Re: [PATCH v2 1/4] mm,hwpoison,hugetlb,memory_hotplug: hotremove memory section with hwpoisoned hugepage

[Miaohe Lin <linmiaohe@huawei.com>]

On 2022/9/5 14:21, Naoya Horiguchi wrote:
> From: Naoya Horiguchi <naoya.horiguchi@nec.com>
> 
> HWPoisoned page is not supposed to be accessed once marked, but currently
> such accesses can happen during memory hotremove because do_migrate_range()
> can be called before dissolve_free_huge_pages() is called.
> 
> Move dissolve_free_huge_pages() before scan_movable_pages(). Recently
> delayed dissolve has been implemented, so the dissolving can turn
> a hwpoisoned hugepage into 4kB hwpoison page, which memory hotplug can
> handle safely.

Yes, thanks for your work, Naoya. ;)

> 
> Reported-by: Miaohe Lin <linmiaohe@huawei.com>
> Signed-off-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
> ---
>  mm/memory_hotplug.c | 22 +++++++++++-----------
>  1 file changed, 11 insertions(+), 11 deletions(-)
> 
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index fad6d1f2262a..c24735d63b25 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1880,6 +1880,17 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
>  
>  			cond_resched();
>  
> +			/*
> +			 * Dissolve free hugepages in the memory block before doing
> +			 * offlining actually in order to make hugetlbfs's object
> +			 * counting consistent.
> +			 */
> +			ret = dissolve_free_huge_pages(start_pfn, end_pfn);
> +			if (ret) {
> +				reason = "failure to dissolve huge pages";
> +				goto failed_removal_isolated;
> +			}

This change has a side-effect. If hugetlb pages are in-use, dissolve_free_huge_pages() will always return -EBUSY
even if those pages can be migrated. So we fail to hotremove the memory even if they could be offlined.
Or am I miss something?

Thanks,
Miaohe Lin

> +
>  			ret = scan_movable_pages(pfn, end_pfn, &pfn);
>  			if (!ret) {
>  				/*
> @@ -1895,17 +1906,6 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
>  			goto failed_removal_isolated;
>  		}
>  
> -		/*
> -		 * Dissolve free hugepages in the memory block before doing
> -		 * offlining actually in order to make hugetlbfs's object
> -		 * counting consistent.
> -		 */
> -		ret = dissolve_free_huge_pages(start_pfn, end_pfn);
> -		if (ret) {
> -			reason = "failure to dissolve huge pages";
> -			goto failed_removal_isolated;
> -		}
> -
>  		ret = test_pages_isolated(start_pfn, end_pfn, MEMORY_OFFLINE);
>  
>  	} while (ret);
>