xsk: do not remove umem from netdevice on fall-back to copy-mode

Commit c9b47cc1 ("xsk: fix bug when trying to use both copy and
zero-copy on one queue id") stores the umem into the netdev._rx
struct. However, the patch incorrectly removed the umem from the
netdev._rx struct when user-space passed "best-effort" mode
(i.e. select the fastest possible option available), and zero-copy
mode was not available. This commit fixes that.

Fixes: c9b47cc1 ("xsk: fix bug when trying to use both copy and zero-copy on one queue id")
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

net/xdp/xdp_umem.c

@@ -125,9 +125,10 @@ int xdp_umem_assign_dev(struct xdp_umem *umem, struct net_device *dev,
 	return 0;
 
 err_unreg_umem:
-	xdp_clear_umem_at_qid(dev, queue_id);
 	if (!force_zc)
 		err = 0; /* fallback to copy mode */
+	if (err)
+		xdp_clear_umem_at_qid(dev, queue_id);
 out_rtnl_unlock:
 	rtnl_unlock();
 	return err;