aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/libs/curl/lib/cf-socket.c
blob: e4d6a5b86d73b7c49119acdf63e77b14f3c2d3b4 (plain) (blame)
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
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
/***************************************************************************
 *                                  _   _ ____  _
 *  Project                     ___| | | |  _ \| |
 *                             / __| | | | |_) | |
 *                            | (__| |_| |  _ <| |___
 *                             \___|\___/|_| \_\_____|
 *
 * Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
 *
 * This software is licensed as described in the file COPYING, which
 * you should have received as part of this distribution. The terms
 * are also available at https://curl.se/docs/copyright.html.
 *
 * You may opt to use, copy, modify, merge, publish, distribute and/or sell
 * copies of the Software, and permit persons to whom the Software is
 * furnished to do so, under the terms of the COPYING file.
 *
 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
 * KIND, either express or implied.
 *
 * SPDX-License-Identifier: curl
 *
 ***************************************************************************/

#include "curl_setup.h"

#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h> /* <netinet/tcp.h> may need it */
#endif
#ifdef HAVE_SYS_UN_H
#include <sys/un.h> /* for sockaddr_un */
#endif
#ifdef HAVE_LINUX_TCP_H
#include <linux/tcp.h>
#elif defined(HAVE_NETINET_TCP_H)
#include <netinet/tcp.h>
#endif
#ifdef HAVE_NETINET_UDP_H
#include <netinet/udp.h>
#endif
#ifdef HAVE_SYS_IOCTL_H
#include <sys/ioctl.h>
#endif
#ifdef HAVE_NETDB_H
#include <netdb.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#ifdef HAVE_ARPA_INET_H
#include <arpa/inet.h>
#endif

#ifdef __VMS
#include <in.h>
#include <inet.h>
#endif

#ifdef __DragonFly__
/* Required for __DragonFly_version */
#include <sys/param.h>
#endif

#include "urldata.h"
#include "bufq.h"
#include "sendf.h"
#include "if2ip.h"
#include "strerror.h"
#include "cfilters.h"
#include "cf-socket.h"
#include "connect.h"
#include "select.h"
#include "url.h" /* for Curl_safefree() */
#include "multiif.h"
#include "sockaddr.h" /* required for Curl_sockaddr_storage */
#include "inet_ntop.h"
#include "inet_pton.h"
#include "progress.h"
#include "warnless.h"
#include "conncache.h"
#include "multihandle.h"
#include "rand.h"
#include "share.h"
#include "strdup.h"
#include "version_win32.h"

/* The last 3 #include files should be in this order */
#include "curl_printf.h"
#include "curl_memory.h"
#include "memdebug.h"


#if defined(USE_IPV6) && defined(IPV6_V6ONLY) && defined(_WIN32)
/* It makes support for IPv4-mapped IPv6 addresses.
 * Linux kernel, NetBSD, FreeBSD and Darwin: default is off;
 * Windows Vista and later: default is on;
 * DragonFly BSD: acts like off, and dummy setting;
 * OpenBSD and earlier Windows: unsupported.
 * Linux: controlled by /proc/sys/net/ipv6/bindv6only.
 */
static void set_ipv6_v6only(curl_socket_t sockfd, int on)
{
  (void)setsockopt(sockfd, IPPROTO_IPV6, IPV6_V6ONLY, (void *)&on, sizeof(on));
}
#else
#define set_ipv6_v6only(x,y)
#endif

static void tcpnodelay(struct Curl_easy *data, curl_socket_t sockfd)
{
#if defined(TCP_NODELAY)
  curl_socklen_t onoff = (curl_socklen_t) 1;
  int level = IPPROTO_TCP;
  char buffer[STRERROR_LEN];

  if(setsockopt(sockfd, level, TCP_NODELAY, (void *)&onoff,
                sizeof(onoff)) < 0)
    infof(data, "Could not set TCP_NODELAY: %s",
          Curl_strerror(SOCKERRNO, buffer, sizeof(buffer)));
#else
  (void)data;
  (void)sockfd;
#endif
}

#ifdef SO_NOSIGPIPE
/* The preferred method on macOS (10.2 and later) to prevent SIGPIPEs when
   sending data to a dead peer (instead of relying on the 4th argument to send
   being MSG_NOSIGNAL). Possibly also existing and in use on other BSD
   systems? */
static void nosigpipe(struct Curl_easy *data,
                      curl_socket_t sockfd)
{
  int onoff = 1;
  (void)data;
  if(setsockopt(sockfd, SOL_SOCKET, SO_NOSIGPIPE, (void *)&onoff,
                sizeof(onoff)) < 0) {
#if !defined(CURL_DISABLE_VERBOSE_STRINGS)
    char buffer[STRERROR_LEN];
    infof(data, "Could not set SO_NOSIGPIPE: %s",
          Curl_strerror(SOCKERRNO, buffer, sizeof(buffer)));
#endif
  }
}
#else
#define nosigpipe(x,y) Curl_nop_stmt
#endif

#if defined(USE_WINSOCK) && \
    defined(TCP_KEEPIDLE) && defined(TCP_KEEPINTVL) && defined(TCP_KEEPCNT)
/*  Win 10, v 1709 (10.0.16299) and later can use SetSockOpt TCP_KEEP____
 *  so should use seconds */
#define CURL_WINSOCK_KEEP_SSO
#define KEEPALIVE_FACTOR(x)
#elif defined(USE_WINSOCK) || \
   (defined(__sun) && !defined(TCP_KEEPIDLE)) || \
   (defined(__DragonFly__) && __DragonFly_version < 500702) || \
   (defined(_WIN32) && !defined(TCP_KEEPIDLE))
/* Solaris < 11.4, DragonFlyBSD < 500702 and Windows < 10.0.16299
 * use millisecond units. */
#define KEEPALIVE_FACTOR(x) (x *= 1000)
#else
#define KEEPALIVE_FACTOR(x)
#endif

#if defined(USE_WINSOCK) && !defined(SIO_KEEPALIVE_VALS)
#define SIO_KEEPALIVE_VALS    _WSAIOW(IOC_VENDOR,4)

struct tcp_keepalive {
  u_long onoff;
  u_long keepalivetime;
  u_long keepaliveinterval;
};
#endif

static void
tcpkeepalive(struct Curl_easy *data,
             curl_socket_t sockfd)
{
  int optval = data->set.tcp_keepalive?1:0;

  /* only set IDLE and INTVL if setting KEEPALIVE is successful */
  if(setsockopt(sockfd, SOL_SOCKET, SO_KEEPALIVE,
        (void *)&optval, sizeof(optval)) < 0) {
    infof(data, "Failed to set SO_KEEPALIVE on fd "
          "%" FMT_SOCKET_T ": errno %d",
          sockfd, SOCKERRNO);
  }
  else {
#if defined(SIO_KEEPALIVE_VALS) /* Windows */
/* Windows 10, version 1709 (10.0.16299) and later versions */
#if defined(CURL_WINSOCK_KEEP_SSO)
    optval = curlx_sltosi(data->set.tcp_keepidle);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPIDLE,
                (const char *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPIDLE on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
    optval = curlx_sltosi(data->set.tcp_keepintvl);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPINTVL,
                (const char *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPINTVL on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
    optval = curlx_sltosi(data->set.tcp_keepcnt);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPCNT,
                (const char *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPCNT on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
#else /* Windows < 10.0.16299 */
    struct tcp_keepalive vals;
    DWORD dummy;
    vals.onoff = 1;
    optval = curlx_sltosi(data->set.tcp_keepidle);
    KEEPALIVE_FACTOR(optval);
    vals.keepalivetime = (u_long)optval;
    optval = curlx_sltosi(data->set.tcp_keepintvl);
    KEEPALIVE_FACTOR(optval);
    vals.keepaliveinterval = (u_long)optval;
    if(WSAIoctl(sockfd, SIO_KEEPALIVE_VALS, (LPVOID) &vals, sizeof(vals),
                NULL, 0, &dummy, NULL, NULL) != 0) {
      infof(data, "Failed to set SIO_KEEPALIVE_VALS on fd "
            "%" FMT_SOCKET_T ": errno %d", sockfd, SOCKERRNO);
    }
#endif
#else /* !Windows */
#ifdef TCP_KEEPIDLE
    optval = curlx_sltosi(data->set.tcp_keepidle);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPIDLE,
          (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPIDLE on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
#elif defined(TCP_KEEPALIVE)
    /* macOS style */
    optval = curlx_sltosi(data->set.tcp_keepidle);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPALIVE,
      (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPALIVE on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
#elif defined(TCP_KEEPALIVE_THRESHOLD)
    /* Solaris <11.4 style */
    optval = curlx_sltosi(data->set.tcp_keepidle);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPALIVE_THRESHOLD,
      (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPALIVE_THRESHOLD on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
#endif
#ifdef TCP_KEEPINTVL
    optval = curlx_sltosi(data->set.tcp_keepintvl);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPINTVL,
          (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPINTVL on fd "
            "%" FMT_SOCKET_T ": errno %d",
            sockfd, SOCKERRNO);
    }
#elif defined(TCP_KEEPALIVE_ABORT_THRESHOLD)
    /* Solaris <11.4 style */
    /* TCP_KEEPALIVE_ABORT_THRESHOLD should equal to
     * TCP_KEEPCNT * TCP_KEEPINTVL on other platforms.
     * The default value of TCP_KEEPCNT is 9 on Linux,
     * 8 on *BSD/macOS, 5 or 10 on Windows. We use the
     * default config for Solaris <11.4 because there is
     * no default value for TCP_KEEPCNT on Solaris 11.4.
     *
     * Note that the consequent probes will not be sent
     * at equal intervals on Solaris, but will be sent
     * using the exponential backoff algorithm. */
    optval = curlx_sltosi(data->set.tcp_keepcnt) *
             curlx_sltosi(data->set.tcp_keepintvl);
    KEEPALIVE_FACTOR(optval);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPALIVE_ABORT_THRESHOLD,
          (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPALIVE_ABORT_THRESHOLD on fd "
            "%" FMT_SOCKET_T ": errno %d", sockfd, SOCKERRNO);
    }
#endif
#ifdef TCP_KEEPCNT
    optval = curlx_sltosi(data->set.tcp_keepcnt);
    if(setsockopt(sockfd, IPPROTO_TCP, TCP_KEEPCNT,
                (void *)&optval, sizeof(optval)) < 0) {
      infof(data, "Failed to set TCP_KEEPCNT on fd "
            "%" FMT_SOCKET_T ": errno %d", sockfd, SOCKERRNO);
    }
#endif
#endif
  }
}

/**
 * Assign the address `ai` to the Curl_sockaddr_ex `dest` and
 * set the transport used.
 */
void Curl_sock_assign_addr(struct Curl_sockaddr_ex *dest,
                           const struct Curl_addrinfo *ai,
                           int transport)
{
  /*
   * The Curl_sockaddr_ex structure is basically libcurl's external API
   * curl_sockaddr structure with enough space available to directly hold
   * any protocol-specific address structures. The variable declared here
   * will be used to pass / receive data to/from the fopensocket callback
   * if this has been set, before that, it is initialized from parameters.
   */
  dest->family = ai->ai_family;
  switch(transport) {
  case TRNSPRT_TCP:
    dest->socktype = SOCK_STREAM;
    dest->protocol = IPPROTO_TCP;
    break;
  case TRNSPRT_UNIX:
    dest->socktype = SOCK_STREAM;
    dest->protocol = IPPROTO_IP;
    break;
  default: /* UDP and QUIC */
    dest->socktype = SOCK_DGRAM;
    dest->protocol = IPPROTO_UDP;
    break;
  }
  dest->addrlen = (unsigned int)ai->ai_addrlen;

  if(dest->addrlen > sizeof(struct Curl_sockaddr_storage))
    dest->addrlen = sizeof(struct Curl_sockaddr_storage);
  memcpy(&dest->sa_addr, ai->ai_addr, dest->addrlen);
}

static CURLcode socket_open(struct Curl_easy *data,
                            struct Curl_sockaddr_ex *addr,
                            curl_socket_t *sockfd)
{
  DEBUGASSERT(data);
  DEBUGASSERT(data->conn);
  if(data->set.fopensocket) {
   /*
    * If the opensocket callback is set, all the destination address
    * information is passed to the callback. Depending on this information the
    * callback may opt to abort the connection, this is indicated returning
    * CURL_SOCKET_BAD; otherwise it will return a not-connected socket. When
    * the callback returns a valid socket the destination address information
    * might have been changed and this 'new' address will actually be used
    * here to connect.
    */
    Curl_set_in_callback(data, true);
    *sockfd = data->set.fopensocket(data->set.opensocket_client,
                                    CURLSOCKTYPE_IPCXN,
                                    (struct curl_sockaddr *)addr);
    Curl_set_in_callback(data, false);
  }
  else {
    /* opensocket callback not set, so simply create the socket now */
    *sockfd = socket(addr->family, addr->socktype, addr->protocol);
  }

  if(*sockfd == CURL_SOCKET_BAD)
    /* no socket, no connection */
    return CURLE_COULDNT_CONNECT;

#if defined(USE_IPV6) && defined(HAVE_SOCKADDR_IN6_SIN6_SCOPE_ID)
  if(data->conn->scope_id && (addr->family == AF_INET6)) {
    struct sockaddr_in6 * const sa6 = (void *)&addr->sa_addr;
    sa6->sin6_scope_id = data->conn->scope_id;
  }
#endif
  return CURLE_OK;
}

/*
 * Create a socket based on info from 'conn' and 'ai'.
 *
 * 'addr' should be a pointer to the correct struct to get data back, or NULL.
 * 'sockfd' must be a pointer to a socket descriptor.
 *
 * If the open socket callback is set, used that!
 *
 */
CURLcode Curl_socket_open(struct Curl_easy *data,
                            const struct Curl_addrinfo *ai,
                            struct Curl_sockaddr_ex *addr,
                            int transport,
                            curl_socket_t *sockfd)
{
  struct Curl_sockaddr_ex dummy;

  if(!addr)
    /* if the caller does not want info back, use a local temp copy */
    addr = &dummy;

  Curl_sock_assign_addr(addr, ai, transport);
  return socket_open(data, addr, sockfd);
}

static int socket_close(struct Curl_easy *data, struct connectdata *conn,
                        int use_callback, curl_socket_t sock)
{
  if(CURL_SOCKET_BAD == sock)
    return 0;

  if(use_callback && conn && conn->fclosesocket) {
    int rc;
    Curl_multi_closed(data, sock);
    Curl_set_in_callback(data, true);
    rc = conn->fclosesocket(conn->closesocket_client, sock);
    Curl_set_in_callback(data, false);
    return rc;
  }

  if(conn)
    /* tell the multi-socket code about this */
    Curl_multi_closed(data, sock);

  sclose(sock);

  return 0;
}

/*
 * Close a socket.
 *
 * 'conn' can be NULL, beware!
 */
int Curl_socket_close(struct Curl_easy *data, struct connectdata *conn,
                      curl_socket_t sock)
{
  return socket_close(data, conn, FALSE, sock);
}

#ifdef USE_WINSOCK
/* When you run a program that uses the Windows Sockets API, you may
   experience slow performance when you copy data to a TCP server.

   https://support.microsoft.com/kb/823764

   Work-around: Make the Socket Send Buffer Size Larger Than the Program Send
   Buffer Size

   The problem described in this knowledge-base is applied only to pre-Vista
   Windows. Following function trying to detect OS version and skips
   SO_SNDBUF adjustment for Windows Vista and above.
*/
#define DETECT_OS_NONE 0
#define DETECT_OS_PREVISTA 1
#define DETECT_OS_VISTA_OR_LATER 2

void Curl_sndbuf_init(curl_socket_t sockfd)
{
  int val = CURL_MAX_WRITE_SIZE + 32;
  int curval = 0;
  int curlen = sizeof(curval);

  static int detectOsState = DETECT_OS_NONE;

  if(detectOsState == DETECT_OS_NONE) {
    if(curlx_verify_windows_version(6, 0, 0, PLATFORM_WINNT,
                                    VERSION_GREATER_THAN_EQUAL))
      detectOsState = DETECT_OS_VISTA_OR_LATER;
    else
      detectOsState = DETECT_OS_PREVISTA;
  }

  if(detectOsState == DETECT_OS_VISTA_OR_LATER)
    return;

  if(getsockopt(sockfd, SOL_SOCKET, SO_SNDBUF, (char *)&curval, &curlen) == 0)
    if(curval > val)
      return;

  setsockopt(sockfd, SOL_SOCKET, SO_SNDBUF, (const char *)&val, sizeof(val));
}
#endif /* USE_WINSOCK */

/*
 * Curl_parse_interface()
 *
 * This is used to parse interface argument in the following formats.
 * In all the examples, `host` can be an IP address or a hostname.
 *
 *   <iface_or_host> - can be either an interface name or a host.
 *   if!<iface> - interface name.
 *   host!<host> - hostname.
 *   ifhost!<iface>!<host> - interface name and hostname.
 *
 * Parameters:
 *
 * input  [in]     - input string.
 * len    [in]     - length of the input string.
 * dev    [in/out] - address where a pointer to newly allocated memory
 *                   holding the interface-or-host will be stored upon
 *                   completion.
 * iface  [in/out] - address where a pointer to newly allocated memory
 *                   holding the interface will be stored upon completion.
 * host   [in/out] - address where a pointer to newly allocated memory
 *                   holding the host will be stored upon completion.
 *
 * Returns CURLE_OK on success.
 */
CURLcode Curl_parse_interface(const char *input,
                              char **dev, char **iface, char **host)
{
  static const char if_prefix[] = "if!";
  static const char host_prefix[] = "host!";
  static const char if_host_prefix[] = "ifhost!";
  size_t len;

  DEBUGASSERT(dev);
  DEBUGASSERT(iface);
  DEBUGASSERT(host);

  len = strlen(input);
  if(len > 512)
    return CURLE_BAD_FUNCTION_ARGUMENT;

  if(!strncmp(if_prefix, input, strlen(if_prefix))) {
    input += strlen(if_prefix);
    if(!*input)
      return CURLE_BAD_FUNCTION_ARGUMENT;
    *iface = Curl_memdup0(input, len - strlen(if_prefix));
    return *iface ? CURLE_OK : CURLE_OUT_OF_MEMORY;
  }
  else if(!strncmp(host_prefix, input, strlen(host_prefix))) {
    input += strlen(host_prefix);
    if(!*input)
      return CURLE_BAD_FUNCTION_ARGUMENT;
    *host = Curl_memdup0(input, len - strlen(host_prefix));
    return *host ? CURLE_OK : CURLE_OUT_OF_MEMORY;
  }
  else if(!strncmp(if_host_prefix, input, strlen(if_host_prefix))) {
    const char *host_part;
    input += strlen(if_host_prefix);
    len -= strlen(if_host_prefix);
    host_part = memchr(input, '!', len);
    if(!host_part || !*(host_part + 1))
      return CURLE_BAD_FUNCTION_ARGUMENT;
    *iface = Curl_memdup0(input, host_part - input);
    if(!*iface)
      return CURLE_OUT_OF_MEMORY;
    ++host_part;
    *host = Curl_memdup0(host_part, len - (host_part - input));
    if(!*host) {
      free(*iface);
      *iface = NULL;
      return CURLE_OUT_OF_MEMORY;
    }
    return CURLE_OK;
  }

  if(!*input)
    return CURLE_BAD_FUNCTION_ARGUMENT;
  *dev = Curl_memdup0(input, len);
  return *dev ? CURLE_OK : CURLE_OUT_OF_MEMORY;
}

#ifndef CURL_DISABLE_BINDLOCAL
static CURLcode bindlocal(struct Curl_easy *data, struct connectdata *conn,
                          curl_socket_t sockfd, int af, unsigned int scope)
{
  struct Curl_sockaddr_storage sa;
  struct sockaddr *sock = (struct sockaddr *)&sa;  /* bind to this address */
  curl_socklen_t sizeof_sa = 0; /* size of the data sock points to */
  struct sockaddr_in *si4 = (struct sockaddr_in *)&sa;
#ifdef USE_IPV6
  struct sockaddr_in6 *si6 = (struct sockaddr_in6 *)&sa;
#endif

  struct Curl_dns_entry *h = NULL;
  unsigned short port = data->set.localport; /* use this port number, 0 for
                                                "random" */
  /* how many port numbers to try to bind to, increasing one at a time */
  int portnum = data->set.localportrange;
  const char *dev = data->set.str[STRING_DEVICE];
  const char *iface_input = data->set.str[STRING_INTERFACE];
  const char *host_input = data->set.str[STRING_BINDHOST];
  const char *iface = iface_input ? iface_input : dev;
  const char *host = host_input ? host_input : dev;
  int error;
#ifdef IP_BIND_ADDRESS_NO_PORT
  int on = 1;
#endif
#ifndef USE_IPV6
  (void)scope;
#endif

  /*************************************************************
   * Select device to bind socket to
   *************************************************************/
  if(!iface && !host && !port)
    /* no local kind of binding was requested */
    return CURLE_OK;

  memset(&sa, 0, sizeof(struct Curl_sockaddr_storage));

  if(iface && (strlen(iface)<255) ) {
    char myhost[256] = "";
    int done = 0; /* -1 for error, 1 for address found */
    if2ip_result_t if2ip_result = IF2IP_NOT_FOUND;

    /* interface */
#ifdef SO_BINDTODEVICE
    /*
      * This binds the local socket to a particular interface. This will
      * force even requests to other local interfaces to go out the external
      * interface. Only bind to the interface when specified as interface,
      * not just as a hostname or ip address.
      *
      * The interface might be a VRF, eg: vrf-blue, which means it cannot be
      * converted to an IP address and would fail Curl_if2ip. Simply try to
      * use it straight away.
      */
    if(setsockopt(sockfd, SOL_SOCKET, SO_BINDTODEVICE,
                  iface, (curl_socklen_t)strlen(iface) + 1) == 0) {
      /* This is often "errno 1, error: Operation not permitted" if you are
        * not running as root or another suitable privileged user. If it
        * succeeds it means the parameter was a valid interface and not an IP
        * address. Return immediately.
        */
      if(!host_input) {
        infof(data, "socket successfully bound to interface '%s'", iface);
        return CURLE_OK;
      }
    }
#endif
    if(!host_input) {
      /* Discover IP from input device, then bind to it */
      if2ip_result = Curl_if2ip(af,
#ifdef USE_IPV6
                      scope, conn->scope_id,
#endif
                      iface, myhost, sizeof(myhost));
    }
    switch(if2ip_result) {
      case IF2IP_NOT_FOUND:
        if(iface_input && !host_input) {
          /* Do not fall back to treating it as a hostname */
          char buffer[STRERROR_LEN];
          data->state.os_errno = error = SOCKERRNO;
          failf(data, "Couldn't bind to interface '%s' with errno %d: %s",
                iface, error, Curl_strerror(error, buffer, sizeof(buffer)));
          return CURLE_INTERFACE_FAILED;
        }
        break;
      case IF2IP_AF_NOT_SUPPORTED:
        /* Signal the caller to try another address family if available */
        return CURLE_UNSUPPORTED_PROTOCOL;
      case IF2IP_FOUND:
        /*
          * We now have the numerical IP address in the 'myhost' buffer
          */
        host = myhost;
        infof(data, "Local Interface %s is ip %s using address family %i",
              iface, host, af);
        done = 1;
        break;
    }
    if(!iface_input || host_input) {
      /*
       * This was not an interface, resolve the name as a hostname
       * or IP number
       *
       * Temporarily force name resolution to use only the address type
       * of the connection. The resolve functions should really be changed
       * to take a type parameter instead.
       */
      unsigned char ipver = conn->ip_version;
      int rc;

      if(af == AF_INET)
        conn->ip_version = CURL_IPRESOLVE_V4;
#ifdef USE_IPV6
      else if(af == AF_INET6)
        conn->ip_version = CURL_IPRESOLVE_V6;
#endif

      rc = Curl_resolv(data, host, 80, FALSE, &h);
      if(rc == CURLRESOLV_PENDING)
        (void)Curl_resolver_wait_resolv(data, &h);
      conn->ip_version = ipver;

      if(h) {
        int h_af = h->addr->ai_family;
        /* convert the resolved address, sizeof myhost >= INET_ADDRSTRLEN */
        Curl_printable_address(h->addr, myhost, sizeof(myhost));
        infof(data, "Name '%s' family %i resolved to '%s' family %i",
              host, af, myhost, h_af);
        Curl_resolv_unlink(data, &h); /* this will NULL, potential free h */
        if(af != h_af) {
          /* bad IP version combo, signal the caller to try another address
             family if available */
          return CURLE_UNSUPPORTED_PROTOCOL;
        }
        done = 1;
      }
      else {
        /*
         * provided dev was no interface (or interfaces are not supported
         * e.g. Solaris) no ip address and no domain we fail here
         */
        done = -1;
      }
    }

    if(done > 0) {
#ifdef USE_IPV6
      /* IPv6 address */
      if(af == AF_INET6) {
#ifdef HAVE_SOCKADDR_IN6_SIN6_SCOPE_ID
        char *scope_ptr = strchr(myhost, '%');
        if(scope_ptr)
          *(scope_ptr++) = '\0';
#endif
        if(Curl_inet_pton(AF_INET6, myhost, &si6->sin6_addr) > 0) {
          si6->sin6_family = AF_INET6;
          si6->sin6_port = htons(port);
#ifdef HAVE_SOCKADDR_IN6_SIN6_SCOPE_ID
          if(scope_ptr) {
            /* The "myhost" string either comes from Curl_if2ip or from
               Curl_printable_address. The latter returns only numeric scope
               IDs and the former returns none at all. So the scope ID, if
               present, is known to be numeric */
            unsigned long scope_id = strtoul(scope_ptr, NULL, 10);
            if(scope_id > UINT_MAX)
              return CURLE_UNSUPPORTED_PROTOCOL;

            si6->sin6_scope_id = (unsigned int)scope_id;
          }
#endif
        }
        sizeof_sa = sizeof(struct sockaddr_in6);
      }
      else
#endif
      /* IPv4 address */
      if((af == AF_INET) &&
         (Curl_inet_pton(AF_INET, myhost, &si4->sin_addr) > 0)) {
        si4->sin_family = AF_INET;
        si4->sin_port = htons(port);
        sizeof_sa = sizeof(struct sockaddr_in);
      }
    }

    if(done < 1) {
      /* errorbuf is set false so failf will overwrite any message already in
         the error buffer, so the user receives this error message instead of a
         generic resolve error. */
      char buffer[STRERROR_LEN];
      data->state.errorbuf = FALSE;
      data->state.os_errno = error = SOCKERRNO;
      failf(data, "Couldn't bind to '%s' with errno %d: %s",
            host, error, Curl_strerror(error, buffer, sizeof(buffer)));
      return CURLE_INTERFACE_FAILED;
    }
  }
  else {
    /* no device was given, prepare sa to match af's needs */
#ifdef USE_IPV6
    if(af == AF_INET6) {
      si6->sin6_family = AF_INET6;
      si6->sin6_port = htons(port);
      sizeof_sa = sizeof(struct sockaddr_in6);
    }
    else
#endif
    if(af == AF_INET) {
      si4->sin_family = AF_INET;
      si4->sin_port = htons(port);
      sizeof_sa = sizeof(struct sockaddr_in);
    }
  }
#ifdef IP_BIND_ADDRESS_NO_PORT
  (void)setsockopt(sockfd, SOL_IP, IP_BIND_ADDRESS_NO_PORT, &on, sizeof(on));
#endif
  for(;;) {
    if(bind(sockfd, sock, sizeof_sa) >= 0) {
      /* we succeeded to bind */
      infof(data, "Local port: %hu", port);
      conn->bits.bound = TRUE;
      return CURLE_OK;
    }

    if(--portnum > 0) {
      port++; /* try next port */
      if(port == 0)
        break;
      infof(data, "Bind to local port %d failed, trying next", port - 1);
      /* We reuse/clobber the port variable here below */
      if(sock->sa_family == AF_INET)
        si4->sin_port = ntohs(port);
#ifdef USE_IPV6
      else
        si6->sin6_port = ntohs(port);
#endif
    }
    else
      break;
  }
  {
    char buffer[STRERROR_LEN];
    data->state.os_errno = error = SOCKERRNO;
    failf(data, "bind failed with errno %d: %s",
          error, Curl_strerror(error, buffer, sizeof(buffer)));
  }

  return CURLE_INTERFACE_FAILED;
}
#endif

/*
 * verifyconnect() returns TRUE if the connect really has happened.
 */
static bool verifyconnect(curl_socket_t sockfd, int *error)
{
  bool rc = TRUE;
#ifdef SO_ERROR
  int err = 0;
  curl_socklen_t errSize = sizeof(err);

#ifdef _WIN32
  /*
   * In October 2003 we effectively nullified this function on Windows due to
   * problems with it using all CPU in multi-threaded cases.
   *
   * In May 2004, we bring it back to offer more info back on connect failures.
   * Gisle Vanem could reproduce the former problems with this function, but
   * could avoid them by adding this SleepEx() call below:
   *
   *    "I do not have Rational Quantify, but the hint from his post was
   *    ntdll::NtRemoveIoCompletion(). I would assume the SleepEx (or maybe
   *    just Sleep(0) would be enough?) would release whatever
   *    mutex/critical-section the ntdll call is waiting on.
   *
   *    Someone got to verify this on Win-NT 4.0, 2000."
   */

#ifdef _WIN32_WCE
  Sleep(0);
#else
  SleepEx(0, FALSE);
#endif

#endif

  if(0 != getsockopt(sockfd, SOL_SOCKET, SO_ERROR, (void *)&err, &errSize))
    err = SOCKERRNO;
#ifdef _WIN32_WCE
  /* Old Windows CE versions do not support SO_ERROR */
  if(WSAENOPROTOOPT == err) {
    SET_SOCKERRNO(0);
    err = 0;
  }
#endif
#if defined(EBADIOCTL) && defined(__minix)
  /* Minix 3.1.x does not support getsockopt on UDP sockets */
  if(EBADIOCTL == err) {
    SET_SOCKERRNO(0);
    err = 0;
  }
#endif
  if((0 == err) || (EISCONN == err))
    /* we are connected, awesome! */
    rc = TRUE;
  else
    /* This was not a successful connect */
    rc = FALSE;
  if(error)
    *error = err;
#else
  (void)sockfd;
  if(error)
    *error = SOCKERRNO;
#endif
  return rc;
}

/**
 * Determine the curl code for a socket connect() == -1 with errno.
 */
static CURLcode socket_connect_result(struct Curl_easy *data,
                                      const char *ipaddress, int error)
{
  switch(error) {
  case EINPROGRESS:
  case EWOULDBLOCK:
#if defined(EAGAIN)
#if (EAGAIN) != (EWOULDBLOCK)
    /* On some platforms EAGAIN and EWOULDBLOCK are the
     * same value, and on others they are different, hence
     * the odd #if
     */
  case EAGAIN:
#endif
#endif
    return CURLE_OK;

  default:
    /* unknown error, fallthrough and try another address! */
#ifdef CURL_DISABLE_VERBOSE_STRINGS
    (void)ipaddress;
#else
    {
      char buffer[STRERROR_LEN];
      infof(data, "Immediate connect fail for %s: %s",
            ipaddress, Curl_strerror(error, buffer, sizeof(buffer)));
    }
#endif
    data->state.os_errno = error;
    /* connect failed */
    return CURLE_COULDNT_CONNECT;
  }
}

/* We have a recv buffer to enhance reads with len < NW_SMALL_READS.
 * This happens often on TLS connections where the TLS implementation
 * tries to read the head of a TLS record, determine the length of the
 * full record and then make a subsequent read for that.
 * On large reads, we will not fill the buffer to avoid the double copy. */
#define NW_RECV_CHUNK_SIZE    (64 * 1024)
#define NW_RECV_CHUNKS         1
#define NW_SMALL_READS        (1024)

struct cf_socket_ctx {
  int transport;
  struct Curl_sockaddr_ex addr;      /* address to connect to */
  curl_socket_t sock;                /* current attempt socket */
  struct ip_quadruple ip;            /* The IP quadruple 2x(addr+port) */
  struct curltime started_at;        /* when socket was created */
  struct curltime connected_at;      /* when socket connected/got first byte */
  struct curltime first_byte_at;     /* when first byte was recvd */
#ifdef USE_WINSOCK
  struct curltime last_sndbuf_query_at;  /* when SO_SNDBUF last queried */
  ULONG sndbuf_size;                     /* the last set SO_SNDBUF size */
#endif
  int error;                         /* errno of last failure or 0 */
#ifdef DEBUGBUILD
  int wblock_percent;                /* percent of writes doing EAGAIN */
  int wpartial_percent;              /* percent of bytes written in send */
  int rblock_percent;                /* percent of reads doing EAGAIN */
  size_t recv_max;                  /* max enforced read size */
#endif
  BIT(got_first_byte);               /* if first byte was received */
  BIT(listening);                    /* socket is listening */
  BIT(accepted);                     /* socket was accepted, not connected */
  BIT(sock_connected);               /* socket is "connected", e.g. in UDP */
  BIT(active);
};

static void cf_socket_ctx_init(struct cf_socket_ctx *ctx,
                               const struct Curl_addrinfo *ai,
                               int transport)
{
  memset(ctx, 0, sizeof(*ctx));
  ctx->sock = CURL_SOCKET_BAD;
  ctx->transport = transport;
  Curl_sock_assign_addr(&ctx->addr, ai, transport);
#ifdef DEBUGBUILD
  {
    char *p = getenv("CURL_DBG_SOCK_WBLOCK");
    if(p) {
      long l = strtol(p, NULL, 10);
      if(l >= 0 && l <= 100)
        ctx->wblock_percent = (int)l;
    }
    p = getenv("CURL_DBG_SOCK_WPARTIAL");
    if(p) {
      long l = strtol(p, NULL, 10);
      if(l >= 0 && l <= 100)
        ctx->wpartial_percent = (int)l;
    }
    p = getenv("CURL_DBG_SOCK_RBLOCK");
    if(p) {
      long l = strtol(p, NULL, 10);
      if(l >= 0 && l <= 100)
        ctx->rblock_percent = (int)l;
    }
    p = getenv("CURL_DBG_SOCK_RMAX");
    if(p) {
      long l = strtol(p, NULL, 10);
      if(l >= 0)
        ctx->recv_max = (size_t)l;
    }
  }
#endif
}

static void cf_socket_close(struct Curl_cfilter *cf, struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  if(ctx && CURL_SOCKET_BAD != ctx->sock) {
    CURL_TRC_CF(data, cf, "cf_socket_close(%" FMT_SOCKET_T ")", ctx->sock);
    if(ctx->sock == cf->conn->sock[cf->sockindex])
      cf->conn->sock[cf->sockindex] = CURL_SOCKET_BAD;
    socket_close(data, cf->conn, !ctx->accepted, ctx->sock);
    ctx->sock = CURL_SOCKET_BAD;
    if(ctx->active && cf->sockindex == FIRSTSOCKET)
      cf->conn->remote_addr = NULL;
    ctx->active = FALSE;
    memset(&ctx->started_at, 0, sizeof(ctx->started_at));
    memset(&ctx->connected_at, 0, sizeof(ctx->connected_at));
  }

  cf->connected = FALSE;
}

static CURLcode cf_socket_shutdown(struct Curl_cfilter *cf,
                                   struct Curl_easy *data,
                                   bool *done)
{
  if(cf->connected) {
    struct cf_socket_ctx *ctx = cf->ctx;

    CURL_TRC_CF(data, cf, "cf_socket_shutdown(%" FMT_SOCKET_T ")", ctx->sock);
    /* On TCP, and when the socket looks well and non-blocking mode
     * can be enabled, receive dangling bytes before close to avoid
     * entering RST states unnecessarily. */
    if(ctx->sock != CURL_SOCKET_BAD &&
       ctx->transport == TRNSPRT_TCP &&
       (curlx_nonblock(ctx->sock, TRUE) >= 0)) {
      unsigned char buf[1024];
      (void)sread(ctx->sock, buf, sizeof(buf));
    }
  }
  *done = TRUE;
  return CURLE_OK;
}

static void cf_socket_destroy(struct Curl_cfilter *cf, struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  cf_socket_close(cf, data);
  CURL_TRC_CF(data, cf, "destroy");
  free(ctx);
  cf->ctx = NULL;
}

static CURLcode set_local_ip(struct Curl_cfilter *cf,
                             struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;

#ifdef HAVE_GETSOCKNAME
  if((ctx->sock != CURL_SOCKET_BAD) &&
     !(data->conn->handler->protocol & CURLPROTO_TFTP)) {
    /* TFTP does not connect, so it cannot get the IP like this */

    char buffer[STRERROR_LEN];
    struct Curl_sockaddr_storage ssloc;
    curl_socklen_t slen = sizeof(struct Curl_sockaddr_storage);

    memset(&ssloc, 0, sizeof(ssloc));
    if(getsockname(ctx->sock, (struct sockaddr*) &ssloc, &slen)) {
      int error = SOCKERRNO;
      failf(data, "getsockname() failed with errno %d: %s",
            error, Curl_strerror(error, buffer, sizeof(buffer)));
      return CURLE_FAILED_INIT;
    }
    if(!Curl_addr2string((struct sockaddr*)&ssloc, slen,
                         ctx->ip.local_ip, &ctx->ip.local_port)) {
      failf(data, "ssloc inet_ntop() failed with errno %d: %s",
            errno, Curl_strerror(errno, buffer, sizeof(buffer)));
      return CURLE_FAILED_INIT;
    }
  }
#else
  (void)data;
  ctx->ip.local_ip[0] = 0;
  ctx->ip.local_port = -1;
#endif
  return CURLE_OK;
}

static CURLcode set_remote_ip(struct Curl_cfilter *cf,
                              struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  /* store remote address and port used in this connection attempt */
  if(!Curl_addr2string(&ctx->addr.sa_addr, (curl_socklen_t)ctx->addr.addrlen,
                       ctx->ip.remote_ip, &ctx->ip.remote_port)) {
    char buffer[STRERROR_LEN];

    ctx->error = errno;
    /* malformed address or bug in inet_ntop, try next address */
    failf(data, "sa_addr inet_ntop() failed with errno %d: %s",
          errno, Curl_strerror(errno, buffer, sizeof(buffer)));
    return CURLE_FAILED_INIT;
  }
  return CURLE_OK;
}

static CURLcode cf_socket_open(struct Curl_cfilter *cf,
                              struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  int error = 0;
  bool isconnected = FALSE;
  CURLcode result = CURLE_COULDNT_CONNECT;
  bool is_tcp;

  (void)data;
  DEBUGASSERT(ctx->sock == CURL_SOCKET_BAD);
  ctx->started_at = Curl_now();
#ifdef SOCK_NONBLOCK
  /* Do not tuck SOCK_NONBLOCK into socktype when opensocket callback is set
   * because we would not know how socketype is about to be used in the
   * callback, SOCK_NONBLOCK might get factored out before calling socket().
   */
  if(!data->set.fopensocket)
    ctx->addr.socktype |= SOCK_NONBLOCK;
#endif
  result = socket_open(data, &ctx->addr, &ctx->sock);
#ifdef SOCK_NONBLOCK
  /* Restore the socktype after the socket is created. */
  if(!data->set.fopensocket)
    ctx->addr.socktype &= ~SOCK_NONBLOCK;
#endif
  if(result)
    goto out;

  result = set_remote_ip(cf, data);
  if(result)
    goto out;

#ifdef USE_IPV6
  if(ctx->addr.family == AF_INET6) {
    set_ipv6_v6only(ctx->sock, 0);
    infof(data, "  Trying [%s]:%d...", ctx->ip.remote_ip, ctx->ip.remote_port);
  }
  else
#endif
    infof(data, "  Trying %s:%d...", ctx->ip.remote_ip, ctx->ip.remote_port);

#ifdef USE_IPV6
  is_tcp = (ctx->addr.family == AF_INET
            || ctx->addr.family == AF_INET6) &&
           ctx->addr.socktype == SOCK_STREAM;
#else
  is_tcp = (ctx->addr.family == AF_INET) &&
           ctx->addr.socktype == SOCK_STREAM;
#endif
  if(is_tcp && data->set.tcp_nodelay)
    tcpnodelay(data, ctx->sock);

  nosigpipe(data, ctx->sock);

  Curl_sndbuf_init(ctx->sock);

  if(is_tcp && data->set.tcp_keepalive)
    tcpkeepalive(data, ctx->sock);

  if(data->set.fsockopt) {
    /* activate callback for setting socket options */
    Curl_set_in_callback(data, true);
    error = data->set.fsockopt(data->set.sockopt_client,
                               ctx->sock,
                               CURLSOCKTYPE_IPCXN);
    Curl_set_in_callback(data, false);

    if(error == CURL_SOCKOPT_ALREADY_CONNECTED)
      isconnected = TRUE;
    else if(error) {
      result = CURLE_ABORTED_BY_CALLBACK;
      goto out;
    }
  }

#ifndef CURL_DISABLE_BINDLOCAL
  /* possibly bind the local end to an IP, interface or port */
  if(ctx->addr.family == AF_INET
#ifdef USE_IPV6
     || ctx->addr.family == AF_INET6
#endif
    ) {
    result = bindlocal(data, cf->conn, ctx->sock, ctx->addr.family,
                       Curl_ipv6_scope(&ctx->addr.sa_addr));
    if(result) {
      if(result == CURLE_UNSUPPORTED_PROTOCOL) {
        /* The address family is not supported on this interface.
           We can continue trying addresses */
        result = CURLE_COULDNT_CONNECT;
      }
      goto out;
    }
  }
#endif

#ifndef SOCK_NONBLOCK
  /* Set socket non-blocking, must be a non-blocking socket for
   * a non-blocking connect. */
  error = curlx_nonblock(ctx->sock, TRUE);
  if(error < 0) {
    result = CURLE_UNSUPPORTED_PROTOCOL;
    ctx->error = SOCKERRNO;
    goto out;
  }
#else
  if(data->set.fopensocket) {
    /* Set socket non-blocking, must be a non-blocking socket for
     * a non-blocking connect. */
    error = curlx_nonblock(ctx->sock, TRUE);
    if(error < 0) {
      result = CURLE_UNSUPPORTED_PROTOCOL;
      ctx->error = SOCKERRNO;
      goto out;
    }
  }
#endif
  ctx->sock_connected = (ctx->addr.socktype != SOCK_DGRAM);
out:
  if(result) {
    if(ctx->sock != CURL_SOCKET_BAD) {
      socket_close(data, cf->conn, TRUE, ctx->sock);
      ctx->sock = CURL_SOCKET_BAD;
    }
  }
  else if(isconnected) {
    set_local_ip(cf, data);
    ctx->connected_at = Curl_now();
    cf->connected = TRUE;
  }
  CURL_TRC_CF(data, cf, "cf_socket_open() -> %d, fd=%" FMT_SOCKET_T,
              result, ctx->sock);
  return result;
}

static int do_connect(struct Curl_cfilter *cf, struct Curl_easy *data,
                      bool is_tcp_fastopen)
{
  struct cf_socket_ctx *ctx = cf->ctx;
#ifdef TCP_FASTOPEN_CONNECT
  int optval = 1;
#endif
  int rc = -1;

  (void)data;
  if(is_tcp_fastopen) {
#if defined(CONNECT_DATA_IDEMPOTENT) /* Darwin */
#  if defined(HAVE_BUILTIN_AVAILABLE)
    /* while connectx function is available since macOS 10.11 / iOS 9,
       it did not have the interface declared correctly until
       Xcode 9 / macOS SDK 10.13 */
    if(__builtin_available(macOS 10.11, iOS 9.0, tvOS 9.0, watchOS 2.0, *)) {
      sa_endpoints_t endpoints;
      endpoints.sae_srcif = 0;
      endpoints.sae_srcaddr = NULL;
      endpoints.sae_srcaddrlen = 0;
      endpoints.sae_dstaddr = &ctx->addr.sa_addr;
      endpoints.sae_dstaddrlen = ctx->addr.addrlen;

      rc = connectx(ctx->sock, &endpoints, SAE_ASSOCID_ANY,
                    CONNECT_RESUME_ON_READ_WRITE | CONNECT_DATA_IDEMPOTENT,
                    NULL, 0, NULL, NULL);
    }
    else {
      rc = connect(ctx->sock, &ctx->addr.sa_addr, ctx->addr.addrlen);
    }
#  else
    rc = connect(ctx->sock, &ctx->addr.sa_addr, ctx->addr.addrlen);
#  endif /* HAVE_BUILTIN_AVAILABLE */
#elif defined(TCP_FASTOPEN_CONNECT) /* Linux >= 4.11 */
    if(setsockopt(ctx->sock, IPPROTO_TCP, TCP_FASTOPEN_CONNECT,
                  (void *)&optval, sizeof(optval)) < 0)
      infof(data, "Failed to enable TCP Fast Open on fd %" FMT_SOCKET_T,
            ctx->sock);

    rc = connect(ctx->sock, &ctx->addr.sa_addr, ctx->addr.addrlen);
#elif defined(MSG_FASTOPEN) /* old Linux */
    if(cf->conn->given->flags & PROTOPT_SSL)
      rc = connect(ctx->sock, &ctx->addr.sa_addr, ctx->addr.addrlen);
    else
      rc = 0; /* Do nothing */
#endif
  }
  else {
    rc = connect(ctx->sock, &ctx->addr.sa_addr,
                 (curl_socklen_t)ctx->addr.addrlen);
  }
  return rc;
}

static CURLcode cf_tcp_connect(struct Curl_cfilter *cf,
                               struct Curl_easy *data,
                               bool blocking, bool *done)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  CURLcode result = CURLE_COULDNT_CONNECT;
  int rc = 0;

  (void)data;
  if(cf->connected) {
    *done = TRUE;
    return CURLE_OK;
  }

  /* TODO: need to support blocking connect? */
  if(blocking)
    return CURLE_UNSUPPORTED_PROTOCOL;

  *done = FALSE; /* a very negative world view is best */
  if(ctx->sock == CURL_SOCKET_BAD) {
    int error;

    result = cf_socket_open(cf, data);
    if(result)
      goto out;

    if(cf->connected) {
      *done = TRUE;
      return CURLE_OK;
    }

    /* Connect TCP socket */
    rc = do_connect(cf, data, cf->conn->bits.tcp_fastopen);
    error = SOCKERRNO;
    set_local_ip(cf, data);
    CURL_TRC_CF(data, cf, "local address %s port %d...",
                ctx->ip.local_ip, ctx->ip.local_port);
    if(-1 == rc) {
      result = socket_connect_result(data, ctx->ip.remote_ip, error);
      goto out;
    }
  }

#ifdef mpeix
  /* Call this function once now, and ignore the results. We do this to
     "clear" the error state on the socket so that we can later read it
     reliably. This is reported necessary on the MPE/iX operating
     system. */
  (void)verifyconnect(ctx->sock, NULL);
#endif
  /* check socket for connect */
  rc = SOCKET_WRITABLE(ctx->sock, 0);

  if(rc == 0) { /* no connection yet */
    CURL_TRC_CF(data, cf, "not connected yet");
    return CURLE_OK;
  }
  else if(rc == CURL_CSELECT_OUT || cf->conn->bits.tcp_fastopen) {
    if(verifyconnect(ctx->sock, &ctx->error)) {
      /* we are connected with TCP, awesome! */
      ctx->connected_at = Curl_now();
      set_local_ip(cf, data);
      *done = TRUE;
      cf->connected = TRUE;
      CURL_TRC_CF(data, cf, "connected");
      return CURLE_OK;
    }
  }
  else if(rc & CURL_CSELECT_ERR) {
    (void)verifyconnect(ctx->sock, &ctx->error);
    result = CURLE_COULDNT_CONNECT;
  }

out:
  if(result) {
    if(ctx->error) {
      set_local_ip(cf, data);
      data->state.os_errno = ctx->error;
      SET_SOCKERRNO(ctx->error);
#ifndef CURL_DISABLE_VERBOSE_STRINGS
      {
        char buffer[STRERROR_LEN];
        infof(data, "connect to %s port %u from %s port %d failed: %s",
              ctx->ip.remote_ip, ctx->ip.remote_port,
              ctx->ip.local_ip, ctx->ip.local_port,
              Curl_strerror(ctx->error, buffer, sizeof(buffer)));
      }
#endif
    }
    if(ctx->sock != CURL_SOCKET_BAD) {
      socket_close(data, cf->conn, TRUE, ctx->sock);
      ctx->sock = CURL_SOCKET_BAD;
    }
    *done = FALSE;
  }
  return result;
}

static void cf_socket_get_host(struct Curl_cfilter *cf,
                               struct Curl_easy *data,
                               const char **phost,
                               const char **pdisplay_host,
                               int *pport)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  (void)data;
  *phost = cf->conn->host.name;
  *pdisplay_host = cf->conn->host.dispname;
  *pport = ctx->ip.remote_port;
}

static void cf_socket_adjust_pollset(struct Curl_cfilter *cf,
                                      struct Curl_easy *data,
                                      struct easy_pollset *ps)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  if(ctx->sock != CURL_SOCKET_BAD) {
    /* A listening socket filter needs to be connected before the accept
     * for some weird FTP interaction. This should be rewritten, so that
     * FTP no longer does the socket checks and accept calls and delegates
     * all that to the filter. TODO. */
    if(ctx->listening) {
      Curl_pollset_set_in_only(data, ps, ctx->sock);
      CURL_TRC_CF(data, cf, "adjust_pollset, listening, POLLIN fd=%"
                  FMT_SOCKET_T, ctx->sock);
    }
    else if(!cf->connected) {
      Curl_pollset_set_out_only(data, ps, ctx->sock);
      CURL_TRC_CF(data, cf, "adjust_pollset, !connected, POLLOUT fd=%"
                  FMT_SOCKET_T, ctx->sock);
    }
    else if(!ctx->active) {
      Curl_pollset_add_in(data, ps, ctx->sock);
      CURL_TRC_CF(data, cf, "adjust_pollset, !active, POLLIN fd=%"
                  FMT_SOCKET_T, ctx->sock);
    }
  }
}

static bool cf_socket_data_pending(struct Curl_cfilter *cf,
                                   const struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  int readable;

  (void)data;
  readable = SOCKET_READABLE(ctx->sock, 0);
  return (readable > 0 && (readable & CURL_CSELECT_IN));
}

#ifdef USE_WINSOCK

#ifndef SIO_IDEAL_SEND_BACKLOG_QUERY
#define SIO_IDEAL_SEND_BACKLOG_QUERY 0x4004747B
#endif

static void win_update_sndbuf_size(struct cf_socket_ctx *ctx)
{
  ULONG ideal;
  DWORD ideallen;
  struct curltime n = Curl_now();

  if(Curl_timediff(n, ctx->last_sndbuf_query_at) > 1000) {
    if(!WSAIoctl(ctx->sock, SIO_IDEAL_SEND_BACKLOG_QUERY, 0, 0,
                  &ideal, sizeof(ideal), &ideallen, 0, 0) &&
       ideal != ctx->sndbuf_size &&
       !setsockopt(ctx->sock, SOL_SOCKET, SO_SNDBUF,
                   (const char *)&ideal, sizeof(ideal))) {
      ctx->sndbuf_size = ideal;
    }
    ctx->last_sndbuf_query_at = n;
  }
}

#endif /* USE_WINSOCK */

static ssize_t cf_socket_send(struct Curl_cfilter *cf, struct Curl_easy *data,
                              const void *buf, size_t len, bool eos,
                              CURLcode *err)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  curl_socket_t fdsave;
  ssize_t nwritten;
  size_t orig_len = len;

  (void)eos; /* unused */
  *err = CURLE_OK;
  fdsave = cf->conn->sock[cf->sockindex];
  cf->conn->sock[cf->sockindex] = ctx->sock;

#ifdef DEBUGBUILD
  /* simulate network blocking/partial writes */
  if(ctx->wblock_percent > 0) {
    unsigned char c = 0;
    Curl_rand_bytes(data, FALSE, &c, 1);
    if(c >= ((100-ctx->wblock_percent)*256/100)) {
      CURL_TRC_CF(data, cf, "send(len=%zu) SIMULATE EWOULDBLOCK", orig_len);
      *err = CURLE_AGAIN;
      nwritten = -1;
      cf->conn->sock[cf->sockindex] = fdsave;
      return nwritten;
    }
  }
  if(cf->cft != &Curl_cft_udp && ctx->wpartial_percent > 0 && len > 8) {
    len = len * ctx->wpartial_percent / 100;
    if(!len)
      len = 1;
    CURL_TRC_CF(data, cf, "send(len=%zu) SIMULATE partial write of %zu bytes",
                orig_len, len);
  }
#endif

#if defined(MSG_FASTOPEN) && !defined(TCP_FASTOPEN_CONNECT) /* Linux */
  if(cf->conn->bits.tcp_fastopen) {
    nwritten = sendto(ctx->sock, buf, len, MSG_FASTOPEN,
                      &cf->conn->remote_addr->sa_addr,
                      cf->conn->remote_addr->addrlen);
    cf->conn->bits.tcp_fastopen = FALSE;
  }
  else
#endif
    nwritten = swrite(ctx->sock, buf, len);

  if(-1 == nwritten) {
    int sockerr = SOCKERRNO;

    if(
#ifdef WSAEWOULDBLOCK
      /* This is how Windows does it */
      (WSAEWOULDBLOCK == sockerr)
#else
      /* errno may be EWOULDBLOCK or on some systems EAGAIN when it returned
         due to its inability to send off data without blocking. We therefore
         treat both error codes the same here */
      (EWOULDBLOCK == sockerr) || (EAGAIN == sockerr) || (EINTR == sockerr) ||
      (EINPROGRESS == sockerr)
#endif
      ) {
      /* this is just a case of EWOULDBLOCK */
      *err = CURLE_AGAIN;
    }
    else {
      char buffer[STRERROR_LEN];
      failf(data, "Send failure: %s",
            Curl_strerror(sockerr, buffer, sizeof(buffer)));
      data->state.os_errno = sockerr;
      *err = CURLE_SEND_ERROR;
    }
  }

#if defined(USE_WINSOCK)
  if(!*err)
    win_update_sndbuf_size(ctx);
#endif

  CURL_TRC_CF(data, cf, "send(len=%zu) -> %d, err=%d",
              orig_len, (int)nwritten, *err);
  cf->conn->sock[cf->sockindex] = fdsave;
  return nwritten;
}

static ssize_t cf_socket_recv(struct Curl_cfilter *cf, struct Curl_easy *data,
                              char *buf, size_t len, CURLcode *err)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  ssize_t nread;

  *err = CURLE_OK;

#ifdef DEBUGBUILD
  /* simulate network blocking/partial reads */
  if(cf->cft != &Curl_cft_udp && ctx->rblock_percent > 0) {
    unsigned char c = 0;
    Curl_rand(data, &c, 1);
    if(c >= ((100-ctx->rblock_percent)*256/100)) {
      CURL_TRC_CF(data, cf, "recv(len=%zu) SIMULATE EWOULDBLOCK", len);
      *err = CURLE_AGAIN;
      return -1;
    }
  }
  if(cf->cft != &Curl_cft_udp && ctx->recv_max && ctx->recv_max < len) {
    size_t orig_len = len;
    len = ctx->recv_max;
    CURL_TRC_CF(data, cf, "recv(len=%zu) SIMULATE max read of %zu bytes",
                orig_len, len);
  }
#endif

  *err = CURLE_OK;
  nread = sread(ctx->sock, buf, len);

  if(-1 == nread) {
    int sockerr = SOCKERRNO;

    if(
#ifdef WSAEWOULDBLOCK
      /* This is how Windows does it */
      (WSAEWOULDBLOCK == sockerr)
#else
      /* errno may be EWOULDBLOCK or on some systems EAGAIN when it returned
         due to its inability to send off data without blocking. We therefore
         treat both error codes the same here */
      (EWOULDBLOCK == sockerr) || (EAGAIN == sockerr) || (EINTR == sockerr)
#endif
      ) {
      /* this is just a case of EWOULDBLOCK */
      *err = CURLE_AGAIN;
    }
    else {
      char buffer[STRERROR_LEN];

      failf(data, "Recv failure: %s",
            Curl_strerror(sockerr, buffer, sizeof(buffer)));
      data->state.os_errno = sockerr;
      *err = CURLE_RECV_ERROR;
    }
  }

  CURL_TRC_CF(data, cf, "recv(len=%zu) -> %d, err=%d", len, (int)nread,
              *err);
  if(nread > 0 && !ctx->got_first_byte) {
    ctx->first_byte_at = Curl_now();
    ctx->got_first_byte = TRUE;
  }
  return nread;
}

static void cf_socket_update_data(struct Curl_cfilter *cf,
                                  struct Curl_easy *data)
{
  /* Update the IP info held in the transfer, if we have that. */
  if(cf->connected && (cf->sockindex == FIRSTSOCKET)) {
    struct cf_socket_ctx *ctx = cf->ctx;
    data->info.primary = ctx->ip;
    /* not sure if this is redundant... */
    data->info.conn_remote_port = cf->conn->remote_port;
  }
}

static void cf_socket_active(struct Curl_cfilter *cf, struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  /* use this socket from now on */
  cf->conn->sock[cf->sockindex] = ctx->sock;
  set_local_ip(cf, data);
  if(cf->sockindex == FIRSTSOCKET) {
    cf->conn->primary = ctx->ip;
    cf->conn->remote_addr = &ctx->addr;
  #ifdef USE_IPV6
    cf->conn->bits.ipv6 = (ctx->addr.family == AF_INET6)? TRUE : FALSE;
  #endif
  }
  else {
    cf->conn->secondary = ctx->ip;
  }
  ctx->active = TRUE;
}

static CURLcode cf_socket_cntrl(struct Curl_cfilter *cf,
                                struct Curl_easy *data,
                                int event, int arg1, void *arg2)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  (void)arg1;
  (void)arg2;
  switch(event) {
  case CF_CTRL_CONN_INFO_UPDATE:
    cf_socket_active(cf, data);
    cf_socket_update_data(cf, data);
    break;
  case CF_CTRL_DATA_SETUP:
    cf_socket_update_data(cf, data);
    break;
  case CF_CTRL_FORGET_SOCKET:
    ctx->sock = CURL_SOCKET_BAD;
    break;
  }
  return CURLE_OK;
}

static bool cf_socket_conn_is_alive(struct Curl_cfilter *cf,
                                    struct Curl_easy *data,
                                    bool *input_pending)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  struct pollfd pfd[1];
  int r;

  *input_pending = FALSE;
  (void)data;
  if(!ctx || ctx->sock == CURL_SOCKET_BAD)
    return FALSE;

  /* Check with 0 timeout if there are any events pending on the socket */
  pfd[0].fd = ctx->sock;
  pfd[0].events = POLLRDNORM|POLLIN|POLLRDBAND|POLLPRI;
  pfd[0].revents = 0;

  r = Curl_poll(pfd, 1, 0);
  if(r < 0) {
    CURL_TRC_CF(data, cf, "is_alive: poll error, assume dead");
    return FALSE;
  }
  else if(r == 0) {
    CURL_TRC_CF(data, cf, "is_alive: poll timeout, assume alive");
    return TRUE;
  }
  else if(pfd[0].revents & (POLLERR|POLLHUP|POLLPRI|POLLNVAL)) {
    CURL_TRC_CF(data, cf, "is_alive: err/hup/etc events, assume dead");
    return FALSE;
  }

  CURL_TRC_CF(data, cf, "is_alive: valid events, looks alive");
  *input_pending = TRUE;
  return TRUE;
}

static CURLcode cf_socket_query(struct Curl_cfilter *cf,
                                struct Curl_easy *data,
                                int query, int *pres1, void *pres2)
{
  struct cf_socket_ctx *ctx = cf->ctx;

  switch(query) {
  case CF_QUERY_SOCKET:
    DEBUGASSERT(pres2);
    *((curl_socket_t *)pres2) = ctx->sock;
    return CURLE_OK;
  case CF_QUERY_CONNECT_REPLY_MS:
    if(ctx->got_first_byte) {
      timediff_t ms = Curl_timediff(ctx->first_byte_at, ctx->started_at);
      *pres1 = (ms < INT_MAX)? (int)ms : INT_MAX;
    }
    else
      *pres1 = -1;
    return CURLE_OK;
  case CF_QUERY_TIMER_CONNECT: {
    struct curltime *when = pres2;
    switch(ctx->transport) {
    case TRNSPRT_UDP:
    case TRNSPRT_QUIC:
      /* Since UDP connected sockets work different from TCP, we use the
       * time of the first byte from the peer as the "connect" time. */
      if(ctx->got_first_byte) {
        *when = ctx->first_byte_at;
        break;
      }
      FALLTHROUGH();
    default:
      *when = ctx->connected_at;
      break;
    }
    return CURLE_OK;
  }
  case CF_QUERY_IP_INFO:
#ifdef USE_IPV6
    *pres1 = (ctx->addr.family == AF_INET6)? TRUE : FALSE;
#else
    *pres1 = FALSE;
#endif
    *(struct ip_quadruple *)pres2 = ctx->ip;
    return CURLE_OK;
  default:
    break;
  }
  return cf->next?
    cf->next->cft->query(cf->next, data, query, pres1, pres2) :
    CURLE_UNKNOWN_OPTION;
}

struct Curl_cftype Curl_cft_tcp = {
  "TCP",
  CF_TYPE_IP_CONNECT,
  CURL_LOG_LVL_NONE,
  cf_socket_destroy,
  cf_tcp_connect,
  cf_socket_close,
  cf_socket_shutdown,
  cf_socket_get_host,
  cf_socket_adjust_pollset,
  cf_socket_data_pending,
  cf_socket_send,
  cf_socket_recv,
  cf_socket_cntrl,
  cf_socket_conn_is_alive,
  Curl_cf_def_conn_keep_alive,
  cf_socket_query,
};

CURLcode Curl_cf_tcp_create(struct Curl_cfilter **pcf,
                            struct Curl_easy *data,
                            struct connectdata *conn,
                            const struct Curl_addrinfo *ai,
                            int transport)
{
  struct cf_socket_ctx *ctx = NULL;
  struct Curl_cfilter *cf = NULL;
  CURLcode result;

  (void)data;
  (void)conn;
  DEBUGASSERT(transport == TRNSPRT_TCP);
  ctx = calloc(1, sizeof(*ctx));
  if(!ctx) {
    result = CURLE_OUT_OF_MEMORY;
    goto out;
  }
  cf_socket_ctx_init(ctx, ai, transport);

  result = Curl_cf_create(&cf, &Curl_cft_tcp, ctx);

out:
  *pcf = (!result)? cf : NULL;
  if(result) {
    Curl_safefree(cf);
    Curl_safefree(ctx);
  }

  return result;
}

static CURLcode cf_udp_setup_quic(struct Curl_cfilter *cf,
                                  struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  int rc;
  int one = 1;

  (void)one;

  /* QUIC needs a connected socket, nonblocking */
  DEBUGASSERT(ctx->sock != CURL_SOCKET_BAD);

  rc = connect(ctx->sock, &ctx->addr.sa_addr,
               (curl_socklen_t)ctx->addr.addrlen);
  if(-1 == rc) {
    return socket_connect_result(data, ctx->ip.remote_ip, SOCKERRNO);
  }
  ctx->sock_connected = TRUE;
  set_local_ip(cf, data);
  CURL_TRC_CF(data, cf, "%s socket %" FMT_SOCKET_T
              " connected: [%s:%d] -> [%s:%d]",
              (ctx->transport == TRNSPRT_QUIC)? "QUIC" : "UDP",
              ctx->sock, ctx->ip.local_ip, ctx->ip.local_port,
              ctx->ip.remote_ip, ctx->ip.remote_port);

  /* Currently, cf->ctx->sock is always non-blocking because the only
   * caller to cf_udp_setup_quic() is cf_udp_connect() that passes the
   * non-blocking socket created by cf_socket_open() to it. Thus, we
   * do not need to call curlx_nonblock() in cf_udp_setup_quic() anymore.
   */
  switch(ctx->addr.family) {
#if defined(__linux__) && defined(IP_MTU_DISCOVER)
  case AF_INET: {
    int val = IP_PMTUDISC_DO;
    (void)setsockopt(ctx->sock, IPPROTO_IP, IP_MTU_DISCOVER, &val,
                     sizeof(val));
    break;
  }
#endif
#if defined(__linux__) && defined(IPV6_MTU_DISCOVER)
  case AF_INET6: {
    int val = IPV6_PMTUDISC_DO;
    (void)setsockopt(ctx->sock, IPPROTO_IPV6, IPV6_MTU_DISCOVER, &val,
                     sizeof(val));
    break;
  }
#endif
  }

#if defined(__linux__) && defined(UDP_GRO) &&                                 \
  (defined(HAVE_SENDMMSG) || defined(HAVE_SENDMSG)) &&                        \
  ((defined(USE_NGTCP2) && defined(USE_NGHTTP3)) || defined(USE_QUICHE))
  (void)setsockopt(ctx->sock, IPPROTO_UDP, UDP_GRO, &one,
                   (socklen_t)sizeof(one));
#endif

  return CURLE_OK;
}

static CURLcode cf_udp_connect(struct Curl_cfilter *cf,
                               struct Curl_easy *data,
                               bool blocking, bool *done)
{
  struct cf_socket_ctx *ctx = cf->ctx;
  CURLcode result = CURLE_COULDNT_CONNECT;

  (void)blocking;
  if(cf->connected) {
    *done = TRUE;
    return CURLE_OK;
  }
  *done = FALSE;
  if(ctx->sock == CURL_SOCKET_BAD) {
    result = cf_socket_open(cf, data);
    if(result) {
      CURL_TRC_CF(data, cf, "cf_udp_connect(), open failed -> %d", result);
      goto out;
    }

    if(ctx->transport == TRNSPRT_QUIC) {
      result = cf_udp_setup_quic(cf, data);
      if(result)
        goto out;
      CURL_TRC_CF(data, cf, "cf_udp_connect(), opened socket=%"
                  FMT_SOCKET_T " (%s:%d)",
                  ctx->sock, ctx->ip.local_ip, ctx->ip.local_port);
    }
    else {
      CURL_TRC_CF(data, cf, "cf_udp_connect(), opened socket=%"
                  FMT_SOCKET_T " (unconnected)", ctx->sock);
    }
    *done = TRUE;
    cf->connected = TRUE;
  }
out:
  return result;
}

struct Curl_cftype Curl_cft_udp = {
  "UDP",
  CF_TYPE_IP_CONNECT,
  CURL_LOG_LVL_NONE,
  cf_socket_destroy,
  cf_udp_connect,
  cf_socket_close,
  cf_socket_shutdown,
  cf_socket_get_host,
  cf_socket_adjust_pollset,
  cf_socket_data_pending,
  cf_socket_send,
  cf_socket_recv,
  cf_socket_cntrl,
  cf_socket_conn_is_alive,
  Curl_cf_def_conn_keep_alive,
  cf_socket_query,
};

CURLcode Curl_cf_udp_create(struct Curl_cfilter **pcf,
                            struct Curl_easy *data,
                            struct connectdata *conn,
                            const struct Curl_addrinfo *ai,
                            int transport)
{
  struct cf_socket_ctx *ctx = NULL;
  struct Curl_cfilter *cf = NULL;
  CURLcode result;

  (void)data;
  (void)conn;
  DEBUGASSERT(transport == TRNSPRT_UDP || transport == TRNSPRT_QUIC);
  ctx = calloc(1, sizeof(*ctx));
  if(!ctx) {
    result = CURLE_OUT_OF_MEMORY;
    goto out;
  }
  cf_socket_ctx_init(ctx, ai, transport);

  result = Curl_cf_create(&cf, &Curl_cft_udp, ctx);

out:
  *pcf = (!result)? cf : NULL;
  if(result) {
    Curl_safefree(cf);
    Curl_safefree(ctx);
  }

  return result;
}

/* this is the TCP filter which can also handle this case */
struct Curl_cftype Curl_cft_unix = {
  "UNIX",
  CF_TYPE_IP_CONNECT,
  CURL_LOG_LVL_NONE,
  cf_socket_destroy,
  cf_tcp_connect,
  cf_socket_close,
  cf_socket_shutdown,
  cf_socket_get_host,
  cf_socket_adjust_pollset,
  cf_socket_data_pending,
  cf_socket_send,
  cf_socket_recv,
  cf_socket_cntrl,
  cf_socket_conn_is_alive,
  Curl_cf_def_conn_keep_alive,
  cf_socket_query,
};

CURLcode Curl_cf_unix_create(struct Curl_cfilter **pcf,
                             struct Curl_easy *data,
                             struct connectdata *conn,
                             const struct Curl_addrinfo *ai,
                             int transport)
{
  struct cf_socket_ctx *ctx = NULL;
  struct Curl_cfilter *cf = NULL;
  CURLcode result;

  (void)data;
  (void)conn;
  DEBUGASSERT(transport == TRNSPRT_UNIX);
  ctx = calloc(1, sizeof(*ctx));
  if(!ctx) {
    result = CURLE_OUT_OF_MEMORY;
    goto out;
  }
  cf_socket_ctx_init(ctx, ai, transport);

  result = Curl_cf_create(&cf, &Curl_cft_unix, ctx);

out:
  *pcf = (!result)? cf : NULL;
  if(result) {
    Curl_safefree(cf);
    Curl_safefree(ctx);
  }

  return result;
}

static CURLcode cf_tcp_accept_connect(struct Curl_cfilter *cf,
                                      struct Curl_easy *data,
                                      bool blocking, bool *done)
{
  /* we start accepted, if we ever close, we cannot go on */
  (void)data;
  (void)blocking;
  if(cf->connected) {
    *done = TRUE;
    return CURLE_OK;
  }
  return CURLE_FAILED_INIT;
}

struct Curl_cftype Curl_cft_tcp_accept = {
  "TCP-ACCEPT",
  CF_TYPE_IP_CONNECT,
  CURL_LOG_LVL_NONE,
  cf_socket_destroy,
  cf_tcp_accept_connect,
  cf_socket_close,
  cf_socket_shutdown,
  cf_socket_get_host,              /* TODO: not accurate */
  cf_socket_adjust_pollset,
  cf_socket_data_pending,
  cf_socket_send,
  cf_socket_recv,
  cf_socket_cntrl,
  cf_socket_conn_is_alive,
  Curl_cf_def_conn_keep_alive,
  cf_socket_query,
};

CURLcode Curl_conn_tcp_listen_set(struct Curl_easy *data,
                                  struct connectdata *conn,
                                  int sockindex, curl_socket_t *s)
{
  CURLcode result;
  struct Curl_cfilter *cf = NULL;
  struct cf_socket_ctx *ctx = NULL;

  /* replace any existing */
  Curl_conn_cf_discard_all(data, conn, sockindex);
  DEBUGASSERT(conn->sock[sockindex] == CURL_SOCKET_BAD);

  ctx = calloc(1, sizeof(*ctx));
  if(!ctx) {
    result = CURLE_OUT_OF_MEMORY;
    goto out;
  }
  ctx->transport = conn->transport;
  ctx->sock = *s;
  ctx->listening = TRUE;
  ctx->accepted = FALSE;
  result = Curl_cf_create(&cf, &Curl_cft_tcp_accept, ctx);
  if(result)
    goto out;
  Curl_conn_cf_add(data, conn, sockindex, cf);

  conn->sock[sockindex] = ctx->sock;
  set_local_ip(cf, data);
  ctx->active = TRUE;
  ctx->connected_at = Curl_now();
  cf->connected = TRUE;
  CURL_TRC_CF(data, cf, "Curl_conn_tcp_listen_set(%" FMT_SOCKET_T ")",
              ctx->sock);

out:
  if(result) {
    Curl_safefree(cf);
    Curl_safefree(ctx);
  }
  return result;
}

static void set_accepted_remote_ip(struct Curl_cfilter *cf,
                                   struct Curl_easy *data)
{
  struct cf_socket_ctx *ctx = cf->ctx;
#ifdef HAVE_GETPEERNAME
  char buffer[STRERROR_LEN];
  struct Curl_sockaddr_storage ssrem;
  curl_socklen_t plen;

  ctx->ip.remote_ip[0] = 0;
  ctx->ip.remote_port = 0;
  plen = sizeof(ssrem);
  memset(&ssrem, 0, plen);
  if(getpeername(ctx->sock, (struct sockaddr*) &ssrem, &plen)) {
    int error = SOCKERRNO;
    failf(data, "getpeername() failed with errno %d: %s",
          error, Curl_strerror(error, buffer, sizeof(buffer)));
    return;
  }
  if(!Curl_addr2string((struct sockaddr*)&ssrem, plen,
                       ctx->ip.remote_ip, &ctx->ip.remote_port)) {
    failf(data, "ssrem inet_ntop() failed with errno %d: %s",
          errno, Curl_strerror(errno, buffer, sizeof(buffer)));
    return;
  }
#else
  ctx->ip.remote_ip[0] = 0;
  ctx->ip.remote_port = 0;
  (void)data;
#endif
}

CURLcode Curl_conn_tcp_accepted_set(struct Curl_easy *data,
                                    struct connectdata *conn,
                                    int sockindex, curl_socket_t *s)
{
  struct Curl_cfilter *cf = NULL;
  struct cf_socket_ctx *ctx = NULL;

  cf = conn->cfilter[sockindex];
  if(!cf || cf->cft != &Curl_cft_tcp_accept)
    return CURLE_FAILED_INIT;

  ctx = cf->ctx;
  DEBUGASSERT(ctx->listening);
  /* discard the listen socket */
  socket_close(data, conn, TRUE, ctx->sock);
  ctx->listening = FALSE;
  ctx->sock = *s;
  conn->sock[sockindex] = ctx->sock;
  set_accepted_remote_ip(cf, data);
  set_local_ip(cf, data);
  ctx->active = TRUE;
  ctx->accepted = TRUE;
  ctx->connected_at = Curl_now();
  cf->connected = TRUE;
  CURL_TRC_CF(data, cf, "accepted_set(sock=%" FMT_SOCKET_T
              ", remote=%s port=%d)",
              ctx->sock, ctx->ip.remote_ip, ctx->ip.remote_port);

  return CURLE_OK;
}

/**
 * Return TRUE iff `cf` is a socket filter.
 */
static bool cf_is_socket(struct Curl_cfilter *cf)
{
  return cf && (cf->cft == &Curl_cft_tcp ||
                cf->cft == &Curl_cft_udp ||
                cf->cft == &Curl_cft_unix ||
                cf->cft == &Curl_cft_tcp_accept);
}

CURLcode Curl_cf_socket_peek(struct Curl_cfilter *cf,
                             struct Curl_easy *data,
                             curl_socket_t *psock,
                             const struct Curl_sockaddr_ex **paddr,
                             struct ip_quadruple *pip)
{
  (void)data;
  if(cf_is_socket(cf) && cf->ctx) {
    struct cf_socket_ctx *ctx = cf->ctx;

    if(psock)
      *psock = ctx->sock;
    if(paddr)
      *paddr = &ctx->addr;
    if(pip)
      *pip = ctx->ip;
    return CURLE_OK;
  }
  return CURLE_FAILED_INIT;
}