Converting the hotplug locking, i.e. get_online_cpus(), to a percpu rwsem unearthed a circular lock dependency which was hidden from lockdep due to the lockdep annotation of get_online_cpus() which prevents lockdep from creating full dependency chains. CPU0 CPU1 ---- ---- lock((&wfc.work)); lock(cpu_hotplug_lock.rw_sem); lock((&wfc.work)); lock(cpu_hotplug_lock.rw_sem); This dependency is established via acpi_processor_start() which calls into the work queue code. And the work queue code establishes the reverse dependency. This is not a problem of get_online_cpus() recursion, it's a possible deadlock undetected by lockdep so far. The cure is to use cpu_hotplug_disable() instead of get_online_cpus() to protect the probing from acpi_processor_start(). There is a side effect to this: cpu_hotplug_disable() makes a concurrent cpu hotplug attempt via the sysfs interfaces fail with -EBUSY, but that probing usually happens during the boot process where no interaction is possible. Any later invocations are infrequent enough and concurrent hotplug attempts are so unlikely that the danger of user space visible regressions is very close to zero. Anyway, thats preferrable over a real deadlock. Signed-off-by: Thomas Gleixner Acked-by: Rafael J. Wysocki Cc: Peter Zijlstra Cc: Sebastian Siewior Cc: Steven Rostedt Cc: linux-acpi@vger.kernel.org Cc: Len Brown --- drivers/acpi/processor_driver.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) Index: b/drivers/acpi/processor_driver.c =================================================================== --- a/drivers/acpi/processor_driver.c +++ b/drivers/acpi/processor_driver.c @@ -268,9 +268,9 @@ static int acpi_processor_start(struct d return -ENODEV; /* Protect against concurrent CPU hotplug operations */ - get_online_cpus(); + cpu_hotplug_disable(); ret = __acpi_processor_start(device); - put_online_cpus(); + cpu_hotplug_enable(); return ret; }