[Devel] [PATCH rh7 2/3] ve: Wait till ve's attaching work finished

Kirill Tkhai ktkhai at odin.com
Wed Oct 28 07:37:05 PDT 2015


On 28.10.2015 15:51, Vladimir Davydov wrote:
> On Wed, Oct 28, 2015 at 12:54:10PM +0300, Kirill Tkhai wrote:
> ...
>>>> +static void ve_wait_work(struct callback_head *head)
>>>> +{
>>>> +	atomic_t *nr = &current->task_ve_nr_works;
>>>> +
>>>> +	while (atomic_read(nr))
>>>> +		schedule_timeout(1);
>>>
>>> Let's not introduce such temporary solutions. They look misleading and
>>> don't ease review. I think it'd be OK to merge patches 2 and 3.
>>
>> It's not a new interface, it's just a stub. Could you please explain,
>> how it mislead a reviewer?
> 
> You don't mention it's a stub anywhere. People usually don't start
> reviewing a patch set from the end, you know. I do sometimes though :-)

I thought, if two patches go one after one, it's seen. Yeah, maybe
better I wrote "it's a stub" in description.

>>
>> The thing is I splited patches 2 and 3 after I saw the big one is difficult
>> to review.
> 
> Try simplifying patch 3 then?

I don't think it's complicated. It follows the idea from patch 1/3 and 2/3,
and it seems it's extremely simple.
 
>>  
>>>> +
>>>> +	init_task_work(&current->task_ve_work, NULL);
>>>>  }
>>>>  
>>>>  static int ve_can_attach(struct cgroup *cg, struct cgroup_taskset *tset)
>>



More information about the Devel mailing list