[RFC,18/18] mm/pgtable: pass ptdesc in pte_free_defer

Commit Message

[e-mail] July 30, 2024, 7:27 a.m. UTC

From: Alex Shi <[e-mail]>

pass ptdesc in pte_free_defer() and use ptdesc in collapse_huge_page().

This patch is immature, there is a issue from pmd_pgtable() conversion
in few archs. The problem need a fix.

Signed-off-by: Alex Shi <[e-mail]>
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: Mike Rapoport  <[e-mail]>
Cc: Barry Song <[e-mail]>
Cc: Peter Xu <[e-mail]>
Cc: Lance Yang <[e-mail]>
Cc: Kemeng Shi <[e-mail]>
Cc: Aneesh Kumar K.V <[e-mail]>
Cc: Qi Zheng <[e-mail]>
Cc: Sven Schnelle <[e-mail]>
Cc: Christian Borntraeger <[e-mail]>
Cc: Vasily Gorbik <[e-mail]>
Cc: Heiko Carstens <[e-mail]>
Cc: Naveen N Rao <[e-mail]>
Cc: Nicholas Piggin <[e-mail]>
Cc: Michael Ellerman <[e-mail]>
Cc: Matthew Wilcox  <[e-mail]>
Cc: Ryan Roberts <[e-mail]>
Cc: David Hildenbrand <[e-mail]>
Cc: Vishal Moola  <[e-mail]>
Cc: Hugh Dickins <[e-mail]>
Cc: Gerald Schaefer <[e-mail]>
Cc: Alexander Gordeev <[e-mail]>
Cc: Christophe Leroy <[e-mail]>
---
 arch/powerpc/include/asm/pgalloc.h |  2 +-
 arch/s390/include/asm/pgalloc.h    |  2 +-
 arch/s390/mm/pgalloc.c             |  2 +-
 include/linux/pgtable.h            |  2 +-
 mm/khugepaged.c                    | 10 +++++-----
 mm/pgtable-generic.c               |  4 +---
 6 files changed, 10 insertions(+), 12 deletions(-)

Patch

diff --git a/arch/powerpc/include/asm/pgalloc.h b/arch/powerpc/include/asm/pgalloc.h
index 12520521163e..ca21b67c593f 100644
--- a/arch/powerpc/include/asm/pgalloc.h
+++ b/arch/powerpc/include/asm/pgalloc.h
@@ -47,7 +47,7 @@  static inline void pte_free(struct mm_struct *mm, struct ptdesc *ptepage)
 
 /* arch use pte_free_defer() implementation in arch/powerpc/mm/pgtable-frag.c */
 #define pte_free_defer pte_free_defer
-void pte_free_defer(struct mm_struct *mm, pgtable_t pgtable);
+void pte_free_defer(struct mm_struct *mm, struct ptdesc *pgtable);
 
 /*
  * Functions that deal with pagetables that could be at any level of
diff --git a/arch/s390/include/asm/pgalloc.h b/arch/s390/include/asm/pgalloc.h
index 771494526f6e..a229cee11bbd 100644
--- a/arch/s390/include/asm/pgalloc.h
+++ b/arch/s390/include/asm/pgalloc.h
@@ -144,7 +144,7 @@  static inline void pmd_populate(struct mm_struct *mm,
 
 /* arch use pte_free_defer() implementation in arch/s390/mm/pgalloc.c */
 #define pte_free_defer pte_free_defer
-void pte_free_defer(struct mm_struct *mm, pgtable_t pgtable);
+void pte_free_defer(struct mm_struct *mm, struct ptdesc *pgtable);
 
 void vmem_map_init(void);
 void *vmem_crst_alloc(unsigned long val);
diff --git a/arch/s390/mm/pgalloc.c b/arch/s390/mm/pgalloc.c
index f691e0fb66a2..c7bb38d85d81 100644
--- a/arch/s390/mm/pgalloc.c
+++ b/arch/s390/mm/pgalloc.c
@@ -214,7 +214,7 @@  static void pte_free_now(struct rcu_head *head)
 	pagetable_pte_dtor_free(ptdesc);
 }
 
-void pte_free_defer(struct mm_struct *mm, pgtable_t pgtable)
+void pte_free_defer(struct mm_struct *mm, struct ptdesc *pgtable)
 {
 	struct ptdesc *ptdesc = virt_to_ptdesc(pgtable);
 
diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
index 9d256c548f5e..e7b018de1d0f 100644
--- a/include/linux/pgtable.h
+++ b/include/linux/pgtable.h
@@ -116,7 +116,7 @@  static inline void pte_unmap(pte_t *pte)
 }
 #endif
 
-void pte_free_defer(struct mm_struct *mm, pgtable_t pgtable);
+void pte_free_defer(struct mm_struct *mm, struct ptdesc *ptdesc);
 
 /* Find an entry in the second-level page table.. */
 #ifndef pmd_offset
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 5b466a1c2136..30cf61d02c1c 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -1094,7 +1094,7 @@  static int collapse_huge_page(struct mm_struct *mm, unsigned long address,
 	LIST_HEAD(compound_pagelist);
 	pmd_t *pmd, _pmd;
 	pte_t *pte;
-	pgtable_t pgtable;
+	struct ptdesc *ptdesc;
 	struct folio *folio;
 	spinlock_t *pmd_ptl, *pte_ptl;
 	int result = SCAN_FAIL;
@@ -1223,7 +1223,7 @@  static int collapse_huge_page(struct mm_struct *mm, unsigned long address,
 	 * write.
 	 */
 	__folio_mark_uptodate(folio);
-	pgtable = pmd_pgtable(_pmd);
+	ptdesc = pmd_ptdesc(&_pmd);
 
 	_pmd = mk_huge_pmd(&folio->page, vma->vm_page_prot);
 	_pmd = maybe_pmd_mkwrite(pmd_mkdirty(_pmd), vma);
@@ -1232,7 +1232,7 @@  static int collapse_huge_page(struct mm_struct *mm, unsigned long address,
 	BUG_ON(!pmd_none(*pmd));
 	folio_add_new_anon_rmap(folio, vma, address, RMAP_EXCLUSIVE);
 	folio_add_lru_vma(folio, vma);
-	pgtable_trans_huge_deposit(mm, pmd, page_ptdesc(pgtable));
+	pgtable_trans_huge_deposit(mm, pmd, ptdesc);
 	set_pmd_at(mm, address, pmd, _pmd);
 	update_mmu_cache_pmd(vma, address, pmd);
 	spin_unlock(pmd_ptl);
@@ -1664,7 +1664,7 @@  int collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr,
 
 	mm_dec_nr_ptes(mm);
 	page_table_check_pte_clear_range(mm, haddr, pgt_pmd);
-	pte_free_defer(mm, pmd_pgtable(pgt_pmd));
+	pte_free_defer(mm, pmd_ptdesc(&pgt_pmd));
 
 maybe_install_pmd:
 	/* step 5: install pmd entry */
@@ -1777,7 +1777,7 @@  static void retract_page_tables(struct address_space *mapping, pgoff_t pgoff)
 		if (retracted) {
 			mm_dec_nr_ptes(mm);
 			page_table_check_pte_clear_range(mm, addr, pgt_pmd);
-			pte_free_defer(mm, pmd_pgtable(pgt_pmd));
+			pte_free_defer(mm, pmd_ptdesc(&pgt_pmd));
 		}
 	}
 	i_mmap_unlock_read(mapping);
diff --git a/mm/pgtable-generic.c b/mm/pgtable-generic.c
index 5e763682941d..f3bc2b17893a 100644
--- a/mm/pgtable-generic.c
+++ b/mm/pgtable-generic.c
@@ -244,10 +244,8 @@  static void pte_free_now(struct rcu_head *head)
 	pte_free(NULL /* mm not passed and not used */, ptdesc);
 }
 
-void pte_free_defer(struct mm_struct *mm, pgtable_t pgtable)
+void pte_free_defer(struct mm_struct *mm, struct ptdesc *ptdesc)
 {
-	struct ptdesc *ptdesc = page_ptdesc(pgtable);
-
 	call_rcu(&ptdesc->pt_rcu_head, pte_free_now);
 }
 #endif /* pte_free_defer */