-----BEGIN PGP SIGNATURE-----
iIYEABYKAC4WIQSVyBthFV4iTW/VU1/l49DojIL20gUCZ+bGgBAcbWljQGRpZ2lr
b2QubmV0AAoJEOXj0OiMgvbSKmgBAICZsmQTuKMHIXdB7kwA+BX5k++SZcyA+qHN
0hrJTSMsAP0Uv6NpiPT4CTduqBMRbuMwNhujBczRiok6yaHDbC8eCw==
=K8XL
-----END PGP SIGNATURE-----
Merge tag 'landlock-6.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux
Pull landlock updates from Mickaël Salaün:
"This brings two main changes to Landlock:
- A signal scoping fix with a new interface for user space to know if
it is compatible with the running kernel.
- Audit support to give visibility on why access requests are denied,
including the origin of the security policy, missing access rights,
and description of object(s). This was designed to limit log spam
as much as possible while still alerting about unexpected blocked
access.
With these changes come new and improved documentation, and a lot of
new tests"
* tag 'landlock-6.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux: (36 commits)
landlock: Add audit documentation
selftests/landlock: Add audit tests for network
selftests/landlock: Add audit tests for filesystem
selftests/landlock: Add audit tests for abstract UNIX socket scoping
selftests/landlock: Add audit tests for ptrace
selftests/landlock: Test audit with restrict flags
selftests/landlock: Add tests for audit flags and domain IDs
selftests/landlock: Extend tests for landlock_restrict_self(2)'s flags
selftests/landlock: Add test for invalid ruleset file descriptor
samples/landlock: Enable users to log sandbox denials
landlock: Add LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF
landlock: Add LANDLOCK_RESTRICT_SELF_LOG_*_EXEC_* flags
landlock: Log scoped denials
landlock: Log TCP bind and connect denials
landlock: Log truncate and IOCTL denials
landlock: Factor out IOCTL hooks
landlock: Log file-related denials
landlock: Log mount-related denials
landlock: Add AUDIT_LANDLOCK_DOMAIN and log domain status
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials
...
Add LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF for the case of sandboxer
tools, init systems, or runtime containers launching programs sandboxing
themselves in an inconsistent way. Setting this flag should only
depends on runtime configuration (i.e. not hardcoded).
We don't create a new ruleset's option because this should not be part
of the security policy: only the task that enforces the policy (not the
one that create it) knows if itself or its children may request denied
actions.
This is the first and only flag that can be set without actually
restricting the caller (i.e. without providing a ruleset).
Extend struct landlock_cred_security with a u8 log_subdomains_off.
struct landlock_file_security is still 16 bytes.
Cc: Günther Noack <gnoack@google.com>
Cc: Paul Moore <paul@paul-moore.com>
Closes: https://github.com/landlock-lsm/linux/issues/3
Link: https://lore.kernel.org/r/20250320190717.2287696-19-mic@digikod.net
[mic: Fix comment]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Most of the time we want to log denied access because they should not
happen and such information helps diagnose issues. However, when
sandboxing processes that we know will try to access denied resources
(e.g. unknown, bogus, or malicious binary), we might want to not log
related access requests that might fill up logs.
By default, denied requests are logged until the task call execve(2).
If the LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF flag is set, denied
requests will not be logged for the same executed file.
If the LANDLOCK_RESTRICT_SELF_LOG_NEW_EXEC_ON flag is set, denied
requests from after an execve(2) call will be logged.
The rationale is that a program should know its own behavior, but not
necessarily the behavior of other programs.
Because LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF is set for a specific
Landlock domain, it makes it possible to selectively mask some access
requests that would be logged by a parent domain, which might be handy
for unprivileged processes to limit logs. However, system
administrators should still use the audit filtering mechanism. There is
intentionally no audit nor sysctl configuration to re-enable these logs.
This is delegated to the user space program.
Increment the Landlock ABI version to reflect this interface change.
Cc: Günther Noack <gnoack@google.com>
Cc: Paul Moore <paul@paul-moore.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-18-mic@digikod.net
[mic: Rename variables and fix __maybe_unused]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Add audit support for unix_stream_connect, unix_may_send, task_kill, and
file_send_sigiotask hooks.
The related blockers are:
- scope.abstract_unix_socket
- scope.signal
Audit event sample for abstract unix socket:
type=LANDLOCK_DENY msg=audit(1729738800.268:30): domain=195ba459b blockers=scope.abstract_unix_socket path=00666F6F
Audit event sample for signal:
type=LANDLOCK_DENY msg=audit(1729738800.291:31): domain=195ba459b blockers=scope.signal opid=1 ocomm="systemd"
Refactor and simplify error handling in LSM hooks.
Extend struct landlock_file_security with fown_layer and use it to log
the blocking domain. The struct aligned size is still 16 bytes.
Cc: Günther Noack <gnoack@google.com>
Cc: Tahera Fahimi <fahimitahera@gmail.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-17-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Add audit support to the file_truncate and file_ioctl hooks.
Add a deny_masks_t type and related helpers to store the domain's layer
level per optional access rights (i.e. LANDLOCK_ACCESS_FS_TRUNCATE and
LANDLOCK_ACCESS_FS_IOCTL_DEV) when opening a file, which cannot be
inferred later. In practice, the landlock_file_security aligned blob size is
still 16 bytes because this new one-byte deny_masks field follows the
existing two-bytes allowed_access field and precede the packed
fown_subject.
Implementing deny_masks_t with a bitfield instead of a struct enables a
generic implementation to store and extract layer levels.
Add KUnit tests to check the identification of a layer level from a
deny_masks_t, and the computation of a deny_masks_t from an access right
with its layer level or a layer_mask_t array.
Audit event sample:
type=LANDLOCK_DENY msg=audit(1729738800.349:44): domain=195ba459b blockers=fs.ioctl_dev path="/dev/tty" dev="devtmpfs" ino=9 ioctlcmd=0x5401
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-15-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Compat and non-compat IOCTL hooks are almost the same, except to compare
the IOCTL command. Factor out these two IOCTL hooks to highlight the
difference and minimize audit changes (see next commit).
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-14-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Add audit support for path_mkdir, path_mknod, path_symlink, path_unlink,
path_rmdir, path_truncate, path_link, path_rename, and file_open hooks.
The dedicated blockers are:
- fs.execute
- fs.write_file
- fs.read_file
- fs.read_dir
- fs.remove_dir
- fs.remove_file
- fs.make_char
- fs.make_dir
- fs.make_reg
- fs.make_sock
- fs.make_fifo
- fs.make_block
- fs.make_sym
- fs.refer
- fs.truncate
- fs.ioctl_dev
Audit event sample for a denied link action:
type=LANDLOCK_DENY msg=audit(1729738800.349:44): domain=195ba459b blockers=fs.refer path="/usr/bin" dev="vda2" ino=351
type=LANDLOCK_DENY msg=audit(1729738800.349:44): domain=195ba459b blockers=fs.make_reg,fs.refer path="/usr/local" dev="vda2" ino=365
We could pack blocker names (e.g. "fs:make_reg,refer") but that would
increase complexity for the kernel and log parsers. Moreover, this
could not handle blockers of different classes (e.g. fs and net). Make
it simple and flexible instead.
Add KUnit tests to check the identification from a layer_mask_t array of
the first layer level denying such request.
Cc: Günther Noack <gnoack@google.com>
Depends-on: 058518c20920 ("landlock: Align partial refer access checks with final ones")
Depends-on: d617f0d72d80 ("landlock: Optimize file path walks and prepare for audit support")
Link: https://lore.kernel.org/r/20250320190717.2287696-13-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Add audit support for sb_mount, move_mount, sb_umount, sb_remount, and
sb_pivot_root hooks.
The new related blocker is "fs.change_topology".
Audit event sample:
type=LANDLOCK_DENY msg=audit(1729738800.349:44): domain=195ba459b blockers=fs.change_topology name="/" dev="tmpfs" ino=1
Remove landlock_get_applicable_domain() and get_current_fs_domain()
which are now fully replaced with landlock_get_applicable_subject().
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-12-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Asynchronously log domain information when it first denies an access.
This minimize the amount of generated logs, which makes it possible to
always log denials for the current execution since they should not
happen. These records are identified with the new AUDIT_LANDLOCK_DOMAIN
type.
The AUDIT_LANDLOCK_DOMAIN message contains:
- the "domain" ID which is described;
- the "status" which can either be "allocated" or "deallocated";
- the "mode" which is for now only "enforcing";
- for the "allocated" status, a minimal set of properties to easily
identify the task that loaded the domain's policy with
landlock_restrict_self(2): "pid", "uid", executable path ("exe"), and
command line ("comm");
- for the "deallocated" state, the number of "denials" accounted to this
domain, which is at least 1.
This requires each domain to save these task properties at creation
time in the new struct landlock_details. A reference to the PID is kept
for the lifetime of the domain to avoid race conditions when
investigating the related task. The executable path is resolved and
stored to not keep a reference to the filesystem and block related
actions. All these metadata are stored for the lifetime of the related
domain and should then be minimal. The required memory is not accounted
to the task calling landlock_restrict_self(2) contrary to most other
Landlock allocations (see related comment).
The AUDIT_LANDLOCK_DOMAIN record follows the first AUDIT_LANDLOCK_ACCESS
record for the same domain, which is always followed by AUDIT_SYSCALL
and AUDIT_PROCTITLE. This is in line with the audit logic to first
record the cause of an event, and then add context with other types of
record.
Audit event sample for a first denial:
type=LANDLOCK_ACCESS msg=audit(1732186800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd"
type=LANDLOCK_DOMAIN msg=audit(1732186800.349:44): domain=195ba459b status=allocated mode=enforcing pid=300 uid=0 exe="/root/sandboxer" comm="sandboxer"
type=SYSCALL msg=audit(1732186800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0
Audit event sample for a following denial:
type=LANDLOCK_ACCESS msg=audit(1732186800.372:45): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd"
type=SYSCALL msg=audit(1732186800.372:45): arch=c000003e syscall=101 success=no [...] pid=300 auid=0
Log domain deletion with the "deallocated" state when a domain was
previously logged. This makes it possible for log parsers to free
potential resources when a domain ID will never show again.
The number of denied access requests is useful to easily check how many
access requests a domain blocked and potentially if some of them are
missing in logs because of audit rate limiting, audit rules, or Landlock
log configuration flags (see following commit).
Audit event sample for a deletion of a domain that denied something:
type=LANDLOCK_DOMAIN msg=audit(1732186800.393:46): domain=195ba459b status=deallocated denials=2
Cc: Günther Noack <gnoack@google.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-11-mic@digikod.net
[mic: Update comment and GFP flag for landlock_log_drop_domain()]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access
request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates
that something unexpected happened.
For now, only denied access are logged, which means that any
AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with
"success=no". However, log parsers should check this syscall property
because this is the only sign that a request was denied. Indeed, we
could have "success=yes" if Landlock would support a "permissive" mode.
We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode
(see following commit).
By default, the only logged access requests are those coming from the
same executed program that enforced the Landlock restriction on itself.
In other words, no audit record are created for a task after it called
execve(2). This is required to avoid log spam because programs may only
be aware of their own restrictions, but not the inherited ones.
Following commits will allow to conditionally generate
AUDIT_LANDLOCK_ACCESS records according to dedicated
landlock_restrict_self(2)'s flags.
The AUDIT_LANDLOCK_ACCESS message contains:
- the "domain" ID restricting the action on an object,
- the "blockers" that are missing to allow the requested access,
- a set of fields identifying the related object (e.g. task identified
with "opid" and "ocomm").
The blockers are implicit restrictions (e.g. ptrace), or explicit access
rights (e.g. filesystem), or explicit scopes (e.g. signal). This field
contains a list of at least one element, each separated with a comma.
The initial blocker is "ptrace", which describe all implicit Landlock
restrictions related to ptrace (e.g. deny tracing of tasks outside a
sandbox).
Add audit support to ptrace_access_check and ptrace_traceme hooks. For
the ptrace_access_check case, we log the current/parent domain and the
child task. For the ptrace_traceme case, we log the parent domain and
the current/child task. Indeed, the requester and the target are the
current task, but the action would be performed by the parent task.
Audit event sample:
type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd"
type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0
A following commit adds user documentation.
Add KUnit tests to check reading of domain ID relative to layer level.
The quick return for non-landlocked tasks is moved from task_ptrace() to
each LSM hooks.
It is not useful to inline the audit_enabled check because other
computation are performed by landlock_log_denial().
Use scoped guards for RCU read-side critical sections.
Cc: Günther Noack <gnoack@google.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Extend struct landlock_cred_security with a domain_exec bitmask to
identify which Landlock domain were created by the current task's bprm.
The whole bitmask is reset on each execve(2) call.
Cc: Günther Noack <gnoack@google.com>
Cc: Paul Moore <paul@paul-moore.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-9-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
This cosmetic change is needed for audit support, specifically to be
able to filter according to cross-execution boundaries.
struct landlock_file_security's size stay the same for now but it will
increase with struct landlock_cred_security's size.
Only save Landlock domain in hook_file_set_fowner() if the current
domain has LANDLOCK_SCOPE_SIGNAL, which was previously done for each
hook_file_send_sigiotask() calls. This should improve a bit
performance.
Replace hardcoded LANDLOCK_SCOPE_SIGNAL with the signal_scope.scope
variable.
Use scoped guards for RCU read-side critical sections.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-8-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
This cosmetic change that is needed for audit support, specifically to
be able to filter according to cross-execution boundaries.
Replace hardcoded LANDLOCK_SCOPE_SIGNAL with the signal_scope.scope
variable.
Use scoped guards for RCU read-side critical sections.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-7-mic@digikod.net
[mic: Update headers]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
This cosmetic change that is needed for audit support, specifically to
be able to filter according to cross-execution boundaries.
Optimize current_check_access_socket() to only handle the access
request.
Remove explicit domain->num_layers check which is now part of the
landlock_get_applicable_subject() call.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-6-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
This cosmetic change is needed for audit support, specifically to be
able to filter according to cross-execution boundaries.
Add landlock_get_applicable_subject(), mainly a copy of
landlock_get_applicable_domain(), which will fully replace it in a
following commit.
Optimize current_check_access_path() to only handle the access request.
Partially replace get_current_fs_domain() with explicit calls to
landlock_get_applicable_subject(). The remaining ones will follow with
more changes.
Remove explicit domain->num_layers check which is now part of the
landlock_get_applicable_subject() call.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-5-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Create a new domain.h file containing the struct landlock_hierarchy
definition and helpers. This type will grow with audit support. This
also prepares for a new domain type.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-4-mic@digikod.net
Reviewed-by: Günther Noack <gnoack3000@gmail.com>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Landlock IDs can be generated to uniquely identify Landlock objects.
For now, only Landlock domains get an ID at creation time. These IDs
map to immutable domain hierarchies.
Landlock IDs have important properties:
- They are unique during the lifetime of the running system thanks to
the 64-bit values: at worse, 2^60 - 2*2^32 useful IDs.
- They are always greater than 2^32 and must then be stored in 64-bit
integer types.
- The initial ID (at boot time) is randomly picked between 2^32 and
2^33, which limits collisions in logs across different boots.
- IDs are sequential, which enables users to order them.
- IDs may not be consecutive but increase with a random 2^4 step, which
limits side channels.
Such IDs can be exposed to unprivileged processes, even if it is not the
case with this audit patch series. The domain IDs will be useful for
user space to identify sandboxes and get their properties.
These Landlock IDs are more secure that other absolute kernel IDs such
as pipe's inodes which rely on a shared global counter.
For checkpoint/restore features (i.e. CRIU), we could easily implement a
privileged interface (e.g. sysfs) to set the next ID counter.
IDR/IDA are not used because we only need a bijection from Landlock
objects to Landlock IDs, and we must not recycle IDs. This enables us
to identify all Landlock objects during the lifetime of the system (e.g.
in logs), but not to access an object from an ID nor know if an ID is
assigned. Using a counter is simpler, it scales (i.e. avoids growing
memory footprint), and it does not require locking. We'll use proper
file descriptors (with IDs used as inode numbers) to access Landlock
objects.
Cc: Günther Noack <gnoack@google.com>
Cc: Paul Moore <paul@paul-moore.com>
Link: https://lore.kernel.org/r/20250320190717.2287696-3-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Because Linux credentials are managed per thread, user space relies on
some hack to synchronize credential update across threads from the same
process. This is required by the Native POSIX Threads Library and
implemented by set*id(2) wrappers and libcap(3) to use tgkill(2) to
synchronize threads. See nptl(7) and libpsx(3). Furthermore, some
runtimes like Go do not enable developers to have control over threads
[1].
To avoid potential issues, and because threads are not security
boundaries, let's relax the Landlock (optional) signal scoping to always
allow signals sent between threads of the same process. This exception
is similar to the __ptrace_may_access() one.
hook_file_set_fowner() now checks if the target task is part of the same
process as the caller. If this is the case, then the related signal
triggered by the socket will always be allowed.
Scoping of abstract UNIX sockets is not changed because kernel objects
(e.g. sockets) should be tied to their creator's domain at creation
time.
Note that creating one Landlock domain per thread puts each of these
threads (and their future children) in their own scope, which is
probably not what users expect, especially in Go where we do not control
threads. However, being able to drop permissions on all threads should
not be restricted by signal scoping. We are working on a way to make it
possible to atomically restrict all threads of a process with the same
domain [2].
Add erratum for signal scoping.
Closes: https://github.com/landlock-lsm/go-landlock/issues/36
Fixes: 54a6e6bbf3be ("landlock: Add signal scoping")
Fixes: c8994965013e ("selftests/landlock: Test signal scoping for threads")
Depends-on: 26f204380a3c ("fs: Fix file_set_fowner LSM hook inconsistencies")
Link: https://pkg.go.dev/kernel.org/pub/linux/libs/security/libcap/psx [1]
Link: https://github.com/landlock-lsm/linux/issues/2 [2]
Cc: Günther Noack <gnoack@google.com>
Cc: Paul Moore <paul@paul-moore.com>
Cc: Serge Hallyn <serge@hallyn.com>
Cc: Tahera Fahimi <fahimitahera@gmail.com>
Cc: stable@vger.kernel.org
Acked-by: Christian Brauner <brauner@kernel.org>
Link: https://lore.kernel.org/r/20250318161443.279194-6-mic@digikod.net
[mic: Add extra pointer check and RCU guard, and ease backport]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
-----BEGIN PGP SIGNATURE-----
iHUEABYKAB0WIQRAhzRXHqcMeLMyaSiRxhvAZXjcogUCZ90p4AAKCRCRxhvAZXjc
ojMIAP9atkG3u7+490+NGWLdulQlaHnD51Owa9MiW87UfKpsTQEArwi/NrJqXJNT
PFQ2xIa5TxG+9haChR89w3kjZ6b/hgs=
=iDkx
-----END PGP SIGNATURE-----
Merge tag 'vfs-6.15-rc1.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Pull misc vfs updates from Christian Brauner:
"Features:
- Add CONFIG_DEBUG_VFS infrastucture:
- Catch invalid modes in open
- Use the new debug macros in inode_set_cached_link()
- Use debug-only asserts around fd allocation and install
- Place f_ref to 3rd cache line in struct file to resolve false
sharing
Cleanups:
- Start using anon_inode_getfile_fmode() helper in various places
- Don't take f_lock during SEEK_CUR if exclusion is guaranteed by
f_pos_lock
- Add unlikely() to kcmp()
- Remove legacy ->remount_fs method from ecryptfs after port to the
new mount api
- Remove invalidate_inodes() in favour of evict_inodes()
- Simplify ep_busy_loopER by removing unused argument
- Avoid mmap sem relocks when coredumping with many missing pages
- Inline getname()
- Inline new_inode_pseudo() and de-staticize alloc_inode()
- Dodge an atomic in putname if ref == 1
- Consistently deref the files table with rcu_dereference_raw()
- Dedup handling of struct filename init and refcounts bumps
- Use wq_has_sleeper() in end_dir_add()
- Drop the lock trip around I_NEW wake up in evict()
- Load the ->i_sb pointer once in inode_sb_list_{add,del}
- Predict not reaching the limit in alloc_empty_file()
- Tidy up do_sys_openat2() with likely/unlikely
- Call inode_sb_list_add() outside of inode hash lock
- Sort out fd allocation vs dup2 race commentary
- Turn page_offset() into a wrapper around folio_pos()
- Remove locking in exportfs around ->get_parent() call
- try_lookup_one_len() does not need any locks in autofs
- Fix return type of several functions from long to int in open
- Fix return type of several functions from long to int in ioctls
Fixes:
- Fix watch queue accounting mismatch"
* tag 'vfs-6.15-rc1.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: (30 commits)
fs: sort out fd allocation vs dup2 race commentary, take 2
fs: call inode_sb_list_add() outside of inode hash lock
fs: tidy up do_sys_openat2() with likely/unlikely
fs: predict not reaching the limit in alloc_empty_file()
fs: load the ->i_sb pointer once in inode_sb_list_{add,del}
fs: drop the lock trip around I_NEW wake up in evict()
fs: use wq_has_sleeper() in end_dir_add()
VFS/autofs: try_lookup_one_len() does not need any locks
fs: dedup handling of struct filename init and refcounts bumps
fs: consistently deref the files table with rcu_dereference_raw()
exportfs: remove locking around ->get_parent() call.
fs: use debug-only asserts around fd allocation and install
fs: dodge an atomic in putname if ref == 1
vfs: Remove invalidate_inodes()
ecryptfs: remove NULL remount_fs from super_operations
watch_queue: fix pipe accounting mismatch
fs: place f_ref to 3rd cache line in struct file to resolve false sharing
epoll: simplify ep_busy_loop by removing always 0 argument
fs: Turn page_offset() into a wrapper around folio_pos()
kcmp: improve performance adding an unlikely hint to task comparisons
...
Potentially include errata for Landlock ABI v5 (Linux 6.10) and v6
(Linux 6.12). That will be useful for the following signal scoping
erratum.
As explained in errata.h, this commit should be backportable without
conflict down to ABI v5. It must then not include the errata/abi-6.h
file.
Fixes: 54a6e6bbf3be ("landlock: Add signal scoping")
Cc: Günther Noack <gnoack@google.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20250318161443.279194-5-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Some fixes may require user space to check if they are applied on the
running kernel before using a specific feature. For instance, this
applies when a restriction was previously too restrictive and is now
getting relaxed (e.g. for compatibility reasons). However, non-visible
changes for legitimate use (e.g. security fixes) do not require an
erratum.
Because fixes are backported down to a specific Landlock ABI, we need a
way to avoid cherry-pick conflicts. The solution is to only update a
file related to the lower ABI impacted by this issue. All the ABI files
are then used to create a bitmask of fixes.
The new errata interface is similar to the one used to get the supported
Landlock ABI version, but it returns a bitmask instead because the order
of fixes may not match the order of versions, and not all fixes may
apply to all versions.
The actual errata will come with dedicated commits. The description is
not actually used in the code but serves as documentation.
Create the landlock_abi_version symbol and use its value to check errata
consistency.
Update test_base's create_ruleset_checks_ordering tests and add errata
tests.
This commit is backportable down to the first version of Landlock.
Fixes: 3532b0b4352c ("landlock: Enable user space to infer supported features")
Cc: Günther Noack <gnoack@google.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20250318161443.279194-3-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
To ease backports in setup.c, let's group changes from
__lsm_ro_after_init to __ro_after_init with commit f22f9aaf6c3d
("selinux: remove the runtime disable functionality"), and the
landlock_lsmid addition with commit f3b8788cde61 ("LSM: Identify modules
by more than name").
That will help to backport the following errata.
Cc: Günther Noack <gnoack@google.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20250318161443.279194-2-mic@digikod.net
Fixes: f3b8788cde61 ("LSM: Identify modules by more than name")
Signed-off-by: Mickaël Salaün <mic@digikod.net>
The function can be replaced by evict_inodes. The only difference is
that evict_inodes() skips the inodes with positive refcount without
touching ->i_lock, but they are equivalent as evict_inodes() repeats the
refcount check after having grabbed ->i_lock.
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20250307144318.28120-2-jack@suse.cz
Signed-off-by: Christian Brauner <brauner@kernel.org>
Use sk_is_tcp() to check if socket is TCP in bind(2) and connect(2)
hooks.
SMC, MPTCP, SCTP protocols are currently restricted by TCP access
rights. The purpose of TCP access rights is to provide control over
ports that can be used by userland to establish a TCP connection.
Therefore, it is incorrect to deny bind(2) and connect(2) requests for a
socket of another protocol.
However, SMC, MPTCP and RDS implementations use TCP internal sockets to
establish communication or even to exchange packets over a TCP
connection [1]. Landlock rules that configure bind(2) and connect(2)
usage for TCP sockets should not cover requests for sockets of such
protocols. These protocols have different set of security issues and
security properties, therefore, it is necessary to provide the userland
with the ability to distinguish between them (eg. [2]).
Control over TCP connection used by other protocols can be achieved with
upcoming support of socket creation control [3].
[1] https://lore.kernel.org/all/62336067-18c2-3493-d0ec-6dd6a6d3a1b5@huawei-partners.com/
[2] https://lore.kernel.org/all/20241204.fahVio7eicim@digikod.net/
[3] https://lore.kernel.org/all/20240904104824.1844082-1-ivanov.mikhail1@huawei-partners.com/
Closes: https://github.com/landlock-lsm/linux/issues/40
Fixes: fff69fb03dde ("landlock: Support network rules with TCP bind and connect")
Signed-off-by: Mikhail Ivanov <ivanov.mikhail1@huawei-partners.com>
Link: https://lore.kernel.org/r/20250205093651.1424339-2-ivanov.mikhail1@huawei-partners.com
[mic: Format commit message to 72 columns]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Always synchronize access_masked_parent* with access_request_parent*
according to allowed_parent*. This is required for audit support to be
able to get back to the reason of denial.
In a rename/link action, instead of always checking a rule two times for
the same parent directory of the source and the destination files, only
check it when an action on a child was not already allowed. This also
enables us to keep consistent allowed_parent* status, which is required
to get back to the reason of denial.
For internal mount points, only upgrade allowed_parent* to true but do
not wrongfully set both of them to false otherwise. This is also
required to get back to the reason of denial.
This does not impact the current behavior but slightly optimize code and
prepare for audit support that needs to know the exact reason why an
access was denied.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250108154338.1129069-14-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Fix a logical issue that could have been visible if the source or the
destination of a rename/link action was allowed for either the source or
the destination but not both. However, this logical bug is unreachable
because either:
- the rename/link action is allowed by the access rights tied to the
same mount point (without relying on access rights in a parent mount
point) and the access request is allowed (i.e. allow_parent1 and
allow_parent2 are true in current_check_refer_path),
- or a common rule in a parent mount point updates the access check for
the source and the destination (cf. is_access_to_paths_allowed).
See the following layout1.refer_part_mount_tree_is_allowed test that
work with and without this fix.
This fix does not impact current code but it is required for the audit
support.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250108154338.1129069-12-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Upgrade domain's handled access masks when creating a domain from a
ruleset, instead of converting them at runtime. This is more consistent
and helps with audit support.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250108154338.1129069-7-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Move LANDLOCK_ACCESS_FS_INITIALLY_DENIED, access_mask_t, struct
access_mask, and struct access_masks_all to a dedicated access.h file.
Rename LANDLOCK_ACCESS_FS_INITIALLY_DENIED to
_LANDLOCK_ACCESS_FS_INITIALLY_DENIED to make it clear that it's not part
of UAPI. Add some newlines when appropriate.
This file will be extended with following commits, and it will help to
avoid dependency loops.
Cc: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250108154338.1129069-6-mic@digikod.net
[mic: Fix rebase conflict because of the new cleanup headers]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Simplify error handling by replacing goto statements with automatic
calls to landlock_put_ruleset() when going out of scope.
This change depends on the TCP support.
Cc: Konstantin Meskhidze <konstantin.meskhidze@huawei.com>
Cc: Mikhail Ivanov <ivanov.mikhail1@huawei-partners.com>
Reviewed-by: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250113161112.452505-3-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Simplify error handling by replacing goto statements with automatic
calls to landlock_put_ruleset() when going out of scope.
This change will be easy to backport to v6.6 if needed, only the
kernel.h include line conflicts. As for any other similar changes, we
should be careful when backporting without goto statements.
Add missing include file.
Reviewed-by: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20250113161112.452505-2-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Making sure that struct fd instances are destroyed in the same
scope where they'd been created, getting rid of reassignments
and passing them by reference, converting to CLASS(fd{,_pos,_raw}).
We are getting very close to having the memory safety of that stuff
trivial to verify.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQQqUNBr3gm4hGXdBJlZ7Krx/gZQ6wUCZzdikAAKCRBZ7Krx/gZQ
69nJAQCmbQHK3TGUbQhOw6MJXOK9ezpyEDN3FZb4jsu38vTIdgEA6OxAYDO2m2g9
CN18glYmD3wRyU6Bwl4vGODouSJvDgA=
=gVH3
-----END PGP SIGNATURE-----
Merge tag 'pull-fd' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull 'struct fd' class updates from Al Viro:
"The bulk of struct fd memory safety stuff
Making sure that struct fd instances are destroyed in the same scope
where they'd been created, getting rid of reassignments and passing
them by reference, converting to CLASS(fd{,_pos,_raw}).
We are getting very close to having the memory safety of that stuff
trivial to verify"
* tag 'pull-fd' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (28 commits)
deal with the last remaing boolean uses of fd_file()
css_set_fork(): switch to CLASS(fd_raw, ...)
memcg_write_event_control(): switch to CLASS(fd)
assorted variants of irqfd setup: convert to CLASS(fd)
do_pollfd(): convert to CLASS(fd)
convert do_select()
convert vfs_dedupe_file_range().
convert cifs_ioctl_copychunk()
convert media_request_get_by_fd()
convert spu_run(2)
switch spufs_calls_{get,put}() to CLASS() use
convert cachestat(2)
convert do_preadv()/do_pwritev()
fdget(), more trivial conversions
fdget(), trivial conversions
privcmd_ioeventfd_assign(): don't open-code eventfd_ctx_fdget()
o2hb_region_dev_store(): avoid goto around fdget()/fdput()
introduce "fd_pos" class, convert fdget_pos() users to it.
fdget_raw() users: switch to CLASS(fd_raw)
convert vmsplice() to CLASS(fd)
...
Do not walk through the domain hierarchy when the required scope is not
supported by this domain. This is the same approach as for filesystem
and network restrictions.
Cc: Mikhail Ivanov <ivanov.mikhail1@huawei-partners.com>
Cc: Tahera Fahimi <fahimitahera@gmail.com>
Reviewed-by: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20241109110856.222842-4-mic@digikod.net
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Replace get_raw_handled_fs_accesses() with a generic
landlock_union_access_masks(), and replace get_fs_domain() with a
generic landlock_get_applicable_domain(). These helpers will also be
useful for other types of access.
Cc: Mikhail Ivanov <ivanov.mikhail1@huawei-partners.com>
Reviewed-by: Günther Noack <gnoack@google.com>
Link: https://lore.kernel.org/r/20241109110856.222842-2-mic@digikod.net
[mic: Slightly improve doc as suggested by Günther]
Signed-off-by: Mickaël Salaün <mic@digikod.net>
fdget() is the first thing done in scope, all matching fdput() are
immediately followed by leaving the scope.
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
-----BEGIN PGP SIGNATURE-----
iIYEABYKAC4WIQSVyBthFV4iTW/VU1/l49DojIL20gUCZvGpchAcbWljQGRpZ2lr
b2QubmV0AAoJEOXj0OiMgvbSTzMBAIpcYKf75IyC4DXqiXlko508YdyI2YfYeWdd
5yVZbSHgAP0aEFO4AOvJ26pPlGF+8zVIHq+HNAhrAalZBulxASePCA==
=nsAF
-----END PGP SIGNATURE-----
Merge tag 'landlock-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux
Pull landlock updates from Mickaël Salaün:
"We can now scope a Landlock domain thanks to a new "scoped" field that
can deny interactions with resources outside of this domain.
The LANDLOCK_SCOPE_ABSTRACT_UNIX_SOCKET flag denies connections to an
abstract UNIX socket created outside of the current scoped domain, and
the LANDLOCK_SCOPE_SIGNAL flag denies sending a signal to processes
outside of the current scoped domain.
These restrictions also apply to nested domains according to their
scope. The related changes will also be useful to support other kind
of IPC isolations"
* tag 'landlock-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux:
landlock: Document LANDLOCK_SCOPE_SIGNAL
samples/landlock: Add support for signal scoping
selftests/landlock: Test signal created by out-of-bound message
selftests/landlock: Test signal scoping for threads
selftests/landlock: Test signal scoping
landlock: Add signal scoping
landlock: Document LANDLOCK_SCOPE_ABSTRACT_UNIX_SOCKET
samples/landlock: Add support for abstract UNIX socket scoping
selftests/landlock: Test inherited restriction of abstract UNIX socket
selftests/landlock: Test connected and unconnected datagram UNIX socket
selftests/landlock: Test UNIX sockets with any address formats
selftests/landlock: Test abstract UNIX socket scoping
selftests/landlock: Test handling of unknown scope
landlock: Add abstract UNIX socket scoping
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQQqUNBr3gm4hGXdBJlZ7Krx/gZQ6wUCZvDNmgAKCRBZ7Krx/gZQ
63zrAP9vI0rf55v27twiabe9LnI7aSx5ckoqXxFIFxyT3dOYpQD/bPmoApnWDD3d
592+iDgLsema/H/0/CqfqlaNtDNY8Q0=
=HUl5
-----END PGP SIGNATURE-----
Merge tag 'pull-stable-struct_fd' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull 'struct fd' updates from Al Viro:
"Just the 'struct fd' layout change, with conversion to accessor
helpers"
* tag 'pull-stable-struct_fd' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
add struct fd constructors, get rid of __to_fd()
struct fd: representation change
introduce fd_file(), convert all accessors to it.
Currently, a sandbox process is not restricted to sending a signal (e.g.
SIGKILL) to a process outside the sandbox environment. The ability to
send a signal for a sandboxed process should be scoped the same way
abstract UNIX sockets are scoped. Therefore, we extend the "scoped"
field in a ruleset with LANDLOCK_SCOPE_SIGNAL to specify that a ruleset
will deny sending any signal from within a sandbox process to its parent
(i.e. any parent sandbox or non-sandboxed processes).
This patch adds file_set_fowner and file_free_security hooks to set and
release a pointer to the file owner's domain. This pointer, fown_domain
in landlock_file_security will be used in file_send_sigiotask to check
if the process can send a signal.
The ruleset_with_unknown_scope test is updated to support
LANDLOCK_SCOPE_SIGNAL.
This depends on two new changes:
- commit 1934b212615d ("file: reclaim 24 bytes from f_owner"): replace
container_of(fown, struct file, f_owner) with fown->file .
- commit 26f204380a3c ("fs: Fix file_set_fowner LSM hook
inconsistencies"): lock before calling the hook.
Signed-off-by: Tahera Fahimi <fahimitahera@gmail.com>
Closes: https://github.com/landlock-lsm/linux/issues/8
Link: https://lore.kernel.org/r/df2b4f880a2ed3042992689a793ea0951f6798a5.1725657727.git.fahimitahera@gmail.com
[mic: Update landlock_get_current_domain()'s return type, improve and
fix locking in hook_file_set_fowner(), simplify and fix sleepable call
and locking issue in hook_file_send_sigiotask() and rebase on the latest
VFS tree, simplify hook_task_kill() and quickly return when not
sandboxed, improve comments, rename LANDLOCK_SCOPED_SIGNAL]
Co-developed-by: Mickaël Salaün <mic@digikod.net>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
Introduce a new "scoped" member to landlock_ruleset_attr that can
specify LANDLOCK_SCOPE_ABSTRACT_UNIX_SOCKET to restrict connection to
abstract UNIX sockets from a process outside of the socket's domain.
Two hooks are implemented to enforce these restrictions:
unix_stream_connect and unix_may_send.
Closes: https://github.com/landlock-lsm/linux/issues/7
Signed-off-by: Tahera Fahimi <fahimitahera@gmail.com>
Link: https://lore.kernel.org/r/5f7ad85243b78427242275b93481cfc7c127764b.1725494372.git.fahimitahera@gmail.com
[mic: Fix commit message formatting, improve documentation, simplify
hook_unix_may_send(), and cosmetic fixes including rename of
LANDLOCK_SCOPED_ABSTRACT_UNIX_SOCKET]
Co-developed-by: Mickaël Salaün <mic@digikod.net>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
For any changes of struct fd representation we need to
turn existing accesses to fields into calls of wrappers.
Accesses to struct fd::flags are very few (3 in linux/file.h,
1 in net/socket.c, 3 in fs/overlayfs/file.c and 3 more in
explicit initializers).
Those can be dealt with in the commit converting to
new layout; accesses to struct fd::file are too many for that.
This commit converts (almost) all of f.file to
fd_file(f). It's not entirely mechanical ('file' is used as
a member name more than just in struct fd) and it does not
even attempt to distinguish the uses in pointer context from
those in boolean context; the latter will be eventually turned
into a separate helper (fd_empty()).
NOTE: mass conversion to fd_empty(), tempting as it
might be, is a bad idea; better do that piecewise in commit
that convert from fdget...() to CLASS(...).
[conflicts in fs/fhandle.c, kernel/bpf/syscall.c, mm/memcontrol.c
caught by git; fs/stat.c one got caught by git grep]
[fs/xattr.c conflict]
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The LSM framework has an existing inode_free_security() hook which
is used by LSMs that manage state associated with an inode, but
due to the use of RCU to protect the inode, special care must be
taken to ensure that the LSMs do not fully release the inode state
until it is safe from a RCU perspective.
This patch implements a new inode_free_security_rcu() implementation
hook which is called when it is safe to free the LSM's internal inode
state. Unfortunately, this new hook does not have access to the inode
itself as it may already be released, so the existing
inode_free_security() hook is retained for those LSMs which require
access to the inode.
Cc: stable@vger.kernel.org
Reported-by: syzbot+5446fbf332b0602ede0b@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/r/00000000000076ba3b0617f65cc8@google.com
Signed-off-by: Paul Moore <paul@paul-moore.com>
-----BEGIN PGP SIGNATURE-----
iIYEABYKAC4WIQSVyBthFV4iTW/VU1/l49DojIL20gUCZqFEchAcbWljQGRpZ2lr
b2QubmV0AAoJEOXj0OiMgvbSULcBAPEV5Viu/zox2FdS87EGTqWxEQJcBRvc3ahj
MQk44WtMAP4o2CnwrOoMyZXeq9npteL5lQsVhEzeI+p8oN9C9bThBg==
=zizo
-----END PGP SIGNATURE-----
Merge tag 'landlock-6.11-rc1-houdini-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux
Pull landlock fix from Mickaël Salaün:
"Jann Horn reported a sandbox bypass for Landlock. This includes the
fix and new tests. This should be backported"
* tag 'landlock-6.11-rc1-houdini-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux:
selftests/landlock: Add cred_transfer test
landlock: Don't lose track of restrictions on cred_transfer
When a process' cred struct is replaced, this _almost_ always invokes
the cred_prepare LSM hook; but in one special case (when
KEYCTL_SESSION_TO_PARENT updates the parent's credentials), the
cred_transfer LSM hook is used instead. Landlock only implements the
cred_prepare hook, not cred_transfer, so KEYCTL_SESSION_TO_PARENT causes
all information on Landlock restrictions to be lost.
This basically means that a process with the ability to use the fork()
and keyctl() syscalls can get rid of all Landlock restrictions on
itself.
Fix it by adding a cred_transfer hook that does the same thing as the
existing cred_prepare hook. (Implemented by having hook_cred_prepare()
call hook_cred_transfer() so that the two functions are less likely to
accidentally diverge in the future.)
Cc: stable@kernel.org
Fixes: 385975dca53e ("landlock: Set up the security framework and manage credentials")
Signed-off-by: Jann Horn <jannh@google.com>
Link: https://lore.kernel.org/r/20240724-landlock-houdini-fix-v1-1-df89a4560ca3@google.com
Signed-off-by: Mickaël Salaün <mic@digikod.net>
