ovl: whiteout inode sharing

Share inode with different whiteout files for saving inode and speeding up
delete operation.

If EMLINK is encountered when linking a shared whiteout, create a new one.
In case of any other error, disable sharing for this super block.

Note: ofs->whiteout is protected by inode lock on workdir.
Signed-off-by: Chengguang Xu <cgxu519@mykernel.net>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>

fs/overlayfs/dir.c

@@ -62,35 +62,59 @@ struct dentry *ovl_lookup_temp(struct dentry *workdir)
 }
 
 /* caller holds i_mutex on workdir */
-static struct dentry *ovl_whiteout(struct dentry *workdir)
+static struct dentry *ovl_whiteout(struct ovl_fs *ofs)
 {
 	int err;
 	struct dentry *whiteout;
+	struct dentry *workdir = ofs->workdir;
 	struct inode *wdir = workdir->d_inode;
 
+	if (!ofs->whiteout) {
 		whiteout = ovl_lookup_temp(workdir);
 		if (IS_ERR(whiteout))
-		return whiteout;
+			goto out;
 
 		err = ovl_do_whiteout(wdir, whiteout);
 		if (err) {
 			dput(whiteout);
 			whiteout = ERR_PTR(err);
+			goto out;
+		}
+		ofs->whiteout = whiteout;
 	}
 
+	if (ofs->share_whiteout) {
+		whiteout = ovl_lookup_temp(workdir);
+		if (IS_ERR(whiteout))
+			goto out;
+
+		err = ovl_do_link(ofs->whiteout, wdir, whiteout);
+		if (!err)
+			goto out;
+
+		if (err != -EMLINK) {
+			pr_warn("Failed to link whiteout - disabling whiteout inode sharing(nlink=%u, err=%i)\n",
+				ofs->whiteout->d_inode->i_nlink, err);
+			ofs->share_whiteout = false;
+		}
+		dput(whiteout);
+	}
+	whiteout = ofs->whiteout;
+	ofs->whiteout = NULL;
+out:
 	return whiteout;
 }
 
 /* Caller must hold i_mutex on both workdir and dir */
-int ovl_cleanup_and_whiteout(struct dentry *workdir, struct inode *dir,
+int ovl_cleanup_and_whiteout(struct ovl_fs *ofs, struct inode *dir,
 			     struct dentry *dentry)
 {
-	struct inode *wdir = workdir->d_inode;
+	struct inode *wdir = ofs->workdir->d_inode;
 	struct dentry *whiteout;
 	int err;
 	int flags = 0;
 
-	whiteout = ovl_whiteout(workdir);
+	whiteout = ovl_whiteout(ofs);
 	err = PTR_ERR(whiteout);
 	if (IS_ERR(whiteout))
 		return err;
@@ -715,6 +739,7 @@ static bool ovl_matches_upper(struct dentry *dentry, struct dentry *upper)
 static int ovl_remove_and_whiteout(struct dentry *dentry,
 				   struct list_head *list)
 {
+	struct ovl_fs *ofs = OVL_FS(dentry->d_sb);
 	struct dentry *workdir = ovl_workdir(dentry);
 	struct dentry *upperdir = ovl_dentry_upper(dentry->d_parent);
 	struct dentry *upper;
@@ -748,7 +773,7 @@ static int ovl_remove_and_whiteout(struct dentry *dentry,
 		goto out_dput_upper;
 	}
 
-	err = ovl_cleanup_and_whiteout(workdir, d_inode(upperdir), upper);
+	err = ovl_cleanup_and_whiteout(ofs, d_inode(upperdir), upper);
 	if (err)
 		goto out_d_drop;
 

fs/overlayfs/overlayfs.h

@@ -455,7 +455,7 @@ static inline void ovl_copyflags(struct inode *from, struct inode *to)
 
 /* dir.c */
 extern const struct inode_operations ovl_dir_inode_operations;
-int ovl_cleanup_and_whiteout(struct dentry *workdir, struct inode *dir,
+int ovl_cleanup_and_whiteout(struct ovl_fs *ofs, struct inode *dir,
 			     struct dentry *dentry);
 struct ovl_cattr {
 	dev_t rdev;

fs/overlayfs/ovl_entry.h

@@ -68,6 +68,7 @@ struct ovl_fs {
 	/* Did we take the inuse lock? */
 	bool upperdir_locked;
 	bool workdir_locked;
+	bool share_whiteout;
 	/* Traps in ovl inode cache */
 	struct inode *upperdir_trap;
 	struct inode *workbasedir_trap;
@@ -77,6 +78,8 @@ struct ovl_fs {
 	int xino_mode;
 	/* For allocation of non-persistent inode numbers */
 	atomic_long_t last_ino;
+	/* Whiteout dentry cache */
+	struct dentry *whiteout;
 };
 
 static inline struct ovl_fs *OVL_FS(struct super_block *sb)

fs/overlayfs/readdir.c

@@ -1154,7 +1154,7 @@ int ovl_indexdir_cleanup(struct ovl_fs *ofs)
 			 * Whiteout orphan index to block future open by
 			 * handle after overlay nlink dropped to zero.
 			 */
-			err = ovl_cleanup_and_whiteout(indexdir, dir, index);
+			err = ovl_cleanup_and_whiteout(ofs, dir, index);
 		} else {
 			/* Cleanup orphan index entries */
 			err = ovl_cleanup(dir, index);

fs/overlayfs/super.c

@@ -217,6 +217,7 @@ static void ovl_free_fs(struct ovl_fs *ofs)
 	iput(ofs->indexdir_trap);
 	iput(ofs->workdir_trap);
 	iput(ofs->upperdir_trap);
+	dput(ofs->whiteout);
 	dput(ofs->indexdir);
 	dput(ofs->workdir);
 	if (ofs->workdir_locked)
@@ -1776,6 +1777,9 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	if (!cred)
 		goto out_err;
 
+	/* Is there a reason anyone would want not to share whiteouts? */
+	ofs->share_whiteout = true;
+
 	ofs->config.index = ovl_index_def;
 	ofs->config.nfs_export = ovl_nfs_export_def;
 	ofs->config.xino = ovl_xino_def();

fs/overlayfs/util.c

@@ -707,7 +707,8 @@ static void ovl_cleanup_index(struct dentry *dentry)
 		index = NULL;
 	} else if (ovl_index_all(dentry->d_sb)) {
 		/* Whiteout orphan index to block future open by handle */
-		err = ovl_cleanup_and_whiteout(indexdir, dir, index);
+		err = ovl_cleanup_and_whiteout(OVL_FS(dentry->d_sb),
+					       dir, index);
 	} else {
 		/* Cleanup orphan index entries */
 		err = ovl_cleanup(dir, index);