Re: [PATCH] memcg: fix memcg_cache_name() to use cgroup_name()

[Glauber Costa <glommer@parallels.com>]

On 03/27/2013 08:27 PM, Michal Hocko wrote:
> On Wed 27-03-13 09:21:02, Tejun Heo wrote:
>> On Wed, Mar 27, 2013 at 9:19 AM, Michal Hocko <mhocko@suse.cz> wrote:
>>>> Maybe the name could signify it's part of memcg?
>>>
>>> kmem_ prefix is used for all CONFIG_MEMCG_KMEM functions. I understand
>>> it clashes with sl?b naming but this is out of scope of this patch IMO.
>>
>> Oh, it's not using kmemcg? I see. Maybe we can rename later.
> 
> Some parts use memcg_kmem_* other kmem_. A cleanup would be nice.
> Glauber?
> 
I have been using kmem_ only in functions that will deal directly with
the slab caches and with the single purpose of operating them.

kmem_cache_destroy_work_func => worker interface to kmem_cache_destroy
kmem_cache_destroy_memcg_children => cache destructor iterator
kmem_cache_dup => interface to kmem_cache_create

All the other functions start with memcg_
Analogously, all slab-side functions that deal with memcg *ends* with
_memcg. except the functions that are only there to operate memcg data
structures:

memcg_update_all_caches.

In general, those functions could very well live in the other file (slab
or memcg), but they need to take locks or manipulate data structures
that are internal to slab/memcg.

I believe this is a sound convention.