group memnodemap and memnodeshift in a memnode structure From: Eric Dumazet pfn_to_page() and others need to access both memnode_shift and the very first bytes of memnodemap[]. If we force memnode_shift to be just before the memnodemap array, we can reduce the memory footprint to one cache line instead of two for most setups. This patch introduce a 'memnode' structure where shift and map[] are carefully placed. Signed-off-by: Eric Dumazet Signed-off-by: Andi Kleen --- arch/x86_64/mm/numa.c | 6 ++---- include/asm-x86_64/mmzone.h | 9 +++++++-- 2 files changed, 9 insertions(+), 6 deletions(-) Index: linux/include/asm-x86_64/mmzone.h =================================================================== --- linux.orig/include/asm-x86_64/mmzone.h +++ linux/include/asm-x86_64/mmzone.h @@ -15,8 +15,13 @@ #define NODEMAPSIZE 0xfff /* Simple perfect hash to map physical addresses to node numbers */ -extern int memnode_shift; -extern u8 memnodemap[NODEMAPSIZE]; +struct memnode { + int shift; + u8 map[NODEMAPSIZE]; +} ____cacheline_aligned; +extern struct memnode memnode; +#define memnode_shift memnode.shift +#define memnodemap memnode.map extern struct pglist_data *node_data[]; Index: linux/arch/x86_64/mm/numa.c =================================================================== --- linux.orig/arch/x86_64/mm/numa.c +++ linux/arch/x86_64/mm/numa.c @@ -25,8 +25,7 @@ struct pglist_data *node_data[MAX_NUMNODES] __read_mostly; bootmem_data_t plat_node_bdata[MAX_NUMNODES]; -int memnode_shift; -u8 memnodemap[NODEMAPSIZE]; +struct memnode memnode; unsigned char cpu_to_node[NR_CPUS] __read_mostly = { [0 ... NR_CPUS-1] = NUMA_NO_NODE @@ -390,8 +389,7 @@ void __init init_cpu_to_node(void) EXPORT_SYMBOL(cpu_to_node); EXPORT_SYMBOL(node_to_cpumask); -EXPORT_SYMBOL(memnode_shift); -EXPORT_SYMBOL(memnodemap); +EXPORT_SYMBOL(memnode); EXPORT_SYMBOL(node_data); #ifdef CONFIG_DISCONTIGMEM