[16/32] xfs: hide metadata inodes from everyone because they are special

Commit Message

Darrick J. Wong Dec. 31, 2023, 9:10 p.m. UTC

From: Darrick J. Wong <[e-mail]>

Metadata inodes are private files and therefore cannot be exposed to
userspace.  This means no bulkstat, no open-by-handle, no linking them
into the directory tree, and no feeding them to LSMs.  As such, we mark
them S_PRIVATE, which stops all that.

While we're at it, put them in a separate lockdep class so that it won't
get confused by "recursive" i_rwsem locking such as what happens when we
write to a rt file and need to allocate from the rt bitmap file.

Signed-off-by: Darrick J. Wong <[e-mail]>
---
 fs/xfs/scrub/tempfile.c |    8 ++++++++
 fs/xfs/xfs_iops.c       |   34 ++++++++++++++++++++++++++++++++--
 2 files changed, 40 insertions(+), 2 deletions(-)

Patch

diff --git a/fs/xfs/scrub/tempfile.c b/fs/xfs/scrub/tempfile.c
index db3a9a47e1f82..583bbd5a02bc9 100644
--- a/fs/xfs/scrub/tempfile.c
+++ b/fs/xfs/scrub/tempfile.c
@@ -867,6 +867,14 @@  xrep_is_tempfile(
 	const struct xfs_inode	*ip)
 {
 	const struct inode	*inode = &ip->i_vnode;
+	struct xfs_mount	*mp = ip->i_mount;
+
+	/*
+	 * Files in the metadata directory tree also have S_PRIVATE set and
+	 * IOP_XATTR unset, so we must distinguish them separately.
+	 */
+	if (xfs_has_metadir(mp) && (ip->i_diflags2 & XFS_DIFLAG2_METADIR))
+		return false;
 
 	if (IS_PRIVATE(inode) && !(inode->i_opflags & IOP_XATTR))
 		return true;
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index 90296b3c838aa..461e77dd54e38 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -45,6 +45,15 @@ 
 static struct lock_class_key xfs_nondir_ilock_class;
 static struct lock_class_key xfs_dir_ilock_class;
 
+/*
+ * Metadata directories and files are not exposed to userspace, which means
+ * that they never access any of the VFS IO locks and never experience page
+ * faults.  Give them separate locking classes so that lockdep will not
+ * complain about conflicts that cannot happen.
+ */
+static struct lock_class_key xfs_metadata_file_ilock_class;
+static struct lock_class_key xfs_metadata_dir_ilock_class;
+
 static int
 xfs_initxattrs(
 	struct inode		*inode,
@@ -1291,6 +1300,7 @@  xfs_setup_inode(
 {
 	struct inode		*inode = &ip->i_vnode;
 	gfp_t			gfp_mask;
+	bool			is_meta = xfs_is_metadata_inode(ip);
 
 	inode->i_ino = ip->i_ino;
 	inode->i_state |= I_NEW;
@@ -1302,6 +1312,16 @@  xfs_setup_inode(
 	i_size_write(inode, ip->i_disk_size);
 	xfs_diflags_to_iflags(ip, true);
 
+	/*
+	 * Mark our metadata files as private so that LSMs and the ACL code
+	 * don't try to add their own metadata or reason about these files,
+	 * and users cannot ever obtain file handles to them.
+	 */
+	if (is_meta) {
+		inode->i_flags |= S_PRIVATE;
+		inode->i_opflags &= ~IOP_XATTR;
+	}
+
 	if (S_ISDIR(inode->i_mode)) {
 		/*
 		 * We set the i_rwsem class here to avoid potential races with
@@ -1311,9 +1331,19 @@  xfs_setup_inode(
 		 */
 		lockdep_set_class(&inode->i_rwsem,
 				  &inode->i_sb->s_type->i_mutex_dir_key);
-		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_dir_ilock_class);
+		if (is_meta)
+			lockdep_set_class(&ip->i_lock.mr_lock,
+					  &xfs_metadata_dir_ilock_class);
+		else
+			lockdep_set_class(&ip->i_lock.mr_lock,
+					  &xfs_dir_ilock_class);
 	} else {
-		lockdep_set_class(&ip->i_lock.mr_lock, &xfs_nondir_ilock_class);
+		if (is_meta)
+			lockdep_set_class(&ip->i_lock.mr_lock,
+					  &xfs_metadata_file_ilock_class);
+		else
+			lockdep_set_class(&ip->i_lock.mr_lock,
+					  &xfs_nondir_ilock_class);
 	}
 
 	/*