reiserfs: Fix permissions on .reiserfs_priv

Commit 677c9b2e393a0cd203bd54e9c18b012b2c73305a removed the magic from the lookup code to hide the .reiserfs_priv directory since it was getting loaded at mount-time instead. The intent was that the entry would be hidden from the user via a poisoned d_compare, but this was faulty.

This introduced a security issue where unpriviledged users could access and modify extended attributes or ACLs belonging to other users, including root.

This patch resolves the issue by properly hiding .reiserfs_priv. This was the intent of the xattr poisoning code, but it appears to have never worked as expected. This is fixed by using d_revalidate instead of d_compare.

This patch makes -oexpose_privroot a no-op. I'm fine leaving it this way. The effort involved in working out the corner cases wrt permissions and caching outweigh the benefit of the feature.

Signed-off-by Jeff Mahoney <[email protected]>

---

 fs/reiserfs/dir.c   |    2 --
 fs/reiserfs/xattr.c |   17 ++++-------------
 2 files changed, 4 insertions(+), 15 deletions(-)

fs/reiserfs/dir.c +0 -2

@@ -45,8 +45,6 @@ static inline bool is_privroot_deh(struc
 				   struct reiserfs_de_head *deh)
 {
 	struct dentry *privroot = REISERFS_SB(dir->d_sb)->priv_root;
-	if (reiserfs_expose_privroot(dir->d_sb))
-		return 0;
 	return (dir == dir->d_parent && privroot->d_inode &&
 	        deh->deh_objectid == INODE_PKEY(privroot->d_inode)->k_objectid);
 }

fs/reiserfs/xattr.c +4 -13

@@ -972,21 +972,13 @@ int reiserfs_permission(struct inode *in
 	return generic_permission(inode, mask, NULL);
 }
 
-/* This will catch lookups from the fs root to .reiserfs_priv */
-static int
-xattr_lookup_poison(struct dentry *dentry, struct qstr *q1, struct qstr *name)
+static int xattr_hide_revalidate(struct dentry *dentry, struct nameidata *nd)
 {
-	struct dentry *priv_root = REISERFS_SB(dentry->d_sb)->priv_root;
-	if (container_of(q1, struct dentry, d_name) == priv_root)
-		return -ENOENT;
-	if (q1->len == name->len &&
-		   !memcmp(q1->name, name->name, name->len))
-		return 0;
-	return 1;
+	return -EPERM;
 }
 
 static const struct dentry_operations xattr_lookup_poison_ops = {
-	.d_compare = xattr_lookup_poison,
+	.d_revalidate = xattr_hide_revalidate,
 };
 
 int reiserfs_lookup_privroot(struct super_block *s)

@@ -1000,8 +992,7 @@ int reiserfs_lookup_privroot(struct supe
 				strlen(PRIVROOT_NAME));
 	if (!IS_ERR(dentry)) {
 		REISERFS_SB(s)->priv_root = dentry;
-		if (!reiserfs_expose_privroot(s))
-			s->s_root->d_op = &xattr_lookup_poison_ops;
+		dentry->d_op = &xattr_lookup_poison_ops;
 		if (dentry->d_inode)
 			dentry->d_inode->i_flags |= S_PRIVATE;
 	} else

Quoted 4 lines · Jeff Mahoney

[...] This patch resolves the issue by properly hiding .reiserfs_priv. This was the intent of the xattr poisoning code, but it appears to have never worked as expected. This is fixed by using d_revalidate instead of d_compare. [...]

Quoted 5 lines · Edward Shishkin

Quoted 4 lines · Jeff Mahoney

[...] This patch resolves the issue by properly hiding .reiserfs_priv. This was the intent of the xattr poisoning code, but it appears to have never worked as expected. This is fixed by using d_revalidate instead of d_compare. [...]

The patch seems to rely on the fact that the priv_root dentry is stuck in the dcache for as long as the filesystem is mounted, so that attempts to follow ".reiserfs_priv" are guaranteed to match that dentry in the cache and be refused by xattr_hide_revalidate. Is my understanding correct?