Re: [PATCH v9 04/11] mm/hugetlb: Defer freeing of HugeTLB pages

[Mike Kravetz <mike.kravetz@oracle.com>]

On 12/13/20 7:45 AM, Muchun Song wrote:
> In the subsequent patch, we will allocate the vmemmap pages when free
> HugeTLB pages. But update_and_free_page() is called from a non-task
> context(and hold hugetlb_lock), so we can defer the actual freeing in
> a workqueue to prevent use GFP_ATOMIC to allocate the vmemmap pages.
> 
> Signed-off-by: Muchun Song <songmuchun@bytedance.com>

It is unfortunate we need to add this complexitty, but I can not think
of another way.  One small comment (no required change) below.

Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>

> ---
>  mm/hugetlb.c         | 77 ++++++++++++++++++++++++++++++++++++++++++++++++----
>  mm/hugetlb_vmemmap.c | 12 --------
>  mm/hugetlb_vmemmap.h | 17 ++++++++++++
>  3 files changed, 88 insertions(+), 18 deletions(-)
> 
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 140135fc8113..0ff9b90e524f 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -1292,15 +1292,76 @@ static inline void destroy_compound_gigantic_page(struct page *page,
>  						unsigned int order) { }
>  #endif
>  
> -static void update_and_free_page(struct hstate *h, struct page *page)
> +static void __free_hugepage(struct hstate *h, struct page *page);
> +
> +/*
> + * As update_and_free_page() is be called from a non-task context(and hold
> + * hugetlb_lock), we can defer the actual freeing in a workqueue to prevent
> + * use GFP_ATOMIC to allocate a lot of vmemmap pages.
> + *
> + * update_hpage_vmemmap_workfn() locklessly retrieves the linked list of
> + * pages to be freed and frees them one-by-one. As the page->mapping pointer
> + * is going to be cleared in update_hpage_vmemmap_workfn() anyway, it is
> + * reused as the llist_node structure of a lockless linked list of huge
> + * pages to be freed.
> + */
> +static LLIST_HEAD(hpage_update_freelist);
> +
> +static void update_hpage_vmemmap_workfn(struct work_struct *work)
>  {
> -	int i;
> +	struct llist_node *node;
> +	struct page *page;
> +
> +	node = llist_del_all(&hpage_update_freelist);
>  
> +	while (node) {
> +		page = container_of((struct address_space **)node,
> +				     struct page, mapping);
> +		node = node->next;
> +		page->mapping = NULL;
> +		__free_hugepage(page_hstate(page), page);
> +
> +		cond_resched();
> +	}
> +}
> +static DECLARE_WORK(hpage_update_work, update_hpage_vmemmap_workfn);
> +
> +static inline void __update_and_free_page(struct hstate *h, struct page *page)
> +{
> +	/* No need to allocate vmemmap pages */
> +	if (!free_vmemmap_pages_per_hpage(h)) {
> +		__free_hugepage(h, page);
> +		return;
> +	}
> +
> +	/*
> +	 * Defer freeing to avoid using GFP_ATOMIC to allocate vmemmap
> +	 * pages.
> +	 *
> +	 * Only call schedule_work() if hpage_update_freelist is previously
> +	 * empty. Otherwise, schedule_work() had been called but the workfn
> +	 * hasn't retrieved the list yet.
> +	 */
> +	if (llist_add((struct llist_node *)&page->mapping,
> +		      &hpage_update_freelist))
> +		schedule_work(&hpage_update_work);
> +}
> +
> +static void update_and_free_page(struct hstate *h, struct page *page)
> +{
>  	if (hstate_is_gigantic(h) && !gigantic_page_runtime_supported())
>  		return;
>  
>  	h->nr_huge_pages--;
>  	h->nr_huge_pages_node[page_to_nid(page)]--;
> +
> +	__update_and_free_page(h, page);
> +}
> +
> +static void __free_hugepage(struct hstate *h, struct page *page)
> +{
> +	int i;
> +

Can we add a comment here saying that this is where the call to allocate
vmemmmap pages will be inserted in a later patch.  Such a comment would
help a bit to understand the restructuring of the code.

-- 
Mike Kravetz

>  	for (i = 0; i < pages_per_huge_page(h); i++) {
>  		page[i].flags &= ~(1 << PG_locked | 1 << PG_error |
>  				1 << PG_referenced | 1 << PG_dirty |
> @@ -1313,13 +1374,17 @@ static void update_and_free_page(struct hstate *h, struct page *page)
>  	set_page_refcounted(page);
>  	if (hstate_is_gigantic(h)) {
>  		/*
> -		 * Temporarily drop the hugetlb_lock, because
> -		 * we might block in free_gigantic_page().
> +		 * Temporarily drop the hugetlb_lock only when this type of
> +		 * HugeTLB page does not support vmemmap optimization (which
> +		 * contex do not hold the hugetlb_lock), because we might block
> +		 * in free_gigantic_page().
>  		 */
> -		spin_unlock(&hugetlb_lock);
> +		if (!free_vmemmap_pages_per_hpage(h))
> +			spin_unlock(&hugetlb_lock);
>  		destroy_compound_gigantic_page(page, huge_page_order(h));
>  		free_gigantic_page(page, huge_page_order(h));
> -		spin_lock(&hugetlb_lock);
> +		if (!free_vmemmap_pages_per_hpage(h))
> +			spin_lock(&hugetlb_lock);
>  	} else {
>  		__free_pages(page, huge_page_order(h));
>  	}