1 .. SPDX-License-Identifier: GPL-2.0
2 .. Copyright © 2017-2020 Mickaël Salaün <mic@digikod.net>
3 .. Copyright © 2019-2020 ANSSI
4 .. Copyright © 2021-2022 Microsoft Corporation
6 =====================================
7 Landlock: unprivileged access control
8 =====================================
10 :Author: Mickaël Salaün
13 The goal of Landlock is to enable to restrict ambient rights (e.g. global
14 filesystem or network access) for a set of processes. Because Landlock
15 is a stackable LSM, it makes possible to create safe security sandboxes as new
16 security layers in addition to the existing system-wide access-controls. This
17 kind of sandbox is expected to help mitigate the security impact of bugs or
18 unexpected/malicious behaviors in user space applications. Landlock empowers
19 any process, including unprivileged ones, to securely restrict themselves.
21 We can quickly make sure that Landlock is enabled in the running system by
22 looking for "landlock: Up and running" in kernel logs (as root): ``dmesg | grep
23 landlock || journalctl -kg landlock`` . Developers can also easily check for
24 Landlock support with a :ref:`related system call <landlock_abi_versions>`. If
25 Landlock is not currently supported, we need to :ref:`configure the kernel
26 appropriately <kernel_support>`.
31 A Landlock rule describes an action on an object which the process intends to
32 perform. A set of rules is aggregated in a ruleset, which can then restrict
33 the thread enforcing it, and its future children.
35 The two existing types of rules are:
38 For these rules, the object is a file hierarchy,
39 and the related filesystem actions are defined with
40 `filesystem access rights`.
42 Network rules (since ABI v4)
43 For these rules, the object is a TCP port,
44 and the related actions are defined with `network access rights`.
46 Defining and enforcing a security policy
47 ----------------------------------------
49 We first need to define the ruleset that will contain our rules.
51 For this example, the ruleset will contain rules that only allow filesystem
52 read actions and establish a specific TCP connection. Filesystem write
53 actions and other TCP actions will be denied.
55 The ruleset then needs to handle both these kinds of actions. This is
56 required for backward and forward compatibility (i.e. the kernel and user
57 space may not know each other's supported restrictions), hence the need
58 to be explicit about the denied-by-default access rights.
62 struct landlock_ruleset_attr ruleset_attr = {
64 LANDLOCK_ACCESS_FS_EXECUTE |
65 LANDLOCK_ACCESS_FS_WRITE_FILE |
66 LANDLOCK_ACCESS_FS_READ_FILE |
67 LANDLOCK_ACCESS_FS_READ_DIR |
68 LANDLOCK_ACCESS_FS_REMOVE_DIR |
69 LANDLOCK_ACCESS_FS_REMOVE_FILE |
70 LANDLOCK_ACCESS_FS_MAKE_CHAR |
71 LANDLOCK_ACCESS_FS_MAKE_DIR |
72 LANDLOCK_ACCESS_FS_MAKE_REG |
73 LANDLOCK_ACCESS_FS_MAKE_SOCK |
74 LANDLOCK_ACCESS_FS_MAKE_FIFO |
75 LANDLOCK_ACCESS_FS_MAKE_BLOCK |
76 LANDLOCK_ACCESS_FS_MAKE_SYM |
77 LANDLOCK_ACCESS_FS_REFER |
78 LANDLOCK_ACCESS_FS_TRUNCATE,
80 LANDLOCK_ACCESS_NET_BIND_TCP |
81 LANDLOCK_ACCESS_NET_CONNECT_TCP,
84 Because we may not know on which kernel version an application will be
85 executed, it is safer to follow a best-effort security approach. Indeed, we
86 should try to protect users as much as possible whatever the kernel they are
87 using. To avoid binary enforcement (i.e. either all security features or
88 none), we can leverage a dedicated Landlock command to get the current version
89 of the Landlock ABI and adapt the handled accesses. Let's check if we should
90 remove access rights which are only supported in higher versions of the ABI.
96 abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
98 /* Degrades gracefully if Landlock is not handled. */
99 perror("The running kernel does not enable to use Landlock");
104 /* Removes LANDLOCK_ACCESS_FS_REFER for ABI < 2 */
105 ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_REFER;
106 __attribute__((fallthrough));
108 /* Removes LANDLOCK_ACCESS_FS_TRUNCATE for ABI < 3 */
109 ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_TRUNCATE;
110 __attribute__((fallthrough));
112 /* Removes network support for ABI < 4 */
113 ruleset_attr.handled_access_net &=
114 ~(LANDLOCK_ACCESS_NET_BIND_TCP |
115 LANDLOCK_ACCESS_NET_CONNECT_TCP);
118 This enables to create an inclusive ruleset that will contain our rules.
124 ruleset_fd = landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0);
125 if (ruleset_fd < 0) {
126 perror("Failed to create a ruleset");
130 We can now add a new rule to this ruleset thanks to the returned file
131 descriptor referring to this ruleset. The rule will only allow reading the
132 file hierarchy ``/usr``. Without another rule, write actions would then be
133 denied by the ruleset. To add ``/usr`` to the ruleset, we open it with the
134 ``O_PATH`` flag and fill the &struct landlock_path_beneath_attr with this file
140 struct landlock_path_beneath_attr path_beneath = {
142 LANDLOCK_ACCESS_FS_EXECUTE |
143 LANDLOCK_ACCESS_FS_READ_FILE |
144 LANDLOCK_ACCESS_FS_READ_DIR,
147 path_beneath.parent_fd = open("/usr", O_PATH | O_CLOEXEC);
148 if (path_beneath.parent_fd < 0) {
149 perror("Failed to open file");
153 err = landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH,
155 close(path_beneath.parent_fd);
157 perror("Failed to update ruleset");
162 It may also be required to create rules following the same logic as explained
163 for the ruleset creation, by filtering access rights according to the Landlock
164 ABI version. In this example, this is not required because all of the requested
165 ``allowed_access`` rights are already available in ABI 1.
167 For network access-control, we can add a set of rules that allow to use a port
168 number for a specific action: HTTPS connections.
172 struct landlock_net_port_attr net_port = {
173 .allowed_access = LANDLOCK_ACCESS_NET_CONNECT_TCP,
177 err = landlock_add_rule(ruleset_fd, LANDLOCK_RULE_NET_PORT,
180 The next step is to restrict the current thread from gaining more privileges
181 (e.g. through a SUID binary). We now have a ruleset with the first rule
182 allowing read access to ``/usr`` while denying all other handled accesses for
183 the filesystem, and a second rule allowing HTTPS connections.
187 if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
188 perror("Failed to restrict privileges");
193 The current thread is now ready to sandbox itself with the ruleset.
197 if (landlock_restrict_self(ruleset_fd, 0)) {
198 perror("Failed to enforce ruleset");
204 If the ``landlock_restrict_self`` system call succeeds, the current thread is
205 now restricted and this policy will be enforced on all its subsequently created
206 children as well. Once a thread is landlocked, there is no way to remove its
207 security policy; only adding more restrictions is allowed. These threads are
208 now in a new Landlock domain, merge of their parent one (if any) with the new
211 Full working code can be found in `samples/landlock/sandboxer.c`_.
216 It is recommended setting access rights to file hierarchy leaves as much as
217 possible. For instance, it is better to be able to have ``~/doc/`` as a
218 read-only hierarchy and ``~/tmp/`` as a read-write hierarchy, compared to
219 ``~/`` as a read-only hierarchy and ``~/tmp/`` as a read-write hierarchy.
220 Following this good practice leads to self-sufficient hierarchies that do not
221 depend on their location (i.e. parent directories). This is particularly
222 relevant when we want to allow linking or renaming. Indeed, having consistent
223 access rights per directory enables to change the location of such directory
224 without relying on the destination directory access rights (except those that
225 are required for this operation, see ``LANDLOCK_ACCESS_FS_REFER``
227 Having self-sufficient hierarchies also helps to tighten the required access
228 rights to the minimal set of data. This also helps avoid sinkhole directories,
229 i.e. directories where data can be linked to but not linked from. However,
230 this depends on data organization, which might not be controlled by developers.
231 In this case, granting read-write access to ``~/tmp/``, instead of write-only
232 access, would potentially allow to move ``~/tmp/`` to a non-readable directory
233 and still keep the ability to list the content of ``~/tmp/``.
235 Layers of file path access rights
236 ---------------------------------
238 Each time a thread enforces a ruleset on itself, it updates its Landlock domain
239 with a new layer of policy. Indeed, this complementary policy is stacked with
240 the potentially other rulesets already restricting this thread. A sandboxed
241 thread can then safely add more constraints to itself with a new enforced
244 One policy layer grants access to a file path if at least one of its rules
245 encountered on the path grants the access. A sandboxed thread can only access
246 a file path if all its enforced policy layers grant the access as well as all
247 the other system access controls (e.g. filesystem DAC, other LSM policies,
250 Bind mounts and OverlayFS
251 -------------------------
253 Landlock enables to restrict access to file hierarchies, which means that these
254 access rights can be propagated with bind mounts (cf.
255 Documentation/filesystems/sharedsubtree.rst) but not with
256 Documentation/filesystems/overlayfs.rst.
258 A bind mount mirrors a source file hierarchy to a destination. The destination
259 hierarchy is then composed of the exact same files, on which Landlock rules can
260 be tied, either via the source or the destination path. These rules restrict
261 access when they are encountered on a path, which means that they can restrict
262 access to multiple file hierarchies at the same time, whether these hierarchies
263 are the result of bind mounts or not.
265 An OverlayFS mount point consists of upper and lower layers. These layers are
266 combined in a merge directory, result of the mount point. This merge hierarchy
267 may include files from the upper and lower layers, but modifications performed
268 on the merge hierarchy only reflects on the upper layer. From a Landlock
269 policy point of view, each OverlayFS layers and merge hierarchies are
270 standalone and contains their own set of files and directories, which is
271 different from bind mounts. A policy restricting an OverlayFS layer will not
272 restrict the resulted merged hierarchy, and vice versa. Landlock users should
273 then only think about file hierarchies they want to allow access to, regardless
274 of the underlying filesystem.
279 Every new thread resulting from a :manpage:`clone(2)` inherits Landlock domain
280 restrictions from its parent. This is similar to the seccomp inheritance (cf.
281 Documentation/userspace-api/seccomp_filter.rst) or any other LSM dealing with
282 task's :manpage:`credentials(7)`. For instance, one process's thread may apply
283 Landlock rules to itself, but they will not be automatically applied to other
284 sibling threads (unlike POSIX thread credential changes, cf.
287 When a thread sandboxes itself, we have the guarantee that the related security
288 policy will stay enforced on all this thread's descendants. This allows
289 creating standalone and modular security policies per application, which will
290 automatically be composed between themselves according to their runtime parent
296 A sandboxed process has less privileges than a non-sandboxed process and must
297 then be subject to additional restrictions when manipulating another process.
298 To be allowed to use :manpage:`ptrace(2)` and related syscalls on a target
299 process, a sandboxed process should have a subset of the target process rules,
300 which means the tracee must be in a sub-domain of the tracer.
305 The operations covered by ``LANDLOCK_ACCESS_FS_WRITE_FILE`` and
306 ``LANDLOCK_ACCESS_FS_TRUNCATE`` both change the contents of a file and sometimes
307 overlap in non-intuitive ways. It is recommended to always specify both of
310 A particularly surprising example is :manpage:`creat(2)`. The name suggests
311 that this system call requires the rights to create and write files. However,
312 it also requires the truncate right if an existing file under the same name is
315 It should also be noted that truncating files does not require the
316 ``LANDLOCK_ACCESS_FS_WRITE_FILE`` right. Apart from the :manpage:`truncate(2)`
317 system call, this can also be done through :manpage:`open(2)` with the flags
318 ``O_RDONLY | O_TRUNC``.
320 When opening a file, the availability of the ``LANDLOCK_ACCESS_FS_TRUNCATE``
321 right is associated with the newly created file descriptor and will be used for
322 subsequent truncation attempts using :manpage:`ftruncate(2)`. The behavior is
323 similar to opening a file for reading or writing, where permissions are checked
324 during :manpage:`open(2)`, but not during the subsequent :manpage:`read(2)` and
325 :manpage:`write(2)` calls.
327 As a consequence, it is possible to have multiple open file descriptors for the
328 same file, where one grants the right to truncate the file and the other does
329 not. It is also possible to pass such file descriptors between processes,
330 keeping their Landlock properties, even when these processes do not have an
331 enforced Landlock ruleset.
336 Backward and forward compatibility
337 ----------------------------------
339 Landlock is designed to be compatible with past and future versions of the
340 kernel. This is achieved thanks to the system call attributes and the
341 associated bitflags, particularly the ruleset's ``handled_access_fs``. Making
342 handled access right explicit enables the kernel and user space to have a clear
343 contract with each other. This is required to make sure sandboxing will not
344 get stricter with a system update, which could break applications.
346 Developers can subscribe to the `Landlock mailing list
347 <https://subspace.kernel.org/lists.linux.dev.html>`_ to knowingly update and
348 test their applications with the latest available features. In the interest of
349 users, and because they may use different kernel versions, it is strongly
350 encouraged to follow a best-effort security approach by checking the Landlock
351 ABI version at runtime and only enforcing the supported features.
353 .. _landlock_abi_versions:
355 Landlock ABI versions
356 ---------------------
358 The Landlock ABI version can be read with the sys_landlock_create_ruleset()
365 abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
369 printf("Landlock is not supported by the current kernel.\n");
372 printf("Landlock is currently disabled.\n");
378 printf("Landlock supports LANDLOCK_ACCESS_FS_REFER.\n");
381 The following kernel interfaces are implicitly supported by the first ABI
382 version. Features only supported from a specific version are explicitly marked
391 .. kernel-doc:: include/uapi/linux/landlock.h
392 :identifiers: fs_access net_access
394 Creating a new ruleset
395 ----------------------
397 .. kernel-doc:: security/landlock/syscalls.c
398 :identifiers: sys_landlock_create_ruleset
400 .. kernel-doc:: include/uapi/linux/landlock.h
401 :identifiers: landlock_ruleset_attr
406 .. kernel-doc:: security/landlock/syscalls.c
407 :identifiers: sys_landlock_add_rule
409 .. kernel-doc:: include/uapi/linux/landlock.h
410 :identifiers: landlock_rule_type landlock_path_beneath_attr
411 landlock_net_port_attr
416 .. kernel-doc:: security/landlock/syscalls.c
417 :identifiers: sys_landlock_restrict_self
422 Filesystem topology modification
423 --------------------------------
425 Threads sandboxed with filesystem restrictions cannot modify filesystem
426 topology, whether via :manpage:`mount(2)` or :manpage:`pivot_root(2)`.
427 However, :manpage:`chroot(2)` calls are not denied.
432 Access to regular files and directories can be restricted by Landlock,
433 according to the handled accesses of a ruleset. However, files that do not
434 come from a user-visible filesystem (e.g. pipe, socket), but can still be
435 accessed through ``/proc/<pid>/fd/*``, cannot currently be explicitly
436 restricted. Likewise, some special kernel filesystems such as nsfs, which can
437 be accessed through ``/proc/<pid>/ns/*``, cannot currently be explicitly
438 restricted. However, thanks to the `ptrace restrictions`_, access to such
439 sensitive ``/proc`` files are automatically restricted according to domain
440 hierarchies. Future Landlock evolutions could still enable to explicitly
441 restrict such paths with dedicated ruleset flags.
446 There is a limit of 16 layers of stacked rulesets. This can be an issue for a
447 task willing to enforce a new ruleset in complement to its 16 inherited
448 rulesets. Once this limit is reached, sys_landlock_restrict_self() returns
449 E2BIG. It is then strongly suggested to carefully build rulesets once in the
450 life of a thread, especially for applications able to launch other applications
451 that may also want to sandbox themselves (e.g. shells, container managers,
457 Kernel memory allocated to create rulesets is accounted and can be restricted
458 by the Documentation/admin-guide/cgroup-v1/memory.rst.
463 File renaming and linking (ABI < 2)
464 -----------------------------------
466 Because Landlock targets unprivileged access controls, it needs to properly
467 handle composition of rules. Such property also implies rules nesting.
468 Properly handling multiple layers of rulesets, each one of them able to
469 restrict access to files, also implies inheritance of the ruleset restrictions
470 from a parent to its hierarchy. Because files are identified and restricted by
471 their hierarchy, moving or linking a file from one directory to another implies
472 propagation of the hierarchy constraints, or restriction of these actions
473 according to the potentially lost constraints. To protect against privilege
474 escalations through renaming or linking, and for the sake of simplicity,
475 Landlock previously limited linking and renaming to the same directory.
476 Starting with the Landlock ABI version 2, it is now possible to securely
477 control renaming and linking thanks to the new ``LANDLOCK_ACCESS_FS_REFER``
480 File truncation (ABI < 3)
481 -------------------------
483 File truncation could not be denied before the third Landlock ABI, so it is
484 always allowed when using a kernel that only supports the first or second ABI.
486 Starting with the Landlock ABI version 3, it is now possible to securely control
487 truncation thanks to the new ``LANDLOCK_ACCESS_FS_TRUNCATE`` access right.
489 Network support (ABI < 4)
490 -------------------------
492 Starting with the Landlock ABI version 4, it is now possible to restrict TCP
493 bind and connect actions to only a set of allowed ports thanks to the new
494 ``LANDLOCK_ACCESS_NET_BIND_TCP`` and ``LANDLOCK_ACCESS_NET_CONNECT_TCP``
502 Landlock was first introduced in Linux 5.13 but it must be configured at build
503 time with ``CONFIG_SECURITY_LANDLOCK=y``. Landlock must also be enabled at boot
504 time as the other security modules. The list of security modules enabled by
505 default is set with ``CONFIG_LSM``. The kernel configuration should then
506 contains ``CONFIG_LSM=landlock,[...]`` with ``[...]`` as the list of other
507 potentially useful security modules for the running system (see the
508 ``CONFIG_LSM`` help).
510 If the running kernel does not have ``landlock`` in ``CONFIG_LSM``, then we can
511 still enable it by adding ``lsm=landlock,[...]`` to
512 Documentation/admin-guide/kernel-parameters.rst thanks to the bootloader
515 To be able to explicitly allow TCP operations (e.g., adding a network rule with
516 ``LANDLOCK_ACCESS_NET_BIND_TCP``), the kernel must support TCP
517 (``CONFIG_INET=y``). Otherwise, sys_landlock_add_rule() returns an
518 ``EAFNOSUPPORT`` error, which can safely be ignored because this kind of TCP
519 operation is already not possible.
521 Questions and answers
522 =====================
524 What about user space sandbox managers?
525 ---------------------------------------
527 Using user space process to enforce restrictions on kernel resources can lead
528 to race conditions or inconsistent evaluations (i.e. `Incorrect mirroring of
529 the OS code and state
530 <https://www.ndss-symposium.org/ndss2003/traps-and-pitfalls-practical-problems-system-call-interposition-based-security-tools/>`_).
532 What about namespaces and containers?
533 -------------------------------------
535 Namespaces can help create sandboxes but they are not designed for
536 access-control and then miss useful features for such use case (e.g. no
537 fine-grained restrictions). Moreover, their complexity can lead to security
538 issues, especially when untrusted processes can manipulate them (cf.
539 `Controlling access to user namespaces <https://lwn.net/Articles/673597/>`_).
541 Additional documentation
542 ========================
544 * Documentation/security/landlock.rst
545 * https://landlock.io
548 .. _samples/landlock/sandboxer.c:
549 https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/samples/landlock/sandboxer.c