xref: /linux/Documentation/userspace-api/landlock.rst (revision f14aa5ea415b8add245e976bfab96a12986c6843)
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
5
6=====================================
7Landlock: unprivileged access control
8=====================================
9
10:Author: Mickaël Salaün
11:Date: April 2024
12
13The goal of Landlock is to enable to restrict ambient rights (e.g. global
14filesystem or network access) for a set of processes.  Because Landlock
15is a stackable LSM, it makes possible to create safe security sandboxes as new
16security layers in addition to the existing system-wide access-controls. This
17kind of sandbox is expected to help mitigate the security impact of bugs or
18unexpected/malicious behaviors in user space applications.  Landlock empowers
19any process, including unprivileged ones, to securely restrict themselves.
20
21We can quickly make sure that Landlock is enabled in the running system by
22looking for "landlock: Up and running" in kernel logs (as root):
23``dmesg | grep landlock || journalctl -kb -g landlock`` .
24Developers can also easily check for Landlock support with a
25:ref:`related system call <landlock_abi_versions>`.
26If Landlock is not currently supported, we need to
27:ref:`configure the kernel appropriately <kernel_support>`.
28
29Landlock rules
30==============
31
32A Landlock rule describes an action on an object which the process intends to
33perform.  A set of rules is aggregated in a ruleset, which can then restrict
34the thread enforcing it, and its future children.
35
36The two existing types of rules are:
37
38Filesystem rules
39    For these rules, the object is a file hierarchy,
40    and the related filesystem actions are defined with
41    `filesystem access rights`.
42
43Network rules (since ABI v4)
44    For these rules, the object is a TCP port,
45    and the related actions are defined with `network access rights`.
46
47Defining and enforcing a security policy
48----------------------------------------
49
50We first need to define the ruleset that will contain our rules.
51
52For this example, the ruleset will contain rules that only allow filesystem
53read actions and establish a specific TCP connection. Filesystem write
54actions and other TCP actions will be denied.
55
56The ruleset then needs to handle both these kinds of actions.  This is
57required for backward and forward compatibility (i.e. the kernel and user
58space may not know each other's supported restrictions), hence the need
59to be explicit about the denied-by-default access rights.
60
61.. code-block:: c
62
63    struct landlock_ruleset_attr ruleset_attr = {
64        .handled_access_fs =
65            LANDLOCK_ACCESS_FS_EXECUTE |
66            LANDLOCK_ACCESS_FS_WRITE_FILE |
67            LANDLOCK_ACCESS_FS_READ_FILE |
68            LANDLOCK_ACCESS_FS_READ_DIR |
69            LANDLOCK_ACCESS_FS_REMOVE_DIR |
70            LANDLOCK_ACCESS_FS_REMOVE_FILE |
71            LANDLOCK_ACCESS_FS_MAKE_CHAR |
72            LANDLOCK_ACCESS_FS_MAKE_DIR |
73            LANDLOCK_ACCESS_FS_MAKE_REG |
74            LANDLOCK_ACCESS_FS_MAKE_SOCK |
75            LANDLOCK_ACCESS_FS_MAKE_FIFO |
76            LANDLOCK_ACCESS_FS_MAKE_BLOCK |
77            LANDLOCK_ACCESS_FS_MAKE_SYM |
78            LANDLOCK_ACCESS_FS_REFER |
79            LANDLOCK_ACCESS_FS_TRUNCATE |
80            LANDLOCK_ACCESS_FS_IOCTL_DEV,
81        .handled_access_net =
82            LANDLOCK_ACCESS_NET_BIND_TCP |
83            LANDLOCK_ACCESS_NET_CONNECT_TCP,
84    };
85
86Because we may not know on which kernel version an application will be
87executed, it is safer to follow a best-effort security approach.  Indeed, we
88should try to protect users as much as possible whatever the kernel they are
89using.
90
91To be compatible with older Linux versions, we detect the available Landlock ABI
92version, and only use the available subset of access rights:
93
94.. code-block:: c
95
96    int abi;
97
98    abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
99    if (abi < 0) {
100        /* Degrades gracefully if Landlock is not handled. */
101        perror("The running kernel does not enable to use Landlock");
102        return 0;
103    }
104    switch (abi) {
105    case 1:
106        /* Removes LANDLOCK_ACCESS_FS_REFER for ABI < 2 */
107        ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_REFER;
108        __attribute__((fallthrough));
109    case 2:
110        /* Removes LANDLOCK_ACCESS_FS_TRUNCATE for ABI < 3 */
111        ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_TRUNCATE;
112        __attribute__((fallthrough));
113    case 3:
114        /* Removes network support for ABI < 4 */
115        ruleset_attr.handled_access_net &=
116            ~(LANDLOCK_ACCESS_NET_BIND_TCP |
117              LANDLOCK_ACCESS_NET_CONNECT_TCP);
118        __attribute__((fallthrough));
119    case 4:
120        /* Removes LANDLOCK_ACCESS_FS_IOCTL_DEV for ABI < 5 */
121        ruleset_attr.handled_access_fs &= ~LANDLOCK_ACCESS_FS_IOCTL_DEV;
122    }
123
124This enables to create an inclusive ruleset that will contain our rules.
125
126.. code-block:: c
127
128    int ruleset_fd;
129
130    ruleset_fd = landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0);
131    if (ruleset_fd < 0) {
132        perror("Failed to create a ruleset");
133        return 1;
134    }
135
136We can now add a new rule to this ruleset thanks to the returned file
137descriptor referring to this ruleset.  The rule will only allow reading the
138file hierarchy ``/usr``.  Without another rule, write actions would then be
139denied by the ruleset.  To add ``/usr`` to the ruleset, we open it with the
140``O_PATH`` flag and fill the &struct landlock_path_beneath_attr with this file
141descriptor.
142
143.. code-block:: c
144
145    int err;
146    struct landlock_path_beneath_attr path_beneath = {
147        .allowed_access =
148            LANDLOCK_ACCESS_FS_EXECUTE |
149            LANDLOCK_ACCESS_FS_READ_FILE |
150            LANDLOCK_ACCESS_FS_READ_DIR,
151    };
152
153    path_beneath.parent_fd = open("/usr", O_PATH | O_CLOEXEC);
154    if (path_beneath.parent_fd < 0) {
155        perror("Failed to open file");
156        close(ruleset_fd);
157        return 1;
158    }
159    err = landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH,
160                            &path_beneath, 0);
161    close(path_beneath.parent_fd);
162    if (err) {
163        perror("Failed to update ruleset");
164        close(ruleset_fd);
165        return 1;
166    }
167
168It may also be required to create rules following the same logic as explained
169for the ruleset creation, by filtering access rights according to the Landlock
170ABI version.  In this example, this is not required because all of the requested
171``allowed_access`` rights are already available in ABI 1.
172
173For network access-control, we can add a set of rules that allow to use a port
174number for a specific action: HTTPS connections.
175
176.. code-block:: c
177
178    struct landlock_net_port_attr net_port = {
179        .allowed_access = LANDLOCK_ACCESS_NET_CONNECT_TCP,
180        .port = 443,
181    };
182
183    err = landlock_add_rule(ruleset_fd, LANDLOCK_RULE_NET_PORT,
184                            &net_port, 0);
185
186The next step is to restrict the current thread from gaining more privileges
187(e.g. through a SUID binary).  We now have a ruleset with the first rule
188allowing read access to ``/usr`` while denying all other handled accesses for
189the filesystem, and a second rule allowing HTTPS connections.
190
191.. code-block:: c
192
193    if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
194        perror("Failed to restrict privileges");
195        close(ruleset_fd);
196        return 1;
197    }
198
199The current thread is now ready to sandbox itself with the ruleset.
200
201.. code-block:: c
202
203    if (landlock_restrict_self(ruleset_fd, 0)) {
204        perror("Failed to enforce ruleset");
205        close(ruleset_fd);
206        return 1;
207    }
208    close(ruleset_fd);
209
210If the ``landlock_restrict_self`` system call succeeds, the current thread is
211now restricted and this policy will be enforced on all its subsequently created
212children as well.  Once a thread is landlocked, there is no way to remove its
213security policy; only adding more restrictions is allowed.  These threads are
214now in a new Landlock domain, merge of their parent one (if any) with the new
215ruleset.
216
217Full working code can be found in `samples/landlock/sandboxer.c`_.
218
219Good practices
220--------------
221
222It is recommended setting access rights to file hierarchy leaves as much as
223possible.  For instance, it is better to be able to have ``~/doc/`` as a
224read-only hierarchy and ``~/tmp/`` as a read-write hierarchy, compared to
225``~/`` as a read-only hierarchy and ``~/tmp/`` as a read-write hierarchy.
226Following this good practice leads to self-sufficient hierarchies that do not
227depend on their location (i.e. parent directories).  This is particularly
228relevant when we want to allow linking or renaming.  Indeed, having consistent
229access rights per directory enables to change the location of such directory
230without relying on the destination directory access rights (except those that
231are required for this operation, see ``LANDLOCK_ACCESS_FS_REFER``
232documentation).
233
234Having self-sufficient hierarchies also helps to tighten the required access
235rights to the minimal set of data.  This also helps avoid sinkhole directories,
236i.e.  directories where data can be linked to but not linked from.  However,
237this depends on data organization, which might not be controlled by developers.
238In this case, granting read-write access to ``~/tmp/``, instead of write-only
239access, would potentially allow to move ``~/tmp/`` to a non-readable directory
240and still keep the ability to list the content of ``~/tmp/``.
241
242Layers of file path access rights
243---------------------------------
244
245Each time a thread enforces a ruleset on itself, it updates its Landlock domain
246with a new layer of policy.  Indeed, this complementary policy is stacked with
247the potentially other rulesets already restricting this thread.  A sandboxed
248thread can then safely add more constraints to itself with a new enforced
249ruleset.
250
251One policy layer grants access to a file path if at least one of its rules
252encountered on the path grants the access.  A sandboxed thread can only access
253a file path if all its enforced policy layers grant the access as well as all
254the other system access controls (e.g. filesystem DAC, other LSM policies,
255etc.).
256
257Bind mounts and OverlayFS
258-------------------------
259
260Landlock enables to restrict access to file hierarchies, which means that these
261access rights can be propagated with bind mounts (cf.
262Documentation/filesystems/sharedsubtree.rst) but not with
263Documentation/filesystems/overlayfs.rst.
264
265A bind mount mirrors a source file hierarchy to a destination.  The destination
266hierarchy is then composed of the exact same files, on which Landlock rules can
267be tied, either via the source or the destination path.  These rules restrict
268access when they are encountered on a path, which means that they can restrict
269access to multiple file hierarchies at the same time, whether these hierarchies
270are the result of bind mounts or not.
271
272An OverlayFS mount point consists of upper and lower layers.  These layers are
273combined in a merge directory, result of the mount point.  This merge hierarchy
274may include files from the upper and lower layers, but modifications performed
275on the merge hierarchy only reflects on the upper layer.  From a Landlock
276policy point of view, each OverlayFS layers and merge hierarchies are
277standalone and contains their own set of files and directories, which is
278different from bind mounts.  A policy restricting an OverlayFS layer will not
279restrict the resulted merged hierarchy, and vice versa.  Landlock users should
280then only think about file hierarchies they want to allow access to, regardless
281of the underlying filesystem.
282
283Inheritance
284-----------
285
286Every new thread resulting from a :manpage:`clone(2)` inherits Landlock domain
287restrictions from its parent.  This is similar to the seccomp inheritance (cf.
288Documentation/userspace-api/seccomp_filter.rst) or any other LSM dealing with
289task's :manpage:`credentials(7)`.  For instance, one process's thread may apply
290Landlock rules to itself, but they will not be automatically applied to other
291sibling threads (unlike POSIX thread credential changes, cf.
292:manpage:`nptl(7)`).
293
294When a thread sandboxes itself, we have the guarantee that the related security
295policy will stay enforced on all this thread's descendants.  This allows
296creating standalone and modular security policies per application, which will
297automatically be composed between themselves according to their runtime parent
298policies.
299
300Ptrace restrictions
301-------------------
302
303A sandboxed process has less privileges than a non-sandboxed process and must
304then be subject to additional restrictions when manipulating another process.
305To be allowed to use :manpage:`ptrace(2)` and related syscalls on a target
306process, a sandboxed process should have a subset of the target process rules,
307which means the tracee must be in a sub-domain of the tracer.
308
309Truncating files
310----------------
311
312The operations covered by ``LANDLOCK_ACCESS_FS_WRITE_FILE`` and
313``LANDLOCK_ACCESS_FS_TRUNCATE`` both change the contents of a file and sometimes
314overlap in non-intuitive ways.  It is recommended to always specify both of
315these together.
316
317A particularly surprising example is :manpage:`creat(2)`.  The name suggests
318that this system call requires the rights to create and write files.  However,
319it also requires the truncate right if an existing file under the same name is
320already present.
321
322It should also be noted that truncating files does not require the
323``LANDLOCK_ACCESS_FS_WRITE_FILE`` right.  Apart from the :manpage:`truncate(2)`
324system call, this can also be done through :manpage:`open(2)` with the flags
325``O_RDONLY | O_TRUNC``.
326
327The truncate right is associated with the opened file (see below).
328
329Rights associated with file descriptors
330---------------------------------------
331
332When opening a file, the availability of the ``LANDLOCK_ACCESS_FS_TRUNCATE`` and
333``LANDLOCK_ACCESS_FS_IOCTL_DEV`` rights is associated with the newly created
334file descriptor and will be used for subsequent truncation and ioctl attempts
335using :manpage:`ftruncate(2)` and :manpage:`ioctl(2)`.  The behavior is similar
336to opening a file for reading or writing, where permissions are checked during
337:manpage:`open(2)`, but not during the subsequent :manpage:`read(2)` and
338:manpage:`write(2)` calls.
339
340As a consequence, it is possible that a process has multiple open file
341descriptors referring to the same file, but Landlock enforces different things
342when operating with these file descriptors.  This can happen when a Landlock
343ruleset gets enforced and the process keeps file descriptors which were opened
344both before and after the enforcement.  It is also possible to pass such file
345descriptors between processes, keeping their Landlock properties, even when some
346of the involved processes do not have an enforced Landlock ruleset.
347
348Compatibility
349=============
350
351Backward and forward compatibility
352----------------------------------
353
354Landlock is designed to be compatible with past and future versions of the
355kernel.  This is achieved thanks to the system call attributes and the
356associated bitflags, particularly the ruleset's ``handled_access_fs``.  Making
357handled access right explicit enables the kernel and user space to have a clear
358contract with each other.  This is required to make sure sandboxing will not
359get stricter with a system update, which could break applications.
360
361Developers can subscribe to the `Landlock mailing list
362<https://subspace.kernel.org/lists.linux.dev.html>`_ to knowingly update and
363test their applications with the latest available features.  In the interest of
364users, and because they may use different kernel versions, it is strongly
365encouraged to follow a best-effort security approach by checking the Landlock
366ABI version at runtime and only enforcing the supported features.
367
368.. _landlock_abi_versions:
369
370Landlock ABI versions
371---------------------
372
373The Landlock ABI version can be read with the sys_landlock_create_ruleset()
374system call:
375
376.. code-block:: c
377
378    int abi;
379
380    abi = landlock_create_ruleset(NULL, 0, LANDLOCK_CREATE_RULESET_VERSION);
381    if (abi < 0) {
382        switch (errno) {
383        case ENOSYS:
384            printf("Landlock is not supported by the current kernel.\n");
385            break;
386        case EOPNOTSUPP:
387            printf("Landlock is currently disabled.\n");
388            break;
389        }
390        return 0;
391    }
392    if (abi >= 2) {
393        printf("Landlock supports LANDLOCK_ACCESS_FS_REFER.\n");
394    }
395
396The following kernel interfaces are implicitly supported by the first ABI
397version.  Features only supported from a specific version are explicitly marked
398as such.
399
400Kernel interface
401================
402
403Access rights
404-------------
405
406.. kernel-doc:: include/uapi/linux/landlock.h
407    :identifiers: fs_access net_access
408
409Creating a new ruleset
410----------------------
411
412.. kernel-doc:: security/landlock/syscalls.c
413    :identifiers: sys_landlock_create_ruleset
414
415.. kernel-doc:: include/uapi/linux/landlock.h
416    :identifiers: landlock_ruleset_attr
417
418Extending a ruleset
419-------------------
420
421.. kernel-doc:: security/landlock/syscalls.c
422    :identifiers: sys_landlock_add_rule
423
424.. kernel-doc:: include/uapi/linux/landlock.h
425    :identifiers: landlock_rule_type landlock_path_beneath_attr
426                  landlock_net_port_attr
427
428Enforcing a ruleset
429-------------------
430
431.. kernel-doc:: security/landlock/syscalls.c
432    :identifiers: sys_landlock_restrict_self
433
434Current limitations
435===================
436
437Filesystem topology modification
438--------------------------------
439
440Threads sandboxed with filesystem restrictions cannot modify filesystem
441topology, whether via :manpage:`mount(2)` or :manpage:`pivot_root(2)`.
442However, :manpage:`chroot(2)` calls are not denied.
443
444Special filesystems
445-------------------
446
447Access to regular files and directories can be restricted by Landlock,
448according to the handled accesses of a ruleset.  However, files that do not
449come from a user-visible filesystem (e.g. pipe, socket), but can still be
450accessed through ``/proc/<pid>/fd/*``, cannot currently be explicitly
451restricted.  Likewise, some special kernel filesystems such as nsfs, which can
452be accessed through ``/proc/<pid>/ns/*``, cannot currently be explicitly
453restricted.  However, thanks to the `ptrace restrictions`_, access to such
454sensitive ``/proc`` files are automatically restricted according to domain
455hierarchies.  Future Landlock evolutions could still enable to explicitly
456restrict such paths with dedicated ruleset flags.
457
458Ruleset layers
459--------------
460
461There is a limit of 16 layers of stacked rulesets.  This can be an issue for a
462task willing to enforce a new ruleset in complement to its 16 inherited
463rulesets.  Once this limit is reached, sys_landlock_restrict_self() returns
464E2BIG.  It is then strongly suggested to carefully build rulesets once in the
465life of a thread, especially for applications able to launch other applications
466that may also want to sandbox themselves (e.g. shells, container managers,
467etc.).
468
469Memory usage
470------------
471
472Kernel memory allocated to create rulesets is accounted and can be restricted
473by the Documentation/admin-guide/cgroup-v1/memory.rst.
474
475IOCTL support
476-------------
477
478The ``LANDLOCK_ACCESS_FS_IOCTL_DEV`` right restricts the use of
479:manpage:`ioctl(2)`, but it only applies to *newly opened* device files.  This
480means specifically that pre-existing file descriptors like stdin, stdout and
481stderr are unaffected.
482
483Users should be aware that TTY devices have traditionally permitted to control
484other processes on the same TTY through the ``TIOCSTI`` and ``TIOCLINUX`` IOCTL
485commands.  Both of these require ``CAP_SYS_ADMIN`` on modern Linux systems, but
486the behavior is configurable for ``TIOCSTI``.
487
488On older systems, it is therefore recommended to close inherited TTY file
489descriptors, or to reopen them from ``/proc/self/fd/*`` without the
490``LANDLOCK_ACCESS_FS_IOCTL_DEV`` right, if possible.
491
492Landlock's IOCTL support is coarse-grained at the moment, but may become more
493fine-grained in the future.  Until then, users are advised to establish the
494guarantees that they need through the file hierarchy, by only allowing the
495``LANDLOCK_ACCESS_FS_IOCTL_DEV`` right on files where it is really required.
496
497Previous limitations
498====================
499
500File renaming and linking (ABI < 2)
501-----------------------------------
502
503Because Landlock targets unprivileged access controls, it needs to properly
504handle composition of rules.  Such property also implies rules nesting.
505Properly handling multiple layers of rulesets, each one of them able to
506restrict access to files, also implies inheritance of the ruleset restrictions
507from a parent to its hierarchy.  Because files are identified and restricted by
508their hierarchy, moving or linking a file from one directory to another implies
509propagation of the hierarchy constraints, or restriction of these actions
510according to the potentially lost constraints.  To protect against privilege
511escalations through renaming or linking, and for the sake of simplicity,
512Landlock previously limited linking and renaming to the same directory.
513Starting with the Landlock ABI version 2, it is now possible to securely
514control renaming and linking thanks to the new ``LANDLOCK_ACCESS_FS_REFER``
515access right.
516
517File truncation (ABI < 3)
518-------------------------
519
520File truncation could not be denied before the third Landlock ABI, so it is
521always allowed when using a kernel that only supports the first or second ABI.
522
523Starting with the Landlock ABI version 3, it is now possible to securely control
524truncation thanks to the new ``LANDLOCK_ACCESS_FS_TRUNCATE`` access right.
525
526Network support (ABI < 4)
527-------------------------
528
529Starting with the Landlock ABI version 4, it is now possible to restrict TCP
530bind and connect actions to only a set of allowed ports thanks to the new
531``LANDLOCK_ACCESS_NET_BIND_TCP`` and ``LANDLOCK_ACCESS_NET_CONNECT_TCP``
532access rights.
533
534IOCTL (ABI < 5)
535---------------
536
537IOCTL operations could not be denied before the fifth Landlock ABI, so
538:manpage:`ioctl(2)` is always allowed when using a kernel that only supports an
539earlier ABI.
540
541Starting with the Landlock ABI version 5, it is possible to restrict the use of
542:manpage:`ioctl(2)` using the new ``LANDLOCK_ACCESS_FS_IOCTL_DEV`` right.
543
544.. _kernel_support:
545
546Kernel support
547==============
548
549Build time configuration
550------------------------
551
552Landlock was first introduced in Linux 5.13 but it must be configured at build
553time with ``CONFIG_SECURITY_LANDLOCK=y``.  Landlock must also be enabled at boot
554time as the other security modules.  The list of security modules enabled by
555default is set with ``CONFIG_LSM``.  The kernel configuration should then
556contains ``CONFIG_LSM=landlock,[...]`` with ``[...]``  as the list of other
557potentially useful security modules for the running system (see the
558``CONFIG_LSM`` help).
559
560Boot time configuration
561-----------------------
562
563If the running kernel does not have ``landlock`` in ``CONFIG_LSM``, then we can
564enable Landlock by adding ``lsm=landlock,[...]`` to
565Documentation/admin-guide/kernel-parameters.rst in the boot loader
566configuration.
567
568For example, if the current built-in configuration is:
569
570.. code-block:: console
571
572    $ zgrep -h "^CONFIG_LSM=" "/boot/config-$(uname -r)" /proc/config.gz 2>/dev/null
573    CONFIG_LSM="lockdown,yama,integrity,apparmor"
574
575...and if the cmdline doesn't contain ``landlock`` either:
576
577.. code-block:: console
578
579    $ sed -n 's/.*\(\<lsm=\S\+\).*/\1/p' /proc/cmdline
580    lsm=lockdown,yama,integrity,apparmor
581
582...we should configure the boot loader to set a cmdline extending the ``lsm``
583list with the ``landlock,`` prefix::
584
585  lsm=landlock,lockdown,yama,integrity,apparmor
586
587After a reboot, we can check that Landlock is up and running by looking at
588kernel logs:
589
590.. code-block:: console
591
592    # dmesg | grep landlock || journalctl -kb -g landlock
593    [    0.000000] Command line: [...] lsm=landlock,lockdown,yama,integrity,apparmor
594    [    0.000000] Kernel command line: [...] lsm=landlock,lockdown,yama,integrity,apparmor
595    [    0.000000] LSM: initializing lsm=lockdown,capability,landlock,yama,integrity,apparmor
596    [    0.000000] landlock: Up and running.
597
598The kernel may be configured at build time to always load the ``lockdown`` and
599``capability`` LSMs.  In that case, these LSMs will appear at the beginning of
600the ``LSM: initializing`` log line as well, even if they are not configured in
601the boot loader.
602
603Network support
604---------------
605
606To be able to explicitly allow TCP operations (e.g., adding a network rule with
607``LANDLOCK_ACCESS_NET_BIND_TCP``), the kernel must support TCP
608(``CONFIG_INET=y``).  Otherwise, sys_landlock_add_rule() returns an
609``EAFNOSUPPORT`` error, which can safely be ignored because this kind of TCP
610operation is already not possible.
611
612Questions and answers
613=====================
614
615What about user space sandbox managers?
616---------------------------------------
617
618Using user space process to enforce restrictions on kernel resources can lead
619to race conditions or inconsistent evaluations (i.e. `Incorrect mirroring of
620the OS code and state
621<https://www.ndss-symposium.org/ndss2003/traps-and-pitfalls-practical-problems-system-call-interposition-based-security-tools/>`_).
622
623What about namespaces and containers?
624-------------------------------------
625
626Namespaces can help create sandboxes but they are not designed for
627access-control and then miss useful features for such use case (e.g. no
628fine-grained restrictions).  Moreover, their complexity can lead to security
629issues, especially when untrusted processes can manipulate them (cf.
630`Controlling access to user namespaces <https://lwn.net/Articles/673597/>`_).
631
632Additional documentation
633========================
634
635* Documentation/security/landlock.rst
636* https://landlock.io
637
638.. Links
639.. _samples/landlock/sandboxer.c:
640   https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/samples/landlock/sandboxer.c
641