IPVS: init and cleanup restructuring

DESCRIPTION
This patch tries to restore the initial init and cleanup
sequences that was before namspace patch.
Netns also requires action when net devices unregister
which has never been implemented. I.e this patch also
covers when a device moves into a network namespace,
and has to be released.

IMPLEMENTATION
The number of calls to register_pernet_device have been
reduced to one for the ip_vs.ko
Schedulers still have their own calls.

This patch adds a function __ip_vs_service_cleanup()
and an enable flag for the netfilter hooks.

The nf hooks will be enabled when the first service is loaded
and never disabled again, except when a namespace exit starts.

Signed-off-by: Hans Schillstrom <[e-mail]>
Acked-by: Julian Anastasov <[e-mail]>
[[e-mail]: minor edit to changelog]
Signed-off-by: Simon Horman <[e-mail]>

include/net/ip_vs.h

···
 /* IPVS in network namespace */
 struct netns_ipvs {
 	int			gen;		/* Generation */
+	int			enable;		/* enable like nf_hooks do */
 	/*
 	 *	Hash table: for real service lookups
 	 */
···
 	atomic_inc(&ctl_cp->n_control);
 }
 
+/*
+ * IPVS netns init & cleanup functions
+ */
+extern int __ip_vs_estimator_init(struct net *net);
+extern int __ip_vs_control_init(struct net *net);
+extern int __ip_vs_protocol_init(struct net *net);
+extern int __ip_vs_app_init(struct net *net);
+extern int __ip_vs_conn_init(struct net *net);
+extern int __ip_vs_sync_init(struct net *net);
+extern void __ip_vs_conn_cleanup(struct net *net);
+extern void __ip_vs_app_cleanup(struct net *net);
+extern void __ip_vs_protocol_cleanup(struct net *net);
+extern void __ip_vs_control_cleanup(struct net *net);
+extern void __ip_vs_estimator_cleanup(struct net *net);
+extern void __ip_vs_sync_cleanup(struct net *net);
+extern void __ip_vs_service_cleanup(struct net *net);
 
 /*
  *      IPVS application functions

net/netfilter/ipvs/ip_vs_app.c

···
 };
 #endif
 
-static int __net_init __ip_vs_app_init(struct net *net)
+int __net_init __ip_vs_app_init(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 
···
 	return 0;
 }
 
-static void __net_exit __ip_vs_app_cleanup(struct net *net)
+void __net_exit __ip_vs_app_cleanup(struct net *net)
 {
 	proc_net_remove(net, "ip_vs_app");
 }
 
-static struct pernet_operations ip_vs_app_ops = {
-	.init = __ip_vs_app_init,
-	.exit = __ip_vs_app_cleanup,
-};
-
 int __init ip_vs_app_init(void)
 {
-	int rv;
-
-	rv = register_pernet_subsys(&ip_vs_app_ops);
-	return rv;
+	return 0;
 }
 
 
 void ip_vs_app_cleanup(void)
 {
-	unregister_pernet_subsys(&ip_vs_app_ops);
 }

net/netfilter/ipvs/ip_vs_conn.c

···
 	return 0;
 }
 
-static void __net_exit __ip_vs_conn_cleanup(struct net *net)
+void __net_exit __ip_vs_conn_cleanup(struct net *net)
 {
 	/* flush all the connection entries first */
 	ip_vs_conn_flush(net);
 	proc_net_remove(net, "ip_vs_conn");
 	proc_net_remove(net, "ip_vs_conn_sync");
 }
-static struct pernet_operations ipvs_conn_ops = {
-	.init = __ip_vs_conn_init,
-	.exit = __ip_vs_conn_cleanup,
-};
 
 int __init ip_vs_conn_init(void)
 {
 	int idx;
-	int retc;
 
 	/* Compute size and mask */
 	ip_vs_conn_tab_size = 1 << ip_vs_conn_tab_bits;
···
 		rwlock_init(&__ip_vs_conntbl_lock_array[idx].l);
 	}
 
-	retc = register_pernet_subsys(&ipvs_conn_ops);
-
 	/* calculate the random value for connection hash */
 	get_random_bytes(&ip_vs_conn_rnd, sizeof(ip_vs_conn_rnd));
 
-	return retc;
+	return 0;
 }
 
 void ip_vs_conn_cleanup(void)
 {
-	unregister_pernet_subsys(&ipvs_conn_ops);
 	/* Release the empty cache */
 	kmem_cache_destroy(ip_vs_conn_cachep);
 	vfree(ip_vs_conn_tab);

net/netfilter/ipvs/ip_vs_core.c

···
 		return NF_ACCEPT;
 
 	net = skb_net(skb);
+	if (!net_ipvs(net)->enable)
+		return NF_ACCEPT;
+
 	ip_vs_fill_iphdr(af, skb_network_header(skb), &iph);
 #ifdef CONFIG_IP_VS_IPV6
 	if (af == AF_INET6) {
···
 		return NF_ACCEPT; /* The packet looks wrong, ignore */
 
 	net = skb_net(skb);
+
 	pd = ip_vs_proto_data_get(net, cih->protocol);
 	if (!pd)
 		return NF_ACCEPT;
···
 			      IP_VS_DBG_ADDR(af, &iph.daddr), hooknum);
 		return NF_ACCEPT;
 	}
+	/* ipvs enabled in this netns ? */
+	net = skb_net(skb);
+	if (!net_ipvs(net)->enable)
+		return NF_ACCEPT;
+
 	ip_vs_fill_iphdr(af, skb_network_header(skb), &iph);
 
 	/* Bad... Do not break raw sockets */
···
 			ip_vs_fill_iphdr(af, skb_network_header(skb), &iph);
 		}
 
-	net = skb_net(skb);
 	/* Protocol supported? */
 	pd = ip_vs_proto_data_get(net, iph.protocol);
 	if (unlikely(!pd))
···
 	}
 
 	IP_VS_DBG_PKT(11, af, pp, skb, 0, "Incoming packet");
-	net = skb_net(skb);
 	ipvs = net_ipvs(net);
 	/* Check the server status */
 	if (cp->dest && !(cp->dest->flags & IP_VS_DEST_F_AVAILABLE)) {
···
 		   int (*okfn)(struct sk_buff *))
 {
 	int r;
+	struct net *net;
 
 	if (ip_hdr(skb)->protocol != IPPROTO_ICMP)
+		return NF_ACCEPT;
+
+	/* ipvs enabled in this netns ? */
+	net = skb_net(skb);
+	if (!net_ipvs(net)->enable)
 		return NF_ACCEPT;
 
 	return ip_vs_in_icmp(skb, &r, hooknum);
···
 		      int (*okfn)(struct sk_buff *))
 {
 	int r;
+	struct net *net;
 
 	if (ipv6_hdr(skb)->nexthdr != IPPROTO_ICMPV6)
+		return NF_ACCEPT;
+
+	/* ipvs enabled in this netns ? */
+	net = skb_net(skb);
+	if (!net_ipvs(net)->enable)
 		return NF_ACCEPT;
 
 	return ip_vs_in_icmp_v6(skb, &r, hooknum);
···
 		pr_err("%s(): no memory.\n", __func__);
 		return -ENOMEM;
 	}
+	/* Hold the beast until a service is registerd */
+	ipvs->enable = 0;
 	ipvs->net = net;
 	/* Counters used for creating unique names */
 	ipvs->gen = atomic_read(&ipvs_netns_cnt);
 	atomic_inc(&ipvs_netns_cnt);
 	net->ipvs = ipvs;
+
+	if (__ip_vs_estimator_init(net) < 0)
+		goto estimator_fail;
+
+	if (__ip_vs_control_init(net) < 0)
+		goto control_fail;
+
+	if (__ip_vs_protocol_init(net) < 0)
+		goto protocol_fail;
+
+	if (__ip_vs_app_init(net) < 0)
+		goto app_fail;
+
+	if (__ip_vs_conn_init(net) < 0)
+		goto conn_fail;
+
+	if (__ip_vs_sync_init(net) < 0)
+		goto sync_fail;
+
 	printk(KERN_INFO "IPVS: Creating netns size=%zu id=%d\n",
 			 sizeof(struct netns_ipvs), ipvs->gen);
 	return 0;
+/*
+ * Error handling
+ */
+
+sync_fail:
+	__ip_vs_conn_cleanup(net);
+conn_fail:
+	__ip_vs_app_cleanup(net);
+app_fail:
+	__ip_vs_protocol_cleanup(net);
+protocol_fail:
+	__ip_vs_control_cleanup(net);
+control_fail:
+	__ip_vs_estimator_cleanup(net);
+estimator_fail:
+	return -ENOMEM;
 }
 
 static void __net_exit __ip_vs_cleanup(struct net *net)
 {
+	__ip_vs_service_cleanup(net);	/* ip_vs_flush() with locks */
+	__ip_vs_conn_cleanup(net);
+	__ip_vs_app_cleanup(net);
+	__ip_vs_protocol_cleanup(net);
+	__ip_vs_control_cleanup(net);
+	__ip_vs_estimator_cleanup(net);
 	IP_VS_DBG(2, "ipvs netns %d released\n", net_ipvs(net)->gen);
+}
+
+static void __net_exit __ip_vs_dev_cleanup(struct net *net)
+{
+	EnterFunction(2);
+	net_ipvs(net)->enable = 0;	/* Disable packet reception */
+	__ip_vs_sync_cleanup(net);
+	LeaveFunction(2);
 }
 
 static struct pernet_operations ipvs_core_ops = {
···
 	.size = sizeof(struct netns_ipvs),
 };
 
+static struct pernet_operations ipvs_core_dev_ops = {
+	.exit = __ip_vs_dev_cleanup,
+};
+
 /*
  *	Initialize IP Virtual Server
  */
 static int __init ip_vs_init(void)
 {
 	int ret;
-
-	ret = register_pernet_subsys(&ipvs_core_ops);	/* Alloc ip_vs struct */
-	if (ret < 0)
-		return ret;
 
 	ip_vs_estimator_init();
 	ret = ip_vs_control_init();
···
 		goto cleanup_conn;
 	}
 
+	ret = register_pernet_subsys(&ipvs_core_ops);	/* Alloc ip_vs struct */
+	if (ret < 0)
+		goto cleanup_sync;
+
+	ret = register_pernet_device(&ipvs_core_dev_ops);
+	if (ret < 0)
+		goto cleanup_sub;
+
 	ret = nf_register_hooks(ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
 	if (ret < 0) {
 		pr_err("can't register hooks.\n");
-		goto cleanup_sync;
+		goto cleanup_dev;
 	}
 
 	pr_info("ipvs loaded.\n");
+
 	return ret;
 
+cleanup_dev:
+	unregister_pernet_device(&ipvs_core_dev_ops);
+cleanup_sub:
+	unregister_pernet_subsys(&ipvs_core_ops);
 cleanup_sync:
 	ip_vs_sync_cleanup();
   cleanup_conn:
···
 	ip_vs_control_cleanup();
   cleanup_estimator:
 	ip_vs_estimator_cleanup();
-	unregister_pernet_subsys(&ipvs_core_ops);	/* free ip_vs struct */
 	return ret;
 }
 
 static void __exit ip_vs_cleanup(void)
 {
 	nf_unregister_hooks(ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
+	unregister_pernet_device(&ipvs_core_dev_ops);
+	unregister_pernet_subsys(&ipvs_core_ops);	/* free ip_vs struct */
 	ip_vs_sync_cleanup();
 	ip_vs_conn_cleanup();
 	ip_vs_app_cleanup();
 	ip_vs_protocol_cleanup();
 	ip_vs_control_cleanup();
 	ip_vs_estimator_cleanup();
-	unregister_pernet_subsys(&ipvs_core_ops);	/* free ip_vs struct */
 	pr_info("ipvs unloaded.\n");
 }
 

net/netfilter/ipvs/ip_vs_ctl.c

···
 }
 #endif
 
+
+/*  Protos */
+static void __ip_vs_del_service(struct ip_vs_service *svc);
+
+
 #ifdef CONFIG_IP_VS_IPV6
 /* Taken from rt6_fill_node() in net/ipv6/route.c, is there a better way? */
 static int __ip_vs_addr_is_local_v6(struct net *net,
···
 	write_unlock_bh(&__ip_vs_svc_lock);
 
 	*svc_p = svc;
+	/* Now there is a service - full throttle */
+	ipvs->enable = 1;
 	return 0;
 
 
···
 	return 0;
 }
 
+/*
+ *	Delete service by {netns} in the service table.
+ *	Called by __ip_vs_cleanup()
+ */
+void __ip_vs_service_cleanup(struct net *net)
+{
+	EnterFunction(2);
+	/* Check for "full" addressed entries */
+	mutex_lock(&__ip_vs_mutex);
+	ip_vs_flush(net);
+	mutex_unlock(&__ip_vs_mutex);
+	LeaveFunction(2);
+}
+/*
+ * Release dst hold by dst_cache
+ */
+static inline void
+__ip_vs_dev_reset(struct ip_vs_dest *dest, struct net_device *dev)
+{
+	spin_lock_bh(&dest->dst_lock);
+	if (dest->dst_cache && dest->dst_cache->dev == dev) {
+		IP_VS_DBG_BUF(3, "Reset dev:%s dest %s:%u ,dest->refcnt=%d\n",
+			      dev->name,
+			      IP_VS_DBG_ADDR(dest->af, &dest->addr),
+			      ntohs(dest->port),
+			      atomic_read(&dest->refcnt));
+		ip_vs_dst_reset(dest);
+	}
+	spin_unlock_bh(&dest->dst_lock);
+
+}
+/*
+ * Netdev event receiver
+ * Currently only NETDEV_UNREGISTER is handled, i.e. if we hold a reference to
+ * a device that is "unregister" it must be released.
+ */
+static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
+			    void *ptr)
+{
+	struct net_device *dev = ptr;
+	struct net *net = dev_net(dev);
+	struct ip_vs_service *svc;
+	struct ip_vs_dest *dest;
+	unsigned int idx;
+
+	if (event != NETDEV_UNREGISTER)
+		return NOTIFY_DONE;
+	IP_VS_DBG(3, "%s() dev=%s\n", __func__, dev->name);
+	EnterFunction(2);
+	mutex_lock(&__ip_vs_mutex);
+	for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
+		list_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
+			if (net_eq(svc->net, net)) {
+				list_for_each_entry(dest, &svc->destinations,
+						    n_list) {
+					__ip_vs_dev_reset(dest, dev);
+				}
+			}
+		}
+
+		list_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
+			if (net_eq(svc->net, net)) {
+				list_for_each_entry(dest, &svc->destinations,
+						    n_list) {
+					__ip_vs_dev_reset(dest, dev);
+				}
+			}
+
+		}
+	}
+
+	list_for_each_entry(dest, &net_ipvs(net)->dest_trash, n_list) {
+		__ip_vs_dev_reset(dest, dev);
+	}
+	mutex_unlock(&__ip_vs_mutex);
+	LeaveFunction(2);
+	return NOTIFY_DONE;
+}
 
 /*
  *	Zero counters in a service or all services
···
 
 #endif
 
+static struct notifier_block ip_vs_dst_notifier = {
+	.notifier_call = ip_vs_dst_event,
+};
+
 int __net_init __ip_vs_control_init(struct net *net)
 {
 	int idx;
···
 	return -ENOMEM;
 }
 
-static void __net_exit __ip_vs_control_cleanup(struct net *net)
+void __net_exit __ip_vs_control_cleanup(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 
···
 	proc_net_remove(net, "ip_vs");
 	free_percpu(ipvs->tot_stats.cpustats);
 }
-
-static struct pernet_operations ipvs_control_ops = {
-	.init = __ip_vs_control_init,
-	.exit = __ip_vs_control_cleanup,
-};
 
 int __init ip_vs_control_init(void)
 {
···
 		INIT_LIST_HEAD(&ip_vs_svc_fwm_table[idx]);
 	}
 
-	ret = register_pernet_subsys(&ipvs_control_ops);
-	if (ret) {
-		pr_err("cannot register namespace.\n");
-		goto err;
-	}
-
 	smp_wmb();	/* Do we really need it now ? */
 
 	ret = nf_register_sockopt(&ip_vs_sockopts);
 	if (ret) {
 		pr_err("cannot register sockopt.\n");
-		goto err_net;
+		goto err_sock;
 	}
 
 	ret = ip_vs_genl_register();
 	if (ret) {
 		pr_err("cannot register Generic Netlink interface.\n");
-		nf_unregister_sockopt(&ip_vs_sockopts);
-		goto err_net;
+		goto err_genl;
 	}
+
+	ret = register_netdevice_notifier(&ip_vs_dst_notifier);
+	if (ret < 0)
+		goto err_notf;
 
 	LeaveFunction(2);
 	return 0;
 
-err_net:
-	unregister_pernet_subsys(&ipvs_control_ops);
-err:
+err_notf:
+	ip_vs_genl_unregister();
+err_genl:
+	nf_unregister_sockopt(&ip_vs_sockopts);
+err_sock:
 	return ret;
 }
 
···
 void ip_vs_control_cleanup(void)
 {
 	EnterFunction(2);
-	unregister_pernet_subsys(&ipvs_control_ops);
 	ip_vs_genl_unregister();
 	nf_unregister_sockopt(&ip_vs_sockopts);
 	LeaveFunction(2);

net/netfilter/ipvs/ip_vs_est.c

···
 	dst->outbps = (e->outbps + 0xF) >> 5;
 }
 
-static int __net_init __ip_vs_estimator_init(struct net *net)
+int __net_init __ip_vs_estimator_init(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 
···
 	return 0;
 }
 
-static void __net_exit __ip_vs_estimator_exit(struct net *net)
+void __net_exit __ip_vs_estimator_cleanup(struct net *net)
 {
 	del_timer_sync(&net_ipvs(net)->est_timer);
 }
-static struct pernet_operations ip_vs_app_ops = {
-	.init = __ip_vs_estimator_init,
-	.exit = __ip_vs_estimator_exit,
-};
 
 int __init ip_vs_estimator_init(void)
 {
-	int rv;
-
-	rv = register_pernet_subsys(&ip_vs_app_ops);
-	return rv;
+	return 0;
 }
 
 void ip_vs_estimator_cleanup(void)
 {
-	unregister_pernet_subsys(&ip_vs_app_ops);
 }

net/netfilter/ipvs/ip_vs_proto.c

···
 /*
  * per network name-space init
  */
-static int __net_init __ip_vs_protocol_init(struct net *net)
+int __net_init __ip_vs_protocol_init(struct net *net)
 {
 #ifdef CONFIG_IP_VS_PROTO_TCP
 	register_ip_vs_proto_netns(net, &ip_vs_protocol_tcp);
···
 	return 0;
 }
 
-static void __net_exit __ip_vs_protocol_cleanup(struct net *net)
+void __net_exit __ip_vs_protocol_cleanup(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 	struct ip_vs_proto_data *pd;
···
 			unregister_ip_vs_proto_netns(net, pd);
 	}
 }
-
-static struct pernet_operations ipvs_proto_ops = {
-	.init = __ip_vs_protocol_init,
-	.exit = __ip_vs_protocol_cleanup,
-};
 
 int __init ip_vs_protocol_init(void)
 {
···
 	REGISTER_PROTOCOL(&ip_vs_protocol_esp);
 #endif
 	pr_info("Registered protocols (%s)\n", &protocols[2]);
-	return register_pernet_subsys(&ipvs_proto_ops);
 
 	return 0;
 }
···
 	struct ip_vs_protocol *pp;
 	int i;
 
-	unregister_pernet_subsys(&ipvs_proto_ops);
 	/* unregister all the ipvs protocols */
 	for (i = 0; i < IP_VS_PROTO_TAB_SIZE; i++) {
 		while ((pp = ip_vs_proto_table[i]) != NULL)

net/netfilter/ipvs/ip_vs_sync.c

···
 /*
  * Initialize data struct for each netns
  */
-static int __net_init __ip_vs_sync_init(struct net *net)
+int __net_init __ip_vs_sync_init(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 
···
 	return 0;
 }
 
-static void __ip_vs_sync_cleanup(struct net *net)
+void __ip_vs_sync_cleanup(struct net *net)
 {
 	int retc;
 
···
 		pr_err("Failed to stop Backup Daemon\n");
 }
 
-static struct pernet_operations ipvs_sync_ops = {
-	.init = __ip_vs_sync_init,
-	.exit = __ip_vs_sync_cleanup,
-};
-
-
 int __init ip_vs_sync_init(void)
 {
-	return register_pernet_device(&ipvs_sync_ops);
+	return 0;
 }
 
 void ip_vs_sync_cleanup(void)
 {
-	unregister_pernet_device(&ipvs_sync_ops);
 }