Patchwork [-V9,06/20] mm/THP: deposit the transpare huge pgtable before set_pmd

login
register
mail settings
Submitter Aneesh Kumar K.V
Date May 22, 2013, 4:57 p.m.
Message ID <1369241872-15700-7-git-send-email-aneesh.kumar@linux.vnet.ibm.com>
Download mbox | patch
Permalink /patch/245667/
State Superseded
Delegated to: Benjamin Herrenschmidt
Headers show

Comments

Aneesh Kumar K.V - May 22, 2013, 4:57 p.m.
From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>

Architectures like powerpc use the deposited pgtable to store
hash index values. We need to make the deposted pgtable visible
to other cpus before we are ready to take a hash fault.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
 mm/huge_memory.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

Patch

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 5c4fac2..59d9384 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -729,8 +729,8 @@  static int __do_huge_pmd_anonymous_page(struct mm_struct *mm,
 		pmd_t entry;
 		entry = mk_huge_pmd(page, vma);
 		page_add_new_anon_rmap(page, vma, haddr);
-		set_pmd_at(mm, haddr, pmd, entry);
 		pgtable_trans_huge_deposit(mm, pmd, pgtable);
+		set_pmd_at(mm, haddr, pmd, entry);
 		add_mm_counter(mm, MM_ANONPAGES, HPAGE_PMD_NR);
 		mm->nr_ptes++;
 		spin_unlock(&mm->page_table_lock);
@@ -771,8 +771,8 @@  static bool set_huge_zero_page(pgtable_t pgtable, struct mm_struct *mm,
 	entry = mk_pmd(zero_page, vma->vm_page_prot);
 	entry = pmd_wrprotect(entry);
 	entry = pmd_mkhuge(entry);
-	set_pmd_at(mm, haddr, pmd, entry);
 	pgtable_trans_huge_deposit(mm, pmd, pgtable);
+	set_pmd_at(mm, haddr, pmd, entry);
 	mm->nr_ptes++;
 	return true;
 }
@@ -916,8 +916,8 @@  int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 
 	pmdp_set_wrprotect(src_mm, addr, src_pmd);
 	pmd = pmd_mkold(pmd_wrprotect(pmd));
-	set_pmd_at(dst_mm, addr, dst_pmd, pmd);
 	pgtable_trans_huge_deposit(dst_mm, dst_pmd, pgtable);
+	set_pmd_at(dst_mm, addr, dst_pmd, pmd);
 	dst_mm->nr_ptes++;
 
 	ret = 0;
@@ -2367,9 +2367,9 @@  static void collapse_huge_page(struct mm_struct *mm,
 	spin_lock(&mm->page_table_lock);
 	BUG_ON(!pmd_none(*pmd));
 	page_add_new_anon_rmap(new_page, vma, address);
+	pgtable_trans_huge_deposit(mm, pmd, pgtable);
 	set_pmd_at(mm, address, pmd, _pmd);
 	update_mmu_cache_pmd(vma, address, pmd);
-	pgtable_trans_huge_deposit(mm, pmd, pgtable);
 	spin_unlock(&mm->page_table_lock);
 
 	*hpage = NULL;