Skip to contentSkip to conversation

landlock: Fix LOG_SUBDOMAINS_OFF inheritance across fork()

PATCH In review Mickaël Salaün 2 participants 3 messages Apr 7

Subsystems (4)
lore
Share & tools
lore

Mickaël Salaün

author maintainer
#0 lore raw

PATCH v2 1/2 landlock: Fix LOG_SUBDOMAINS_OFF inheritance across fork()

hook_cred_transfer() only copies the Landlock security blob when the source credential has a domain. This is inconsistent with landlock_restrict_self() which can set LOG_SUBDOMAINS_OFF on a credential without creating a domain (via the ruleset_fd=-1 path): the field is committed but not preserved across fork() because the child's prepare_creds() calls hook_cred_transfer() which skips the copy when domain is NULL.

This breaks the documented use case where a process mutes subdomain logs before forking sandboxed children: the children lose the muting and their domains produce unexpected audit records.

Fix this by unconditionally copying the Landlock credential blob.

Cc Günther Noack <[email protected]>
Cc Jann Horn <[email protected]>
Fixes ead9079f7569 ("landlock: Add LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF")

---

Changes since v1:
https://lore.kernel.org/r/[email protected]
- Improve subject.
- Add Reviewed-by Günther.

---
 security/landlock/cred.c                      |  6 +-
 tools/testing/selftests/landlock/audit_test.c | 88 +++++++++++++++++++
 2 files changed, 90 insertions(+), 4 deletions(-)
security/landlock/cred.c +2 -4
index 0cb3edde4d18..cc419de75cd6 100644
@@ -22,10 +22,8 @@ static void hook_cred_transfer(struct cred *const new,
 	const struct landlock_cred_security *const old_llcred =
 		landlock_cred(old);
 
-	if (old_llcred->domain) {
-		landlock_get_ruleset(old_llcred->domain);
-		*landlock_cred(new) = *old_llcred;
-	}
+	landlock_get_ruleset(old_llcred->domain);
+	*landlock_cred(new) = *old_llcred;
 }
 
 static int hook_cred_prepare(struct cred *const new,
tools/testing/selftests/landlock/audit_test.c +88 -0

94 lines not on this page · raw message

Signature · 1 line
2.53.0
author maintainer
#1 lore raw

PATCH v2 2/2 landlock: Allow TSYNC with LOG_SUBDOMAINS_OFF and fd=-1

LANDLOCK_RESTRICT_SELF_TSYNC does not allow LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF with ruleset_fd=-1, preventing a multithreaded process from atomically propagating subdomain log muting to all threads without creating a domain layer. Relax the fd=-1 condition to accept TSYNC alongside LOG_SUBDOMAINS_OFF, and update the documentation accordingly.

Add flag validation tests for all TSYNC combinations with ruleset_fd=-1, and audit tests verifying both transition directions: muting via TSYNC (logged to not logged) and override via TSYNC (not logged to logged).

Cc Günther Noack <[email protected]>
Fixes 42fc7e6543f6 ("landlock: Multithreading support for landlock_restrict_self()")

---

Changes since v1:
https://lore.kernel.org/r/[email protected]
- Zero-initialize struct thread_data to avoid flaky uninitialized mute_subdomains field (pointed out by Günther Noack).
- Drop capabilities and set no_new_privs in tsync_without_ruleset fixture setup instead of relying on ambient CAP_SYS_ADMIN (pointed out by Günther Noack).
- Add Reviewed-by Günther.

---
 include/uapi/linux/landlock.h                 |   4 +-
 security/landlock/syscalls.c                  |  14 +-
 tools/testing/selftests/landlock/audit_test.c | 233 ++++++++++++++++++
 tools/testing/selftests/landlock/tsync_test.c |  77 ++++++
 4 files changed, 322 insertions(+), 6 deletions(-)
include/uapi/linux/landlock.h +3 -1
index f88fa1f68b77..d37603efc273 100644
@@ -116,7 +116,9 @@ struct landlock_ruleset_attr {
  *     ``LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF``, this flag only affects
  *     future nested domains, not the one being created. It can also be used
  *     with a @ruleset_fd value of -1 to mute subdomain logs without creating a
- *     domain.
+ *     domain.  When combined with %LANDLOCK_RESTRICT_SELF_TSYNC and a
+ *     @ruleset_fd value of -1, this configuration is propagated to all threads
+ *     of the current process.
  *
  * The following flag supports policy enforcement in multithreaded processes:
  *
security/landlock/syscalls.c +9 -5
index 0d66a68677b7..a0bb664e0d31 100644
@@ -512,10 +512,13 @@ SYSCALL_DEFINE2(landlock_restrict_self, const int, ruleset_fd, const __u32,
 
 	/*
 	 * It is allowed to set LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF with
-	 * -1 as ruleset_fd, but no other flag must be set.
+	 * -1 as ruleset_fd, optionally combined with
+	 * LANDLOCK_RESTRICT_SELF_TSYNC to propagate this configuration to all
+	 * threads.  No other flag must be set.
 	 */
 	if (!(ruleset_fd == -1 &&
-	      flags == LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF)) {
+	      (flags & ~LANDLOCK_RESTRICT_SELF_TSYNC) ==
+		      LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF)) {
 		/* Gets and checks the ruleset. */
 		ruleset = get_ruleset_from_fd(ruleset_fd, FMODE_CAN_READ);
 		if (IS_ERR(ruleset))
@@ -537,9 +540,10 @@ SYSCALL_DEFINE2(landlock_restrict_self, const int, ruleset_fd, const __u32,
 
 	/*
 	 * The only case when a ruleset may not be set is if
-	 * LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF is set and ruleset_fd is -1.
-	 * We could optimize this case by not calling commit_creds() if this flag
-	 * was already set, but it is not worth the complexity.
+	 * LANDLOCK_RESTRICT_SELF_LOG_SUBDOMAINS_OFF is set (optionally with
+	 * LANDLOCK_RESTRICT_SELF_TSYNC) and ruleset_fd is -1.  We could
+	 * optimize this case by not calling commit_creds() if this flag was
+	 * already set, but it is not worth the complexity.
 	 */
 	if (ruleset) {
 		/*
tools/testing/selftests/landlock/audit_test.c +233 -0

245 lines not on this page · raw message

tools/testing/selftests/landlock/tsync_test.c +77 -0

81 lines not on this page · raw message

Signature · 1 line
2.53.0
Reviewed-by on v2 1/2
#2 lore raw
In reply to Mickaël Salaün
Quoted 8 lines · Mickaël Salaün

hook_cred_transfer() only copies the Landlock security blob when the source credential has a domain. This is inconsistent with landlock_restrict_self() which can set LOG_SUBDOMAINS_OFF on a credential without creating a domain (via the ruleset_fd=-1 path): the field is committed but not preserved across fork() because the child's prepare_creds() calls hook_cred_transfer() which skips the copy when domain is NULL.

Quoted 5 lines

This breaks the documented use case where a process mutes subdomain logs before forking sandboxed children: the children lose the muting and their domains produce unexpected audit records.

Fix this by unconditionally copying the Landlock credential blob.

As before, LGTM for both patches. Thanks for the fixes!

–Günther