VirtualBox

source: vbox/trunk/src/VBox/Devices/Network/slirp/slirp.c@ 18584

Last change on this file since 18584 was 18584, checked in by vboxsync, 16 years ago

NAT: adding name domains improved.
multibytes handling in IPRT style
memory leak seems to be illuminated
  • Property svn:eol-style set to native
File size: 57.6 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# include <Winnls.h>
14# define _WINSOCK2API_
15# include <IPHlpApi.h>
16#endif
17
18#if !defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) || !defined(RT_OS_WINDOWS)
19
20# ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
21# define DO_ENGAGE_EVENT1(so, fdset, label) \
22 do { \
23 FD_SET((so)->s, (fdset)); \
24 UPD_NFDS((so)->s); \
25 } while(0)
26
27
28# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
29 do { \
30 FD_SET((so)->s, (fdset1)); \
31 FD_SET((so)->s, (fdset2)); \
32 UPD_NFDS((so)->s); \
33 } while(0)
34
35# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
36
37# define DO_CHECK_FD_SET(so, events, fdset) (FD_ISSET((so)->s, fdset))
38# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) 0 /*specific for Unix API */
39# else /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
40# define DO_ENGAGE_EVENT1(so, fdset, label) \
41 do { \
42 if( so->so_poll_index != -1 \
43 && so->s == polls[so->so_poll_index].fd) { \
44 polls[so->so_poll_index].events |= N_(fdset ## _poll); \
45 break; /* out of this loop */ \
46 } \
47 AssertRelease(poll_index < (nfds)); \
48 AssertRelease(poll_index >= 0 && poll_index < (nfds)); \
49 polls[poll_index].fd = (so)->s; \
50 (so)->so_poll_index = poll_index; \
51 polls[poll_index].events = N_(fdset ## _poll); \
52 polls[poll_index].revents = 0; \
53 poll_index++; \
54 } while(0)
55
56
57# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
58 do { \
59 if( so->so_poll_index != -1 \
60 && so->s == polls[so->so_poll_index].fd) { \
61 polls[so->so_poll_index].events |= \
62 N_(fdset1 ## _poll) | N_(fdset1 ## _poll); \
63 break; /* out of this loop */ \
64 } \
65 AssertRelease(poll_index < (nfds)); \
66 polls[poll_index].fd = (so)->s; \
67 (so)->so_poll_index = poll_index; \
68 polls[poll_index].events = \
69 N_(fdset1 ## _poll) | N_(fdset1 ## _poll); \
70 poll_index++; \
71 } while(0)
72
73# define DO_POLL_EVENTS(rc, error, so, events, label) do {} while (0)
74
75# define DO_CHECK_FD_SET(so, events, fdset) ( ((so)->so_poll_index != -1) \
76 && ((so)->so_poll_index <= ndfs) \
77 && ((so)->s == polls[so->so_poll_index].fd) \
78 && (polls[(so)->so_poll_index].revents & N_(fdset ## _poll)))
79# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset) /*specific for Unix API */
80# define DO_WIN_CHECK_FD_SET(so, events, fdset ) 0 /* specific for Windows Winsock API */
81# endif /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
82
83# ifndef RT_OS_WINDOWS
84
85# ifndef RT_OS_LINUX
86# define readfds_poll (POLLRDNORM)
87# define writefds_poll (POLLWRNORM)
88# define xfds_poll (POLLRDBAND|POLLWRBAND|POLLPRI)
89# else
90# define readfds_poll (POLLIN)
91# define writefds_poll (POLLOUT)
92# define xfds_poll (POLLPRI)
93# endif
94# define rderr_poll (POLLERR)
95# define rdhup_poll (POLLHUP)
96# define nval_poll (POLLNVAL)
97
98# define ICMP_ENGAGE_EVENT(so, fdset) \
99 do { \
100 if (pData->icmp_socket.s != -1) \
101 DO_ENGAGE_EVENT1((so), fdset, ICMP); \
102 } while (0)
103# else /* !RT_OS_WINDOWS */
104# ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
105# define DO_WIN_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset)
106# else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
107# define DO_WIN_CHECK_FD_SET(so, events, fdset ) 0
108# endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
109# define ICMP_ENGAGE_EVENT(so, fdset) do {} while(0)
110#endif /* RT_OS_WINDOWS */
111
112#else /* defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS) */
113
114/*
115 * On Windows, we will be notified by IcmpSendEcho2() when the response arrives.
116 * So no call to WSAEventSelect necessary.
117 */
118# define ICMP_ENGAGE_EVENT(so, fdset) do {} while(0)
119
120# define DO_ENGAGE_EVENT1(so, fdset1, label) \
121 do { \
122 rc = WSAEventSelect((so)->s, VBOX_SOCKET_EVENT, FD_ALL_EVENTS); \
123 if (rc == SOCKET_ERROR) \
124 { \
125 /* This should not happen */ \
126 error = WSAGetLastError(); \
127 LogRel(("WSAEventSelect (" #label ") error %d (so=%x, socket=%s, event=%x)\n", \
128 error, (so), (so)->s, VBOX_SOCKET_EVENT)); \
129 } \
130 } while(0); \
131 CONTINUE(label)
132
133# define DO_ENGAGE_EVENT2(so, fdset1, fdset2, label) \
134 DO_ENGAGE_EVENT1((so), (fdset1), label)
135
136# define DO_POLL_EVENTS(rc, error, so, events, label) \
137 (rc) = WSAEnumNetworkEvents((so)->s, VBOX_SOCKET_EVENT, (events)); \
138 if ((rc) == SOCKET_ERROR) \
139 { \
140 (error) = WSAGetLastError(); \
141 LogRel(("WSAEnumNetworkEvents " #label " error %d\n", (error))); \
142 CONTINUE(label); \
143 }
144
145# define acceptds_win FD_ACCEPT
146# define acceptds_win_bit FD_ACCEPT_BIT
147
148# define readfds_win FD_READ
149# define readfds_win_bit FD_READ_BIT
150
151# define writefds_win FD_WRITE
152# define writefds_win_bit FD_WRITE_BIT
153
154# define xfds_win FD_OOB
155# define xfds_win_bit FD_OOB_BIT
156
157# define DO_CHECK_FD_SET(so, events, fdset) \
158 (((events).lNetworkEvents & fdset ## _win) && ((events).iErrorCode[fdset ## _win_bit] == 0))
159
160# define DO_WIN_CHECK_FD_SET(so, events, fdset ) DO_CHECK_FD_SET((so), (events), fdset)
161# define DO_UNIX_CHECK_FD_SET(so, events, fdset ) 1 /*specific for Unix API */
162
163#endif /* defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS) */
164
165#define TCP_ENGAGE_EVENT1(so, fdset) \
166 DO_ENGAGE_EVENT1((so), fdset, tcp)
167
168#define TCP_ENGAGE_EVENT2(so, fdset1, fdset2) \
169 DO_ENGAGE_EVENT2((so), fdset1, fdset2, tcp)
170
171#define UDP_ENGAGE_EVENT(so, fdset) \
172 DO_ENGAGE_EVENT1((so), fdset, udp)
173
174#define POLL_TCP_EVENTS(rc, error, so, events) \
175 DO_POLL_EVENTS((rc), (error), (so), (events), tcp)
176
177#define POLL_UDP_EVENTS(rc, error, so, events) \
178 DO_POLL_EVENTS((rc), (error), (so), (events), udp)
179
180#define CHECK_FD_SET(so, events, set) \
181 (DO_CHECK_FD_SET((so), (events), set))
182
183#define WIN_CHECK_FD_SET(so, events, set) \
184 (DO_WIN_CHECK_FD_SET((so), (events), set))
185#define UNIX_CHECK_FD_SET(so, events, set) \
186 (DO_UNIX_CHECK_FD_SET(so, events, set))
187
188/*
189 * Loging macros
190 */
191#if VBOX_WITH_DEBUG_NAT_SOCKETS
192# if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
193# if defined(RT_OS_WINDOWS)
194# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
195 do { \
196 LogRel((" " #proto " %R[natsock] %R[natwinnetevents]\n", (so), (winevent))); \
197 } while (0)
198# else /* RT_OS_WINDOWS */
199# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
200 do { \
201 LogRel((" " #proto " %R[natsock] %s %s %s er: %s, %s, %s\n", (so), \
202 CHECK_FD_SET(so, ign ,r_fdset) ? "READ":"", \
203 CHECK_FD_SET(so, ign, w_fdset) ? "WRITE":"", \
204 CHECK_FD_SET(so, ign, x_fdset) ? "OOB":"", \
205 CHECK_FD_SET(so, ign, rderr) ? "RDERR":"", \
206 CHECK_FD_SET(so, ign, rdhup) ? "RDHUP":"", \
207 CHECK_FD_SET(so, ign, nval) ? "RDNVAL":"")); \
208 } while (0)
209# endif /* !RT_OS_WINDOWS */
210# else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
211# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) \
212 do { \
213 LogRel((" " #proto " %R[natsock] %s %s %s\n", (so), \
214 FD_ISSET((so)->s, (r_fdset))?"READ":"", \
215 FD_ISSET((so)->s, (w_fdset))?"WRITE":"", \
216 FD_ISSET((so)->s, (x_fdset))?"OOB":"")); \
217 } while (0)
218# endif
219#else /* VBOX_WITH_DEBUG_NAT_SOCKETS */
220# define DO_LOG_NAT_SOCK(so, proto, winevent, r_fdset, w_fdset, x_fdset) do {} while (0)
221#endif /* !VBOX_WITH_DEBUG_NAT_SOCKETS */
222
223#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)
224
225static const uint8_t special_ethaddr[6] =
226{
227 0x52, 0x54, 0x00, 0x12, 0x35, 0x00
228};
229
230#ifdef RT_OS_WINDOWS
231# ifndef VBOX_WITH_MULTI_DNS
232static int get_dns_addr_domain(PNATState pData, bool fVerbose,
233 struct in_addr *pdns_addr,
234 const char **ppszDomain)
235{
236 int rc = 0;
237 FIXED_INFO *FixedInfo = NULL;
238 ULONG BufLen;
239 DWORD ret;
240 IP_ADDR_STRING *pIPAddr;
241 struct in_addr tmp_addr;
242
243 FixedInfo = (FIXED_INFO *)GlobalAlloc(GPTR, sizeof(FIXED_INFO));
244 BufLen = sizeof(FIXED_INFO);
245
246 /** @todo: this API returns all DNS servers, no matter whether the
247 * corresponding network adapter is disabled or not. Maybe replace
248 * this by GetAdapterAddresses(), which is XP/Vista only though. */
249 if (ERROR_BUFFER_OVERFLOW == GetNetworkParams(FixedInfo, &BufLen))
250 {
251 if (FixedInfo)
252 {
253 GlobalFree(FixedInfo);
254 FixedInfo = NULL;
255 }
256 FixedInfo = GlobalAlloc(GPTR, BufLen);
257 }
258
259 if ((ret = GetNetworkParams(FixedInfo, &BufLen)) != ERROR_SUCCESS)
260 {
261 Log(("GetNetworkParams failed. ret = %08x\n", (u_int)ret ));
262 if (FixedInfo)
263 {
264 GlobalFree(FixedInfo);
265 FixedInfo = NULL;
266 }
267 rc = -1;
268 goto get_dns_prefix;
269 }
270
271 pIPAddr = &(FixedInfo->DnsServerList);
272 inet_aton(pIPAddr->IpAddress.String, &tmp_addr);
273 Log(("nat: DNS Servers:\n"));
274 if (fVerbose || pdns_addr->s_addr != tmp_addr.s_addr)
275 LogRel(("NAT: DNS address: %s\n", pIPAddr->IpAddress.String));
276 *pdns_addr = tmp_addr;
277
278 pIPAddr = FixedInfo -> DnsServerList.Next;
279 while (pIPAddr)
280 {
281 if (fVerbose)
282 LogRel(("NAT: ignored DNS address: %s\n", pIPAddr ->IpAddress.String));
283 pIPAddr = pIPAddr ->Next;
284 }
285 if (FixedInfo)
286 {
287 GlobalFree(FixedInfo);
288 FixedInfo = NULL;
289 }
290
291get_dns_prefix:
292 if (ppszDomain)
293 {
294 OSVERSIONINFO ver;
295 char szDnsDomain[256];
296 DWORD dwSize = sizeof(szDnsDomain);
297
298 *ppszDomain = NULL;
299 GetVersionEx(&ver);
300 if (ver.dwMajorVersion >= 5)
301 {
302 /* GetComputerNameEx exists in Windows versions starting with 2000. */
303 if (GetComputerNameEx(ComputerNameDnsDomain, szDnsDomain, &dwSize))
304 {
305 if (szDnsDomain[0])
306 {
307 /* Just non-empty strings are valid. */
308 *ppszDomain = RTStrDup(szDnsDomain);
309 if (pData->fPassDomain)
310 {
311 if (fVerbose)
312 LogRel(("NAT: passing domain name %s\n", szDnsDomain));
313 }
314 else
315 Log(("nat: ignoring domain %s\n", szDnsDomain));
316 }
317 }
318 else
319 Log(("nat: GetComputerNameEx failed (%d)\n", GetLastError()));
320 }
321 }
322 return rc;
323}
324# else /* !VBOX_WITH_MULTI_DNS */
325static int get_dns_addr_domain(PNATState pData, bool fVerbose,
326 struct in_addr *pdns_addr,
327 const char **ppszDomain)
328{
329 /* Get amount of memory required for operation */
330 ULONG flags = GAA_FLAG_INCLUDE_PREFIX; /*GAA_FLAG_INCLUDE_ALL_INTERFACES;*/ /* all interfaces registered in NDIS */
331 PIP_ADAPTER_ADDRESSES addresses = NULL;
332 PIP_ADAPTER_ADDRESSES addr = NULL;
333 PIP_ADAPTER_DNS_SERVER_ADDRESS dns = NULL;
334 ULONG size = 0;
335 int wlen = 0;
336 char *suffix;
337 struct dns_entry *da = NULL;
338 struct dns_domain_entry *dd = NULL;
339 ULONG ret = ERROR_SUCCESS;
340
341 /* @todo add SKIPing flags to get only required information */
342
343 ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, addresses, &size);
344 if (ret != ERROR_BUFFER_OVERFLOW)
345 {
346 LogRel(("NAT: error %lu occured on capacity detection operation\n", ret));
347 return -1;
348 }
349
350 if (size == 0)
351 {
352 LogRel(("NAT: Win socket API returns non capacity\n"));
353 return -1;
354 }
355
356 addresses = RTMemAllocZ(size);
357 if (addresses == NULL)
358 {
359 LogRel(("NAT: No memory available \n"));
360 return -1;
361 }
362
363 ret = pData->pfGetAdaptersAddresses(AF_INET, 0, NULL /* reserved */, addresses, &size);
364 if (ret != ERROR_SUCCESS)
365 {
366 LogRel(("NAT: error %lu occured on fetching adapters info\n", ret));
367 return -1;
368 }
369 addr = addresses;
370 while(addr != NULL)
371 {
372 int found;
373 if (addr->OperStatus != IfOperStatusUp)
374 goto next;
375 dns = addr->FirstDnsServerAddress;
376 while (dns != NULL)
377 {
378 struct sockaddr *saddr = dns->Address.lpSockaddr;
379 if (saddr->sa_family != AF_INET)
380 goto next_dns;
381 /* add dns server to list */
382 da = RTMemAllocZ(sizeof(struct dns_entry));
383 if (da == NULL)
384 {
385 LogRel(("NAT: Can't allocate buffer for DNS entry\n"));
386 return VERR_NO_MEMORY;
387 }
388 LogRel(("NAT: adding %R[IP4] to DNS server list\n", &((struct sockaddr_in *)saddr)->sin_addr));
389 if ((((struct sockaddr_in *)saddr)->sin_addr.s_addr & htonl(IN_CLASSA_NET)) == ntohl(INADDR_LOOPBACK & IN_CLASSA_NET)) {
390 da->de_addr.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_ALIAS);
391 }
392 else
393 {
394 da->de_addr.s_addr = ((struct sockaddr_in *)saddr)->sin_addr.s_addr;
395 }
396 LIST_INSERT_HEAD(&pData->dns_list_head, da, de_list);
397
398 if (addr->DnsSuffix == NULL)
399 goto next_dns;
400
401 /*uniq*/
402 RTUtf16ToUtf8(addr->DnsSuffix, &suffix);
403 found = 0;
404 LIST_FOREACH(dd, &pData->dns_domain_list_head, dd_list)
405 {
406 if ( dd->dd_pszDomain != NULL
407 && strcmp(dd->dd_pszDomain, suffix) == 0)
408 {
409 found = 1;
410 break;
411 }
412 }
413 if (found == 0)
414 {
415 dd = RTMemAllocZ(sizeof(struct dns_domain_entry));
416 if (dd == NULL)
417 {
418 LogRel(("NAT: not enought memory\n"));
419 return VERR_NO_MEMORY;
420 }
421 dd->dd_pszDomain = suffix;
422 LogRel(("NAT: adding domain name %s to search list\n", dd->dd_pszDomain));
423 LIST_INSERT_HEAD(&pData->dns_domain_list_head, dd, dd_list);
424 }
425 next_dns:
426 dns = dns->Next;
427 }
428 next:
429 addr = addr->Next;
430 }
431 return 0;
432}
433# endif /* VBOX_WITH_MULTI_DNS */
434
435#else /* !RT_OS_WINDOWS */
436
437static int get_dns_addr_domain(PNATState pData, bool fVerbose,
438 struct in_addr *pdns_addr,
439 const char **ppszDomain)
440{
441 char buff[512];
442 char buff2[256];
443 FILE *f;
444 int found = 0;
445 struct in_addr tmp_addr;
446
447#ifdef RT_OS_OS2
448 /* Try various locations. */
449 char *etc = getenv("ETC");
450 f = NULL;
451 if (etc)
452 {
453 snprintf(buff, sizeof(buff), "%s/RESOLV2", etc);
454 f = fopen(buff, "rt");
455 }
456 if (!f)
457 {
458 snprintf(buff, sizeof(buff), "%s/RESOLV2", _PATH_ETC);
459 f = fopen(buff, "rt");
460 }
461 if (!f)
462 {
463 snprintf(buff, sizeof(buff), "%s/resolv.conf", _PATH_ETC);
464 f = fopen(buff, "rt");
465 }
466#else
467 f = fopen("/etc/resolv.conf", "r");
468#endif
469 if (!f)
470 return -1;
471
472 if (ppszDomain)
473 *ppszDomain = NULL;
474 Log(("nat: DNS Servers:\n"));
475 while (fgets(buff, 512, f) != NULL)
476 {
477#ifdef VBOX_WITH_MULTI_DNS
478 struct dns_entry *da = NULL;
479#endif
480 if (sscanf(buff, "nameserver%*[ \t]%256s", buff2) == 1)
481 {
482 if (!inet_aton(buff2, &tmp_addr))
483 continue;
484#ifndef VBOX_WITH_MULTI_DNS
485 /* If it's the first one, set it to dns_addr */
486 if (!found)
487 {
488 if (fVerbose || pdns_addr->s_addr != tmp_addr.s_addr)
489 LogRel(("NAT: DNS address: %s\n", buff2));
490 *pdns_addr = tmp_addr;
491 }
492 else
493 {
494 if (fVerbose)
495 LogRel(("NAT: ignored DNS address: %s\n", buff2));
496 }
497#else
498 /*localhost mask */
499 da = RTMemAllocZ(sizeof (struct dns_entry));
500 if (da == NULL)
501 {
502 LogRel(("can't alloc memory for DNS entry\n"));
503 return -1;
504 }
505 /*check */
506 da->de_addr.s_addr = tmp_addr.s_addr;
507 if ((da->de_addr.s_addr & htonl(IN_CLASSA_NET)) == ntohl(INADDR_LOOPBACK & IN_CLASSA_NET)) {
508 da->de_addr.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_ALIAS);
509 }
510 LIST_INSERT_HEAD(&pData->dns_list_head, da, de_list);
511#endif
512 found++;
513 }
514#ifndef VBOX_WITH_MULTI_DNS
515 if ( ppszDomain
516 && (!strncmp(buff, "domain", 6) || !strncmp(buff, "search", 6)))
517 {
518 /* Domain name/search list present. Pick first entry */
519 if (*ppszDomain == NULL)
520 {
521 char *tok;
522 char *saveptr;
523 tok = strtok_r(&buff[6], " \t\n", &saveptr);
524 if (tok)
525 {
526 *ppszDomain = RTStrDup(tok);
527 if (pData->fPassDomain)
528 {
529 if (fVerbose)
530 LogRel(("NAT: passing domain name %s\n", tok));
531 }
532 else
533 Log(("nat: ignoring domain %s\n", tok));
534 }
535 }
536 }
537#else
538 if ((!strncmp(buff, "domain", 6) || !strncmp(buff, "search", 6)))
539 {
540 char *tok;
541 char *saveptr;
542 struct dns_domain_entry *dd = NULL;
543 int found = 0;
544 tok = strtok_r(&buff[6], " \t\n", &saveptr);
545 LIST_FOREACH(dd, &pData->dns_domain_list_head, dd_list)
546 {
547 if( tok != NULL
548 && strcmp(tok, dd->dd_pszDomain) == 0)
549 {
550 found = 1;
551 break;
552 }
553 }
554 if (tok != NULL && found == 0) {
555 dd = RTMemAllocZ(sizeof(struct dns_domain_entry));
556 if (dd == NULL)
557 {
558 LogRel(("NAT: not enought memory to add domain list\n"));
559 return VERR_NO_MEMORY;
560 }
561 dd->dd_pszDomain = RTStrDup(tok);
562 LogRel(("NAT: adding domain name %s to search list\n", dd->dd_pszDomain));
563 LIST_INSERT_HEAD(&pData->dns_domain_list_head, dd, dd_list);
564 }
565 }
566#endif
567 }
568 fclose(f);
569 if (!found)
570 return -1;
571 return 0;
572}
573
574#endif
575#ifdef VBOX_WITH_MULTI_DNS
576static int slirp_init_dns_list(PNATState pData)
577{
578 LIST_INIT(&pData->dns_list_head);
579 LIST_INIT(&pData->dns_domain_list_head);
580 return get_dns_addr_domain(pData, true, NULL, NULL);
581}
582
583static void slirp_release_dns_list(PNATState pData)
584{
585 struct dns_entry *de = NULL;
586 struct dns_domain_entry *dd = NULL;
587 while(!LIST_EMPTY(&pData->dns_domain_list_head)) {
588 dd = LIST_FIRST(&pData->dns_domain_list_head);
589 LIST_REMOVE(dd, dd_list);
590 if (dd->dd_pszDomain != NULL)
591 RTStrFree(dd->dd_pszDomain);
592 RTMemFree(dd);
593 }
594 while(!LIST_EMPTY(&pData->dns_domain_list_head)) {
595 dd = LIST_FIRST(&pData->dns_domain_list_head);
596 LIST_REMOVE(dd, dd_list);
597 if (dd->dd_pszDomain != NULL)
598 RTStrFree(dd->dd_pszDomain);
599 RTMemFree(dd);
600 }
601}
602#endif
603
604int get_dns_addr(PNATState pData, struct in_addr *pdns_addr)
605{
606 return get_dns_addr_domain(pData, false, pdns_addr, NULL);
607}
608
609int slirp_init(PNATState *ppData, const char *pszNetAddr, uint32_t u32Netmask,
610 bool fPassDomain, void *pvUser)
611{
612 int fNATfailed = 0;
613 int rc;
614 PNATState pData = RTMemAllocZ(sizeof(NATState));
615 *ppData = pData;
616 if (!pData)
617 return VERR_NO_MEMORY;
618 if (u32Netmask & 0x1f)
619 /* CTL is x.x.x.15, bootp passes up to 16 IPs (15..31) */
620 return VERR_INVALID_PARAMETER;
621 pData->fPassDomain = fPassDomain;
622 pData->pvUser = pvUser;
623 pData->netmask = u32Netmask;
624
625#ifdef RT_OS_WINDOWS
626 {
627 WSADATA Data;
628 WSAStartup(MAKEWORD(2, 0), &Data);
629 }
630# if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
631 pData->phEvents[VBOX_SOCKET_EVENT_INDEX] = CreateEvent(NULL, FALSE, FALSE, NULL);
632# endif
633#endif
634#ifdef VBOX_WITH_SLIRP_MT
635 QSOCKET_LOCK_CREATE(tcb);
636 QSOCKET_LOCK_CREATE(udb);
637 rc = RTReqCreateQueue(&pData->pReqQueue);
638 AssertReleaseRC(rc);
639#endif
640
641 link_up = 1;
642
643 debug_init();
644 if_init(pData);
645 ip_init(pData);
646 icmp_init(pData);
647
648 /* Initialise mbufs *after* setting the MTU */
649 m_init(pData);
650
651 inet_aton(pszNetAddr, &special_addr);
652 alias_addr.s_addr = special_addr.s_addr | htonl(CTL_ALIAS);
653 /* @todo: add ability to configure this staff */
654
655 /* set default addresses */
656 inet_aton("127.0.0.1", &loopback_addr);
657#ifndef VBOX_WITH_MULTI_DNS
658 inet_aton("127.0.0.1", &dns_addr);
659
660 if (get_dns_addr_domain(pData, true, &dns_addr, &pData->pszDomain) < 0)
661#else
662 if (slirp_init_dns_list(pData) < 0)
663#endif
664 fNATfailed = 1;
665#ifdef VBOX_WITH_SLIRP_DNS_PROXY
666 dnsproxy_init(pData);
667#endif
668
669 getouraddr(pData);
670 return fNATfailed ? VINF_NAT_DNS : VINF_SUCCESS;
671}
672
673/**
674 * Statistics counters.
675 */
676void slirp_register_timers(PNATState pData, PPDMDRVINS pDrvIns)
677{
678#ifdef VBOX_WITH_STATISTICS
679 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatFill, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
680 STAMUNIT_TICKS_PER_CALL, "Profiling slirp fills", "/Drivers/NAT%d/Fill", pDrvIns->iInstance);
681 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatPoll, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
682 STAMUNIT_TICKS_PER_CALL, "Profiling slirp polls", "/Drivers/NAT%d/Poll", pDrvIns->iInstance);
683 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatFastTimer, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
684 STAMUNIT_TICKS_PER_CALL, "Profiling slirp fast timer", "/Drivers/NAT%d/TimerFast", pDrvIns->iInstance);
685 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatSlowTimer, STAMTYPE_PROFILE, STAMVISIBILITY_ALWAYS,
686 STAMUNIT_TICKS_PER_CALL, "Profiling slirp slow timer", "/Drivers/NAT%d/TimerSlow", pDrvIns->iInstance);
687 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatTCP, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
688 STAMUNIT_COUNT, "TCP sockets", "/Drivers/NAT%d/SockTCP", pDrvIns->iInstance);
689 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatTCPHot, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
690 STAMUNIT_COUNT, "TCP sockets active", "/Drivers/NAT%d/SockTCPHot", pDrvIns->iInstance);
691 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatUDP, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
692 STAMUNIT_COUNT, "UDP sockets", "/Drivers/NAT%d/SockUDP", pDrvIns->iInstance);
693 PDMDrvHlpSTAMRegisterF(pDrvIns, &pData->StatUDPHot, STAMTYPE_COUNTER, STAMVISIBILITY_ALWAYS,
694 STAMUNIT_COUNT, "UDP sockets active", "/Drivers/NAT%d/SockUDPHot", pDrvIns->iInstance);
695#endif /* VBOX_WITH_STATISTICS */
696}
697
698/**
699 * Marks the link as up, making it possible to establish new connections.
700 */
701void slirp_link_up(PNATState pData)
702{
703 link_up = 1;
704}
705
706/**
707 * Marks the link as down and cleans up the current connections.
708 */
709void slirp_link_down(PNATState pData)
710{
711 struct socket *so;
712
713 while ((so = tcb.so_next) != &tcb)
714 {
715 if (so->so_state & SS_NOFDREF || so->s == -1)
716 sofree(pData, so);
717 else
718 tcp_drop(pData, sototcpcb(so), 0);
719 }
720
721 while ((so = udb.so_next) != &udb)
722 udp_detach(pData, so);
723
724 link_up = 0;
725}
726
727/**
728 * Terminates the slirp component.
729 */
730void slirp_term(PNATState pData)
731{
732#ifndef VBOX_WITH_MULTI_DNS
733 if (pData->pszDomain)
734 RTStrFree((char *)(void *)pData->pszDomain);
735#endif
736
737#ifdef RT_OS_WINDOWS
738 pData->pfIcmpCloseHandle(pData->icmp_socket.sh);
739 FreeLibrary(pData->hmIcmpLibrary);
740 RTMemFree(pData->pvIcmpBuffer);
741# else
742 closesocket(pData->icmp_socket.s);
743#endif
744
745 slirp_link_down(pData);
746#ifdef VBOX_WITH_MULTI_DNS
747 slirp_release_dns_list(pData);
748#endif
749#ifdef RT_OS_WINDOWS
750 WSACleanup();
751#endif
752#ifdef LOG_ENABLED
753 Log(("\n"
754 "NAT statistics\n"
755 "--------------\n"
756 "\n"));
757 ipstats(pData);
758 tcpstats(pData);
759 udpstats(pData);
760 icmpstats(pData);
761 mbufstats(pData);
762 sockstats(pData);
763 Log(("\n"
764 "\n"
765 "\n"));
766#endif
767 RTMemFree(pData);
768}
769
770
771#define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
772#define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
773#define UPD_NFDS(x) do { if (nfds < (x)) nfds = (x); } while (0)
774
775/*
776 * curtime kept to an accuracy of 1ms
777 */
778static void updtime(PNATState pData)
779{
780#ifdef RT_OS_WINDOWS
781 struct _timeb tb;
782
783 _ftime(&tb);
784 curtime = (u_int)tb.time * (u_int)1000;
785 curtime += (u_int)tb.millitm;
786#else
787 gettimeofday(&tt, 0);
788
789 curtime = (u_int)tt.tv_sec * (u_int)1000;
790 curtime += (u_int)tt.tv_usec / (u_int)1000;
791
792 if ((tt.tv_usec % 1000) >= 500)
793 curtime++;
794#endif
795}
796
797#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
798void slirp_select_fill(PNATState pData, int *pnfds,
799 fd_set *readfds, fd_set *writefds, fd_set *xfds)
800#else /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
801# ifdef RT_OS_WINDOWS
802void slirp_select_fill(PNATState pData, int *pnfds)
803# else /* RT_OS_WINDOWS */
804void slirp_select_fill(PNATState pData, int *pnfds, struct pollfd *polls)
805# endif /* !RT_OS_WINDOWS */
806#endif /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
807{
808 struct socket *so, *so_next;
809 int nfds;
810#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
811 int rc;
812 int error;
813#endif
814#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
815 int poll_index = 0;
816#endif
817 int i;
818
819 STAM_PROFILE_START(&pData->StatFill, a);
820
821 nfds = *pnfds;
822
823 /*
824 * First, TCP sockets
825 */
826 do_slowtimo = 0;
827 if (link_up)
828 {
829 /*
830 * *_slowtimo needs calling if there are IP fragments
831 * in the fragment queue, or there are TCP connections active
832 */
833 /* XXX:
834 * triggering of fragment expiration should be the same but use new macroses
835 */
836 do_slowtimo = (tcb.so_next != &tcb);
837 if (!do_slowtimo)
838 {
839 for (i = 0; i < IPREASS_NHASH; i++)
840 {
841 if (!TAILQ_EMPTY(&ipq[i]))
842 {
843 do_slowtimo = 1;
844 break;
845 }
846 }
847 }
848 ICMP_ENGAGE_EVENT(&pData->icmp_socket, readfds);
849
850 STAM_COUNTER_RESET(&pData->StatTCP);
851 STAM_COUNTER_RESET(&pData->StatTCPHot);
852
853 QSOCKET_FOREACH(so, so_next, tcp)
854 /* { */
855#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
856 so->so_poll_index = -1;
857#endif
858 STAM_COUNTER_INC(&pData->StatTCP);
859
860 /*
861 * See if we need a tcp_fasttimo
862 */
863 if ( time_fasttimo == 0
864 && so->so_tcpcb != NULL
865 && so->so_tcpcb->t_flags & TF_DELACK)
866 time_fasttimo = curtime; /* Flag when we want a fasttimo */
867
868 /*
869 * NOFDREF can include still connecting to local-host,
870 * newly socreated() sockets etc. Don't want to select these.
871 */
872 if (so->so_state & SS_NOFDREF || so->s == -1)
873 CONTINUE(tcp);
874
875 /*
876 * Set for reading sockets which are accepting
877 */
878 if (so->so_state & SS_FACCEPTCONN)
879 {
880 STAM_COUNTER_INC(&pData->StatTCPHot);
881 TCP_ENGAGE_EVENT1(so, readfds);
882 CONTINUE(tcp);
883 }
884
885 /*
886 * Set for writing sockets which are connecting
887 */
888 if (so->so_state & SS_ISFCONNECTING)
889 {
890 Log2(("connecting %R[natsock] engaged\n",so));
891 STAM_COUNTER_INC(&pData->StatTCPHot);
892 TCP_ENGAGE_EVENT1(so, writefds);
893 }
894
895 /*
896 * Set for writing if we are connected, can send more, and
897 * we have something to send
898 */
899 if (CONN_CANFSEND(so) && so->so_rcv.sb_cc)
900 {
901 STAM_COUNTER_INC(&pData->StatTCPHot);
902 TCP_ENGAGE_EVENT1(so, writefds);
903 }
904
905 /*
906 * Set for reading (and urgent data) if we are connected, can
907 * receive more, and we have room for it XXX /2 ?
908 */
909 if (CONN_CANFRCV(so) && (so->so_snd.sb_cc < (so->so_snd.sb_datalen/2)))
910 {
911 STAM_COUNTER_INC(&pData->StatTCPHot);
912 TCP_ENGAGE_EVENT2(so, readfds, xfds);
913 }
914 LOOP_LABEL(tcp, so, so_next);
915 }
916
917 /*
918 * UDP sockets
919 */
920 STAM_COUNTER_RESET(&pData->StatUDP);
921 STAM_COUNTER_RESET(&pData->StatUDPHot);
922
923 QSOCKET_FOREACH(so, so_next, udp)
924 /* { */
925
926 STAM_COUNTER_INC(&pData->StatUDP);
927#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
928 so->so_poll_index = -1;
929#endif
930
931 /*
932 * See if it's timed out
933 */
934 if (so->so_expire)
935 {
936 if (so->so_expire <= curtime)
937 {
938#ifdef VBOX_WITH_SLIRP_DNS_PROXY
939 Log2(("NAT: %R[natsock] expired\n", so));
940 if (so->so_timeout != NULL)
941 {
942 so->so_timeout(pData, so, so->so_timeout_arg);
943 }
944#endif
945#ifdef VBOX_WITH_SLIRP_MT
946 /* we need so_next for continue our cycle*/
947 so_next = so->so_next;
948#endif
949 UDP_DETACH(pData, so, so_next);
950 CONTINUE_NO_UNLOCK(udp);
951 }
952 else
953 do_slowtimo = 1; /* Let socket expire */
954 }
955
956 /*
957 * When UDP packets are received from over the link, they're
958 * sendto()'d straight away, so no need for setting for writing
959 * Limit the number of packets queued by this session to 4.
960 * Note that even though we try and limit this to 4 packets,
961 * the session could have more queued if the packets needed
962 * to be fragmented.
963 *
964 * (XXX <= 4 ?)
965 */
966 if ((so->so_state & SS_ISFCONNECTED) && so->so_queued <= 4)
967 {
968 STAM_COUNTER_INC(&pData->StatUDPHot);
969 UDP_ENGAGE_EVENT(so, readfds);
970 }
971 LOOP_LABEL(udp, so, so_next);
972 }
973
974 }
975
976#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
977# if defined(RT_OS_WINDOWS)
978 *pnfds = VBOX_EVENT_COUNT;
979# else /* RT_OS_WINDOWS */
980 AssertRelease(poll_index <= *pnfds);
981 *pnfds = poll_index;
982# endif /* !RT_OS_WINDOWS */
983#else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
984 *pnfds = nfds;
985#endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
986
987 STAM_PROFILE_STOP(&pData->StatFill, a);
988}
989
990#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC)
991# if defined(RT_OS_WINDOWS)
992void slirp_select_poll(PNATState pData, int fTimeout, int fIcmp)
993# else /* RT_OS_WINDOWS */
994void slirp_select_poll(PNATState pData, struct pollfd *polls, int ndfs)
995# endif /* !RT_OS_WINDOWS */
996#else /* VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
997void slirp_select_poll(PNATState pData, fd_set *readfds, fd_set *writefds, fd_set *xfds)
998#endif /* !VBOX_WITH_SIMPLIFIED_SLIRP_SYNC */
999{
1000 struct socket *so, *so_next;
1001 int ret;
1002#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1003 WSANETWORKEVENTS NetworkEvents;
1004 int rc;
1005 int error;
1006#endif
1007#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && !defined(RT_OS_WINDOWS)
1008 int poll_index = 0;
1009#endif
1010
1011 STAM_PROFILE_START(&pData->StatPoll, a);
1012
1013 /* Update time */
1014 updtime(pData);
1015
1016 /*
1017 * See if anything has timed out
1018 */
1019 if (link_up)
1020 {
1021 if (time_fasttimo && ((curtime - time_fasttimo) >= 2))
1022 {
1023 STAM_PROFILE_START(&pData->StatFastTimer, a);
1024 tcp_fasttimo(pData);
1025 time_fasttimo = 0;
1026 STAM_PROFILE_STOP(&pData->StatFastTimer, a);
1027 }
1028 if (do_slowtimo && ((curtime - last_slowtimo) >= 499))
1029 {
1030 STAM_PROFILE_START(&pData->StatSlowTimer, a);
1031 ip_slowtimo(pData);
1032 tcp_slowtimo(pData);
1033 last_slowtimo = curtime;
1034 STAM_PROFILE_STOP(&pData->StatSlowTimer, a);
1035 }
1036 }
1037#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1038 if (fTimeout)
1039 return; /* only timer update */
1040#endif
1041
1042 /*
1043 * Check sockets
1044 */
1045 if (link_up)
1046 {
1047#if defined(RT_OS_WINDOWS)
1048 /*XXX: before renaming please make see define
1049 * fIcmp in slirp_state.h
1050 */
1051 if (fIcmp)
1052 sorecvfrom(pData, &pData->icmp_socket);
1053#else
1054 if ( (pData->icmp_socket.s != -1)
1055 && CHECK_FD_SET(&pData->icmp_socket, ignored, readfds))
1056 sorecvfrom(pData, &pData->icmp_socket);
1057#endif
1058 /*
1059 * Check TCP sockets
1060 */
1061 QSOCKET_FOREACH(so, so_next, tcp)
1062 /* { */
1063
1064#ifdef VBOX_WITH_SLIRP_MT
1065 if ( so->so_state & SS_NOFDREF
1066 && so->so_deleted == 1)
1067 {
1068 struct socket *son, *sop = NULL;
1069 QSOCKET_LOCK(tcb);
1070 if (so->so_next != NULL)
1071 {
1072 if (so->so_next != &tcb)
1073 SOCKET_LOCK(so->so_next);
1074 son = so->so_next;
1075 }
1076 if ( so->so_prev != &tcb
1077 && so->so_prev != NULL)
1078 {
1079 SOCKET_LOCK(so->so_prev);
1080 sop = so->so_prev;
1081 }
1082 QSOCKET_UNLOCK(tcb);
1083 remque(pData, so);
1084 NSOCK_DEC();
1085 SOCKET_UNLOCK(so);
1086 SOCKET_LOCK_DESTROY(so);
1087 RTMemFree(so);
1088 so_next = son;
1089 if (sop != NULL)
1090 SOCKET_UNLOCK(sop);
1091 CONTINUE_NO_UNLOCK(tcp);
1092 }
1093#endif
1094 /*
1095 * FD_ISSET is meaningless on these sockets
1096 * (and they can crash the program)
1097 */
1098 if (so->so_state & SS_NOFDREF || so->s == -1)
1099 CONTINUE(tcp);
1100
1101 POLL_TCP_EVENTS(rc, error, so, &NetworkEvents);
1102
1103 LOG_NAT_SOCK(so, TCP, &NetworkEvents, readfds, writefds, xfds);
1104
1105
1106 /*
1107 * Check for URG data
1108 * This will soread as well, so no need to
1109 * test for readfds below if this succeeds
1110 */
1111
1112 /* out-of-band data */
1113 if (CHECK_FD_SET(so, NetworkEvents, xfds))
1114 {
1115 sorecvoob(pData, so);
1116 }
1117
1118 /*
1119 * Check sockets for reading
1120 */
1121 else if ( CHECK_FD_SET(so, NetworkEvents, readfds)
1122 || WIN_CHECK_FD_SET(so, NetworkEvents, acceptds))
1123 {
1124 /*
1125 * Check for incoming connections
1126 */
1127 if (so->so_state & SS_FACCEPTCONN)
1128 {
1129 TCP_CONNECT(pData, so);
1130#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1131 if (!(NetworkEvents.lNetworkEvents & FD_CLOSE))
1132#endif
1133 CONTINUE(tcp);
1134 }
1135
1136 ret = soread(pData, so);
1137 /* Output it if we read something */
1138 if (ret > 0)
1139 TCP_OUTPUT(pData, sototcpcb(so));
1140 }
1141
1142#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1143 /*
1144 * Check for FD_CLOSE events.
1145 * in some cases once FD_CLOSE engaged on socket it could be flashed latter (for some reasons)
1146 */
1147 if ( (NetworkEvents.lNetworkEvents & FD_CLOSE)
1148 || (so->so_close == 1))
1149 {
1150 so->so_close = 1; /* mark it */
1151 /*
1152 * drain the socket
1153 */
1154 for (;;)
1155 {
1156 ret = soread(pData, so);
1157 if (ret > 0)
1158 TCP_OUTPUT(pData, sototcpcb(so));
1159 else
1160 break;
1161 }
1162 CONTINUE(tcp);
1163 }
1164#endif
1165
1166 /*
1167 * Check sockets for writing
1168 */
1169 if (CHECK_FD_SET(so, NetworkEvents, writefds))
1170 {
1171 /*
1172 * Check for non-blocking, still-connecting sockets
1173 */
1174 if (so->so_state & SS_ISFCONNECTING)
1175 {
1176 Log2(("connecting %R[natsock] catched\n", so));
1177 /* Connected */
1178 so->so_state &= ~SS_ISFCONNECTING;
1179
1180 /*
1181 * This should be probably guarded by PROBE_CONN too. Anyway,
1182 * we disable it on OS/2 because the below send call returns
1183 * EFAULT which causes the opened TCP socket to close right
1184 * after it has been opened and connected.
1185 */
1186#ifndef RT_OS_OS2
1187 ret = send(so->s, (const char *)&ret, 0, 0);
1188 if (ret < 0)
1189 {
1190 /* XXXXX Must fix, zero bytes is a NOP */
1191 if ( errno == EAGAIN
1192 || errno == EWOULDBLOCK
1193 || errno == EINPROGRESS
1194 || errno == ENOTCONN)
1195 CONTINUE(tcp);
1196
1197 /* else failed */
1198 so->so_state = SS_NOFDREF;
1199 }
1200 /* else so->so_state &= ~SS_ISFCONNECTING; */
1201#endif
1202
1203 /*
1204 * Continue tcp_input
1205 */
1206 TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1207 /* continue; */
1208 }
1209 else
1210 SOWRITE(ret, pData, so);
1211 /*
1212 * XXX If we wrote something (a lot), there could be the need
1213 * for a window update. In the worst case, the remote will send
1214 * a window probe to get things going again.
1215 */
1216 }
1217
1218 /*
1219 * Probe a still-connecting, non-blocking socket
1220 * to check if it's still alive
1221 */
1222#ifdef PROBE_CONN
1223 if (so->so_state & SS_ISFCONNECTING)
1224 {
1225 ret = recv(so->s, (char *)&ret, 0, 0);
1226
1227 if (ret < 0)
1228 {
1229 /* XXX */
1230 if ( errno == EAGAIN
1231 || errno == EWOULDBLOCK
1232 || errno == EINPROGRESS
1233 || errno == ENOTCONN)
1234 {
1235 CONTINUE(tcp); /* Still connecting, continue */
1236 }
1237
1238 /* else failed */
1239 so->so_state = SS_NOFDREF;
1240
1241 /* tcp_input will take care of it */
1242 }
1243 else
1244 {
1245 ret = send(so->s, &ret, 0, 0);
1246 if (ret < 0)
1247 {
1248 /* XXX */
1249 if ( errno == EAGAIN
1250 || errno == EWOULDBLOCK
1251 || errno == EINPROGRESS
1252 || errno == ENOTCONN)
1253 {
1254 CONTINUE(tcp);
1255 }
1256 /* else failed */
1257 so->so_state = SS_NOFDREF;
1258 }
1259 else
1260 so->so_state &= ~SS_ISFCONNECTING;
1261
1262 }
1263 TCP_INPUT((struct mbuf *)NULL, sizeof(struct ip),so);
1264 } /* SS_ISFCONNECTING */
1265#endif
1266#ifndef RT_OS_WINDOWS
1267 if ( UNIX_CHECK_FD_SET(so, NetworkEvents, rdhup)
1268 || UNIX_CHECK_FD_SET(so, NetworkEvents, rderr))
1269 {
1270 int err;
1271 int inq, outq;
1272 int status;
1273 socklen_t optlen = sizeof(int);
1274 inq = outq = 0;
1275 status = getsockopt(so->s, SOL_SOCKET, SO_ERROR, &err, &optlen);
1276 if (status != 0)
1277 Log(("NAT: can't get error status from %R[natsock]\n", so));
1278#ifndef RT_OS_SOLARIS
1279 status = ioctl(so->s, FIONREAD, &inq); /* tcp(7) recommends SIOCINQ which is Linux specific */
1280 if (status != 0 || status != EINVAL)
1281 {
1282 /* EINVAL returned if socket in listen state tcp(7)*/
1283 Log(("NAT: can't get depth of IN queue status from %R[natsock]\n", so));
1284 }
1285 status = ioctl(so->s, TIOCOUTQ, &outq); /* SIOCOUTQ see previous comment */
1286 if (status != 0)
1287 Log(("NAT: can't get depth of OUT queue from %R[natsock]\n", so));
1288#else
1289 /*
1290 * Solaris has bit different ioctl commands and its handlings
1291 * hint: streamio(7) I_NREAD
1292 */
1293#endif
1294 if ( so->so_state & SS_ISFCONNECTING
1295 || UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1296 {
1297 /**
1298 * Check if we need here take care about gracefull connection
1299 * @todo try with proxy server
1300 */
1301 if (UNIX_CHECK_FD_SET(so, NetworkEvents, readfds))
1302 {
1303 /*
1304 * Never meet inq != 0 or outq != 0, anyway let it stay for a while
1305 * in case it happens we'll able to detect it.
1306 * Give TCP/IP stack wait or expire the socket.
1307 */
1308 Log(("NAT: %R[natsock] err(%d:%s) s(in:%d,out:%d)happens on read I/O, "
1309 "other side close connection \n", so, err, strerror(err), inq, outq));
1310 CONTINUE(tcp);
1311 }
1312 goto tcp_input_close;
1313 }
1314 if ( !UNIX_CHECK_FD_SET(so, NetworkEvents, readfds)
1315 && !UNIX_CHECK_FD_SET(so, NetworkEvents, writefds)
1316 && !UNIX_CHECK_FD_SET(so, NetworkEvents, xfds))
1317 {
1318 Log(("NAT: system expires the socket %R[natsock] err(%d:%s) s(in:%d,out:%d) happens on non-I/O. ",
1319 so, err, strerror(err), inq, outq));
1320 goto tcp_input_close;
1321 }
1322 Log(("NAT: %R[natsock] we've met(%d:%s) s(in:%d, out:%d) unhandled combination hup (%d) "
1323 "rederr(%d) on (r:%d, w:%d, x:%d)\n",
1324 so, err, strerror(err),
1325 inq, outq,
1326 UNIX_CHECK_FD_SET(so, ign, rdhup),
1327 UNIX_CHECK_FD_SET(so, ign, rderr),
1328 UNIX_CHECK_FD_SET(so, ign, readfds),
1329 UNIX_CHECK_FD_SET(so, ign, writefds),
1330 UNIX_CHECK_FD_SET(so, ign, xfds)));
1331 /*
1332 * Give OS's TCP/IP stack a chance to resolve an issue or expire the socket.
1333 */
1334 CONTINUE(tcp);
1335tcp_input_close:
1336 so->so_state = SS_NOFDREF; /*cause connection valid tcp connection termination and socket closing */
1337 TCP_INPUT(pData, (struct mbuf *)NULL, sizeof(struct ip), so);
1338 CONTINUE(tcp);
1339 }
1340#endif
1341 LOOP_LABEL(tcp, so, so_next);
1342 }
1343
1344 /*
1345 * Now UDP sockets.
1346 * Incoming packets are sent straight away, they're not buffered.
1347 * Incoming UDP data isn't buffered either.
1348 */
1349 QSOCKET_FOREACH(so, so_next, udp)
1350 /* { */
1351#ifdef VBOX_WITH_SLIRP_MT
1352 if ( so->so_state & SS_NOFDREF
1353 && so->so_deleted == 1)
1354 {
1355 struct socket *son, *sop = NULL;
1356 QSOCKET_LOCK(udb);
1357 if (so->so_next != NULL)
1358 {
1359 if (so->so_next != &udb)
1360 SOCKET_LOCK(so->so_next);
1361 son = so->so_next;
1362 }
1363 if ( so->so_prev != &udb
1364 && so->so_prev != NULL)
1365 {
1366 SOCKET_LOCK(so->so_prev);
1367 sop = so->so_prev;
1368 }
1369 QSOCKET_UNLOCK(udb);
1370 remque(pData, so);
1371 NSOCK_DEC();
1372 SOCKET_UNLOCK(so);
1373 SOCKET_LOCK_DESTROY(so);
1374 RTMemFree(so);
1375 so_next = son;
1376 if (sop != NULL)
1377 SOCKET_UNLOCK(sop);
1378 CONTINUE_NO_UNLOCK(udp);
1379 }
1380#endif
1381 POLL_UDP_EVENTS(rc, error, so, &NetworkEvents);
1382
1383 LOG_NAT_SOCK(so, UDP, &NetworkEvents, readfds, writefds, xfds);
1384
1385 if (so->s != -1 && CHECK_FD_SET(so, NetworkEvents, readfds))
1386 {
1387 SORECVFROM(pData, so);
1388 }
1389 LOOP_LABEL(udp, so, so_next);
1390 }
1391
1392 }
1393
1394#ifndef VBOX_WITH_SLIRP_MT
1395 /*
1396 * See if we can start outputting
1397 */
1398 if (if_queued && link_up)
1399 if_start(pData);
1400#endif
1401
1402 STAM_PROFILE_STOP(&pData->StatPoll, a);
1403}
1404
1405#define ETH_ALEN 6
1406#define ETH_HLEN 14
1407
1408#define ARPOP_REQUEST 1 /* ARP request */
1409#define ARPOP_REPLY 2 /* ARP reply */
1410
1411struct ethhdr
1412{
1413 unsigned char h_dest[ETH_ALEN]; /* destination eth addr */
1414 unsigned char h_source[ETH_ALEN]; /* source ether addr */
1415 unsigned short h_proto; /* packet type ID field */
1416};
1417AssertCompileSize(struct ethhdr, 14);
1418
1419struct arphdr
1420{
1421 unsigned short ar_hrd; /* format of hardware address */
1422 unsigned short ar_pro; /* format of protocol address */
1423 unsigned char ar_hln; /* length of hardware address */
1424 unsigned char ar_pln; /* length of protocol address */
1425 unsigned short ar_op; /* ARP opcode (command) */
1426
1427 /*
1428 * Ethernet looks like this : This bit is variable sized however...
1429 */
1430 unsigned char ar_sha[ETH_ALEN]; /* sender hardware address */
1431 unsigned char ar_sip[4]; /* sender IP address */
1432 unsigned char ar_tha[ETH_ALEN]; /* target hardware address */
1433 unsigned char ar_tip[4]; /* target IP address */
1434};
1435AssertCompileSize(struct arphdr, 28);
1436
1437#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1438static void arp_input(PNATState pData, struct mbuf *m)
1439#else
1440static void arp_input(PNATState pData, const uint8_t *pkt, int pkt_len)
1441#endif
1442{
1443 struct ethhdr *eh;
1444 struct ethhdr *reh;
1445 struct arphdr *ah;
1446 struct arphdr *rah;
1447 int ar_op;
1448 struct ex_list *ex_ptr;
1449 uint32_t htip;
1450#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1451 uint8_t arp_reply[sizeof(struct arphdr) + ETH_HLEN];
1452 eh = (struct ethhdr *)pkt;
1453#else
1454 struct mbuf *mr;
1455 eh = mtod(m, struct ethhdr *);
1456#endif
1457 ah = (struct arphdr *)&eh[1];
1458 htip = ntohl(*(uint32_t*)ah->ar_tip);
1459
1460#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1461 mr = m_get(pData);
1462 mr->m_data += if_maxlinkhdr;
1463 mr->m_len = sizeof(struct arphdr);
1464 rah = mtod(mr, struct arphdr *);
1465#else
1466 reh = (struct ethhdr *)arp_reply;
1467 rah = (struct arphdr *)&reh[1];
1468#endif
1469
1470 ar_op = ntohs(ah->ar_op);
1471 switch(ar_op)
1472 {
1473 case ARPOP_REQUEST:
1474 if ((htip & pData->netmask) == ntohl(special_addr.s_addr))
1475 {
1476 if ( CTL_CHECK(htip, CTL_DNS)
1477 || CTL_CHECK(htip, CTL_ALIAS)
1478 || CTL_CHECK(htip, CTL_TFTP))
1479 goto arp_ok;
1480 for (ex_ptr = exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next)
1481 {
1482 if ((htip & ~pData->netmask) == ex_ptr->ex_addr)
1483 goto arp_ok;
1484 }
1485 return;
1486 arp_ok:
1487
1488#ifndef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1489 memcpy(reh->h_dest, eh->h_source, ETH_ALEN);
1490 memcpy(reh->h_source, &special_addr, ETH_ALEN);
1491 reh->h_source[5] = ah->ar_tip[3];
1492 reh->h_proto = htons(ETH_P_ARP);
1493#endif
1494 rah->ar_hrd = htons(1);
1495 rah->ar_pro = htons(ETH_P_IP);
1496 rah->ar_hln = ETH_ALEN;
1497 rah->ar_pln = 4;
1498 rah->ar_op = htons(ARPOP_REPLY);
1499 memcpy(rah->ar_sha, special_ethaddr, ETH_ALEN);
1500
1501 switch (htip & ~pData->netmask)
1502 {
1503 case CTL_DNS:
1504 case CTL_ALIAS:
1505 rah->ar_sha[5] = (uint8_t)(htip & ~pData->netmask);
1506 break;
1507 default:;
1508 }
1509
1510 memcpy(rah->ar_sip, ah->ar_tip, 4);
1511 memcpy(rah->ar_tha, ah->ar_sha, ETH_ALEN);
1512 memcpy(rah->ar_tip, ah->ar_sip, 4);
1513#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1514 if_encap(pData, ETH_P_ARP, mr);
1515 m_free(pData, m);
1516#else
1517 slirp_output(pData->pvUser, arp_reply, sizeof(arp_reply));
1518#endif
1519 }
1520 break;
1521 default:
1522 break;
1523 }
1524}
1525
1526void slirp_input(PNATState pData, const uint8_t *pkt, int pkt_len)
1527{
1528 struct mbuf *m;
1529 int proto;
1530 static bool fWarnedIpv6;
1531
1532 if (pkt_len < ETH_HLEN)
1533 {
1534 LogRel(("NAT: packet having size %d has been ingnored\n", pkt_len));
1535 return;
1536 }
1537
1538 m = m_get(pData);
1539 if (!m)
1540 {
1541 LogRel(("NAT: can't allocate new mbuf\n"));
1542 return;
1543 }
1544
1545 /* Note: we add to align the IP header */
1546
1547 if (M_FREEROOM(m) < pkt_len)
1548 m_inc(m, pkt_len);
1549
1550 m->m_len = pkt_len ;
1551 memcpy(m->m_data, pkt, pkt_len);
1552
1553 proto = ntohs(*(uint16_t *)(pkt + 12));
1554 switch(proto)
1555 {
1556 case ETH_P_ARP:
1557#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1558 arp_input(pData, m);
1559#else
1560 arp_input(pData, pkt, pkt_len);
1561 m_free(pData, m);
1562#endif
1563 break;
1564 case ETH_P_IP:
1565 /* Update time. Important if the network is very quiet, as otherwise
1566 * the first outgoing connection gets an incorrect timestamp. */
1567 updtime(pData);
1568 m->m_data += ETH_HLEN;
1569 m->m_len -= ETH_HLEN;
1570 ip_input(pData, m);
1571 break;
1572 case ETH_P_IPV6:
1573 m_free(pData, m);
1574 if (!fWarnedIpv6)
1575 {
1576 LogRel(("NAT: IPv6 not supported\n"));
1577 fWarnedIpv6 = true;
1578 }
1579 break;
1580 default:
1581 LogRel(("NAT: Unsupported protocol %x\n", proto));
1582 m_free(pData, m);
1583 break;
1584 }
1585 RTMemFree((void *)pkt);
1586}
1587
1588/* output the IP packet to the ethernet device */
1589#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1590void if_encap(PNATState pData, uint16_t eth_proto, struct mbuf *m)
1591#else
1592void if_encap(PNATState pData, uint8_t *ip_data, int ip_data_len)
1593#endif
1594{
1595#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1596 struct ethhdr *eh;
1597 uint8_t *buf = RTMemAlloc(1600);
1598 m->m_data -= if_maxlinkhdr;
1599 m->m_len += ETH_HLEN;
1600 eh = mtod(m, struct ethhdr *);
1601#else
1602 uint8_t buf[1600];
1603 struct ethhdr *eh = (struct ethhdr *)buf;
1604
1605 if (ip_data_len + ETH_HLEN > sizeof(buf))
1606 return;
1607
1608 memcpy(buf + sizeof(struct ethhdr), ip_data, ip_data_len);
1609#endif
1610
1611
1612 memcpy(eh->h_dest, client_ethaddr, ETH_ALEN);
1613 memcpy(eh->h_source, special_ethaddr, ETH_ALEN - 1);
1614 /* XXX: not correct */
1615 eh->h_source[5] = CTL_ALIAS;
1616#ifdef VBOX_WITH_SIMPLIFIED_SLIRP_SYNC
1617 eh->h_proto = htons(eth_proto);
1618#if 0
1619 slirp_output(pData->pvUser, m, mtod(m, uint8_t *), m->m_len);
1620#else
1621 memcpy(buf, mtod(m, uint8_t *), m->m_len);
1622 slirp_output(pData->pvUser, NULL, buf, m->m_len);
1623 m_free(pData, m);
1624#endif
1625#else
1626 eh->h_proto = htons(ETH_P_IP);
1627 slirp_output(pData->pvUser, buf, ip_data_len + ETH_HLEN);
1628#endif
1629}
1630
1631int slirp_redir(PNATState pData, int is_udp, int host_port,
1632 struct in_addr guest_addr, int guest_port)
1633{
1634 struct socket *so;
1635 Log2(("NAT: set redirect %s hp:%d gp:%d\n", (is_udp?"UDP":"TCP"), host_port, guest_port));
1636 if (is_udp)
1637 {
1638 so = udp_listen(pData, htons(host_port), guest_addr.s_addr,
1639 htons(guest_port), 0);
1640 }
1641 else
1642 {
1643 so = solisten(pData, htons(host_port), guest_addr.s_addr,
1644 htons(guest_port), 0);
1645 }
1646 Log2(("NAT: redirecting socket %R[natsock]\n", so));
1647 return (so != NULL ? 0 : -1);
1648}
1649
1650int slirp_add_exec(PNATState pData, int do_pty, const char *args, int addr_low_byte,
1651 int guest_port)
1652{
1653 return add_exec(&exec_list, do_pty, (char *)args,
1654 addr_low_byte, htons(guest_port));
1655}
1656
1657void slirp_set_ethaddr(PNATState pData, const uint8_t *ethaddr)
1658{
1659 memcpy(client_ethaddr, ethaddr, ETH_ALEN);
1660}
1661
1662#if defined(VBOX_WITH_SIMPLIFIED_SLIRP_SYNC) && defined(RT_OS_WINDOWS)
1663HANDLE *slirp_get_events(PNATState pData)
1664{
1665 return pData->phEvents;
1666}
1667void slirp_register_external_event(PNATState pData, HANDLE hEvent, int index)
1668{
1669 pData->phEvents[index] = hEvent;
1670}
1671#endif
1672
1673unsigned int slirp_get_timeout_ms(PNATState pData)
1674{
1675 if (link_up)
1676 {
1677 if (time_fasttimo)
1678 return 2;
1679 if (do_slowtimo)
1680 return 500; /* see PR_SLOWHZ */
1681 }
1682 return 0;
1683}
1684
1685#ifndef RT_OS_WINDOWS
1686int slirp_get_nsock(PNATState pData)
1687{
1688 return pData->nsock;
1689}
1690#endif
1691
1692/*
1693 * this function called from NAT thread
1694 */
1695void slirp_post_sent(PNATState pData, void *pvArg)
1696{
1697 struct socket *so = 0;
1698 struct tcpcb *tp = 0;
1699 struct mbuf *m = (struct mbuf *)pvArg;
1700 m_free(pData, m);
1701}
1702#ifdef VBOX_WITH_SLIRP_MT
1703void slirp_process_queue(PNATState pData)
1704{
1705 RTReqProcess(pData->pReqQueue, RT_INDEFINITE_WAIT);
1706}
1707void *slirp_get_queue(PNATState pData)
1708{
1709 return pData->pReqQueue;
1710}
1711#endif
1712
1713uint16_t slirp_get_service(int proto, uint16_t dport, uint16_t sport)
1714{
1715 uint16_t hdport, hsport, service;
1716 hdport = ntohs(dport);
1717 hsport = ntohs(sport);
1718 Log2(("proto: %d, dport: %d sport: %d\n", proto, hdport, hsport));
1719 service = 0;
1720#if 0
1721 /* Always return 0 here */
1722 switch (hdport)
1723 {
1724 case 500:
1725 /* service = sport; */
1726 break;
1727 }
1728#endif
1729 Log2(("service : %d\n", service));
1730 return htons(service);
1731}
1732
1733void slirp_set_dhcp_TFTP_prefix(PNATState pData, const char *tftpPrefix)
1734{
1735 Log2(("tftp_prefix:%s\n", tftpPrefix));
1736 tftp_prefix = tftpPrefix;
1737}
1738
1739void slirp_set_dhcp_TFTP_bootfile(PNATState pData, const char *bootFile)
1740{
1741 Log2(("bootFile:%s\n", bootFile));
1742 bootp_filename = bootFile;
1743}
1744
1745void slirp_set_dhcp_next_server(PNATState pData, const char *next_server)
1746{
1747 Log2(("next_server:%s\n", next_server));
1748 if (next_server == NULL)
1749 pData->tftp_server.s_addr = htonl(ntohl(special_addr.s_addr) | CTL_TFTP);
1750 else
1751 inet_aton(next_server, &pData->tftp_server);
1752}
1753#ifdef VBOX_WITH_SLIRP_DNS_PROXY
1754void slirp_set_dhcp_dns_proxy(PNATState pData, bool fDNSProxy)
1755{
1756 pData->use_dns_proxy = fDNSProxy;
1757}
1758#endif
Note: See TracBrowser for help on using the repository browser.

© 2025 Oracle Support Privacy / Do Not Sell My Info Terms of Use Trademark Policy Automated Access Etiquette