]> pilppa.com Git - linux-2.6-omap-h63xx.git/commitdiff
[IPSEC] flow: Remove an unnecessary ____cacheline_aligned
authorEric Dumazet <dada1@cosmosbay.com>
Fri, 8 Feb 2008 02:03:18 +0000 (18:03 -0800)
committerDavid S. Miller <davem@davemloft.net>
Fri, 8 Feb 2008 02:03:18 +0000 (18:03 -0800)
We use a percpu variable named flow_hash_info, which holds 12 bytes.

It is currently marked as ____cacheline_aligned, which makes linker
skip space to properly align this variable.

Before :
c065cc90 D per_cpu__softnet_data
c065cd00 d per_cpu__flow_tables
<Here, hole of 124 bytes>
c065cd80 d per_cpu__flow_hash_info
<Here, hole of 116 bytes>
c065ce00 d per_cpu__flow_flush_tasklets
c065ce14 d per_cpu__rt_cache_stat

This alignement is quite unproductive, and removing it reduces the
size of percpu data (by 240 bytes on my x86 machine), and improves
performance (flow_tables & flow_hash_info can share a single cache
line)

After patch :
c065cc04 D per_cpu__softnet_data
c065cc4c d per_cpu__flow_tables
c065cc50 d per_cpu__flow_hash_info
c065cc5c d per_cpu__flow_flush_tasklets
c065cc70 d per_cpu__rt_cache_stat

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/core/flow.c

index 46b38e06e0d75601bde3fbef5dbf3048008bc263..9cfe84571ca55554284ac617478333f2d3ba26d4 100644 (file)
@@ -52,7 +52,7 @@ struct flow_percpu_info {
        int hash_rnd_recalc;
        u32 hash_rnd;
        int count;
-} ____cacheline_aligned;
+};
 static DEFINE_PER_CPU(struct flow_percpu_info, flow_hash_info) = { 0 };
 
 #define flow_hash_rnd_recalc(cpu) \