]>
Commit | Line | Data |
---|---|---|
15ae422b LP |
1 | /*** |
2 | This file is part of systemd. | |
3 | ||
4 | Copyright 2010 Lennart Poettering | |
5 | ||
6 | systemd is free software; you can redistribute it and/or modify it | |
5430f7f2 LP |
7 | under the terms of the GNU Lesser General Public License as published by |
8 | the Free Software Foundation; either version 2.1 of the License, or | |
15ae422b LP |
9 | (at your option) any later version. |
10 | ||
11 | systemd is distributed in the hope that it will be useful, but | |
12 | WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
5430f7f2 | 14 | Lesser General Public License for more details. |
15ae422b | 15 | |
5430f7f2 | 16 | You should have received a copy of the GNU Lesser General Public License |
15ae422b LP |
17 | along with systemd; If not, see <http://www.gnu.org/licenses/>. |
18 | ***/ | |
19 | ||
20 | #include <errno.h> | |
07630cea | 21 | #include <sched.h> |
15ae422b | 22 | #include <stdio.h> |
07630cea LP |
23 | #include <string.h> |
24 | #include <sys/mount.h> | |
15ae422b | 25 | #include <sys/stat.h> |
07630cea | 26 | #include <unistd.h> |
25e870b5 | 27 | #include <linux/fs.h> |
15ae422b | 28 | |
b5efdb8a | 29 | #include "alloc-util.h" |
7f112f50 | 30 | #include "dev-setup.h" |
3ffd4af2 | 31 | #include "fd-util.h" |
d944dc95 | 32 | #include "fs-util.h" |
07630cea LP |
33 | #include "loopback-setup.h" |
34 | #include "missing.h" | |
35 | #include "mkdir.h" | |
4349cd7c | 36 | #include "mount-util.h" |
3ffd4af2 | 37 | #include "namespace.h" |
07630cea | 38 | #include "path-util.h" |
d7b8eec7 | 39 | #include "selinux-util.h" |
2583fbea | 40 | #include "socket-util.h" |
8b43440b | 41 | #include "string-table.h" |
07630cea LP |
42 | #include "string-util.h" |
43 | #include "strv.h" | |
affb60b1 | 44 | #include "umask-util.h" |
ee104e11 | 45 | #include "user-util.h" |
07630cea | 46 | #include "util.h" |
15ae422b | 47 | |
737ba3c8 | 48 | #define DEV_MOUNT_OPTIONS (MS_NOSUID|MS_STRICTATIME|MS_NOEXEC) |
49 | ||
c17ec25e | 50 | typedef enum MountMode { |
15ae422b LP |
51 | /* This is ordered by priority! */ |
52 | INACCESSIBLE, | |
53 | READONLY, | |
ac0930c8 LP |
54 | PRIVATE_TMP, |
55 | PRIVATE_VAR_TMP, | |
7f112f50 | 56 | PRIVATE_DEV, |
59eeb84b | 57 | READWRITE, |
c17ec25e | 58 | } MountMode; |
15ae422b | 59 | |
34de407a | 60 | typedef struct MountEntry { |
5327c910 | 61 | const char *path_const; /* Memory allocated on stack or static */ |
cfbeb4ef | 62 | MountMode mode:5; |
5327c910 LP |
63 | bool ignore:1; /* Ignore if path does not exist? */ |
64 | bool has_prefix:1; /* Already is prefixed by the root dir? */ | |
cfbeb4ef | 65 | bool read_only:1; /* Shall this mount point be read-only? */ |
5327c910 | 66 | char *path_malloc; /* Use this instead of 'path' if we had to allocate memory */ |
34de407a | 67 | } MountEntry; |
15ae422b | 68 | |
f471b2af DH |
69 | /* |
70 | * The following Protect tables are to protect paths and mark some of them | |
71 | * READONLY, in case a path is covered by an option from another table, then | |
72 | * it is marked READWRITE in the current one, and the more restrictive mode is | |
73 | * applied from that other table. This way all options can be combined in a | |
74 | * safe and comprehensible way for users. | |
75 | */ | |
76 | ||
11a30cec | 77 | /* ProtectKernelTunables= option and the related filesystem APIs */ |
34de407a | 78 | static const MountEntry protect_kernel_tunables_table[] = { |
c6232fb0 LP |
79 | { "/proc/sys", READONLY, false }, |
80 | { "/proc/sysrq-trigger", READONLY, true }, | |
81 | { "/proc/latency_stats", READONLY, true }, | |
82 | { "/proc/mtrr", READONLY, true }, | |
aa70f38b | 83 | { "/proc/apm", READONLY, true }, /* Obsolete API, there's no point in permitting access to this, ever */ |
c6232fb0 LP |
84 | { "/proc/acpi", READONLY, true }, |
85 | { "/proc/timer_stats", READONLY, true }, | |
86 | { "/proc/asound", READONLY, true }, | |
87 | { "/proc/bus", READONLY, true }, | |
88 | { "/proc/fs", READONLY, true }, | |
89 | { "/proc/irq", READONLY, true }, | |
90 | { "/sys", READONLY, false }, | |
91 | { "/sys/kernel/debug", READONLY, true }, | |
92 | { "/sys/kernel/tracing", READONLY, true }, | |
93 | { "/sys/fs/cgroup", READWRITE, false }, /* READONLY is set by ProtectControlGroups= option */ | |
11a30cec DH |
94 | }; |
95 | ||
c575770b | 96 | /* ProtectKernelModules= option */ |
34de407a | 97 | static const MountEntry protect_kernel_modules_table[] = { |
c575770b | 98 | #ifdef HAVE_SPLIT_USR |
c6232fb0 | 99 | { "/lib/modules", INACCESSIBLE, true }, |
c575770b | 100 | #endif |
c6232fb0 | 101 | { "/usr/lib/modules", INACCESSIBLE, true }, |
c575770b DH |
102 | }; |
103 | ||
b6c432ca DH |
104 | /* |
105 | * ProtectHome=read-only table, protect $HOME and $XDG_RUNTIME_DIR and rest of | |
106 | * system should be protected by ProtectSystem= | |
107 | */ | |
34de407a | 108 | static const MountEntry protect_home_read_only_table[] = { |
c6232fb0 LP |
109 | { "/home", READONLY, true }, |
110 | { "/run/user", READONLY, true }, | |
111 | { "/root", READONLY, true }, | |
b6c432ca DH |
112 | }; |
113 | ||
114 | /* ProtectHome=yes table */ | |
34de407a | 115 | static const MountEntry protect_home_yes_table[] = { |
c6232fb0 LP |
116 | { "/home", INACCESSIBLE, true }, |
117 | { "/run/user", INACCESSIBLE, true }, | |
118 | { "/root", INACCESSIBLE, true }, | |
b6c432ca DH |
119 | }; |
120 | ||
f471b2af | 121 | /* ProtectSystem=yes table */ |
34de407a | 122 | static const MountEntry protect_system_yes_table[] = { |
c6232fb0 LP |
123 | { "/usr", READONLY, false }, |
124 | { "/boot", READONLY, true }, | |
125 | { "/efi", READONLY, true }, | |
f471b2af DH |
126 | }; |
127 | ||
128 | /* ProtectSystem=full includes ProtectSystem=yes */ | |
34de407a | 129 | static const MountEntry protect_system_full_table[] = { |
c6232fb0 LP |
130 | { "/usr", READONLY, false }, |
131 | { "/boot", READONLY, true }, | |
132 | { "/efi", READONLY, true }, | |
133 | { "/etc", READONLY, false }, | |
f471b2af DH |
134 | }; |
135 | ||
136 | /* | |
137 | * ProtectSystem=strict table. In this strict mode, we mount everything | |
138 | * read-only, except for /proc, /dev, /sys which are the kernel API VFS, | |
139 | * which are left writable, but PrivateDevices= + ProtectKernelTunables= | |
140 | * protect those, and these options should be fully orthogonal. | |
141 | * (And of course /home and friends are also left writable, as ProtectHome= | |
142 | * shall manage those, orthogonally). | |
143 | */ | |
34de407a | 144 | static const MountEntry protect_system_strict_table[] = { |
ddbe0412 LP |
145 | { "/", READONLY, false }, |
146 | { "/proc", READWRITE, false }, /* ProtectKernelTunables= */ | |
147 | { "/sys", READWRITE, false }, /* ProtectKernelTunables= */ | |
148 | { "/dev", READWRITE, false }, /* PrivateDevices= */ | |
149 | { "/home", READWRITE, true }, /* ProtectHome= */ | |
150 | { "/run/user", READWRITE, true }, /* ProtectHome= */ | |
151 | { "/root", READWRITE, true }, /* ProtectHome= */ | |
f471b2af DH |
152 | }; |
153 | ||
34de407a | 154 | static const char *mount_entry_path(const MountEntry *p) { |
f0a4feb0 DH |
155 | assert(p); |
156 | ||
5327c910 LP |
157 | /* Returns the path of this bind mount. If the malloc()-allocated ->path_buffer field is set we return that, |
158 | * otherwise the stack/static ->path field is returned. */ | |
f0a4feb0 | 159 | |
5327c910 | 160 | return p->path_malloc ?: p->path_const; |
f0a4feb0 DH |
161 | } |
162 | ||
34de407a | 163 | static bool mount_entry_read_only(const MountEntry *p) { |
cfbeb4ef LP |
164 | assert(p); |
165 | ||
166 | return p->read_only || IN_SET(p->mode, READONLY, INACCESSIBLE); | |
167 | } | |
168 | ||
34de407a | 169 | static int append_access_mounts(MountEntry **p, char **strv, MountMode mode) { |
15ae422b LP |
170 | char **i; |
171 | ||
613b411c LP |
172 | assert(p); |
173 | ||
5327c910 LP |
174 | /* Adds a list of user-supplied READWRITE/READONLY/INACCESSIBLE entries */ |
175 | ||
15ae422b | 176 | STRV_FOREACH(i, strv) { |
5327c910 LP |
177 | bool ignore = false, needs_prefix = false; |
178 | const char *e = *i; | |
15ae422b | 179 | |
5327c910 LP |
180 | /* Look for any prefixes */ |
181 | if (startswith(e, "-")) { | |
182 | e++; | |
9c94d52e | 183 | ignore = true; |
ea92ae33 | 184 | } |
5327c910 LP |
185 | if (startswith(e, "+")) { |
186 | e++; | |
187 | needs_prefix = true; | |
188 | } | |
ea92ae33 | 189 | |
5327c910 | 190 | if (!path_is_absolute(e)) |
15ae422b LP |
191 | return -EINVAL; |
192 | ||
34de407a | 193 | *((*p)++) = (MountEntry) { |
5327c910 LP |
194 | .path_const = e, |
195 | .mode = mode, | |
196 | .ignore = ignore, | |
197 | .has_prefix = !needs_prefix, | |
198 | }; | |
15ae422b LP |
199 | } |
200 | ||
201 | return 0; | |
202 | } | |
203 | ||
34de407a | 204 | static int append_static_mounts(MountEntry **p, const MountEntry *mounts, unsigned n, bool ignore_protect) { |
f471b2af | 205 | unsigned i; |
11a30cec DH |
206 | |
207 | assert(p); | |
f471b2af | 208 | assert(mounts); |
11a30cec | 209 | |
5327c910 | 210 | /* Adds a list of static pre-defined entries */ |
f471b2af | 211 | |
5327c910 | 212 | for (i = 0; i < n; i++) |
34de407a LP |
213 | *((*p)++) = (MountEntry) { |
214 | .path_const = mount_entry_path(mounts+i), | |
5327c910 LP |
215 | .mode = mounts[i].mode, |
216 | .ignore = mounts[i].ignore || ignore_protect, | |
217 | }; | |
f471b2af DH |
218 | |
219 | return 0; | |
220 | } | |
221 | ||
34de407a | 222 | static int append_protect_home(MountEntry **p, ProtectHome protect_home, bool ignore_protect) { |
c575770b DH |
223 | assert(p); |
224 | ||
5327c910 | 225 | switch (protect_home) { |
b6c432ca | 226 | |
5327c910 | 227 | case PROTECT_HOME_NO: |
b6c432ca DH |
228 | return 0; |
229 | ||
b6c432ca | 230 | case PROTECT_HOME_READ_ONLY: |
5327c910 LP |
231 | return append_static_mounts(p, protect_home_read_only_table, ELEMENTSOF(protect_home_read_only_table), ignore_protect); |
232 | ||
b6c432ca | 233 | case PROTECT_HOME_YES: |
5327c910 LP |
234 | return append_static_mounts(p, protect_home_yes_table, ELEMENTSOF(protect_home_yes_table), ignore_protect); |
235 | ||
b6c432ca | 236 | default: |
5327c910 | 237 | assert_not_reached("Unexpected ProtectHome= value"); |
b6c432ca | 238 | } |
b6c432ca DH |
239 | } |
240 | ||
34de407a | 241 | static int append_protect_system(MountEntry **p, ProtectSystem protect_system, bool ignore_protect) { |
f471b2af DH |
242 | assert(p); |
243 | ||
5327c910 LP |
244 | switch (protect_system) { |
245 | ||
246 | case PROTECT_SYSTEM_NO: | |
f471b2af DH |
247 | return 0; |
248 | ||
f471b2af | 249 | case PROTECT_SYSTEM_STRICT: |
5327c910 LP |
250 | return append_static_mounts(p, protect_system_strict_table, ELEMENTSOF(protect_system_strict_table), ignore_protect); |
251 | ||
f471b2af | 252 | case PROTECT_SYSTEM_YES: |
5327c910 LP |
253 | return append_static_mounts(p, protect_system_yes_table, ELEMENTSOF(protect_system_yes_table), ignore_protect); |
254 | ||
f471b2af | 255 | case PROTECT_SYSTEM_FULL: |
5327c910 LP |
256 | return append_static_mounts(p, protect_system_full_table, ELEMENTSOF(protect_system_full_table), ignore_protect); |
257 | ||
f471b2af | 258 | default: |
5327c910 | 259 | assert_not_reached("Unexpected ProtectSystem= value"); |
f471b2af | 260 | } |
11a30cec DH |
261 | } |
262 | ||
c17ec25e | 263 | static int mount_path_compare(const void *a, const void *b) { |
34de407a | 264 | const MountEntry *p = a, *q = b; |
a0827e2b | 265 | int d; |
15ae422b | 266 | |
6ee1a919 | 267 | /* If the paths are not equal, then order prefixes first */ |
34de407a | 268 | d = path_compare(mount_entry_path(p), mount_entry_path(q)); |
6ee1a919 LP |
269 | if (d != 0) |
270 | return d; | |
15ae422b | 271 | |
6ee1a919 LP |
272 | /* If the paths are equal, check the mode */ |
273 | if (p->mode < q->mode) | |
274 | return -1; | |
15ae422b | 275 | |
6ee1a919 LP |
276 | if (p->mode > q->mode) |
277 | return 1; | |
15ae422b | 278 | |
6ee1a919 | 279 | return 0; |
15ae422b LP |
280 | } |
281 | ||
34de407a | 282 | static int prefix_where_needed(MountEntry *m, unsigned n, const char *root_directory) { |
5327c910 LP |
283 | unsigned i; |
284 | ||
285 | /* Prefixes all paths in the bind mount table with the root directory if it is specified and the entry needs | |
286 | * that. */ | |
287 | ||
288 | if (!root_directory) | |
289 | return 0; | |
290 | ||
291 | for (i = 0; i < n; i++) { | |
292 | char *s; | |
293 | ||
294 | if (m[i].has_prefix) | |
295 | continue; | |
296 | ||
34de407a | 297 | s = prefix_root(root_directory, mount_entry_path(m+i)); |
5327c910 LP |
298 | if (!s) |
299 | return -ENOMEM; | |
300 | ||
301 | free(m[i].path_malloc); | |
302 | m[i].path_malloc = s; | |
303 | ||
304 | m[i].has_prefix = true; | |
305 | } | |
306 | ||
307 | return 0; | |
308 | } | |
309 | ||
34de407a LP |
310 | static void drop_duplicates(MountEntry *m, unsigned *n) { |
311 | MountEntry *f, *t, *previous; | |
15ae422b | 312 | |
c17ec25e | 313 | assert(m); |
15ae422b | 314 | assert(n); |
15ae422b | 315 | |
fe3c2583 LP |
316 | /* Drops duplicate entries. Expects that the array is properly ordered already. */ |
317 | ||
1d54cd5d | 318 | for (f = m, t = m, previous = NULL; f < m + *n; f++) { |
15ae422b | 319 | |
fe3c2583 LP |
320 | /* The first one wins (which is the one with the more restrictive mode), see mount_path_compare() |
321 | * above. */ | |
34de407a LP |
322 | if (previous && path_equal(mount_entry_path(f), mount_entry_path(previous))) { |
323 | log_debug("%s is duplicate.", mount_entry_path(f)); | |
324 | previous->read_only = previous->read_only || mount_entry_read_only(f); /* Propagate the read-only flag to the remaining entry */ | |
5327c910 | 325 | f->path_malloc = mfree(f->path_malloc); |
15ae422b | 326 | continue; |
fe3c2583 | 327 | } |
15ae422b | 328 | |
e2d7c1a0 | 329 | *t = *f; |
15ae422b | 330 | previous = t; |
fe3c2583 LP |
331 | t++; |
332 | } | |
333 | ||
334 | *n = t - m; | |
335 | } | |
336 | ||
34de407a LP |
337 | static void drop_inaccessible(MountEntry *m, unsigned *n) { |
338 | MountEntry *f, *t; | |
fe3c2583 LP |
339 | const char *clear = NULL; |
340 | ||
341 | assert(m); | |
342 | assert(n); | |
343 | ||
344 | /* Drops all entries obstructed by another entry further up the tree. Expects that the array is properly | |
345 | * ordered already. */ | |
346 | ||
1d54cd5d | 347 | for (f = m, t = m; f < m + *n; f++) { |
fe3c2583 LP |
348 | |
349 | /* If we found a path set for INACCESSIBLE earlier, and this entry has it as prefix we should drop | |
350 | * it, as inaccessible paths really should drop the entire subtree. */ | |
34de407a LP |
351 | if (clear && path_startswith(mount_entry_path(f), clear)) { |
352 | log_debug("%s is masked by %s.", mount_entry_path(f), clear); | |
5327c910 | 353 | f->path_malloc = mfree(f->path_malloc); |
fe3c2583 LP |
354 | continue; |
355 | } | |
15ae422b | 356 | |
34de407a | 357 | clear = f->mode == INACCESSIBLE ? mount_entry_path(f) : NULL; |
fe3c2583 LP |
358 | |
359 | *t = *f; | |
15ae422b LP |
360 | t++; |
361 | } | |
362 | ||
c17ec25e | 363 | *n = t - m; |
15ae422b LP |
364 | } |
365 | ||
34de407a LP |
366 | static void drop_nop(MountEntry *m, unsigned *n) { |
367 | MountEntry *f, *t; | |
7648a565 LP |
368 | |
369 | assert(m); | |
370 | assert(n); | |
371 | ||
372 | /* Drops all entries which have an immediate parent that has the same type, as they are redundant. Assumes the | |
373 | * list is ordered by prefixes. */ | |
374 | ||
1d54cd5d | 375 | for (f = m, t = m; f < m + *n; f++) { |
7648a565 LP |
376 | |
377 | /* Only suppress such subtrees for READONLY and READWRITE entries */ | |
378 | if (IN_SET(f->mode, READONLY, READWRITE)) { | |
34de407a | 379 | MountEntry *p; |
7648a565 LP |
380 | bool found = false; |
381 | ||
382 | /* Now let's find the first parent of the entry we are looking at. */ | |
383 | for (p = t-1; p >= m; p--) { | |
34de407a | 384 | if (path_startswith(mount_entry_path(f), mount_entry_path(p))) { |
7648a565 LP |
385 | found = true; |
386 | break; | |
387 | } | |
388 | } | |
389 | ||
390 | /* We found it, let's see if it's the same mode, if so, we can drop this entry */ | |
391 | if (found && p->mode == f->mode) { | |
34de407a | 392 | log_debug("%s is redundant by %s", mount_entry_path(f), mount_entry_path(p)); |
5327c910 | 393 | f->path_malloc = mfree(f->path_malloc); |
7648a565 LP |
394 | continue; |
395 | } | |
396 | } | |
397 | ||
398 | *t = *f; | |
399 | t++; | |
400 | } | |
401 | ||
402 | *n = t - m; | |
403 | } | |
404 | ||
34de407a LP |
405 | static void drop_outside_root(const char *root_directory, MountEntry *m, unsigned *n) { |
406 | MountEntry *f, *t; | |
cd2902c9 LP |
407 | |
408 | assert(m); | |
409 | assert(n); | |
410 | ||
1d54cd5d | 411 | /* Nothing to do */ |
cd2902c9 LP |
412 | if (!root_directory) |
413 | return; | |
414 | ||
415 | /* Drops all mounts that are outside of the root directory. */ | |
416 | ||
1d54cd5d | 417 | for (f = m, t = m; f < m + *n; f++) { |
cd2902c9 | 418 | |
34de407a LP |
419 | if (!path_startswith(mount_entry_path(f), root_directory)) { |
420 | log_debug("%s is outside of root directory.", mount_entry_path(f)); | |
5327c910 | 421 | f->path_malloc = mfree(f->path_malloc); |
cd2902c9 LP |
422 | continue; |
423 | } | |
424 | ||
425 | *t = *f; | |
426 | t++; | |
427 | } | |
428 | ||
429 | *n = t - m; | |
430 | } | |
431 | ||
34de407a | 432 | static int mount_dev(MountEntry *m) { |
7f112f50 LP |
433 | static const char devnodes[] = |
434 | "/dev/null\0" | |
435 | "/dev/zero\0" | |
436 | "/dev/full\0" | |
437 | "/dev/random\0" | |
438 | "/dev/urandom\0" | |
439 | "/dev/tty\0"; | |
440 | ||
2b85f4e1 | 441 | char temporary_mount[] = "/tmp/namespace-dev-XXXXXX"; |
63cc4c31 | 442 | const char *d, *dev = NULL, *devpts = NULL, *devshm = NULL, *devhugepages = NULL, *devmqueue = NULL, *devlog = NULL, *devptmx = NULL; |
7f112f50 LP |
443 | _cleanup_umask_ mode_t u; |
444 | int r; | |
445 | ||
446 | assert(m); | |
447 | ||
448 | u = umask(0000); | |
449 | ||
2b85f4e1 LP |
450 | if (!mkdtemp(temporary_mount)) |
451 | return -errno; | |
452 | ||
63c372cb | 453 | dev = strjoina(temporary_mount, "/dev"); |
dc751688 | 454 | (void) mkdir(dev, 0755); |
737ba3c8 | 455 | if (mount("tmpfs", dev, "tmpfs", DEV_MOUNT_OPTIONS, "mode=755") < 0) { |
2b85f4e1 LP |
456 | r = -errno; |
457 | goto fail; | |
458 | } | |
459 | ||
63c372cb | 460 | devpts = strjoina(temporary_mount, "/dev/pts"); |
dc751688 | 461 | (void) mkdir(devpts, 0755); |
2b85f4e1 LP |
462 | if (mount("/dev/pts", devpts, NULL, MS_BIND, NULL) < 0) { |
463 | r = -errno; | |
464 | goto fail; | |
465 | } | |
466 | ||
63c372cb | 467 | devptmx = strjoina(temporary_mount, "/dev/ptmx"); |
3164e3cb ZJS |
468 | if (symlink("pts/ptmx", devptmx) < 0) { |
469 | r = -errno; | |
470 | goto fail; | |
471 | } | |
e06b6479 | 472 | |
63c372cb | 473 | devshm = strjoina(temporary_mount, "/dev/shm"); |
dc751688 | 474 | (void) mkdir(devshm, 01777); |
2b85f4e1 LP |
475 | r = mount("/dev/shm", devshm, NULL, MS_BIND, NULL); |
476 | if (r < 0) { | |
477 | r = -errno; | |
478 | goto fail; | |
479 | } | |
480 | ||
63c372cb | 481 | devmqueue = strjoina(temporary_mount, "/dev/mqueue"); |
dc751688 | 482 | (void) mkdir(devmqueue, 0755); |
3164e3cb | 483 | (void) mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL); |
2b85f4e1 | 484 | |
63c372cb | 485 | devhugepages = strjoina(temporary_mount, "/dev/hugepages"); |
dc751688 | 486 | (void) mkdir(devhugepages, 0755); |
3164e3cb | 487 | (void) mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL); |
2b85f4e1 | 488 | |
63c372cb | 489 | devlog = strjoina(temporary_mount, "/dev/log"); |
3164e3cb | 490 | (void) symlink("/run/systemd/journal/dev-log", devlog); |
82d25240 | 491 | |
7f112f50 | 492 | NULSTR_FOREACH(d, devnodes) { |
2b85f4e1 LP |
493 | _cleanup_free_ char *dn = NULL; |
494 | struct stat st; | |
495 | ||
496 | r = stat(d, &st); | |
7f112f50 | 497 | if (r < 0) { |
2b85f4e1 LP |
498 | |
499 | if (errno == ENOENT) | |
500 | continue; | |
501 | ||
502 | r = -errno; | |
503 | goto fail; | |
7f112f50 LP |
504 | } |
505 | ||
2b85f4e1 LP |
506 | if (!S_ISBLK(st.st_mode) && |
507 | !S_ISCHR(st.st_mode)) { | |
508 | r = -EINVAL; | |
509 | goto fail; | |
510 | } | |
511 | ||
512 | if (st.st_rdev == 0) | |
513 | continue; | |
514 | ||
515 | dn = strappend(temporary_mount, d); | |
516 | if (!dn) { | |
517 | r = -ENOMEM; | |
518 | goto fail; | |
519 | } | |
520 | ||
ecabcf8b | 521 | mac_selinux_create_file_prepare(d, st.st_mode); |
2b85f4e1 | 522 | r = mknod(dn, st.st_mode, st.st_rdev); |
ecabcf8b | 523 | mac_selinux_create_file_clear(); |
dd078a1e | 524 | |
2b85f4e1 LP |
525 | if (r < 0) { |
526 | r = -errno; | |
527 | goto fail; | |
528 | } | |
7f112f50 LP |
529 | } |
530 | ||
03cfe0d5 | 531 | dev_setup(temporary_mount, UID_INVALID, GID_INVALID); |
7f112f50 | 532 | |
ee818b89 AC |
533 | /* Create the /dev directory if missing. It is more likely to be |
534 | * missing when the service is started with RootDirectory. This is | |
535 | * consistent with mount units creating the mount points when missing. | |
536 | */ | |
34de407a | 537 | (void) mkdir_p_label(mount_entry_path(m), 0755); |
ee818b89 | 538 | |
9e5f8252 | 539 | /* Unmount everything in old /dev */ |
34de407a LP |
540 | umount_recursive(mount_entry_path(m), 0); |
541 | if (mount(dev, mount_entry_path(m), NULL, MS_MOVE, NULL) < 0) { | |
2b85f4e1 LP |
542 | r = -errno; |
543 | goto fail; | |
544 | } | |
7f112f50 | 545 | |
2b85f4e1 LP |
546 | rmdir(dev); |
547 | rmdir(temporary_mount); | |
7f112f50 | 548 | |
2b85f4e1 | 549 | return 0; |
7f112f50 | 550 | |
2b85f4e1 LP |
551 | fail: |
552 | if (devpts) | |
553 | umount(devpts); | |
7f112f50 | 554 | |
2b85f4e1 LP |
555 | if (devshm) |
556 | umount(devshm); | |
7f112f50 | 557 | |
2b85f4e1 LP |
558 | if (devhugepages) |
559 | umount(devhugepages); | |
7f112f50 | 560 | |
2b85f4e1 LP |
561 | if (devmqueue) |
562 | umount(devmqueue); | |
7f112f50 | 563 | |
d267c5aa ZJS |
564 | umount(dev); |
565 | rmdir(dev); | |
2b85f4e1 | 566 | rmdir(temporary_mount); |
7f112f50 | 567 | |
2b85f4e1 | 568 | return r; |
7f112f50 LP |
569 | } |
570 | ||
ac0930c8 | 571 | static int apply_mount( |
34de407a | 572 | MountEntry *m, |
ac0930c8 | 573 | const char *tmp_dir, |
c17ec25e | 574 | const char *var_tmp_dir) { |
ac0930c8 | 575 | |
15ae422b | 576 | const char *what; |
15ae422b | 577 | int r; |
15ae422b | 578 | |
c17ec25e | 579 | assert(m); |
15ae422b | 580 | |
34de407a | 581 | log_debug("Applying namespace mount on %s", mount_entry_path(m)); |
fe3c2583 | 582 | |
c17ec25e | 583 | switch (m->mode) { |
15ae422b | 584 | |
160cfdbe LP |
585 | case INACCESSIBLE: { |
586 | struct stat target; | |
6d313367 LP |
587 | |
588 | /* First, get rid of everything that is below if there | |
589 | * is anything... Then, overmount it with an | |
c4b41707 | 590 | * inaccessible path. */ |
34de407a | 591 | (void) umount_recursive(mount_entry_path(m), 0); |
6d313367 | 592 | |
34de407a LP |
593 | if (lstat(mount_entry_path(m), &target) < 0) |
594 | return log_debug_errno(errno, "Failed to lstat() %s to determine what to mount over it: %m", mount_entry_path(m)); | |
15ae422b | 595 | |
c4b41707 | 596 | what = mode_to_inaccessible_node(target.st_mode); |
5fd7cf6f LP |
597 | if (!what) { |
598 | log_debug("File type not supported for inaccessible mounts. Note that symlinks are not allowed"); | |
c4b41707 AP |
599 | return -ELOOP; |
600 | } | |
601 | break; | |
160cfdbe | 602 | } |
fe3c2583 | 603 | |
15ae422b | 604 | case READONLY: |
15ae422b | 605 | case READWRITE: |
6b7c9f8b | 606 | |
34de407a | 607 | r = path_is_mount_point(mount_entry_path(m), NULL, 0); |
d944dc95 | 608 | if (r < 0) |
34de407a | 609 | return log_debug_errno(r, "Failed to determine whether %s is already a mount point: %m", mount_entry_path(m)); |
6b7c9f8b LP |
610 | if (r > 0) /* Nothing to do here, it is already a mount. We just later toggle the MS_RDONLY bit for the mount point if needed. */ |
611 | return 0; | |
6b7c9f8b | 612 | /* This isn't a mount point yet, let's make it one. */ |
34de407a | 613 | what = mount_entry_path(m); |
6b7c9f8b | 614 | break; |
15ae422b | 615 | |
ac0930c8 LP |
616 | case PRIVATE_TMP: |
617 | what = tmp_dir; | |
618 | break; | |
619 | ||
620 | case PRIVATE_VAR_TMP: | |
621 | what = var_tmp_dir; | |
15ae422b | 622 | break; |
e364ad06 | 623 | |
d6797c92 LP |
624 | case PRIVATE_DEV: |
625 | return mount_dev(m); | |
626 | ||
e364ad06 LP |
627 | default: |
628 | assert_not_reached("Unknown mode"); | |
15ae422b LP |
629 | } |
630 | ||
ac0930c8 | 631 | assert(what); |
15ae422b | 632 | |
34de407a LP |
633 | if (mount(what, mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0) |
634 | return log_debug_errno(errno, "Failed to mount %s to %s: %m", what, mount_entry_path(m)); | |
6b7c9f8b | 635 | |
34de407a | 636 | log_debug("Successfully mounted %s to %s", what, mount_entry_path(m)); |
6b7c9f8b | 637 | return 0; |
ac0930c8 | 638 | } |
15ae422b | 639 | |
34de407a | 640 | static int make_read_only(MountEntry *m, char **blacklist) { |
6b7c9f8b | 641 | int r = 0; |
15ae422b | 642 | |
c17ec25e | 643 | assert(m); |
ac0930c8 | 644 | |
34de407a LP |
645 | if (mount_entry_read_only(m)) |
646 | r = bind_remount_recursive(mount_entry_path(m), true, blacklist); | |
6b7c9f8b | 647 | else if (m->mode == PRIVATE_DEV) { /* Can be readonly but the submounts can't*/ |
34de407a | 648 | if (mount(NULL, mount_entry_path(m), NULL, MS_REMOUNT|DEV_MOUNT_OPTIONS|MS_RDONLY, NULL) < 0) |
6b7c9f8b | 649 | r = -errno; |
737ba3c8 | 650 | } else |
6b7c9f8b LP |
651 | return 0; |
652 | ||
653 | /* Not that we only turn on the MS_RDONLY flag here, we never turn it off. Something that was marked read-only | |
654 | * already stays this way. This improves compatibility with container managers, where we won't attempt to undo | |
655 | * read-only mounts already applied. */ | |
ac0930c8 | 656 | |
d6797c92 | 657 | return r; |
15ae422b LP |
658 | } |
659 | ||
1d54cd5d | 660 | /* Chase symlinks and remove failed paths from mounts */ |
34de407a LP |
661 | static int chase_all_symlinks(const char *root_directory, MountEntry *m, unsigned *n) { |
662 | MountEntry *f, *t; | |
1d54cd5d | 663 | int r = 0; |
d944dc95 LP |
664 | |
665 | assert(m); | |
666 | assert(n); | |
667 | ||
668 | /* Since mount() will always follow symlinks and we need to take the different root directory into account we | |
669 | * chase the symlinks on our own first. This call wil do so for all entries and remove all entries where we | |
670 | * can't resolve the path, and which have been marked for such removal. */ | |
671 | ||
49fedb40 | 672 | for (f = m, t = m; f < m + *n; f++) { |
f0a4feb0 | 673 | _cleanup_free_ char *chased = NULL; |
5327c910 | 674 | int k; |
49fedb40 | 675 | |
34de407a | 676 | k = chase_symlinks(mount_entry_path(f), root_directory, 0, &chased); |
1d54cd5d DH |
677 | if (k < 0) { |
678 | /* Get only real errors */ | |
679 | if (r >= 0 && (k != -ENOENT || !f->ignore)) | |
680 | r = k; | |
681 | ||
1d54cd5d | 682 | /* Doesn't exist or failed? Then remove it and continue! */ |
34de407a | 683 | log_debug_errno(k, "Failed to chase symlinks for %s: %m", mount_entry_path(f)); |
5327c910 | 684 | f->path_malloc = mfree(f->path_malloc); |
d944dc95 | 685 | continue; |
f0a4feb0 | 686 | } |
d944dc95 | 687 | |
34de407a LP |
688 | if (!path_equal(mount_entry_path(f), chased)) { |
689 | log_debug("Chased %s → %s", mount_entry_path(f), chased); | |
5327c910 LP |
690 | |
691 | free(f->path_malloc); | |
692 | f->path_malloc = chased; | |
693 | chased = NULL; | |
d944dc95 LP |
694 | } |
695 | ||
696 | *t = *f; | |
697 | t++; | |
698 | } | |
699 | ||
700 | *n = t - m; | |
1d54cd5d | 701 | return r; |
d944dc95 LP |
702 | } |
703 | ||
2652c6c1 | 704 | static unsigned namespace_calculate_mounts( |
c575770b | 705 | const NameSpaceInfo *ns_info, |
2652c6c1 DH |
706 | char** read_write_paths, |
707 | char** read_only_paths, | |
708 | char** inaccessible_paths, | |
709 | const char* tmp_dir, | |
710 | const char* var_tmp_dir, | |
2652c6c1 DH |
711 | ProtectHome protect_home, |
712 | ProtectSystem protect_system) { | |
713 | ||
b6c432ca | 714 | unsigned protect_home_cnt; |
f471b2af DH |
715 | unsigned protect_system_cnt = |
716 | (protect_system == PROTECT_SYSTEM_STRICT ? | |
717 | ELEMENTSOF(protect_system_strict_table) : | |
718 | ((protect_system == PROTECT_SYSTEM_FULL) ? | |
719 | ELEMENTSOF(protect_system_full_table) : | |
720 | ((protect_system == PROTECT_SYSTEM_YES) ? | |
721 | ELEMENTSOF(protect_system_yes_table) : 0))); | |
722 | ||
b6c432ca DH |
723 | protect_home_cnt = |
724 | (protect_home == PROTECT_HOME_YES ? | |
725 | ELEMENTSOF(protect_home_yes_table) : | |
726 | ((protect_home == PROTECT_HOME_READ_ONLY) ? | |
727 | ELEMENTSOF(protect_home_read_only_table) : 0)); | |
728 | ||
2652c6c1 DH |
729 | return !!tmp_dir + !!var_tmp_dir + |
730 | strv_length(read_write_paths) + | |
731 | strv_length(read_only_paths) + | |
732 | strv_length(inaccessible_paths) + | |
c575770b DH |
733 | ns_info->private_dev + |
734 | (ns_info->protect_kernel_tunables ? ELEMENTSOF(protect_kernel_tunables_table) : 0) + | |
735 | (ns_info->protect_control_groups ? 1 : 0) + | |
736 | (ns_info->protect_kernel_modules ? ELEMENTSOF(protect_kernel_modules_table) : 0) + | |
b6c432ca | 737 | protect_home_cnt + protect_system_cnt; |
2652c6c1 DH |
738 | } |
739 | ||
613b411c | 740 | int setup_namespace( |
ee818b89 | 741 | const char* root_directory, |
c575770b | 742 | const NameSpaceInfo *ns_info, |
2a624c36 AP |
743 | char** read_write_paths, |
744 | char** read_only_paths, | |
745 | char** inaccessible_paths, | |
a004cb4c LP |
746 | const char* tmp_dir, |
747 | const char* var_tmp_dir, | |
1b8689f9 LP |
748 | ProtectHome protect_home, |
749 | ProtectSystem protect_system, | |
e6547662 | 750 | unsigned long mount_flags) { |
15ae422b | 751 | |
34de407a | 752 | MountEntry *m, *mounts = NULL; |
d944dc95 | 753 | bool make_slave = false; |
f0a4feb0 | 754 | unsigned n_mounts; |
c17ec25e | 755 | int r = 0; |
15ae422b | 756 | |
613b411c | 757 | if (mount_flags == 0) |
c17ec25e | 758 | mount_flags = MS_SHARED; |
ac0930c8 | 759 | |
cfbeb4ef LP |
760 | n_mounts = namespace_calculate_mounts( |
761 | ns_info, | |
762 | read_write_paths, | |
763 | read_only_paths, | |
764 | inaccessible_paths, | |
765 | tmp_dir, var_tmp_dir, | |
766 | protect_home, protect_system); | |
613b411c | 767 | |
2652c6c1 | 768 | /* Set mount slave mode */ |
f0a4feb0 | 769 | if (root_directory || n_mounts > 0) |
d944dc95 LP |
770 | make_slave = true; |
771 | ||
f0a4feb0 | 772 | if (n_mounts > 0) { |
34de407a | 773 | m = mounts = (MountEntry *) alloca0(n_mounts * sizeof(MountEntry)); |
5327c910 | 774 | r = append_access_mounts(&m, read_write_paths, READWRITE); |
613b411c | 775 | if (r < 0) |
f0a4feb0 | 776 | goto finish; |
613b411c | 777 | |
5327c910 | 778 | r = append_access_mounts(&m, read_only_paths, READONLY); |
613b411c | 779 | if (r < 0) |
f0a4feb0 | 780 | goto finish; |
613b411c | 781 | |
5327c910 | 782 | r = append_access_mounts(&m, inaccessible_paths, INACCESSIBLE); |
613b411c | 783 | if (r < 0) |
f0a4feb0 | 784 | goto finish; |
7ff7394d | 785 | |
613b411c | 786 | if (tmp_dir) { |
34de407a | 787 | *(m++) = (MountEntry) { |
5327c910 LP |
788 | .path_const = "/tmp", |
789 | .mode = PRIVATE_TMP, | |
790 | }; | |
613b411c | 791 | } |
7ff7394d | 792 | |
613b411c | 793 | if (var_tmp_dir) { |
34de407a | 794 | *(m++) = (MountEntry) { |
5327c910 LP |
795 | .path_const = "/var/tmp", |
796 | .mode = PRIVATE_VAR_TMP, | |
797 | }; | |
7ff7394d | 798 | } |
ac0930c8 | 799 | |
c575770b | 800 | if (ns_info->private_dev) { |
34de407a | 801 | *(m++) = (MountEntry) { |
5327c910 LP |
802 | .path_const = "/dev", |
803 | .mode = PRIVATE_DEV, | |
804 | }; | |
7f112f50 LP |
805 | } |
806 | ||
c575770b | 807 | if (ns_info->protect_kernel_tunables) { |
5327c910 | 808 | r = append_static_mounts(&m, protect_kernel_tunables_table, ELEMENTSOF(protect_kernel_tunables_table), ns_info->ignore_protect_paths); |
c575770b | 809 | if (r < 0) |
f0a4feb0 | 810 | goto finish; |
c575770b DH |
811 | } |
812 | ||
813 | if (ns_info->protect_kernel_modules) { | |
5327c910 | 814 | r = append_static_mounts(&m, protect_kernel_modules_table, ELEMENTSOF(protect_kernel_modules_table), ns_info->ignore_protect_paths); |
c575770b | 815 | if (r < 0) |
f0a4feb0 | 816 | goto finish; |
c575770b | 817 | } |
59eeb84b | 818 | |
c575770b | 819 | if (ns_info->protect_control_groups) { |
34de407a | 820 | *(m++) = (MountEntry) { |
5327c910 LP |
821 | .path_const = "/sys/fs/cgroup", |
822 | .mode = READONLY, | |
823 | }; | |
59eeb84b LP |
824 | } |
825 | ||
5327c910 | 826 | r = append_protect_home(&m, protect_home, ns_info->ignore_protect_paths); |
b6c432ca | 827 | if (r < 0) |
f0a4feb0 | 828 | goto finish; |
417116f2 | 829 | |
5327c910 | 830 | r = append_protect_system(&m, protect_system, false); |
f471b2af | 831 | if (r < 0) |
f0a4feb0 | 832 | goto finish; |
417116f2 | 833 | |
f0a4feb0 | 834 | assert(mounts + n_mounts == m); |
ac0930c8 | 835 | |
5327c910 LP |
836 | /* Prepend the root directory where that's necessary */ |
837 | r = prefix_where_needed(mounts, n_mounts, root_directory); | |
838 | if (r < 0) | |
839 | goto finish; | |
840 | ||
d944dc95 LP |
841 | /* Resolve symlinks manually first, as mount() will always follow them relative to the host's |
842 | * root. Moreover we want to suppress duplicates based on the resolved paths. This of course is a bit | |
843 | * racy. */ | |
f0a4feb0 | 844 | r = chase_all_symlinks(root_directory, mounts, &n_mounts); |
d944dc95 LP |
845 | if (r < 0) |
846 | goto finish; | |
847 | ||
34de407a | 848 | qsort(mounts, n_mounts, sizeof(MountEntry), mount_path_compare); |
fe3c2583 | 849 | |
f0a4feb0 DH |
850 | drop_duplicates(mounts, &n_mounts); |
851 | drop_outside_root(root_directory, mounts, &n_mounts); | |
852 | drop_inaccessible(mounts, &n_mounts); | |
853 | drop_nop(mounts, &n_mounts); | |
15ae422b LP |
854 | } |
855 | ||
d944dc95 LP |
856 | if (unshare(CLONE_NEWNS) < 0) { |
857 | r = -errno; | |
858 | goto finish; | |
859 | } | |
1e4e94c8 | 860 | |
d944dc95 | 861 | if (make_slave) { |
c2c13f2d LP |
862 | /* Remount / as SLAVE so that nothing now mounted in the namespace |
863 | shows up in the parent */ | |
d944dc95 LP |
864 | if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) { |
865 | r = -errno; | |
866 | goto finish; | |
867 | } | |
ee818b89 AC |
868 | } |
869 | ||
870 | if (root_directory) { | |
8f1ad200 | 871 | /* Turn directory into bind mount, if it isn't one yet */ |
e1873695 | 872 | r = path_is_mount_point(root_directory, NULL, AT_SYMLINK_FOLLOW); |
8f1ad200 | 873 | if (r < 0) |
d944dc95 | 874 | goto finish; |
8f1ad200 LP |
875 | if (r == 0) { |
876 | if (mount(root_directory, root_directory, NULL, MS_BIND|MS_REC, NULL) < 0) { | |
877 | r = -errno; | |
878 | goto finish; | |
879 | } | |
d944dc95 | 880 | } |
ee818b89 | 881 | } |
c2c13f2d | 882 | |
f0a4feb0 | 883 | if (n_mounts > 0) { |
6b7c9f8b LP |
884 | char **blacklist; |
885 | unsigned j; | |
886 | ||
887 | /* First round, add in all special mounts we need */ | |
f0a4feb0 | 888 | for (m = mounts; m < mounts + n_mounts; ++m) { |
c2c13f2d LP |
889 | r = apply_mount(m, tmp_dir, var_tmp_dir); |
890 | if (r < 0) | |
d944dc95 | 891 | goto finish; |
c2c13f2d | 892 | } |
15ae422b | 893 | |
6b7c9f8b | 894 | /* Create a blacklist we can pass to bind_mount_recursive() */ |
f0a4feb0 DH |
895 | blacklist = newa(char*, n_mounts+1); |
896 | for (j = 0; j < n_mounts; j++) | |
34de407a | 897 | blacklist[j] = (char*) mount_entry_path(mounts+j); |
6b7c9f8b LP |
898 | blacklist[j] = NULL; |
899 | ||
900 | /* Second round, flip the ro bits if necessary. */ | |
f0a4feb0 | 901 | for (m = mounts; m < mounts + n_mounts; ++m) { |
6b7c9f8b | 902 | r = make_read_only(m, blacklist); |
c2c13f2d | 903 | if (r < 0) |
d944dc95 | 904 | goto finish; |
c2c13f2d | 905 | } |
15ae422b LP |
906 | } |
907 | ||
ee818b89 AC |
908 | if (root_directory) { |
909 | /* MS_MOVE does not work on MS_SHARED so the remount MS_SHARED will be done later */ | |
910 | r = mount_move_root(root_directory); | |
d944dc95 LP |
911 | if (r < 0) |
912 | goto finish; | |
ee818b89 AC |
913 | } |
914 | ||
c2c13f2d LP |
915 | /* Remount / as the desired mode. Not that this will not |
916 | * reestablish propagation from our side to the host, since | |
917 | * what's disconnected is disconnected. */ | |
d944dc95 LP |
918 | if (mount(NULL, "/", NULL, mount_flags | MS_REC, NULL) < 0) { |
919 | r = -errno; | |
920 | goto finish; | |
921 | } | |
15ae422b | 922 | |
d944dc95 | 923 | r = 0; |
15ae422b | 924 | |
d944dc95 | 925 | finish: |
f0a4feb0 | 926 | for (m = mounts; m < mounts + n_mounts; m++) |
5327c910 | 927 | free(m->path_malloc); |
613b411c LP |
928 | |
929 | return r; | |
930 | } | |
931 | ||
932 | static int setup_one_tmp_dir(const char *id, const char *prefix, char **path) { | |
933 | _cleanup_free_ char *x = NULL; | |
6b46ea73 LP |
934 | char bid[SD_ID128_STRING_MAX]; |
935 | sd_id128_t boot_id; | |
936 | int r; | |
613b411c LP |
937 | |
938 | assert(id); | |
939 | assert(prefix); | |
940 | assert(path); | |
941 | ||
6b46ea73 LP |
942 | /* We include the boot id in the directory so that after a |
943 | * reboot we can easily identify obsolete directories. */ | |
944 | ||
945 | r = sd_id128_get_boot(&boot_id); | |
946 | if (r < 0) | |
947 | return r; | |
948 | ||
605405c6 | 949 | x = strjoin(prefix, "/systemd-private-", sd_id128_to_string(boot_id, bid), "-", id, "-XXXXXX"); |
613b411c LP |
950 | if (!x) |
951 | return -ENOMEM; | |
952 | ||
953 | RUN_WITH_UMASK(0077) | |
954 | if (!mkdtemp(x)) | |
955 | return -errno; | |
956 | ||
957 | RUN_WITH_UMASK(0000) { | |
958 | char *y; | |
959 | ||
63c372cb | 960 | y = strjoina(x, "/tmp"); |
613b411c LP |
961 | |
962 | if (mkdir(y, 0777 | S_ISVTX) < 0) | |
963 | return -errno; | |
c17ec25e | 964 | } |
15ae422b | 965 | |
613b411c LP |
966 | *path = x; |
967 | x = NULL; | |
968 | ||
969 | return 0; | |
970 | } | |
971 | ||
972 | int setup_tmp_dirs(const char *id, char **tmp_dir, char **var_tmp_dir) { | |
973 | char *a, *b; | |
974 | int r; | |
975 | ||
976 | assert(id); | |
977 | assert(tmp_dir); | |
978 | assert(var_tmp_dir); | |
979 | ||
980 | r = setup_one_tmp_dir(id, "/tmp", &a); | |
981 | if (r < 0) | |
982 | return r; | |
983 | ||
984 | r = setup_one_tmp_dir(id, "/var/tmp", &b); | |
985 | if (r < 0) { | |
986 | char *t; | |
987 | ||
63c372cb | 988 | t = strjoina(a, "/tmp"); |
613b411c LP |
989 | rmdir(t); |
990 | rmdir(a); | |
991 | ||
992 | free(a); | |
993 | return r; | |
994 | } | |
995 | ||
996 | *tmp_dir = a; | |
997 | *var_tmp_dir = b; | |
998 | ||
999 | return 0; | |
1000 | } | |
1001 | ||
1002 | int setup_netns(int netns_storage_socket[2]) { | |
1003 | _cleanup_close_ int netns = -1; | |
3ee897d6 | 1004 | int r, q; |
613b411c LP |
1005 | |
1006 | assert(netns_storage_socket); | |
1007 | assert(netns_storage_socket[0] >= 0); | |
1008 | assert(netns_storage_socket[1] >= 0); | |
1009 | ||
1010 | /* We use the passed socketpair as a storage buffer for our | |
76cd584b LP |
1011 | * namespace reference fd. Whatever process runs this first |
1012 | * shall create a new namespace, all others should just join | |
1013 | * it. To serialize that we use a file lock on the socket | |
1014 | * pair. | |
613b411c LP |
1015 | * |
1016 | * It's a bit crazy, but hey, works great! */ | |
1017 | ||
1018 | if (lockf(netns_storage_socket[0], F_LOCK, 0) < 0) | |
1019 | return -errno; | |
1020 | ||
3ee897d6 LP |
1021 | netns = receive_one_fd(netns_storage_socket[0], MSG_DONTWAIT); |
1022 | if (netns == -EAGAIN) { | |
613b411c LP |
1023 | /* Nothing stored yet, so let's create a new namespace */ |
1024 | ||
1025 | if (unshare(CLONE_NEWNET) < 0) { | |
1026 | r = -errno; | |
1027 | goto fail; | |
1028 | } | |
1029 | ||
1030 | loopback_setup(); | |
1031 | ||
1032 | netns = open("/proc/self/ns/net", O_RDONLY|O_CLOEXEC|O_NOCTTY); | |
1033 | if (netns < 0) { | |
1034 | r = -errno; | |
1035 | goto fail; | |
1036 | } | |
1037 | ||
1038 | r = 1; | |
613b411c | 1039 | |
3ee897d6 LP |
1040 | } else if (netns < 0) { |
1041 | r = netns; | |
1042 | goto fail; | |
613b411c | 1043 | |
3ee897d6 LP |
1044 | } else { |
1045 | /* Yay, found something, so let's join the namespace */ | |
613b411c LP |
1046 | if (setns(netns, CLONE_NEWNET) < 0) { |
1047 | r = -errno; | |
1048 | goto fail; | |
1049 | } | |
1050 | ||
1051 | r = 0; | |
1052 | } | |
1053 | ||
3ee897d6 LP |
1054 | q = send_one_fd(netns_storage_socket[1], netns, MSG_DONTWAIT); |
1055 | if (q < 0) { | |
1056 | r = q; | |
613b411c LP |
1057 | goto fail; |
1058 | } | |
1059 | ||
1060 | fail: | |
fe048ce5 | 1061 | (void) lockf(netns_storage_socket[0], F_ULOCK, 0); |
15ae422b LP |
1062 | return r; |
1063 | } | |
417116f2 | 1064 | |
1b8689f9 LP |
1065 | static const char *const protect_home_table[_PROTECT_HOME_MAX] = { |
1066 | [PROTECT_HOME_NO] = "no", | |
1067 | [PROTECT_HOME_YES] = "yes", | |
1068 | [PROTECT_HOME_READ_ONLY] = "read-only", | |
417116f2 LP |
1069 | }; |
1070 | ||
1b8689f9 LP |
1071 | DEFINE_STRING_TABLE_LOOKUP(protect_home, ProtectHome); |
1072 | ||
1073 | static const char *const protect_system_table[_PROTECT_SYSTEM_MAX] = { | |
1074 | [PROTECT_SYSTEM_NO] = "no", | |
1075 | [PROTECT_SYSTEM_YES] = "yes", | |
1076 | [PROTECT_SYSTEM_FULL] = "full", | |
3f815163 | 1077 | [PROTECT_SYSTEM_STRICT] = "strict", |
1b8689f9 LP |
1078 | }; |
1079 | ||
1080 | DEFINE_STRING_TABLE_LOOKUP(protect_system, ProtectSystem); |