Linux-libre 4.2-gnu

3 files changed, 129 insertions, 96 deletions

diff --git a/drivers/net/ipvlan/ipvlan.h b/drivers/net/ipvlan/ipvlan.h
index 54549a622..9542b7bac 100644
--- a/drivers/net/ipvlan/ipvlan.h
+++ b/drivers/net/ipvlan/ipvlan.h
@@ -39,6 +39,8 @@
 #define IPVLAN_MAC_FILTER_SIZE	(1 << IPVLAN_MAC_FILTER_BITS)
 #define IPVLAN_MAC_FILTER_MASK	(IPVLAN_MAC_FILTER_SIZE - 1)
 
+#define IPVLAN_QBACKLOG_LIMIT	1000
+
 typedef enum {
 	IPVL_IPV6 = 0,
 	IPVL_ICMPV6,
@@ -65,8 +67,6 @@ struct ipvl_dev {
 	struct ipvl_port	*port;
 	struct net_device	*phy_dev;
 	struct list_head	addrs;
-	int			ipv4cnt;
-	int			ipv6cnt;
 	struct ipvl_pcpu_stats	__percpu *pcpu_stats;
 	DECLARE_BITMAP(mac_filters, IPVLAN_MAC_FILTER_SIZE);
 	netdev_features_t	sfeatures;
@@ -93,6 +93,8 @@ struct ipvl_port {
 	struct hlist_head	hlhead[IPVLAN_HASH_SIZE];
 	struct list_head	ipvlans;
 	struct rcu_head		rcu;
+	struct work_struct	wq;
+	struct sk_buff_head	backlog;
 	int			count;
 	u16			mode;
 };
@@ -102,6 +104,11 @@ static inline struct ipvl_port *ipvlan_port_get_rcu(const struct net_device *d)
 	return rcu_dereference(d->rx_handler_data);
 }
 
+static inline struct ipvl_port *ipvlan_port_get_rcu_bh(const struct net_device *d)
+{
+	return rcu_dereference_bh(d->rx_handler_data);
+}
+
 static inline struct ipvl_port *ipvlan_port_get_rtnl(const struct net_device *d)
 {
 	return rtnl_dereference(d->rx_handler_data);
@@ -112,6 +119,7 @@ void ipvlan_set_port_mode(struct ipvl_port *port, u32 nval);
 void ipvlan_init_secret(void);
 unsigned int ipvlan_mac_hash(const unsigned char *addr);
 rx_handler_result_t ipvlan_handle_frame(struct sk_buff **pskb);
+void ipvlan_process_multicast(struct work_struct *work);
 int ipvlan_queue_xmit(struct sk_buff *skb, struct net_device *dev);
 void ipvlan_ht_addr_add(struct ipvl_dev *ipvlan, struct ipvl_addr *addr);
 struct ipvl_addr *ipvlan_find_addr(const struct ipvl_dev *ipvlan,
@@ -119,5 +127,5 @@ struct ipvl_addr *ipvlan_find_addr(const struct ipvl_dev *ipvlan,
 bool ipvlan_addr_busy(struct ipvl_port *port, void *iaddr, bool is_v6);
 struct ipvl_addr *ipvlan_ht_addr_lookup(const struct ipvl_port *port,
 					const void *iaddr, bool is_v6);
-void ipvlan_ht_addr_del(struct ipvl_addr *addr, bool sync);
+void ipvlan_ht_addr_del(struct ipvl_addr *addr);
 #endif /* __IPVLAN_H */
diff --git a/drivers/net/ipvlan/ipvlan_core.c b/drivers/net/ipvlan/ipvlan_core.c
index c30b5c300..207f62e8d 100644
--- a/drivers/net/ipvlan/ipvlan_core.c
+++ b/drivers/net/ipvlan/ipvlan_core.c
@@ -85,11 +85,9 @@ void ipvlan_ht_addr_add(struct ipvl_dev *ipvlan, struct ipvl_addr *addr)
 		hlist_add_head_rcu(&addr->hlnode, &port->hlhead[hash]);
 }
 
-void ipvlan_ht_addr_del(struct ipvl_addr *addr, bool sync)
+void ipvlan_ht_addr_del(struct ipvl_addr *addr)
 {
 	hlist_del_init_rcu(&addr->hlnode);
-	if (sync)
-		synchronize_rcu();
 }
 
 struct ipvl_addr *ipvlan_find_addr(const struct ipvl_dev *ipvlan,
@@ -189,62 +187,69 @@ unsigned int ipvlan_mac_hash(const unsigned char *addr)
 	return hash & IPVLAN_MAC_FILTER_MASK;
 }
 
-static void ipvlan_multicast_frame(struct ipvl_port *port, struct sk_buff *skb,
-				   const struct ipvl_dev *in_dev, bool local)
+void ipvlan_process_multicast(struct work_struct *work)
 {
-	struct ethhdr *eth = eth_hdr(skb);
+	struct ipvl_port *port = container_of(work, struct ipvl_port, wq);
+	struct ethhdr *ethh;
 	struct ipvl_dev *ipvlan;
-	struct sk_buff *nskb;
+	struct sk_buff *skb, *nskb;
+	struct sk_buff_head list;
 	unsigned int len;
 	unsigned int mac_hash;
 	int ret;
+	u8 pkt_type;
+	bool hlocal, dlocal;
 
-	if (skb->protocol == htons(ETH_P_PAUSE))
-		return;
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(ipvlan, &port->ipvlans, pnode) {
-		if (local && (ipvlan == in_dev))
-			continue;
+	__skb_queue_head_init(&list);
 
-		mac_hash = ipvlan_mac_hash(eth->h_dest);
-		if (!test_bit(mac_hash, ipvlan->mac_filters))
-			continue;
+	spin_lock_bh(&port->backlog.lock);
+	skb_queue_splice_tail_init(&port->backlog, &list);
+	spin_unlock_bh(&port->backlog.lock);
 
-		ret = NET_RX_DROP;
-		len = skb->len + ETH_HLEN;
-		nskb = skb_clone(skb, GFP_ATOMIC);
-		if (!nskb)
-			goto mcast_acct;
+	while ((skb = __skb_dequeue(&list)) != NULL) {
+		ethh = eth_hdr(skb);
+		hlocal = ether_addr_equal(ethh->h_source, port->dev->dev_addr);
+		mac_hash = ipvlan_mac_hash(ethh->h_dest);
 
-		if (ether_addr_equal(eth->h_dest, ipvlan->phy_dev->broadcast))
-			nskb->pkt_type = PACKET_BROADCAST;
+		if (ether_addr_equal(ethh->h_dest, port->dev->broadcast))
+			pkt_type = PACKET_BROADCAST;
 		else
-			nskb->pkt_type = PACKET_MULTICAST;
-
-		nskb->dev = ipvlan->dev;
-		if (local)
-			ret = dev_forward_skb(ipvlan->dev, nskb);
-		else
-			ret = netif_rx(nskb);
-mcast_acct:
-		ipvlan_count_rx(ipvlan, len, ret == NET_RX_SUCCESS, true);
-	}
-	rcu_read_unlock();
-
-	/* Locally generated? ...Forward a copy to the main-device as
-	 * well. On the RX side we'll ignore it (wont give it to any
-	 * of the virtual devices.
-	 */
-	if (local) {
-		nskb = skb_clone(skb, GFP_ATOMIC);
-		if (nskb) {
-			if (ether_addr_equal(eth->h_dest, port->dev->broadcast))
-				nskb->pkt_type = PACKET_BROADCAST;
+			pkt_type = PACKET_MULTICAST;
+
+		dlocal = false;
+		rcu_read_lock();
+		list_for_each_entry_rcu(ipvlan, &port->ipvlans, pnode) {
+			if (hlocal && (ipvlan->dev == skb->dev)) {
+				dlocal = true;
+				continue;
+			}
+			if (!test_bit(mac_hash, ipvlan->mac_filters))
+				continue;
+
+			ret = NET_RX_DROP;
+			len = skb->len + ETH_HLEN;
+			nskb = skb_clone(skb, GFP_ATOMIC);
+			if (!nskb)
+				goto acct;
+
+			nskb->pkt_type = pkt_type;
+			nskb->dev = ipvlan->dev;
+			if (hlocal)
+				ret = dev_forward_skb(ipvlan->dev, nskb);
 			else
-				nskb->pkt_type = PACKET_MULTICAST;
-
-			dev_forward_skb(port->dev, nskb);
+				ret = netif_rx(nskb);
+acct:
+			ipvlan_count_rx(ipvlan, len, ret == NET_RX_SUCCESS, true);
+		}
+		rcu_read_unlock();
+
+		if (dlocal) {
+			/* If the packet originated here, send it out. */
+			skb->dev = port->dev;
+			skb->pkt_type = pkt_type;
+			dev_queue_xmit(skb);
+		} else {
+			kfree_skb(skb);
 		}
 	}
 }
@@ -446,6 +451,26 @@ out:
 	return ret;
 }
 
+static void ipvlan_multicast_enqueue(struct ipvl_port *port,
+				     struct sk_buff *skb)
+{
+	if (skb->protocol == htons(ETH_P_PAUSE)) {
+		kfree_skb(skb);
+		return;
+	}
+
+	spin_lock(&port->backlog.lock);
+	if (skb_queue_len(&port->backlog) < IPVLAN_QBACKLOG_LIMIT) {
+		__skb_queue_tail(&port->backlog, skb);
+		spin_unlock(&port->backlog.lock);
+		schedule_work(&port->wq);
+	} else {
+		spin_unlock(&port->backlog.lock);
+		atomic_long_inc(&skb->dev->rx_dropped);
+		kfree_skb(skb);
+	}
+}
+
 static int ipvlan_xmit_mode_l3(struct sk_buff *skb, struct net_device *dev)
 {
 	const struct ipvl_dev *ipvlan = netdev_priv(dev);
@@ -493,11 +518,8 @@ static int ipvlan_xmit_mode_l2(struct sk_buff *skb, struct net_device *dev)
 		return dev_forward_skb(ipvlan->phy_dev, skb);
 
 	} else if (is_multicast_ether_addr(eth->h_dest)) {
-		u8 ip_summed = skb->ip_summed;
-
-		skb->ip_summed = CHECKSUM_UNNECESSARY;
-		ipvlan_multicast_frame(ipvlan->port, skb, ipvlan, true);
-		skb->ip_summed = ip_summed;
+		ipvlan_multicast_enqueue(ipvlan->port, skb);
+		return NET_XMIT_SUCCESS;
 	}
 
 	skb->dev = ipvlan->phy_dev;
@@ -507,7 +529,7 @@ static int ipvlan_xmit_mode_l2(struct sk_buff *skb, struct net_device *dev)
 int ipvlan_queue_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
-	struct ipvl_port *port = ipvlan_port_get_rcu(ipvlan->phy_dev);
+	struct ipvl_port *port = ipvlan_port_get_rcu_bh(ipvlan->phy_dev);
 
 	if (!port)
 		goto out;
@@ -581,8 +603,18 @@ static rx_handler_result_t ipvlan_handle_mode_l2(struct sk_buff **pskb,
 	int addr_type;
 
 	if (is_multicast_ether_addr(eth->h_dest)) {
-		if (ipvlan_external_frame(skb, port))
-			ipvlan_multicast_frame(port, skb, NULL, false);
+		if (ipvlan_external_frame(skb, port)) {
+			struct sk_buff *nskb = skb_clone(skb, GFP_ATOMIC);
+
+			/* External frames are queued for device local
+			 * distribution, but a copy is given to master
+			 * straight away to avoid sending duplicates later
+			 * when work-queue processes this frame. This is
+			 * achieved by returning RX_HANDLER_PASS.
+			 */
+			if (nskb)
+				ipvlan_multicast_enqueue(port, nskb);
+		}
 	} else {
 		struct ipvl_addr *addr;
 
diff --git a/drivers/net/ipvlan/ipvlan_main.c b/drivers/net/ipvlan/ipvlan_main.c
index 77b92a0fe..20b58bdec 100644
--- a/drivers/net/ipvlan/ipvlan_main.c
+++ b/drivers/net/ipvlan/ipvlan_main.c
@@ -54,6 +54,9 @@ static int ipvlan_port_create(struct net_device *dev)
 	for (idx = 0; idx < IPVLAN_HASH_SIZE; idx++)
 		INIT_HLIST_HEAD(&port->hlhead[idx]);
 
+	skb_queue_head_init(&port->backlog);
+	INIT_WORK(&port->wq, ipvlan_process_multicast);
+
 	err = netdev_rx_handler_register(dev, ipvlan_handle_frame, port);
 	if (err)
 		goto err;
@@ -72,6 +75,8 @@ static void ipvlan_port_destroy(struct net_device *dev)
 
 	dev->priv_flags &= ~IFF_IPVLAN_MASTER;
 	netdev_rx_handler_unregister(dev);
+	cancel_work_sync(&port->wq);
+	__skb_queue_purge(&port->backlog);
 	kfree_rcu(port, rcu);
 }
 
@@ -148,10 +153,9 @@ static int ipvlan_open(struct net_device *dev)
 	else
 		dev->flags &= ~IFF_NOARP;
 
-	if (ipvlan->ipv6cnt > 0 || ipvlan->ipv4cnt > 0) {
-		list_for_each_entry(addr, &ipvlan->addrs, anode)
-			ipvlan_ht_addr_add(ipvlan, addr);
-	}
+	list_for_each_entry(addr, &ipvlan->addrs, anode)
+		ipvlan_ht_addr_add(ipvlan, addr);
+
 	return dev_uc_add(phy_dev, phy_dev->dev_addr);
 }
 
@@ -166,10 +170,9 @@ static int ipvlan_stop(struct net_device *dev)
 
 	dev_uc_del(phy_dev, phy_dev->dev_addr);
 
-	if (ipvlan->ipv6cnt > 0 || ipvlan->ipv4cnt > 0) {
-		list_for_each_entry(addr, &ipvlan->addrs, anode)
-			ipvlan_ht_addr_del(addr, !dev->dismantle);
-	}
+	list_for_each_entry(addr, &ipvlan->addrs, anode)
+		ipvlan_ht_addr_del(addr);
+
 	return 0;
 }
 
@@ -213,17 +216,6 @@ static void ipvlan_change_rx_flags(struct net_device *dev, int change)
 		dev_set_allmulti(phy_dev, dev->flags & IFF_ALLMULTI? 1 : -1);
 }
 
-static void ipvlan_set_broadcast_mac_filter(struct ipvl_dev *ipvlan, bool set)
-{
-	struct net_device *dev = ipvlan->dev;
-	unsigned int hashbit = ipvlan_mac_hash(dev->broadcast);
-
-	if (set && !test_bit(hashbit, ipvlan->mac_filters))
-		__set_bit(hashbit, ipvlan->mac_filters);
-	else if (!set && test_bit(hashbit, ipvlan->mac_filters))
-		__clear_bit(hashbit, ipvlan->mac_filters);
-}
-
 static void ipvlan_set_multicast_mac_filter(struct net_device *dev)
 {
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
@@ -238,6 +230,12 @@ static void ipvlan_set_multicast_mac_filter(struct net_device *dev)
 		netdev_for_each_mc_addr(ha, dev)
 			__set_bit(ipvlan_mac_hash(ha->addr), mc_filters);
 
+		/* Turn-on broadcast bit irrespective of address family,
+		 * since broadcast is deferred to a work-queue, hence no
+		 * impact on fast-path processing.
+		 */
+		__set_bit(ipvlan_mac_hash(dev->broadcast), mc_filters);
+
 		bitmap_copy(ipvlan->mac_filters, mc_filters,
 			    IPVLAN_MAC_FILTER_SIZE);
 	}
@@ -471,8 +469,6 @@ static int ipvlan_link_new(struct net *src_net, struct net_device *dev,
 	ipvlan->port = port;
 	ipvlan->sfeatures = IPVLAN_FEATURES;
 	INIT_LIST_HEAD(&ipvlan->addrs);
-	ipvlan->ipv4cnt = 0;
-	ipvlan->ipv6cnt = 0;
 
 	/* TODO Probably put random address here to be presented to the
 	 * world but keep using the physical-dev address for the outgoing
@@ -508,12 +504,12 @@ static void ipvlan_link_delete(struct net_device *dev, struct list_head *head)
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
 	struct ipvl_addr *addr, *next;
 
-	if (ipvlan->ipv6cnt > 0 || ipvlan->ipv4cnt > 0) {
-		list_for_each_entry_safe(addr, next, &ipvlan->addrs, anode) {
-			ipvlan_ht_addr_del(addr, !dev->dismantle);
-			list_del(&addr->anode);
-		}
+	list_for_each_entry_safe(addr, next, &ipvlan->addrs, anode) {
+		ipvlan_ht_addr_del(addr);
+		list_del(&addr->anode);
+		kfree_rcu(addr, rcu);
 	}
+
 	list_del_rcu(&ipvlan->pnode);
 	unregister_netdevice_queue(dev, head);
 	netdev_upper_dev_unlink(ipvlan->phy_dev, dev);
@@ -627,7 +623,7 @@ static int ipvlan_add_addr6(struct ipvl_dev *ipvlan, struct in6_addr *ip6_addr)
 	memcpy(&addr->ip6addr, ip6_addr, sizeof(struct in6_addr));
 	addr->atype = IPVL_IPV6;
 	list_add_tail(&addr->anode, &ipvlan->addrs);
-	ipvlan->ipv6cnt++;
+
 	/* If the interface is not up, the address will be added to the hash
 	 * list by ipvlan_open.
 	 */
@@ -645,10 +641,8 @@ static void ipvlan_del_addr6(struct ipvl_dev *ipvlan, struct in6_addr *ip6_addr)
 	if (!addr)
 		return;
 
-	ipvlan_ht_addr_del(addr, true);
+	ipvlan_ht_addr_del(addr);
 	list_del(&addr->anode);
-	ipvlan->ipv6cnt--;
-	WARN_ON(ipvlan->ipv6cnt < 0);
 	kfree_rcu(addr, rcu);
 
 	return;
@@ -661,6 +655,10 @@ static int ipvlan_addr6_event(struct notifier_block *unused,
 	struct net_device *dev = (struct net_device *)if6->idev->dev;
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
 
+	/* FIXME IPv6 autoconf calls us from bh without RTNL */
+	if (in_softirq())
+		return NOTIFY_DONE;
+
 	if (!netif_is_ipvlan(dev))
 		return NOTIFY_DONE;
 
@@ -699,13 +697,12 @@ static int ipvlan_add_addr4(struct ipvl_dev *ipvlan, struct in_addr *ip4_addr)
 	memcpy(&addr->ip4addr, ip4_addr, sizeof(struct in_addr));
 	addr->atype = IPVL_IPV4;
 	list_add_tail(&addr->anode, &ipvlan->addrs);
-	ipvlan->ipv4cnt++;
+
 	/* If the interface is not up, the address will be added to the hash
 	 * list by ipvlan_open.
 	 */
 	if (netif_running(ipvlan->dev))
 		ipvlan_ht_addr_add(ipvlan, addr);
-	ipvlan_set_broadcast_mac_filter(ipvlan, true);
 
 	return 0;
 }
@@ -718,12 +715,8 @@ static void ipvlan_del_addr4(struct ipvl_dev *ipvlan, struct in_addr *ip4_addr)
 	if (!addr)
 		return;
 
-	ipvlan_ht_addr_del(addr, true);
+	ipvlan_ht_addr_del(addr);
 	list_del(&addr->anode);
-	ipvlan->ipv4cnt--;
-	WARN_ON(ipvlan->ipv4cnt < 0);
-	if (!ipvlan->ipv4cnt)
-	    ipvlan_set_broadcast_mac_filter(ipvlan, false);
 	kfree_rcu(addr, rcu);
 
 	return;