1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
|
/* fhandler_socket_inet.cc.
See fhandler.h for a description of the fhandler classes.
This file is part of Cygwin.
This software is a copyrighted work licensed under the terms of the
Cygwin license. Please consult the file "CYGWIN_LICENSE" for
details. */
#define __INSIDE_CYGWIN_NET__
#define USE_SYS_TYPES_FD_SET
#include "winsup.h"
#ifdef __x86_64__
/* 2014-04-24: Current Mingw headers define sockaddr_in6 using u_long (8 byte)
because a redefinition for LP64 systems is missing. This leads to a wrong
definition and size of sockaddr_in6 when building with winsock headers.
This definition is also required to use the right u_long type in subsequent
function calls. */
#undef u_long
#define u_long __ms_u_long
#endif
#include <w32api/ws2tcpip.h>
#include <w32api/mswsock.h>
#include <unistd.h>
#include <asm/byteorder.h>
#include <sys/socket.h>
#include <sys/param.h>
#include <sys/statvfs.h>
#include <cygwin/acl.h>
#include "cygerrno.h"
#include "path.h"
#include "fhandler.h"
#include "dtable.h"
#include "cygheap.h"
#include "shared_info.h"
#include "wininfo.h"
#define ASYNC_MASK (FD_READ|FD_WRITE|FD_OOB|FD_ACCEPT|FD_CONNECT)
#define EVENT_MASK (FD_READ|FD_WRITE|FD_OOB|FD_ACCEPT|FD_CONNECT|FD_CLOSE)
#define LOCK_EVENTS \
if (wsock_mtx && \
WaitForSingleObject (wsock_mtx, INFINITE) != WAIT_FAILED) \
{
#define UNLOCK_EVENTS \
ReleaseMutex (wsock_mtx); \
}
/* Maximum number of concurrently opened sockets from all Cygwin processes
per session. Note that shared sockets (through dup/fork/exec) are
counted as one socket. */
#define NUM_SOCKS 2048U
#define LOCK_EVENTS \
if (wsock_mtx && \
WaitForSingleObject (wsock_mtx, INFINITE) != WAIT_FAILED) \
{
#define UNLOCK_EVENTS \
ReleaseMutex (wsock_mtx); \
}
static wsa_event wsa_events[NUM_SOCKS] __attribute__((section (".cygwin_dll_common"), shared));
static LONG socket_serial_number __attribute__((section (".cygwin_dll_common"), shared));
static HANDLE wsa_slot_mtx;
static PWCHAR
sock_shared_name (PWCHAR buf, LONG num)
{
__small_swprintf (buf, L"socket.%d", num);
return buf;
}
static wsa_event *
search_wsa_event_slot (LONG new_serial_number)
{
WCHAR name[32], searchname[32];
UNICODE_STRING uname;
OBJECT_ATTRIBUTES attr;
NTSTATUS status;
if (!wsa_slot_mtx)
{
RtlInitUnicodeString (&uname, sock_shared_name (name, 0));
InitializeObjectAttributes (&attr, &uname, OBJ_INHERIT | OBJ_OPENIF,
get_session_parent_dir (),
everyone_sd (CYG_MUTANT_ACCESS));
status = NtCreateMutant (&wsa_slot_mtx, CYG_MUTANT_ACCESS, &attr, FALSE);
if (!NT_SUCCESS (status))
api_fatal ("Couldn't create/open shared socket mutex %S, %y",
&uname, status);
}
switch (WaitForSingleObject (wsa_slot_mtx, INFINITE))
{
case WAIT_OBJECT_0:
case WAIT_ABANDONED:
break;
default:
api_fatal ("WFSO failed for shared socket mutex, %E");
break;
}
unsigned int slot = new_serial_number % NUM_SOCKS;
while (wsa_events[slot].serial_number)
{
HANDLE searchmtx;
RtlInitUnicodeString (&uname, sock_shared_name (searchname,
wsa_events[slot].serial_number));
InitializeObjectAttributes (&attr, &uname, 0, get_session_parent_dir (),
NULL);
status = NtOpenMutant (&searchmtx, READ_CONTROL, &attr);
if (!NT_SUCCESS (status))
break;
/* Mutex still exists, attached socket is active, try next slot. */
NtClose (searchmtx);
slot = (slot + 1) % NUM_SOCKS;
if (slot == (new_serial_number % NUM_SOCKS))
{
/* Did the whole array once. Too bad. */
debug_printf ("No free socket slot");
ReleaseMutex (wsa_slot_mtx);
return NULL;
}
}
memset (&wsa_events[slot], 0, sizeof (wsa_event));
wsa_events[slot].serial_number = new_serial_number;
ReleaseMutex (wsa_slot_mtx);
return wsa_events + slot;
}
/* cygwin internal: map sockaddr into internet domain address */
static int
get_inet_addr_inet (const struct sockaddr *in, int inlen,
struct sockaddr_storage *out, int *outlen)
{
switch (in->sa_family)
{
case AF_INET:
memcpy (out, in, inlen);
*outlen = inlen;
/* If the peer address given in connect or sendto is the ANY address,
Winsock fails with WSAEADDRNOTAVAIL, while Linux converts that into
a connection/send attempt to LOOPBACK. We're doing the same here. */
if (((struct sockaddr_in *) out)->sin_addr.s_addr == htonl (INADDR_ANY))
((struct sockaddr_in *) out)->sin_addr.s_addr = htonl (INADDR_LOOPBACK);
return 0;
case AF_INET6:
memcpy (out, in, inlen);
*outlen = inlen;
/* See comment in AF_INET case. */
if (IN6_IS_ADDR_UNSPECIFIED (&((struct sockaddr_in6 *) out)->sin6_addr))
((struct sockaddr_in6 *) out)->sin6_addr = in6addr_loopback;
return 0;
default:
set_errno (EAFNOSUPPORT);
return SOCKET_ERROR;
}
}
/* There's no DLL which exports the symbol WSARecvMsg. One has to call
WSAIoctl as below to fetch the function pointer. Why on earth did the
MS developers decide not to export a normal symbol for these extension
functions? */
inline int
get_ext_funcptr (SOCKET sock, void *funcptr)
{
DWORD bret;
const GUID guid = WSAID_WSARECVMSG;
return WSAIoctl (sock, SIO_GET_EXTENSION_FUNCTION_POINTER,
(void *) &guid, sizeof (GUID), funcptr, sizeof (void *),
&bret, NULL, NULL);
}
static int
convert_ws1_ip_optname (int optname)
{
static int ws2_optname[] =
{
0,
IP_OPTIONS,
IP_MULTICAST_IF,
IP_MULTICAST_TTL,
IP_MULTICAST_LOOP,
IP_ADD_MEMBERSHIP,
IP_DROP_MEMBERSHIP,
IP_TTL,
IP_TOS,
IP_DONTFRAGMENT
};
return (optname < 1 || optname > _WS1_IP_DONTFRAGMENT)
? optname
: ws2_optname[optname];
}
fhandler_socket_wsock::fhandler_socket_wsock () :
fhandler_socket (),
wsock_events (NULL),
wsock_mtx (NULL),
wsock_evt (NULL),
status (),
prot_info_ptr (NULL)
{
need_fork_fixup (true);
}
fhandler_socket_wsock::~fhandler_socket_wsock ()
{
if (prot_info_ptr)
cfree (prot_info_ptr);
}
bool
fhandler_socket_wsock::init_events ()
{
LONG new_serial_number;
WCHAR name[32];
UNICODE_STRING uname;
OBJECT_ATTRIBUTES attr;
NTSTATUS status;
do
{
new_serial_number =
InterlockedIncrement (&socket_serial_number);
if (!new_serial_number) /* 0 is reserved for global mutex */
InterlockedIncrement (&socket_serial_number);
set_ino (new_serial_number);
RtlInitUnicodeString (&uname, sock_shared_name (name, new_serial_number));
InitializeObjectAttributes (&attr, &uname, OBJ_INHERIT | OBJ_OPENIF,
get_session_parent_dir (),
everyone_sd (CYG_MUTANT_ACCESS));
status = NtCreateMutant (&wsock_mtx, CYG_MUTANT_ACCESS, &attr, FALSE);
if (!NT_SUCCESS (status))
{
debug_printf ("NtCreateMutant(%S), %y", &uname, status);
set_errno (ENOBUFS);
return false;
}
if (status == STATUS_OBJECT_NAME_EXISTS)
NtClose (wsock_mtx);
}
while (status == STATUS_OBJECT_NAME_EXISTS);
if ((wsock_evt = CreateEvent (&sec_all, TRUE, FALSE, NULL))
== WSA_INVALID_EVENT)
{
debug_printf ("CreateEvent, %E");
set_errno (ENOBUFS);
NtClose (wsock_mtx);
return false;
}
if (WSAEventSelect (get_socket (), wsock_evt, EVENT_MASK) == SOCKET_ERROR)
{
debug_printf ("WSAEventSelect, %E");
set_winsock_errno ();
NtClose (wsock_evt);
NtClose (wsock_mtx);
return false;
}
if (!(wsock_events = search_wsa_event_slot (new_serial_number)))
{
set_errno (ENOBUFS);
NtClose (wsock_evt);
NtClose (wsock_mtx);
return false;
}
if (get_socket_type () == SOCK_DGRAM)
wsock_events->events = FD_WRITE;
return true;
}
int
fhandler_socket_wsock::evaluate_events (const long event_mask, long &events,
const bool erase)
{
int ret = 0;
long events_now = 0;
WSANETWORKEVENTS evts = { 0 };
if (!(WSAEnumNetworkEvents (get_socket (), wsock_evt, &evts)))
{
if (evts.lNetworkEvents)
{
LOCK_EVENTS;
wsock_events->events |= evts.lNetworkEvents;
events_now = (wsock_events->events & event_mask);
if (evts.lNetworkEvents & FD_CONNECT)
{
wsock_events->connect_errorcode = evts.iErrorCode[FD_CONNECT_BIT];
/* Setting the connect_state and calling the AF_LOCAL handshake
here allows to handle this stuff from a single point. This
is independent of FD_CONNECT being requested. Consider a
server calling connect(2) and then immediately poll(2) with
only polling for POLLIN (example: postfix), or select(2) just
asking for descriptors ready to read.
Something weird occurs in Winsock: If you fork off and call
recv/send on the duplicated, already connected socket, another
FD_CONNECT event is generated in the child process. This
would trigger a call to af_local_connect which obviously fail.
Avoid this by calling set_connect_state only if connect_state
is connect_pending. */
if (connect_state () == connect_pending)
{
if (wsock_events->connect_errorcode)
connect_state (connect_failed);
else if (af_local_connect ())
{
wsock_events->connect_errorcode = WSAGetLastError ();
connect_state (connect_failed);
}
else
connect_state (connected);
}
}
UNLOCK_EVENTS;
if ((evts.lNetworkEvents & FD_OOB) && wsock_events->owner)
kill (wsock_events->owner, SIGURG);
}
}
LOCK_EVENTS;
if ((events = events_now) != 0
|| (events = (wsock_events->events & event_mask)) != 0)
{
if (events & FD_CONNECT)
{
int wsa_err = wsock_events->connect_errorcode;
if (wsa_err)
{
/* CV 2014-04-23: This is really weird. If you call connect
asynchronously on a socket and then select, an error like
"Connection refused" is set in the event and in the SO_ERROR
socket option. If you call connect, then dup, then select,
the error is set in the event, but not in the SO_ERROR socket
option, despite the dup'ed socket handle referring to the same
socket. We're trying to workaround this problem here by
taking the connect errorcode from the event and write it back
into the SO_ERROR socket option.
CV 2014-06-16: Call WSASetLastError *after* setsockopt since,
apparently, setsockopt sets the last WSA error code to 0 on
success. */
::setsockopt (get_socket (), SOL_SOCKET, SO_ERROR,
(const char *) &wsa_err, sizeof wsa_err);
WSASetLastError (wsa_err);
ret = SOCKET_ERROR;
}
else
wsock_events->events |= FD_WRITE;
wsock_events->events &= ~FD_CONNECT;
wsock_events->connect_errorcode = 0;
}
/* This test makes accept/connect behave as on Linux when accept/connect
is called on a socket for which shutdown has been called. The second
half of this code is in the shutdown method. */
if (events & FD_CLOSE)
{
if ((event_mask & FD_ACCEPT) && saw_shutdown_read ())
{
WSASetLastError (WSAEINVAL);
ret = SOCKET_ERROR;
}
if (event_mask & FD_CONNECT)
{
WSASetLastError (WSAECONNRESET);
ret = SOCKET_ERROR;
}
}
if (erase)
wsock_events->events &= ~(events & ~(FD_WRITE | FD_CLOSE));
}
UNLOCK_EVENTS;
return ret;
}
int
fhandler_socket_wsock::wait_for_events (const long event_mask,
const DWORD flags)
{
if (async_io ())
return 0;
int ret;
long events = 0;
DWORD wfmo_timeout = 50;
DWORD timeout;
WSAEVENT ev[3] = { wsock_evt, NULL, NULL };
wait_signal_arrived here (ev[1]);
DWORD ev_cnt = 2;
if ((ev[2] = pthread::get_cancel_event ()) != NULL)
++ev_cnt;
if (is_nonblocking () || (flags & MSG_DONTWAIT))
timeout = 0;
else if (event_mask & FD_READ)
timeout = rcvtimeo ();
else if (event_mask & FD_WRITE)
timeout = sndtimeo ();
else
timeout = INFINITE;
while (!(ret = evaluate_events (event_mask, events, !(flags & MSG_PEEK)))
&& !events)
{
if (timeout == 0)
{
WSASetLastError (WSAEWOULDBLOCK);
return SOCKET_ERROR;
}
if (timeout < wfmo_timeout)
wfmo_timeout = timeout;
switch (WSAWaitForMultipleEvents (ev_cnt, ev, FALSE, wfmo_timeout, FALSE))
{
case WSA_WAIT_TIMEOUT:
case WSA_WAIT_EVENT_0:
if (timeout != INFINITE)
timeout -= wfmo_timeout;
break;
case WSA_WAIT_EVENT_0 + 1:
if (_my_tls.call_signal_handler ())
break;
WSASetLastError (WSAEINTR);
return SOCKET_ERROR;
case WSA_WAIT_EVENT_0 + 2:
pthread::static_cancel_self ();
break;
default:
/* wsock_evt can be NULL. We're generating the same errno values
as for sockets on which shutdown has been called. */
if (WSAGetLastError () != WSA_INVALID_HANDLE)
WSASetLastError (WSAEFAULT);
else
WSASetLastError ((event_mask & FD_CONNECT) ? WSAECONNRESET
: WSAEINVAL);
return SOCKET_ERROR;
}
}
return ret;
}
void
fhandler_socket_wsock::release_events ()
{
if (WaitForSingleObject (wsock_mtx, INFINITE) != WAIT_FAILED)
{
HANDLE evt = wsock_evt;
HANDLE mtx = wsock_mtx;
wsock_evt = wsock_mtx = NULL;
ReleaseMutex (mtx);
NtClose (evt);
NtClose (mtx);
}
}
void
fhandler_socket_wsock::set_close_on_exec (bool val)
{
set_no_inheritance (wsock_mtx, val);
set_no_inheritance (wsock_evt, val);
if (need_fixup_before ())
{
close_on_exec (val);
debug_printf ("set close_on_exec for %s to %d", get_name (), val);
}
else
fhandler_base::set_close_on_exec (val);
}
/* Called if a freshly created socket is not inheritable. In that case we
have to use fixup_before_fork_exec. See comment in set_socket_handle for
a description of the problem. */
void
fhandler_socket_wsock::init_fixup_before ()
{
prot_info_ptr = (LPWSAPROTOCOL_INFOW)
cmalloc_abort (HEAP_BUF, sizeof (WSAPROTOCOL_INFOW));
cygheap->fdtab.inc_need_fixup_before ();
}
int
fhandler_socket_wsock::fixup_before_fork_exec (DWORD win_pid)
{
SOCKET ret = WSADuplicateSocketW (get_socket (), win_pid, prot_info_ptr);
if (ret)
set_winsock_errno ();
else
debug_printf ("WSADuplicateSocket succeeded (%x)", prot_info_ptr->dwProviderReserved);
return (int) ret;
}
void
fhandler_socket_wsock::fixup_after_fork (HANDLE parent)
{
fork_fixup (parent, wsock_mtx, "wsock_mtx");
fork_fixup (parent, wsock_evt, "wsock_evt");
if (!need_fixup_before ())
{
fhandler_base::fixup_after_fork (parent);
return;
}
SOCKET new_sock = WSASocketW (FROM_PROTOCOL_INFO, FROM_PROTOCOL_INFO,
FROM_PROTOCOL_INFO, prot_info_ptr, 0,
WSA_FLAG_OVERLAPPED);
if (new_sock == INVALID_SOCKET)
{
set_winsock_errno ();
set_io_handle ((HANDLE) INVALID_SOCKET);
}
else
{
/* Even though the original socket was not inheritable, the duplicated
socket is potentially inheritable again. */
SetHandleInformation ((HANDLE) new_sock, HANDLE_FLAG_INHERIT, 0);
set_io_handle ((HANDLE) new_sock);
debug_printf ("WSASocket succeeded (%p)", new_sock);
}
}
void
fhandler_socket_wsock::fixup_after_exec ()
{
if (need_fixup_before () && !close_on_exec ())
fixup_after_fork (NULL);
}
int
fhandler_socket_wsock::dup (fhandler_base *child, int flags)
{
debug_printf ("here");
fhandler_socket_wsock *fhs = (fhandler_socket_wsock *) child;
if (!DuplicateHandle (GetCurrentProcess (), wsock_mtx,
GetCurrentProcess (), &fhs->wsock_mtx,
0, TRUE, DUPLICATE_SAME_ACCESS))
{
__seterrno ();
return -1;
}
if (!DuplicateHandle (GetCurrentProcess (), wsock_evt,
GetCurrentProcess (), &fhs->wsock_evt,
0, TRUE, DUPLICATE_SAME_ACCESS))
{
__seterrno ();
NtClose (fhs->wsock_mtx);
return -1;
}
if (!need_fixup_before ())
{
int ret = fhandler_base::dup (child, flags);
if (ret)
{
NtClose (fhs->wsock_evt);
NtClose (fhs->wsock_mtx);
}
return ret;
}
cygheap->user.deimpersonate ();
fhs->init_fixup_before ();
fhs->set_io_handle (get_io_handle ());
int ret = fhs->fixup_before_fork_exec (GetCurrentProcessId ());
cygheap->user.reimpersonate ();
if (!ret)
{
fhs->fixup_after_fork (GetCurrentProcess ());
if (fhs->get_io_handle() != (HANDLE) INVALID_SOCKET)
return 0;
}
cygheap->fdtab.dec_need_fixup_before ();
NtClose (fhs->wsock_evt);
NtClose (fhs->wsock_mtx);
return -1;
}
int
fhandler_socket_wsock::set_socket_handle (SOCKET sock, int af, int type,
int flags)
{
DWORD hdl_flags;
bool lsp_fixup = false;
int file_flags = O_RDWR | O_BINARY;
/* Usually sockets are inheritable IFS objects. Unfortunately some virus
scanners or other network-oriented software replace normal sockets
with their own kind, which is running through a filter driver called
"layered service provider" (LSP) which, fortunately, are deprecated.
LSP sockets are not kernel objects. They are typically not marked as
inheritable, nor are they IFS handles. They are in fact not inheritable
to child processes, and it does not help to mark them inheritable via
SetHandleInformation. Subsequent socket calls in the child process fail
with error 10038, WSAENOTSOCK.
There's a neat way to workaround these annoying LSP sockets. WSAIoctl
allows to fetch the underlying base socket, which is a normal, inheritable
IFS handle. So we fetch the base socket, duplicate it, and close the
original socket. Now we have a standard IFS socket which (hopefully)
works as expected.
If that doesn't work for some reason, mark the sockets for duplication
via WSADuplicateSocket/WSASocket. This requires to start the child
process in SUSPENDED state so we only do this if really necessary. */
if (!GetHandleInformation ((HANDLE) sock, &hdl_flags)
|| !(hdl_flags & HANDLE_FLAG_INHERIT))
{
int ret;
SOCKET base_sock;
DWORD bret;
lsp_fixup = true;
debug_printf ("LSP handle: %p", sock);
ret = WSAIoctl (sock, SIO_BASE_HANDLE, NULL, 0, (void *) &base_sock,
sizeof (base_sock), &bret, NULL, NULL);
if (ret)
debug_printf ("WSAIoctl: %u", WSAGetLastError ());
else if (base_sock != sock)
{
if (GetHandleInformation ((HANDLE) base_sock, &hdl_flags)
&& (flags & HANDLE_FLAG_INHERIT))
{
if (!DuplicateHandle (GetCurrentProcess (), (HANDLE) base_sock,
GetCurrentProcess (), (PHANDLE) &base_sock,
0, TRUE, DUPLICATE_SAME_ACCESS))
debug_printf ("DuplicateHandle failed, %E");
else
{
::closesocket (sock);
sock = base_sock;
lsp_fixup = false;
}
}
}
}
set_io_handle ((HANDLE) sock);
set_addr_family (af);
set_socket_type (type);
if (!init_events ())
return -1;
if (flags & SOCK_NONBLOCK)
file_flags |= O_NONBLOCK;
if (flags & SOCK_CLOEXEC)
{
set_close_on_exec (true);
file_flags |= O_CLOEXEC;
}
set_flags (file_flags);
if (lsp_fixup)
init_fixup_before ();
set_unique_id ();
if (get_socket_type () == SOCK_DGRAM)
{
/* Workaround the problem that a missing listener on a UDP socket
in a call to sendto will result in select/WSAEnumNetworkEvents
reporting that the socket has pending data and a subsequent call
to recvfrom will return -1 with error set to WSAECONNRESET.
This problem is a regression introduced in Windows 2000.
Instead of fixing the problem, a new socket IOCTL code has
been added, see http://support.microsoft.com/kb/263823 */
BOOL cr = FALSE;
DWORD blen;
if (WSAIoctl (sock, SIO_UDP_CONNRESET, &cr, sizeof cr, NULL, 0,
&blen, NULL, NULL) == SOCKET_ERROR)
debug_printf ("Reset SIO_UDP_CONNRESET: WinSock error %u",
WSAGetLastError ());
}
#ifdef __x86_64__
rmem () = 212992;
wmem () = 212992;
#else
rmem () = 64512;
wmem () = 64512;
#endif
return 0;
}
fhandler_socket_inet::fhandler_socket_inet () :
fhandler_socket_wsock (),
oobinline (false)
{
}
fhandler_socket_inet::~fhandler_socket_inet ()
{
}
int
fhandler_socket_inet::socket (int af, int type, int protocol, int flags)
{
SOCKET sock;
int ret;
/* This test should be covered by ::socket, but make sure we don't
accidentally try anything else. */
if (type != SOCK_STREAM && type != SOCK_DGRAM && type != SOCK_RAW)
{
set_errno (EINVAL);
return -1;
}
sock = ::socket (af, type, protocol);
if (sock == INVALID_SOCKET)
{
set_winsock_errno ();
return -1;
}
ret = set_socket_handle (sock, af, type, flags);
if (ret < 0)
::closesocket (sock);
return ret;
}
int
fhandler_socket_inet::socketpair (int af, int type, int protocol, int flags,
fhandler_socket *fh_out)
{
set_errno (EAFNOSUPPORT);
return -1;
}
int
fhandler_socket_inet::bind (const struct sockaddr *name, int namelen)
{
int res = -1;
if (!saw_reuseaddr ())
{
/* If the application didn't explicitely request SO_REUSEADDR,
enforce POSIX standard socket binding behaviour by setting the
SO_EXCLUSIVEADDRUSE socket option. See cygwin_setsockopt()
for a more detailed description. */
int on = 1;
int ret = ::setsockopt (get_socket (), SOL_SOCKET,
SO_EXCLUSIVEADDRUSE,
(const char *) &on, sizeof on);
debug_printf ("%d = setsockopt(SO_EXCLUSIVEADDRUSE), %E", ret);
}
if (::bind (get_socket (), name, namelen))
set_winsock_errno ();
else
res = 0;
return res;
}
int
fhandler_socket_inet::connect (const struct sockaddr *name, int namelen)
{
struct sockaddr_storage sst;
if (get_inet_addr_inet (name, namelen, &sst, &namelen) == SOCKET_ERROR)
return SOCKET_ERROR;
/* Initialize connect state to "connect_pending". State is ultimately set
to "connected" or "connect_failed" in wait_for_events when the FD_CONNECT
event occurs. Note that the underlying OS sockets are always non-blocking
and a successfully initiated non-blocking Winsock connect always returns
WSAEWOULDBLOCK. Thus it's safe to rely on event handling.
Check for either unconnected or connect_failed since in both cases it's
allowed to retry connecting the socket. It's also ok (albeit ugly) to
call connect to check if a previous non-blocking connect finished.
Set connect_state before calling connect, otherwise a race condition with
an already running select or poll might occur. */
if (connect_state () == unconnected || connect_state () == connect_failed)
connect_state (connect_pending);
int res = ::connect (get_socket (), (struct sockaddr *) &sst, namelen);
if (!is_nonblocking ()
&& res == SOCKET_ERROR
&& WSAGetLastError () == WSAEWOULDBLOCK)
res = wait_for_events (FD_CONNECT | FD_CLOSE, 0);
if (res)
{
DWORD err = WSAGetLastError ();
/* Some applications use the ugly technique to check if a non-blocking
connect succeeded by calling connect again, until it returns EISCONN.
This circumvents the event handling and connect_state is never set.
Thus we check for this situation here. */
if (err == WSAEISCONN)
connect_state (connected);
/* Winsock returns WSAEWOULDBLOCK if the non-blocking socket cannot be
conected immediately. Convert to POSIX/Linux compliant EINPROGRESS. */
else if (is_nonblocking () && err == WSAEWOULDBLOCK)
WSASetLastError (WSAEINPROGRESS);
/* Winsock returns WSAEINVAL if the socket is already a listener.
Convert to POSIX/Linux compliant EISCONN. */
else if (err == WSAEINVAL && connect_state () == listener)
WSASetLastError (WSAEISCONN);
/* Any other error except WSAEALREADY during connect_pending means the
connect failed. */
else if (connect_state () == connect_pending && err != WSAEALREADY)
connect_state (connect_failed);
set_winsock_errno ();
}
return res;
}
int
fhandler_socket_inet::listen (int backlog)
{
int res = ::listen (get_socket (), backlog);
if (res && WSAGetLastError () == WSAEINVAL)
{
/* It's perfectly valid to call listen on an unbound INET socket.
In this case the socket is automatically bound to an unused
port number, listening on all interfaces. On WinSock, listen
fails with WSAEINVAL when it's called on an unbound socket.
So we have to bind manually here to have POSIX semantics. */
if (get_addr_family () == AF_INET)
{
struct sockaddr_in sin;
sin.sin_family = AF_INET;
sin.sin_port = 0;
sin.sin_addr.s_addr = INADDR_ANY;
if (!::bind (get_socket (), (struct sockaddr *) &sin, sizeof sin))
res = ::listen (get_socket (), backlog);
}
else if (get_addr_family () == AF_INET6)
{
struct sockaddr_in6 sin6;
memset (&sin6, 0, sizeof sin6);
sin6.sin6_family = AF_INET6;
if (!::bind (get_socket (), (struct sockaddr *) &sin6, sizeof sin6))
res = ::listen (get_socket (), backlog);
}
}
if (!res)
connect_state (listener); /* gets set to connected on accepted socket. */
else
set_winsock_errno ();
return res;
}
int
fhandler_socket_inet::accept4 (struct sockaddr *peer, int *len, int flags)
{
int ret = -1;
/* Allows NULL peer and len parameters. */
struct sockaddr_storage lpeer;
int llen = sizeof (struct sockaddr_storage);
/* Windows event handling does not check for the validity of the desired
flags so we have to do it here. */
if (connect_state () != listener)
{
WSASetLastError (WSAEINVAL);
set_winsock_errno ();
return -1;
}
SOCKET res = INVALID_SOCKET;
while (!(res = wait_for_events (FD_ACCEPT | FD_CLOSE, 0))
&& (res = ::accept (get_socket (), (struct sockaddr *) &lpeer, &llen))
== INVALID_SOCKET
&& WSAGetLastError () == WSAEWOULDBLOCK)
;
if (res == INVALID_SOCKET)
set_winsock_errno ();
else
{
cygheap_fdnew fd;
if (fd >= 0)
{
fhandler_socket_inet *sock = (fhandler_socket_inet *)
build_fh_dev (dev ());
if (sock && sock->set_socket_handle (res, get_addr_family (),
get_socket_type (),
get_socket_flags ()) == 0)
{
sock->async_io (false); /* set_socket_handle disables async. */
/* No locking necessary at this point. */
sock->wsock_events->events = wsock_events->events | FD_WRITE;
sock->wsock_events->owner = wsock_events->owner;
sock->connect_state (connected);
fd = sock;
if (fd <= 2)
set_std_handle (fd);
ret = fd;
if (peer)
{
memcpy (peer, &lpeer, MIN (*len, llen));
*len = llen;
}
}
else
delete sock;
}
if (ret == -1)
::closesocket (res);
}
return ret;
}
int
fhandler_socket_inet::getsockname (struct sockaddr *name, int *namelen)
{
int res = -1;
/* WinSock just returns WSAEFAULT if the buffer is too small. Use a
big enough local buffer and truncate later as necessary, per POSIX. */
struct sockaddr_storage sock;
int len = sizeof sock;
res = ::getsockname (get_socket (), (struct sockaddr *) &sock, &len);
if (!res)
{
memcpy (name, &sock, MIN (*namelen, len));
*namelen = len;
}
else
{
if (WSAGetLastError () == WSAEINVAL)
{
/* WinSock returns WSAEINVAL if the socket is locally
unbound. Per SUSv3 this is not an error condition.
We're faking a valid return value here by creating the
same content in the sockaddr structure as on Linux. */
memset (&sock, 0, sizeof sock);
sock.ss_family = get_addr_family ();
switch (get_addr_family ())
{
case AF_INET:
res = 0;
len = (int) sizeof (struct sockaddr_in);
break;
case AF_INET6:
res = 0;
len = (int) sizeof (struct sockaddr_in6);
break;
default:
WSASetLastError (WSAEOPNOTSUPP);
break;
}
if (!res)
{
memcpy (name, &sock, MIN (*namelen, len));
*namelen = len;
}
}
if (res)
set_winsock_errno ();
}
return res;
}
int
fhandler_socket_inet::getpeername (struct sockaddr *name, int *namelen)
{
/* Always use a local big enough buffer and truncate later as necessary
per POSIX. WinSock unfortunately only returns WSAEFAULT if the buffer
is too small. */
struct sockaddr_storage sock;
int len = sizeof sock;
int res = ::getpeername (get_socket (), (struct sockaddr *) &sock, &len);
if (res)
set_winsock_errno ();
else
{
memcpy (name, &sock, MIN (*namelen, len));
*namelen = len;
}
return res;
}
int
fhandler_socket_wsock::shutdown (int how)
{
int res = ::shutdown (get_socket (), how);
/* Linux allows to call shutdown for any socket, even if it's not connected.
This also disables to call accept on this socket, if shutdown has been
called with the SHUT_RD or SHUT_RDWR parameter. In contrast, WinSock
only allows to call shutdown on a connected socket. The accept function
is in no way affected. So, what we do here is to fake success, and to
change the event settings so that an FD_CLOSE event is triggered for the
calling Cygwin function. The evaluate_events method handles the call
from accept specially to generate a Linux-compatible behaviour. */
if (res && WSAGetLastError () != WSAENOTCONN)
set_winsock_errno ();
else
{
res = 0;
switch (how)
{
case SHUT_RD:
saw_shutdown_read (true);
wsock_events->events |= FD_CLOSE;
SetEvent (wsock_evt);
break;
case SHUT_WR:
saw_shutdown_write (true);
break;
case SHUT_RDWR:
saw_shutdown_read (true);
saw_shutdown_write (true);
wsock_events->events |= FD_CLOSE;
SetEvent (wsock_evt);
break;
}
}
return res;
}
int
fhandler_socket_wsock::close ()
{
int res = 0;
release_events ();
while ((res = ::closesocket (get_socket ())) != 0)
{
if (WSAGetLastError () != WSAEWOULDBLOCK)
{
set_winsock_errno ();
res = -1;
break;
}
if (cygwait (10) == WAIT_SIGNALED)
{
set_errno (EINTR);
res = -1;
break;
}
WSASetLastError (0);
}
return res;
}
ssize_t
fhandler_socket_inet::recv_internal (LPWSAMSG wsamsg, bool use_recvmsg)
{
ssize_t res = 0;
DWORD ret = 0, wret;
int evt_mask = (wsamsg->dwFlags & MSG_OOB) ? FD_OOB : FD_READ;
LPWSABUF &wsabuf = wsamsg->lpBuffers;
ULONG &wsacnt = wsamsg->dwBufferCount;
static NO_COPY LPFN_WSARECVMSG WSARecvMsg;
bool read_oob = false;
/* CV 2014-10-26: Do not check for the connect_state at this point. In
certain scenarios there's no way to check the connect state reliably.
Example (hexchat): Parent process creates socket, forks, child process
calls connect, parent process calls read. Even if the event handling
allows to check for FD_CONNECT in the parent, there is always yet another
scenario we can easily break. */
DWORD wait_flags = wsamsg->dwFlags;
bool waitall = !!(wait_flags & MSG_WAITALL);
wsamsg->dwFlags &= (MSG_OOB | MSG_PEEK | MSG_DONTROUTE);
if (use_recvmsg)
{
if (!WSARecvMsg
&& get_ext_funcptr (get_socket (), &WSARecvMsg) == SOCKET_ERROR)
{
if (wsamsg->Control.len > 0)
{
set_winsock_errno ();
return SOCKET_ERROR;
}
use_recvmsg = false;
}
else /* Only MSG_PEEK is supported by WSARecvMsg. */
wsamsg->dwFlags &= MSG_PEEK;
}
if (waitall)
{
if (get_socket_type () != SOCK_STREAM)
{
WSASetLastError (WSAEOPNOTSUPP);
set_winsock_errno ();
return SOCKET_ERROR;
}
if (is_nonblocking () || (wsamsg->dwFlags & (MSG_OOB | MSG_PEEK)))
waitall = false;
}
/* recv() returns EINVAL if MSG_OOB flag is set in inline mode. */
if (oobinline && (wsamsg->dwFlags & MSG_OOB))
{
set_errno (EINVAL);
return SOCKET_ERROR;
}
/* Check whether OOB data is ready or not */
if (get_socket_type () == SOCK_STREAM)
if ((wsamsg->dwFlags & MSG_OOB) || oobinline)
{
u_long atmark = 0;
#ifdef __x86_64__
/* SIOCATMARK = _IOR('s',7,u_long) */
int err = ::ioctlsocket (get_socket (), _IOR('s',7,u_long), &atmark);
#else
int err = ::ioctlsocket (get_socket (), SIOCATMARK, &atmark);
#endif
if (err)
{
set_winsock_errno ();
return SOCKET_ERROR;
}
/* If there is no OOB data, recv() with MSG_OOB returns EINVAL.
Note: The return value of SIOCATMARK in non-inline mode of
winsock is FALSE if OOB data exists, TRUE otherwise. */
if (atmark && (wsamsg->dwFlags & MSG_OOB))
{
/* No OOB data */
set_errno (EINVAL);
return SOCKET_ERROR;
}
/* Inline mode for out-of-band (OOB) data of winsock is
completely broken. That is, SIOCATMARK always returns
TRUE in inline mode. Due to this problem, application
cannot determine OOB data at all. Therefore the behavior
of a socket with SO_OOBINLINE set is simulated using
a socket with SO_OOBINLINE not set. In this fake inline
mode, the order of the OOB and non-OOB data is not
preserved. OOB data is read before non-OOB data sent
prior to the OOB data. However, this most likely is
not a problem in most cases. */
/* If there is OOB data, read OOB data using MSG_OOB in
fake inline mode. */
if (!atmark && oobinline)
{
read_oob = true;
evt_mask = FD_OOB;
}
}
/* Note: Don't call WSARecvFrom(MSG_PEEK) without actually having data
waiting in the buffers, otherwise the event handling gets messed up
for some reason. */
while (!(res = wait_for_events (evt_mask | FD_CLOSE, wait_flags))
|| saw_shutdown_read ())
{
DWORD dwFlags = wsamsg->dwFlags | (read_oob ? MSG_OOB : 0);
if (use_recvmsg)
res = WSARecvMsg (get_socket (), wsamsg, &wret, NULL, NULL);
/* This is working around a really weird problem in WinSock.
Assume you create a socket, fork the process (thus duplicating
the socket), connect the socket in the child, then call recv
on the original socket handle in the parent process.
In this scenario, calls to WinSock's recvfrom and WSARecvFrom
in the parent will fail with WSAEINVAL, regardless whether both
address parameters, name and namelen, are NULL or point to valid
storage. However, calls to recv and WSARecv succeed as expected.
Per MSDN, WSAEINVAL in the context of recv means "The socket has not
been bound". It is as if the recvfrom functions test if the socket
is bound locally, but in the parent process, WinSock doesn't know
about that and fails, while the same test is omitted in the recv
functions.
This also covers another weird case: WinSock returns WSAEFAULT if
namelen is a valid pointer while name is NULL. Both parameters are
ignored for TCP sockets, so this only occurs when using UDP socket. */
else if (!wsamsg->name || get_socket_type () == SOCK_STREAM)
res = WSARecv (get_socket (), wsabuf, wsacnt, &wret, &dwFlags,
NULL, NULL);
else
res = WSARecvFrom (get_socket (), wsabuf, wsacnt, &wret,
&dwFlags, wsamsg->name, &wsamsg->namelen,
NULL, NULL);
if (!res)
{
ret += wret;
if (!waitall)
break;
while (wret && wsacnt)
{
if (wsabuf->len > wret)
{
wsabuf->len -= wret;
wsabuf->buf += wret;
wret = 0;
}
else
{
wret -= wsabuf->len;
++wsabuf;
--wsacnt;
}
}
if (!wret)
break;
}
else if (WSAGetLastError () != WSAEWOULDBLOCK)
break;
}
if (res)
{
/* According to SUSv3, errno isn't set in that case and no error
condition is returned. */
if (WSAGetLastError () == WSAEMSGSIZE)
ret += wret;
else if (!ret)
{
/* ESHUTDOWN isn't defined for recv in SUSv3. Simply EOF is returned
in this case. */
if (WSAGetLastError () == WSAESHUTDOWN)
ret = 0;
else
{
set_winsock_errno ();
return SOCKET_ERROR;
}
}
}
return ret;
}
ssize_t
fhandler_socket_wsock::recvfrom (void *in_ptr, size_t len, int flags,
struct sockaddr *from, int *fromlen)
{
char *ptr = (char *) in_ptr;
#ifdef __x86_64__
/* size_t is 64 bit, but the len member in WSABUF is 32 bit.
Split buffer if necessary. */
DWORD bufcnt = len / UINT32_MAX + ((!len || (len % UINT32_MAX)) ? 1 : 0);
WSABUF wsabuf[bufcnt];
WSAMSG wsamsg = { from, from && fromlen ? *fromlen : 0,
wsabuf, bufcnt,
{ 0, NULL },
(DWORD) flags };
/* Don't use len as loop condition, it could be 0. */
for (WSABUF *wsaptr = wsabuf; bufcnt--; ++wsaptr)
{
wsaptr->len = MIN (len, UINT32_MAX);
wsaptr->buf = ptr;
len -= wsaptr->len;
ptr += wsaptr->len;
}
#else
WSABUF wsabuf = { len, ptr };
WSAMSG wsamsg = { from, from && fromlen ? *fromlen : 0,
&wsabuf, 1,
{ 0, NULL},
(DWORD) flags };
#endif
ssize_t ret = recv_internal (&wsamsg, false);
if (fromlen)
*fromlen = wsamsg.namelen;
return ret;
}
ssize_t
fhandler_socket_wsock::recvmsg (struct msghdr *msg, int flags)
{
/* Disappointing but true: Even if WSARecvMsg is supported, it's only
supported for datagram and raw sockets. */
bool use_recvmsg = true;
if (get_socket_type () == SOCK_STREAM || get_addr_family () == AF_LOCAL)
{
use_recvmsg = false;
msg->msg_controllen = 0;
}
WSABUF wsabuf[msg->msg_iovlen];
WSABUF *wsaptr = wsabuf + msg->msg_iovlen;
const struct iovec *iovptr = msg->msg_iov + msg->msg_iovlen;
while (--wsaptr >= wsabuf)
{
wsaptr->len = (--iovptr)->iov_len;
wsaptr->buf = (char *) iovptr->iov_base;
}
WSAMSG wsamsg = { (struct sockaddr *) msg->msg_name, msg->msg_namelen,
wsabuf, (DWORD) msg->msg_iovlen,
{ (DWORD) msg->msg_controllen, (char *) msg->msg_control },
(DWORD) flags };
ssize_t ret = recv_internal (&wsamsg, use_recvmsg);
if (ret >= 0)
{
msg->msg_namelen = wsamsg.namelen;
msg->msg_controllen = wsamsg.Control.len;
if (!CYGWIN_VERSION_CHECK_FOR_USING_ANCIENT_MSGHDR)
msg->msg_flags = wsamsg.dwFlags;
}
return ret;
}
void __reg3
fhandler_socket_wsock::read (void *in_ptr, size_t& len)
{
char *ptr = (char *) in_ptr;
#ifdef __x86_64__
/* size_t is 64 bit, but the len member in WSABUF is 32 bit.
Split buffer if necessary. */
DWORD bufcnt = len / UINT32_MAX + ((!len || (len % UINT32_MAX)) ? 1 : 0);
WSABUF wsabuf[bufcnt];
WSAMSG wsamsg = { NULL, 0, wsabuf, bufcnt, { 0, NULL }, 0 };
/* Don't use len as loop condition, it could be 0. */
for (WSABUF *wsaptr = wsabuf; bufcnt--; ++wsaptr)
{
wsaptr->len = MIN (len, UINT32_MAX);
wsaptr->buf = ptr;
len -= wsaptr->len;
ptr += wsaptr->len;
}
#else
WSABUF wsabuf = { len, ptr };
WSAMSG wsamsg = { NULL, 0, &wsabuf, 1, { 0, NULL }, 0 };
#endif
len = recv_internal (&wsamsg, false);
}
ssize_t
fhandler_socket_wsock::readv (const struct iovec *const iov, const int iovcnt,
ssize_t tot)
{
WSABUF wsabuf[iovcnt];
WSABUF *wsaptr = wsabuf + iovcnt;
const struct iovec *iovptr = iov + iovcnt;
while (--wsaptr >= wsabuf)
{
wsaptr->len = (--iovptr)->iov_len;
wsaptr->buf = (char *) iovptr->iov_base;
}
WSAMSG wsamsg = { NULL, 0, wsabuf, (DWORD) iovcnt, { 0, NULL}, 0 };
return recv_internal (&wsamsg, false);
}
ssize_t
fhandler_socket_wsock::send_internal (struct _WSAMSG *wsamsg, int flags)
{
ssize_t res = 0;
DWORD ret = 0, sum = 0;
WSABUF out_buf[wsamsg->dwBufferCount];
bool use_sendmsg = false;
DWORD wait_flags = flags & MSG_DONTWAIT;
bool nosignal = !!(flags & MSG_NOSIGNAL);
flags &= (MSG_OOB | MSG_DONTROUTE);
if (wsamsg->Control.len > 0)
use_sendmsg = true;
/* Workaround for MSDN KB 823764: Split a message into chunks <= SO_SNDBUF.
in_idx is the index of the current lpBuffers from the input wsamsg buffer.
in_off is used to keep track of the next byte to write from a wsamsg
buffer which only gets partially written. */
for (DWORD in_idx = 0, in_off = 0;
in_idx < wsamsg->dwBufferCount;
in_off >= wsamsg->lpBuffers[in_idx].len && (++in_idx, in_off = 0))
{
/* Split a message into the least number of pieces to minimize the
number of WsaSendTo calls. Don't split datagram messages (bad idea).
out_idx is the index of the next buffer in the out_buf WSABUF,
also the number of buffers given to WSASendTo.
out_len is the number of bytes in the buffers given to WSASendTo.
Don't split datagram messages (very bad idea). */
DWORD out_idx = 0;
DWORD out_len = 0;
if (get_socket_type () == SOCK_STREAM)
{
do
{
out_buf[out_idx].buf = wsamsg->lpBuffers[in_idx].buf + in_off;
out_buf[out_idx].len = wsamsg->lpBuffers[in_idx].len - in_off;
out_len += out_buf[out_idx].len;
out_idx++;
}
while (out_len < (unsigned) wmem ()
&& (in_off = 0, ++in_idx < wsamsg->dwBufferCount));
/* Tweak len of the last out_buf buffer so the entire number of bytes
is (less than or) equal to wmem (). Fix out_len as well since it's
used in a subsequent test expression. */
if (out_len > (unsigned) wmem ())
{
out_buf[out_idx - 1].len -= out_len - (unsigned) wmem ();
out_len = (unsigned) wmem ();
}
/* Add the bytes written from the current last buffer to in_off,
so in_off points to the next byte to be written from that buffer,
or beyond which lets the outper loop skip to the next buffer. */
in_off += out_buf[out_idx - 1].len;
}
do
{
if (use_sendmsg)
res = WSASendMsg (get_socket (), wsamsg, flags, &ret, NULL, NULL);
else if (get_socket_type () == SOCK_STREAM)
res = WSASendTo (get_socket (), out_buf, out_idx, &ret, flags,
wsamsg->name, wsamsg->namelen, NULL, NULL);
else
res = WSASendTo (get_socket (), wsamsg->lpBuffers,
wsamsg->dwBufferCount, &ret, flags,
wsamsg->name, wsamsg->namelen, NULL, NULL);
if (res && (WSAGetLastError () == WSAEWOULDBLOCK))
{
LOCK_EVENTS;
wsock_events->events &= ~FD_WRITE;
UNLOCK_EVENTS;
}
}
while (res && (WSAGetLastError () == WSAEWOULDBLOCK)
&& !(res = wait_for_events (FD_WRITE | FD_CLOSE, wait_flags)));
if (!res)
{
sum += ret;
/* For streams, return to application if the number of bytes written
is less than the number of bytes we intended to write in a single
call to WSASendTo. Otherwise we would have to add code to
backtrack in the input buffers, which is questionable. There was
probably a good reason we couldn't write more. */
if (get_socket_type () != SOCK_STREAM || ret < out_len)
break;
}
else if (is_nonblocking () || WSAGetLastError() != WSAEWOULDBLOCK)
break;
}
if (sum)
res = sum;
else if (res == SOCKET_ERROR)
{
set_winsock_errno ();
/* Special handling for EPIPE and SIGPIPE.
EPIPE is generated if the local end has been shut down on a connection
oriented socket. In this case the process will also receive a SIGPIPE
unless MSG_NOSIGNAL is set. */
if ((get_errno () == ECONNABORTED || get_errno () == ESHUTDOWN)
&& get_socket_type () == SOCK_STREAM)
{
set_errno (EPIPE);
if (!nosignal)
raise (SIGPIPE);
}
}
return res;
}
ssize_t
fhandler_socket_inet::sendto (const void *in_ptr, size_t len, int flags,
const struct sockaddr *to, int tolen)
{
char *ptr = (char *) in_ptr;
struct sockaddr_storage sst;
if (to && get_inet_addr_inet (to, tolen, &sst, &tolen) == SOCKET_ERROR)
return SOCKET_ERROR;
#ifdef __x86_64__
/* size_t is 64 bit, but the len member in WSABUF is 32 bit.
Split buffer if necessary. */
DWORD bufcnt = len / UINT32_MAX + ((!len || (len % UINT32_MAX)) ? 1 : 0);
WSABUF wsabuf[bufcnt];
WSAMSG wsamsg = { to ? (struct sockaddr *) &sst : NULL, tolen,
wsabuf, bufcnt,
{ 0, NULL },
0 };
/* Don't use len as loop condition, it could be 0. */
for (WSABUF *wsaptr = wsabuf; bufcnt--; ++wsaptr)
{
wsaptr->len = MIN (len, UINT32_MAX);
wsaptr->buf = ptr;
len -= wsaptr->len;
ptr += wsaptr->len;
}
#else
WSABUF wsabuf = { len, ptr };
WSAMSG wsamsg = { to ? (struct sockaddr *) &sst : NULL, tolen,
&wsabuf, 1,
{ 0, NULL},
0 };
#endif
return send_internal (&wsamsg, flags);
}
ssize_t
fhandler_socket_inet::sendmsg (const struct msghdr *msg, int flags)
{
struct sockaddr_storage sst;
int len = 0;
if (msg->msg_name
&& get_inet_addr_inet ((struct sockaddr *) msg->msg_name,
msg->msg_namelen, &sst, &len) == SOCKET_ERROR)
return SOCKET_ERROR;
WSABUF wsabuf[msg->msg_iovlen];
WSABUF *wsaptr = wsabuf;
const struct iovec *iovptr = msg->msg_iov;
for (int i = 0; i < msg->msg_iovlen; ++i)
{
wsaptr->len = iovptr->iov_len;
(wsaptr++)->buf = (char *) (iovptr++)->iov_base;
}
/* Disappointing but true: Even if WSASendMsg is supported, it's only
supported for datagram and raw sockets. */
DWORD controllen = (DWORD) ((get_socket_type () == SOCK_STREAM)
? 0 : msg->msg_controllen);
WSAMSG wsamsg = { msg->msg_name ? (struct sockaddr *) &sst : NULL, len,
wsabuf, (DWORD) msg->msg_iovlen,
{ controllen, (char *) msg->msg_control },
0 };
return send_internal (&wsamsg, flags);
}
ssize_t
fhandler_socket_wsock::write (const void *in_ptr, size_t len)
{
char *ptr = (char *) in_ptr;
#ifdef __x86_64__
/* size_t is 64 bit, but the len member in WSABUF is 32 bit.
Split buffer if necessary. */
DWORD bufcnt = len / UINT32_MAX + ((!len || (len % UINT32_MAX)) ? 1 : 0);
WSABUF wsabuf[bufcnt];
WSAMSG wsamsg = { NULL, 0, wsabuf, bufcnt, { 0, NULL }, 0 };
/* Don't use len as loop condition, it could be 0. */
for (WSABUF *wsaptr = wsabuf; bufcnt--; ++wsaptr)
{
wsaptr->len = MIN (len, UINT32_MAX);
wsaptr->buf = ptr;
len -= wsaptr->len;
ptr += wsaptr->len;
}
#else
WSABUF wsabuf = { len, ptr };
WSAMSG wsamsg = { NULL, 0, &wsabuf, 1, { 0, NULL }, 0 };
#endif
return send_internal (&wsamsg, 0);
}
ssize_t
fhandler_socket_wsock::writev (const struct iovec *const iov, const int iovcnt,
ssize_t tot)
{
WSABUF wsabuf[iovcnt];
WSABUF *wsaptr = wsabuf;
const struct iovec *iovptr = iov;
for (int i = 0; i < iovcnt; ++i)
{
wsaptr->len = iovptr->iov_len;
(wsaptr++)->buf = (char *) (iovptr++)->iov_base;
}
WSAMSG wsamsg = { NULL, 0, wsabuf, (DWORD) iovcnt, { 0, NULL}, 0 };
return send_internal (&wsamsg, 0);
}
int
fhandler_socket_inet::setsockopt (int level, int optname, const void *optval,
socklen_t optlen)
{
bool ignore = false;
int ret = -1;
/* Preprocessing setsockopt. Set ignore to true if setsockopt call should
get skipped entirely. */
switch (level)
{
case SOL_SOCKET:
switch (optname)
{
case SO_PEERCRED:
set_errno (ENOPROTOOPT);
return -1;
case SO_REUSEADDR:
/* Per POSIX we must not be able to reuse a complete duplicate of a
local TCP address (same IP, same port), even if SO_REUSEADDR has
been set. This behaviour is maintained in WinSock for backward
compatibility, while the WinSock standard behaviour of stream
socket binding is equivalent to the POSIX behaviour as if
SO_REUSEADDR has been set. The SO_EXCLUSIVEADDRUSE option has
been added to allow an application to request POSIX standard
behaviour in the non-SO_REUSEADDR case.
To emulate POSIX socket binding behaviour, note that SO_REUSEADDR
has been set but don't call setsockopt. Instead
fhandler_socket::bind sets SO_EXCLUSIVEADDRUSE if the application
did not set SO_REUSEADDR. */
if (optlen < (socklen_t) sizeof (int))
{
set_errno (EINVAL);
return ret;
}
if (get_socket_type () == SOCK_STREAM)
ignore = true;
break;
case SO_RCVTIMEO:
case SO_SNDTIMEO:
if (optlen < (socklen_t) sizeof (struct timeval))
{
set_errno (EINVAL);
return ret;
}
if (timeval_to_ms ((struct timeval *) optval,
(optname == SO_RCVTIMEO) ? rcvtimeo ()
: sndtimeo ()))
ret = 0;
else
set_errno (EDOM);
return ret;
case SO_OOBINLINE:
/* Inline mode for out-of-band (OOB) data of winsock is
completely broken. That is, SIOCATMARK always returns
TRUE in inline mode. Due to this problem, application
cannot determine OOB data at all. Therefore the behavior
of a socket with SO_OOBINLINE set is simulated using
a socket with SO_OOBINLINE not set. In this fake inline
mode, the order of the OOB and non-OOB data is not
preserved. OOB data is read before non-OOB data sent
prior to the OOB data. However, this most likely is
not a problem in most cases. */
/* Here, instead of actually setting inline mode, simply
set the variable oobinline. */
oobinline = *(int *) optval ? true : false;
ignore = true;
break;
default:
break;
}
break;
case IPPROTO_IP:
/* Old applications still use the old WinSock1 IPPROTO_IP values. */
if (CYGWIN_VERSION_CHECK_FOR_USING_WINSOCK1_VALUES)
optname = convert_ws1_ip_optname (optname);
switch (optname)
{
case IP_TOS:
/* Winsock doesn't support setting the IP_TOS field with setsockopt
and TOS was never implemented for TCP anyway. setsockopt returns
WinSock error 10022, WSAEINVAL when trying to set the IP_TOS
field. We just return 0 instead. */
ignore = true;
break;
default:
break;
}
break;
case IPPROTO_IPV6:
{
switch (optname)
{
case IPV6_TCLASS:
/* Unsupported */
ignore = true;
break;
default:
break;
}
}
default:
break;
}
/* Call Winsock setsockopt (or not) */
if (ignore)
ret = 0;
else
{
ret = ::setsockopt (get_socket (), level, optname, (const char *) optval,
optlen);
if (ret == SOCKET_ERROR)
{
set_winsock_errno ();
return ret;
}
}
if (optlen == (socklen_t) sizeof (int))
debug_printf ("setsockopt optval=%x", *(int *) optval);
/* Postprocessing setsockopt, setting fhandler_socket members, etc. */
switch (level)
{
case SOL_SOCKET:
switch (optname)
{
case SO_REUSEADDR:
saw_reuseaddr (*(int *) optval);
break;
case SO_RCVBUF:
rmem (*(int *) optval);
break;
case SO_SNDBUF:
wmem (*(int *) optval);
break;
default:
break;
}
break;
default:
break;
}
return ret;
}
int
fhandler_socket_inet::getsockopt (int level, int optname, const void *optval,
socklen_t *optlen)
{
bool onebyte = false;
int ret = -1;
/* Preprocessing getsockopt. */
switch (level)
{
case SOL_SOCKET:
switch (optname)
{
case SO_PEERCRED:
set_errno (ENOPROTOOPT);
return -1;
case SO_REUSEADDR:
{
unsigned int *reuseaddr = (unsigned int *) optval;
if (*optlen < (socklen_t) sizeof *reuseaddr)
{
set_errno (EINVAL);
return -1;
}
*reuseaddr = saw_reuseaddr();
*optlen = (socklen_t) sizeof *reuseaddr;
return 0;
}
case SO_RCVTIMEO:
case SO_SNDTIMEO:
{
struct timeval *time_out = (struct timeval *) optval;
if (*optlen < (socklen_t) sizeof *time_out)
{
set_errno (EINVAL);
return -1;
}
DWORD ms = (optname == SO_RCVTIMEO) ? rcvtimeo () : sndtimeo ();
if (ms == 0 || ms == INFINITE)
{
time_out->tv_sec = 0;
time_out->tv_usec = 0;
}
else
{
time_out->tv_sec = ms / MSPERSEC;
time_out->tv_usec = ((ms % MSPERSEC) * USPERSEC) / MSPERSEC;
}
*optlen = (socklen_t) sizeof *time_out;
return 0;
}
case SO_TYPE:
{
unsigned int *type = (unsigned int *) optval;
*type = get_socket_type ();
*optlen = (socklen_t) sizeof *type;
return 0;
}
case SO_OOBINLINE:
*(int *) optval = oobinline ? 1 : 0;
return 0;
default:
break;
}
break;
case IPPROTO_IP:
/* Old applications still use the old WinSock1 IPPROTO_IP values. */
if (CYGWIN_VERSION_CHECK_FOR_USING_WINSOCK1_VALUES)
optname = convert_ws1_ip_optname (optname);
break;
default:
break;
}
/* Call Winsock getsockopt */
ret = ::getsockopt (get_socket (), level, optname, (char *) optval,
(int *) optlen);
if (ret == SOCKET_ERROR)
{
set_winsock_errno ();
return ret;
}
/* Postprocessing getsockopt, setting fhandler_socket members, etc. Set
onebyte true for options returning BOOLEAN instead of a boolean DWORD. */
switch (level)
{
case SOL_SOCKET:
switch (optname)
{
case SO_ERROR:
{
int *e = (int *) optval;
debug_printf ("WinSock SO_ERROR = %d", *e);
*e = find_winsock_errno (*e);
}
break;
case SO_KEEPALIVE:
case SO_DONTROUTE:
onebyte = true;
break;
default:
break;
}
break;
case IPPROTO_TCP:
switch (optname)
{
case TCP_NODELAY:
onebyte = true;
break;
default:
break;
}
default:
break;
}
if (onebyte)
{
/* Regression in Vista and later: instead of a 4 byte BOOL value, a
1 byte BOOLEAN value is returned, in contrast to older systems and
the documentation. Since an int type is expected by the calling
application, we convert the result here. For some reason only three
BSD-compatible socket options seem to be affected. */
BOOLEAN *in = (BOOLEAN *) optval;
int *out = (int *) optval;
*out = *in;
*optlen = 4;
}
return ret;
}
int
fhandler_socket_wsock::ioctl (unsigned int cmd, void *p)
{
int res;
switch (cmd)
{
/* Here we handle only ioctl commands which are understood by Winsock.
However, we have a problem, which is, the different size of u_long
in Windows and 64 bit Cygwin. This affects the definitions of
FIOASYNC, etc, because they are defined in terms of sizeof(u_long).
So we have to use case labels which are independent of the sizeof
u_long. Since we're redefining u_long at the start of this file to
matching Winsock's idea of u_long, we can use the real definitions in
calls to Windows. In theory we also have to make sure to convert the
different ideas of u_long between the application and Winsock, but
fortunately, the parameters defined as u_long pointers are on Linux
and BSD systems defined as int pointer, so the applications will
use a type of the expected size. Hopefully. */
case FIOASYNC:
#ifdef __x86_64__
case _IOW('f', 125, u_long):
#endif
res = WSAAsyncSelect (get_socket (), winmsg, WM_ASYNCIO,
*(int *) p ? ASYNC_MASK : 0);
syscall_printf ("Async I/O on socket %s",
*(int *) p ? "started" : "cancelled");
async_io (*(int *) p != 0);
/* If async_io is switched off, revert the event handling. */
if (*(int *) p == 0)
WSAEventSelect (get_socket (), wsock_evt, EVENT_MASK);
break;
case FIONREAD:
#ifdef __x86_64__
case _IOR('f', 127, u_long):
#endif
/* Make sure to use the Winsock definition of FIONREAD. */
res = ::ioctlsocket (get_socket (), _IOR('f', 127, u_long), (u_long *) p);
if (res == SOCKET_ERROR)
set_winsock_errno ();
break;
case FIONBIO:
case SIOCATMARK:
/* Sockets are always non-blocking internally. So we just note the
state here. */
#ifdef __x86_64__
/* Convert the different idea of u_long in the definition of cmd. */
if (((cmd >> 16) & IOCPARM_MASK) == sizeof (unsigned long))
cmd = (cmd & ~(IOCPARM_MASK << 16)) | (sizeof (u_long) << 16);
#endif
if (cmd == FIONBIO)
{
syscall_printf ("socket is now %sblocking",
*(int *) p ? "non" : "");
set_nonblocking (*(int *) p);
res = 0;
}
else
res = ::ioctlsocket (get_socket (), cmd, (u_long *) p);
/* In winsock, the return value of SIOCATMARK is FALSE if
OOB data exists, TRUE otherwise. This is almost opposite
to expectation. */
#ifdef __x86_64__
/* SIOCATMARK = _IOR('s',7,u_long) */
if (cmd == _IOR('s',7,u_long) && !res)
*(u_long *)p = !*(u_long *)p;
#else
if (cmd == SIOCATMARK && !res)
*(u_long *)p = !*(u_long *)p;
#endif
break;
default:
res = fhandler_socket::ioctl (cmd, p);
break;
}
syscall_printf ("%d = ioctl_socket(%x, %p)", res, cmd, p);
return res;
}
int
fhandler_socket_wsock::fcntl (int cmd, intptr_t arg)
{
int res = 0;
switch (cmd)
{
case F_SETOWN:
{
pid_t pid = (pid_t) arg;
LOCK_EVENTS;
wsock_events->owner = pid;
UNLOCK_EVENTS;
debug_printf ("owner set to %d", pid);
}
break;
case F_GETOWN:
res = wsock_events->owner;
break;
default:
res = fhandler_socket::fcntl (cmd, arg);
break;
}
return res;
}
|