nagai/kernel
3
0
Fork 0
forked from mirrors/linux
Code Pull requests Activity

ipip: percpu stats accounting

Browse source 
Maintain per_cpu tx_bytes, tx_packets, rx_bytes, rx_packets.

Other seldom used fields are kept in netdev->stats structure, possibly
unsafe.

This is a preliminary work to support lockless transmit path, and
correct RX stats, that are already unsafe.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Eric Dumazet 2010-09-27 00:35:50 +00:00 committed by David S. Miller
parent e985aad723
commit 3c97af99a5
1 changed files with 93 additions and 34 deletions
repo.diff.show_diff_stats Download patch file Download diff file Expand all files Collapse all files

127
net/ipv4/ipip.c
View file

@ -131,8 +131,9 @@ struct ipip_net {
struct net_device *fb_tunnel_dev; struct net_device *fb_tunnel_dev;
}; };
static void ipip_tunnel_init(struct net_device *dev); static int ipip_tunnel_init(struct net_device *dev);
static void ipip_tunnel_setup(struct net_device *dev); static void ipip_tunnel_setup(struct net_device *dev);
static void ipip_dev_free(struct net_device *dev);
/* /*
* Locking : hash tables are protected by RCU and RTNL * Locking : hash tables are protected by RCU and RTNL
@ -141,6 +142,34 @@ static void ipip_tunnel_setup(struct net_device *dev);
#define for_each_ip_tunnel_rcu(start) \ #define for_each_ip_tunnel_rcu(start) \
for (t = rcu_dereference(start); t; t = rcu_dereference(t->next)) for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
/* often modified stats are per cpu, other are shared (netdev->stats) */
struct pcpu_tstats {
unsigned long rx_packets;
unsigned long rx_bytes;
unsigned long tx_packets;
unsigned long tx_bytes;
};
static struct net_device_stats *ipip_get_stats(struct net_device *dev)
{
struct pcpu_tstats sum = { 0 };
int i;
for_each_possible_cpu(i) {
const struct pcpu_tstats *tstats = per_cpu_ptr(dev->tstats, i);
sum.rx_packets += tstats->rx_packets;
sum.rx_bytes += tstats->rx_bytes;
sum.tx_packets += tstats->tx_packets;
sum.tx_bytes += tstats->tx_bytes;
}
dev->stats.rx_packets = sum.rx_packets;
dev->stats.rx_bytes = sum.rx_bytes;
dev->stats.tx_packets = sum.tx_packets;
dev->stats.tx_bytes = sum.tx_bytes;
return &dev->stats;
}
static struct ip_tunnel * ipip_tunnel_lookup(struct net *net, static struct ip_tunnel * ipip_tunnel_lookup(struct net *net,
__be32 remote, __be32 local) __be32 remote, __be32 local)
{ {
@ -239,7 +268,7 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
if (parms->name[0]) if (parms->name[0])
strlcpy(name, parms->name, IFNAMSIZ); strlcpy(name, parms->name, IFNAMSIZ);
else else
sprintf(name, "tunl%%d"); strcpy(name, "tunl%d");
dev = alloc_netdev(sizeof(*t), name, ipip_tunnel_setup); dev = alloc_netdev(sizeof(*t), name, ipip_tunnel_setup);
if (dev == NULL) if (dev == NULL)
@ -255,7 +284,8 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
nt = netdev_priv(dev); nt = netdev_priv(dev);
nt->parms = *parms; nt->parms = *parms;
ipip_tunnel_init(dev); if (ipip_tunnel_init(dev) < 0)
goto failed_free;
if (register_netdevice(dev) < 0) if (register_netdevice(dev) < 0)
goto failed_free; goto failed_free;
@ -265,7 +295,7 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
return nt; return nt;
failed_free: failed_free:
free_netdev(dev); ipip_dev_free(dev);
return NULL; return NULL;
} }
@ -359,8 +389,10 @@ static int ipip_rcv(struct sk_buff *skb)
const struct iphdr *iph = ip_hdr(skb); const struct iphdr *iph = ip_hdr(skb);
rcu_read_lock(); rcu_read_lock();
if ((tunnel = ipip_tunnel_lookup(dev_net(skb->dev), tunnel = ipip_tunnel_lookup(dev_net(skb->dev), iph->saddr, iph->daddr);
iph->saddr, iph->daddr)) != NULL) { if (tunnel != NULL) {
struct pcpu_tstats *tstats;
if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) { if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
rcu_read_unlock(); rcu_read_unlock();
kfree_skb(skb); kfree_skb(skb);
@ -374,7 +406,11 @@ static int ipip_rcv(struct sk_buff *skb)
skb->protocol = htons(ETH_P_IP); skb->protocol = htons(ETH_P_IP);
skb->pkt_type = PACKET_HOST; skb->pkt_type = PACKET_HOST;
skb_tunnel_rx(skb, tunnel->dev); tstats = this_cpu_ptr(tunnel->dev->tstats);
tstats->rx_packets++;
tstats->rx_bytes += skb->len;
__skb_tunnel_rx(skb, tunnel->dev);
ipip_ecn_decapsulate(iph, skb); ipip_ecn_decapsulate(iph, skb);
@ -397,13 +433,12 @@ static int ipip_rcv(struct sk_buff *skb)
static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev) static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
{ {
struct ip_tunnel *tunnel = netdev_priv(dev); struct ip_tunnel *tunnel = netdev_priv(dev);
struct net_device_stats *stats = &dev->stats; struct pcpu_tstats *tstats;
struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
struct iphdr *tiph = &tunnel->parms.iph; struct iphdr *tiph = &tunnel->parms.iph;
u8 tos = tunnel->parms.iph.tos; u8 tos = tunnel->parms.iph.tos;
__be16 df = tiph->frag_off; __be16 df = tiph->frag_off;
struct rtable *rt; /* Route to the other host */ struct rtable *rt; /* Route to the other host */
struct net_device *tdev; /* Device to other host */ struct net_device *tdev; /* Device to other host */
struct iphdr *old_iph = ip_hdr(skb); struct iphdr *old_iph = ip_hdr(skb);
struct iphdr *iph; /* Our new IP header */ struct iphdr *iph; /* Our new IP header */
unsigned int max_headroom; /* The extra header space needed */ unsigned int max_headroom; /* The extra header space needed */
@ -413,13 +448,13 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
if (skb->protocol != htons(ETH_P_IP)) if (skb->protocol != htons(ETH_P_IP))
goto tx_error; goto tx_error;
if (tos&1) if (tos & 1)
tos = old_iph->tos; tos = old_iph->tos;
if (!dst) { if (!dst) {
/* NBMA tunnel */ /* NBMA tunnel */
if ((rt = skb_rtable(skb)) == NULL) { if ((rt = skb_rtable(skb)) == NULL) {
stats->tx_fifo_errors++; dev->stats.tx_fifo_errors++;
goto tx_error; goto tx_error;
} }
if ((dst = rt->rt_gateway) == 0) if ((dst = rt->rt_gateway) == 0)
@ -427,14 +462,20 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
} }
{ {
struct flowi fl = { .oif = tunnel->parms.link, struct flowi fl = {
.nl_u = { .ip4_u = .oif = tunnel->parms.link,
{ .daddr = dst, .nl_u = {
.saddr = tiph->saddr, .ip4_u = {
.tos = RT_TOS(tos) } }, .daddr = dst,
.proto = IPPROTO_IPIP }; .saddr = tiph->saddr,
.tos = RT_TOS(tos)
}
},
.proto = IPPROTO_IPIP
};
if (ip_route_output_key(dev_net(dev), &rt, &fl)) { if (ip_route_output_key(dev_net(dev), &rt, &fl)) {
stats->tx_carrier_errors++; dev->stats.tx_carrier_errors++;
goto tx_error_icmp; goto tx_error_icmp;
} }
} }
@ -442,7 +483,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
if (tdev == dev) { if (tdev == dev) {
ip_rt_put(rt); ip_rt_put(rt);
stats->collisions++; dev->stats.collisions++;
goto tx_error; goto tx_error;
} }
@ -452,7 +493,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
mtu = dst_mtu(&rt->dst) - sizeof(struct iphdr); mtu = dst_mtu(&rt->dst) - sizeof(struct iphdr);
if (mtu < 68) { if (mtu < 68) {
stats->collisions++; dev->stats.collisions++;
ip_rt_put(rt); ip_rt_put(rt);
goto tx_error; goto tx_error;
} }
@ -488,7 +529,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
struct sk_buff *new_skb = skb_realloc_headroom(skb, max_headroom); struct sk_buff *new_skb = skb_realloc_headroom(skb, max_headroom);
if (!new_skb) { if (!new_skb) {
ip_rt_put(rt); ip_rt_put(rt);
txq->tx_dropped++; dev->stats.tx_dropped++;
dev_kfree_skb(skb); dev_kfree_skb(skb);
return NETDEV_TX_OK; return NETDEV_TX_OK;
} }
@ -525,14 +566,14 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
iph->ttl = old_iph->ttl; iph->ttl = old_iph->ttl;
nf_reset(skb); nf_reset(skb);
tstats = this_cpu_ptr(dev->tstats);
IPTUNNEL_XMIT(); __IPTUNNEL_XMIT(tstats, &dev->stats);
return NETDEV_TX_OK; return NETDEV_TX_OK;
tx_error_icmp: tx_error_icmp:
dst_link_failure(skb); dst_link_failure(skb);
tx_error: tx_error:
stats->tx_errors++; dev->stats.tx_errors++;
dev_kfree_skb(skb); dev_kfree_skb(skb);
return NETDEV_TX_OK; return NETDEV_TX_OK;
} }
@ -547,13 +588,19 @@ static void ipip_tunnel_bind_dev(struct net_device *dev)
iph = &tunnel->parms.iph; iph = &tunnel->parms.iph;
if (iph->daddr) { if (iph->daddr) {
struct flowi fl = { .oif = tunnel->parms.link, struct flowi fl = {
.nl_u = { .ip4_u = .oif = tunnel->parms.link,
{ .daddr = iph->daddr, .nl_u = {
.saddr = iph->saddr, .ip4_u = {
.tos = RT_TOS(iph->tos) } }, .daddr = iph->daddr,
.proto = IPPROTO_IPIP }; .saddr = iph->saddr,
.tos = RT_TOS(iph->tos)
}
},
.proto = IPPROTO_IPIP
};
struct rtable *rt; struct rtable *rt;
if (!ip_route_output_key(dev_net(dev), &rt, &fl)) { if (!ip_route_output_key(dev_net(dev), &rt, &fl)) {
tdev = rt->dst.dev; tdev = rt->dst.dev;
ip_rt_put(rt); ip_rt_put(rt);
@ -699,13 +746,19 @@ static const struct net_device_ops ipip_netdev_ops = {
.ndo_start_xmit = ipip_tunnel_xmit, .ndo_start_xmit = ipip_tunnel_xmit,
.ndo_do_ioctl = ipip_tunnel_ioctl, .ndo_do_ioctl = ipip_tunnel_ioctl,
.ndo_change_mtu = ipip_tunnel_change_mtu, .ndo_change_mtu = ipip_tunnel_change_mtu,
.ndo_get_stats = ipip_get_stats,
}; };
static void ipip_dev_free(struct net_device *dev)
{
free_percpu(dev->tstats);
free_netdev(dev);
}
static void ipip_tunnel_setup(struct net_device *dev) static void ipip_tunnel_setup(struct net_device *dev)
{ {
dev->netdev_ops = &ipip_netdev_ops; dev->netdev_ops = &ipip_netdev_ops;
dev->destructor = free_netdev; dev->destructor = ipip_dev_free;
dev->type = ARPHRD_TUNNEL; dev->type = ARPHRD_TUNNEL;
dev->hard_header_len = LL_MAX_HEADER + sizeof(struct iphdr); dev->hard_header_len = LL_MAX_HEADER + sizeof(struct iphdr);
@ -717,7 +770,7 @@ static void ipip_tunnel_setup(struct net_device *dev)
dev->priv_flags &= ~IFF_XMIT_DST_RELEASE; dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
} }
static void ipip_tunnel_init(struct net_device *dev) static int ipip_tunnel_init(struct net_device *dev)
{ {
struct ip_tunnel *tunnel = netdev_priv(dev); struct ip_tunnel *tunnel = netdev_priv(dev);
@ -728,6 +781,12 @@ static void ipip_tunnel_init(struct net_device *dev)
memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4); memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4);
ipip_tunnel_bind_dev(dev); ipip_tunnel_bind_dev(dev);
dev->tstats = alloc_percpu(struct pcpu_tstats);
if (!dev->tstats)
return -ENOMEM;
return 0;
} }
static void __net_init ipip_fb_tunnel_init(struct net_device *dev) static void __net_init ipip_fb_tunnel_init(struct net_device *dev)