* [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT
@ 2025-12-14 17:05 Justin Suess
2025-12-14 17:05 ` [PATCH v5 1/6] landlock: " Justin Suess
` (5 more replies)
0 siblings, 6 replies; 9+ messages in thread
From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw)
To: Mickaël Salaün
Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara,
Abhinav Saxena, linux-security-module
Hi,
This is version 5 of the LANDLOCK_ADD_RULE_NO_INHERIT series, which
implements a new flag to suppress inheritance of access rights and
flags from parent objects.
This version of the series focuses on simplification and cleanup.
Behavior of the flag is identical to the previous patch.
This version has 1136 insertions(+), 72 deletions(-), down from
1285 insertions(+), 9 deletions(-), while adding documentation
and retaining all existing tests.
This series is still based on v6 of Tingmao Wang's "quiet flag" series.
Previous patch summary:
The new flag enables policies where a parent directory needs broader
access than its children. For example, a sandbox may permit read-write
access to /home/user but still prohibit writes to ~/.bashrc or
~/.ssh, even though they are nested beneath the parent. Today this is
not possible because access rights always propagate from parent to
child inodes.
When a rule is added with LANDLOCK_ADD_RULE_NO_INHERIT:
* access rights on parent inodes are ignored for that inode and its
descendants; and
* operations that change the filesystem ancestors of such objects
(via rename, rmdir, link) are denied up to the VFS root.
* parent flags do not propagate below a NO_INHERIT rule.
These parent-directory restrictions help mitigate sandbox-restart
attacks: a sandboxed process could otherwise move a protected
directory before exit, causing the next sandbox instance to apply its
policy to the wrong path.
Changes since v4:
1. Trimmed ~120 lines from core implementation in fs.c.
2. Centralized path traversal logic with a helper function
landlock_walk_path_up.
3. Fixed bug in test on applying LANDLOCK_ADD_RULE_NO_INHERIT on
a file, giving it valid access rights.
4. Restructured commits to allow independent builds.
5. Adds userspace API documentation for the flag.
Changes since v3:
1. Trimmed core implementation in fs.c by removing redundant functions.
2. Fixed placement/inclusion of prototypes.
3. Added 4 new selftests for bind mount cases.
4. Protections now apply up to the VFS root instead of the mountpoint
root.
Changes since v2:
1. Add six new selftests for the new flag.
2. Add an optimization to stop permission harvesting when all
relevant layers are tagged with NO_INHERIT.
3. Suppress inheritance of parent flags.
4. Rebase onto v5 of the quiet-flag series.
5. Remove the xarray structure used for flag tracking in favor of
blank rule insertion, simplifying the implementation.
6. Fix edge cases involving flag inheritance across multiple
NO_INHERIT layers.
7. Add documenting comments to new functions.
Links:
v1:
https://lore.kernel.org/linux-security-module/20251105180019.1432367-1-utilityemal77@gmail.com/
v2:
https://lore.kernel.org/linux-security-module/20251120222346.1157004-1-utilityemal77@gmail.com/
v3:
https://lore.kernel.org/linux-security-module/20251126122039.3832162-1-utilityemal77@gmail.com/
v4:
https://lore.kernel.org/linux-security-module/20251207015132.800576-1-utilityemal77@gmail.com/
quiet-flag v6:
https://lore.kernel.org/linux-security-module/cover.1765040503.git.m@maowtm.org/
Example usage:
# LL_FS_RO="/a/b/c" LL_FS_RW="/" LL_FS_NO_INHERIT="/a/b/c"
landlock-sandboxer sh
# touch /a/b/c/fi # denied; / RW does not inherit
# rmdir /a/b/c # denied by ancestor protections
# mv /a /bad # denied
# mkdir /a/good; touch /a/good/fi # allowed; unrelated path
About 120 lines of code have been removed from the fs.c file, achieved by
removing/streamlining many of the previous functions, and adding shared
path traversal logic.
Simplifying the path handling has a nice side effect of making some hairy
functions (is_access_to_paths_allowed) more readable.
All tests added by this series, and all other existing landlock tests,
are passing. This patch was also validated through checkpatch.pl.
Special thanks to Tingmao Wang and Mickaël Salaün for your valuable
feedback.
Thank you for your time and review.
Regards,
Justin Suess
Justin Suess (6):
landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT
landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT userspace api
samples/landlock: Add LANDLOCK_ADD_RULE_NO_INHERIT to
landlock-sandboxer
selftests/landlock: Implement selftests for
LANDLOCK_ADD_RULE_NO_INHERIT
landlock: Implement KUnit test for LANDLOCK_ADD_RULE_NO_INHERIT
landlock: Add documentation for LANDLOCK_ADD_RULE_NO_INHERIT
Documentation/userspace-api/landlock.rst | 17 +
include/uapi/linux/landlock.h | 29 +
samples/landlock/sandboxer.c | 13 +-
security/landlock/fs.c | 266 ++++++--
security/landlock/ruleset.c | 108 ++-
security/landlock/ruleset.h | 29 +-
security/landlock/syscalls.c | 16 +-
tools/testing/selftests/landlock/fs_test.c | 730 +++++++++++++++++++++
8 files changed, 1136 insertions(+), 72 deletions(-)
base-commit: 92f98eb2cc08c6e2d093d4682f1cd1204728e97e
--
2.51.0
^ permalink raw reply [flat|nested] 9+ messages in thread* [PATCH v5 1/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess @ 2025-12-14 17:05 ` Justin Suess 2025-12-14 22:53 ` Tingmao Wang 2025-12-14 17:05 ` [PATCH v5 2/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT userspace api Justin Suess ` (4 subsequent siblings) 5 siblings, 1 reply; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Implements a flag to prevent access grant inheritance within the filesystem hierarchy for landlock rules. If a landlock rule on an inode has this flag, any access grants on parent inodes will be ignored. Moreover, operations that involve altering the ancestors of the subject with LANDLOCK_ADD_RULE_NO_INHERIT will be denied up to the VFS root. For example, if /a/b/c/ = read only + LANDLOCK_ADD_RULE_NO_INHERIT and / = read write, writes to files in /a/b/c will be denied. Moreover, moving /a to /bad, removing /a/b/c, or creating links to /a will be prohibited. Parent flag inheritance is automatically suppressed by the permission harvesting logic, which will finish processing early if all relevant layers are tagged with NO_INHERIT. The parent directory restrictions mitigate sandbox-restart attacks. For example, if a sandboxed program is able to move a LANDLOCK_ADD_RULE_NO_INHERIT restricted directory, upon sandbox restart, the policy applied naively on the same filenames would be invalid. Preventing these operations mitigates these attacks. Because this requires path walking, to centralize the logic in a single location, the path walking logic is moved into a helper function landlock_walk_path_up, which takes a path as a parameter, and returns an enum corresponding to whether the path is a mount root, real root, or other. Signed-off-by: Justin Suess <utilityemal77@gmail.com> Cc: Tingmao Wang <m@maowtm.org> Cc: Mickaël Salaün <mic@digikod.net> --- v4..v5 changes: * Centralized path walking logic with landlock_walk_path_up. * Removed redundant functions in fs.c, and streamlined core logic, removing ~120 lines of code. * Removed mark_no_inherit_ancestors, replacing with direct flag setting in append_fs_rule. * Removed micro-optimization of skipping ancestor processing when all layers have no_inherit, as it complicated the code significantly for little gain. v3..v4 changes: * Rebased on v6 of Tingmao Wang's "quiet flag" series. * Removed unnecessary mask_no_inherit_descendant_layers and related code at Tingmao Wang's suggestion, simplifying patch. * Updated to use new disconnected directory handling. * Improved WARN_ON_ONCE usage. (Thanks Tingmao Wang!) * Removed redundant loop for single-layer rulesets (again thanks Tingmao Wang!) * Protections now apply up to the VFS root, not just the mountpoint. * Indentation fixes. * Removed redundant flag marker blocked_flag_masks. v2..v3 changes: * Parent directory topology protections now work by lazily inserting blank rules on parent inodes if they do not exist. This replaces the previous xarray implementation with simplified logic. * Added an optimization to skip further processing if all layers collected have no inherit. * Added support to block flag inheritance. include/uapi/linux/landlock.h | 29 ++++ security/landlock/fs.c | 266 +++++++++++++++++++++++++--------- security/landlock/ruleset.h | 29 +++- 3 files changed, 258 insertions(+), 66 deletions(-) diff --git a/include/uapi/linux/landlock.h b/include/uapi/linux/landlock.h index d4f47d20361a..6ab3e7bd1c81 100644 --- a/include/uapi/linux/landlock.h +++ b/include/uapi/linux/landlock.h @@ -127,10 +127,39 @@ struct landlock_ruleset_attr { * allowed_access in the passed in rule_attr. When this flag is * present, the caller is also allowed to pass in an empty * allowed_access. + * %LANDLOCK_ADD_RULE_NO_INHERIT + * When set on a rule being added to a ruleset, this flag disables the + * inheritance of access rights and flags from parent objects. + * + * This flag currently applies only to filesystem rules. Adding it to + * non-filesystem rules will return -EINVAL, unless future extensions + * of Landlock define other hierarchical object types. + * + * By default, Landlock filesystem rules inherit allowed accesses from + * ancestor directories: if a parent directory grants certain rights, + * those rights also apply to its children. A rule marked with + * LANDLOCK_ADD_RULE_NO_INHERIT stops this propagation at the directory + * covered by the rule. Descendants of that directory continue to inherit + * normally unless they also have rules using this flag. + * + * If a regular file is marked with this flag, it will not inherit any + * access rights from its parent directories; only the accesses explicitly + * allowed by the rule will apply to that file. + * + * This flag also enforces parent-directory restrictions: rename, rmdir, + * link, and other operations that would change the directory's immediate + * parent subtree are denied up to the VFS root. This prevents + * sandboxed processes from manipulating the filesystem hierarchy to evade + * restrictions (e.g., via sandbox-restart attacks). + * + * In addition, this flag blocks the inheritance of rule-layer flags + * (such as the quiet flag) from parent directories to the object covered + * by this rule. */ /* clang-format off */ #define LANDLOCK_ADD_RULE_QUIET (1U << 0) +#define LANDLOCK_ADD_RULE_NO_INHERIT (1U << 1) /* clang-format on */ /** diff --git a/security/landlock/fs.c b/security/landlock/fs.c index 0b589263ea42..8d8623ea857f 100644 --- a/security/landlock/fs.c +++ b/security/landlock/fs.c @@ -317,6 +317,37 @@ static struct landlock_object *get_inode_object(struct inode *const inode) LANDLOCK_ACCESS_FS_IOCTL_DEV) /* clang-format on */ +enum landlock_walk_result { + LANDLOCK_WALK_CONTINUE, + LANDLOCK_WALK_STOP_REAL_ROOT, + LANDLOCK_WALK_MOUNT_ROOT, +}; + +static enum landlock_walk_result landlock_walk_path_up(struct path *const path) +{ + while (path->dentry == path->mnt->mnt_root) { + if (!follow_up(path)) + return LANDLOCK_WALK_STOP_REAL_ROOT; + } + + if (unlikely(IS_ROOT(path->dentry))) { + if (likely(path->mnt->mnt_flags & MNT_INTERNAL)) + return LANDLOCK_WALK_MOUNT_ROOT; + dput(path->dentry); + path->dentry = dget(path->mnt->mnt_root); + return LANDLOCK_WALK_CONTINUE; + } + + struct dentry *const parent = dget_parent(path->dentry); + + dput(path->dentry); + path->dentry = parent; + return LANDLOCK_WALK_CONTINUE; +} + +static const struct landlock_rule *find_rule(const struct landlock_ruleset *const domain, + const struct dentry *const dentry); + /* * @path: Should have been checked by get_path_from_fd(). */ @@ -344,6 +375,48 @@ int landlock_append_fs_rule(struct landlock_ruleset *const ruleset, return PTR_ERR(id.key.object); mutex_lock(&ruleset->lock); err = landlock_insert_rule(ruleset, id, access_rights, flags); + if (err || !(flags & LANDLOCK_ADD_RULE_NO_INHERIT)) + goto out_unlock; + + /* Create ancestor rules and set has_no_inherit_descendant flags */ + struct path walker = *path; + + path_get(&walker); + while (landlock_walk_path_up(&walker) != LANDLOCK_WALK_STOP_REAL_ROOT) { + struct landlock_rule *ancestor_rule; + + if (WARN_ON_ONCE(!walker.dentry || d_is_negative(walker.dentry))) { + err = -EIO; + break; + } + + ancestor_rule = (struct landlock_rule *)find_rule(ruleset, walker.dentry); + if (!ancestor_rule) { + struct landlock_id ancestor_id = { + .type = LANDLOCK_KEY_INODE, + .key.object = get_inode_object(d_backing_inode(walker.dentry)), + }; + + if (IS_ERR(ancestor_id.key.object)) { + err = PTR_ERR(ancestor_id.key.object); + break; + } + err = landlock_insert_rule(ruleset, ancestor_id, 0, 0); + landlock_put_object(ancestor_id.key.object); + if (err) + break; + + ancestor_rule = (struct landlock_rule *) + find_rule(ruleset, walker.dentry); + } + if (WARN_ON_ONCE(!ancestor_rule || ancestor_rule->num_layers != 1)) { + err = -EIO; + break; + } + ancestor_rule->layers[0].flags.has_no_inherit_descendant = true; + } + path_put(&walker); +out_unlock: mutex_unlock(&ruleset->lock); /* * No need to check for an error because landlock_insert_rule() @@ -772,8 +845,10 @@ static bool is_access_to_paths_allowed( _layer_masks_child2[LANDLOCK_NUM_ACCESS_FS]; layer_mask_t(*layer_masks_child1)[LANDLOCK_NUM_ACCESS_FS] = NULL, (*layer_masks_child2)[LANDLOCK_NUM_ACCESS_FS] = NULL; - struct collected_rule_flags *rule_flags_parent1 = &log_request_parent1->rule_flags; - struct collected_rule_flags *rule_flags_parent2 = &log_request_parent2->rule_flags; + struct collected_rule_flags *rule_flags_parent1 = + &log_request_parent1->rule_flags; + struct collected_rule_flags *rule_flags_parent2 = + log_request_parent2 ? &log_request_parent2->rule_flags : NULL; if (!access_request_parent1 && !access_request_parent2) return true; @@ -784,7 +859,7 @@ static bool is_access_to_paths_allowed( if (is_nouser_or_private(path->dentry)) return true; - if (WARN_ON_ONCE(!layer_masks_parent1)) + if (WARN_ON_ONCE(!layer_masks_parent1 || !log_request_parent1)) return false; allowed_parent1 = is_layer_masks_allowed(layer_masks_parent1); @@ -851,6 +926,7 @@ static bool is_access_to_paths_allowed( */ while (true) { const struct landlock_rule *rule; + enum landlock_walk_result walk_res; /* * If at least all accesses allowed on the destination are @@ -910,46 +986,14 @@ static bool is_access_to_paths_allowed( if (allowed_parent1 && allowed_parent2) break; -jump_up: - if (walker_path.dentry == walker_path.mnt->mnt_root) { - if (follow_up(&walker_path)) { - /* Ignores hidden mount points. */ - goto jump_up; - } else { - /* - * Stops at the real root. Denies access - * because not all layers have granted access. - */ - break; - } - } - - if (unlikely(IS_ROOT(walker_path.dentry))) { - if (likely(walker_path.mnt->mnt_flags & MNT_INTERNAL)) { - /* - * Stops and allows access when reaching disconnected root - * directories that are part of internal filesystems (e.g. nsfs, - * which is reachable through /proc/<pid>/ns/<namespace>). - */ - allowed_parent1 = true; - allowed_parent2 = true; - break; - } - - /* - * We reached a disconnected root directory from a bind mount. - * Let's continue the walk with the mount point we missed. - */ - dput(walker_path.dentry); - walker_path.dentry = walker_path.mnt->mnt_root; - dget(walker_path.dentry); - } else { - struct dentry *const parent_dentry = - dget_parent(walker_path.dentry); - - dput(walker_path.dentry); - walker_path.dentry = parent_dentry; + walk_res = landlock_walk_path_up(&walker_path); + if (walk_res == LANDLOCK_WALK_MOUNT_ROOT) { + allowed_parent1 = true; + allowed_parent2 = true; + break; } + if (walk_res != LANDLOCK_WALK_CONTINUE) + break; } path_put(&walker_path); @@ -963,7 +1007,7 @@ static bool is_access_to_paths_allowed( ARRAY_SIZE(*layer_masks_parent1); } - if (!allowed_parent2) { + if (!allowed_parent2 && log_request_parent2) { log_request_parent2->type = LANDLOCK_REQUEST_FS_ACCESS; log_request_parent2->audit.type = LSM_AUDIT_DATA_PATH; log_request_parent2->audit.u.path = *path; @@ -1037,8 +1081,8 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) * collect_domain_accesses - Walk through a file path and collect accesses * * @domain: Domain to check against. - * @mnt_root: Last directory to check. - * @dir: Directory to start the walk from. + * @mnt_root: Last path element to check. + * @dir: Directory path to start the walk from. * @layer_masks_dom: Where to store the collected accesses. * * This helper is useful to begin a path walk from the @dir directory to a @@ -1060,29 +1104,31 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) */ static bool collect_domain_accesses( const struct landlock_ruleset *const domain, - const struct dentry *const mnt_root, struct dentry *dir, + const struct path *const mnt_root, const struct path *const dir, layer_mask_t (*const layer_masks_dom)[LANDLOCK_NUM_ACCESS_FS], struct collected_rule_flags *const rule_flags) { unsigned long access_dom; bool ret = false; + struct path walker; if (WARN_ON_ONCE(!domain || !mnt_root || !dir || !layer_masks_dom)) return true; - if (is_nouser_or_private(dir)) + if (is_nouser_or_private(dir->dentry)) return true; access_dom = landlock_init_layer_masks(domain, LANDLOCK_MASK_ACCESS_FS, layer_masks_dom, LANDLOCK_KEY_INODE); - dget(dir); + walker = *dir; + path_get(&walker); while (true) { - struct dentry *parent_dentry; + enum landlock_walk_result walk_res; /* Gets all layers allowing all domain accesses. */ if (landlock_unmask_layers( - find_rule(domain, dir), access_dom, layer_masks_dom, + find_rule(domain, walker.dentry), access_dom, layer_masks_dom, ARRAY_SIZE(*layer_masks_dom), rule_flags)) { /* * Stops when all handled accesses are allowed by at @@ -1091,22 +1137,69 @@ static bool collect_domain_accesses( ret = true; break; } - - /* - * Stops at the mount point or the filesystem root for a disconnected - * directory. - */ - if (dir == mnt_root || unlikely(IS_ROOT(dir))) + if (walker.dentry == mnt_root->dentry && walker.mnt == mnt_root->mnt) + break; + walk_res = landlock_walk_path_up(&walker); + if (walk_res != LANDLOCK_WALK_CONTINUE) break; - - parent_dentry = dget_parent(dir); - dput(dir); - dir = parent_dentry; } - dput(dir); + path_put(&walker); return ret; } +/** + * deny_no_inherit_topology_change - deny topology changes on sealed paths + * @subject: Subject performing the operation (contains the domain). + * @path: Path whose dentry is the target of the topology modification. + * + * Checks whether any domain layers are sealed against topology changes at + * @path. If so, emit an audit record and return -EACCES. Otherwise return 0. + */ +static int deny_no_inherit_topology_change(const struct landlock_cred_security + *subject, + const struct path *const path) +{ + layer_mask_t sealed_layers = 0; + layer_mask_t override_layers = 0; + const struct landlock_rule *rule; + u32 layer_index; + unsigned long audit_layer_index; + + if (WARN_ON_ONCE(!subject || !path || !path->dentry || !path->mnt || + d_is_negative(path->dentry))) + return 0; + + rule = find_rule(subject->domain, path->dentry); + if (!rule) + return 0; + + for (layer_index = 0; layer_index < rule->num_layers; layer_index++) { + const struct landlock_layer *layer = &rule->layers[layer_index]; + layer_mask_t layer_bit = BIT_ULL(layer->level - 1); + + if (layer->flags.no_inherit || + layer->flags.has_no_inherit_descendant) + sealed_layers |= layer_bit; + else + override_layers |= layer_bit; + } + + sealed_layers &= ~override_layers; + if (!sealed_layers) + return 0; + + audit_layer_index = __ffs((unsigned long)sealed_layers); + landlock_log_denial(subject, &(struct landlock_request) { + .type = LANDLOCK_REQUEST_FS_CHANGE_TOPOLOGY, + .audit = { + .type = LSM_AUDIT_DATA_DENTRY, + .u.dentry = path->dentry, + }, + .layer_plus_one = audit_layer_index + 1, + }); + return -EACCES; +} + /** * current_check_refer_path - Check if a rename or link action is allowed * @@ -1191,6 +1284,21 @@ static int current_check_refer_path(struct dentry *const old_dentry, access_request_parent2 = get_mode_access(d_backing_inode(old_dentry)->i_mode); if (removable) { + int err = deny_no_inherit_topology_change(subject, + &(struct path) + { .mnt = new_dir->mnt, + .dentry = old_dentry }); + + if (err) + return err; + if (exchange) { + err = deny_no_inherit_topology_change(subject, + &(struct path) + { .mnt = new_dir->mnt, + .dentry = new_dentry }); + if (err) + return err; + } access_request_parent1 |= maybe_remove(old_dentry); access_request_parent2 |= maybe_remove(new_dentry); } @@ -1232,12 +1340,15 @@ static int current_check_refer_path(struct dentry *const old_dentry, old_dentry->d_parent; /* new_dir->dentry is equal to new_dentry->d_parent */ - allow_parent1 = collect_domain_accesses(subject->domain, mnt_dir.dentry, - old_parent, + allow_parent1 = collect_domain_accesses(subject->domain, + &mnt_dir, + &(struct path){ .mnt = new_dir->mnt, + .dentry = old_parent }, &layer_masks_parent1, &request1.rule_flags); - allow_parent2 = collect_domain_accesses(subject->domain, mnt_dir.dentry, - new_dir->dentry, + allow_parent2 = collect_domain_accesses(subject->domain, &mnt_dir, + &(struct path){ .mnt = new_dir->mnt, + .dentry = new_dir->dentry }, &layer_masks_parent2, &request2.rule_flags); @@ -1583,12 +1694,37 @@ static int hook_path_symlink(const struct path *const dir, static int hook_path_unlink(const struct path *const dir, struct dentry *const dentry) { + const struct landlock_cred_security *const subject = + landlock_get_applicable_subject(current_cred(), any_fs, NULL); + int err; + + if (subject) { + err = deny_no_inherit_topology_change(subject, + &(struct path) + { .mnt = dir->mnt, + .dentry = dentry }); + if (err) + return err; + } return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_FILE); } static int hook_path_rmdir(const struct path *const dir, struct dentry *const dentry) { + const struct landlock_cred_security *const subject = + landlock_get_applicable_subject(current_cred(), any_fs, NULL); + int err; + + if (subject) { + err = deny_no_inherit_topology_change(subject, + &(struct path) + { .mnt = dir->mnt, + .dentry = dentry }); + if (err) + return err; + } + return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_DIR); } diff --git a/security/landlock/ruleset.h b/security/landlock/ruleset.h index eb60db646422..81df6c56a152 100644 --- a/security/landlock/ruleset.h +++ b/security/landlock/ruleset.h @@ -40,6 +40,21 @@ struct landlock_layer { * down the file hierarchy. */ bool quiet:1; + /** + * @no_inherit: Prevents this rule from being inherited by + * descendant directories in the filesystem layer. Only used + * for filesystem rules. + */ + bool no_inherit:1; + /** + * @has_no_inherit_descendant: Marker to indicate that this layer + * has at least one descendant directory with a rule having the + * no_inherit flag. Only used for filesystem rules. + * This "flag" is not set by the user, but by Landlock on + * parent directories of rules when the child rule has + * a rule with the no_inherit flag. + */ + bool has_no_inherit_descendant:1; } flags; /** * @access: Bitfield of allowed actions on the kernel object. They are @@ -49,13 +64,25 @@ struct landlock_layer { }; /** - * struct collected_rule_flags - Hold accumulated flags for each layer. + * struct collected_rule_flags - Hold accumulated flags and their markers for each layer. */ struct collected_rule_flags { /** * @quiet_masks: Layers for which the quiet flag is effective. */ layer_mask_t quiet_masks; + /** + * @no_inherit_masks: Layers for which the no_inherit flag is effective. + */ + layer_mask_t no_inherit_masks; + /** + * @no_inherit_desc_masks: Layers for which the + * has_no_inherit_descendant tag is effective. + * This is not a flag itself, but a marker set on ancestors + * of rules with the no_inherit flag to deny topology changes + * in the direct parent path. + */ + layer_mask_t no_inherit_desc_masks; }; /** -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH v5 1/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 17:05 ` [PATCH v5 1/6] landlock: " Justin Suess @ 2025-12-14 22:53 ` Tingmao Wang 2025-12-15 22:21 ` Justin Suess 0 siblings, 1 reply; 9+ messages in thread From: Tingmao Wang @ 2025-12-14 22:53 UTC (permalink / raw) To: Justin Suess Cc: Mickaël Salaün, Günther Noack, Jan Kara, Abhinav Saxena, linux-security-module On 12/14/25 17:05, Justin Suess wrote: > [...] > diff --git a/include/uapi/linux/landlock.h b/include/uapi/linux/landlock.h > index d4f47d20361a..6ab3e7bd1c81 100644 > --- a/include/uapi/linux/landlock.h > +++ b/include/uapi/linux/landlock.h > @@ -127,10 +127,39 @@ struct landlock_ruleset_attr { > * allowed_access in the passed in rule_attr. When this flag is > * present, the caller is also allowed to pass in an empty > * allowed_access. > + * %LANDLOCK_ADD_RULE_NO_INHERIT > + * When set on a rule being added to a ruleset, this flag disables the > + * inheritance of access rights and flags from parent objects. > + * > + * This flag currently applies only to filesystem rules. Adding it to > + * non-filesystem rules will return -EINVAL, unless future extensions > + * of Landlock define other hierarchical object types. > + * > + * By default, Landlock filesystem rules inherit allowed accesses from > + * ancestor directories: if a parent directory grants certain rights, > + * those rights also apply to its children. A rule marked with > + * LANDLOCK_ADD_RULE_NO_INHERIT stops this propagation at the directory > + * covered by the rule. Descendants of that directory continue to inherit > + * normally unless they also have rules using this flag. > + * > + * If a regular file is marked with this flag, it will not inherit any > + * access rights from its parent directories; only the accesses explicitly > + * allowed by the rule will apply to that file. > + * > + * This flag also enforces parent-directory restrictions: rename, rmdir, > + * link, and other operations that would change the directory's immediate > + * parent subtree are denied up to the VFS root. This prevents > + * sandboxed processes from manipulating the filesystem hierarchy to evade > + * restrictions (e.g., via sandbox-restart attacks). > + * > + * In addition, this flag blocks the inheritance of rule-layer flags tbh I feel that it's less confusing to just say "rule flags" (instead of "rule-layer flags"). > + * (such as the quiet flag) from parent directories to the object covered > + * by this rule. > */ > > /* clang-format off */ > #define LANDLOCK_ADD_RULE_QUIET (1U << 0) > +#define LANDLOCK_ADD_RULE_NO_INHERIT (1U << 1) > /* clang-format on */ > > /** > diff --git a/security/landlock/fs.c b/security/landlock/fs.c > index 0b589263ea42..8d8623ea857f 100644 > --- a/security/landlock/fs.c > +++ b/security/landlock/fs.c > @@ -317,6 +317,37 @@ static struct landlock_object *get_inode_object(struct inode *const inode) > LANDLOCK_ACCESS_FS_IOCTL_DEV) > /* clang-format on */ > > +enum landlock_walk_result { > + LANDLOCK_WALK_CONTINUE, > + LANDLOCK_WALK_STOP_REAL_ROOT, > + LANDLOCK_WALK_MOUNT_ROOT, > +}; > + > +static enum landlock_walk_result landlock_walk_path_up(struct path *const path) > +{ > + while (path->dentry == path->mnt->mnt_root) { > + if (!follow_up(path)) > + return LANDLOCK_WALK_STOP_REAL_ROOT; > + } > + > + if (unlikely(IS_ROOT(path->dentry))) { > + if (likely(path->mnt->mnt_flags & MNT_INTERNAL)) > + return LANDLOCK_WALK_MOUNT_ROOT; imo, LANDLOCK_WALK_MOUNT_ROOT is a somewhat confusing name for this, especially in the context that if we see this in is_access_to_paths_allowed() we allow access unconditionally. Would LANDLOCK_WALK_INTERNAL be a better name here? > + dput(path->dentry); > + path->dentry = dget(path->mnt->mnt_root); > + return LANDLOCK_WALK_CONTINUE; > + } > + > + struct dentry *const parent = dget_parent(path->dentry); > + > + dput(path->dentry); > + path->dentry = parent; > + return LANDLOCK_WALK_CONTINUE; > +} > + > +static const struct landlock_rule *find_rule(const struct landlock_ruleset *const domain, > + const struct dentry *const dentry); > + > /* > * @path: Should have been checked by get_path_from_fd(). > */ > @@ -344,6 +375,48 @@ int landlock_append_fs_rule(struct landlock_ruleset *const ruleset, > return PTR_ERR(id.key.object); > mutex_lock(&ruleset->lock); > err = landlock_insert_rule(ruleset, id, access_rights, flags); > + if (err || !(flags & LANDLOCK_ADD_RULE_NO_INHERIT)) > + goto out_unlock; > + > + /* Create ancestor rules and set has_no_inherit_descendant flags */ > + struct path walker = *path; > + > + path_get(&walker); > + while (landlock_walk_path_up(&walker) != LANDLOCK_WALK_STOP_REAL_ROOT) { Why not landlock_walk_path_up(&walker) == LANDLOCK_WALK_CONTINUE here? I'm not sure if it's actually possible to end up with an infinite loop by ignoring LANDLOCK_WALK_MOUNT_ROOT (i.e. not sure if "internal" mounts can have disconnected dentries), but it seems safer to write to loop in a way such that if that happens, we exit. > + struct landlock_rule *ancestor_rule; > + > + if (WARN_ON_ONCE(!walker.dentry || d_is_negative(walker.dentry))) { > + err = -EIO; > + break; > + } > + > + ancestor_rule = (struct landlock_rule *)find_rule(ruleset, walker.dentry); > + if (!ancestor_rule) { > + struct landlock_id ancestor_id = { > + .type = LANDLOCK_KEY_INODE, > + .key.object = get_inode_object(d_backing_inode(walker.dentry)), > + }; > + > + if (IS_ERR(ancestor_id.key.object)) { > + err = PTR_ERR(ancestor_id.key.object); > + break; > + } > + err = landlock_insert_rule(ruleset, ancestor_id, 0, 0); > + landlock_put_object(ancestor_id.key.object); > + if (err) > + break; > + > + ancestor_rule = (struct landlock_rule *) > + find_rule(ruleset, walker.dentry); > + } > + if (WARN_ON_ONCE(!ancestor_rule || ancestor_rule->num_layers != 1)) { > + err = -EIO; > + break; > + } > + ancestor_rule->layers[0].flags.has_no_inherit_descendant = true; > + } > + path_put(&walker); > +out_unlock: > mutex_unlock(&ruleset->lock); > /* > * No need to check for an error because landlock_insert_rule() > @@ -772,8 +845,10 @@ static bool is_access_to_paths_allowed( > _layer_masks_child2[LANDLOCK_NUM_ACCESS_FS]; > layer_mask_t(*layer_masks_child1)[LANDLOCK_NUM_ACCESS_FS] = NULL, > (*layer_masks_child2)[LANDLOCK_NUM_ACCESS_FS] = NULL; > - struct collected_rule_flags *rule_flags_parent1 = &log_request_parent1->rule_flags; > - struct collected_rule_flags *rule_flags_parent2 = &log_request_parent2->rule_flags; > + struct collected_rule_flags *rule_flags_parent1 = > + &log_request_parent1->rule_flags; > + struct collected_rule_flags *rule_flags_parent2 = > + log_request_parent2 ? &log_request_parent2->rule_flags : NULL; Good point, I think the original was still safe because it would not be used by landlock_unmask_layers anyway, but this is better. I will take this in the next version, thanks! > > if (!access_request_parent1 && !access_request_parent2) > return true; > @@ -784,7 +859,7 @@ static bool is_access_to_paths_allowed( > if (is_nouser_or_private(path->dentry)) > return true; > > - if (WARN_ON_ONCE(!layer_masks_parent1)) > + if (WARN_ON_ONCE(!layer_masks_parent1 || !log_request_parent1)) > return false; > > allowed_parent1 = is_layer_masks_allowed(layer_masks_parent1); > @@ -851,6 +926,7 @@ static bool is_access_to_paths_allowed( > */ > while (true) { > const struct landlock_rule *rule; > + enum landlock_walk_result walk_res; > > /* > * If at least all accesses allowed on the destination are > @@ -910,46 +986,14 @@ static bool is_access_to_paths_allowed( > if (allowed_parent1 && allowed_parent2) > break; > > -jump_up: > - if (walker_path.dentry == walker_path.mnt->mnt_root) { > - if (follow_up(&walker_path)) { > - /* Ignores hidden mount points. */ > - goto jump_up; > - } else { > - /* > - * Stops at the real root. Denies access > - * because not all layers have granted access. > - */ > - break; > - } > - } > - > - if (unlikely(IS_ROOT(walker_path.dentry))) { > - if (likely(walker_path.mnt->mnt_flags & MNT_INTERNAL)) { > - /* > - * Stops and allows access when reaching disconnected root > - * directories that are part of internal filesystems (e.g. nsfs, > - * which is reachable through /proc/<pid>/ns/<namespace>). > - */ > - allowed_parent1 = true; > - allowed_parent2 = true; > - break; > - } > - > - /* > - * We reached a disconnected root directory from a bind mount. > - * Let's continue the walk with the mount point we missed. > - */ I think we might want to preserve these comments. > - dput(walker_path.dentry); > - walker_path.dentry = walker_path.mnt->mnt_root; > - dget(walker_path.dentry); > - } else { > - struct dentry *const parent_dentry = > - dget_parent(walker_path.dentry); > - > - dput(walker_path.dentry); > - walker_path.dentry = parent_dentry; > + walk_res = landlock_walk_path_up(&walker_path); > + if (walk_res == LANDLOCK_WALK_MOUNT_ROOT) { > + allowed_parent1 = true; > + allowed_parent2 = true; > + break; > } > + if (walk_res != LANDLOCK_WALK_CONTINUE) > + break; > } > path_put(&walker_path); > > @@ -963,7 +1007,7 @@ static bool is_access_to_paths_allowed( > ARRAY_SIZE(*layer_masks_parent1); > } > > - if (!allowed_parent2) { > + if (!allowed_parent2 && log_request_parent2) { > log_request_parent2->type = LANDLOCK_REQUEST_FS_ACCESS; > log_request_parent2->audit.type = LSM_AUDIT_DATA_PATH; > log_request_parent2->audit.u.path = *path; > @@ -1037,8 +1081,8 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) > * collect_domain_accesses - Walk through a file path and collect accesses > * > * @domain: Domain to check against. > - * @mnt_root: Last directory to check. > - * @dir: Directory to start the walk from. > + * @mnt_root: Last path element to check. > + * @dir: Directory path to start the walk from. > * @layer_masks_dom: Where to store the collected accesses. > * > * This helper is useful to begin a path walk from the @dir directory to a > @@ -1060,29 +1104,31 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) > */ > static bool collect_domain_accesses( > const struct landlock_ruleset *const domain, > - const struct dentry *const mnt_root, struct dentry *dir, > + const struct path *const mnt_root, const struct path *const dir, > layer_mask_t (*const layer_masks_dom)[LANDLOCK_NUM_ACCESS_FS], > struct collected_rule_flags *const rule_flags) > { This function only walks up to the mountpoint of dir. If dir is changed from a *dentry to a *path, wouldn't mnt_root be redundant? Since mnt_root->dentry is always going to be dir->mnt->mnt_root. This also means that they can't accidentally not be the same. > unsigned long access_dom; > bool ret = false; > + struct path walker; > > if (WARN_ON_ONCE(!domain || !mnt_root || !dir || !layer_masks_dom)) > return true; > - if (is_nouser_or_private(dir)) > + if (is_nouser_or_private(dir->dentry)) > return true; > > access_dom = landlock_init_layer_masks(domain, LANDLOCK_MASK_ACCESS_FS, > layer_masks_dom, > LANDLOCK_KEY_INODE); > > - dget(dir); > + walker = *dir; > + path_get(&walker); > while (true) { > - struct dentry *parent_dentry; > + enum landlock_walk_result walk_res; > > /* Gets all layers allowing all domain accesses. */ > if (landlock_unmask_layers( > - find_rule(domain, dir), access_dom, layer_masks_dom, > + find_rule(domain, walker.dentry), access_dom, layer_masks_dom, > ARRAY_SIZE(*layer_masks_dom), rule_flags)) { > /* > * Stops when all handled accesses are allowed by at > @@ -1091,22 +1137,69 @@ static bool collect_domain_accesses( > ret = true; > break; > } > - > - /* > - * Stops at the mount point or the filesystem root for a disconnected > - * directory. > - */ > - if (dir == mnt_root || unlikely(IS_ROOT(dir))) > + if (walker.dentry == mnt_root->dentry && walker.mnt == mnt_root->mnt) > + break; > + walk_res = landlock_walk_path_up(&walker); > + if (walk_res != LANDLOCK_WALK_CONTINUE) > break; > - > - parent_dentry = dget_parent(dir); > - dput(dir); > - dir = parent_dentry; > } > - dput(dir); > + path_put(&walker); > return ret; > } > > +/** > + * deny_no_inherit_topology_change - deny topology changes on sealed paths > + * @subject: Subject performing the operation (contains the domain). > + * @path: Path whose dentry is the target of the topology modification. > + * > + * Checks whether any domain layers are sealed against topology changes at > + * @path. If so, emit an audit record and return -EACCES. Otherwise return 0. > + */ > +static int deny_no_inherit_topology_change(const struct landlock_cred_security > + *subject, > + const struct path *const path) Since you're not using path->mnt here (except for a NULL check), would it be easier to just pass the dentry instead? In that case you wouldn't have to do an inline initializer in current_check_refer_path / hook_path_* below as well. > +{ > + layer_mask_t sealed_layers = 0; > + layer_mask_t override_layers = 0; > + const struct landlock_rule *rule; > + u32 layer_index; > + unsigned long audit_layer_index; > + > + if (WARN_ON_ONCE(!subject || !path || !path->dentry || !path->mnt || > + d_is_negative(path->dentry))) > + return 0; > + > + rule = find_rule(subject->domain, path->dentry); > + if (!rule) > + return 0; > + > + for (layer_index = 0; layer_index < rule->num_layers; layer_index++) { > + const struct landlock_layer *layer = &rule->layers[layer_index]; > + layer_mask_t layer_bit = BIT_ULL(layer->level - 1); > + > + if (layer->flags.no_inherit || > + layer->flags.has_no_inherit_descendant) > + sealed_layers |= layer_bit; > + else > + override_layers |= layer_bit; > + } > + > + sealed_layers &= ~override_layers; > + if (!sealed_layers) > + return 0; > + > + audit_layer_index = __ffs((unsigned long)sealed_layers); > + landlock_log_denial(subject, &(struct landlock_request) { > + .type = LANDLOCK_REQUEST_FS_CHANGE_TOPOLOGY, > + .audit = { > + .type = LSM_AUDIT_DATA_DENTRY, > + .u.dentry = path->dentry, > + }, > + .layer_plus_one = audit_layer_index + 1, > + }); > + return -EACCES; > +} > + > /** > * current_check_refer_path - Check if a rename or link action is allowed > * > @@ -1191,6 +1284,21 @@ static int current_check_refer_path(struct dentry *const old_dentry, > access_request_parent2 = > get_mode_access(d_backing_inode(old_dentry)->i_mode); > if (removable) { > + int err = deny_no_inherit_topology_change(subject, > + &(struct path) > + { .mnt = new_dir->mnt, > + .dentry = old_dentry }); > + > + if (err) > + return err; > + if (exchange) { > + err = deny_no_inherit_topology_change(subject, > + &(struct path) > + { .mnt = new_dir->mnt, > + .dentry = new_dentry }); > + if (err) > + return err; > + } > access_request_parent1 |= maybe_remove(old_dentry); > access_request_parent2 |= maybe_remove(new_dentry); > } > @@ -1232,12 +1340,15 @@ static int current_check_refer_path(struct dentry *const old_dentry, > old_dentry->d_parent; > > /* new_dir->dentry is equal to new_dentry->d_parent */ > - allow_parent1 = collect_domain_accesses(subject->domain, mnt_dir.dentry, > - old_parent, > + allow_parent1 = collect_domain_accesses(subject->domain, > + &mnt_dir, > + &(struct path){ .mnt = new_dir->mnt, > + .dentry = old_parent }, > &layer_masks_parent1, > &request1.rule_flags); > - allow_parent2 = collect_domain_accesses(subject->domain, mnt_dir.dentry, > - new_dir->dentry, > + allow_parent2 = collect_domain_accesses(subject->domain, &mnt_dir, > + &(struct path){ .mnt = new_dir->mnt, > + .dentry = new_dir->dentry }, > &layer_masks_parent2, > &request2.rule_flags); > > @@ -1583,12 +1694,37 @@ static int hook_path_symlink(const struct path *const dir, > static int hook_path_unlink(const struct path *const dir, > struct dentry *const dentry) > { > + const struct landlock_cred_security *const subject = > + landlock_get_applicable_subject(current_cred(), any_fs, NULL); > + int err; > + > + if (subject) { > + err = deny_no_inherit_topology_change(subject, > + &(struct path) > + { .mnt = dir->mnt, > + .dentry = dentry }); > + if (err) > + return err; > + } > return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_FILE); > } > > static int hook_path_rmdir(const struct path *const dir, > struct dentry *const dentry) > { > + const struct landlock_cred_security *const subject = > + landlock_get_applicable_subject(current_cred(), any_fs, NULL); > + int err; > + > + if (subject) { > + err = deny_no_inherit_topology_change(subject, > + &(struct path) > + { .mnt = dir->mnt, > + .dentry = dentry }); > + if (err) > + return err; > + } > + > return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_DIR); > } > > [...] ^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5 1/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 22:53 ` Tingmao Wang @ 2025-12-15 22:21 ` Justin Suess 0 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-15 22:21 UTC (permalink / raw) To: m; +Cc: gnoack, jack, linux-security-module, mic, utilityemal77, xandfury On 12/14/25 17:53, Tingmao Wang wrote: > On 12/14/25 17:05, Justin Suess wrote: >> [...] >> diff --git a/include/uapi/linux/landlock.h b/include/uapi/linux/landlock.h >> index d4f47d20361a..6ab3e7bd1c81 100644 >> --- a/include/uapi/linux/landlock.h >> +++ b/include/uapi/linux/landlock.h >> @@ -127,10 +127,39 @@ struct landlock_ruleset_attr { >> * allowed_access in the passed in rule_attr. When this flag is >> * present, the caller is also allowed to pass in an empty >> * allowed_access. >> + * %LANDLOCK_ADD_RULE_NO_INHERIT >> + * When set on a rule being added to a ruleset, this flag disables the >> + * inheritance of access rights and flags from parent objects. >> + * >> + * This flag currently applies only to filesystem rules. Adding it to >> + * non-filesystem rules will return -EINVAL, unless future extensions >> + * of Landlock define other hierarchical object types. >> + * >> + * By default, Landlock filesystem rules inherit allowed accesses from >> + * ancestor directories: if a parent directory grants certain rights, >> + * those rights also apply to its children. A rule marked with >> + * LANDLOCK_ADD_RULE_NO_INHERIT stops this propagation at the directory >> + * covered by the rule. Descendants of that directory continue to inherit >> + * normally unless they also have rules using this flag. >> + * >> + * If a regular file is marked with this flag, it will not inherit any >> + * access rights from its parent directories; only the accesses explicitly >> + * allowed by the rule will apply to that file. >> + * >> + * This flag also enforces parent-directory restrictions: rename, rmdir, >> + * link, and other operations that would change the directory's immediate >> + * parent subtree are denied up to the VFS root. This prevents >> + * sandboxed processes from manipulating the filesystem hierarchy to evade >> + * restrictions (e.g., via sandbox-restart attacks). >> + * >> + * In addition, this flag blocks the inheritance of rule-layer flags > tbh I feel that it's less confusing to just say "rule flags" (instead of > "rule-layer flags"). Agreed. I'll change it here and in any other locations it pops up, I'll have to see. >> + * (such as the quiet flag) from parent directories to the object covered >> + * by this rule. >> */ >> >> /* clang-format off */ >> #define LANDLOCK_ADD_RULE_QUIET (1U << 0) >> +#define LANDLOCK_ADD_RULE_NO_INHERIT (1U << 1) >> /* clang-format on */ >> >> /** >> diff --git a/security/landlock/fs.c b/security/landlock/fs.c >> index 0b589263ea42..8d8623ea857f 100644 >> --- a/security/landlock/fs.c >> +++ b/security/landlock/fs.c >> @@ -317,6 +317,37 @@ static struct landlock_object *get_inode_object(struct inode *const inode) >> LANDLOCK_ACCESS_FS_IOCTL_DEV) >> /* clang-format on */ >> >> +enum landlock_walk_result { >> + LANDLOCK_WALK_CONTINUE, >> + LANDLOCK_WALK_STOP_REAL_ROOT, >> + LANDLOCK_WALK_MOUNT_ROOT, >> +}; >> + >> +static enum landlock_walk_result landlock_walk_path_up(struct path *const path) >> +{ >> + while (path->dentry == path->mnt->mnt_root) { >> + if (!follow_up(path)) >> + return LANDLOCK_WALK_STOP_REAL_ROOT; >> + } >> + >> + if (unlikely(IS_ROOT(path->dentry))) { >> + if (likely(path->mnt->mnt_flags & MNT_INTERNAL)) >> + return LANDLOCK_WALK_MOUNT_ROOT; > imo, LANDLOCK_WALK_MOUNT_ROOT is a somewhat confusing name for this, > especially in the context that if we see this in > is_access_to_paths_allowed() we allow access unconditionally. > > Would LANDLOCK_WALK_INTERNAL be a better name here? > Yeah that seems better. LANDLOCK_WALK_INTERNAL seems like a better name. Plus some documenting comments in the landlock_walk_result are warranted. I'll fix it in the next version. >> + dput(path->dentry); >> + path->dentry = dget(path->mnt->mnt_root); >> + return LANDLOCK_WALK_CONTINUE; >> + } >> + >> + struct dentry *const parent = dget_parent(path->dentry); >> + >> + dput(path->dentry); >> + path->dentry = parent; >> + return LANDLOCK_WALK_CONTINUE; >> +} >> + >> +static const struct landlock_rule *find_rule(const struct landlock_ruleset *const domain, >> + const struct dentry *const dentry); >> + >> /* >> * @path: Should have been checked by get_path_from_fd(). >> */ >> @@ -344,6 +375,48 @@ int landlock_append_fs_rule(struct landlock_ruleset *const ruleset, >> return PTR_ERR(id.key.object); >> mutex_lock(&ruleset->lock); >> err = landlock_insert_rule(ruleset, id, access_rights, flags); >> + if (err || !(flags & LANDLOCK_ADD_RULE_NO_INHERIT)) >> + goto out_unlock; >> + >> + /* Create ancestor rules and set has_no_inherit_descendant flags */ >> + struct path walker = *path; >> + >> + path_get(&walker); >> + while (landlock_walk_path_up(&walker) != LANDLOCK_WALK_STOP_REAL_ROOT) { > Why not landlock_walk_path_up(&walker) == LANDLOCK_WALK_CONTINUE here? > I'm not sure if it's actually possible to end up with an infinite loop by > ignoring LANDLOCK_WALK_MOUNT_ROOT (i.e. not sure if "internal" mounts can > have disconnected dentries), but it seems safer to write to loop in a way > such that if that happens, we exit. I don't *think* it's possible to end up in an infinite loop this way, but you never know. I'll definitely take your suggestion because it's semantically clearer at the very least. > >> + struct landlock_rule *ancestor_rule; >> + >> + if (WARN_ON_ONCE(!walker.dentry || d_is_negative(walker.dentry))) { >> + err = -EIO; >> + break; >> + } >> + >> + ancestor_rule = (struct landlock_rule *)find_rule(ruleset, walker.dentry); >> + if (!ancestor_rule) { >> + struct landlock_id ancestor_id = { >> + .type = LANDLOCK_KEY_INODE, >> + .key.object = get_inode_object(d_backing_inode(walker.dentry)), >> + }; >> + >> + if (IS_ERR(ancestor_id.key.object)) { >> + err = PTR_ERR(ancestor_id.key.object); >> + break; >> + } >> + err = landlock_insert_rule(ruleset, ancestor_id, 0, 0); >> + landlock_put_object(ancestor_id.key.object); >> + if (err) >> + break; >> + >> + ancestor_rule = (struct landlock_rule *) >> + find_rule(ruleset, walker.dentry); >> + } >> + if (WARN_ON_ONCE(!ancestor_rule || ancestor_rule->num_layers != 1)) { >> + err = -EIO; >> + break; >> + } >> + ancestor_rule->layers[0].flags.has_no_inherit_descendant = true; >> + } >> + path_put(&walker); >> +out_unlock: >> mutex_unlock(&ruleset->lock); >> /* >> * No need to check for an error because landlock_insert_rule() >> @@ -772,8 +845,10 @@ static bool is_access_to_paths_allowed( >> _layer_masks_child2[LANDLOCK_NUM_ACCESS_FS]; >> layer_mask_t(*layer_masks_child1)[LANDLOCK_NUM_ACCESS_FS] = NULL, >> (*layer_masks_child2)[LANDLOCK_NUM_ACCESS_FS] = NULL; >> - struct collected_rule_flags *rule_flags_parent1 = &log_request_parent1->rule_flags; >> - struct collected_rule_flags *rule_flags_parent2 = &log_request_parent2->rule_flags; >> + struct collected_rule_flags *rule_flags_parent1 = >> + &log_request_parent1->rule_flags; >> + struct collected_rule_flags *rule_flags_parent2 = >> + log_request_parent2 ? &log_request_parent2->rule_flags : NULL; > Good point, I think the original was still safe because it would not be > used by landlock_unmask_layers anyway, but this is better. I will take > this in the next version, thanks! No problem. I actually meant to put this as a review under your patch as a comment but I pulled it in accidentally. Rebasing off your patch has been a breeze btw 🙂 > >> if (!access_request_parent1 && !access_request_parent2) >> return true; >> @@ -784,7 +859,7 @@ static bool is_access_to_paths_allowed( >> if (is_nouser_or_private(path->dentry)) >> return true; >> >> - if (WARN_ON_ONCE(!layer_masks_parent1)) >> + if (WARN_ON_ONCE(!layer_masks_parent1 || !log_request_parent1)) >> return false; >> >> allowed_parent1 = is_layer_masks_allowed(layer_masks_parent1); >> @@ -851,6 +926,7 @@ static bool is_access_to_paths_allowed( >> */ >> while (true) { >> const struct landlock_rule *rule; >> + enum landlock_walk_result walk_res; >> >> /* >> * If at least all accesses allowed on the destination are >> @@ -910,46 +986,14 @@ static bool is_access_to_paths_allowed( >> if (allowed_parent1 && allowed_parent2) >> break; >> >> -jump_up: >> - if (walker_path.dentry == walker_path.mnt->mnt_root) { >> - if (follow_up(&walker_path)) { >> - /* Ignores hidden mount points. */ >> - goto jump_up; >> - } else { >> - /* >> - * Stops at the real root. Denies access >> - * because not all layers have granted access. >> - */ >> - break; >> - } >> - } >> - >> - if (unlikely(IS_ROOT(walker_path.dentry))) { >> - if (likely(walker_path.mnt->mnt_flags & MNT_INTERNAL)) { >> - /* >> - * Stops and allows access when reaching disconnected root >> - * directories that are part of internal filesystems (e.g. nsfs, >> - * which is reachable through /proc/<pid>/ns/<namespace>). >> - */ >> - allowed_parent1 = true; >> - allowed_parent2 = true; >> - break; >> - } >> - >> - /* >> - * We reached a disconnected root directory from a bind mount. >> - * Let's continue the walk with the mount point we missed. >> - */ > I think we might want to preserve these comments. Agreed. Thank you, I missed those. I'll preserve them in the next version. > >> - dput(walker_path.dentry); >> - walker_path.dentry = walker_path.mnt->mnt_root; >> - dget(walker_path.dentry); >> - } else { >> - struct dentry *const parent_dentry = >> - dget_parent(walker_path.dentry); >> - >> - dput(walker_path.dentry); >> - walker_path.dentry = parent_dentry; >> + walk_res = landlock_walk_path_up(&walker_path); >> + if (walk_res == LANDLOCK_WALK_MOUNT_ROOT) { >> + allowed_parent1 = true; >> + allowed_parent2 = true; >> + break; >> } >> + if (walk_res != LANDLOCK_WALK_CONTINUE) >> + break; >> } >> path_put(&walker_path); >> >> @@ -963,7 +1007,7 @@ static bool is_access_to_paths_allowed( >> ARRAY_SIZE(*layer_masks_parent1); >> } >> >> - if (!allowed_parent2) { >> + if (!allowed_parent2 && log_request_parent2) { >> log_request_parent2->type = LANDLOCK_REQUEST_FS_ACCESS; >> log_request_parent2->audit.type = LSM_AUDIT_DATA_PATH; >> log_request_parent2->audit.u.path = *path; >> @@ -1037,8 +1081,8 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) >> * collect_domain_accesses - Walk through a file path and collect accesses >> * >> * @domain: Domain to check against. >> - * @mnt_root: Last directory to check. >> - * @dir: Directory to start the walk from. >> + * @mnt_root: Last path element to check. >> + * @dir: Directory path to start the walk from. >> * @layer_masks_dom: Where to store the collected accesses. >> * >> * This helper is useful to begin a path walk from the @dir directory to a >> @@ -1060,29 +1104,31 @@ static access_mask_t maybe_remove(const struct dentry *const dentry) >> */ >> static bool collect_domain_accesses( >> const struct landlock_ruleset *const domain, >> - const struct dentry *const mnt_root, struct dentry *dir, >> + const struct path *const mnt_root, const struct path *const dir, >> layer_mask_t (*const layer_masks_dom)[LANDLOCK_NUM_ACCESS_FS], >> struct collected_rule_flags *const rule_flags) >> { > This function only walks up to the mountpoint of dir. If dir is changed > from a *dentry to a *path, wouldn't mnt_root be redundant? Since > mnt_root->dentry is always going to be dir->mnt->mnt_root. This also > means that they can't accidentally not be the same. Good catch, yeah they should be redundant. I'll remove the mnt_root parameter in the next version. > >> unsigned long access_dom; >> bool ret = false; >> + struct path walker; >> >> if (WARN_ON_ONCE(!domain || !mnt_root || !dir || !layer_masks_dom)) >> return true; >> - if (is_nouser_or_private(dir)) >> + if (is_nouser_or_private(dir->dentry)) >> return true; >> >> access_dom = landlock_init_layer_masks(domain, LANDLOCK_MASK_ACCESS_FS, >> layer_masks_dom, >> LANDLOCK_KEY_INODE); >> >> - dget(dir); >> + walker = *dir; >> + path_get(&walker); >> while (true) { >> - struct dentry *parent_dentry; >> + enum landlock_walk_result walk_res; >> >> /* Gets all layers allowing all domain accesses. */ >> if (landlock_unmask_layers( >> - find_rule(domain, dir), access_dom, layer_masks_dom, >> + find_rule(domain, walker.dentry), access_dom, layer_masks_dom, >> ARRAY_SIZE(*layer_masks_dom), rule_flags)) { >> /* >> * Stops when all handled accesses are allowed by at >> @@ -1091,22 +1137,69 @@ static bool collect_domain_accesses( >> ret = true; >> break; >> } >> - >> - /* >> - * Stops at the mount point or the filesystem root for a disconnected >> - * directory. >> - */ >> - if (dir == mnt_root || unlikely(IS_ROOT(dir))) >> + if (walker.dentry == mnt_root->dentry && walker.mnt == mnt_root->mnt) >> + break; >> + walk_res = landlock_walk_path_up(&walker); >> + if (walk_res != LANDLOCK_WALK_CONTINUE) >> break; >> - >> - parent_dentry = dget_parent(dir); >> - dput(dir); >> - dir = parent_dentry; >> } >> - dput(dir); >> + path_put(&walker); >> return ret; >> } >> >> +/** >> + * deny_no_inherit_topology_change - deny topology changes on sealed paths >> + * @subject: Subject performing the operation (contains the domain). >> + * @path: Path whose dentry is the target of the topology modification. >> + * >> + * Checks whether any domain layers are sealed against topology changes at >> + * @path. If so, emit an audit record and return -EACCES. Otherwise return 0. >> + */ >> +static int deny_no_inherit_topology_change(const struct landlock_cred_security >> + *subject, >> + const struct path *const path) > Since you're not using path->mnt here (except for a NULL check), would it > be easier to just pass the dentry instead? In that case you wouldn't have > to do an inline initializer in current_check_refer_path / hook_path_* > below as well. Yeah, this was leftover before I did some refactoring and removed the mark_no_inherit_ancestors. Good catch. I'll address this in the next version. > >> +{ >> + layer_mask_t sealed_layers = 0; >> + layer_mask_t override_layers = 0; >> + const struct landlock_rule *rule; >> + u32 layer_index; >> + unsigned long audit_layer_index; >> + >> + if (WARN_ON_ONCE(!subject || !path || !path->dentry || !path->mnt || >> + d_is_negative(path->dentry))) >> + return 0; >> + >> + rule = find_rule(subject->domain, path->dentry); >> + if (!rule) >> + return 0; >> + >> + for (layer_index = 0; layer_index < rule->num_layers; layer_index++) { >> + const struct landlock_layer *layer = &rule->layers[layer_index]; >> + layer_mask_t layer_bit = BIT_ULL(layer->level - 1); >> + >> + if (layer->flags.no_inherit || >> + layer->flags.has_no_inherit_descendant) >> + sealed_layers |= layer_bit; >> + else >> + override_layers |= layer_bit; >> + } >> + >> + sealed_layers &= ~override_layers; >> + if (!sealed_layers) >> + return 0; >> + >> + audit_layer_index = __ffs((unsigned long)sealed_layers); >> + landlock_log_denial(subject, &(struct landlock_request) { >> + .type = LANDLOCK_REQUEST_FS_CHANGE_TOPOLOGY, >> + .audit = { >> + .type = LSM_AUDIT_DATA_DENTRY, >> + .u.dentry = path->dentry, >> + }, >> + .layer_plus_one = audit_layer_index + 1, >> + }); >> + return -EACCES; >> +} >> + >> /** >> * current_check_refer_path - Check if a rename or link action is allowed >> * >> @@ -1191,6 +1284,21 @@ static int current_check_refer_path(struct dentry *const old_dentry, >> access_request_parent2 = >> get_mode_access(d_backing_inode(old_dentry)->i_mode); >> if (removable) { >> + int err = deny_no_inherit_topology_change(subject, >> + &(struct path) >> + { .mnt = new_dir->mnt, >> + .dentry = old_dentry }); >> + >> + if (err) >> + return err; >> + if (exchange) { >> + err = deny_no_inherit_topology_change(subject, >> + &(struct path) >> + { .mnt = new_dir->mnt, >> + .dentry = new_dentry }); >> + if (err) >> + return err; >> + } >> access_request_parent1 |= maybe_remove(old_dentry); >> access_request_parent2 |= maybe_remove(new_dentry); >> } >> @@ -1232,12 +1340,15 @@ static int current_check_refer_path(struct dentry *const old_dentry, >> old_dentry->d_parent; >> >> /* new_dir->dentry is equal to new_dentry->d_parent */ >> - allow_parent1 = collect_domain_accesses(subject->domain, mnt_dir.dentry, >> - old_parent, >> + allow_parent1 = collect_domain_accesses(subject->domain, >> + &mnt_dir, >> + &(struct path){ .mnt = new_dir->mnt, >> + .dentry = old_parent }, >> &layer_masks_parent1, >> &request1.rule_flags); >> - allow_parent2 = collect_domain_accesses(subject->domain, mnt_dir.dentry, >> - new_dir->dentry, >> + allow_parent2 = collect_domain_accesses(subject->domain, &mnt_dir, >> + &(struct path){ .mnt = new_dir->mnt, >> + .dentry = new_dir->dentry }, >> &layer_masks_parent2, >> &request2.rule_flags); >> >> @@ -1583,12 +1694,37 @@ static int hook_path_symlink(const struct path *const dir, >> static int hook_path_unlink(const struct path *const dir, >> struct dentry *const dentry) >> { >> + const struct landlock_cred_security *const subject = >> + landlock_get_applicable_subject(current_cred(), any_fs, NULL); >> + int err; >> + >> + if (subject) { >> + err = deny_no_inherit_topology_change(subject, >> + &(struct path) >> + { .mnt = dir->mnt, >> + .dentry = dentry }); >> + if (err) >> + return err; >> + } >> return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_FILE); >> } >> >> static int hook_path_rmdir(const struct path *const dir, >> struct dentry *const dentry) >> { >> + const struct landlock_cred_security *const subject = >> + landlock_get_applicable_subject(current_cred(), any_fs, NULL); >> + int err; >> + >> + if (subject) { >> + err = deny_no_inherit_topology_change(subject, >> + &(struct path) >> + { .mnt = dir->mnt, >> + .dentry = dentry }); >> + if (err) >> + return err; >> + } >> + >> return current_check_access_path(dir, LANDLOCK_ACCESS_FS_REMOVE_DIR); >> } >> >> [...] Overall I'm feeling pretty good about this series, but if either you or Mickaël have any more feedback I'd like to hear it. I'll wait until your next quiet flag version comes and do a rebase before sending the revisions. Sorry for the double tap Tingmao, I forgot to cc the mailing list :( Regards, Justin ^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v5 2/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT userspace api 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess 2025-12-14 17:05 ` [PATCH v5 1/6] landlock: " Justin Suess @ 2025-12-14 17:05 ` Justin Suess 2025-12-14 17:05 ` [PATCH v5 3/6] samples/landlock: Add LANDLOCK_ADD_RULE_NO_INHERIT to landlock-sandboxer Justin Suess ` (3 subsequent siblings) 5 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Implements the syscall side flag handling and kernel api headers for the LANDLOCK_ADD_RULE_NO_INHERIT flag. Signed-off-by: Justin Suess <utilityemal77@gmail.com> --- v4..v5 changes: * Moved syscall handling to this patch and moved out flag definition to allow independent build. v3..v4 changes: * Changed documentation to reflect protections now apply to VFS root instead of the mountpoint. v2..v3 changes: * Extended documentation for flag inheritance suppression on LANDLOCK_ADD_RULE_NO_INHERIT. * Extended the flag validation rules in the syscall. * Added mention of no inherit in empty rules in add_rule_path_beneath as per Tingmao Wang's suggestion. * Added check for useless no-inherit flag in networking rules. security/landlock/ruleset.c | 19 ++++++++++++++++++- security/landlock/syscalls.c | 16 ++++++++++++---- 2 files changed, 30 insertions(+), 5 deletions(-) diff --git a/security/landlock/ruleset.c b/security/landlock/ruleset.c index 750a444e1983..9152a939d79a 100644 --- a/security/landlock/ruleset.c +++ b/security/landlock/ruleset.c @@ -255,8 +255,13 @@ static int insert_rule(struct landlock_ruleset *const ruleset, return -EINVAL; if (WARN_ON_ONCE(this->layers[0].level != 0)) return -EINVAL; + /* Merge the flags into the rules */ this->layers[0].access |= (*layers)[0].access; this->layers[0].flags.quiet |= (*layers)[0].flags.quiet; + this->layers[0].flags.no_inherit |= + (*layers)[0].flags.no_inherit; + this->layers[0].flags.has_no_inherit_descendant |= + (*layers)[0].flags.has_no_inherit_descendant; return 0; } @@ -315,7 +320,10 @@ int landlock_insert_rule(struct landlock_ruleset *const ruleset, .level = 0, .flags = { .quiet = !!(flags & LANDLOCK_ADD_RULE_QUIET), - }, + .no_inherit = !!(flags & LANDLOCK_ADD_RULE_NO_INHERIT), + .has_no_inherit_descendant = + !!(flags & LANDLOCK_ADD_RULE_NO_INHERIT), + } } }; build_check_layer(); @@ -662,9 +670,18 @@ bool landlock_unmask_layers(const struct landlock_rule *const rule, unsigned long access_bit; bool is_empty; + /* Skip layers that already have no inherit flags. */ + if (rule_flags && + (rule_flags->no_inherit_masks & layer_bit)) + continue; + /* Collect rule flags for each layer. */ if (rule_flags && layer->flags.quiet) rule_flags->quiet_masks |= layer_bit; + if (rule_flags && layer->flags.no_inherit) + rule_flags->no_inherit_masks |= layer_bit; + if (rule_flags && layer->flags.has_no_inherit_descendant) + rule_flags->no_inherit_desc_masks |= layer_bit; /* * Records in @layer_masks which layer grants access to each requested diff --git a/security/landlock/syscalls.c b/security/landlock/syscalls.c index 5cf1183bb596..cffe7d944ae5 100644 --- a/security/landlock/syscalls.c +++ b/security/landlock/syscalls.c @@ -352,7 +352,7 @@ static int add_rule_path_beneath(struct landlock_ruleset *const ruleset, /* * Informs about useless rule: empty allowed_access (i.e. deny rules) * are ignored in path walks. However, the rule is not useless if it - * is there to hold a quiet flag + * is there to hold a quiet or no inherit flag. */ if (!flags && !path_beneath_attr.allowed_access) return -ENOMSG; @@ -407,6 +407,10 @@ static int add_rule_net_port(struct landlock_ruleset *ruleset, if (flags & LANDLOCK_ADD_RULE_QUIET && !ruleset->quiet_masks.net) return -EINVAL; + /* No inherit is always useless for this scope */ + if (flags & LANDLOCK_ADD_RULE_NO_INHERIT) + return -EINVAL; + /* Denies inserting a rule with port greater than 65535. */ if (net_port_attr.port > U16_MAX) return -EINVAL; @@ -424,7 +428,7 @@ static int add_rule_net_port(struct landlock_ruleset *ruleset, * @rule_type: Identify the structure type pointed to by @rule_attr: * %LANDLOCK_RULE_PATH_BENEATH or %LANDLOCK_RULE_NET_PORT. * @rule_attr: Pointer to a rule (matching the @rule_type). - * @flags: Must be 0 or %LANDLOCK_ADD_RULE_QUIET. + * @flags: Must be 0 or %LANDLOCK_ADD_RULE_QUIET and/or %LANDLOCK_ADD_RULE_NO_INHERIT. * * This system call enables to define a new rule and add it to an existing * ruleset. @@ -462,8 +466,12 @@ SYSCALL_DEFINE4(landlock_add_rule, const int, ruleset_fd, if (!is_initialized()) return -EOPNOTSUPP; - - if (flags && flags != LANDLOCK_ADD_RULE_QUIET) + /* Checks flag existence */ + if (flags & ~(LANDLOCK_ADD_RULE_QUIET | LANDLOCK_ADD_RULE_NO_INHERIT)) + return -EINVAL; + /* No inherit may only apply on path_beneath rules. */ + if ((flags & LANDLOCK_ADD_RULE_NO_INHERIT) && + rule_type != LANDLOCK_RULE_PATH_BENEATH) return -EINVAL; /* Gets and checks the ruleset. */ -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v5 3/6] samples/landlock: Add LANDLOCK_ADD_RULE_NO_INHERIT to landlock-sandboxer 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess 2025-12-14 17:05 ` [PATCH v5 1/6] landlock: " Justin Suess 2025-12-14 17:05 ` [PATCH v5 2/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT userspace api Justin Suess @ 2025-12-14 17:05 ` Justin Suess 2025-12-14 17:05 ` [PATCH v5 4/6] selftests/landlock: Implement selftests for LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess ` (2 subsequent siblings) 5 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Adds support to landlock-sandboxer with environment variable LL_FS_NO_INHERIT, which can be tagged on any filesystem object to suppress access right inheritance. Cc: Tingmao Wang <m@maowtm.org> Signed-off-by: Justin Suess <utilityemal77@gmail.com> --- v4..v5 changes: * None v3..v4 changes: * Modified LL_FS_R(O/W)_NO_INHERIT variables to a single variable to allow access rule combination. (credit to Tingmao Wang) v2..v3 changes: * Minor formatting fixes samples/landlock/sandboxer.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/samples/landlock/sandboxer.c b/samples/landlock/sandboxer.c index 07dc0013ff19..852ffa413c75 100644 --- a/samples/landlock/sandboxer.c +++ b/samples/landlock/sandboxer.c @@ -60,6 +60,7 @@ static inline int landlock_restrict_self(const int ruleset_fd, #define ENV_FS_RW_NAME "LL_FS_RW" #define ENV_FS_QUIET_NAME "LL_FS_QUIET" #define ENV_FS_QUIET_ACCESS_NAME "LL_FS_QUIET_ACCESS" +#define ENV_FS_NO_INHERIT_NAME "LL_FS_NO_INHERIT" #define ENV_TCP_BIND_NAME "LL_TCP_BIND" #define ENV_TCP_CONNECT_NAME "LL_TCP_CONNECT" #define ENV_NET_QUIET_NAME "LL_NET_QUIET" @@ -383,6 +384,7 @@ static const char help[] = "but to test audit we can set " ENV_FORCE_LOG_NAME "=1\n" ENV_FS_QUIET_NAME " and " ENV_NET_QUIET_NAME ", both optional, can then be used " "to make access to some denied paths or network ports not trigger audit logging.\n" + ENV_FS_NO_INHERIT_NAME " can be used to suppress access right propagation (ABI >= 8).\n" ENV_FS_QUIET_ACCESS_NAME " and " ENV_NET_QUIET_ACCESS_NAME " can be used to specify " "which accesses should be quieted (defaults to all):\n" "* " ENV_FS_QUIET_ACCESS_NAME ": file system accesses to quiet\n" @@ -430,6 +432,7 @@ int main(const int argc, char *const argv[], char *const *const envp) }; bool quiet_supported = true; + bool no_inherit_supported = true; int supported_restrict_flags = LANDLOCK_RESTRICT_SELF_LOG_NEW_EXEC_ON; int set_restrict_flags = 0; @@ -517,8 +520,9 @@ int main(const int argc, char *const argv[], char *const *const envp) LANDLOCK_ABI_LAST, abi); __attribute__((fallthrough)); case 7: - /* Don't add quiet flags for ABI < 8 later on. */ + /* Don't add quiet/no_inherit flags for ABI < 8 later on. */ quiet_supported = false; + no_inherit_supported = false; __attribute__((fallthrough)); case LANDLOCK_ABI_LAST: @@ -605,6 +609,13 @@ int main(const int argc, char *const argv[], char *const *const envp) goto err_close_ruleset; } + /* Don't require this env to be present. */ + if (no_inherit_supported && getenv(ENV_FS_NO_INHERIT_NAME)) { + if (populate_ruleset_fs(ENV_FS_NO_INHERIT_NAME, ruleset_fd, 0, + LANDLOCK_ADD_RULE_NO_INHERIT)) + goto err_close_ruleset; + } + if (populate_ruleset_net(ENV_TCP_BIND_NAME, ruleset_fd, LANDLOCK_ACCESS_NET_BIND_TCP, 0)) { goto err_close_ruleset; -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v5 4/6] selftests/landlock: Implement selftests for LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess ` (2 preceding siblings ...) 2025-12-14 17:05 ` [PATCH v5 3/6] samples/landlock: Add LANDLOCK_ADD_RULE_NO_INHERIT to landlock-sandboxer Justin Suess @ 2025-12-14 17:05 ` Justin Suess 2025-12-14 17:05 ` [PATCH v5 5/6] landlock: Implement KUnit test " Justin Suess 2025-12-14 17:05 ` [PATCH v5 6/6] landlock: Add documentation " Justin Suess 5 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Implements 15 selftests for the flag, covering allowed and disallowed operations on parent and child directories when this flag is set, as well as multi-layer configurations and flag inheritance / audit logging. Also tests a bind mount configuration. Signed-off-by: Justin Suess <utilityemal77@gmail.com> --- v4..v5 changes: * Fixed a bug in a test applying invalid access rights to a file. v3..v4 changes: * Added 4 new tests for bind mount handling, increasing selftests from 11 -> 15. v2..v3 changes: * Also covers flag inheritance, audit logging and LANDLOCK_ADD_RULE_QUIET suppression. * Increases number of selftests from 5 -> 11. tools/testing/selftests/landlock/fs_test.c | 730 +++++++++++++++++++++ 1 file changed, 730 insertions(+) diff --git a/tools/testing/selftests/landlock/fs_test.c b/tools/testing/selftests/landlock/fs_test.c index 44e131957fba..211c3b206710 100644 --- a/tools/testing/selftests/landlock/fs_test.c +++ b/tools/testing/selftests/landlock/fs_test.c @@ -1484,6 +1484,111 @@ TEST_F_FORK(layout1, inherit_superset) ASSERT_EQ(0, test_open(file1_s1d3, O_RDONLY)); } +TEST_F_FORK(layout1, inherit_no_inherit_flag) +{ + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW, + }; + int ruleset_fd; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + add_path_beneath(_metadata, ruleset_fd, ACCESS_RW, dir_s1d1, 0); + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d2, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* Parent directory still grants write access to its direct children. */ + EXPECT_EQ(0, test_open(dir_s1d1, O_RDONLY | O_DIRECTORY)); + EXPECT_EQ(0, test_open(file1_s1d1, O_WRONLY)); + + /* dir_s1d2 gets only its explicit read-only access rights. */ + EXPECT_EQ(0, test_open(dir_s1d2, O_RDONLY | O_DIRECTORY)); + EXPECT_EQ(0, test_open(file1_s1d2, O_RDONLY)); + EXPECT_EQ(EACCES, test_open(file1_s1d2, O_WRONLY)); + + /* Descendants of dir_s1d2 inherit the reduced access mask. */ + EXPECT_EQ(0, test_open(dir_s1d3, O_RDONLY | O_DIRECTORY)); + EXPECT_EQ(0, test_open(file1_s1d3, O_RDONLY)); + EXPECT_EQ(EACCES, test_open(file1_s1d3, O_WRONLY)); +} + +TEST_F_FORK(layout1, inherit_no_inherit_nested_levels) +{ + int ruleset_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Level 1: s1d1 (RW + REFER + REMOVE + NO_INHERIT) */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + dir_s1d1, LANDLOCK_ADD_RULE_NO_INHERIT); + + /* Level 2: s1d2 (RO + NO_INHERIT) */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d2, + LANDLOCK_ADD_RULE_NO_INHERIT); + + /* Level 3: s1d3 (RW + REFER + REMOVE + NO_INHERIT) */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + dir_s1d3, LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * Level 3: s1d3 + * - RW allowed (unlink file) + * - REFER allowed (rename file) + * - REMOVE_DIR denied (parent s1d2 is part of direct parent tree) + */ + ASSERT_EQ(0, unlink(file1_s1d3)); + ASSERT_EQ(0, rename(file2_s1d3, file1_s1d3)); + ASSERT_EQ(0, rename(file1_s1d3, file2_s1d3)); + ASSERT_EQ(-1, rmdir(dir_s1d3)); + ASSERT_EQ(EACCES, errno); + + /* + * Level 2: s1d2 + * - RW denied (unlink file), layer is RO + * - REFER denied (rename file) + * - REMOVE_DIR of s1d2 not allowed (parent s1d1 is part of direct parent tree) + */ + ASSERT_EQ(-1, unlink(file1_s1d2)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rename(file2_s1d2, file1_s1d2)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rmdir(dir_s1d2)); + ASSERT_EQ(EACCES, errno); + + /* + * Level 1: s1d1 + * - RW allowed + * - Rename allowed (except for direct parent tree s1d2) + * - REMOVE_DIR denied (parent tmp is denied) + */ + ASSERT_EQ(0, unlink(file1_s1d1)); + ASSERT_EQ(0, rename(file2_s1d1, file1_s1d1)); + ASSERT_EQ(0, rename(file1_s1d1, file2_s1d1)); + ASSERT_EQ(-1, rmdir(dir_s1d1)); + ASSERT_EQ(EACCES, errno); +} + TEST_F_FORK(layout0, max_layers) { int i, err; @@ -4408,6 +4513,266 @@ TEST_F_FORK(layout1, named_unix_domain_socket_ioctl) ASSERT_EQ(0, close(cli_fd)); } +TEST_F_FORK(layout1, inherit_no_inherit_topology_dir) +{ + const struct rule rules[] = { + { + .path = TMP_DIR, + .access = ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + }, + {}, + }; + int ruleset_fd; + + ruleset_fd = create_ruleset(_metadata, + ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + rules); + ASSERT_LE(0, ruleset_fd); + + /* Adds a no-inherit rule on a leaf directory. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d3, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * Topology modifications of the rule path and its parents are denied. + */ + + /* Target directory s1d3 */ + ASSERT_EQ(-1, rmdir(dir_s1d3)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rename(dir_s1d3, dir_s2d3)); + ASSERT_EQ(EACCES, errno); + + /* Parent directory s1d2 */ + ASSERT_EQ(-1, rmdir(dir_s1d2)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rename(dir_s1d2, dir_s2d2)); + ASSERT_EQ(EACCES, errno); + + /* Grandparent directory s1d1 */ + ASSERT_EQ(-1, rmdir(dir_s1d1)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rename(dir_s1d1, dir_s2d1)); + ASSERT_EQ(EACCES, errno); + + /* + * Sibling operations are allowed. + */ + /* Sibling of s1d3 */ + ASSERT_EQ(0, unlink(file1_s1d2)); + /* Sibling of s1d2 */ + ASSERT_EQ(0, unlink(file1_s1d1)); + + /* + * Content of the no-inherit directory is restricted by the rule (RO). + */ + ASSERT_EQ(-1, unlink(file1_s1d3)); + ASSERT_EQ(EACCES, errno); +} + +TEST_F_FORK(layout1, no_inherit_allow_inner_removal) +{ + int ruleset_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, dir_s1d2, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * Content of the no-inherit directory is mutable (RW). + * This checks that the no-inherit flag does not seal the content. + */ + ASSERT_EQ(0, unlink(file1_s1d2)); + + /* + * Topology modifications of the rule path are denied. + */ + ASSERT_EQ(-1, rmdir(dir_s1d2)); + ASSERT_EQ(EACCES, errno); + ASSERT_EQ(-1, rename(dir_s1d2, dir_s2d2)); + ASSERT_EQ(EACCES, errno); +} + +TEST_F_FORK(layout1, inherit_no_inherit_topology_unrelated) +{ + const struct rule rules[] = { + { + .path = TMP_DIR, + .access = ACCESS_RW, + }, + {}, + }; + static const char unrelated_dir[] = TMP_DIR "/s2d1/unrelated"; + static const char unrelated_file[] = TMP_DIR "/s2d1/unrelated/f1"; + int ruleset_fd; + + ruleset_fd = create_ruleset(_metadata, ACCESS_RW, rules); + ASSERT_LE(0, ruleset_fd); + + /* Adds a no-inherit rule on a leaf directory unrelated to s2. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d3, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* Ensure we can still create and delete files outside the sealed branch. */ + ASSERT_EQ(0, mkdir(unrelated_dir, 0700)); + ASSERT_EQ(0, mknod(unrelated_file, S_IFREG | 0600, 0)); + ASSERT_EQ(0, unlink(unrelated_file)); + ASSERT_EQ(0, rmdir(unrelated_dir)); + + /* Existing siblings in s2 remain modifiable. */ + ASSERT_EQ(0, unlink(file1_s2d1)); + ASSERT_EQ(0, mknod(file1_s2d1, S_IFREG | 0700, 0)); +} + +TEST_F_FORK(layout1, inherit_no_inherit_descendant_rw) +{ + const struct rule rules[] = { + { + .path = TMP_DIR, + .access = ACCESS_RO, + }, + {}, + }; + const __u64 handled_access = ACCESS_RW | LANDLOCK_ACCESS_FS_MAKE_REG | + LANDLOCK_ACCESS_FS_REMOVE_FILE; + static const char child_file[] = + TMP_DIR "/s1d1/s1d2/s1d3/rw_descendant"; + int ruleset_fd; + + ruleset_fd = create_ruleset(_metadata, handled_access, rules); + ASSERT_LE(0, ruleset_fd); + + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d2, + LANDLOCK_ADD_RULE_NO_INHERIT); + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_MAKE_REG | + LANDLOCK_ACCESS_FS_REMOVE_FILE, + dir_s1d3, 0); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + ASSERT_EQ(0, mknod(child_file, S_IFREG | 0600, 0)); + ASSERT_EQ(0, unlink(child_file)); +} + +TEST_F_FORK(layout1, inherit_no_inherit_topology_file) +{ + const struct rule rules[] = { + { + .path = TMP_DIR, + .access = ACCESS_RW, + }, + {}, + }; + int ruleset_fd; + + /* + * Both file1_s1d2 and file2_s1d2 already exist from the fixture. + * file2_s1d2 is in the same directory as file1_s1d2 and will be + * used to test inheritance vs. NO_INHERIT behavior. + */ + + ruleset_fd = create_ruleset(_metadata, ACCESS_RW, rules); + ASSERT_LE(0, ruleset_fd); + + /* + * Add a NO_INHERIT rule on file1_s1d2 with READ_FILE access. + * This should succeed (files can have NO_INHERIT). + * Use READ_FILE (not ACCESS_RO which includes READ_DIR) since + * directory access rights don't make sense for files. + */ + add_path_beneath(_metadata, ruleset_fd, LANDLOCK_ACCESS_FS_READ_FILE, + file1_s1d2, LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * file1_s1d2 has NO_INHERIT with READ_FILE access only, + * so it should only be readable (not inheriting RW from parent TMP_DIR). + */ + ASSERT_EQ(0, test_open(file1_s1d2, O_RDONLY)); + ASSERT_EQ(EACCES, test_open(file1_s1d2, O_WRONLY)); + + /* + * file2_s1d2 does not have NO_INHERIT, so it should inherit + * RW access from parent TMP_DIR rule. + */ + ASSERT_EQ(0, test_open(file2_s1d2, O_RDONLY)); + ASSERT_EQ(0, test_open(file2_s1d2, O_WRONLY)); +} + +TEST_F_FORK(layout1, inherit_no_inherit_layered) +{ + const struct rule layer1_and_2[] = { + { + .path = TMP_DIR, + .access = ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + }, + {}, + }; + int ruleset_fd; + static const char unrelated_dir[] = TMP_DIR "/s2d1/unrelated"; + static const char unrelated_file[] = TMP_DIR "/s2d1/unrelated/f1"; + + /* Layer 1: RW on TMP_DIR */ + ruleset_fd = create_ruleset(_metadata, + ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + layer1_and_2); + ASSERT_LE(0, ruleset_fd); + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* Layer 2: Add no-inherit RO rule on s1d2 */ + ruleset_fd = create_ruleset(_metadata, + ACCESS_RW | LANDLOCK_ACCESS_FS_REMOVE_FILE, + layer1_and_2); + ASSERT_LE(0, ruleset_fd); + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d2, + LANDLOCK_ADD_RULE_NO_INHERIT); + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* Operations in unrelated areas should still work */ + ASSERT_EQ(0, mkdir(unrelated_dir, 0700)); + ASSERT_EQ(0, mknod(unrelated_file, S_IFREG | 0600, 0)); + ASSERT_EQ(0, unlink(unrelated_file)); + ASSERT_EQ(0, rmdir(unrelated_dir)); + + /* Creating in s1d1 should be allowed (parent still has RW) */ + ASSERT_EQ(0, mknod(TMP_DIR "/s1d1/newfile", S_IFREG | 0600, 0)); + ASSERT_EQ(0, unlink(TMP_DIR "/s1d1/newfile")); + + /* Content of s1d2 should be read-only */ + ASSERT_EQ(-1, unlink(file1_s1d2)); + ASSERT_EQ(EACCES, errno); + + /* Topology changes to s1d2 should be denied */ + ASSERT_EQ(-1, rename(dir_s1d2, TMP_DIR "/s2d1/renamed")); + ASSERT_EQ(EACCES, errno); + + /* Renaming s1d1 should also be denied (it's an ancestor) */ + ASSERT_EQ(-1, rename(dir_s1d1, TMP_DIR "/s2d1/renamed")); + ASSERT_EQ(EACCES, errno); +} + /* clang-format off */ FIXTURE(ioctl) {}; @@ -5747,6 +6112,277 @@ TEST_F_FORK(layout4_disconnected_leafs, read_rename_exchange) test_renameat(s1d42_bind_fd, "f4", s1d42_bind_fd, "f5")); } +/* + * Test that LANDLOCK_ADD_RULE_NO_INHERIT on a directory accessed via a mount + * point protects the parent hierarchy within the mount from topology changes. + * + * Layout (after bind mount s1d2 -> s2d2): + * tmp + * ├── s1d1 + * │ └── s1d2 [source of bind mount] + * │ ├── s1d31 + * │ │ └── s1d41 + * │ │ ├── f1 + * │ │ └── f2 + * │ └── s1d32 + * │ └── s1d42 + * │ ├── f3 + * │ └── f4 + * └── s2d1 + * └── s2d2 [bind mount destination from s1d2] + * ├── s1d31 <- parent of protected dir, should be immovable + * │ └── s1d41 <- protected with NO_INHERIT + * │ ├── f1 + * │ └── f2 + * └── s1d32 + * └── s1d42 + * ├── f3 + * └── f4 + * + * When s1d41 (accessed via the mount at s2d2) is protected with NO_INHERIT, + * its parent directories within the mount (s1d31) should be immovable. + */ +TEST_F_FORK(layout4_disconnected_leafs, no_inherit_mount_parent_rename) +{ + int ruleset_fd, s1d41_bind_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Allow full access to TMP_DIR. */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + TMP_DIR, 0); + + /* + * Access s1d41 through the bind mount at s2d2 and protect it with + * NO_INHERIT. This should seal the parent hierarchy through the mount. + */ + s1d41_bind_fd = open(TMP_DIR "/s2d1/s2d2/s1d31/s1d41", + O_DIRECTORY | O_PATH | O_CLOEXEC); + ASSERT_LE(0, s1d41_bind_fd); + + ASSERT_EQ(0, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH, + &(struct landlock_path_beneath_attr){ + .parent_fd = s1d41_bind_fd, + .allowed_access = ACCESS_RO, + }, + LANDLOCK_ADD_RULE_NO_INHERIT)); + EXPECT_EQ(0, close(s1d41_bind_fd)); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * s1d31 is the parent of s1d41 within the mount. Renaming it should + * be denied because it is part of the protected parent hierarchy. + * Test via the mount path. + */ + ASSERT_EQ(-1, rename(TMP_DIR "/s2d1/s2d2/s1d31", + TMP_DIR "/s2d1/s2d2/s1d31_renamed")); + ASSERT_EQ(EACCES, errno); + + /* + * s1d32 is a sibling directory (not in the protected parent chain), + * so renaming it should be allowed. + */ + ASSERT_EQ(0, rename(TMP_DIR "/s2d1/s2d2/s1d32", + TMP_DIR "/s2d1/s2d2/s1d32_renamed")); + ASSERT_EQ(0, rename(TMP_DIR "/s2d1/s2d2/s1d32_renamed", + TMP_DIR "/s2d1/s2d2/s1d32")); + + /* + * Renaming directories not in the protected parent hierarchy should + * still be allowed. + */ + ASSERT_EQ(0, rename(TMP_DIR "/s3d1", TMP_DIR "/s3d1_renamed")); + ASSERT_EQ(0, rename(TMP_DIR "/s3d1_renamed", TMP_DIR "/s3d1")); +} + +TEST_F_FORK(layout4_disconnected_leafs, no_inherit_mount_parent_rmdir) +{ + int ruleset_fd, s1d41_bind_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Allow full access to TMP_DIR. */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + TMP_DIR, 0); + + /* + * Access s1d41 through the bind mount at s2d2 and protect it with + * NO_INHERIT. This should seal the parent hierarchy through the mount. + */ + s1d41_bind_fd = open(TMP_DIR "/s2d1/s2d2/s1d31/s1d41", + O_DIRECTORY | O_PATH | O_CLOEXEC); + ASSERT_LE(0, s1d41_bind_fd); + + ASSERT_EQ(0, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH, + &(struct landlock_path_beneath_attr){ + .parent_fd = s1d41_bind_fd, + .allowed_access = ACCESS_RO, + }, + LANDLOCK_ADD_RULE_NO_INHERIT)); + EXPECT_EQ(0, close(s1d41_bind_fd)); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * s1d31 is the parent of s1d41 within the mount. Removing it should + * be denied because it is part of the protected parent hierarchy. + */ + ASSERT_EQ(-1, rmdir(TMP_DIR "/s2d1/s2d2/s1d31")); + ASSERT_EQ(EACCES, errno); + + /* + * Removing an unrelated directory should still be allowed (if empty). + */ + ASSERT_EQ(0, rmdir(TMP_DIR "/s3d1")); + ASSERT_EQ(0, mkdir(TMP_DIR "/s3d1", 0755)); +} + +TEST_F_FORK(layout4_disconnected_leafs, no_inherit_mount_parent_link) +{ + int ruleset_fd, s1d41_bind_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR | + LANDLOCK_ACCESS_FS_MAKE_REG, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Allow full access to TMP_DIR. */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR | + LANDLOCK_ACCESS_FS_MAKE_REG, + TMP_DIR, 0); + + /* + * Access s1d41 through the bind mount at s2d2 and protect it with + * NO_INHERIT. This should seal the parent hierarchy through the mount. + */ + s1d41_bind_fd = open(TMP_DIR "/s2d1/s2d2/s1d31/s1d41", + O_DIRECTORY | O_PATH | O_CLOEXEC); + ASSERT_LE(0, s1d41_bind_fd); + + ASSERT_EQ(0, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH, + &(struct landlock_path_beneath_attr){ + .parent_fd = s1d41_bind_fd, + .allowed_access = ACCESS_RO, + }, + LANDLOCK_ADD_RULE_NO_INHERIT)); + EXPECT_EQ(0, close(s1d41_bind_fd)); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * Creating a hard link within the protected NO_INHERIT directory should + * be denied because NO_INHERIT grants only ACCESS_RO (no MAKE_REG). + */ + ASSERT_EQ(-1, linkat(AT_FDCWD, TMP_DIR "/s2d1/s2d2/s1d31/s1d41/f1", + AT_FDCWD, TMP_DIR "/s2d1/s2d2/s1d31/s1d41/f1_link", + 0)); + ASSERT_EQ(EACCES, errno); + + /* + * Creating links within directories outside the protected chain + * (using the mount source path to avoid EXDEV) should still be allowed. + */ + ASSERT_EQ(0, linkat(AT_FDCWD, TMP_DIR "/s1d1/s1d2/s1d32/s1d42/f3", + AT_FDCWD, TMP_DIR "/s1d1/s1d2/s1d32/s1d42/f3_link", + 0)); + ASSERT_EQ(0, unlink(TMP_DIR "/s1d1/s1d2/s1d32/s1d42/f3_link")); +} + +/* + * Test that NO_INHERIT protection extends to the mount source hierarchy. + * If a directory is protected via a mount path, its parents within the + * mount source should also be protected from topology changes. + */ +TEST_F_FORK(layout4_disconnected_leafs, no_inherit_source_parent_rename) +{ + int ruleset_fd, s1d41_bind_fd; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + }; + + ruleset_fd = + landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Allow full access to TMP_DIR. */ + add_path_beneath(_metadata, ruleset_fd, + ACCESS_RW | LANDLOCK_ACCESS_FS_REFER | + LANDLOCK_ACCESS_FS_REMOVE_FILE | + LANDLOCK_ACCESS_FS_REMOVE_DIR, + TMP_DIR, 0); + + /* + * Access s1d41 through the bind mount at s2d2 and protect it with + * NO_INHERIT. The source mount path parents should also be protected. + */ + s1d41_bind_fd = open(TMP_DIR "/s2d1/s2d2/s1d31/s1d41", + O_DIRECTORY | O_PATH | O_CLOEXEC); + ASSERT_LE(0, s1d41_bind_fd); + + ASSERT_EQ(0, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_PATH_BENEATH, + &(struct landlock_path_beneath_attr){ + .parent_fd = s1d41_bind_fd, + .allowed_access = ACCESS_RO, + }, + LANDLOCK_ADD_RULE_NO_INHERIT)); + EXPECT_EQ(0, close(s1d41_bind_fd)); + + enforce_ruleset(_metadata, ruleset_fd); + ASSERT_EQ(0, close(ruleset_fd)); + + /* + * The mount source is s1d1/s1d2. The protected directory s1d41 is at + * s1d1/s1d2/s1d31/s1d41. The parent s1d31 within the mount source + * should be protected from topology changes. + */ + ASSERT_EQ(-1, rename(TMP_DIR "/s1d1/s1d2/s1d31", + TMP_DIR "/s1d1/s1d2/s1d31_renamed")); + ASSERT_EQ(EACCES, errno); + + /* + * s1d32 is a sibling, not in the protected parent chain. It should + * be renamable. + */ + ASSERT_EQ(0, rename(TMP_DIR "/s1d1/s1d2/s1d32", + TMP_DIR "/s1d1/s1d2/s1d32_renamed")); + ASSERT_EQ(0, rename(TMP_DIR "/s1d1/s1d2/s1d32_renamed", + TMP_DIR "/s1d1/s1d2/s1d32")); +} + /* * layout5_disconnected_branch before rename: * @@ -7231,6 +7867,100 @@ TEST_F(audit_layout1, write_file) EXPECT_EQ(1, records.domain); } +TEST_F(audit_layout1, no_inherit_parent_is_logged) +{ + struct audit_records records; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW, + }; + int ruleset_fd; + + ruleset_fd = landlock_create_ruleset(&ruleset_attr, + sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Base read-only rule at s1d1. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d1, 0); + /* Descendant s1d1/s1d2/s1d3 forbids inheritance but should still log. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d3, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + + EXPECT_EQ(EACCES, test_open(file1_s1d2, O_WRONLY)); + EXPECT_EQ(0, matches_log_fs(_metadata, self->audit_fd, + "fs\\.write_file", file1_s1d2)); + EXPECT_EQ(0, audit_count_records(self->audit_fd, &records)); + EXPECT_EQ(0, records.access); + EXPECT_EQ(1, records.domain); + + EXPECT_EQ(0, close(ruleset_fd)); +} + +TEST_F(audit_layout1, no_inherit_blocks_quiet_flag_inheritance) +{ + struct audit_records records; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW, + .quiet_access_fs = ACCESS_RW, + }; + int ruleset_fd; + + ruleset_fd = landlock_create_ruleset(&ruleset_attr, + sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Base read-only rule at tmp/s1d1 with quiet flag. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d1, + LANDLOCK_ADD_RULE_QUIET); + /* Descendant tmp/s1d1/s1d2/s1d3 forbids inheritance of quiet flag and should still log. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d3, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + + EXPECT_EQ(EACCES, test_open(file1_s1d3, O_WRONLY)); + EXPECT_EQ(0, matches_log_fs(_metadata, self->audit_fd, + "fs\\.write_file", file1_s1d3)); + EXPECT_EQ(0, audit_count_records(self->audit_fd, &records)); + EXPECT_EQ(0, records.access); + EXPECT_EQ(1, records.domain); + + EXPECT_EQ(0, close(ruleset_fd)); +} + +TEST_F(audit_layout1, no_inherit_quiet_parent) +{ + struct audit_records records; + struct landlock_ruleset_attr ruleset_attr = { + .handled_access_fs = ACCESS_RW, + .quiet_access_fs = ACCESS_RW, + }; + int ruleset_fd; + + ruleset_fd = landlock_create_ruleset(&ruleset_attr, + sizeof(ruleset_attr), 0); + ASSERT_LE(0, ruleset_fd); + + /* Base read-only rule at tmp/s1d1 with quiet flag. */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d1, + LANDLOCK_ADD_RULE_QUIET); + /* Access to dir_s1d1 shouldn't log */ + add_path_beneath(_metadata, ruleset_fd, ACCESS_RO, dir_s1d3, + LANDLOCK_ADD_RULE_NO_INHERIT); + + enforce_ruleset(_metadata, ruleset_fd); + + EXPECT_EQ(EACCES, test_open(file1_s1d1, O_WRONLY)); + EXPECT_NE(0, matches_log_fs(_metadata, self->audit_fd, + "fs\\.write_file", file1_s1d1)); + EXPECT_EQ(0, audit_count_records(self->audit_fd, &records)); + EXPECT_EQ(0, records.access); + EXPECT_EQ(0, records.domain); + + EXPECT_EQ(0, close(ruleset_fd)); +} + TEST_F(audit_layout1, read_file) { struct audit_records records; -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v5 5/6] landlock: Implement KUnit test for LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess ` (3 preceding siblings ...) 2025-12-14 17:05 ` [PATCH v5 4/6] selftests/landlock: Implement selftests for LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess @ 2025-12-14 17:05 ` Justin Suess 2025-12-14 17:05 ` [PATCH v5 6/6] landlock: Add documentation " Justin Suess 5 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Add a unit test for rule_flag collection, ensuring that access masks are properly propagated with the flags. Signed-off-by: Justin Suess <utilityemal77@gmail.com> --- v4..v5 changes: * None v2..v3 changes: * Removing erroneously misplaced code and placed in the proper patch. security/landlock/ruleset.c | 89 +++++++++++++++++++++++++++++++++++++ 1 file changed, 89 insertions(+) diff --git a/security/landlock/ruleset.c b/security/landlock/ruleset.c index 9152a939d79a..8064139fde8f 100644 --- a/security/landlock/ruleset.c +++ b/security/landlock/ruleset.c @@ -22,6 +22,7 @@ #include <linux/spinlock.h> #include <linux/workqueue.h> #include <uapi/linux/landlock.h> +#include <kunit/test.h> #include "access.h" #include "audit.h" @@ -770,3 +771,91 @@ landlock_init_layer_masks(const struct landlock_ruleset *const domain, } return handled_accesses; } + +#ifdef CONFIG_SECURITY_LANDLOCK_KUNIT_TEST + +/** + * test_unmask_layers_no_inherit - Test landlock_unmask_layers() with no_inherit + * @test: The KUnit test context. + */ +static void test_unmask_layers_no_inherit(struct kunit *const test) +{ + struct landlock_rule *rule; + layer_mask_t layer_masks[LANDLOCK_NUM_ACCESS_FS]; + struct collected_rule_flags rule_flags; + const access_mask_t access_request = BIT_ULL(0) | BIT_ULL(1); + const layer_mask_t layers_initialized = BIT_ULL(0) | BIT_ULL(1); + size_t i; + + rule = kzalloc(struct_size(rule, layers, 2), GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, rule); + + rule->num_layers = 2; + + /* Layer 1: allows access 0, no_inherit */ + rule->layers[0].level = 1; + rule->layers[0].access = BIT_ULL(0); + rule->layers[0].flags.no_inherit = 1; + + /* Layer 2: allows access 1 */ + rule->layers[1].level = 2; + rule->layers[1].access = BIT_ULL(1); + + /* Case 1: No rule_flags provided (should behave normally) */ + for (i = 0; i < ARRAY_SIZE(layer_masks); i++) + layer_masks[i] = layers_initialized; + + landlock_unmask_layers(rule, access_request, &layer_masks, + ARRAY_SIZE(layer_masks), NULL); + + /* Access 0 should be unmasked by layer 1 */ + KUNIT_EXPECT_EQ(test, layer_masks[0], layers_initialized & ~BIT_ULL(0)); + /* Access 1 should be unmasked by layer 2 */ + KUNIT_EXPECT_EQ(test, layer_masks[1], layers_initialized & ~BIT_ULL(1)); + + /* Case 2: rule_flags provided, no existing no_inherit_masks */ + for (i = 0; i < ARRAY_SIZE(layer_masks); i++) + layer_masks[i] = layers_initialized; + memset(&rule_flags, 0, sizeof(rule_flags)); + + landlock_unmask_layers(rule, access_request, &layer_masks, + ARRAY_SIZE(layer_masks), &rule_flags); + + /* Access 0 should be unmasked by layer 1 */ + KUNIT_EXPECT_EQ(test, layer_masks[0], layers_initialized & ~BIT_ULL(0)); + /* Access 1 should be unmasked by layer 2 */ + KUNIT_EXPECT_EQ(test, layer_masks[1], layers_initialized & ~BIT_ULL(1)); + + /* rule_flags should collect no_inherit from layer 1 */ + KUNIT_EXPECT_EQ(test, rule_flags.no_inherit_masks, (layer_mask_t)BIT_ULL(0)); + + /* Case 3: rule_flags provided, layer 1 is masked by no_inherit_masks */ + for (i = 0; i < ARRAY_SIZE(layer_masks); i++) + layer_masks[i] = layers_initialized; + memset(&rule_flags, 0, sizeof(rule_flags)); + rule_flags.no_inherit_masks = BIT_ULL(0); /* Mask layer 1 */ + + landlock_unmask_layers(rule, access_request, &layer_masks, + ARRAY_SIZE(layer_masks), &rule_flags); + + /* Access 0 should NOT be unmasked by layer 1 because it is skipped */ + KUNIT_EXPECT_EQ(test, layer_masks[0], layers_initialized); + /* Access 1 should be unmasked by layer 2 */ + KUNIT_EXPECT_EQ(test, layer_masks[1], layers_initialized & ~BIT_ULL(1)); + + kfree(rule); +} + +static struct kunit_case ruleset_test_cases[] = { + KUNIT_CASE(test_unmask_layers_no_inherit), + {} +}; + +static struct kunit_suite ruleset_test_suite = { + .name = "landlock_ruleset", + .test_cases = ruleset_test_cases, +}; + +kunit_test_suite(ruleset_test_suite); + +#endif /* CONFIG_SECURITY_LANDLOCK_KUNIT_TEST */ -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v5 6/6] landlock: Add documentation for LANDLOCK_ADD_RULE_NO_INHERIT 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess ` (4 preceding siblings ...) 2025-12-14 17:05 ` [PATCH v5 5/6] landlock: Implement KUnit test " Justin Suess @ 2025-12-14 17:05 ` Justin Suess 5 siblings, 0 replies; 9+ messages in thread From: Justin Suess @ 2025-12-14 17:05 UTC (permalink / raw) To: Mickaël Salaün Cc: Tingmao Wang, Günther Noack, Justin Suess, Jan Kara, Abhinav Saxena, linux-security-module Adds documentation of the flag to the userspace api, describing the functionality of the flag and parent directory protections. Signed-off-by: Justin Suess <utilityemal77@gmail.com> --- v1..v5 changes: * Initial addition Documentation/userspace-api/landlock.rst | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/Documentation/userspace-api/landlock.rst b/Documentation/userspace-api/landlock.rst index 1d0c2c15c22e..3671cd90fbe2 100644 --- a/Documentation/userspace-api/landlock.rst +++ b/Documentation/userspace-api/landlock.rst @@ -604,6 +604,23 @@ Landlock audit events with the ``LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF``, sys_landlock_restrict_self(). See Documentation/admin-guide/LSM/landlock.rst for more details on audit. +Filesystem inheritance suppression (ABI < 8) +----------------- + +Starting with the Landlock ABI version 8, it is possible to prevent a directory +or file from inheriting it's parent's access grants by using the +``LANDLOCK_ADD_RULE_NO_INHERIT`` flag passed to sys_landlock_add_rule(). This +can be useful for policies where a parent directory needs broader access than its +children. + +To mitigate sandbox-restart attacks, the inode itself, and ancestors of inodes +tagged with ``LANDLOCK_ADD_RULE_NO_INHERIT`` cannot be removed, renamed, +reparented, or linked into/from other directories. + +These parent directory protections propagate up to the root. Further inheritance +for grants originating beneath a ``LANDLOCK_ADD_RULE_NO_INHERIT`` tagged inode +are not affected unless also tagged with this flag. + .. _kernel_support: Kernel support -- 2.51.0 ^ permalink raw reply related [flat|nested] 9+ messages in thread
end of thread, other threads:[~2025-12-15 22:21 UTC | newest] Thread overview: 9+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2025-12-14 17:05 [PATCH v5 0/6] Implement LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess 2025-12-14 17:05 ` [PATCH v5 1/6] landlock: " Justin Suess 2025-12-14 22:53 ` Tingmao Wang 2025-12-15 22:21 ` Justin Suess 2025-12-14 17:05 ` [PATCH v5 2/6] landlock: Implement LANDLOCK_ADD_RULE_NO_INHERIT userspace api Justin Suess 2025-12-14 17:05 ` [PATCH v5 3/6] samples/landlock: Add LANDLOCK_ADD_RULE_NO_INHERIT to landlock-sandboxer Justin Suess 2025-12-14 17:05 ` [PATCH v5 4/6] selftests/landlock: Implement selftests for LANDLOCK_ADD_RULE_NO_INHERIT Justin Suess 2025-12-14 17:05 ` [PATCH v5 5/6] landlock: Implement KUnit test " Justin Suess 2025-12-14 17:05 ` [PATCH v5 6/6] landlock: Add documentation " Justin Suess
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).