Re: [PATCH v1 2/4] mm: smaller folio_pte_batch() improvements

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Fri, Jun 27, 2025 at 01:55:08PM +0200, David Hildenbrand wrote:
> Let's clean up a bit:
>
> (1) No need for start_ptep vs. ptep anymore, we can simply use ptep
>
> (2) Let's switch to "unsigned int" for everything
>
> (3) We can simplify the code by leaving the pte unchanged after the
>     pte_same() check.
>
> (4) Clarify that we should never exceed a single VMA; it indicates a
>     problem in the caller.
>
> No functional change intended.
>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  mm/internal.h | 37 +++++++++++++++----------------------
>  1 file changed, 15 insertions(+), 22 deletions(-)
>
> diff --git a/mm/internal.h b/mm/internal.h
> index 9690c75063881..ca6590c6d9eab 100644
> --- a/mm/internal.h
> +++ b/mm/internal.h
> @@ -221,7 +221,7 @@ static inline pte_t __pte_batch_clear_ignored(pte_t pte, fpb_t flags)
>   * folio_pte_batch - detect a PTE batch for a large folio
>   * @folio: The large folio to detect a PTE batch for.
>   * @addr: The user virtual address the first page is mapped at.
> - * @start_ptep: Page table pointer for the first entry.
> + * @ptep: Page table pointer for the first entry.
>   * @pte: Page table entry for the first page.
>   * @max_nr: The maximum number of table entries to consider.
>   * @flags: Flags to modify the PTE batch semantics.
> @@ -233,24 +233,24 @@ static inline pte_t __pte_batch_clear_ignored(pte_t pte, fpb_t flags)
>   *		  first one is dirty.
>   *
>   * Detect a PTE batch: consecutive (present) PTEs that map consecutive
> - * pages of the same large folio.
> + * pages of the same large folio in a single VMA and a single page table.
>   *
>   * All PTEs inside a PTE batch have the same PTE bits set, excluding the PFN,
>   * the accessed bit, writable bit, dirty bit (unless FPB_HONOR_DIRTY is set) and
>   * soft-dirty bit (unless FPB_HONOR_SOFT_DIRTY is set).
>   *
> - * start_ptep must map any page of the folio. max_nr must be at least one and
> - * must be limited by the caller so scanning cannot exceed a single page table.
> + * @ptep must map any page of the folio. max_nr must be at least one and
> + * must be limited by the caller so scanning cannot exceed a single VMA and
> + * a single page table.
>   *
>   * Return: the number of table entries in the batch.
>   */
> -static inline int folio_pte_batch(struct folio *folio, unsigned long addr,
> -		pte_t *start_ptep, pte_t pte, int max_nr, fpb_t flags,
> +static inline unsigned int folio_pte_batch(struct folio *folio, unsigned long addr,

Do we need to worry about propagating this type change?

mremap_folio_pte_batch() and zap_present_ptes() return the value as an int for example.

I mean I doubt we're going to be seeing an overflow here :) but maybe worth
propagating this everywhere.

> +		pte_t *ptep, pte_t pte, unsigned int max_nr, fpb_t flags,
>  		bool *any_writable, bool *any_young, bool *any_dirty)
>  {
> -	pte_t expected_pte, *ptep;
> -	bool writable, young, dirty;
> -	int nr, cur_nr;
> +	unsigned int nr, cur_nr;
> +	pte_t expected_pte;
>
>  	if (any_writable)
>  		*any_writable = false;
> @@ -267,29 +267,22 @@ static inline int folio_pte_batch(struct folio *folio, unsigned long addr,
>  	max_nr = min_t(unsigned long, max_nr,
>  		       folio_pfn(folio) + folio_nr_pages(folio) - pte_pfn(pte));
>
> -	nr = pte_batch_hint(start_ptep, pte);
> +	nr = pte_batch_hint(ptep, pte);
>  	expected_pte = __pte_batch_clear_ignored(pte_advance_pfn(pte, nr), flags);
> -	ptep = start_ptep + nr;
> +	ptep = ptep + nr;
>
>  	while (nr < max_nr) {
>  		pte = ptep_get(ptep);
> -		if (any_writable)
> -			writable = !!pte_write(pte);
> -		if (any_young)
> -			young = !!pte_young(pte);
> -		if (any_dirty)
> -			dirty = !!pte_dirty(pte);
> -		pte = __pte_batch_clear_ignored(pte, flags);
>
> -		if (!pte_same(pte, expected_pte))
> +		if (!pte_same(__pte_batch_clear_ignored(pte, flags), expected_pte))

Doing this here will change the output of any_writable, any_young:

static inline pte_t __pte_batch_clear_ignored(pte_t pte, fpb_t flags)
{
	...
	return pte_wrprotect(pte_mkold(pte));
}

So we probably need to get these values earlier?

>  			break;
>
>  		if (any_writable)
> -			*any_writable |= writable;
> +			*any_writable |= pte_write(pte);
>  		if (any_young)
> -			*any_young |= young;
> +			*any_young |= pte_young(pte);
>  		if (any_dirty)
> -			*any_dirty |= dirty;
> +			*any_dirty |= pte_dirty(pte);
>
>  		cur_nr = pte_batch_hint(ptep, pte);
>  		expected_pte = pte_advance_pfn(expected_pte, cur_nr);
> --
> 2.49.0
>