slirp.c@ 59420

Last change on this file since 59420 was 59219, checked in by vboxsync, 9 years ago
NAT: Improve dns mapping in host resolver - pattern match should ignore case; make literal match take priority over pattern match if both are applicable.
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 63.2 KB

Line
1	/* $Id: slirp.c 59219 2015-12-24 22:45:49Z vboxsync $ */
2	/** @file
3	* NAT - slirp glue.
4	*/
5
6	/*
7	* Copyright (C) 2006-2015 Oracle Corporation
8	*
9	* This file is part of VirtualBox Open Source Edition (OSE), as
10	* available from http://www.virtualbox.org. This file is free software;
11	* you can redistribute it and/or modify it under the terms of the GNU
12	* General Public License (GPL) as published by the Free Software
13	* Foundation, in version 2 as it comes in the "COPYING" file of the
14	* VirtualBox OSE distribution. VirtualBox OSE is distributed in the
15	* hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
16	*/
17
18	/*
19	* This code is based on:
20	*
21	* libslirp glue
22	*
23	* Copyright (c) 2004-2008 Fabrice Bellard
24	*
25	* Permission is hereby granted, free of charge, to any person obtaining a copy
26	* of this software and associated documentation files (the "Software"), to deal
27	* in the Software without restriction, including without limitation the rights
28	* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
29	* copies of the Software, and to permit persons to whom the Software is
30	* furnished to do so, subject to the following conditions:
31	*
32	* The above copyright notice and this permission notice shall be included in
33	* all copies or substantial portions of the Software.
34	*
35	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
36	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
37	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
38	* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
39	* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
40	* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
41	* THE SOFTWARE.
42	*/
43
44	#include "slirp.h"
45	#ifdef RT_OS_OS2
46	# include <paths.h>
47	#endif
48
49	#include <VBox/err.h>
50	#include <VBox/vmm/dbgf.h>
51	#include <VBox/vmm/pdmdrv.h>
52	#include <iprt/assert.h>
53	#include <iprt/file.h>
54	#ifndef RT_OS_WINDOWS
55	# include <sys/ioctl.h>
56	# include <poll.h>
57	# include <netinet/in.h>
58	#else
59	# include <Winnls.h>
60	# define _WINSOCK2API_
61	# include <IPHlpApi.h>
62	#endif
63	#include <alias.h>
64
65	#ifndef RT_OS_WINDOWS
66	/**
67	* XXX: It shouldn't be non-Windows specific.
68	* resolv_conf_parser.h client's structure isn't OS specific, it's just need to be generalized a
69	* a bit to replace slirp_state.h DNS server (domain) lists with rcp_state like structure.
70	*/
71	# include "resolv_conf_parser.h"
72	#endif
73
74	#ifndef RT_OS_WINDOWS
75	# define DO_ENGAGE_EVENT1(so, fdset, label) \
76	do { \
77	if ( so->so_poll_index != -1 \
78	&& so->s == polls[so->so_poll_index].fd) \
79	{ \
80	polls[so->so_poll_index].events \|= N_(fdset ## _poll); \
81	break; \
82	} \
83	AssertRelease(poll_index < (nfds)); \
84	AssertRelease(poll_index >= 0 && poll_index < (nfds)); \
85	polls[poll_index].fd = (so)->s; \
86	(so)->so_poll_index = poll_index; \
87	polls[poll_index].events = N_(fdset ## _poll); \
88	polls[poll_index].revents = 0; \
89	poll_index++; \
90	} while (0)
91
92	# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
93	do { \
94	if ( so->so_poll_index != -1 \
95	&& so->s == polls[so->so_poll_index].fd) \
96	{ \
97	polls[so->so_poll_index].events \|= \
98	N_(fdset1 ## _poll) \| N_(fdset2 ## _poll); \
99	break; \
100	} \
101	AssertRelease(poll_index < (nfds)); \
102	polls[poll_index].fd = (so)->s; \
103	(so)->so_poll_index = poll_index; \
104	polls[poll_index].events = \
105	N_(fdset1 ## _poll) \| N_(fdset2 ## _poll); \
106	poll_index++; \
107	} while (0)
108
109	# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
110
111	/*
112	* DO_CHECK_FD_SET is used in dumping events on socket, including POLLNVAL.
113	* gcc warns about attempts to log POLLNVAL so construction in a last to lines
114	* used to catch POLLNVAL while logging and return false in case of error while
115	* normal usage.
116	*/
117	# define DO_CHECK_FD_SET(so, events, fdset) \
118	( ((so)->so_poll_index != -1) \
119	&& ((so)->so_poll_index <= ndfs) \
120	&& ((so)->s == polls[so->so_poll_index].fd) \
121	&& (polls[(so)->so_poll_index].revents & N_(fdset ## _poll)) \
122	&& ( N_(fdset ## _poll) == POLLNVAL \
123	\|\| !(polls[(so)->so_poll_index].revents & POLLNVAL)))
124
125	/* specific for Windows Winsock API */
126	# define DO_WIN_CHECK_FD_SET(so, events, fdset) 0
127
128	# ifndef RT_OS_LINUX
129	# define readfds_poll (POLLRDNORM)
130	# define writefds_poll (POLLWRNORM)
131	# else
132	# define readfds_poll (POLLIN)
133	# define writefds_poll (POLLOUT)
134	# endif
135	# define xfds_poll (POLLPRI)
136	# define closefds_poll (POLLHUP)
137	# define rderr_poll (POLLERR)
138	# if 0 /* unused yet */
139	# define rdhup_poll (POLLHUP)
140	# define nval_poll (POLLNVAL)
141	# endif
142
143	# define ICMP_ENGAGE_EVENT(so, fdset) \
144	do { \
145	if (pData->icmp_socket.s != -1) \
146	DO_ENGAGE_EVENT1((so), fdset, ICMP); \
147	} while (0)
148
149	#else /* RT_OS_WINDOWS */
150
151	/*
152	* On Windows, we will be notified by IcmpSendEcho2() when the response arrives.
153	* So no call to WSAEventSelect necessary.
154	*/
155	# define ICMP_ENGAGE_EVENT(so, fdset) do {} while (0)
156
157	/*
158	* On Windows we use FD_ALL_EVENTS to ensure that we don't miss any event.
159	*/
160	# define DO_ENGAGE_EVENT1(so, fdset1, label) \
161	do { \
162	rc = WSAEventSelect((so)->s, VBOX_SOCKET_EVENT, FD_ALL_EVENTS); \
163	if (rc == SOCKET_ERROR) \
164	{ \
165	/* This should not happen */ \
166	error = WSAGetLastError(); \
167	LogRel(("WSAEventSelect (" #label ") error %d (so=%x, socket=%s, event=%x)\n", \
168	error, (so), (so)->s, VBOX_SOCKET_EVENT)); \
169	} \
170	} while (0); \
171	CONTINUE(label)
172
173	# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
174	DO_ENGAGE_EVENT1((so), (fdset1), label)
175
176	# define DO_POLL_EVENTS(rc, error, so, events, label) \
177	(rc) = WSAEnumNetworkEvents((so)->s, VBOX_SOCKET_EVENT, (events)); \
178	if ((rc) == SOCKET_ERROR) \
179	{ \
180	(error) = WSAGetLastError(); \
181	LogRel(("WSAEnumNetworkEvents %R[natsock] " #label " error %d\n", (so), (error))); \
182	LogFunc(("WSAEnumNetworkEvents %R[natsock] " #label " error %d\n", (so), (error))); \
183	CONTINUE(label); \
184	}
185
186	# define acceptds_win FD_ACCEPT
187	# define acceptds_win_bit FD_ACCEPT_BIT
188	# define readfds_win FD_READ
189	# define readfds_win_bit FD_READ_BIT
190	# define writefds_win FD_WRITE
191	# define writefds_win_bit FD_WRITE_BIT
192	# define xfds_win FD_OOB
193	# define xfds_win_bit FD_OOB_BIT
194	# define closefds_win FD_CLOSE
195	# define closefds_win_bit FD_CLOSE_BIT
196	# define connectfds_win FD_CONNECT
197	# define connectfds_win_bit FD_CONNECT_BIT
198
199	# define closefds_win FD_CLOSE
200	# define closefds_win_bit FD_CLOSE_BIT
201
202	# define DO_CHECK_FD_SET(so, events, fdset) \
203	((events).lNetworkEvents & fdset ## _win)
204
205	# define DO_WIN_CHECK_FD_SET(so, events, fdset) DO_CHECK_FD_SET((so), (events), fdset)
206	# define DO_UNIX_CHECK_FD_SET(so, events, fdset) 1 /specific for Unix API /
207
208	#endif /* RT_OS_WINDOWS */
209
210	#define TCP_ENGAGE_EVENT1(so, fdset) \
211	DO_ENGAGE_EVENT1((so), fdset, tcp)
212
213	#define TCP_ENGAGE_EVENT2(so, fdset1, fdset2) \
214	DO_ENGAGE_EVENT2((so), fdset1, fdset2, tcp)
215
216	#ifdef RT_OS_WINDOWS
217	# define WIN_TCP_ENGAGE_EVENT2(so, fdset, fdset2) TCP_ENGAGE_EVENT2(so, fdset1, fdset2)
218	#endif
219
220	#define UDP_ENGAGE_EVENT(so, fdset) \
221	DO_ENGAGE_EVENT1((so), fdset, udp)
222
223	#define POLL_TCP_EVENTS(rc, error, so, events) \
224	DO_POLL_EVENTS((rc), (error), (so), (events), tcp)
225
226	#define POLL_UDP_EVENTS(rc, error, so, events) \
227	DO_POLL_EVENTS((rc), (error), (so), (events), udp)
228
229	#define CHECK_FD_SET(so, events, set) \
230	(DO_CHECK_FD_SET((so), (events), set))
231
232	#define WIN_CHECK_FD_SET(so, events, set) \
233	(DO_WIN_CHECK_FD_SET((so), (events), set))
234
235	/*
236	* Loging macros
237	*/
238	#if VBOX_WITH_DEBUG_NAT_SOCKETS
239	# if defined(RT_OS_WINDOWS)
240	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
241	do { \
242	LogRel((" " #proto " %R[natsock] %R[natwinnetevents]\n", (so), (winevent))); \
243	} while (0)
244	# else /* !RT_OS_WINDOWS */
245	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
246	do { \
247	LogRel((" " #proto " %R[natsock] %s %s %s er: %s, %s, %s\n", (so), \
248	CHECK_FD_SET(so, ign ,r_fdset) ? "READ":"", \
249	CHECK_FD_SET(so, ign, w_fdset) ? "WRITE":"", \
250	CHECK_FD_SET(so, ign, x_fdset) ? "OOB":"", \
251	CHECK_FD_SET(so, ign, rderr) ? "RDERR":"", \
252	CHECK_FD_SET(so, ign, rdhup) ? "RDHUP":"", \
253	CHECK_FD_SET(so, ign, nval) ? "RDNVAL":"")); \
254	} while (0)
255	# endif /* !RT_OS_WINDOWS */
256	#else /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
257	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) do {} while (0)
258	#endif /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
259
260	#define LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
261	DO_LOG_NAT_SOCK((so), proto, (winevent), r_fdset, w_fdset, x_fdset)
262
263	static const uint8_t special_ethaddr[6] =
264	{
265	0x52, 0x54, 0x00, 0x12, 0x35, 0x00
266	};
267
268	static const uint8_t broadcast_ethaddr[6] =
269	{
270	0xff, 0xff, 0xff, 0xff, 0xff, 0xff
271	};
272
273	const uint8_t zerro_ethaddr[6] =
274	{
275	0x0, 0x0, 0x0, 0x0, 0x0, 0x0
276	};
277
278	/**
279	* This helper routine do the checks in descriptions to
280	* ''fUnderPolling'' and ''fShouldBeRemoved'' flags
281	* @returns 1 if socket removed and 0 if no changes was made.
282	*/
283	static int slirpVerifyAndFreeSocket(PNATState pData, struct socket *pSocket)
284	{
285	AssertPtrReturn(pData, 0);
286	AssertPtrReturn(pSocket, 0);
287	AssertReturn(pSocket->fUnderPolling, 0);
288	if (pSocket->fShouldBeRemoved)
289	{
290	pSocket->fUnderPolling = 0;
291	sofree(pData, pSocket);
292	/* pSocket is PHANTOM, now */
293	return 1;
294	}
295	return 0;
296	}
297
298	int slirp_init(PNATState *ppData, uint32_t u32NetAddr, uint32_t u32Netmask,
299	bool fPassDomain, bool fUseHostResolver, int i32AliasMode,
300	int iIcmpCacheLimit, void *pvUser)
301	{
302	int rc;
303	PNATState pData;
304	if (u32Netmask & 0x1f)
305	{
306	/* CTL is x.x.x.15, bootp passes up to 16 IPs (15..31) */
307	LogRel(("NAT: The last 5 bits of the netmask (%RTnaipv4) need to be unset\n", RT_BE2H_U32(u32Netmask)));
308	return VERR_INVALID_PARAMETER;
309	}
310	pData = RTMemAllocZ(RT_ALIGN_Z(sizeof(NATState), sizeof(uint64_t)));
311	*ppData = pData;
312	if (!pData)
313	return VERR_NO_MEMORY;
314	pData->fPassDomain = !fUseHostResolver ? fPassDomain : false;
315	pData->fUseHostResolver = fUseHostResolver;
316	pData->fUseHostResolverPermanent = fUseHostResolver;
317	pData->pvUser = pvUser;
318	pData->netmask = u32Netmask;
319
320	rc = RTCritSectRwInit(&pData->CsRwHandlerChain);
321	if (RT_FAILURE(rc))
322	return rc;
323
324	/* sockets & TCP defaults */
325	pData->socket_rcv = 64 * _1K;
326	pData->socket_snd = 64 * _1K;
327	tcp_sndspace = 64 * _1K;
328	tcp_rcvspace = 64 * _1K;
329
330	/*
331	* Use the same default here as in DevNAT.cpp (SoMaxConnection CFGM value)
332	* to avoid release log noise.
333	*/
334	pData->soMaxConn = 10;
335
336	#ifdef RT_OS_WINDOWS
337	{
338	WSADATA Data;
339	RTLDRMOD hLdrMod;
340
341	WSAStartup(MAKEWORD(2, 0), &Data);
342
343	rc = RTLdrLoadSystem("Iphlpapi.dll", /* :fNoUnload */ true, &hLdrMod);
344	if (RT_SUCCESS(rc))
345	{
346	rc = RTLdrGetSymbol(hLdrMod, "GetAdaptersAddresses", (void **)&pData->pfGetAdaptersAddresses);
347	if (RT_FAILURE(rc))
348	LogRel(("NAT: Can't find GetAdapterAddresses in Iphlpapi.dll\n"));
349
350	RTLdrClose(hLdrMod);
351	}
352	}
353	pData->phEvents[VBOX_SOCKET_EVENT_INDEX] = CreateEvent(NULL, FALSE, FALSE, NULL);
354	#endif
355
356	rc = bootp_dhcp_init(pData);
357	if (RT_FAILURE(rc))
358	{
359	Log(("NAT: DHCP server initialization failed\n"));
360	RTMemFree(pData);
361	*ppData = NULL;
362	return rc;
363	}
364	debug_init(pData);
365	if_init(pData);
366	ip_init(pData);
367	icmp_init(pData, iIcmpCacheLimit);
368
369	/* Initialise mbufs after setting the MTU */
370	mbuf_init(pData);
371
372	pData->special_addr.s_addr = u32NetAddr;
373	pData->slirp_ethaddr = &special_ethaddr[0];
374	alias_addr.s_addr = pData->special_addr.s_addr \| RT_H2N_U32_C(CTL_ALIAS);
375	/* @todo: add ability to configure this staff */
376
377	/* set default addresses */
378	inet_aton("127.0.0.1", &loopback_addr);
379
380	rc = slirpTftpInit(pData);
381	AssertRCReturn(rc, VINF_NAT_DNS);
382
383	if (i32AliasMode & ~(PKT_ALIAS_LOG\|PKT_ALIAS_SAME_PORTS\|PKT_ALIAS_PROXY_ONLY))
384	{
385	Log(("NAT: alias mode %x is ignored\n", i32AliasMode));
386	i32AliasMode = 0;
387	}
388	pData->i32AliasMode = i32AliasMode;
389	getouraddr(pData);
390	{
391	int flags = 0;
392	struct in_addr proxy_addr;
393	pData->proxy_alias = LibAliasInit(pData, NULL);
394	if (pData->proxy_alias == NULL)
395	{
396	Log(("NAT: LibAlias default rule wasn't initialized\n"));
397	AssertMsgFailed(("NAT: LibAlias default rule wasn't initialized\n"));
398	}
399	flags = LibAliasSetMode(pData->proxy_alias, 0, 0);
400	#ifndef NO_FW_PUNCH
401	flags \|= PKT_ALIAS_PUNCH_FW;
402	#endif
403	flags \|= pData->i32AliasMode; /* do transparent proxying */
404	flags = LibAliasSetMode(pData->proxy_alias, flags, ~0);
405	proxy_addr.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
406	LibAliasSetAddress(pData->proxy_alias, proxy_addr);
407	ftp_alias_load(pData);
408	nbt_alias_load(pData);
409	}
410	#ifdef VBOX_WITH_NAT_SEND2HOME
411	/* @todo: we should know all interfaces available on host. */
412	pData->pInSockAddrHomeAddress = RTMemAllocZ(sizeof(struct sockaddr));
413	pData->cInHomeAddressSize = 1;
414	inet_aton("192.168.1.25", &pData->pInSockAddrHomeAddress[0].sin_addr);
415	pData->pInSockAddrHomeAddress[0].sin_family = AF_INET;
416	# ifdef RT_OS_DARWIN
417	pData->pInSockAddrHomeAddress[0].sin_len = sizeof(struct sockaddr_in);
418	# endif
419	#endif
420
421	#ifdef VBOX_WITH_DNSMAPPING_IN_HOSTRESOLVER
422	STAILQ_INIT(&pData->DNSMapNames);
423	STAILQ_INIT(&pData->DNSMapPatterns);
424	#endif
425
426	slirp_link_up(pData);
427	return VINF_SUCCESS;
428	}
429
430	/**
431	* Register statistics.
432	*/
433	void slirp_register_statistics(PNATState pData, PPDMDRVINS pDrvIns)
434	{
435	#ifdef VBOX_WITH_STATISTICS
436	# define PROFILE_COUNTER(name, dsc) REGISTER_COUNTER(name, pData, STAMTYPE_PROFILE, STAMUNIT_TICKS_PER_CALL, dsc)
437	# define COUNTING_COUNTER(name, dsc) REGISTER_COUNTER(name, pData, STAMTYPE_COUNTER, STAMUNIT_COUNT, dsc)
438	# include "counters.h"
439	# undef COUNTER
440	/** @todo register statistics for the variables dumped by:
441	* ipstats(pData); tcpstats(pData); udpstats(pData); icmpstats(pData);
442	* mbufstats(pData); sockstats(pData); */
443	#else /* VBOX_WITH_STATISTICS */
444	NOREF(pData);
445	NOREF(pDrvIns);
446	#endif /* !VBOX_WITH_STATISTICS */
447	}
448
449	/**
450	* Deregister statistics.
451	*/
452	void slirp_deregister_statistics(PNATState pData, PPDMDRVINS pDrvIns)
453	{
454	if (pData == NULL)
455	return;
456	#ifdef VBOX_WITH_STATISTICS
457	# define PROFILE_COUNTER(name, dsc) DEREGISTER_COUNTER(name, pData)
458	# define COUNTING_COUNTER(name, dsc) DEREGISTER_COUNTER(name, pData)
459	# include "counters.h"
460	#else /* VBOX_WITH_STATISTICS */
461	NOREF(pData);
462	NOREF(pDrvIns);
463	#endif /* !VBOX_WITH_STATISTICS */
464	}
465
466	/**
467	* Marks the link as up, making it possible to establish new connections.
468	*/
469	void slirp_link_up(PNATState pData)
470	{
471	struct arp_cache_entry *ac;
472
473	if (link_up == 1)
474	return;
475
476	link_up = 1;
477
478	if (!pData->fUseHostResolverPermanent)
479	slirpInitializeDnsSettings(pData);
480	}
481
482	/**
483	* Marks the link as down and cleans up the current connections.
484	*/
485	void slirp_link_down(PNATState pData)
486	{
487	struct port_forward_rule *rule;
488
489	if (link_up == 0)
490	return;
491
492	slirpReleaseDnsSettings(pData);
493
494	link_up = 0;
495	}
496
497	/**
498	* Terminates the slirp component.
499	*/
500	void slirp_term(PNATState pData)
501	{
502	struct socket *so;
503
504	if (pData == NULL)
505	return;
506
507	icmp_finit(pData);
508
509	while ((so = tcb.so_next) != &tcb)
510	{
511	/* Don't miss TCB releasing */
512	if ( !sototcpcb(so)
513	&& ( so->so_state & SS_NOFDREF
514	\|\| so->s == -1))
515	sofree(pData, so);
516	else
517	tcp_close(pData, sototcpcb(so));
518	}
519
520	while ((so = udb.so_next) != &udb)
521	udp_detach(pData, so);
522
523	slirp_link_down(pData);
524	ftp_alias_unload(pData);
525	nbt_alias_unload(pData);
526
527	#ifdef VBOX_WITH_DNSMAPPING_IN_HOSTRESOLVER
528	{
529	DNSMAPPINGHEAD *heads[2];
530	int i;
531
532	heads[0] = &pData->DNSMapNames;
533	heads[1] = &pData->DNSMapPatterns;
534	for (i = 0; i < RT_ELEMENTS(heads); ++i)
535	{
536	while (!STAILQ_EMPTY(heads[i]))
537	{
538	PDNSMAPPINGENTRY pDnsEntry = STAILQ_FIRST(heads[i]);
539	STAILQ_REMOVE_HEAD(heads[i], MapList);
540	RTStrFree(pDnsEntry->pszName);
541	RTMemFree(pDnsEntry);
542	}
543	}
544	}
545	#endif
546
547	while (!LIST_EMPTY(&instancehead))
548	{
549	struct libalias *la = LIST_FIRST(&instancehead);
550	/* libalias do all clean up */
551	LibAliasUninit(la);
552	}
553	while (!LIST_EMPTY(&pData->arp_cache))
554	{
555	struct arp_cache_entry *ac = LIST_FIRST(&pData->arp_cache);
556	LIST_REMOVE(ac, list);
557	RTMemFree(ac);
558	}
559	slirpTftpTerm(pData);
560	bootp_dhcp_fini(pData);
561	m_fini(pData);
562	#ifdef RT_OS_WINDOWS
563	WSACleanup();
564	#endif
565	#ifdef LOG_ENABLED
566	Log(("\n"
567	"NAT statistics\n"
568	"--------------\n"
569	"\n"));
570	ipstats(pData);
571	tcpstats(pData);
572	udpstats(pData);
573	icmpstats(pData);
574	mbufstats(pData);
575	sockstats(pData);
576	Log(("\n"
577	"\n"
578	"\n"));
579	#endif
580	RTCritSectRwDelete(&pData->CsRwHandlerChain);
581	RTMemFree(pData);
582	}
583
584
585	#define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE\|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
586	#define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE\|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
587
588	/*
589	* curtime kept to an accuracy of 1ms
590	*/
591	static void updtime(PNATState pData)
592	{
593	#ifdef RT_OS_WINDOWS
594	struct _timeb tb;
595
596	_ftime(&tb);
597	curtime = (u_int)tb.time * (u_int)1000;
598	curtime += (u_int)tb.millitm;
599	#else
600	gettimeofday(&tt, 0);
601
602	curtime = (u_int)tt.tv_sec * (u_int)1000;
603	curtime += (u_int)tt.tv_usec / (u_int)1000;
604
605	if ((tt.tv_usec % 1000) >= 500)
606	curtime++;
607	#endif
608	}
609
610	#ifdef RT_OS_WINDOWS
611	void slirp_select_fill(PNATState pData, int *pnfds)
612	#else /* RT_OS_WINDOWS */
613	void slirp_select_fill(PNATState pData, int pnfds, struct pollfd polls)
614	#endif /* !RT_OS_WINDOWS */
615	{
616	struct socket so, so_next;
617	int nfds;
618	#if defined(RT_OS_WINDOWS)
619	int rc;
620	int error;
621	#else
622	int poll_index = 0;
623	#endif
624	int i;
625
626	STAM_PROFILE_START(&pData->StatFill, a);
627
628	nfds = *pnfds;
629
630	/*
631	* First, TCP sockets
632	*/
633	do_slowtimo = 0;
634	if (!link_up)
635	goto done;
636
637	/*
638	* *_slowtimo needs calling if there are IP fragments
639	* in the fragment queue, or there are TCP connections active
640	*/
641	/* XXX:
642	* triggering of fragment expiration should be the same but use new macroses
643	*/
644	do_slowtimo = (tcb.so_next != &tcb);
645	if (!do_slowtimo)
646	{
647	for (i = 0; i < IPREASS_NHASH; i++)
648	{
649	if (!TAILQ_EMPTY(&ipq[i]))
650	{
651	do_slowtimo = 1;
652	break;
653	}
654	}
655	}
656	/* always add the ICMP socket */
657	#ifndef RT_OS_WINDOWS
658	pData->icmp_socket.so_poll_index = -1;
659	#endif
660	ICMP_ENGAGE_EVENT(&pData->icmp_socket, readfds);
661
662	STAM_COUNTER_RESET(&pData->StatTCP);
663	STAM_COUNTER_RESET(&pData->StatTCPHot);
664
665	QSOCKET_FOREACH(so, so_next, tcp)
666	/* { */
667	Assert(so->so_type == IPPROTO_TCP);
668	#if !defined(RT_OS_WINDOWS)
669	so->so_poll_index = -1;
670	#endif
671	STAM_COUNTER_INC(&pData->StatTCP);
672	#ifdef VBOX_WITH_NAT_UDP_SOCKET_CLONE
673	/* TCP socket can't be cloned */
674	Assert((!so->so_cloneOf));
675	#endif
676	/*
677	* See if we need a tcp_fasttimo
678	*/
679	if ( time_fasttimo == 0
680	&& so->so_tcpcb != NULL
681	&& so->so_tcpcb->t_flags & TF_DELACK)
682	{
683	time_fasttimo = curtime; /* Flag when we want a fasttimo */
684	}
685
686	/*
687	* NOFDREF can include still connecting to local-host,
688	* newly socreated() sockets etc. Don't want to select these.
689	*/
690	if (so->so_state & SS_NOFDREF \|\| so->s == -1)
691	CONTINUE(tcp);
692
693	/*
694	* Set for reading sockets which are accepting
695	*/
696	if (so->so_state & SS_FACCEPTCONN)
697	{
698	STAM_COUNTER_INC(&pData->StatTCPHot);
699	TCP_ENGAGE_EVENT1(so, readfds);
700	CONTINUE(tcp);
701	}
702
703	/*
704	* Set for writing sockets which are connecting
705	*/
706	if (so->so_state & SS_ISFCONNECTING)
707	{
708	Log2(("connecting %R[natsock] engaged\n",so));
709	STAM_COUNTER_INC(&pData->StatTCPHot);
710	#ifdef RT_OS_WINDOWS
711	WIN_TCP_ENGAGE_EVENT2(so, writefds, connectfds);
712	#else
713	TCP_ENGAGE_EVENT1(so, writefds);
714	#endif
715	}
716
717	/*
718	* Set for writing if we are connected, can send more, and
719	* we have something to send
720	*/
721	if (CONN_CANFSEND(so) && SBUF_LEN(&so->so_rcv))
722	{
723	STAM_COUNTER_INC(&pData->StatTCPHot);
724	TCP_ENGAGE_EVENT1(so, writefds);
725	}
726
727	/*
728	* Set for reading (and urgent data) if we are connected, can
729	* receive more, and we have room for it XXX /2 ?
730	*/
731	/* @todo: vvl - check which predicat here will be more useful here in rerm of new sbufs. */
732	if ( CONN_CANFRCV(so)
733	&& (SBUF_LEN(&so->so_snd) < (SBUF_SIZE(&so->so_snd)/2))
734	#ifdef RT_OS_WINDOWS
735	&& !(so->so_state & SS_ISFCONNECTING)
736	#endif
737	)
738	{
739	STAM_COUNTER_INC(&pData->StatTCPHot);
740	TCP_ENGAGE_EVENT2(so, readfds, xfds);
741	}
742	LOOP_LABEL(tcp, so, so_next);
743	}
744
745	/*
746	* UDP sockets
747	*/
748	STAM_COUNTER_RESET(&pData->StatUDP);
749	STAM_COUNTER_RESET(&pData->StatUDPHot);
750
751	QSOCKET_FOREACH(so, so_next, udp)
752	/* { */
753
754	Assert(so->so_type == IPPROTO_UDP);
755	STAM_COUNTER_INC(&pData->StatUDP);
756	#if !defined(RT_OS_WINDOWS)
757	so->so_poll_index = -1;
758	#endif
759
760	/*
761	* See if it's timed out
762	*/
763	if (so->so_expire)
764	{
765	if (so->so_expire <= curtime)
766	{
767	Log2(("NAT: %R[natsock] expired\n", so));
768	if (so->so_timeout != NULL)
769	{
770	/* so_timeout - might change the so_expire value or
771	* drop so_timeout* from so.
772	*/
773	so->so_timeout(pData, so, so->so_timeout_arg);
774	/* on 4.2 so->
775	*/
776	if ( so_next->so_prev != so /* so_timeout freed the socket */
777	\|\| so->so_timeout) /* so_timeout just freed so_timeout */
778	CONTINUE_NO_UNLOCK(udp);
779	}
780	UDP_DETACH(pData, so, so_next);
781	CONTINUE_NO_UNLOCK(udp);
782	}
783	}
784	#ifdef VBOX_WITH_NAT_UDP_SOCKET_CLONE
785	if (so->so_cloneOf)
786	CONTINUE_NO_UNLOCK(udp);
787	#endif
788
789	/*
790	* When UDP packets are received from over the link, they're
791	* sendto()'d straight away, so no need for setting for writing
792	* Limit the number of packets queued by this session to 4.
793	* Note that even though we try and limit this to 4 packets,
794	* the session could have more queued if the packets needed
795	* to be fragmented.
796	*
797	* (XXX <= 4 ?)
798	*/
799	if ((so->so_state & SS_ISFCONNECTED) && so->so_queued <= 4)
800	{
801	STAM_COUNTER_INC(&pData->StatUDPHot);
802	UDP_ENGAGE_EVENT(so, readfds);
803	}
804	LOOP_LABEL(udp, so, so_next);
805	}
806	done:
807
808	#if defined(RT_OS_WINDOWS)
809	*pnfds = VBOX_EVENT_COUNT;
810	#else /* RT_OS_WINDOWS */
811	AssertRelease(poll_index <= *pnfds);
812	*pnfds = poll_index;
813	#endif /* !RT_OS_WINDOWS */
814
815	STAM_PROFILE_STOP(&pData->StatFill, a);
816	}
817
818
819	/**
820	* This function do Connection or sending tcp sequence to.
821	* @returns if true operation completed
822	* @note: functions call tcp_input that potentially could lead to tcp_drop
823	*/
824	static bool slirpConnectOrWrite(PNATState pData, struct socket *so, bool fConnectOnly)
825	{
826	int ret;
827	LogFlowFunc(("ENTER: so:%R[natsock], fConnectOnly:%RTbool\n", so, fConnectOnly));
828	/*
829	* Check for non-blocking, still-connecting sockets
830	*/
831	if (so->so_state & SS_ISFCONNECTING)
832	{
833	Log2(("connecting %R[natsock] catched\n", so));
834	/* Connected */
835	so->so_state &= ~SS_ISFCONNECTING;
836
837	/*
838	* This should be probably guarded by PROBE_CONN too. Anyway,
839	* we disable it on OS/2 because the below send call returns
840	* EFAULT which causes the opened TCP socket to close right
841	* after it has been opened and connected.
842	*/
843	#ifndef RT_OS_OS2
844	ret = send(so->s, (const char *)&ret, 0, 0);
845	if (ret < 0)
846	{
847	/* XXXXX Must fix, zero bytes is a NOP */
848	if ( soIgnorableErrorCode(errno)
849	\|\| errno == ENOTCONN)
850	{
851	LogFlowFunc(("LEAVE: false\n"));
852	return false;
853	}
854
855	/* else failed */
856	so->so_state = SS_NOFDREF;
857	}
858	/* else so->so_state &= ~SS_ISFCONNECTING; */
859	#endif
860
861	/*
862	* Continue tcp_input
863	*/
864	TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
865	/* continue; */
866	}
867	else if (!fConnectOnly)
868	{
869	SOWRITE(ret, pData, so);
870	if (RT_LIKELY(ret > 0))
871	{
872	/*
873	* Make sure we will send window update to peer. This is
874	* a moral equivalent of calling tcp_output() for PRU_RCVD
875	* in tcp_usrreq() of the real stack.
876	*/
877	struct tcpcb *tp = sototcpcb(so);
878	if (RT_LIKELY(tp != NULL))
879	tp->t_flags \|= TF_DELACK;
880	}
881	}
882
883	LogFlowFunc(("LEAVE: true\n"));
884	return true;
885	}
886
887	#if defined(RT_OS_WINDOWS)
888	void slirp_select_poll(PNATState pData, int fTimeout)
889	#else /* RT_OS_WINDOWS */
890	void slirp_select_poll(PNATState pData, struct pollfd *polls, int ndfs)
891	#endif /* !RT_OS_WINDOWS */
892	{
893	struct socket so, so_next;
894	int ret;
895	#if defined(RT_OS_WINDOWS)
896	WSANETWORKEVENTS NetworkEvents;
897	int rc;
898	int error;
899	#endif
900
901	STAM_PROFILE_START(&pData->StatPoll, a);
902
903	/* Update time */
904	updtime(pData);
905
906	/*
907	* See if anything has timed out
908	*/
909	if (link_up)
910	{
911	if (time_fasttimo && ((curtime - time_fasttimo) >= 2))
912	{
913	STAM_PROFILE_START(&pData->StatFastTimer, b);
914	tcp_fasttimo(pData);
915	time_fasttimo = 0;
916	STAM_PROFILE_STOP(&pData->StatFastTimer, b);
917	}
918	if (do_slowtimo && ((curtime - last_slowtimo) >= 499))
919	{
920	STAM_PROFILE_START(&pData->StatSlowTimer, c);
921	ip_slowtimo(pData);
922	tcp_slowtimo(pData);
923	last_slowtimo = curtime;
924	STAM_PROFILE_STOP(&pData->StatSlowTimer, c);
925	}
926	}
927	#if defined(RT_OS_WINDOWS)
928	if (fTimeout)
929	return; /* only timer update */
930	#endif
931
932	/*
933	* Check sockets
934	*/
935	if (!link_up)
936	goto done;
937	#if defined(RT_OS_WINDOWS)
938	icmpwin_process(pData);
939	#else
940	if ( (pData->icmp_socket.s != -1)
941	&& CHECK_FD_SET(&pData->icmp_socket, ignored, readfds))
942	sorecvfrom(pData, &pData->icmp_socket);
943	#endif
944	/*
945	* Check TCP sockets
946	*/
947	QSOCKET_FOREACH(so, so_next, tcp)
948	/* { */
949	/* TCP socket can't be cloned */
950	#ifdef VBOX_WITH_NAT_UDP_SOCKET_CLONE
951	Assert((!so->so_cloneOf));
952	#endif
953	Assert(!so->fUnderPolling);
954	so->fUnderPolling = 1;
955	if (slirpVerifyAndFreeSocket(pData, so))
956	CONTINUE(tcp);
957	/*
958	* FD_ISSET is meaningless on these sockets
959	* (and they can crash the program)
960	*/
961	if (so->so_state & SS_NOFDREF \|\| so->s == -1)
962	{
963	so->fUnderPolling = 0;
964	CONTINUE(tcp);
965	}
966
967	POLL_TCP_EVENTS(rc, error, so, &NetworkEvents);
968
969	LOG_NAT_SOCK(so, TCP, &NetworkEvents, readfds, writefds, xfds);
970
971	if (so->so_state & SS_ISFCONNECTING)
972	{
973	int sockerr = 0;
974	#if !defined(RT_OS_WINDOWS)
975	{
976	int revents = 0;
977
978	/*
979	* Failed connect(2) is reported by poll(2) on
980	* different OSes with different combinations of
981	* POLLERR, POLLHUP, and POLLOUT.
982	*/
983	if ( CHECK_FD_SET(so, NetworkEvents, closefds) /* POLLHUP */
984	\|\| CHECK_FD_SET(so, NetworkEvents, rderr)) /* POLLERR */
985	{
986	revents = POLLHUP; /* squash to single "failed" flag */
987	}
988	#if defined(RT_OS_SOLARIS) \|\| defined(RT_OS_NETBSD)
989	/* Solaris and NetBSD report plain POLLOUT even on error */
990	else if (CHECK_FD_SET(so, NetworkEvents, writefds)) /* POLLOUT */
991	{
992	revents = POLLOUT;
993	}
994	#endif
995
996	if (revents != 0)
997	{
998	socklen_t optlen = (socklen_t)sizeof(sockerr);
999	ret = getsockopt(so->s, SOL_SOCKET, SO_ERROR, &sockerr, &optlen);
1000
1001	if ( RT_UNLIKELY(ret < 0)
1002	\|\| ( (revents & POLLHUP)
1003	&& RT_UNLIKELY(sockerr == 0)))
1004	sockerr = ETIMEDOUT;
1005	}
1006	}
1007	#else /* RT_OS_WINDOWS */
1008	{
1009	if (NetworkEvents.lNetworkEvents & FD_CONNECT)
1010	sockerr = NetworkEvents.iErrorCode[FD_CONNECT_BIT];
1011	}
1012	#endif
1013	if (sockerr != 0)
1014	{
1015	tcp_fconnect_failed(pData, so, sockerr);
1016	ret = slirpVerifyAndFreeSocket(pData, so);
1017	Assert(ret == 1); /* freed */
1018	CONTINUE(tcp);
1019	}
1020
1021	/*
1022	* XXX: For now just fall through to the old code to
1023	* handle successful connect(2).
1024	*/
1025	}
1026
1027	/*
1028	* Check for URG data
1029	* This will soread as well, so no need to
1030	* test for readfds below if this succeeds
1031	*/
1032
1033	/* out-of-band data */
1034	if ( CHECK_FD_SET(so, NetworkEvents, xfds)
1035	#ifdef RT_OS_DARWIN
1036	/* Darwin and probably BSD hosts generates POLLPRI\|POLLHUP event on receiving TCP.flags.{ACK\|URG\|FIN} this
1037	* combination on other Unixs hosts doesn't enter to this branch
1038	*/
1039	&& !CHECK_FD_SET(so, NetworkEvents, closefds)
1040	#endif
1041	#ifdef RT_OS_WINDOWS
1042	/**
1043	* In some cases FD_CLOSE comes with FD_OOB, that confuse tcp processing.
1044	*/
1045	&& !WIN_CHECK_FD_SET(so, NetworkEvents, closefds)
1046	#endif
1047	)
1048	{
1049	sorecvoob(pData, so);
1050	if (slirpVerifyAndFreeSocket(pData, so))
1051	CONTINUE(tcp);
1052	}
1053
1054	/*
1055	* Check sockets for reading
1056	*/
1057	else if ( CHECK_FD_SET(so, NetworkEvents, readfds)
1058	\|\| WIN_CHECK_FD_SET(so, NetworkEvents, acceptds))
1059	{
1060
1061	#ifdef RT_OS_WINDOWS
1062	if (WIN_CHECK_FD_SET(so, NetworkEvents, connectfds))
1063	{
1064	/* Finish connection first */
1065	/* should we ignore return value? */
1066	bool fRet = slirpConnectOrWrite(pData, so, true);
1067	LogFunc(("fRet:%RTbool\n", fRet));
1068	if (slirpVerifyAndFreeSocket(pData, so))
1069	CONTINUE(tcp);
1070	}
1071	#endif
1072	/*
1073	* Check for incoming connections
1074	*/
1075	if (so->so_state & SS_FACCEPTCONN)
1076	{
1077	TCP_CONNECT(pData, so);
1078	if (slirpVerifyAndFreeSocket(pData, so))
1079	CONTINUE(tcp);
1080	if (!CHECK_FD_SET(so, NetworkEvents, closefds))
1081	{
1082	so->fUnderPolling = 0;
1083	CONTINUE(tcp);
1084	}
1085	}
1086
1087	ret = soread(pData, so);
1088	if (slirpVerifyAndFreeSocket(pData, so))
1089	CONTINUE(tcp);
1090	/* Output it if we read something */
1091	if (RT_LIKELY(ret > 0))
1092	TCP_OUTPUT(pData, sototcpcb(so));
1093
1094	if (slirpVerifyAndFreeSocket(pData, so))
1095	CONTINUE(tcp);
1096	}
1097
1098	/*
1099	* Check for FD_CLOSE events.
1100	* in some cases once FD_CLOSE engaged on socket it could be flashed latter (for some reasons)
1101	*/
1102	if ( CHECK_FD_SET(so, NetworkEvents, closefds)
1103	\|\| (so->so_close == 1))
1104	{
1105	/*
1106	* drain the socket
1107	*/
1108	for (; so_next->so_prev == so
1109	&& !slirpVerifyAndFreeSocket(pData, so);)
1110	{
1111	ret = soread(pData, so);
1112	if (slirpVerifyAndFreeSocket(pData, so))
1113	break;
1114
1115	if (ret > 0)
1116	TCP_OUTPUT(pData, sototcpcb(so));
1117	else if (so_next->so_prev == so)
1118	{
1119	Log2(("%R[natsock] errno %d (%s)\n", so, errno, strerror(errno)));
1120	break;
1121	}
1122	}
1123
1124	/* if socket freed ''so'' is PHANTOM and next socket isn't points on it */
1125	if (so_next->so_prev == so)
1126	{
1127	/* mark the socket for termination _after_ it was drained */
1128	so->so_close = 1;
1129	/* No idea about Windows but on Posix, POLLHUP means that we can't send more.
1130	* Actually in the specific error scenario, POLLERR is set as well. */
1131	#ifndef RT_OS_WINDOWS
1132	if (CHECK_FD_SET(so, NetworkEvents, rderr))
1133	sofcantsendmore(so);
1134	#endif
1135	}
1136	if (so_next->so_prev == so)
1137	so->fUnderPolling = 0;
1138	CONTINUE(tcp);
1139	}
1140
1141	/*
1142	* Check sockets for writing
1143	*/
1144	if ( CHECK_FD_SET(so, NetworkEvents, writefds)
1145	#ifdef RT_OS_WINDOWS
1146	\|\| WIN_CHECK_FD_SET(so, NetworkEvents, connectfds)
1147	#endif
1148	)
1149	{
1150	int fConnectOrWriteSuccess = slirpConnectOrWrite(pData, so, false);
1151	/* slirpConnectOrWrite could return true even if tcp_input called tcp_drop,
1152	* so we should be ready to such situations.
1153	*/
1154	if (slirpVerifyAndFreeSocket(pData, so))
1155	CONTINUE(tcp);
1156	else if (!fConnectOrWriteSuccess)
1157	{
1158	so->fUnderPolling = 0;
1159	CONTINUE(tcp);
1160	}
1161	/* slirpConnectionOrWrite succeeded and socket wasn't dropped */
1162	}
1163
1164	/*
1165	* Probe a still-connecting, non-blocking socket
1166	* to check if it's still alive
1167	*/
1168	#ifdef PROBE_CONN
1169	if (so->so_state & SS_ISFCONNECTING)
1170	{
1171	ret = recv(so->s, (char *)&ret, 0, 0);
1172
1173	if (ret < 0)
1174	{
1175	/* XXX */
1176	if ( soIgnorableErrorCode(errno)
1177	\|\| errno == ENOTCONN)
1178	{
1179	CONTINUE(tcp); /* Still connecting, continue */
1180	}
1181
1182	/* else failed */
1183	so->so_state = SS_NOFDREF;
1184
1185	/* tcp_input will take care of it */
1186	}
1187	else
1188	{
1189	ret = send(so->s, &ret, 0, 0);
1190	if (ret < 0)
1191	{
1192	/* XXX */
1193	if ( soIgnorableErrorCode(errno)
1194	\|\| errno == ENOTCONN)
1195	{
1196	CONTINUE(tcp);
1197	}
1198	/* else failed */
1199	so->so_state = SS_NOFDREF;
1200	}
1201	else
1202	so->so_state &= ~SS_ISFCONNECTING;
1203
1204	}
1205	TCP_INPUT((struct mbuf *)NULL, sizeof(struct ip),so);
1206	} /* SS_ISFCONNECTING */
1207	#endif
1208	if (!slirpVerifyAndFreeSocket(pData, so))
1209	so->fUnderPolling = 0;
1210	LOOP_LABEL(tcp, so, so_next);
1211	}
1212
1213	/*
1214	* Now UDP sockets.
1215	* Incoming packets are sent straight away, they're not buffered.
1216	* Incoming UDP data isn't buffered either.
1217	*/
1218	QSOCKET_FOREACH(so, so_next, udp)
1219	/* { */
1220	#ifdef VBOX_WITH_NAT_UDP_SOCKET_CLONE
1221	if (so->so_cloneOf)
1222	CONTINUE_NO_UNLOCK(udp);
1223	#endif
1224	#if 0
1225	so->fUnderPolling = 1;
1226	if(slirpVerifyAndFreeSocket(pData, so));
1227	CONTINUE(udp);
1228	so->fUnderPolling = 0;
1229	#endif
1230
1231	POLL_UDP_EVENTS(rc, error, so, &NetworkEvents);
1232
1233	LOG_NAT_SOCK(so, UDP, &NetworkEvents, readfds, writefds, xfds);
1234
1235	if (so->s != -1 && CHECK_FD_SET(so, NetworkEvents, readfds))
1236	{
1237	SORECVFROM(pData, so);
1238	}
1239	LOOP_LABEL(udp, so, so_next);
1240	}
1241
1242	done:
1243
1244	STAM_PROFILE_STOP(&pData->StatPoll, a);
1245	}
1246
1247
1248	struct arphdr
1249	{
1250	unsigned short ar_hrd; /* format of hardware address */
1251	unsigned short ar_pro; /* format of protocol address */
1252	unsigned char ar_hln; /* length of hardware address */
1253	unsigned char ar_pln; /* length of protocol address */
1254	unsigned short ar_op; /* ARP opcode (command) */
1255
1256	/*
1257	* Ethernet looks like this : This bit is variable sized however...
1258	*/
1259	unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
1260	unsigned char ar_sip[4]; /* sender IP address */
1261	unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
1262	unsigned char ar_tip[4]; /* target IP address */
1263	};
1264	AssertCompileSize(struct arphdr, 28);
1265
1266	static void arp_output(PNATState pData, const uint8_t pcu8EtherSource, const struct arphdr pcARPHeaderSource, uint32_t ip4TargetAddress)
1267	{
1268	struct ethhdr *pEtherHeaderResponse;
1269	struct arphdr *pARPHeaderResponse;
1270	uint32_t ip4TargetAddressInHostFormat;
1271	struct mbuf *pMbufResponse;
1272
1273	Assert((pcu8EtherSource));
1274	if (!pcu8EtherSource)
1275	return;
1276	ip4TargetAddressInHostFormat = RT_N2H_U32(ip4TargetAddress);
1277
1278	pMbufResponse = m_getcl(pData, M_NOWAIT, MT_HEADER, M_PKTHDR);
1279	if (!pMbufResponse)
1280	return;
1281	pEtherHeaderResponse = mtod(pMbufResponse, struct ethhdr *);
1282	/* @note: if_encap will swap src and dst*/
1283	memcpy(pEtherHeaderResponse->h_source, pcu8EtherSource, ETH_ALEN);
1284	pMbufResponse->m_data += ETH_HLEN;
1285	pARPHeaderResponse = mtod(pMbufResponse, struct arphdr *);
1286	pMbufResponse->m_len = sizeof(struct arphdr);
1287
1288	pARPHeaderResponse->ar_hrd = RT_H2N_U16_C(1);
1289	pARPHeaderResponse->ar_pro = RT_H2N_U16_C(ETH_P_IP);
1290	pARPHeaderResponse->ar_hln = ETH_ALEN;
1291	pARPHeaderResponse->ar_pln = 4;
1292	pARPHeaderResponse->ar_op = RT_H2N_U16_C(ARPOP_REPLY);
1293	memcpy(pARPHeaderResponse->ar_sha, special_ethaddr, ETH_ALEN);
1294
1295	if (!slirpMbufTagService(pData, pMbufResponse, (uint8_t)(ip4TargetAddressInHostFormat & ~pData->netmask)))
1296	{
1297	static bool fTagErrorReported;
1298	if (!fTagErrorReported)
1299	{
1300	LogRel(("NAT: Couldn't add the tag(PACKET_SERVICE:%d)\n",
1301	(uint8_t)(ip4TargetAddressInHostFormat & ~pData->netmask)));
1302	fTagErrorReported = true;
1303	}
1304	}
1305	pARPHeaderResponse->ar_sha[5] = (uint8_t)(ip4TargetAddressInHostFormat & ~pData->netmask);
1306
1307	memcpy(pARPHeaderResponse->ar_sip, pcARPHeaderSource->ar_tip, 4);
1308	memcpy(pARPHeaderResponse->ar_tha, pcARPHeaderSource->ar_sha, ETH_ALEN);
1309	memcpy(pARPHeaderResponse->ar_tip, pcARPHeaderSource->ar_sip, 4);
1310	if_encap(pData, ETH_P_ARP, pMbufResponse, ETH_ENCAP_URG);
1311	}
1312
1313	/**
1314	* @note This function will free m!
1315	*/
1316	static void arp_input(PNATState pData, struct mbuf *m)
1317	{
1318	struct ethhdr *pEtherHeader;
1319	struct arphdr *pARPHeader;
1320	uint32_t ip4TargetAddress;
1321
1322	int ar_op;
1323	pEtherHeader = mtod(m, struct ethhdr *);
1324	pARPHeader = (struct arphdr *)&pEtherHeader[1];
1325
1326	ar_op = RT_N2H_U16(pARPHeader->ar_op);
1327	ip4TargetAddress = (uint32_t)pARPHeader->ar_tip;
1328
1329	switch (ar_op)
1330	{
1331	case ARPOP_REQUEST:
1332	if ( CTL_CHECK(ip4TargetAddress, CTL_DNS)
1333	\|\| CTL_CHECK(ip4TargetAddress, CTL_ALIAS)
1334	\|\| CTL_CHECK(ip4TargetAddress, CTL_TFTP))
1335	{
1336	slirp_update_guest_addr_guess(pData, (uint32_t )pARPHeader->ar_sip, "arp request");
1337	arp_output(pData, pEtherHeader->h_source, pARPHeader, ip4TargetAddress);
1338	break;
1339	}
1340
1341	/* Gratuitous ARP */
1342	if ( (uint32_t )pARPHeader->ar_sip == (uint32_t )pARPHeader->ar_tip
1343	&& ( memcmp(pARPHeader->ar_tha, zerro_ethaddr, ETH_ALEN) == 0
1344	\|\| memcmp(pARPHeader->ar_tha, broadcast_ethaddr, ETH_ALEN) == 0)
1345	&& memcmp(pEtherHeader->h_dest, broadcast_ethaddr, ETH_ALEN) == 0)
1346	{
1347	LogRel2(("NAT: Gratuitous ARP from %RTnaipv4 at %RTmac\n",
1348	(uint32_t )pARPHeader->ar_sip, pARPHeader->ar_sha));
1349	slirp_update_guest_addr_guess(pData, (uint32_t )pARPHeader->ar_sip, "gratuitous arp");
1350	slirp_arp_cache_update_or_add(pData, (uint32_t )pARPHeader->ar_sip, &pARPHeader->ar_sha[0]);
1351	}
1352	break;
1353
1354	case ARPOP_REPLY:
1355	slirp_arp_cache_update_or_add(pData, (uint32_t )pARPHeader->ar_sip, &pARPHeader->ar_sha[0]);
1356	break;
1357
1358	default:
1359	break;
1360	}
1361
1362	m_freem(pData, m);
1363	}
1364
1365	/**
1366	* Feed a packet into the slirp engine.
1367	*
1368	* @param m Data buffer, m_len is not valid.
1369	* @param cbBuf The length of the data in m.
1370	*/
1371	void slirp_input(PNATState pData, struct mbuf *m, size_t cbBuf)
1372	{
1373	int proto;
1374	static bool fWarnedIpv6;
1375	struct ethhdr *eh;
1376	uint8_t au8Ether[ETH_ALEN];
1377
1378	m->m_len = cbBuf;
1379	if (cbBuf < ETH_HLEN)
1380	{
1381	Log(("NAT: packet having size %d has been ignored\n", m->m_len));
1382	m_freem(pData, m);
1383	return;
1384	}
1385	eh = mtod(m, struct ethhdr *);
1386	proto = RT_N2H_U16(eh->h_proto);
1387
1388	memcpy(au8Ether, eh->h_source, ETH_ALEN);
1389
1390	switch(proto)
1391	{
1392	case ETH_P_ARP:
1393	arp_input(pData, m);
1394	break;
1395
1396	case ETH_P_IP:
1397	/* Update time. Important if the network is very quiet, as otherwise
1398	* the first outgoing connection gets an incorrect timestamp. */
1399	updtime(pData);
1400	m_adj(m, ETH_HLEN);
1401	M_ASSERTPKTHDR(m);
1402	m->m_pkthdr.header = mtod(m, void *);
1403	ip_input(pData, m);
1404	break;
1405
1406	case ETH_P_IPV6:
1407	m_freem(pData, m);
1408	if (!fWarnedIpv6)
1409	{
1410	LogRel(("NAT: IPv6 not supported\n"));
1411	fWarnedIpv6 = true;
1412	}
1413	break;
1414
1415	default:
1416	Log(("NAT: Unsupported protocol %x\n", proto));
1417	m_freem(pData, m);
1418	break;
1419	}
1420	}
1421
1422	/**
1423	* Output the IP packet to the ethernet device.
1424	*
1425	* @note This function will free m!
1426	*/
1427	void if_encap(PNATState pData, uint16_t eth_proto, struct mbuf *m, int flags)
1428	{
1429	struct ethhdr *eh;
1430	uint8_t *mbuf = NULL;
1431	size_t mlen = 0;
1432	STAM_PROFILE_START(&pData->StatIF_encap, a);
1433	LogFlowFunc(("ENTER: pData:%p, eth_proto:%RX16, m:%p, flags:%d\n",
1434	pData, eth_proto, m, flags));
1435
1436	M_ASSERTPKTHDR(m);
1437
1438	Assert(M_LEADINGSPACE(m) >= ETH_HLEN);
1439	m->m_data -= ETH_HLEN;
1440	m->m_len += ETH_HLEN;
1441	eh = mtod(m, struct ethhdr *);
1442	mlen = m->m_len;
1443
1444	if (memcmp(eh->h_source, special_ethaddr, ETH_ALEN) != 0)
1445	{
1446	struct m_tag *t = m_tag_first(m);
1447	uint8_t u8ServiceId = CTL_ALIAS;
1448	memcpy(eh->h_dest, eh->h_source, ETH_ALEN);
1449	memcpy(eh->h_source, special_ethaddr, ETH_ALEN);
1450	Assert(memcmp(eh->h_dest, special_ethaddr, ETH_ALEN) != 0);
1451	if (memcmp(eh->h_dest, zerro_ethaddr, ETH_ALEN) == 0)
1452	{
1453	/* don't do anything */
1454	m_freem(pData, m);
1455	goto done;
1456	}
1457	if ( t
1458	&& (t = m_tag_find(m, PACKET_SERVICE, NULL)))
1459	{
1460	Assert(t);
1461	u8ServiceId = (uint8_t )&t[1];
1462	}
1463	eh->h_source[5] = u8ServiceId;
1464	}
1465	/*
1466	* we're processing the chain, that isn't not expected.
1467	*/
1468	Assert((!m->m_next));
1469	if (m->m_next)
1470	{
1471	Log(("NAT: if_encap's recived the chain, dropping...\n"));
1472	m_freem(pData, m);
1473	goto done;
1474	}
1475	mbuf = mtod(m, uint8_t *);
1476	eh->h_proto = RT_H2N_U16(eth_proto);
1477	LogFunc(("eh(dst:%RTmac, src:%RTmac)\n", eh->h_dest, eh->h_source));
1478	if (flags & ETH_ENCAP_URG)
1479	slirp_urg_output(pData->pvUser, m, mbuf, mlen);
1480	else
1481	slirp_output(pData->pvUser, m, mbuf, mlen);
1482	done:
1483	STAM_PROFILE_STOP(&pData->StatIF_encap, a);
1484	LogFlowFuncLeave();
1485	}
1486
1487
1488	void
1489	slirp_update_guest_addr_guess(PNATState pData, uint32_t guess, const char *msg)
1490	{
1491	Assert(msg != NULL);
1492
1493	if (pData->guest_addr_guess.s_addr == guess)
1494	{
1495	LogRel2(("NAT: Guest address guess %RTnaipv4 re-confirmed by %s\n",
1496	pData->guest_addr_guess.s_addr, msg));
1497	return;
1498	}
1499
1500	if (pData->guest_addr_guess.s_addr == INADDR_ANY)
1501	{
1502	pData->guest_addr_guess.s_addr = guess;
1503	LogRel(("NAT: Guest address guess set to %RTnaipv4 by %s\n",
1504	pData->guest_addr_guess.s_addr, msg));
1505	return;
1506	}
1507	else
1508	{
1509	LogRel(("NAT: Guest address guess changed from %RTnaipv4 to %RTnaipv4 by %s\n",
1510	pData->guest_addr_guess.s_addr, guess, msg));
1511	pData->guest_addr_guess.s_addr = guess;
1512	return;
1513	}
1514	}
1515
1516
1517	static struct port_forward_rule *
1518	slirp_find_redirect(PNATState pData,
1519	int is_udp,
1520	struct in_addr host_addr, int host_port,
1521	struct in_addr guest_addr, int guest_port)
1522	{
1523	struct port_forward_rule *rule;
1524	uint16_t proto = (is_udp ? IPPROTO_UDP : IPPROTO_TCP);
1525
1526	LIST_FOREACH(rule, &pData->port_forward_rule_head, list)
1527	{
1528	if ( rule->proto == proto
1529	&& rule->host_port == host_port
1530	&& rule->bind_ip.s_addr == host_addr.s_addr
1531	&& rule->guest_port == guest_port
1532	&& rule->guest_addr.s_addr == guest_addr.s_addr)
1533	{
1534	return rule;
1535	}
1536	}
1537
1538	return NULL;
1539	}
1540
1541
1542	int slirp_add_redirect(PNATState pData, int is_udp, struct in_addr host_addr, int host_port,
1543	struct in_addr guest_addr, int guest_port)
1544	{
1545	struct port_forward_rule *rule;
1546
1547	rule = slirp_find_redirect(pData, is_udp, host_addr, host_port, guest_addr, guest_port);
1548	if (rule != NULL) /* rule has been already registered */
1549	{
1550	/* XXX: this shouldn't happen */
1551	return 0;
1552	}
1553
1554	rule = RTMemAllocZ(sizeof(struct port_forward_rule));
1555	if (rule == NULL)
1556	return 1;
1557
1558	rule->proto = (is_udp ? IPPROTO_UDP : IPPROTO_TCP);
1559	rule->bind_ip.s_addr = host_addr.s_addr;
1560	rule->host_port = host_port;
1561	rule->guest_addr.s_addr = guest_addr.s_addr;
1562	rule->guest_port = guest_port;
1563
1564	if (rule->proto == IPPROTO_UDP)
1565	rule->so = udp_listen(pData, rule->bind_ip.s_addr, RT_H2N_U16(rule->host_port),
1566	rule->guest_addr.s_addr, RT_H2N_U16(rule->guest_port), 0);
1567	else
1568	rule->so = solisten(pData, rule->bind_ip.s_addr, RT_H2N_U16(rule->host_port),
1569	rule->guest_addr.s_addr, RT_H2N_U16(rule->guest_port), 0);
1570
1571	if (rule->so == NULL)
1572	{
1573	LogRel(("NAT: Failed to redirect %s %RTnaipv4:%d -> %RTnaipv4:%d\n",
1574	rule->proto == IPPROTO_UDP ? "UDP" : "TCP",
1575	rule->bind_ip.s_addr, rule->host_port,
1576	guest_addr, rule->guest_port));
1577	RTMemFree(rule);
1578	return 1;
1579	}
1580
1581	LogRel(("NAT: Set redirect %s %RTnaipv4:%d -> %RTnaipv4:%d\n",
1582	rule->proto == IPPROTO_UDP ? "UDP" : "TCP",
1583	rule->bind_ip.s_addr, rule->host_port,
1584	guest_addr, rule->guest_port));
1585
1586	LIST_INSERT_HEAD(&pData->port_forward_rule_head, rule, list);
1587	return 0;
1588	}
1589
1590
1591	int slirp_remove_redirect(PNATState pData, int is_udp, struct in_addr host_addr, int host_port,
1592	struct in_addr guest_addr, int guest_port)
1593	{
1594	struct port_forward_rule *rule;
1595
1596	rule = slirp_find_redirect(pData, is_udp, host_addr, host_port, guest_addr, guest_port);
1597	if (rule == NULL)
1598	{
1599	LogRel(("NAT: Unable to find redirect %s %RTnaipv4:%d -> %RTnaipv4:%d\n",
1600	rule->proto == IPPROTO_UDP ? "UDP" : "TCP",
1601	rule->bind_ip.s_addr, rule->host_port,
1602	guest_addr.s_addr, rule->guest_port));
1603	return 0;
1604	}
1605
1606	LogRel(("NAT: Remove redirect %s %RTnaipv4:%d -> %RTnaipv4:%d\n",
1607	rule->proto == IPPROTO_UDP ? "UDP" : "TCP",
1608	rule->bind_ip.s_addr, rule->host_port,
1609	guest_addr.s_addr, rule->guest_port));
1610
1611	if (rule->so != NULL)
1612	{
1613	if (is_udp)
1614	udp_detach(pData, rule->so);
1615	else
1616	tcp_close(pData, sototcpcb(rule->so));
1617	}
1618
1619	LIST_REMOVE(rule, list);
1620	RTMemFree(rule);
1621	return 0;
1622	}
1623
1624
1625	#if defined(RT_OS_WINDOWS)
1626	HANDLE *slirp_get_events(PNATState pData)
1627	{
1628	return pData->phEvents;
1629	}
1630	void slirp_register_external_event(PNATState pData, HANDLE hEvent, int index)
1631	{
1632	pData->phEvents[index] = hEvent;
1633	}
1634	#endif
1635
1636	unsigned int slirp_get_timeout_ms(PNATState pData)
1637	{
1638	if (link_up)
1639	{
1640	if (time_fasttimo)
1641	return 2;
1642	if (do_slowtimo)
1643	return 500; /* see PR_SLOWHZ */
1644	}
1645	return 36001000; / one hour */
1646	}
1647
1648	#ifndef RT_OS_WINDOWS
1649	int slirp_get_nsock(PNATState pData)
1650	{
1651	return pData->nsock;
1652	}
1653	#endif
1654
1655	/*
1656	* this function called from NAT thread
1657	*/
1658	void slirp_post_sent(PNATState pData, void *pvArg)
1659	{
1660	struct mbuf m = (struct mbuf )pvArg;
1661	m_freem(pData, m);
1662	}
1663
1664	void slirp_set_dhcp_TFTP_prefix(PNATState pData, const char *tftpPrefix)
1665	{
1666	Log2(("tftp_prefix: %s\n", tftpPrefix));
1667	tftp_prefix = tftpPrefix;
1668	}
1669
1670	void slirp_set_dhcp_TFTP_bootfile(PNATState pData, const char *bootFile)
1671	{
1672	Log2(("bootFile: %s\n", bootFile));
1673	bootp_filename = bootFile;
1674	}
1675
1676	void slirp_set_dhcp_next_server(PNATState pData, const char *next_server)
1677	{
1678	Log2(("next_server: %s\n", next_server));
1679	if (next_server == NULL)
1680	pData->tftp_server.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_TFTP);
1681	else
1682	inet_aton(next_server, &pData->tftp_server);
1683	}
1684
1685	int slirp_set_binding_address(PNATState pData, char *addr)
1686	{
1687	if (addr == NULL \|\| (inet_aton(addr, &pData->bindIP) == 0))
1688	{
1689	pData->bindIP.s_addr = INADDR_ANY;
1690	return 1;
1691	}
1692	return 0;
1693	}
1694
1695	void slirp_set_dhcp_dns_proxy(PNATState pData, bool fDNSProxy)
1696	{
1697	if (!pData->fUseHostResolver)
1698	{
1699	Log2(("NAT: DNS proxy switched %s\n", (fDNSProxy ? "on" : "off")));
1700	pData->fUseDnsProxy = fDNSProxy;
1701	}
1702	else if (fDNSProxy)
1703	LogRel(("NAT: Host Resolver conflicts with DNS proxy, the last one was forcely ignored\n"));
1704	}
1705
1706	#define CHECK_ARG(name, val, lim_min, lim_max) \
1707	do { \
1708	if ((val) < (lim_min) \|\| (val) > (lim_max)) \
1709	{ \
1710	LogRel(("NAT: (" #name ":%d) has been ignored, " \
1711	"because out of range (%d, %d)\n", (val), (lim_min), (lim_max))); \
1712	return; \
1713	} \
1714	else \
1715	LogRel(("NAT: (" #name ":%d)\n", (val))); \
1716	} while (0)
1717
1718	void slirp_set_somaxconn(PNATState pData, int iSoMaxConn)
1719	{
1720	LogFlowFunc(("iSoMaxConn:%d\n", iSoMaxConn));
1721	/* Conditions */
1722	if (iSoMaxConn > SOMAXCONN)
1723	{
1724	LogRel(("NAT: value of somaxconn(%d) bigger than SOMAXCONN(%d)\n", iSoMaxConn, SOMAXCONN));
1725	iSoMaxConn = SOMAXCONN;
1726	}
1727
1728	if (iSoMaxConn < 1)
1729	{
1730	LogRel(("NAT: proposed value(%d) of somaxconn is invalid, default value is used (%d)\n", iSoMaxConn, pData->soMaxConn));
1731	LogFlowFuncLeave();
1732	return;
1733	}
1734
1735	/* Asignment */
1736	if (pData->soMaxConn != iSoMaxConn)
1737	{
1738	LogRel(("NAT: value of somaxconn has been changed from %d to %d\n",
1739	pData->soMaxConn, iSoMaxConn));
1740	pData->soMaxConn = iSoMaxConn;
1741	}
1742	LogFlowFuncLeave();
1743	}
1744	/* don't allow user set less 8kB and more than 1M values */
1745	#define _8K_1M_CHECK_ARG(name, val) CHECK_ARG(name, (val), 8, 1024)
1746	void slirp_set_rcvbuf(PNATState pData, int kilobytes)
1747	{
1748	_8K_1M_CHECK_ARG("SOCKET_RCVBUF", kilobytes);
1749	pData->socket_rcv = kilobytes;
1750	}
1751	void slirp_set_sndbuf(PNATState pData, int kilobytes)
1752	{
1753	_8K_1M_CHECK_ARG("SOCKET_SNDBUF", kilobytes);
1754	pData->socket_snd = kilobytes * _1K;
1755	}
1756	void slirp_set_tcp_rcvspace(PNATState pData, int kilobytes)
1757	{
1758	_8K_1M_CHECK_ARG("TCP_RCVSPACE", kilobytes);
1759	tcp_rcvspace = kilobytes * _1K;
1760	}
1761	void slirp_set_tcp_sndspace(PNATState pData, int kilobytes)
1762	{
1763	_8K_1M_CHECK_ARG("TCP_SNDSPACE", kilobytes);
1764	tcp_sndspace = kilobytes * _1K;
1765	}
1766
1767	/*
1768	* Looking for Ether by ip in ARP-cache
1769	* Note: it´s responsible of caller to allocate buffer for result
1770	* @returns iprt status code
1771	*/
1772	int slirp_arp_lookup_ether_by_ip(PNATState pData, uint32_t ip, uint8_t *ether)
1773	{
1774	struct arp_cache_entry *ac;
1775
1776	if (ether == NULL)
1777	return VERR_INVALID_PARAMETER;
1778
1779	if (LIST_EMPTY(&pData->arp_cache))
1780	return VERR_NOT_FOUND;
1781
1782	LIST_FOREACH(ac, &pData->arp_cache, list)
1783	{
1784	if ( ac->ip == ip
1785	&& memcmp(ac->ether, broadcast_ethaddr, ETH_ALEN) != 0)
1786	{
1787	memcpy(ether, ac->ether, ETH_ALEN);
1788	return VINF_SUCCESS;
1789	}
1790	}
1791	return VERR_NOT_FOUND;
1792	}
1793
1794	/*
1795	* Looking for IP by Ether in ARP-cache
1796	* Note: it´s responsible of caller to allocate buffer for result
1797	* @returns 0 - if found, 1 - otherwise
1798	*/
1799	int slirp_arp_lookup_ip_by_ether(PNATState pData, const uint8_t ether, uint32_t ip)
1800	{
1801	struct arp_cache_entry *ac;
1802	*ip = INADDR_ANY;
1803
1804	if (LIST_EMPTY(&pData->arp_cache))
1805	return VERR_NOT_FOUND;
1806
1807	LIST_FOREACH(ac, &pData->arp_cache, list)
1808	{
1809	if (memcmp(ether, ac->ether, ETH_ALEN) == 0)
1810	{
1811	*ip = ac->ip;
1812	return VINF_SUCCESS;
1813	}
1814	}
1815	return VERR_NOT_FOUND;
1816	}
1817
1818	void slirp_arp_who_has(PNATState pData, uint32_t dst)
1819	{
1820	struct mbuf *m;
1821	struct ethhdr *ehdr;
1822	struct arphdr *ahdr;
1823	static bool fWarned = false;
1824	LogFlowFunc(("ENTER: %RTnaipv4\n", dst));
1825
1826	/* ARP request WHO HAS 0.0.0.0 is one of the signals
1827	* that something has been broken at Slirp. Investigating
1828	* pcap dumps it's easy to miss warning ARP requests being
1829	* focused on investigation of other protocols flow.
1830	*/
1831	#ifdef DEBUG_vvl
1832	Assert((dst != INADDR_ANY));
1833	NOREF(fWarned);
1834	#else
1835	if ( dst == INADDR_ANY
1836	&& !fWarned)
1837	{
1838	LogRel(("NAT: ARP: \"WHO HAS INADDR_ANY\" request has been detected\n"));
1839	fWarned = true;
1840	}
1841	#endif /* !DEBUG_vvl */
1842
1843	m = m_getcl(pData, M_NOWAIT, MT_HEADER, M_PKTHDR);
1844	if (m == NULL)
1845	{
1846	Log(("NAT: Can't alloc mbuf for ARP request\n"));
1847	LogFlowFuncLeave();
1848	return;
1849	}
1850	ehdr = mtod(m, struct ethhdr *);
1851	memset(ehdr->h_source, 0xff, ETH_ALEN);
1852	ahdr = (struct arphdr *)&ehdr[1];
1853	ahdr->ar_hrd = RT_H2N_U16_C(1);
1854	ahdr->ar_pro = RT_H2N_U16_C(ETH_P_IP);
1855	ahdr->ar_hln = ETH_ALEN;
1856	ahdr->ar_pln = 4;
1857	ahdr->ar_op = RT_H2N_U16_C(ARPOP_REQUEST);
1858	memcpy(ahdr->ar_sha, special_ethaddr, ETH_ALEN);
1859	/* we assume that this request come from gw, but not from DNS or TFTP */
1860	ahdr->ar_sha[5] = CTL_ALIAS;
1861	(uint32_t )ahdr->ar_sip = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
1862	memset(ahdr->ar_tha, 0xff, ETH_ALEN); /broadcast/
1863	(uint32_t )ahdr->ar_tip = dst;
1864	/* warn!!! should falls in mbuf minimal size */
1865	m->m_len = sizeof(struct arphdr) + ETH_HLEN;
1866	m->m_data += ETH_HLEN;
1867	m->m_len -= ETH_HLEN;
1868	if_encap(pData, ETH_P_ARP, m, ETH_ENCAP_URG);
1869	LogFlowFuncLeave();
1870	}
1871
1872
1873	/* updates the arp cache
1874	* @note: this is helper function, slirp_arp_cache_update_or_add should be used.
1875	* @returns 0 - if has found and updated
1876	* 1 - if hasn't found.
1877	*/
1878	static inline int slirp_arp_cache_update(PNATState pData, uint32_t dst, const uint8_t *mac)
1879	{
1880	struct arp_cache_entry *ac;
1881	Assert(( memcmp(mac, broadcast_ethaddr, ETH_ALEN)
1882	&& memcmp(mac, zerro_ethaddr, ETH_ALEN)));
1883	LIST_FOREACH(ac, &pData->arp_cache, list)
1884	{
1885	if (ac->ip == dst)
1886	{
1887	memcpy(ac->ether, mac, ETH_ALEN);
1888	return 0;
1889	}
1890	}
1891	return 1;
1892	}
1893
1894	/**
1895	* add entry to the arp cache
1896	* @note: this is helper function, slirp_arp_cache_update_or_add should be used.
1897	*/
1898	static inline void slirp_arp_cache_add(PNATState pData, uint32_t ip, const uint8_t *ether)
1899	{
1900	struct arp_cache_entry *ac = NULL;
1901	Assert(( memcmp(ether, broadcast_ethaddr, ETH_ALEN)
1902	&& memcmp(ether, zerro_ethaddr, ETH_ALEN)));
1903	ac = RTMemAllocZ(sizeof(struct arp_cache_entry));
1904	if (ac == NULL)
1905	{
1906	Log(("NAT: Can't allocate arp cache entry\n"));
1907	return;
1908	}
1909	ac->ip = ip;
1910	memcpy(ac->ether, ether, ETH_ALEN);
1911	LIST_INSERT_HEAD(&pData->arp_cache, ac, list);
1912	}
1913
1914	/* updates or adds entry to the arp cache
1915	* @returns 0 - if has found and updated
1916	* 1 - if hasn't found.
1917	*/
1918	int slirp_arp_cache_update_or_add(PNATState pData, uint32_t dst, const uint8_t *mac)
1919	{
1920	if ( !memcmp(mac, broadcast_ethaddr, ETH_ALEN)
1921	\|\| !memcmp(mac, zerro_ethaddr, ETH_ALEN))
1922	{
1923	static bool fBroadcastEtherAddReported;
1924	if (!fBroadcastEtherAddReported)
1925	{
1926	LogRel(("NAT: Attempt to add pair [%RTmac:%RTnaipv4] in ARP cache was ignored\n",
1927	mac, dst));
1928	fBroadcastEtherAddReported = true;
1929	}
1930	return 1;
1931	}
1932	if (slirp_arp_cache_update(pData, dst, mac))
1933	slirp_arp_cache_add(pData, dst, mac);
1934
1935	return 0;
1936	}
1937
1938
1939	void slirp_set_mtu(PNATState pData, int mtu)
1940	{
1941	if (mtu < 20 \|\| mtu >= 16000)
1942	{
1943	LogRel(("NAT: MTU(%d) is out of range (20;16000] mtu forcely assigned to 1500\n", mtu));
1944	mtu = 1500;
1945	}
1946	/* MTU is maximum transition unit on */
1947	if_mtu =
1948	if_mru = mtu;
1949	}
1950
1951	/**
1952	* Info handler.
1953	*/
1954	void slirp_info(PNATState pData, const void pvArg, const char pszArgs)
1955	{
1956	struct socket so, so_next;
1957	struct arp_cache_entry *ac;
1958	struct port_forward_rule *rule;
1959	PCDBGFINFOHLP pHlp = (PCDBGFINFOHLP)pvArg;
1960	NOREF(pszArgs);
1961
1962	pHlp->pfnPrintf(pHlp, "NAT parameters: MTU=%d\n", if_mtu);
1963	pHlp->pfnPrintf(pHlp, "NAT TCP ports:\n");
1964	QSOCKET_FOREACH(so, so_next, tcp)
1965	/* { */
1966	pHlp->pfnPrintf(pHlp, " %R[natsock]\n", so);
1967	}
1968
1969	pHlp->pfnPrintf(pHlp, "NAT UDP ports:\n");
1970	QSOCKET_FOREACH(so, so_next, udp)
1971	/* { */
1972	pHlp->pfnPrintf(pHlp, " %R[natsock]\n", so);
1973	}
1974
1975	pHlp->pfnPrintf(pHlp, "NAT ARP cache:\n");
1976	LIST_FOREACH(ac, &pData->arp_cache, list)
1977	{
1978	pHlp->pfnPrintf(pHlp, " %RTnaipv4 %RTmac\n", ac->ip, &ac->ether);
1979	}
1980
1981	pHlp->pfnPrintf(pHlp, "NAT rules:\n");
1982	LIST_FOREACH(rule, &pData->port_forward_rule_head, list)
1983	{
1984	pHlp->pfnPrintf(pHlp, " %s %d => %RTnaipv4:%d %c\n",
1985	rule->proto == IPPROTO_UDP ? "UDP" : "TCP",
1986	rule->host_port, rule->guest_addr.s_addr, rule->guest_port,
1987	rule->activated ? ' ' : '*');
1988	}
1989	}
1990
1991	/**
1992	* @note: NATState::fUseHostResolver could be changed in bootp.c::dhcp_decode
1993	* @note: this function is executed on GUI/VirtualBox or main/VBoxHeadless thread.
1994	* @note: this function can potentially race with bootp.c::dhcp_decode (except Darwin)
1995	*/
1996	int slirp_host_network_configuration_change_strategy_selector(const PNATState pData)
1997	{
1998	if (pData->fUseHostResolverPermanent)
1999	return VBOX_NAT_DNS_HOSTRESOLVER;
2000
2001	if (pData->fUseDnsProxy) {
2002	#if HAVE_NOTIFICATION_FOR_DNS_UPDATE /* XXX */ && !defined(RT_OS_WINDOWS)
2003	/* We dont conflict with bootp.c::dhcp_decode */
2004	struct rcp_state rcp_state;
2005	int rc;
2006
2007	rcp_state.rcps_flags \|= RCPSF_IGNORE_IPV6;
2008	rc = rcp_parse(&rcp_state, RESOLV_CONF_FILE);
2009	LogRelFunc(("NAT: rcp_parse:%Rrc old domain:%s new domain:%s\n",
2010	rc, LIST_EMPTY(&pData->pDomainList)
2011	? "(null)"
2012	: LIST_FIRST(&pData->pDomainList)->dd_pszDomain,
2013	rcp_state.rcps_domain));
2014	if ( RT_FAILURE(rc)
2015	\|\| LIST_EMPTY(&pData->pDomainList))
2016	return VBOX_NAT_DNS_DNSPROXY;
2017
2018	if ( rcp_state.rcps_domain
2019	&& strcmp(rcp_state.rcps_domain, LIST_FIRST(&pData->pDomainList)->dd_pszDomain) == 0)
2020	return VBOX_NAT_DNS_DNSPROXY;
2021	else
2022	return VBOX_NAT_DNS_EXTERNAL;
2023	#else
2024	/* copy domain name */
2025	/* domain only compare with coy version */
2026	return VBOX_NAT_DNS_DNSPROXY;
2027	#endif
2028	}
2029	return VBOX_NAT_DNS_EXTERNAL;
2030	}

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

source: vbox/trunk/src/VBox/Devices/Network/slirp/slirp.c@ 59420

Download in other formats: