Re: [RFC] [PATCH] memory controller statistics

[Balbir Singh <balbir-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org>]

Balbir Singh wrote:
> YAMAMOTO Takashi wrote:
>>> hi,
>>>
>>> i implemented some statistics for your memory controller.
>>>
>>> it's tested with 2.6.23-rc2-mm2 + memory controller v7.
>>> i think it can be applied to 2.6.23-rc4-mm1 as well.
>>>
>>> YAMOMOTO Takshi
>>>
>>> todo: something like nr_active/inactive in /proc/vmstat.
>> here's the version i'm working on currently.  any comments?
>>
>> changes from the previous version:
>>
>> - adapt to 2.6.23-rc8-mm2, container -> cgroup rename.
>> - reflect some of comments on this list.
>> 	- rename some macros as suggested by balbir
>> 	- sprinkle some inlines as suggested by balbir.
>> 	- remove "isolate" statistics
>> 	- remove PAGE_CONTAINER_CACHE hack and
>> 	  add "flags" member to page_container instead.
>> 	- make counters atomic_long_t.
>> 	- add some comments.
>> 	- coding style.
>> - implement nr_active/nr_inactive.  they show numbers of pages on
>>   per-cgroup lru lists.
>>
>> todo:
>> - consider to make counters per-cpu.
>> - more statistics.
>>
>> YAMAMOTO Takashi
> 
> Hi, YAMAMOTO-San
> 
> Looks much better, I did a quick review, it looks nice to me.
> I'll test the patches and get back.
> 
> Thank you for working on this,
> 

Forgot to mention one other detail, could we track the statistics
in bytes? We track usage and limit in bytes currently. I am open
to opinions on this, but the last time we discussed numbers,
tracking bytes was the preferred approach.

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL