On 2/2/2023 1:38 AM, Matthew Wilcox wrote:
> On Wed, Feb 01, 2023 at 04:17:36PM +0800, Yin Fengwei wrote:
>> do_set_pte_range() allows to setup page table entries for a
>> specific range. It calls page_add_file_rmap_range() to take
>> advantage of batched rmap update for large folio.
>
> How about something more like this? Yes, we need to define
> flush_icache_pages() and PTE_STRIDE.
Yes. This could remove more duplicated operations here. Let me
try to add flush_icache_pages() in next version.
But for pte_next(), it needs be added to handle all architectures.
I suppose it will take long journey to make it. What about a
followup patch for pte_next()?
Regards
Yin, Fengwei
>
> (we could also do for (i = 0; i < nr; i++) flush_icache_page(...) but
> given that some architectures already implement flush_icache_range(),
> I think they may appreciate being given one large range to flush)
>
> +++ b/mm/memory.c
> @@ -4277,15 +4277,19 @@ vm_fault_t do_set_pmd(struct vm_fault *vmf, struct page
> *page)
> }
> #endif
>
> -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 int start, unsigned int nr,
> + unsigned long addr)
> {
> + struct page *page = folio_page(page, start);
> struct vm_area_struct *vma = vmf->vma;
> bool uffd_wp = pte_marker_uffd_wp(vmf->orig_pte);
> bool write = vmf->flags & FAULT_FLAG_WRITE;
> bool prefault = vmf->address != addr;
> pte_t entry;
> + unsigned int i;
>
> - flush_icache_page(vma, page);
> + flush_icache_pages(vma, page, nr);
> entry = mk_pte(page, vma->vm_page_prot);
>
> if (prefault && arch_wants_old_prefaulted_pte())
> @@ -4299,14 +4303,23 @@ void do_set_pte(struct vm_fault *vmf, struct page *page, unsigned long addr)
> entry = pte_mkuffd_wp(pte_wrprotect(entry));
> /* copy-on-write page */
> if (write && !(vma->vm_flags & VM_SHARED)) {
> - inc_mm_counter(vma->vm_mm, MM_ANONPAGES);
> - page_add_new_anon_rmap(page, vma, addr);
> - lru_cache_add_inactive_or_unevictable(page, vma);
> + add_mm_counter(vma->vm_mm, MM_ANONPAGES, nr);
> + for (i = 0; i < nr; i++) {
> + page_add_new_anon_rmap(page + i, vma, addr);
> + lru_cache_add_inactive_or_unevictable(page + i, vma);
> + }
> } else {
> - inc_mm_counter(vma->vm_mm, mm_counter_file(page));
> - page_add_file_rmap(page, vma, false);
> + add_mm_counter(vma->vm_mm, mm_counter_file(page), nr);
> + folio_add_file_rmap(folio, start, n, vma);
> + }
> +
> + for (i = 0; i < nr; i++) {
> + set_pte_at(vma->vm_mm, addr, vmf->pte + i, entry);
> + /* no need to invalidate: a not-present page won't be cached */
> + update_mmu_cache(vma, addr, vmf->pte + i);
> + addr += PAGE_SIZE;
> + entry += PTE_STRIDE;
> }
> - set_pte_at(vma->vm_mm, addr, vmf->pte, entry);
> }
>
> static bool vmf_pte_changed(struct vm_fault *vmf)
>
