Revert "af_unix: fix hard linked sockets on overlay"

This reverts commit eb0a4a47.

Since commit 51f7e52d ("ovl: share inode for hard link") there's no
need to call d_real_inode() to check two overlay inodes for equality.

Side effect of this revert is that it's no longer possible to connect one
socket on overlayfs to one on the underlying layer (something which didn't
make sense anyway).
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>

net/unix/af_unix.c

@@ -315,7 +315,7 @@ static struct sock *unix_find_socket_byinode(struct inode *i)
 		    &unix_socket_table[i->i_ino & (UNIX_HASH_SIZE - 1)]) {
 		struct dentry *dentry = unix_sk(s)->path.dentry;
 
-		if (dentry && d_real_inode(dentry) == i) {
+		if (dentry && d_backing_inode(dentry) == i) {
 			sock_hold(s);
 			goto found;
 		}
@@ -913,7 +913,7 @@ static struct sock *unix_find_other(struct net *net,
 		err = kern_path(sunname->sun_path, LOOKUP_FOLLOW, &path);
 		if (err)
 			goto fail;
-		inode = d_real_inode(path.dentry);
+		inode = d_backing_inode(path.dentry);
 		err = inode_permission(inode, MAY_WRITE);
 		if (err)
 			goto put_fail;
@@ -1040,7 +1040,7 @@ static int unix_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 			goto out_up;
 		}
 		addr->hash = UNIX_HASH_SIZE;
-		hash = d_real_inode(path.dentry)->i_ino & (UNIX_HASH_SIZE - 1);
+		hash = d_backing_inode(path.dentry)->i_ino & (UNIX_HASH_SIZE - 1);
 		spin_lock(&unix_table_lock);
 		u->path = path;
 		list = &unix_socket_table[hash];