net: better pcpu data alignment

Tunnels can force an alignment of their percpu data to reduce number of
cache lines used in fast path, or read in .ndo_get_stats()

percpu_alloc() is a very fine grained allocator, so any small hole will
be used anyway.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/veth.c

@@ -27,8 +27,8 @@
 
 struct veth_net_stats {
 	u64			rx_packets;
-	u64			tx_packets;
 	u64			rx_bytes;
+	u64			tx_packets;
 	u64			tx_bytes;
 	u64			rx_dropped;
 	struct u64_stats_sync	syncp;

net/ipv4/ip_gre.c

@@ -171,7 +171,7 @@ struct pcpu_tstats {
 	unsigned long	rx_bytes;
 	unsigned long	tx_packets;
 	unsigned long	tx_bytes;
-};
+} __attribute__((aligned(4*sizeof(unsigned long))));
 
 static struct net_device_stats *ipgre_get_stats(struct net_device *dev)
 {

net/ipv4/ipip.c

@@ -148,7 +148,7 @@ struct pcpu_tstats {
 	unsigned long	rx_bytes;
 	unsigned long	tx_packets;
 	unsigned long	tx_bytes;
-};
+} __attribute__((aligned(4*sizeof(unsigned long))));
 
 static struct net_device_stats *ipip_get_stats(struct net_device *dev)
 {

net/ipv6/ip6_tunnel.c

@@ -93,7 +93,7 @@ struct pcpu_tstats {
 	unsigned long	rx_bytes;
 	unsigned long	tx_packets;
 	unsigned long	tx_bytes;
-};
+} __attribute__((aligned(4*sizeof(unsigned long))));
 
 static struct net_device_stats *ip6_get_stats(struct net_device *dev)
 {

net/ipv6/sit.c

@@ -91,7 +91,7 @@ struct pcpu_tstats {
 	unsigned long	rx_bytes;
 	unsigned long	tx_packets;
 	unsigned long	tx_bytes;
-};
+} __attribute__((aligned(4*sizeof(unsigned long))));
 
 static struct net_device_stats *ipip6_get_stats(struct net_device *dev)
 {