Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1200263

[PATCH 08/11] thp: Fix zap_huge_pmd() for DAX

Show all headers | View raw

From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>

The original DAX code assumed that pgtable_t was a pointer, which isn't
true on all architectures.  Restructure the code to not rely on that
assumption.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
[further fixes from Matthew integrated into this patch]
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
---
 mm/huge_memory.c | 71 +++++++++++++++++++++++++-------------------------------
 1 file changed, 31 insertions(+), 40 deletions(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 326d17e..b51bed1 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1426,50 +1426,41 @@ out:
 int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
 		 pmd_t *pmd, unsigned long addr)
 {
+	pmd_t orig_pmd;
 	spinlock_t *ptl;
-	int ret = 0;
 
-	if (__pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
-		pgtable_t pgtable;
-		pmd_t orig_pmd;
-		/*
-		 * For architectures like ppc64 we look at deposited pgtable
-		 * when calling pmdp_huge_get_and_clear. So do the
-		 * pgtable_trans_huge_withdraw after finishing pmdp related
-		 * operations.
-		 */
-		orig_pmd = pmdp_huge_get_and_clear_full(tlb->mm, addr, pmd,
-							tlb->fullmm);
-		tlb_remove_pmd_tlb_entry(tlb, pmd, addr);
-		if (vma_is_dax(vma)) {
-			if (is_huge_zero_pmd(orig_pmd)) {
-				pgtable = NULL;
-			} else {
-				spin_unlock(ptl);
-				return 1;
-			}
-		} else {
-			pgtable = pgtable_trans_huge_withdraw(tlb->mm, pmd);
-		}
-		if (is_huge_zero_pmd(orig_pmd)) {
-			atomic_long_dec(&tlb->mm->nr_ptes);
-			spin_unlock(ptl);
+	if (__pmd_trans_huge_lock(pmd, vma, &ptl) != 1)
+		return 0;
+	/*
+	 * For architectures like ppc64 we look at deposited pgtable
+	 * when calling pmdp_huge_get_and_clear. So do the
+	 * pgtable_trans_huge_withdraw after finishing pmdp related
+	 * operations.
+	 */
+	orig_pmd = pmdp_huge_get_and_clear_full(tlb->mm, addr, pmd,
+			tlb->fullmm);
+	tlb_remove_pmd_tlb_entry(tlb, pmd, addr);
+	if (vma_is_dax(vma)) {
+		spin_unlock(ptl);
+		if (is_huge_zero_pmd(orig_pmd))
 			put_huge_zero_page();
-		} else {
-			struct page *page = pmd_page(orig_pmd);
-			page_remove_rmap(page);
-			VM_BUG_ON_PAGE(page_mapcount(page) < 0, page);
-			add_mm_counter(tlb->mm, MM_ANONPAGES, -HPAGE_PMD_NR);
-			VM_BUG_ON_PAGE(!PageHead(page), page);
-			atomic_long_dec(&tlb->mm->nr_ptes);
-			spin_unlock(ptl);
-			tlb_remove_page(tlb, page);
-		}
-		if (pgtable)
-			pte_free(tlb->mm, pgtable);
-		ret = 1;
+	} else if (is_huge_zero_pmd(orig_pmd)) {
+		pte_free(tlb->mm, pgtable_trans_huge_withdraw(tlb->mm, pmd));
+		atomic_long_dec(&tlb->mm->nr_ptes);
+		spin_unlock(ptl);
+		put_huge_zero_page();
+	} else {
+		struct page *page = pmd_page(orig_pmd);
+		page_remove_rmap(page);
+		VM_BUG_ON_PAGE(page_mapcount(page) < 0, page);
+		add_mm_counter(tlb->mm, MM_ANONPAGES, -HPAGE_PMD_NR);
+		VM_BUG_ON_PAGE(!PageHead(page), page);
+		pte_free(tlb->mm, pgtable_trans_huge_withdraw(tlb->mm, pmd));
+		atomic_long_dec(&tlb->mm->nr_ptes);
+		spin_unlock(ptl);
+		tlb_remove_page(tlb, page);
 	}
-	return ret;
+	return 1;
 }
 
 int move_huge_pmd(struct vm_area_struct *vma, struct vm_area_struct *new_vma,
-- 
2.1.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

[PATCH 00/11] DAX fixes for 4.3 Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:00 +0200
  [PATCH 10/11] dax: Ensure that zero pages are removed from other processes Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:00 +0200
  [PATCH 09/11] dax: Don't use set_huge_zero_page() Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:00 +0200
  [PATCH 07/11] thp: Decrement refcount on huge zero page if it is split Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:00 +0200
  [PATCH 08/11] thp: Fix zap_huge_pmd() for DAX Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:00 +0200
  [PATCH 01/11] ext4: Use ext4_get_block_write() for DAX Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:10 +0200
  [PATCH 06/11] dax: Fix race between simultaneous faults Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:10 +0200
    Re: [PATCH 06/11] dax: Fix race between simultaneous faults "Kirill A. Shutemov" <kirill@shutemov.name> - 2015-08-05 13:50 +0200
  [PATCH 11/11] dax: Use linear_page_index() Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:10 +0200
  [PATCH 04/11] ext4: Add ext4_get_block_dax() Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:10 +0200
    Re: [PATCH 04/11] ext4: Add ext4_get_block_dax() Dave Chinner <david@fromorbit.com> - 2015-08-05 04:10 +0200
      Re: [PATCH 04/11] ext4: Add ext4_get_block_dax() Matthew Wilcox <willy@linux.intel.com> - 2015-08-05 17:20 +0200
  [PATCH 05/11] ext4: Start transaction before calling into DAX Matthew Wilcox <matthew.r.wilcox@intel.com> - 2015-08-04 22:10 +0200

csiph-web