Re: [PATCH v2 05/12] mm/hugetlb: unify hugetlb migration callback function

[Michal Hocko]

On Wed 27-05-20 15:44:56, Joonsoo Kim wrote:
[...]
> -/* page migration callback function */
>  struct page *alloc_huge_page_nodemask(struct hstate *h,
>  				struct alloc_control *ac)
>  {
>  	ac->gfp_mask |= htlb_alloc_mask(h);
> +	if (ac->nid == NUMA_NO_NODE)
> +		ac->gfp_mask &= ~__GFP_THISNODE;

Is this really needed? alloc_huge_page_node is currently only called
from numa migration code and the target node should be always defined.

>  
>  	spin_lock(&hugetlb_lock);
>  	if (h->free_huge_pages - h->resv_huge_pages > 0) {
> diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> index 3b6b551..e705efd 100644
> --- a/mm/mempolicy.c
> +++ b/mm/mempolicy.c
> @@ -1073,9 +1073,10 @@ struct page *alloc_new_node_page(struct page *page, unsigned long node)
>  		struct alloc_control ac = {
>  			.nid = node,
>  			.nmask = NULL,
> +			.gfp_mask = __GFP_THISNODE,
>  		};
>  
> -		return alloc_huge_page_node(h, &ac);
> +		return alloc_huge_page_nodemask(h, &ac);
>  	} else if (PageTransHuge(page)) {
>  		struct page *thp;
>  
> -- 
> 2.7.4
> 

-- 
Michal Hocko
SUSE Labs