openvswitch: Make tunnel set action attach a metadata dst

Utilize the new metadata dst to attach encapsulation instructions to
the skb. The existing egress_tun_info via the OVS_CB() is left in
place until all tunnel vports have been converted to the new method.

Signed-off-by: Thomas Graf <[e-mail]>
Signed-off-by: Pravin B Shelar <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

net/openvswitch/actions.c

···
 {
 	/* Only tunnel set execution is supported without a mask. */
 	if (nla_type(a) == OVS_KEY_ATTR_TUNNEL_INFO) {
-		OVS_CB(skb)->egress_tun_info = nla_data(a);
+		struct ovs_tunnel_info *tun = nla_data(a);
+
+		skb_dst_drop(skb);
+		dst_hold((struct dst_entry *)tun->tun_dst);
+		skb_dst_set(skb, (struct dst_entry *)tun->tun_dst);
+
+		/* FIXME: Remove when all vports have been converted */
+		OVS_CB(skb)->egress_tun_info = &tun->tun_dst->u.tun_info;
+
 		return 0;
 	}
 

net/openvswitch/datapath.c

···
 		}
 		ovs_unlock();
 
-		ovs_nla_free_flow_actions(old_acts);
+		ovs_nla_free_flow_actions_rcu(old_acts);
 		ovs_flow_free(new_flow, false);
 	}
 
···
 	ovs_unlock();
 	kfree_skb(reply);
 err_kfree_acts:
-	kfree(acts);
+	ovs_nla_free_flow_actions(acts);
 err_kfree_flow:
 	ovs_flow_free(new_flow, false);
 error:
···
 	if (reply)
 		ovs_notify(&dp_flow_genl_family, reply, info);
 	if (old_acts)
-		ovs_nla_free_flow_actions(old_acts);
+		ovs_nla_free_flow_actions_rcu(old_acts);
 
 	return 0;
 
···
 	ovs_unlock();
 	kfree_skb(reply);
 err_kfree_acts:
-	kfree(acts);
+	ovs_nla_free_flow_actions(acts);
 error:
 	return error;
 }

net/openvswitch/flow.h

···
 #include <linux/flex_array.h>
 #include <net/inet_ecn.h>
 #include <net/ip_tunnels.h>
+#include <net/dst_metadata.h>
 
 struct sk_buff;
 
···
 	(FIELD_SIZEOF(struct sw_flow_key, tun_opts) - opt_len)
 #define TUN_METADATA_OPTS(flow_key, opt_len) \
 	((void *)((flow_key)->tun_opts + TUN_METADATA_OFFSET(opt_len)))
+
+struct ovs_tunnel_info {
+	struct metadata_dst	*tun_dst;
+};
 
 #define OVS_SW_FLOW_KEY_METADATA_SIZE			\
 	(offsetof(struct sw_flow_key, recirc_id) +	\

net/openvswitch/flow_netlink.c

···
 	return sfa;
 }
 
-/* Schedules 'sf_acts' to be freed after the next RCU grace period.
- * The caller must hold rcu_read_lock for this to be sensible. */
+static void ovs_nla_free_set_action(const struct nlattr *a)
+{
+	const struct nlattr *ovs_key = nla_data(a);
+	struct ovs_tunnel_info *ovs_tun;
+
+	switch (nla_type(ovs_key)) {
+	case OVS_KEY_ATTR_TUNNEL_INFO:
+		ovs_tun = nla_data(ovs_key);
+		dst_release((struct dst_entry *)ovs_tun->tun_dst);
+		break;
+	}
+}
+
 void ovs_nla_free_flow_actions(struct sw_flow_actions *sf_acts)
 {
-	kfree_rcu(sf_acts, rcu);
+	const struct nlattr *a;
+	int rem;
+
+	if (!sf_acts)
+		return;
+
+	nla_for_each_attr(a, sf_acts->actions, sf_acts->actions_len, rem) {
+		switch (nla_type(a)) {
+		case OVS_ACTION_ATTR_SET:
+			ovs_nla_free_set_action(a);
+			break;
+		}
+	}
+
+	kfree(sf_acts);
+}
+
+static void __ovs_nla_free_flow_actions(struct rcu_head *head)
+{
+	ovs_nla_free_flow_actions(container_of(head, struct sw_flow_actions, rcu));
+}
+
+/* Schedules 'sf_acts' to be freed after the next RCU grace period.
+ * The caller must hold rcu_read_lock for this to be sensible. */
+void ovs_nla_free_flow_actions_rcu(struct sw_flow_actions *sf_acts)
+{
+	call_rcu(&sf_acts->rcu, __ovs_nla_free_flow_actions);
 }
 
 static struct nlattr *reserve_sfa_size(struct sw_flow_actions **sfa,
···
 {
 	struct sw_flow_match match;
 	struct sw_flow_key key;
+	struct metadata_dst *tun_dst;
 	struct ip_tunnel_info *tun_info;
+	struct ovs_tunnel_info *ovs_tun;
 	struct nlattr *a;
 	int err = 0, start, opts_type;
 
···
 	if (start < 0)
 		return start;
 
-	a = __add_action(sfa, OVS_KEY_ATTR_TUNNEL_INFO, NULL,
-			 sizeof(*tun_info) + key.tun_opts_len, log);
-	if (IS_ERR(a))
-		return PTR_ERR(a);
+	tun_dst = metadata_dst_alloc(key.tun_opts_len, GFP_KERNEL);
+	if (!tun_dst)
+		return -ENOMEM;
 
-	tun_info = nla_data(a);
+	a = __add_action(sfa, OVS_KEY_ATTR_TUNNEL_INFO, NULL,
+			 sizeof(*ovs_tun), log);
+	if (IS_ERR(a)) {
+		dst_release((struct dst_entry *)tun_dst);
+		return PTR_ERR(a);
+	}
+
+	ovs_tun = nla_data(a);
+	ovs_tun->tun_dst = tun_dst;
+
+	tun_info = &tun_dst->u.tun_info;
+	tun_info->mode = IP_TUNNEL_INFO_TX;
 	tun_info->key = key.tun_key;
 	tun_info->options_len = key.tun_opts_len;
 
···
 	err = __ovs_nla_copy_actions(attr, key, 0, sfa, key->eth.type,
 				     key->eth.tci, log);
 	if (err)
-		kfree(*sfa);
+		ovs_nla_free_flow_actions(*sfa);
 
 	return err;
 }
···
 
 	switch (key_type) {
 	case OVS_KEY_ATTR_TUNNEL_INFO: {
-		struct ip_tunnel_info *tun_info = nla_data(ovs_key);
+		struct ovs_tunnel_info *ovs_tun = nla_data(ovs_key);
+		struct ip_tunnel_info *tun_info = &ovs_tun->tun_dst->u.tun_info;
 
 		start = nla_nest_start(skb, OVS_ACTION_ATTR_SET);
 		if (!start)

net/openvswitch/flow_netlink.h

···
 			int len, struct sk_buff *skb);
 
 void ovs_nla_free_flow_actions(struct sw_flow_actions *);
+void ovs_nla_free_flow_actions_rcu(struct sw_flow_actions *);
 
 #endif /* flow_netlink.h */

net/openvswitch/flow_table.c

···
 
 #include "flow.h"
 #include "datapath.h"
+#include "flow_netlink.h"
 #include <linux/uaccess.h>
 #include <linux/netdevice.h>
 #include <linux/etherdevice.h>
···
 
 	if (ovs_identifier_is_key(&flow->id))
 		kfree(flow->id.unmasked_key);
-	kfree((struct sw_flow_actions __force *)flow->sf_acts);
+	if (flow->sf_acts)
+		ovs_nla_free_flow_actions((struct sw_flow_actions __force *)flow->sf_acts);
 	for_each_node(node)
 		if (flow->stats[node])
 			kmem_cache_free(flow_stats_cache,