On Sat, 2024-03-16 at 23:57 +0000, Al Viro wrote:

On Fri, Mar 15, 2024 at 12:52:54PM -0400, Jeff Layton wrote:

quoted

@@ -4603,9 +4606,12 @@ int vfs_link(struct dentry *old_dentry, struct mnt_idmap *idmap,
 	else if (max_links && inode->i_nlink >= max_links)
 		error = -EMLINK;
 	else {
-		error = try_break_deleg(inode, delegated_inode);
-		if (!error)
-			error = dir->i_op->link(old_dentry, dir, new_dentry);
+		error = try_break_deleg(dir, delegated_inode);
+		if (!error) {
+			error = try_break_deleg(inode, delegated_inode);
+			if (!error)
+				error = dir->i_op->link(old_dentry, dir, new_dentry);
+		}

A minor nit: that might be easier to follow as
		error = try_break_deleg(dir, delegated_inode);
		if (!error)
			error = try_break_deleg(inode, delegated_inode);
		if (!error)
			error = dir->i_op->link(old_dentry, dir, new_dentry);

and let the compiler deal with optimizing it - any C compiler is going to be
able to figure out that one out.  vfs_link() is a mix of those styles anyway -
we have
        if (!error && (inode->i_state & I_LINKABLE)) {
                spin_lock(&inode->i_lock);
                inode->i_state &= ~I_LINKABLE;
                spin_unlock(&inode->i_lock);
        }
immediately afterwards; might as well make that consistent, especially since
you are getting more shallow nesting that way.

Sounds good. Fixed in my tree.

Thanks for the review so far!
-- 
Jeff Layton [off-list ref]