slirp.c@ 26598

Last change on this file since 26598 was 26574, checked in by vboxsync, 15 years ago
Networking: Preparing to make the driver return a send buffer to the device emulation.
Property svn:eol-style set to `native`
File size: 65.9 KB

Line
1	#include "slirp.h"
2	#ifdef RT_OS_OS2
3	# include <paths.h>
4	#endif
5
6	#include <VBox/err.h>
7	#include <VBox/pdmdrv.h>
8	#include <iprt/assert.h>
9	#include <iprt/file.h>
10	#ifndef RT_OS_WINDOWS
11	# include <sys/ioctl.h>
12	# include <poll.h>
13	#else
14	# include <Winnls.h>
15	# define _WINSOCK2API_
16	# include <IPHlpApi.h>
17	#endif
18	#include <alias.h>
19
20	#ifndef RT_OS_WINDOWS
21
22	# define DO_ENGAGE_EVENT1(so, fdset, label) \
23	do { \
24	if ( so->so_poll_index != -1 \
25	&& so->s == polls[so->so_poll_index].fd) \
26	{ \
27	polls[so->so_poll_index].events \|= N_(fdset ## _poll); \
28	break; \
29	} \
30	AssertRelease(poll_index < (nfds)); \
31	AssertRelease(poll_index >= 0 && poll_index < (nfds)); \
32	polls[poll_index].fd = (so)->s; \
33	(so)->so_poll_index = poll_index; \
34	polls[poll_index].events = N_(fdset ## _poll); \
35	polls[poll_index].revents = 0; \
36	poll_index++; \
37	} while (0)
38
39	# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
40	do { \
41	if ( so->so_poll_index != -1 \
42	&& so->s == polls[so->so_poll_index].fd) \
43	{ \
44	polls[so->so_poll_index].events \|= \
45	N_(fdset1 ## _poll) \| N_(fdset2 ## _poll); \
46	break; \
47	} \
48	AssertRelease(poll_index < (nfds)); \
49	polls[poll_index].fd = (so)->s; \
50	(so)->so_poll_index = poll_index; \
51	polls[poll_index].events = \
52	N_(fdset1 ## _poll) \| N_(fdset2 ## _poll); \
53	poll_index++; \
54	} while (0)
55
56	# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
57
58	# define DO_CHECK_FD_SET(so, events, fdset) \
59	( ((so)->so_poll_index != -1) \
60	&& ((so)->so_poll_index <= ndfs) \
61	&& ((so)->s == polls[so->so_poll_index].fd) \
62	&& (polls[(so)->so_poll_index].revents & N_(fdset ## _poll)))
63
64	/* specific for Unix API */
65	# define DO_UNIX_CHECK_FD_SET(so, events, fdset) DO_CHECK_FD_SET((so), (events), fdset)
66	/* specific for Windows Winsock API */
67	# define DO_WIN_CHECK_FD_SET(so, events, fdset) 0
68
69	# ifndef RT_OS_LINUX
70	# define readfds_poll (POLLRDNORM)
71	# define writefds_poll (POLLWRNORM)
72	# else
73	# define readfds_poll (POLLIN)
74	# define writefds_poll (POLLOUT)
75	# endif
76	# define xfds_poll (POLLPRI)
77	# define closefds_poll (POLLHUP)
78	# define rderr_poll (POLLERR)
79	# define rdhup_poll (POLLHUP)
80	# define nval_poll (POLLNVAL)
81
82	# define ICMP_ENGAGE_EVENT(so, fdset) \
83	do { \
84	if (pData->icmp_socket.s != -1) \
85	DO_ENGAGE_EVENT1((so), fdset, ICMP); \
86	} while (0)
87
88	#else /* RT_OS_WINDOWS */
89
90	/*
91	* On Windows, we will be notified by IcmpSendEcho2() when the response arrives.
92	* So no call to WSAEventSelect necessary.
93	*/
94	# define ICMP_ENGAGE_EVENT(so, fdset) do {} while (0)
95
96	/*
97	* On Windows we use FD_ALL_EVENTS to ensure that we don't miss any event.
98	*/
99	# define DO_ENGAGE_EVENT1(so, fdset1, label) \
100	do { \
101	rc = WSAEventSelect((so)->s, VBOX_SOCKET_EVENT, FD_ALL_EVENTS); \
102	if (rc == SOCKET_ERROR) \
103	{ \
104	/* This should not happen */ \
105	error = WSAGetLastError(); \
106	LogRel(("WSAEventSelect (" #label ") error %d (so=%x, socket=%s, event=%x)\n", \
107	error, (so), (so)->s, VBOX_SOCKET_EVENT)); \
108	} \
109	} while (0); \
110	CONTINUE(label)
111
112	# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
113	DO_ENGAGE_EVENT1((so), (fdset1), label)
114
115	# define DO_POLL_EVENTS(rc, error, so, events, label) \
116	(rc) = WSAEnumNetworkEvents((so)->s, VBOX_SOCKET_EVENT, (events)); \
117	if ((rc) == SOCKET_ERROR) \
118	{ \
119	(error) = WSAGetLastError(); \
120	LogRel(("WSAEnumNetworkEvents " #label " error %d\n", (error))); \
121	CONTINUE(label); \
122	}
123
124	# define acceptds_win FD_ACCEPT
125	# define acceptds_win_bit FD_ACCEPT_BIT
126	# define readfds_win FD_READ
127	# define readfds_win_bit FD_READ_BIT
128	# define writefds_win FD_WRITE
129	# define writefds_win_bit FD_WRITE_BIT
130	# define xfds_win FD_OOB
131	# define xfds_win_bit FD_OOB_BIT
132	# define closefds_win FD_CLOSE
133	# define closefds_win_bit FD_CLOSE_BIT
134
135	# define DO_CHECK_FD_SET(so, events, fdset) \
136	(((events).lNetworkEvents & fdset ## _win) && ((events).iErrorCode[fdset ## _win_bit] == 0))
137
138	# define DO_WIN_CHECK_FD_SET(so, events, fdset) DO_CHECK_FD_SET((so), (events), fdset)
139	# define DO_UNIX_CHECK_FD_SET(so, events, fdset) 1 /specific for Unix API /
140
141	#endif /* RT_OS_WINDOWS */
142
143	#define TCP_ENGAGE_EVENT1(so, fdset) \
144	DO_ENGAGE_EVENT1((so), fdset, tcp)
145
146	#define TCP_ENGAGE_EVENT2(so, fdset1, fdset2) \
147	DO_ENGAGE_EVENT2((so), fdset1, fdset2, tcp)
148
149	#define UDP_ENGAGE_EVENT(so, fdset) \
150	DO_ENGAGE_EVENT1((so), fdset, udp)
151
152	#define POLL_TCP_EVENTS(rc, error, so, events) \
153	DO_POLL_EVENTS((rc), (error), (so), (events), tcp)
154
155	#define POLL_UDP_EVENTS(rc, error, so, events) \
156	DO_POLL_EVENTS((rc), (error), (so), (events), udp)
157
158	#define CHECK_FD_SET(so, events, set) \
159	(DO_CHECK_FD_SET((so), (events), set))
160
161	#define WIN_CHECK_FD_SET(so, events, set) \
162	(DO_WIN_CHECK_FD_SET((so), (events), set))
163
164	#define UNIX_CHECK_FD_SET(so, events, set) \
165	(DO_UNIX_CHECK_FD_SET(so, events, set))
166
167	/*
168	* Loging macros
169	*/
170	#if VBOX_WITH_DEBUG_NAT_SOCKETS
171	# if defined(RT_OS_WINDOWS)
172	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
173	do { \
174	LogRel((" " #proto " %R[natsock] %R[natwinnetevents]\n", (so), (winevent))); \
175	} while (0)
176	# else /* !RT_OS_WINDOWS */
177	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
178	do { \
179	LogRel((" " #proto " %R[natsock] %s %s %s er: %s, %s, %s\n", (so), \
180	CHECK_FD_SET(so, ign ,r_fdset) ? "READ":"", \
181	CHECK_FD_SET(so, ign, w_fdset) ? "WRITE":"", \
182	CHECK_FD_SET(so, ign, x_fdset) ? "OOB":"", \
183	CHECK_FD_SET(so, ign, rderr) ? "RDERR":"", \
184	CHECK_FD_SET(so, ign, rdhup) ? "RDHUP":"", \
185	CHECK_FD_SET(so, ign, nval) ? "RDNVAL":"")); \
186	} while (0)
187	# endif /* !RT_OS_WINDOWS */
188	#else /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
189	# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) do {} while (0)
190	#endif /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
191
192	#define LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
193	DO_LOG_NAT_SOCK((so), proto, (winevent), r_fdset, w_fdset, x_fdset)
194
195	static void activate_port_forwarding(PNATState, const uint8_t *pEther);
196
197	static const uint8_t special_ethaddr[6] =
198	{
199	0x52, 0x54, 0x00, 0x12, 0x35, 0x00
200	};
201
202	static const uint8_t broadcast_ethaddr[6] =
203	{
204	0xff, 0xff, 0xff, 0xff, 0xff, 0xff
205	};
206
207	const uint8_t zerro_ethaddr[6] =
208	{
209	0x0, 0x0, 0x0, 0x0, 0x0, 0x0
210	};
211
212	#ifdef RT_OS_WINDOWS
213	static int get_dns_addr_domain(PNATState pData, bool fVerbose,
214	struct in_addr *pdns_addr,
215	const char **ppszDomain)
216	{
217	ULONG flags = GAA_FLAG_INCLUDE_PREFIX; /GAA_FLAG_INCLUDE_ALL_INTERFACES;/ /* all interfaces registered in NDIS */
218	PIP_ADAPTER_ADDRESSES pAdapterAddr = NULL;
219	PIP_ADAPTER_ADDRESSES pAddr = NULL;
220	PIP_ADAPTER_DNS_SERVER_ADDRESS pDnsAddr = NULL;
221	ULONG size;
222	int wlen = 0;
223	char *pszSuffix;
224	struct dns_domain_entry *pDomain = NULL;
225	ULONG ret = ERROR_SUCCESS;
226
227	/* @todo add SKIPing flags to get only required information */
228
229	/* determine size of buffer */
230	size = 0;
231	ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, pAdapterAddr, &size);
232	if (ret != ERROR_BUFFER_OVERFLOW)
233	{
234	LogRel(("NAT: error %lu occurred on capacity detection operation\n", ret));
235	return -1;
236	}
237	if (size == 0)
238	{
239	LogRel(("NAT: Win socket API returns non capacity\n"));
240	return -1;
241	}
242
243	pAdapterAddr = RTMemAllocZ(size);
244	if (!pAdapterAddr)
245	{
246	LogRel(("NAT: No memory available \n"));
247	return -1;
248	}
249	ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, pAdapterAddr, &size);
250	if (ret != ERROR_SUCCESS)
251	{
252	LogRel(("NAT: error %lu occurred on fetching adapters info\n", ret));
253	RTMemFree(pAdapterAddr);
254	return -1;
255	}
256
257	for (pAddr = pAdapterAddr; pAddr != NULL; pAddr = pAddr->Next)
258	{
259	int found;
260	if (pAddr->OperStatus != IfOperStatusUp)
261	continue;
262
263	for (pDnsAddr = pAddr->FirstDnsServerAddress; pDnsAddr != NULL; pDnsAddr = pDnsAddr->Next)
264	{
265	struct sockaddr *SockAddr = pDnsAddr->Address.lpSockaddr;
266	struct in_addr InAddr;
267	struct dns_entry *pDns;
268
269	if (SockAddr->sa_family != AF_INET)
270	continue;
271
272	InAddr = ((struct sockaddr_in *)SockAddr)->sin_addr;
273
274	/* add dns server to list */
275	pDns = RTMemAllocZ(sizeof(struct dns_entry));
276	if (!pDns)
277	{
278	LogRel(("NAT: Can't allocate buffer for DNS entry\n"));
279	RTMemFree(pAdapterAddr);
280	return VERR_NO_MEMORY;
281	}
282
283	LogRel(("NAT: adding %R[IP4] to DNS server list\n", &InAddr));
284	if ((InAddr.s_addr & RT_H2N_U32_C(IN_CLASSA_NET)) == RT_N2H_U32_C(INADDR_LOOPBACK & IN_CLASSA_NET))
285	pDns->de_addr.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
286	else
287	pDns->de_addr.s_addr = InAddr.s_addr;
288
289	TAILQ_INSERT_HEAD(&pData->pDnsList, pDns, de_list);
290
291	if (pAddr->DnsSuffix == NULL)
292	continue;
293
294	/* uniq */
295	RTUtf16ToUtf8(pAddr->DnsSuffix, &pszSuffix);
296	if (!pszSuffix \|\| strlen(pszSuffix) == 0)
297	{
298	RTStrFree(pszSuffix);
299	continue;
300	}
301
302	found = 0;
303	LIST_FOREACH(pDomain, &pData->pDomainList, dd_list)
304	{
305	if ( pDomain->dd_pszDomain != NULL
306	&& strcmp(pDomain->dd_pszDomain, pszSuffix) == 0)
307	{
308	found = 1;
309	RTStrFree(pszSuffix);
310	break;
311	}
312	}
313	if (!found)
314	{
315	pDomain = RTMemAllocZ(sizeof(struct dns_domain_entry));
316	if (!pDomain)
317	{
318	LogRel(("NAT: not enough memory\n"));
319	RTStrFree(pszSuffix);
320	RTMemFree(pAdapterAddr);
321	return VERR_NO_MEMORY;
322	}
323	pDomain->dd_pszDomain = pszSuffix;
324	LogRel(("NAT: adding domain name %s to search list\n", pDomain->dd_pszDomain));
325	LIST_INSERT_HEAD(&pData->pDomainList, pDomain, dd_list);
326	}
327	}
328	}
329	RTMemFree(pAdapterAddr);
330	return 0;
331	}
332
333	#else /* !RT_OS_WINDOWS */
334
335	static int RTFileGets(RTFILE File, void pvBuf, size_t cbBufSize, size_t pcbRead)
336	{
337	size_t cbRead;
338	char bTest;
339	int rc = VERR_NO_MEMORY;
340	char pu8Buf = (char )pvBuf;
341	*pcbRead = 0;
342
343	while ( RT_SUCCESS(rc = RTFileRead(File, &bTest, 1, &cbRead))
344	&& (pu8Buf - (char *)pvBuf) < cbBufSize)
345	{
346	if (cbRead == 0)
347	return VERR_EOF;
348
349	if (bTest == '\r' \|\| bTest == '\n')
350	{
351	*pu8Buf = 0;
352	return VINF_SUCCESS;
353	}
354	*pu8Buf = bTest;
355	pu8Buf++;
356	(*pcbRead)++;
357	}
358	return rc;
359	}
360
361	static int get_dns_addr_domain(PNATState pData, bool fVerbose,
362	struct in_addr *pdns_addr,
363	const char **ppszDomain)
364	{
365	char buff[512];
366	char buff2[256];
367	RTFILE f;
368	int fFoundNameserver = 0;
369	struct in_addr tmp_addr;
370	int rc;
371	size_t bytes;
372
373	# ifdef RT_OS_OS2
374	/* Try various locations. */
375	char *etc = getenv("ETC");
376	if (etc)
377	{
378	RTStrmPrintf(buff, sizeof(buff), "%s/RESOLV2", etc);
379	rc = RTFileOpen(&f, buff, RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
380	}
381	if (RT_FAILURE(rc))
382	{
383	RTStrmPrintf(buff, sizeof(buff), "%s/RESOLV2", _PATH_ETC);
384	rc = RTFileOpen(&f, buff, RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
385	}
386	if (RT_FAILURE(rc))
387	{
388	RTStrmPrintf(buff, sizeof(buff), "%s/resolv.conf", _PATH_ETC);
389	rc = RTFileOpen(&f, buff, RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
390	}
391	# else /* !RT_OS_OS2 */
392	# ifndef DEBUG_vvl
393	rc = RTFileOpen(&f, "/etc/resolv.conf", RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
394	# else
395	char *home = getenv("HOME");
396	RTStrPrintf(buff, sizeof(buff), "%s/resolv.conf", home);
397	rc = RTFileOpen(&f, buff, RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
398	if (RT_SUCCESS(rc))
399	{
400	Log(("NAT: DNS we're using %s\n", buff));
401	}
402	else
403	{
404	rc = RTFileOpen(&f, "/etc/resolv.conf", RTFILE_O_READ \| RTFILE_O_OPEN \| RTFILE_O_DENY_NONE);
405	Log(("NAT: DNS we're using %s\n", buff));
406	}
407	# endif
408	# endif /* !RT_OS_OS2 */
409	if (RT_FAILURE(rc))
410	return -1;
411
412	if (ppszDomain)
413	*ppszDomain = NULL;
414
415	Log(("NAT: DNS Servers:\n"));
416	while ( RT_SUCCESS(rc = RTFileGets(f, buff, 512, &bytes))
417	&& rc != VERR_EOF)
418	{
419	struct dns_entry *pDns = NULL;
420	if (sscanf(buff, "nameserver%*[ \t]%256s", buff2) == 1)
421	{
422	if (!inet_aton(buff2, &tmp_addr))
423	continue;
424
425	/* localhost mask */
426	pDns = RTMemAllocZ(sizeof (struct dns_entry));
427	if (!pDns)
428	{
429	LogRel(("can't alloc memory for DNS entry\n"));
430	return -1;
431	}
432
433	/* check */
434	pDns->de_addr.s_addr = tmp_addr.s_addr;
435	if ((pDns->de_addr.s_addr & RT_H2N_U32_C(IN_CLASSA_NET)) == RT_N2H_U32_C(INADDR_LOOPBACK & IN_CLASSA_NET))
436	{
437	pDns->de_addr.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
438	}
439	TAILQ_INSERT_HEAD(&pData->pDnsList, pDns, de_list);
440	fFoundNameserver++;
441	}
442	if ((!strncmp(buff, "domain", 6) \|\| !strncmp(buff, "search", 6)))
443	{
444	char *tok;
445	char *saveptr;
446	struct dns_domain_entry *pDomain = NULL;
447	int fFoundDomain = 0;
448	tok = strtok_r(&buff[6], " \t\n", &saveptr);
449	LIST_FOREACH(pDomain, &pData->pDomainList, dd_list)
450	{
451	if ( tok != NULL
452	&& strcmp(tok, pDomain->dd_pszDomain) == 0)
453	{
454	fFoundDomain = 1;
455	break;
456	}
457	}
458	if (tok != NULL && !fFoundDomain)
459	{
460	pDomain = RTMemAllocZ(sizeof(struct dns_domain_entry));
461	if (!pDomain)
462	{
463	LogRel(("NAT: not enought memory to add domain list\n"));
464	return VERR_NO_MEMORY;
465	}
466	pDomain->dd_pszDomain = RTStrDup(tok);
467	LogRel(("NAT: adding domain name %s to search list\n", pDomain->dd_pszDomain));
468	LIST_INSERT_HEAD(&pData->pDomainList, pDomain, dd_list);
469	}
470	}
471	}
472	RTFileClose(f);
473	if (!fFoundNameserver)
474	return -1;
475	return 0;
476	}
477
478	#endif /* !RT_OS_WINDOWS */
479
480	static int slirp_init_dns_list(PNATState pData)
481	{
482	TAILQ_INIT(&pData->pDnsList);
483	LIST_INIT(&pData->pDomainList);
484	return get_dns_addr_domain(pData, true, NULL, NULL);
485	}
486
487	static void slirp_release_dns_list(PNATState pData)
488	{
489	struct dns_entry *pDns = NULL;
490	struct dns_domain_entry *pDomain = NULL;
491
492	while (!TAILQ_EMPTY(&pData->pDnsList))
493	{
494	pDns = TAILQ_FIRST(&pData->pDnsList);
495	TAILQ_REMOVE(&pData->pDnsList, pDns, de_list);
496	RTMemFree(pDns);
497	}
498
499	while (!LIST_EMPTY(&pData->pDomainList))
500	{
501	pDomain = LIST_FIRST(&pData->pDomainList);
502	LIST_REMOVE(pDomain, dd_list);
503	if (pDomain->dd_pszDomain != NULL)
504	RTStrFree(pDomain->dd_pszDomain);
505	RTMemFree(pDomain);
506	}
507	}
508
509	int get_dns_addr(PNATState pData, struct in_addr *pdns_addr)
510	{
511	return get_dns_addr_domain(pData, false, pdns_addr, NULL);
512	}
513
514	#ifndef VBOX_WITH_NAT_SERVICE
515	int slirp_init(PNATState ppData, const char pszNetAddr, uint32_t u32Netmask,
516	bool fPassDomain, bool fUseHostResolver, void *pvUser)
517	#else
518	int slirp_init(PNATState *ppData, uint32_t u32NetAddr, uint32_t u32Netmask,
519	bool fPassDomain, void *pvUser)
520	#endif
521	{
522	int fNATfailed = 0;
523	int rc;
524	PNATState pData = RTMemAllocZ(sizeof(NATState));
525	*ppData = pData;
526	if (!pData)
527	return VERR_NO_MEMORY;
528	if (u32Netmask & 0x1f)
529	/* CTL is x.x.x.15, bootp passes up to 16 IPs (15..31) */
530	return VERR_INVALID_PARAMETER;
531	pData->fPassDomain = !fUseHostResolver ? fPassDomain : false;
532	pData->use_host_resolver = fUseHostResolver;
533	pData->pvUser = pvUser;
534	pData->netmask = u32Netmask;
535
536	/* sockets & TCP defaults */
537	pData->socket_rcv = 64 * _1K;
538	pData->socket_snd = 64 * _1K;
539	tcp_sndspace = 64 * _1K;
540	tcp_rcvspace = 64 * _1K;
541
542	#ifdef RT_OS_WINDOWS
543	{
544	WSADATA Data;
545	WSAStartup(MAKEWORD(2, 0), &Data);
546	}
547	pData->phEvents[VBOX_SOCKET_EVENT_INDEX] = CreateEvent(NULL, FALSE, FALSE, NULL);
548	#endif
549	#ifdef VBOX_WITH_SLIRP_MT
550	QSOCKET_LOCK_CREATE(tcb);
551	QSOCKET_LOCK_CREATE(udb);
552	rc = RTReqCreateQueue(&pData->pReqQueue);
553	AssertReleaseRC(rc);
554	#endif
555
556	link_up = 1;
557
558	rc = bootp_dhcp_init(pData);
559	if (rc != 0)
560	{
561	LogRel(("NAT: DHCP server initialization was failed\n"));
562	return VINF_NAT_DNS;
563	}
564	debug_init();
565	if_init(pData);
566	ip_init(pData);
567	icmp_init(pData);
568
569	/* Initialise mbufs after setting the MTU */
570	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
571	m_init(pData);
572	#else
573	mbuf_init(pData);
574	#endif
575
576	#ifndef VBOX_WITH_NAT_SERVICE
577	inet_aton(pszNetAddr, &pData->special_addr);
578	#else
579	pData->special_addr.s_addr = u32NetAddr;
580	#endif
581	pData->slirp_ethaddr = &special_ethaddr[0];
582	alias_addr.s_addr = pData->special_addr.s_addr \| RT_H2N_U32_C(CTL_ALIAS);
583	/* @todo: add ability to configure this staff */
584
585	/* set default addresses */
586	inet_aton("127.0.0.1", &loopback_addr);
587	if (!pData->use_host_resolver)
588	{
589	if (slirp_init_dns_list(pData) < 0)
590	fNATfailed = 1;
591
592	dnsproxy_init(pData);
593	}
594
595	getouraddr(pData);
596	{
597	int flags = 0;
598	struct in_addr proxy_addr;
599	pData->proxy_alias = LibAliasInit(pData, NULL);
600	if (pData->proxy_alias == NULL)
601	{
602	LogRel(("NAT: LibAlias default rule wasn't initialized\n"));
603	AssertMsgFailed(("NAT: LibAlias default rule wasn't initialized\n"));
604	}
605	flags = LibAliasSetMode(pData->proxy_alias, 0, 0);
606	#ifndef NO_FW_PUNCH
607	flags \|= PKT_ALIAS_PUNCH_FW;
608	#endif
609	flags \|= PKT_ALIAS_LOG; /* set logging */
610	flags = LibAliasSetMode(pData->proxy_alias, flags, ~0);
611	proxy_addr.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
612	LibAliasSetAddress(pData->proxy_alias, proxy_addr);
613	ftp_alias_load(pData);
614	nbt_alias_load(pData);
615	if (pData->use_host_resolver)
616	dns_alias_load(pData);
617	}
618	return fNATfailed ? VINF_NAT_DNS : VINF_SUCCESS;
619	}
620
621	/**
622	* Register statistics.
623	*/
624	void slirp_register_statistics(PNATState pData, PPDMDRVINS pDrvIns)
625	{
626	#ifdef VBOX_WITH_STATISTICS
627	# define PROFILE_COUNTER(name, dsc) REGISTER_COUNTER(name, pData, STAMTYPE_PROFILE, STAMUNIT_TICKS_PER_CALL, dsc)
628	# define COUNTING_COUNTER(name, dsc) REGISTER_COUNTER(name, pData, STAMTYPE_COUNTER, STAMUNIT_COUNT, dsc)
629	# include "counters.h"
630	# undef COUNTER
631	/** @todo register statistics for the variables dumped by:
632	* ipstats(pData); tcpstats(pData); udpstats(pData); icmpstats(pData);
633	* mbufstats(pData); sockstats(pData); */
634	#endif /* VBOX_WITH_STATISTICS */
635	}
636
637	/**
638	* Deregister statistics.
639	*/
640	void slirp_deregister_statistics(PNATState pData, PPDMDRVINS pDrvIns)
641	{
642	#ifdef VBOX_WITH_STATISTICS
643	# define PROFILE_COUNTER(name, dsc) DEREGISTER_COUNTER(name, pData)
644	# define COUNTING_COUNTER(name, dsc) DEREGISTER_COUNTER(name, pData)
645	# include "counters.h"
646	#endif /* VBOX_WITH_STATISTICS */
647	}
648
649	/**
650	* Marks the link as up, making it possible to establish new connections.
651	*/
652	void slirp_link_up(PNATState pData)
653	{
654	struct arp_cache_entry *ac;
655	link_up = 1;
656
657	if (LIST_EMPTY(&pData->arp_cache))
658	return;
659
660	LIST_FOREACH(ac, &pData->arp_cache, list)
661	{
662	activate_port_forwarding(pData, ac->ether);
663	}
664	}
665
666	/**
667	* Marks the link as down and cleans up the current connections.
668	*/
669	void slirp_link_down(PNATState pData)
670	{
671	struct socket *so;
672	struct port_forward_rule *rule;
673
674	while ((so = tcb.so_next) != &tcb)
675	{
676	if (so->so_state & SS_NOFDREF \|\| so->s == -1)
677	sofree(pData, so);
678	else
679	tcp_drop(pData, sototcpcb(so), 0);
680	}
681
682	while ((so = udb.so_next) != &udb)
683	udp_detach(pData, so);
684
685	/*
686	* Clear the active state of port-forwarding rules to force
687	* re-setup on restoration of communications.
688	*/
689	LIST_FOREACH(rule, &pData->port_forward_rule_head, list)
690	{
691	rule->activated = 0;
692	}
693	pData->cRedirectionsActive = 0;
694
695	link_up = 0;
696	}
697
698	/**
699	* Terminates the slirp component.
700	*/
701	void slirp_term(PNATState pData)
702	{
703	#ifdef RT_OS_WINDOWS
704	pData->pfIcmpCloseHandle(pData->icmp_socket.sh);
705	FreeLibrary(pData->hmIcmpLibrary);
706	RTMemFree(pData->pvIcmpBuffer);
707	#else
708	closesocket(pData->icmp_socket.s);
709	#endif
710
711	slirp_link_down(pData);
712	slirp_release_dns_list(pData);
713	ftp_alias_unload(pData);
714	nbt_alias_unload(pData);
715	if (pData->use_host_resolver)
716	dns_alias_unload(pData);
717	while (!LIST_EMPTY(&instancehead))
718	{
719	struct libalias *la = LIST_FIRST(&instancehead);
720	/* libalias do all clean up */
721	LibAliasUninit(la);
722	}
723	while (!LIST_EMPTY(&pData->arp_cache))
724	{
725	struct arp_cache_entry *ac = LIST_FIRST(&pData->arp_cache);
726	LIST_REMOVE(ac, list);
727	RTMemFree(ac);
728	}
729	bootp_dhcp_fini(pData);
730	m_fini(pData);
731	#ifdef RT_OS_WINDOWS
732	WSACleanup();
733	#endif
734	#ifdef LOG_ENABLED
735	Log(("\n"
736	"NAT statistics\n"
737	"--------------\n"
738	"\n"));
739	ipstats(pData);
740	tcpstats(pData);
741	udpstats(pData);
742	icmpstats(pData);
743	mbufstats(pData);
744	sockstats(pData);
745	Log(("\n"
746	"\n"
747	"\n"));
748	#endif
749	RTMemFree(pData);
750	}
751
752
753	#define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE\|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
754	#define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE\|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
755
756	/*
757	* curtime kept to an accuracy of 1ms
758	*/
759	static void updtime(PNATState pData)
760	{
761	#ifdef RT_OS_WINDOWS
762	struct _timeb tb;
763
764	_ftime(&tb);
765	curtime = (u_int)tb.time * (u_int)1000;
766	curtime += (u_int)tb.millitm;
767	#else
768	gettimeofday(&tt, 0);
769
770	curtime = (u_int)tt.tv_sec * (u_int)1000;
771	curtime += (u_int)tt.tv_usec / (u_int)1000;
772
773	if ((tt.tv_usec % 1000) >= 500)
774	curtime++;
775	#endif
776	}
777
778	#ifdef RT_OS_WINDOWS
779	void slirp_select_fill(PNATState pData, int *pnfds)
780	#else /* RT_OS_WINDOWS */
781	void slirp_select_fill(PNATState pData, int pnfds, struct pollfd polls)
782	#endif /* !RT_OS_WINDOWS */
783	{
784	struct socket so, so_next;
785	int nfds;
786	#if defined(RT_OS_WINDOWS)
787	int rc;
788	int error;
789	#else
790	int poll_index = 0;
791	#endif
792	int i;
793
794	STAM_PROFILE_START(&pData->StatFill, a);
795
796	nfds = *pnfds;
797
798	/*
799	* First, TCP sockets
800	*/
801	do_slowtimo = 0;
802	if (!link_up)
803	goto done;
804
805	/*
806	* *_slowtimo needs calling if there are IP fragments
807	* in the fragment queue, or there are TCP connections active
808	*/
809	/* XXX:
810	* triggering of fragment expiration should be the same but use new macroses
811	*/
812	do_slowtimo = (tcb.so_next != &tcb);
813	if (!do_slowtimo)
814	{
815	for (i = 0; i < IPREASS_NHASH; i++)
816	{
817	if (!TAILQ_EMPTY(&ipq[i]))
818	{
819	do_slowtimo = 1;
820	slirp_arm_slow_timer(pData->pvUser);
821	break;
822	}
823	}
824	}
825	ICMP_ENGAGE_EVENT(&pData->icmp_socket, readfds);
826
827	STAM_COUNTER_RESET(&pData->StatTCP);
828	STAM_COUNTER_RESET(&pData->StatTCPHot);
829
830	QSOCKET_FOREACH(so, so_next, tcp)
831	/* { */
832	#if !defined(RT_OS_WINDOWS)
833	so->so_poll_index = -1;
834	#endif
835	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
836	if (pData->fmbuf_water_line == 1)
837	{
838	if (mbuf_alloced < pData->mbuf_water_line_limit/2)
839	{
840	pData->fmbuf_water_warn_sent = 0;
841	pData->fmbuf_water_line = 0;
842	}
843	# ifndef RT_OS_WINDOWS
844	poll_index = 0;
845	# endif
846	goto done;
847	}
848	#endif /* !VBOX_WITH_SLIRP_BSD_MBUF */
849	STAM_COUNTER_INC(&pData->StatTCP);
850
851	/*
852	* See if we need a tcp_fasttimo
853	*/
854	if ( time_fasttimo == 0
855	&& so->so_tcpcb != NULL
856	&& so->so_tcpcb->t_flags & TF_DELACK)
857	{
858	time_fasttimo = curtime; /* Flag when we want a fasttimo */
859	slirp_arm_fast_timer(pData->pvUser);
860	}
861
862	/*
863	* NOFDREF can include still connecting to local-host,
864	* newly socreated() sockets etc. Don't want to select these.
865	*/
866	if (so->so_state & SS_NOFDREF \|\| so->s == -1)
867	CONTINUE(tcp);
868
869	/*
870	* Set for reading sockets which are accepting
871	*/
872	if (so->so_state & SS_FACCEPTCONN)
873	{
874	STAM_COUNTER_INC(&pData->StatTCPHot);
875	TCP_ENGAGE_EVENT1(so, readfds);
876	CONTINUE(tcp);
877	}
878
879	/*
880	* Set for writing sockets which are connecting
881	*/
882	if (so->so_state & SS_ISFCONNECTING)
883	{
884	Log2(("connecting %R[natsock] engaged\n",so));
885	STAM_COUNTER_INC(&pData->StatTCPHot);
886	TCP_ENGAGE_EVENT1(so, writefds);
887	}
888
889	/*
890	* Set for writing if we are connected, can send more, and
891	* we have something to send
892	*/
893	if (CONN_CANFSEND(so) && so->so_rcv.sb_cc)
894	{
895	STAM_COUNTER_INC(&pData->StatTCPHot);
896	TCP_ENGAGE_EVENT1(so, writefds);
897	}
898
899	/*
900	* Set for reading (and urgent data) if we are connected, can
901	* receive more, and we have room for it XXX /2 ?
902	*/
903	if (CONN_CANFRCV(so) && (so->so_snd.sb_cc < (so->so_snd.sb_datalen/2)))
904	{
905	STAM_COUNTER_INC(&pData->StatTCPHot);
906	TCP_ENGAGE_EVENT2(so, readfds, xfds);
907	}
908	LOOP_LABEL(tcp, so, so_next);
909	}
910
911	/*
912	* UDP sockets
913	*/
914	STAM_COUNTER_RESET(&pData->StatUDP);
915	STAM_COUNTER_RESET(&pData->StatUDPHot);
916
917	QSOCKET_FOREACH(so, so_next, udp)
918	/* { */
919
920	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
921	if (pData->fmbuf_water_line == 1)
922	{
923	if (mbuf_alloced < pData->mbuf_water_line_limit/2)
924	{
925	pData->fmbuf_water_line = 0;
926	pData->fmbuf_water_warn_sent = 0;
927	}
928	# ifndef RT_OS_WINDOWS
929	poll_index = 0;
930	# endif
931	goto done;
932	}
933	#endif /* !VBOX_WITH_SLIRP_BSD_MBUF */
934	STAM_COUNTER_INC(&pData->StatUDP);
935	#if !defined(RT_OS_WINDOWS)
936	so->so_poll_index = -1;
937	#endif
938
939	/*
940	* See if it's timed out
941	*/
942	if (so->so_expire)
943	{
944	if (so->so_expire <= curtime)
945	{
946	Log2(("NAT: %R[natsock] expired\n", so));
947	if (so->so_timeout != NULL)
948	{
949	so->so_timeout(pData, so, so->so_timeout_arg);
950	}
951	#ifdef VBOX_WITH_SLIRP_MT
952	/* we need so_next for continue our cycle*/
953	so_next = so->so_next;
954	#endif
955	UDP_DETACH(pData, so, so_next);
956	CONTINUE_NO_UNLOCK(udp);
957	}
958	else
959	{
960	do_slowtimo = 1; /* Let socket expire */
961	slirp_arm_slow_timer(pData->pvUser);
962	}
963	}
964
965	/*
966	* When UDP packets are received from over the link, they're
967	* sendto()'d straight away, so no need for setting for writing
968	* Limit the number of packets queued by this session to 4.
969	* Note that even though we try and limit this to 4 packets,
970	* the session could have more queued if the packets needed
971	* to be fragmented.
972	*
973	* (XXX <= 4 ?)
974	*/
975	if ((so->so_state & SS_ISFCONNECTED) && so->so_queued <= 4)
976	{
977	STAM_COUNTER_INC(&pData->StatUDPHot);
978	UDP_ENGAGE_EVENT(so, readfds);
979	}
980	LOOP_LABEL(udp, so, so_next);
981	}
982	done:
983
984	#if defined(RT_OS_WINDOWS)
985	*pnfds = VBOX_EVENT_COUNT;
986	#else /* RT_OS_WINDOWS */
987	AssertRelease(poll_index <= *pnfds);
988	*pnfds = poll_index;
989	#endif /* !RT_OS_WINDOWS */
990
991	STAM_PROFILE_STOP(&pData->StatFill, a);
992	}
993
994	#if defined(RT_OS_WINDOWS)
995	void slirp_select_poll(PNATState pData, int fTimeout, int fIcmp)
996	#else /* RT_OS_WINDOWS */
997	void slirp_select_poll(PNATState pData, struct pollfd *polls, int ndfs)
998	#endif /* !RT_OS_WINDOWS */
999	{
1000	struct socket so, so_next;
1001	int ret;
1002	#if defined(RT_OS_WINDOWS)
1003	WSANETWORKEVENTS NetworkEvents;
1004	int rc;
1005	int error;
1006	#else
1007	int poll_index = 0;
1008	#endif
1009
1010	STAM_PROFILE_START(&pData->StatPoll, a);
1011
1012	/* Update time */
1013	updtime(pData);
1014
1015	/*
1016	* See if anything has timed out
1017	*/
1018	if (link_up)
1019	{
1020	if (time_fasttimo && ((curtime - time_fasttimo) >= 2))
1021	{
1022	STAM_PROFILE_START(&pData->StatFastTimer, b);
1023	tcp_fasttimo(pData);
1024	time_fasttimo = 0;
1025	STAM_PROFILE_STOP(&pData->StatFastTimer, b);
1026	}
1027	if (do_slowtimo && ((curtime - last_slowtimo) >= 499))
1028	{
1029	STAM_PROFILE_START(&pData->StatSlowTimer, c);
1030	ip_slowtimo(pData);
1031	tcp_slowtimo(pData);
1032	last_slowtimo = curtime;
1033	STAM_PROFILE_STOP(&pData->StatSlowTimer, c);
1034	}
1035	}
1036	#if defined(RT_OS_WINDOWS)
1037	if (fTimeout)
1038	return; /* only timer update */
1039	#endif
1040
1041	/*
1042	* Check sockets
1043	*/
1044	if (!link_up)
1045	goto done;
1046	#if defined(RT_OS_WINDOWS)
1047	/*XXX: before renaming please make see define
1048	* fIcmp in slirp_state.h
1049	*/
1050	if (fIcmp)
1051	sorecvfrom(pData, &pData->icmp_socket);
1052	#else
1053	if ( (pData->icmp_socket.s != -1)
1054	&& CHECK_FD_SET(&pData->icmp_socket, ignored, readfds))
1055	sorecvfrom(pData, &pData->icmp_socket);
1056	#endif
1057	/*
1058	* Check TCP sockets
1059	*/
1060	QSOCKET_FOREACH(so, so_next, tcp)
1061	/* { */
1062	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
1063	if (pData->fmbuf_water_line == 1)
1064	{
1065	if (mbuf_alloced < pData->mbuf_water_line_limit/2)
1066	{
1067	pData->fmbuf_water_line = 0;
1068	pData->fmbuf_water_warn_sent = 0;
1069	}
1070	goto done;
1071	}
1072	#endif
1073
1074	#ifdef VBOX_WITH_SLIRP_MT
1075	if ( so->so_state & SS_NOFDREF
1076	&& so->so_deleted == 1)
1077	{
1078	struct socket son, sop = NULL;
1079	QSOCKET_LOCK(tcb);
1080	if (so->so_next != NULL)
1081	{
1082	if (so->so_next != &tcb)
1083	SOCKET_LOCK(so->so_next);
1084	son = so->so_next;
1085	}
1086	if ( so->so_prev != &tcb
1087	&& so->so_prev != NULL)
1088	{
1089	SOCKET_LOCK(so->so_prev);
1090	sop = so->so_prev;
1091	}
1092	QSOCKET_UNLOCK(tcb);
1093	remque(pData, so);
1094	NSOCK_DEC();
1095	SOCKET_UNLOCK(so);
1096	SOCKET_LOCK_DESTROY(so);
1097	RTMemFree(so);
1098	so_next = son;
1099	if (sop != NULL)
1100	SOCKET_UNLOCK(sop);
1101	CONTINUE_NO_UNLOCK(tcp);
1102	}
1103	#endif
1104	/*
1105	* FD_ISSET is meaningless on these sockets
1106	* (and they can crash the program)
1107	*/
1108	if (so->so_state & SS_NOFDREF \|\| so->s == -1)
1109	CONTINUE(tcp);
1110
1111	POLL_TCP_EVENTS(rc, error, so, &NetworkEvents);
1112
1113	LOG_NAT_SOCK(so, TCP, &NetworkEvents, readfds, writefds, xfds);
1114
1115
1116	/*
1117	* Check for URG data
1118	* This will soread as well, so no need to
1119	* test for readfds below if this succeeds
1120	*/
1121
1122	/* out-of-band data */
1123	if (CHECK_FD_SET(so, NetworkEvents, xfds))
1124	{
1125	sorecvoob(pData, so);
1126	}
1127
1128	/*
1129	* Check sockets for reading
1130	*/
1131	else if ( CHECK_FD_SET(so, NetworkEvents, readfds)
1132	\|\| WIN_CHECK_FD_SET(so, NetworkEvents, acceptds))
1133	{
1134	/*
1135	* Check for incoming connections
1136	*/
1137	if (so->so_state & SS_FACCEPTCONN)
1138	{
1139	TCP_CONNECT(pData, so);
1140	#if defined(RT_OS_WINDOWS)
1141	if (!CHECK_FD_SET(so, NetworkEvents, closefds))
1142	#endif
1143	CONTINUE(tcp);
1144	}
1145
1146	ret = soread(pData, so);
1147	/* Output it if we read something */
1148	if (RT_LIKELY(ret > 0))
1149	TCP_OUTPUT(pData, sototcpcb(so));
1150	}
1151
1152	#if defined(RT_OS_WINDOWS)
1153	/*
1154	* Check for FD_CLOSE events.
1155	* in some cases once FD_CLOSE engaged on socket it could be flashed latter (for some reasons)
1156	*/
1157	if ( CHECK_FD_SET(so, NetworkEvents, closefds)
1158	\|\| (so->so_close == 1))
1159	{
1160	so->so_close = 1; /* mark it */
1161	/*
1162	* drain the socket
1163	*/
1164	for (;;)
1165	{
1166	ret = soread(pData, so);
1167	if (ret > 0)
1168	TCP_OUTPUT(pData, sototcpcb(so));
1169	else
1170	break;
1171	}
1172	CONTINUE(tcp);
1173	}
1174	#endif
1175
1176	/*
1177	* Check sockets for writing
1178	*/
1179	if (CHECK_FD_SET(so, NetworkEvents, writefds))
1180	{
1181	/*
1182	* Check for non-blocking, still-connecting sockets
1183	*/
1184	if (so->so_state & SS_ISFCONNECTING)
1185	{
1186	Log2(("connecting %R[natsock] catched\n", so));
1187	/* Connected */
1188	so->so_state &= ~SS_ISFCONNECTING;
1189
1190	/*
1191	* This should be probably guarded by PROBE_CONN too. Anyway,
1192	* we disable it on OS/2 because the below send call returns
1193	* EFAULT which causes the opened TCP socket to close right
1194	* after it has been opened and connected.
1195	*/
1196	#ifndef RT_OS_OS2
1197	ret = send(so->s, (const char *)&ret, 0, 0);
1198	if (ret < 0)
1199	{
1200	/* XXXXX Must fix, zero bytes is a NOP */
1201	if ( errno == EAGAIN
1202	\|\| errno == EWOULDBLOCK
1203	\|\| errno == EINPROGRESS
1204	\|\| errno == ENOTCONN)
1205	CONTINUE(tcp);
1206
1207	/* else failed */
1208	so->so_state = SS_NOFDREF;
1209	}
1210	/* else so->so_state &= ~SS_ISFCONNECTING; */
1211	#endif
1212
1213	/*
1214	* Continue tcp_input
1215	*/
1216	TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1217	/* continue; */
1218	}
1219	else
1220	SOWRITE(ret, pData, so);
1221	/*
1222	* XXX If we wrote something (a lot), there could be the need
1223	* for a window update. In the worst case, the remote will send
1224	* a window probe to get things going again.
1225	*/
1226	}
1227
1228	/*
1229	* Probe a still-connecting, non-blocking socket
1230	* to check if it's still alive
1231	*/
1232	#ifdef PROBE_CONN
1233	if (so->so_state & SS_ISFCONNECTING)
1234	{
1235	ret = recv(so->s, (char *)&ret, 0, 0);
1236
1237	if (ret < 0)
1238	{
1239	/* XXX */
1240	if ( errno == EAGAIN
1241	\|\| errno == EWOULDBLOCK
1242	\|\| errno == EINPROGRESS
1243	\|\| errno == ENOTCONN)
1244	{
1245	CONTINUE(tcp); /* Still connecting, continue */
1246	}
1247
1248	/* else failed */
1249	so->so_state = SS_NOFDREF;
1250
1251	/* tcp_input will take care of it */
1252	}
1253	else
1254	{
1255	ret = send(so->s, &ret, 0, 0);
1256	if (ret < 0)
1257	{
1258	/* XXX */
1259	if ( errno == EAGAIN
1260	\|\| errno == EWOULDBLOCK
1261	\|\| errno == EINPROGRESS
1262	\|\| errno == ENOTCONN)
1263	{
1264	CONTINUE(tcp);
1265	}
1266	/* else failed */
1267	so->so_state = SS_NOFDREF;
1268	}
1269	else
1270	so->so_state &= ~SS_ISFCONNECTING;
1271
1272	}
1273	TCP_INPUT((struct mbuf *)NULL, sizeof(struct ip),so);
1274	} /* SS_ISFCONNECTING */
1275	#endif
1276	#ifndef RT_OS_WINDOWS
1277	if ( UNIX_CHECK_FD_SET(so, NetworkEvents, rdhup)
1278	\|\| UNIX_CHECK_FD_SET(so, NetworkEvents, rderr))
1279	{
1280	int err;
1281	int inq, outq;
1282	int status;
1283	socklen_t optlen = sizeof(int);
1284	inq = outq = 0;
1285	status = getsockopt(so->s, SOL_SOCKET, SO_ERROR, &err, &optlen);
1286	if (status != 0)
1287	Log(("NAT: can't get error status from %R[natsock]\n", so));
1288	#ifndef RT_OS_SOLARIS
1289	status = ioctl(so->s, FIONREAD, &inq); /* tcp(7) recommends SIOCINQ which is Linux specific */
1290	if (status != 0 \|\| status != EINVAL)
1291	{
1292	/* EINVAL returned if socket in listen state tcp(7)*/
1293	Log(("NAT: can't get depth of IN queue status from %R[natsock]\n", so));
1294	}
1295	status = ioctl(so->s, TIOCOUTQ, &outq); /* SIOCOUTQ see previous comment */
1296	if (status != 0)
1297	Log(("NAT: can't get depth of OUT queue from %R[natsock]\n", so));
1298	#else
1299	/*
1300	* Solaris has bit different ioctl commands and its handlings
1301	* hint: streamio(7) I_NREAD
1302	*/
1303	#endif
1304	if ( so->so_state & SS_ISFCONNECTING
1305	\|\| UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1306	{
1307	/**
1308	* Check if we need here take care about gracefull connection
1309	* @todo try with proxy server
1310	*/
1311	if (UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1312	{
1313	/*
1314	* Never meet inq != 0 or outq != 0, anyway let it stay for a while
1315	* in case it happens we'll able to detect it.
1316	* Give TCP/IP stack wait or expire the socket.
1317	*/
1318	Log(("NAT: %R[natsock] err(%d:%s) s(in:%d,out:%d)happens on read I/O, "
1319	"other side close connection \n", so, err, strerror(err), inq, outq));
1320	CONTINUE(tcp);
1321	}
1322	goto tcp_input_close;
1323	}
1324	if ( !UNIX_CHECK_FD_SET(so, NetworkEvents, readfds)
1325	&& !UNIX_CHECK_FD_SET(so, NetworkEvents, writefds)
1326	&& !UNIX_CHECK_FD_SET(so, NetworkEvents, xfds))
1327	{
1328	Log(("NAT: system expires the socket %R[natsock] err(%d:%s) s(in:%d,out:%d) happens on non-I/O. ",
1329	so, err, strerror(err), inq, outq));
1330	goto tcp_input_close;
1331	}
1332	Log(("NAT: %R[natsock] we've met(%d:%s) s(in:%d, out:%d) unhandled combination hup (%d) "
1333	"rederr(%d) on (r:%d, w:%d, x:%d)\n",
1334	so, err, strerror(err),
1335	inq, outq,
1336	UNIX_CHECK_FD_SET(so, ign, rdhup),
1337	UNIX_CHECK_FD_SET(so, ign, rderr),
1338	UNIX_CHECK_FD_SET(so, ign, readfds),
1339	UNIX_CHECK_FD_SET(so, ign, writefds),
1340	UNIX_CHECK_FD_SET(so, ign, xfds)));
1341	/*
1342	* Give OS's TCP/IP stack a chance to resolve an issue or expire the socket.
1343	*/
1344	CONTINUE(tcp);
1345	tcp_input_close:
1346	so->so_state = SS_NOFDREF; /cause connection valid tcp connection termination and socket closing /
1347	TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1348	CONTINUE(tcp);
1349	}
1350	#endif
1351	LOOP_LABEL(tcp, so, so_next);
1352	}
1353
1354	/*
1355	* Now UDP sockets.
1356	* Incoming packets are sent straight away, they're not buffered.
1357	* Incoming UDP data isn't buffered either.
1358	*/
1359	QSOCKET_FOREACH(so, so_next, udp)
1360	/* { */
1361	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
1362	if (pData->fmbuf_water_line == 1)
1363	{
1364	if (mbuf_alloced < pData->mbuf_water_line_limit/2)
1365	{
1366	pData->fmbuf_water_line = 0;
1367	pData->fmbuf_water_warn_sent = 0;
1368	}
1369	goto done;
1370	}
1371	#endif
1372	#ifdef VBOX_WITH_SLIRP_MT
1373	if ( so->so_state & SS_NOFDREF
1374	&& so->so_deleted == 1)
1375	{
1376	struct socket son, sop = NULL;
1377	QSOCKET_LOCK(udb);
1378	if (so->so_next != NULL)
1379	{
1380	if (so->so_next != &udb)
1381	SOCKET_LOCK(so->so_next);
1382	son = so->so_next;
1383	}
1384	if ( so->so_prev != &udb
1385	&& so->so_prev != NULL)
1386	{
1387	SOCKET_LOCK(so->so_prev);
1388	sop = so->so_prev;
1389	}
1390	QSOCKET_UNLOCK(udb);
1391	remque(pData, so);
1392	NSOCK_DEC();
1393	SOCKET_UNLOCK(so);
1394	SOCKET_LOCK_DESTROY(so);
1395	RTMemFree(so);
1396	so_next = son;
1397	if (sop != NULL)
1398	SOCKET_UNLOCK(sop);
1399	CONTINUE_NO_UNLOCK(udp);
1400	}
1401	#endif
1402	POLL_UDP_EVENTS(rc, error, so, &NetworkEvents);
1403
1404	LOG_NAT_SOCK(so, UDP, &NetworkEvents, readfds, writefds, xfds);
1405
1406	if (so->s != -1 && CHECK_FD_SET(so, NetworkEvents, readfds))
1407	{
1408	SORECVFROM(pData, so);
1409	}
1410	LOOP_LABEL(udp, so, so_next);
1411	}
1412
1413	done:
1414	#if 0
1415	/*
1416	* See if we can start outputting
1417	*/
1418	if (if_queued && link_up)
1419	if_start(pData);
1420	#endif
1421
1422	STAM_PROFILE_STOP(&pData->StatPoll, a);
1423	}
1424
1425
1426	struct arphdr
1427	{
1428	unsigned short ar_hrd; /* format of hardware address */
1429	unsigned short ar_pro; /* format of protocol address */
1430	unsigned char ar_hln; /* length of hardware address */
1431	unsigned char ar_pln; /* length of protocol address */
1432	unsigned short ar_op; /* ARP opcode (command) */
1433
1434	/*
1435	* Ethernet looks like this : This bit is variable sized however...
1436	*/
1437	unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
1438	unsigned char ar_sip[4]; /* sender IP address */
1439	unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
1440	unsigned char ar_tip[4]; /* target IP address */
1441	};
1442	AssertCompileSize(struct arphdr, 28);
1443
1444	static void arp_input(PNATState pData, struct mbuf *m)
1445	{
1446	struct ethhdr *eh;
1447	struct ethhdr *reh;
1448	struct arphdr *ah;
1449	struct arphdr *rah;
1450	int ar_op;
1451	struct ex_list *ex_ptr;
1452	uint32_t htip;
1453	uint32_t tip;
1454	struct mbuf *mr;
1455	eh = mtod(m, struct ethhdr *);
1456	ah = (struct arphdr *)&eh[1];
1457	htip = RT_N2H_U32((uint32_t)ah->ar_tip);
1458	tip = (uint32_t)ah->ar_tip;
1459
1460	ar_op = RT_N2H_U16(ah->ar_op);
1461
1462	switch (ar_op)
1463	{
1464	case ARPOP_REQUEST:
1465	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
1466	mr = m_get(pData);
1467
1468	reh = mtod(mr, struct ethhdr *);
1469	memcpy(reh->h_source, eh->h_source, ETH_ALEN); /* XXX: if_encap will swap src and dst*/
1470	Log4(("NAT: arp:%R[ether]->%R[ether]\n",
1471	reh->h_source, reh->h_dest));
1472	Log4(("NAT: arp: %R[IP4]\n", &tip));
1473
1474	mr->m_data += if_maxlinkhdr;
1475	mr->m_len = sizeof(struct arphdr);
1476	rah = mtod(mr, struct arphdr *);
1477	#else
1478	mr = m_getcl(pData, M_NOWAIT, MT_HEADER, M_PKTHDR);
1479	reh = mtod(mr, struct ethhdr *);
1480	mr->m_data += ETH_HLEN;
1481	rah = mtod(mr, struct arphdr *);
1482	mr->m_len = sizeof(struct arphdr);
1483	Assert(mr);
1484	memcpy(reh->h_source, eh->h_source, ETH_ALEN); /* XXX: if_encap will swap src and dst*/
1485	#endif
1486	#ifdef VBOX_WITH_NAT_SERVICE
1487	if (tip == pData->special_addr.s_addr)
1488	goto arp_ok;
1489	#endif
1490	if ((htip & pData->netmask) == RT_N2H_U32(pData->special_addr.s_addr))
1491	{
1492	if ( CTL_CHECK(htip, CTL_DNS)
1493	\|\| CTL_CHECK(htip, CTL_ALIAS)
1494	\|\| CTL_CHECK(htip, CTL_TFTP))
1495	goto arp_ok;
1496	for (ex_ptr = exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next)
1497	{
1498	if ((htip & ~pData->netmask) == ex_ptr->ex_addr)
1499	{
1500	goto arp_ok;
1501	}
1502	}
1503	m_free(pData, m);
1504	m_free(pData, mr);
1505	return;
1506
1507	arp_ok:
1508	rah->ar_hrd = RT_H2N_U16_C(1);
1509	rah->ar_pro = RT_H2N_U16_C(ETH_P_IP);
1510	rah->ar_hln = ETH_ALEN;
1511	rah->ar_pln = 4;
1512	rah->ar_op = RT_H2N_U16_C(ARPOP_REPLY);
1513	memcpy(rah->ar_sha, special_ethaddr, ETH_ALEN);
1514
1515	switch (htip & ~pData->netmask)
1516	{
1517	case CTL_DNS:
1518	case CTL_ALIAS:
1519	rah->ar_sha[5] = (uint8_t)(htip & ~pData->netmask);
1520	break;
1521	default:;
1522	}
1523
1524	memcpy(rah->ar_sip, ah->ar_tip, 4);
1525	memcpy(rah->ar_tha, ah->ar_sha, ETH_ALEN);
1526	memcpy(rah->ar_tip, ah->ar_sip, 4);
1527	if_encap(pData, ETH_P_ARP, mr, ETH_ENCAP_URG);
1528	m_free(pData, m);
1529	}
1530	/* Gratuitous ARP */
1531	if ( (uint32_t )ah->ar_sip == (uint32_t )ah->ar_tip
1532	&& memcmp(ah->ar_tha, broadcast_ethaddr, ETH_ALEN) == 0
1533	&& memcmp(eh->h_dest, broadcast_ethaddr, ETH_ALEN) == 0)
1534	{
1535	/* we've received anounce about address asignment
1536	* Let's do ARP cache update
1537	*/
1538	if (slirp_arp_cache_update(pData, (uint32_t )ah->ar_tip, &eh->h_dest[0]) == 0)
1539	{
1540	m_free(pData, mr);
1541	m_free(pData, m);
1542	break;
1543	}
1544	slirp_arp_cache_add(pData, (uint32_t )ah->ar_tip, &eh->h_dest[0]);
1545	}
1546	break;
1547
1548	case ARPOP_REPLY:
1549	if (slirp_arp_cache_update(pData, (uint32_t )ah->ar_sip, &ah->ar_sha[0]) == 0)
1550	{
1551	m_free(pData, m);
1552	break;
1553	}
1554	slirp_arp_cache_add(pData, (uint32_t )ah->ar_sip, ah->ar_sha);
1555	m_free(pData, m);
1556	break;
1557
1558	default:
1559	break;
1560	}
1561	}
1562
1563	/**
1564	* Feed a packet into the slirp engine.
1565	*
1566	* @param m Data buffer, m_len is not valid.
1567	* @param cbBuf The length of the data in m.
1568	*/
1569	void slirp_input(PNATState pData, struct mbuf *m, size_t cbBuf)
1570	{
1571	int proto;
1572	static bool fWarnedIpv6;
1573	struct ethhdr *eh;
1574	uint8_t au8Ether[ETH_ALEN];
1575
1576	m->m_len = cbBuf;
1577	if (cbBuf < ETH_HLEN)
1578	{
1579	LogRel(("NAT: packet having size %d has been ignored\n", m->m_len));
1580	m_free(pData, m);
1581	return;
1582	}
1583	eh = mtod(m, struct ethhdr *);
1584	proto = RT_N2H_U16(eh->h_proto);
1585
1586	memcpy(au8Ether, eh->h_source, ETH_ALEN);
1587
1588	switch(proto)
1589	{
1590	case ETH_P_ARP:
1591	arp_input(pData, m);
1592	break;
1593
1594	case ETH_P_IP:
1595	/* Update time. Important if the network is very quiet, as otherwise
1596	* the first outgoing connection gets an incorrect timestamp. */
1597	updtime(pData);
1598	m_adj(m, ETH_HLEN);
1599	#ifdef VBOX_WITH_SLIRP_BSD_MBUF
1600	M_ASSERTPKTHDR(m);
1601	m->m_pkthdr.header = mtod(m, void *);
1602	#else /* !VBOX_WITH_SLIRP_BSD_MBUF */
1603	if ( pData->fmbuf_water_line
1604	&& pData->fmbuf_water_warn_sent == 0
1605	&& (curtime - pData->tsmbuf_water_warn_sent) > 500)
1606	{
1607	icmp_error(pData, m, ICMP_SOURCEQUENCH, 0, 0, "Out of resources!!!");
1608	pData->fmbuf_water_warn_sent = 1;
1609	pData->tsmbuf_water_warn_sent = curtime;
1610	}
1611	#endif /* !VBOX_WITH_SLIRP_BSD_MBUF */
1612	ip_input(pData, m);
1613	break;
1614
1615	case ETH_P_IPV6:
1616	m_free(pData, m);
1617	if (!fWarnedIpv6)
1618	{
1619	LogRel(("NAT: IPv6 not supported\n"));
1620	fWarnedIpv6 = true;
1621	}
1622	break;
1623
1624	default:
1625	Log(("NAT: Unsupported protocol %x\n", proto));
1626	m_free(pData, m);
1627	break;
1628	}
1629
1630	if (pData->cRedirectionsActive != pData->cRedirectionsStored)
1631	activate_port_forwarding(pData, au8Ether);
1632	}
1633
1634	/* output the IP packet to the ethernet device */
1635	void if_encap(PNATState pData, uint16_t eth_proto, struct mbuf *m, int flags)
1636	{
1637	struct ethhdr *eh;
1638	uint8_t *buf = NULL;
1639	size_t mlen = 0;
1640	STAM_PROFILE_START(&pData->StatIF_encap, a);
1641
1642	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
1643	m->m_data -= if_maxlinkhdr;
1644	m->m_len += ETH_HLEN;
1645	eh = mtod(m, struct ethhdr *);
1646
1647	if (MBUF_HEAD(m) != m->m_data)
1648	{
1649	LogRel(("NAT: ethernet detects corruption of the packet"));
1650	AssertMsgFailed(("!!Ethernet frame corrupted!!"));
1651	}
1652	#else
1653	M_ASSERTPKTHDR(m);
1654	m->m_data -= ETH_HLEN;
1655	m->m_len += ETH_HLEN;
1656	eh = mtod(m, struct ethhdr *);
1657	#endif
1658
1659	if (memcmp(eh->h_source, special_ethaddr, ETH_ALEN) != 0)
1660	{
1661	memcpy(eh->h_dest, eh->h_source, ETH_ALEN);
1662	memcpy(eh->h_source, special_ethaddr, ETH_ALEN);
1663	Assert(memcmp(eh->h_dest, special_ethaddr, ETH_ALEN) != 0);
1664	if (memcmp(eh->h_dest, zerro_ethaddr, ETH_ALEN) == 0)
1665	{
1666	/* don't do anything */
1667	m_free(pData, m);
1668	goto done;
1669	}
1670	}
1671	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
1672	mlen = m->m_len;
1673	#else
1674	mlen = m_length(m, NULL);
1675	buf = RTMemAlloc(mlen);
1676	if (buf == NULL)
1677	{
1678	LogRel(("NAT: Can't alloc memory for outgoing buffer\n"));
1679	m_free(pData, m);
1680	goto done;
1681	}
1682	#endif
1683	eh->h_proto = RT_H2N_U16(eth_proto);
1684	#ifdef VBOX_WITH_SLIRP_BSD_MBUF
1685	m_copydata(m, 0, mlen, (char *)buf);
1686	if (flags & ETH_ENCAP_URG)
1687	slirp_urg_output(pData->pvUser, m, buf, mlen);
1688	else
1689	slirp_output(pData->pvUser, m, buf, mlen);
1690	#else
1691	if (flags & ETH_ENCAP_URG)
1692	slirp_urg_output(pData->pvUser, m, mtod(m, const uint8_t *), mlen);
1693	else
1694	slirp_output(pData->pvUser, m, mtod(m, const uint8_t *), mlen);
1695	#endif
1696	done:
1697	STAM_PROFILE_STOP(&pData->StatIF_encap, a);
1698	}
1699
1700	/**
1701	* Still we're using dhcp server leasing to map ether to IP
1702	* @todo see rt_lookup_in_cache
1703	*/
1704	static uint32_t find_guest_ip(PNATState pData, const uint8_t *eth_addr)
1705	{
1706	uint32_t ip = INADDR_ANY;
1707	int rc;
1708
1709	if (eth_addr == NULL)
1710	return INADDR_ANY;
1711
1712	if ( memcmp(eth_addr, zerro_ethaddr, ETH_ALEN) == 0
1713	\|\| memcmp(eth_addr, broadcast_ethaddr, ETH_ALEN) == 0)
1714	return INADDR_ANY;
1715
1716	rc = slirp_arp_lookup_ip_by_ether(pData, eth_addr, &ip);
1717	if (RT_SUCCESS(rc))
1718	return ip;
1719
1720	bootp_cache_lookup_ip_by_ether(pData, eth_addr, &ip);
1721	/* ignore return code, ip will be set to INADDR_ANY on error */
1722	return ip;
1723	}
1724
1725	/**
1726	* We need check if we've activated port forwarding
1727	* for specific machine ... that of course relates to
1728	* service mode
1729	* @todo finish this for service case
1730	*/
1731	static void activate_port_forwarding(PNATState pData, const uint8_t *h_source)
1732	{
1733	struct port_forward_rule *rule;
1734
1735	/* check mac here */
1736	LIST_FOREACH(rule, &pData->port_forward_rule_head, list)
1737	{
1738	struct socket *so;
1739	struct alias_link *alias_link;
1740	struct libalias *lib;
1741	int flags;
1742	struct sockaddr sa;
1743	struct sockaddr_in *psin;
1744	socklen_t socketlen;
1745	struct in_addr alias;
1746	int rc;
1747	uint32_t guest_addr; /* need to understand if we already give address to guest */
1748
1749	if (rule->activated)
1750	continue;
1751
1752	#ifdef VBOX_WITH_NAT_SERVICE
1753	if (memcmp(rule->mac_address, h_source, ETH_ALEN) != 0)
1754	continue; /not right mac, @todo: it'd be better do the list port forwarding per mac /
1755	guest_addr = find_guest_ip(pData, h_source);
1756	#else
1757	#if 0
1758	if (memcmp(client_ethaddr, h_source, ETH_ALEN) != 0)
1759	continue;
1760	#endif
1761	guest_addr = find_guest_ip(pData, h_source);
1762	#endif
1763	if (guest_addr == INADDR_ANY)
1764	{
1765	/* the address wasn't granted */
1766	return;
1767	}
1768
1769	#if !defined(VBOX_WITH_NAT_SERVICE)
1770	if (rule->guest_addr.s_addr != guest_addr)
1771	continue;
1772	#endif
1773
1774	LogRel(("NAT: set redirect %s host port %d => guest port %d @ %R[IP4]\n",
1775	(rule->proto == IPPROTO_UDP?"UDP":"TCP"),
1776	rule->host_port, rule->guest_port, &guest_addr));
1777
1778	if (rule->proto == IPPROTO_UDP)
1779	so = udp_listen(pData, rule->bind_ip.s_addr, RT_H2N_U16(rule->host_port), guest_addr,
1780	RT_H2N_U16(rule->guest_port), 0);
1781	else
1782	so = solisten(pData, rule->bind_ip.s_addr, RT_H2N_U16(rule->host_port), guest_addr,
1783	RT_H2N_U16(rule->guest_port), 0);
1784
1785	if (so == NULL)
1786	goto remove_port_forwarding;
1787
1788	psin = (struct sockaddr_in *)&sa;
1789	psin->sin_family = AF_INET;
1790	psin->sin_port = 0;
1791	psin->sin_addr.s_addr = INADDR_ANY;
1792	socketlen = sizeof(struct sockaddr);
1793
1794	rc = getsockname(so->s, &sa, &socketlen);
1795	if (rc < 0 \|\| sa.sa_family != AF_INET)
1796	goto remove_port_forwarding;
1797
1798	psin = (struct sockaddr_in *)&sa;
1799
1800	lib = LibAliasInit(pData, NULL);
1801	flags = LibAliasSetMode(lib, 0, 0);
1802	flags \|= PKT_ALIAS_LOG; /* set logging */
1803	flags \|= PKT_ALIAS_REVERSE; /* set logging */
1804	flags = LibAliasSetMode(lib, flags, ~0);
1805
1806	alias.s_addr = RT_H2N_U32(RT_N2H_U32(guest_addr) \| CTL_ALIAS);
1807	alias_link = LibAliasRedirectPort(lib, psin->sin_addr, RT_H2N_U16(rule->host_port),
1808	alias, RT_H2N_U16(rule->guest_port),
1809	pData->special_addr, -1, /* not very clear for now */
1810	rule->proto);
1811	if (!alias_link)
1812	goto remove_port_forwarding;
1813
1814	so->so_la = lib;
1815	rule->activated = 1;
1816	pData->cRedirectionsActive++;
1817	continue;
1818
1819	remove_port_forwarding:
1820	LogRel(("NAT: failed to redirect %s %d => %d\n",
1821	(rule->proto == IPPROTO_UDP?"UDP":"TCP"), rule->host_port, rule->guest_port));
1822	LIST_REMOVE(rule, list);
1823	pData->cRedirectionsStored--;
1824	RTMemFree(rule);
1825	}
1826	}
1827
1828	/**
1829	* Changes in 3.1 instead of opening new socket do the following:
1830	* gain more information:
1831	* 1. bind IP
1832	* 2. host port
1833	* 3. guest port
1834	* 4. proto
1835	* 5. guest MAC address
1836	* the guest's MAC address is rather important for service, but we easily
1837	* could get it from VM configuration in DrvNAT or Service, the idea is activating
1838	* corresponding port-forwarding
1839	*/
1840	int slirp_redir(PNATState pData, int is_udp, struct in_addr host_addr, int host_port,
1841	struct in_addr guest_addr, int guest_port, const uint8_t *ethaddr)
1842	{
1843	struct port_forward_rule *rule = NULL;
1844	Assert(memcmp(ethaddr, zerro_ethaddr, ETH_ALEN) == 0);
1845
1846	rule = RTMemAllocZ(sizeof(struct port_forward_rule));
1847	if (rule == NULL)
1848	return 1;
1849
1850	rule->proto = (is_udp ? IPPROTO_UDP : IPPROTO_TCP);
1851	rule->host_port = host_port;
1852	rule->guest_port = guest_port;
1853	#ifndef VBOX_WITH_NAT_SERVICE
1854	rule->guest_addr.s_addr = guest_addr.s_addr;
1855	#endif
1856	rule->bind_ip.s_addr = host_addr.s_addr;
1857	memcpy(rule->mac_address, ethaddr, ETH_ALEN);
1858	/* @todo add mac address */
1859	LIST_INSERT_HEAD(&pData->port_forward_rule_head, rule, list);
1860	pData->cRedirectionsStored++;
1861	return 0;
1862	}
1863
1864	int slirp_add_exec(PNATState pData, int do_pty, const char *args, int addr_low_byte,
1865	int guest_port)
1866	{
1867	return add_exec(&exec_list, do_pty, (char *)args,
1868	addr_low_byte, RT_H2N_U16(guest_port));
1869	}
1870
1871	void slirp_set_ethaddr_and_activate_port_forwarding(PNATState pData, const uint8_t *ethaddr, uint32_t GuestIP)
1872	{
1873	#ifndef VBOX_WITH_NAT_SERVICE
1874	memcpy(client_ethaddr, ethaddr, ETH_ALEN);
1875	#endif
1876	if (GuestIP != INADDR_ANY)
1877	{
1878	slirp_arp_cache_update_or_add(pData, GuestIP, ethaddr);
1879	activate_port_forwarding(pData, ethaddr);
1880	}
1881	}
1882
1883	#if defined(RT_OS_WINDOWS)
1884	HANDLE *slirp_get_events(PNATState pData)
1885	{
1886	return pData->phEvents;
1887	}
1888	void slirp_register_external_event(PNATState pData, HANDLE hEvent, int index)
1889	{
1890	pData->phEvents[index] = hEvent;
1891	}
1892	#endif
1893
1894	unsigned int slirp_get_timeout_ms(PNATState pData)
1895	{
1896	if (link_up)
1897	{
1898	if (time_fasttimo)
1899	return 2;
1900	if (do_slowtimo)
1901	return 500; /* see PR_SLOWHZ */
1902	}
1903	return 0;
1904	}
1905
1906	#ifndef RT_OS_WINDOWS
1907	int slirp_get_nsock(PNATState pData)
1908	{
1909	return pData->nsock;
1910	}
1911	#endif
1912
1913	/*
1914	* this function called from NAT thread
1915	*/
1916	void slirp_post_sent(PNATState pData, void *pvArg)
1917	{
1918	struct socket *so = 0;
1919	struct tcpcb *tp = 0;
1920	struct mbuf m = (struct mbuf )pvArg;
1921	m_free(pData, m);
1922	}
1923	#ifdef VBOX_WITH_SLIRP_MT
1924	void slirp_process_queue(PNATState pData)
1925	{
1926	RTReqProcess(pData->pReqQueue, RT_INDEFINITE_WAIT);
1927	}
1928	void *slirp_get_queue(PNATState pData)
1929	{
1930	return pData->pReqQueue;
1931	}
1932	#endif
1933
1934	void slirp_set_dhcp_TFTP_prefix(PNATState pData, const char *tftpPrefix)
1935	{
1936	Log2(("tftp_prefix:%s\n", tftpPrefix));
1937	tftp_prefix = tftpPrefix;
1938	}
1939
1940	void slirp_set_dhcp_TFTP_bootfile(PNATState pData, const char *bootFile)
1941	{
1942	Log2(("bootFile:%s\n", bootFile));
1943	bootp_filename = bootFile;
1944	}
1945
1946	void slirp_set_dhcp_next_server(PNATState pData, const char *next_server)
1947	{
1948	Log2(("next_server:%s\n", next_server));
1949	if (next_server == NULL)
1950	pData->tftp_server.s_addr = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_TFTP);
1951	else
1952	inet_aton(next_server, &pData->tftp_server);
1953	}
1954
1955	int slirp_set_binding_address(PNATState pData, char *addr)
1956	{
1957	if (addr == NULL \|\| (inet_aton(addr, &pData->bindIP) == 0))
1958	{
1959	pData->bindIP.s_addr = INADDR_ANY;
1960	return 1;
1961	}
1962	return 0;
1963	}
1964
1965	void slirp_set_dhcp_dns_proxy(PNATState pData, bool fDNSProxy)
1966	{
1967	if (!pData->use_host_resolver)
1968	{
1969	Log2(("NAT: DNS proxy switched %s\n", (fDNSProxy ? "on" : "off")));
1970	pData->use_dns_proxy = fDNSProxy;
1971	}
1972	else
1973	LogRel(("NAT: Host Resolver conflicts with DNS proxy, the last one was forcely ignored\n"));
1974	}
1975
1976	#define CHECK_ARG(name, val, lim_min, lim_max) \
1977	do { \
1978	if ((val) < (lim_min) \|\| (val) > (lim_max)) \
1979	{ \
1980	LogRel(("NAT: (" #name ":%d) has been ignored, " \
1981	"because out of range (%d, %d)\n", (val), (lim_min), (lim_max))); \
1982	return; \
1983	} \
1984	else \
1985	LogRel(("NAT: (" #name ":%d)\n", (val))); \
1986	} while (0)
1987
1988	/* don't allow user set less 8kB and more than 1M values */
1989	#define _8K_1M_CHECK_ARG(name, val) CHECK_ARG(name, (val), 8, 1024)
1990	void slirp_set_rcvbuf(PNATState pData, int kilobytes)
1991	{
1992	_8K_1M_CHECK_ARG("SOCKET_RCVBUF", kilobytes);
1993	pData->socket_rcv = kilobytes;
1994	}
1995	void slirp_set_sndbuf(PNATState pData, int kilobytes)
1996	{
1997	_8K_1M_CHECK_ARG("SOCKET_SNDBUF", kilobytes);
1998	pData->socket_snd = kilobytes * _1K;
1999	}
2000	void slirp_set_tcp_rcvspace(PNATState pData, int kilobytes)
2001	{
2002	_8K_1M_CHECK_ARG("TCP_RCVSPACE", kilobytes);
2003	tcp_rcvspace = kilobytes * _1K;
2004	}
2005	void slirp_set_tcp_sndspace(PNATState pData, int kilobytes)
2006	{
2007	_8K_1M_CHECK_ARG("TCP_SNDSPACE", kilobytes);
2008	tcp_sndspace = kilobytes * _1K;
2009	}
2010
2011	/*
2012	* Looking for Ether by ip in ARP-cache
2013	* Note: it´s responsible of caller to allocate buffer for result
2014	* @returns iprt status code
2015	*/
2016	int slirp_arp_lookup_ether_by_ip(PNATState pData, uint32_t ip, uint8_t *ether)
2017	{
2018	struct arp_cache_entry *ac;
2019
2020	if (ether == NULL)
2021	return VERR_INVALID_PARAMETER;
2022
2023	if (LIST_EMPTY(&pData->arp_cache))
2024	return VERR_NOT_FOUND;
2025
2026	LIST_FOREACH(ac, &pData->arp_cache, list)
2027	{
2028	if (ac->ip == ip)
2029	{
2030	memcpy(ether, ac->ether, ETH_ALEN);
2031	return VINF_SUCCESS;
2032	}
2033	}
2034	return VERR_NOT_FOUND;
2035	}
2036
2037	/*
2038	* Looking for IP by Ether in ARP-cache
2039	* Note: it´s responsible of caller to allocate buffer for result
2040	* @returns 0 - if found, 1 - otherwise
2041	*/
2042	int slirp_arp_lookup_ip_by_ether(PNATState pData, const uint8_t ether, uint32_t ip)
2043	{
2044	struct arp_cache_entry *ac;
2045	*ip = INADDR_ANY;
2046
2047	if (LIST_EMPTY(&pData->arp_cache))
2048	return VERR_NOT_FOUND;
2049
2050	LIST_FOREACH(ac, &pData->arp_cache, list)
2051	{
2052	if (memcmp(ether, ac->ether, ETH_ALEN) == 0)
2053	{
2054	*ip = ac->ip;
2055	return VINF_SUCCESS;
2056	}
2057	}
2058	return VERR_NOT_FOUND;
2059	}
2060
2061	void slirp_arp_who_has(PNATState pData, uint32_t dst)
2062	{
2063	struct mbuf *m;
2064	struct ethhdr *ehdr;
2065	struct arphdr *ahdr;
2066
2067	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
2068	m = m_get(pData);
2069	#else
2070	m = m_getcl(pData, M_NOWAIT, MT_HEADER, M_PKTHDR);
2071	#endif
2072	if (m == NULL)
2073	{
2074	LogRel(("NAT: Can't alloc mbuf for ARP request\n"));
2075	return;
2076	}
2077	ehdr = mtod(m, struct ethhdr *);
2078	memset(ehdr->h_source, 0xff, ETH_ALEN);
2079	ahdr = (struct arphdr *)&ehdr[1];
2080	ahdr->ar_hrd = RT_H2N_U16_C(1);
2081	ahdr->ar_pro = RT_H2N_U16_C(ETH_P_IP);
2082	ahdr->ar_hln = ETH_ALEN;
2083	ahdr->ar_pln = 4;
2084	ahdr->ar_op = RT_H2N_U16_C(ARPOP_REQUEST);
2085	memcpy(ahdr->ar_sha, special_ethaddr, ETH_ALEN);
2086	(uint32_t )ahdr->ar_sip = RT_H2N_U32(RT_N2H_U32(pData->special_addr.s_addr) \| CTL_ALIAS);
2087	memset(ahdr->ar_tha, 0xff, ETH_ALEN); /broadcast/
2088	(uint32_t )ahdr->ar_tip = dst;
2089	#ifndef VBOX_WITH_SLIRP_BSD_MBUF
2090	m->m_data += if_maxlinkhdr;
2091	m->m_len = sizeof(struct arphdr);
2092	#else
2093	/* warn!!! should falls in mbuf minimal size */
2094	m->m_len = sizeof(struct arphdr) + ETH_HLEN;
2095	#endif
2096	if_encap(pData, ETH_P_ARP, m, ETH_ENCAP_URG);
2097	}
2098
2099	int slirp_arp_cache_update_or_add(PNATState pData, uint32_t dst, const uint8_t *mac)
2100	{
2101	if (slirp_arp_cache_update(pData, dst, mac))
2102	slirp_arp_cache_add(pData, dst, mac);
2103
2104	return 0;
2105	}
2106
2107	/* updates the arp cache
2108	* @returns 0 - if has found and updated
2109	* 1 - if hasn't found.
2110	*/
2111	int slirp_arp_cache_update(PNATState pData, uint32_t dst, const uint8_t *mac)
2112	{
2113	struct arp_cache_entry *ac;
2114	LIST_FOREACH(ac, &pData->arp_cache, list)
2115	{
2116	if (memcmp(ac->ether, mac, ETH_ALEN) == 0)
2117	{
2118	ac->ip = dst;
2119	return 0;
2120	}
2121	}
2122	return 1;
2123	}
2124
2125	void slirp_arp_cache_add(PNATState pData, uint32_t ip, const uint8_t *ether)
2126	{
2127	struct arp_cache_entry *ac = NULL;
2128	ac = RTMemAllocZ(sizeof(struct arp_cache_entry));
2129	if (ac == NULL)
2130	{
2131	LogRel(("NAT: Can't allocate arp cache entry\n"));
2132	return;
2133	}
2134	ac->ip = ip;
2135	memcpy(ac->ether, ether, ETH_ALEN);
2136	LIST_INSERT_HEAD(&pData->arp_cache, ac, list);
2137	}
2138
2139	#ifdef VBOX_WITH_SLIRP_BSD_MBUF
2140	void slirp_set_mtu(PNATState pData, int mtu)
2141	{
2142	if (mtu < 20 \|\| mtu >= 16000)
2143	{
2144	LogRel(("NAT: mtu(%d) is out of range (20;16000] mtu forcely assigned to 1500\n", mtu));
2145	mtu = 1500;
2146	}
2147	if_mtu =
2148	if_mru = mtu;
2149	}
2150	#endif

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

source: vbox/trunk/src/VBox/Devices/Network/slirp/slirp.c@ 26598

Download in other formats: