Repository navigation
Expand file tree
/
Copy path_posixsubprocess.c
More file actions
1322 lines (1183 loc) · 43.1 KB
/
Copy path_posixsubprocess.c
File metadata and controls
1322 lines (1183 loc) · 43.1 KB
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
/* Authors: Gregory P. Smith & Jeffrey Yasskin */
#ifndef Py_BUILD_CORE_BUILTIN
# define Py_BUILD_CORE_MODULE 1
#endif
#include "Python.h"
#include "pycore_fileutils.h"
#include "pycore_pystate.h"
#if defined(HAVE_PIPE2) && !defined(_GNU_SOURCE)
# define _GNU_SOURCE
#endif
#include <unistd.h>
#include <fcntl.h>
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#if defined(HAVE_SYS_STAT_H)
#include <sys/stat.h>
#endif
#ifdef HAVE_SYS_SYSCALL_H
#include <sys/syscall.h>
#endif
#if defined(HAVE_SYS_RESOURCE_H)
#include <sys/resource.h>
#endif
#ifdef HAVE_DIRENT_H
#include <dirent.h>
#endif
#ifdef HAVE_GRP_H
#include <grp.h>
#endif /* HAVE_GRP_H */
#include "posixmodule.h"
#ifdef _Py_MEMORY_SANITIZER
# include <sanitizer/msan_interface.h>
#endif
#if defined(__ANDROID__) && __ANDROID_API__ < 21 && !defined(SYS_getdents64)
# include <sys/linux-syscalls.h>
# define SYS_getdents64 __NR_getdents64
#endif
#if defined(__linux__) && defined(HAVE_VFORK) && defined(HAVE_SIGNAL_H) && \
defined(HAVE_PTHREAD_SIGMASK) && !defined(HAVE_BROKEN_PTHREAD_SIGMASK)
/* If this is ever expanded to non-Linux platforms, verify what calls are
* allowed after vfork(). Ex: setsid() may be disallowed on macOS? */
# include <signal.h>
# define VFORK_USABLE 1
#endif
#if defined(__sun) && defined(__SVR4)
/* readdir64 is used to work around Solaris 9 bug 6395699. */
# define readdir readdir64
# define dirent dirent64
# if !defined(HAVE_DIRFD)
/* Some versions of Solaris lack dirfd(). */
# define dirfd(dirp) ((dirp)->dd_fd)
# define HAVE_DIRFD
# endif
#endif
#if defined(__FreeBSD__) || (defined(__APPLE__) && defined(__MACH__)) || defined(__DragonFly__)
# define FD_DIR "/dev/fd"
#else
# define FD_DIR "/proc/self/fd"
#endif
#ifdef NGROUPS_MAX
#define MAX_GROUPS NGROUPS_MAX
#else
#define MAX_GROUPS 64
#endif
#define POSIX_CALL(call) do { if ((call) == -1) goto error; } while (0)
static struct PyModuleDef _posixsubprocessmodule;
/*[clinic input]
module _posixsubprocess
[clinic start generated code]*/
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=c62211df27cf7334]*/
/*[python input]
class pid_t_converter(CConverter):
type = 'pid_t'
format_unit = '" _Py_PARSE_PID "'
def parse_arg(self, argname, displayname):
return """
{paramname} = PyLong_AsPid({argname});
if ({paramname} == -1 && PyErr_Occurred()) {{{{
goto exit;
}}}}
""".format(argname=argname, paramname=self.parser_name)
[python start generated code]*/
/*[python end generated code: output=da39a3ee5e6b4b0d input=5af1c116d56cbb5a]*/
#include "clinic/_posixsubprocess.c.h"
/* Convert ASCII to a positive int, no libc call. no overflow. -1 on error. */
static int
_pos_int_from_ascii(const char *name)
{
int num = 0;
while (*name >= '0' && *name <= '9') {
num = num * 10 + (*name - '0');
++name;
}
if (*name)
return -1; /* Non digit found, not a number. */
return num;
}
#if defined(__FreeBSD__) || defined(__DragonFly__)
/* When /dev/fd isn't mounted it is often a static directory populated
* with 0 1 2 or entries for 0 .. 63 on FreeBSD, NetBSD, OpenBSD and DragonFlyBSD.
* NetBSD and OpenBSD have a /proc fs available (though not necessarily
* mounted) and do not have fdescfs for /dev/fd. MacOS X has a devfs
* that properly supports /dev/fd.
*/
static int
_is_fdescfs_mounted_on_dev_fd(void)
{
struct stat dev_stat;
struct stat dev_fd_stat;
if (stat("/dev", &dev_stat) != 0)
return 0;
if (stat(FD_DIR, &dev_fd_stat) != 0)
return 0;
if (dev_stat.st_dev == dev_fd_stat.st_dev)
return 0; /* / == /dev == /dev/fd means it is static. #fail */
return 1;
}
#endif
/* Returns 1 if there is a problem with fd_sequence, 0 otherwise. */
static int
_sanity_check_python_fd_sequence(PyObject *fd_sequence)
{
Py_ssize_t seq_idx;
long prev_fd = -1;
for (seq_idx = 0; seq_idx < PyTuple_GET_SIZE(fd_sequence); ++seq_idx) {
PyObject* py_fd = PyTuple_GET_ITEM(fd_sequence, seq_idx);
long iter_fd;
if (!PyLong_Check(py_fd)) {
return 1;
}
iter_fd = PyLong_AsLong(py_fd);
if (iter_fd < 0 || iter_fd <= prev_fd || iter_fd > INT_MAX) {
/* Negative, overflow, unsorted, too big for a fd. */
return 1;
}
prev_fd = iter_fd;
}
return 0;
}
/* Is fd found in the sorted Python Sequence? */
static int
_is_fd_in_sorted_fd_sequence(int fd, int *fd_sequence,
Py_ssize_t fd_sequence_len)
{
/* Binary search. */
Py_ssize_t search_min = 0;
Py_ssize_t search_max = fd_sequence_len - 1;
if (search_max < 0)
return 0;
do {
long middle = (search_min + search_max) / 2;
long middle_fd = fd_sequence[middle];
if (fd == middle_fd)
return 1;
if (fd > middle_fd)
search_min = middle + 1;
else
search_max = middle - 1;
} while (search_min <= search_max);
return 0;
}
// Forward declaration
static void _Py_FreeCharPArray(char *const array[]);
/*
* Flatten a sequence of bytes() objects into a C array of
* NULL terminated string pointers with a NULL char* terminating the array.
* (ie: an argv or env list)
*
* Memory allocated for the returned list is allocated using PyMem_Malloc()
* and MUST be freed by _Py_FreeCharPArray().
*/
static char *const *
_PySequence_BytesToCharpArray(PyObject* self)
{
char **array;
Py_ssize_t i, argc;
PyObject *item = NULL;
Py_ssize_t size;
argc = PySequence_Size(self);
if (argc == -1)
return NULL;
assert(argc >= 0);
if ((size_t)argc > (PY_SSIZE_T_MAX-sizeof(char *)) / sizeof(char *)) {
PyErr_NoMemory();
return NULL;
}
array = PyMem_Malloc((argc + 1) * sizeof(char *));
if (array == NULL) {
PyErr_NoMemory();
return NULL;
}
for (i = 0; i < argc; ++i) {
char *data;
item = PySequence_GetItem(self, i);
if (item == NULL) {
/* NULL terminate before freeing. */
array[i] = NULL;
goto fail;
}
/* check for embedded null bytes */
if (PyBytes_AsStringAndSize(item, &data, NULL) < 0) {
/* NULL terminate before freeing. */
array[i] = NULL;
goto fail;
}
size = PyBytes_GET_SIZE(item) + 1;
array[i] = PyMem_Malloc(size);
if (!array[i]) {
PyErr_NoMemory();
goto fail;
}
memcpy(array[i], data, size);
Py_DECREF(item);
}
array[argc] = NULL;
return array;
fail:
Py_XDECREF(item);
_Py_FreeCharPArray(array);
return NULL;
}
/* Free's a NULL terminated char** array of C strings. */
static void
_Py_FreeCharPArray(char *const array[])
{
Py_ssize_t i;
for (i = 0; array[i] != NULL; ++i) {
PyMem_Free(array[i]);
}
PyMem_Free((void*)array);
}
/*
* Do all the Python C API calls in the parent process to turn the pass_fds
* "py_fds_to_keep" tuple into a C array. The caller owns allocation and
* freeing of the array.
*
* On error an unknown number of array elements may have been filled in.
* A Python exception has been set when an error is returned.
*
* Returns: -1 on error, 0 on success.
*/
static int
convert_fds_to_keep_to_c(PyObject *py_fds_to_keep, int *c_fds_to_keep)
{
Py_ssize_t i, len;
len = PyTuple_GET_SIZE(py_fds_to_keep);
for (i = 0; i < len; ++i) {
PyObject* fdobj = PyTuple_GET_ITEM(py_fds_to_keep, i);
long fd = PyLong_AsLong(fdobj);
if (fd == -1 && PyErr_Occurred()) {
return -1;
}
if (fd < 0 || fd > INT_MAX) {
PyErr_SetString(PyExc_ValueError,
"fd out of range in fds_to_keep.");
return -1;
}
c_fds_to_keep[i] = (int)fd;
}
return 0;
}
/* This function must be async-signal-safe as it is called from child_exec()
* after fork() or vfork().
*/
static int
make_inheritable(int *c_fds_to_keep, Py_ssize_t len, int errpipe_write)
{
Py_ssize_t i;
for (i = 0; i < len; ++i) {
int fd = c_fds_to_keep[i];
if (fd == errpipe_write) {
/* errpipe_write is part of fds_to_keep. It must be closed at
exec(), but kept open in the child process until exec() is
called. */
continue;
}
if (_Py_set_inheritable_async_safe(fd, 1, NULL) < 0)
return -1;
}
return 0;
}
/* Get the maximum file descriptor that could be opened by this process.
* This function is async signal safe for use between fork() and exec().
*/
static long
safe_get_max_fd(void)
{
long local_max_fd;
#if defined(__NetBSD__)
local_max_fd = fcntl(0, F_MAXFD);
if (local_max_fd >= 0)
return local_max_fd;
#endif
#if defined(HAVE_SYS_RESOURCE_H) && defined(__OpenBSD__)
struct rlimit rl;
/* Not on the POSIX async signal safe functions list but likely
* safe. TODO - Someone should audit OpenBSD to make sure. */
if (getrlimit(RLIMIT_NOFILE, &rl) >= 0)
return (long) rl.rlim_max;
#endif
#ifdef _SC_OPEN_MAX
local_max_fd = sysconf(_SC_OPEN_MAX);
if (local_max_fd == -1)
#endif
local_max_fd = 256; /* Matches legacy Lib/subprocess.py behavior. */
return local_max_fd;
}
/* Close all file descriptors in the given range except for those in
* fds_to_keep by invoking closer on each subrange.
*
* If end_fd == -1, it's guessed via safe_get_max_fd(), but it isn't
* possible to know for sure what the max fd to go up to is for
* processes with the capability of raising their maximum, or in case
* a process opened a high fd and then lowered its maximum.
*/
static int
_close_range_except(int start_fd,
int end_fd,
int *fds_to_keep,
Py_ssize_t fds_to_keep_len,
int (*closer)(int, int))
{
if (end_fd == -1) {
end_fd = Py_MIN(safe_get_max_fd(), INT_MAX);
}
Py_ssize_t keep_seq_idx;
/* As fds_to_keep is sorted we can loop through the list closing
* fds in between any in the keep list falling within our range. */
for (keep_seq_idx = 0; keep_seq_idx < fds_to_keep_len; ++keep_seq_idx) {
int keep_fd = fds_to_keep[keep_seq_idx];
if (keep_fd < start_fd)
continue;
if (closer(start_fd, keep_fd - 1) != 0)
return -1;
start_fd = keep_fd + 1;
}
if (start_fd <= end_fd) {
if (closer(start_fd, end_fd) != 0)
return -1;
}
return 0;
}
#if defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)
/* It doesn't matter if d_name has room for NAME_MAX chars; we're using this
* only to read a directory of short file descriptor number names. The kernel
* will return an error if we didn't give it enough space. Highly Unlikely.
* This structure is very old and stable: It will not change unless the kernel
* chooses to break compatibility with all existing binaries. Highly Unlikely.
*/
struct linux_dirent64 {
unsigned long long d_ino;
long long d_off;
unsigned short d_reclen; /* Length of this linux_dirent */
unsigned char d_type;
char d_name[256]; /* Filename (null-terminated) */
};
static int
_brute_force_closer(int first, int last)
{
for (int i = first; i <= last; i++) {
/* Ignore errors */
(void)close(i);
}
return 0;
}
/* Close all open file descriptors in the range from start_fd and higher
* Do not close any in the sorted fds_to_keep list.
*
* This version is async signal safe as it does not make any unsafe C library
* calls, malloc calls or handle any locks. It is _unfortunate_ to be forced
* to resort to making a kernel system call directly but this is the ONLY api
* available that does no harm. opendir/readdir/closedir perform memory
* allocation and locking so while they usually work they are not guaranteed
* to (especially if you have replaced your malloc implementation). A version
* of this function that uses those can be found in the _maybe_unsafe variant.
*
* This is Linux specific because that is all I am ready to test it on. It
* should be easy to add OS specific dirent or dirent64 structures and modify
* it with some cpp #define magic to work on other OSes as well if you want.
*/
static void
_close_open_fds_safe(int start_fd, int *fds_to_keep, Py_ssize_t fds_to_keep_len)
{
int fd_dir_fd;
fd_dir_fd = _Py_open_noraise(FD_DIR, O_RDONLY);
if (fd_dir_fd == -1) {
/* No way to get a list of open fds. */
_close_range_except(start_fd, -1,
fds_to_keep, fds_to_keep_len,
_brute_force_closer);
return;
} else {
char buffer[sizeof(struct linux_dirent64)];
int bytes;
while ((bytes = syscall(SYS_getdents64, fd_dir_fd,
(struct linux_dirent64 *)buffer,
sizeof(buffer))) > 0) {
struct linux_dirent64 *entry;
int offset;
#ifdef _Py_MEMORY_SANITIZER
__msan_unpoison(buffer, bytes);
#endif
for (offset = 0; offset < bytes; offset += entry->d_reclen) {
int fd;
entry = (struct linux_dirent64 *)(buffer + offset);
if ((fd = _pos_int_from_ascii(entry->d_name)) < 0)
continue; /* Not a number. */
if (fd != fd_dir_fd && fd >= start_fd &&
!_is_fd_in_sorted_fd_sequence(fd, fds_to_keep,
fds_to_keep_len)) {
close(fd);
}
}
}
close(fd_dir_fd);
}
}
#define _close_open_fds_fallback _close_open_fds_safe
#else /* NOT (defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)) */
static int
_unsafe_closer(int first, int last)
{
_Py_closerange(first, last);
return 0;
}
/* Close all open file descriptors from start_fd and higher.
* Do not close any in the sorted fds_to_keep tuple.
*
* This function violates the strict use of async signal safe functions. :(
* It calls opendir(), readdir() and closedir(). Of these, the one most
* likely to ever cause a problem is opendir() as it performs an internal
* malloc(). Practically this should not be a problem. The Java VM makes the
* same calls between fork and exec in its own UNIXProcess_md.c implementation.
*
* readdir_r() is not used because it provides no benefit. It is typically
* implemented as readdir() followed by memcpy(). See also:
* http://womble.decadent.org.uk/readdir_r-advisory.html
*/
static void
_close_open_fds_maybe_unsafe(int start_fd, int *fds_to_keep,
Py_ssize_t fds_to_keep_len)
{
DIR *proc_fd_dir;
#ifndef HAVE_DIRFD
while (_is_fd_in_sorted_fd_sequence(start_fd, fds_to_keep,
fds_to_keep_len)) {
++start_fd;
}
/* Close our lowest fd before we call opendir so that it is likely to
* reuse that fd otherwise we might close opendir's file descriptor in
* our loop. This trick assumes that fd's are allocated on a lowest
* available basis. */
close(start_fd);
++start_fd;
#endif
#if defined(__FreeBSD__) || defined(__DragonFly__)
if (!_is_fdescfs_mounted_on_dev_fd())
proc_fd_dir = NULL;
else
#endif
proc_fd_dir = opendir(FD_DIR);
if (!proc_fd_dir) {
/* No way to get a list of open fds. */
_close_range_except(start_fd, -1, fds_to_keep, fds_to_keep_len,
_unsafe_closer);
} else {
struct dirent *dir_entry;
#ifdef HAVE_DIRFD
int fd_used_by_opendir = dirfd(proc_fd_dir);
#else
int fd_used_by_opendir = start_fd - 1;
#endif
errno = 0;
while ((dir_entry = readdir(proc_fd_dir))) {
int fd;
if ((fd = _pos_int_from_ascii(dir_entry->d_name)) < 0)
continue; /* Not a number. */
if (fd != fd_used_by_opendir && fd >= start_fd &&
!_is_fd_in_sorted_fd_sequence(fd, fds_to_keep,
fds_to_keep_len)) {
close(fd);
}
errno = 0;
}
if (errno) {
/* readdir error, revert behavior. Highly Unlikely. */
_close_range_except(start_fd, -1, fds_to_keep, fds_to_keep_len,
_unsafe_closer);
}
closedir(proc_fd_dir);
}
}
#define _close_open_fds_fallback _close_open_fds_maybe_unsafe
#endif /* else NOT (defined(__linux__) && defined(HAVE_SYS_SYSCALL_H)) */
/* We can use close_range() library function only if it's known to be
* async-signal-safe.
*
* On Linux, glibc explicitly documents it to be a thin wrapper over
* the system call, and other C libraries are likely to follow glibc.
*/
#if defined(HAVE_CLOSE_RANGE) && \
(defined(__linux__) || defined(__FreeBSD__))
#define HAVE_ASYNC_SAFE_CLOSE_RANGE
static int
_close_range_closer(int first, int last)
{
return close_range(first, last, 0);
}
#endif
static void
_close_open_fds(int start_fd, int *fds_to_keep, Py_ssize_t fds_to_keep_len)
{
#ifdef HAVE_ASYNC_SAFE_CLOSE_RANGE
if (_close_range_except(
start_fd, INT_MAX, fds_to_keep, fds_to_keep_len,
_close_range_closer) == 0) {
return;
}
#endif
_close_open_fds_fallback(start_fd, fds_to_keep, fds_to_keep_len);
}
#ifdef VFORK_USABLE
/* Reset dispositions for all signals to SIG_DFL except for ignored
* signals. This way we ensure that no signal handlers can run
* after we unblock signals in a child created by vfork().
*/
static void
reset_signal_handlers(const sigset_t *child_sigmask)
{
struct sigaction sa_dfl = {.sa_handler = SIG_DFL};
for (int sig = 1; sig < _NSIG; sig++) {
/* Dispositions for SIGKILL and SIGSTOP can't be changed. */
if (sig == SIGKILL || sig == SIGSTOP) {
continue;
}
/* There is no need to reset the disposition of signals that will
* remain blocked across execve() since the kernel will do it. */
if (sigismember(child_sigmask, sig) == 1) {
continue;
}
struct sigaction sa;
/* C libraries usually return EINVAL for signals used
* internally (e.g. for thread cancellation), so simply
* skip errors here. */
if (sigaction(sig, NULL, &sa) == -1) {
continue;
}
/* void *h works as these fields are both pointer types already. */
void *h = (sa.sa_flags & SA_SIGINFO ? (void *)sa.sa_sigaction :
(void *)sa.sa_handler);
if (h == SIG_IGN || h == SIG_DFL) {
continue;
}
/* This call can't reasonably fail, but if it does, terminating
* the child seems to be too harsh, so ignore errors. */
(void) sigaction(sig, &sa_dfl, NULL);
}
}
#endif /* VFORK_USABLE */
/*
* This function is code executed in the child process immediately after
* (v)fork to set things up and call exec().
*
* All of the code in this function must only use async-signal-safe functions,
* listed at `man 7 signal` or
* http://www.opengroup.org/onlinepubs/009695399/functions/xsh_chap02_04.html.
*
* This restriction is documented at
* http://www.opengroup.org/onlinepubs/009695399/functions/fork.html.
*
* If this function is called after vfork(), even more care must be taken.
* The lack of preparations that C libraries normally take on fork(),
* as well as sharing the address space with the parent, might make even
* async-signal-safe functions vfork-unsafe. In particular, on Linux,
* set*id() and setgroups() library functions must not be called, since
* they have to interact with the library-level thread list and send
* library-internal signals to implement per-process credentials semantics
* required by POSIX but not supported natively on Linux. Another reason to
* avoid this family of functions is that sharing an address space between
* processes running with different privileges is inherently insecure.
* See https://bugs.python.org/issue35823 for discussion and references.
*
* In some C libraries, setrlimit() has the same thread list/signalling
* behavior since resource limits were per-thread attributes before
* Linux 2.6.10. Musl, as of 1.2.1, is known to have this issue
* (https://www.openwall.com/lists/musl/2020/10/15/6).
*
* If vfork-unsafe functionality is desired after vfork(), consider using
* syscall() to obtain it.
*/
Py_NO_INLINE static void
child_exec(char *const exec_array[],
char *const argv[],
char *const envp[],
const char *cwd,
int p2cread, int p2cwrite,
int c2pread, int c2pwrite,
int errread, int errwrite,
int errpipe_read, int errpipe_write,
int close_fds, int restore_signals,
int call_setsid, pid_t pgid_to_set,
gid_t gid,
Py_ssize_t extra_group_size, const gid_t *extra_groups,
uid_t uid, int child_umask,
const void *child_sigmask,
int *fds_to_keep, Py_ssize_t fds_to_keep_len,
PyObject *preexec_fn,
PyObject *preexec_fn_args_tuple)
{
int i, saved_errno, reached_preexec = 0;
PyObject *result;
const char* err_msg = "";
/* Buffer large enough to hold a hex integer. We can't malloc. */
char hex_errno[sizeof(saved_errno)*2+1];
if (make_inheritable(fds_to_keep, fds_to_keep_len, errpipe_write) < 0)
goto error;
/* Close parent's pipe ends. */
if (p2cwrite != -1)
POSIX_CALL(close(p2cwrite));
if (c2pread != -1)
POSIX_CALL(close(c2pread));
if (errread != -1)
POSIX_CALL(close(errread));
POSIX_CALL(close(errpipe_read));
/* When duping fds, if there arises a situation where one of the fds is
either 0, 1 or 2, it is possible that it is overwritten (#12607). */
if (c2pwrite == 0) {
POSIX_CALL(c2pwrite = dup(c2pwrite));
/* issue32270 */
if (_Py_set_inheritable_async_safe(c2pwrite, 0, NULL) < 0) {
goto error;
}
}
while (errwrite == 0 || errwrite == 1) {
POSIX_CALL(errwrite = dup(errwrite));
/* issue32270 */
if (_Py_set_inheritable_async_safe(errwrite, 0, NULL) < 0) {
goto error;
}
}
/* Dup fds for child.
dup2() removes the CLOEXEC flag but we must do it ourselves if dup2()
would be a no-op (issue #10806). */
if (p2cread == 0) {
if (_Py_set_inheritable_async_safe(p2cread, 1, NULL) < 0)
goto error;
}
else if (p2cread != -1)
POSIX_CALL(dup2(p2cread, 0)); /* stdin */
if (c2pwrite == 1) {
if (_Py_set_inheritable_async_safe(c2pwrite, 1, NULL) < 0)
goto error;
}
else if (c2pwrite != -1)
POSIX_CALL(dup2(c2pwrite, 1)); /* stdout */
if (errwrite == 2) {
if (_Py_set_inheritable_async_safe(errwrite, 1, NULL) < 0)
goto error;
}
else if (errwrite != -1)
POSIX_CALL(dup2(errwrite, 2)); /* stderr */
/* We no longer manually close p2cread, c2pwrite, and errwrite here as
* _close_open_fds takes care when it is not already non-inheritable. */
if (cwd)
POSIX_CALL(chdir(cwd));
if (child_umask >= 0)
umask(child_umask); /* umask() always succeeds. */
if (restore_signals)
_Py_RestoreSignals();
#ifdef VFORK_USABLE
if (child_sigmask) {
reset_signal_handlers(child_sigmask);
if ((errno = pthread_sigmask(SIG_SETMASK, child_sigmask, NULL))) {
goto error;
}
}
#endif
#ifdef HAVE_SETSID
if (call_setsid)
POSIX_CALL(setsid());
#endif
#ifdef HAVE_SETPGID
static_assert(_Py_IS_TYPE_SIGNED(pid_t), "pid_t is unsigned");
if (pgid_to_set >= 0) {
POSIX_CALL(setpgid(0, pgid_to_set));
}
#endif
#ifdef HAVE_SETGROUPS
if (extra_group_size > 0)
POSIX_CALL(setgroups(extra_group_size, extra_groups));
#endif /* HAVE_SETGROUPS */
#ifdef HAVE_SETREGID
if (gid != (gid_t)-1)
POSIX_CALL(setregid(gid, gid));
#endif /* HAVE_SETREGID */
#ifdef HAVE_SETREUID
if (uid != (uid_t)-1)
POSIX_CALL(setreuid(uid, uid));
#endif /* HAVE_SETREUID */
reached_preexec = 1;
if (preexec_fn != Py_None && preexec_fn_args_tuple) {
/* This is where the user has asked us to deadlock their program. */
result = PyObject_Call(preexec_fn, preexec_fn_args_tuple, NULL);
if (result == NULL) {
/* Stringifying the exception or traceback would involve
* memory allocation and thus potential for deadlock.
* We've already faced potential deadlock by calling back
* into Python in the first place, so it probably doesn't
* matter but we avoid it to minimize the possibility. */
err_msg = "Exception occurred in preexec_fn.";
errno = 0; /* We don't want to report an OSError. */
goto error;
}
/* Py_DECREF(result); - We're about to exec so why bother? */
}
/* close FDs after executing preexec_fn, which might open FDs */
if (close_fds) {
/* TODO HP-UX could use pstat_getproc() if anyone cares about it. */
_close_open_fds(3, fds_to_keep, fds_to_keep_len);
}
/* This loop matches the Lib/os.py _execvpe()'s PATH search when */
/* given the executable_list generated by Lib/subprocess.py. */
saved_errno = 0;
for (i = 0; exec_array[i] != NULL; ++i) {
const char *executable = exec_array[i];
if (envp) {
execve(executable, argv, envp);
} else {
execv(executable, argv);
}
if (errno != ENOENT && errno != ENOTDIR && saved_errno == 0) {
saved_errno = errno;
}
}
/* Report the first exec error, not the last. */
if (saved_errno)
errno = saved_errno;
error:
saved_errno = errno;
/* Report the posix error to our parent process. */
/* We ignore all write() return values as the total size of our writes is
less than PIPEBUF and we cannot do anything about an error anyways.
Use _Py_write_noraise() to retry write() if it is interrupted by a
signal (fails with EINTR). */
if (saved_errno) {
char *cur;
_Py_write_noraise(errpipe_write, "OSError:", 8);
cur = hex_errno + sizeof(hex_errno);
while (saved_errno != 0 && cur != hex_errno) {
*--cur = Py_hexdigits[saved_errno % 16];
saved_errno /= 16;
}
_Py_write_noraise(errpipe_write, cur, hex_errno + sizeof(hex_errno) - cur);
_Py_write_noraise(errpipe_write, ":", 1);
if (!reached_preexec) {
/* Indicate to the parent that the error happened before exec(). */
_Py_write_noraise(errpipe_write, "noexec", 6);
}
/* We can't call strerror(saved_errno). It is not async signal safe.
* The parent process will look the error message up. */
} else {
_Py_write_noraise(errpipe_write, "SubprocessError:0:", 18);
_Py_write_noraise(errpipe_write, err_msg, strlen(err_msg));
}
}
/* The main purpose of this wrapper function is to isolate vfork() from both
* subprocess_fork_exec() and child_exec(). A child process created via
* vfork() executes on the same stack as the parent process while the latter is
* suspended, so this function should not be inlined to avoid compiler bugs
* that might clobber data needed by the parent later. Additionally,
* child_exec() should not be inlined to avoid spurious -Wclobber warnings from
* GCC (see bpo-35823).
*/
Py_NO_INLINE static pid_t
do_fork_exec(char *const exec_array[],
char *const argv[],
char *const envp[],
const char *cwd,
int p2cread, int p2cwrite,
int c2pread, int c2pwrite,
int errread, int errwrite,
int errpipe_read, int errpipe_write,
int close_fds, int restore_signals,
int call_setsid, pid_t pgid_to_set,
gid_t gid,
Py_ssize_t extra_group_size, const gid_t *extra_groups,
uid_t uid, int child_umask,
const void *child_sigmask,
int *fds_to_keep, Py_ssize_t fds_to_keep_len,
PyObject *preexec_fn,
PyObject *preexec_fn_args_tuple)
{
pid_t pid;
#ifdef VFORK_USABLE
PyThreadState *vfork_tstate_save;
if (child_sigmask) {
/* These are checked by our caller; verify them in debug builds. */
assert(uid == (uid_t)-1);
assert(gid == (gid_t)-1);
assert(extra_group_size < 0);
assert(preexec_fn == Py_None);
/* Drop the GIL so that other threads can continue execution while this
* thread in the parent remains blocked per vfork-semantics on the
* child's exec syscall outcome. Exec does filesystem access which
* can take an arbitrarily long time. This addresses GH-104372.
*
* The vfork'ed child still runs in our address space. Per POSIX it
* must be limited to nothing but exec, but the Linux implementation
* is a little more usable. See the child_exec() comment - The child
* MUST NOT re-acquire the GIL.
*/
vfork_tstate_save = PyEval_SaveThread();
pid = vfork();
if (pid != 0) {
// Not in the child process, reacquire the GIL.
PyEval_RestoreThread(vfork_tstate_save);
}
if (pid == (pid_t)-1) {
/* If vfork() fails, fall back to using fork(). When it isn't
* allowed in a process by the kernel, vfork can return -1
* with errno EINVAL. https://bugs.python.org/issue47151. */
pid = fork();
}
} else
#endif
{
pid = fork();
}
if (pid != 0) {
// Parent process.
return pid;
}
/* Child process.
* See the comment above child_exec() for restrictions imposed on
* the code below.
*/
if (preexec_fn != Py_None) {
/* We'll be calling back into Python later so we need to do this.
* This call may not be async-signal-safe but neither is calling
* back into Python. The user asked us to use hope as a strategy
* to avoid deadlock... */
PyOS_AfterFork_Child();
}
child_exec(exec_array, argv, envp, cwd,
p2cread, p2cwrite, c2pread, c2pwrite,
errread, errwrite, errpipe_read, errpipe_write,
close_fds, restore_signals, call_setsid, pgid_to_set,
gid, extra_group_size, extra_groups,
uid, child_umask, child_sigmask,
fds_to_keep, fds_to_keep_len,
preexec_fn, preexec_fn_args_tuple);
_exit(255);
return 0; /* Dead code to avoid a potential compiler warning. */
}
/*[clinic input]
_posixsubprocess.fork_exec as subprocess_fork_exec
args as process_args: object
executable_list: object
close_fds: bool
pass_fds as py_fds_to_keep: object(subclass_of='&PyTuple_Type')
cwd as cwd_obj: object
env as env_list: object
p2cread: int
p2cwrite: int
c2pread: int
c2pwrite: int
errread: int
errwrite: int
errpipe_read: int
errpipe_write: int
restore_signals: bool
call_setsid: bool
pgid_to_set: pid_t
gid as gid_object: object
extra_groups as extra_groups_packed: object
uid as uid_object: object
child_umask: int
preexec_fn: object
allow_vfork: bool
/
Spawn a fresh new child process.
Fork a child process, close parent file descriptors as appropriate in the
child and duplicate the few that are needed before calling exec() in the
child process.
If close_fds is True, close file descriptors 3 and higher, except those listed
in the sorted tuple pass_fds.
The preexec_fn, if supplied, will be called immediately before closing file
descriptors and exec.
WARNING: preexec_fn is NOT SAFE if your application uses threads.
It may trigger infrequent, difficult to debug deadlocks.
If an error occurs in the child process before the exec, it is
serialized and written to the errpipe_write fd per subprocess.py.
Returns: the child process's PID.
Raises: Only on an error in the parent process.
[clinic start generated code]*/
static PyObject *