[PATCH] x86_64: Early initialization of cpu_to_node

Patch enables early intialization of cpu_to_node.
apicid_to_node is built by reading the SRAT table, from acpi_numa_init with
ACPI_NUMA and k8_scan_nodes with K8_NUMA.
x86_cpu_to_apicid is built by parsing the ACPI MADT table, from acpi_boot_init.
We combine these two tables and setup cpu_to_node.

Early intialization helps the static per_cpu_areas in getting pages from
correct node.

Change since last release:
Do not initialize early init_cpu_to_node for faking node cases.

Patch tested on TYAN dual core 4P board with K8 only, ACPI_NUMA.
Tested on EM64T NUMA. Also tested with numa=off, numa=fake, and  running
a kernel compiled with NUMA on a regular EM64 2 way SMP.

Signed-off-by: Alok N Kataria <alokk@calsoftinc.com>
Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Signed-off-by: Shai Fultheim <shai@scalex86.org>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

diff --git a/arch/x86_64/kernel/setup.c b/arch/x86_64/kernel/setup.c
index 46bf556efc301cb69bc5b66de65dcb8f214d5da0..173bdc55113b1f11026832cd95d2224d95325cd6 100644 (file)
--- a/arch/x86_64/kernel/setup.c
+++ b/arch/x86_64/kernel/setup.c
@@ -713,6 +713,8 @@ void __init setup_arch(char **cmdline_p)
        acpi_boot_init();
 #endif
 
+       init_cpu_to_node();
+
 #ifdef CONFIG_X86_LOCAL_APIC
        /*
         * get boot-time SMP configuration:

diff --git a/arch/x86_64/mm/numa.c b/arch/x86_64/mm/numa.c
index 42ec1f4f491a3e44ce5482033b346a2ac70f4641..876e343744128f57d10b6ea2c70165da3cf62e64 100644 (file)
--- a/arch/x86_64/mm/numa.c
+++ b/arch/x86_64/mm/numa.c
@@ -330,6 +330,31 @@ __init int numa_setup(char *opt)
        return 1;
 } 
 
+/*
+ * Setup early cpu_to_node.
+ *
+ * Populate cpu_to_node[] only if x86_cpu_to_apicid[],
+ * and apicid_to_node[] tables have valid entries for a CPU.
+ * This means we skip cpu_to_node[] initialisation for NUMA
+ * emulation and faking node case (when running a kernel compiled
+ * for NUMA on a non NUMA box), which is OK as cpu_to_node[]
+ * is already initialized in a round robin manner at numa_init_array,
+ * prior to this call, and this initialization is good enough
+ * for the fake NUMA cases.
+ */
+void __init init_cpu_to_node(void)
+{
+       int i;
+       for (i = 0; i < NR_CPUS; i++) {
+               u8 apicid = x86_cpu_to_apicid[i];
+               if (apicid == BAD_APICID)
+                       continue;
+               if (apicid_to_node[apicid] == NUMA_NO_NODE)
+                       continue;
+               cpu_to_node[i] = apicid_to_node[apicid];
+       }
+}
+
 EXPORT_SYMBOL(cpu_to_node);
 EXPORT_SYMBOL(node_to_cpumask);
 EXPORT_SYMBOL(memnode_shift);

diff --git a/include/asm-x86_64/numa.h b/include/asm-x86_64/numa.h
index d51e56fdc3da04d534a5cf618548b0956f1d7868..34e434ce3268409b7cc90f5fb1c6c925d051086c 100644 (file)
--- a/include/asm-x86_64/numa.h
+++ b/include/asm-x86_64/numa.h
@@ -20,6 +20,11 @@ extern int numa_off;
 extern void numa_set_node(int cpu, int node);
 
 extern unsigned char apicid_to_node[256];
+#ifdef CONFIG_NUMA
+extern void __init init_cpu_to_node(void);
+#else
+#define init_cpu_to_node() do {} while (0)
+#endif
 
 #define NUMA_NO_NODE 0xff