When one CPU is going down, and smpboot_register_percpu_thread is called, there is the race issue below: T1(CPUA): T2(CPUB): _cpu_down() smpboot_register_percpu_thread() smpboot_park_threads() ... __stop_machine() __smpboot_create_thread(CPU_Dying) [Currently, the being down CPU is online yet] take_cpu_down() smpboot_unpark_thread(CPU_Dying) __cpu_disable() .... native_cpu_disable() .... Here the new kthread will get running based on the CPU_Dying set_cpu_online(cpu, false) .... cpu_notify(CPU_DYING)
After notified the CPU_DYING, the new created kthead for dying CPU will be migrated to another CPU in migration_call(). Here we need use get_online_cpus()/put_online_cpus() when calling function smpboot_register_percpu_thread(). Signed-off-by: liu chuansheng <chuansheng....@intel.com> --- kernel/smpboot.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/kernel/smpboot.c b/kernel/smpboot.c index d6c5fc0..3fe708a 100644 --- a/kernel/smpboot.c +++ b/kernel/smpboot.c @@ -266,6 +266,7 @@ int smpboot_register_percpu_thread(struct smp_hotplug_thread *plug_thread) unsigned int cpu; int ret = 0; + get_online_cpus(); mutex_lock(&smpboot_threads_lock); for_each_online_cpu(cpu) { ret = __smpboot_create_thread(plug_thread, cpu); @@ -278,6 +279,7 @@ int smpboot_register_percpu_thread(struct smp_hotplug_thread *plug_thread) list_add(&plug_thread->list, &hotplug_threads); out: mutex_unlock(&smpboot_threads_lock); + put_online_cpus(); return ret; } EXPORT_SYMBOL_GPL(smpboot_register_percpu_thread); -- 1.7.0.4 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/