net: better IFF_XMIT_DST_RELEASE support

Testing xmit_more support with netperf and connected UDP sockets,
I found strange dst refcount false sharing.

Current handling of IFF_XMIT_DST_RELEASE is not optimal.

Dropping dst in validate_xmit_skb() is certainly too late in case
packet was queued by cpu X but dequeued by cpu Y

The logical point to take care of drop/force is in __dev_queue_xmit()
before even taking qdisc lock.

As Julian Anastasov pointed out, need for skb_dst() might come from some
packet schedulers or classifiers.

This patch adds new helper to cleanly express needs of various drivers
or qdiscs/classifiers.

Drivers that need skb_dst() in their ndo_start_xmit() should call
following helper in their setup instead of the prior :

	dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
->
	netif_keep_dst(dev);

Instead of using a single bit, we use two bits, one being
eventually rebuilt in bonding/team drivers.

The other one, is permanent and blocks IFF_XMIT_DST_RELEASE being
rebuilt in bonding/team. Eventually, we could add something
smarter later.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 files changed, 23 insertions, 16 deletions

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 13e6e0431592..58b5aa3b6f2d 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -1364,7 +1364,7 @@ void ipoib_setup(struct net_device *dev)
 	dev->tx_queue_len	 = ipoib_sendq_size * 2;
 	dev->features		 = (NETIF_F_VLAN_CHALLENGED	|
 				    NETIF_F_HIGHDMA);
-	dev->priv_flags		&= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 
 	memcpy(dev->broadcast, ipv4_bcast_addr, INFINIBAND_ALEN);
 
diff --git a/drivers/net/appletalk/ipddp.c b/drivers/net/appletalk/ipddp.c
index 10d0dba572c2..e90c6a7333d7 100644
--- a/drivers/net/appletalk/ipddp.c
+++ b/drivers/net/appletalk/ipddp.c
@@ -74,7 +74,7 @@ static struct net_device * __init ipddp_init(void)
 	if (!dev)
 		return ERR_PTR(-ENOMEM);
 
-	dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 	strcpy(dev->name, "ipddp%d");
 
 	if (version_printed++ == 0)
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 3ad5413d4f57..c9ac06cfe6b7 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -1002,7 +1002,8 @@ static netdev_features_t bond_fix_features(struct net_device *dev,
 
 static void bond_compute_features(struct bonding *bond)
 {
-	unsigned int flags, dst_release_flag = IFF_XMIT_DST_RELEASE;
+	unsigned int dst_release_flag = IFF_XMIT_DST_RELEASE |
+					IFF_XMIT_DST_RELEASE_PERM;
 	netdev_features_t vlan_features = BOND_VLAN_FEATURES;
 	netdev_features_t enc_features  = BOND_ENC_FEATURES;
 	struct net_device *bond_dev = bond->dev;
@@ -1038,8 +1039,10 @@ done:
 	bond_dev->gso_max_segs = gso_max_segs;
 	netif_set_gso_max_size(bond_dev, gso_max_size);
 
-	flags = bond_dev->priv_flags & ~IFF_XMIT_DST_RELEASE;
-	bond_dev->priv_flags = flags | dst_release_flag;
+	bond_dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	if ((bond_dev->priv_flags & IFF_XMIT_DST_RELEASE_PERM) &&
+	    dst_release_flag == (IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM))
+		bond_dev->priv_flags |= IFF_XMIT_DST_RELEASE;
 
 	netdev_change_features(bond_dev);
 }
diff --git a/drivers/net/eql.c b/drivers/net/eql.c
index 957e5c0cede3..a10ad74cc8d2 100644
--- a/drivers/net/eql.c
+++ b/drivers/net/eql.c
@@ -199,7 +199,7 @@ static void __init eql_setup(struct net_device *dev)
 
 	dev->type       	= ARPHRD_SLIP;
 	dev->tx_queue_len 	= 5;		/* Hands them off fast */
-	dev->priv_flags	       &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 }
 
 static int eql_open(struct net_device *dev)
diff --git a/drivers/net/ifb.c b/drivers/net/ifb.c
index d2d4a3d2237f..34f846b4bd05 100644
--- a/drivers/net/ifb.c
+++ b/drivers/net/ifb.c
@@ -185,7 +185,8 @@ static void ifb_setup(struct net_device *dev)
 
 	dev->flags |= IFF_NOARP;
 	dev->flags &= ~IFF_MULTICAST;
-	dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_TX_SKB_SHARING);
+	dev->priv_flags &= ~IFF_TX_SKB_SHARING;
+	netif_keep_dst(dev);
 	eth_hw_addr_random(dev);
 }
 
diff --git a/drivers/net/loopback.c b/drivers/net/loopback.c
index 8f2262540561..c76283c2f84a 100644
--- a/drivers/net/loopback.c
+++ b/drivers/net/loopback.c
@@ -169,7 +169,7 @@ static void loopback_setup(struct net_device *dev)
 	dev->type		= ARPHRD_LOOPBACK;	/* 0x0001*/
 	dev->flags		= IFF_LOOPBACK;
 	dev->priv_flags		|= IFF_LIVE_ADDR_CHANGE;
-	dev->priv_flags	       &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 	dev->hw_features	= NETIF_F_ALL_TSO | NETIF_F_UFO;
 	dev->features 		= NETIF_F_SG | NETIF_F_FRAGLIST
 		| NETIF_F_ALL_TSO
diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
index e8a453f1b458..38b4fae61f04 100644
--- a/drivers/net/macvlan.c
+++ b/drivers/net/macvlan.c
@@ -1025,7 +1025,8 @@ void macvlan_common_setup(struct net_device *dev)
 {
 	ether_setup(dev);
 
-	dev->priv_flags	       &= ~(IFF_XMIT_DST_RELEASE | IFF_TX_SKB_SHARING);
+	dev->priv_flags	       &= ~IFF_TX_SKB_SHARING;
+	netif_keep_dst(dev);
 	dev->priv_flags	       |= IFF_UNICAST_FLT;
 	dev->netdev_ops		= &macvlan_netdev_ops;
 	dev->destructor		= free_netdev;
diff --git a/drivers/net/ppp/ppp_generic.c b/drivers/net/ppp/ppp_generic.c
index fa0d71727894..80e6f3430f65 100644
--- a/drivers/net/ppp/ppp_generic.c
+++ b/drivers/net/ppp/ppp_generic.c
@@ -1103,7 +1103,7 @@ static void ppp_setup(struct net_device *dev)
 	dev->type = ARPHRD_PPP;
 	dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
 	dev->features |= NETIF_F_NETNS_LOCAL;
-	dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 }
 
 /*
diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c
index 2277c3679a51..a94a9df3e6bd 100644
--- a/drivers/net/team/team.c
+++ b/drivers/net/team/team.c
@@ -970,7 +970,8 @@ static void __team_compute_features(struct team *team)
 	struct team_port *port;
 	u32 vlan_features = TEAM_VLAN_FEATURES & NETIF_F_ALL_FOR_ALL;
 	unsigned short max_hard_header_len = ETH_HLEN;
-	unsigned int flags, dst_release_flag = IFF_XMIT_DST_RELEASE;
+	unsigned int dst_release_flag = IFF_XMIT_DST_RELEASE |
+					IFF_XMIT_DST_RELEASE_PERM;
 
 	list_for_each_entry(port, &team->port_list, list) {
 		vlan_features = netdev_increment_features(vlan_features,
@@ -985,8 +986,9 @@ static void __team_compute_features(struct team *team)
 	team->dev->vlan_features = vlan_features;
 	team->dev->hard_header_len = max_hard_header_len;
 
-	flags = team->dev->priv_flags & ~IFF_XMIT_DST_RELEASE;
-	team->dev->priv_flags = flags | dst_release_flag;
+	team->dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	if (dst_release_flag == (IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM))
+		team->dev->priv_flags |= IFF_XMIT_DST_RELEASE;
 
 	netdev_change_features(team->dev);
 }
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 2af795d6ba05..2a51e6e48e1e 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -2193,7 +2193,7 @@ static void vxlan_setup(struct net_device *dev)
 	dev->hw_features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_RXCSUM;
 	dev->hw_features |= NETIF_F_GSO_SOFTWARE;
 	dev->hw_features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
-	dev->priv_flags	&= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 	dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 
 	INIT_LIST_HEAD(&vxlan->next);
diff --git a/drivers/net/wan/hdlc_fr.c b/drivers/net/wan/hdlc_fr.c
index e5c7e6165a4b..3ebed1c40abb 100644
--- a/drivers/net/wan/hdlc_fr.c
+++ b/drivers/net/wan/hdlc_fr.c
@@ -1047,7 +1047,7 @@ static void pvc_setup(struct net_device *dev)
 	dev->flags = IFF_POINTOPOINT;
 	dev->hard_header_len = 10;
 	dev->addr_len = 2;
-	dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(dev);
 }
 
 static const struct net_device_ops pvc_ops = {
diff --git a/drivers/s390/net/qeth_l3_main.c b/drivers/s390/net/qeth_l3_main.c
index f8427a2c4840..afebb9709763 100644
--- a/drivers/s390/net/qeth_l3_main.c
+++ b/drivers/s390/net/qeth_l3_main.c
@@ -3306,7 +3306,7 @@ static int qeth_l3_setup_netdev(struct qeth_card *card)
 	card->dev->features |=	NETIF_F_HW_VLAN_CTAG_TX |
 				NETIF_F_HW_VLAN_CTAG_RX |
 				NETIF_F_HW_VLAN_CTAG_FILTER;
-	card->dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
+	netif_keep_dst(card->dev);
 	card->dev->gso_max_size = 15 * PAGE_SIZE;
 
 	SET_NETDEV_DEV(card->dev, &card->gdev->dev);