[PATCH v2 1/4] mm/hugetlb: Refactor unmap_ref_private() to take folio instead of page

[PATCH v2 1/4] mm/hugetlb: Refactor unmap_ref_private() to take folio instead of page

[nifan.cxl]

From: Fan Ni <fan.ni@samsung.com>

The function unmap_ref_private() has only user, which passes in
&folio->page. Let it take folio directly.

Signed-off-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Muchun Song <muchun.song@linux.dev>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
---
v2:
Picked up Reviewed-by tags;

v1:
https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#mbf9f3e8b49497755b414e1887b2376b3902ffb76
---
 mm/hugetlb.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index ccc4f08f8481..b5d1ac8290a7 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -6064,7 +6064,7 @@ void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
  * same region.
  */
 static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
-			      struct page *page, unsigned long address)
+			      struct folio *folio, unsigned long address)
 {
 	struct hstate *h = hstate_vma(vma);
 	struct vm_area_struct *iter_vma;
@@ -6108,7 +6108,8 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
 		 */
 		if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
 			unmap_hugepage_range(iter_vma, address,
-					     address + huge_page_size(h), page, 0);
+					     address + huge_page_size(h),
+					     folio_page(folio, 0), 0);
 	}
 	i_mmap_unlock_write(mapping);
 }
@@ -6231,8 +6232,7 @@ static vm_fault_t hugetlb_wp(struct folio *pagecache_folio,
 			hugetlb_vma_unlock_read(vma);
 			mutex_unlock(&hugetlb_fault_mutex_table[hash]);
 
-			unmap_ref_private(mm, vma, &old_folio->page,
-					vmf->address);
+			unmap_ref_private(mm, vma, old_folio, vmf->address);
 
 			mutex_lock(&hugetlb_fault_mutex_table[hash]);
 			hugetlb_vma_lock_read(vma);
-- 
2.47.2

[PATCH v2 2/4] mm/hugetlb: Refactor unmap_hugepage_range() to take folio instead of page

[nifan.cxl]

From: Fan Ni <fan.ni@samsung.com>

The function unmap_hugepage_range() has two kinds of users:
1) unmap_ref_private(), which passes in the head page of a folio.  Since
   unmap_ref_private() already takes folio and there are no other uses
   of the folio struct in the function, it is natural for
   unmap_hugepage_range() to take folio also.
2) All other uses, which pass in NULL pointer.

In both cases, we can pass in folio. Refactor unmap_hugepage_range() to
take folio.

Signed-off-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Muchun Song <muchun.song@linux.dev>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
---
v2: 
Picked up Reviewed-by tags;

v1:
https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m7ba630d783e08ae708453279579093d481c6e721
---
 include/linux/hugetlb.h | 2 +-
 mm/hugetlb.c            | 7 ++++---
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index a57bed83c657..b7699f35c87f 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -128,7 +128,7 @@ int move_hugetlb_page_tables(struct vm_area_struct *vma,
 int copy_hugetlb_page_range(struct mm_struct *, struct mm_struct *,
 			    struct vm_area_struct *, struct vm_area_struct *);
 void unmap_hugepage_range(struct vm_area_struct *,
-			  unsigned long, unsigned long, struct page *,
+			  unsigned long, unsigned long, struct folio *folio,
 			  zap_flags_t);
 void __unmap_hugepage_range(struct mmu_gather *tlb,
 			  struct vm_area_struct *vma,
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index b5d1ac8290a7..3181dbe0c4bb 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -6039,7 +6039,7 @@ void __hugetlb_zap_end(struct vm_area_struct *vma,
 }
 
 void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
-			  unsigned long end, struct page *ref_page,
+			  unsigned long end, struct folio *ref_folio,
 			  zap_flags_t zap_flags)
 {
 	struct mmu_notifier_range range;
@@ -6051,7 +6051,8 @@ void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
 	mmu_notifier_invalidate_range_start(&range);
 	tlb_gather_mmu(&tlb, vma->vm_mm);
 
-	__unmap_hugepage_range(&tlb, vma, start, end, ref_page, zap_flags);
+	__unmap_hugepage_range(&tlb, vma, start, end,
+			       folio_page(ref_folio, 0), zap_flags);
 
 	mmu_notifier_invalidate_range_end(&range);
 	tlb_finish_mmu(&tlb);
@@ -6109,7 +6110,7 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
 		if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
 			unmap_hugepage_range(iter_vma, address,
 					     address + huge_page_size(h),
-					     folio_page(folio, 0), 0);
+					     folio, 0);
 	}
 	i_mmap_unlock_write(mapping);
 }
-- 
2.47.2

[PATCH v2 3/4] mm/hugetlb: Refactor __unmap_hugepage_range() to take folio instead of page

[nifan.cxl]

From: Fan Ni <fan.ni@samsung.com>

The function __unmap_hugepage_range() has two kinds of users:
1) unmap_hugepage_range(), which passes in the head page of a folio.
   Since unmap_hugepage_range() already takes folio and there are no other
   uses of the folio struct in the function, it is natural for
   __unmap_hugepage_range() to take folio also.
2) All other uses, which pass in NULL pointer.

In both cases, we can pass in folio. Refactor __unmap_hugepage_range() to
take folio.

Signed-off-by: Fan Ni <fan.ni@samsung.com>
---
v2: fixed issue mentioned here:
https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m2b9cc1743e1907e52658815b297b9d249474f387

v1:
https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m738e9e7f7d7fe4aab6b21782b4658dd65abf8fc4
---

 include/linux/hugetlb.h |  4 ++--
 mm/hugetlb.c            | 10 +++++-----
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index b7699f35c87f..ebaf95231934 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -133,7 +133,7 @@ void unmap_hugepage_range(struct vm_area_struct *,
 void __unmap_hugepage_range(struct mmu_gather *tlb,
 			  struct vm_area_struct *vma,
 			  unsigned long start, unsigned long end,
-			  struct page *ref_page, zap_flags_t zap_flags);
+			  struct folio *ref_folio, zap_flags_t zap_flags);
 void hugetlb_report_meminfo(struct seq_file *);
 int hugetlb_report_node_meminfo(char *buf, int len, int nid);
 void hugetlb_show_meminfo_node(int nid);
@@ -452,7 +452,7 @@ static inline long hugetlb_change_protection(
 
 static inline void __unmap_hugepage_range(struct mmu_gather *tlb,
 			struct vm_area_struct *vma, unsigned long start,
-			unsigned long end, struct page *ref_page,
+			unsigned long end, struct folio *ref_folio,
 			zap_flags_t zap_flags)
 {
 	BUG();
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 3181dbe0c4bb..7d280ab23784 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -5833,7 +5833,7 @@ int move_hugetlb_page_tables(struct vm_area_struct *vma,
 
 void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 			    unsigned long start, unsigned long end,
-			    struct page *ref_page, zap_flags_t zap_flags)
+			    struct folio *ref_folio, zap_flags_t zap_flags)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	unsigned long address;
@@ -5910,8 +5910,8 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 		 * page is being unmapped, not a range. Ensure the page we
 		 * are about to unmap is the actual page of interest.
 		 */
-		if (ref_page) {
-			if (page != ref_page) {
+		if (ref_folio) {
+			if (page != folio_page(ref_folio, 0)) {
 				spin_unlock(ptl);
 				continue;
 			}
@@ -5977,7 +5977,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 		/*
 		 * Bail out after unmapping reference page if supplied
 		 */
-		if (ref_page)
+		if (ref_folio)
 			break;
 	}
 	tlb_end_vma(tlb, vma);
@@ -6052,7 +6052,7 @@ void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
 	tlb_gather_mmu(&tlb, vma->vm_mm);
 
 	__unmap_hugepage_range(&tlb, vma, start, end,
-			       folio_page(ref_folio, 0), zap_flags);
+			       ref_folio, zap_flags);
 
 	mmu_notifier_invalidate_range_end(&range);
 	tlb_finish_mmu(&tlb);
-- 
2.47.2

[PATCH v2 4/4] mm/hugetlb: Convert use of struct page to folio in __unmap_hugepage_range()

[nifan.cxl]

From: Fan Ni <fan.ni@samsung.com>

In __unmap_hugepage_range(), the "page" pointer always points to the
first page of a huge page, which guarantees there is a folio associating
with it.  Convert the "page" pointer to use folio.

Signed-off-by: Fan Ni <fan.ni@samsung.com>
---
This is a new patch added to the series based on the discussion here:
https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m2b9cc1743e1907e52658815b297b9d249474f387
---
 mm/hugetlb.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 7d280ab23784..8177a3fe47d7 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -5840,7 +5840,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 	pte_t *ptep;
 	pte_t pte;
 	spinlock_t *ptl;
-	struct page *page;
+	struct folio *folio;
 	struct hstate *h = hstate_vma(vma);
 	unsigned long sz = huge_page_size(h);
 	bool adjust_reservation = false;
@@ -5904,14 +5904,14 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 			continue;
 		}
 
-		page = pte_page(pte);
+		folio = page_folio(pte_page(pte));
 		/*
 		 * If a reference page is supplied, it is because a specific
 		 * page is being unmapped, not a range. Ensure the page we
 		 * are about to unmap is the actual page of interest.
 		 */
 		if (ref_folio) {
-			if (page != folio_page(ref_folio, 0)) {
+			if (folio != ref_folio) {
 				spin_unlock(ptl);
 				continue;
 			}
@@ -5926,7 +5926,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 		pte = huge_ptep_get_and_clear(mm, address, ptep, sz);
 		tlb_remove_huge_tlb_entry(h, tlb, ptep, address);
 		if (huge_pte_dirty(pte))
-			set_page_dirty(page);
+			folio_mark_dirty(folio);
 		/* Leave a uffd-wp pte marker if needed */
 		if (huge_pte_uffd_wp(pte) &&
 		    !(zap_flags & ZAP_FLAG_DROP_MARKER))
@@ -5934,7 +5934,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 					make_pte_marker(PTE_MARKER_UFFD_WP),
 					sz);
 		hugetlb_count_sub(pages_per_huge_page(h), mm);
-		hugetlb_remove_rmap(page_folio(page));
+		hugetlb_remove_rmap(folio);
 
 		/*
 		 * Restore the reservation for anonymous page, otherwise the
@@ -5943,8 +5943,8 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 		 * reservation bit.
 		 */
 		if (!h->surplus_huge_pages && __vma_private_lock(vma) &&
-		    folio_test_anon(page_folio(page))) {
-			folio_set_hugetlb_restore_reserve(page_folio(page));
+		    folio_test_anon(folio)) {
+			folio_set_hugetlb_restore_reserve(folio);
 			/* Reservation to be adjusted after the spin lock */
 			adjust_reservation = true;
 		}
@@ -5968,12 +5968,12 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 				 * count will not be incremented by free_huge_folio.
 				 * Act as if we consumed the reservation.
 				 */
-				folio_clear_hugetlb_restore_reserve(page_folio(page));
+				folio_clear_hugetlb_restore_reserve(folio);
 			else if (rc)
 				vma_add_reservation(h, vma, address);
 		}
 
-		tlb_remove_page_size(tlb, page, huge_page_size(h));
+		tlb_remove_page_size(tlb, folio_page(folio, 0), huge_page_size(h));
 		/*
 		 * Bail out after unmapping reference page if supplied
 		 */
-- 
2.47.2

Re: [PATCH v2 4/4] mm/hugetlb: Convert use of struct page to folio in __unmap_hugepage_range()

[Sidhartha Kumar]

On 4/18/25 12:57 PM, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> In __unmap_hugepage_range(), the "page" pointer always points to the
> first page of a huge page, which guarantees there is a folio associating
> with it.  Convert the "page" pointer to use folio.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> ---
> This is a new patch added to the series based on the discussion here:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m2b9cc1743e1907e52658815b297b9d249474f387
> ---
>   mm/hugetlb.c | 18 +++++++++---------
>   1 file changed, 9 insertions(+), 9 deletions(-)
> 
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 7d280ab23784..8177a3fe47d7 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -5840,7 +5840,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   	pte_t *ptep;
>   	pte_t pte;
>   	spinlock_t *ptl;
> -	struct page *page;
> +	struct folio *folio;
>   	struct hstate *h = hstate_vma(vma);
>   	unsigned long sz = huge_page_size(h);
>   	bool adjust_reservation = false;
> @@ -5904,14 +5904,14 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   			continue;
>   		}
>   
> -		page = pte_page(pte);
> +		folio = page_folio(pte_page(pte));
>   		/*
>   		 * If a reference page is supplied, it is because a specific
>   		 * page is being unmapped, not a range. Ensure the page we
>   		 * are about to unmap is the actual page of interest.
>   		 */
>   		if (ref_folio) {
> -			if (page != folio_page(ref_folio, 0)) {
> +			if (folio != ref_folio) {
>   				spin_unlock(ptl);
>   				continue;
>   			}
> @@ -5926,7 +5926,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   		pte = huge_ptep_get_and_clear(mm, address, ptep, sz);
>   		tlb_remove_huge_tlb_entry(h, tlb, ptep, address);
>   		if (huge_pte_dirty(pte))
> -			set_page_dirty(page);
> +			folio_mark_dirty(folio);
>   		/* Leave a uffd-wp pte marker if needed */
>   		if (huge_pte_uffd_wp(pte) &&
>   		    !(zap_flags & ZAP_FLAG_DROP_MARKER))
> @@ -5934,7 +5934,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   					make_pte_marker(PTE_MARKER_UFFD_WP),
>   					sz);
>   		hugetlb_count_sub(pages_per_huge_page(h), mm);
> -		hugetlb_remove_rmap(page_folio(page));
> +		hugetlb_remove_rmap(folio);
>   
>   		/*
>   		 * Restore the reservation for anonymous page, otherwise the
> @@ -5943,8 +5943,8 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   		 * reservation bit.
>   		 */
>   		if (!h->surplus_huge_pages && __vma_private_lock(vma) &&
> -		    folio_test_anon(page_folio(page))) {
> -			folio_set_hugetlb_restore_reserve(page_folio(page));
> +		    folio_test_anon(folio)) {
> +			folio_set_hugetlb_restore_reserve(folio);
>   			/* Reservation to be adjusted after the spin lock */
>   			adjust_reservation = true;
>   		}
> @@ -5968,12 +5968,12 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   				 * count will not be incremented by free_huge_folio.
>   				 * Act as if we consumed the reservation.
>   				 */
> -				folio_clear_hugetlb_restore_reserve(page_folio(page));
> +				folio_clear_hugetlb_restore_reserve(folio);
>   			else if (rc)
>   				vma_add_reservation(h, vma, address);
>   		}
>   
> -		tlb_remove_page_size(tlb, page, huge_page_size(h));
> +		tlb_remove_page_size(tlb, folio_page(folio, 0), huge_page_size(h));
>   		/*
>   		 * Bail out after unmapping reference page if supplied
>   		 */
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>

Re: [PATCH v2 1/4] mm/hugetlb: Refactor unmap_ref_private() to take folio instead of page

[David Hildenbrand]

On 18.04.25 18:57, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> The function unmap_ref_private() has only user, which passes in
> &folio->page. Let it take folio directly.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> Reviewed-by: Muchun Song <muchun.song@linux.dev>
> Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
> ---
> v2:
> Picked up Reviewed-by tags;
> 
> v1:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#mbf9f3e8b49497755b414e1887b2376b3902ffb76
> ---

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 2/4] mm/hugetlb: Refactor unmap_hugepage_range() to take folio instead of page

[David Hildenbrand]

On 18.04.25 18:57, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> The function unmap_hugepage_range() has two kinds of users:
> 1) unmap_ref_private(), which passes in the head page of a folio.  Since
>     unmap_ref_private() already takes folio and there are no other uses
>     of the folio struct in the function, it is natural for
>     unmap_hugepage_range() to take folio also.
> 2) All other uses, which pass in NULL pointer.
> 
> In both cases, we can pass in folio. Refactor unmap_hugepage_range() to
> take folio.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> Reviewed-by: Muchun Song <muchun.song@linux.dev>
> Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
> ---
> v2:
> Picked up Reviewed-by tags;
> 
> v1:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m7ba630d783e08ae708453279579093d481c6e721
> ---
>   include/linux/hugetlb.h | 2 +-
>   mm/hugetlb.c            | 7 ++++---
>   2 files changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> index a57bed83c657..b7699f35c87f 100644
> --- a/include/linux/hugetlb.h
> +++ b/include/linux/hugetlb.h
> @@ -128,7 +128,7 @@ int move_hugetlb_page_tables(struct vm_area_struct *vma,
>   int copy_hugetlb_page_range(struct mm_struct *, struct mm_struct *,
>   			    struct vm_area_struct *, struct vm_area_struct *);
>   void unmap_hugepage_range(struct vm_area_struct *,
> -			  unsigned long, unsigned long, struct page *,
> +			  unsigned long, unsigned long, struct folio *folio,
>   			  zap_flags_t);

Note that no parameters are using names. While I am not a big fan of 
that in general, we should keep it consistent.

Either name all or none.


Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 2/4] mm/hugetlb: Refactor unmap_hugepage_range() to take folio instead of page

[David Hildenbrand]

On 18.04.25 18:57, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> The function unmap_hugepage_range() has two kinds of users:
> 1) unmap_ref_private(), which passes in the head page of a folio.  Since
>     unmap_ref_private() already takes folio and there are no other uses
>     of the folio struct in the function, it is natural for
>     unmap_hugepage_range() to take folio also.
> 2) All other uses, which pass in NULL pointer.
> 
> In both cases, we can pass in folio. Refactor unmap_hugepage_range() to
> take folio.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> Reviewed-by: Muchun Song <muchun.song@linux.dev>
> Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
> ---
> v2:
> Picked up Reviewed-by tags;
> 
> v1:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m7ba630d783e08ae708453279579093d481c6e721
> ---
>   include/linux/hugetlb.h | 2 +-
>   mm/hugetlb.c            | 7 ++++---
>   2 files changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> index a57bed83c657..b7699f35c87f 100644
> --- a/include/linux/hugetlb.h
> +++ b/include/linux/hugetlb.h
> @@ -128,7 +128,7 @@ int move_hugetlb_page_tables(struct vm_area_struct *vma,
>   int copy_hugetlb_page_range(struct mm_struct *, struct mm_struct *,
>   			    struct vm_area_struct *, struct vm_area_struct *);
>   void unmap_hugepage_range(struct vm_area_struct *,
> -			  unsigned long, unsigned long, struct page *,
> +			  unsigned long, unsigned long, struct folio *folio,
>   			  zap_flags_t);
>   void __unmap_hugepage_range(struct mmu_gather *tlb,
>   			  struct vm_area_struct *vma,
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index b5d1ac8290a7..3181dbe0c4bb 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -6039,7 +6039,7 @@ void __hugetlb_zap_end(struct vm_area_struct *vma,
>   }
>   
>   void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
> -			  unsigned long end, struct page *ref_page,
> +			  unsigned long end, struct folio *ref_folio,
>   			  zap_flags_t zap_flags)

Oh, and now I realize that you call it "folio" above and "ref_folio" here.

Can we just rename it to "folio" consistently or is there a reason to 
use the weird "ref_" prefix?

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 3/4] mm/hugetlb: Refactor __unmap_hugepage_range() to take folio instead of page

[David Hildenbrand]

On 18.04.25 18:57, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> The function __unmap_hugepage_range() has two kinds of users:
> 1) unmap_hugepage_range(), which passes in the head page of a folio.
>     Since unmap_hugepage_range() already takes folio and there are no other
>     uses of the folio struct in the function, it is natural for
>     __unmap_hugepage_range() to take folio also.
> 2) All other uses, which pass in NULL pointer.
> 
> In both cases, we can pass in folio. Refactor __unmap_hugepage_range() to
> take folio.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> ---
> v2: fixed issue mentioned here:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m2b9cc1743e1907e52658815b297b9d249474f387
> 
> v1:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m738e9e7f7d7fe4aab6b21782b4658dd65abf8fc4
> ---
> 
>   include/linux/hugetlb.h |  4 ++--
>   mm/hugetlb.c            | 10 +++++-----
>   2 files changed, 7 insertions(+), 7 deletions(-)
> 
> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> index b7699f35c87f..ebaf95231934 100644
> --- a/include/linux/hugetlb.h
> +++ b/include/linux/hugetlb.h
> @@ -133,7 +133,7 @@ void unmap_hugepage_range(struct vm_area_struct *,
>   void __unmap_hugepage_range(struct mmu_gather *tlb,
>   			  struct vm_area_struct *vma,
>   			  unsigned long start, unsigned long end,
> -			  struct page *ref_page, zap_flags_t zap_flags);
> +			  struct folio *ref_folio, zap_flags_t zap_flags);
>   void hugetlb_report_meminfo(struct seq_file *);
>   int hugetlb_report_node_meminfo(char *buf, int len, int nid);
>   void hugetlb_show_meminfo_node(int nid);
> @@ -452,7 +452,7 @@ static inline long hugetlb_change_protection(
>   
>   static inline void __unmap_hugepage_range(struct mmu_gather *tlb,
>   			struct vm_area_struct *vma, unsigned long start,
> -			unsigned long end, struct page *ref_page,
> +			unsigned long end, struct folio *ref_folio,
>   			zap_flags_t zap_flags)
>   {

Same comment as for the previous patch, can we just call this "folio" 
instead of "ref_folio" if we are already touching it?


-- 
Cheers,

David / dhildenb

Re: [PATCH v2 4/4] mm/hugetlb: Convert use of struct page to folio in __unmap_hugepage_range()

[David Hildenbrand]

On 18.04.25 18:57, nifan.cxl@gmail.com wrote:
> From: Fan Ni <fan.ni@samsung.com>
> 
> In __unmap_hugepage_range(), the "page" pointer always points to the
> first page of a huge page, which guarantees there is a folio associating
> with it.  Convert the "page" pointer to use folio.
> 
> Signed-off-by: Fan Ni <fan.ni@samsung.com>
> ---
> This is a new patch added to the series based on the discussion here:
> https://lore.kernel.org/linux-mm/aAHUluy7T32ZlYg7@debian/T/#m2b9cc1743e1907e52658815b297b9d249474f387
> ---
>   mm/hugetlb.c | 18 +++++++++---------
>   1 file changed, 9 insertions(+), 9 deletions(-)
> 
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 7d280ab23784..8177a3fe47d7 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -5840,7 +5840,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   	pte_t *ptep;
>   	pte_t pte;
>   	spinlock_t *ptl;
> -	struct page *page;
> +	struct folio *folio;
>   	struct hstate *h = hstate_vma(vma);
>   	unsigned long sz = huge_page_size(h);
>   	bool adjust_reservation = false;
> @@ -5904,14 +5904,14 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>   			continue;
>   		}
>   
> -		page = pte_page(pte);
> +		folio = page_folio(pte_page(pte));
>   		/*
>   		 * If a reference page is supplied, it is because a specific
>   		 * page is being unmapped, not a range. Ensure the page we
>   		 * are about to unmap is the actual page of interest.
>   		 */
>   		if (ref_folio) {
> -			if (page != folio_page(ref_folio, 0)) {
> +			if (folio != ref_folio) {
>   				spin_unlock(ptl);
>   				continue;
>   			}

What about something like (keeping in mind that I suggest renaming 
"ref_folio" -> "folio" in previous patches)


const bool folio_provided = !!folio;

...

if (folio_provided) {
	if (folio != page_folio(pte_page(pte))) {
		spin_unlock(ptl);
		continue;
	}
	...
} else {
	folio = page_folio(pte_page(pte);
}

...

if (folio_supplied)
	break;
...

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 2/4] mm/hugetlb: Refactor unmap_hugepage_range() to take folio instead of page

[Matthew Wilcox]

On Fri, Apr 18, 2025 at 09:57:40AM -0700, nifan.cxl@gmail.com wrote:
>  void unmap_hugepage_range(struct vm_area_struct *,
> -			  unsigned long, unsigned long, struct page *,
> +			  unsigned long, unsigned long, struct folio *folio,

I'm fine with leaving the vma and folio unnamed, but it is a crime
against our fellow programmers to leave the two 'unsigned long's
unnamed.  What the hell are they?

>  void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
> -			  unsigned long end, struct page *ref_page,
> +			  unsigned long end, struct folio *ref_folio,

... start and end.  I'd happily see a patch which only named those
parameters and left the struct folio unnamed.

> -	__unmap_hugepage_range(&tlb, vma, start, end, ref_page, zap_flags);
> +	__unmap_hugepage_range(&tlb, vma, start, end,
> +			       folio_page(ref_folio, 0), zap_flags);

I do not like this.  Why should we pass in the first page here?  It
seems to me that this is just "Now we will call a function which still
takes a struct page", and we *SHOULD* use &folio->page here to indicate
that we just haven't done the conversion yet.

Re: [PATCH v2 1/4] mm/hugetlb: Refactor unmap_ref_private() to take folio instead of page

[Matthew Wilcox]

With regard to the Subject: line, you're not "refactoring".  You're
passing a folio to the function.

> @@ -6108,7 +6108,8 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
>  		 */
>  		if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
>  			unmap_hugepage_range(iter_vma, address,
> -					     address + huge_page_size(h), page, 0);
> +					     address + huge_page_size(h),
> +					     folio_page(folio, 0), 0);

As previously, use &folio->page here.

Re: [PATCH v2 3/4] mm/hugetlb: Refactor __unmap_hugepage_range() to take folio instead of page

[Matthew Wilcox]

On Fri, Apr 18, 2025 at 09:57:41AM -0700, nifan.cxl@gmail.com wrote:
> @@ -5910,8 +5910,8 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
>  		 * page is being unmapped, not a range. Ensure the page we
>  		 * are about to unmap is the actual page of interest.
>  		 */
> -		if (ref_page) {
> -			if (page != ref_page) {
> +		if (ref_folio) {
> +			if (page != folio_page(ref_folio, 0)) {

I think this is actually wrong.  Shouldn't it be:

			if (page_folio(page) != ref_folio)

so that it doesn't matter which page in the folio we found; we just need
to know if the page belongs to the same folio or not?

Re: [PATCH v2 4/4] mm/hugetlb: Convert use of struct page to folio in __unmap_hugepage_range()

[Matthew Wilcox]

On Fri, Apr 18, 2025 at 09:57:42AM -0700, nifan.cxl@gmail.com wrote:
>  
> -		tlb_remove_page_size(tlb, page, huge_page_size(h));
> +		tlb_remove_page_size(tlb, folio_page(folio, 0), huge_page_size(h));

Surely this should be:
		tlb_remove_page_size(tlb, folio_page(folio, 0), folio_size(folio));

Re: [PATCH v2 4/4] mm/hugetlb: Convert use of struct page to folio in __unmap_hugepage_range()

[Andrew Morton]

On Wed, 23 Apr 2025 04:22:42 +0100 Matthew Wilcox <willy@infradead.org> wrote:

> On Fri, Apr 18, 2025 at 09:57:42AM -0700, nifan.cxl@gmail.com wrote:
> >  
> > -		tlb_remove_page_size(tlb, page, huge_page_size(h));
> > +		tlb_remove_page_size(tlb, folio_page(folio, 0), huge_page_size(h));
> 
> Surely this should be:
> 		tlb_remove_page_size(tlb, folio_page(folio, 0), folio_size(folio));

Thanks, I removed this version of the series from mm.git's mm-new branch.

Re: [PATCH v2 1/4] mm/hugetlb: Refactor unmap_ref_private() to take folio instead of page

[Fan Ni]

On Wed, Apr 23, 2025 at 04:17:25AM +0100, Matthew Wilcox wrote:
> 
> With regard to the Subject: line, you're not "refactoring".  You're
> passing a folio to the function.

Hi Matthew,

Thanks for the review. 

Are you suggesting something like "Passing folio instead of page to unmap_ref_private()"?

We do not change the behavior of the function, but only the interface, it
is kind of "refactoring" to me.

Fan

> 
> > @@ -6108,7 +6108,8 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
> >  		 */
> >  		if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
> >  			unmap_hugepage_range(iter_vma, address,
> > -					     address + huge_page_size(h), page, 0);
> > +					     address + huge_page_size(h),
> > +					     folio_page(folio, 0), 0);
> 
> As previously, use &folio->page here.
> 
>