Re: [PATCH] mm/memcg: return -EBUSY when oom-kill-disable modified and memcg use_hierarchy, has children

[Wanpeng Li <liwp.linux@gmail.com>]

On Mon, Jul 09, 2012 at 01:43:23PM +0900, Kamezawa Hiroyuki wrote:
>(2012/07/05 19:55), Wanpeng Li wrote:
>> From: Wanpeng Li <liwp@linux.vnet.ibm.com>
>> 
>> When oom-kill-disable modified by the user and current memcg use_hierarchy,
>> the change can occur, provided the current memcg has no children. If it
>> has children, return -EBUSY is enough.
>> 
>> Signed-off-by: Wanpeng Li <liwp.linux@gmail.com>
>
>I'm sorry what is the point ? You think -EBUSY should be returned in this case 
>rather than -EINVAl ? Then, why ?

just like in function cmem_cgroup_hierarchy_write:

if((!parent_memcg || !parent_memcg->use_hierarchy) &&
		(val == 1 || val == 0) {
		if (list_empty(&cont->children))
			memcg->use_hierarchy = val;
		else
			return -EBUSY;
} else
		return = -EINVAL;

If memcg->use_hierarchy && has children memcg, the user can try again
if children memcg disappear. Or I miss something ....

Regards,
Wanpeng Li

>
>
>> ---
>>   mm/memcontrol.c |    7 +++++--
>>   1 files changed, 5 insertions(+), 2 deletions(-)
>> 
>> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
>> index 63e36e7..4b64fe0 100644
>> --- a/mm/memcontrol.c
>> +++ b/mm/memcontrol.c
>> @@ -4521,11 +4521,14 @@ static int mem_cgroup_oom_control_write(struct cgroup *cgrp,
>>   
>>   	cgroup_lock();
>>   	/* oom-kill-disable is a flag for subhierarchy. */
>> -	if ((parent->use_hierarchy) ||
>> -	    (memcg->use_hierarchy && !list_empty(&cgrp->children))) {
>> +	if (parent->use_hierarchy) {
>>   		cgroup_unlock();
>>   		return -EINVAL;
>> +	} else if (memcg->use_hierarchy && !list_empty(&cgrp->children)) {
>> +		cgroup_unlock();
>> +		return -EBUSY;
>>   	}
>> +
>>   	memcg->oom_kill_disable = val;
>>   	if (!val)
>>   		memcg_oom_recover(memcg);
>> 
>
>