tcp_subr.c@ 14529

Last change on this file since 14529 was 14522, checked in by vboxsync, 16 years ago
removed unused global tcbhead
Property svn:eol-style set to `native`
File size: 37.4 KB

Line
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.
13	* 3. All advertising materials mentioning features or use of this software
14	* must display the following acknowledgement:
15	* This product includes software developed by the University of
16	* California, Berkeley and its contributors.
17	* 4. Neither the name of the University nor the names of its contributors
18	* may be used to endorse or promote products derived from this software
19	* without specific prior written permission.
20	*
21	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31	* SUCH DAMAGE.
32	*
33	* @(#)tcp_subr.c 8.1 (Berkeley) 6/10/93
34	* tcp_subr.c,v 1.5 1994/10/08 22:39:58 phk Exp
35	*/
36
37	/*
38	* Changes and additions relating to SLiRP
39	* Copyright (c) 1995 Danny Gasparovski.
40	*
41	* Please read the file COPYRIGHT for the
42	* terms and conditions of the copyright.
43	*/
44
45	#define WANT_SYS_IOCTL_H
46	#include <slirp.h>
47
48
49	/*
50	* Tcp initialization
51	*/
52	void
53	tcp_init(PNATState pData)
54	{
55	tcp_iss = 1; /* wrong */
56	tcb.so_next = tcb.so_prev = &tcb;
57	tcp_last_so = &tcb;
58	#ifdef VBOX_WITH_BSD_TCP_REASS
59	tcp_reass_maxqlen = 48;
60	tcp_reass_maxseg = 256;
61	#endif /* VBOX_WITH_BSD_TCP_REASS */
62	}
63
64	/*
65	* Create template to be used to send tcp packets on a connection.
66	* Call after host entry created, fills
67	* in a skeletal tcp/ip header, minimizing the amount of work
68	* necessary when the connection is used.
69	*/
70	/* struct tcpiphdr * */
71	void
72	tcp_template(tp)
73	struct tcpcb *tp;
74	{
75	struct socket *so = tp->t_socket;
76	register struct tcpiphdr *n = &tp->t_template;
77
78	n->ti_next = n->ti_prev = 0;
79	n->ti_x1 = 0;
80	n->ti_pr = IPPROTO_TCP;
81	n->ti_len = htons(sizeof (struct tcpiphdr) - sizeof (struct ip));
82	n->ti_src = so->so_faddr;
83	n->ti_dst = so->so_laddr;
84	n->ti_sport = so->so_fport;
85	n->ti_dport = so->so_lport;
86
87	n->ti_seq = 0;
88	n->ti_ack = 0;
89	n->ti_x2 = 0;
90	n->ti_off = 5;
91	n->ti_flags = 0;
92	n->ti_win = 0;
93	n->ti_sum = 0;
94	n->ti_urp = 0;
95	}
96
97	/*
98	* Send a single message to the TCP at address specified by
99	* the given TCP/IP header. If m == 0, then we make a copy
100	* of the tcpiphdr at ti and send directly to the addressed host.
101	* This is used to force keep alive messages out using the TCP
102	* template for a connection tp->t_template. If flags are given
103	* then we send a message back to the TCP which originated the
104	* segment ti, and discard the mbuf containing it and any other
105	* attached mbufs.
106	*
107	* In any case the ack and sequence number of the transmitted
108	* segment are as specified by the parameters.
109	*/
110	void
111	tcp_respond(PNATState pData, struct tcpcb tp, struct tcpiphdr ti, struct mbuf *m, tcp_seq ack, tcp_seq seq, int flags)
112	{
113	register int tlen;
114	int win = 0;
115
116	DEBUG_CALL("tcp_respond");
117	DEBUG_ARG("tp = %lx", (long)tp);
118	DEBUG_ARG("ti = %lx", (long)ti);
119	DEBUG_ARG("m = %lx", (long)m);
120	DEBUG_ARG("ack = %u", ack);
121	DEBUG_ARG("seq = %u", seq);
122	DEBUG_ARG("flags = %x", flags);
123
124	if (tp)
125	win = sbspace(&tp->t_socket->so_rcv);
126	if (m == 0) {
127	if ((m = m_get(pData)) == NULL)
128	return;
129	#ifdef TCP_COMPAT_42
130	tlen = 1;
131	#else
132	tlen = 0;
133	#endif
134	m->m_data += if_maxlinkhdr;
135	mtod(m, struct tcpiphdr ) = *ti;
136	ti = mtod(m, struct tcpiphdr *);
137	flags = TH_ACK;
138	} else {
139	/*
140	* ti points into m so the next line is just making
141	* the mbuf point to ti
142	*/
143	m->m_data = (caddr_t)ti;
144
145	m->m_len = sizeof (struct tcpiphdr);
146	tlen = 0;
147	#define xchg(a,b,type) { type t; t=a; a=b; b=t; }
148	xchg(ti->ti_dst.s_addr, ti->ti_src.s_addr, u_int32_t);
149	xchg(ti->ti_dport, ti->ti_sport, u_int16_t);
150	#undef xchg
151	}
152	ti->ti_len = htons((u_short)(sizeof (struct tcphdr) + tlen));
153	tlen += sizeof (struct tcpiphdr);
154	m->m_len = tlen;
155
156	ti->ti_next = ti->ti_prev = 0;
157	ti->ti_x1 = 0;
158	ti->ti_seq = htonl(seq);
159	ti->ti_ack = htonl(ack);
160	ti->ti_x2 = 0;
161	ti->ti_off = sizeof (struct tcphdr) >> 2;
162	ti->ti_flags = flags;
163	if (tp)
164	ti->ti_win = htons((u_int16_t) (win >> tp->rcv_scale));
165	else
166	ti->ti_win = htons((u_int16_t)win);
167	ti->ti_urp = 0;
168	ti->ti_sum = 0;
169	ti->ti_sum = cksum(m, tlen);
170	((struct ip *)ti)->ip_len = tlen;
171
172	if(flags & TH_RST)
173	((struct ip *)ti)->ip_ttl = MAXTTL;
174	else
175	((struct ip *)ti)->ip_ttl = ip_defttl;
176
177	(void) ip_output(pData, (struct socket *)0, m);
178	}
179
180	/*
181	* Create a new TCP control block, making an
182	* empty reassembly queue and hooking it to the argument
183	* protocol control block.
184	*/
185	struct tcpcb *
186	tcp_newtcpcb(PNATState pData, struct socket *so)
187	{
188	register struct tcpcb *tp;
189
190	tp = (struct tcpcb )malloc(sizeof(tp));
191	if (tp == NULL)
192	return ((struct tcpcb *)0);
193
194	memset((char *) tp, 0, sizeof(struct tcpcb));
195	#ifndef VBOX_WITH_BSD_TCP_REASS
196	tp->seg_next = tp->seg_prev = ptr_to_u32(pData, (struct tcpiphdr *)tp);
197	#endif /* !VBOX_WITH_BSD_TCP_REASS */
198	tp->t_maxseg = tcp_mssdflt;
199
200	tp->t_flags = tcp_do_rfc1323 ? (TF_REQ_SCALE\|TF_REQ_TSTMP) : 0;
201	tp->t_socket = so;
202
203	/*
204	* Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no
205	* rtt estimate. Set rttvar so that srtt + 2 * rttvar gives
206	* reasonable initial retransmit time.
207	*/
208	tp->t_srtt = TCPTV_SRTTBASE;
209	tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << 2;
210	tp->t_rttmin = TCPTV_MIN;
211
212	TCPT_RANGESET(tp->t_rxtcur,
213	((TCPTV_SRTTBASE >> 2) + (TCPTV_SRTTDFLT << 2)) >> 1,
214	TCPTV_MIN, TCPTV_REXMTMAX);
215
216	tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT;
217	tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT;
218	tp->t_state = TCPS_CLOSED;
219
220	so->so_tcpcb = tp;
221
222	return (tp);
223	}
224
225	/*
226	* Drop a TCP connection, reporting
227	* the specified error. If connection is synchronized,
228	* then send a RST to peer.
229	*/
230	struct tcpcb tcp_drop(PNATState pData, struct tcpcb tp, int err)
231	{
232	/* tcp_drop(tp, errno)
233	register struct tcpcb *tp;
234	int errno;
235	{
236	*/
237
238	DEBUG_CALL("tcp_drop");
239	DEBUG_ARG("tp = %lx", (long)tp);
240	DEBUG_ARG("errno = %d", errno);
241
242	if (TCPS_HAVERCVDSYN(tp->t_state)) {
243	tp->t_state = TCPS_CLOSED;
244	(void) tcp_output(pData, tp);
245	tcpstat.tcps_drops++;
246	} else
247	tcpstat.tcps_conndrops++;
248	/* if (errno == ETIMEDOUT && tp->t_softerror)
249	* errno = tp->t_softerror;
250	*/
251	/* so->so_error = errno; */
252	return (tcp_close(pData, tp));
253	}
254
255	/*
256	* Close a TCP control block:
257	* discard all space held by the tcp
258	* discard internet protocol block
259	* wake up any sleepers
260	*/
261	struct tcpcb *
262	tcp_close(PNATState pData, register struct tcpcb *tp)
263	{
264	register struct tcpiphdr *t;
265	struct socket *so = tp->t_socket;
266	register struct mbuf *m;
267
268	#ifndef VBOX_WITH_BSD_TCP_REASS
269	DEBUG_CALL("tcp_close");
270	DEBUG_ARG("tp = %lx", (long )tp);
271
272	/* free the reassembly queue, if any */
273	t = u32_to_ptr(pData, tp->seg_next, struct tcpiphdr *);
274	while (t != (struct tcpiphdr *)tp) {
275	t = u32_to_ptr(pData, t->ti_next, struct tcpiphdr *);
276	m = REASS_MBUF_GET(u32_to_ptr(pData, t->ti_prev, struct tcpiphdr *));
277	remque_32(pData, u32_to_ptr(pData, t->ti_prev, struct tcpiphdr *));
278	m_freem(pData, m);
279	}
280	/* It's static */
281	/* if (tp->t_template)
282	* (void) m_free(dtom(tp->t_template));
283	*/
284	/* free(tp, M_PCB); */
285	u32ptr_done(pData, ptr_to_u32(pData, tp), tp);
286	#else /* VBOX_WITH_BSD_TCP_REASS */
287	struct tseg_qent *te;
288	DEBUG_CALL("tcp_close");
289	DEBUG_ARG("tp = %lx", (long )tp);
290	/XXX: freeing the reassembly queue /
291	LIST_FOREACH(te, &tp->t_segq, tqe_q) {
292	LIST_REMOVE(te, tqe_q);
293	m_freem(pData, te->tqe_m);
294	free(te);
295	tcp_reass_qsize--;
296	}
297	#endif /* VBOX_WITH_BSD_TCP_REASS */
298	free(tp);
299	so->so_tcpcb = 0;
300	soisfdisconnected(so);
301	/* clobber input socket cache if we're closing the cached connection */
302	if (so == tcp_last_so)
303	tcp_last_so = &tcb;
304	closesocket(so->s);
305	sbfree(&so->so_rcv);
306	sbfree(&so->so_snd);
307	sofree(pData, so);
308	tcpstat.tcps_closed++;
309	return ((struct tcpcb *)0);
310	}
311
312	void
313	tcp_drain()
314	{
315	/* XXX */
316	}
317
318	/*
319	* When a source quench is received, close congestion window
320	* to one segment. We will gradually open it again as we proceed.
321	*/
322
323	#ifdef notdef
324
325	void
326	tcp_quench(i, errno)
327
328	int errno;
329	{
330	struct tcpcb *tp = intotcpcb(inp);
331
332	if (tp)
333	tp->snd_cwnd = tp->t_maxseg;
334	}
335
336	#endif /* notdef */
337
338	/*
339	* TCP protocol interface to socket abstraction.
340	*/
341
342	/*
343	* User issued close, and wish to trail through shutdown states:
344	* if never received SYN, just forget it. If got a SYN from peer,
345	* but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
346	* If already got a FIN from peer, then almost done; go to LAST_ACK
347	* state. In all other cases, have already sent FIN to peer (e.g.
348	* after PRU_SHUTDOWN), and just have to play tedious game waiting
349	* for peer to send FIN or not respond to keep-alives, etc.
350	* We can let the user exit from the close as soon as the FIN is acked.
351	*/
352	void
353	tcp_sockclosed(PNATState pData, struct tcpcb *tp)
354	{
355
356	DEBUG_CALL("tcp_sockclosed");
357	DEBUG_ARG("tp = %lx", (long)tp);
358
359	switch (tp->t_state) {
360
361	case TCPS_CLOSED:
362	case TCPS_LISTEN:
363	case TCPS_SYN_SENT:
364	tp->t_state = TCPS_CLOSED;
365	tp = tcp_close(pData, tp);
366	break;
367
368	case TCPS_SYN_RECEIVED:
369	case TCPS_ESTABLISHED:
370	tp->t_state = TCPS_FIN_WAIT_1;
371	break;
372
373	case TCPS_CLOSE_WAIT:
374	tp->t_state = TCPS_LAST_ACK;
375	break;
376	}
377	/* soisfdisconnecting(tp->t_socket); */
378	if (tp && tp->t_state >= TCPS_FIN_WAIT_2)
379	soisfdisconnected(tp->t_socket);
380	if (tp)
381	tcp_output(pData, tp);
382	}
383
384	/*
385	* Connect to a host on the Internet
386	* Called by tcp_input
387	* Only do a connect, the tcp fields will be set in tcp_input
388	* return 0 if there's a result of the connect,
389	* else return -1 means we're still connecting
390	* The return value is almost always -1 since the socket is
391	* nonblocking. Connect returns after the SYN is sent, and does
392	* not wait for ACK+SYN.
393	*/
394	int tcp_fconnect(PNATState pData, struct socket *so)
395	{
396	int ret=0;
397
398	DEBUG_CALL("tcp_fconnect");
399	DEBUG_ARG("so = %lx", (long )so);
400
401	if( (ret=so->s=socket(AF_INET,SOCK_STREAM,0)) >= 0) {
402	int opt, s=so->s;
403	struct sockaddr_in addr;
404
405	fd_nonblock(s);
406	opt = 1;
407	setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&opt,sizeof(opt ));
408	opt = 1;
409	setsockopt(s,SOL_SOCKET,SO_OOBINLINE,(char *)&opt,sizeof(opt ));
410
411	addr.sin_family = AF_INET;
412	if ((so->so_faddr.s_addr & htonl(pData->netmask)) == special_addr.s_addr) {
413	/* It's an alias */
414	switch(ntohl(so->so_faddr.s_addr) & ~pData->netmask) {
415	case CTL_DNS:
416	if (!get_dns_addr(pData, &dns_addr))
417	addr.sin_addr = dns_addr;
418	else
419	addr.sin_addr = loopback_addr;
420	break;
421	case CTL_ALIAS:
422	default:
423	addr.sin_addr = loopback_addr;
424	break;
425	}
426	} else
427	addr.sin_addr = so->so_faddr;
428	addr.sin_port = so->so_fport;
429
430	DEBUG_MISC((dfd, " connect()ing, addr.sin_port=%d, "
431	"addr.sin_addr.s_addr=%.16s\n",
432	ntohs(addr.sin_port), inet_ntoa(addr.sin_addr)));
433	/* We don't care what port we get */
434	ret = connect(s,(struct sockaddr *)&addr,sizeof (addr));
435
436	/*
437	* If it's not in progress, it failed, so we just return 0,
438	* without clearing SS_NOFDREF
439	*/
440	soisfconnecting(so);
441	}
442
443	return(ret);
444	}
445
446	/*
447	* Accept the socket and connect to the local-host
448	*
449	* We have a problem. The correct thing to do would be
450	* to first connect to the local-host, and only if the
451	* connection is accepted, then do an accept() here.
452	* But, a) we need to know who's trying to connect
453	* to the socket to be able to SYN the local-host, and
454	* b) we are already connected to the foreign host by
455	* the time it gets to accept(), so... We simply accept
456	* here and SYN the local-host.
457	*/
458	void
459	tcp_connect(PNATState pData, struct socket *inso)
460	{
461	struct socket *so;
462	struct sockaddr_in addr;
463	socklen_t addrlen = sizeof(struct sockaddr_in);
464	struct tcpcb *tp;
465	int s, opt;
466
467	DEBUG_CALL("tcp_connect");
468	DEBUG_ARG("inso = %lx", (long)inso);
469
470	/*
471	* If it's an SS_ACCEPTONCE socket, no need to socreate()
472	* another socket, just use the accept() socket.
473	*/
474	if (inso->so_state & SS_FACCEPTONCE) {
475	/* FACCEPTONCE already have a tcpcb */
476	so = inso;
477	} else {
478	if ((so = socreate()) == NULL) {
479	/* If it failed, get rid of the pending connection */
480	closesocket(accept(inso->s,(struct sockaddr *)&addr,&addrlen));
481	return;
482	}
483	if (tcp_attach(pData, so) < 0) {
484	free(so); /* NOT sofree */
485	return;
486	}
487	so->so_laddr = inso->so_laddr;
488	so->so_lport = inso->so_lport;
489	}
490
491	(void) tcp_mss(pData, sototcpcb(so), 0);
492
493	if ((s = accept(inso->s,(struct sockaddr *)&addr,&addrlen)) < 0) {
494	tcp_close(pData, sototcpcb(so)); /* This will sofree() as well */
495	return;
496	}
497	fd_nonblock(s);
498	opt = 1;
499	setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&opt,sizeof(int));
500	opt = 1;
501	setsockopt(s,SOL_SOCKET,SO_OOBINLINE,(char *)&opt,sizeof(int));
502	opt = 1;
503	setsockopt(s,IPPROTO_TCP,TCP_NODELAY,(char *)&opt,sizeof(int));
504
505	so->so_fport = addr.sin_port;
506	so->so_faddr = addr.sin_addr;
507	/* Translate connections from localhost to the real hostname */
508	if (so->so_faddr.s_addr == 0 \|\| so->so_faddr.s_addr == loopback_addr.s_addr)
509	so->so_faddr = alias_addr;
510
511	/* Close the accept() socket, set right state */
512	if (inso->so_state & SS_FACCEPTONCE) {
513	closesocket(so->s); /* If we only accept once, close the accept() socket */
514	so->so_state = SS_NOFDREF; /* Don't select it yet, even though we have an FD */
515	/* if it's not FACCEPTONCE, it's already NOFDREF */
516	}
517	so->s = s;
518
519	so->so_iptos = tcp_tos(so);
520	tp = sototcpcb(so);
521
522	tcp_template(tp);
523
524	/* Compute window scaling to request. */
525	/* while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
526	* (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat)
527	* tp->request_r_scale++;
528	*/
529
530	/* soisconnecting(so); / / NOFDREF used instead */
531	tcpstat.tcps_connattempt++;
532
533	tp->t_state = TCPS_SYN_SENT;
534	tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
535	tp->iss = tcp_iss;
536	tcp_iss += TCP_ISSINCR/2;
537	tcp_sendseqinit(tp);
538	tcp_output(pData, tp);
539	}
540
541	/*
542	* Attach a TCPCB to a socket.
543	*/
544	int
545	tcp_attach(PNATState pData, struct socket *so)
546	{
547	if ((so->so_tcpcb = tcp_newtcpcb(pData, so)) == NULL)
548	return -1;
549
550	insque(pData, so, &tcb);
551
552	return 0;
553	}
554
555	/*
556	* Set the socket's type of service field
557	*/
558	static const struct tos_t tcptos[] = {
559	{0, 20, IPTOS_THROUGHPUT, 0}, /* ftp data */
560	{21, 21, IPTOS_LOWDELAY, EMU_FTP}, /* ftp control */
561	{0, 23, IPTOS_LOWDELAY, 0}, /* telnet */
562	{0, 80, IPTOS_THROUGHPUT, 0}, /* WWW */
563	{0, 513, IPTOS_LOWDELAY, EMU_RLOGIN\|EMU_NOCONNECT}, /* rlogin */
564	{0, 514, IPTOS_LOWDELAY, EMU_RSH\|EMU_NOCONNECT}, /* shell */
565	{0, 544, IPTOS_LOWDELAY, EMU_KSH}, /* kshell */
566	{0, 543, IPTOS_LOWDELAY, 0}, /* klogin */
567	{0, 6667, IPTOS_THROUGHPUT, EMU_IRC}, /* IRC */
568	{0, 6668, IPTOS_THROUGHPUT, EMU_IRC}, /* IRC undernet */
569	{0, 7070, IPTOS_LOWDELAY, EMU_REALAUDIO }, /* RealAudio control */
570	{0, 113, IPTOS_LOWDELAY, EMU_IDENT }, /* identd protocol */
571	{0, 0, 0, 0}
572	};
573
574	/*
575	* Return TOS according to the above table
576	*/
577	u_int8_t
578	tcp_tos(so)
579	struct socket *so;
580	{
581	int i = 0;
582
583	while(tcptos[i].tos) {
584	if ((tcptos[i].fport && (ntohs(so->so_fport) == tcptos[i].fport)) \|\|
585	(tcptos[i].lport && (ntohs(so->so_lport) == tcptos[i].lport))) {
586	so->so_emu = tcptos[i].emu;
587	return tcptos[i].tos;
588	}
589	i++;
590	}
591
592	return 0;
593	}
594
595	/*
596	* Emulate programs that try and connect to us
597	* This includes ftp (the data connection is
598	* initiated by the server) and IRC (DCC CHAT and
599	* DCC SEND) for now
600	*
601	* NOTE: It's possible to crash SLiRP by sending it
602	* unstandard strings to emulate... if this is a problem,
603	* more checks are needed here
604	*
605	* XXX Assumes the whole command came in one packet
606	*
607	* XXX Some ftp clients will have their TOS set to
608	* LOWDELAY and so Nagel will kick in. Because of this,
609	* we'll get the first letter, followed by the rest, so
610	* we simply scan for ORT instead of PORT...
611	* DCC doesn't have this problem because there's other stuff
612	* in the packet before the DCC command.
613	*
614	* Return 1 if the mbuf m is still valid and should be
615	* sbappend()ed
616	*
617	* NOTE: if you return 0 you MUST m_free() the mbuf!
618	*/
619	int
620	tcp_emu(PNATState pData, struct socket so, struct mbuf m)
621	{
622	u_int n1, n2, n3, n4, n5, n6;
623	char buff[256];
624	u_int32_t laddr;
625	u_int lport;
626	char *bptr;
627
628	DEBUG_CALL("tcp_emu");
629	DEBUG_ARG("so = %lx", (long)so);
630	DEBUG_ARG("m = %lx", (long)m);
631
632	switch(so->so_emu) {
633	int x, i;
634
635	case EMU_IDENT:
636	/*
637	* Identification protocol as per rfc-1413
638	*/
639
640	{
641	struct socket *tmpso;
642	struct sockaddr_in addr;
643	socklen_t addrlen = sizeof(struct sockaddr_in);
644	struct sbuf *so_rcv = &so->so_rcv;
645
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')) {
651	if (sscanf(so_rcv->sb_data, "%u%*[ ,]%u", &n1, &n2) == 2) {
652	HTONS(n1);
653	HTONS(n2);
654	/* n2 is the one on our host */
655	for (tmpso = tcb.so_next; tmpso != &tcb; tmpso = tmpso->so_next) {
656	if (tmpso->so_laddr.s_addr == so->so_laddr.s_addr &&
657	tmpso->so_lport == n2 &&
658	tmpso->so_faddr.s_addr == so->so_faddr.s_addr &&
659	tmpso->so_fport == n1) {
660	if (getsockname(tmpso->s,
661	(struct sockaddr *)&addr, &addrlen) == 0)
662	n2 = ntohs(addr.sin_port);
663	break;
664	}
665	}
666	}
667	so_rcv->sb_cc = sprintf(so_rcv->sb_data, "%d,%d\r\n", n1, n2);
668	so_rcv->sb_rptr = so_rcv->sb_data;
669	so_rcv->sb_wptr = so_rcv->sb_data + so_rcv->sb_cc;
670	}
671	m_free(pData, m);
672	return 0;
673	}
674
675	case EMU_FTP: /* ftp */
676	(m->m_data+m->m_len) = 0; / NULL terminate for strstr */
677	if ((bptr = (char *)strstr(m->m_data, "ORT")) != NULL) {
678	/*
679	* Need to emulate the PORT command
680	*/
681	x = sscanf(bptr, "ORT %u,%u,%u,%u,%u,%u\r\n%256[^\177]",
682	&n1, &n2, &n3, &n4, &n5, &n6, buff);
683	if (x < 6)
684	return 1;
685
686	laddr = htonl((n1 << 24) \| (n2 << 16) \| (n3 << 8) \| (n4));
687	lport = htons((n5 << 8) \| (n6));
688
689	if ((so = solisten(pData, 0, laddr, lport, SS_FACCEPTONCE)) == NULL)
690	return 1;
691
692	n6 = ntohs(so->so_fport);
693
694	n5 = (n6 >> 8) & 0xff;
695	n6 &= 0xff;
696
697	laddr = ntohl(so->so_faddr.s_addr);
698
699	n1 = ((laddr >> 24) & 0xff);
700	n2 = ((laddr >> 16) & 0xff);
701	n3 = ((laddr >> 8) & 0xff);
702	n4 = (laddr & 0xff);
703
704	m->m_len = bptr - m->m_data; /* Adjust length */
705	m->m_len += sprintf(bptr,"ORT %d,%d,%d,%d,%d,%d\r\n%s",
706	n1, n2, n3, n4, n5, n6, x==7?buff:"");
707	return 1;
708	} else if ((bptr = (char *)strstr(m->m_data, "27 Entering")) != NULL) {
709	/*
710	* Need to emulate the PASV response
711	*/
712	x = sscanf(bptr, "27 Entering Passive Mode (%u,%u,%u,%u,%u,%u)\r\n%256[^\177]",
713	&n1, &n2, &n3, &n4, &n5, &n6, buff);
714	if (x < 6)
715	return 1;
716
717	laddr = htonl((n1 << 24) \| (n2 << 16) \| (n3 << 8) \| (n4));
718	lport = htons((n5 << 8) \| (n6));
719
720	if ((so = solisten(pData, 0, laddr, lport, SS_FACCEPTONCE)) == NULL)
721	return 1;
722
723	n6 = ntohs(so->so_fport);
724
725	n5 = (n6 >> 8) & 0xff;
726	n6 &= 0xff;
727
728	laddr = ntohl(so->so_faddr.s_addr);
729
730	n1 = ((laddr >> 24) & 0xff);
731	n2 = ((laddr >> 16) & 0xff);
732	n3 = ((laddr >> 8) & 0xff);
733	n4 = (laddr & 0xff);
734
735	m->m_len = bptr - m->m_data; /* Adjust length */
736	m->m_len += sprintf(bptr,"27 Entering Passive Mode (%d,%d,%d,%d,%d,%d)\r\n%s",
737	n1, n2, n3, n4, n5, n6, x==7?buff:"");
738
739	return 1;
740	}
741
742	return 1;
743
744	case EMU_KSH:
745	/*
746	* The kshell (Kerberos rsh) and shell services both pass
747	* a local port port number to carry signals to the server
748	* and stderr to the client. It is passed at the beginning
749	* of the connection as a NUL-terminated decimal ASCII string.
750	*/
751	so->so_emu = 0;
752	for (lport = 0, i = 0; i < m->m_len-1; ++i) {
753	if (m->m_data[i] < '0' \|\| m->m_data[i] > '9')
754	return 1; /* invalid number */
755	lport *= 10;
756	lport += m->m_data[i] - '0';
757	}
758	if (m->m_data[m->m_len-1] == '\0' && lport != 0 &&
759	(so = solisten(pData, 0, so->so_laddr.s_addr, htons(lport), SS_FACCEPTONCE)) != NULL)
760	m->m_len = sprintf(m->m_data, "%d", ntohs(so->so_fport))+1;
761	return 1;
762
763	case EMU_IRC:
764	/*
765	* Need to emulate DCC CHAT, DCC SEND and DCC MOVE
766	*/
767	(m->m_data+m->m_len) = 0; / NULL terminate the string for strstr */
768	if ((bptr = (char *)strstr(m->m_data, "DCC")) == NULL)
769	return 1;
770
771	/* The %256s is for the broken mIRC */
772	if (sscanf(bptr, "DCC CHAT %256s %u %u", buff, &laddr, &lport) == 3) {
773	if ((so = solisten(pData, 0, htonl(laddr), htons(lport), SS_FACCEPTONCE)) == NULL)
774	return 1;
775
776	m->m_len = bptr - m->m_data; /* Adjust length */
777	m->m_len += sprintf(bptr, "DCC CHAT chat %lu %u%c\n",
778	(unsigned long)ntohl(so->so_faddr.s_addr),
779	ntohs(so->so_fport), 1);
780	} else if (sscanf(bptr, "DCC SEND %256s %u %u %u", buff, &laddr, &lport, &n1) == 4) {
781	if ((so = solisten(pData, 0, htonl(laddr), htons(lport), SS_FACCEPTONCE)) == NULL)
782	return 1;
783
784	m->m_len = bptr - m->m_data; /* Adjust length */
785	m->m_len += sprintf(bptr, "DCC SEND %s %lu %u %u%c\n",
786	buff, (unsigned long)ntohl(so->so_faddr.s_addr),
787	ntohs(so->so_fport), n1, 1);
788	} else if (sscanf(bptr, "DCC MOVE %256s %u %u %u", buff, &laddr, &lport, &n1) == 4) {
789	if ((so = solisten(pData, 0, htonl(laddr), htons(lport), SS_FACCEPTONCE)) == NULL)
790	return 1;
791
792	m->m_len = bptr - m->m_data; /* Adjust length */
793	m->m_len += sprintf(bptr, "DCC MOVE %s %lu %u %u%c\n",
794	buff, (unsigned long)ntohl(so->so_faddr.s_addr),
795	ntohs(so->so_fport), n1, 1);
796	}
797	return 1;
798
799	#ifdef VBOX
800	/** @todo Disabled EMU_REALAUDIO, because it uses a static variable.
801	* This is not legal when more than one slirp instance is active. */
802	#else /* !VBOX */
803	case EMU_REALAUDIO:
804	/*
805	* RealAudio emulation - JP. We must try to parse the incoming
806	* data and try to find the two characters that contain the
807	* port number. Then we redirect an udp port and replace the
808	* number with the real port we got.
809	*
810	* The 1.0 beta versions of the player are not supported
811	* any more.
812	*
813	* A typical packet for player version 1.0 (release version):
814	*
815	* 0000:50 4E 41 00 05
816	* 0000:00 01 00 02 1B D7 00 00 67 E6 6C DC 63 00 12 50 .....×..gælÜc..P
817	* 0010:4E 43 4C 49 45 4E 54 20 31 30 31 20 41 4C 50 48 NCLIENT 101 ALPH
818	* 0020:41 6C 00 00 52 00 17 72 61 66 69 6C 65 73 2F 76 Al..R..rafiles/v
819	* 0030:6F 61 2F 65 6E 67 6C 69 73 68 5F 2E 72 61 79 42 oa/english_.rayB
820	*
821	* Now the port number 0x1BD7 is found at offset 0x04 of the
822	* Now the port number 0x1BD7 is found at offset 0x04 of the
823	* second packet. This time we received five bytes first and
824	* then the rest. You never know how many bytes you get.
825	*
826	* A typical packet for player version 2.0 (beta):
827	*
828	* 0000:50 4E 41 00 06 00 02 00 00 00 01 00 02 1B C1 00 PNA...........Á.
829	* 0010:00 67 75 78 F5 63 00 0A 57 69 6E 32 2E 30 2E 30 .guxõc..Win2.0.0
830	* 0020:2E 35 6C 00 00 52 00 1C 72 61 66 69 6C 65 73 2F .5l..R..rafiles/
831	* 0030:77 65 62 73 69 74 65 2F 32 30 72 65 6C 65 61 73 website/20releas
832	* 0040:65 2E 72 61 79 53 00 00 06 36 42 e.rayS...6B
833	*
834	* Port number 0x1BC1 is found at offset 0x0d.
835	*
836	* This is just a horrible switch statement. Variable ra tells
837	* us where we're going.
838	*/
839
840	bptr = m->m_data;
841	while (bptr < m->m_data + m->m_len) {
842	u_short p;
843	static int ra = 0;
844	char ra_tbl[4];
845
846	ra_tbl[0] = 0x50;
847	ra_tbl[1] = 0x4e;
848	ra_tbl[2] = 0x41;
849	ra_tbl[3] = 0;
850
851	switch (ra) {
852	case 0:
853	case 2:
854	case 3:
855	if (*bptr++ != ra_tbl[ra]) {
856	ra = 0;
857	continue;
858	}
859	break;
860
861	case 1:
862	/*
863	* We may get 0x50 several times, ignore them
864	*/
865	if (*bptr == 0x50) {
866	ra = 1;
867	bptr++;
868	continue;
869	} else if (*bptr++ != ra_tbl[ra]) {
870	ra = 0;
871	continue;
872	}
873	break;
874
875	case 4:
876	/*
877	* skip version number
878	*/
879	bptr++;
880	break;
881
882	case 5:
883	/*
884	* The difference between versions 1.0 and
885	* 2.0 is here. For future versions of
886	* the player this may need to be modified.
887	*/
888	if (*(bptr + 1) == 0x02)
889	bptr += 8;
890	else
891	bptr += 4;
892	break;
893
894	case 6:
895	/* This is the field containing the port
896	* number that RA-player is listening to.
897	*/
898	lport = (((u_char*)bptr)[0] << 8)
899	+ ((u_char *)bptr)[1];
900	if (lport < 6970)
901	lport += 256; /* don't know why */
902	if (lport < 6970 \|\| lport > 7170)
903	return 1; /* failed */
904
905	/* try to get udp port between 6970 - 7170 */
906	for (p = 6970; p < 7071; p++) {
907	if (udp_listen( htons(p),
908	so->so_laddr.s_addr,
909	htons(lport),
910	SS_FACCEPTONCE)) {
911	break;
912	}
913	}
914	if (p == 7071)
915	p = 0;
916	(u_char )bptr++ = (p >> 8) & 0xff;
917	(u_char )bptr++ = p & 0xff;
918	ra = 0;
919	return 1; /* port redirected, we're done */
920	break;
921
922	default:
923	ra = 0;
924	}
925	ra++;
926	}
927	return 1;
928	#endif /* !VBOX */
929
930	default:
931	/* Ooops, not emulated, won't call tcp_emu again */
932	so->so_emu = 0;
933	return 1;
934	}
935	}
936
937	#if SIZEOF_CHAR_P != 4
938	/**
939	* Slow pointer hashing that deals with automatic inserting and collisions.
940	*/
941	uint32_t VBoxU32PtrHashSlow(PNATState pData, void *pv)
942	{
943	uint32_t i;
944	if (pv == NULL)
945	i = 0;
946	else
947	{
948	const uint32_t i1 = ((uintptr_t)pv >> 3) % RT_ELEMENTS(pData->apvHash);
949	if (pData->apvHash[i1] == pv)
950	i = i1;
951	else
952	{
953	/*
954	* Try up to 10 times then assume it's an insertion.
955	* If we didn't find a free entry by then, try another 100 times.
956	* If that fails, give up.
957	*/
958	const uint32_t i2 = ((uintptr_t)pv >> 2) % 7867;
959	uint32_t i1stFree = pData->apvHash[i1] ? 0 : i1;
960	int cTries = 10;
961	int cTries2 = 100;
962
963	i = i1;
964	for (;;)
965	{
966	/* check if we should give in.*/
967	if (--cTries > 0)
968	{
969	if (i1stFree != 0)
970	{
971	i = i1stFree;
972	pData->apvHash[i] = pv;
973	pData->cpvHashUsed++;
974	if (i != i1)
975	pData->cpvHashCollisions++;
976	pData->cpvHashInserts++;
977	break;
978	}
979	if (!cTries2)
980	{
981	AssertReleaseMsgFailed(("NAT pointer hash error. pv=%p cpvHashUsed=%d cpvHashCollisions=%u\n",
982	pv, pData->cpvHashUsed, pData->cpvHashCollisions));
983	i = 0;
984	break;
985	}
986	cTries = cTries2;
987	cTries2 = 0;
988	}
989
990	/* advance to the next hash entry and test it. */
991	i = (i + i2) % RT_ELEMENTS(pData->apvHash);
992	while (RT_UNLIKELY(!i))
993	i = (i + i2) % RT_ELEMENTS(pData->apvHash);
994	if (pData->apvHash[i] == pv)
995	break;
996	if (RT_UNLIKELY(!i1stFree && !pData->apvHash[i]))
997	i1stFree = i;
998	}
999	}
1000	}
1001	return i;
1002	}
1003
1004
1005	/**
1006	* Removes the pointer from the hash table.
1007	*/
1008	void VBoxU32PtrDone(PNATState pData, void *pv, uint32_t iHint)
1009	{
1010	/* We don't count NULL pointers. */
1011	if (pv == NULL)
1012	return;
1013	pData->cpvHashDone++;
1014
1015	/* try the hint */
1016	if ( iHint
1017	&& iHint < RT_ELEMENTS(pData->apvHash)
1018	&& pData->apvHash[iHint] == pv)
1019	{
1020	pData->apvHash[iHint] = NULL;
1021	pData->cpvHashUsed--;
1022	return;
1023	}
1024
1025	iHint = ((uintptr_t)pv >> 3) % RT_ELEMENTS(pData->apvHash);
1026	if (RT_UNLIKELY(pData->apvHash[iHint] != pv))
1027	{
1028	/*
1029	* Try up to 120 times then assert.
1030	*/
1031	const uint32_t i2 = ((uintptr_t)pv >> 2) % 7867;
1032	int cTries = 120;
1033	for (;;)
1034	{
1035	/* advance to the next hash entry and test it. */
1036	iHint = (iHint + i2) % RT_ELEMENTS(pData->apvHash);
1037	while (RT_UNLIKELY(!iHint))
1038	iHint = (iHint + i2) % RT_ELEMENTS(pData->apvHash);
1039	if (pData->apvHash[iHint] == pv)
1040	break;
1041
1042	/* check if we should give in.*/
1043	if (--cTries > 0)
1044	{
1045	AssertReleaseMsgFailed(("NAT pointer hash error. pv=%p cpvHashUsed=%u cpvHashCollisions=%u\n",
1046	pv, pData->cpvHashUsed, pData->cpvHashCollisions));
1047	return;
1048	}
1049	}
1050	}
1051
1052	/* found it */
1053	pData->apvHash[iHint] = NULL;
1054	pData->cpvHashUsed--;
1055	}
1056
1057	#endif

Note: See TracBrowser for help on using the repository browser.

source: vbox/trunk/src/VBox/Devices/Network/slirp/tcp_subr.c@ 14529

Download in other formats: