aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorEric Dumazet <eric.dumazet@gmail.com>2011-11-30 19:00:53 +0000
committerDavid S. Miller <davem@davemloft.net>2011-12-01 13:28:54 -0500
commit84f9307c5da84a7c8513e7607dc8427d2cbd63e3 (patch)
treef34fb0525284d97e2b10870afd8e021cc45ef44c
parentdccp: Evaluate ip_hdr() only once in dccp_v4_route_skb(). (diff)
downloadlinux-dev-84f9307c5da84a7c8513e7607dc8427d2cbd63e3.tar.xz
linux-dev-84f9307c5da84a7c8513e7607dc8427d2cbd63e3.zip
ipv4: use a 64bit load/store in output path
gcc compiler is smart enough to use a single load/store if we memcpy(dptr, sptr, 8) on x86_64, regardless of CONFIG_CC_OPTIMIZE_FOR_SIZE In IP header, daddr immediately follows saddr, this wont change in the future. We only need to make sure our flowi4 (saddr,daddr) fields wont break the rule. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-rw-r--r--include/net/flow.h5
-rw-r--r--net/ipv4/ip_output.c21
2 files changed, 21 insertions, 5 deletions
diff --git a/include/net/flow.h b/include/net/flow.h
index a09447749e2d..9192d690b562 100644
--- a/include/net/flow.h
+++ b/include/net/flow.h
@@ -59,8 +59,11 @@ struct flowi4 {
#define flowi4_proto __fl_common.flowic_proto
#define flowi4_flags __fl_common.flowic_flags
#define flowi4_secid __fl_common.flowic_secid
- __be32 daddr;
+
+ /* (saddr,daddr) must be grouped, same order as in IP header */
__be32 saddr;
+ __be32 daddr;
+
union flowi_uli uli;
#define fl4_sport uli.ports.sport
#define fl4_dport uli.ports.dport
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 0bc95f3977d2..0d5e5672f3d1 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -319,6 +319,20 @@ int ip_output(struct sk_buff *skb)
!(IPCB(skb)->flags & IPSKB_REROUTED));
}
+/*
+ * copy saddr and daddr, possibly using 64bit load/stores
+ * Equivalent to :
+ * iph->saddr = fl4->saddr;
+ * iph->daddr = fl4->daddr;
+ */
+static void ip_copy_addrs(struct iphdr *iph, const struct flowi4 *fl4)
+{
+ BUILD_BUG_ON(offsetof(typeof(*fl4), daddr) !=
+ offsetof(typeof(*fl4), saddr) + sizeof(fl4->saddr));
+ memcpy(&iph->saddr, &fl4->saddr,
+ sizeof(fl4->saddr) + sizeof(fl4->daddr));
+}
+
int ip_queue_xmit(struct sk_buff *skb, struct flowi *fl)
{
struct sock *sk = skb->sk;
@@ -381,8 +395,8 @@ packet_routed:
iph->frag_off = 0;
iph->ttl = ip_select_ttl(inet, &rt->dst);
iph->protocol = sk->sk_protocol;
- iph->saddr = fl4->saddr;
- iph->daddr = fl4->daddr;
+ ip_copy_addrs(iph, fl4);
+
/* Transport layer set skb->h.foo itself. */
if (inet_opt && inet_opt->opt.optlen) {
@@ -1337,8 +1351,7 @@ struct sk_buff *__ip_make_skb(struct sock *sk,
ip_select_ident(iph, &rt->dst, sk);
iph->ttl = ttl;
iph->protocol = sk->sk_protocol;
- iph->saddr = fl4->saddr;
- iph->daddr = fl4->daddr;
+ ip_copy_addrs(iph, fl4);
if (opt) {
iph->ihl += opt->optlen>>2;