xsk: remove AF_XDP socket from map when the socket is released

When an AF_XDP socket is released/closed the XSKMAP still holds a
reference to the socket in a "released" state. The socket will still
use the netdev queue resource, and block newly created sockets from
attaching to that queue, but no user application can access the
fill/complete/rx/tx queues. This results in that all applications need
to explicitly clear the map entry from the old "zombie state"
socket. This should be done automatically.

In this patch, the sockets tracks, and have a reference to, which maps
it resides in. When the socket is released, it will remove itself from
all maps.

Suggested-by: Bruce Richardson <bruce.richardson@intel.com>
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

1 files changed, 50 insertions, 0 deletions

diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
index 9f900b56b15b..ee4428a892fa 100644
--- a/net/xdp/xsk.c
+++ b/net/xdp/xsk.c
@@ -429,6 +429,52 @@ static void xsk_unbind_dev(struct xdp_sock *xs)
 	dev_put(dev);
 }
 
+static struct xsk_map *xsk_get_map_list_entry(struct xdp_sock *xs,
+					      struct xdp_sock ***map_entry)
+{
+	struct xsk_map *map = NULL;
+	struct xsk_map_node *node;
+
+	*map_entry = NULL;
+
+	spin_lock_bh(&xs->map_list_lock);
+	node = list_first_entry_or_null(&xs->map_list, struct xsk_map_node,
+					node);
+	if (node) {
+		WARN_ON(xsk_map_inc(node->map));
+		map = node->map;
+		*map_entry = node->map_entry;
+	}
+	spin_unlock_bh(&xs->map_list_lock);
+	return map;
+}
+
+static void xsk_delete_from_maps(struct xdp_sock *xs)
+{
+	/* This function removes the current XDP socket from all the
+	 * maps it resides in. We need to take extra care here, due to
+	 * the two locks involved. Each map has a lock synchronizing
+	 * updates to the entries, and each socket has a lock that
+	 * synchronizes access to the list of maps (map_list). For
+	 * deadlock avoidance the locks need to be taken in the order
+	 * "map lock"->"socket map list lock". We start off by
+	 * accessing the socket map list, and take a reference to the
+	 * map to guarantee existence between the
+	 * xsk_get_map_list_entry() and xsk_map_try_sock_delete()
+	 * calls. Then we ask the map to remove the socket, which
+	 * tries to remove the socket from the map. Note that there
+	 * might be updates to the map between
+	 * xsk_get_map_list_entry() and xsk_map_try_sock_delete().
+	 */
+	struct xdp_sock **map_entry = NULL;
+	struct xsk_map *map;
+
+	while ((map = xsk_get_map_list_entry(xs, &map_entry))) {
+		xsk_map_try_sock_delete(map, xs, map_entry);
+		xsk_map_put(map);
+	}
+}
+
 static int xsk_release(struct socket *sock)
 {
 	struct sock *sk = sock->sk;
@@ -448,6 +494,7 @@ static int xsk_release(struct socket *sock)
 	sock_prot_inuse_add(net, sk->sk_prot, -1);
 	local_bh_enable();
 
+	xsk_delete_from_maps(xs);
 	xsk_unbind_dev(xs);
 
 	xskq_destroy(xs->rx);
@@ -964,6 +1011,9 @@ static int xsk_create(struct net *net, struct socket *sock, int protocol,
 	spin_lock_init(&xs->rx_lock);
 	spin_lock_init(&xs->tx_completion_lock);
 
+	INIT_LIST_HEAD(&xs->map_list);
+	spin_lock_init(&xs->map_list_lock);
+
 	mutex_lock(&net->xdp.lock);
 	sk_add_node_rcu(sk, &net->xdp.list);
 	mutex_unlock(&net->xdp.lock);