Re: [PATCH] mm: vmscan: correct nr_reclaimed for THP

[Yang Shi <yang.shi@linux.alibaba.com>]

On 5/9/19 8:03 PM, Huang, Ying wrote:
> Yang Shi <yang.shi@linux.alibaba.com> writes:
>
>> On 5/9/19 7:12 PM, Huang, Ying wrote:
>>> Yang Shi <yang.shi@linux.alibaba.com> writes:
>>>
>>>> Since commit bd4c82c22c36 ("mm, THP, swap: delay splitting THP after
>>>> swapped out"), THP can be swapped out in a whole.  But, nr_reclaimed
>>>> still gets inc'ed by one even though a whole THP (512 pages) gets
>>>> swapped out.
>>>>
>>>> This doesn't make too much sense to memory reclaim.  For example, direct
>>>> reclaim may just need reclaim SWAP_CLUSTER_MAX pages, reclaiming one THP
>>>> could fulfill it.  But, if nr_reclaimed is not increased correctly,
>>>> direct reclaim may just waste time to reclaim more pages,
>>>> SWAP_CLUSTER_MAX * 512 pages in worst case.
>>>>
>>>> This change may result in more reclaimed pages than scanned pages showed
>>>> by /proc/vmstat since scanning one head page would reclaim 512 base pages.
>>>>
>>>> Cc: "Huang, Ying" <ying.huang@intel.com>
>>>> Cc: Johannes Weiner <hannes@cmpxchg.org>
>>>> Cc: Michal Hocko <mhocko@suse.com>
>>>> Cc: Mel Gorman <mgorman@techsingularity.net>
>>>> Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
>>>> Cc: Hugh Dickins <hughd@google.com>
>>>> Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
>>>> ---
>>>> I'm not quite sure if it was the intended behavior or just omission. I tried
>>>> to dig into the review history, but didn't find any clue. I may miss some
>>>> discussion.
>>>>
>>>>    mm/vmscan.c | 6 +++++-
>>>>    1 file changed, 5 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>>> index fd9de50..7e026ec 100644
>>>> --- a/mm/vmscan.c
>>>> +++ b/mm/vmscan.c
>>>> @@ -1446,7 +1446,11 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>>>>      		unlock_page(page);
>>>>    free_it:
>>>> -		nr_reclaimed++;
>>>> +		/*
>>>> +		 * THP may get swapped out in a whole, need account
>>>> +		 * all base pages.
>>>> +		 */
>>>> +		nr_reclaimed += (1 << compound_order(page));
>>>>      		/*
>>>>    		 * Is there need to periodically free_page_list? It would
>>> Good catch!  Thanks!
>>>
>>> How about to change this to
>>>
>>>
>>>           nr_reclaimed += hpage_nr_pages(page);
>> Either is fine to me. Is this faster than "1 << compound_order(page)"?
> I think the readability is a little better.  And this will become
>
>          nr_reclaimed += 1
>
> if CONFIG_TRANSPARENT_HUAGEPAGE is disabled.

Good point. Will update in v2 soon.

>
> Best Regards,
> Huang, Ying
>
>>> Best Regards,
>>> Huang, Ying