diff --git a/arch/powerpc/include/asm/topology.h b/arch/powerpc/include/asm/topology.h index 89e3ef2496ac72c4e03d17657833930de22f6b8e..d0b5fca6b0776fc8ef220483b91b1557c73a0835 100644 --- a/arch/powerpc/include/asm/topology.h +++ b/arch/powerpc/include/asm/topology.h @@ -22,7 +22,15 @@ struct device_node; static inline int cpu_to_node(int cpu) { - return numa_cpu_lookup_table[cpu]; + int nid; + + nid = numa_cpu_lookup_table[cpu]; + + /* + * During early boot, the numa-cpu lookup table might not have been + * setup for all CPUs yet. In such cases, default to node 0. + */ + return (nid < 0) ? 0 : nid; } #define parent_node(node) (node) diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c index 078d3e00a6168fe7d56f212c0822800ee77bc9f8..6847d509162f11b3b33bad62f1996d4a83d30549 100644 --- a/arch/powerpc/mm/numa.c +++ b/arch/powerpc/mm/numa.c @@ -31,6 +31,8 @@ #include #include #include +#include +#include #include #include #include @@ -152,9 +154,22 @@ static void __init get_node_active_region(unsigned long pfn, } } -static void map_cpu_to_node(int cpu, int node) +static void reset_numa_cpu_lookup_table(void) +{ + unsigned int cpu; + + for_each_possible_cpu(cpu) + numa_cpu_lookup_table[cpu] = -1; +} + +static void update_numa_cpu_lookup_table(unsigned int cpu, int node) { numa_cpu_lookup_table[cpu] = node; +} + +static void map_cpu_to_node(int cpu, int node) +{ + update_numa_cpu_lookup_table(cpu, node); dbg("adding cpu %d to node %d\n", cpu, node); @@ -522,11 +537,24 @@ static int of_drconf_to_nid_single(struct of_drconf_cell *drmem, */ static int numa_setup_cpu(unsigned long lcpu) { - int nid = 0; - struct device_node *cpu = of_get_cpu_node(lcpu, NULL); + int nid; + struct device_node *cpu; + + /* + * If a valid cpu-to-node mapping is already available, use it + * directly instead of querying the firmware, since it represents + * the most recent mapping notified to us by the platform (eg: VPHN). + */ + if ((nid = numa_cpu_lookup_table[lcpu]) >= 0) { + map_cpu_to_node(lcpu, nid); + return nid; + } + + cpu = of_get_cpu_node(lcpu, NULL); if (!cpu) { WARN_ON(1); + nid = 0; goto out; } @@ -1067,6 +1095,7 @@ void __init do_init_bootmem(void) */ setup_node_to_cpumask_map(); + reset_numa_cpu_lookup_table(); register_cpu_notifier(&ppc64_numa_nb); cpu_numa_callback(&ppc64_numa_nb, CPU_UP_PREPARE, (void *)(unsigned long)boot_cpuid); @@ -1445,6 +1474,33 @@ static int update_cpu_topology(void *data) return 0; } +static int update_lookup_table(void *data) +{ + struct topology_update_data *update; + + if (!data) + return -EINVAL; + + /* + * Upon topology update, the numa-cpu lookup table needs to be updated + * for all threads in the core, including offline CPUs, to ensure that + * future hotplug operations respect the cpu-to-node associativity + * properly. + */ + for (update = data; update; update = update->next) { + int nid, base, j; + + nid = update->new_nid; + base = cpu_first_thread_sibling(update->cpu); + + for (j = 0; j < threads_per_core; j++) { + update_numa_cpu_lookup_table(base + j, nid); + } + } + + return 0; +} + /* * Update the node maps and sysfs entries for each cpu whose home node * has changed. Returns 1 when the topology has changed, and 0 otherwise. @@ -1513,6 +1569,14 @@ int arch_update_cpu_topology(void) stop_machine(update_cpu_topology, &updates[0], &updated_cpus); + /* + * Update the numa-cpu lookup table with the new mappings, even for + * offline CPUs. It is best to perform this update from the stop- + * machine context. + */ + stop_machine(update_lookup_table, &updates[0], + cpumask_of(raw_smp_processor_id())); + for (ud = &updates[0]; ud; ud = ud->next) { unregister_cpu_under_node(ud->cpu, ud->old_nid); register_cpu_under_node(ud->cpu, ud->new_nid);