Re: [PATCH 2/3] stop_machine: dequeue work before signal completion

[Namhyung Kim <namhyung@kernel.org>]

Hi Tejun and Hillf,

On Wed, 6 Feb 2013 10:47:49 -0800, Tejun Heo wrote:
> On Wed, Feb 06, 2013 at 08:38:43PM +0800, Hillf Danton wrote:
>> As handled by the kernel thread, work is dequeued first for further actions.
>
> Ditto as the previous patch.
>
>> Signed-off-by: Hillf Danton <dhillf@gmail.com>
>> ---
>> 
>> --- a/kernel/stop_machine.c	Wed Feb  6 19:57:12 2013
>> +++ b/kernel/stop_machine.c	Wed Feb  6 20:02:12 2013
>> @@ -334,23 +334,24 @@ static int __cpuinit cpu_stop_cpu_callba
>>  #ifdef CONFIG_HOTPLUG_CPU
>>  	case CPU_UP_CANCELED:
>>  	case CPU_POST_DEAD:
>> -	{
>> -		struct cpu_stop_work *work;
>> -
>>  		sched_set_stop_task(cpu, NULL);
>>  		/* kill the stopper */
>>  		kthread_stop(stopper->thread);
>>  		/* drain remaining works */
>>  		spin_lock_irq(&stopper->lock);
>> -		list_for_each_entry(work, &stopper->works, list)
>> +		while (!list_empty(&stopper->works)) {
>> +			struct cpu_stop_work *work;
>> +			work = list_first_entry(&stopper->works,
>> +					struct cpu_stop_work, list);
>> +			list_del_init(&work->list);
>>  			cpu_stop_signal_done(work->done, false, 0);
>> +		}
>>  		stopper->enabled = false;
>>  		spin_unlock_irq(&stopper->lock);
>
> Why does this matter?  It's inside spinlock.  What's being made better
> by this change?

IIUC the work should be deleted from the list, otherwise it'd trigger
BUG_ON when the cpu gets online again.

Thanks,
Namhyung