Merge tag 'landlock-6.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux

Pull landlock updates from Mickaël Salaün:
 "This improves documentation, and makes some tests more flexible to be
  able to run on systems without overlayfs or with Yama restrictions"

* tag 'landlock-6.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mic/linux:
  MAINTAINERS: Update Landlock repository
  selftests/landlock: Test ptrace as much as possible with Yama
  selftests/landlock: Skip overlayfs tests when not supported
  landlock: Explain file descriptor access rights

Documentation/security/landlock.rst

@@ -7,7 +7,7 @@ Landlock LSM: kernel documentation
 ==================================
 
 :Author: Mickaël Salaün
-:Date: September 2022
+:Date: December 2022
 
 Landlock's goal is to create scoped access-control (i.e. sandboxing).  To
 harden a whole system, this feature should be available to any process,
@@ -41,12 +41,16 @@ Guiding principles for safe access controls
   processes.
 * Computation related to Landlock operations (e.g. enforcing a ruleset) shall
   only impact the processes requesting them.
+* Resources (e.g. file descriptors) directly obtained from the kernel by a
+  sandboxed process shall retain their scoped accesses (at the time of resource
+  acquisition) whatever process use them.
+  Cf. `File descriptor access rights`_.
 
 Design choices
 ==============
 
-Filesystem access rights
-------------------------
+Inode access rights
+-------------------
 
 All access rights are tied to an inode and what can be accessed through it.
 Reading the content of a directory does not imply to be allowed to read the
@@ -57,6 +61,30 @@ directory, not the unlinked inode.  This is the reason why
 ``LANDLOCK_ACCESS_FS_REMOVE_FILE`` or ``LANDLOCK_ACCESS_FS_REFER`` are not
 allowed to be tied to files but only to directories.
 
+File descriptor access rights
+-----------------------------
+
+Access rights are checked and tied to file descriptors at open time.  The
+underlying principle is that equivalent sequences of operations should lead to
+the same results, when they are executed under the same Landlock domain.
+
+Taking the ``LANDLOCK_ACCESS_FS_TRUNCATE`` right as an example, it may be
+allowed to open a file for writing without being allowed to
+:manpage:`ftruncate` the resulting file descriptor if the related file
+hierarchy doesn't grant such access right.  The following sequences of
+operations have the same semantic and should then have the same result:
+
+* ``truncate(path);``
+* ``int fd = open(path, O_WRONLY); ftruncate(fd); close(fd);``
+
+Similarly to file access modes (e.g. ``O_RDWR``), Landlock access rights
+attached to file descriptors are retained even if they are passed between
+processes (e.g. through a Unix domain socket).  Such access rights will then be
+enforced even if the receiving process is not sandboxed by Landlock.  Indeed,
+this is required to keep a consistent access control over the whole system, and
+this avoids unattended bypasses through file descriptor passing (i.e. confused
+deputy attack).
+
 Tests
 =====
 

MAINTAINERS

@@ -11592,7 +11592,7 @@ M:	Mickaël Salaün <mic@digikod.net>
 L:	linux-security-module@vger.kernel.org
 S:	Supported
 W:	https://landlock.io
-T:	git https://github.com/landlock-lsm/linux.git
+T:	git https://git.kernel.org/pub/scm/linux/kernel/git/mic/linux.git
 F:	Documentation/security/landlock.rst
 F:	Documentation/userspace-api/landlock.rst
 F:	include/uapi/linux/landlock.h

tools/testing/selftests/landlock/fs_test.c

@@ -11,6 +11,7 @@
 #include <fcntl.h>
 #include <linux/landlock.h>
 #include <sched.h>
+#include <stdio.h>
 #include <string.h>
 #include <sys/capability.h>
 #include <sys/mount.h>
@@ -89,6 +90,40 @@ static const char dir_s3d3[] = TMP_DIR "/s3d1/s3d2/s3d3";
  *         └── s3d3
  */
 
+static bool fgrep(FILE *const inf, const char *const str)
+{
+	char line[32];
+	const int slen = strlen(str);
+
+	while (!feof(inf)) {
+		if (!fgets(line, sizeof(line), inf))
+			break;
+		if (strncmp(line, str, slen))
+			continue;
+
+		return true;
+	}
+
+	return false;
+}
+
+static bool supports_overlayfs(void)
+{
+	bool res;
+	FILE *const inf = fopen("/proc/filesystems", "r");
+
+	/*
+	 * Consider that the filesystem is supported if we cannot get the
+	 * supported ones.
+	 */
+	if (!inf)
+		return true;
+
+	res = fgrep(inf, "nodev\toverlay\n");
+	fclose(inf);
+	return res;
+}
+
 static void mkdir_parents(struct __test_metadata *const _metadata,
 			  const char *const path)
 {
@@ -4001,6 +4036,9 @@ FIXTURE(layout2_overlay) {};
 
 FIXTURE_SETUP(layout2_overlay)
 {
+	if (!supports_overlayfs())
+		SKIP(return, "overlayfs is not supported");
+
 	prepare_layout(_metadata);
 
 	create_directory(_metadata, LOWER_BASE);
@@ -4037,6 +4075,9 @@ FIXTURE_SETUP(layout2_overlay)
 
 FIXTURE_TEARDOWN(layout2_overlay)
 {
+	if (!supports_overlayfs())
+		SKIP(return, "overlayfs is not supported");
+
 	EXPECT_EQ(0, remove_path(lower_do1_fl3));
 	EXPECT_EQ(0, remove_path(lower_dl1_fl2));
 	EXPECT_EQ(0, remove_path(lower_fl1));
@@ -4068,6 +4109,9 @@ FIXTURE_TEARDOWN(layout2_overlay)
 
 TEST_F_FORK(layout2_overlay, no_restriction)
 {
+	if (!supports_overlayfs())
+		SKIP(return, "overlayfs is not supported");
+
 	ASSERT_EQ(0, test_open(lower_fl1, O_RDONLY));
 	ASSERT_EQ(0, test_open(lower_dl1, O_RDONLY));
 	ASSERT_EQ(0, test_open(lower_dl1_fl2, O_RDONLY));
@@ -4231,6 +4275,9 @@ TEST_F_FORK(layout2_overlay, same_content_different_file)
 	size_t i;
 	const char *path_entry;
 
+	if (!supports_overlayfs())
+		SKIP(return, "overlayfs is not supported");
+
 	/* Sets rules on base directories (i.e. outside overlay scope). */
 	ruleset_fd = create_ruleset(_metadata, ACCESS_RW, layer1_base);
 	ASSERT_LE(0, ruleset_fd);

tools/testing/selftests/landlock/ptrace_test.c

@@ -19,6 +19,12 @@
 
 #include "common.h"
 
+/* Copied from security/yama/yama_lsm.c */
+#define YAMA_SCOPE_DISABLED 0
+#define YAMA_SCOPE_RELATIONAL 1
+#define YAMA_SCOPE_CAPABILITY 2
+#define YAMA_SCOPE_NO_ATTACH 3
+
 static void create_domain(struct __test_metadata *const _metadata)
 {
 	int ruleset_fd;
@@ -60,6 +66,25 @@ static int test_ptrace_read(const pid_t pid)
 	return 0;
 }
 
+static int get_yama_ptrace_scope(void)
+{
+	int ret;
+	char buf[2] = {};
+	const int fd = open("/proc/sys/kernel/yama/ptrace_scope", O_RDONLY);
+
+	if (fd < 0)
+		return 0;
+
+	if (read(fd, buf, 1) < 0) {
+		close(fd);
+		return -1;
+	}
+
+	ret = atoi(buf);
+	close(fd);
+	return ret;
+}
+
 /* clang-format off */
 FIXTURE(hierarchy) {};
 /* clang-format on */
@@ -232,8 +257,51 @@ TEST_F(hierarchy, trace)
 	pid_t child, parent;
 	int status, err_proc_read;
 	int pipe_child[2], pipe_parent[2];
+	int yama_ptrace_scope;
 	char buf_parent;
 	long ret;
+	bool can_read_child, can_trace_child, can_read_parent, can_trace_parent;
+
+	yama_ptrace_scope = get_yama_ptrace_scope();
+	ASSERT_LE(0, yama_ptrace_scope);
+
+	if (yama_ptrace_scope > YAMA_SCOPE_DISABLED)
+		TH_LOG("Incomplete tests due to Yama restrictions (scope %d)",
+		       yama_ptrace_scope);
+
+	/*
+	 * can_read_child is true if a parent process can read its child
+	 * process, which is only the case when the parent process is not
+	 * isolated from the child with a dedicated Landlock domain.
+	 */
+	can_read_child = !variant->domain_parent;
+
+	/*
+	 * can_trace_child is true if a parent process can trace its child
+	 * process.  This depends on two conditions:
+	 * - The parent process is not isolated from the child with a dedicated
+	 *   Landlock domain.
+	 * - Yama allows tracing children (up to YAMA_SCOPE_RELATIONAL).
+	 */
+	can_trace_child = can_read_child &&
+			  yama_ptrace_scope <= YAMA_SCOPE_RELATIONAL;
+
+	/*
+	 * can_read_parent is true if a child process can read its parent
+	 * process, which is only the case when the child process is not
+	 * isolated from the parent with a dedicated Landlock domain.
+	 */
+	can_read_parent = !variant->domain_child;
+
+	/*
+	 * can_trace_parent is true if a child process can trace its parent
+	 * process.  This depends on two conditions:
+	 * - The child process is not isolated from the parent with a dedicated
+	 *   Landlock domain.
+	 * - Yama is disabled (YAMA_SCOPE_DISABLED).
+	 */
+	can_trace_parent = can_read_parent &&
+			   yama_ptrace_scope <= YAMA_SCOPE_DISABLED;
 
 	/*
 	 * Removes all effective and permitted capabilities to not interfere
@@ -264,16 +332,21 @@ TEST_F(hierarchy, trace)
 		/* Waits for the parent to be in a domain, if any. */
 		ASSERT_EQ(1, read(pipe_parent[0], &buf_child, 1));
 
-		/* Tests PTRACE_ATTACH and PTRACE_MODE_READ on the parent. */
+		/* Tests PTRACE_MODE_READ on the parent. */
 		err_proc_read = test_ptrace_read(parent);
+		if (can_read_parent) {
+			EXPECT_EQ(0, err_proc_read);
+		} else {
+			EXPECT_EQ(EACCES, err_proc_read);
+		}
+
+		/* Tests PTRACE_ATTACH on the parent. */
 		ret = ptrace(PTRACE_ATTACH, parent, NULL, 0);
-		if (variant->domain_child) {
+		if (can_trace_parent) {
+			EXPECT_EQ(0, ret);
+		} else {
 			EXPECT_EQ(-1, ret);
 			EXPECT_EQ(EPERM, errno);
-			EXPECT_EQ(EACCES, err_proc_read);
-		} else {
-			EXPECT_EQ(0, ret);
-			EXPECT_EQ(0, err_proc_read);
 		}
 		if (ret == 0) {
 			ASSERT_EQ(parent, waitpid(parent, &status, 0));
@@ -283,11 +356,11 @@ TEST_F(hierarchy, trace)
 
 		/* Tests child PTRACE_TRACEME. */
 		ret = ptrace(PTRACE_TRACEME);
-		if (variant->domain_parent) {
+		if (can_trace_child) {
+			EXPECT_EQ(0, ret);
+		} else {
 			EXPECT_EQ(-1, ret);
 			EXPECT_EQ(EPERM, errno);
-		} else {
-			EXPECT_EQ(0, ret);
 		}
 
 		/*
@@ -296,7 +369,7 @@ TEST_F(hierarchy, trace)
 		 */
 		ASSERT_EQ(1, write(pipe_child[1], ".", 1));
 
-		if (!variant->domain_parent) {
+		if (can_trace_child) {
 			ASSERT_EQ(0, raise(SIGSTOP));
 		}
 
@@ -321,7 +394,7 @@ TEST_F(hierarchy, trace)
 	ASSERT_EQ(1, read(pipe_child[0], &buf_parent, 1));
 
 	/* Tests child PTRACE_TRACEME. */
-	if (!variant->domain_parent) {
+	if (can_trace_child) {
 		ASSERT_EQ(child, waitpid(child, &status, 0));
 		ASSERT_EQ(1, WIFSTOPPED(status));
 		ASSERT_EQ(0, ptrace(PTRACE_DETACH, child, NULL, 0));
@@ -331,17 +404,23 @@ TEST_F(hierarchy, trace)
 		EXPECT_EQ(ESRCH, errno);
 	}
 
-	/* Tests PTRACE_ATTACH and PTRACE_MODE_READ on the child. */
+	/* Tests PTRACE_MODE_READ on the child. */
 	err_proc_read = test_ptrace_read(child);
+	if (can_read_child) {
+		EXPECT_EQ(0, err_proc_read);
+	} else {
+		EXPECT_EQ(EACCES, err_proc_read);
+	}
+
+	/* Tests PTRACE_ATTACH on the child. */
 	ret = ptrace(PTRACE_ATTACH, child, NULL, 0);
-	if (variant->domain_parent) {
+	if (can_trace_child) {
+		EXPECT_EQ(0, ret);
+	} else {
 		EXPECT_EQ(-1, ret);
 		EXPECT_EQ(EPERM, errno);
-		EXPECT_EQ(EACCES, err_proc_read);
-	} else {
-		EXPECT_EQ(0, ret);
-		EXPECT_EQ(0, err_proc_read);
 	}
+
 	if (ret == 0) {
 		ASSERT_EQ(child, waitpid(child, &status, 0));
 		ASSERT_EQ(1, WIFSTOPPED(status));