[06/11] mm/ksm: use folio in write_protect_page

Commit Message

[e-mail] March 20, 2024, 7:40 a.m. UTC

From: "Alex Shi (tencent)" <[e-mail]>

Compound page is checked and skipped before write_protect_page() called,
use folio to save few compound_head checking and also remove duplicated
compound checking again in the func.

Signed-off-by: Alex Shi (tencent) <[e-mail]>
Cc: Izik Eidus <[e-mail]>
Cc: Matthew Wilcox <[e-mail]>
Cc: Andrea Arcangeli <[e-mail]>
Cc: Hugh Dickins <[e-mail]>
Cc: Chris Wright <[e-mail]>
---
 mm/ksm.c | 22 ++++++++++------------
 1 file changed, 10 insertions(+), 12 deletions(-)

Comments

Matthew Wilcox March 20, 2024, 2:57 p.m. UTC | #1

On Wed, Mar 20, 2024 at 03:40:42PM +0800, [e-mail] wrote:
> -static int write_protect_page(struct vm_area_struct *vma, struct page *page,
> +static int write_protect_page(struct vm_area_struct *vma, struct folio *folio,
>  			      pte_t *orig_pte)
>  {
>  	struct mm_struct *mm = vma->vm_mm;
> -	DEFINE_PAGE_VMA_WALK(pvmw, page, vma, 0, 0);
> +	DEFINE_PAGE_VMA_WALK(pvmw, &folio->page, vma, 0, 0);

We have a DEFINE_FOLIO_VMA_WALK

> -	pvmw.address = page_address_in_vma(page, vma);
> +	pvmw.address = page_address_in_vma(&folio->page, vma);

We don't yet have a folio_address_in_vma().  This needs more study,
so I approve of how you've converted this line.

> -	BUG_ON(PageTransCompound(page));

I might make this a VM_BUG_ON(folio_test_large(folio))

> @@ -1505,7 +1503,7 @@ static int try_to_merge_one_page(struct vm_area_struct *vma,
>  	 * ptes are necessarily already write-protected.  But in either
>  	 * case, we need to lock and check page_count is not raised.
>  	 */
> -	if (write_protect_page(vma, page, &orig_pte) == 0) {
> +	if (write_protect_page(vma, (struct folio *)page, &orig_pte) == 0) {

I don't love this cast.  I see why it's safe (called split_huge_page()
above), but I'd rather see a call to page_folio() just to keep things
tidy.

Patch

diff --git a/mm/ksm.c b/mm/ksm.c
index 165a3e4162bf..ad3a0294a2ec 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -1275,23 +1275,21 @@  static u32 calc_checksum(struct page *page)
 	return checksum;
 }
 
-static int write_protect_page(struct vm_area_struct *vma, struct page *page,
+static int write_protect_page(struct vm_area_struct *vma, struct folio *folio,
 			      pte_t *orig_pte)
 {
 	struct mm_struct *mm = vma->vm_mm;
-	DEFINE_PAGE_VMA_WALK(pvmw, page, vma, 0, 0);
+	DEFINE_PAGE_VMA_WALK(pvmw, &folio->page, vma, 0, 0);
 	int swapped;
 	int err = -EFAULT;
 	struct mmu_notifier_range range;
 	bool anon_exclusive;
 	pte_t entry;
 
-	pvmw.address = page_address_in_vma(page, vma);
+	pvmw.address = page_address_in_vma(&folio->page, vma);
 	if (pvmw.address == -EFAULT)
 		goto out;
 
-	BUG_ON(PageTransCompound(page));
-
 	mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, mm, pvmw.address,
 				pvmw.address + PAGE_SIZE);
 	mmu_notifier_invalidate_range_start(&range);
@@ -1301,12 +1299,12 @@  static int write_protect_page(struct vm_area_struct *vma, struct page *page,
 	if (WARN_ONCE(!pvmw.pte, "Unexpected PMD mapping?"))
 		goto out_unlock;
 
-	anon_exclusive = PageAnonExclusive(page);
+	anon_exclusive = PageAnonExclusive(&folio->page);
 	entry = ptep_get(pvmw.pte);
 	if (pte_write(entry) || pte_dirty(entry) ||
 	    anon_exclusive || mm_tlb_flush_pending(mm)) {
-		swapped = PageSwapCache(page);
-		flush_cache_page(vma, pvmw.address, page_to_pfn(page));
+		swapped = folio_test_swapcache(folio);
+		flush_cache_page(vma, pvmw.address, folio_pfn(folio));
 		/*
 		 * Ok this is tricky, when get_user_pages_fast() run it doesn't
 		 * take any lock, therefore the check that we are going to make
@@ -1326,20 +1324,20 @@  static int write_protect_page(struct vm_area_struct *vma, struct page *page,
 		 * Check that no O_DIRECT or similar I/O is in progress on the
 		 * page
 		 */
-		if (page_mapcount(page) + 1 + swapped != page_count(page)) {
+		if (folio_mapcount(folio) + 1 + swapped != folio_ref_count(folio)) {
 			set_pte_at(mm, pvmw.address, pvmw.pte, entry);
 			goto out_unlock;
 		}
 
 		/* See folio_try_share_anon_rmap_pte(): clear PTE first. */
 		if (anon_exclusive &&
-		    folio_try_share_anon_rmap_pte(page_folio(page), page)) {
+		    folio_try_share_anon_rmap_pte(folio, &folio->page)) {
 			set_pte_at(mm, pvmw.address, pvmw.pte, entry);
 			goto out_unlock;
 		}
 
 		if (pte_dirty(entry))
-			set_page_dirty(page);
+			folio_mark_dirty(folio);
 		entry = pte_mkclean(entry);
 
 		if (pte_write(entry))
@@ -1505,7 +1503,7 @@  static int try_to_merge_one_page(struct vm_area_struct *vma,
 	 * ptes are necessarily already write-protected.  But in either
 	 * case, we need to lock and check page_count is not raised.
 	 */
-	if (write_protect_page(vma, page, &orig_pte) == 0) {
+	if (write_protect_page(vma, (struct folio *)page, &orig_pte) == 0) {
 		if (!kpage) {
 			/*
 			 * While we hold page lock, upgrade page from