Re: [PATCH 11/11] mm/mm_init: move pageblock migratetype init to memmap helpers

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] mm/mm_init: commit message cites nonexistent
  memmap_init_page_frozen()
It also noticed 1 related pre-existing issue this patch does not
introduce:
- [Critical] mm: uninitialized migratetype in
  pageblock_migratetype_init_range() [1]
--

commit e81bf248ba8e6f6857b5a558e23fdf0fa0e68ad6
Author: Muchun Song <muchun.song@linux.dev>

mm/mm_init: move pageblock migratetype init to memmap helpers

This commit moves MIGRATE_MOVABLE pageblock initialization into
memmap_init_compound_page_frozen() and memmap_init_zone_device_basepages(),
and removes hugetlb_bootmem_init_migratetype().

> Move the MIGRATE_MOVABLE pageblock initialization into the frozen memmap
> initialization helpers instead. This makes memmap_init_page_frozen()
> initialize the pageblock for single pages and
> memmap_init_compound_page_frozen() initialize the whole range for compound
> pages, so the ZONE_DEVICE and hugetlb bootmem paths can share the same
> initialization.

[Severity: Low]
This isn't a bug, but the commit message mentions a nonexistent
memmap_init_page_frozen() helper.  The patch updates
memmap_init_zone_device_basepages() to initialize the pageblock range for
ZONE_DEVICE base pages instead.

> diff --git a/mm/mm_init.c b/mm/mm_init.c
> index 61a720bb44dce..e05db8abde2fd 100644
> --- a/mm/mm_init.c
> +++ b/mm/mm_init.c
> @@ -751,6 +751,9 @@ void __meminit memmap_init_compound_page_frozen(struct page *head,
>  	for (unsigned long i = 1; i < nr_pages; i++)
>  		memmap_init_page_from_template(head + i, pfn + i, zone, &template);
>  	prep_compound_head(head, order);
> +
> +	pageblock_migratetype_init_range(pfn, 1UL << order, MIGRATE_MOVABLE,
> +					 /* isolate */ false, /* atomic */ false);
>  }
>  
>  struct zone __meminit *pfn_to_zone(unsigned long pfn, int nid)
> @@ -1110,6 +1113,9 @@ static void memmap_init_zone_device_basepages(unsigned long start, unsigned long
>  		if (IS_ALIGNED(pfn, PAGES_PER_SECTION))
>  			cond_resched();
>  	}
> +
> +	pageblock_migratetype_init_range(start, end - start, MIGRATE_MOVABLE,
> +					 /* isolate */ false, /* atomic */ false);
>  }

[Severity: Critical]
This problem wasn't introduced by this patch, but can
pageblock_migratetype_init_range() leave the first partial pageblock's
migratetype uninitialized [1] when the starting PFN is not aligned to
a pageblock boundary?

mm/mm_init.c:pageblock_migratetype_init_range() {
    ...
	for (pfn = pageblock_align(pfn); pfn < end; pfn += pageblock_nr_pages) {
		enum migratetype mt = kho_bootmem_migratetype(pfn, migratetype);

		init_pageblock_migratetype(pfn_to_page(pfn), mt, isolate);
    ...
}

pageblock_align(pfn) rounds the starting PFN upward to the next pageblock
boundary.  When a caller such as ZONE_DEVICE initialization or memory
hotplug passes a range starting in the middle of a pageblock, the loop
skips that initial partial pageblock, leaving its pageblock flags
uninitialized.

[1] https://sashiko.dev/#/bug/linux-492df7cc-d097-43e5-9701-9954aee1d064

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261109095807.41547-1-songmuchun@bytedance.com?part=11