Re: [patch 3/3] mm: page_alloc: fair zone allocator policy

[Rik van Riel <riel@redhat.com>]

On 07/22/2013 05:04 PM, Johannes Weiner wrote:
> On Mon, Jul 22, 2013 at 04:21:07PM -0400, Rik van Riel wrote:
>> On 07/19/2013 04:55 PM, Johannes Weiner wrote:
>>
>>> @@ -1984,7 +1992,8 @@ this_zone_full:
>>>   		goto zonelist_scan;
>>>   	}
>>>
>>> -	if (page)
>>> +	if (page) {
>>> +		atomic_sub(1U << order, &zone->alloc_batch);
>>>   		/*
>>>   		 * page->pfmemalloc is set when ALLOC_NO_WATERMARKS was
>>>   		 * necessary to allocate the page. The expectation is
>>
>> Could this be moved into the slow path in buffered_rmqueue and
>> rmqueue_bulk, or would the effect of ignoring the pcp buffers be
>> too detrimental to keeping the balance between zones?
>
> What I'm worried about is not the inaccury that comes from the buffer
> size but the fact that there are no guaranteed buffer empty+refill
> cycles.  The reclaimer could end up feeding the pcp list that the
> allocator is using indefinitely, which brings us back to the original
> problem.  If you have >= NR_CPU jobs running, the kswapds are bound to
> share CPUs with the allocating tasks, so the scenario is not unlikely.

You are absolutely right.  Thinking about it some more,
I cannot think of a better way to do this than your patch.

Reviewed-by: Rik van Riel <riel@redhat.com>

-- 
All rights reversed

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>