ovl: hash directory inodes for fsnotify

fsnotify pins a watched directory inode in cache, but if directory dentry
is released, new lookup will allocate a new dentry and a new inode.
Directory events will be notified on the new inode, while fsnotify listener
is watching the old pinned inode.

Hash all directory inodes to reuse the pinned inode on lookup. Pure upper
dirs are hashes by real upper inode, merge and lower dirs are hashed by
real lower inode.

The reference to lower inode was being held by the lower dentry object
in the overlay dentry (oe->lowerstack[0]). Releasing the overlay dentry
may drop lower inode refcount to zero. Add a refcount on behalf of the
overlay inode to prevent that.

As a by-product, hashing directory inodes also detects multiple
redirected dirs to the same lower dir and uncovered redirected dir
target on and returns -ESTALE on lookup.

The reported issue dates back to initial version of overlayfs, but this
patch depends on ovl_inode code that was introduced in kernel v4.13.

Cc: <[email protected]> #v4.13
Reported-by: Niklas Cassel <[email protected]>
Signed-off-by: Amir Goldstein <[email protected]>
Signed-off-by: Miklos Szeredi <[email protected]>
Tested-by: Niklas Cassel <[email protected]>

Author: amir73il

fs/overlayfs/inode.c

@@ -606,6 +606,16 @@ static int ovl_inode_set(struct inode *inode, void *data)
 static bool ovl_verify_inode(struct inode *inode, struct dentry *lowerdentry,
 			     struct dentry *upperdentry)
 {
+	if (S_ISDIR(inode->i_mode)) {
+		/* Real lower dir moved to upper layer under us? */
+		if (!lowerdentry && ovl_inode_lower(inode))
+			return false;
+
+		/* Lookup of an uncovered redirect origin? */
+		if (!upperdentry && ovl_inode_upper(inode))
+			return false;
+	}
+
 	/*
 	 * Allow non-NULL lower inode in ovl_inode even if lowerdentry is NULL.
 	 * This happens when finding a copied up overlay inode for a renamed
@@ -633,6 +643,8 @@ struct inode *ovl_get_inode(struct dentry *dentry, struct dentry *upperdentry,
 	struct inode *inode;
 	/* Already indexed or could be indexed on copy up? */
 	bool indexed = (index || (ovl_indexdir(dentry->d_sb) && !upperdentry));
+	struct dentry *origin = indexed ? lowerdentry : NULL;
+	bool is_dir;
 
 	if (WARN_ON(upperdentry && indexed && !lowerdentry))
 		return ERR_PTR(-EIO);
@@ -641,15 +653,19 @@ struct inode *ovl_get_inode(struct dentry *dentry, struct dentry *upperdentry,
 		realinode = d_inode(lowerdentry);
 
 	/*
-	 * Copy up origin (lower) may exist for non-indexed upper, but we must
-	 * not use lower as hash key in that case.
-	 * Hash inodes that are or could be indexed by origin inode and
-	 * non-indexed upper inodes that could be hard linked by upper inode.
+	 * Copy up origin (lower) may exist for non-indexed non-dir upper, but
+	 * we must not use lower as hash key in that case.
+	 * Hash non-dir that is or could be indexed by origin inode.
+	 * Hash dir that is or could be merged by origin inode.
+	 * Hash pure upper and non-indexed non-dir by upper inode.
 	 */
-	if (!S_ISDIR(realinode->i_mode) && (upperdentry || indexed)) {
-		struct inode *key = d_inode(indexed ? lowerdentry :
-						      upperdentry);
-		unsigned int nlink;
+	is_dir = S_ISDIR(realinode->i_mode);
+	if (is_dir)
+		origin = lowerdentry;
+
+	if (upperdentry || origin) {
+		struct inode *key = d_inode(origin ?: upperdentry);
+		unsigned int nlink = is_dir ? 1 : realinode->i_nlink;
 
 		inode = iget5_locked(dentry->d_sb, (unsigned long) key,
 				     ovl_inode_test, ovl_inode_set, key);
@@ -670,8 +686,9 @@ struct inode *ovl_get_inode(struct dentry *dentry, struct dentry *upperdentry,
 			goto out;
 		}
 
-		nlink = ovl_get_nlink(lowerdentry, upperdentry,
-				      realinode->i_nlink);
+		/* Recalculate nlink for non-dir due to indexing */
+		if (!is_dir)
+			nlink = ovl_get_nlink(lowerdentry, upperdentry, nlink);
 		set_nlink(inode, nlink);
 	} else {
 		inode = new_inode(dentry->d_sb);
@@ -685,7 +702,7 @@ struct inode *ovl_get_inode(struct dentry *dentry, struct dentry *upperdentry,
 		ovl_set_flag(OVL_IMPURE, inode);
 
 	/* Check for non-merge dir that may have whiteouts */
-	if (S_ISDIR(realinode->i_mode)) {
+	if (is_dir) {
 		struct ovl_entry *oe = dentry->d_fsdata;
 
 		if (((upperdentry && lowerdentry) || oe->numlower > 1) ||

fs/overlayfs/super.c

@@ -211,6 +211,7 @@ static void ovl_destroy_inode(struct inode *inode)
 	struct ovl_inode *oi = OVL_I(inode);
 
 	dput(oi->__upperdentry);
+	iput(oi->lower);
 	kfree(oi->redirect);
 	ovl_dir_cache_free(inode);
 	mutex_destroy(&oi->lock);

fs/overlayfs/util.c

@@ -257,7 +257,7 @@ void ovl_inode_init(struct inode *inode, struct dentry *upperdentry,
 	if (upperdentry)
 		OVL_I(inode)->__upperdentry = upperdentry;
 	if (lowerdentry)
-		OVL_I(inode)->lower = d_inode(lowerdentry);
+		OVL_I(inode)->lower = igrab(d_inode(lowerdentry));
 
 	ovl_copyattr(d_inode(upperdentry ?: lowerdentry), inode);
 }
@@ -273,7 +273,7 @@ void ovl_inode_update(struct inode *inode, struct dentry *upperdentry)
 	 */
 	smp_wmb();
 	OVL_I(inode)->__upperdentry = upperdentry;
-	if (!S_ISDIR(upperinode->i_mode) && inode_unhashed(inode)) {
+	if (inode_unhashed(inode)) {
 		inode->i_private = upperinode;
 		__insert_inode_hash(inode, (unsigned long) upperinode);
 	}