On 8/29/19 2:50 PM, Joao Martins wrote:
> On 8/29/19 12:56 PM, Marcelo Tosatti wrote:
>> Hi Joao,
>>
>> On Wed, Aug 28, 2019 at 07:56:50PM +0100, Joao Martins wrote:
>>> +static void haltpoll_uninit(void)
>>> +{
>>> + unsigned int cpu;
>>> +
>>> + cpus_read_lock();
>>> +
>>> + for_each_online_cpu(cpu) {
>>> + struct cpuidle_device *dev =
>>> + per_cpu_ptr(haltpoll_cpuidle_devices, cpu);
>>> +
>>> + if (!dev->registered)
>>> + continue;
>>> +
>>> + arch_haltpoll_disable(cpu);
>>> + cpuidle_unregister_device(dev);
>>> + }
>>
>> 1)
>>
>>> +
>>> + cpuidle_unregister(&haltpoll_driver);
>>
>> cpuidle_unregister_driver.
>
> Will fix -- this was an oversight.
>
>>
>>> + free_percpu(haltpoll_cpuidle_devices);
>>> + haltpoll_cpuidle_devices = NULL;
>>> +
>>> + cpus_read_unlock();
>>
>> Any reason you can't cpus_read_unlock() at 1) ?
>>
> No, let me adjust that too.
>
>> Looks good otherwise.
>>
BTW, should I take this as a Acked-by, Reviewed-by, or neither? :)
Joao
