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
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
|
// Copyright 2018 The gVisor Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package linux
import (
"syscall"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/context"
"gvisor.dev/gvisor/pkg/sentry/arch"
"gvisor.dev/gvisor/pkg/sentry/fs"
"gvisor.dev/gvisor/pkg/sentry/fs/lock"
"gvisor.dev/gvisor/pkg/sentry/fs/tmpfs"
"gvisor.dev/gvisor/pkg/sentry/kernel"
"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
"gvisor.dev/gvisor/pkg/sentry/kernel/fasync"
ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
"gvisor.dev/gvisor/pkg/sentry/limits"
"gvisor.dev/gvisor/pkg/syserror"
"gvisor.dev/gvisor/pkg/usermem"
)
// fileOpAt performs an operation on the second last component in the path.
func fileOpAt(t *kernel.Task, dirFD int32, path string, fn func(root *fs.Dirent, d *fs.Dirent, name string, remainingTraversals uint) error) error {
// Extract the last component.
dir, name := fs.SplitLast(path)
if dir == "/" {
// Common case: we are accessing a file in the root.
root := t.FSContext().RootDirectory()
err := fn(root, root, name, linux.MaxSymlinkTraversals)
root.DecRef()
return err
} else if dir == "." && dirFD == linux.AT_FDCWD {
// Common case: we are accessing a file relative to the current
// working directory; skip the look-up.
wd := t.FSContext().WorkingDirectory()
root := t.FSContext().RootDirectory()
err := fn(root, wd, name, linux.MaxSymlinkTraversals)
wd.DecRef()
root.DecRef()
return err
}
return fileOpOn(t, dirFD, dir, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, remainingTraversals uint) error {
return fn(root, d, name, remainingTraversals)
})
}
// fileOpOn performs an operation on the last entry of the path.
func fileOpOn(t *kernel.Task, dirFD int32, path string, resolve bool, fn func(root *fs.Dirent, d *fs.Dirent, remainingTraversals uint) error) error {
var (
d *fs.Dirent // The file.
wd *fs.Dirent // The working directory (if required.)
rel *fs.Dirent // The relative directory for search (if required.)
f *fs.File // The file corresponding to dirFD (if required.)
err error
)
// Extract the working directory (maybe).
if len(path) > 0 && path[0] == '/' {
// Absolute path; rel can be nil.
} else if dirFD == linux.AT_FDCWD {
// Need to reference the working directory.
wd = t.FSContext().WorkingDirectory()
rel = wd
} else {
// Need to extract the given FD.
f = t.GetFile(dirFD)
if f == nil {
return syserror.EBADF
}
rel = f.Dirent
if !fs.IsDir(rel.Inode.StableAttr) {
return syserror.ENOTDIR
}
}
// Grab the root (always required.)
root := t.FSContext().RootDirectory()
// Lookup the node.
remainingTraversals := uint(linux.MaxSymlinkTraversals)
if resolve {
d, err = t.MountNamespace().FindInode(t, root, rel, path, &remainingTraversals)
} else {
d, err = t.MountNamespace().FindLink(t, root, rel, path, &remainingTraversals)
}
root.DecRef()
if wd != nil {
wd.DecRef()
}
if f != nil {
f.DecRef()
}
if err != nil {
return err
}
err = fn(root, d, remainingTraversals)
d.DecRef()
return err
}
// copyInPath copies a path in.
func copyInPath(t *kernel.Task, addr usermem.Addr, allowEmpty bool) (path string, dirPath bool, err error) {
path, err = t.CopyInString(addr, linux.PATH_MAX)
if err != nil {
return "", false, err
}
if path == "" && !allowEmpty {
return "", false, syserror.ENOENT
}
// If the path ends with a /, then checks must be enforced in various
// ways in the different callers. We pass this back to the caller.
path, dirPath = fs.TrimTrailingSlashes(path)
return path, dirPath, nil
}
func openAt(t *kernel.Task, dirFD int32, addr usermem.Addr, flags uint) (fd uintptr, err error) {
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, err
}
resolve := flags&linux.O_NOFOLLOW == 0
err = fileOpOn(t, dirFD, path, resolve, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// First check a few things about the filesystem before trying to get the file
// reference.
//
// It's required that Check does not try to open files not that aren't backed by
// this dirent (e.g. pipes and sockets) because this would result in opening these
// files an extra time just to check permissions.
if err := d.Inode.CheckPermission(t, flagsToPermissions(flags)); err != nil {
return err
}
if fs.IsSymlink(d.Inode.StableAttr) && !resolve {
return syserror.ELOOP
}
fileFlags := linuxToFlags(flags)
// Linux always adds the O_LARGEFILE flag when running in 64-bit mode.
fileFlags.LargeFile = true
if fs.IsDir(d.Inode.StableAttr) {
// Don't allow directories to be opened writable.
if fileFlags.Write {
return syserror.EISDIR
}
} else {
// If O_DIRECTORY is set, but the file is not a directory, then fail.
if fileFlags.Directory {
return syserror.ENOTDIR
}
// If it's a directory, then make sure.
if dirPath {
return syserror.ENOTDIR
}
}
// Truncate is called when O_TRUNC is specified for any kind of
// existing Dirent. Behavior is delegated to the entry's Truncate
// implementation.
if flags&linux.O_TRUNC != 0 {
if err := d.Inode.Truncate(t, d, 0); err != nil {
return err
}
}
file, err := d.Inode.GetFile(t, d, fileFlags)
if err != nil {
return syserror.ConvertIntr(err, kernel.ERESTARTSYS)
}
defer file.DecRef()
// Success.
newFD, err := t.NewFDFrom(0, file, kernel.FDFlags{
CloseOnExec: flags&linux.O_CLOEXEC != 0,
})
if err != nil {
return err
}
// Set return result in frame.
fd = uintptr(newFD)
// Generate notification for opened file.
d.InotifyEvent(linux.IN_OPEN, 0)
return nil
})
return fd, err // Use result in frame.
}
func mknodAt(t *kernel.Task, dirFD int32, addr usermem.Addr, mode linux.FileMode) error {
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
if dirPath {
return syserror.ENOENT
}
return fileOpAt(t, dirFD, path, func(root *fs.Dirent, d *fs.Dirent, name string, _ uint) error {
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Do we have the appropriate permissions on the parent?
if err := d.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
// Attempt a creation.
perms := fs.FilePermsFromMode(mode &^ linux.FileMode(t.FSContext().Umask()))
switch mode.FileType() {
case 0:
// "Zero file type is equivalent to type S_IFREG." - mknod(2)
fallthrough
case linux.ModeRegular:
// We are not going to return the file, so the actual
// flags used don't matter, but they cannot be empty or
// Create will complain.
flags := fs.FileFlags{Read: true, Write: true}
file, err := d.Create(t, root, name, flags, perms)
if err != nil {
return err
}
file.DecRef()
return nil
case linux.ModeNamedPipe:
return d.CreateFifo(t, root, name, perms)
case linux.ModeSocket:
// While it is possible create a unix domain socket file on linux
// using mknod(2), in practice this is pretty useless from an
// application. Linux internally uses mknod() to create the socket
// node during bind(2), but we implement bind(2) independently. If
// an application explicitly creates a socket node using mknod(),
// you can't seem to bind() or connect() to the resulting socket.
//
// Instead of emulating this seemingly useless behaviour, we'll
// indicate that the filesystem doesn't support the creation of
// sockets.
return syserror.EOPNOTSUPP
case linux.ModeCharacterDevice:
fallthrough
case linux.ModeBlockDevice:
// TODO(b/72101894): We don't support creating block or character
// devices at the moment.
//
// When we start supporting block and character devices, we'll
// need to check for CAP_MKNOD here.
return syserror.EPERM
default:
// "EINVAL - mode requested creation of something other than a
// regular file, device special file, FIFO or socket." - mknod(2)
return syserror.EINVAL
}
})
}
// Mknod implements the linux syscall mknod(2).
func Mknod(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
path := args[0].Pointer()
mode := linux.FileMode(args[1].ModeT())
// We don't need this argument until we support creation of device nodes.
_ = args[2].Uint() // dev
return 0, nil, mknodAt(t, linux.AT_FDCWD, path, mode)
}
// Mknodat implements the linux syscall mknodat(2).
func Mknodat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
path := args[1].Pointer()
mode := linux.FileMode(args[2].ModeT())
// We don't need this argument until we support creation of device nodes.
_ = args[3].Uint() // dev
return 0, nil, mknodAt(t, dirFD, path, mode)
}
func createAt(t *kernel.Task, dirFD int32, addr usermem.Addr, flags uint, mode linux.FileMode) (fd uintptr, err error) {
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, err
}
if dirPath {
return 0, syserror.ENOENT
}
fileFlags := linuxToFlags(flags)
// Linux always adds the O_LARGEFILE flag when running in 64-bit mode.
fileFlags.LargeFile = true
err = fileOpAt(t, dirFD, path, func(root *fs.Dirent, parent *fs.Dirent, name string, remainingTraversals uint) error {
// Resolve the name to see if it exists, and follow any
// symlinks along the way. We must do the symlink resolution
// manually because if the symlink target does not exist, we
// must create the target (and not the symlink itself).
var (
found *fs.Dirent
err error
)
for {
if !fs.IsDir(parent.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Start by looking up the dirent at 'name'.
found, err = t.MountNamespace().FindLink(t, root, parent, name, &remainingTraversals)
if err != nil {
break
}
defer found.DecRef()
// We found something (possibly a symlink). If the
// O_EXCL flag was passed, then we can immediately
// return EEXIST.
if flags&linux.O_EXCL != 0 {
return syserror.EEXIST
}
// If we have a non-symlink, then we can proceed.
if !fs.IsSymlink(found.Inode.StableAttr) {
break
}
// If O_NOFOLLOW was passed, then don't try to resolve
// anything.
if flags&linux.O_NOFOLLOW != 0 {
return syserror.ELOOP
}
// Try to resolve the symlink directly to a Dirent.
var resolved *fs.Dirent
resolved, err = found.Inode.Getlink(t)
if err == nil {
// No more resolution necessary.
defer resolved.DecRef()
break
}
if err != fs.ErrResolveViaReadlink {
return err
}
// Are we able to resolve further?
if remainingTraversals == 0 {
return syscall.ELOOP
}
// Resolve the symlink to a path via Readlink.
var path string
path, err = found.Inode.Readlink(t)
if err != nil {
break
}
remainingTraversals--
// Get the new parent from the target path.
var newParent *fs.Dirent
newParentPath, newName := fs.SplitLast(path)
newParent, err = t.MountNamespace().FindInode(t, root, parent, newParentPath, &remainingTraversals)
if err != nil {
break
}
defer newParent.DecRef()
// Repeat the process with the parent and name of the
// symlink target.
parent = newParent
name = newName
}
var newFile *fs.File
switch err {
case nil:
// Like sys_open, check for a few things about the
// filesystem before trying to get a reference to the
// fs.File. The same constraints on Check apply.
if err := found.Inode.CheckPermission(t, flagsToPermissions(flags)); err != nil {
return err
}
// Truncate is called when O_TRUNC is specified for any kind of
// existing Dirent. Behavior is delegated to the entry's Truncate
// implementation.
if flags&linux.O_TRUNC != 0 {
if err := found.Inode.Truncate(t, found, 0); err != nil {
return err
}
}
// Create a new fs.File.
newFile, err = found.Inode.GetFile(t, found, fileFlags)
if err != nil {
return syserror.ConvertIntr(err, kernel.ERESTARTSYS)
}
defer newFile.DecRef()
case syserror.ENOENT:
// File does not exist. Proceed with creation.
// Do we have write permissions on the parent?
if err := parent.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
// Attempt a creation.
perms := fs.FilePermsFromMode(mode &^ linux.FileMode(t.FSContext().Umask()))
newFile, err = parent.Create(t, root, name, fileFlags, perms)
if err != nil {
// No luck, bail.
return err
}
defer newFile.DecRef()
found = newFile.Dirent
default:
return err
}
// Success.
newFD, err := t.NewFDFrom(0, newFile, kernel.FDFlags{
CloseOnExec: flags&linux.O_CLOEXEC != 0,
})
if err != nil {
return err
}
// Set result in frame.
fd = uintptr(newFD)
// Queue the open inotify event. The creation event is
// automatically queued when the dirent is found. The open
// events are implemented at the syscall layer so we need to
// manually queue one here.
found.InotifyEvent(linux.IN_OPEN, 0)
return nil
})
return fd, err // Use result in frame.
}
// Open implements linux syscall open(2).
func Open(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
flags := uint(args[1].Uint())
if flags&linux.O_CREAT != 0 {
mode := linux.FileMode(args[2].ModeT())
n, err := createAt(t, linux.AT_FDCWD, addr, flags, mode)
return n, nil, err
}
n, err := openAt(t, linux.AT_FDCWD, addr, flags)
return n, nil, err
}
// Openat implements linux syscall openat(2).
func Openat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
flags := uint(args[2].Uint())
if flags&linux.O_CREAT != 0 {
mode := linux.FileMode(args[3].ModeT())
n, err := createAt(t, dirFD, addr, flags, mode)
return n, nil, err
}
n, err := openAt(t, dirFD, addr, flags)
return n, nil, err
}
// Creat implements linux syscall creat(2).
func Creat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
mode := linux.FileMode(args[1].ModeT())
n, err := createAt(t, linux.AT_FDCWD, addr, linux.O_WRONLY|linux.O_TRUNC, mode)
return n, nil, err
}
// accessContext is a context that overrides the credentials used, but
// otherwise carries the same values as the embedded context.
//
// accessContext should only be used for access(2).
type accessContext struct {
context.Context
creds *auth.Credentials
}
// Value implements context.Context.
func (ac accessContext) Value(key interface{}) interface{} {
switch key {
case auth.CtxCredentials:
return ac.creds
default:
return ac.Context.Value(key)
}
}
func accessAt(t *kernel.Task, dirFD int32, addr usermem.Addr, resolve bool, mode uint) error {
const rOK = 4
const wOK = 2
const xOK = 1
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
// Sanity check the mode.
if mode&^(rOK|wOK|xOK) != 0 {
return syserror.EINVAL
}
return fileOpOn(t, dirFD, path, resolve, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// access(2) and faccessat(2) check permissions using real
// UID/GID, not effective UID/GID.
//
// "access() needs to use the real uid/gid, not the effective
// uid/gid. We do this by temporarily clearing all FS-related
// capabilities and switching the fsuid/fsgid around to the
// real ones." -fs/open.c:faccessat
creds := t.Credentials().Fork()
creds.EffectiveKUID = creds.RealKUID
creds.EffectiveKGID = creds.RealKGID
if creds.EffectiveKUID.In(creds.UserNamespace) == auth.RootUID {
creds.EffectiveCaps = creds.PermittedCaps
} else {
creds.EffectiveCaps = 0
}
ctx := &accessContext{
Context: t,
creds: creds,
}
return d.Inode.CheckPermission(ctx, fs.PermMask{
Read: mode&rOK != 0,
Write: mode&wOK != 0,
Execute: mode&xOK != 0,
})
})
}
// Access implements linux syscall access(2).
func Access(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
mode := args[1].ModeT()
return 0, nil, accessAt(t, linux.AT_FDCWD, addr, true, mode)
}
// Faccessat implements linux syscall faccessat(2).
func Faccessat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
mode := args[2].ModeT()
flags := args[3].Int()
return 0, nil, accessAt(t, dirFD, addr, flags&linux.AT_SYMLINK_NOFOLLOW == 0, mode)
}
// Ioctl implements linux syscall ioctl(2).
func Ioctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
request := int(args[1].Int())
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
// Shared flags between file and socket.
switch request {
case linux.FIONCLEX:
t.FDTable().SetFlags(fd, kernel.FDFlags{
CloseOnExec: false,
})
return 0, nil, nil
case linux.FIOCLEX:
t.FDTable().SetFlags(fd, kernel.FDFlags{
CloseOnExec: true,
})
return 0, nil, nil
case linux.FIONBIO:
var set int32
if _, err := t.CopyIn(args[2].Pointer(), &set); err != nil {
return 0, nil, err
}
flags := file.Flags()
if set != 0 {
flags.NonBlocking = true
} else {
flags.NonBlocking = false
}
file.SetFlags(flags.Settable())
return 0, nil, nil
case linux.FIOASYNC:
var set int32
if _, err := t.CopyIn(args[2].Pointer(), &set); err != nil {
return 0, nil, err
}
flags := file.Flags()
if set != 0 {
flags.Async = true
} else {
flags.Async = false
}
file.SetFlags(flags.Settable())
return 0, nil, nil
case linux.FIOSETOWN, linux.SIOCSPGRP:
var set int32
if _, err := t.CopyIn(args[2].Pointer(), &set); err != nil {
return 0, nil, err
}
fSetOwn(t, file, set)
return 0, nil, nil
case linux.FIOGETOWN, linux.SIOCGPGRP:
who := fGetOwn(t, file)
_, err := t.CopyOut(args[2].Pointer(), &who)
return 0, nil, err
default:
ret, err := file.FileOperations.Ioctl(t, file, t.MemoryManager(), args)
if err != nil {
return 0, nil, err
}
return ret, nil, nil
}
}
// Getcwd implements the linux syscall getcwd(2).
func Getcwd(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
size := args[1].SizeT()
cwd := t.FSContext().WorkingDirectory()
defer cwd.DecRef()
root := t.FSContext().RootDirectory()
defer root.DecRef()
// Get our fullname from the root and preprend unreachable if the root was
// unreachable from our current dirent this is the same behavior as on linux.
s, reachable := cwd.FullName(root)
if !reachable {
s = "(unreachable)" + s
}
// Note this is >= because we need a terminator.
if uint(len(s)) >= size {
return 0, nil, syserror.ERANGE
}
// Copy out the path name for the node.
bytes, err := t.CopyOutBytes(addr, []byte(s))
if err != nil {
return 0, nil, err
}
// Top it off with a terminator.
_, err = t.CopyOut(addr+usermem.Addr(bytes), []byte("\x00"))
return uintptr(bytes + 1), nil, err
}
// Chroot implements the linux syscall chroot(2).
func Chroot(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
if !t.HasCapability(linux.CAP_SYS_CHROOT) {
return 0, nil, syserror.EPERM
}
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, nil, err
}
return 0, nil, fileOpOn(t, linux.AT_FDCWD, path, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// Is it a directory?
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Does it have execute permissions?
if err := d.Inode.CheckPermission(t, fs.PermMask{Execute: true}); err != nil {
return err
}
t.FSContext().SetRootDirectory(d)
return nil
})
}
// Chdir implements the linux syscall chdir(2).
func Chdir(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, nil, err
}
return 0, nil, fileOpOn(t, linux.AT_FDCWD, path, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// Is it a directory?
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Does it have execute permissions?
if err := d.Inode.CheckPermission(t, fs.PermMask{Execute: true}); err != nil {
return err
}
t.FSContext().SetWorkingDirectory(d)
return nil
})
}
// Fchdir implements the linux syscall fchdir(2).
func Fchdir(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
// Is it a directory?
if !fs.IsDir(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.ENOTDIR
}
// Does it have execute permissions?
if err := file.Dirent.Inode.CheckPermission(t, fs.PermMask{Execute: true}); err != nil {
return 0, nil, err
}
t.FSContext().SetWorkingDirectory(file.Dirent)
return 0, nil, nil
}
// Close implements linux syscall close(2).
func Close(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
// Note that Remove provides a reference on the file that we may use to
// flush. It is still active until we drop the final reference below
// (and other reference-holding operations complete).
file := t.FDTable().Remove(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
err := file.Flush(t)
return 0, nil, handleIOError(t, false /* partial */, err, syserror.EINTR, "close", file)
}
// Dup implements linux syscall dup(2).
func Dup(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
newFD, err := t.NewFDFrom(0, file, kernel.FDFlags{})
if err != nil {
return 0, nil, syserror.EMFILE
}
return uintptr(newFD), nil, nil
}
// Dup2 implements linux syscall dup2(2).
func Dup2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldfd := args[0].Int()
newfd := args[1].Int()
// If oldfd is a valid file descriptor, and newfd has the same value as oldfd,
// then dup2() does nothing, and returns newfd.
if oldfd == newfd {
oldFile := t.GetFile(oldfd)
if oldFile == nil {
return 0, nil, syserror.EBADF
}
defer oldFile.DecRef()
return uintptr(newfd), nil, nil
}
// Zero out flags arg to be used by Dup3.
args[2].Value = 0
return Dup3(t, args)
}
// Dup3 implements linux syscall dup3(2).
func Dup3(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldfd := args[0].Int()
newfd := args[1].Int()
flags := args[2].Uint()
if oldfd == newfd {
return 0, nil, syserror.EINVAL
}
oldFile := t.GetFile(oldfd)
if oldFile == nil {
return 0, nil, syserror.EBADF
}
defer oldFile.DecRef()
err := t.NewFDAt(newfd, oldFile, kernel.FDFlags{CloseOnExec: flags&linux.O_CLOEXEC != 0})
if err != nil {
return 0, nil, err
}
return uintptr(newfd), nil, nil
}
func fGetOwnEx(t *kernel.Task, file *fs.File) linux.FOwnerEx {
ma := file.Async(nil)
if ma == nil {
return linux.FOwnerEx{}
}
a := ma.(*fasync.FileAsync)
ot, otg, opg := a.Owner()
switch {
case ot != nil:
return linux.FOwnerEx{
Type: linux.F_OWNER_TID,
PID: int32(t.PIDNamespace().IDOfTask(ot)),
}
case otg != nil:
return linux.FOwnerEx{
Type: linux.F_OWNER_PID,
PID: int32(t.PIDNamespace().IDOfThreadGroup(otg)),
}
case opg != nil:
return linux.FOwnerEx{
Type: linux.F_OWNER_PGRP,
PID: int32(t.PIDNamespace().IDOfProcessGroup(opg)),
}
default:
return linux.FOwnerEx{}
}
}
func fGetOwn(t *kernel.Task, file *fs.File) int32 {
owner := fGetOwnEx(t, file)
if owner.Type == linux.F_OWNER_PGRP {
return -owner.PID
}
return owner.PID
}
// fSetOwn sets the file's owner with the semantics of F_SETOWN in Linux.
//
// If who is positive, it represents a PID. If negative, it represents a PGID.
// If the PID or PGID is invalid, the owner is silently unset.
func fSetOwn(t *kernel.Task, file *fs.File, who int32) {
a := file.Async(fasync.New).(*fasync.FileAsync)
if who < 0 {
pg := t.PIDNamespace().ProcessGroupWithID(kernel.ProcessGroupID(-who))
a.SetOwnerProcessGroup(t, pg)
}
tg := t.PIDNamespace().ThreadGroupWithID(kernel.ThreadID(who))
a.SetOwnerThreadGroup(t, tg)
}
// Fcntl implements linux syscall fcntl(2).
func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
cmd := args[1].Int()
file, flags := t.FDTable().Get(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
switch cmd {
case linux.F_DUPFD, linux.F_DUPFD_CLOEXEC:
from := args[2].Int()
fd, err := t.NewFDFrom(from, file, kernel.FDFlags{
CloseOnExec: cmd == linux.F_DUPFD_CLOEXEC,
})
if err != nil {
return 0, nil, err
}
return uintptr(fd), nil, nil
case linux.F_GETFD:
return uintptr(flags.ToLinuxFDFlags()), nil, nil
case linux.F_SETFD:
flags := args[2].Uint()
t.FDTable().SetFlags(fd, kernel.FDFlags{
CloseOnExec: flags&linux.FD_CLOEXEC != 0,
})
return 0, nil, nil
case linux.F_GETFL:
return uintptr(file.Flags().ToLinux()), nil, nil
case linux.F_SETFL:
flags := uint(args[2].Uint())
file.SetFlags(linuxToFlags(flags).Settable())
return 0, nil, nil
case linux.F_SETLK, linux.F_SETLKW:
// In Linux the file system can choose to provide lock operations for an inode.
// Normally pipe and socket types lack lock operations. We diverge and use a heavy
// hammer by only allowing locks on files and directories.
if !fs.IsFile(file.Dirent.Inode.StableAttr) && !fs.IsDir(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.EBADF
}
// Copy in the lock request.
flockAddr := args[2].Pointer()
var flock linux.Flock
if _, err := t.CopyIn(flockAddr, &flock); err != nil {
return 0, nil, err
}
// Compute the lock whence.
var sw fs.SeekWhence
switch flock.Whence {
case 0:
sw = fs.SeekSet
case 1:
sw = fs.SeekCurrent
case 2:
sw = fs.SeekEnd
default:
return 0, nil, syserror.EINVAL
}
// Compute the lock offset.
var off int64
switch sw {
case fs.SeekSet:
off = 0
case fs.SeekCurrent:
// Note that Linux does not hold any mutexes while retrieving the file offset,
// see fs/locks.c:flock_to_posix_lock and fs/locks.c:fcntl_setlk.
off = file.Offset()
case fs.SeekEnd:
uattr, err := file.Dirent.Inode.UnstableAttr(t)
if err != nil {
return 0, nil, err
}
off = uattr.Size
default:
return 0, nil, syserror.EINVAL
}
// Compute the lock range.
rng, err := lock.ComputeRange(flock.Start, flock.Len, off)
if err != nil {
return 0, nil, err
}
// The lock uid is that of the Task's FDTable.
lockUniqueID := lock.UniqueID(t.FDTable().ID())
// These locks don't block; execute the non-blocking operation using the inode's lock
// context directly.
switch flock.Type {
case linux.F_RDLCK:
if !file.Flags().Read {
return 0, nil, syserror.EBADF
}
if cmd == linux.F_SETLK {
// Non-blocking lock, provide a nil lock.Blocker.
if !file.Dirent.Inode.LockCtx.Posix.LockRegion(lockUniqueID, lock.ReadLock, rng, nil) {
return 0, nil, syserror.EAGAIN
}
} else {
// Blocking lock, pass in the task to satisfy the lock.Blocker interface.
if !file.Dirent.Inode.LockCtx.Posix.LockRegion(lockUniqueID, lock.ReadLock, rng, t) {
return 0, nil, syserror.EINTR
}
}
return 0, nil, nil
case linux.F_WRLCK:
if !file.Flags().Write {
return 0, nil, syserror.EBADF
}
if cmd == linux.F_SETLK {
// Non-blocking lock, provide a nil lock.Blocker.
if !file.Dirent.Inode.LockCtx.Posix.LockRegion(lockUniqueID, lock.WriteLock, rng, nil) {
return 0, nil, syserror.EAGAIN
}
} else {
// Blocking lock, pass in the task to satisfy the lock.Blocker interface.
if !file.Dirent.Inode.LockCtx.Posix.LockRegion(lockUniqueID, lock.WriteLock, rng, t) {
return 0, nil, syserror.EINTR
}
}
return 0, nil, nil
case linux.F_UNLCK:
file.Dirent.Inode.LockCtx.Posix.UnlockRegion(lockUniqueID, rng)
return 0, nil, nil
default:
return 0, nil, syserror.EINVAL
}
case linux.F_GETOWN:
return uintptr(fGetOwn(t, file)), nil, nil
case linux.F_SETOWN:
fSetOwn(t, file, args[2].Int())
return 0, nil, nil
case linux.F_GETOWN_EX:
addr := args[2].Pointer()
owner := fGetOwnEx(t, file)
_, err := t.CopyOut(addr, &owner)
return 0, nil, err
case linux.F_SETOWN_EX:
addr := args[2].Pointer()
var owner linux.FOwnerEx
n, err := t.CopyIn(addr, &owner)
if err != nil {
return 0, nil, err
}
a := file.Async(fasync.New).(*fasync.FileAsync)
switch owner.Type {
case linux.F_OWNER_TID:
task := t.PIDNamespace().TaskWithID(kernel.ThreadID(owner.PID))
if task == nil {
return 0, nil, syserror.ESRCH
}
a.SetOwnerTask(t, task)
return uintptr(n), nil, nil
case linux.F_OWNER_PID:
tg := t.PIDNamespace().ThreadGroupWithID(kernel.ThreadID(owner.PID))
if tg == nil {
return 0, nil, syserror.ESRCH
}
a.SetOwnerThreadGroup(t, tg)
return uintptr(n), nil, nil
case linux.F_OWNER_PGRP:
pg := t.PIDNamespace().ProcessGroupWithID(kernel.ProcessGroupID(owner.PID))
if pg == nil {
return 0, nil, syserror.ESRCH
}
a.SetOwnerProcessGroup(t, pg)
return uintptr(n), nil, nil
default:
return 0, nil, syserror.EINVAL
}
case linux.F_GET_SEALS:
val, err := tmpfs.GetSeals(file.Dirent.Inode)
return uintptr(val), nil, err
case linux.F_ADD_SEALS:
if !file.Flags().Write {
return 0, nil, syserror.EPERM
}
err := tmpfs.AddSeals(file.Dirent.Inode, args[2].Uint())
return 0, nil, err
case linux.F_GETPIPE_SZ:
sz, ok := file.FileOperations.(fs.FifoSizer)
if !ok {
return 0, nil, syserror.EINVAL
}
size, err := sz.FifoSize(t, file)
return uintptr(size), nil, err
case linux.F_SETPIPE_SZ:
sz, ok := file.FileOperations.(fs.FifoSizer)
if !ok {
return 0, nil, syserror.EINVAL
}
n, err := sz.SetFifoSize(int64(args[2].Int()))
return uintptr(n), nil, err
default:
// Everything else is not yet supported.
return 0, nil, syserror.EINVAL
}
}
const (
_FADV_NORMAL = 0
_FADV_RANDOM = 1
_FADV_SEQUENTIAL = 2
_FADV_WILLNEED = 3
_FADV_DONTNEED = 4
_FADV_NOREUSE = 5
)
// Fadvise64 implements linux syscall fadvise64(2).
// This implementation currently ignores the provided advice.
func Fadvise64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
length := args[2].Int64()
advice := args[3].Int()
// Note: offset is allowed to be negative.
if length < 0 {
return 0, nil, syserror.EINVAL
}
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
// If the FD refers to a pipe or FIFO, return error.
if fs.IsPipe(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.ESPIPE
}
switch advice {
case _FADV_NORMAL:
case _FADV_RANDOM:
case _FADV_SEQUENTIAL:
case _FADV_WILLNEED:
case _FADV_DONTNEED:
case _FADV_NOREUSE:
default:
return 0, nil, syserror.EINVAL
}
// Sure, whatever.
return 0, nil, nil
}
func mkdirAt(t *kernel.Task, dirFD int32, addr usermem.Addr, mode linux.FileMode) error {
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
return fileOpAt(t, dirFD, path, func(root *fs.Dirent, d *fs.Dirent, name string, _ uint) error {
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Does this directory exist already?
remainingTraversals := uint(linux.MaxSymlinkTraversals)
f, err := t.MountNamespace().FindInode(t, root, d, name, &remainingTraversals)
switch err {
case nil:
// The directory existed.
defer f.DecRef()
return syserror.EEXIST
case syserror.EACCES:
// Permission denied while walking to the directory.
return err
default:
// Do we have write permissions on the parent?
if err := d.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
// Create the directory.
perms := fs.FilePermsFromMode(mode &^ linux.FileMode(t.FSContext().Umask()))
return d.CreateDirectory(t, root, name, perms)
}
})
}
// Mkdir implements linux syscall mkdir(2).
func Mkdir(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
mode := linux.FileMode(args[1].ModeT())
return 0, nil, mkdirAt(t, linux.AT_FDCWD, addr, mode)
}
// Mkdirat implements linux syscall mkdirat(2).
func Mkdirat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
mode := linux.FileMode(args[2].ModeT())
return 0, nil, mkdirAt(t, dirFD, addr, mode)
}
func rmdirAt(t *kernel.Task, dirFD int32, addr usermem.Addr) error {
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
// Special case: removing the root always returns EBUSY.
if path == "/" {
return syserror.EBUSY
}
return fileOpAt(t, dirFD, path, func(root *fs.Dirent, d *fs.Dirent, name string, _ uint) error {
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Linux returns different ernos when the path ends in single
// dot vs. double dots.
switch name {
case ".":
return syserror.EINVAL
case "..":
return syserror.ENOTEMPTY
}
if err := fs.MayDelete(t, root, d, name); err != nil {
return err
}
return d.RemoveDirectory(t, root, name)
})
}
// Rmdir implements linux syscall rmdir(2).
func Rmdir(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
return 0, nil, rmdirAt(t, linux.AT_FDCWD, addr)
}
func symlinkAt(t *kernel.Task, dirFD int32, newAddr usermem.Addr, oldAddr usermem.Addr) error {
newPath, dirPath, err := copyInPath(t, newAddr, false /* allowEmpty */)
if err != nil {
return err
}
if dirPath {
return syserror.ENOENT
}
// The oldPath is copied in verbatim. This is because the symlink
// will include all details, including trailing slashes.
oldPath, err := t.CopyInString(oldAddr, linux.PATH_MAX)
if err != nil {
return err
}
if oldPath == "" {
return syserror.ENOENT
}
return fileOpAt(t, dirFD, newPath, func(root *fs.Dirent, d *fs.Dirent, name string, _ uint) error {
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Make sure we have write permissions on the parent directory.
if err := d.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
return d.CreateLink(t, root, oldPath, name)
})
}
// Symlink implements linux syscall symlink(2).
func Symlink(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldAddr := args[0].Pointer()
newAddr := args[1].Pointer()
return 0, nil, symlinkAt(t, linux.AT_FDCWD, newAddr, oldAddr)
}
// Symlinkat implements linux syscall symlinkat(2).
func Symlinkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldAddr := args[0].Pointer()
dirFD := args[1].Int()
newAddr := args[2].Pointer()
return 0, nil, symlinkAt(t, dirFD, newAddr, oldAddr)
}
// mayLinkAt determines whether t can create a hard link to target.
//
// This corresponds to Linux's fs/namei.c:may_linkat.
func mayLinkAt(t *kernel.Task, target *fs.Inode) error {
// Linux will impose the following restrictions on hard links only if
// sysctl_protected_hardlinks is enabled. The kernel disables this
// setting by default for backward compatibility (see commit
// 561ec64ae67e), but also recommends that distributions enable it (and
// Debian does:
// https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=889098).
//
// gVisor currently behaves as though sysctl_protected_hardlinks is
// always enabled, and thus imposes the following restrictions on hard
// links.
if target.CheckOwnership(t) {
// fs/namei.c:may_linkat: "Source inode owner (or CAP_FOWNER)
// can hardlink all they like."
return nil
}
// If we are not the owner, then the file must be regular and have
// Read+Write permissions.
if !fs.IsRegular(target.StableAttr) {
return syserror.EPERM
}
if target.CheckPermission(t, fs.PermMask{Read: true, Write: true}) != nil {
return syserror.EPERM
}
return nil
}
// linkAt creates a hard link to the target specified by oldDirFD and oldAddr,
// specified by newDirFD and newAddr. If resolve is true, then the symlinks
// will be followed when evaluating the target.
func linkAt(t *kernel.Task, oldDirFD int32, oldAddr usermem.Addr, newDirFD int32, newAddr usermem.Addr, resolve, allowEmpty bool) error {
oldPath, _, err := copyInPath(t, oldAddr, allowEmpty)
if err != nil {
return err
}
newPath, dirPath, err := copyInPath(t, newAddr, false /* allowEmpty */)
if err != nil {
return err
}
if dirPath {
return syserror.ENOENT
}
if allowEmpty && oldPath == "" {
target := t.GetFile(oldDirFD)
if target == nil {
return syserror.EBADF
}
defer target.DecRef()
if err := mayLinkAt(t, target.Dirent.Inode); err != nil {
return err
}
// Resolve the target directory.
return fileOpAt(t, newDirFD, newPath, func(root *fs.Dirent, newParent *fs.Dirent, newName string, _ uint) error {
if !fs.IsDir(newParent.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Make sure we have write permissions on the parent directory.
if err := newParent.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
return newParent.CreateHardLink(t, root, target.Dirent, newName)
})
}
// Resolve oldDirFD and oldAddr to a dirent. The "resolve" argument
// only applies to this name.
return fileOpOn(t, oldDirFD, oldPath, resolve, func(root *fs.Dirent, target *fs.Dirent, _ uint) error {
if err := mayLinkAt(t, target.Inode); err != nil {
return err
}
// Next resolve newDirFD and newAddr to the parent dirent and name.
return fileOpAt(t, newDirFD, newPath, func(root *fs.Dirent, newParent *fs.Dirent, newName string, _ uint) error {
if !fs.IsDir(newParent.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Make sure we have write permissions on the parent directory.
if err := newParent.Inode.CheckPermission(t, fs.PermMask{Write: true, Execute: true}); err != nil {
return err
}
return newParent.CreateHardLink(t, root, target, newName)
})
})
}
// Link implements linux syscall link(2).
func Link(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldAddr := args[0].Pointer()
newAddr := args[1].Pointer()
// man link(2):
// POSIX.1-2001 says that link() should dereference oldpath if it is a
// symbolic link. However, since kernel 2.0, Linux does not do so: if
// oldpath is a symbolic link, then newpath is created as a (hard) link
// to the same symbolic link file (i.e., newpath becomes a symbolic
// link to the same file that oldpath refers to).
resolve := false
return 0, nil, linkAt(t, linux.AT_FDCWD, oldAddr, linux.AT_FDCWD, newAddr, resolve, false /* allowEmpty */)
}
// Linkat implements linux syscall linkat(2).
func Linkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldDirFD := args[0].Int()
oldAddr := args[1].Pointer()
newDirFD := args[2].Int()
newAddr := args[3].Pointer()
// man linkat(2):
// By default, linkat(), does not dereference oldpath if it is a
// symbolic link (like link(2)). Since Linux 2.6.18, the flag
// AT_SYMLINK_FOLLOW can be specified in flags to cause oldpath to be
// dereferenced if it is a symbolic link.
flags := args[4].Int()
// Sanity check flags.
if flags&^(linux.AT_SYMLINK_FOLLOW|linux.AT_EMPTY_PATH) != 0 {
return 0, nil, syserror.EINVAL
}
resolve := flags&linux.AT_SYMLINK_FOLLOW == linux.AT_SYMLINK_FOLLOW
allowEmpty := flags&linux.AT_EMPTY_PATH == linux.AT_EMPTY_PATH
if allowEmpty && !t.HasCapabilityIn(linux.CAP_DAC_READ_SEARCH, t.UserNamespace().Root()) {
return 0, nil, syserror.ENOENT
}
return 0, nil, linkAt(t, oldDirFD, oldAddr, newDirFD, newAddr, resolve, allowEmpty)
}
func readlinkAt(t *kernel.Task, dirFD int32, addr usermem.Addr, bufAddr usermem.Addr, size uint) (copied uintptr, err error) {
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, err
}
if dirPath {
return 0, syserror.ENOENT
}
err = fileOpOn(t, dirFD, path, false /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// Check for Read permission.
if err := d.Inode.CheckPermission(t, fs.PermMask{Read: true}); err != nil {
return err
}
s, err := d.Inode.Readlink(t)
if err == syserror.ENOLINK {
return syserror.EINVAL
}
if err != nil {
return err
}
buffer := []byte(s)
if uint(len(buffer)) > size {
buffer = buffer[:size]
}
n, err := t.CopyOutBytes(bufAddr, buffer)
// Update frame return value.
copied = uintptr(n)
return err
})
return copied, err // Return frame value.
}
// Readlink implements linux syscall readlink(2).
func Readlink(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
bufAddr := args[1].Pointer()
size := args[2].SizeT()
n, err := readlinkAt(t, linux.AT_FDCWD, addr, bufAddr, size)
return n, nil, err
}
// Readlinkat implements linux syscall readlinkat(2).
func Readlinkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
bufAddr := args[2].Pointer()
size := args[3].SizeT()
n, err := readlinkAt(t, dirFD, addr, bufAddr, size)
return n, nil, err
}
func unlinkAt(t *kernel.Task, dirFD int32, addr usermem.Addr) error {
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
return fileOpAt(t, dirFD, path, func(root *fs.Dirent, d *fs.Dirent, name string, _ uint) error {
if !fs.IsDir(d.Inode.StableAttr) {
return syserror.ENOTDIR
}
if err := fs.MayDelete(t, root, d, name); err != nil {
return err
}
return d.Remove(t, root, name, dirPath)
})
}
// Unlink implements linux syscall unlink(2).
func Unlink(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
return 0, nil, unlinkAt(t, linux.AT_FDCWD, addr)
}
// Unlinkat implements linux syscall unlinkat(2).
func Unlinkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
flags := args[2].Uint()
if flags&linux.AT_REMOVEDIR != 0 {
return 0, nil, rmdirAt(t, dirFD, addr)
}
return 0, nil, unlinkAt(t, dirFD, addr)
}
// Truncate implements linux syscall truncate(2).
func Truncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
length := args[1].Int64()
if length < 0 {
return 0, nil, syserror.EINVAL
}
path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return 0, nil, err
}
if dirPath {
return 0, nil, syserror.EINVAL
}
if uint64(length) >= t.ThreadGroup().Limits().Get(limits.FileSize).Cur {
t.SendSignal(&arch.SignalInfo{
Signo: int32(linux.SIGXFSZ),
Code: arch.SignalInfoUser,
})
return 0, nil, syserror.EFBIG
}
return 0, nil, fileOpOn(t, linux.AT_FDCWD, path, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
if fs.IsDir(d.Inode.StableAttr) {
return syserror.EISDIR
}
// In contrast to open(O_TRUNC), truncate(2) is only valid for file
// types.
if !fs.IsFile(d.Inode.StableAttr) {
return syserror.EINVAL
}
// Reject truncation if the access permissions do not allow truncation.
// This is different from the behavior of sys_ftruncate, see below.
if err := d.Inode.CheckPermission(t, fs.PermMask{Write: true}); err != nil {
return err
}
if err := d.Inode.Truncate(t, d, length); err != nil {
return err
}
// File length modified, generate notification.
d.InotifyEvent(linux.IN_MODIFY, 0)
return nil
})
}
// Ftruncate implements linux syscall ftruncate(2).
func Ftruncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
length := args[1].Int64()
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
// Reject truncation if the file flags do not permit this operation.
// This is different from truncate(2) above.
if !file.Flags().Write {
return 0, nil, syserror.EINVAL
}
// In contrast to open(O_TRUNC), truncate(2) is only valid for file
// types. Note that this is different from truncate(2) above, where a
// directory returns EISDIR.
if !fs.IsFile(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.EINVAL
}
if length < 0 {
return 0, nil, syserror.EINVAL
}
if uint64(length) >= t.ThreadGroup().Limits().Get(limits.FileSize).Cur {
t.SendSignal(&arch.SignalInfo{
Signo: int32(linux.SIGXFSZ),
Code: arch.SignalInfoUser,
})
return 0, nil, syserror.EFBIG
}
if err := file.Dirent.Inode.Truncate(t, file.Dirent, length); err != nil {
return 0, nil, err
}
// File length modified, generate notification.
file.Dirent.InotifyEvent(linux.IN_MODIFY, 0)
return 0, nil, nil
}
// Umask implements linux syscall umask(2).
func Umask(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
mask := args[0].ModeT()
mask = t.FSContext().SwapUmask(mask & 0777)
return uintptr(mask), nil, nil
}
// Change ownership of a file.
//
// uid and gid may be -1, in which case they will not be changed.
func chown(t *kernel.Task, d *fs.Dirent, uid auth.UID, gid auth.GID) error {
owner := fs.FileOwner{
UID: auth.NoID,
GID: auth.NoID,
}
uattr, err := d.Inode.UnstableAttr(t)
if err != nil {
return err
}
c := t.Credentials()
hasCap := d.Inode.CheckCapability(t, linux.CAP_CHOWN)
isOwner := uattr.Owner.UID == c.EffectiveKUID
if uid.Ok() {
kuid := c.UserNamespace.MapToKUID(uid)
// Valid UID must be supplied if UID is to be changed.
if !kuid.Ok() {
return syserror.EINVAL
}
// "Only a privileged process (CAP_CHOWN) may change the owner
// of a file." -chown(2)
//
// Linux also allows chown if you own the file and are
// explicitly not changing its UID.
isNoop := uattr.Owner.UID == kuid
if !(hasCap || (isOwner && isNoop)) {
return syserror.EPERM
}
owner.UID = kuid
}
if gid.Ok() {
kgid := c.UserNamespace.MapToKGID(gid)
// Valid GID must be supplied if GID is to be changed.
if !kgid.Ok() {
return syserror.EINVAL
}
// "The owner of a file may change the group of the file to any
// group of which that owner is a member. A privileged process
// (CAP_CHOWN) may change the group arbitrarily." -chown(2)
isNoop := uattr.Owner.GID == kgid
isMemberGroup := c.InGroup(kgid)
if !(hasCap || (isOwner && (isNoop || isMemberGroup))) {
return syserror.EPERM
}
owner.GID = kgid
}
// FIXME(b/62949101): This is racy; the inode's owner may have changed in
// the meantime. (Linux holds i_mutex while calling
// fs/attr.c:notify_change() => inode_operations::setattr =>
// inode_change_ok().)
if err := d.Inode.SetOwner(t, d, owner); err != nil {
return err
}
// When the owner or group are changed by an unprivileged user,
// chown(2) also clears the set-user-ID and set-group-ID bits, but
// we do not support them.
return nil
}
func chownAt(t *kernel.Task, fd int32, addr usermem.Addr, resolve, allowEmpty bool, uid auth.UID, gid auth.GID) error {
path, _, err := copyInPath(t, addr, allowEmpty)
if err != nil {
return err
}
if path == "" {
// Annoying. What's wrong with fchown?
file := t.GetFile(fd)
if file == nil {
return syserror.EBADF
}
defer file.DecRef()
return chown(t, file.Dirent, uid, gid)
}
return fileOpOn(t, fd, path, resolve, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
return chown(t, d, uid, gid)
})
}
// Chown implements linux syscall chown(2).
func Chown(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
uid := auth.UID(args[1].Uint())
gid := auth.GID(args[2].Uint())
return 0, nil, chownAt(t, linux.AT_FDCWD, addr, true /* resolve */, false /* allowEmpty */, uid, gid)
}
// Lchown implements linux syscall lchown(2).
func Lchown(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
uid := auth.UID(args[1].Uint())
gid := auth.GID(args[2].Uint())
return 0, nil, chownAt(t, linux.AT_FDCWD, addr, false /* resolve */, false /* allowEmpty */, uid, gid)
}
// Fchown implements linux syscall fchown(2).
func Fchown(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
uid := auth.UID(args[1].Uint())
gid := auth.GID(args[2].Uint())
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
return 0, nil, chown(t, file.Dirent, uid, gid)
}
// Fchownat implements Linux syscall fchownat(2).
func Fchownat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
addr := args[1].Pointer()
uid := auth.UID(args[2].Uint())
gid := auth.GID(args[3].Uint())
flags := args[4].Int()
if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
return 0, nil, syserror.EINVAL
}
return 0, nil, chownAt(t, dirFD, addr, flags&linux.AT_SYMLINK_NOFOLLOW == 0, flags&linux.AT_EMPTY_PATH != 0, uid, gid)
}
func chmod(t *kernel.Task, d *fs.Dirent, mode linux.FileMode) error {
// Must own file to change mode.
if !d.Inode.CheckOwnership(t) {
return syserror.EPERM
}
p := fs.FilePermsFromMode(mode)
if !d.Inode.SetPermissions(t, d, p) {
return syserror.EPERM
}
// File attribute changed, generate notification.
d.InotifyEvent(linux.IN_ATTRIB, 0)
return nil
}
func chmodAt(t *kernel.Task, fd int32, addr usermem.Addr, mode linux.FileMode) error {
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
return fileOpOn(t, fd, path, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
return chmod(t, d, mode)
})
}
// Chmod implements linux syscall chmod(2).
func Chmod(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
mode := linux.FileMode(args[1].ModeT())
return 0, nil, chmodAt(t, linux.AT_FDCWD, addr, mode)
}
// Fchmod implements linux syscall fchmod(2).
func Fchmod(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
mode := linux.FileMode(args[1].ModeT())
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
return 0, nil, chmod(t, file.Dirent, mode)
}
// Fchmodat implements linux syscall fchmodat(2).
func Fchmodat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
addr := args[1].Pointer()
mode := linux.FileMode(args[2].ModeT())
return 0, nil, chmodAt(t, fd, addr, mode)
}
// defaultSetToSystemTimeSpec returns a TimeSpec that will set ATime and MTime
// to the system time.
func defaultSetToSystemTimeSpec() fs.TimeSpec {
return fs.TimeSpec{
ATimeSetSystemTime: true,
MTimeSetSystemTime: true,
}
}
func utimes(t *kernel.Task, dirFD int32, addr usermem.Addr, ts fs.TimeSpec, resolve bool) error {
setTimestamp := func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
// Does the task own the file?
if !d.Inode.CheckOwnership(t) {
// Trying to set a specific time? Must be owner.
if (ts.ATimeOmit || !ts.ATimeSetSystemTime) && (ts.MTimeOmit || !ts.MTimeSetSystemTime) {
return syserror.EPERM
}
// Trying to set to current system time? Must have write access.
if err := d.Inode.CheckPermission(t, fs.PermMask{Write: true}); err != nil {
return err
}
}
if err := d.Inode.SetTimestamps(t, d, ts); err != nil {
return err
}
// File attribute changed, generate notification.
d.InotifyEvent(linux.IN_ATTRIB, 0)
return nil
}
// From utimes.c:
// "If filename is NULL and dfd refers to an open file, then operate on
// the file. Otherwise look up filename, possibly using dfd as a
// starting point."
if addr == 0 && dirFD != linux.AT_FDCWD {
if !resolve {
// Linux returns EINVAL in this case. See utimes.c.
return syserror.EINVAL
}
f := t.GetFile(dirFD)
if f == nil {
return syserror.EBADF
}
defer f.DecRef()
root := t.FSContext().RootDirectory()
defer root.DecRef()
return setTimestamp(root, f.Dirent, linux.MaxSymlinkTraversals)
}
path, _, err := copyInPath(t, addr, false /* allowEmpty */)
if err != nil {
return err
}
return fileOpOn(t, dirFD, path, resolve, setTimestamp)
}
// Utime implements linux syscall utime(2).
func Utime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
filenameAddr := args[0].Pointer()
timesAddr := args[1].Pointer()
// No timesAddr argument will be interpreted as current system time.
ts := defaultSetToSystemTimeSpec()
if timesAddr != 0 {
var times linux.Utime
if _, err := t.CopyIn(timesAddr, ×); err != nil {
return 0, nil, err
}
ts = fs.TimeSpec{
ATime: ktime.FromSeconds(times.Actime),
MTime: ktime.FromSeconds(times.Modtime),
}
}
return 0, nil, utimes(t, linux.AT_FDCWD, filenameAddr, ts, true)
}
// Utimes implements linux syscall utimes(2).
func Utimes(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
filenameAddr := args[0].Pointer()
timesAddr := args[1].Pointer()
// No timesAddr argument will be interpreted as current system time.
ts := defaultSetToSystemTimeSpec()
if timesAddr != 0 {
var times [2]linux.Timeval
if _, err := t.CopyIn(timesAddr, ×); err != nil {
return 0, nil, err
}
ts = fs.TimeSpec{
ATime: ktime.FromTimeval(times[0]),
MTime: ktime.FromTimeval(times[1]),
}
}
return 0, nil, utimes(t, linux.AT_FDCWD, filenameAddr, ts, true)
}
// timespecIsValid checks that the timespec is valid for use in utimensat.
func timespecIsValid(ts linux.Timespec) bool {
// Nsec must be UTIME_OMIT, UTIME_NOW, or less than 10^9.
return ts.Nsec == linux.UTIME_OMIT || ts.Nsec == linux.UTIME_NOW || ts.Nsec < 1e9
}
// Utimensat implements linux syscall utimensat(2).
func Utimensat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
pathnameAddr := args[1].Pointer()
timesAddr := args[2].Pointer()
flags := args[3].Int()
// No timesAddr argument will be interpreted as current system time.
ts := defaultSetToSystemTimeSpec()
if timesAddr != 0 {
var times [2]linux.Timespec
if _, err := t.CopyIn(timesAddr, ×); err != nil {
return 0, nil, err
}
if !timespecIsValid(times[0]) || !timespecIsValid(times[1]) {
return 0, nil, syserror.EINVAL
}
// If both are UTIME_OMIT, this is a noop.
if times[0].Nsec == linux.UTIME_OMIT && times[1].Nsec == linux.UTIME_OMIT {
return 0, nil, nil
}
ts = fs.TimeSpec{
ATime: ktime.FromTimespec(times[0]),
ATimeOmit: times[0].Nsec == linux.UTIME_OMIT,
ATimeSetSystemTime: times[0].Nsec == linux.UTIME_NOW,
MTime: ktime.FromTimespec(times[1]),
MTimeOmit: times[1].Nsec == linux.UTIME_OMIT,
MTimeSetSystemTime: times[0].Nsec == linux.UTIME_NOW,
}
}
return 0, nil, utimes(t, dirFD, pathnameAddr, ts, flags&linux.AT_SYMLINK_NOFOLLOW == 0)
}
// Futimesat implements linux syscall futimesat(2).
func Futimesat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
dirFD := args[0].Int()
pathnameAddr := args[1].Pointer()
timesAddr := args[2].Pointer()
// No timesAddr argument will be interpreted as current system time.
ts := defaultSetToSystemTimeSpec()
if timesAddr != 0 {
var times [2]linux.Timeval
if _, err := t.CopyIn(timesAddr, ×); err != nil {
return 0, nil, err
}
if times[0].Usec >= 1e6 || times[0].Usec < 0 ||
times[1].Usec >= 1e6 || times[1].Usec < 0 {
return 0, nil, syserror.EINVAL
}
ts = fs.TimeSpec{
ATime: ktime.FromTimeval(times[0]),
MTime: ktime.FromTimeval(times[1]),
}
}
return 0, nil, utimes(t, dirFD, pathnameAddr, ts, true)
}
func renameAt(t *kernel.Task, oldDirFD int32, oldAddr usermem.Addr, newDirFD int32, newAddr usermem.Addr) error {
newPath, _, err := copyInPath(t, newAddr, false /* allowEmpty */)
if err != nil {
return err
}
oldPath, _, err := copyInPath(t, oldAddr, false /* allowEmpty */)
if err != nil {
return err
}
return fileOpAt(t, oldDirFD, oldPath, func(root *fs.Dirent, oldParent *fs.Dirent, oldName string, _ uint) error {
if !fs.IsDir(oldParent.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Rename rejects paths that end in ".", "..", or empty (i.e.
// the root) with EBUSY.
switch oldName {
case "", ".", "..":
return syserror.EBUSY
}
return fileOpAt(t, newDirFD, newPath, func(root *fs.Dirent, newParent *fs.Dirent, newName string, _ uint) error {
if !fs.IsDir(newParent.Inode.StableAttr) {
return syserror.ENOTDIR
}
// Rename rejects paths that end in ".", "..", or empty
// (i.e. the root) with EBUSY.
switch newName {
case "", ".", "..":
return syserror.EBUSY
}
return fs.Rename(t, root, oldParent, oldName, newParent, newName)
})
})
}
// Rename implements linux syscall rename(2).
func Rename(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldPathAddr := args[0].Pointer()
newPathAddr := args[1].Pointer()
return 0, nil, renameAt(t, linux.AT_FDCWD, oldPathAddr, linux.AT_FDCWD, newPathAddr)
}
// Renameat implements linux syscall renameat(2).
func Renameat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
oldDirFD := args[0].Int()
oldPathAddr := args[1].Pointer()
newDirFD := args[2].Int()
newPathAddr := args[3].Pointer()
return 0, nil, renameAt(t, oldDirFD, oldPathAddr, newDirFD, newPathAddr)
}
// Fallocate implements linux system call fallocate(2).
func Fallocate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
mode := args[1].Int64()
offset := args[2].Int64()
length := args[3].Int64()
file := t.GetFile(fd)
if file == nil {
return 0, nil, syserror.EBADF
}
defer file.DecRef()
if offset < 0 || length <= 0 {
return 0, nil, syserror.EINVAL
}
if mode != 0 {
t.Kernel().EmitUnimplementedEvent(t)
return 0, nil, syserror.ENOTSUP
}
if !file.Flags().Write {
return 0, nil, syserror.EBADF
}
if fs.IsPipe(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.ESPIPE
}
if fs.IsDir(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.EISDIR
}
if !fs.IsRegular(file.Dirent.Inode.StableAttr) {
return 0, nil, syserror.ENODEV
}
size := offset + length
if size < 0 {
return 0, nil, syserror.EFBIG
}
if uint64(size) >= t.ThreadGroup().Limits().Get(limits.FileSize).Cur {
t.SendSignal(&arch.SignalInfo{
Signo: int32(linux.SIGXFSZ),
Code: arch.SignalInfoUser,
})
return 0, nil, syserror.EFBIG
}
if err := file.Dirent.Inode.Allocate(t, file.Dirent, offset, length); err != nil {
return 0, nil, err
}
// File length modified, generate notification.
file.Dirent.InotifyEvent(linux.IN_MODIFY, 0)
return 0, nil, nil
}
// Flock implements linux syscall flock(2).
func Flock(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
fd := args[0].Int()
operation := args[1].Int()
file := t.GetFile(fd)
if file == nil {
// flock(2): EBADF fd is not an open file descriptor.
return 0, nil, syserror.EBADF
}
defer file.DecRef()
nonblocking := operation&linux.LOCK_NB != 0
operation &^= linux.LOCK_NB
// flock(2):
// Locks created by flock() are associated with an open file table entry. This means that
// duplicate file descriptors (created by, for example, fork(2) or dup(2)) refer to the
// same lock, and this lock may be modified or released using any of these descriptors. Furthermore,
// the lock is released either by an explicit LOCK_UN operation on any of these duplicate
// descriptors, or when all such descriptors have been closed.
//
// If a process uses open(2) (or similar) to obtain more than one descriptor for the same file,
// these descriptors are treated independently by flock(). An attempt to lock the file using
// one of these file descriptors may be denied by a lock that the calling process has already placed via
// another descriptor.
//
// We use the File UniqueID as the lock UniqueID because it needs to reference the same lock across dup(2)
// and fork(2).
lockUniqueID := lock.UniqueID(file.UniqueID)
// A BSD style lock spans the entire file.
rng := lock.LockRange{
Start: 0,
End: lock.LockEOF,
}
switch operation {
case linux.LOCK_EX:
if nonblocking {
// Since we're nonblocking we pass a nil lock.Blocker implementation.
if !file.Dirent.Inode.LockCtx.BSD.LockRegion(lockUniqueID, lock.WriteLock, rng, nil) {
return 0, nil, syserror.EWOULDBLOCK
}
} else {
// Because we're blocking we will pass the task to satisfy the lock.Blocker interface.
if !file.Dirent.Inode.LockCtx.BSD.LockRegion(lockUniqueID, lock.WriteLock, rng, t) {
return 0, nil, syserror.EINTR
}
}
case linux.LOCK_SH:
if nonblocking {
// Since we're nonblocking we pass a nil lock.Blocker implementation.
if !file.Dirent.Inode.LockCtx.BSD.LockRegion(lockUniqueID, lock.ReadLock, rng, nil) {
return 0, nil, syserror.EWOULDBLOCK
}
} else {
// Because we're blocking we will pass the task to satisfy the lock.Blocker interface.
if !file.Dirent.Inode.LockCtx.BSD.LockRegion(lockUniqueID, lock.ReadLock, rng, t) {
return 0, nil, syserror.EINTR
}
}
case linux.LOCK_UN:
file.Dirent.Inode.LockCtx.BSD.UnlockRegion(lockUniqueID, rng)
default:
// flock(2): EINVAL operation is invalid.
return 0, nil, syserror.EINVAL
}
return 0, nil, nil
}
const (
memfdPrefix = "/memfd:"
memfdAllFlags = uint32(linux.MFD_CLOEXEC | linux.MFD_ALLOW_SEALING)
memfdMaxNameLen = linux.NAME_MAX - len(memfdPrefix) + 1
)
// MemfdCreate implements the linux syscall memfd_create(2).
func MemfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
addr := args[0].Pointer()
flags := args[1].Uint()
if flags&^memfdAllFlags != 0 {
// Unknown bits in flags.
return 0, nil, syserror.EINVAL
}
allowSeals := flags&linux.MFD_ALLOW_SEALING != 0
cloExec := flags&linux.MFD_CLOEXEC != 0
name, err := t.CopyInString(addr, syscall.PathMax-len(memfdPrefix))
if err != nil {
return 0, nil, err
}
if len(name) > memfdMaxNameLen {
return 0, nil, syserror.EINVAL
}
name = memfdPrefix + name
inode := tmpfs.NewMemfdInode(t, allowSeals)
dirent := fs.NewDirent(t, inode, name)
// Per Linux, mm/shmem.c:__shmem_file_setup(), memfd files are set up with
// FMODE_READ | FMODE_WRITE.
file, err := inode.GetFile(t, dirent, fs.FileFlags{Read: true, Write: true})
if err != nil {
return 0, nil, err
}
defer dirent.DecRef()
defer file.DecRef()
newFD, err := t.NewFDFrom(0, file, kernel.FDFlags{
CloseOnExec: cloExec,
})
if err != nil {
return 0, nil, err
}
return uintptr(newFD), nil, nil
}
|