VirtualBox

source: vbox/trunk/src/VBox/Devices/Network/slirp/slirp.c@ 17656

Last change on this file since 17656 was 17656, checked in by vboxsync, 16 years ago

NAT: disable service calculation
  • Property svn:eol-style set to native
File size: 55.4 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#ifndef RT_OS_WINDOWS
10# include <sys/ioctl.h>
11# include <poll.h>
12#else
13# define _WINSOCK2API_
14# include <IPHlpApi.h>
15#endif
16
17#if !defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) || !defined(RT_OS_WINDOWS)
18
19# ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
20# define DO_ENGAGE_EVENT1(so, fdset, label) \
21 do { \
22 FD_SET((so)->s, (fdset)); \
23 UPD_NFDS((so)->s); \
24 } while(0)
25
26
27# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
28 do { \
29 FD_SET((so)->s, (fdset1)); \
30 FD_SET((so)->s, (fdset2)); \
31 UPD_NFDS((so)->s); \
32 } while(0)
33
34# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
35
36# define DO_CHECK_FD_SET(so, events, fdset) (FD_ISSET((so)->s, fdset))
37# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) 0 /*specific for Unix API */
38# else /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
39# define DO_ENGAGE_EVENT1(so, fdset, label) \
40 do { \
41 if( so->so_poll_index != -1 \
42 && so->s == polls[so->so_poll_index].fd) { \
43 polls[so->so_poll_index].events |= N_(fdset ## _poll); \
44 break; /* out of this loop */ \
45 } \
46 AssertRelease(poll_index < (nfds)); \
47 AssertRelease(poll_index >= 0 && poll_index < (nfds)); \
48 polls[poll_index].fd = (so)->s; \
49 (so)->so_poll_index = poll_index; \
50 polls[poll_index].events = N_(fdset ## _poll); \
51 polls[poll_index].revents = 0; \
52 poll_index++; \
53 } while(0)
54
55
56# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
57 do { \
58 if( so->so_poll_index != -1 \
59 && so->s == polls[so->so_poll_index].fd) { \
60 polls[so->so_poll_index].events |= \
61 N_(fdset1 ## _poll) | N_(fdset1 ## _poll); \
62 break; /* out of this loop */ \
63 } \
64 AssertRelease(poll_index < (nfds)); \
65 polls[poll_index].fd = (so)->s; \
66 (so)->so_poll_index = poll_index; \
67 polls[poll_index].events = \
68 N_(fdset1 ## _poll) | N_(fdset1 ## _poll); \
69 poll_index++; \
70 } while(0)
71
72# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
73
74# define DO_CHECK_FD_SET(so, events, fdset) ( ((so)->so_poll_index != -1) \
75 && ((so)->so_poll_index <= ndfs) \
76 && ((so)->s == polls[so->so_poll_index].fd) \
77 && (polls[(so)->so_poll_index].revents & N_(fdset ## _poll)))
78# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset) /*specific for Unix API */
79# define DO_WIN_CHECK_FD_SET(so, events, fdset ) 0 /* specific for Windows Winsock API */
80# endif /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
81
82# ifndef RT_OS_WINDOWS
83
84# ifndef RT_OS_LINUX
85# define readfds_poll (POLLRDNORM)
86# define writefds_poll (POLLWRNORM)
87# define xfds_poll (POLLRDBAND|POLLWRBAND|POLLPRI)
88# else
89# define readfds_poll (POLLIN)
90# define writefds_poll (POLLOUT)
91# define xfds_poll (POLLPRI)
92# endif
93# define rderr_poll (POLLERR)
94# define rdhup_poll (POLLHUP)
95# define nval_poll (POLLNVAL)
96
97# define ICMP_ENGAGE_EVENT(so, fdset) \
98 do { \
99 if (pData->icmp_socket.s != -1) \
100 DO_ENGAGE_EVENT1((so), fdset, ICMP); \
101 } while (0)
102# else /* !RT_OS_WINDOWS */
103# ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
104# define DO_WIN_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset)
105# else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
106# define DO_WIN_CHECK_FD_SET(so, events, fdset ) 0
107# endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
108# define ICMP_ENGAGE_EVENT(so, fdset) do {} while(0)
109#endif /* RT_OS_WINDOWS */
110
111#else /* defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS) */
112
113/*
114 * On Windows, we will be notified by IcmpSendEcho2() when the response arrives.
115 * So no call to WSAEventSelect necessary.
116 */
117# define ICMP_ENGAGE_EVENT(so, fdset) do {} while(0)
118
119# define DO_ENGAGE_EVENT1(so, fdset1, label) \
120 do { \
121 rc = WSAEventSelect((so)->s, VBOX_SOCKET_EVENT, FD_ALL_EVENTS); \
122 if (rc == SOCKET_ERROR) \
123 { \
124 /* This should not happen */ \
125 error = WSAGetLastError(); \
126 LogRel(("WSAEventSelect (" #label ") error %d (so=%x, socket=%s, event=%x)\n", \
127 error, (so), (so)->s, VBOX_SOCKET_EVENT)); \
128 } \
129 } while(0); \
130 CONTINUE(label)
131
132# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
133 DO_ENGAGE_EVENT1((so), (fdset1), label)
134
135# define DO_POLL_EVENTS(rc, error, so, events, label) \
136 (rc) = WSAEnumNetworkEvents((so)->s, VBOX_SOCKET_EVENT, (events)); \
137 if ((rc) == SOCKET_ERROR) \
138 { \
139 (error) = WSAGetLastError(); \
140 LogRel(("WSAEnumNetworkEvents " #label " error %d\n", (error))); \
141 CONTINUE(label); \
142 }
143
144# define acceptds_win FD_ACCEPT
145# define acceptds_win_bit FD_ACCEPT_BIT
146
147# define readfds_win FD_READ
148# define readfds_win_bit FD_READ_BIT
149
150# define writefds_win FD_WRITE
151# define writefds_win_bit FD_WRITE_BIT
152
153# define xfds_win FD_OOB
154# define xfds_win_bit FD_OOB_BIT
155
156# define DO_CHECK_FD_SET(so, events, fdset) \
157 (((events).lNetworkEvents & fdset ## _win) && ((events).iErrorCode[fdset ## _win_bit] == 0))
158
159# define DO_WIN_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset)
160# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) 1 /*specific for Unix API */
161
162#endif /* defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS) */
163
164#define TCP_ENGAGE_EVENT1(so, fdset) \
165 DO_ENGAGE_EVENT1((so), fdset, tcp)
166
167#define TCP_ENGAGE_EVENT2(so, fdset1, fdset2) \
168 DO_ENGAGE_EVENT2((so), fdset1, fdset2, tcp)
169
170#define UDP_ENGAGE_EVENT(so, fdset) \
171 DO_ENGAGE_EVENT1((so), fdset, udp)
172
173#define POLL_TCP_EVENTS(rc, error, so, events) \
174 DO_POLL_EVENTS((rc), (error), (so), (events), tcp)
175
176#define POLL_UDP_EVENTS(rc, error, so, events) \
177 DO_POLL_EVENTS((rc), (error), (so), (events), udp)
178
179#define CHECK_FD_SET(so, events, set) \
180 (DO_CHECK_FD_SET((so), (events), set))
181
182#define WIN_CHECK_FD_SET(so, events, set) \
183 (DO_WIN_CHECK_FD_SET((so), (events), set))
184#define UNIX_CHECK_FD_SET(so, events, set) \
185 (DO_UNIX_CHECK_FD_SET(so, events, set))
186
187/*
188 * Loging macros
189 */
190#if VBOX_WITH_DEBUG_NAT_SOCKETS
191# if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
192# if defined(RT_OS_WINDOWS)
193# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
194 do { \
195 LogRel((" " #proto " %R[natsock] %R[natwinnetevents]\n", (so), (winevent))); \
196 } while (0)
197# else /* RT_OS_WINDOWS */
198# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
199 do { \
200 LogRel((" " #proto " %R[natsock] %s %s %s er: %s, %s, %s\n", (so), \
201 CHECK_FD_SET(so, ign ,r_fdset) ? "READ":"", \
202 CHECK_FD_SET(so, ign, w_fdset) ? "WRITE":"", \
203 CHECK_FD_SET(so, ign, x_fdset) ? "OOB":"", \
204 CHECK_FD_SET(so, ign, rderr) ? "RDERR":"", \
205 CHECK_FD_SET(so, ign, rdhup) ? "RDHUP":"", \
206 CHECK_FD_SET(so, ign, nval) ? "RDNVAL":"")); \
207 } while (0)
208# endif /* !RT_OS_WINDOWS */
209# else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
210# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
211 do { \
212 LogRel((" " #proto " %R[natsock] %s %s %s\n", (so), \
213 FD_ISSET((so)->s, (r_fdset))?"READ":"", \
214 FD_ISSET((so)->s, (w_fdset))?"WRITE":"", \
215 FD_ISSET((so)->s, (x_fdset))?"OOB":"")); \
216 } while (0)
217# endif
218#else /* VBOX_WITH_DEBUG_NAT_SOCKETS */
219# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) do {} while (0)
220#endif /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
221
222#define LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) DO_LOG_NAT_SOCK((so), proto, (winevent), r_fdset, w_fdset, x_fdset)
223
224static const uint8_t special_ethaddr[6] =
225{
226 0x52, 0x54, 0x00, 0x12, 0x35, 0x00
227};
228
229#ifdef RT_OS_WINDOWS
230# ifndef VBOX_WITH_MULTI_DNS
231static int get_dns_addr_domain(PNATState pData, bool fVerbose,
232 struct in_addr *pdns_addr,
233 const char **ppszDomain)
234{
235 int rc = 0;
236 FIXED_INFO *FixedInfo = NULL;
237 ULONG BufLen;
238 DWORD ret;
239 IP_ADDR_STRING *pIPAddr;
240 struct in_addr tmp_addr;
241
242 FixedInfo = (FIXED_INFO *)GlobalAlloc(GPTR, sizeof(FIXED_INFO));
243 BufLen = sizeof(FIXED_INFO);
244
245 /** @todo: this API returns all DNS servers, no matter whether the
246 * corresponding network adapter is disabled or not. Maybe replace
247 * this by GetAdapterAddresses(), which is XP/Vista only though. */
248 if (ERROR_BUFFER_OVERFLOW == GetNetworkParams(FixedInfo, &BufLen))
249 {
250 if (FixedInfo)
251 {
252 GlobalFree(FixedInfo);
253 FixedInfo = NULL;
254 }
255 FixedInfo = GlobalAlloc(GPTR, BufLen);
256 }
257
258 if ((ret = GetNetworkParams(FixedInfo, &BufLen)) != ERROR_SUCCESS)
259 {
260 Log(("GetNetworkParams failed. ret = %08x\n", (u_int)ret ));
261 if (FixedInfo)
262 {
263 GlobalFree(FixedInfo);
264 FixedInfo = NULL;
265 }
266 rc = -1;
267 goto get_dns_prefix;
268 }
269
270 pIPAddr = &(FixedInfo->DnsServerList);
271 inet_aton(pIPAddr->IpAddress.String, &tmp_addr);
272 Log(("nat: DNS Servers:\n"));
273 if (fVerbose || pdns_addr->s_addr != tmp_addr.s_addr)
274 LogRel(("NAT: DNS address: %s\n", pIPAddr->IpAddress.String));
275 *pdns_addr = tmp_addr;
276
277 pIPAddr = FixedInfo -> DnsServerList.Next;
278 while (pIPAddr)
279 {
280 if (fVerbose)
281 LogRel(("NAT: ignored DNS address: %s\n", pIPAddr ->IpAddress.String));
282 pIPAddr = pIPAddr ->Next;
283 }
284 if (FixedInfo)
285 {
286 GlobalFree(FixedInfo);
287 FixedInfo = NULL;
288 }
289
290get_dns_prefix:
291 if (ppszDomain)
292 {
293 OSVERSIONINFO ver;
294 char szDnsDomain[256];
295 DWORD dwSize = sizeof(szDnsDomain);
296
297 *ppszDomain = NULL;
298 GetVersionEx(&ver);
299 if (ver.dwMajorVersion >= 5)
300 {
301 /* GetComputerNameEx exists in Windows versions starting with 2000. */
302 if (GetComputerNameEx(ComputerNameDnsDomain, szDnsDomain, &dwSize))
303 {
304 if (szDnsDomain[0])
305 {
306 /* Just non-empty strings are valid. */
307 *ppszDomain = RTStrDup(szDnsDomain);
308 if (pData->fPassDomain)
309 {
310 if (fVerbose)
311 LogRel(("NAT: passing domain name %s\n", szDnsDomain));
312 }
313 else
314 Log(("nat: ignoring domain %s\n", szDnsDomain));
315 }
316 }
317 else
318 Log(("nat: GetComputerNameEx failed (%d)\n", GetLastError()));
319 }
320 }
321 return rc;
322}
323# else /* !VBOX_WITH_MULTI_DNS */
324static int get_dns_addr_domain(PNATState pData, bool fVerbose,
325 struct in_addr *pdns_addr,
326 const char **ppszDomain)
327{
328 /* Get amount of memory required for operation */
329 ULONG flags = GAA_FLAG_INCLUDE_PREFIX; /*GAA_FLAG_INCLUDE_ALL_INTERFACES;*/ /* all interfaces registered in NDIS */
330 PIP_ADAPTER_ADDRESSES addresses = NULL;
331 PIP_ADAPTER_ADDRESSES addr = NULL;
332 PIP_ADAPTER_DNS_SERVER_ADDRESS dns = NULL;
333 ULONG size = 0;
334 char *suffix;
335 struct dns_entry *da = NULL;
336 ULONG ret = ERROR_SUCCESS;
337
338 /* @todo add SKIPing flags to get only required information */
339
340 ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, addresses, &size);
341 if (ret != ERROR_BUFFER_OVERFLOW)
342 {
343 LogRel(("NAT: error %lu occured on capacity detection operation\n", ret));
344 return -1;
345 }
346
347 if (size == 0)
348 {
349 LogRel(("NAT: Win socket API returns non capacity\n"));
350 return -1;
351 }
352
353 addresses = RTMemAllocZ(size);
354 if (addresses == NULL)
355 {
356 LogRel(("NAT: No memory available \n"));
357 return -1;
358 }
359
360 ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, addresses, &size);
361 if (ret != ERROR_SUCCESS)
362 {
363 LogRel(("NAT: error %lu occured on fetching adapters info\n", ret));
364 return -1;
365 }
366 addr = addresses;
367 while(addr != NULL)
368 {
369 size_t buff_size;
370 if (addr->OperStatus != IfOperStatusUp)
371 goto next;
372 dns = addr->FirstDnsServerAddress;
373 while (dns != NULL)
374 {
375 struct sockaddr *saddr = dns->Address.lpSockaddr;
376 if (saddr->sa_family != AF_INET)
377 goto next_dns;
378 /* add dns server to list */
379 da = RTMemAllocZ(sizeof(struct dns_entry));
380 if (da == NULL)
381 {
382 LogRel(("NAT: Can't allocate buffer for DNS entry\n"));
383 return VERR_NO_MEMORY;
384 }
385 LogRel(("NAT: adding %R[IP4] to DNS server list\n", &((struct sockaddr_in *)saddr)->sin_addr));
386 if ((((struct sockaddr_in *)saddr)->sin_addr.s_addr & htonl(IN_CLASSA_NET)) == ntohl(INADDR_LOOPBACK & IN_CLASSA_NET)) {
387 da->de_addr.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_ALIAS);
388 }
389 else
390 {
391 da->de_addr.s_addr = ((struct sockaddr_in *)saddr)->sin_addr.s_addr;
392 }
393 LIST_INSERT_HEAD(&pData->dns_list_head, da, de_list);
394 next_dns:
395 dns = dns->Next;
396 }
397 buff_size = wcstombs(NULL, addr->DnsSuffix, 0);
398 if (buff_size == 0)
399 goto next;
400 suffix = RTMemAllocZ(buff_size);
401 wcstombs(suffix, addr->DnsSuffix, buff_size);
402 LogRel(("NAT: adding %s to DNS suffix list\n", suffix));
403 *ppszDomain = suffix;
404 next:
405 addr = addr->Next;
406 }
407 /*@todo add dns suffix if required */
408 LogRel(("NAT: adding dns suffix %s to the list \n", ppszDomain));
409 return 0;
410}
411# endif /* VBOX_WITH_MULTI_DNS */
412
413#else /* !RT_OS_WINDOWS */
414
415static int get_dns_addr_domain(PNATState pData, bool fVerbose,
416 struct in_addr *pdns_addr,
417 const char **ppszDomain)
418{
419 char buff[512];
420 char buff2[256];
421 FILE *f;
422 int found = 0;
423 struct in_addr tmp_addr;
424
425#ifdef RT_OS_OS2
426 /* Try various locations. */
427 char *etc = getenv("ETC");
428 f = NULL;
429 if (etc)
430 {
431 snprintf(buff, sizeof(buff), "%s/RESOLV2", etc);
432 f = fopen(buff, "rt");
433 }
434 if (!f)
435 {
436 snprintf(buff, sizeof(buff), "%s/RESOLV2", _PATH_ETC);
437 f = fopen(buff, "rt");
438 }
439 if (!f)
440 {
441 snprintf(buff, sizeof(buff), "%s/resolv.conf", _PATH_ETC);
442 f = fopen(buff, "rt");
443 }
444#else
445 f = fopen("/etc/resolv.conf", "r");
446#endif
447 if (!f)
448 return -1;
449
450 if (ppszDomain)
451 *ppszDomain = NULL;
452 Log(("nat: DNS Servers:\n"));
453 while (fgets(buff, 512, f) != NULL)
454 {
455#ifdef VBOX_WITH_MULTI_DNS
456 struct dns_entry *da = NULL;
457#endif
458 if (sscanf(buff, "nameserver%*[ \t]%256s", buff2) == 1)
459 {
460 if (!inet_aton(buff2, &tmp_addr))
461 continue;
462#ifndef VBOX_WITH_MULTI_DNS
463 /* If it's the first one, set it to dns_addr */
464 if (!found)
465 {
466 if (fVerbose || pdns_addr->s_addr != tmp_addr.s_addr)
467 LogRel(("NAT: DNS address: %s\n", buff2));
468 *pdns_addr = tmp_addr;
469 }
470 else
471 {
472 if (fVerbose)
473 LogRel(("NAT: ignored DNS address: %s\n", buff2));
474 }
475#else
476 /*localhost mask */
477 da = RTMemAllocZ(sizeof (struct dns_entry));
478 if (da == NULL)
479 {
480 LogRel(("can't alloc memory for DNS entry\n"));
481 return -1;
482 }
483 /*check */
484 da->de_addr.s_addr = tmp_addr.s_addr;
485 if ((da->de_addr.s_addr & htonl(IN_CLASSA_NET)) == ntohl(INADDR_LOOPBACK & IN_CLASSA_NET)) {
486 da->de_addr.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_ALIAS);
487 }
488 LIST_INSERT_HEAD(&pData->dns_list_head, da, de_list);
489#endif
490 found++;
491 }
492 if ( ppszDomain
493 && (!strncmp(buff, "domain", 6) || !strncmp(buff, "search", 6)))
494 {
495 /* Domain name/search list present. Pick first entry */
496 if (*ppszDomain == NULL)
497 {
498 char *tok;
499 char *saveptr;
500 tok = strtok_r(&buff[6], " \t\n", &saveptr);
501 if (tok)
502 {
503 *ppszDomain = RTStrDup(tok);
504 if (pData->fPassDomain)
505 {
506 if (fVerbose)
507 LogRel(("NAT: passing domain name %s\n", tok));
508 }
509 else
510 Log(("nat: ignoring domain %s\n", tok));
511 }
512 }
513 }
514 }
515 fclose(f);
516 if (!found)
517 return -1;
518 return 0;
519}
520
521#endif
522#ifdef VBOX_WITH_MULTI_DNS
523static int slirp_init_dns_list(PNATState pData)
524{
525 LIST_INIT(&pData->dns_list_head);
526 return get_dns_addr_domain(pData, true, NULL, &pData->pszDomain);
527}
528static void slirp_release_dns_list(PNATState pData)
529{
530 struct dns_entry *de = NULL;
531 while(!LIST_EMPTY(&pData->dns_list_head)) {
532 de = LIST_FIRST(&pData->dns_list_head);
533 LIST_REMOVE(de, de_list);
534 RTMemFree(de);
535 }
536}
537#endif
538
539int get_dns_addr(PNATState pData, struct in_addr *pdns_addr)
540{
541 return get_dns_addr_domain(pData, false, pdns_addr, NULL);
542}
543
544int slirp_init(PNATState *ppData, const char *pszNetAddr, uint32_t u32Netmask,
545 bool fPassDomain, void *pvUser)
546{
547 int fNATfailed = 0;
548 int rc;
549 PNATState pData = RTMemAllocZ(sizeof(NATState));
550 *ppData = pData;
551 if (!pData)
552 return VERR_NO_MEMORY;
553 if (u32Netmask & 0x1f)
554 /* CTL is x.x.x.15, bootp passes up to 16 IPs (15..31) */
555 return VERR_INVALID_PARAMETER;
556 pData->fPassDomain = fPassDomain;
557 pData->pvUser = pvUser;
558 pData->netmask = u32Netmask;
559
560#ifdef RT_OS_WINDOWS
561 {
562 WSADATA Data;
563 WSAStartup(MAKEWORD(2, 0), &Data);
564 }
565# if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
566 pData->phEvents[VBOX_SOCKET_EVENT_INDEX] = CreateEvent(NULL, FALSE, FALSE, NULL);
567# endif
568#endif
569#ifdef VBOX_WITH_SLIRP_MT
570 QSOCKET_LOCK_CREATE(tcb);
571 QSOCKET_LOCK_CREATE(udb);
572 rc = RTReqCreateQueue(&pData->pReqQueue);
573 AssertReleaseRC(rc);
574#endif
575
576 link_up = 1;
577
578 debug_init();
579 if_init(pData);
580 ip_init(pData);
581 icmp_init(pData);
582
583 /* Initialise mbufs *after* setting the MTU */
584 m_init(pData);
585
586 inet_aton(pszNetAddr, &special_addr);
587 alias_addr.s_addr = special_addr.s_addr | htonl(CTL_ALIAS);
588 /* @todo: add ability to configure this staff */
589
590 /* set default addresses */
591 inet_aton("127.0.0.1", &loopback_addr);
592#ifndef VBOX_WITH_MULTI_DNS
593 inet_aton("127.0.0.1", &dns_addr);
594
595 if (get_dns_addr_domain(pData, true, &dns_addr, &pData->pszDomain) < 0)
596#else
597 if (slirp_init_dns_list(pData) < 0)
598#endif
599 fNATfailed = 1;
600#ifdef VBOX_WITH_SLIRP_DNS_PROXY
601 dnsproxy_init(pData);
602#endif
603
604 getouraddr(pData);
605 return fNATfailed ? VINF_NAT_DNS : VINF_SUCCESS;
606}
607
608/**
609 * Statistics counters.
610 */
611void slirp_register_timers(PNATState pData, PPDMDRVINS pDrvIns)
612{
613#ifdef VBOX_WITH_STATISTICS
614 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatFill, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
615 STAMUNIT_TICKS_PER_CALL, "Profiling slirp fills", "/Drivers/NAT%d/Fill", pDrvIns->iInstance);
616 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatPoll, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
617 STAMUNIT_TICKS_PER_CALL, "Profiling slirp polls", "/Drivers/NAT%d/Poll", pDrvIns->iInstance);
618 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatFastTimer, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
619 STAMUNIT_TICKS_PER_CALL, "Profiling slirp fast timer", "/Drivers/NAT%d/TimerFast", pDrvIns->iInstance);
620 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatSlowTimer, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
621 STAMUNIT_TICKS_PER_CALL, "Profiling slirp slow timer", "/Drivers/NAT%d/TimerSlow", pDrvIns->iInstance);
622 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatTCP, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
623 STAMUNIT_COUNT, "TCP sockets", "/Drivers/NAT%d/SockTCP", pDrvIns->iInstance);
624 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatTCPHot, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
625 STAMUNIT_COUNT, "TCP sockets active", "/Drivers/NAT%d/SockTCPHot", pDrvIns->iInstance);
626 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatUDP, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
627 STAMUNIT_COUNT, "UDP sockets", "/Drivers/NAT%d/SockUDP", pDrvIns->iInstance);
628 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatUDPHot, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
629 STAMUNIT_COUNT, "UDP sockets active", "/Drivers/NAT%d/SockUDPHot", pDrvIns->iInstance);
630#endif /* VBOX_WITH_STATISTICS */
631}
632
633/**
634 * Marks the link as up, making it possible to establish new connections.
635 */
636void slirp_link_up(PNATState pData)
637{
638 link_up = 1;
639}
640
641/**
642 * Marks the link as down and cleans up the current connections.
643 */
644void slirp_link_down(PNATState pData)
645{
646 struct socket *so;
647
648 while ((so = tcb.so_next) != &tcb)
649 {
650 if (so->so_state & SS_NOFDREF || so->s == -1)
651 sofree(pData, so);
652 else
653 tcp_drop(pData, sototcpcb(so), 0);
654 }
655
656 while ((so = udb.so_next) != &udb)
657 udp_detach(pData, so);
658
659 link_up = 0;
660}
661
662/**
663 * Terminates the slirp component.
664 */
665void slirp_term(PNATState pData)
666{
667 if (pData->pszDomain)
668 RTStrFree((char *)(void *)pData->pszDomain);
669
670#ifdef RT_OS_WINDOWS
671 pData->pfIcmpCloseHandle(pData->icmp_socket.sh);
672 FreeLibrary(pData->hmIcmpLibrary);
673 RTMemFree(pData->pvIcmpBuffer);
674# else
675 closesocket(pData->icmp_socket.s);
676#endif
677
678 slirp_link_down(pData);
679#ifdef VBOX_WITH_MULTI_DNS
680 slirp_release_dns_list(pData);
681#endif
682#ifdef RT_OS_WINDOWS
683 WSACleanup();
684#endif
685#ifdef LOG_ENABLED
686 Log(("\n"
687 "NAT statistics\n"
688 "--------------\n"
689 "\n"));
690 ipstats(pData);
691 tcpstats(pData);
692 udpstats(pData);
693 icmpstats(pData);
694 mbufstats(pData);
695 sockstats(pData);
696 Log(("\n"
697 "\n"
698 "\n"));
699#endif
700 RTMemFree(pData);
701}
702
703
704#define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
705#define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
706#define UPD_NFDS(x) do { if (nfds < (x)) nfds = (x); } while (0)
707
708/*
709 * curtime kept to an accuracy of 1ms
710 */
711static void updtime(PNATState pData)
712{
713#ifdef RT_OS_WINDOWS
714 struct _timeb tb;
715
716 _ftime(&tb);
717 curtime = (u_int)tb.time * (u_int)1000;
718 curtime += (u_int)tb.millitm;
719#else
720 gettimeofday(&tt, 0);
721
722 curtime = (u_int)tt.tv_sec * (u_int)1000;
723 curtime += (u_int)tt.tv_usec / (u_int)1000;
724
725 if ((tt.tv_usec % 1000) >= 500)
726 curtime++;
727#endif
728}
729
730#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
731void slirp_select_fill(PNATState pData, int *pnfds,
732 fd_set *readfds, fd_set *writefds, fd_set *xfds)
733#else /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
734# ifdef RT_OS_WINDOWS
735void slirp_select_fill(PNATState pData, int *pnfds)
736# else /* RT_OS_WINDOWS */
737void slirp_select_fill(PNATState pData, int *pnfds, struct pollfd *polls)
738# endif /* !RT_OS_WINDOWS */
739#endif /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
740{
741 struct socket *so, *so_next;
742 int nfds;
743#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
744 int rc;
745 int error;
746#endif
747#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
748 int poll_index = 0;
749#endif
750 int i;
751
752 STAM_PROFILE_START(&pData->StatFill, a);
753
754 nfds = *pnfds;
755
756 /*
757 * First, TCP sockets
758 */
759 do_slowtimo = 0;
760 if (link_up)
761 {
762 /*
763 * *_slowtimo needs calling if there are IP fragments
764 * in the fragment queue, or there are TCP connections active
765 */
766 /* XXX:
767 * triggering of fragment expiration should be the same but use new macroses
768 */
769 do_slowtimo = (tcb.so_next != &tcb);
770 if (!do_slowtimo)
771 {
772 for (i = 0; i < IPREASS_NHASH; i++)
773 {
774 if (!TAILQ_EMPTY(&ipq[i]))
775 {
776 do_slowtimo = 1;
777 break;
778 }
779 }
780 }
781 ICMP_ENGAGE_EVENT(&pData->icmp_socket, readfds);
782
783 STAM_COUNTER_RESET(&pData->StatTCP);
784 STAM_COUNTER_RESET(&pData->StatTCPHot);
785
786 QSOCKET_FOREACH(so, so_next, tcp)
787 /* { */
788#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
789 so->so_poll_index = -1;
790#endif
791 STAM_COUNTER_INC(&pData->StatTCP);
792
793 /*
794 * See if we need a tcp_fasttimo
795 */
796 if ( time_fasttimo == 0
797 && so->so_tcpcb != NULL
798 && so->so_tcpcb->t_flags & TF_DELACK)
799 time_fasttimo = curtime; /* Flag when we want a fasttimo */
800
801 /*
802 * NOFDREF can include still connecting to local-host,
803 * newly socreated() sockets etc. Don't want to select these.
804 */
805 if (so->so_state & SS_NOFDREF || so->s == -1)
806 CONTINUE(tcp);
807
808 /*
809 * Set for reading sockets which are accepting
810 */
811 if (so->so_state & SS_FACCEPTCONN)
812 {
813 STAM_COUNTER_INC(&pData->StatTCPHot);
814 TCP_ENGAGE_EVENT1(so, readfds);
815 CONTINUE(tcp);
816 }
817
818 /*
819 * Set for writing sockets which are connecting
820 */
821 if (so->so_state & SS_ISFCONNECTING)
822 {
823 Log2(("connecting %R[natsock] engaged\n",so));
824 STAM_COUNTER_INC(&pData->StatTCPHot);
825 TCP_ENGAGE_EVENT1(so, writefds);
826 }
827
828 /*
829 * Set for writing if we are connected, can send more, and
830 * we have something to send
831 */
832 if (CONN_CANFSEND(so) && so->so_rcv.sb_cc)
833 {
834 STAM_COUNTER_INC(&pData->StatTCPHot);
835 TCP_ENGAGE_EVENT1(so, writefds);
836 }
837
838 /*
839 * Set for reading (and urgent data) if we are connected, can
840 * receive more, and we have room for it XXX /2 ?
841 */
842 if (CONN_CANFRCV(so) && (so->so_snd.sb_cc < (so->so_snd.sb_datalen/2)))
843 {
844 STAM_COUNTER_INC(&pData->StatTCPHot);
845 TCP_ENGAGE_EVENT2(so, readfds, xfds);
846 }
847 LOOP_LABEL(tcp, so, so_next);
848 }
849
850 /*
851 * UDP sockets
852 */
853 STAM_COUNTER_RESET(&pData->StatUDP);
854 STAM_COUNTER_RESET(&pData->StatUDPHot);
855
856 QSOCKET_FOREACH(so, so_next, udp)
857 /* { */
858
859 STAM_COUNTER_INC(&pData->StatUDP);
860#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
861 so->so_poll_index = -1;
862#endif
863
864 /*
865 * See if it's timed out
866 */
867 if (so->so_expire)
868 {
869 if (so->so_expire <= curtime)
870 {
871#ifdef VBOX_WITH_SLIRP_DNS_PROXY
872 Log2(("NAT: %R[natsock] expired\n", so));
873 if (so->so_timeout != NULL)
874 {
875 so->so_timeout(pData, so, so->so_timeout_arg);
876 }
877#endif
878#ifdef VBOX_WITH_SLIRP_MT
879 /* we need so_next for continue our cycle*/
880 so_next = so->so_next;
881#endif
882 UDP_DETACH(pData, so, so_next);
883 CONTINUE_NO_UNLOCK(udp);
884 }
885 else
886 do_slowtimo = 1; /* Let socket expire */
887 }
888
889 /*
890 * When UDP packets are received from over the link, they're
891 * sendto()'d straight away, so no need for setting for writing
892 * Limit the number of packets queued by this session to 4.
893 * Note that even though we try and limit this to 4 packets,
894 * the session could have more queued if the packets needed
895 * to be fragmented.
896 *
897 * (XXX <= 4 ?)
898 */
899 if ((so->so_state & SS_ISFCONNECTED) && so->so_queued <= 4)
900 {
901 STAM_COUNTER_INC(&pData->StatUDPHot);
902 UDP_ENGAGE_EVENT(so, readfds);
903 }
904 LOOP_LABEL(udp, so, so_next);
905 }
906
907 }
908
909#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
910# if defined(RT_OS_WINDOWS)
911 *pnfds = VBOX_EVENT_COUNT;
912# else /* RT_OS_WINDOWS */
913 AssertRelease(poll_index <= *pnfds);
914 *pnfds = poll_index;
915# endif /* !RT_OS_WINDOWS */
916#else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
917 *pnfds = nfds;
918#endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
919
920 STAM_PROFILE_STOP(&pData->StatFill, a);
921}
922
923#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
924# if defined(RT_OS_WINDOWS)
925void slirp_select_poll(PNATState pData, int fTimeout, int fIcmp)
926# else /* RT_OS_WINDOWS */
927void slirp_select_poll(PNATState pData, struct pollfd *polls, int ndfs)
928# endif /* !RT_OS_WINDOWS */
929#else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
930void slirp_select_poll(PNATState pData, fd_set *readfds, fd_set *writefds, fd_set *xfds)
931#endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
932{
933 struct socket *so, *so_next;
934 int ret;
935#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
936 WSANETWORKEVENTS NetworkEvents;
937 int rc;
938 int error;
939#endif
940#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
941 int poll_index = 0;
942#endif
943
944 STAM_PROFILE_START(&pData->StatPoll, a);
945
946 /* Update time */
947 updtime(pData);
948
949 /*
950 * See if anything has timed out
951 */
952 if (link_up)
953 {
954 if (time_fasttimo && ((curtime - time_fasttimo) >= 2))
955 {
956 STAM_PROFILE_START(&pData->StatFastTimer, a);
957 tcp_fasttimo(pData);
958 time_fasttimo = 0;
959 STAM_PROFILE_STOP(&pData->StatFastTimer, a);
960 }
961 if (do_slowtimo && ((curtime - last_slowtimo) >= 499))
962 {
963 STAM_PROFILE_START(&pData->StatSlowTimer, a);
964 ip_slowtimo(pData);
965 tcp_slowtimo(pData);
966 last_slowtimo = curtime;
967 STAM_PROFILE_STOP(&pData->StatSlowTimer, a);
968 }
969 }
970#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
971 if (fTimeout)
972 return; /* only timer update */
973#endif
974
975 /*
976 * Check sockets
977 */
978 if (link_up)
979 {
980#if defined(RT_OS_WINDOWS)
981 /*XXX: before renaming please make see define
982 * fIcmp in slirp_state.h
983 */
984 if (fIcmp)
985 sorecvfrom(pData, &pData->icmp_socket);
986#else
987 if ( (pData->icmp_socket.s != -1)
988 && CHECK_FD_SET(&pData->icmp_socket, ignored, readfds))
989 sorecvfrom(pData, &pData->icmp_socket);
990#endif
991 /*
992 * Check TCP sockets
993 */
994 QSOCKET_FOREACH(so, so_next, tcp)
995 /* { */
996
997#ifdef VBOX_WITH_SLIRP_MT
998 if ( so->so_state & SS_NOFDREF
999 && so->so_deleted == 1)
1000 {
1001 struct socket *son, *sop = NULL;
1002 QSOCKET_LOCK(tcb);
1003 if (so->so_next != NULL)
1004 {
1005 if (so->so_next != &tcb)
1006 SOCKET_LOCK(so->so_next);
1007 son = so->so_next;
1008 }
1009 if ( so->so_prev != &tcb
1010 && so->so_prev != NULL)
1011 {
1012 SOCKET_LOCK(so->so_prev);
1013 sop = so->so_prev;
1014 }
1015 QSOCKET_UNLOCK(tcb);
1016 remque(pData, so);
1017 NSOCK_DEC();
1018 SOCKET_UNLOCK(so);
1019 SOCKET_LOCK_DESTROY(so);
1020 RTMemFree(so);
1021 so_next = son;
1022 if (sop != NULL)
1023 SOCKET_UNLOCK(sop);
1024 CONTINUE_NO_UNLOCK(tcp);
1025 }
1026#endif
1027 /*
1028 * FD_ISSET is meaningless on these sockets
1029 * (and they can crash the program)
1030 */
1031 if (so->so_state & SS_NOFDREF || so->s == -1)
1032 CONTINUE(tcp);
1033
1034 POLL_TCP_EVENTS(rc, error, so, &NetworkEvents);
1035
1036 LOG_NAT_SOCK(so, TCP, &NetworkEvents, readfds, writefds, xfds);
1037
1038
1039 /*
1040 * Check for URG data
1041 * This will soread as well, so no need to
1042 * test for readfds below if this succeeds
1043 */
1044
1045 /* out-of-band data */
1046 if (CHECK_FD_SET(so, NetworkEvents, xfds))
1047 {
1048 sorecvoob(pData, so);
1049 }
1050
1051 /*
1052 * Check sockets for reading
1053 */
1054 else if ( CHECK_FD_SET(so, NetworkEvents, readfds)
1055 || WIN_CHECK_FD_SET(so, NetworkEvents, acceptds))
1056 {
1057 /*
1058 * Check for incoming connections
1059 */
1060 if (so->so_state & SS_FACCEPTCONN)
1061 {
1062 TCP_CONNECT(pData, so);
1063#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1064 if (!(NetworkEvents.lNetworkEvents & FD_CLOSE))
1065#endif
1066 CONTINUE(tcp);
1067 }
1068
1069 ret = soread(pData, so);
1070 /* Output it if we read something */
1071 if (ret > 0)
1072 TCP_OUTPUT(pData, sototcpcb(so));
1073 }
1074
1075#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1076 /*
1077 * Check for FD_CLOSE events.
1078 * in some cases once FD_CLOSE engaged on socket it could be flashed latter (for some reasons)
1079 */
1080 if ( (NetworkEvents.lNetworkEvents & FD_CLOSE)
1081 || (so->so_close == 1))
1082 {
1083 so->so_close = 1; /* mark it */
1084 /*
1085 * drain the socket
1086 */
1087 for (;;)
1088 {
1089 ret = soread(pData, so);
1090 if (ret > 0)
1091 TCP_OUTPUT(pData, sototcpcb(so));
1092 else
1093 break;
1094 }
1095 CONTINUE(tcp);
1096 }
1097#endif
1098
1099 /*
1100 * Check sockets for writing
1101 */
1102 if (CHECK_FD_SET(so, NetworkEvents, writefds))
1103 {
1104 /*
1105 * Check for non-blocking, still-connecting sockets
1106 */
1107 if (so->so_state & SS_ISFCONNECTING)
1108 {
1109 Log2(("connecting %R[natsock] catched\n", so));
1110 /* Connected */
1111 so->so_state &= ~SS_ISFCONNECTING;
1112
1113 /*
1114 * This should be probably guarded by PROBE_CONN too. Anyway,
1115 * we disable it on OS/2 because the below send call returns
1116 * EFAULT which causes the opened TCP socket to close right
1117 * after it has been opened and connected.
1118 */
1119#ifndef RT_OS_OS2
1120 ret = send(so->s, (const char *)&ret, 0, 0);
1121 if (ret < 0)
1122 {
1123 /* XXXXX Must fix, zero bytes is a NOP */
1124 if ( errno == EAGAIN
1125 || errno == EWOULDBLOCK
1126 || errno == EINPROGRESS
1127 || errno == ENOTCONN)
1128 CONTINUE(tcp);
1129
1130 /* else failed */
1131 so->so_state = SS_NOFDREF;
1132 }
1133 /* else so->so_state &= ~SS_ISFCONNECTING; */
1134#endif
1135
1136 /*
1137 * Continue tcp_input
1138 */
1139 TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1140 /* continue; */
1141 }
1142 else
1143 SOWRITE(ret, pData, so);
1144 /*
1145 * XXX If we wrote something (a lot), there could be the need
1146 * for a window update. In the worst case, the remote will send
1147 * a window probe to get things going again.
1148 */
1149 }
1150
1151 /*
1152 * Probe a still-connecting, non-blocking socket
1153 * to check if it's still alive
1154 */
1155#ifdef PROBE_CONN
1156 if (so->so_state & SS_ISFCONNECTING)
1157 {
1158 ret = recv(so->s, (char *)&ret, 0, 0);
1159
1160 if (ret < 0)
1161 {
1162 /* XXX */
1163 if ( errno == EAGAIN
1164 || errno == EWOULDBLOCK
1165 || errno == EINPROGRESS
1166 || errno == ENOTCONN)
1167 {
1168 CONTINUE(tcp); /* Still connecting, continue */
1169 }
1170
1171 /* else failed */
1172 so->so_state = SS_NOFDREF;
1173
1174 /* tcp_input will take care of it */
1175 }
1176 else
1177 {
1178 ret = send(so->s, &ret, 0, 0);
1179 if (ret < 0)
1180 {
1181 /* XXX */
1182 if ( errno == EAGAIN
1183 || errno == EWOULDBLOCK
1184 || errno == EINPROGRESS
1185 || errno == ENOTCONN)
1186 {
1187 CONTINUE(tcp);
1188 }
1189 /* else failed */
1190 so->so_state = SS_NOFDREF;
1191 }
1192 else
1193 so->so_state &= ~SS_ISFCONNECTING;
1194
1195 }
1196 TCP_INPUT((struct mbuf *)NULL, sizeof(struct ip),so);
1197 } /* SS_ISFCONNECTING */
1198#endif
1199#ifndef RT_OS_WINDOWS
1200 if ( UNIX_CHECK_FD_SET(so, NetworkEvents, rdhup)
1201 || UNIX_CHECK_FD_SET(so, NetworkEvents, rderr))
1202 {
1203 int err;
1204 int inq, outq;
1205 int status;
1206 socklen_t optlen = sizeof(int);
1207 inq = outq = 0;
1208 status = getsockopt(so->s, SOL_SOCKET, SO_ERROR, &err, &optlen);
1209 if (status != 0)
1210 Log(("NAT: can't get error status from %R[natsock]\n", so));
1211#ifndef RT_OS_SOLARIS
1212 status = ioctl(so->s, FIONREAD, &inq); /* tcp(7) recommends SIOCINQ which is Linux specific */
1213 if (status != 0 || status != EINVAL)
1214 {
1215 /* EINVAL returned if socket in listen state tcp(7)*/
1216 Log(("NAT: can't get depth of IN queue status from %R[natsock]\n", so));
1217 }
1218 status = ioctl(so->s, TIOCOUTQ, &outq); /* SIOCOUTQ see previous comment */
1219 if (status != 0)
1220 Log(("NAT: can't get depth of OUT queue from %R[natsock]\n", so));
1221#else
1222 /*
1223 * Solaris has bit different ioctl commands and its handlings
1224 * hint: streamio(7) I_NREAD
1225 */
1226#endif
1227 if ( so->so_state & SS_ISFCONNECTING
1228 || UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1229 {
1230 /**
1231 * Check if we need here take care about gracefull connection
1232 * @todo try with proxy server
1233 */
1234 if (UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1235 {
1236 /*
1237 * Never meet inq != 0 or outq != 0, anyway let it stay for a while
1238 * in case it happens we'll able to detect it.
1239 * Give TCP/IP stack wait or expire the socket.
1240 */
1241 Log(("NAT: %R[natsock] err(%d:%s) s(in:%d,out:%d)happens on read I/O, "
1242 "other side close connection \n", so, err, strerror(err), inq, outq));
1243 CONTINUE(tcp);
1244 }
1245 goto tcp_input_close;
1246 }
1247 if ( !UNIX_CHECK_FD_SET(so, NetworkEvents, readfds)
1248 && !UNIX_CHECK_FD_SET(so, NetworkEvents, writefds)
1249 && !UNIX_CHECK_FD_SET(so, NetworkEvents, xfds))
1250 {
1251 Log(("NAT: system expires the socket %R[natsock] err(%d:%s) s(in:%d,out:%d) happens on non-I/O. ",
1252 so, err, strerror(err), inq, outq));
1253 goto tcp_input_close;
1254 }
1255 Log(("NAT: %R[natsock] we've met(%d:%s) s(in:%d, out:%d) unhandled combination hup (%d) "
1256 "rederr(%d) on (r:%d, w:%d, x:%d)\n",
1257 so, err, strerror(err),
1258 inq, outq,
1259 UNIX_CHECK_FD_SET(so, ign, rdhup),
1260 UNIX_CHECK_FD_SET(so, ign, rderr),
1261 UNIX_CHECK_FD_SET(so, ign, readfds),
1262 UNIX_CHECK_FD_SET(so, ign, writefds),
1263 UNIX_CHECK_FD_SET(so, ign, xfds)));
1264 /*
1265 * Give OS's TCP/IP stack a chance to resolve an issue or expire the socket.
1266 */
1267 CONTINUE(tcp);
1268tcp_input_close:
1269 so->so_state = SS_NOFDREF; /*cause connection valid tcp connection termination and socket closing */
1270 TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1271 CONTINUE(tcp);
1272 }
1273#endif
1274 LOOP_LABEL(tcp, so, so_next);
1275 }
1276
1277 /*
1278 * Now UDP sockets.
1279 * Incoming packets are sent straight away, they're not buffered.
1280 * Incoming UDP data isn't buffered either.
1281 */
1282 QSOCKET_FOREACH(so, so_next, udp)
1283 /* { */
1284#ifdef VBOX_WITH_SLIRP_MT
1285 if ( so->so_state & SS_NOFDREF
1286 && so->so_deleted == 1)
1287 {
1288 struct socket *son, *sop = NULL;
1289 QSOCKET_LOCK(udb);
1290 if (so->so_next != NULL)
1291 {
1292 if (so->so_next != &udb)
1293 SOCKET_LOCK(so->so_next);
1294 son = so->so_next;
1295 }
1296 if ( so->so_prev != &udb
1297 && so->so_prev != NULL)
1298 {
1299 SOCKET_LOCK(so->so_prev);
1300 sop = so->so_prev;
1301 }
1302 QSOCKET_UNLOCK(udb);
1303 remque(pData, so);
1304 NSOCK_DEC();
1305 SOCKET_UNLOCK(so);
1306 SOCKET_LOCK_DESTROY(so);
1307 RTMemFree(so);
1308 so_next = son;
1309 if (sop != NULL)
1310 SOCKET_UNLOCK(sop);
1311 CONTINUE_NO_UNLOCK(udp);
1312 }
1313#endif
1314 POLL_UDP_EVENTS(rc, error, so, &NetworkEvents);
1315
1316 LOG_NAT_SOCK(so, UDP, &NetworkEvents, readfds, writefds, xfds);
1317
1318 if (so->s != -1 && CHECK_FD_SET(so, NetworkEvents, readfds))
1319 {
1320 SORECVFROM(pData, so);
1321 }
1322 LOOP_LABEL(udp, so, so_next);
1323 }
1324
1325 }
1326
1327#ifndef VBOX_WITH_SLIRP_MT
1328 /*
1329 * See if we can start outputting
1330 */
1331 if (if_queued && link_up)
1332 if_start(pData);
1333#endif
1334
1335 STAM_PROFILE_STOP(&pData->StatPoll, a);
1336}
1337
1338#define ETH_ALEN 6
1339#define ETH_HLEN 14
1340
1341#define ARPOP_REQUEST 1 /* ARP request */
1342#define ARPOP_REPLY 2 /* ARP reply */
1343
1344struct ethhdr
1345{
1346 unsigned char h_dest[ETH_ALEN]; /* destination eth addr */
1347 unsigned char h_source[ETH_ALEN]; /* source ether addr */
1348 unsigned short h_proto; /* packet type ID field */
1349};
1350AssertCompileSize(struct ethhdr, 14);
1351
1352struct arphdr
1353{
1354 unsigned short ar_hrd; /* format of hardware address */
1355 unsigned short ar_pro; /* format of protocol address */
1356 unsigned char ar_hln; /* length of hardware address */
1357 unsigned char ar_pln; /* length of protocol address */
1358 unsigned short ar_op; /* ARP opcode (command) */
1359
1360 /*
1361 * Ethernet looks like this : This bit is variable sized however...
1362 */
1363 unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
1364 unsigned char ar_sip[4]; /* sender IP address */
1365 unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
1366 unsigned char ar_tip[4]; /* target IP address */
1367};
1368AssertCompileSize(struct arphdr, 28);
1369
1370#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1371static void arp_input(PNATState pData, struct mbuf *m)
1372#else
1373static void arp_input(PNATState pData, const uint8_t *pkt, int pkt_len)
1374#endif
1375{
1376 struct ethhdr *eh;
1377 struct ethhdr *reh;
1378 struct arphdr *ah;
1379 struct arphdr *rah;
1380 int ar_op;
1381 struct ex_list *ex_ptr;
1382 uint32_t htip;
1383#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1384 uint8_t arp_reply[sizeof(struct arphdr) + ETH_HLEN];
1385 eh = (struct ethhdr *)pkt;
1386#else
1387 struct mbuf *mr;
1388 eh = mtod(m, struct ethhdr *);
1389#endif
1390 ah = (struct arphdr *)&eh[1];
1391 htip = ntohl(*(uint32_t*)ah->ar_tip);
1392
1393#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1394 mr = m_get(pData);
1395 mr->m_data += if_maxlinkhdr;
1396 mr->m_len = sizeof(struct arphdr);
1397 rah = mtod(mr, struct arphdr *);
1398#else
1399 reh = (struct ethhdr *)arp_reply;
1400 rah = (struct arphdr *)&reh[1];
1401#endif
1402
1403 ar_op = ntohs(ah->ar_op);
1404 switch(ar_op)
1405 {
1406 case ARPOP_REQUEST:
1407 if ((htip & pData->netmask) == ntohl(special_addr.s_addr))
1408 {
1409 if ( CTL_CHECK(htip, CTL_DNS)
1410 || CTL_CHECK(htip, CTL_ALIAS)
1411 || CTL_CHECK(htip, CTL_TFTP))
1412 goto arp_ok;
1413 for (ex_ptr = exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next)
1414 {
1415 if ((htip & ~pData->netmask) == ex_ptr->ex_addr)
1416 goto arp_ok;
1417 }
1418 return;
1419 arp_ok:
1420
1421#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1422 memcpy(reh->h_dest, eh->h_source, ETH_ALEN);
1423 memcpy(reh->h_source, &special_addr, ETH_ALEN);
1424 reh->h_source[5] = ah->ar_tip[3];
1425 reh->h_proto = htons(ETH_P_ARP);
1426#endif
1427 rah->ar_hrd = htons(1);
1428 rah->ar_pro = htons(ETH_P_IP);
1429 rah->ar_hln = ETH_ALEN;
1430 rah->ar_pln = 4;
1431 rah->ar_op = htons(ARPOP_REPLY);
1432 memcpy(rah->ar_sha, special_ethaddr, ETH_ALEN);
1433
1434 switch (htip & ~pData->netmask)
1435 {
1436 case CTL_DNS:
1437 case CTL_ALIAS:
1438 rah->ar_sha[5] = (uint8_t)(htip & ~pData->netmask);
1439 break;
1440 default:;
1441 }
1442
1443 memcpy(rah->ar_sip, ah->ar_tip, 4);
1444 memcpy(rah->ar_tha, ah->ar_sha, ETH_ALEN);
1445 memcpy(rah->ar_tip, ah->ar_sip, 4);
1446#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1447 if_encap(pData, ETH_P_ARP, mr);
1448 m_free(pData, m);
1449#else
1450 slirp_output(pData->pvUser, arp_reply, sizeof(arp_reply));
1451#endif
1452 }
1453 break;
1454 default:
1455 break;
1456 }
1457}
1458
1459void slirp_input(PNATState pData, const uint8_t *pkt, int pkt_len)
1460{
1461 struct mbuf *m;
1462 int proto;
1463 static bool fWarnedIpv6;
1464
1465 if (pkt_len < ETH_HLEN)
1466 {
1467 LogRel(("NAT: packet having size %d has been ingnored\n", pkt_len));
1468 return;
1469 }
1470
1471 m = m_get(pData);
1472 if (!m)
1473 {
1474 LogRel(("NAT: can't allocate new mbuf\n"));
1475 return;
1476 }
1477
1478 /* Note: we add to align the IP header */
1479
1480 if (M_FREEROOM(m) < pkt_len)
1481 m_inc(m, pkt_len);
1482
1483 m->m_len = pkt_len ;
1484 memcpy(m->m_data, pkt, pkt_len);
1485
1486 proto = ntohs(*(uint16_t *)(pkt + 12));
1487 switch(proto)
1488 {
1489 case ETH_P_ARP:
1490#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1491 arp_input(pData, m);
1492#else
1493 arp_input(pData, pkt, pkt_len);
1494 m_free(pData, m);
1495#endif
1496 break;
1497 case ETH_P_IP:
1498 /* Update time. Important if the network is very quiet, as otherwise
1499 * the first outgoing connection gets an incorrect timestamp. */
1500 updtime(pData);
1501 m->m_data += ETH_HLEN;
1502 m->m_len -= ETH_HLEN;
1503 ip_input(pData, m);
1504 break;
1505 case ETH_P_IPV6:
1506 m_free(pData, m);
1507 if (!fWarnedIpv6)
1508 {
1509 LogRel(("NAT: IPv6 not supported\n"));
1510 fWarnedIpv6 = true;
1511 }
1512 break;
1513 default:
1514 LogRel(("NAT: Unsupported protocol %x\n", proto));
1515 m_free(pData, m);
1516 break;
1517 }
1518 RTMemFree((void *)pkt);
1519}
1520
1521/* output the IP packet to the ethernet device */
1522#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1523void if_encap(PNATState pData, uint16_t eth_proto, struct mbuf *m)
1524#else
1525void if_encap(PNATState pData, uint8_t *ip_data, int ip_data_len)
1526#endif
1527{
1528#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1529 struct ethhdr *eh;
1530 uint8_t *buf = RTMemAlloc(1600);
1531 m->m_data -= if_maxlinkhdr;
1532 m->m_len += ETH_HLEN;
1533 eh = mtod(m, struct ethhdr *);
1534#else
1535 uint8_t buf[1600];
1536 struct ethhdr *eh = (struct ethhdr *)buf;
1537
1538 if (ip_data_len + ETH_HLEN > sizeof(buf))
1539 return;
1540
1541 memcpy(buf + sizeof(struct ethhdr), ip_data, ip_data_len);
1542#endif
1543
1544
1545 memcpy(eh->h_dest, client_ethaddr, ETH_ALEN);
1546 memcpy(eh->h_source, special_ethaddr, ETH_ALEN - 1);
1547 /* XXX: not correct */
1548 eh->h_source[5] = CTL_ALIAS;
1549#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1550 eh->h_proto = htons(eth_proto);
1551#if 0
1552 slirp_output(pData->pvUser, m, mtod(m, uint8_t *), m->m_len);
1553#else
1554 memcpy(buf, mtod(m, uint8_t *), m->m_len);
1555 slirp_output(pData->pvUser, NULL, buf, m->m_len);
1556 m_free(pData, m);
1557#endif
1558#else
1559 eh->h_proto = htons(ETH_P_IP);
1560 slirp_output(pData->pvUser, buf, ip_data_len + ETH_HLEN);
1561#endif
1562}
1563
1564int slirp_redir(PNATState pData, int is_udp, int host_port,
1565 struct in_addr guest_addr, int guest_port)
1566{
1567 struct socket *so;
1568 Log2(("NAT: set redirect %s hp:%d gp:%d\n", (is_udp?"UDP":"TCP"), host_port, guest_port));
1569 if (is_udp)
1570 {
1571 so = udp_listen(pData, htons(host_port), guest_addr.s_addr,
1572 htons(guest_port), 0);
1573 }
1574 else
1575 {
1576 so = solisten(pData, htons(host_port), guest_addr.s_addr,
1577 htons(guest_port), 0);
1578 }
1579 Log2(("NAT: redirecting socket %R[natsock]\n", so));
1580 return (so != NULL ? 0 : -1);
1581}
1582
1583int slirp_add_exec(PNATState pData, int do_pty, const char *args, int addr_low_byte,
1584 int guest_port)
1585{
1586 return add_exec(&exec_list, do_pty, (char *)args,
1587 addr_low_byte, htons(guest_port));
1588}
1589
1590void slirp_set_ethaddr(PNATState pData, const uint8_t *ethaddr)
1591{
1592 memcpy(client_ethaddr, ethaddr, ETH_ALEN);
1593}
1594
1595#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1596HANDLE *slirp_get_events(PNATState pData)
1597{
1598 return pData->phEvents;
1599}
1600void slirp_register_external_event(PNATState pData, HANDLE hEvent, int index)
1601{
1602 pData->phEvents[index] = hEvent;
1603}
1604#endif
1605
1606unsigned int slirp_get_timeout_ms(PNATState pData)
1607{
1608 if (link_up)
1609 {
1610 if (time_fasttimo)
1611 return 2;
1612 if (do_slowtimo)
1613 return 500; /* see PR_SLOWHZ */
1614 }
1615 return 0;
1616}
1617
1618#ifndef RT_OS_WINDOWS
1619int slirp_get_nsock(PNATState pData)
1620{
1621 return pData->nsock;
1622}
1623#endif
1624
1625/*
1626 * this function called from NAT thread
1627 */
1628void slirp_post_sent(PNATState pData, void *pvArg)
1629{
1630 struct socket *so = 0;
1631 struct tcpcb *tp = 0;
1632 struct mbuf *m = (struct mbuf *)pvArg;
1633 m_free(pData, m);
1634}
1635#ifdef VBOX_WITH_SLIRP_MT
1636void slirp_process_queue(PNATState pData)
1637{
1638 RTReqProcess(pData->pReqQueue, RT_INDEFINITE_WAIT);
1639}
1640void *slirp_get_queue(PNATState pData)
1641{
1642 return pData->pReqQueue;
1643}
1644#endif
1645
1646uint16_t slirp_get_service(int proto, uint16_t dport, uint16_t sport)
1647{
1648 uint16_t hdport, hsport, service;
1649 hdport = ntohs(dport);
1650 hsport = ntohs(sport);
1651 Log2(("proto: %d, dport: %d sport: %d\n", proto, hdport, hsport));
1652 service = 0;
1653#if 0
1654 /* Always return 0 here */
1655 switch (hdport)
1656 {
1657 case 500:
1658 /* service = sport; */
1659 break;
1660 }
1661#endif
1662 Log2(("service : %d\n", service));
1663 return htons(service);
1664}
1665
1666void slirp_set_dhcp_TFTP_prefix(PNATState pData, const char *tftpPrefix)
1667{
1668 Log2(("tftp_prefix:%s\n", tftpPrefix));
1669 tftp_prefix = tftpPrefix;
1670}
1671
1672void slirp_set_dhcp_TFTP_bootfile(PNATState pData, const char *bootFile)
1673{
1674 Log2(("bootFile:%s\n", bootFile));
1675 bootp_filename = bootFile;
1676}
1677
1678void slirp_set_dhcp_next_server(PNATState pData, const char *next_server)
1679{
1680 Log2(("next_server:%s\n", next_server));
1681 if (next_server == NULL)
1682 pData->tftp_server.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_TFTP);
1683 else
1684 inet_aton(next_server, &pData->tftp_server);
1685}
1686#ifdef VBOX_WITH_SLIRP_DNS_PROXY
1687void slirp_set_dhcp_dns_proxy(PNATState pData, bool fDNSProxy)
1688{
1689 pData->use_dns_proxy = fDNSProxy;
1690}
1691#endif
Note: See TracBrowser for help on using the repository browser.

© 2024 Oracle Support Privacy / Do Not Sell My Info Terms of Use Trademark Policy Automated Access Etiquette