mm: pgtable: introduce pagetable_dtor()

The pagetable_p*_dtor() are exactly the same except for the handling of
ptlock. If we make ptlock_free() handle the case where ptdesc->ptl is
NULL and remove VM_BUG_ON_PAGE() from pmd_ptlock_free(), we can unify
pagetable_p*_dtor() into one function. Let's introduce pagetable_dtor()
to do this.

Later, pagetable_dtor() will be moved to tlb_remove_ptdesc(), so that
ptlock and page table pages can be freed together (regardless of whether
RCU is used). This prevents the use-after-free problem where the ptlock
is freed immediately but the page table pages is freed later via RCU.

Link: https://lkml.kernel.org/r/47f4[e-mail]
Signed-off-by: Qi Zheng <[e-mail]>
Originally-by: Peter Zijlstra (Intel) <[e-mail]>
Reviewed-by: Kevin Brodsky <[e-mail]>
Acked-by: Alexander Gordeev <[e-mail]> [s390]
Cc: Alexandre Ghiti <[e-mail]>
Cc: Alexandre Ghiti <[e-mail]>
Cc: Andreas Larsson <[e-mail]>
Cc: Aneesh Kumar K.V (Arm) <[e-mail]>
Cc: Arnd Bergmann <[e-mail]>
Cc: Dave Hansen <[e-mail]>
Cc: David Hildenbrand <[e-mail]>
Cc: David Rientjes <[e-mail]>
Cc: Hugh Dickins <[e-mail]>
Cc: Jann Horn <[e-mail]>
Cc: Lorenzo Stoakes <[e-mail]>
Cc: Matthew Wilcox (Oracle) <[e-mail]>
Cc: Mike Rapoport (Microsoft) <[e-mail]>
Cc: Muchun Song <[e-mail]>
Cc: Nicholas Piggin <[e-mail]>
Cc: Palmer Dabbelt <[e-mail]>
Cc: Ryan Roberts <[e-mail]>
Cc: Thomas Gleixner <[e-mail]>
Cc: Vishal Moola (Oracle) <[e-mail]>
Cc: Will Deacon <[e-mail]>
Cc: Yu Zhao <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>

Documentation/mm/split_page_table_lock.rst

···
 ===================================================
 
 There's no need in special enabling of PTE split page table lock: everything
-required is done by pagetable_pte_ctor() and pagetable_pte_dtor(), which
+required is done by pagetable_pte_ctor() and pagetable_dtor(), which
 must be called on PTE table allocation / freeing.
 
 Make sure the architecture doesn't use slab allocator for page table
···
 levels.
 
 PMD split lock enabling requires pagetable_pmd_ctor() call on PMD table
-allocation and pagetable_pmd_dtor() on freeing.
+allocation and pagetable_dtor() on freeing.
 
 Allocation usually happens in pmd_alloc_one(), freeing in pmd_free() and
 pmd_free_tlb(), but make sure you cover all PMD table allocation / freeing

arch/arm/include/asm/tlb.h

···
 {
 	struct ptdesc *ptdesc = page_ptdesc(pte);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 
 #ifndef CONFIG_ARM_LPAE
 	/*
···
 #ifdef CONFIG_ARM_LPAE
 	struct ptdesc *ptdesc = virt_to_ptdesc(pmdp);
 
-	pagetable_pmd_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	tlb_remove_ptdesc(tlb, ptdesc);
 #endif
 }

arch/arm64/include/asm/tlb.h

···
 {
 	struct ptdesc *ptdesc = page_ptdesc(pte);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	tlb_remove_ptdesc(tlb, ptdesc);
 }
 
···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pmdp);
 
-	pagetable_pmd_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	tlb_remove_ptdesc(tlb, ptdesc);
 }
 #endif
···
 	if (!pgtable_l4_enabled())
 		return;
 
-	pagetable_pud_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	tlb_remove_ptdesc(tlb, ptdesc);
 }
 #endif
···
 	if (!pgtable_l5_enabled())
 		return;
 
-	pagetable_p4d_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	tlb_remove_ptdesc(tlb, ptdesc);
 }
 #endif

arch/csky/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, address)		\
 do {							\
-	pagetable_pte_dtor(page_ptdesc(pte));		\
+	pagetable_dtor(page_ptdesc(pte));		\
 	tlb_remove_page_ptdesc(tlb, page_ptdesc(pte));	\
 } while (0)
 

arch/hexagon/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, addr)				\
 do {								\
-	pagetable_pte_dtor((page_ptdesc(pte)));			\
+	pagetable_dtor((page_ptdesc(pte)));			\
 	tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));	\
 } while (0)
 

arch/loongarch/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, address)			\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), page_ptdesc(pte));	\
 } while (0)
 

arch/m68k/include/asm/mcf_pgalloc.h

···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pgtable);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 
···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pgtable);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 

arch/m68k/include/asm/sun3_pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, addr)				\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), page_ptdesc(pte));	\
 } while (0)
 

arch/m68k/mm/motorola.c

···
 		list_del(dp);
 		mmu_page_dtor((void *)page);
 		if (type == TABLE_PTE)
-			pagetable_pte_dtor(virt_to_ptdesc((void *)page));
+			pagetable_dtor(virt_to_ptdesc((void *)page));
 		free_page (page);
 		return 1;
 	} else if (ptable_list[type].next != dp) {

arch/mips/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, address)			\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), page_ptdesc(pte));	\
 } while (0)
 

arch/nios2/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, addr)					\
 	do {								\
-		pagetable_pte_dtor(page_ptdesc(pte));			\
+		pagetable_dtor(page_ptdesc(pte));			\
 		tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));	\
 	} while (0)
 

arch/openrisc/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, addr)				\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));	\
 } while (0)
 

arch/powerpc/mm/book3s64/mmu_context.c

···
 	count = ((unsigned long)pmd_frag & ~PAGE_MASK) >> PMD_FRAG_SIZE_SHIFT;
 	/* We allow PTE_FRAG_NR fragments from a PTE page */
 	if (atomic_sub_and_test(PMD_FRAG_NR - count, &ptdesc->pt_frag_refcount)) {
-		pagetable_pmd_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 		pagetable_free(ptdesc);
 	}
 }

arch/powerpc/mm/book3s64/pgtable.c

···
 
 	BUG_ON(atomic_read(&ptdesc->pt_frag_refcount) <= 0);
 	if (atomic_dec_and_test(&ptdesc->pt_frag_refcount)) {
-		pagetable_pmd_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 		pagetable_free(ptdesc);
 	}
 }

arch/powerpc/mm/pgtable-frag.c

···
 	count = ((unsigned long)pte_frag & ~PAGE_MASK) >> PTE_FRAG_SIZE_SHIFT;
 	/* We allow PTE_FRAG_NR fragments from a PTE page */
 	if (atomic_sub_and_test(PTE_FRAG_NR - count, &ptdesc->pt_frag_refcount)) {
-		pagetable_pte_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 		pagetable_free(ptdesc);
 	}
 }
···
 	struct ptdesc *ptdesc;
 
 	ptdesc = container_of(head, struct ptdesc, pt_rcu_head);
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 

arch/riscv/include/asm/pgalloc.h

···
 	if (pgtable_l4_enabled) {
 		struct ptdesc *ptdesc = virt_to_ptdesc(pud);
 
-		pagetable_pud_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 		riscv_tlb_remove_ptdesc(tlb, ptdesc);
 	}
 }
···
 	if (pgtable_l5_enabled) {
 		struct ptdesc *ptdesc = virt_to_ptdesc(p4d);
 
-		pagetable_p4d_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 		riscv_tlb_remove_ptdesc(tlb, virt_to_ptdesc(p4d));
 	}
 }
···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pmd);
 
-	pagetable_pmd_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	riscv_tlb_remove_ptdesc(tlb, ptdesc);
 }
 
···
 {
 	struct ptdesc *ptdesc = page_ptdesc(pte);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	riscv_tlb_remove_ptdesc(tlb, ptdesc);
 }
 #endif /* CONFIG_MMU */

arch/riscv/mm/init.c

···
 			return;
 	}
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	if (PageReserved(page))
 		free_reserved_page(page);
 	else
···
 	}
 
 	if (!is_vmemmap)
-		pagetable_pmd_dtor(ptdesc);
+		pagetable_dtor(ptdesc);
 	if (PageReserved(page))
 		free_reserved_page(page);
 	else

arch/s390/include/asm/pgalloc.h

···
 	if (mm_p4d_folded(mm))
 		return;
 
-	pagetable_p4d_dtor(virt_to_ptdesc(p4d));
+	pagetable_dtor(virt_to_ptdesc(p4d));
 	crst_table_free(mm, (unsigned long *) p4d);
 }
 
···
 	if (mm_pud_folded(mm))
 		return;
 
-	pagetable_pud_dtor(virt_to_ptdesc(pud));
+	pagetable_dtor(virt_to_ptdesc(pud));
 	crst_table_free(mm, (unsigned long *) pud);
 }
 
···
 {
 	if (mm_pmd_folded(mm))
 		return;
-	pagetable_pmd_dtor(virt_to_ptdesc(pmd));
+	pagetable_dtor(virt_to_ptdesc(pmd));
 	crst_table_free(mm, (unsigned long *) pmd);
 }
 

arch/s390/include/asm/tlb.h

···
 {
 	if (mm_pmd_folded(tlb->mm))
 		return;
-	pagetable_pmd_dtor(virt_to_ptdesc(pmd));
+	pagetable_dtor(virt_to_ptdesc(pmd));
 	__tlb_adjust_range(tlb, address, PAGE_SIZE);
 	tlb->mm->context.flush_mm = 1;
 	tlb->freed_tables = 1;
···
 {
 	if (mm_p4d_folded(tlb->mm))
 		return;
-	pagetable_p4d_dtor(virt_to_ptdesc(p4d));
+	pagetable_dtor(virt_to_ptdesc(p4d));
 	__tlb_adjust_range(tlb, address, PAGE_SIZE);
 	tlb->mm->context.flush_mm = 1;
 	tlb->freed_tables = 1;
···
 {
 	if (mm_pud_folded(tlb->mm))
 		return;
-	pagetable_pud_dtor(virt_to_ptdesc(pud));
+	pagetable_dtor(virt_to_ptdesc(pud));
 	tlb->mm->context.flush_mm = 1;
 	tlb->freed_tables = 1;
 	tlb->cleared_p4ds = 1;

arch/s390/mm/pgalloc.c

···
 
 static void pagetable_pte_dtor_free(struct ptdesc *ptdesc)
 {
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 

arch/sh/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, addr)				\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));	\
 } while (0)
 

arch/sparc/mm/init_64.c

···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pte);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 

arch/sparc/mm/srmmu.c

···
 	page = pfn_to_page(__nocache_pa((unsigned long)ptep) >> PAGE_SHIFT);
 	spin_lock(&mm->page_table_lock);
 	if (page_ref_dec_return(page) == 1)
-		pagetable_pte_dtor(page_ptdesc(page));
+		pagetable_dtor(page_ptdesc(page));
 	spin_unlock(&mm->page_table_lock);
 
 	srmmu_free_nocache(ptep, SRMMU_PTE_TABLE_SIZE);

arch/um/include/asm/pgalloc.h

···
 
 #define __pte_free_tlb(tlb, pte, address)			\
 do {								\
-	pagetable_pte_dtor(page_ptdesc(pte));			\
+	pagetable_dtor(page_ptdesc(pte));			\
 	tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));	\
 } while (0)
 
···
 
 #define __pmd_free_tlb(tlb, pmd, address)			\
 do {								\
-	pagetable_pmd_dtor(virt_to_ptdesc(pmd));			\
+	pagetable_dtor(virt_to_ptdesc(pmd));			\
 	tlb_remove_page_ptdesc((tlb), virt_to_ptdesc(pmd));	\
 } while (0)
 
···
 
 #define __pud_free_tlb(tlb, pud, address)			\
 do {								\
-	pagetable_pud_dtor(virt_to_ptdesc(pud));		\
+	pagetable_dtor(virt_to_ptdesc(pud));		\
 	tlb_remove_page_ptdesc((tlb), virt_to_ptdesc(pud));	\
 } while (0)
 

arch/x86/mm/pgtable.c

···
 
 void ___pte_free_tlb(struct mmu_gather *tlb, struct page *pte)
 {
-	pagetable_pte_dtor(page_ptdesc(pte));
+	pagetable_dtor(page_ptdesc(pte));
 	paravirt_release_pte(page_to_pfn(pte));
 	paravirt_tlb_remove_table(tlb, pte);
 }
···
 #ifdef CONFIG_X86_PAE
 	tlb->need_flush_all = 1;
 #endif
-	pagetable_pmd_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	paravirt_tlb_remove_table(tlb, ptdesc_page(ptdesc));
 }
 
···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pud);
 
-	pagetable_pud_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	paravirt_release_pud(__pa(pud) >> PAGE_SHIFT);
 	paravirt_tlb_remove_table(tlb, virt_to_page(pud));
 }
···
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(p4d);
 
-	pagetable_p4d_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	paravirt_release_p4d(__pa(p4d) >> PAGE_SHIFT);
 	paravirt_tlb_remove_table(tlb, virt_to_page(p4d));
 }
···
 		if (pmds[i]) {
 			ptdesc = virt_to_ptdesc(pmds[i]);
 
-			pagetable_pmd_dtor(ptdesc);
+			pagetable_dtor(ptdesc);
 			pagetable_free(ptdesc);
 			mm_dec_nr_pmds(mm);
 		}
···
 
 	free_page((unsigned long)pmd_sv);
 
-	pagetable_pmd_dtor(virt_to_ptdesc(pmd));
+	pagetable_dtor(virt_to_ptdesc(pmd));
 	free_page((unsigned long)pmd);
 
 	return 1;

include/asm-generic/pgalloc.h

···
 {
 	struct ptdesc *ptdesc = page_ptdesc(pte_page);
 
-	pagetable_pte_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 
···
 	struct ptdesc *ptdesc = virt_to_ptdesc(pmd);
 
 	BUG_ON((unsigned long)pmd & (PAGE_SIZE-1));
-	pagetable_pmd_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 #endif
···
 	struct ptdesc *ptdesc = virt_to_ptdesc(pud);
 
 	BUG_ON((unsigned long)pud & (PAGE_SIZE-1));
-	pagetable_pud_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 
···
 	struct ptdesc *ptdesc = virt_to_ptdesc(p4d);
 
 	BUG_ON((unsigned long)p4d & (PAGE_SIZE-1));
-	pagetable_p4d_dtor(ptdesc);
+	pagetable_dtor(ptdesc);
 	pagetable_free(ptdesc);
 }
 

include/linux/mm.h

···
 static inline void ptlock_free(struct ptdesc *ptdesc) {}
 #endif /* defined(CONFIG_SPLIT_PTE_PTLOCKS) */
 
+static inline void pagetable_dtor(struct ptdesc *ptdesc)
+{
+	struct folio *folio = ptdesc_folio(ptdesc);
+
+	ptlock_free(ptdesc);
+	__folio_clear_pgtable(folio);
+	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
+}
+
 static inline bool pagetable_pte_ctor(struct ptdesc *ptdesc)
 {
 	struct folio *folio = ptdesc_folio(ptdesc);
···
 	__folio_set_pgtable(folio);
 	lruvec_stat_add_folio(folio, NR_PAGETABLE);
 	return true;
-}
-
-static inline void pagetable_pte_dtor(struct ptdesc *ptdesc)
-{
-	struct folio *folio = ptdesc_folio(ptdesc);
-
-	ptlock_free(ptdesc);
-	__folio_clear_pgtable(folio);
-	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
 }
 
 pte_t *___pte_offset_map(pmd_t *pmd, unsigned long addr, pmd_t *pmdvalp);
···
 	return ptlock_init(ptdesc);
 }
 
-static inline void pmd_ptlock_free(struct ptdesc *ptdesc)
-{
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
-	VM_BUG_ON_PAGE(ptdesc->pmd_huge_pte, ptdesc_page(ptdesc));
-#endif
-	ptlock_free(ptdesc);
-}
-
 #define pmd_huge_pte(mm, pmd) (pmd_ptdesc(pmd)->pmd_huge_pte)
 
 #else
···
 }
 
 static inline bool pmd_ptlock_init(struct ptdesc *ptdesc) { return true; }
-static inline void pmd_ptlock_free(struct ptdesc *ptdesc) {}
 
 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
 
···
 	ptdesc_pmd_pts_init(ptdesc);
 	lruvec_stat_add_folio(folio, NR_PAGETABLE);
 	return true;
-}
-
-static inline void pagetable_pmd_dtor(struct ptdesc *ptdesc)
-{
-	struct folio *folio = ptdesc_folio(ptdesc);
-
-	pmd_ptlock_free(ptdesc);
-	__folio_clear_pgtable(folio);
-	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
 }
 
 /*
···
 	lruvec_stat_add_folio(folio, NR_PAGETABLE);
 }
 
-static inline void pagetable_pud_dtor(struct ptdesc *ptdesc)
-{
-	struct folio *folio = ptdesc_folio(ptdesc);
-
-	__folio_clear_pgtable(folio);
-	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
-}
-
 static inline void pagetable_p4d_ctor(struct ptdesc *ptdesc)
 {
 	struct folio *folio = ptdesc_folio(ptdesc);
 
 	__folio_set_pgtable(folio);
 	lruvec_stat_add_folio(folio, NR_PAGETABLE);
-}
-
-static inline void pagetable_p4d_dtor(struct ptdesc *ptdesc)
-{
-	struct folio *folio = ptdesc_folio(ptdesc);
-
-	__folio_clear_pgtable(folio);
-	lruvec_stat_sub_folio(folio, NR_PAGETABLE);
 }
 
 extern void __init pagecache_init(void);

mm/memory.c

···
 
 void ptlock_free(struct ptdesc *ptdesc)
 {
-	kmem_cache_free(page_ptl_cachep, ptdesc->ptl);
+	if (ptdesc->ptl)
+		kmem_cache_free(page_ptl_cachep, ptdesc->ptl);
 }
 #endif