ovs: Turn vports with dependencies into separate modules

The internal and netdev vport remain part of openvswitch.ko. Encap
vports including vxlan, gre, and geneve can be built as separate
modules and are loaded on demand. Modules can be unloaded after use.
Datapath ports keep a reference to the vport module during their
lifetime.

Allows to remove the error prone maintenance of the global list
vport_ops_list.

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 70 insertions, 32 deletions

diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index 6015802..8168ef0 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -28,6 +28,7 @@
 #include <linux/rtnetlink.h>
 #include <linux/compat.h>
 #include <net/net_namespace.h>
+#include <linux/module.h>
 
 #include "datapath.h"
 #include "vport.h"
@@ -36,22 +37,7 @@
 static void ovs_vport_record_error(struct vport *,
 				   enum vport_err_type err_type);
 
-/* List of statically compiled vport implementations.  Don't forget to also
- * add yours to the list at the bottom of vport.h. */
-static const struct vport_ops *vport_ops_list[] = {
-	&ovs_netdev_vport_ops,
-	&ovs_internal_vport_ops,
-
-#ifdef CONFIG_OPENVSWITCH_GRE
-	&ovs_gre_vport_ops,
-#endif
-#ifdef CONFIG_OPENVSWITCH_VXLAN
-	&ovs_vxlan_vport_ops,
-#endif
-#ifdef CONFIG_OPENVSWITCH_GENEVE
-	&ovs_geneve_vport_ops,
-#endif
-};
+static LIST_HEAD(vport_ops_list);
 
 /* Protected by RCU read lock for reading, ovs_mutex for writing. */
 static struct hlist_head *dev_table;
@@ -88,6 +74,32 @@ static struct hlist_head *hash_bucket(struct net *net, const char *name)
 	return &dev_table[hash & (VPORT_HASH_BUCKETS - 1)];
 }
 
+int ovs_vport_ops_register(struct vport_ops *ops)
+{
+	int err = -EEXIST;
+	struct vport_ops *o;
+
+	ovs_lock();
+	list_for_each_entry(o, &vport_ops_list, list)
+		if (ops->type == o->type)
+			goto errout;
+
+	list_add_tail(&ops->list, &vport_ops_list);
+	err = 0;
+errout:
+	ovs_unlock();
+	return err;
+}
+EXPORT_SYMBOL(ovs_vport_ops_register);
+
+void ovs_vport_ops_unregister(struct vport_ops *ops)
+{
+	ovs_lock();
+	list_del(&ops->list);
+	ovs_unlock();
+}
+EXPORT_SYMBOL(ovs_vport_ops_unregister);
+
 /**
  *	ovs_vport_locate - find a port that has already been created
  *
@@ -153,6 +165,7 @@ struct vport *ovs_vport_alloc(int priv_size, const struct vport_ops *ops,
 
 	return vport;
 }
+EXPORT_SYMBOL(ovs_vport_alloc);
 
 /**
  *	ovs_vport_free - uninitialize and free vport
@@ -173,6 +186,18 @@ void ovs_vport_free(struct vport *vport)
 	free_percpu(vport->percpu_stats);
 	kfree(vport);
 }
+EXPORT_SYMBOL(ovs_vport_free);
+
+static struct vport_ops *ovs_vport_lookup(const struct vport_parms *parms)
+{
+	struct vport_ops *ops;
+
+	list_for_each_entry(ops, &vport_ops_list, list)
+		if (ops->type == parms->type)
+			return ops;
+
+	return NULL;
+}
 
 /**
  *	ovs_vport_add - add vport device (for kernel callers)
@@ -184,31 +209,40 @@ void ovs_vport_free(struct vport *vport)
  */
 struct vport *ovs_vport_add(const struct vport_parms *parms)
 {
+	struct vport_ops *ops;
 	struct vport *vport;
-	int err = 0;
-	int i;
 
-	for (i = 0; i < ARRAY_SIZE(vport_ops_list); i++) {
-		if (vport_ops_list[i]->type == parms->type) {
-			struct hlist_head *bucket;
+	ops = ovs_vport_lookup(parms);
+	if (ops) {
+		struct hlist_head *bucket;
 
-			vport = vport_ops_list[i]->create(parms);
-			if (IS_ERR(vport)) {
-				err = PTR_ERR(vport);
-				goto out;
-			}
+		if (!try_module_get(ops->owner))
+			return ERR_PTR(-EAFNOSUPPORT);
 
-			bucket = hash_bucket(ovs_dp_get_net(vport->dp),
-					     vport->ops->get_name(vport));
-			hlist_add_head_rcu(&vport->hash_node, bucket);
+		vport = ops->create(parms);
+		if (IS_ERR(vport)) {
+			module_put(ops->owner);
 			return vport;
 		}
+
+		bucket = hash_bucket(ovs_dp_get_net(vport->dp),
+				     vport->ops->get_name(vport));
+		hlist_add_head_rcu(&vport->hash_node, bucket);
+		return vport;
 	}
 
-	err = -EAFNOSUPPORT;
+	/* Unlock to attempt module load and return -EAGAIN if load
+	 * was successful as we need to restart the port addition
+	 * workflow.
+	 */
+	ovs_unlock();
+	request_module("vport-type-%d", parms->type);
+	ovs_lock();
 
-out:
-	return ERR_PTR(err);
+	if (!ovs_vport_lookup(parms))
+		return ERR_PTR(-EAFNOSUPPORT);
+	else
+		return ERR_PTR(-EAGAIN);
 }
 
 /**
@@ -242,6 +276,8 @@ void ovs_vport_del(struct vport *vport)
 	hlist_del_rcu(&vport->hash_node);
 
 	vport->ops->destroy(vport);
+
+	module_put(vport->ops->owner);
 }
 
 /**
@@ -457,6 +493,7 @@ void ovs_vport_receive(struct vport *vport, struct sk_buff *skb,
 	}
 	ovs_dp_process_packet(skb, &key);
 }
+EXPORT_SYMBOL(ovs_vport_receive);
 
 /**
  *	ovs_vport_send - send a packet on a device
@@ -535,3 +572,4 @@ void ovs_vport_deferred_free(struct vport *vport)
 
 	call_rcu(&vport->rcu, free_vport_rcu);
 }
+EXPORT_SYMBOL(ovs_vport_deferred_free);