summaryrefslogtreecommitdiff
path: root/src/basic/fileio.c
blob: 51b11ab9c7d12e85dc76500c3216e82deb9cc43b (plain)
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
/* SPDX-License-Identifier: LGPL-2.1-or-later */

#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <stdarg.h>
#include <stdint.h>
#include <stdio_ext.h>
#include <stdlib.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <unistd.h>

#include "alloc-util.h"
#include "chase-symlinks.h"
#include "fd-util.h"
#include "fileio.h"
#include "fs-util.h"
#include "hexdecoct.h"
#include "log.h"
#include "macro.h"
#include "mkdir.h"
#include "nulstr-util.h"
#include "parse-util.h"
#include "path-util.h"
#include "socket-util.h"
#include "stdio-util.h"
#include "string-util.h"
#include "sync-util.h"
#include "tmpfile-util.h"

/* The maximum size of the file we'll read in one go in read_full_file() (64M). */
#define READ_FULL_BYTES_MAX (64U*1024U*1024U - 1U)

/* The maximum size of virtual files (i.e. procfs, sysfs, and other virtual "API" files) we'll read in one go
 * in read_virtual_file(). Note that this limit is different (and much lower) than the READ_FULL_BYTES_MAX
 * limit. This reflects the fact that we use different strategies for reading virtual and regular files:
 * virtual files we generally have to read in a single read() syscall since the kernel doesn't support
 * continuation read()s for them. Thankfully they are somewhat size constrained. Thus we can allocate the
 * full potential buffer in advance. Regular files OTOH can be much larger, and there we grow the allocations
 * exponentially in a loop. We use a size limit of 4M-2 because 4M-1 is the maximum buffer that /proc/sys/
 * allows us to read() (larger reads will fail with ENOMEM), and we want to read one extra byte so that we
 * can detect EOFs. */
#define READ_VIRTUAL_BYTES_MAX (4U*1024U*1024U - 2U)

int fopen_unlocked_at(int dir_fd, const char *path, const char *options, int flags, FILE **ret) {
        int r;

        assert(ret);

        r = xfopenat(dir_fd, path, options, flags, ret);
        if (r < 0)
                return r;

        (void) __fsetlocking(*ret, FSETLOCKING_BYCALLER);

        return 0;
}

int fdopen_unlocked(int fd, const char *options, FILE **ret) {
        assert(ret);

        FILE *f = fdopen(fd, options);
        if (!f)
                return -errno;

        (void) __fsetlocking(f, FSETLOCKING_BYCALLER);

        *ret = f;
        return 0;
}

int take_fdopen_unlocked(int *fd, const char *options, FILE **ret) {
        int r;

        assert(fd);

        r = fdopen_unlocked(*fd, options, ret);
        if (r < 0)
                return r;

        *fd = -EBADF;

        return 0;
}

FILE* take_fdopen(int *fd, const char *options) {
        assert(fd);

        FILE *f = fdopen(*fd, options);
        if (!f)
                return NULL;

        *fd = -EBADF;

        return f;
}

DIR* take_fdopendir(int *dfd) {
        assert(dfd);

        DIR *d = fdopendir(*dfd);
        if (!d)
                return NULL;

        *dfd = -EBADF;

        return d;
}

FILE* open_memstream_unlocked(char **ptr, size_t *sizeloc) {
        FILE *f = open_memstream(ptr, sizeloc);
        if (!f)
                return NULL;

        (void) __fsetlocking(f, FSETLOCKING_BYCALLER);

        return f;
}

FILE* fmemopen_unlocked(void *buf, size_t size, const char *mode) {
        FILE *f = fmemopen(buf, size, mode);
        if (!f)
                return NULL;

        (void) __fsetlocking(f, FSETLOCKING_BYCALLER);

        return f;
}

int write_string_stream_ts(
                FILE *f,
                const char *line,
                WriteStringFileFlags flags,
                const struct timespec *ts) {

        bool needs_nl;
        int r, fd = -EBADF;

        assert(f);
        assert(line);

        if (ferror(f))
                return -EIO;

        if (ts) {
                /* If we shall set the timestamp we need the fd. But fmemopen() streams generally don't have
                 * an fd. Let's fail early in that case. */
                fd = fileno(f);
                if (fd < 0)
                        return -EBADF;
        }

        if (flags & WRITE_STRING_FILE_SUPPRESS_REDUNDANT_VIRTUAL) {
                _cleanup_free_ char *t = NULL;

                /* If value to be written is same as that of the existing value, then suppress the write. */

                if (fd < 0) {
                        fd = fileno(f);
                        if (fd < 0)
                                return -EBADF;
                }

                /* Read an additional byte to detect cases where the prefix matches but the rest
                 * doesn't. Also, 0 returned by read_virtual_file_fd() means the read was truncated and
                 * it won't be equal to the new value. */
                if (read_virtual_file_fd(fd, strlen(line)+1, &t, NULL) > 0 &&
                    streq_skip_trailing_chars(line, t, NEWLINE)) {
                        log_debug("No change in value '%s', suppressing write", line);
                        return 0;
                }

                if (lseek(fd, 0, SEEK_SET) < 0)
                        return -errno;
        }

        needs_nl = !(flags & WRITE_STRING_FILE_AVOID_NEWLINE) && !endswith(line, "\n");

        if (needs_nl && (flags & WRITE_STRING_FILE_DISABLE_BUFFER)) {
                /* If STDIO buffering was disabled, then let's append the newline character to the string
                 * itself, so that the write goes out in one go, instead of two */

                line = strjoina(line, "\n");
                needs_nl = false;
        }

        if (fputs(line, f) == EOF)
                return -errno;

        if (needs_nl)
                if (fputc('\n', f) == EOF)
                        return -errno;

        if (flags & WRITE_STRING_FILE_SYNC)
                r = fflush_sync_and_check(f);
        else
                r = fflush_and_check(f);
        if (r < 0)
                return r;

        if (ts) {
                const struct timespec twice[2] = {*ts, *ts};

                assert(fd >= 0);
                if (futimens(fd, twice) < 0)
                        return -errno;
        }

        return 0;
}

static int write_string_file_atomic_at(
                int dir_fd,
                const char *fn,
                const char *line,
                WriteStringFileFlags flags,
                const struct timespec *ts) {

        _cleanup_fclose_ FILE *f = NULL;
        _cleanup_free_ char *p = NULL;
        int r;

        assert(fn);
        assert(line);

        /* Note that we'd really like to use O_TMPFILE here, but can't really, since we want replacement
         * semantics here, and O_TMPFILE can't offer that. i.e. rename() replaces but linkat() doesn't. */

        r = fopen_temporary_at(dir_fd, fn, &f, &p);
        if (r < 0)
                return r;

        r = write_string_stream_ts(f, line, flags, ts);
        if (r < 0)
                goto fail;

        r = fchmod_umask(fileno(f), FLAGS_SET(flags, WRITE_STRING_FILE_MODE_0600) ? 0600 : 0644);
        if (r < 0)
                goto fail;

        if (renameat(dir_fd, p, dir_fd, fn) < 0) {
                r = -errno;
                goto fail;
        }

        if (FLAGS_SET(flags, WRITE_STRING_FILE_SYNC)) {
                /* Sync the rename, too */
                r = fsync_directory_of_file(fileno(f));
                if (r < 0)
                        return r;
        }

        return 0;

fail:
        (void) unlinkat(dir_fd, p, 0);
        return r;
}

int write_string_file_ts_at(
                int dir_fd,
                const char *fn,
                const char *line,
                WriteStringFileFlags flags,
                const struct timespec *ts) {

        _cleanup_fclose_ FILE *f = NULL;
        int q, r, fd;

        assert(fn);
        assert(line);

        /* We don't know how to verify whether the file contents was already on-disk. */
        assert(!((flags & WRITE_STRING_FILE_VERIFY_ON_FAILURE) && (flags & WRITE_STRING_FILE_SYNC)));

        if (flags & WRITE_STRING_FILE_MKDIR_0755) {
                r = mkdirat_parents(dir_fd, fn, 0755);
                if (r < 0)
                        return r;
        }

        if (flags & WRITE_STRING_FILE_ATOMIC) {
                assert(flags & WRITE_STRING_FILE_CREATE);

                r = write_string_file_atomic_at(dir_fd, fn, line, flags, ts);
                if (r < 0)
                        goto fail;

                return r;
        } else
                assert(!ts);

        /* We manually build our own version of fopen(..., "we") that works without O_CREAT and with O_NOFOLLOW if needed. */
        fd = openat(dir_fd, fn, O_CLOEXEC|O_NOCTTY |
                    (FLAGS_SET(flags, WRITE_STRING_FILE_NOFOLLOW) ? O_NOFOLLOW : 0) |
                    (FLAGS_SET(flags, WRITE_STRING_FILE_CREATE) ? O_CREAT : 0) |
                    (FLAGS_SET(flags, WRITE_STRING_FILE_TRUNCATE) ? O_TRUNC : 0) |
                    (FLAGS_SET(flags, WRITE_STRING_FILE_SUPPRESS_REDUNDANT_VIRTUAL) ? O_RDWR : O_WRONLY),
                    (FLAGS_SET(flags, WRITE_STRING_FILE_MODE_0600) ? 0600 : 0666));
        if (fd < 0) {
                r = -errno;
                goto fail;
        }

        r = fdopen_unlocked(fd, "w", &f);
        if (r < 0) {
                safe_close(fd);
                goto fail;
        }

        if (flags & WRITE_STRING_FILE_DISABLE_BUFFER)
                setvbuf(f, NULL, _IONBF, 0);

        r = write_string_stream_ts(f, line, flags, ts);
        if (r < 0)
                goto fail;

        return 0;

fail:
        if (!(flags & WRITE_STRING_FILE_VERIFY_ON_FAILURE))
                return r;

        f = safe_fclose(f);

        /* OK, the operation failed, but let's see if the right
         * contents in place already. If so, eat up the error. */

        q = verify_file(fn, line, !(flags & WRITE_STRING_FILE_AVOID_NEWLINE) || (flags & WRITE_STRING_FILE_VERIFY_IGNORE_NEWLINE));
        if (q <= 0)
                return r;

        return 0;
}

int write_string_filef(
                const char *fn,
                WriteStringFileFlags flags,
                const char *format, ...) {

        _cleanup_free_ char *p = NULL;
        va_list ap;
        int r;

        va_start(ap, format);
        r = vasprintf(&p, format, ap);
        va_end(ap);

        if (r < 0)
                return -ENOMEM;

        return write_string_file(fn, p, flags);
}

int read_one_line_file(const char *fn, char **line) {
        _cleanup_fclose_ FILE *f = NULL;
        int r;

        assert(fn);
        assert(line);

        r = fopen_unlocked(fn, "re", &f);
        if (r < 0)
                return r;

        return read_line(f, LONG_LINE_MAX, line);
}

int verify_file_at(int dir_fd, const char *fn, const char *blob, bool accept_extra_nl) {
        _cleanup_fclose_ FILE *f = NULL;
        _cleanup_free_ char *buf = NULL;
        size_t l, k;
        int r;

        assert(fn);
        assert(blob);

        l = strlen(blob);

        if (accept_extra_nl && endswith(blob, "\n"))
                accept_extra_nl = false;

        buf = malloc(l + accept_extra_nl + 1);
        if (!buf)
                return -ENOMEM;

        r = fopen_unlocked_at(dir_fd, fn, "re", 0, &f);
        if (r < 0)
                return r;

        /* We try to read one byte more than we need, so that we know whether we hit eof */
        errno = 0;
        k = fread(buf, 1, l + accept_extra_nl + 1, f);
        if (ferror(f))
                return errno_or_else(EIO);

        if (k != l && k != l + accept_extra_nl)
                return 0;
        if (memcmp(buf, blob, l) != 0)
                return 0;
        if (k > l && buf[l] != '\n')
                return 0;

        return 1;
}

int read_virtual_file_fd(int fd, size_t max_size, char **ret_contents, size_t *ret_size) {
        _cleanup_free_ char *buf = NULL;
        size_t n, size;
        int n_retries;
        bool truncated = false;

        /* Virtual filesystems such as sysfs or procfs use kernfs, and kernfs can work with two sorts of
         * virtual files. One sort uses "seq_file", and the results of the first read are buffered for the
         * second read. The other sort uses "raw" reads which always go direct to the device. In the latter
         * case, the content of the virtual file must be retrieved with a single read otherwise a second read
         * might get the new value instead of finding EOF immediately. That's the reason why the usage of
         * fread(3) is prohibited in this case as it always performs a second call to read(2) looking for
         * EOF. See issue #13585.
         *
         * max_size specifies a limit on the bytes read. If max_size is SIZE_MAX, the full file is read. If
         * the full file is too large to read, an error is returned. For other values of max_size, *partial
         * contents* may be returned. (Though the read is still done using one syscall.) Returns 0 on
         * partial success, 1 if untruncated contents were read. */

        assert(fd >= 0);
        assert(max_size <= READ_VIRTUAL_BYTES_MAX || max_size == SIZE_MAX);

        /* Limit the number of attempts to read the number of bytes returned by fstat(). */
        n_retries = 3;

        for (;;) {
                struct stat st;

                if (fstat(fd, &st) < 0)
                        return -errno;

                if (!S_ISREG(st.st_mode))
                        return -EBADF;

                /* Be prepared for files from /proc which generally report a file size of 0. */
                assert_cc(READ_VIRTUAL_BYTES_MAX < SSIZE_MAX);
                if (st.st_size > 0 && n_retries > 1) {
                        /* Let's use the file size if we have more than 1 attempt left. On the last attempt
                         * we'll ignore the file size */

                        if (st.st_size > SSIZE_MAX) { /* Avoid overflow with 32-bit size_t and 64-bit off_t. */

                                if (max_size == SIZE_MAX)
                                        return -EFBIG;

                                size = max_size;
                        } else {
                                size = MIN((size_t) st.st_size, max_size);

                                if (size > READ_VIRTUAL_BYTES_MAX)
                                        return -EFBIG;
                        }

                        n_retries--;
                } else if (n_retries > 1) {
                        /* Files in /proc are generally smaller than the page size so let's start with
                         * a page size buffer from malloc and only use the max buffer on the final try. */
                        size = MIN3(page_size() - 1, READ_VIRTUAL_BYTES_MAX, max_size);
                        n_retries = 1;
                } else {
                        size = MIN(READ_VIRTUAL_BYTES_MAX, max_size);
                        n_retries = 0;
                }

                buf = malloc(size + 1);
                if (!buf)
                        return -ENOMEM;

                /* Use a bigger allocation if we got it anyway, but not more than the limit. */
                size = MIN3(MALLOC_SIZEOF_SAFE(buf) - 1, max_size, READ_VIRTUAL_BYTES_MAX);

                for (;;) {
                        ssize_t k;

                        /* Read one more byte so we can detect whether the content of the
                         * file has already changed or the guessed size for files from /proc
                         * wasn't large enough . */
                        k = read(fd, buf, size + 1);
                        if (k >= 0) {
                                n = k;
                                break;
                        }

                        if (errno != EINTR)
                                return -errno;
                }

                /* Consider a short read as EOF */
                if (n <= size)
                        break;

                /* If a maximum size is specified and we already read more we know the file is larger, and
                 * can handle this as truncation case. Note that if the size of what we read equals the
                 * maximum size then this doesn't mean truncation, the file might or might not end on that
                 * byte. We need to rerun the loop in that case, with a larger buffer size, so that we read
                 * at least one more byte to be able to distinguish EOF from truncation. */
                if (max_size != SIZE_MAX && n > max_size) {
                        n = size; /* Make sure we never use more than what we sized the buffer for (so that
                                   * we have one free byte in it for the trailing NUL we add below).*/
                        truncated = true;
                        break;
                }

                /* We have no further attempts left? Then the file is apparently larger than our limits. Give up. */
                if (n_retries <= 0)
                        return -EFBIG;

                /* Hmm... either we read too few bytes from /proc or less likely the content of the file
                 * might have been changed (and is now bigger) while we were processing, let's try again
                 * either with the new file size. */

                if (lseek(fd, 0, SEEK_SET) < 0)
                        return -errno;

                buf = mfree(buf);
        }

        if (ret_contents) {

                /* Safety check: if the caller doesn't want to know the size of what we just read it will
                 * rely on the trailing NUL byte. But if there's an embedded NUL byte, then we should refuse
                 * operation as otherwise there'd be ambiguity about what we just read. */
                if (!ret_size && memchr(buf, 0, n))
                        return -EBADMSG;

                if (n < size) {
                        char *p;

                        /* Return rest of the buffer to libc */
                        p = realloc(buf, n + 1);
                        if (!p)
                                return -ENOMEM;
                        buf = p;
                }

                buf[n] = 0;
                *ret_contents = TAKE_PTR(buf);
        }

        if (ret_size)
                *ret_size = n;

        return !truncated;
}

int read_virtual_file_at(
                int dir_fd,
                const char *filename,
                size_t max_size,
                char **ret_contents,
                size_t *ret_size) {

        _cleanup_close_ int fd = -EBADF;

        assert(dir_fd >= 0 || dir_fd == AT_FDCWD);

        if (!filename) {
                if (dir_fd == AT_FDCWD)
                        return -EBADF;

                return read_virtual_file_fd(dir_fd, max_size, ret_contents, ret_size);
        }

        fd = openat(dir_fd, filename, O_RDONLY | O_NOCTTY | O_CLOEXEC);
        if (fd < 0)
                return -errno;

        return read_virtual_file_fd(fd, max_size, ret_contents, ret_size);
}

int read_full_stream_full(
                FILE *f,
                const char *filename,
                uint64_t offset,
                size_t size,
                ReadFullFileFlags flags,
                char **ret_contents,
                size_t *ret_size) {

        _cleanup_free_ char *buf = NULL;
        size_t n, n_next = 0, l;
        int fd, r;

        assert(f);
        assert(ret_contents);
        assert(!FLAGS_SET(flags, READ_FULL_FILE_UNBASE64 | READ_FULL_FILE_UNHEX));
        assert(size != SIZE_MAX || !FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER));

        if (offset != UINT64_MAX && offset > LONG_MAX) /* fseek() can only deal with "long" offsets */
                return -ERANGE;

        fd = fileno(f);
        if (fd >= 0) { /* If the FILE* object is backed by an fd (as opposed to memory or such, see
                        * fmemopen()), let's optimize our buffering */
                struct stat st;

                if (fstat(fd, &st) < 0)
                        return -errno;

                if (S_ISREG(st.st_mode)) {

                        /* Try to start with the right file size if we shall read the file in full. Note
                         * that we increase the size to read here by one, so that the first read attempt
                         * already makes us notice the EOF. If the reported size of the file is zero, we
                         * avoid this logic however, since quite likely it might be a virtual file in procfs
                         * that all report a zero file size. */

                        if (st.st_size > 0 &&
                            (size == SIZE_MAX || FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER))) {

                                uint64_t rsize =
                                        LESS_BY((uint64_t) st.st_size, offset == UINT64_MAX ? 0 : offset);

                                if (rsize < SIZE_MAX) /* overflow check */
                                        n_next = rsize + 1;
                        }

                        if (flags & READ_FULL_FILE_WARN_WORLD_READABLE)
                                (void) warn_file_is_world_accessible(filename, &st, NULL, 0);
                }
        }

        /* If we don't know how much to read, figure it out now. If we shall read a part of the file, then
         * allocate the requested size. If we shall load the full file start with LINE_MAX. Note that if
         * READ_FULL_FILE_FAIL_WHEN_LARGER we consider the specified size a safety limit, and thus also start
         * with LINE_MAX, under assumption the file is most likely much shorter. */
        if (n_next == 0)
                n_next = size != SIZE_MAX && !FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER) ? size : LINE_MAX;

        /* Never read more than we need to determine that our own limit is hit */
        if (n_next > READ_FULL_BYTES_MAX)
                n_next = READ_FULL_BYTES_MAX + 1;

        if (offset != UINT64_MAX && fseek(f, offset, SEEK_SET) < 0)
                return -errno;

        n = l = 0;
        for (;;) {
                char *t;
                size_t k;

                /* If we shall fail when reading overly large data, then read exactly one byte more than the
                 * specified size at max, since that'll tell us if there's anymore data beyond the limit*/
                if (FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER) && n_next > size)
                        n_next = size + 1;

                if (flags & READ_FULL_FILE_SECURE) {
                        t = malloc(n_next + 1);
                        if (!t) {
                                r = -ENOMEM;
                                goto finalize;
                        }
                        memcpy_safe(t, buf, n);
                        explicit_bzero_safe(buf, n);
                        free(buf);
                } else {
                        t = realloc(buf, n_next + 1);
                        if (!t)
                                return -ENOMEM;
                }

                buf = t;
                /* Unless a size has been explicitly specified, try to read as much as fits into the memory
                 * we allocated (minus 1, to leave one byte for the safety NUL byte) */
                n = size == SIZE_MAX ? MALLOC_SIZEOF_SAFE(buf) - 1 : n_next;

                errno = 0;
                k = fread(buf + l, 1, n - l, f);

                assert(k <= n - l);
                l += k;

                if (ferror(f)) {
                        r = errno_or_else(EIO);
                        goto finalize;
                }
                if (feof(f))
                        break;

                if (size != SIZE_MAX && !FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER)) { /* If we got asked to read some specific size, we already sized the buffer right, hence leave */
                        assert(l == size);
                        break;
                }

                assert(k > 0); /* we can't have read zero bytes because that would have been EOF */

                if (FLAGS_SET(flags, READ_FULL_FILE_FAIL_WHEN_LARGER) && l > size) {
                        r = -E2BIG;
                        goto finalize;
                }

                if (n >= READ_FULL_BYTES_MAX) {
                        r = -E2BIG;
                        goto finalize;
                }

                n_next = MIN(n * 2, READ_FULL_BYTES_MAX);
        }

        if (flags & (READ_FULL_FILE_UNBASE64 | READ_FULL_FILE_UNHEX)) {
                _cleanup_free_ void *decoded = NULL;
                size_t decoded_size;

                buf[l++] = 0;
                if (flags & READ_FULL_FILE_UNBASE64)
                        r = unbase64mem_full(buf, l, flags & READ_FULL_FILE_SECURE, &decoded, &decoded_size);
                else
                        r = unhexmem_full(buf, l, flags & READ_FULL_FILE_SECURE, &decoded, &decoded_size);
                if (r < 0)
                        goto finalize;

                if (flags & READ_FULL_FILE_SECURE)
                        explicit_bzero_safe(buf, n);
                free_and_replace(buf, decoded);
                n = l = decoded_size;
        }

        if (!ret_size) {
                /* Safety check: if the caller doesn't want to know the size of what we just read it will rely on the
                 * trailing NUL byte. But if there's an embedded NUL byte, then we should refuse operation as otherwise
                 * there'd be ambiguity about what we just read. */

                if (memchr(buf, 0, l)) {
                        r = -EBADMSG;
                        goto finalize;
                }
        }

        buf[l] = 0;
        *ret_contents = TAKE_PTR(buf);

        if (ret_size)
                *ret_size = l;

        return 0;

finalize:
        if (flags & READ_FULL_FILE_SECURE)
                explicit_bzero_safe(buf, n);

        return r;
}

int read_full_file_full(
                int dir_fd,
                const char *filename,
                uint64_t offset,
                size_t size,
                ReadFullFileFlags flags,
                const char *bind_name,
                char **ret_contents,
                size_t *ret_size) {

        _cleanup_fclose_ FILE *f = NULL;
        int r;

        assert(filename);
        assert(ret_contents);

        r = xfopenat(dir_fd, filename, "re", 0, &f);
        if (r < 0) {
                _cleanup_close_ int sk = -EBADF;

                /* ENXIO is what Linux returns if we open a node that is an AF_UNIX socket */
                if (r != -ENXIO)
                        return r;

                /* If this is enabled, let's try to connect to it */
                if (!FLAGS_SET(flags, READ_FULL_FILE_CONNECT_SOCKET))
                        return -ENXIO;

                /* Seeking is not supported on AF_UNIX sockets */
                if (offset != UINT64_MAX)
                        return -ENXIO;

                sk = socket(AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
                if (sk < 0)
                        return -errno;

                if (bind_name) {
                        /* If the caller specified a socket name to bind to, do so before connecting. This is
                         * useful to communicate some minor, short meta-information token from the client to
                         * the server. */
                        union sockaddr_union bsa;

                        r = sockaddr_un_set_path(&bsa.un, bind_name);
                        if (r < 0)
                                return r;

                        if (bind(sk, &bsa.sa, r) < 0)
                                return -errno;
                }

                r = connect_unix_path(sk, dir_fd, filename);
                if (IN_SET(r, -ENOTSOCK, -EINVAL)) /* propagate original error if this is not a socket after all */
                        return -ENXIO;
                if (r < 0)
                        return r;

                if (shutdown(sk, SHUT_WR) < 0)
                        return -errno;

                f = fdopen(sk, "r");
                if (!f)
                        return -errno;

                TAKE_FD(sk);
        }

        (void) __fsetlocking(f, FSETLOCKING_BYCALLER);

        return read_full_stream_full(f, filename, offset, size, flags, ret_contents, ret_size);
}

int executable_is_script(const char *path, char **interpreter) {
        _cleanup_free_ char *line = NULL;
        size_t len;
        char *ans;
        int r;

        assert(path);

        r = read_one_line_file(path, &line);
        if (r == -ENOBUFS) /* First line overly long? if so, then it's not a script */
                return 0;
        if (r < 0)
                return r;

        if (!startswith(line, "#!"))
                return 0;

        ans = strstrip(line + 2);
        len = strcspn(ans, " \t");

        if (len == 0)
                return 0;

        ans = strndup(ans, len);
        if (!ans)
                return -ENOMEM;

        *interpreter = ans;
        return 1;
}

/**
 * Retrieve one field from a file like /proc/self/status.  pattern
 * should not include whitespace or the delimiter (':'). pattern matches only
 * the beginning of a line. Whitespace before ':' is skipped. Whitespace and
 * zeros after the ':' will be skipped. field must be freed afterwards.
 * terminator specifies the terminating characters of the field value (not
 * included in the value).
 */
int get_proc_field(const char *filename, const char *pattern, const char *terminator, char **field) {
        _cleanup_free_ char *status = NULL;
        char *t, *f;
        int r;

        assert(terminator);
        assert(filename);
        assert(pattern);
        assert(field);

        r = read_full_virtual_file(filename, &status, NULL);
        if (r < 0)
                return r;

        t = status;

        do {
                bool pattern_ok;

                do {
                        t = strstr(t, pattern);
                        if (!t)
                                return -ENOENT;

                        /* Check that pattern occurs in beginning of line. */
                        pattern_ok = (t == status || t[-1] == '\n');

                        t += strlen(pattern);

                } while (!pattern_ok);

                t += strspn(t, " \t");
                if (!*t)
                        return -ENOENT;

        } while (*t != ':');

        t++;

        if (*t) {
                t += strspn(t, " \t");

                /* Also skip zeros, because when this is used for
                 * capabilities, we don't want the zeros. This way the
                 * same capability set always maps to the same string,
                 * irrespective of the total capability set size. For
                 * other numbers it shouldn't matter. */
                t += strspn(t, "0");
                /* Back off one char if there's nothing but whitespace
                   and zeros */
                if (!*t || isspace(*t))
                        t--;
        }

        f = strdupcspn(t, terminator);
        if (!f)
                return -ENOMEM;

        *field = f;
        return 0;
}

DIR *xopendirat(int fd, const char *name, int flags) {
        int nfd;
        DIR *d;

        assert(!(flags & O_CREAT));

        if (fd == AT_FDCWD && flags == 0)
                return opendir(name);

        nfd = openat(fd, name, O_RDONLY|O_NONBLOCK|O_DIRECTORY|O_CLOEXEC|flags, 0);
        if (nfd < 0)
                return NULL;

        d = fdopendir(nfd);
        if (!d) {
                safe_close(nfd);
                return NULL;
        }

        return d;
}

int fopen_mode_to_flags(const char *mode) {
        const char *p;
        int flags;

        assert(mode);

        if ((p = startswith(mode, "r+")))
                flags = O_RDWR;
        else if ((p = startswith(mode, "r")))
                flags = O_RDONLY;
        else if ((p = startswith(mode, "w+")))
                flags = O_RDWR|O_CREAT|O_TRUNC;
        else if ((p = startswith(mode, "w")))
                flags = O_WRONLY|O_CREAT|O_TRUNC;
        else if ((p = startswith(mode, "a+")))
                flags = O_RDWR|O_CREAT|O_APPEND;
        else if ((p = startswith(mode, "a")))
                flags = O_WRONLY|O_CREAT|O_APPEND;
        else
                return -EINVAL;

        for (; *p != 0; p++) {

                switch (*p) {

                case 'e':
                        flags |= O_CLOEXEC;
                        break;

                case 'x':
                        flags |= O_EXCL;
                        break;

                case 'm':
                        /* ignore this here, fdopen() might care later though */
                        break;

                case 'c': /* not sure what to do about this one */
                default:
                        return -EINVAL;
                }
        }

        return flags;
}

int xfopenat(int dir_fd, const char *path, const char *mode, int flags, FILE **ret) {
        FILE *f;

        /* A combination of fopen() with openat() */

        if (dir_fd == AT_FDCWD && flags == 0) {
                f = fopen(path, mode);
                if (!f)
                        return -errno;
        } else {
                int fd, mode_flags;

                mode_flags = fopen_mode_to_flags(mode);
                if (mode_flags < 0)
                        return mode_flags;

                fd = openat(dir_fd, path, mode_flags | flags);
                if (fd < 0)
                        return -errno;

                f = fdopen(fd, mode);
                if (!f) {
                        safe_close(fd);
                        return -errno;
                }
        }

        *ret = f;
        return 0;
}

static int search_and_fopen_internal(
                const char *path,
                const char *mode,
                const char *root,
                char **search,
                FILE **ret,
                char **ret_path) {

        assert(path);
        assert(mode);
        assert(ret);

        if (!path_strv_resolve_uniq(search, root))
                return -ENOMEM;

        STRV_FOREACH(i, search) {
                _cleanup_free_ char *p = NULL;
                FILE *f;

                p = path_join(root, *i, path);
                if (!p)
                        return -ENOMEM;

                f = fopen(p, mode);
                if (f) {
                        if (ret_path)
                                *ret_path = path_simplify(TAKE_PTR(p));

                        *ret = f;
                        return 0;
                }

                if (errno != ENOENT)
                        return -errno;
        }

        return -ENOENT;
}

int search_and_fopen(
                const char *filename,
                const char *mode,
                const char *root,
                const char **search,
                FILE **ret,
                char **ret_path) {

        _cleanup_strv_free_ char **copy = NULL;

        assert(filename);
        assert(mode);
        assert(ret);

        if (path_is_absolute(filename)) {
                _cleanup_fclose_ FILE *f = NULL;

                f = fopen(filename, mode);
                if (!f)
                        return -errno;

                if (ret_path) {
                        char *p;

                        p = strdup(filename);
                        if (!p)
                                return -ENOMEM;

                        *ret_path = path_simplify(p);
                }

                *ret = TAKE_PTR(f);
                return 0;
        }

        copy = strv_copy((char**) search);
        if (!copy)
                return -ENOMEM;

        return search_and_fopen_internal(filename, mode, root, copy, ret, ret_path);
}

int search_and_fopen_nulstr(
                const char *filename,
                const char *mode,
                const char *root,
                const char *search,
                FILE **ret,
                char **ret_path) {

        _cleanup_strv_free_ char **s = NULL;

        if (path_is_absolute(filename)) {
                _cleanup_fclose_ FILE *f = NULL;

                f = fopen(filename, mode);
                if (!f)
                        return -errno;

                if (ret_path) {
                        char *p;

                        p = strdup(filename);
                        if (!p)
                                return -ENOMEM;

                        *ret_path = path_simplify(p);
                }

                *ret = TAKE_PTR(f);
                return 0;
        }

        s = strv_split_nulstr(search);
        if (!s)
                return -ENOMEM;

        return search_and_fopen_internal(filename, mode, root, s, ret, ret_path);
}

int fflush_and_check(FILE *f) {
        assert(f);

        errno = 0;
        fflush(f);

        if (ferror(f))
                return errno_or_else(EIO);

        return 0;
}

int fflush_sync_and_check(FILE *f) {
        int r, fd;

        assert(f);

        r = fflush_and_check(f);
        if (r < 0)
                return r;

        /* Not all file streams have an fd associated (think: fmemopen()), let's handle this gracefully and
         * assume that in that case we need no explicit syncing */
        fd = fileno(f);
        if (fd < 0)
                return 0;

        r = fsync_full(fd);
        if (r < 0)
                return r;

        return 0;
}

int write_timestamp_file_atomic(const char *fn, usec_t n) {
        char ln[DECIMAL_STR_MAX(n)+2];

        /* Creates a "timestamp" file, that contains nothing but a
         * usec_t timestamp, formatted in ASCII. */

        if (!timestamp_is_set(n))
                return -ERANGE;

        xsprintf(ln, USEC_FMT "\n", n);

        return write_string_file(fn, ln, WRITE_STRING_FILE_CREATE|WRITE_STRING_FILE_ATOMIC);
}

int read_timestamp_file(const char *fn, usec_t *ret) {
        _cleanup_free_ char *ln = NULL;
        uint64_t t;
        int r;

        r = read_one_line_file(fn, &ln);
        if (r < 0)
                return r;

        r = safe_atou64(ln, &t);
        if (r < 0)
                return r;

        if (!timestamp_is_set(t))
                return -ERANGE;

        *ret = (usec_t) t;
        return 0;
}

int fputs_with_space(FILE *f, const char *s, const char *separator, bool *space) {
        int r;

        assert(s);

        /* Outputs the specified string with fputs(), but optionally prefixes it with a separator. The *space parameter
         * when specified shall initially point to a boolean variable initialized to false. It is set to true after the
         * first invocation. This call is supposed to be use in loops, where a separator shall be inserted between each
         * element, but not before the first one. */

        if (!f)
                f = stdout;

        if (space) {
                if (!separator)
                        separator = " ";

                if (*space) {
                        r = fputs(separator, f);
                        if (r < 0)
                                return r;
                }

                *space = true;
        }

        return fputs(s, f);
}

/* A bitmask of the EOL markers we know */
typedef enum EndOfLineMarker {
        EOL_NONE     = 0,
        EOL_ZERO     = 1 << 0,  /* \0 (aka NUL) */
        EOL_TEN      = 1 << 1,  /* \n (aka NL, aka LF)  */
        EOL_THIRTEEN = 1 << 2,  /* \r (aka CR)  */
} EndOfLineMarker;

static EndOfLineMarker categorize_eol(char c, ReadLineFlags flags) {

        if (!FLAGS_SET(flags, READ_LINE_ONLY_NUL)) {
                if (c == '\n')
                        return EOL_TEN;
                if (c == '\r')
                        return EOL_THIRTEEN;
        }

        if (c == '\0')
                return EOL_ZERO;

        return EOL_NONE;
}

DEFINE_TRIVIAL_CLEANUP_FUNC_FULL(FILE*, funlockfile, NULL);

int read_line_full(FILE *f, size_t limit, ReadLineFlags flags, char **ret) {
        _cleanup_free_ char *buffer = NULL;
        size_t n = 0, count = 0;
        int r;

        assert(f);

        /* Something like a bounded version of getline().
         *
         * Considers EOF, \n, \r and \0 end of line delimiters (or combinations of these), and does not include these
         * delimiters in the string returned. Specifically, recognizes the following combinations of markers as line
         * endings:
         *
         *     • \n        (UNIX)
         *     • \r        (old MacOS)
         *     • \0        (C strings)
         *     • \n\0
         *     • \r\0
         *     • \r\n      (Windows)
         *     • \n\r
         *     • \r\n\0
         *     • \n\r\0
         *
         * Returns the number of bytes read from the files (i.e. including delimiters — this hence usually differs from
         * the number of characters in the returned string). When EOF is hit, 0 is returned.
         *
         * The input parameter limit is the maximum numbers of characters in the returned string, i.e. excluding
         * delimiters. If the limit is hit we fail and return -ENOBUFS.
         *
         * If a line shall be skipped ret may be initialized as NULL. */

        if (ret) {
                if (!GREEDY_REALLOC(buffer, 1))
                        return -ENOMEM;
        }

        {
                _unused_ _cleanup_(funlockfilep) FILE *flocked = f;
                EndOfLineMarker previous_eol = EOL_NONE;
                flockfile(f);

                for (;;) {
                        EndOfLineMarker eol;
                        char c;

                        if (n >= limit)
                                return -ENOBUFS;

                        if (count >= INT_MAX) /* We couldn't return the counter anymore as "int", hence refuse this */
                                return -ENOBUFS;

                        r = safe_fgetc(f, &c);
                        if (r < 0)
                                return r;
                        if (r == 0) /* EOF is definitely EOL */
                                break;

                        eol = categorize_eol(c, flags);

                        if (FLAGS_SET(previous_eol, EOL_ZERO) ||
                            (eol == EOL_NONE && previous_eol != EOL_NONE) ||
                            (eol != EOL_NONE && (previous_eol & eol) != 0)) {
                                /* Previous char was a NUL? This is not an EOL, but the previous char was? This type of
                                 * EOL marker has been seen right before?  In either of these three cases we are
                                 * done. But first, let's put this character back in the queue. (Note that we have to
                                 * cast this to (unsigned char) here as ungetc() expects a positive 'int', and if we
                                 * are on an architecture where 'char' equals 'signed char' we need to ensure we don't
                                 * pass a negative value here. That said, to complicate things further ungetc() is
                                 * actually happy with most negative characters and implicitly casts them back to
                                 * positive ones as needed, except for \xff (aka -1, aka EOF), which it refuses. What a
                                 * godawful API!) */
                                assert_se(ungetc((unsigned char) c, f) != EOF);
                                break;
                        }

                        count++;

                        if (eol != EOL_NONE) {
                                /* If we are on a tty, we can't shouldn't wait for more input, because that
                                 * generally means waiting for the user, interactively. In the case of a TTY
                                 * we expect only \n as the single EOL marker, so we are in the lucky
                                 * position that there is no need to wait. We check this condition last, to
                                 * avoid isatty() check if not necessary. */

                                if ((flags & (READ_LINE_IS_A_TTY|READ_LINE_NOT_A_TTY)) == 0) {
                                        int fd;

                                        fd = fileno(f);
                                        if (fd < 0) /* Maybe an fmemopen() stream? Handle this gracefully,
                                                     * and don't call isatty() on an invalid fd */
                                                flags |= READ_LINE_NOT_A_TTY;
                                        else
                                                flags |= isatty(fd) ? READ_LINE_IS_A_TTY : READ_LINE_NOT_A_TTY;
                                }
                                if (FLAGS_SET(flags, READ_LINE_IS_A_TTY))
                                        break;
                        }

                        if (eol != EOL_NONE) {
                                previous_eol |= eol;
                                continue;
                        }

                        if (ret) {
                                if (!GREEDY_REALLOC(buffer, n + 2))
                                        return -ENOMEM;

                                buffer[n] = c;
                        }

                        n++;
                }
        }

        if (ret) {
                buffer[n] = 0;

                *ret = TAKE_PTR(buffer);
        }

        return (int) count;
}

int safe_fgetc(FILE *f, char *ret) {
        int k;

        assert(f);

        /* A safer version of plain fgetc(): let's propagate the error that happened while reading as such, and
         * separate the EOF condition from the byte read, to avoid those confusion signed/unsigned issues fgetc()
         * has. */

        errno = 0;
        k = fgetc(f);
        if (k == EOF) {
                if (ferror(f))
                        return errno_or_else(EIO);

                if (ret)
                        *ret = 0;

                return 0;
        }

        if (ret)
                *ret = k;

        return 1;
}

int warn_file_is_world_accessible(const char *filename, struct stat *st, const char *unit, unsigned line) {
        struct stat _st;

        if (!filename)
                return 0;

        if (!st) {
                if (stat(filename, &_st) < 0)
                        return -errno;
                st = &_st;
        }

        if ((st->st_mode & S_IRWXO) == 0)
                return 0;

        if (unit)
                log_syntax(unit, LOG_WARNING, filename, line, 0,
                           "%s has %04o mode that is too permissive, please adjust the ownership and access mode.",
                           filename, st->st_mode & 07777);
        else
                log_warning("%s has %04o mode that is too permissive, please adjust the ownership and access mode.",
                            filename, st->st_mode & 07777);
        return 0;
}