net: better pcpu data alignment

Eric Dumazet · davem330 · commit 8ce120f11898 · 2011-11-08T15:10:59.000-05:00
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: Eric Dumazet &lt;eric.dumazet@gmail.com&gt;
Signed-off-by: David S. Miller &lt;davem@davemloft.net&gt;
diff --git a/drivers/net/veth.c b/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;
diff --git a/net/ipv4/ip_gre.c b/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)
 {
diff --git a/net/ipv4/ipip.c b/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)
 {
diff --git a/net/ipv6/ip6_tunnel.c b/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)
 {
diff --git a/net/ipv6/sit.c b/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)
 {

Original file line number	Diff line number	Diff line change
`@@ -171,7 +171,7 @@ struct pcpu_tstats {`
`171`	`171`	`unsigned long rx_bytes;`
`172`	`172`	`unsigned long tx_packets;`
`173`	`173`	`unsigned long tx_bytes;`
`174`		`-};`
	`174`	`+} __attribute__((aligned(4*sizeof(unsigned long))));`
`175`	`175`
`176`	`176`	`static struct net_device_stats ipgre_get_stats(struct net_device dev)`
`177`	`177`	`{`
Original file line number	Diff line number	Diff line change
`@@ -148,7 +148,7 @@ struct pcpu_tstats {`
`148`	`148`	`unsigned long rx_bytes;`
`149`	`149`	`unsigned long tx_packets;`
`150`	`150`	`unsigned long tx_bytes;`
`151`		`-};`
	`151`	`+} __attribute__((aligned(4*sizeof(unsigned long))));`
`152`	`152`
`153`	`153`	`static struct net_device_stats ipip_get_stats(struct net_device dev)`
`154`	`154`	`{`
Original file line number	Diff line number	Diff line change
`@@ -93,7 +93,7 @@ struct pcpu_tstats {`
`93`	`93`	`unsigned long rx_bytes;`
`94`	`94`	`unsigned long tx_packets;`
`95`	`95`	`unsigned long tx_bytes;`
`96`		`-};`
	`96`	`+} __attribute__((aligned(4*sizeof(unsigned long))));`
`97`	`97`
`98`	`98`	`static struct net_device_stats ip6_get_stats(struct net_device dev)`
`99`	`99`	`{`
Original file line number	Diff line number	Diff line change
`@@ -91,7 +91,7 @@ struct pcpu_tstats {`
`91`	`91`	`unsigned long rx_bytes;`
`92`	`92`	`unsigned long tx_packets;`
`93`	`93`	`unsigned long tx_bytes;`
`94`		`-};`
	`94`	`+} __attribute__((aligned(4*sizeof(unsigned long))));`
`95`	`95`
`96`	`96`	`static struct net_device_stats ipip6_get_stats(struct net_device dev)`
`97`	`97`	`{`