]> git.proxmox.com Git - mirror_qemu.git/blame - slirp/tcp_subr.c
slirp: add unregister_poll_fd() callback
[mirror_qemu.git] / slirp / tcp_subr.c
CommitLineData
f0cbd3ec
FB
1/*
2 * Copyright (c) 1982, 1986, 1988, 1990, 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.
2f5f8996 13 * 3. Neither the name of the University nor the names of its contributors
f0cbd3ec
FB
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 * @(#)tcp_subr.c 8.1 (Berkeley) 6/10/93
30 * tcp_subr.c,v 1.5 1994/10/08 22:39:58 phk Exp
31 */
32
33/*
34 * Changes and additions relating to SLiRP
35 * Copyright (c) 1995 Danny Gasparovski.
5fafdf24
TS
36 *
37 * Please read the file COPYRIGHT for the
f0cbd3ec
FB
38 * terms and conditions of the copyright.
39 */
40
7df7482b 41#include "qemu/osdep.h"
a9c94277 42#include "slirp.h"
f0cbd3ec
FB
43
44/* patchable/settable parameters for tcp */
9634d903
BS
45/* Don't do rfc1323 performance enhancements */
46#define TCP_DO_RFC1323 0
f0cbd3ec
FB
47
48/*
49 * Tcp initialization
50 */
51void
460fec67 52tcp_init(Slirp *slirp)
f0cbd3ec 53{
460fec67
JK
54 slirp->tcp_iss = 1; /* wrong */
55 slirp->tcb.so_next = slirp->tcb.so_prev = &slirp->tcb;
56 slirp->tcp_last_so = &slirp->tcb;
f0cbd3ec
FB
57}
58
a68adc22
JK
59void tcp_cleanup(Slirp *slirp)
60{
61 while (slirp->tcb.so_next != &slirp->tcb) {
62 tcp_close(sototcpcb(slirp->tcb.so_next));
63 }
64}
65
f0cbd3ec
FB
66/*
67 * Create template to be used to send tcp packets on a connection.
68 * Call after host entry created, fills
69 * in a skeletal tcp/ip header, minimizing the amount of work
70 * necessary when the connection is used.
71 */
f0cbd3ec 72void
511d2b14 73tcp_template(struct tcpcb *tp)
f0cbd3ec
FB
74{
75 struct socket *so = tp->t_socket;
76 register struct tcpiphdr *n = &tp->t_template;
77
429d0a3d 78 n->ti_mbuf = NULL;
98c63057
GS
79 memset(&n->ti, 0, sizeof(n->ti));
80 n->ti_x0 = 0;
9dfbf250
GS
81 switch (so->so_ffamily) {
82 case AF_INET:
1252cf40
GS
83 n->ti_pr = IPPROTO_TCP;
84 n->ti_len = htons(sizeof(struct tcphdr));
85 n->ti_src = so->so_faddr;
86 n->ti_dst = so->so_laddr;
87 n->ti_sport = so->so_fport;
88 n->ti_dport = so->so_lport;
9dfbf250
GS
89 break;
90
3feea444
GS
91 case AF_INET6:
92 n->ti_nh6 = IPPROTO_TCP;
93 n->ti_len = htons(sizeof(struct tcphdr));
94 n->ti_src6 = so->so_faddr6;
95 n->ti_dst6 = so->so_laddr6;
96 n->ti_sport = so->so_fport6;
97 n->ti_dport = so->so_lport6;
98 break;
99
9dfbf250
GS
100 default:
101 g_assert_not_reached();
102 }
5fafdf24 103
f0cbd3ec
FB
104 n->ti_seq = 0;
105 n->ti_ack = 0;
106 n->ti_x2 = 0;
107 n->ti_off = 5;
108 n->ti_flags = 0;
109 n->ti_win = 0;
110 n->ti_sum = 0;
111 n->ti_urp = 0;
112}
113
114/*
115 * Send a single message to the TCP at address specified by
116 * the given TCP/IP header. If m == 0, then we make a copy
117 * of the tcpiphdr at ti and send directly to the addressed host.
118 * This is used to force keep alive messages out using the TCP
119 * template for a connection tp->t_template. If flags are given
120 * then we send a message back to the TCP which originated the
121 * segment ti, and discard the mbuf containing it and any other
122 * attached mbufs.
123 *
124 * In any case the ack and sequence number of the transmitted
125 * segment are as specified by the parameters.
126 */
127void
511d2b14 128tcp_respond(struct tcpcb *tp, struct tcpiphdr *ti, struct mbuf *m,
9dfbf250 129 tcp_seq ack, tcp_seq seq, int flags, unsigned short af)
f0cbd3ec
FB
130{
131 register int tlen;
132 int win = 0;
133
134 DEBUG_CALL("tcp_respond");
c4d12a74
SW
135 DEBUG_ARG("tp = %p", tp);
136 DEBUG_ARG("ti = %p", ti);
137 DEBUG_ARG("m = %p", m);
f0cbd3ec
FB
138 DEBUG_ARG("ack = %u", ack);
139 DEBUG_ARG("seq = %u", seq);
140 DEBUG_ARG("flags = %x", flags);
5fafdf24 141
f0cbd3ec
FB
142 if (tp)
143 win = sbspace(&tp->t_socket->so_rcv);
511d2b14 144 if (m == NULL) {
e56afbc5 145 if (!tp || (m = m_get(tp->t_socket->slirp)) == NULL)
f0cbd3ec 146 return;
f0cbd3ec 147 tlen = 0;
9634d903 148 m->m_data += IF_MAXLINKHDR;
f0cbd3ec
FB
149 *mtod(m, struct tcpiphdr *) = *ti;
150 ti = mtod(m, struct tcpiphdr *);
990132cd
TW
151 switch (af) {
152 case AF_INET:
153 ti->ti.ti_i4.ih_x1 = 0;
154 break;
155 case AF_INET6:
156 ti->ti.ti_i6.ih_x1 = 0;
157 break;
158 default:
159 g_assert_not_reached();
160 }
f0cbd3ec
FB
161 flags = TH_ACK;
162 } else {
5fafdf24 163 /*
f0cbd3ec
FB
164 * ti points into m so the next line is just making
165 * the mbuf point to ti
166 */
167 m->m_data = (caddr_t)ti;
3b46e624 168
f0cbd3ec
FB
169 m->m_len = sizeof (struct tcpiphdr);
170 tlen = 0;
171#define xchg(a,b,type) { type t; t=a; a=b; b=t; }
9dfbf250
GS
172 switch (af) {
173 case AF_INET:
1252cf40
GS
174 xchg(ti->ti_dst.s_addr, ti->ti_src.s_addr, uint32_t);
175 xchg(ti->ti_dport, ti->ti_sport, uint16_t);
9dfbf250 176 break;
3feea444
GS
177 case AF_INET6:
178 xchg(ti->ti_dst6, ti->ti_src6, struct in6_addr);
179 xchg(ti->ti_dport, ti->ti_sport, uint16_t);
180 break;
9dfbf250
GS
181 default:
182 g_assert_not_reached();
183 }
f0cbd3ec
FB
184#undef xchg
185 }
186 ti->ti_len = htons((u_short)(sizeof (struct tcphdr) + tlen));
187 tlen += sizeof (struct tcpiphdr);
188 m->m_len = tlen;
189
98c63057
GS
190 ti->ti_mbuf = NULL;
191 ti->ti_x0 = 0;
f0cbd3ec
FB
192 ti->ti_seq = htonl(seq);
193 ti->ti_ack = htonl(ack);
194 ti->ti_x2 = 0;
195 ti->ti_off = sizeof (struct tcphdr) >> 2;
196 ti->ti_flags = flags;
197 if (tp)
b6dce92e 198 ti->ti_win = htons((uint16_t) (win >> tp->rcv_scale));
f0cbd3ec 199 else
b6dce92e 200 ti->ti_win = htons((uint16_t)win);
f0cbd3ec
FB
201 ti->ti_urp = 0;
202 ti->ti_sum = 0;
203 ti->ti_sum = cksum(m, tlen);
f0cbd3ec 204
98c63057 205 struct tcpiphdr tcpiph_save = *(mtod(m, struct tcpiphdr *));
9dfbf250 206 struct ip *ip;
3feea444 207 struct ip6 *ip6;
9dfbf250
GS
208
209 switch (af) {
210 case AF_INET:
1252cf40
GS
211 m->m_data += sizeof(struct tcpiphdr) - sizeof(struct tcphdr)
212 - sizeof(struct ip);
213 m->m_len -= sizeof(struct tcpiphdr) - sizeof(struct tcphdr)
214 - sizeof(struct ip);
215 ip = mtod(m, struct ip *);
2e30230a 216 ip->ip_len = m->m_len;
1252cf40
GS
217 ip->ip_dst = tcpiph_save.ti_dst;
218 ip->ip_src = tcpiph_save.ti_src;
219 ip->ip_p = tcpiph_save.ti_pr;
220
221 if (flags & TH_RST) {
222 ip->ip_ttl = MAXTTL;
223 } else {
224 ip->ip_ttl = IPDEFTTL;
225 }
226
3feea444
GS
227 ip_output(NULL, m);
228 break;
229
230 case AF_INET6:
231 m->m_data += sizeof(struct tcpiphdr) - sizeof(struct tcphdr)
232 - sizeof(struct ip6);
233 m->m_len -= sizeof(struct tcpiphdr) - sizeof(struct tcphdr)
234 - sizeof(struct ip6);
235 ip6 = mtod(m, struct ip6 *);
2e30230a 236 ip6->ip_pl = tcpiph_save.ti_len;
3feea444
GS
237 ip6->ip_dst = tcpiph_save.ti_dst6;
238 ip6->ip_src = tcpiph_save.ti_src6;
239 ip6->ip_nh = tcpiph_save.ti_nh6;
240
241 ip6_output(NULL, m, 0);
9dfbf250
GS
242 break;
243
244 default:
245 g_assert_not_reached();
246 }
f0cbd3ec
FB
247}
248
249/*
250 * Create a new TCP control block, making an
251 * empty reassembly queue and hooking it to the argument
252 * protocol control block.
253 */
254struct tcpcb *
511d2b14 255tcp_newtcpcb(struct socket *so)
f0cbd3ec
FB
256{
257 register struct tcpcb *tp;
5fafdf24 258
f0cbd3ec
FB
259 tp = (struct tcpcb *)malloc(sizeof(*tp));
260 if (tp == NULL)
261 return ((struct tcpcb *)0);
5fafdf24 262
f0cbd3ec 263 memset((char *) tp, 0, sizeof(struct tcpcb));
429d0a3d 264 tp->seg_next = tp->seg_prev = (struct tcpiphdr*)tp;
3feea444 265 tp->t_maxseg = (so->so_ffamily == AF_INET) ? TCP_MSS : TCP6_MSS;
5fafdf24 266
9634d903 267 tp->t_flags = TCP_DO_RFC1323 ? (TF_REQ_SCALE|TF_REQ_TSTMP) : 0;
f0cbd3ec 268 tp->t_socket = so;
5fafdf24 269
f0cbd3ec
FB
270 /*
271 * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no
272 * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives
273 * reasonable initial retransmit time.
274 */
275 tp->t_srtt = TCPTV_SRTTBASE;
9634d903 276 tp->t_rttvar = TCPTV_SRTTDFLT << 2;
f0cbd3ec
FB
277 tp->t_rttmin = TCPTV_MIN;
278
5fafdf24 279 TCPT_RANGESET(tp->t_rxtcur,
f0cbd3ec
FB
280 ((TCPTV_SRTTBASE >> 2) + (TCPTV_SRTTDFLT << 2)) >> 1,
281 TCPTV_MIN, TCPTV_REXMTMAX);
282
283 tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT;
284 tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT;
285 tp->t_state = TCPS_CLOSED;
5fafdf24 286
f0cbd3ec
FB
287 so->so_tcpcb = tp;
288
289 return (tp);
290}
291
292/*
293 * Drop a TCP connection, reporting
294 * the specified error. If connection is synchronized,
295 * then send a RST to peer.
296 */
5fafdf24 297struct tcpcb *tcp_drop(struct tcpcb *tp, int err)
f0cbd3ec 298{
f0cbd3ec 299 DEBUG_CALL("tcp_drop");
ecc804ca 300 DEBUG_ARG("tp = %p", tp);
f0cbd3ec 301 DEBUG_ARG("errno = %d", errno);
5fafdf24 302
f0cbd3ec
FB
303 if (TCPS_HAVERCVDSYN(tp->t_state)) {
304 tp->t_state = TCPS_CLOSED;
305 (void) tcp_output(tp);
0fe6a7f2 306 }
f0cbd3ec
FB
307 return (tcp_close(tp));
308}
309
310/*
311 * Close a TCP control block:
312 * discard all space held by the tcp
313 * discard internet protocol block
314 * wake up any sleepers
315 */
316struct tcpcb *
511d2b14 317tcp_close(struct tcpcb *tp)
f0cbd3ec
FB
318{
319 register struct tcpiphdr *t;
320 struct socket *so = tp->t_socket;
460fec67 321 Slirp *slirp = so->slirp;
f0cbd3ec
FB
322 register struct mbuf *m;
323
324 DEBUG_CALL("tcp_close");
ecc804ca 325 DEBUG_ARG("tp = %p", tp);
5fafdf24 326
f0cbd3ec 327 /* free the reassembly queue, if any */
429d0a3d
BS
328 t = tcpfrag_list_first(tp);
329 while (!tcpfrag_list_end(t, tp)) {
330 t = tcpiphdr_next(t);
331 m = tcpiphdr_prev(t)->ti_mbuf;
332 remque(tcpiphdr2qlink(tcpiphdr_prev(t)));
3acccfc6 333 m_free(m);
f0cbd3ec 334 }
f0cbd3ec 335 free(tp);
511d2b14 336 so->so_tcpcb = NULL;
f0cbd3ec 337 /* clobber input socket cache if we're closing the cached connection */
460fec67
JK
338 if (so == slirp->tcp_last_so)
339 slirp->tcp_last_so = &slirp->tcb;
f6e5aa36 340 so->slirp->cb->unregister_poll_fd(so->s);
707bd47e 341 slirp_closesocket(so->s);
f0cbd3ec
FB
342 sbfree(&so->so_rcv);
343 sbfree(&so->so_snd);
344 sofree(so);
f0cbd3ec
FB
345 return ((struct tcpcb *)0);
346}
347
f0cbd3ec
FB
348/*
349 * TCP protocol interface to socket abstraction.
350 */
351
352/*
353 * User issued close, and wish to trail through shutdown states:
354 * if never received SYN, just forget it. If got a SYN from peer,
355 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
356 * If already got a FIN from peer, then almost done; go to LAST_ACK
357 * state. In all other cases, have already sent FIN to peer (e.g.
358 * after PRU_SHUTDOWN), and just have to play tedious game waiting
359 * for peer to send FIN or not respond to keep-alives, etc.
360 * We can let the user exit from the close as soon as the FIN is acked.
361 */
362void
511d2b14 363tcp_sockclosed(struct tcpcb *tp)
f0cbd3ec
FB
364{
365
366 DEBUG_CALL("tcp_sockclosed");
ecc804ca 367 DEBUG_ARG("tp = %p", tp);
5fafdf24 368
b5ab6771
SL
369 if (!tp) {
370 return;
371 }
372
f0cbd3ec
FB
373 switch (tp->t_state) {
374
375 case TCPS_CLOSED:
376 case TCPS_LISTEN:
377 case TCPS_SYN_SENT:
378 tp->t_state = TCPS_CLOSED;
379 tp = tcp_close(tp);
380 break;
381
382 case TCPS_SYN_RECEIVED:
383 case TCPS_ESTABLISHED:
384 tp->t_state = TCPS_FIN_WAIT_1;
385 break;
386
387 case TCPS_CLOSE_WAIT:
388 tp->t_state = TCPS_LAST_ACK;
389 break;
390 }
b5ab6771 391 tcp_output(tp);
f0cbd3ec
FB
392}
393
5fafdf24 394/*
f0cbd3ec
FB
395 * Connect to a host on the Internet
396 * Called by tcp_input
397 * Only do a connect, the tcp fields will be set in tcp_input
398 * return 0 if there's a result of the connect,
399 * else return -1 means we're still connecting
400 * The return value is almost always -1 since the socket is
5fafdf24 401 * nonblocking. Connect returns after the SYN is sent, and does
f0cbd3ec
FB
402 * not wait for ACK+SYN.
403 */
cc573a69 404int tcp_fconnect(struct socket *so, unsigned short af)
f0cbd3ec
FB
405{
406 int ret=0;
3b46e624 407
f0cbd3ec 408 DEBUG_CALL("tcp_fconnect");
ecc804ca 409 DEBUG_ARG("so = %p", so);
f0cbd3ec 410
707bd47e 411 ret = so->s = slirp_socket(af, SOCK_STREAM, 0);
cc573a69 412 if (ret >= 0) {
f0cbd3ec 413 int opt, s=so->s;
5379229a 414 struct sockaddr_storage addr;
f0cbd3ec 415
848c7092
MAL
416 slirp_set_nonblock(s);
417 so->slirp->cb->register_poll_fd(so->s);
707bd47e 418 slirp_socket_set_fast_reuse(s);
f0cbd3ec 419 opt = 1;
707bd47e 420 slirp_setsockopt(s, SOL_SOCKET, SO_OOBINLINE, &opt, sizeof(opt));
058665b9 421 opt = 1;
707bd47e 422 slirp_setsockopt(s, IPPROTO_TCP, TCP_NODELAY, &opt, sizeof(opt));
3b46e624 423
5379229a 424 addr = so->fhost.ss;
2afbb788 425 DEBUG_CALL(" connect()ing");
5379229a
GS
426 sotranslate_out(so, &addr);
427
f0cbd3ec 428 /* We don't care what port we get */
0d48dfed 429 ret = connect(s, (struct sockaddr *)&addr, sockaddr_size(&addr));
3b46e624 430
f0cbd3ec
FB
431 /*
432 * If it's not in progress, it failed, so we just return 0,
433 * without clearing SS_NOFDREF
434 */
435 soisfconnecting(so);
436 }
437
438 return(ret);
439}
440
441/*
442 * Accept the socket and connect to the local-host
5fafdf24 443 *
f0cbd3ec
FB
444 * We have a problem. The correct thing to do would be
445 * to first connect to the local-host, and only if the
446 * connection is accepted, then do an accept() here.
5fafdf24 447 * But, a) we need to know who's trying to connect
f0cbd3ec
FB
448 * to the socket to be able to SYN the local-host, and
449 * b) we are already connected to the foreign host by
450 * the time it gets to accept(), so... We simply accept
451 * here and SYN the local-host.
5fafdf24 452 */
4ef7b894 453void tcp_connect(struct socket *inso)
f0cbd3ec 454{
4ef7b894
MK
455 Slirp *slirp = inso->slirp;
456 struct socket *so;
9dfbf250
GS
457 struct sockaddr_storage addr;
458 socklen_t addrlen = sizeof(struct sockaddr_storage);
4ef7b894
MK
459 struct tcpcb *tp;
460 int s, opt;
f0cbd3ec 461
4ef7b894 462 DEBUG_CALL("tcp_connect");
ecc804ca 463 DEBUG_ARG("inso = %p", inso);
5fafdf24 464
4ef7b894
MK
465 /*
466 * If it's an SS_ACCEPTONCE socket, no need to socreate()
467 * another socket, just use the accept() socket.
468 */
469 if (inso->so_state & SS_FACCEPTONCE) {
470 /* FACCEPTONCE already have a tcpcb */
471 so = inso;
472 } else {
473 so = socreate(slirp);
4ef7b894 474 if (tcp_attach(so) < 0) {
84ec9bfa 475 g_free(so); /* NOT sofree */
4ef7b894
MK
476 return;
477 }
9dfbf250
GS
478 so->lhost = inso->lhost;
479 so->so_ffamily = inso->so_ffamily;
4ef7b894 480 }
5fafdf24 481
4ef7b894 482 tcp_mss(sototcpcb(so), 0);
f0cbd3ec 483
4ef7b894
MK
484 s = accept(inso->s, (struct sockaddr *)&addr, &addrlen);
485 if (s < 0) {
486 tcp_close(sototcpcb(so)); /* This will sofree() as well */
487 return;
488 }
848c7092
MAL
489 slirp_set_nonblock(s);
490 so->slirp->cb->register_poll_fd(so->s);
707bd47e 491 slirp_socket_set_fast_reuse(s);
4ef7b894 492 opt = 1;
707bd47e
MAL
493 slirp_setsockopt(s, SOL_SOCKET, SO_OOBINLINE, &opt, sizeof(int));
494 slirp_socket_set_nodelay(s);
4ef7b894 495
9dfbf250 496 so->fhost.ss = addr;
5379229a 497 sotranslate_accept(so);
5fafdf24 498
4ef7b894
MK
499 /* Close the accept() socket, set right state */
500 if (inso->so_state & SS_FACCEPTONCE) {
501 /* If we only accept once, close the accept() socket */
f6e5aa36 502 so->slirp->cb->unregister_poll_fd(so->s);
707bd47e 503 slirp_closesocket(so->s);
4ef7b894
MK
504
505 /* Don't select it yet, even though we have an FD */
506 /* if it's not FACCEPTONCE, it's already NOFDREF */
507 so->so_state = SS_NOFDREF;
508 }
509 so->s = s;
510 so->so_state |= SS_INCOMING;
5fafdf24 511
4ef7b894
MK
512 so->so_iptos = tcp_tos(so);
513 tp = sototcpcb(so);
f0cbd3ec 514
4ef7b894 515 tcp_template(tp);
5fafdf24 516
4ef7b894
MK
517 tp->t_state = TCPS_SYN_SENT;
518 tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
519 tp->iss = slirp->tcp_iss;
520 slirp->tcp_iss += TCP_ISSINCR/2;
521 tcp_sendseqinit(tp);
522 tcp_output(tp);
f0cbd3ec
FB
523}
524
525/*
526 * Attach a TCPCB to a socket.
527 */
528int
511d2b14 529tcp_attach(struct socket *so)
f0cbd3ec
FB
530{
531 if ((so->so_tcpcb = tcp_newtcpcb(so)) == NULL)
532 return -1;
5fafdf24 533
460fec67 534 insque(so, &so->slirp->tcb);
f0cbd3ec
FB
535
536 return 0;
537}
538
539/*
540 * Set the socket's type of service field
541 */
9634d903 542static const struct tos_t tcptos[] = {
f0cbd3ec
FB
543 {0, 20, IPTOS_THROUGHPUT, 0}, /* ftp data */
544 {21, 21, IPTOS_LOWDELAY, EMU_FTP}, /* ftp control */
545 {0, 23, IPTOS_LOWDELAY, 0}, /* telnet */
546 {0, 80, IPTOS_THROUGHPUT, 0}, /* WWW */
547 {0, 513, IPTOS_LOWDELAY, EMU_RLOGIN|EMU_NOCONNECT}, /* rlogin */
f0cbd3ec
FB
548 {0, 544, IPTOS_LOWDELAY, EMU_KSH}, /* kshell */
549 {0, 543, IPTOS_LOWDELAY, 0}, /* klogin */
550 {0, 6667, IPTOS_THROUGHPUT, EMU_IRC}, /* IRC */
551 {0, 6668, IPTOS_THROUGHPUT, EMU_IRC}, /* IRC undernet */
552 {0, 7070, IPTOS_LOWDELAY, EMU_REALAUDIO }, /* RealAudio control */
553 {0, 113, IPTOS_LOWDELAY, EMU_IDENT }, /* identd protocol */
554 {0, 0, 0, 0}
555};
556
0d62c4cf 557static struct emu_t *tcpemu = NULL;
3b46e624 558
f0cbd3ec
FB
559/*
560 * Return TOS according to the above table
561 */
b6dce92e 562uint8_t
511d2b14 563tcp_tos(struct socket *so)
f0cbd3ec
FB
564{
565 int i = 0;
566 struct emu_t *emup;
5fafdf24 567
f0cbd3ec
FB
568 while(tcptos[i].tos) {
569 if ((tcptos[i].fport && (ntohs(so->so_fport) == tcptos[i].fport)) ||
570 (tcptos[i].lport && (ntohs(so->so_lport) == tcptos[i].lport))) {
571 so->so_emu = tcptos[i].emu;
572 return tcptos[i].tos;
573 }
574 i++;
575 }
5fafdf24 576
f0cbd3ec
FB
577 /* Nope, lets see if there's a user-added one */
578 for (emup = tcpemu; emup; emup = emup->next) {
579 if ((emup->fport && (ntohs(so->so_fport) == emup->fport)) ||
580 (emup->lport && (ntohs(so->so_lport) == emup->lport))) {
581 so->so_emu = emup->emu;
582 return emup->tos;
583 }
584 }
5fafdf24 585
f0cbd3ec
FB
586 return 0;
587}
588
f0cbd3ec
FB
589/*
590 * Emulate programs that try and connect to us
591 * This includes ftp (the data connection is
592 * initiated by the server) and IRC (DCC CHAT and
593 * DCC SEND) for now
5fafdf24 594 *
f0cbd3ec
FB
595 * NOTE: It's possible to crash SLiRP by sending it
596 * unstandard strings to emulate... if this is a problem,
597 * more checks are needed here
598 *
599 * XXX Assumes the whole command came in one packet
3b46e624 600 *
f0cbd3ec
FB
601 * XXX Some ftp clients will have their TOS set to
602 * LOWDELAY and so Nagel will kick in. Because of this,
603 * we'll get the first letter, followed by the rest, so
604 * we simply scan for ORT instead of PORT...
605 * DCC doesn't have this problem because there's other stuff
606 * in the packet before the DCC command.
5fafdf24
TS
607 *
608 * Return 1 if the mbuf m is still valid and should be
f0cbd3ec 609 * sbappend()ed
5fafdf24 610 *
f0cbd3ec
FB
611 * NOTE: if you return 0 you MUST m_free() the mbuf!
612 */
613int
511d2b14 614tcp_emu(struct socket *so, struct mbuf *m)
f0cbd3ec 615{
460fec67 616 Slirp *slirp = so->slirp;
f0cbd3ec 617 u_int n1, n2, n3, n4, n5, n6;
363a37d5 618 char buff[257];
b6dce92e 619 uint32_t laddr;
f0cbd3ec
FB
620 u_int lport;
621 char *bptr;
5fafdf24 622
f0cbd3ec 623 DEBUG_CALL("tcp_emu");
ecc804ca
SW
624 DEBUG_ARG("so = %p", so);
625 DEBUG_ARG("m = %p", m);
5fafdf24 626
f0cbd3ec
FB
627 switch(so->so_emu) {
628 int x, i;
3b46e624 629
f0cbd3ec
FB
630 case EMU_IDENT:
631 /*
632 * Identification protocol as per rfc-1413
633 */
3b46e624 634
f0cbd3ec
FB
635 {
636 struct socket *tmpso;
637 struct sockaddr_in addr;
b55266b5 638 socklen_t addrlen = sizeof(struct sockaddr_in);
f0cbd3ec 639 struct sbuf *so_rcv = &so->so_rcv;
3b46e624 640
a7104eda
PP
641 if (m->m_len > so_rcv->sb_datalen
642 - (so_rcv->sb_wptr - so_rcv->sb_data)) {
643 return 1;
644 }
645
f0cbd3ec
FB
646 memcpy(so_rcv->sb_wptr, m->m_data, m->m_len);
647 so_rcv->sb_wptr += m->m_len;
648 so_rcv->sb_rptr += m->m_len;
649 m->m_data[m->m_len] = 0; /* NULL terminate */
650 if (strchr(m->m_data, '\r') || strchr(m->m_data, '\n')) {
9634d903 651 if (sscanf(so_rcv->sb_data, "%u%*[ ,]%u", &n1, &n2) == 2) {
f0cbd3ec
FB
652 HTONS(n1);
653 HTONS(n2);
654 /* n2 is the one on our host */
460fec67
JK
655 for (tmpso = slirp->tcb.so_next;
656 tmpso != &slirp->tcb;
657 tmpso = tmpso->so_next) {
f0cbd3ec
FB
658 if (tmpso->so_laddr.s_addr == so->so_laddr.s_addr &&
659 tmpso->so_lport == n2 &&
660 tmpso->so_faddr.s_addr == so->so_faddr.s_addr &&
661 tmpso->so_fport == n1) {
662 if (getsockname(tmpso->s,
663 (struct sockaddr *)&addr, &addrlen) == 0)
664 n2 = ntohs(addr.sin_port);
665 break;
666 }
667 }
668 }
363a37d5
BS
669 so_rcv->sb_cc = snprintf(so_rcv->sb_data,
670 so_rcv->sb_datalen,
671 "%d,%d\r\n", n1, n2);
f0cbd3ec
FB
672 so_rcv->sb_rptr = so_rcv->sb_data;
673 so_rcv->sb_wptr = so_rcv->sb_data + so_rcv->sb_cc;
674 }
675 m_free(m);
676 return 0;
677 }
3b46e624 678
f0cbd3ec 679 case EMU_FTP: /* ftp */
511d2b14 680 *(m->m_data+m->m_len) = 0; /* NUL terminate for strstr */
f0cbd3ec
FB
681 if ((bptr = (char *)strstr(m->m_data, "ORT")) != NULL) {
682 /*
683 * Need to emulate the PORT command
3b46e624 684 */
9634d903 685 x = sscanf(bptr, "ORT %u,%u,%u,%u,%u,%u\r\n%256[^\177]",
f0cbd3ec
FB
686 &n1, &n2, &n3, &n4, &n5, &n6, buff);
687 if (x < 6)
688 return 1;
3b46e624 689
f0cbd3ec
FB
690 laddr = htonl((n1 << 24) | (n2 << 16) | (n3 << 8) | (n4));
691 lport = htons((n5 << 8) | (n6));
3b46e624 692
460fec67
JK
693 if ((so = tcp_listen(slirp, INADDR_ANY, 0, laddr,
694 lport, SS_FACCEPTONCE)) == NULL) {
f0cbd3ec 695 return 1;
460fec67 696 }
f0cbd3ec 697 n6 = ntohs(so->so_fport);
3b46e624 698
f0cbd3ec
FB
699 n5 = (n6 >> 8) & 0xff;
700 n6 &= 0xff;
3b46e624 701
f0cbd3ec 702 laddr = ntohl(so->so_faddr.s_addr);
3b46e624 703
f0cbd3ec
FB
704 n1 = ((laddr >> 24) & 0xff);
705 n2 = ((laddr >> 16) & 0xff);
706 n3 = ((laddr >> 8) & 0xff);
707 n4 = (laddr & 0xff);
3b46e624 708
f0cbd3ec 709 m->m_len = bptr - m->m_data; /* Adjust length */
0e44486c 710 m->m_len += snprintf(bptr, m->m_size - m->m_len,
363a37d5
BS
711 "ORT %d,%d,%d,%d,%d,%d\r\n%s",
712 n1, n2, n3, n4, n5, n6, x==7?buff:"");
f0cbd3ec
FB
713 return 1;
714 } else if ((bptr = (char *)strstr(m->m_data, "27 Entering")) != NULL) {
715 /*
716 * Need to emulate the PASV response
717 */
9634d903 718 x = sscanf(bptr, "27 Entering Passive Mode (%u,%u,%u,%u,%u,%u)\r\n%256[^\177]",
f0cbd3ec
FB
719 &n1, &n2, &n3, &n4, &n5, &n6, buff);
720 if (x < 6)
721 return 1;
3b46e624 722
f0cbd3ec
FB
723 laddr = htonl((n1 << 24) | (n2 << 16) | (n3 << 8) | (n4));
724 lport = htons((n5 << 8) | (n6));
3b46e624 725
460fec67
JK
726 if ((so = tcp_listen(slirp, INADDR_ANY, 0, laddr,
727 lport, SS_FACCEPTONCE)) == NULL) {
f0cbd3ec 728 return 1;
460fec67 729 }
f0cbd3ec 730 n6 = ntohs(so->so_fport);
3b46e624 731
f0cbd3ec
FB
732 n5 = (n6 >> 8) & 0xff;
733 n6 &= 0xff;
3b46e624 734
f0cbd3ec 735 laddr = ntohl(so->so_faddr.s_addr);
3b46e624 736
f0cbd3ec
FB
737 n1 = ((laddr >> 24) & 0xff);
738 n2 = ((laddr >> 16) & 0xff);
739 n3 = ((laddr >> 8) & 0xff);
740 n4 = (laddr & 0xff);
3b46e624 741
f0cbd3ec 742 m->m_len = bptr - m->m_data; /* Adjust length */
0e44486c 743 m->m_len += snprintf(bptr, m->m_size - m->m_len,
363a37d5
BS
744 "27 Entering Passive Mode (%d,%d,%d,%d,%d,%d)\r\n%s",
745 n1, n2, n3, n4, n5, n6, x==7?buff:"");
3b46e624 746
f0cbd3ec
FB
747 return 1;
748 }
3b46e624 749
f0cbd3ec 750 return 1;
3b46e624 751
f0cbd3ec
FB
752 case EMU_KSH:
753 /*
754 * The kshell (Kerberos rsh) and shell services both pass
755 * a local port port number to carry signals to the server
756 * and stderr to the client. It is passed at the beginning
757 * of the connection as a NUL-terminated decimal ASCII string.
758 */
759 so->so_emu = 0;
760 for (lport = 0, i = 0; i < m->m_len-1; ++i) {
761 if (m->m_data[i] < '0' || m->m_data[i] > '9')
762 return 1; /* invalid number */
763 lport *= 10;
764 lport += m->m_data[i] - '0';
765 }
766 if (m->m_data[m->m_len-1] == '\0' && lport != 0 &&
460fec67
JK
767 (so = tcp_listen(slirp, INADDR_ANY, 0, so->so_laddr.s_addr,
768 htons(lport), SS_FACCEPTONCE)) != NULL)
0e44486c 769 m->m_len = snprintf(m->m_data, m->m_size, "%d",
363a37d5 770 ntohs(so->so_fport)) + 1;
f0cbd3ec 771 return 1;
3b46e624 772
f0cbd3ec
FB
773 case EMU_IRC:
774 /*
775 * Need to emulate DCC CHAT, DCC SEND and DCC MOVE
776 */
777 *(m->m_data+m->m_len) = 0; /* NULL terminate the string for strstr */
778 if ((bptr = (char *)strstr(m->m_data, "DCC")) == NULL)
779 return 1;
3b46e624 780
f0cbd3ec
FB
781 /* The %256s is for the broken mIRC */
782 if (sscanf(bptr, "DCC CHAT %256s %u %u", buff, &laddr, &lport) == 3) {
460fec67
JK
783 if ((so = tcp_listen(slirp, INADDR_ANY, 0,
784 htonl(laddr), htons(lport),
785 SS_FACCEPTONCE)) == NULL) {
f0cbd3ec 786 return 1;
460fec67 787 }
f0cbd3ec 788 m->m_len = bptr - m->m_data; /* Adjust length */
0e44486c 789 m->m_len += snprintf(bptr, m->m_size,
363a37d5
BS
790 "DCC CHAT chat %lu %u%c\n",
791 (unsigned long)ntohl(so->so_faddr.s_addr),
792 ntohs(so->so_fport), 1);
f0cbd3ec 793 } else if (sscanf(bptr, "DCC SEND %256s %u %u %u", buff, &laddr, &lport, &n1) == 4) {
460fec67
JK
794 if ((so = tcp_listen(slirp, INADDR_ANY, 0,
795 htonl(laddr), htons(lport),
796 SS_FACCEPTONCE)) == NULL) {
f0cbd3ec 797 return 1;
460fec67 798 }
f0cbd3ec 799 m->m_len = bptr - m->m_data; /* Adjust length */
0e44486c 800 m->m_len += snprintf(bptr, m->m_size,
363a37d5
BS
801 "DCC SEND %s %lu %u %u%c\n", buff,
802 (unsigned long)ntohl(so->so_faddr.s_addr),
803 ntohs(so->so_fport), n1, 1);
f0cbd3ec 804 } else if (sscanf(bptr, "DCC MOVE %256s %u %u %u", buff, &laddr, &lport, &n1) == 4) {
460fec67
JK
805 if ((so = tcp_listen(slirp, INADDR_ANY, 0,
806 htonl(laddr), htons(lport),
807 SS_FACCEPTONCE)) == NULL) {
f0cbd3ec 808 return 1;
460fec67 809 }
f0cbd3ec 810 m->m_len = bptr - m->m_data; /* Adjust length */
0e44486c 811 m->m_len += snprintf(bptr, m->m_size,
363a37d5
BS
812 "DCC MOVE %s %lu %u %u%c\n", buff,
813 (unsigned long)ntohl(so->so_faddr.s_addr),
814 ntohs(so->so_fport), n1, 1);
f0cbd3ec
FB
815 }
816 return 1;
817
818 case EMU_REALAUDIO:
5fafdf24 819 /*
f0cbd3ec
FB
820 * RealAudio emulation - JP. We must try to parse the incoming
821 * data and try to find the two characters that contain the
822 * port number. Then we redirect an udp port and replace the
823 * number with the real port we got.
824 *
825 * The 1.0 beta versions of the player are not supported
826 * any more.
5fafdf24 827 *
f0cbd3ec 828 * A typical packet for player version 1.0 (release version):
3b46e624 829 *
5fafdf24 830 * 0000:50 4E 41 00 05
0d62c4cf 831 * 0000:00 01 00 02 1B D7 00 00 67 E6 6C DC 63 00 12 50 ........g.l.c..P
f0cbd3ec
FB
832 * 0010:4E 43 4C 49 45 4E 54 20 31 30 31 20 41 4C 50 48 NCLIENT 101 ALPH
833 * 0020:41 6C 00 00 52 00 17 72 61 66 69 6C 65 73 2F 76 Al..R..rafiles/v
834 * 0030:6F 61 2F 65 6E 67 6C 69 73 68 5F 2E 72 61 79 42 oa/english_.rayB
3b46e624 835 *
f0cbd3ec
FB
836 * Now the port number 0x1BD7 is found at offset 0x04 of the
837 * Now the port number 0x1BD7 is found at offset 0x04 of the
838 * second packet. This time we received five bytes first and
839 * then the rest. You never know how many bytes you get.
840 *
841 * A typical packet for player version 2.0 (beta):
3b46e624 842 *
0d62c4cf
JK
843 * 0000:50 4E 41 00 06 00 02 00 00 00 01 00 02 1B C1 00 PNA.............
844 * 0010:00 67 75 78 F5 63 00 0A 57 69 6E 32 2E 30 2E 30 .gux.c..Win2.0.0
f0cbd3ec
FB
845 * 0020:2E 35 6C 00 00 52 00 1C 72 61 66 69 6C 65 73 2F .5l..R..rafiles/
846 * 0030:77 65 62 73 69 74 65 2F 32 30 72 65 6C 65 61 73 website/20releas
847 * 0040:65 2E 72 61 79 53 00 00 06 36 42 e.rayS...6B
3b46e624 848 *
f0cbd3ec 849 * Port number 0x1BC1 is found at offset 0x0d.
3b46e624 850 *
f0cbd3ec
FB
851 * This is just a horrible switch statement. Variable ra tells
852 * us where we're going.
853 */
3b46e624 854
f0cbd3ec
FB
855 bptr = m->m_data;
856 while (bptr < m->m_data + m->m_len) {
857 u_short p;
858 static int ra = 0;
5fafdf24 859 char ra_tbl[4];
3b46e624 860
f0cbd3ec
FB
861 ra_tbl[0] = 0x50;
862 ra_tbl[1] = 0x4e;
863 ra_tbl[2] = 0x41;
864 ra_tbl[3] = 0;
3b46e624 865
f0cbd3ec
FB
866 switch (ra) {
867 case 0:
868 case 2:
869 case 3:
870 if (*bptr++ != ra_tbl[ra]) {
871 ra = 0;
872 continue;
873 }
874 break;
3b46e624 875
f0cbd3ec
FB
876 case 1:
877 /*
878 * We may get 0x50 several times, ignore them
879 */
880 if (*bptr == 0x50) {
881 ra = 1;
882 bptr++;
883 continue;
884 } else if (*bptr++ != ra_tbl[ra]) {
885 ra = 0;
886 continue;
887 }
888 break;
3b46e624 889
5fafdf24
TS
890 case 4:
891 /*
f0cbd3ec
FB
892 * skip version number
893 */
894 bptr++;
895 break;
3b46e624 896
5fafdf24 897 case 5:
f0cbd3ec
FB
898 /*
899 * The difference between versions 1.0 and
900 * 2.0 is here. For future versions of
901 * the player this may need to be modified.
902 */
903 if (*(bptr + 1) == 0x02)
904 bptr += 8;
905 else
906 bptr += 4;
3b46e624
TS
907 break;
908
f0cbd3ec
FB
909 case 6:
910 /* This is the field containing the port
911 * number that RA-player is listening to.
912 */
5fafdf24 913 lport = (((u_char*)bptr)[0] << 8)
f0cbd3ec 914 + ((u_char *)bptr)[1];
3b46e624 915 if (lport < 6970)
f0cbd3ec
FB
916 lport += 256; /* don't know why */
917 if (lport < 6970 || lport > 7170)
918 return 1; /* failed */
3b46e624 919
f0cbd3ec
FB
920 /* try to get udp port between 6970 - 7170 */
921 for (p = 6970; p < 7071; p++) {
460fec67 922 if (udp_listen(slirp, INADDR_ANY,
3c6a0580 923 htons(p),
f0cbd3ec
FB
924 so->so_laddr.s_addr,
925 htons(lport),
926 SS_FACCEPTONCE)) {
927 break;
928 }
929 }
930 if (p == 7071)
931 p = 0;
932 *(u_char *)bptr++ = (p >> 8) & 0xff;
369c86e7 933 *(u_char *)bptr = p & 0xff;
5fafdf24 934 ra = 0;
f0cbd3ec 935 return 1; /* port redirected, we're done */
3b46e624
TS
936 break;
937
f0cbd3ec 938 default:
3b46e624 939 ra = 0;
f0cbd3ec
FB
940 }
941 ra++;
942 }
3b46e624
TS
943 return 1;
944
f0cbd3ec
FB
945 default:
946 /* Ooops, not emulated, won't call tcp_emu again */
947 so->so_emu = 0;
948 return 1;
949 }
950}
951
952/*
953 * Do misc. config of SLiRP while its running.
954 * Return 0 if this connections is to be closed, 1 otherwise,
955 * return 2 if this is a command-line connection
956 */
b35725c5 957int tcp_ctl(struct socket *so)
f0cbd3ec 958{
460fec67 959 Slirp *slirp = so->slirp;
b35725c5 960 struct sbuf *sb = &so->so_snd;
5d300fc9 961 struct gfwd_list *ex_ptr;
b35725c5
JK
962
963 DEBUG_CALL("tcp_ctl");
ecc804ca 964 DEBUG_ARG("so = %p", so);
b35725c5 965
460fec67 966 if (so->so_faddr.s_addr != slirp->vhost_addr.s_addr) {
b35725c5 967 /* Check if it's pty_exec */
5d300fc9 968 for (ex_ptr = slirp->guestfwd_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
b35725c5 969 if (ex_ptr->ex_fport == so->so_fport &&
a13a4126 970 so->so_faddr.s_addr == ex_ptr->ex_addr.s_addr) {
44b4ff24 971 if (ex_ptr->write_cb) {
b35725c5 972 so->s = -1;
44b4ff24 973 so->guestfwd = ex_ptr;
b35725c5
JK
974 return 1;
975 }
226ea7a9 976 DEBUG_MISC(" executing %s", ex_ptr->ex_exec);
43bc7340 977 return fork_exec(so, ex_ptr->ex_exec);
b35725c5
JK
978 }
979 }
980 }
981 sb->sb_cc =
982 snprintf(sb->sb_wptr, sb->sb_datalen - (sb->sb_wptr - sb->sb_data),
983 "Error: No application configured.\r\n");
984 sb->sb_wptr += sb->sb_cc;
985 return 0;
f0cbd3ec 986}