Re: [RFC PATCH v4 3/4] mm: add do_set_pte_range()

[Matthew Wilcox <willy@infradead.org>]

On Mon, Feb 06, 2023 at 10:06:38PM +0800, Yin Fengwei wrote:
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index d6f8f41514cc..93192f04b276 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -1162,6 +1162,9 @@ static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
>  
>  vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page *page);
>  void do_set_pte(struct vm_fault *vmf, struct page *page, unsigned long addr);
> +void do_set_pte_range(struct vm_fault *vmf, struct folio *folio,
> +		unsigned long addr, pte_t *pte,
> +		unsigned long start, unsigned int nr);

There are only two callers of do_set_pte(), and they're both in mm.
I don't think we should retain do_set_pte() as a wrapper, but rather
change both callers to call 'set_pte_range()'.  The 'do' doesn't add
any value, so let's drop that word.

> +	if (!cow) {
> +		folio_add_file_rmap_range(folio, start, nr, vma, false);
> +		add_mm_counter(vma->vm_mm, mm_counter_file(page), nr);
> +	} else {
> +		/*
> +		 * rmap code is not ready to handle COW with anonymous
> +		 * large folio yet. Capture and warn if large folio
> +		 * is given.
> +		 */
> +		VM_WARN_ON_FOLIO(folio_test_large(folio), folio);
> +	}

The handling of cow pages is still very clunky.
folio_add_new_anon_rmap() handles anonymous large folios just fine.  I
think David was looking at current code, not the code in mm-next.

> +		set_pte_at(vma->vm_mm, addr, pte, entry);
> +
> +		/* no need to invalidate: a not-present page won't be cached */
> +		update_mmu_cache(vma, addr, pte);
> +	} while (pte++, page++, addr += PAGE_SIZE, --nr > 0);

There's no need to speed-run this.  Let's do it properly and get the
arch interface right.  This code isn't going to hit linux-next for four
more weeks, which is plenty of time.