Marking PUDs that map a "normal" refcounted folios as special is
against our rules documented for vm_normal_page(). normal (refcounted)
folios shall never have the page table mapping marked as special.

Fortunately, there are not that many pud_special() check that can be
mislead and are right now rather harmless: e.g., none so far
bases decisions whether to grab a folio reference on that decision.

Well, and GUP-fast will fallback to GUP-slow. All in all, so far no big
implications as it seems.

Getting this right will get more important as we introduce
folio_normal_page_pud() and start using it in more place where we
currently special-case based on other VMA flags.

Fix it just like we fixed vmf_insert_folio_pmd().

Add folio_mk_pud() to mimic what we do with folio_mk_pmd().

Fixes: dbe54153296d ("mm/huge_memory: add vmf_insert_folio_pud()")
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/mm.h | 19 ++++++++++++++++-
 mm/huge_memory.c   | 52 ++++++++++++++++++++++++++--------------------
 2 files changed, 47 insertions(+), 24 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index fa538feaa8d95..912b6d40a12d6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1816,7 +1816,24 @@ static inline pmd_t folio_mk_pmd(struct folio *folio, pgprot_t pgprot)
 {
 	return pmd_mkhuge(pfn_pmd(folio_pfn(folio), pgprot));
 }
-#endif
+
+#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
+/**
+ * folio_mk_pud - Create a PUD for this folio
+ * @folio: The folio to create a PUD for
+ * @pgprot: The page protection bits to use
+ *
+ * Create a page table entry for the first page of this folio.
+ * This is suitable for passing to set_pud_at().
+ *
+ * Return: A page table entry suitable for mapping this folio.
+ */
+static inline pud_t folio_mk_pud(struct folio *folio, pgprot_t pgprot)
+{
+	return pud_mkhuge(pfn_pud(folio_pfn(folio), pgprot));
+}
+#endif /* CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD */
+#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif /* CONFIG_MMU */
 
 static inline bool folio_has_pincount(const struct folio *folio)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index d1e3e253c714a..bbc1dab98f2f7 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1536,15 +1536,18 @@ static pud_t maybe_pud_mkwrite(pud_t pud, struct vm_area_struct *vma)
 	return pud;
 }
 
-static void insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
-		pud_t *pud, pfn_t pfn, pgprot_t prot, bool write)
+static void insert_pud(struct vm_area_struct *vma, unsigned long addr,
+		pud_t *pud, struct folio_or_pfn fop, pgprot_t prot, bool write)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	pud_t entry;
 
 	if (!pud_none(*pud)) {
+		const unsigned long pfn = fop.is_folio ? folio_pfn(fop.folio) :
+					  pfn_t_to_pfn(fop.pfn);
+
 		if (write) {
-			if (WARN_ON_ONCE(pud_pfn(*pud) != pfn_t_to_pfn(pfn)))
+			if (WARN_ON_ONCE(pud_pfn(*pud) != pfn))
 				return;
 			entry = pud_mkyoung(*pud);
 			entry = maybe_pud_mkwrite(pud_mkdirty(entry), vma);
@@ -1554,11 +1557,20 @@ static void insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
 		return;
 	}
 
-	entry = pud_mkhuge(pfn_t_pud(pfn, prot));
-	if (pfn_t_devmap(pfn))
-		entry = pud_mkdevmap(entry);
-	else
-		entry = pud_mkspecial(entry);
+	if (fop.is_folio) {
+		entry = folio_mk_pud(fop.folio, vma->vm_page_prot);
+
+		folio_get(fop.folio);
+		folio_add_file_rmap_pud(fop.folio, &fop.folio->page, vma);
+		add_mm_counter(mm, mm_counter_file(fop.folio), HPAGE_PUD_NR);
+	} else {
+		entry = pud_mkhuge(pfn_t_pud(fop.pfn, prot));
+
+		if (pfn_t_devmap(fop.pfn))
+			entry = pud_mkdevmap(entry);
+		else
+			entry = pud_mkspecial(entry);
+	}
 	if (write) {
 		entry = pud_mkyoung(pud_mkdirty(entry));
 		entry = maybe_pud_mkwrite(entry, vma);
@@ -1582,6 +1594,9 @@ vm_fault_t vmf_insert_pfn_pud(struct vm_fault *vmf, pfn_t pfn, bool write)
 	unsigned long addr = vmf->address & PUD_MASK;
 	struct vm_area_struct *vma = vmf->vma;
 	pgprot_t pgprot = vma->vm_page_prot;
+	struct folio_or_pfn fop = {
+		.pfn = pfn,
+	};
 	spinlock_t *ptl;
 
 	/*
@@ -1601,7 +1616,7 @@ vm_fault_t vmf_insert_pfn_pud(struct vm_fault *vmf, pfn_t pfn, bool write)
 	pfnmap_setup_cachemode_pfn(pfn_t_to_pfn(pfn), &pgprot);
 
 	ptl = pud_lock(vma->vm_mm, vmf->pud);
-	insert_pfn_pud(vma, addr, vmf->pud, pfn, pgprot, write);
+	insert_pud(vma, addr, vmf->pud, fop, pgprot, write);
 	spin_unlock(ptl);
 
 	return VM_FAULT_NOPAGE;
@@ -1623,6 +1638,10 @@ vm_fault_t vmf_insert_folio_pud(struct vm_fault *vmf, struct folio *folio,
 	unsigned long addr = vmf->address & PUD_MASK;
 	pud_t *pud = vmf->pud;
 	struct mm_struct *mm = vma->vm_mm;
+	struct folio_or_pfn fop = {
+		.folio = folio,
+		.is_folio = true,
+	};
 	spinlock_t *ptl;
 
 	if (addr < vma->vm_start || addr >= vma->vm_end)
@@ -1632,20 +1651,7 @@ vm_fault_t vmf_insert_folio_pud(struct vm_fault *vmf, struct folio *folio,
 		return VM_FAULT_SIGBUS;
 
 	ptl = pud_lock(mm, pud);
-
-	/*
-	 * If there is already an entry present we assume the folio is
-	 * already mapped, hence no need to take another reference. We
-	 * still call insert_pfn_pud() though in case the mapping needs
-	 * upgrading to writeable.
-	 */
-	if (pud_none(*vmf->pud)) {
-		folio_get(folio);
-		folio_add_file_rmap_pud(folio, &folio->page, vma);
-		add_mm_counter(mm, mm_counter_file(folio), HPAGE_PUD_NR);
-	}
-	insert_pfn_pud(vma, addr, vmf->pud, pfn_to_pfn_t(folio_pfn(folio)),
-		       vma->vm_page_prot, write);
+	insert_pud(vma, addr, vmf->pud, fop, vma->vm_page_prot, write);
 	spin_unlock(ptl);
 
 	return VM_FAULT_NOPAGE;
-- 
2.49.0

On Fri, Jun 13, 2025 at 11:27:02AM +0200, David Hildenbrand wrote:
> Marking PUDs that map a "normal" refcounted folios as special is
> against our rules documented for vm_normal_page(). normal (refcounted)
> folios shall never have the page table mapping marked as special.
> 
> Fortunately, there are not that many pud_special() check that can be
> mislead and are right now rather harmless: e.g., none so far
> bases decisions whether to grab a folio reference on that decision.
> 
> Well, and GUP-fast will fallback to GUP-slow. All in all, so far no big
> implications as it seems.
> 
> Getting this right will get more important as we introduce
> folio_normal_page_pud() and start using it in more place where we
> currently special-case based on other VMA flags.
> 
> Fix it just like we fixed vmf_insert_folio_pmd().
> 
> Add folio_mk_pud() to mimic what we do with folio_mk_pmd().
> 
> Fixes: dbe54153296d ("mm/huge_memory: add vmf_insert_folio_pud()")
> Reviewed-by: Dan Williams <dan.j.williams@intel.com>
> Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
> Tested-by: Dan Williams <dan.j.williams@intel.com>
> Signed-off-by: David Hildenbrand <david@redhat.com>

Reviewed-by: Oscar Salvador <osalvador@suse.de>



-- 
Oscar Salvador
SUSE Labs