[PATCH 10/17] x86-64: Move nodenumber from PDA to per-cpu.

[Brian Gerst <brgerst@gmail.com>]

Signed-off-by: Brian Gerst <brgerst@gmail.com>
---
 arch/x86/include/asm/pda.h      |    1 -
 arch/x86/include/asm/topology.h |    3 ++-
 arch/x86/kernel/cpu/common.c    |   11 ++++++-----
 arch/x86/kernel/setup_percpu.c  |    4 +++-
 4 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/pda.h b/arch/x86/include/asm/pda.h
index 4527d70..b30ef6b 100644
--- a/arch/x86/include/asm/pda.h
+++ b/arch/x86/include/asm/pda.h
@@ -22,7 +22,6 @@ struct x8664_pda {
 					/* gcc-ABI: this canary MUST be at
 					   offset 40!!! */
 #endif
-	short nodenumber;		/* number of current node (32k max) */
 	short in_bootmem;		/* pda lives in bootmem */
 	short isidle;
 } ____cacheline_aligned_in_smp;
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index 87ca3fd..90a8a15 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -83,7 +83,8 @@ extern cpumask_t *node_to_cpumask_map;
 DECLARE_EARLY_PER_CPU(int, x86_cpu_to_node_map);
 
 /* Returns the number of the current Node. */
-#define numa_node_id()		read_pda(nodenumber)
+DECLARE_PER_CPU(int, nodenumber);
+#define numa_node_id()		percpu_read(nodenumber)
 
 #ifdef CONFIG_DEBUG_PER_CPU_MAPS
 extern int cpu_to_node(int cpu);
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 62f9665..ccc6f8a 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -899,11 +899,6 @@ void __cpuinit pda_init(int cpu)
 	loadsegment(gs, 0);
 
 	load_pda_offset(cpu);
-
-	if (cpu != 0) {
-		if (pda->nodenumber == 0 && cpu_to_node(cpu) != NUMA_NO_NODE)
-			pda->nodenumber = cpu_to_node(cpu);
-	}
 }
 
 static DEFINE_PER_CPU_PAGE_ALIGNED(char, exception_stacks
@@ -973,6 +968,12 @@ void __cpuinit cpu_init(void)
 	if (cpu != 0)
 		pda_init(cpu);
 
+#ifdef CONFIG_NUMA
+	if (cpu != 0 && percpu_read(nodenumber) == 0 &&
+	    cpu_to_node(cpu) != NUMA_NO_NODE)
+		percpu_write(nodenumber, cpu_to_node(cpu));
+#endif
+
 	me = current;
 
 	if (cpumask_test_and_set_cpu(cpu, cpu_initialized_mask))
diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c
index bd38127..0ddb184 100644
--- a/arch/x86/kernel/setup_percpu.c
+++ b/arch/x86/kernel/setup_percpu.c
@@ -44,6 +44,8 @@ EXPORT_EARLY_PER_CPU_SYMBOL(x86_bios_cpu_apicid);
 
 #if defined(CONFIG_NUMA) && defined(CONFIG_X86_64)
 #define	X86_64_NUMA	1	/* (used later) */
+DEFINE_PER_CPU(int, nodenumber) = 0;
+EXPORT_PER_CPU_SYMBOL(nodenumber);
 
 /*
  * Map cpu index to node index
@@ -276,7 +278,7 @@ void __cpuinit numa_set_node(int cpu, int node)
 	per_cpu(x86_cpu_to_node_map, cpu) = node;
 
 	if (node != NUMA_NO_NODE)
-		cpu_pda(cpu)->nodenumber = node;
+		per_cpu(nodenumber, cpu) = node;
 }
 
 void __cpuinit numa_clear_node(int cpu)
-- 
1.6.1.rc1