diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c
index 50e4407398d816b769adc3b1b4a5d11dc8a203f1..ddfd93f13cc5f4a7e3948c137252f58bd1e9a370 100644
--- a/fs/overlayfs/file.c
+++ b/fs/overlayfs/file.c
@@ -144,11 +144,47 @@ static int ovl_release(struct inode *inode, struct file *file)
 
 static loff_t ovl_llseek(struct file *file, loff_t offset, int whence)
 {
-	struct inode *realinode = ovl_inode_real(file_inode(file));
+	struct inode *inode = file_inode(file);
+	struct fd real;
+	const struct cred *old_cred;
+	ssize_t ret;
+
+	/*
+	 * The two special cases below do not need to involve real fs,
+	 * so we can optimizing concurrent callers.
+	 */
+	if (offset == 0) {
+		if (whence == SEEK_CUR)
+			return file->f_pos;
+
+		if (whence == SEEK_SET)
+			return vfs_setpos(file, 0, 0);
+	}
+
+	ret = ovl_real_fdget(file, &real);
+	if (ret)
+		return ret;
 
-	return generic_file_llseek_size(file, offset, whence,
-					realinode->i_sb->s_maxbytes,
-					i_size_read(realinode));
+	/*
+	 * Overlay file f_pos is the master copy that is preserved
+	 * through copy up and modified on read/write, but only real
+	 * fs knows how to SEEK_HOLE/SEEK_DATA and real fs may impose
+	 * limitations that are more strict than ->s_maxbytes for specific
+	 * files, so we use the real file to perform seeks.
+	 */
+	inode_lock(inode);
+	real.file->f_pos = file->f_pos;
+
+	old_cred = ovl_override_creds(inode->i_sb);
+	ret = vfs_llseek(real.file, offset, whence);
+	revert_creds(old_cred);
+
+	file->f_pos = real.file->f_pos;
+	inode_unlock(inode);
+
+	fdput(real);
+
+	return ret;
 }
 
 static void ovl_file_accessed(struct file *file)