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 access) for a set of processes. Because Landlock is a stackable
15 LSM, it makes possible to create safe security sandboxes as new security layers
16 in addition to the existing system-wide access-controls. This kind of sandbox
17 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. An object is currently a
32 file hierarchy, and the related filesystem actions are defined with `access
33 rights`_. A set of rules is aggregated in a ruleset, which can then restrict
34 the thread enforcing it, and its future children.
36 Defining and enforcing a security policy
37 ----------------------------------------
39 We first need to define the ruleset that will contain our rules. For this
40 example, the ruleset will contain rules that only allow read actions, but write
41 actions will be denied. The ruleset then needs to handle both of these kind of
42 actions. This is required for backward and forward compatibility (i.e. the
43 kernel and user space may not know each other's supported restrictions), hence
44 the need to be explicit about the denied-by-default access rights.
48 struct landlock_ruleset_attr ruleset_attr = {
50 LANDLOCK_ACCESS_FS_EXECUTE |
51 LANDLOCK_ACCESS_FS_WRITE_FILE |
52 LANDLOCK_ACCESS_FS_READ_FILE |
53 LANDLOCK_ACCESS_FS_READ_DIR |
54 LANDLOCK_ACCESS_FS_REMOVE_DIR |
55 LANDLOCK_ACCESS_FS_REMOVE_FILE |
56 LANDLOCK_ACCESS_FS_MAKE_CHAR |
57 LANDLOCK_ACCESS_FS_MAKE_DIR |
58 LANDLOCK_ACCESS_FS_MAKE_REG |
59 LANDLOCK_ACCESS_FS_MAKE_SOCK |
60 LANDLOCK_ACCESS_FS_MAKE_FIFO |
61 LANDLOCK_ACCESS_FS_MAKE_BLOCK |
62 LANDLOCK_ACCESS_FS_MAKE_SYM |
63 LANDLOCK_ACCESS_FS_REFER,
66 Because we may not know on which kernel version an application will be
67 executed, it is safer to follow a best-effort security approach. Indeed, we
68 should try to protect users as much as possible whatever the kernel they are
69 using. To avoid binary enforcement (i.e. either all security features or
70 none), we can leverage a dedicated Landlock command to get the current version
71 of the Landlock ABI and adapt the handled accesses. Let's check if we should
72 remove the `LANDLOCK_ACCESS_FS_REFER` access right which is only supported
73 starting with the second version of the ABI.
79 abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
81 ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_REFER;
84 This enables to create an inclusive ruleset that will contain our rules.
90 ruleset_fd = landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0);
92 perror("Failed to create a ruleset");
96 We can now add a new rule to this ruleset thanks to the returned file
97 descriptor referring to this ruleset. The rule will only allow reading the
98 file hierarchy ``/usr``. Without another rule, write actions would then be
99 denied by the ruleset. To add ``/usr`` to the ruleset, we open it with the
100 ``O_PATH`` flag and fill the &struct landlock_path_beneath_attr with this file
106 struct landlock_path_beneath_attr path_beneath = {
108 LANDLOCK_ACCESS_FS_EXECUTE |
109 LANDLOCK_ACCESS_FS_READ_FILE |
110 LANDLOCK_ACCESS_FS_READ_DIR,
113 path_beneath.parent_fd = open("/usr", O_PATH | O_CLOEXEC);
114 if (path_beneath.parent_fd < 0) {
115 perror("Failed to open file");
119 err = landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH,
121 close(path_beneath.parent_fd);
123 perror("Failed to update ruleset");
128 It may also be required to create rules following the same logic as explained
129 for the ruleset creation, by filtering access rights according to the Landlock
130 ABI version. In this example, this is not required because
131 `LANDLOCK_ACCESS_FS_REFER` is not allowed by any rule.
133 We now have a ruleset with one rule allowing read access to ``/usr`` while
134 denying all other handled accesses for the filesystem. The next step is to
135 restrict the current thread from gaining more privileges (e.g. thanks to a SUID
140 if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
141 perror("Failed to restrict privileges");
146 The current thread is now ready to sandbox itself with the ruleset.
150 if (landlock_restrict_self(ruleset_fd, 0)) {
151 perror("Failed to enforce ruleset");
157 If the `landlock_restrict_self` system call succeeds, the current thread is now
158 restricted and this policy will be enforced on all its subsequently created
159 children as well. Once a thread is landlocked, there is no way to remove its
160 security policy; only adding more restrictions is allowed. These threads are
161 now in a new Landlock domain, merge of their parent one (if any) with the new
164 Full working code can be found in `samples/landlock/sandboxer.c`_.
169 It is recommended setting access rights to file hierarchy leaves as much as
170 possible. For instance, it is better to be able to have ``~/doc/`` as a
171 read-only hierarchy and ``~/tmp/`` as a read-write hierarchy, compared to
172 ``~/`` as a read-only hierarchy and ``~/tmp/`` as a read-write hierarchy.
173 Following this good practice leads to self-sufficient hierarchies that don't
174 depend on their location (i.e. parent directories). This is particularly
175 relevant when we want to allow linking or renaming. Indeed, having consistent
176 access rights per directory enables to change the location of such directory
177 without relying on the destination directory access rights (except those that
178 are required for this operation, see `LANDLOCK_ACCESS_FS_REFER` documentation).
179 Having self-sufficient hierarchies also helps to tighten the required access
180 rights to the minimal set of data. This also helps avoid sinkhole directories,
181 i.e. directories where data can be linked to but not linked from. However,
182 this depends on data organization, which might not be controlled by developers.
183 In this case, granting read-write access to ``~/tmp/``, instead of write-only
184 access, would potentially allow to move ``~/tmp/`` to a non-readable directory
185 and still keep the ability to list the content of ``~/tmp/``.
187 Layers of file path access rights
188 ---------------------------------
190 Each time a thread enforces a ruleset on itself, it updates its Landlock domain
191 with a new layer of policy. Indeed, this complementary policy is stacked with
192 the potentially other rulesets already restricting this thread. A sandboxed
193 thread can then safely add more constraints to itself with a new enforced
196 One policy layer grants access to a file path if at least one of its rules
197 encountered on the path grants the access. A sandboxed thread can only access
198 a file path if all its enforced policy layers grant the access as well as all
199 the other system access controls (e.g. filesystem DAC, other LSM policies,
202 Bind mounts and OverlayFS
203 -------------------------
205 Landlock enables to restrict access to file hierarchies, which means that these
206 access rights can be propagated with bind mounts (cf.
207 Documentation/filesystems/sharedsubtree.rst) but not with
208 Documentation/filesystems/overlayfs.rst.
210 A bind mount mirrors a source file hierarchy to a destination. The destination
211 hierarchy is then composed of the exact same files, on which Landlock rules can
212 be tied, either via the source or the destination path. These rules restrict
213 access when they are encountered on a path, which means that they can restrict
214 access to multiple file hierarchies at the same time, whether these hierarchies
215 are the result of bind mounts or not.
217 An OverlayFS mount point consists of upper and lower layers. These layers are
218 combined in a merge directory, result of the mount point. This merge hierarchy
219 may include files from the upper and lower layers, but modifications performed
220 on the merge hierarchy only reflects on the upper layer. From a Landlock
221 policy point of view, each OverlayFS layers and merge hierarchies are
222 standalone and contains their own set of files and directories, which is
223 different from bind mounts. A policy restricting an OverlayFS layer will not
224 restrict the resulted merged hierarchy, and vice versa. Landlock users should
225 then only think about file hierarchies they want to allow access to, regardless
226 of the underlying filesystem.
231 Every new thread resulting from a :manpage:`clone(2)` inherits Landlock domain
232 restrictions from its parent. This is similar to the seccomp inheritance (cf.
233 Documentation/userspace-api/seccomp_filter.rst) or any other LSM dealing with
234 task's :manpage:`credentials(7)`. For instance, one process's thread may apply
235 Landlock rules to itself, but they will not be automatically applied to other
236 sibling threads (unlike POSIX thread credential changes, cf.
239 When a thread sandboxes itself, we have the guarantee that the related security
240 policy will stay enforced on all this thread's descendants. This allows
241 creating standalone and modular security policies per application, which will
242 automatically be composed between themselves according to their runtime parent
248 A sandboxed process has less privileges than a non-sandboxed process and must
249 then be subject to additional restrictions when manipulating another process.
250 To be allowed to use :manpage:`ptrace(2)` and related syscalls on a target
251 process, a sandboxed process should have a subset of the target process rules,
252 which means the tracee must be in a sub-domain of the tracer.
257 Backward and forward compatibility
258 ----------------------------------
260 Landlock is designed to be compatible with past and future versions of the
261 kernel. This is achieved thanks to the system call attributes and the
262 associated bitflags, particularly the ruleset's `handled_access_fs`. Making
263 handled access right explicit enables the kernel and user space to have a clear
264 contract with each other. This is required to make sure sandboxing will not
265 get stricter with a system update, which could break applications.
267 Developers can subscribe to the `Landlock mailing list
268 <https://subspace.kernel.org/lists.linux.dev.html>`_ to knowingly update and
269 test their applications with the latest available features. In the interest of
270 users, and because they may use different kernel versions, it is strongly
271 encouraged to follow a best-effort security approach by checking the Landlock
272 ABI version at runtime and only enforcing the supported features.
274 .. _landlock_abi_versions:
276 Landlock ABI versions
277 ---------------------
279 The Landlock ABI version can be read with the sys_landlock_create_ruleset()
286 abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
290 printf("Landlock is not supported by the current kernel.\n");
293 printf("Landlock is currently disabled.\n");
299 printf("Landlock supports LANDLOCK_ACCESS_FS_REFER.\n");
302 The following kernel interfaces are implicitly supported by the first ABI
303 version. Features only supported from a specific version are explicitly marked
312 .. kernel-doc:: include/uapi/linux/landlock.h
313 :identifiers: fs_access
315 Creating a new ruleset
316 ----------------------
318 .. kernel-doc:: security/landlock/syscalls.c
319 :identifiers: sys_landlock_create_ruleset
321 .. kernel-doc:: include/uapi/linux/landlock.h
322 :identifiers: landlock_ruleset_attr
327 .. kernel-doc:: security/landlock/syscalls.c
328 :identifiers: sys_landlock_add_rule
330 .. kernel-doc:: include/uapi/linux/landlock.h
331 :identifiers: landlock_rule_type landlock_path_beneath_attr
336 .. kernel-doc:: security/landlock/syscalls.c
337 :identifiers: sys_landlock_restrict_self
342 Filesystem topology modification
343 --------------------------------
345 As for file renaming and linking, a sandboxed thread cannot modify its
346 filesystem topology, whether via :manpage:`mount(2)` or
347 :manpage:`pivot_root(2)`. However, :manpage:`chroot(2)` calls are not denied.
352 Access to regular files and directories can be restricted by Landlock,
353 according to the handled accesses of a ruleset. However, files that do not
354 come from a user-visible filesystem (e.g. pipe, socket), but can still be
355 accessed through ``/proc/<pid>/fd/*``, cannot currently be explicitly
356 restricted. Likewise, some special kernel filesystems such as nsfs, which can
357 be accessed through ``/proc/<pid>/ns/*``, cannot currently be explicitly
358 restricted. However, thanks to the `ptrace restrictions`_, access to such
359 sensitive ``/proc`` files are automatically restricted according to domain
360 hierarchies. Future Landlock evolutions could still enable to explicitly
361 restrict such paths with dedicated ruleset flags.
366 There is a limit of 16 layers of stacked rulesets. This can be an issue for a
367 task willing to enforce a new ruleset in complement to its 16 inherited
368 rulesets. Once this limit is reached, sys_landlock_restrict_self() returns
369 E2BIG. It is then strongly suggested to carefully build rulesets once in the
370 life of a thread, especially for applications able to launch other applications
371 that may also want to sandbox themselves (e.g. shells, container managers,
377 Kernel memory allocated to create rulesets is accounted and can be restricted
378 by the Documentation/admin-guide/cgroup-v1/memory.rst.
383 File renaming and linking (ABI 1)
384 ---------------------------------
386 Because Landlock targets unprivileged access controls, it needs to properly
387 handle composition of rules. Such property also implies rules nesting.
388 Properly handling multiple layers of rulesets, each one of them able to
389 restrict access to files, also implies inheritance of the ruleset restrictions
390 from a parent to its hierarchy. Because files are identified and restricted by
391 their hierarchy, moving or linking a file from one directory to another implies
392 propagation of the hierarchy constraints, or restriction of these actions
393 according to the potentially lost constraints. To protect against privilege
394 escalations through renaming or linking, and for the sake of simplicity,
395 Landlock previously limited linking and renaming to the same directory.
396 Starting with the Landlock ABI version 2, it is now possible to securely
397 control renaming and linking thanks to the new `LANDLOCK_ACCESS_FS_REFER`
405 Landlock was first introduced in Linux 5.13 but it must be configured at build
406 time with `CONFIG_SECURITY_LANDLOCK=y`. Landlock must also be enabled at boot
407 time as the other security modules. The list of security modules enabled by
408 default is set with `CONFIG_LSM`. The kernel configuration should then
409 contains `CONFIG_LSM=landlock,[...]` with `[...]` as the list of other
410 potentially useful security modules for the running system (see the
413 If the running kernel doesn't have `landlock` in `CONFIG_LSM`, then we can
414 still enable it by adding ``lsm=landlock,[...]`` to
415 Documentation/admin-guide/kernel-parameters.rst thanks to the bootloader
418 Questions and answers
419 =====================
421 What about user space sandbox managers?
422 ---------------------------------------
424 Using user space process to enforce restrictions on kernel resources can lead
425 to race conditions or inconsistent evaluations (i.e. `Incorrect mirroring of
426 the OS code and state
427 <https://www.ndss-symposium.org/ndss2003/traps-and-pitfalls-practical-problems-system-call-interposition-based-security-tools/>`_).
429 What about namespaces and containers?
430 -------------------------------------
432 Namespaces can help create sandboxes but they are not designed for
433 access-control and then miss useful features for such use case (e.g. no
434 fine-grained restrictions). Moreover, their complexity can lead to security
435 issues, especially when untrusted processes can manipulate them (cf.
436 `Controlling access to user namespaces <https://lwn.net/Articles/673597/>`_).
438 Additional documentation
439 ========================
441 * Documentation/security/landlock.rst
442 * https://landlock.io
445 .. _samples/landlock/sandboxer.c:
446 https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/samples/landlock/sandboxer.c