[PATCH 1/4] cgroup freezer: Fix buggy resume test for tasks frozen with cgroup freezer

[Li Zefan]

Serge E. Hallyn wrote:
> Quoting Matt Helsley (matthltc at us.ibm.com):
>> -int cgroup_frozen(struct task_struct *task)
>> +int cgroup_freezing_or_frozen(struct task_struct *task)
>>  {
>>  	struct freezer *freezer;
>>  	enum freezer_state state;
>>
>>  	task_lock(task);
>>  	freezer = task_freezer(task);
>> -	state = freezer->state;
>> +	if (!freezer->css.cgroup->parent)
>> +		state = CGROUP_THAWED; /* root cgroup can't be frozen */
>> +	else
>> +		state = freezer->state;
> 
> Why do you have to add this special-casing now?  I thought
> we were already preventing freezing the root cgroup, so
> freezer->state should be sane in that case anyway?
> 

Yeah, IIRC top_freezer->state is alwasy CGROUP_THAWED, so this
check is redundant.

>>  	task_unlock(task);
>>
>> -	return state == CGROUP_FROZEN;
>> +	return (state == CGROUP_FREEZING) || (state == CGROUP_FROZEN);
>>  }