-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathcgroup_fuse.c
2105 lines (1778 loc) · 45.5 KB
/
cgroup_fuse.c
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
/* SPDX-License-Identifier: LGPL-2.1+ */
#include "config.h"
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <inttypes.h>
#include <libgen.h>
#include <pthread.h>
#include <sched.h>
#include <stdarg.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <wait.h>
#include <linux/magic.h>
#include <linux/sched.h>
#include <sys/epoll.h>
#include <sys/mman.h>
#include <sys/mount.h>
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/syscall.h>
#include <sys/sysinfo.h>
#include <sys/vfs.h>
#include "cgroup_fuse.h"
#include "bindings.h"
#include "cgroups/cgroup.h"
#include "cgroups/cgroup_utils.h"
#include "lxcfs_fuse_compat.h"
#include "memory_utils.h"
#include "utils.h"
struct cgfs_files {
char *name;
uint32_t uid, gid;
uint32_t mode;
};
struct pid_ns_clone_args {
int *cpipe;
int sock;
pid_t tpid;
/* pid_from_ns or pid_to_ns. */
int (*wrapped) (int, pid_t);
};
static inline int get_cgroup_fd_handle_named(const char *controller)
{
if (controller && strcmp(controller, "systemd") == 0)
return get_cgroup_fd("name=systemd");
return get_cgroup_fd(controller);
}
static char *get_pid_cgroup_handle_named(pid_t pid, const char *controller)
{
if (controller && strcmp(controller, "systemd") == 0)
return get_pid_cgroup(pid, "name=systemd");
return get_pid_cgroup(pid, controller);
}
static bool get_cgroup_handle_named(struct cgroup_ops *ops,
const char *controller, const char *cgroup,
const char *file, char **value)
{
if (controller && strcmp(controller, "systemd") == 0)
return cgroup_ops->get(ops, "name=systemd", cgroup, file, value);
return cgroup_ops->get(cgroup_ops, controller, cgroup, file, value);
}
/*
* given /cgroup/freezer/a/b, return "freezer".
* the returned char* should NOT be freed.
*/
static char *pick_controller_from_path(struct fuse_context *fc, const char *path)
{
const char *p1;
char *contr, *slash;
if (strlen(path) < 9) {
errno = EACCES;
return NULL;
}
if (*(path + 7) != '/') {
errno = EINVAL;
return NULL;
}
p1 = path + 8;
contr = strdupa(p1);
if (!contr) {
errno = ENOMEM;
return NULL;
}
slash = strstr(contr, "/");
if (slash)
*slash = '\0';
for (struct hierarchy **h = cgroup_ops->hierarchies; h && *h; h++) {
if ((*h)->__controllers && strcmp((*h)->__controllers, contr) == 0)
return (*h)->__controllers;
}
errno = ENOENT;
return NULL;
}
/*
* Find the start of cgroup in /cgroup/controller/the/cgroup/path
* Note that the returned value may include files (keynames) etc
*/
static const char *find_cgroup_in_path(const char *path)
{
const char *p1;
if (strlen(path) < 9) {
errno = EACCES;
return NULL;
}
p1 = strstr(path + 8, "/");
if (!p1) {
errno = EINVAL;
return NULL;
}
errno = 0;
return p1 + 1;
}
/*
* split the last path element from the path in @cg.
* @dir is newly allocated and should be freed, @last not
*/
static void get_cgdir_and_path(const char *cg, char **dir, char **last)
{
char *p;
do {
*dir = strdup(cg);
} while (!*dir);
*last = strrchr(cg, '/');
if (!*last) {
*last = NULL;
return;
}
p = strrchr(*dir, '/');
*p = '\0';
}
static bool is_child_cgroup(const char *controller, const char *cgroup,
const char *file)
{
__do_free char *path = NULL;
int cfd, ret;
struct stat sb;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return false;
path = must_make_path_relative(cgroup, file, NULL);
ret = fstatat(cfd, path, &sb, 0);
if (ret < 0 || !S_ISDIR(sb.st_mode))
return false;
return true;
}
/*
* If pid is in /a/b/c, they may see that /a exists, but not /b or /a/c.
*/
static bool caller_may_see_dir(pid_t pid, const char *contrl, const char *cg)
{
bool answer = false;
char *c2, *task_cg;
size_t target_len, task_len;
if (strcmp(cg, "/") == 0 || strcmp(cg, "./") == 0)
return true;
c2 = get_pid_cgroup_handle_named(pid, contrl);
if (!c2)
return false;
prune_init_slice(c2);
task_cg = c2 + 1;
target_len = strlen(cg);
task_len = strlen(task_cg);
if (task_len == 0) {
/* Task is in the root cg, it can see everything. This case is
* not handled by the strmcps below, since they test for the
* last /, but that is the first / that we've chopped off
* above.
*/
answer = true;
goto out;
}
if (strcmp(cg, task_cg) == 0) {
answer = true;
goto out;
}
if (target_len < task_len) {
/* looking up a parent dir */
if (strncmp(task_cg, cg, target_len) == 0 && task_cg[target_len] == '/')
answer = true;
goto out;
}
if (target_len > task_len) {
/* looking up a child dir */
if (strncmp(task_cg, cg, task_len) == 0 && cg[task_len] == '/')
answer = true;
goto out;
}
out:
free(c2);
return answer;
}
/*
* taskcg is a/b/c
* querycg is /a/b/c/d/e
* we return 'd'
*/
static char *get_next_cgroup_dir(const char *taskcg, const char *querycg)
{
char *start, *end;
if (strlen(taskcg) <= strlen(querycg)) {
lxcfs_error("%s\n", "I was fed bad input.");
return NULL;
}
if ((strcmp(querycg, "/") == 0) || (strcmp(querycg, "./") == 0))
start = strdup(taskcg + 1);
else
start = strdup(taskcg + strlen(querycg) + 1);
if (!start)
return NULL;
end = strchr(start, '/');
if (end)
*end = '\0';
return start;
}
/*
* If pid is in /a/b/c/d, they may only act on things under cg=/a/b/c/d.
* If pid is in /a, they may act on /a/b, but not on /b.
* if the answer is false and nextcg is not NULL, then *nextcg will point
* to a string containing the next cgroup directory under cg, which must be
* freed by the caller.
*/
static bool caller_is_in_ancestor(pid_t pid, const char *contrl, const char *cg, char **nextcg)
{
bool answer = false;
char *c2;
char *linecmp;
c2 = get_pid_cgroup_handle_named(pid, contrl);
if (!c2)
return false;
prune_init_slice(c2);
/*
* callers pass in '/' or './' (openat()) for root cgroup, otherwise
* they pass in a cgroup without leading '/'
*
* The original line here was:
* linecmp = *cg == '/' ? c2 : c2+1;
* TODO: I'm not sure why you'd want to increment when *cg != '/'?
* Serge, do you know?
*/
if (*cg == '/' || !strncmp(cg, "./", 2))
linecmp = c2;
else
linecmp = c2 + 1;
if (strncmp(linecmp, cg, strlen(linecmp)) != 0) {
if (nextcg) {
*nextcg = get_next_cgroup_dir(linecmp, cg);
}
goto out;
}
answer = true;
out:
free(c2);
return answer;
}
static struct cgfs_files *cgfs_get_key(const char *controller,
const char *cgroup, const char *file)
{
__do_free char *path = NULL;
struct cgfs_files *newkey;
int cfd, ret;
struct stat sb;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return false;
if (file && *file == '/')
file++;
if (file && strchr(file, '/'))
return NULL;
if (file)
path = must_make_path_relative(cgroup, file, NULL);
else
path = must_make_path_relative(cgroup, NULL);
ret = fstatat(cfd, path, &sb, 0);
if (ret < 0)
return NULL;
newkey = must_realloc(NULL, sizeof(struct cgfs_files));
if (file)
newkey->name = must_copy_string(file);
else if (strrchr(cgroup, '/'))
newkey->name = must_copy_string(strrchr(cgroup, '/'));
else
newkey->name = must_copy_string(cgroup);
newkey->uid = sb.st_uid;
newkey->gid = sb.st_gid;
newkey->mode = sb.st_mode;
return newkey;
}
/*
* Given a open file * to /proc/pid/{u,g}id_map, and an id
* valid in the caller's namespace, return the id mapped into
* pid's namespace.
* Returns the mapped id, or -1 on error.
*/
static int convert_id_to_ns(FILE *idfile, unsigned int in_id)
{
unsigned int nsuid, // base id for a range in the idfile's namespace
hostuid, // base id for a range in the caller's namespace
count; // number of ids in this range
char line[400];
int ret;
fseek(idfile, 0L, SEEK_SET);
while (fgets(line, 400, idfile)) {
ret = sscanf(line, "%u %u %u\n", &nsuid, &hostuid, &count);
if (ret != 3)
continue;
if (hostuid + count < hostuid || nsuid + count < nsuid) {
/*
* uids wrapped around - unexpected as this is a procfile,
* so just bail.
*/
lxcfs_error("pid wrapparound at entry %u %u %u in %s\n",
nsuid, hostuid, count, line);
return -1;
}
if (hostuid <= in_id && hostuid+count > in_id) {
/*
* now since hostuid <= in_id < hostuid+count, and
* hostuid+count and nsuid+count do not wrap around,
* we know that nsuid+(in_id-hostuid) which must be
* less that nsuid+(count) must not wrap around
*/
return (in_id - hostuid) + nsuid;
}
}
// no answer found
return -1;
}
/*
* for is_privileged_over,
* specify whether we require the calling uid to be root in their
* namespace
*/
#define NS_ROOT_REQD true
#define NS_ROOT_OPT false
#define PROCLEN 100
static bool is_privileged_over(pid_t pid, uid_t uid, uid_t victim, bool req_ns_root)
{
FILE *f;
char fpath[PROCLEN];
int ret;
bool answer = false;
uid_t nsuid;
if (victim == (uid_t)-1 || uid == (uid_t)-1)
return false;
/*
* If the request is one not requiring root in the namespace,
* then having the same uid suffices. (i.e. uid 1000 has write
* access to files owned by uid 1000
*/
if (!req_ns_root && uid == victim)
return true;
ret = snprintf(fpath, PROCLEN, "/proc/%d/uid_map", pid);
if (ret < 0 || ret >= PROCLEN)
return false;
f = fopen(fpath, "re");
if (!f)
return false;
/* if caller's not root in their namespace, reject */
nsuid = convert_id_to_ns(f, uid);
if (nsuid)
goto out;
/*
* If victim is not mapped into caller's ns, reject.
* XXX I'm not sure this check is needed given that fuse
* will be sending requests where the vfs has converted
*/
nsuid = convert_id_to_ns(f, victim);
if (nsuid == (uid_t)-1)
goto out;
answer = true;
out:
fclose(f);
return answer;
}
static bool perms_include(int fmode, mode_t req_mode)
{
mode_t r;
switch (req_mode & O_ACCMODE) {
case O_RDONLY:
r = S_IROTH;
break;
case O_WRONLY:
r = S_IWOTH;
break;
case O_RDWR:
r = S_IROTH | S_IWOTH;
break;
default:
return false;
}
return ((fmode & r) == r);
}
static void free_key(struct cgfs_files *k)
{
if (k) {
free_disarm(k->name);
free_disarm(k);
}
}
/*
* check whether a fuse context may access a cgroup dir or file
*
* If file is not null, it is a cgroup file to check under cg.
* If file is null, then we are checking perms on cg itself.
*
* For files we can check the mode of the list_keys result.
* For cgroups, we must make assumptions based on the files under the
* cgroup, because cgmanager doesn't tell us ownership/perms of cgroups
* yet.
*/
static bool fc_may_access(struct fuse_context *fc, const char *contrl, const char *cg, const char *file, mode_t mode)
{
struct cgfs_files *k = NULL;
bool ret = false;
k = cgfs_get_key(contrl, cg, file);
if (!k)
return false;
if (is_privileged_over(fc->pid, fc->uid, k->uid, NS_ROOT_OPT)) {
if (perms_include(k->mode >> 6, mode)) {
ret = true;
goto out;
}
}
if (fc->gid == k->gid) {
if (perms_include(k->mode >> 3, mode)) {
ret = true;
goto out;
}
}
ret = perms_include(k->mode, mode);
out:
free_key(k);
return ret;
}
__lxcfs_fuse_ops int cg_getattr(const char *path, struct stat *sb)
{
struct timespec now;
struct fuse_context *fc = fuse_get_context();
char * cgdir = NULL;
char *last = NULL, *path1, *path2;
struct cgfs_files *k = NULL;
const char *cgroup;
const char *controller = NULL;
int ret = -ENOENT;
if (!liblxcfs_functional())
return -EIO;
if (!fc || !cgroup_ops)
return -EIO;
memset(sb, 0, sizeof(struct stat));
if (clock_gettime(CLOCK_REALTIME, &now) < 0)
return -EINVAL;
sb->st_uid = sb->st_gid = 0;
sb->st_atim = sb->st_mtim = sb->st_ctim = now;
sb->st_size = 0;
if (pure_unified_layout(cgroup_ops) || strcmp(path, "/cgroup") == 0) {
sb->st_mode = S_IFDIR | 00755;
sb->st_nlink = 2;
return 0;
}
controller = pick_controller_from_path(fc, path);
if (!controller)
return -errno;
cgroup = find_cgroup_in_path(path);
if (!cgroup) {
/* this is just /cgroup/controller, return it as a dir */
sb->st_mode = S_IFDIR | 00755;
sb->st_nlink = 2;
return 0;
}
get_cgdir_and_path(cgroup, &cgdir, &last);
if (!last) {
path1 = "/";
path2 = cgdir;
} else {
path1 = cgdir;
path2 = last;
}
pid_t initpid = lookup_initpid_in_store(fc->pid);
if (initpid <= 1 || is_shared_pidns(initpid))
initpid = fc->pid;
/* check that cgcopy is either a child cgroup of cgdir, or listed in its keys.
* Then check that caller's cgroup is under path if last is a child
* cgroup, or cgdir if last is a file */
if (is_child_cgroup(controller, path1, path2)) {
if (!caller_may_see_dir(initpid, controller, cgroup)) {
ret = -ENOENT;
goto out;
}
if (!caller_is_in_ancestor(initpid, controller, cgroup, NULL)) {
/* this is just /cgroup/controller, return it as a dir */
sb->st_mode = S_IFDIR | 00555;
sb->st_nlink = 2;
ret = 0;
goto out;
}
if (!fc_may_access(fc, controller, cgroup, NULL, O_RDONLY)) {
ret = -EACCES;
goto out;
}
// get uid, gid, from '/tasks' file and make up a mode
// That is a hack, until cgmanager gains a GetCgroupPerms fn.
sb->st_mode = S_IFDIR | 00755;
k = cgfs_get_key(controller, cgroup, NULL);
if (!k) {
sb->st_uid = sb->st_gid = 0;
} else {
sb->st_uid = k->uid;
sb->st_gid = k->gid;
}
free_key(k);
sb->st_nlink = 2;
ret = 0;
goto out;
}
if ((k = cgfs_get_key(controller, path1, path2)) != NULL) {
sb->st_mode = S_IFREG | k->mode;
sb->st_nlink = 1;
sb->st_uid = k->uid;
sb->st_gid = k->gid;
sb->st_size = 4096;
free_key(k);
if (!caller_is_in_ancestor(initpid, controller, path1, NULL)) {
ret = -ENOENT;
goto out;
}
ret = 0;
}
out:
free(cgdir);
return ret;
}
/*
* Chown all the files in the cgroup directory. We do this when we create a
* cgroup on behalf of a user.
*/
static void chown_all_cgroup_files(const char *dirname, uid_t uid, gid_t gid, int fd)
{
struct dirent *direntp;
char path[MAXPATHLEN];
size_t len;
DIR *d;
int fd1, ret;
len = strlen(dirname);
if (len >= MAXPATHLEN) {
lxcfs_error("Pathname too long: %s\n", dirname);
return;
}
fd1 = openat(fd, dirname, O_DIRECTORY);
if (fd1 < 0)
return;
d = fdopendir(fd1);
if (!d) {
lxcfs_error("Failed to open %s\n", dirname);
return;
}
while ((direntp = readdir(d))) {
if (!strcmp(direntp->d_name, ".") || !strcmp(direntp->d_name, ".."))
continue;
ret = snprintf(path, MAXPATHLEN, "%s/%s", dirname, direntp->d_name);
if (ret < 0 || ret >= MAXPATHLEN) {
lxcfs_error("Pathname too long under %s\n", dirname);
continue;
}
if (fchownat(fd, path, uid, gid, 0) < 0)
lxcfs_error("Failed to chown file %s to %u:%u", path, uid, gid);
}
closedir(d);
}
static int cgfs_create(const char *controller, const char *cg, uid_t uid, gid_t gid)
{
__do_free char *path = NULL;
int cfd;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return -EINVAL;
path = must_make_path_relative(cg, NULL);
if (mkdirat(cfd, path, 0755) < 0)
return -errno;
if (uid == 0 && gid == 0)
return 0;
if (fchownat(cfd, path, uid, gid, 0) < 0)
return -errno;
chown_all_cgroup_files(path, uid, gid, cfd);
return 0;
}
__lxcfs_fuse_ops int cg_mkdir(const char *path, mode_t mode)
{
struct fuse_context *fc = fuse_get_context();
char *last = NULL, *path1, *cgdir = NULL, *controller, *next = NULL;
const char *cgroup;
int ret;
if (!liblxcfs_functional())
return -EIO;
if (!fc || !cgroup_ops || pure_unified_layout(cgroup_ops))
return -EIO;
controller = pick_controller_from_path(fc, path);
if (!controller)
return errno == ENOENT ? -EPERM : -errno;
cgroup = find_cgroup_in_path(path);
if (!cgroup)
return -errno;
get_cgdir_and_path(cgroup, &cgdir, &last);
if (!last)
path1 = "/";
else
path1 = cgdir;
pid_t initpid = lookup_initpid_in_store(fc->pid);
if (initpid <= 1 || is_shared_pidns(initpid))
initpid = fc->pid;
if (!caller_is_in_ancestor(initpid, controller, path1, &next)) {
if (!next)
ret = -EINVAL;
else if (last && strcmp(next, last) == 0)
ret = -EEXIST;
else
ret = -EPERM;
goto out;
}
if (!fc_may_access(fc, controller, path1, NULL, O_RDWR)) {
ret = -EACCES;
goto out;
}
if (!caller_is_in_ancestor(initpid, controller, path1, NULL)) {
ret = -EACCES;
goto out;
}
ret = cgfs_create(controller, cgroup, fc->uid, fc->gid);
out:
free(cgdir);
free(next);
return ret;
}
static bool recursive_rmdir(const char *dirname, int fd, const int cfd)
{
__do_close int dupfd = -EBADF;
__do_closedir DIR *dir = NULL;
bool ret = false;
struct dirent *direntp;
char pathname[MAXPATHLEN];
dupfd = dup(fd);
if (dupfd < 0)
return false;
dir = fdopendir(dupfd);
if (!dir) {
lxcfs_debug("Failed to open %s: %s.\n", dirname, strerror(errno));
return false;
}
move_fd(dupfd);
while ((direntp = readdir(dir))) {
struct stat mystat;
int rc;
if (!strcmp(direntp->d_name, ".") ||
!strcmp(direntp->d_name, ".."))
continue;
rc = snprintf(pathname, MAXPATHLEN, "%s/%s", dirname, direntp->d_name);
if (rc < 0 || rc >= MAXPATHLEN) {
lxcfs_error("%s\n", "Pathname too long.");
continue;
}
rc = fstatat(cfd, pathname, &mystat, AT_SYMLINK_NOFOLLOW);
if (rc) {
lxcfs_debug("Failed to stat %s: %s.\n", pathname, strerror(errno));
continue;
}
if (S_ISDIR(mystat.st_mode))
if (!recursive_rmdir(pathname, fd, cfd))
lxcfs_debug("Error removing %s.\n", pathname);
}
ret = true;
if (unlinkat(cfd, dirname, AT_REMOVEDIR) < 0) {
lxcfs_debug("Failed to delete %s: %s.\n", dirname, strerror(errno));
ret = false;
}
return ret;
}
static bool cgfs_remove(const char *controller, const char *cgroup)
{
__do_close int fd = -EBADF;
__do_free char *path = NULL;
int cfd;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return false;
path = must_make_path_relative(cgroup, NULL);
fd = openat(cfd, path, O_DIRECTORY);
if (fd < 0)
return false;
return recursive_rmdir(path, fd, cfd);
}
__lxcfs_fuse_ops int cg_rmdir(const char *path)
{
struct fuse_context *fc = fuse_get_context();
char *last = NULL, *cgdir = NULL, *controller, *next = NULL;
const char *cgroup;
int ret;
if (!liblxcfs_functional())
return -EIO;
if (!fc || !cgroup_ops || pure_unified_layout(cgroup_ops))
return -EIO;
controller = pick_controller_from_path(fc, path);
if (!controller) /* Someone's trying to delete "/cgroup". */
return -EPERM;
cgroup = find_cgroup_in_path(path);
if (!cgroup) /* Someone's trying to delete a controller e.g. "/blkio". */
return -EPERM;
get_cgdir_and_path(cgroup, &cgdir, &last);
if (!last) {
/* Someone's trying to delete a cgroup on the same level as the
* "/lxc" cgroup e.g. rmdir "/cgroup/blkio/lxc" or
* rmdir "/cgroup/blkio/init.slice".
*/
ret = -EPERM;
goto out;
}
pid_t initpid = lookup_initpid_in_store(fc->pid);
if (initpid <= 1 || is_shared_pidns(initpid))
initpid = fc->pid;
if (!caller_is_in_ancestor(initpid, controller, cgroup, &next)) {
if (!last || (next && (strcmp(next, last) == 0)))
ret = -EBUSY;
else
ret = -ENOENT;
goto out;
}
if (!fc_may_access(fc, controller, cgdir, NULL, O_WRONLY)) {
ret = -EACCES;
goto out;
}
if (!caller_is_in_ancestor(initpid, controller, cgroup, NULL)) {
ret = -EACCES;
goto out;
}
if (!cgfs_remove(controller, cgroup)) {
ret = -EINVAL;
goto out;
}
ret = 0;
out:
free(cgdir);
free(next);
return ret;
}
static bool cgfs_chmod_file(const char *controller, const char *file, mode_t mode)
{
__do_free char *path = NULL;
int cfd;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return false;
path = must_make_path_relative(file, NULL);
if (fchmodat(cfd, path, mode, 0) < 0)
return false;
return true;
}
__lxcfs_fuse_ops int cg_chmod(const char *path, mode_t mode)
{
struct fuse_context *fc = fuse_get_context();
char * cgdir = NULL, *last = NULL, *path1, *path2, *controller;
struct cgfs_files *k = NULL;
const char *cgroup;
int ret;
if (!liblxcfs_functional())
return -EIO;
if (!fc || !cgroup_ops || pure_unified_layout(cgroup_ops))
return -EIO;
if (strcmp(path, "/cgroup") == 0)
return -EPERM;
controller = pick_controller_from_path(fc, path);
if (!controller)
return errno == ENOENT ? -EPERM : -errno;
cgroup = find_cgroup_in_path(path);
if (!cgroup)
/* this is just /cgroup/controller */
return -EPERM;
get_cgdir_and_path(cgroup, &cgdir, &last);
if (!last) {
path1 = "/";
path2 = cgdir;
} else {
path1 = cgdir;
path2 = last;
}
if (is_child_cgroup(controller, path1, path2)) {
// get uid, gid, from '/tasks' file and make up a mode
// That is a hack, until cgmanager gains a GetCgroupPerms fn.
k = cgfs_get_key(controller, cgroup, "tasks");
} else
k = cgfs_get_key(controller, path1, path2);
if (!k) {
ret = -EINVAL;
goto out;
}
/*
* This being a fuse request, the uid and gid must be valid
* in the caller's namespace. So we can just check to make
* sure that the caller is root in their uid, and privileged
* over the file's current owner.
*/
if (!is_privileged_over(fc->pid, fc->uid, k->uid, NS_ROOT_OPT)) {
ret = -EPERM;
goto out;
}
if (!cgfs_chmod_file(controller, cgroup, mode)) {
ret = -EINVAL;
goto out;
}
ret = 0;
out:
free_key(k);
free(cgdir);
return ret;
}
static inline bool is_dir(int dirfd, const char *path)
{
struct stat st;
return fstatat(dirfd, path, &st, 0) == 0 && S_ISDIR(st.st_mode);
}
static int chown_tasks_files(int dirfd, const char *dirname, uid_t uid, gid_t gid)
{
__do_free char *path;
path = must_make_path_relative(dirname, "tasks", NULL);
if (fchownat(dirfd, path, uid, gid, 0) != 0)
return -errno;
free_disarm(path);
path = must_make_path_relative(dirname, "cgroup.procs", NULL);
if (fchownat(dirfd, path, uid, gid, 0) != 0)
return -errno;
return 0;
}
static int cgfs_chown_file(const char *controller, const char *file, uid_t uid, gid_t gid)
{
__do_free char *path = NULL;
int cfd;
cfd = get_cgroup_fd_handle_named(controller);
if (cfd < 0)
return false;
path = must_make_path_relative(file, NULL);
if (fchownat(cfd, path, uid, gid, 0) < 0)
return -errno;
if (is_dir(cfd, path))