[PATCH v2 21/28] vmscan: also shrink slab in memcg pressure

Glauber Costa glommer at parallels.com
Fri Apr 5 08:25:04 UTC 2013


On 04/04/2013 01:35 PM, Sha Zhengju wrote:
> On Wed, Apr 3, 2013 at 6:43 PM, Glauber Costa <glommer at parallels.com> wrote:
>> On 04/03/2013 02:11 PM, Sha Zhengju wrote:
>>>>> +unsigned long
>>>>> +memcg_zone_reclaimable_pages(struct mem_cgroup *memcg, struct zone *zone)
>>>>> +{
>>>>> +       int nid = zone_to_nid(zone);
>>>>> +       int zid = zone_idx(zone);
>>>>> +       unsigned long val;
>>>>> +
>>>>> +       val = mem_cgroup_zone_nr_lru_pages(memcg, nid, zid, LRU_ALL_FILE);
>>>>> +       if (do_swap_account)
>>> IMHO May get_nr_swap_pages() be more appropriate here?
>>>
>>
>> This is a per-memcg number, how would get_nr_swap_pages() help us here?
>>
> 
> I meant to add get_nr_swap_pages() as the if-judgement, that is:
>    if (do_swap_account && get_nr_swap_pages())
>        ....
> since anon pages becomes unreclaimable if swap space is used up.
> 
> 
Well, I believe this is doable.


More information about the Containers mailing list