Re: [PATCH v5 08/11] mm: hwpoison: soft offline supports thp migration
From: Anshuman Khandual
Date: Fri Apr 21 2017 - 04:11:51 EST
On 04/21/2017 02:17 AM, Zi Yan wrote:
> From: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx>
>
> This patch enables thp migration for soft offline.
>
> Signed-off-by: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx>
>
> ChangeLog: v1 -> v5:
> - fix page isolation counting error
>
> Signed-off-by: Zi Yan <zi.yan@xxxxxxxxxxxxxx>
> ---
> mm/memory-failure.c | 35 ++++++++++++++---------------------
> 1 file changed, 14 insertions(+), 21 deletions(-)
>
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index 9b77476ef31f..23ff02eb3ed4 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -1481,7 +1481,17 @@ static struct page *new_page(struct page *p, unsigned long private, int **x)
> if (PageHuge(p))
> return alloc_huge_page_node(page_hstate(compound_head(p)),
> nid);
> - else
> + else if (thp_migration_supported() && PageTransHuge(p)) {
> + struct page *thp;
> +
> + thp = alloc_pages_node(nid,
> + (GFP_TRANSHUGE | __GFP_THISNODE) & ~__GFP_RECLAIM,
Why not __GFP_RECLAIM ? Its soft offline path we wait a bit before
declaring that THP page cannot be allocated and hence should invoke
reclaim methods as well.
> + HPAGE_PMD_ORDER);
> + if (!thp)
> + return NULL;
> + prep_transhuge_page(thp);
> + return thp;
> + } else
> return __alloc_pages_node(nid, GFP_HIGHUSER_MOVABLE, 0);
> }
>
> @@ -1665,8 +1675,8 @@ static int __soft_offline_page(struct page *page, int flags)
> * cannot have PAGE_MAPPING_MOVABLE.
> */
> if (!__PageMovable(page))
> - inc_node_page_state(page, NR_ISOLATED_ANON +
> - page_is_file_cache(page));
> + mod_node_page_state(page_pgdat(page), NR_ISOLATED_ANON +
> + page_is_file_cache(page), hpage_nr_pages(page));
> list_add(&page->lru, &pagelist);
> ret = migrate_pages(&pagelist, new_page, NULL, MPOL_MF_MOVE_ALL,
> MIGRATE_SYNC, MR_MEMORY_FAILURE);
> @@ -1689,28 +1699,11 @@ static int __soft_offline_page(struct page *page, int flags)
> static int soft_offline_in_use_page(struct page *page, int flags)
> {
> int ret;
> - struct page *hpage = compound_head(page);
> -
> - if (!PageHuge(page) && PageTransHuge(hpage)) {
> - lock_page(hpage);
> - if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
> - unlock_page(hpage);
> - if (!PageAnon(hpage))
> - pr_info("soft offline: %#lx: non anonymous thp\n", page_to_pfn(page));
> - else
> - pr_info("soft offline: %#lx: thp split failed\n", page_to_pfn(page));
> - put_hwpoison_page(hpage);
> - return -EBUSY;
> - }
> - unlock_page(hpage);
> - get_hwpoison_page(page);
> - put_hwpoison_page(hpage);
> - }
>
> if (PageHuge(page))
> ret = soft_offline_huge_page(page, flags);
> else
> - ret = __soft_offline_page(page, flags);
> + ret = __soft_offline_page(compound_head(page), flags);
Hmm, what if the THP allocation fails in the new_page() path and
we fallback for general page allocation. In that case we will
always be still calling with the head page ? Because we dont
split the huge page any more.