Re: [PATCH 10/22] mm/khugepaged: restrict collapse to common memory

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Medium] mm/khugepaged: missing folio_is_common_memory() check in
  collapse_file()
--

commit 5cdd71b37e4c9e4d2c7bf61d7c51d7ba89a07fb5
Author: Gregory Price (Meta) <gourry@gourry.net>

mm/khugepaged: restrict collapse to common memory

This commit replaces the zone_device checks in anonymous THP scan and
isolation paths with page_is_common_memory() and adds a
folio_is_common_memory() check to collapse_scan_file().

> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 4bbe319ef86fa..b4bd29f25c145 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c

[ ... ]

> @@ -2741,6 +2741,12 @@ static enum scan_result collapse_scan_file(struct mm_struct *mm,
>  			break;
>  		}
>  
> +		if (unlikely(!folio_is_common_memory(folio))) {
> +			result = SCAN_PAGE_NULL;
> +			failed_pfn = folio_pfn(folio);
> +			folio_put(folio);
> +			break;
> +		}
>  		node = folio_nid(folio);
>  		if (collapse_scan_abort(node, cc)) {
>  			result = SCAN_SCAN_ABORT;

[Severity: Medium]
Should collapse_file() also check folio_is_common_memory(folio) after
locking each folio, matching __collapse_huge_page_isolate()?

collapse_scan_file() drops its folio reference on each iteration and
skips missing page-cache indices and swapped-out shmem entries, and
collapse_single_pmd() drops mmap_read_lock() before calling
collapse_pmd() -> collapse_file().  Folios that are faulted into holes,
migrated, or instantiated by shmem_get_folio() or
page_cache_sync_readahead() on a node outside N_MEMORY_COMMON after the
pre-scan will still be isolated and collapsed by collapse_file().

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261114070959.3490008-1-gourry@gourry.net?part=10