]>
Commit | Line | Data |
---|---|---|
53e1b683 | 1 | /* SPDX-License-Identifier: LGPL-2.1+ */ |
15ae422b LP |
2 | |
3 | #include <errno.h> | |
e08f94ac | 4 | #include <linux/loop.h> |
07630cea | 5 | #include <sched.h> |
15ae422b | 6 | #include <stdio.h> |
07630cea | 7 | #include <sys/mount.h> |
07630cea | 8 | #include <unistd.h> |
25e870b5 | 9 | #include <linux/fs.h> |
15ae422b | 10 | |
b5efdb8a | 11 | #include "alloc-util.h" |
10404d52 | 12 | #include "base-filesystem.h" |
7f112f50 | 13 | #include "dev-setup.h" |
3ffd4af2 | 14 | #include "fd-util.h" |
d944dc95 | 15 | #include "fs-util.h" |
e908468b | 16 | #include "label.h" |
915e6d16 | 17 | #include "loop-util.h" |
07630cea | 18 | #include "loopback-setup.h" |
07630cea | 19 | #include "mkdir.h" |
4349cd7c | 20 | #include "mount-util.h" |
049af8ad | 21 | #include "mountpoint-util.h" |
0cb8e3d1 | 22 | #include "namespace-util.h" |
3ffd4af2 | 23 | #include "namespace.h" |
d8b4d14d | 24 | #include "nulstr-util.h" |
07630cea | 25 | #include "path-util.h" |
d7b8eec7 | 26 | #include "selinux-util.h" |
2583fbea | 27 | #include "socket-util.h" |
760877e9 | 28 | #include "sort-util.h" |
36ce7110 | 29 | #include "stat-util.h" |
8b43440b | 30 | #include "string-table.h" |
07630cea LP |
31 | #include "string-util.h" |
32 | #include "strv.h" | |
a652f050 | 33 | #include "tmpfile-util.h" |
affb60b1 | 34 | #include "umask-util.h" |
ee104e11 | 35 | #include "user-util.h" |
15ae422b | 36 | |
737ba3c8 | 37 | #define DEV_MOUNT_OPTIONS (MS_NOSUID|MS_STRICTATIME|MS_NOEXEC) |
38 | ||
c17ec25e | 39 | typedef enum MountMode { |
15ae422b LP |
40 | /* This is ordered by priority! */ |
41 | INACCESSIBLE, | |
d2d6c096 LP |
42 | BIND_MOUNT, |
43 | BIND_MOUNT_RECURSIVE, | |
ac0930c8 | 44 | PRIVATE_TMP, |
7f112f50 | 45 | PRIVATE_DEV, |
5d997827 | 46 | BIND_DEV, |
6c47cd7d | 47 | EMPTY_DIR, |
5d997827 LP |
48 | SYSFS, |
49 | PROCFS, | |
50 | READONLY, | |
59eeb84b | 51 | READWRITE, |
2abd4e38 | 52 | TMPFS, |
1e05071d | 53 | READWRITE_IMPLICIT, /* Should have the lowest priority. */ |
5beb8688 | 54 | _MOUNT_MODE_MAX, |
c17ec25e | 55 | } MountMode; |
15ae422b | 56 | |
34de407a | 57 | typedef struct MountEntry { |
5327c910 | 58 | const char *path_const; /* Memory allocated on stack or static */ |
cfbeb4ef | 59 | MountMode mode:5; |
5327c910 LP |
60 | bool ignore:1; /* Ignore if path does not exist? */ |
61 | bool has_prefix:1; /* Already is prefixed by the root dir? */ | |
cfbeb4ef | 62 | bool read_only:1; /* Shall this mount point be read-only? */ |
9ce4e4b0 | 63 | bool nosuid:1; /* Shall set MS_NOSUID on the mount itself */ |
088696fe | 64 | bool applied:1; /* Already applied */ |
55fe7432 | 65 | char *path_malloc; /* Use this instead of 'path_const' if we had to allocate memory */ |
d2d6c096 LP |
66 | const char *source_const; /* The source path, for bind mounts */ |
67 | char *source_malloc; | |
2abd4e38 YW |
68 | const char *options_const;/* Mount options for tmpfs */ |
69 | char *options_malloc; | |
70 | unsigned long flags; /* Mount flags used by EMPTY_DIR and TMPFS. Do not include MS_RDONLY here, but please use read_only. */ | |
088696fe | 71 | unsigned n_followed; |
34de407a | 72 | } MountEntry; |
15ae422b | 73 | |
5d997827 | 74 | /* If MountAPIVFS= is used, let's mount /sys and /proc into the it, but only as a fallback if the user hasn't mounted |
3fe91079 | 75 | * something there already. These mounts are hence overridden by any other explicitly configured mounts. */ |
5d997827 LP |
76 | static const MountEntry apivfs_table[] = { |
77 | { "/proc", PROCFS, false }, | |
78 | { "/dev", BIND_DEV, false }, | |
79 | { "/sys", SYSFS, false }, | |
80 | }; | |
f471b2af | 81 | |
11a30cec | 82 | /* ProtectKernelTunables= option and the related filesystem APIs */ |
34de407a | 83 | static const MountEntry protect_kernel_tunables_table[] = { |
1e05071d YW |
84 | { "/proc/acpi", READONLY, true }, |
85 | { "/proc/apm", READONLY, true }, /* Obsolete API, there's no point in permitting access to this, ever */ | |
86 | { "/proc/asound", READONLY, true }, | |
87 | { "/proc/bus", READONLY, true }, | |
88 | { "/proc/fs", READONLY, true }, | |
89 | { "/proc/irq", READONLY, true }, | |
90 | { "/proc/kallsyms", INACCESSIBLE, true }, | |
91 | { "/proc/kcore", INACCESSIBLE, true }, | |
92 | { "/proc/latency_stats", READONLY, true }, | |
93 | { "/proc/mtrr", READONLY, true }, | |
94 | { "/proc/scsi", READONLY, true }, | |
95 | { "/proc/sys", READONLY, false }, | |
96 | { "/proc/sysrq-trigger", READONLY, true }, | |
97 | { "/proc/timer_stats", READONLY, true }, | |
98 | { "/sys", READONLY, false }, | |
99 | { "/sys/fs/bpf", READONLY, true }, | |
100 | { "/sys/fs/cgroup", READWRITE_IMPLICIT, false }, /* READONLY is set by ProtectControlGroups= option */ | |
101 | { "/sys/fs/selinux", READWRITE_IMPLICIT, true }, | |
102 | { "/sys/kernel/debug", READONLY, true }, | |
103 | { "/sys/kernel/tracing", READONLY, true }, | |
11a30cec DH |
104 | }; |
105 | ||
c575770b | 106 | /* ProtectKernelModules= option */ |
34de407a | 107 | static const MountEntry protect_kernel_modules_table[] = { |
349cc4a5 | 108 | #if HAVE_SPLIT_USR |
c6232fb0 | 109 | { "/lib/modules", INACCESSIBLE, true }, |
c575770b | 110 | #endif |
c6232fb0 | 111 | { "/usr/lib/modules", INACCESSIBLE, true }, |
c575770b DH |
112 | }; |
113 | ||
94a7b275 KK |
114 | /* ProtectKernelLogs= option */ |
115 | static const MountEntry protect_kernel_logs_table[] = { | |
116 | { "/proc/kmsg", INACCESSIBLE, true }, | |
117 | { "/dev/kmsg", INACCESSIBLE, true }, | |
118 | }; | |
119 | ||
b6c432ca DH |
120 | /* |
121 | * ProtectHome=read-only table, protect $HOME and $XDG_RUNTIME_DIR and rest of | |
122 | * system should be protected by ProtectSystem= | |
123 | */ | |
34de407a | 124 | static const MountEntry protect_home_read_only_table[] = { |
c6232fb0 LP |
125 | { "/home", READONLY, true }, |
126 | { "/run/user", READONLY, true }, | |
127 | { "/root", READONLY, true }, | |
b6c432ca DH |
128 | }; |
129 | ||
e4da7d8c YW |
130 | /* ProtectHome=tmpfs table */ |
131 | static const MountEntry protect_home_tmpfs_table[] = { | |
132 | { "/home", TMPFS, true, .read_only = true, .options_const = "mode=0755", .flags = MS_NODEV|MS_STRICTATIME }, | |
133 | { "/run/user", TMPFS, true, .read_only = true, .options_const = "mode=0755", .flags = MS_NODEV|MS_STRICTATIME }, | |
134 | { "/root", TMPFS, true, .read_only = true, .options_const = "mode=0700", .flags = MS_NODEV|MS_STRICTATIME }, | |
135 | }; | |
136 | ||
b6c432ca | 137 | /* ProtectHome=yes table */ |
34de407a | 138 | static const MountEntry protect_home_yes_table[] = { |
c6232fb0 LP |
139 | { "/home", INACCESSIBLE, true }, |
140 | { "/run/user", INACCESSIBLE, true }, | |
141 | { "/root", INACCESSIBLE, true }, | |
b6c432ca DH |
142 | }; |
143 | ||
f471b2af | 144 | /* ProtectSystem=yes table */ |
34de407a | 145 | static const MountEntry protect_system_yes_table[] = { |
c6232fb0 LP |
146 | { "/usr", READONLY, false }, |
147 | { "/boot", READONLY, true }, | |
148 | { "/efi", READONLY, true }, | |
7486f305 AB |
149 | #if HAVE_SPLIT_USR |
150 | { "/lib", READONLY, true }, | |
151 | { "/lib64", READONLY, true }, | |
152 | { "/bin", READONLY, true }, | |
671f0f8d | 153 | # if HAVE_SPLIT_BIN |
7486f305 | 154 | { "/sbin", READONLY, true }, |
671f0f8d | 155 | # endif |
7486f305 | 156 | #endif |
f471b2af DH |
157 | }; |
158 | ||
159 | /* ProtectSystem=full includes ProtectSystem=yes */ | |
34de407a | 160 | static const MountEntry protect_system_full_table[] = { |
c6232fb0 LP |
161 | { "/usr", READONLY, false }, |
162 | { "/boot", READONLY, true }, | |
163 | { "/efi", READONLY, true }, | |
164 | { "/etc", READONLY, false }, | |
7486f305 AB |
165 | #if HAVE_SPLIT_USR |
166 | { "/lib", READONLY, true }, | |
167 | { "/lib64", READONLY, true }, | |
168 | { "/bin", READONLY, true }, | |
671f0f8d | 169 | # if HAVE_SPLIT_BIN |
7486f305 | 170 | { "/sbin", READONLY, true }, |
671f0f8d | 171 | # endif |
7486f305 | 172 | #endif |
f471b2af DH |
173 | }; |
174 | ||
175 | /* | |
176 | * ProtectSystem=strict table. In this strict mode, we mount everything | |
177 | * read-only, except for /proc, /dev, /sys which are the kernel API VFS, | |
178 | * which are left writable, but PrivateDevices= + ProtectKernelTunables= | |
179 | * protect those, and these options should be fully orthogonal. | |
180 | * (And of course /home and friends are also left writable, as ProtectHome= | |
181 | * shall manage those, orthogonally). | |
182 | */ | |
34de407a | 183 | static const MountEntry protect_system_strict_table[] = { |
1e05071d YW |
184 | { "/", READONLY, false }, |
185 | { "/proc", READWRITE_IMPLICIT, false }, /* ProtectKernelTunables= */ | |
186 | { "/sys", READWRITE_IMPLICIT, false }, /* ProtectKernelTunables= */ | |
187 | { "/dev", READWRITE_IMPLICIT, false }, /* PrivateDevices= */ | |
188 | { "/home", READWRITE_IMPLICIT, true }, /* ProtectHome= */ | |
189 | { "/run/user", READWRITE_IMPLICIT, true }, /* ProtectHome= */ | |
190 | { "/root", READWRITE_IMPLICIT, true }, /* ProtectHome= */ | |
f471b2af DH |
191 | }; |
192 | ||
5beb8688 YW |
193 | static const char * const mount_mode_table[_MOUNT_MODE_MAX] = { |
194 | [INACCESSIBLE] = "inaccessible", | |
195 | [BIND_MOUNT] = "bind", | |
196 | [BIND_MOUNT_RECURSIVE] = "rbind", | |
197 | [PRIVATE_TMP] = "private-tmp", | |
198 | [PRIVATE_DEV] = "private-dev", | |
199 | [BIND_DEV] = "bind-dev", | |
200 | [EMPTY_DIR] = "empty", | |
201 | [SYSFS] = "sysfs", | |
202 | [PROCFS] = "procfs", | |
203 | [READONLY] = "read-only", | |
204 | [READWRITE] = "read-write", | |
205 | [TMPFS] = "tmpfs", | |
206 | [READWRITE_IMPLICIT] = "rw-implicit", | |
207 | }; | |
208 | ||
209 | DEFINE_PRIVATE_STRING_TABLE_LOOKUP_TO_STRING(mount_mode, MountMode); | |
210 | ||
34de407a | 211 | static const char *mount_entry_path(const MountEntry *p) { |
f0a4feb0 DH |
212 | assert(p); |
213 | ||
5327c910 LP |
214 | /* Returns the path of this bind mount. If the malloc()-allocated ->path_buffer field is set we return that, |
215 | * otherwise the stack/static ->path field is returned. */ | |
f0a4feb0 | 216 | |
5327c910 | 217 | return p->path_malloc ?: p->path_const; |
f0a4feb0 DH |
218 | } |
219 | ||
34de407a | 220 | static bool mount_entry_read_only(const MountEntry *p) { |
cfbeb4ef LP |
221 | assert(p); |
222 | ||
223 | return p->read_only || IN_SET(p->mode, READONLY, INACCESSIBLE); | |
224 | } | |
225 | ||
d2d6c096 LP |
226 | static const char *mount_entry_source(const MountEntry *p) { |
227 | assert(p); | |
228 | ||
229 | return p->source_malloc ?: p->source_const; | |
230 | } | |
231 | ||
2abd4e38 YW |
232 | static const char *mount_entry_options(const MountEntry *p) { |
233 | assert(p); | |
234 | ||
235 | return p->options_malloc ?: p->options_const; | |
236 | } | |
237 | ||
1eb7e08e LP |
238 | static void mount_entry_done(MountEntry *p) { |
239 | assert(p); | |
240 | ||
241 | p->path_malloc = mfree(p->path_malloc); | |
242 | p->source_malloc = mfree(p->source_malloc); | |
2abd4e38 | 243 | p->options_malloc = mfree(p->options_malloc); |
1eb7e08e LP |
244 | } |
245 | ||
d18aff04 | 246 | static int append_access_mounts(MountEntry **p, char **strv, MountMode mode, bool forcibly_require_prefix) { |
15ae422b LP |
247 | char **i; |
248 | ||
613b411c LP |
249 | assert(p); |
250 | ||
1e05071d | 251 | /* Adds a list of user-supplied READWRITE/READWRITE_IMPLICIT/READONLY/INACCESSIBLE entries */ |
5327c910 | 252 | |
15ae422b | 253 | STRV_FOREACH(i, strv) { |
5327c910 LP |
254 | bool ignore = false, needs_prefix = false; |
255 | const char *e = *i; | |
15ae422b | 256 | |
5327c910 LP |
257 | /* Look for any prefixes */ |
258 | if (startswith(e, "-")) { | |
259 | e++; | |
9c94d52e | 260 | ignore = true; |
ea92ae33 | 261 | } |
5327c910 LP |
262 | if (startswith(e, "+")) { |
263 | e++; | |
264 | needs_prefix = true; | |
265 | } | |
ea92ae33 | 266 | |
baaa35ad ZJS |
267 | if (!path_is_absolute(e)) |
268 | return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), | |
269 | "Path is not absolute: %s", e); | |
15ae422b | 270 | |
34de407a | 271 | *((*p)++) = (MountEntry) { |
5327c910 LP |
272 | .path_const = e, |
273 | .mode = mode, | |
274 | .ignore = ignore, | |
d18aff04 | 275 | .has_prefix = !needs_prefix && !forcibly_require_prefix, |
5327c910 | 276 | }; |
15ae422b LP |
277 | } |
278 | ||
279 | return 0; | |
280 | } | |
281 | ||
6c47cd7d LP |
282 | static int append_empty_dir_mounts(MountEntry **p, char **strv) { |
283 | char **i; | |
284 | ||
285 | assert(p); | |
286 | ||
287 | /* Adds tmpfs mounts to provide readable but empty directories. This is primarily used to implement the | |
288 | * "/private/" boundary directories for DynamicUser=1. */ | |
289 | ||
290 | STRV_FOREACH(i, strv) { | |
291 | ||
292 | *((*p)++) = (MountEntry) { | |
293 | .path_const = *i, | |
294 | .mode = EMPTY_DIR, | |
295 | .ignore = false, | |
6c47cd7d | 296 | .read_only = true, |
2abd4e38 YW |
297 | .options_const = "mode=755", |
298 | .flags = MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, | |
6c47cd7d LP |
299 | }; |
300 | } | |
301 | ||
302 | return 0; | |
303 | } | |
304 | ||
da6053d0 LP |
305 | static int append_bind_mounts(MountEntry **p, const BindMount *binds, size_t n) { |
306 | size_t i; | |
d2d6c096 LP |
307 | |
308 | assert(p); | |
309 | ||
310 | for (i = 0; i < n; i++) { | |
311 | const BindMount *b = binds + i; | |
312 | ||
313 | *((*p)++) = (MountEntry) { | |
314 | .path_const = b->destination, | |
315 | .mode = b->recursive ? BIND_MOUNT_RECURSIVE : BIND_MOUNT, | |
316 | .read_only = b->read_only, | |
9ce4e4b0 | 317 | .nosuid = b->nosuid, |
d2d6c096 | 318 | .source_const = b->source, |
4ca763a9 | 319 | .ignore = b->ignore_enoent, |
d2d6c096 LP |
320 | }; |
321 | } | |
322 | ||
323 | return 0; | |
324 | } | |
325 | ||
da6053d0 LP |
326 | static int append_tmpfs_mounts(MountEntry **p, const TemporaryFileSystem *tmpfs, size_t n) { |
327 | size_t i; | |
2abd4e38 YW |
328 | int r; |
329 | ||
330 | assert(p); | |
331 | ||
332 | for (i = 0; i < n; i++) { | |
333 | const TemporaryFileSystem *t = tmpfs + i; | |
334 | _cleanup_free_ char *o = NULL, *str = NULL; | |
ad8e66dc | 335 | unsigned long flags; |
2abd4e38 YW |
336 | bool ro = false; |
337 | ||
baaa35ad ZJS |
338 | if (!path_is_absolute(t->path)) |
339 | return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), | |
340 | "Path is not absolute: %s", | |
341 | t->path); | |
2abd4e38 | 342 | |
ad8e66dc AJ |
343 | str = strjoin("mode=0755,", t->options); |
344 | if (!str) | |
345 | return -ENOMEM; | |
2abd4e38 | 346 | |
ad8e66dc AJ |
347 | r = mount_option_mangle(str, MS_NODEV|MS_STRICTATIME, &flags, &o); |
348 | if (r < 0) | |
349 | return log_debug_errno(r, "Failed to parse mount option '%s': %m", str); | |
2abd4e38 | 350 | |
ad8e66dc AJ |
351 | ro = flags & MS_RDONLY; |
352 | if (ro) | |
353 | flags ^= MS_RDONLY; | |
2abd4e38 YW |
354 | |
355 | *((*p)++) = (MountEntry) { | |
356 | .path_const = t->path, | |
357 | .mode = TMPFS, | |
358 | .read_only = ro, | |
ad8e66dc | 359 | .options_malloc = TAKE_PTR(o), |
2abd4e38 YW |
360 | .flags = flags, |
361 | }; | |
2abd4e38 YW |
362 | } |
363 | ||
364 | return 0; | |
365 | } | |
366 | ||
da6053d0 LP |
367 | static int append_static_mounts(MountEntry **p, const MountEntry *mounts, size_t n, bool ignore_protect) { |
368 | size_t i; | |
11a30cec DH |
369 | |
370 | assert(p); | |
f471b2af | 371 | assert(mounts); |
11a30cec | 372 | |
5327c910 | 373 | /* Adds a list of static pre-defined entries */ |
f471b2af | 374 | |
5327c910 | 375 | for (i = 0; i < n; i++) |
34de407a LP |
376 | *((*p)++) = (MountEntry) { |
377 | .path_const = mount_entry_path(mounts+i), | |
5327c910 LP |
378 | .mode = mounts[i].mode, |
379 | .ignore = mounts[i].ignore || ignore_protect, | |
380 | }; | |
f471b2af DH |
381 | |
382 | return 0; | |
383 | } | |
384 | ||
34de407a | 385 | static int append_protect_home(MountEntry **p, ProtectHome protect_home, bool ignore_protect) { |
c575770b DH |
386 | assert(p); |
387 | ||
5327c910 | 388 | switch (protect_home) { |
b6c432ca | 389 | |
5327c910 | 390 | case PROTECT_HOME_NO: |
b6c432ca DH |
391 | return 0; |
392 | ||
b6c432ca | 393 | case PROTECT_HOME_READ_ONLY: |
5327c910 LP |
394 | return append_static_mounts(p, protect_home_read_only_table, ELEMENTSOF(protect_home_read_only_table), ignore_protect); |
395 | ||
e4da7d8c YW |
396 | case PROTECT_HOME_TMPFS: |
397 | return append_static_mounts(p, protect_home_tmpfs_table, ELEMENTSOF(protect_home_tmpfs_table), ignore_protect); | |
398 | ||
b6c432ca | 399 | case PROTECT_HOME_YES: |
5327c910 LP |
400 | return append_static_mounts(p, protect_home_yes_table, ELEMENTSOF(protect_home_yes_table), ignore_protect); |
401 | ||
b6c432ca | 402 | default: |
5327c910 | 403 | assert_not_reached("Unexpected ProtectHome= value"); |
b6c432ca | 404 | } |
b6c432ca DH |
405 | } |
406 | ||
34de407a | 407 | static int append_protect_system(MountEntry **p, ProtectSystem protect_system, bool ignore_protect) { |
f471b2af DH |
408 | assert(p); |
409 | ||
5327c910 LP |
410 | switch (protect_system) { |
411 | ||
412 | case PROTECT_SYSTEM_NO: | |
f471b2af DH |
413 | return 0; |
414 | ||
f471b2af | 415 | case PROTECT_SYSTEM_STRICT: |
5327c910 LP |
416 | return append_static_mounts(p, protect_system_strict_table, ELEMENTSOF(protect_system_strict_table), ignore_protect); |
417 | ||
f471b2af | 418 | case PROTECT_SYSTEM_YES: |
5327c910 LP |
419 | return append_static_mounts(p, protect_system_yes_table, ELEMENTSOF(protect_system_yes_table), ignore_protect); |
420 | ||
f471b2af | 421 | case PROTECT_SYSTEM_FULL: |
5327c910 LP |
422 | return append_static_mounts(p, protect_system_full_table, ELEMENTSOF(protect_system_full_table), ignore_protect); |
423 | ||
f471b2af | 424 | default: |
5327c910 | 425 | assert_not_reached("Unexpected ProtectSystem= value"); |
f471b2af | 426 | } |
11a30cec DH |
427 | } |
428 | ||
93bab288 | 429 | static int mount_path_compare(const MountEntry *a, const MountEntry *b) { |
a0827e2b | 430 | int d; |
15ae422b | 431 | |
6ee1a919 | 432 | /* If the paths are not equal, then order prefixes first */ |
93bab288 | 433 | d = path_compare(mount_entry_path(a), mount_entry_path(b)); |
6ee1a919 LP |
434 | if (d != 0) |
435 | return d; | |
15ae422b | 436 | |
6ee1a919 | 437 | /* If the paths are equal, check the mode */ |
93bab288 | 438 | return CMP((int) a->mode, (int) b->mode); |
15ae422b LP |
439 | } |
440 | ||
da6053d0 LP |
441 | static int prefix_where_needed(MountEntry *m, size_t n, const char *root_directory) { |
442 | size_t i; | |
5327c910 | 443 | |
4a756839 | 444 | /* Prefixes all paths in the bind mount table with the root directory if the entry needs that. */ |
5327c910 LP |
445 | |
446 | for (i = 0; i < n; i++) { | |
447 | char *s; | |
448 | ||
449 | if (m[i].has_prefix) | |
450 | continue; | |
451 | ||
c6134d3e | 452 | s = path_join(root_directory, mount_entry_path(m+i)); |
5327c910 LP |
453 | if (!s) |
454 | return -ENOMEM; | |
455 | ||
e282f51f | 456 | free_and_replace(m[i].path_malloc, s); |
5327c910 LP |
457 | m[i].has_prefix = true; |
458 | } | |
459 | ||
460 | return 0; | |
461 | } | |
462 | ||
da6053d0 | 463 | static void drop_duplicates(MountEntry *m, size_t *n) { |
34de407a | 464 | MountEntry *f, *t, *previous; |
15ae422b | 465 | |
c17ec25e | 466 | assert(m); |
15ae422b | 467 | assert(n); |
15ae422b | 468 | |
fe3c2583 LP |
469 | /* Drops duplicate entries. Expects that the array is properly ordered already. */ |
470 | ||
1d54cd5d | 471 | for (f = m, t = m, previous = NULL; f < m + *n; f++) { |
15ae422b | 472 | |
fe3c2583 | 473 | /* The first one wins (which is the one with the more restrictive mode), see mount_path_compare() |
088696fe LP |
474 | * above. Note that we only drop duplicates that haven't been mounted yet. */ |
475 | if (previous && | |
476 | path_equal(mount_entry_path(f), mount_entry_path(previous)) && | |
477 | !f->applied && !previous->applied) { | |
5beb8688 | 478 | log_debug("%s (%s) is duplicate.", mount_entry_path(f), mount_mode_to_string(f->mode)); |
34de407a | 479 | previous->read_only = previous->read_only || mount_entry_read_only(f); /* Propagate the read-only flag to the remaining entry */ |
1eb7e08e | 480 | mount_entry_done(f); |
15ae422b | 481 | continue; |
fe3c2583 | 482 | } |
15ae422b | 483 | |
e2d7c1a0 | 484 | *t = *f; |
15ae422b | 485 | previous = t; |
fe3c2583 LP |
486 | t++; |
487 | } | |
488 | ||
489 | *n = t - m; | |
490 | } | |
491 | ||
da6053d0 | 492 | static void drop_inaccessible(MountEntry *m, size_t *n) { |
34de407a | 493 | MountEntry *f, *t; |
fe3c2583 LP |
494 | const char *clear = NULL; |
495 | ||
496 | assert(m); | |
497 | assert(n); | |
498 | ||
499 | /* Drops all entries obstructed by another entry further up the tree. Expects that the array is properly | |
500 | * ordered already. */ | |
501 | ||
1d54cd5d | 502 | for (f = m, t = m; f < m + *n; f++) { |
fe3c2583 LP |
503 | |
504 | /* If we found a path set for INACCESSIBLE earlier, and this entry has it as prefix we should drop | |
505 | * it, as inaccessible paths really should drop the entire subtree. */ | |
34de407a LP |
506 | if (clear && path_startswith(mount_entry_path(f), clear)) { |
507 | log_debug("%s is masked by %s.", mount_entry_path(f), clear); | |
1eb7e08e | 508 | mount_entry_done(f); |
fe3c2583 LP |
509 | continue; |
510 | } | |
15ae422b | 511 | |
34de407a | 512 | clear = f->mode == INACCESSIBLE ? mount_entry_path(f) : NULL; |
fe3c2583 LP |
513 | |
514 | *t = *f; | |
15ae422b LP |
515 | t++; |
516 | } | |
517 | ||
c17ec25e | 518 | *n = t - m; |
15ae422b LP |
519 | } |
520 | ||
da6053d0 | 521 | static void drop_nop(MountEntry *m, size_t *n) { |
34de407a | 522 | MountEntry *f, *t; |
7648a565 LP |
523 | |
524 | assert(m); | |
525 | assert(n); | |
526 | ||
527 | /* Drops all entries which have an immediate parent that has the same type, as they are redundant. Assumes the | |
528 | * list is ordered by prefixes. */ | |
529 | ||
1d54cd5d | 530 | for (f = m, t = m; f < m + *n; f++) { |
7648a565 | 531 | |
1e05071d YW |
532 | /* Only suppress such subtrees for READONLY, READWRITE and READWRITE_IMPLICIT entries */ |
533 | if (IN_SET(f->mode, READONLY, READWRITE, READWRITE_IMPLICIT)) { | |
34de407a | 534 | MountEntry *p; |
7648a565 LP |
535 | bool found = false; |
536 | ||
537 | /* Now let's find the first parent of the entry we are looking at. */ | |
538 | for (p = t-1; p >= m; p--) { | |
34de407a | 539 | if (path_startswith(mount_entry_path(f), mount_entry_path(p))) { |
7648a565 LP |
540 | found = true; |
541 | break; | |
542 | } | |
543 | } | |
544 | ||
545 | /* We found it, let's see if it's the same mode, if so, we can drop this entry */ | |
546 | if (found && p->mode == f->mode) { | |
5beb8688 YW |
547 | log_debug("%s (%s) is made redundant by %s (%s)", |
548 | mount_entry_path(f), mount_mode_to_string(f->mode), | |
549 | mount_entry_path(p), mount_mode_to_string(p->mode)); | |
1eb7e08e | 550 | mount_entry_done(f); |
7648a565 LP |
551 | continue; |
552 | } | |
553 | } | |
554 | ||
555 | *t = *f; | |
556 | t++; | |
557 | } | |
558 | ||
559 | *n = t - m; | |
560 | } | |
561 | ||
da6053d0 | 562 | static void drop_outside_root(const char *root_directory, MountEntry *m, size_t *n) { |
34de407a | 563 | MountEntry *f, *t; |
cd2902c9 LP |
564 | |
565 | assert(m); | |
566 | assert(n); | |
567 | ||
1d54cd5d | 568 | /* Nothing to do */ |
cd2902c9 LP |
569 | if (!root_directory) |
570 | return; | |
571 | ||
572 | /* Drops all mounts that are outside of the root directory. */ | |
573 | ||
1d54cd5d | 574 | for (f = m, t = m; f < m + *n; f++) { |
cd2902c9 | 575 | |
34de407a LP |
576 | if (!path_startswith(mount_entry_path(f), root_directory)) { |
577 | log_debug("%s is outside of root directory.", mount_entry_path(f)); | |
1eb7e08e | 578 | mount_entry_done(f); |
cd2902c9 LP |
579 | continue; |
580 | } | |
581 | ||
582 | *t = *f; | |
583 | t++; | |
584 | } | |
585 | ||
586 | *n = t - m; | |
587 | } | |
588 | ||
b2a60844 LP |
589 | static int clone_device_node( |
590 | const char *d, | |
591 | const char *temporary_mount, | |
592 | bool *make_devnode) { | |
593 | ||
594 | _cleanup_free_ char *sl = NULL; | |
595 | const char *dn, *bn, *t; | |
b5e99f23 ДГ |
596 | struct stat st; |
597 | int r; | |
598 | ||
414b304b | 599 | if (stat(d, &st) < 0) { |
b2a60844 LP |
600 | if (errno == ENOENT) { |
601 | log_debug_errno(errno, "Device node '%s' to clone does not exist, ignoring.", d); | |
af984e13 | 602 | return -ENXIO; |
b2a60844 LP |
603 | } |
604 | ||
605 | return log_debug_errno(errno, "Failed to stat() device node '%s' to clone, ignoring: %m", d); | |
b5e99f23 ДГ |
606 | } |
607 | ||
608 | if (!S_ISBLK(st.st_mode) && | |
baaa35ad ZJS |
609 | !S_ISCHR(st.st_mode)) |
610 | return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), | |
611 | "Device node '%s' to clone is not a device node, ignoring.", | |
612 | d); | |
b5e99f23 | 613 | |
6f7f3a33 | 614 | dn = strjoina(temporary_mount, d); |
b5e99f23 | 615 | |
b2a60844 | 616 | /* First, try to create device node properly */ |
16498617 CB |
617 | if (*make_devnode) { |
618 | mac_selinux_create_file_prepare(d, st.st_mode); | |
619 | r = mknod(dn, st.st_mode, st.st_rdev); | |
620 | mac_selinux_create_file_clear(); | |
b2a60844 LP |
621 | if (r >= 0) |
622 | goto add_symlink; | |
16498617 CB |
623 | if (errno != EPERM) |
624 | return log_debug_errno(errno, "mknod failed for %s: %m", d); | |
625 | ||
b2a60844 | 626 | /* This didn't work, let's not try this again for the next iterations. */ |
16498617 CB |
627 | *make_devnode = false; |
628 | } | |
629 | ||
630 | /* We're about to fallback to bind-mounting the device | |
631 | * node. So create a dummy bind-mount target. */ | |
632 | mac_selinux_create_file_prepare(d, 0); | |
633 | r = mknod(dn, S_IFREG, 0); | |
b5e99f23 | 634 | mac_selinux_create_file_clear(); |
16498617 | 635 | if (r < 0 && errno != EEXIST) |
b2a60844 | 636 | return log_debug_errno(errno, "mknod() fallback failed for '%s': %m", d); |
16498617 CB |
637 | |
638 | /* Fallback to bind-mounting: | |
639 | * The assumption here is that all used device nodes carry standard | |
640 | * properties. Specifically, the devices nodes we bind-mount should | |
641 | * either be owned by root:root or root:tty (e.g. /dev/tty, /dev/ptmx) | |
642 | * and should not carry ACLs. */ | |
643 | if (mount(d, dn, NULL, MS_BIND, NULL) < 0) | |
b2a60844 LP |
644 | return log_debug_errno(errno, "Bind mounting failed for '%s': %m", d); |
645 | ||
646 | add_symlink: | |
647 | bn = path_startswith(d, "/dev/"); | |
648 | if (!bn) | |
649 | return 0; | |
650 | ||
651 | /* Create symlinks like /dev/char/1:9 → ../urandom */ | |
652 | if (asprintf(&sl, "%s/dev/%s/%u:%u", temporary_mount, S_ISCHR(st.st_mode) ? "char" : "block", major(st.st_rdev), minor(st.st_rdev)) < 0) | |
653 | return log_oom(); | |
654 | ||
655 | (void) mkdir_parents(sl, 0755); | |
656 | ||
657 | t = strjoina("../", bn); | |
658 | ||
659 | if (symlink(t, sl) < 0) | |
2e4a4fae | 660 | log_debug_errno(errno, "Failed to symlink '%s' to '%s', ignoring: %m", t, sl); |
b5e99f23 | 661 | |
af984e13 | 662 | return 0; |
b5e99f23 ДГ |
663 | } |
664 | ||
5d997827 | 665 | static int mount_private_dev(MountEntry *m) { |
7f112f50 LP |
666 | static const char devnodes[] = |
667 | "/dev/null\0" | |
668 | "/dev/zero\0" | |
669 | "/dev/full\0" | |
670 | "/dev/random\0" | |
671 | "/dev/urandom\0" | |
672 | "/dev/tty\0"; | |
673 | ||
2b85f4e1 | 674 | char temporary_mount[] = "/tmp/namespace-dev-XXXXXX"; |
63cc4c31 | 675 | const char *d, *dev = NULL, *devpts = NULL, *devshm = NULL, *devhugepages = NULL, *devmqueue = NULL, *devlog = NULL, *devptmx = NULL; |
16498617 | 676 | bool can_mknod = true; |
7f112f50 LP |
677 | _cleanup_umask_ mode_t u; |
678 | int r; | |
679 | ||
680 | assert(m); | |
681 | ||
682 | u = umask(0000); | |
683 | ||
2b85f4e1 | 684 | if (!mkdtemp(temporary_mount)) |
2e4a4fae | 685 | return log_debug_errno(errno, "Failed to create temporary directory '%s': %m", temporary_mount); |
2b85f4e1 | 686 | |
63c372cb | 687 | dev = strjoina(temporary_mount, "/dev"); |
dc751688 | 688 | (void) mkdir(dev, 0755); |
737ba3c8 | 689 | if (mount("tmpfs", dev, "tmpfs", DEV_MOUNT_OPTIONS, "mode=755") < 0) { |
2e4a4fae | 690 | r = log_debug_errno(errno, "Failed to mount tmpfs on '%s': %m", dev); |
2b85f4e1 LP |
691 | goto fail; |
692 | } | |
693 | ||
63c372cb | 694 | devpts = strjoina(temporary_mount, "/dev/pts"); |
dc751688 | 695 | (void) mkdir(devpts, 0755); |
2b85f4e1 | 696 | if (mount("/dev/pts", devpts, NULL, MS_BIND, NULL) < 0) { |
2e4a4fae | 697 | r = log_debug_errno(errno, "Failed to bind mount /dev/pts on '%s': %m", devpts); |
2b85f4e1 LP |
698 | goto fail; |
699 | } | |
700 | ||
2e4a4fae YW |
701 | /* /dev/ptmx can either be a device node or a symlink to /dev/pts/ptmx. |
702 | * When /dev/ptmx a device node, /dev/pts/ptmx has 000 permissions making it inaccessible. | |
703 | * Thus, in that case make a clone. | |
704 | * In nspawn and other containers it will be a symlink, in that case make it a symlink. */ | |
36ce7110 | 705 | r = is_symlink("/dev/ptmx"); |
2e4a4fae YW |
706 | if (r < 0) { |
707 | log_debug_errno(r, "Failed to detect whether /dev/ptmx is a symlink or not: %m"); | |
3164e3cb | 708 | goto fail; |
2e4a4fae | 709 | } else if (r > 0) { |
414b304b ДГ |
710 | devptmx = strjoina(temporary_mount, "/dev/ptmx"); |
711 | if (symlink("pts/ptmx", devptmx) < 0) { | |
2e4a4fae | 712 | r = log_debug_errno(errno, "Failed to create a symlink '%s' to pts/ptmx: %m", devptmx); |
414b304b ДГ |
713 | goto fail; |
714 | } | |
715 | } else { | |
16498617 | 716 | r = clone_device_node("/dev/ptmx", temporary_mount, &can_mknod); |
152c475f LP |
717 | if (r < 0) |
718 | goto fail; | |
414b304b | 719 | } |
e06b6479 | 720 | |
63c372cb | 721 | devshm = strjoina(temporary_mount, "/dev/shm"); |
8d953682 | 722 | (void) mkdir(devshm, 0755); |
2b85f4e1 LP |
723 | r = mount("/dev/shm", devshm, NULL, MS_BIND, NULL); |
724 | if (r < 0) { | |
2e4a4fae | 725 | r = log_debug_errno(errno, "Failed to bind mount /dev/shm on '%s': %m", devshm); |
2b85f4e1 LP |
726 | goto fail; |
727 | } | |
728 | ||
63c372cb | 729 | devmqueue = strjoina(temporary_mount, "/dev/mqueue"); |
dc751688 | 730 | (void) mkdir(devmqueue, 0755); |
2e4a4fae YW |
731 | if (mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL) < 0) |
732 | log_debug_errno(errno, "Failed to bind mount /dev/mqueue on '%s', ignoring: %m", devmqueue); | |
2b85f4e1 | 733 | |
63c372cb | 734 | devhugepages = strjoina(temporary_mount, "/dev/hugepages"); |
dc751688 | 735 | (void) mkdir(devhugepages, 0755); |
2e4a4fae YW |
736 | if (mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL) < 0) |
737 | log_debug_errno(errno, "Failed to bind mount /dev/hugepages on '%s', ignoring: %m", devhugepages); | |
2b85f4e1 | 738 | |
63c372cb | 739 | devlog = strjoina(temporary_mount, "/dev/log"); |
2e4a4fae YW |
740 | if (symlink("/run/systemd/journal/dev-log", devlog) < 0) |
741 | log_debug_errno(errno, "Failed to create a symlink '%s' to /run/systemd/journal/dev-log, ignoring: %m", devlog); | |
82d25240 | 742 | |
7f112f50 | 743 | NULSTR_FOREACH(d, devnodes) { |
16498617 | 744 | r = clone_device_node(d, temporary_mount, &can_mknod); |
af984e13 ZJS |
745 | /* ENXIO means the the *source* is not a device file, skip creation in that case */ |
746 | if (r < 0 && r != -ENXIO) | |
2b85f4e1 | 747 | goto fail; |
7f112f50 LP |
748 | } |
749 | ||
2e4a4fae YW |
750 | r = dev_setup(temporary_mount, UID_INVALID, GID_INVALID); |
751 | if (r < 0) | |
752 | log_debug_errno(r, "Failed to setup basic device tree at '%s', ignoring: %m", temporary_mount); | |
7f112f50 | 753 | |
ee818b89 AC |
754 | /* Create the /dev directory if missing. It is more likely to be |
755 | * missing when the service is started with RootDirectory. This is | |
756 | * consistent with mount units creating the mount points when missing. | |
757 | */ | |
34de407a | 758 | (void) mkdir_p_label(mount_entry_path(m), 0755); |
ee818b89 | 759 | |
9e5f8252 | 760 | /* Unmount everything in old /dev */ |
2e4a4fae YW |
761 | r = umount_recursive(mount_entry_path(m), 0); |
762 | if (r < 0) | |
763 | log_debug_errno(r, "Failed to unmount directories below '%s', ignoring: %m", mount_entry_path(m)); | |
764 | ||
34de407a | 765 | if (mount(dev, mount_entry_path(m), NULL, MS_MOVE, NULL) < 0) { |
2e4a4fae | 766 | r = log_debug_errno(errno, "Failed to move mount point '%s' to '%s': %m", dev, mount_entry_path(m)); |
2b85f4e1 LP |
767 | goto fail; |
768 | } | |
7f112f50 | 769 | |
1019a48f LP |
770 | (void) rmdir(dev); |
771 | (void) rmdir(temporary_mount); | |
7f112f50 | 772 | |
2b85f4e1 | 773 | return 0; |
7f112f50 | 774 | |
2b85f4e1 LP |
775 | fail: |
776 | if (devpts) | |
1019a48f | 777 | (void) umount(devpts); |
7f112f50 | 778 | |
2b85f4e1 | 779 | if (devshm) |
1019a48f | 780 | (void) umount(devshm); |
7f112f50 | 781 | |
2b85f4e1 | 782 | if (devhugepages) |
1019a48f | 783 | (void) umount(devhugepages); |
7f112f50 | 784 | |
2b85f4e1 | 785 | if (devmqueue) |
1019a48f | 786 | (void) umount(devmqueue); |
7f112f50 | 787 | |
1019a48f LP |
788 | (void) umount(dev); |
789 | (void) rmdir(dev); | |
790 | (void) rmdir(temporary_mount); | |
7f112f50 | 791 | |
2b85f4e1 | 792 | return r; |
7f112f50 LP |
793 | } |
794 | ||
2a2969fd | 795 | static int mount_bind_dev(const MountEntry *m) { |
5d997827 LP |
796 | int r; |
797 | ||
798 | assert(m); | |
799 | ||
800 | /* Implements the little brother of mount_private_dev(): simply bind mounts the host's /dev into the service's | |
801 | * /dev. This is only used when RootDirectory= is set. */ | |
802 | ||
645767d6 LP |
803 | (void) mkdir_p_label(mount_entry_path(m), 0755); |
804 | ||
5d997827 LP |
805 | r = path_is_mount_point(mount_entry_path(m), NULL, 0); |
806 | if (r < 0) | |
807 | return log_debug_errno(r, "Unable to determine whether /dev is already mounted: %m"); | |
808 | if (r > 0) /* make this a NOP if /dev is already a mount point */ | |
809 | return 0; | |
810 | ||
811 | if (mount("/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0) | |
812 | return log_debug_errno(errno, "Failed to bind mount %s: %m", mount_entry_path(m)); | |
813 | ||
814 | return 1; | |
815 | } | |
816 | ||
2a2969fd | 817 | static int mount_sysfs(const MountEntry *m) { |
5d997827 LP |
818 | int r; |
819 | ||
820 | assert(m); | |
821 | ||
645767d6 LP |
822 | (void) mkdir_p_label(mount_entry_path(m), 0755); |
823 | ||
5d997827 LP |
824 | r = path_is_mount_point(mount_entry_path(m), NULL, 0); |
825 | if (r < 0) | |
826 | return log_debug_errno(r, "Unable to determine whether /sys is already mounted: %m"); | |
827 | if (r > 0) /* make this a NOP if /sys is already a mount point */ | |
828 | return 0; | |
829 | ||
830 | /* Bind mount the host's version so that we get all child mounts of it, too. */ | |
831 | if (mount("/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0) | |
832 | return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m)); | |
833 | ||
834 | return 1; | |
835 | } | |
836 | ||
2a2969fd | 837 | static int mount_procfs(const MountEntry *m) { |
5d997827 LP |
838 | int r; |
839 | ||
840 | assert(m); | |
841 | ||
645767d6 LP |
842 | (void) mkdir_p_label(mount_entry_path(m), 0755); |
843 | ||
5d997827 LP |
844 | r = path_is_mount_point(mount_entry_path(m), NULL, 0); |
845 | if (r < 0) | |
846 | return log_debug_errno(r, "Unable to determine whether /proc is already mounted: %m"); | |
847 | if (r > 0) /* make this a NOP if /proc is already a mount point */ | |
848 | return 0; | |
849 | ||
850 | /* Mount a new instance, so that we get the one that matches our user namespace, if we are running in one */ | |
851 | if (mount("proc", mount_entry_path(m), "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL) < 0) | |
852 | return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m)); | |
853 | ||
854 | return 1; | |
855 | } | |
856 | ||
2abd4e38 | 857 | static int mount_tmpfs(const MountEntry *m) { |
6c47cd7d LP |
858 | assert(m); |
859 | ||
2abd4e38 | 860 | /* First, get rid of everything that is below if there is anything. Then, overmount with our new tmpfs */ |
6c47cd7d LP |
861 | |
862 | (void) mkdir_p_label(mount_entry_path(m), 0755); | |
863 | (void) umount_recursive(mount_entry_path(m), 0); | |
864 | ||
2abd4e38 | 865 | if (mount("tmpfs", mount_entry_path(m), "tmpfs", m->flags, mount_entry_options(m)) < 0) |
6c47cd7d LP |
866 | return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m)); |
867 | ||
868 | return 1; | |
869 | } | |
870 | ||
088696fe | 871 | static int follow_symlink( |
d2d6c096 | 872 | const char *root_directory, |
088696fe | 873 | MountEntry *m) { |
d2d6c096 | 874 | |
088696fe | 875 | _cleanup_free_ char *target = NULL; |
8fceda93 LP |
876 | int r; |
877 | ||
088696fe LP |
878 | /* Let's chase symlinks, but only one step at a time. That's because depending where the symlink points we |
879 | * might need to change the order in which we mount stuff. Hence: let's normalize piecemeal, and do one step at | |
880 | * a time by specifying CHASE_STEP. This function returns 0 if we resolved one step, and > 0 if we reached the | |
881 | * end and already have a fully normalized name. */ | |
8fceda93 | 882 | |
a5648b80 | 883 | r = chase_symlinks(mount_entry_path(m), root_directory, CHASE_STEP|CHASE_NONEXISTENT, &target, NULL); |
088696fe LP |
884 | if (r < 0) |
885 | return log_debug_errno(r, "Failed to chase symlinks '%s': %m", mount_entry_path(m)); | |
886 | if (r > 0) /* Reached the end, nothing more to resolve */ | |
887 | return 1; | |
8fceda93 | 888 | |
baaa35ad ZJS |
889 | if (m->n_followed >= CHASE_SYMLINKS_MAX) /* put a boundary on things */ |
890 | return log_debug_errno(SYNTHETIC_ERRNO(ELOOP), | |
891 | "Symlink loop on '%s'.", | |
892 | mount_entry_path(m)); | |
8fceda93 | 893 | |
088696fe | 894 | log_debug("Followed mount entry path symlink %s → %s.", mount_entry_path(m), target); |
8fceda93 | 895 | |
088696fe LP |
896 | free_and_replace(m->path_malloc, target); |
897 | m->has_prefix = true; | |
8fceda93 | 898 | |
088696fe LP |
899 | m->n_followed ++; |
900 | ||
901 | return 0; | |
8fceda93 LP |
902 | } |
903 | ||
ac0930c8 | 904 | static int apply_mount( |
8fceda93 | 905 | const char *root_directory, |
89bd586c | 906 | MountEntry *m) { |
ac0930c8 | 907 | |
a227a4be | 908 | bool rbind = true, make = false; |
15ae422b | 909 | const char *what; |
15ae422b | 910 | int r; |
15ae422b | 911 | |
c17ec25e | 912 | assert(m); |
15ae422b | 913 | |
34de407a | 914 | log_debug("Applying namespace mount on %s", mount_entry_path(m)); |
fe3c2583 | 915 | |
c17ec25e | 916 | switch (m->mode) { |
15ae422b | 917 | |
160cfdbe LP |
918 | case INACCESSIBLE: { |
919 | struct stat target; | |
6d313367 LP |
920 | |
921 | /* First, get rid of everything that is below if there | |
922 | * is anything... Then, overmount it with an | |
c4b41707 | 923 | * inaccessible path. */ |
34de407a | 924 | (void) umount_recursive(mount_entry_path(m), 0); |
6d313367 | 925 | |
088696fe LP |
926 | if (lstat(mount_entry_path(m), &target) < 0) { |
927 | if (errno == ENOENT && m->ignore) | |
928 | return 0; | |
929 | ||
34de407a | 930 | return log_debug_errno(errno, "Failed to lstat() %s to determine what to mount over it: %m", mount_entry_path(m)); |
088696fe | 931 | } |
15ae422b | 932 | |
c4b41707 | 933 | what = mode_to_inaccessible_node(target.st_mode); |
baaa35ad ZJS |
934 | if (!what) |
935 | return log_debug_errno(SYNTHETIC_ERRNO(ELOOP), | |
936 | "File type not supported for inaccessible mounts. Note that symlinks are not allowed"); | |
c4b41707 | 937 | break; |
160cfdbe | 938 | } |
fe3c2583 | 939 | |
15ae422b | 940 | case READONLY: |
15ae422b | 941 | case READWRITE: |
1e05071d | 942 | case READWRITE_IMPLICIT: |
8fceda93 | 943 | r = path_is_mount_point(mount_entry_path(m), root_directory, 0); |
088696fe LP |
944 | if (r == -ENOENT && m->ignore) |
945 | return 0; | |
d944dc95 | 946 | if (r < 0) |
34de407a | 947 | return log_debug_errno(r, "Failed to determine whether %s is already a mount point: %m", mount_entry_path(m)); |
6b7c9f8b LP |
948 | 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. */ |
949 | return 0; | |
6b7c9f8b | 950 | /* This isn't a mount point yet, let's make it one. */ |
34de407a | 951 | what = mount_entry_path(m); |
6b7c9f8b | 952 | break; |
15ae422b | 953 | |
d2d6c096 LP |
954 | case BIND_MOUNT: |
955 | rbind = false; | |
d2d6c096 | 956 | |
4831981d | 957 | _fallthrough_; |
088696fe LP |
958 | case BIND_MOUNT_RECURSIVE: { |
959 | _cleanup_free_ char *chased = NULL; | |
5d997827 | 960 | |
088696fe LP |
961 | /* Since mount() will always follow symlinks we chase the symlinks on our own first. Note that bind |
962 | * mount source paths are always relative to the host root, hence we pass NULL as root directory to | |
963 | * chase_symlinks() here. */ | |
964 | ||
a5648b80 | 965 | r = chase_symlinks(mount_entry_source(m), NULL, CHASE_TRAIL_SLASH, &chased, NULL); |
088696fe LP |
966 | if (r == -ENOENT && m->ignore) { |
967 | log_debug_errno(r, "Path %s does not exist, ignoring.", mount_entry_source(m)); | |
968 | return 0; | |
969 | } | |
970 | if (r < 0) | |
971 | return log_debug_errno(r, "Failed to follow symlinks on %s: %m", mount_entry_source(m)); | |
972 | ||
973 | log_debug("Followed source symlinks %s → %s.", mount_entry_source(m), chased); | |
974 | ||
975 | free_and_replace(m->source_malloc, chased); | |
d2d6c096 LP |
976 | |
977 | what = mount_entry_source(m); | |
a227a4be | 978 | make = true; |
d2d6c096 | 979 | break; |
088696fe | 980 | } |
d2d6c096 | 981 | |
6c47cd7d | 982 | case EMPTY_DIR: |
2abd4e38 YW |
983 | case TMPFS: |
984 | return mount_tmpfs(m); | |
6c47cd7d | 985 | |
ac0930c8 | 986 | case PRIVATE_TMP: |
89bd586c | 987 | what = mount_entry_source(m); |
a227a4be | 988 | make = true; |
15ae422b | 989 | break; |
e364ad06 | 990 | |
d6797c92 | 991 | case PRIVATE_DEV: |
5d997827 LP |
992 | return mount_private_dev(m); |
993 | ||
994 | case BIND_DEV: | |
995 | return mount_bind_dev(m); | |
996 | ||
997 | case SYSFS: | |
998 | return mount_sysfs(m); | |
999 | ||
1000 | case PROCFS: | |
1001 | return mount_procfs(m); | |
d6797c92 | 1002 | |
e364ad06 LP |
1003 | default: |
1004 | assert_not_reached("Unknown mode"); | |
15ae422b LP |
1005 | } |
1006 | ||
ac0930c8 | 1007 | assert(what); |
15ae422b | 1008 | |
a227a4be LP |
1009 | if (mount(what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL) < 0) { |
1010 | bool try_again = false; | |
1011 | r = -errno; | |
1012 | ||
1013 | if (r == -ENOENT && make) { | |
1014 | struct stat st; | |
1015 | ||
1016 | /* Hmm, either the source or the destination are missing. Let's see if we can create the destination, then try again */ | |
1017 | ||
e8717862 LP |
1018 | if (stat(what, &st) < 0) |
1019 | log_debug_errno(errno, "Mount point source '%s' is not accessible: %m", what); | |
1020 | else { | |
1021 | int q; | |
a227a4be LP |
1022 | |
1023 | (void) mkdir_parents(mount_entry_path(m), 0755); | |
1024 | ||
1025 | if (S_ISDIR(st.st_mode)) | |
e8717862 | 1026 | q = mkdir(mount_entry_path(m), 0755) < 0 ? -errno : 0; |
a227a4be | 1027 | else |
e8717862 LP |
1028 | q = touch(mount_entry_path(m)); |
1029 | ||
1030 | if (q < 0) | |
1031 | log_debug_errno(q, "Failed to create destination mount point node '%s': %m", mount_entry_path(m)); | |
1032 | else | |
1033 | try_again = true; | |
a227a4be LP |
1034 | } |
1035 | } | |
1036 | ||
1037 | if (try_again) { | |
1038 | if (mount(what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL) < 0) | |
1039 | r = -errno; | |
1040 | else | |
1041 | r = 0; | |
1042 | } | |
1043 | ||
1044 | if (r < 0) | |
1045 | return log_debug_errno(r, "Failed to mount %s to %s: %m", what, mount_entry_path(m)); | |
1046 | } | |
6b7c9f8b | 1047 | |
34de407a | 1048 | log_debug("Successfully mounted %s to %s", what, mount_entry_path(m)); |
6b7c9f8b | 1049 | return 0; |
ac0930c8 | 1050 | } |
15ae422b | 1051 | |
9ce4e4b0 LP |
1052 | /* Change per-mount flags on an existing mount */ |
1053 | static int bind_remount_one(const char *path, unsigned long orig_flags, unsigned long new_flags, unsigned long flags_mask) { | |
1054 | if (mount(NULL, path, NULL, (orig_flags & ~flags_mask) | MS_REMOUNT | MS_BIND | new_flags, NULL) < 0) | |
39e91a27 | 1055 | return -errno; |
69338c3d | 1056 | |
39e91a27 | 1057 | return 0; |
69338c3d AJ |
1058 | } |
1059 | ||
2a2969fd | 1060 | static int make_read_only(const MountEntry *m, char **blacklist, FILE *proc_self_mountinfo) { |
9ce4e4b0 | 1061 | unsigned long new_flags = 0, flags_mask = 0; |
763a260a | 1062 | bool submounts = false; |
6b7c9f8b | 1063 | int r = 0; |
15ae422b | 1064 | |
c17ec25e | 1065 | assert(m); |
ac9de0b3 | 1066 | assert(proc_self_mountinfo); |
ac0930c8 | 1067 | |
9ce4e4b0 LP |
1068 | if (mount_entry_read_only(m) || m->mode == PRIVATE_DEV) { |
1069 | new_flags |= MS_RDONLY; | |
1070 | flags_mask |= MS_RDONLY; | |
1071 | } | |
1072 | ||
1073 | if (m->nosuid) { | |
1074 | new_flags |= MS_NOSUID; | |
1075 | flags_mask |= MS_NOSUID; | |
1076 | } | |
1077 | ||
1078 | if (flags_mask == 0) /* No Change? */ | |
6b7c9f8b LP |
1079 | return 0; |
1080 | ||
9ce4e4b0 LP |
1081 | /* We generally apply these changes recursively, except for /dev, and the cases we know there's |
1082 | * nothing further down. Set /dev readonly, but not submounts like /dev/shm. Also, we only set the | |
1083 | * per-mount read-only flag. We can't set it on the superblock, if we are inside a user namespace | |
1084 | * and running Linux <= 4.17. */ | |
1085 | submounts = | |
1086 | mount_entry_read_only(m) && | |
1087 | !IN_SET(m->mode, EMPTY_DIR, TMPFS); | |
1088 | if (submounts) | |
1089 | r = bind_remount_recursive_with_mountinfo(mount_entry_path(m), new_flags, flags_mask, blacklist, proc_self_mountinfo); | |
1090 | else | |
1091 | r = bind_remount_one(mount_entry_path(m), m->flags, new_flags, flags_mask); | |
1092 | ||
867189b5 LP |
1093 | /* Not that we only turn on the MS_RDONLY flag here, we never turn it off. Something that was marked |
1094 | * read-only already stays this way. This improves compatibility with container managers, where we | |
1095 | * won't attempt to undo read-only mounts already applied. */ | |
ac0930c8 | 1096 | |
8fceda93 | 1097 | if (r == -ENOENT && m->ignore) |
867189b5 | 1098 | return 0; |
763a260a | 1099 | if (r < 0) |
9ce4e4b0 | 1100 | return log_debug_errno(r, "Failed to re-mount '%s'%s: %m", mount_entry_path(m), |
763a260a | 1101 | submounts ? " and its submounts" : ""); |
763a260a | 1102 | return 0; |
d944dc95 LP |
1103 | } |
1104 | ||
9b68367b | 1105 | static bool namespace_info_mount_apivfs(const NamespaceInfo *ns_info) { |
5d997827 LP |
1106 | assert(ns_info); |
1107 | ||
9c988f93 DH |
1108 | /* |
1109 | * ProtectControlGroups= and ProtectKernelTunables= imply MountAPIVFS=, | |
1110 | * since to protect the API VFS mounts, they need to be around in the | |
9b68367b | 1111 | * first place... |
9c988f93 | 1112 | */ |
5d997827 | 1113 | |
9b68367b YW |
1114 | return ns_info->mount_apivfs || |
1115 | ns_info->protect_control_groups || | |
1116 | ns_info->protect_kernel_tunables; | |
5d997827 LP |
1117 | } |
1118 | ||
da6053d0 | 1119 | static size_t namespace_calculate_mounts( |
bb0ff3fb | 1120 | const NamespaceInfo *ns_info, |
2652c6c1 DH |
1121 | char** read_write_paths, |
1122 | char** read_only_paths, | |
1123 | char** inaccessible_paths, | |
6c47cd7d | 1124 | char** empty_directories, |
da6053d0 LP |
1125 | size_t n_bind_mounts, |
1126 | size_t n_temporary_filesystems, | |
2652c6c1 DH |
1127 | const char* tmp_dir, |
1128 | const char* var_tmp_dir, | |
2652c6c1 DH |
1129 | ProtectHome protect_home, |
1130 | ProtectSystem protect_system) { | |
1131 | ||
da6053d0 LP |
1132 | size_t protect_home_cnt; |
1133 | size_t protect_system_cnt = | |
f471b2af DH |
1134 | (protect_system == PROTECT_SYSTEM_STRICT ? |
1135 | ELEMENTSOF(protect_system_strict_table) : | |
1136 | ((protect_system == PROTECT_SYSTEM_FULL) ? | |
1137 | ELEMENTSOF(protect_system_full_table) : | |
1138 | ((protect_system == PROTECT_SYSTEM_YES) ? | |
1139 | ELEMENTSOF(protect_system_yes_table) : 0))); | |
1140 | ||
b6c432ca DH |
1141 | protect_home_cnt = |
1142 | (protect_home == PROTECT_HOME_YES ? | |
1143 | ELEMENTSOF(protect_home_yes_table) : | |
1144 | ((protect_home == PROTECT_HOME_READ_ONLY) ? | |
e4da7d8c YW |
1145 | ELEMENTSOF(protect_home_read_only_table) : |
1146 | ((protect_home == PROTECT_HOME_TMPFS) ? | |
1147 | ELEMENTSOF(protect_home_tmpfs_table) : 0))); | |
b6c432ca | 1148 | |
2652c6c1 DH |
1149 | return !!tmp_dir + !!var_tmp_dir + |
1150 | strv_length(read_write_paths) + | |
1151 | strv_length(read_only_paths) + | |
1152 | strv_length(inaccessible_paths) + | |
6c47cd7d | 1153 | strv_length(empty_directories) + |
d2d6c096 | 1154 | n_bind_mounts + |
2abd4e38 | 1155 | n_temporary_filesystems + |
c575770b DH |
1156 | ns_info->private_dev + |
1157 | (ns_info->protect_kernel_tunables ? ELEMENTSOF(protect_kernel_tunables_table) : 0) + | |
c575770b | 1158 | (ns_info->protect_kernel_modules ? ELEMENTSOF(protect_kernel_modules_table) : 0) + |
94a7b275 KK |
1159 | (ns_info->protect_kernel_logs ? ELEMENTSOF(protect_kernel_logs_table) : 0) + |
1160 | (ns_info->protect_control_groups ? 1 : 0) + | |
5d997827 | 1161 | protect_home_cnt + protect_system_cnt + |
aecd5ac6 | 1162 | (ns_info->protect_hostname ? 2 : 0) + |
9b68367b | 1163 | (namespace_info_mount_apivfs(ns_info) ? ELEMENTSOF(apivfs_table) : 0); |
2652c6c1 DH |
1164 | } |
1165 | ||
da6053d0 | 1166 | static void normalize_mounts(const char *root_directory, MountEntry *mounts, size_t *n_mounts) { |
9b68367b | 1167 | assert(root_directory); |
f8b64b57 LP |
1168 | assert(n_mounts); |
1169 | assert(mounts || *n_mounts == 0); | |
1170 | ||
93bab288 | 1171 | typesafe_qsort(mounts, *n_mounts, mount_path_compare); |
f8b64b57 LP |
1172 | |
1173 | drop_duplicates(mounts, n_mounts); | |
1174 | drop_outside_root(root_directory, mounts, n_mounts); | |
1175 | drop_inaccessible(mounts, n_mounts); | |
1176 | drop_nop(mounts, n_mounts); | |
1177 | } | |
1178 | ||
613b411c | 1179 | int setup_namespace( |
ee818b89 | 1180 | const char* root_directory, |
915e6d16 | 1181 | const char* root_image, |
bb0ff3fb | 1182 | const NamespaceInfo *ns_info, |
2a624c36 AP |
1183 | char** read_write_paths, |
1184 | char** read_only_paths, | |
1185 | char** inaccessible_paths, | |
6c47cd7d | 1186 | char** empty_directories, |
d2d6c096 | 1187 | const BindMount *bind_mounts, |
da6053d0 | 1188 | size_t n_bind_mounts, |
2abd4e38 | 1189 | const TemporaryFileSystem *temporary_filesystems, |
da6053d0 | 1190 | size_t n_temporary_filesystems, |
a004cb4c LP |
1191 | const char* tmp_dir, |
1192 | const char* var_tmp_dir, | |
1b8689f9 LP |
1193 | ProtectHome protect_home, |
1194 | ProtectSystem protect_system, | |
915e6d16 | 1195 | unsigned long mount_flags, |
7cc5ef5f ZJS |
1196 | DissectImageFlags dissect_image_flags, |
1197 | char **error_path) { | |
15ae422b | 1198 | |
915e6d16 | 1199 | _cleanup_(loop_device_unrefp) LoopDevice *loop_device = NULL; |
78ebe980 | 1200 | _cleanup_(decrypted_image_unrefp) DecryptedImage *decrypted_image = NULL; |
915e6d16 | 1201 | _cleanup_(dissected_image_unrefp) DissectedImage *dissected_image = NULL; |
78ebe980 | 1202 | _cleanup_free_ void *root_hash = NULL; |
5f7a690a | 1203 | MountEntry *m = NULL, *mounts = NULL; |
9b68367b | 1204 | size_t n_mounts, root_hash_size = 0; |
d18aff04 | 1205 | bool require_prefix = false; |
9b68367b | 1206 | const char *root; |
c17ec25e | 1207 | int r = 0; |
15ae422b | 1208 | |
915e6d16 LP |
1209 | assert(ns_info); |
1210 | ||
613b411c | 1211 | if (mount_flags == 0) |
c17ec25e | 1212 | mount_flags = MS_SHARED; |
ac0930c8 | 1213 | |
915e6d16 LP |
1214 | if (root_image) { |
1215 | dissect_image_flags |= DISSECT_IMAGE_REQUIRE_ROOT; | |
1216 | ||
c9ef8573 LP |
1217 | if (protect_system == PROTECT_SYSTEM_STRICT && |
1218 | protect_home != PROTECT_HOME_NO && | |
1219 | strv_isempty(read_write_paths)) | |
915e6d16 LP |
1220 | dissect_image_flags |= DISSECT_IMAGE_READ_ONLY; |
1221 | ||
1222 | r = loop_device_make_by_path(root_image, | |
1223 | dissect_image_flags & DISSECT_IMAGE_READ_ONLY ? O_RDONLY : O_RDWR, | |
e08f94ac | 1224 | LO_FLAGS_PARTSCAN, |
915e6d16 LP |
1225 | &loop_device); |
1226 | if (r < 0) | |
763a260a | 1227 | return log_debug_errno(r, "Failed to create loop device for root image: %m"); |
915e6d16 | 1228 | |
78ebe980 LP |
1229 | r = root_hash_load(root_image, &root_hash, &root_hash_size); |
1230 | if (r < 0) | |
763a260a | 1231 | return log_debug_errno(r, "Failed to load root hash: %m"); |
78ebe980 LP |
1232 | |
1233 | r = dissect_image(loop_device->fd, root_hash, root_hash_size, dissect_image_flags, &dissected_image); | |
1234 | if (r < 0) | |
763a260a | 1235 | return log_debug_errno(r, "Failed to dissect image: %m"); |
78ebe980 LP |
1236 | |
1237 | r = dissected_image_decrypt(dissected_image, NULL, root_hash, root_hash_size, dissect_image_flags, &decrypted_image); | |
915e6d16 | 1238 | if (r < 0) |
763a260a | 1239 | return log_debug_errno(r, "Failed to decrypt dissected image: %m"); |
915e6d16 LP |
1240 | } |
1241 | ||
e908468b LP |
1242 | if (root_directory) |
1243 | root = root_directory; | |
0722b359 JS |
1244 | else { |
1245 | /* Always create the mount namespace in a temporary directory, instead of operating | |
1246 | * directly in the root. The temporary directory prevents any mounts from being | |
1247 | * potentially obscured my other mounts we already applied. | |
1248 | * We use the same mount point for all images, which is safe, since they all live | |
1249 | * in their own namespaces after all, and hence won't see each other. */ | |
e908468b LP |
1250 | |
1251 | root = "/run/systemd/unit-root"; | |
1252 | (void) mkdir_label(root, 0700); | |
d18aff04 | 1253 | require_prefix = true; |
0722b359 | 1254 | } |
e908468b | 1255 | |
cfbeb4ef LP |
1256 | n_mounts = namespace_calculate_mounts( |
1257 | ns_info, | |
1258 | read_write_paths, | |
1259 | read_only_paths, | |
1260 | inaccessible_paths, | |
6c47cd7d | 1261 | empty_directories, |
f5c52a77 | 1262 | n_bind_mounts, |
2abd4e38 | 1263 | n_temporary_filesystems, |
cfbeb4ef LP |
1264 | tmp_dir, var_tmp_dir, |
1265 | protect_home, protect_system); | |
613b411c | 1266 | |
f0a4feb0 | 1267 | if (n_mounts > 0) { |
5f7a690a LP |
1268 | m = mounts = new0(MountEntry, n_mounts); |
1269 | if (!mounts) | |
1270 | return -ENOMEM; | |
1271 | ||
d18aff04 | 1272 | r = append_access_mounts(&m, read_write_paths, READWRITE, require_prefix); |
613b411c | 1273 | if (r < 0) |
f0a4feb0 | 1274 | goto finish; |
613b411c | 1275 | |
d18aff04 | 1276 | r = append_access_mounts(&m, read_only_paths, READONLY, require_prefix); |
613b411c | 1277 | if (r < 0) |
f0a4feb0 | 1278 | goto finish; |
613b411c | 1279 | |
d18aff04 | 1280 | r = append_access_mounts(&m, inaccessible_paths, INACCESSIBLE, require_prefix); |
613b411c | 1281 | if (r < 0) |
f0a4feb0 | 1282 | goto finish; |
7ff7394d | 1283 | |
6c47cd7d LP |
1284 | r = append_empty_dir_mounts(&m, empty_directories); |
1285 | if (r < 0) | |
1286 | goto finish; | |
1287 | ||
d2d6c096 LP |
1288 | r = append_bind_mounts(&m, bind_mounts, n_bind_mounts); |
1289 | if (r < 0) | |
1290 | goto finish; | |
1291 | ||
2abd4e38 YW |
1292 | r = append_tmpfs_mounts(&m, temporary_filesystems, n_temporary_filesystems); |
1293 | if (r < 0) | |
1294 | goto finish; | |
1295 | ||
613b411c | 1296 | if (tmp_dir) { |
34de407a | 1297 | *(m++) = (MountEntry) { |
5327c910 LP |
1298 | .path_const = "/tmp", |
1299 | .mode = PRIVATE_TMP, | |
89bd586c | 1300 | .source_const = tmp_dir, |
5327c910 | 1301 | }; |
613b411c | 1302 | } |
7ff7394d | 1303 | |
613b411c | 1304 | if (var_tmp_dir) { |
34de407a | 1305 | *(m++) = (MountEntry) { |
5327c910 | 1306 | .path_const = "/var/tmp", |
89bd586c YW |
1307 | .mode = PRIVATE_TMP, |
1308 | .source_const = var_tmp_dir, | |
5327c910 | 1309 | }; |
7ff7394d | 1310 | } |
ac0930c8 | 1311 | |
c575770b | 1312 | if (ns_info->private_dev) { |
34de407a | 1313 | *(m++) = (MountEntry) { |
5327c910 LP |
1314 | .path_const = "/dev", |
1315 | .mode = PRIVATE_DEV, | |
9ce4e4b0 | 1316 | .flags = DEV_MOUNT_OPTIONS, |
5327c910 | 1317 | }; |
7f112f50 LP |
1318 | } |
1319 | ||
c575770b | 1320 | if (ns_info->protect_kernel_tunables) { |
5327c910 | 1321 | r = append_static_mounts(&m, protect_kernel_tunables_table, ELEMENTSOF(protect_kernel_tunables_table), ns_info->ignore_protect_paths); |
c575770b | 1322 | if (r < 0) |
f0a4feb0 | 1323 | goto finish; |
c575770b DH |
1324 | } |
1325 | ||
1326 | if (ns_info->protect_kernel_modules) { | |
5327c910 | 1327 | r = append_static_mounts(&m, protect_kernel_modules_table, ELEMENTSOF(protect_kernel_modules_table), ns_info->ignore_protect_paths); |
c575770b | 1328 | if (r < 0) |
f0a4feb0 | 1329 | goto finish; |
c575770b | 1330 | } |
59eeb84b | 1331 | |
94a7b275 KK |
1332 | if (ns_info->protect_kernel_logs) { |
1333 | r = append_static_mounts(&m, protect_kernel_logs_table, ELEMENTSOF(protect_kernel_logs_table), ns_info->ignore_protect_paths); | |
1334 | if (r < 0) | |
1335 | goto finish; | |
1336 | } | |
1337 | ||
c575770b | 1338 | if (ns_info->protect_control_groups) { |
34de407a | 1339 | *(m++) = (MountEntry) { |
5327c910 LP |
1340 | .path_const = "/sys/fs/cgroup", |
1341 | .mode = READONLY, | |
1342 | }; | |
59eeb84b LP |
1343 | } |
1344 | ||
5327c910 | 1345 | r = append_protect_home(&m, protect_home, ns_info->ignore_protect_paths); |
b6c432ca | 1346 | if (r < 0) |
f0a4feb0 | 1347 | goto finish; |
417116f2 | 1348 | |
5327c910 | 1349 | r = append_protect_system(&m, protect_system, false); |
f471b2af | 1350 | if (r < 0) |
f0a4feb0 | 1351 | goto finish; |
417116f2 | 1352 | |
9b68367b | 1353 | if (namespace_info_mount_apivfs(ns_info)) { |
5d997827 LP |
1354 | r = append_static_mounts(&m, apivfs_table, ELEMENTSOF(apivfs_table), ns_info->ignore_protect_paths); |
1355 | if (r < 0) | |
1356 | goto finish; | |
1357 | } | |
1358 | ||
aecd5ac6 TM |
1359 | if (ns_info->protect_hostname) { |
1360 | *(m++) = (MountEntry) { | |
1361 | .path_const = "/proc/sys/kernel/hostname", | |
1362 | .mode = READONLY, | |
1363 | }; | |
1364 | *(m++) = (MountEntry) { | |
1365 | .path_const = "/proc/sys/kernel/domainname", | |
1366 | .mode = READONLY, | |
1367 | }; | |
1368 | } | |
1369 | ||
f0a4feb0 | 1370 | assert(mounts + n_mounts == m); |
ac0930c8 | 1371 | |
5327c910 | 1372 | /* Prepend the root directory where that's necessary */ |
e908468b | 1373 | r = prefix_where_needed(mounts, n_mounts, root); |
5327c910 LP |
1374 | if (r < 0) |
1375 | goto finish; | |
1376 | ||
839f1877 | 1377 | normalize_mounts(root, mounts, &n_mounts); |
15ae422b LP |
1378 | } |
1379 | ||
1beab8b0 LP |
1380 | /* All above is just preparation, figuring out what to do. Let's now actually start doing something. */ |
1381 | ||
d944dc95 | 1382 | if (unshare(CLONE_NEWNS) < 0) { |
763a260a | 1383 | r = log_debug_errno(errno, "Failed to unshare the mount namespace: %m"); |
1beab8b0 LP |
1384 | if (IN_SET(r, -EACCES, -EPERM, -EOPNOTSUPP, -ENOSYS)) |
1385 | /* If the kernel doesn't support namespaces, or when there's a MAC or seccomp filter in place | |
1386 | * that doesn't allow us to create namespaces (or a missing cap), then propagate a recognizable | |
1387 | * error back, which the caller can use to detect this case (and only this) and optionally | |
1388 | * continue without namespacing applied. */ | |
1389 | r = -ENOANO; | |
1390 | ||
d944dc95 LP |
1391 | goto finish; |
1392 | } | |
1e4e94c8 | 1393 | |
9b68367b YW |
1394 | /* Remount / as SLAVE so that nothing now mounted in the namespace |
1395 | * shows up in the parent */ | |
1396 | if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) { | |
763a260a | 1397 | r = log_debug_errno(errno, "Failed to remount '/' as SLAVE: %m"); |
9b68367b | 1398 | goto finish; |
ee818b89 AC |
1399 | } |
1400 | ||
915e6d16 | 1401 | if (root_image) { |
e908468b | 1402 | /* A root image is specified, mount it to the right place */ |
2d3a5a73 | 1403 | r = dissected_image_mount(dissected_image, root, UID_INVALID, dissect_image_flags); |
763a260a YW |
1404 | if (r < 0) { |
1405 | log_debug_errno(r, "Failed to mount root image: %m"); | |
915e6d16 | 1406 | goto finish; |
763a260a | 1407 | } |
915e6d16 | 1408 | |
07ce7407 TM |
1409 | if (decrypted_image) { |
1410 | r = decrypted_image_relinquish(decrypted_image); | |
763a260a YW |
1411 | if (r < 0) { |
1412 | log_debug_errno(r, "Failed to relinquish decrypted image: %m"); | |
07ce7407 | 1413 | goto finish; |
763a260a | 1414 | } |
07ce7407 | 1415 | } |
78ebe980 | 1416 | |
915e6d16 LP |
1417 | loop_device_relinquish(loop_device); |
1418 | ||
1419 | } else if (root_directory) { | |
1420 | ||
e908468b LP |
1421 | /* A root directory is specified. Turn its directory into bind mount, if it isn't one yet. */ |
1422 | r = path_is_mount_point(root, NULL, AT_SYMLINK_FOLLOW); | |
763a260a YW |
1423 | if (r < 0) { |
1424 | log_debug_errno(r, "Failed to detect that %s is a mount point or not: %m", root); | |
d944dc95 | 1425 | goto finish; |
763a260a | 1426 | } |
8f1ad200 | 1427 | if (r == 0) { |
e908468b | 1428 | if (mount(root, root, NULL, MS_BIND|MS_REC, NULL) < 0) { |
763a260a | 1429 | r = log_debug_errno(errno, "Failed to bind mount '%s': %m", root); |
8f1ad200 LP |
1430 | goto finish; |
1431 | } | |
d944dc95 | 1432 | } |
e908468b | 1433 | |
9b68367b | 1434 | } else { |
e908468b LP |
1435 | |
1436 | /* Let's mount the main root directory to the root directory to use */ | |
1437 | if (mount("/", root, NULL, MS_BIND|MS_REC, NULL) < 0) { | |
763a260a | 1438 | r = log_debug_errno(errno, "Failed to bind mount '/' on '%s': %m", root); |
e908468b LP |
1439 | goto finish; |
1440 | } | |
ee818b89 | 1441 | } |
c2c13f2d | 1442 | |
4e0c20de LP |
1443 | /* Try to set up the new root directory before mounting anything else there. */ |
1444 | if (root_image || root_directory) | |
1445 | (void) base_filesystem_create(root, UID_INVALID, GID_INVALID); | |
1446 | ||
f0a4feb0 | 1447 | if (n_mounts > 0) { |
ac9de0b3 | 1448 | _cleanup_fclose_ FILE *proc_self_mountinfo = NULL; |
5f7a690a | 1449 | _cleanup_free_ char **blacklist = NULL; |
da6053d0 | 1450 | size_t j; |
6b7c9f8b | 1451 | |
ac9de0b3 TR |
1452 | /* Open /proc/self/mountinfo now as it may become unavailable if we mount anything on top of /proc. |
1453 | * For example, this is the case with the option: 'InaccessiblePaths=/proc' */ | |
1454 | proc_self_mountinfo = fopen("/proc/self/mountinfo", "re"); | |
1455 | if (!proc_self_mountinfo) { | |
763a260a | 1456 | r = log_debug_errno(errno, "Failed to open /proc/self/mountinfo: %m"); |
7cc5ef5f ZJS |
1457 | if (error_path) |
1458 | *error_path = strdup("/proc/self/mountinfo"); | |
ac9de0b3 TR |
1459 | goto finish; |
1460 | } | |
1461 | ||
088696fe LP |
1462 | /* First round, establish all mounts we need */ |
1463 | for (;;) { | |
1464 | bool again = false; | |
1465 | ||
1466 | for (m = mounts; m < mounts + n_mounts; ++m) { | |
1467 | ||
1468 | if (m->applied) | |
1469 | continue; | |
1470 | ||
1471 | r = follow_symlink(root, m); | |
7cc5ef5f ZJS |
1472 | if (r < 0) { |
1473 | if (error_path && mount_entry_path(m)) | |
1474 | *error_path = strdup(mount_entry_path(m)); | |
088696fe | 1475 | goto finish; |
7cc5ef5f | 1476 | } |
088696fe LP |
1477 | if (r == 0) { |
1478 | /* We hit a symlinked mount point. The entry got rewritten and might point to a | |
1479 | * very different place now. Let's normalize the changed list, and start from | |
1480 | * the beginning. After all to mount the entry at the new location we might | |
1481 | * need some other mounts first */ | |
1482 | again = true; | |
1483 | break; | |
1484 | } | |
1485 | ||
1486 | r = apply_mount(root, m); | |
7cc5ef5f ZJS |
1487 | if (r < 0) { |
1488 | if (error_path && mount_entry_path(m)) | |
1489 | *error_path = strdup(mount_entry_path(m)); | |
088696fe | 1490 | goto finish; |
7cc5ef5f | 1491 | } |
088696fe LP |
1492 | |
1493 | m->applied = true; | |
1494 | } | |
1495 | ||
1496 | if (!again) | |
1497 | break; | |
1498 | ||
839f1877 | 1499 | normalize_mounts(root, mounts, &n_mounts); |
c2c13f2d | 1500 | } |
15ae422b | 1501 | |
6b7c9f8b | 1502 | /* Create a blacklist we can pass to bind_mount_recursive() */ |
5f7a690a LP |
1503 | blacklist = new(char*, n_mounts+1); |
1504 | if (!blacklist) { | |
1505 | r = -ENOMEM; | |
1506 | goto finish; | |
1507 | } | |
f0a4feb0 | 1508 | for (j = 0; j < n_mounts; j++) |
34de407a | 1509 | blacklist[j] = (char*) mount_entry_path(mounts+j); |
6b7c9f8b LP |
1510 | blacklist[j] = NULL; |
1511 | ||
1512 | /* Second round, flip the ro bits if necessary. */ | |
f0a4feb0 | 1513 | for (m = mounts; m < mounts + n_mounts; ++m) { |
ac9de0b3 | 1514 | r = make_read_only(m, blacklist, proc_self_mountinfo); |
7cc5ef5f ZJS |
1515 | if (r < 0) { |
1516 | if (error_path && mount_entry_path(m)) | |
1517 | *error_path = strdup(mount_entry_path(m)); | |
d944dc95 | 1518 | goto finish; |
7cc5ef5f | 1519 | } |
c2c13f2d | 1520 | } |
15ae422b LP |
1521 | } |
1522 | ||
9b68367b YW |
1523 | /* MS_MOVE does not work on MS_SHARED so the remount MS_SHARED will be done later */ |
1524 | r = mount_move_root(root); | |
763a260a YW |
1525 | if (r < 0) { |
1526 | log_debug_errno(r, "Failed to mount root with MS_MOVE: %m"); | |
9b68367b | 1527 | goto finish; |
763a260a | 1528 | } |
ee818b89 | 1529 | |
55fe7432 | 1530 | /* Remount / as the desired mode. Note that this will not |
c2c13f2d LP |
1531 | * reestablish propagation from our side to the host, since |
1532 | * what's disconnected is disconnected. */ | |
d944dc95 | 1533 | if (mount(NULL, "/", NULL, mount_flags | MS_REC, NULL) < 0) { |
763a260a | 1534 | r = log_debug_errno(errno, "Failed to remount '/' with desired mount flags: %m"); |
d944dc95 LP |
1535 | goto finish; |
1536 | } | |
15ae422b | 1537 | |
d944dc95 | 1538 | r = 0; |
15ae422b | 1539 | |
d944dc95 | 1540 | finish: |
f0a4feb0 | 1541 | for (m = mounts; m < mounts + n_mounts; m++) |
1eb7e08e | 1542 | mount_entry_done(m); |
613b411c | 1543 | |
5f7a690a LP |
1544 | free(mounts); |
1545 | ||
613b411c LP |
1546 | return r; |
1547 | } | |
1548 | ||
da6053d0 LP |
1549 | void bind_mount_free_many(BindMount *b, size_t n) { |
1550 | size_t i; | |
d2d6c096 LP |
1551 | |
1552 | assert(b || n == 0); | |
1553 | ||
1554 | for (i = 0; i < n; i++) { | |
1555 | free(b[i].source); | |
1556 | free(b[i].destination); | |
1557 | } | |
1558 | ||
1559 | free(b); | |
1560 | } | |
1561 | ||
da6053d0 | 1562 | int bind_mount_add(BindMount **b, size_t *n, const BindMount *item) { |
d2d6c096 LP |
1563 | _cleanup_free_ char *s = NULL, *d = NULL; |
1564 | BindMount *c; | |
1565 | ||
1566 | assert(b); | |
1567 | assert(n); | |
1568 | assert(item); | |
1569 | ||
1570 | s = strdup(item->source); | |
1571 | if (!s) | |
1572 | return -ENOMEM; | |
1573 | ||
1574 | d = strdup(item->destination); | |
1575 | if (!d) | |
1576 | return -ENOMEM; | |
1577 | ||
aa484f35 | 1578 | c = reallocarray(*b, *n + 1, sizeof(BindMount)); |
d2d6c096 LP |
1579 | if (!c) |
1580 | return -ENOMEM; | |
1581 | ||
1582 | *b = c; | |
1583 | ||
1584 | c[(*n) ++] = (BindMount) { | |
1cc6c93a YW |
1585 | .source = TAKE_PTR(s), |
1586 | .destination = TAKE_PTR(d), | |
d2d6c096 | 1587 | .read_only = item->read_only, |
9ce4e4b0 | 1588 | .nosuid = item->nosuid, |
d2d6c096 LP |
1589 | .recursive = item->recursive, |
1590 | .ignore_enoent = item->ignore_enoent, | |
1591 | }; | |
1592 | ||
d2d6c096 LP |
1593 | return 0; |
1594 | } | |
1595 | ||
da6053d0 LP |
1596 | void temporary_filesystem_free_many(TemporaryFileSystem *t, size_t n) { |
1597 | size_t i; | |
2abd4e38 YW |
1598 | |
1599 | assert(t || n == 0); | |
1600 | ||
1601 | for (i = 0; i < n; i++) { | |
1602 | free(t[i].path); | |
1603 | free(t[i].options); | |
1604 | } | |
1605 | ||
1606 | free(t); | |
1607 | } | |
1608 | ||
1609 | int temporary_filesystem_add( | |
1610 | TemporaryFileSystem **t, | |
da6053d0 | 1611 | size_t *n, |
2abd4e38 YW |
1612 | const char *path, |
1613 | const char *options) { | |
1614 | ||
1615 | _cleanup_free_ char *p = NULL, *o = NULL; | |
1616 | TemporaryFileSystem *c; | |
1617 | ||
1618 | assert(t); | |
1619 | assert(n); | |
1620 | assert(path); | |
1621 | ||
1622 | p = strdup(path); | |
1623 | if (!p) | |
1624 | return -ENOMEM; | |
1625 | ||
1626 | if (!isempty(options)) { | |
1627 | o = strdup(options); | |
1628 | if (!o) | |
1629 | return -ENOMEM; | |
1630 | } | |
1631 | ||
aa484f35 | 1632 | c = reallocarray(*t, *n + 1, sizeof(TemporaryFileSystem)); |
2abd4e38 YW |
1633 | if (!c) |
1634 | return -ENOMEM; | |
1635 | ||
1636 | *t = c; | |
1637 | ||
1638 | c[(*n) ++] = (TemporaryFileSystem) { | |
1cc6c93a YW |
1639 | .path = TAKE_PTR(p), |
1640 | .options = TAKE_PTR(o), | |
2abd4e38 YW |
1641 | }; |
1642 | ||
2abd4e38 YW |
1643 | return 0; |
1644 | } | |
1645 | ||
a652f050 JR |
1646 | static int make_tmp_prefix(const char *prefix) { |
1647 | _cleanup_free_ char *t = NULL; | |
1648 | int r; | |
1649 | ||
1650 | /* Don't do anything unless we know the dir is actually missing */ | |
1651 | r = access(prefix, F_OK); | |
1652 | if (r >= 0) | |
1653 | return 0; | |
1654 | if (errno != ENOENT) | |
1655 | return -errno; | |
1656 | ||
1657 | r = mkdir_parents(prefix, 0755); | |
1658 | if (r < 0) | |
1659 | return r; | |
1660 | ||
1661 | r = tempfn_random(prefix, NULL, &t); | |
1662 | if (r < 0) | |
1663 | return r; | |
1664 | ||
1665 | if (mkdir(t, 0777) < 0) | |
1666 | return -errno; | |
1667 | ||
1668 | if (chmod(t, 01777) < 0) { | |
1669 | r = -errno; | |
1670 | (void) rmdir(t); | |
1671 | return r; | |
1672 | } | |
1673 | ||
1674 | if (rename(t, prefix) < 0) { | |
1675 | r = -errno; | |
1676 | (void) rmdir(t); | |
1677 | return r == -EEXIST ? 0 : r; /* it's fine if someone else created the dir by now */ | |
1678 | } | |
1679 | ||
1680 | return 0; | |
1681 | ||
1682 | } | |
1683 | ||
613b411c LP |
1684 | static int setup_one_tmp_dir(const char *id, const char *prefix, char **path) { |
1685 | _cleanup_free_ char *x = NULL; | |
6b46ea73 LP |
1686 | char bid[SD_ID128_STRING_MAX]; |
1687 | sd_id128_t boot_id; | |
1688 | int r; | |
613b411c LP |
1689 | |
1690 | assert(id); | |
1691 | assert(prefix); | |
1692 | assert(path); | |
1693 | ||
6b46ea73 LP |
1694 | /* We include the boot id in the directory so that after a |
1695 | * reboot we can easily identify obsolete directories. */ | |
1696 | ||
1697 | r = sd_id128_get_boot(&boot_id); | |
1698 | if (r < 0) | |
1699 | return r; | |
1700 | ||
605405c6 | 1701 | x = strjoin(prefix, "/systemd-private-", sd_id128_to_string(boot_id, bid), "-", id, "-XXXXXX"); |
613b411c LP |
1702 | if (!x) |
1703 | return -ENOMEM; | |
1704 | ||
a652f050 JR |
1705 | r = make_tmp_prefix(prefix); |
1706 | if (r < 0) | |
1707 | return r; | |
1708 | ||
613b411c LP |
1709 | RUN_WITH_UMASK(0077) |
1710 | if (!mkdtemp(x)) | |
1711 | return -errno; | |
1712 | ||
1713 | RUN_WITH_UMASK(0000) { | |
1714 | char *y; | |
1715 | ||
63c372cb | 1716 | y = strjoina(x, "/tmp"); |
613b411c LP |
1717 | |
1718 | if (mkdir(y, 0777 | S_ISVTX) < 0) | |
1719 | return -errno; | |
c17ec25e | 1720 | } |
15ae422b | 1721 | |
1cc6c93a | 1722 | *path = TAKE_PTR(x); |
613b411c LP |
1723 | |
1724 | return 0; | |
1725 | } | |
1726 | ||
1727 | int setup_tmp_dirs(const char *id, char **tmp_dir, char **var_tmp_dir) { | |
1728 | char *a, *b; | |
1729 | int r; | |
1730 | ||
1731 | assert(id); | |
1732 | assert(tmp_dir); | |
1733 | assert(var_tmp_dir); | |
1734 | ||
1735 | r = setup_one_tmp_dir(id, "/tmp", &a); | |
1736 | if (r < 0) | |
1737 | return r; | |
1738 | ||
1739 | r = setup_one_tmp_dir(id, "/var/tmp", &b); | |
1740 | if (r < 0) { | |
1741 | char *t; | |
1742 | ||
63c372cb | 1743 | t = strjoina(a, "/tmp"); |
6990fb6b LP |
1744 | (void) rmdir(t); |
1745 | (void) rmdir(a); | |
613b411c LP |
1746 | |
1747 | free(a); | |
1748 | return r; | |
1749 | } | |
1750 | ||
1751 | *tmp_dir = a; | |
1752 | *var_tmp_dir = b; | |
1753 | ||
1754 | return 0; | |
1755 | } | |
1756 | ||
2caa38e9 | 1757 | int setup_netns(const int netns_storage_socket[static 2]) { |
613b411c | 1758 | _cleanup_close_ int netns = -1; |
3ee897d6 | 1759 | int r, q; |
613b411c LP |
1760 | |
1761 | assert(netns_storage_socket); | |
1762 | assert(netns_storage_socket[0] >= 0); | |
1763 | assert(netns_storage_socket[1] >= 0); | |
1764 | ||
1765 | /* We use the passed socketpair as a storage buffer for our | |
76cd584b LP |
1766 | * namespace reference fd. Whatever process runs this first |
1767 | * shall create a new namespace, all others should just join | |
1768 | * it. To serialize that we use a file lock on the socket | |
1769 | * pair. | |
613b411c LP |
1770 | * |
1771 | * It's a bit crazy, but hey, works great! */ | |
1772 | ||
1773 | if (lockf(netns_storage_socket[0], F_LOCK, 0) < 0) | |
1774 | return -errno; | |
1775 | ||
3ee897d6 LP |
1776 | netns = receive_one_fd(netns_storage_socket[0], MSG_DONTWAIT); |
1777 | if (netns == -EAGAIN) { | |
44ffcbae | 1778 | /* Nothing stored yet, so let's create a new namespace. */ |
613b411c LP |
1779 | |
1780 | if (unshare(CLONE_NEWNET) < 0) { | |
1781 | r = -errno; | |
1782 | goto fail; | |
1783 | } | |
1784 | ||
44ffcbae | 1785 | (void) loopback_setup(); |
613b411c LP |
1786 | |
1787 | netns = open("/proc/self/ns/net", O_RDONLY|O_CLOEXEC|O_NOCTTY); | |
1788 | if (netns < 0) { | |
1789 | r = -errno; | |
1790 | goto fail; | |
1791 | } | |
1792 | ||
1793 | r = 1; | |
613b411c | 1794 | |
3ee897d6 LP |
1795 | } else if (netns < 0) { |
1796 | r = netns; | |
1797 | goto fail; | |
613b411c | 1798 | |
3ee897d6 LP |
1799 | } else { |
1800 | /* Yay, found something, so let's join the namespace */ | |
613b411c LP |
1801 | if (setns(netns, CLONE_NEWNET) < 0) { |
1802 | r = -errno; | |
1803 | goto fail; | |
1804 | } | |
1805 | ||
1806 | r = 0; | |
1807 | } | |
1808 | ||
3ee897d6 LP |
1809 | q = send_one_fd(netns_storage_socket[1], netns, MSG_DONTWAIT); |
1810 | if (q < 0) { | |
1811 | r = q; | |
613b411c LP |
1812 | goto fail; |
1813 | } | |
1814 | ||
1815 | fail: | |
fe048ce5 | 1816 | (void) lockf(netns_storage_socket[0], F_ULOCK, 0); |
15ae422b LP |
1817 | return r; |
1818 | } | |
417116f2 | 1819 | |
2caa38e9 | 1820 | int open_netns_path(const int netns_storage_socket[static 2], const char *path) { |
51af7fb2 LP |
1821 | _cleanup_close_ int netns = -1; |
1822 | int q, r; | |
1823 | ||
1824 | assert(netns_storage_socket); | |
1825 | assert(netns_storage_socket[0] >= 0); | |
1826 | assert(netns_storage_socket[1] >= 0); | |
1827 | assert(path); | |
1828 | ||
1829 | /* If the storage socket doesn't contain a netns fd yet, open one via the file system and store it in | |
1830 | * it. This is supposed to be called ahead of time, i.e. before setup_netns() which will allocate a | |
1831 | * new anonymous netns if needed. */ | |
1832 | ||
1833 | if (lockf(netns_storage_socket[0], F_LOCK, 0) < 0) | |
1834 | return -errno; | |
1835 | ||
1836 | netns = receive_one_fd(netns_storage_socket[0], MSG_DONTWAIT); | |
1837 | if (netns == -EAGAIN) { | |
1838 | /* Nothing stored yet. Open the file from the file system. */ | |
1839 | ||
1840 | netns = open(path, O_RDONLY|O_NOCTTY|O_CLOEXEC); | |
1841 | if (netns < 0) { | |
1842 | r = -errno; | |
1843 | goto fail; | |
1844 | } | |
1845 | ||
1846 | r = fd_is_network_ns(netns); | |
1847 | if (r == 0) { /* Not a netns? Refuse early. */ | |
1848 | r = -EINVAL; | |
1849 | goto fail; | |
1850 | } | |
1851 | if (r < 0 && r != -EUCLEAN) /* EUCLEAN: we don't know */ | |
1852 | goto fail; | |
1853 | ||
1854 | r = 1; | |
1855 | ||
1856 | } else if (netns < 0) { | |
1857 | r = netns; | |
1858 | goto fail; | |
1859 | } else | |
1860 | r = 0; /* Already allocated */ | |
1861 | ||
1862 | q = send_one_fd(netns_storage_socket[1], netns, MSG_DONTWAIT); | |
1863 | if (q < 0) { | |
1864 | r = q; | |
1865 | goto fail; | |
1866 | } | |
1867 | ||
1868 | fail: | |
1869 | (void) lockf(netns_storage_socket[0], F_ULOCK, 0); | |
1870 | return r; | |
1871 | } | |
1872 | ||
6e2d7c4f MS |
1873 | bool ns_type_supported(NamespaceType type) { |
1874 | const char *t, *ns_proc; | |
1875 | ||
0fa5b831 LP |
1876 | t = namespace_type_to_string(type); |
1877 | if (!t) /* Don't know how to translate this? Then it's not supported */ | |
6e2d7c4f MS |
1878 | return false; |
1879 | ||
6e2d7c4f | 1880 | ns_proc = strjoina("/proc/self/ns/", t); |
6e2d7c4f MS |
1881 | return access(ns_proc, F_OK) == 0; |
1882 | } | |
1883 | ||
1b8689f9 LP |
1884 | static const char *const protect_home_table[_PROTECT_HOME_MAX] = { |
1885 | [PROTECT_HOME_NO] = "no", | |
1886 | [PROTECT_HOME_YES] = "yes", | |
1887 | [PROTECT_HOME_READ_ONLY] = "read-only", | |
e4da7d8c | 1888 | [PROTECT_HOME_TMPFS] = "tmpfs", |
417116f2 LP |
1889 | }; |
1890 | ||
1e8c7bd5 | 1891 | DEFINE_STRING_TABLE_LOOKUP_WITH_BOOLEAN(protect_home, ProtectHome, PROTECT_HOME_YES); |
5e1c6154 | 1892 | |
1b8689f9 LP |
1893 | static const char *const protect_system_table[_PROTECT_SYSTEM_MAX] = { |
1894 | [PROTECT_SYSTEM_NO] = "no", | |
1895 | [PROTECT_SYSTEM_YES] = "yes", | |
1896 | [PROTECT_SYSTEM_FULL] = "full", | |
3f815163 | 1897 | [PROTECT_SYSTEM_STRICT] = "strict", |
1b8689f9 LP |
1898 | }; |
1899 | ||
1e8c7bd5 | 1900 | DEFINE_STRING_TABLE_LOOKUP_WITH_BOOLEAN(protect_system, ProtectSystem, PROTECT_SYSTEM_YES); |
03c791aa | 1901 | |
6e2d7c4f MS |
1902 | static const char* const namespace_type_table[] = { |
1903 | [NAMESPACE_MOUNT] = "mnt", | |
1904 | [NAMESPACE_CGROUP] = "cgroup", | |
1905 | [NAMESPACE_UTS] = "uts", | |
1906 | [NAMESPACE_IPC] = "ipc", | |
1907 | [NAMESPACE_USER] = "user", | |
1908 | [NAMESPACE_PID] = "pid", | |
1909 | [NAMESPACE_NET] = "net", | |
1910 | }; | |
1911 | ||
1912 | DEFINE_STRING_TABLE_LOOKUP(namespace_type, NamespaceType); |