Re: [PATCH v2 3/5] mm: memory_hotplug: check hwpoisoned page firstly in do_migrate_range()

[David Hildenbrand <david@redhat.com>]

On 17.08.24 10:49, Kefeng Wang wrote:
> The commit b15c87263a69 ("hwpoison, memory_hotplug: allow hwpoisoned
> pages to be offlined") don't handle the hugetlb pages, the endless
> loop still occur if offline a hwpoison hugetlb, luckly, with the
> commit e591ef7d96d6 ("mm,hwpoison,hugetlb,memory_hotplug: hotremove
> memory section with hwpoisoned hugepage") section with hwpoisoned
> hugepage"), the HPageMigratable of hugetlb page will be clear, and
> the hwpoison hugetlb page will be skipped in scan_movable_pages(),
> so the endless loop issue is fixed.
> 
> However if the HPageMigratable() check passed(without reference and
> lock), the hugetlb page may be hwpoisoned, it won't cause issue since
> the hwpoisoned page will be handled correctly in the next movable
> pages scan loop, and it will be isolated in do_migrate_range() but
> fails to migrate. In order to avoid the unnecessary isolation and
> unify all hwpoisoned page handling, let's unconditionally check hwpoison
> firstly, and if it is a hwpoisoned hugetlb page, try to unmap it as
> the catch all safety net like normal page does.
> 
> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
> ---
>   mm/memory_hotplug.c | 17 +++++++++--------
>   1 file changed, 9 insertions(+), 8 deletions(-)
> 
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index dc19b0e28fbc..02a0d4fbc3fe 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1793,13 +1793,8 @@ static void do_migrate_range(unsigned long start_pfn, unsigned long end_pfn)
>   		 * but out loop could handle that as it revisits the split
>   		 * folio later.
>   		 */
> -		if (folio_test_large(folio)) {
> +		if (folio_test_large(folio))
>   			pfn = folio_pfn(folio) + folio_nr_pages(folio) - 1;
> -			if (folio_test_hugetlb(folio)) {
> -				isolate_hugetlb(folio, &source);
> -				continue;
> -			}
> -		}
>   
>   		/*
>   		 * HWPoison pages have elevated reference counts so the migration would
> @@ -1808,11 +1803,17 @@ static void do_migrate_range(unsigned long start_pfn, unsigned long end_pfn)
>   		 * (e.g. current hwpoison implementation doesn't unmap KSM pages but keep
>   		 * the unmap as the catch all safety net).
>   		 */
> -		if (PageHWPoison(page)) {
> +		if (folio_test_hwpoison(folio) ||
> +		    (folio_test_large(folio) && folio_test_has_hwpoisoned(folio))) {

We have the exact same check already in mm/shmem.c now.

Likely this should be factored out ... but no idea what function name we 
should use that won't add even more confusion :D

>   			if (WARN_ON(folio_test_lru(folio)))
>   				folio_isolate_lru(folio);
>   			if (folio_mapped(folio))
> -				try_to_unmap(folio, TTU_IGNORE_MLOCK);
> +				unmap_posioned_folio(folio, TTU_IGNORE_MLOCK);
> +			continue;
> +		}
> +
> +		if (folio_test_hugetlb(folio)) {
> +			isolate_hugetlb(folio, &source);
>   			continue;
>   		}
>   

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb