Re: [PATCH v2] mm/migrate: improve mlock tracking for PTE-mapped large folios

["David Hildenbrand (Arm)" <david@kernel.org>]

On 11/10/26 13:32, Wandun Chen wrote:
> From: Wandun Chen <chenwandun@lixiang.com>
> 
> Commit ab521b4142aa ("mm/rmap: improve mlock tracking for large folios")
> mlocks a large folio when all its pages are added in one rmap operation.
> However, when a PTE-mapped large folio is migrated, remove_migration_pte()
> restores PTEs individually, passing nr_pages = 1 to the rmap helpers. As a
> result, PG_mlocked is not set on the destination folio.
> 
> Reclaim can restore mlock in try_to_unmap(), but Mlocked in /proc/meminfo
> remains undercounted until that scan occurs.
> 
> Restore mlock once the full mapping has been restored in a locked VMA
> under one page table lock, rather than waiting for try_to_unmap() to
> restore the mlock state.
> 
> Signed-off-by: Wandun Chen <chenwandun@lixiang.com>
> ---
>  mm/migrate.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 94033c222e04..140e291d4587 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -353,6 +353,7 @@ static bool remove_migration_pte(struct folio *folio,
>  {
>  	struct rmap_walk_arg *rmap_walk_arg = arg;
>  	DEFINE_FOLIO_VMA_WALK(pvmw, rmap_walk_arg->folio, vma, addr, PVMW_SYNC | PVMW_MIGRATION);
> +	unsigned long nr_ptes = 0;
>  
>  	while (page_vma_mapped_walk(&pvmw)) {
>  		rmap_t rmap_flags = RMAP_NONE;
> @@ -446,9 +447,16 @@ static bool remove_migration_pte(struct folio *folio,
>  			else
>  				folio_add_file_rmap_pte(folio, new, vma);
>  			set_pte_at(vma->vm_mm, pvmw.address, pvmw.pte, pte);
> +
> +			nr_ptes++;
>  		}
> -		if (READ_ONCE(vma->vm_flags) & VM_LOCKED)
> +		if (READ_ONCE(vma->vm_flags) & VM_LOCKED) {
> +			if (folio_test_large(folio) &&
> +			    nr_ptes == folio_nr_pages(folio) &&
> +			    !(pvmw.flags & PVMW_PGTABLE_CROSSED))
> +				mlock_vma_folio(folio, vma);
>  			mlock_drain_local();
> +		}
>  
>  		trace_remove_migration_pte(pvmw.address, pte_val(pte),
>  					   compound_order(new));

I guess pte batching would be one way of handling it, but that requires more churn.

The spirit is similar as in try_to_unmap_one(), where we count the number of
ptes in one VMA. For consistency, I would call it "ptes".

@Kiryl, can you take a look?

BTW, why don't we care about that in try_to_migrate?

-- 
Cheers,

David