1 | /* |
---|
2 | * Copyright (c) 1982, 1986, 1988, 1993 |
---|
3 | * The Regents of the University of California. All rights reserved. |
---|
4 | * |
---|
5 | * Redistribution and use in source and binary forms, with or without |
---|
6 | * modification, are permitted provided that the following conditions |
---|
7 | * are met: |
---|
8 | * 1. Redistributions of source code must retain the above copyright |
---|
9 | * notice, this list of conditions and the following disclaimer. |
---|
10 | * 2. Redistributions in binary form must reproduce the above copyright |
---|
11 | * notice, this list of conditions and the following disclaimer in the |
---|
12 | * documentation and/or other materials provided with the distribution. |
---|
13 | * 4. Neither the name of the University nor the names of its contributors |
---|
14 | * may be used to endorse or promote products derived from this software |
---|
15 | * without specific prior written permission. |
---|
16 | * |
---|
17 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
---|
18 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
---|
19 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
---|
20 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
---|
21 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
---|
22 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
---|
23 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
---|
24 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
---|
25 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
---|
26 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
---|
27 | * SUCH DAMAGE. |
---|
28 | * |
---|
29 | * From: @(#)tcp_usrreq.c 8.2 (Berkeley) 1/3/94 |
---|
30 | * $FreeBSD: src/sys/netinet/tcp_usrreq.c,v 1.120 2005/05/01 14:01:38 rwatson Exp $ |
---|
31 | */ |
---|
32 | |
---|
33 | /* |
---|
34 | * $Id$ |
---|
35 | */ |
---|
36 | |
---|
37 | #include "opt_tcpdebug.h" |
---|
38 | |
---|
39 | #include <sys/param.h> |
---|
40 | #include <sys/queue.h> |
---|
41 | #include <sys/systm.h> |
---|
42 | #include <sys/kernel.h> |
---|
43 | #include <sys/sysctl.h> |
---|
44 | #include <sys/malloc.h> |
---|
45 | #include <sys/mbuf.h> |
---|
46 | #include <sys/socket.h> |
---|
47 | #include <sys/socketvar.h> |
---|
48 | #include <sys/protosw.h> |
---|
49 | #include <sys/errno.h> |
---|
50 | #include <sys/stat.h> |
---|
51 | |
---|
52 | #include <net/if.h> |
---|
53 | #include <net/route.h> |
---|
54 | |
---|
55 | #include <netinet/in.h> |
---|
56 | #include <netinet/in_systm.h> |
---|
57 | #include <netinet/ip.h> |
---|
58 | #include <netinet/in_pcb.h> |
---|
59 | #include <netinet/in_var.h> |
---|
60 | #include <netinet/ip_var.h> |
---|
61 | #include <netinet/tcp.h> |
---|
62 | #include <netinet/tcp_fsm.h> |
---|
63 | #include <netinet/tcp_seq.h> |
---|
64 | #include <netinet/tcp_timer.h> |
---|
65 | #include <netinet/tcp_var.h> |
---|
66 | #include <netinet/tcpip.h> |
---|
67 | #ifdef TCPDEBUG |
---|
68 | #include <netinet/tcp_debug.h> |
---|
69 | #endif |
---|
70 | |
---|
71 | /* |
---|
72 | * TCP protocol interface to socket abstraction. |
---|
73 | */ |
---|
74 | extern char *tcpstates[]; |
---|
75 | |
---|
76 | static int tcp_attach(struct socket *); |
---|
77 | static int tcp_connect __P((struct tcpcb *, struct mbuf *)); |
---|
78 | static struct tcpcb * |
---|
79 | tcp_disconnect(struct tcpcb *); |
---|
80 | static struct tcpcb * |
---|
81 | tcp_usrclosed __P((struct tcpcb *)); |
---|
82 | |
---|
83 | #ifdef TCPDEBUG |
---|
84 | #define TCPDEBUG0 int ostate |
---|
85 | #define TCPDEBUG1() ostate = tp ? tp->t_state : 0 |
---|
86 | #define TCPDEBUG2(req) if (tp && (so->so_options & SO_DEBUG)) \ |
---|
87 | tcp_trace(TA_USER, ostate, tp, 0, req) |
---|
88 | #else |
---|
89 | #define TCPDEBUG0 |
---|
90 | #define TCPDEBUG1() |
---|
91 | #define TCPDEBUG2(req) |
---|
92 | #endif |
---|
93 | |
---|
94 | /* |
---|
95 | * TCP attaches to socket via pru_attach(), reserving space, |
---|
96 | * and an internet control block. |
---|
97 | */ |
---|
98 | static int |
---|
99 | tcp_usr_attach(struct socket *so, int proto) |
---|
100 | { |
---|
101 | int s = splnet(); |
---|
102 | int error; |
---|
103 | struct inpcb *inp = sotoinpcb(so); |
---|
104 | struct tcpcb *tp = 0; |
---|
105 | TCPDEBUG0; |
---|
106 | |
---|
107 | TCPDEBUG1(); |
---|
108 | if (inp) { |
---|
109 | error = EISCONN; |
---|
110 | goto out; |
---|
111 | } |
---|
112 | |
---|
113 | error = tcp_attach(so); |
---|
114 | if (error) |
---|
115 | goto out; |
---|
116 | |
---|
117 | if ((so->so_options & SO_LINGER) && so->so_linger == 0) |
---|
118 | so->so_linger = TCP_LINGERTIME * hz; |
---|
119 | tp = sototcpcb(so); |
---|
120 | out: |
---|
121 | TCPDEBUG2(PRU_ATTACH); |
---|
122 | splx(s); |
---|
123 | return error; |
---|
124 | } |
---|
125 | |
---|
126 | /* |
---|
127 | * pru_detach() detaches the TCP protocol from the socket. |
---|
128 | * If the protocol state is non-embryonic, then can't |
---|
129 | * do this directly: have to initiate a pru_disconnect(), |
---|
130 | * which may finish later; embryonic TCB's can just |
---|
131 | * be discarded here. |
---|
132 | */ |
---|
133 | static int |
---|
134 | tcp_usr_detach(struct socket *so) |
---|
135 | { |
---|
136 | int s = splnet(); |
---|
137 | int error = 0; |
---|
138 | struct inpcb *inp = sotoinpcb(so); |
---|
139 | struct tcpcb *tp; |
---|
140 | TCPDEBUG0; |
---|
141 | |
---|
142 | if (inp == 0) { |
---|
143 | splx(s); |
---|
144 | return EINVAL; /* XXX */ |
---|
145 | } |
---|
146 | tp = intotcpcb(inp); |
---|
147 | TCPDEBUG1(); |
---|
148 | if (tp->t_state > TCPS_LISTEN) |
---|
149 | tp = tcp_disconnect(tp); |
---|
150 | else |
---|
151 | tp = tcp_close(tp); |
---|
152 | |
---|
153 | TCPDEBUG2(PRU_DETACH); |
---|
154 | splx(s); |
---|
155 | return error; |
---|
156 | } |
---|
157 | |
---|
158 | #define COMMON_START() TCPDEBUG0; \ |
---|
159 | do { \ |
---|
160 | if (inp == 0) { \ |
---|
161 | splx(s); \ |
---|
162 | return EINVAL; \ |
---|
163 | } \ |
---|
164 | tp = intotcpcb(inp); \ |
---|
165 | TCPDEBUG1(); \ |
---|
166 | } while(0) |
---|
167 | |
---|
168 | #define COMMON_END(req) out: TCPDEBUG2(req); splx(s); return error; goto out |
---|
169 | |
---|
170 | |
---|
171 | /* |
---|
172 | * Give the socket an address. |
---|
173 | */ |
---|
174 | static int |
---|
175 | tcp_usr_bind(struct socket *so, struct mbuf *nam) |
---|
176 | { |
---|
177 | int s = splnet(); |
---|
178 | int error = 0; |
---|
179 | struct inpcb *inp = sotoinpcb(so); |
---|
180 | struct tcpcb *tp; |
---|
181 | struct sockaddr_in *sinp; |
---|
182 | |
---|
183 | COMMON_START(); |
---|
184 | |
---|
185 | /* |
---|
186 | * Must check for multicast addresses and disallow binding |
---|
187 | * to them. |
---|
188 | */ |
---|
189 | sinp = mtod(nam, struct sockaddr_in *); |
---|
190 | if (sinp->sin_family == AF_INET && |
---|
191 | IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) { |
---|
192 | error = EAFNOSUPPORT; |
---|
193 | goto out; |
---|
194 | } |
---|
195 | error = in_pcbbind(inp, nam); |
---|
196 | if (error) |
---|
197 | goto out; |
---|
198 | COMMON_END(PRU_BIND); |
---|
199 | |
---|
200 | } |
---|
201 | |
---|
202 | /* |
---|
203 | * Prepare to accept connections. |
---|
204 | */ |
---|
205 | static int |
---|
206 | tcp_usr_listen(struct socket *so) |
---|
207 | { |
---|
208 | int s = splnet(); |
---|
209 | int error = 0; |
---|
210 | struct inpcb *inp = sotoinpcb(so); |
---|
211 | struct tcpcb *tp; |
---|
212 | |
---|
213 | COMMON_START(); |
---|
214 | if (inp->inp_lport == 0) |
---|
215 | error = in_pcbbind(inp, NULL); |
---|
216 | if (error == 0) |
---|
217 | tp->t_state = TCPS_LISTEN; |
---|
218 | COMMON_END(PRU_LISTEN); |
---|
219 | } |
---|
220 | |
---|
221 | /* |
---|
222 | * Initiate connection to peer. |
---|
223 | * Create a template for use in transmissions on this connection. |
---|
224 | * Enter SYN_SENT state, and mark socket as connecting. |
---|
225 | * Start keep-alive timer, and seed output sequence space. |
---|
226 | * Send initial segment on connection. |
---|
227 | */ |
---|
228 | static int |
---|
229 | tcp_usr_connect(struct socket *so, struct mbuf *nam) |
---|
230 | { |
---|
231 | int s = splnet(); |
---|
232 | int error = 0; |
---|
233 | struct inpcb *inp = sotoinpcb(so); |
---|
234 | struct tcpcb *tp; |
---|
235 | struct sockaddr_in *sinp; |
---|
236 | |
---|
237 | COMMON_START(); |
---|
238 | |
---|
239 | /* |
---|
240 | * Must disallow TCP ``connections'' to multicast addresses. |
---|
241 | */ |
---|
242 | sinp = mtod(nam, struct sockaddr_in *); |
---|
243 | if (sinp->sin_family == AF_INET |
---|
244 | && IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) { |
---|
245 | error = EAFNOSUPPORT; |
---|
246 | goto out; |
---|
247 | } |
---|
248 | |
---|
249 | if ((error = tcp_connect(tp, nam)) != 0) |
---|
250 | goto out; |
---|
251 | error = tcp_output(tp); |
---|
252 | COMMON_END(PRU_CONNECT); |
---|
253 | } |
---|
254 | |
---|
255 | /* |
---|
256 | * Initiate disconnect from peer. |
---|
257 | * If connection never passed embryonic stage, just drop; |
---|
258 | * else if don't need to let data drain, then can just drop anyways, |
---|
259 | * else have to begin TCP shutdown process: mark socket disconnecting, |
---|
260 | * drain unread data, state switch to reflect user close, and |
---|
261 | * send segment (e.g. FIN) to peer. Socket will be really disconnected |
---|
262 | * when peer sends FIN and acks ours. |
---|
263 | * |
---|
264 | * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. |
---|
265 | */ |
---|
266 | static int |
---|
267 | tcp_usr_disconnect(struct socket *so) |
---|
268 | { |
---|
269 | int s = splnet(); |
---|
270 | int error = 0; |
---|
271 | struct inpcb *inp = sotoinpcb(so); |
---|
272 | struct tcpcb *tp; |
---|
273 | |
---|
274 | COMMON_START(); |
---|
275 | tp = tcp_disconnect(tp); |
---|
276 | COMMON_END(PRU_DISCONNECT); |
---|
277 | } |
---|
278 | |
---|
279 | /* |
---|
280 | * Accept a connection. Essentially all the work is |
---|
281 | * done at higher levels; just return the address |
---|
282 | * of the peer, storing through addr. |
---|
283 | */ |
---|
284 | static int |
---|
285 | tcp_usr_accept(struct socket *so, struct mbuf *nam) |
---|
286 | { |
---|
287 | int s = splnet(); |
---|
288 | int error = 0; |
---|
289 | struct inpcb *inp = sotoinpcb(so); |
---|
290 | struct tcpcb *tp; |
---|
291 | |
---|
292 | COMMON_START(); |
---|
293 | in_setpeeraddr(inp, nam); |
---|
294 | COMMON_END(PRU_ACCEPT); |
---|
295 | } |
---|
296 | |
---|
297 | /* |
---|
298 | * Mark the connection as being incapable of further output. |
---|
299 | */ |
---|
300 | static int |
---|
301 | tcp_usr_shutdown(struct socket *so) |
---|
302 | { |
---|
303 | int s = splnet(); |
---|
304 | int error = 0; |
---|
305 | struct inpcb *inp = sotoinpcb(so); |
---|
306 | struct tcpcb *tp; |
---|
307 | |
---|
308 | COMMON_START(); |
---|
309 | socantsendmore(so); |
---|
310 | tp = tcp_usrclosed(tp); |
---|
311 | if (tp) |
---|
312 | error = tcp_output(tp); |
---|
313 | COMMON_END(PRU_SHUTDOWN); |
---|
314 | } |
---|
315 | |
---|
316 | /* |
---|
317 | * After a receive, possibly send window update to peer. |
---|
318 | */ |
---|
319 | static int |
---|
320 | tcp_usr_rcvd(struct socket *so, int flags) |
---|
321 | { |
---|
322 | int s = splnet(); |
---|
323 | int error = 0; |
---|
324 | struct inpcb *inp = sotoinpcb(so); |
---|
325 | struct tcpcb *tp; |
---|
326 | |
---|
327 | COMMON_START(); |
---|
328 | tcp_output(tp); |
---|
329 | COMMON_END(PRU_RCVD); |
---|
330 | } |
---|
331 | |
---|
332 | /* |
---|
333 | * Do a send by putting data in output queue and updating urgent |
---|
334 | * marker if URG set. Possibly send more data. |
---|
335 | */ |
---|
336 | static int |
---|
337 | tcp_usr_send(struct socket *so, int flags, struct mbuf *m, struct mbuf *nam, |
---|
338 | struct mbuf *control) |
---|
339 | { |
---|
340 | int s = splnet(); |
---|
341 | int error = 0; |
---|
342 | struct inpcb *inp = sotoinpcb(so); |
---|
343 | struct tcpcb *tp; |
---|
344 | |
---|
345 | COMMON_START(); |
---|
346 | if (control && control->m_len) { |
---|
347 | m_freem(control); /* XXX shouldn't caller do this??? */ |
---|
348 | if (m) |
---|
349 | m_freem(m); |
---|
350 | error = EINVAL; |
---|
351 | goto out; |
---|
352 | } |
---|
353 | |
---|
354 | if(!(flags & PRUS_OOB)) { |
---|
355 | sbappend(&so->so_snd, m); |
---|
356 | if (nam && tp->t_state < TCPS_SYN_SENT) { |
---|
357 | /* |
---|
358 | * Do implied connect if not yet connected, |
---|
359 | * initialize window to default value, and |
---|
360 | * initialize maxseg/maxopd using peer's cached |
---|
361 | * MSS. |
---|
362 | */ |
---|
363 | error = tcp_connect(tp, nam); |
---|
364 | if (error) |
---|
365 | goto out; |
---|
366 | tp->snd_wnd = TTCP_CLIENT_SND_WND; |
---|
367 | tcp_mss(tp, -1); |
---|
368 | } |
---|
369 | |
---|
370 | if (flags & PRUS_EOF) { |
---|
371 | /* |
---|
372 | * Close the send side of the connection after |
---|
373 | * the data is sent. |
---|
374 | */ |
---|
375 | socantsendmore(so); |
---|
376 | tp = tcp_usrclosed(tp); |
---|
377 | } |
---|
378 | if (tp != NULL) |
---|
379 | error = tcp_output(tp); |
---|
380 | } else { |
---|
381 | if (sbspace(&so->so_snd) < -512) { |
---|
382 | m_freem(m); |
---|
383 | error = ENOBUFS; |
---|
384 | goto out; |
---|
385 | } |
---|
386 | /* |
---|
387 | * According to RFC961 (Assigned Protocols), |
---|
388 | * the urgent pointer points to the last octet |
---|
389 | * of urgent data. We continue, however, |
---|
390 | * to consider it to indicate the first octet |
---|
391 | * of data past the urgent section. |
---|
392 | * Otherwise, snd_up should be one lower. |
---|
393 | */ |
---|
394 | sbappend(&so->so_snd, m); |
---|
395 | if (nam && tp->t_state < TCPS_SYN_SENT) { |
---|
396 | /* |
---|
397 | * Do implied connect if not yet connected, |
---|
398 | * initialize window to default value, and |
---|
399 | * initialize maxseg/maxopd using peer's cached |
---|
400 | * MSS. |
---|
401 | */ |
---|
402 | error = tcp_connect(tp, nam); |
---|
403 | if (error) |
---|
404 | goto out; |
---|
405 | tp->snd_wnd = TTCP_CLIENT_SND_WND; |
---|
406 | tcp_mss(tp, -1); |
---|
407 | } |
---|
408 | tp->snd_up = tp->snd_una + so->so_snd.sb_cc; |
---|
409 | tp->t_force = 1; |
---|
410 | error = tcp_output(tp); |
---|
411 | tp->t_force = 0; |
---|
412 | } |
---|
413 | COMMON_END((flags & PRUS_OOB) ? PRU_SENDOOB : |
---|
414 | ((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND)); |
---|
415 | } |
---|
416 | |
---|
417 | /* |
---|
418 | * Abort the TCP. |
---|
419 | */ |
---|
420 | static int |
---|
421 | tcp_usr_abort(struct socket *so) |
---|
422 | { |
---|
423 | int s = splnet(); |
---|
424 | int error = 0; |
---|
425 | struct inpcb *inp = sotoinpcb(so); |
---|
426 | struct tcpcb *tp; |
---|
427 | |
---|
428 | COMMON_START(); |
---|
429 | tp = tcp_drop(tp, ECONNABORTED); |
---|
430 | COMMON_END(PRU_ABORT); |
---|
431 | } |
---|
432 | |
---|
433 | /* |
---|
434 | * Fill in st_bklsize for fstat() operations on a socket. |
---|
435 | */ |
---|
436 | static int |
---|
437 | tcp_usr_sense(struct socket *so, struct stat *sb) |
---|
438 | { |
---|
439 | int s = splnet(); |
---|
440 | |
---|
441 | sb->st_blksize = so->so_snd.sb_hiwat; |
---|
442 | splx(s); |
---|
443 | return 0; |
---|
444 | } |
---|
445 | |
---|
446 | /* |
---|
447 | * Receive out-of-band data. |
---|
448 | */ |
---|
449 | static int |
---|
450 | tcp_usr_rcvoob(struct socket *so, struct mbuf *m, int flags) |
---|
451 | { |
---|
452 | int s = splnet(); |
---|
453 | int error = 0; |
---|
454 | struct inpcb *inp = sotoinpcb(so); |
---|
455 | struct tcpcb *tp; |
---|
456 | |
---|
457 | COMMON_START(); |
---|
458 | if ((so->so_oobmark == 0 && |
---|
459 | (so->so_state & SS_RCVATMARK) == 0) || |
---|
460 | so->so_options & SO_OOBINLINE || |
---|
461 | tp->t_oobflags & TCPOOB_HADDATA) { |
---|
462 | error = EINVAL; |
---|
463 | goto out; |
---|
464 | } |
---|
465 | if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { |
---|
466 | error = EWOULDBLOCK; |
---|
467 | goto out; |
---|
468 | } |
---|
469 | m->m_len = 1; |
---|
470 | *mtod(m, caddr_t) = tp->t_iobc; |
---|
471 | if ((flags & MSG_PEEK) == 0) |
---|
472 | tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); |
---|
473 | COMMON_END(PRU_RCVOOB); |
---|
474 | } |
---|
475 | |
---|
476 | static int |
---|
477 | tcp_usr_sockaddr(struct socket *so, struct mbuf *nam) |
---|
478 | { |
---|
479 | int s = splnet(); |
---|
480 | int error = 0; |
---|
481 | struct inpcb *inp = sotoinpcb(so); |
---|
482 | struct tcpcb *tp; |
---|
483 | |
---|
484 | COMMON_START(); |
---|
485 | in_setsockaddr(inp, nam); |
---|
486 | COMMON_END(PRU_SOCKADDR); |
---|
487 | } |
---|
488 | |
---|
489 | static int |
---|
490 | tcp_usr_peeraddr(struct socket *so, struct mbuf *nam) |
---|
491 | { |
---|
492 | int s = splnet(); |
---|
493 | int error = 0; |
---|
494 | struct inpcb *inp = sotoinpcb(so); |
---|
495 | struct tcpcb *tp; |
---|
496 | |
---|
497 | COMMON_START(); |
---|
498 | in_setpeeraddr(inp, nam); |
---|
499 | COMMON_END(PRU_PEERADDR); |
---|
500 | } |
---|
501 | |
---|
502 | /* |
---|
503 | * XXX - this should just be a call to in_control, but we need to get |
---|
504 | * the types worked out. |
---|
505 | */ |
---|
506 | static int |
---|
507 | tcp_usr_control(struct socket *so, int cmd, caddr_t arg, struct ifnet *ifp) |
---|
508 | { |
---|
509 | return in_control(so, cmd, arg, ifp); |
---|
510 | } |
---|
511 | |
---|
512 | /* xxx - should be const */ |
---|
513 | struct pr_usrreqs tcp_usrreqs = { |
---|
514 | tcp_usr_abort, tcp_usr_accept, tcp_usr_attach, tcp_usr_bind, |
---|
515 | tcp_usr_connect, pru_connect2_notsupp, tcp_usr_control, tcp_usr_detach, |
---|
516 | tcp_usr_disconnect, tcp_usr_listen, tcp_usr_peeraddr, tcp_usr_rcvd, |
---|
517 | tcp_usr_rcvoob, tcp_usr_send, tcp_usr_sense, tcp_usr_shutdown, |
---|
518 | tcp_usr_sockaddr |
---|
519 | }; |
---|
520 | |
---|
521 | /* |
---|
522 | * Common subroutine to open a TCP connection to remote host specified |
---|
523 | * by struct sockaddr_in in mbuf *nam. Call in_pcbbind to assign a local |
---|
524 | * port number if needed. Call in_pcbladdr to do the routing and to choose |
---|
525 | * a local host address (interface). If there is an existing incarnation |
---|
526 | * of the same connection in TIME-WAIT state and if the remote host was |
---|
527 | * sending CC options and if the connection duration was < MSL, then |
---|
528 | * truncate the previous TIME-WAIT state and proceed. |
---|
529 | * Initialize connection parameters and enter SYN-SENT state. |
---|
530 | */ |
---|
531 | static int |
---|
532 | tcp_connect(tp, nam) |
---|
533 | register struct tcpcb *tp; |
---|
534 | struct mbuf *nam; |
---|
535 | { |
---|
536 | struct inpcb *inp = tp->t_inpcb, *oinp; |
---|
537 | struct socket *so = inp->inp_socket; |
---|
538 | struct tcpcb *otp; |
---|
539 | struct sockaddr_in *sin = mtod(nam, struct sockaddr_in *); |
---|
540 | struct sockaddr_in *ifaddr; |
---|
541 | int error; |
---|
542 | struct rmxp_tao *taop; |
---|
543 | struct rmxp_tao tao_noncached; |
---|
544 | |
---|
545 | if (inp->inp_lport == 0) { |
---|
546 | error = in_pcbbind(inp, NULL); |
---|
547 | if (error) |
---|
548 | return error; |
---|
549 | } |
---|
550 | |
---|
551 | /* |
---|
552 | * Cannot simply call in_pcbconnect, because there might be an |
---|
553 | * earlier incarnation of this same connection still in |
---|
554 | * TIME_WAIT state, creating an ADDRINUSE error. |
---|
555 | */ |
---|
556 | error = in_pcbladdr(inp, nam, &ifaddr); |
---|
557 | if (error) |
---|
558 | return error; |
---|
559 | oinp = in_pcblookuphash(inp->inp_pcbinfo, |
---|
560 | sin->sin_addr, sin->sin_port, |
---|
561 | inp->inp_laddr.s_addr != INADDR_ANY ? inp->inp_laddr |
---|
562 | : ifaddr->sin_addr, |
---|
563 | inp->inp_lport, 0); |
---|
564 | if (oinp) { |
---|
565 | if (oinp != inp && (otp = intotcpcb(oinp)) != NULL && |
---|
566 | otp->t_state == TCPS_TIME_WAIT && |
---|
567 | otp->t_duration < TCPTV_MSL && |
---|
568 | (otp->t_flags & TF_RCVD_CC)) |
---|
569 | otp = tcp_close(otp); |
---|
570 | else |
---|
571 | return EADDRINUSE; |
---|
572 | } |
---|
573 | if (inp->inp_laddr.s_addr == INADDR_ANY) |
---|
574 | inp->inp_laddr = ifaddr->sin_addr; |
---|
575 | inp->inp_faddr = sin->sin_addr; |
---|
576 | inp->inp_fport = sin->sin_port; |
---|
577 | in_pcbrehash(inp); |
---|
578 | |
---|
579 | tp->t_template = tcp_template(tp); |
---|
580 | if (tp->t_template == 0) { |
---|
581 | in_pcbdisconnect(inp); |
---|
582 | return ENOBUFS; |
---|
583 | } |
---|
584 | |
---|
585 | /* Compute window scaling to request. */ |
---|
586 | while (tp->request_r_scale < TCP_MAX_WINSHIFT && |
---|
587 | (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) |
---|
588 | tp->request_r_scale++; |
---|
589 | |
---|
590 | soisconnecting(so); |
---|
591 | tcpstat.tcps_connattempt++; |
---|
592 | tp->t_state = TCPS_SYN_SENT; |
---|
593 | tp->t_timer[TCPT_KEEP] = tcp_keepinit; |
---|
594 | tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2; |
---|
595 | tcp_sendseqinit(tp); |
---|
596 | |
---|
597 | /* |
---|
598 | * Generate a CC value for this connection and |
---|
599 | * check whether CC or CCnew should be used. |
---|
600 | */ |
---|
601 | if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) { |
---|
602 | taop = &tao_noncached; |
---|
603 | bzero(taop, sizeof(*taop)); |
---|
604 | } |
---|
605 | |
---|
606 | tp->cc_send = CC_INC(tcp_ccgen); |
---|
607 | if (taop->tao_ccsent != 0 && |
---|
608 | CC_GEQ(tp->cc_send, taop->tao_ccsent)) { |
---|
609 | taop->tao_ccsent = tp->cc_send; |
---|
610 | } else { |
---|
611 | taop->tao_ccsent = 0; |
---|
612 | tp->t_flags |= TF_SENDCCNEW; |
---|
613 | } |
---|
614 | |
---|
615 | return 0; |
---|
616 | } |
---|
617 | |
---|
618 | int |
---|
619 | tcp_ctloutput(op, so, level, optname, mp) |
---|
620 | int op; |
---|
621 | struct socket *so; |
---|
622 | int level, optname; |
---|
623 | struct mbuf **mp; |
---|
624 | { |
---|
625 | int error = 0, s; |
---|
626 | struct inpcb *inp; |
---|
627 | register struct tcpcb *tp; |
---|
628 | register struct mbuf *m; |
---|
629 | register int i; |
---|
630 | |
---|
631 | s = splnet(); |
---|
632 | inp = sotoinpcb(so); |
---|
633 | if (inp == NULL) { |
---|
634 | splx(s); |
---|
635 | if (op == PRCO_SETOPT && *mp) |
---|
636 | (void) m_free(*mp); |
---|
637 | return (ECONNRESET); |
---|
638 | } |
---|
639 | if (level != IPPROTO_TCP) { |
---|
640 | error = ip_ctloutput(op, so, level, optname, mp); |
---|
641 | splx(s); |
---|
642 | return (error); |
---|
643 | } |
---|
644 | tp = intotcpcb(inp); |
---|
645 | |
---|
646 | switch (op) { |
---|
647 | |
---|
648 | case PRCO_SETOPT: |
---|
649 | m = *mp; |
---|
650 | switch (optname) { |
---|
651 | |
---|
652 | case TCP_NODELAY: |
---|
653 | if (m == NULL || m->m_len < sizeof (int)) |
---|
654 | error = EINVAL; |
---|
655 | else if (*mtod(m, int *)) |
---|
656 | tp->t_flags |= TF_NODELAY; |
---|
657 | else |
---|
658 | tp->t_flags &= ~TF_NODELAY; |
---|
659 | break; |
---|
660 | |
---|
661 | case TCP_MAXSEG: |
---|
662 | if (m && (i = *mtod(m, int *)) > 0 && i <= tp->t_maxseg) |
---|
663 | tp->t_maxseg = i; |
---|
664 | else |
---|
665 | error = EINVAL; |
---|
666 | break; |
---|
667 | |
---|
668 | case TCP_NOOPT: |
---|
669 | if (m == NULL || m->m_len < sizeof (int)) |
---|
670 | error = EINVAL; |
---|
671 | else if (*mtod(m, int *)) |
---|
672 | tp->t_flags |= TF_NOOPT; |
---|
673 | else |
---|
674 | tp->t_flags &= ~TF_NOOPT; |
---|
675 | break; |
---|
676 | |
---|
677 | case TCP_NOPUSH: |
---|
678 | if (m == NULL || m->m_len < sizeof (int)) |
---|
679 | error = EINVAL; |
---|
680 | else if (*mtod(m, int *)) |
---|
681 | tp->t_flags |= TF_NOPUSH; |
---|
682 | else |
---|
683 | tp->t_flags &= ~TF_NOPUSH; |
---|
684 | break; |
---|
685 | |
---|
686 | default: |
---|
687 | error = ENOPROTOOPT; |
---|
688 | break; |
---|
689 | } |
---|
690 | if (m) |
---|
691 | (void) m_free(m); |
---|
692 | break; |
---|
693 | |
---|
694 | case PRCO_GETOPT: |
---|
695 | *mp = m = m_get(M_WAIT, MT_SOOPTS); |
---|
696 | m->m_len = sizeof(int); |
---|
697 | |
---|
698 | switch (optname) { |
---|
699 | case TCP_NODELAY: |
---|
700 | *mtod(m, int *) = tp->t_flags & TF_NODELAY; |
---|
701 | break; |
---|
702 | case TCP_MAXSEG: |
---|
703 | *mtod(m, int *) = tp->t_maxseg; |
---|
704 | break; |
---|
705 | case TCP_NOOPT: |
---|
706 | *mtod(m, int *) = tp->t_flags & TF_NOOPT; |
---|
707 | break; |
---|
708 | case TCP_NOPUSH: |
---|
709 | *mtod(m, int *) = tp->t_flags & TF_NOPUSH; |
---|
710 | break; |
---|
711 | default: |
---|
712 | error = ENOPROTOOPT; |
---|
713 | break; |
---|
714 | } |
---|
715 | break; |
---|
716 | } |
---|
717 | splx(s); |
---|
718 | return (error); |
---|
719 | } |
---|
720 | |
---|
721 | /* |
---|
722 | * tcp_sendspace and tcp_recvspace are the default send and receive window |
---|
723 | * sizes, respectively. These are obsolescent (this information should |
---|
724 | * be set by the route). |
---|
725 | */ |
---|
726 | u_long tcp_sendspace = 1024*16; |
---|
727 | SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, |
---|
728 | CTLFLAG_RW, &tcp_sendspace , 0, ""); |
---|
729 | u_long tcp_recvspace = 1024*16; |
---|
730 | SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, |
---|
731 | CTLFLAG_RW, &tcp_recvspace , 0, ""); |
---|
732 | |
---|
733 | /* |
---|
734 | * Attach TCP protocol to socket, allocating |
---|
735 | * internet protocol control block, tcp control block, |
---|
736 | * bufer space, and entering LISTEN state if to accept connections. |
---|
737 | */ |
---|
738 | static int |
---|
739 | tcp_attach(so) |
---|
740 | struct socket *so; |
---|
741 | { |
---|
742 | register struct tcpcb *tp; |
---|
743 | struct inpcb *inp; |
---|
744 | int error; |
---|
745 | |
---|
746 | if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { |
---|
747 | error = soreserve(so, tcp_sendspace, tcp_recvspace); |
---|
748 | if (error) |
---|
749 | return (error); |
---|
750 | } |
---|
751 | error = in_pcballoc(so, &tcbinfo); |
---|
752 | if (error) |
---|
753 | return (error); |
---|
754 | inp = sotoinpcb(so); |
---|
755 | tp = tcp_newtcpcb(inp); |
---|
756 | if (tp == 0) { |
---|
757 | int nofd = so->so_state & SS_NOFDREF; /* XXX */ |
---|
758 | |
---|
759 | so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ |
---|
760 | in_pcbdetach(inp); |
---|
761 | so->so_state |= nofd; |
---|
762 | return (ENOBUFS); |
---|
763 | } |
---|
764 | tp->t_state = TCPS_CLOSED; |
---|
765 | return (0); |
---|
766 | } |
---|
767 | |
---|
768 | /* |
---|
769 | * Initiate (or continue) disconnect. |
---|
770 | * If embryonic state, just send reset (once). |
---|
771 | * If in ``let data drain'' option and linger null, just drop. |
---|
772 | * Otherwise (hard), mark socket disconnecting and drop |
---|
773 | * current input data; switch states based on user close, and |
---|
774 | * send segment to peer (with FIN). |
---|
775 | */ |
---|
776 | static struct tcpcb * |
---|
777 | tcp_disconnect(tp) |
---|
778 | register struct tcpcb *tp; |
---|
779 | { |
---|
780 | struct socket *so = tp->t_inpcb->inp_socket; |
---|
781 | |
---|
782 | if (tp->t_state < TCPS_ESTABLISHED) |
---|
783 | tp = tcp_close(tp); |
---|
784 | else if ((so->so_options & SO_LINGER) && so->so_linger == 0) |
---|
785 | tp = tcp_drop(tp, 0); |
---|
786 | else { |
---|
787 | soisdisconnecting(so); |
---|
788 | sbflush(&so->so_rcv); |
---|
789 | tp = tcp_usrclosed(tp); |
---|
790 | if (tp) |
---|
791 | (void) tcp_output(tp); |
---|
792 | } |
---|
793 | return (tp); |
---|
794 | } |
---|
795 | |
---|
796 | /* |
---|
797 | * User issued close, and wish to trail through shutdown states: |
---|
798 | * if never received SYN, just forget it. If got a SYN from peer, |
---|
799 | * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. |
---|
800 | * If already got a FIN from peer, then almost done; go to LAST_ACK |
---|
801 | * state. In all other cases, have already sent FIN to peer (e.g. |
---|
802 | * after PRU_SHUTDOWN), and just have to play tedious game waiting |
---|
803 | * for peer to send FIN or not respond to keep-alives, etc. |
---|
804 | * We can let the user exit from the close as soon as the FIN is acked. |
---|
805 | */ |
---|
806 | static struct tcpcb * |
---|
807 | tcp_usrclosed(tp) |
---|
808 | register struct tcpcb *tp; |
---|
809 | { |
---|
810 | |
---|
811 | switch (tp->t_state) { |
---|
812 | |
---|
813 | case TCPS_CLOSED: |
---|
814 | case TCPS_LISTEN: |
---|
815 | tp->t_state = TCPS_CLOSED; |
---|
816 | tp = tcp_close(tp); |
---|
817 | break; |
---|
818 | |
---|
819 | case TCPS_SYN_SENT: |
---|
820 | case TCPS_SYN_RECEIVED: |
---|
821 | tp->t_flags |= TF_NEEDFIN; |
---|
822 | break; |
---|
823 | |
---|
824 | case TCPS_ESTABLISHED: |
---|
825 | tp->t_state = TCPS_FIN_WAIT_1; |
---|
826 | break; |
---|
827 | |
---|
828 | case TCPS_CLOSE_WAIT: |
---|
829 | tp->t_state = TCPS_LAST_ACK; |
---|
830 | break; |
---|
831 | } |
---|
832 | if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { |
---|
833 | soisdisconnected(tp->t_inpcb->inp_socket); |
---|
834 | /* To prevent the connection hanging in FIN_WAIT_2 forever. */ |
---|
835 | if (tp->t_state == TCPS_FIN_WAIT_2) |
---|
836 | tp->t_timer[TCPT_2MSL] = tcp_maxidle; |
---|
837 | } |
---|
838 | return (tp); |
---|
839 | } |
---|