->t_iobc;
1197 so->so_state &= ~SS_RCVATMARK;
1198 if ((flags & MSG_PEEK) == 0)
1199 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA);
1200 COMMON_END(PRU_RCVOOB);
1201 }
1202
1203 static int
1204 tcp_usr_preconnect(struct socket *so)
1205 {
1206 struct inpcb *inp = sotoinpcb(so);
1207 int error = 0;
1208
1209 #if NECP
1210 if (necp_socket_should_use_flow_divert(inp)) {
1211 /* May happen, if in tcp_usr_connect we did not had a chance
1212 * to set the usrreqs (due to some error). So, let's get out
1213 * of here.
1214 */
1215 goto out;
1216 }
1217 #endif /* NECP */
1218
1219 error = tcp_output(sototcpcb(so));
1220
1221 /* One read has been done. This was enough. Get back to "normal" behavior. */
1222 so->so_flags1 &= ~SOF1_PRECONNECT_DATA;
1223
1224 COMMON_END(PRU_PRECONNECT);
1225 }
1226
1227 /* xxx - should be const */
1228 struct pr_usrreqs tcp_usrreqs = {
1229 .pru_abort = tcp_usr_abort,
1230 .pru_accept = tcp_usr_accept,
1231 .pru_attach = tcp_usr_attach,
1232 .pru_bind = tcp_usr_bind,
1233 .pru_connect = tcp_usr_connect,
1234 .pru_connectx = tcp_usr_connectx,
1235 .pru_control = in_control,
1236 .pru_detach = tcp_usr_detach,
1237 .pru_disconnect = tcp_usr_disconnect,
1238 .pru_disconnectx = tcp_usr_disconnectx,
1239 .pru_listen = tcp_usr_listen,
1240 .pru_peeraddr = in_getpeeraddr,
1241 .pru_rcvd = tcp_usr_rcvd,
1242 .pru_rcvoob = tcp_usr_rcvoob,
1243 .pru_send = tcp_usr_send,
1244 .pru_shutdown = tcp_usr_shutdown,
1245 .pru_sockaddr = in_getsockaddr,
1246 .pru_sosend = sosend,
1247 .pru_soreceive = soreceive,
1248 .pru_preconnect = tcp_usr_preconnect,
1249 };
1250
1251 #if INET6
1252 struct pr_usrreqs tcp6_usrreqs = {
1253 .pru_abort = tcp_usr_abort,
1254 .pru_accept = tcp6_usr_accept,
1255 .pru_attach = tcp_usr_attach,
1256 .pru_bind = tcp6_usr_bind,
1257 .pru_connect = tcp6_usr_connect,
1258 .pru_connectx = tcp6_usr_connectx,
1259 .pru_control = in6_control,
1260 .pru_detach = tcp_usr_detach,
1261 .pru_disconnect = tcp_usr_disconnect,
1262 .pru_disconnectx = tcp_usr_disconnectx,
1263 .pru_listen = tcp6_usr_listen,
1264 .pru_peeraddr = in6_mapped_peeraddr,
1265 .pru_rcvd = tcp_usr_rcvd,
1266 .pru_rcvoob = tcp_usr_rcvoob,
1267 .pru_send = tcp_usr_send,
1268 .pru_shutdown = tcp_usr_shutdown,
1269 .pru_sockaddr = in6_mapped_sockaddr,
1270 .pru_sosend = sosend,
1271 .pru_soreceive = soreceive,
1272 .pru_preconnect = tcp_usr_preconnect,
1273 };
1274 #endif /* INET6 */
1275
1276 /*
1277 * Common subroutine to open a TCP connection to remote host specified
1278 * by struct sockaddr_in in mbuf *nam. Call in_pcbbind to assign a local
1279 * port number if needed. Call in_pcbladdr to do the routing and to choose
1280 * a local host address (interface). If there is an existing incarnation
1281 * of the same connection in TIME-WAIT state and if the remote host was
1282 * sending CC options and if the connection duration was < MSL, then
1283 * truncate the previous TIME-WAIT state and proceed.
1284 * Initialize connection parameters and enter SYN-SENT state.
1285 *
1286 * Returns: 0 Success
1287 * EADDRINUSE
1288 * EINVAL
1289 * in_pcbbind:EADDRNOTAVAIL Address not available.
1290 * in_pcbbind:EINVAL Invalid argument
1291 * in_pcbbind:EAFNOSUPPORT Address family not supported [notdef]
1292 * in_pcbbind:EACCES Permission denied
1293 * in_pcbbind:EADDRINUSE Address in use
1294 * in_pcbbind:EAGAIN Resource unavailable, try again
1295 * in_pcbbind:EPERM Operation not permitted
1296 * in_pcbladdr:EINVAL Invalid argument
1297 * in_pcbladdr:EAFNOSUPPORT Address family not supported
1298 * in_pcbladdr:EADDRNOTAVAIL Address not available
1299 */
1300 static int
1301 tcp_connect(struct tcpcb *tp, struct sockaddr *nam, struct proc *p)
1302 {
1303 struct inpcb *inp = tp->t_inpcb, *oinp;
1304 struct socket *so = inp->inp_socket;
1305 struct tcpcb *otp;
1306 struct sockaddr_in *sin = (struct sockaddr_in *)(void *)nam;
1307 struct in_addr laddr;
1308 int error = 0;
1309 struct ifnet *outif = NULL;
1310
1311 if (inp->inp_lport == 0) {
1312 error = in_pcbbind(inp, NULL, p);
1313 if (error)
1314 goto done;
1315 }
1316
1317 /*
1318 * Cannot simply call in_pcbconnect, because there might be an
1319 * earlier incarnation of this same connection still in
1320 * TIME_WAIT state, creating an ADDRINUSE error.
1321 */
1322 error = in_pcbladdr(inp, nam, &laddr, IFSCOPE_NONE, &outif, 0);
1323 if (error)
1324 goto done;
1325
1326 tcp_unlock(inp->inp_socket, 0, 0);
1327 oinp = in_pcblookup_hash(inp->inp_pcbinfo,
1328 sin->sin_addr, sin->sin_port,
1329 inp->inp_laddr.s_addr != INADDR_ANY ? inp->inp_laddr : laddr,
1330 inp->inp_lport, 0, NULL);
1331
1332 tcp_lock(inp->inp_socket, 0, 0);
1333 if (oinp) {
1334 if (oinp != inp) /* 4143933: avoid deadlock if inp == oinp */
1335 tcp_lock(oinp->inp_socket, 1, 0);
1336 if (in_pcb_checkstate(oinp, WNT_RELEASE, 1) == WNT_STOPUSING) {
1337 if (oinp != inp)
1338 tcp_unlock(oinp->inp_socket, 1, 0);
1339 goto skip_oinp;
1340 }
1341
1342 if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
1343 otp->t_state == TCPS_TIME_WAIT &&
1344 ((int)(tcp_now - otp->t_starttime)) < tcp_msl &&
1345 (otp->t_flags & TF_RCVD_CC)) {
1346 otp = tcp_close(otp);
1347 } else {
1348 printf("tcp_connect: inp=0x%llx err=EADDRINUSE\n",
1349 (uint64_t)VM_KERNEL_ADDRPERM(inp));
1350 if (oinp != inp)
1351 tcp_unlock(oinp->inp_socket, 1, 0);
1352 error = EADDRINUSE;
1353 goto done;
1354 }
1355 if (oinp != inp)
1356 tcp_unlock(oinp->inp_socket, 1, 0);
1357 }
1358 skip_oinp:
1359 if ((inp->inp_laddr.s_addr == INADDR_ANY ? laddr.s_addr :
1360 inp->inp_laddr.s_addr) == sin->sin_addr.s_addr &&
1361 inp->inp_lport == sin->sin_port) {
1362 error = EINVAL;
1363 goto done;
1364 }
1365 if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->ipi_lock)) {
1366 /*lock inversion issue, mostly with udp multicast packets */
1367 socket_unlock(inp->inp_socket, 0);
1368 lck_rw_lock_exclusive(inp->inp_pcbinfo->ipi_lock);
1369 socket_lock(inp->inp_socket, 0);
1370 }
1371 if (inp->inp_laddr.s_addr == INADDR_ANY) {
1372 inp->inp_laddr = laddr;
1373 /* no reference needed */
1374 inp->inp_last_outifp = outif;
1375 inp->inp_flags |= INP_INADDR_ANY;
1376 }
1377 inp->inp_faddr = sin->sin_addr;
1378 inp->inp_fport = sin->sin_port;
1379 in_pcbrehash(inp);
1380 lck_rw_done(inp->inp_pcbinfo->ipi_lock);
1381
1382 if (inp->inp_flowhash == 0)
1383 inp->inp_flowhash = inp_calc_flowhash(inp);
1384
1385 tcp_set_max_rwinscale(tp, so, TCP_AUTORCVBUF_MAX(outif));
1386
1387 soisconnecting(so);
1388 tcpstat.tcps_connattempt++;
1389 tp->t_state = TCPS_SYN_SENT;
1390 tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp, TCP_CONN_KEEPINIT(tp));
1391 tp->iss = tcp_new_isn(tp);
1392 tcp_sendseqinit(tp);
1393 if (nstat_collect)
1394 nstat_route_connect_attempt(inp->inp_route.ro_rt);
1395
1396 done:
1397 if (outif != NULL)
1398 ifnet_release(outif);
1399
1400 return (error);
1401 }
1402
1403 #if INET6
1404 static int
1405 tcp6_connect(struct tcpcb *tp, struct sockaddr *nam, struct proc *p)
1406 {
1407 struct inpcb *inp = tp->t_inpcb, *oinp;
1408 struct socket *so = inp->inp_socket;
1409 struct tcpcb *otp;
1410 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)(void *)nam;
1411 struct in6_addr addr6;
1412 int error = 0;
1413 struct ifnet *outif = NULL;
1414
1415 if (inp->inp_lport == 0) {
1416 error = in6_pcbbind(inp, NULL, p);
1417 if (error)
1418 goto done;
1419 }
1420
1421 /*
1422 * Cannot simply call in_pcbconnect, because there might be an
1423 * earlier incarnation of this same connection still in
1424 * TIME_WAIT state, creating an ADDRINUSE error.
1425 *
1426 * in6_pcbladdr() might return an ifp with its reference held
1427 * even in the error case, so make sure that it's released
1428 * whenever it's non-NULL.
1429 */
1430 error = in6_pcbladdr(inp, nam, &addr6, &outif);
1431 if (error)
1432 goto done;
1433 tcp_unlock(inp->inp_socket, 0, 0);
1434 oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
1435 &sin6->sin6_addr, sin6->sin6_port,
1436 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
1437 ? &addr6
1438 : &inp->in6p_laddr,
1439 inp->inp_lport, 0, NULL);
1440 tcp_lock(inp->inp_socket, 0, 0);
1441 if (oinp) {
1442 if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
1443 otp->t_state == TCPS_TIME_WAIT &&
1444 ((int)(tcp_now - otp->t_starttime)) < tcp_msl &&
1445 (otp->t_flags & TF_RCVD_CC)) {
1446 otp = tcp_close(otp);
1447 } else {
1448 error = EADDRINUSE;
1449 goto done;
1450 }
1451 }
1452 if (!lck_rw_try_lock_exclusive(inp->inp_pcbinfo->ipi_lock)) {
1453 /*lock inversion issue, mostly with udp multicast packets */
1454 socket_unlock(inp->inp_socket, 0);
1455 lck_rw_lock_exclusive(inp->inp_pcbinfo->ipi_lock);
1456 socket_lock(inp->inp_socket, 0);
1457 }
1458 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1459 inp->in6p_laddr = addr6;
1460 inp->in6p_last_outifp = outif; /* no reference needed */
1461 inp->in6p_flags |= INP_IN6ADDR_ANY;
1462 }
1463 inp->in6p_faddr = sin6->sin6_addr;
1464 inp->inp_fport = sin6->sin6_port;
1465 if ((sin6->sin6_flowinfo & IPV6_FLOWINFO_MASK) != 0)
1466 inp->inp_flow = sin6->sin6_flowinfo;
1467 in_pcbrehash(inp);
1468 lck_rw_done(inp->inp_pcbinfo->ipi_lock);
1469
1470 if (inp->inp_flowhash == 0)
1471 inp->inp_flowhash = inp_calc_flowhash(inp);
1472 /* update flowinfo - RFC 6437 */
1473 if (inp->inp_flow == 0 && inp->in6p_flags & IN6P_AUTOFLOWLABEL) {
1474 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
1475 inp->inp_flow |=
1476 (htonl(inp->inp_flowhash) & IPV6_FLOWLABEL_MASK);
1477 }
1478
1479 tcp_set_max_rwinscale(tp, so, TCP_AUTORCVBUF_MAX(outif));
1480
1481 soisconnecting(so);
1482 tcpstat.tcps_connattempt++;
1483 tp->t_state = TCPS_SYN_SENT;
1484 tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp,
1485 TCP_CONN_KEEPINIT(tp));
1486 tp->iss = tcp_new_isn(tp);
1487 tcp_sendseqinit(tp);
1488 if (nstat_collect)
1489 nstat_route_connect_attempt(inp->inp_route.ro_rt);
1490
1491 done:
1492 if (outif != NULL)
1493 ifnet_release(outif);
1494
1495 return (error);
1496 }
1497 #endif /* INET6 */
1498
1499 /*
1500 * Export TCP internal state information via a struct tcp_info
1501 */
1502 __private_extern__ void
1503 tcp_fill_info(struct tcpcb *tp, struct tcp_info *ti)
1504 {
1505 struct inpcb *inp = tp->t_inpcb;
1506
1507 bzero(ti, sizeof(*ti));
1508
1509 ti->tcpi_state = tp->t_state;
1510 ti->tcpi_flowhash = inp->inp_flowhash;
1511
1512 if (tp->t_state > TCPS_LISTEN) {
1513 if (TSTMP_SUPPORTED(tp))
1514 ti->tcpi_options |= TCPI_OPT_TIMESTAMPS;
1515 if (SACK_ENABLED(tp))
1516 ti->tcpi_options |= TCPI_OPT_SACK;
1517 if (TCP_WINDOW_SCALE_ENABLED(tp)) {
1518 ti->tcpi_options |= TCPI_OPT_WSCALE;
1519 ti->tcpi_snd_wscale = tp->snd_scale;
1520 ti->tcpi_rcv_wscale = tp->rcv_scale;
1521 }
1522 if (TCP_ECN_ENABLED(tp))
1523 ti->tcpi_options |= TCPI_OPT_ECN;
1524
1525 /* Are we in retranmission episode */
1526 if (IN_FASTRECOVERY(tp) || tp->t_rxtshift > 0)
1527 ti->tcpi_flags |= TCPI_FLAG_LOSSRECOVERY;
1528
1529 if (tp->t_flags & TF_STREAMING_ON)
1530 ti->tcpi_flags |= TCPI_FLAG_STREAMING_ON;
1531
1532 ti->tcpi_rto = tp->t_timer[TCPT_REXMT] ? tp->t_rxtcur : 0;
1533 ti->tcpi_snd_mss = tp->t_maxseg;
1534 ti->tcpi_rcv_mss = tp->t_maxseg;
1535
1536 ti->tcpi_rttcur = tp->t_rttcur;
1537 ti->tcpi_srtt = tp->t_srtt >> TCP_RTT_SHIFT;
1538 ti->tcpi_rttvar = tp->t_rttvar >> TCP_RTTVAR_SHIFT;
1539 ti->tcpi_rttbest = tp->t_rttbest >> TCP_RTT_SHIFT;
1540
1541 ti->tcpi_snd_ssthresh = tp->snd_ssthresh;
1542 ti->tcpi_snd_cwnd = tp->snd_cwnd;
1543 ti->tcpi_snd_sbbytes = inp->inp_socket->so_snd.sb_cc;
1544
1545 ti->tcpi_rcv_space = tp->rcv_wnd;
1546
1547 ti->tcpi_snd_wnd = tp->snd_wnd;
1548 ti->tcpi_snd_nxt = tp->snd_nxt;
1549 ti->tcpi_rcv_nxt = tp->rcv_nxt;
1550
1551 /* convert bytes/msec to bits/sec */
1552 if ((tp->t_flagsext & TF_MEASURESNDBW) != 0 &&
1553 tp->t_bwmeas != NULL) {
1554 ti->tcpi_snd_bw = (tp->t_bwmeas->bw_sndbw * 8000);
1555 }
1556
1557 ti->tcpi_last_outif = (tp->t_inpcb->inp_last_outifp == NULL) ? 0 :
1558 tp->t_inpcb->inp_last_outifp->if_index;
1559
1560 //atomic_get_64(ti->tcpi_txbytes, &inp->inp_stat->txbytes);
1561 ti->tcpi_txpackets = inp->inp_stat->txpackets;
1562 ti->tcpi_txbytes = inp->inp_stat->txbytes;
1563 ti->tcpi_txretransmitbytes = tp->t_stat.txretransmitbytes;
1564 ti->tcpi_txretransmitpackets = tp->t_stat.rxmitpkts;
1565 ti->tcpi_txunacked = tp->snd_max - tp->snd_una;
1566
1567 //atomic_get_64(ti->tcpi_rxbytes, &inp->inp_stat->rxbytes);
1568 ti->tcpi_rxpackets = inp->inp_stat->rxpackets;
1569 ti->tcpi_rxbytes = inp->inp_stat->rxbytes;
1570 ti->tcpi_rxduplicatebytes = tp->t_stat.rxduplicatebytes;
1571 ti->tcpi_rxoutoforderbytes = tp->t_stat.rxoutoforderbytes;
1572
1573 if (tp->t_state > TCPS_LISTEN) {
1574 ti->tcpi_synrexmits = tp->t_stat.synrxtshift;
1575 }
1576 ti->tcpi_cell_rxpackets = inp->inp_cstat->rxpackets;
1577 ti->tcpi_cell_rxbytes = inp->inp_cstat->rxbytes;
1578 ti->tcpi_cell_txpackets = inp->inp_cstat->txpackets;
1579 ti->tcpi_cell_txbytes = inp->inp_cstat->txbytes;
1580
1581 ti->tcpi_wifi_rxpackets = inp->inp_wstat->rxpackets;
1582 ti->tcpi_wifi_rxbytes = inp->inp_wstat->rxbytes;
1583 ti->tcpi_wifi_txpackets = inp->inp_wstat->txpackets;
1584 ti->tcpi_wifi_txbytes = inp->inp_wstat->txbytes;
1585
1586 ti->tcpi_wired_rxpackets = inp->inp_Wstat->rxpackets;
1587 ti->tcpi_wired_rxbytes = inp->inp_Wstat->rxbytes;
1588 ti->tcpi_wired_txpackets = inp->inp_Wstat->txpackets;
1589 ti->tcpi_wired_txbytes = inp->inp_Wstat->txbytes;
1590 tcp_get_connectivity_status(tp, &ti->tcpi_connstatus);
1591
1592 ti->tcpi_tfo_syn_data_rcv = !!(tp->t_tfo_stats & TFO_S_SYNDATA_RCV);
1593 ti->tcpi_tfo_cookie_req_rcv = !!(tp->t_tfo_stats & TFO_S_COOKIEREQ_RECV);
1594 ti->tcpi_tfo_cookie_sent = !!(tp->t_tfo_stats & TFO_S_COOKIE_SENT);
1595 ti->tcpi_tfo_cookie_invalid = !!(tp->t_tfo_stats & TFO_S_COOKIE_INVALID);
1596
1597 ti->tcpi_tfo_cookie_req = !!(tp->t_tfo_stats & TFO_S_COOKIE_REQ);
1598 ti->tcpi_tfo_cookie_rcv = !!(tp->t_tfo_stats & TFO_S_COOKIE_RCV);
1599 ti->tcpi_tfo_syn_data_sent = !!(tp->t_tfo_stats & TFO_S_SYN_DATA_SENT);
1600 ti->tcpi_tfo_syn_data_acked = !!(tp->t_tfo_stats & TFO_S_SYN_DATA_ACKED);
1601 ti->tcpi_tfo_syn_loss = !!(tp->t_tfo_stats & TFO_S_SYN_LOSS);
1602 ti->tcpi_tfo_cookie_wrong = !!(tp->t_tfo_stats & TFO_S_COOKIE_WRONG);
1603 ti->tcpi_tfo_no_cookie_rcv = !!(tp->t_tfo_stats & TFO_S_NO_COOKIE_RCV);
1604 ti->tcpi_tfo_heuristics_disable = !!(tp->t_tfo_stats & TFO_S_HEURISTICS_DISABLE);
1605 ti->tcpi_tfo_send_blackhole = !!(tp->t_tfo_stats & TFO_S_SEND_BLACKHOLE);
1606 ti->tcpi_tfo_recv_blackhole = !!(tp->t_tfo_stats & TFO_S_RECV_BLACKHOLE);
1607
1608 ti->tcpi_ecn_client_setup = !!(tp->ecn_flags & TE_SETUPSENT);
1609 ti->tcpi_ecn_server_setup = !!(tp->ecn_flags & TE_SETUPRECEIVED);
1610 ti->tcpi_ecn_success = (tp->ecn_flags & TE_ECN_ON) == TE_ECN_ON ? 1 : 0;
1611 ti->tcpi_ecn_lost_syn = !!(tp->ecn_flags & TE_LOST_SYN);
1612 ti->tcpi_ecn_lost_synack = !!(tp->ecn_flags & TE_LOST_SYNACK);
1613
1614 ti->tcpi_local_peer = !!(tp->t_flags & TF_LOCAL);
1615
1616 if (tp->t_inpcb->inp_last_outifp != NULL) {
1617 if (IFNET_IS_CELLULAR(tp->t_inpcb->inp_last_outifp))
1618 ti->tcpi_if_cell = 1;
1619 else if (IFNET_IS_WIFI(tp->t_inpcb->inp_last_outifp))
1620 ti->tcpi_if_wifi = 1;
1621 }
1622
1623 ti->tcpi_ecn_recv_ce = tp->t_ecn_recv_ce;
1624 ti->tcpi_ecn_recv_cwr = tp->t_ecn_recv_cwr;
1625
1626 ti->tcpi_rcvoopack = tp->t_rcvoopack;
1627 ti->tcpi_pawsdrop = tp->t_pawsdrop;
1628 ti->tcpi_sack_recovery_episode = tp->t_sack_recovery_episode;
1629 ti->tcpi_reordered_pkts = tp->t_reordered_pkts;
1630 ti->tcpi_dsack_sent = tp->t_dsack_sent;
1631 ti->tcpi_dsack_recvd = tp->t_dsack_recvd;
1632 }
1633 }
1634
1635 __private_extern__ errno_t
1636 tcp_fill_info_for_info_tuple(struct info_tuple *itpl, struct tcp_info *ti)
1637 {
1638 struct inpcbinfo *pcbinfo = NULL;
1639 struct inpcb *inp = NULL;
1640 struct socket *so;
1641 struct tcpcb *tp;
1642
1643 if (itpl->itpl_proto == IPPROTO_TCP)
1644 pcbinfo = &tcbinfo;
1645 else
1646 return EINVAL;
1647
1648 if (itpl->itpl_local_sa.sa_family == AF_INET &&
1649 itpl->itpl_remote_sa.sa_family == AF_INET) {
1650 inp = in_pcblookup_hash(pcbinfo,
1651 itpl->itpl_remote_sin.sin_addr,
1652 itpl->itpl_remote_sin.sin_port,
1653 itpl->itpl_local_sin.sin_addr,
1654 itpl->itpl_local_sin.sin_port,
1655 0, NULL);
1656 } else if (itpl->itpl_local_sa.sa_family == AF_INET6 &&
1657 itpl->itpl_remote_sa.sa_family == AF_INET6) {
1658 struct in6_addr ina6_local;
1659 struct in6_addr ina6_remote;
1660
1661 ina6_local = itpl->itpl_local_sin6.sin6_addr;
1662 if (IN6_IS_SCOPE_LINKLOCAL(&ina6_local) &&
1663 itpl->itpl_local_sin6.sin6_scope_id)
1664 ina6_local.s6_addr16[1] = htons(itpl->itpl_local_sin6.sin6_scope_id);
1665
1666 ina6_remote = itpl->itpl_remote_sin6.sin6_addr;
1667 if (IN6_IS_SCOPE_LINKLOCAL(&ina6_remote) &&
1668 itpl->itpl_remote_sin6.sin6_scope_id)
1669 ina6_remote.s6_addr16[1] = htons(itpl->itpl_remote_sin6.sin6_scope_id);
1670
1671 inp = in6_pcblookup_hash(pcbinfo,
1672 &ina6_remote,
1673 itpl->itpl_remote_sin6.sin6_port,
1674 &ina6_local,
1675 itpl->itpl_local_sin6.sin6_port,
1676 0, NULL);
1677 } else {
1678 return EINVAL;
1679 }
1680 if (inp == NULL || (so = inp->inp_socket) == NULL)
1681 return ENOENT;
1682
1683 socket_lock(so, 0);
1684 if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
1685 socket_unlock(so, 0);
1686 return ENOENT;
1687 }
1688 tp = intotcpcb(inp);
1689
1690 tcp_fill_info(tp, ti);
1691 socket_unlock(so, 0);
1692
1693 return 0;
1694 }
1695
1696 static void
1697 tcp_connection_fill_info(struct tcpcb *tp, struct tcp_connection_info *tci)
1698 {
1699 struct inpcb *inp = tp->t_inpcb;
1700
1701 bzero(tci, sizeof(*tci));
1702 tci->tcpi_state = tp->t_state;
1703 if (tp->t_state > TCPS_LISTEN) {
1704 if (TSTMP_SUPPORTED(tp))
1705 tci->tcpi_options |= TCPCI_OPT_TIMESTAMPS;
1706 if (SACK_ENABLED(tp))
1707 tci->tcpi_options |= TCPCI_OPT_SACK;
1708 if (TCP_WINDOW_SCALE_ENABLED(tp)) {
1709 tci->tcpi_options |= TCPCI_OPT_WSCALE;
1710 tci->tcpi_snd_wscale = tp->snd_scale;
1711 tci->tcpi_rcv_wscale = tp->rcv_scale;
1712 }
1713 if (TCP_ECN_ENABLED(tp))
1714 tci->tcpi_options |= TCPCI_OPT_ECN;
1715 if (IN_FASTRECOVERY(tp) || tp->t_rxtshift > 0)
1716 tci->tcpi_flags |= TCPCI_FLAG_LOSSRECOVERY;
1717 if (tp->t_flagsext & TF_PKTS_REORDERED)
1718 tci->tcpi_flags |= TCPCI_FLAG_REORDERING_DETECTED;
1719 tci->tcpi_rto = (tp->t_timer[TCPT_REXMT] > 0) ?
1720 tp->t_rxtcur : 0;
1721 tci->tcpi_maxseg = tp->t_maxseg;
1722 tci->tcpi_snd_ssthresh = tp->snd_ssthresh;
1723 tci->tcpi_snd_cwnd = tp->snd_cwnd;
1724 tci->tcpi_snd_wnd = tp->snd_wnd;
1725 tci->tcpi_snd_sbbytes = inp->inp_socket->so_snd.sb_cc;
1726 tci->tcpi_rcv_wnd = tp->rcv_wnd;
1727 tci->tcpi_rttcur = tp->t_rttcur;
1728 tci->tcpi_srtt = (tp->t_srtt >> TCP_RTT_SHIFT);
1729 tci->tcpi_rttvar = (tp->t_rttvar >> TCP_RTTVAR_SHIFT);
1730 tci->tcpi_txpackets = inp->inp_stat->txpackets;
1731 tci->tcpi_txbytes = inp->inp_stat->txbytes;
1732 tci->tcpi_txretransmitbytes = tp->t_stat.txretransmitbytes;
1733 tci->tcpi_txretransmitpackets = tp->t_stat.rxmitpkts;
1734 tci->tcpi_rxpackets = inp->inp_stat->rxpackets;
1735 tci->tcpi_rxbytes = inp->inp_stat->rxbytes;
1736 tci->tcpi_rxoutoforderbytes = tp->t_stat.rxoutoforderbytes;
1737
1738 tci->tcpi_tfo_syn_data_rcv = !!(tp->t_tfo_stats & TFO_S_SYNDATA_RCV);
1739 tci->tcpi_tfo_cookie_req_rcv = !!(tp->t_tfo_stats & TFO_S_COOKIEREQ_RECV);
1740 tci->tcpi_tfo_cookie_sent = !!(tp->t_tfo_stats & TFO_S_COOKIE_SENT);
1741 tci->tcpi_tfo_cookie_invalid = !!(tp->t_tfo_stats & TFO_S_COOKIE_INVALID);
1742 tci->tcpi_tfo_cookie_req = !!(tp->t_tfo_stats & TFO_S_COOKIE_REQ);
1743 tci->tcpi_tfo_cookie_rcv = !!(tp->t_tfo_stats & TFO_S_COOKIE_RCV);
1744 tci->tcpi_tfo_syn_data_sent = !!(tp->t_tfo_stats & TFO_S_SYN_DATA_SENT);
1745 tci->tcpi_tfo_syn_data_acked = !!(tp->t_tfo_stats & TFO_S_SYN_DATA_ACKED);
1746 tci->tcpi_tfo_syn_loss = !!(tp->t_tfo_stats & TFO_S_SYN_LOSS);
1747 tci->tcpi_tfo_cookie_wrong = !!(tp->t_tfo_stats & TFO_S_COOKIE_WRONG);
1748 tci->tcpi_tfo_no_cookie_rcv = !!(tp->t_tfo_stats & TFO_S_NO_COOKIE_RCV);
1749 tci->tcpi_tfo_heuristics_disable = !!(tp->t_tfo_stats & TFO_S_HEURISTICS_DISABLE);
1750 tci->tcpi_tfo_send_blackhole = !!(tp->t_tfo_stats & TFO_S_SEND_BLACKHOLE);
1751 tci->tcpi_tfo_recv_blackhole = !!(tp->t_tfo_stats & TFO_S_RECV_BLACKHOLE);
1752 }
1753 }
1754
1755
1756 __private_extern__ int
1757 tcp_sysctl_info(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1758 {
1759 int error;
1760 struct tcp_info ti;
1761 struct info_tuple itpl;
1762 proc_t caller = PROC_NULL;
1763 proc_t caller_parent = PROC_NULL;
1764 char command_name[MAXCOMLEN + 1] = "";
1765 char parent_name[MAXCOMLEN + 1] = "";
1766
1767 if ((caller = proc_self()) != PROC_NULL) {
1768 /* get process name */
1769 strlcpy(command_name, caller->p_comm, sizeof(command_name));
1770
1771 /* get parent process name if possible */
1772 if ((caller_parent = proc_find(caller->p_ppid)) != PROC_NULL) {
1773 strlcpy(parent_name, caller_parent->p_comm,
1774 sizeof(parent_name));
1775 proc_rele(caller_parent);
1776 }
1777
1778 if ((escape_str(command_name, strlen(command_name),
1779 sizeof(command_name)) == 0) &&
1780 (escape_str(parent_name, strlen(parent_name),
1781 sizeof(parent_name)) == 0)) {
1782 kern_asl_msg(LOG_DEBUG, "messagetracer",
1783 5,
1784 "com.apple.message.domain",
1785 "com.apple.kernel.tcpstat", /* 1 */
1786 "com.apple.message.signature",
1787 "tcpinfo", /* 2 */
1788 "com.apple.message.signature2", command_name, /* 3 */
1789 "com.apple.message.signature3", parent_name, /* 4 */
1790 "com.apple.message.summarize", "YES", /* 5 */
1791 NULL);
1792 }
1793 }
1794
1795 if (caller != PROC_NULL)
1796 proc_rele(caller);
1797
1798 if (req->newptr == USER_ADDR_NULL) {
1799 return EINVAL;
1800 }
1801 if (req->newlen < sizeof(struct info_tuple)) {
1802 return EINVAL;
1803 }
1804 error = SYSCTL_IN(req, &itpl, sizeof(struct info_tuple));
1805 if (error != 0) {
1806 return error;
1807 }
1808 error = tcp_fill_info_for_info_tuple(&itpl, &ti);
1809 if (error != 0) {
1810 return error;
1811 }
1812 error = SYSCTL_OUT(req, &ti, sizeof(struct tcp_info));
1813 if (error != 0) {
1814 return error;
1815 }
1816
1817 return 0;
1818 }
1819
1820 static int
1821 tcp_lookup_peer_pid_locked(struct socket *so, pid_t *out_pid)
1822 {
1823 int error = EHOSTUNREACH;
1824 *out_pid = -1;
1825 if ((so->so_state & SS_ISCONNECTED) == 0) return ENOTCONN;
1826
1827 struct inpcb *inp = (struct inpcb*)so->so_pcb;
1828 uint16_t lport = inp->inp_lport;
1829 uint16_t fport = inp->inp_fport;
1830 struct inpcb *finp = NULL;
1831
1832 if (inp->inp_vflag & INP_IPV6) {
1833 struct in6_addr laddr6 = inp->in6p_laddr;
1834 struct in6_addr faddr6 = inp->in6p_faddr;
1835 socket_unlock(so, 0);
1836 finp = in6_pcblookup_hash(&tcbinfo, &laddr6, lport, &faddr6, fport, 0, NULL);
1837 socket_lock(so, 0);
1838 } else if (inp->inp_vflag & INP_IPV4) {
1839 struct in_addr laddr4 = inp->inp_laddr;
1840 struct in_addr faddr4 = inp->inp_faddr;
1841 socket_unlock(so, 0);
1842 finp = in_pcblookup_hash(&tcbinfo, laddr4, lport, faddr4, fport, 0, NULL);
1843 socket_lock(so, 0);
1844 }
1845
1846 if (finp) {
1847 *out_pid = finp->inp_socket->last_pid;
1848 error = 0;
1849 in_pcb_checkstate(finp, WNT_RELEASE, 0);
1850 }
1851
1852 return error;
1853 }
1854
1855 void
1856 tcp_getconninfo(struct socket *so, struct conninfo_tcp *tcp_ci)
1857 {
1858 (void) tcp_lookup_peer_pid_locked(so, &tcp_ci->tcpci_peer_pid);
1859 tcp_fill_info(sototcpcb(so), &tcp_ci->tcpci_tcp_info);
1860 }
1861
1862 /*
1863 * The new sockopt interface makes it possible for us to block in the
1864 * copyin/out step (if we take a page fault). Taking a page fault at
1865 * splnet() is probably a Bad Thing. (Since sockets and pcbs both now
1866 * use TSM, there probably isn't any need for this function to run at
1867 * splnet() any more. This needs more examination.)
1868 */
1869 int
1870 tcp_ctloutput(struct socket *so, struct sockopt *sopt)
1871 {
1872 int error, opt, optval;
1873 struct inpcb *inp;
1874 struct tcpcb *tp;
1875
1876 error = 0;
1877 inp = sotoinpcb(so);
1878 if (inp == NULL) {
1879 return (ECONNRESET);
1880 }
1881 /* Allow <SOL_SOCKET,SO_FLUSH/SO_TRAFFIC_MGT_BACKGROUND> at this level */
1882 if (sopt->sopt_level != IPPROTO_TCP &&
1883 !(sopt->sopt_level == SOL_SOCKET && (sopt->sopt_name == SO_FLUSH ||
1884 sopt->sopt_name == SO_TRAFFIC_MGT_BACKGROUND))) {
1885 #if INET6
1886 if (SOCK_CHECK_DOM(so, PF_INET6))
1887 error = ip6_ctloutput(so, sopt);
1888 else
1889 #endif /* INET6 */
1890 error = ip_ctloutput(so, sopt);
1891 return (error);
1892 }
1893 tp = intotcpcb(inp);
1894 if (tp == NULL) {
1895 return (ECONNRESET);
1896 }
1897
1898 calculate_tcp_clock();
1899
1900 switch (sopt->sopt_dir) {
1901 case SOPT_SET:
1902 switch (sopt->sopt_name) {
1903 case TCP_NODELAY:
1904 case TCP_NOOPT:
1905 case TCP_NOPUSH:
1906 error = sooptcopyin(sopt, &optval, sizeof optval,
1907 sizeof optval);
1908 if (error)
1909 break;
1910
1911 switch (sopt->sopt_name) {
1912 case TCP_NODELAY:
1913 opt = TF_NODELAY;
1914 break;
1915 case TCP_NOOPT:
1916 opt = TF_NOOPT;
1917 break;
1918 case TCP_NOPUSH:
1919 opt = TF_NOPUSH;
1920 break;
1921 default:
1922 opt = 0; /* dead code to fool gcc */
1923 break;
1924 }
1925
1926 if (optval)
1927 tp->t_flags |= opt;
1928 else
1929 tp->t_flags &= ~opt;
1930 break;
1931 case TCP_RXT_FINDROP:
1932 case TCP_NOTIMEWAIT:
1933 error = sooptcopyin(sopt, &optval, sizeof optval,
1934 sizeof optval);
1935 if (error)
1936 break;
1937 switch (sopt->sopt_name) {
1938 case TCP_RXT_FINDROP:
1939 opt = TF_RXTFINDROP;
1940 break;
1941 case TCP_NOTIMEWAIT:
1942 opt = TF_NOTIMEWAIT;
1943 break;
1944 default:
1945 opt = 0;
1946 break;
1947 }
1948 if (optval)
1949 tp->t_flagsext |= opt;
1950 else
1951 tp->t_flagsext &= ~opt;
1952 break;
1953 case TCP_MEASURE_SND_BW:
1954 error = sooptcopyin(sopt, &optval, sizeof optval,
1955 sizeof optval);
1956 if (error)
1957 break;
1958 opt = TF_MEASURESNDBW;
1959 if (optval) {
1960 if (tp->t_bwmeas == NULL) {
1961 tp->t_bwmeas = tcp_bwmeas_alloc(tp);
1962 if (tp->t_bwmeas == NULL) {
1963 error = ENOMEM;
1964 break;
1965 }
1966 }
1967 tp->t_flagsext |= opt;
1968 } else {
1969 tp->t_flagsext &= ~opt;
1970 /* Reset snd bw measurement state */
1971 tp->t_flagsext &= ~(TF_BWMEAS_INPROGRESS);
1972 if (tp->t_bwmeas != NULL) {
1973 tcp_bwmeas_free(tp);
1974 }
1975 }
1976 break;
1977 case TCP_MEASURE_BW_BURST: {
1978 struct tcp_measure_bw_burst in;
1979 uint32_t minpkts, maxpkts;
1980 bzero(&in, sizeof(in));
1981
1982 error = sooptcopyin(sopt, &in, sizeof(in),
1983 sizeof(in));
1984 if (error)
1985 break;
1986 if ((tp->t_flagsext & TF_MEASURESNDBW) == 0 ||
1987 tp->t_bwmeas == NULL) {
1988 error = EINVAL;
1989 break;
1990 }
1991 minpkts = (in.min_burst_size != 0) ? in.min_burst_size :
1992 tp->t_bwmeas->bw_minsizepkts;
1993 maxpkts = (in.max_burst_size != 0) ? in.max_burst_size :
1994 tp->t_bwmeas->bw_maxsizepkts;
1995 if (minpkts > maxpkts) {
1996 error = EINVAL;
1997 break;
1998 }
1999 tp->t_bwmeas->bw_minsizepkts = minpkts;
2000 tp->t_bwmeas->bw_maxsizepkts = maxpkts;
2001 tp->t_bwmeas->bw_minsize = (minpkts * tp->t_maxseg);
2002 tp->t_bwmeas->bw_maxsize = (maxpkts * tp->t_maxseg);
2003 break;
2004 }
2005 case TCP_MAXSEG:
2006 error = sooptcopyin(sopt, &optval, sizeof optval,
2007 sizeof optval);
2008 if (error)
2009 break;
2010
2011 if (optval > 0 && optval <= tp->t_maxseg &&
2012 optval + 40 >= tcp_minmss)
2013 tp->t_maxseg = optval;
2014 else
2015 error = EINVAL;
2016 break;
2017
2018 case TCP_KEEPALIVE:
2019 error = sooptcopyin(sopt, &optval, sizeof optval,
2020 sizeof optval);
2021 if (error)
2022 break;
2023 if (optval < 0 || optval > UINT32_MAX/TCP_RETRANSHZ) {
2024 error = EINVAL;
2025 } else {
2026 tp->t_keepidle = optval * TCP_RETRANSHZ;
2027 /* reset the timer to new value */
2028 tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp,
2029 TCP_CONN_KEEPIDLE(tp));
2030 tcp_check_timer_state(tp);
2031 }
2032 break;
2033
2034 case TCP_CONNECTIONTIMEOUT:
2035 error = sooptcopyin(sopt, &optval, sizeof optval,
2036 sizeof optval);
2037 if (error)
2038 break;
2039 if (optval < 0 || optval > UINT32_MAX/TCP_RETRANSHZ) {
2040 error = EINVAL;
2041 } else {
2042 tp->t_keepinit = optval * TCP_RETRANSHZ;
2043 if (tp->t_state == TCPS_SYN_RECEIVED ||
2044 tp->t_state == TCPS_SYN_SENT) {
2045 tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp,
2046 TCP_CONN_KEEPINIT(tp));
2047 tcp_check_timer_state(tp);
2048 }
2049 }
2050 break;
2051
2052 case TCP_KEEPINTVL:
2053 error = sooptcopyin(sopt, &optval, sizeof(optval),
2054 sizeof(optval));
2055 if (error)
2056 break;
2057 if (optval < 0 || optval > UINT32_MAX/TCP_RETRANSHZ) {
2058 error = EINVAL;
2059 } else {
2060 tp->t_keepintvl = optval * TCP_RETRANSHZ;
2061 if (tp->t_state == TCPS_FIN_WAIT_2 &&
2062 TCP_CONN_MAXIDLE(tp) > 0) {
2063 tp->t_timer[TCPT_2MSL] = OFFSET_FROM_START(tp,
2064 TCP_CONN_MAXIDLE(tp));
2065 tcp_check_timer_state(tp);
2066 }
2067 }
2068 break;
2069
2070 case TCP_KEEPCNT:
2071 error = sooptcopyin(sopt, &optval, sizeof(optval),
2072 sizeof(optval));
2073 if (error)
2074 break;
2075 if (optval < 0 || optval > INT32_MAX) {
2076 error = EINVAL;
2077 } else {
2078 tp->t_keepcnt = optval;
2079 if (tp->t_state == TCPS_FIN_WAIT_2 &&
2080 TCP_CONN_MAXIDLE(tp) > 0) {
2081 tp->t_timer[TCPT_2MSL] = OFFSET_FROM_START(tp,
2082 TCP_CONN_MAXIDLE(tp));
2083 tcp_check_timer_state(tp);
2084 }
2085 }
2086 break;
2087
2088 case TCP_KEEPALIVE_OFFLOAD:
2089 error = sooptcopyin(sopt, &optval, sizeof(optval),
2090 sizeof(optval));
2091 if (error)
2092 break;
2093 if (optval < 0 || optval > INT32_MAX) {
2094 error = EINVAL;
2095 break;
2096 }
2097 if (optval != 0)
2098 inp->inp_flags2 |= INP2_KEEPALIVE_OFFLOAD;
2099 else
2100 inp->inp_flags2 &= ~INP2_KEEPALIVE_OFFLOAD;
2101 break;
2102
2103 case PERSIST_TIMEOUT:
2104 error = sooptcopyin(sopt, &optval, sizeof optval,
2105 sizeof optval);
2106 if (error)
2107 break;
2108 if (optval < 0)
2109 error = EINVAL;
2110 else
2111 tp->t_persist_timeout = optval * TCP_RETRANSHZ;
2112 break;
2113 case TCP_RXT_CONNDROPTIME:
2114 error = sooptcopyin(sopt, &optval, sizeof(optval),
2115 sizeof(optval));
2116 if (error)
2117 break;
2118 if (optval < 0)
2119 error = EINVAL;
2120 else
2121 tp->t_rxt_conndroptime = optval * TCP_RETRANSHZ;
2122 break;
2123 case TCP_NOTSENT_LOWAT:
2124 error = sooptcopyin(sopt, &optval, sizeof(optval),
2125 sizeof(optval));
2126 if (error)
2127 break;
2128 if (optval < 0) {
2129 error = EINVAL;
2130 break;
2131 } else {
2132 if (optval == 0) {
2133 so->so_flags &= ~(SOF_NOTSENT_LOWAT);
2134 tp->t_notsent_lowat = 0;
2135 } else {
2136 so->so_flags |= SOF_NOTSENT_LOWAT;
2137 tp->t_notsent_lowat = optval;
2138 }
2139 }
2140 break;
2141 case TCP_ADAPTIVE_READ_TIMEOUT:
2142 error = sooptcopyin(sopt, &optval, sizeof (optval),
2143 sizeof(optval));
2144 if (error)
2145 break;
2146 if (optval < 0 ||
2147 optval > TCP_ADAPTIVE_TIMEOUT_MAX) {
2148 error = EINVAL;
2149 break;
2150 } else if (optval == 0) {
2151 tp->t_adaptive_rtimo = 0;
2152 tcp_keepalive_reset(tp);
2153 } else {
2154 tp->t_adaptive_rtimo = optval;
2155 }
2156 break;
2157 case TCP_ADAPTIVE_WRITE_TIMEOUT:
2158 error = sooptcopyin(sopt, &optval, sizeof (optval),
2159 sizeof (optval));
2160 if (error)
2161 break;
2162 if (optval < 0 ||
2163 optval > TCP_ADAPTIVE_TIMEOUT_MAX) {
2164 error = EINVAL;
2165 break;
2166 } else {
2167 tp->t_adaptive_wtimo = optval;
2168 }
2169 break;
2170 case TCP_ENABLE_MSGS:
2171 error = sooptcopyin(sopt, &optval, sizeof(optval),
2172 sizeof(optval));
2173 if (error)
2174 break;
2175 if (optval < 0 || optval > 1) {
2176 error = EINVAL;
2177 } else if (optval == 1) {
2178 /*
2179 * Check if messages option is already
2180 * enabled, if so return.
2181 */
2182 if (so->so_flags & SOF_ENABLE_MSGS) {
2183 VERIFY(so->so_msg_state != NULL);
2184 break;
2185 }
2186
2187 /*
2188 * allocate memory for storing message
2189 * related state
2190 */
2191 VERIFY(so->so_msg_state == NULL);
2192 MALLOC(so->so_msg_state,
2193 struct msg_state *,
2194 sizeof(struct msg_state),
2195 M_TEMP, M_WAITOK | M_ZERO);
2196 if (so->so_msg_state == NULL) {
2197 error = ENOMEM;
2198 break;
2199 }
2200
2201 /* Enable message delivery */
2202 so->so_flags |= SOF_ENABLE_MSGS;
2203 } else {
2204 /*
2205 * Can't disable message delivery on socket
2206 * because of restrictions imposed by
2207 * encoding/decoding
2208 */
2209 error = EINVAL;
2210 }
2211 break;
2212 case TCP_SENDMOREACKS:
2213 error = sooptcopyin(sopt, &optval, sizeof(optval),
2214 sizeof(optval));
2215 if (error)
2216 break;
2217 if (optval < 0 || optval > 1) {
2218 error = EINVAL;
2219 } else if (optval == 0) {
2220 tp->t_flagsext &= ~(TF_NOSTRETCHACK);
2221 } else {
2222 tp->t_flagsext |= TF_NOSTRETCHACK;
2223 }
2224 break;
2225 case TCP_DISABLE_BLACKHOLE_DETECTION:
2226 error = sooptcopyin(sopt, &optval, sizeof(optval),
2227 sizeof(optval));
2228 if (error)
2229 break;
2230 if (optval < 0 || optval > 1) {
2231 error = EINVAL;
2232 } else if (optval == 0) {
2233 tp->t_flagsext &= ~TF_NOBLACKHOLE_DETECTION;
2234 } else {
2235 tp->t_flagsext |= TF_NOBLACKHOLE_DETECTION;
2236 if ((tp->t_flags & TF_BLACKHOLE) &&
2237 tp->t_pmtud_saved_maxopd > 0)
2238 tcp_pmtud_revert_segment_size(tp);
2239 }
2240 break;
2241 case TCP_FASTOPEN:
2242 if (!(tcp_fastopen & TCP_FASTOPEN_SERVER)) {
2243 error = ENOTSUP;
2244 break;
2245 }
2246
2247 error = sooptcopyin(sopt, &optval, sizeof(optval),
2248 sizeof(optval));
2249 if (error)
2250 break;
2251 if (optval < 0 || optval > 1) {
2252 error = EINVAL;
2253 break;
2254 }
2255 if (tp->t_state != TCPS_LISTEN) {
2256 error = EINVAL;
2257 break;
2258 }
2259 if (optval)
2260 tp->t_flagsext |= TF_FASTOPEN;
2261 else
2262 tcp_disable_tfo(tp);
2263 break;
2264 case TCP_ENABLE_ECN:
2265 error = sooptcopyin(sopt, &optval, sizeof optval,
2266 sizeof optval);
2267 if (error)
2268 break;
2269 if (optval) {
2270 tp->ecn_flags |= TE_ECN_MODE_ENABLE;
2271 tp->ecn_flags &= ~TE_ECN_MODE_DISABLE;
2272 } else {
2273 tp->ecn_flags &= ~TE_ECN_MODE_ENABLE;
2274 }
2275 break;
2276 case TCP_ECN_MODE:
2277 error = sooptcopyin(sopt, &optval, sizeof optval,
2278 sizeof optval);
2279 if (error)
2280 break;
2281 if (optval == ECN_MODE_DEFAULT) {
2282 tp->ecn_flags &= ~TE_ECN_MODE_ENABLE;
2283 tp->ecn_flags &= ~TE_ECN_MODE_DISABLE;
2284 } else if (optval == ECN_MODE_ENABLE) {
2285 tp->ecn_flags |= TE_ECN_MODE_ENABLE;
2286 tp->ecn_flags &= ~TE_ECN_MODE_DISABLE;
2287 } else if (optval == ECN_MODE_DISABLE) {
2288 tp->ecn_flags &= ~TE_ECN_MODE_ENABLE;
2289 tp->ecn_flags |= TE_ECN_MODE_DISABLE;
2290 } else {
2291 error = EINVAL;
2292 }
2293 break;
2294 case TCP_NOTIFY_ACKNOWLEDGEMENT:
2295 error = sooptcopyin(sopt, &optval,
2296 sizeof(optval), sizeof(optval));
2297 if (error)
2298 break;
2299 if (optval <= 0) {
2300 error = EINVAL;
2301 break;
2302 }
2303 if (tp->t_notify_ack_count >= TCP_MAX_NOTIFY_ACK) {
2304 error = ETOOMANYREFS;
2305 break;
2306 }
2307
2308 /*
2309 * validate that the given marker id is not
2310 * a duplicate to avoid ambiguity
2311 */
2312 if ((error = tcp_notify_ack_id_valid(tp, so,
2313 optval)) != 0) {
2314 break;
2315 }
2316 error = tcp_add_notify_ack_marker(tp, optval);
2317 break;
2318 case SO_FLUSH:
2319 if ((error = sooptcopyin(sopt, &optval, sizeof (optval),
2320 sizeof (optval))) != 0)
2321 break;
2322
2323 error = inp_flush(inp, optval);
2324 break;
2325
2326 case SO_TRAFFIC_MGT_BACKGROUND:
2327 if ((error = sooptcopyin(sopt, &optval, sizeof (optval),
2328 sizeof (optval))) != 0)
2329 break;
2330
2331 if (optval) {
2332 socket_set_traffic_mgt_flags_locked(so,
2333 TRAFFIC_MGT_SO_BACKGROUND);
2334 } else {
2335 socket_clear_traffic_mgt_flags_locked(so,
2336 TRAFFIC_MGT_SO_BACKGROUND);
2337 }
2338 break;
2339
2340 default:
2341 error = ENOPROTOOPT;
2342 break;
2343 }
2344 break;
2345
2346 case SOPT_GET:
2347 switch (sopt->sopt_name) {
2348 case TCP_NODELAY:
2349 optval = tp->t_flags & TF_NODELAY;
2350 break;
2351 case TCP_MAXSEG:
2352 optval = tp->t_maxseg;
2353 break;
2354 case TCP_KEEPALIVE:
2355 if (tp->t_keepidle > 0)
2356 optval = tp->t_keepidle / TCP_RETRANSHZ;
2357 else
2358 optval = tcp_keepidle / TCP_RETRANSHZ;
2359 break;
2360 case TCP_KEEPINTVL:
2361 if (tp->t_keepintvl > 0)
2362 optval = tp->t_keepintvl / TCP_RETRANSHZ;
2363 else
2364 optval = tcp_keepintvl / TCP_RETRANSHZ;
2365 break;
2366 case TCP_KEEPCNT:
2367 if (tp->t_keepcnt > 0)
2368 optval = tp->t_keepcnt;
2369 else
2370 optval = tcp_keepcnt;
2371 break;
2372 case TCP_KEEPALIVE_OFFLOAD:
2373 optval = !!(inp->inp_flags2 & INP2_KEEPALIVE_OFFLOAD);
2374 break;
2375 case TCP_NOOPT:
2376 optval = tp->t_flags & TF_NOOPT;
2377 break;
2378 case TCP_NOPUSH:
2379 optval = tp->t_flags & TF_NOPUSH;
2380 break;
2381 case TCP_ENABLE_ECN:
2382 optval = (tp->ecn_flags & TE_ECN_MODE_ENABLE) ? 1 : 0;
2383 break;
2384 case TCP_ECN_MODE:
2385 if (tp->ecn_flags & TE_ECN_MODE_ENABLE)
2386 optval = ECN_MODE_ENABLE;
2387 else if (tp->ecn_flags & TE_ECN_MODE_DISABLE)
2388 optval = ECN_MODE_DISABLE;
2389 else
2390 optval = ECN_MODE_DEFAULT;
2391 break;
2392 case TCP_CONNECTIONTIMEOUT:
2393 optval = tp->t_keepinit / TCP_RETRANSHZ;
2394 break;
2395 case PERSIST_TIMEOUT:
2396 optval = tp->t_persist_timeout / TCP_RETRANSHZ;
2397 break;
2398 case TCP_RXT_CONNDROPTIME:
2399 optval = tp->t_rxt_conndroptime / TCP_RETRANSHZ;
2400 break;
2401 case TCP_RXT_FINDROP:
2402 optval = tp->t_flagsext & TF_RXTFINDROP;
2403 break;
2404 case TCP_NOTIMEWAIT:
2405 optval = (tp->t_flagsext & TF_NOTIMEWAIT) ? 1 : 0;
2406 break;
2407 case TCP_FASTOPEN:
2408 if (tp->t_state != TCPS_LISTEN ||
2409 !(tcp_fastopen & TCP_FASTOPEN_SERVER)) {
2410 error = ENOTSUP;
2411 break;
2412 }
2413 optval = tfo_enabled(tp);
2414 break;
2415 case TCP_MEASURE_SND_BW:
2416 optval = tp->t_flagsext & TF_MEASURESNDBW;
2417 break;
2418 case TCP_INFO: {
2419 struct tcp_info ti;
2420
2421 tcp_fill_info(tp, &ti);
2422 error = sooptcopyout(sopt, &ti, sizeof(struct tcp_info));
2423 goto done;
2424 /* NOT REACHED */
2425 }
2426 case TCP_CONNECTION_INFO: {
2427 struct tcp_connection_info tci;
2428 tcp_connection_fill_info(tp, &tci);
2429 error = sooptcopyout(sopt, &tci,
2430 sizeof(struct tcp_connection_info));
2431 goto done;
2432 }
2433 case TCP_MEASURE_BW_BURST: {
2434 struct tcp_measure_bw_burst out;
2435 if ((tp->t_flagsext & TF_MEASURESNDBW) == 0 ||
2436 tp->t_bwmeas == NULL) {
2437 error = EINVAL;
2438 break;
2439 }
2440 out.min_burst_size = tp->t_bwmeas->bw_minsizepkts;
2441 out.max_burst_size = tp->t_bwmeas->bw_maxsizepkts;
2442 error = sooptcopyout(sopt, &out, sizeof(out));
2443 goto done;
2444 }
2445 case TCP_NOTSENT_LOWAT:
2446 if ((so->so_flags & SOF_NOTSENT_LOWAT) != 0) {
2447 optval = tp->t_notsent_lowat;
2448 } else {
2449 optval = 0;
2450 }
2451 break;
2452
2453 case TCP_ENABLE_MSGS:
2454 if (so->so_flags & SOF_ENABLE_MSGS) {
2455 optval = 1;
2456 } else {
2457 optval = 0;
2458 }
2459 break;
2460 case TCP_SENDMOREACKS:
2461 if (tp->t_flagsext & TF_NOSTRETCHACK)
2462 optval = 1;
2463 else
2464 optval = 0;
2465 break;
2466 case TCP_DISABLE_BLACKHOLE_DETECTION:
2467 if (tp->t_flagsext & TF_NOBLACKHOLE_DETECTION)
2468 optval = 1;
2469 else
2470 optval = 0;
2471 break;
2472 case TCP_PEER_PID: {
2473 pid_t pid;
2474 error = tcp_lookup_peer_pid_locked(so, &pid);
2475 if (error == 0)
2476 error = sooptcopyout(sopt, &pid, sizeof(pid));
2477 goto done;
2478 }
2479 case TCP_ADAPTIVE_READ_TIMEOUT:
2480 optval = tp->t_adaptive_rtimo;
2481 break;
2482 case TCP_ADAPTIVE_WRITE_TIMEOUT:
2483 optval = tp->t_adaptive_wtimo;
2484 break;
2485 case SO_TRAFFIC_MGT_BACKGROUND:
2486 optval = (so->so_flags1 &
2487 SOF1_TRAFFIC_MGT_SO_BACKGROUND) ? 1 : 0;
2488 break;
2489 case TCP_NOTIFY_ACKNOWLEDGEMENT: {
2490 struct tcp_notify_ack_complete retid;
2491
2492 if (sopt->sopt_valsize != sizeof (retid)) {
2493 error = EINVAL;
2494 break;
2495 }
2496 bzero(&retid, sizeof (retid));
2497 tcp_get_notify_ack_count(tp, &retid);
2498 if (retid.notify_complete_count > 0)
2499 tcp_get_notify_ack_ids(tp, &retid);
2500
2501 error = sooptcopyout(sopt, &retid, sizeof (retid));
2502 goto done;
2503 }
2504 default:
2505 error = ENOPROTOOPT;
2506 break;
2507 }
2508 if (error == 0)
2509 error = sooptcopyout(sopt, &optval, sizeof optval);
2510 break;
2511 }
2512 done:
2513 return (error);
2514 }
2515
2516 /*
2517 * tcp_sendspace and tcp_recvspace are the default send and receive window
2518 * sizes, respectively. These are obsolescent (this information should
2519 * be set by the route).
2520 */
2521 u_int32_t tcp_sendspace = 1448*256;
2522 u_int32_t tcp_recvspace = 1448*384;
2523
2524 /* During attach, the size of socket buffer allocated is limited to
2525 * sb_max in sbreserve. Disallow setting the tcp send and recv space
2526 * to be more than sb_max because that will cause tcp_attach to fail
2527 * (see radar 5713060)
2528 */
2529 static int
2530 sysctl_tcp_sospace(struct sysctl_oid *oidp, __unused void *arg1,
2531 __unused int arg2, struct sysctl_req *req) {
2532 u_int32_t new_value = 0, *space_p = NULL;
2533 int changed = 0, error = 0;
2534 u_quad_t sb_effective_max = (sb_max / (MSIZE+MCLBYTES)) * MCLBYTES;
2535
2536 switch (oidp->oid_number) {
2537 case TCPCTL_SENDSPACE:
2538 space_p = &tcp_sendspace;
2539 break;
2540 case TCPCTL_RECVSPACE:
2541 space_p = &tcp_recvspace;
2542 break;
2543 default:
2544 return EINVAL;
2545 }
2546 error = sysctl_io_number(req, *space_p, sizeof(u_int32_t),
2547 &new_value, &changed);
2548 if (changed) {
2549 if (new_value > 0 && new_value <= sb_effective_max) {
2550 *space_p = new_value;
2551 } else {
2552 error = ERANGE;
2553 }
2554 }
2555 return error;
2556 }
2557
2558 SYSCTL_PROC(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2559 &tcp_sendspace , 0, &sysctl_tcp_sospace, "IU", "Maximum outgoing TCP datagram size");
2560 SYSCTL_PROC(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2561 &tcp_recvspace , 0, &sysctl_tcp_sospace, "IU", "Maximum incoming TCP datagram size");
2562
2563 /*
2564 * Attach TCP protocol to socket, allocating
2565 * internet protocol control block, tcp control block,
2566 * bufer space, and entering LISTEN state if to accept connections.
2567 *
2568 * Returns: 0 Success
2569 * in_pcballoc:ENOBUFS
2570 * in_pcballoc:ENOMEM
2571 * in_pcballoc:??? [IPSEC specific]
2572 * soreserve:ENOBUFS
2573 */
2574 static int
2575 tcp_attach(struct socket *so, struct proc *p)
2576 {
2577 struct tcpcb *tp;
2578 struct inpcb *inp;
2579 int error;
2580 #if INET6
2581 int isipv6 = SOCK_CHECK_DOM(so, PF_INET6) != 0;
2582 #endif
2583
2584 error = in_pcballoc(so, &tcbinfo, p);
2585 if (error)
2586 return (error);
2587
2588 inp = sotoinpcb(so);
2589
2590 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
2591 error = soreserve(so, tcp_sendspace, tcp_recvspace);
2592 if (error)
2593 return (error);
2594 }
2595
2596 if (so->so_snd.sb_preconn_hiwat == 0) {
2597 soreserve_preconnect(so, 2048);
2598 }
2599
2600 if ((so->so_rcv.sb_flags & SB_USRSIZE) == 0)
2601 so->so_rcv.sb_flags |= SB_AUTOSIZE;
2602 if ((so->so_snd.sb_flags & SB_USRSIZE) == 0)
2603 so->so_snd.sb_flags |= SB_AUTOSIZE;
2604
2605 #if INET6
2606 if (isipv6) {
2607 inp->inp_vflag |= INP_IPV6;
2608 inp->in6p_hops = -1; /* use kernel default */
2609 }
2610 else
2611 #endif /* INET6 */
2612 inp->inp_vflag |= INP_IPV4;
2613 tp = tcp_newtcpcb(inp);
2614 if (tp == NULL) {
2615 int nofd = so->so_state & SS_NOFDREF; /* XXX */
2616
2617 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */
2618 #if INET6
2619 if (isipv6)
2620 in6_pcbdetach(inp);
2621 else
2622 #endif /* INET6 */
2623 in_pcbdetach(inp);
2624 so->so_state |= nofd;
2625 return (ENOBUFS);
2626 }
2627 if (nstat_collect)
2628 nstat_tcp_new_pcb(inp);
2629 tp->t_state = TCPS_CLOSED;
2630 return (0);
2631 }
2632
2633 /*
2634 * Initiate (or continue) disconnect.
2635 * If embryonic state, just send reset (once).
2636 * If in ``let data drain'' option and linger null, just drop.
2637 * Otherwise (hard), mark socket disconnecting and drop
2638 * current input data; switch states based on user close, and
2639 * send segment to peer (with FIN).
2640 */
2641 static struct tcpcb *
2642 tcp_disconnect(struct tcpcb *tp)
2643 {
2644 struct socket *so = tp->t_inpcb->inp_socket;
2645
2646 if (tp->t_state < TCPS_ESTABLISHED)
2647 tp = tcp_close(tp);
2648 else if ((so->so_options & SO_LINGER) && so->so_linger == 0)
2649 tp = tcp_drop(tp, 0);
2650 else {
2651 soisdisconnecting(so);
2652 sbflush(&so->so_rcv);
2653 tp = tcp_usrclosed(tp);
2654 #if MPTCP
2655 /* A reset has been sent but socket exists, do not send FIN */
2656 if ((so->so_flags & SOF_MP_SUBFLOW) &&
2657 (tp) && (tp->t_mpflags & TMPF_RESET))
2658 return (tp);
2659 #endif
2660 if (tp)
2661 (void) tcp_output(tp);
2662 }
2663 return (tp);
2664 }
2665
2666 /*
2667 * User issued close, and wish to trail through shutdown states:
2668 * if never received SYN, just forget it. If got a SYN from peer,
2669 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
2670 * If already got a FIN from peer, then almost done; go to LAST_ACK
2671 * state. In all other cases, have already sent FIN to peer (e.g.
2672 * after PRU_SHUTDOWN), and just have to play tedious game waiting
2673 * for peer to send FIN or not respond to keep-alives, etc.
2674 * We can let the user exit from the close as soon as the FIN is acked.
2675 */
2676 static struct tcpcb *
2677 tcp_usrclosed(struct tcpcb *tp)
2678 {
2679 switch (tp->t_state) {
2680
2681 case TCPS_CLOSED:
2682 case TCPS_LISTEN:
2683 tp = tcp_close(tp);
2684 break;
2685
2686 case TCPS_SYN_SENT:
2687 case TCPS_SYN_RECEIVED:
2688 tp->t_flags |= TF_NEEDFIN;
2689 break;
2690
2691 case TCPS_ESTABLISHED:
2692 DTRACE_TCP4(state__change, void, NULL,
2693 struct inpcb *, tp->t_inpcb,
2694 struct tcpcb *, tp,
2695 int32_t, TCPS_FIN_WAIT_1);
2696 tp->t_state = TCPS_FIN_WAIT_1;
2697 break;
2698
2699 case TCPS_CLOSE_WAIT:
2700 DTRACE_TCP4(state__change, void, NULL,
2701 struct inpcb *, tp->t_inpcb,
2702 struct tcpcb *, tp,
2703 int32_t, TCPS_LAST_ACK);
2704 tp->t_state = TCPS_LAST_ACK;
2705 break;
2706 }
2707 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) {
2708 soisdisconnected(tp->t_inpcb->inp_socket);
2709 /* To prevent the connection hanging in FIN_WAIT_2 forever. */
2710 if (tp->t_state == TCPS_FIN_WAIT_2)
2711 tp->t_timer[TCPT_2MSL] = OFFSET_FROM_START(tp,
2712 TCP_CONN_MAXIDLE(tp));
2713 }
2714 return (tp);
2715 }
2716
2717 void
2718 tcp_in_cksum_stats(u_int32_t len)
2719 {
2720 tcpstat.tcps_rcv_swcsum++;
2721 tcpstat.tcps_rcv_swcsum_bytes += len;
2722 }
2723
2724 void
2725 tcp_out_cksum_stats(u_int32_t len)
2726 {
2727 tcpstat.tcps_snd_swcsum++;
2728 tcpstat.tcps_snd_swcsum_bytes += len;
2729 }
2730
2731 #if INET6
2732 void
2733 tcp_in6_cksum_stats(u_int32_t len)
2734 {
2735 tcpstat.tcps_rcv6_swcsum++;
2736 tcpstat.tcps_rcv6_swcsum_bytes += len;
2737 }
2738
2739 void
2740 tcp_out6_cksum_stats(u_int32_t len)
2741 {
2742 tcpstat.tcps_snd6_swcsum++;
2743 tcpstat.tcps_snd6_swcsum_bytes += len;
2744 }
2745
2746 /*
2747 * When messages are enabled on a TCP socket, the message priority
2748 * is sent as a control message. This function will extract it.
2749 */
2750 int
2751 tcp_get_msg_priority(struct mbuf *control, uint32_t *msgpri)
2752 {
2753 struct cmsghdr *cm;
2754 if (control == NULL)
2755 return(EINVAL);
2756
2757 for (cm = M_FIRST_CMSGHDR(control); cm;
2758 cm = M_NXT_CMSGHDR(control, cm)) {
2759 if (cm->cmsg_len < sizeof(struct cmsghdr) ||
2760 cm->cmsg_len > control->m_len) {
2761 return (EINVAL);
2762 }
2763 if (cm->cmsg_level == SOL_SOCKET &&
2764 cm->cmsg_type == SCM_MSG_PRIORITY) {
2765 *msgpri = *(unsigned int *)(void *)CMSG_DATA(cm);
2766 break;
2767 }
2768 }
2769
2770 VERIFY(*msgpri >= MSG_PRI_MIN && *msgpri <= MSG_PRI_MAX);
2771 return (0);
2772 }
2773 #endif /* INET6 */
mirror of XNU