[PATCH 0/2] hugetlb minor improvements

[PATCH 0/2] hugetlb minor improvements

[Naoya Horiguchi]

This patchset does minor improvements and cleanups for mm/hugetlb.c.
It's based on mmotm-2014-09-25-16-28 and shows no regression in libhugetlbfs test.

Tree: git@github.com:Naoya-Horiguchi/linux.git
Branch: mmotm-2014-09-25-16-28/hugetlb_minor_improvements
---
Summary:

Naoya Horiguchi (2):
      mm/hugetlb: improve suboptimal migration/hwpoisoned entry check
      mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)()

 mm/hugetlb.c | 60 ++++++++++++++++++++++--------------------------------------
 1 file changed, 22 insertions(+), 38 deletions(-)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check

[Naoya Horiguchi]

Currently hugetlb_fault() checks at first whether pte of the faulted
address is a migration or hwpoisoned entry. The reason of this approach
is that without the checks, the BUG_ON() in huge_pte_alloc() is triggered,
because it assumes that when pte is not none, it always points to a
normal hugepage, which was correct originally but not after hugetlb
supports page migration or hwpoison.

In order to iron out this weird workaround, this patch changes the
wrongly assumed BUG_ON() in huge_pte_alloc(). This allows us to check
pte_present() case only in proper place, which makes code simpler.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
 mm/hugetlb.c | 28 ++++++++++++----------------
 1 file changed, 12 insertions(+), 16 deletions(-)

diff --git mmotm-2014-09-25-16-28.orig/mm/hugetlb.c mmotm-2014-09-25-16-28/mm/hugetlb.c
index 1ecb625bc498..e6543359be4d 100644
--- mmotm-2014-09-25-16-28.orig/mm/hugetlb.c
+++ mmotm-2014-09-25-16-28/mm/hugetlb.c
@@ -3130,20 +3130,10 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	struct hstate *h = hstate_vma(vma);
 	struct address_space *mapping;
 	int need_wait_lock = 0;
+	int need_wait_migration = 0;
 
 	address &= huge_page_mask(h);
 
-	ptep = huge_pte_offset(mm, address);
-	if (ptep) {
-		entry = huge_ptep_get(ptep);
-		if (unlikely(is_hugetlb_entry_migration(entry))) {
-			migration_entry_wait_huge(vma, mm, ptep);
-			return 0;
-		} else if (unlikely(is_hugetlb_entry_hwpoisoned(entry)))
-			return VM_FAULT_HWPOISON_LARGE |
-				VM_FAULT_SET_HINDEX(hstate_index(h));
-	}
-
 	ptep = huge_pte_alloc(mm, address, huge_page_size(h));
 	if (!ptep)
 		return VM_FAULT_OOM;
@@ -3169,12 +3159,16 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	/*
 	 * entry could be a migration/hwpoison entry at this point, so this
 	 * check prevents the kernel from going below assuming that we have
-	 * a active hugepage in pagecache. This goto expects the 2nd page fault,
-	 * and is_hugetlb_entry_(migration|hwpoisoned) check will properly
-	 * handle it.
+	 * a active hugepage in pagecache.
 	 */
-	if (!pte_present(entry))
+	if (!pte_present(entry)) {
+		if (is_hugetlb_entry_migration(entry))
+			need_wait_migration = 1;
+		else if (is_hugetlb_entry_hwpoisoned(entry))
+			ret = VM_FAULT_HWPOISON_LARGE |
+				VM_FAULT_SET_HINDEX(hstate_index(h));
 		goto out_mutex;
+	}
 
 	/*
 	 * If we are going to COW the mapping later, we examine the pending
@@ -3242,6 +3236,8 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	}
 out_mutex:
 	mutex_unlock(&htlb_fault_mutex_table[hash]);
+	if (need_wait_migration)
+		migration_entry_wait_huge(vma, mm, ptep);
 	if (need_wait_lock)
 		wait_on_page_locked(page);
 	return ret;
@@ -3664,7 +3660,7 @@ pte_t *huge_pte_alloc(struct mm_struct *mm,
 				pte = (pte_t *)pmd_alloc(mm, pud, addr);
 		}
 	}
-	BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte));
+	BUG_ON(pte && !pte_none(*pte) && pte_present(*pte) && !pte_huge(*pte));
 
 	return pte;
 }
-- 
1.9.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[mmotm][PATCH 2/2] mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)()

[Naoya Horiguchi]

non_swap_entry() returns true if a given swp_entry_t is a migration
entry or hwpoisoned entry. So non_swap_entry() && is_migration_entry() is
identical with just is_migration_entry(). By removing non_swap_entry(),
we can write is_hugetlb_entry_(migration|hwpoisoned)() more simply.

And the name is_hugetlb_entry_(migration|hwpoisoned) is lengthy and
it's not predictable from naming convention around pte_* family.
Just pte_migration() looks better, but these function contains hugetlb
specific (so architecture dependent) huge_pte_none() check, so let's
rename them as huge_pte_(migration|hwpoisoned).

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
 mm/hugetlb.c | 36 ++++++++++++------------------------
 1 file changed, 12 insertions(+), 24 deletions(-)

diff --git mmotm-2014-09-25-16-28.orig/mm/hugetlb.c mmotm-2014-09-25-16-28/mm/hugetlb.c
index e6543359be4d..e70da7ae36ed 100644
--- mmotm-2014-09-25-16-28.orig/mm/hugetlb.c
+++ mmotm-2014-09-25-16-28/mm/hugetlb.c
@@ -2516,30 +2516,18 @@ static void set_huge_ptep_writable(struct vm_area_struct *vma,
 		update_mmu_cache(vma, address, ptep);
 }
 
-static int is_hugetlb_entry_migration(pte_t pte)
+static inline int huge_pte_migration(pte_t pte)
 {
-	swp_entry_t swp;
-
 	if (huge_pte_none(pte) || pte_present(pte))
 		return 0;
-	swp = pte_to_swp_entry(pte);
-	if (non_swap_entry(swp) && is_migration_entry(swp))
-		return 1;
-	else
-		return 0;
+	return is_migration_entry(pte_to_swp_entry(pte));
 }
 
-static int is_hugetlb_entry_hwpoisoned(pte_t pte)
+static inline int huge_pte_hwpoisoned(pte_t pte)
 {
-	swp_entry_t swp;
-
 	if (huge_pte_none(pte) || pte_present(pte))
 		return 0;
-	swp = pte_to_swp_entry(pte);
-	if (non_swap_entry(swp) && is_hwpoison_entry(swp))
-		return 1;
-	else
-		return 0;
+	return is_hwpoison_entry(pte_to_swp_entry(pte));
 }
 
 int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
@@ -2583,8 +2571,8 @@ int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
 		entry = huge_ptep_get(src_pte);
 		if (huge_pte_none(entry)) { /* skip none entry */
 			;
-		} else if (unlikely(is_hugetlb_entry_migration(entry) ||
-				    is_hugetlb_entry_hwpoisoned(entry))) {
+		} else if (unlikely(huge_pte_migration(entry) ||
+				    huge_pte_hwpoisoned(entry))) {
 			swp_entry_t swp_entry = pte_to_swp_entry(entry);
 
 			if (is_write_migration_entry(swp_entry) && cow) {
@@ -3162,9 +3150,9 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	 * a active hugepage in pagecache.
 	 */
 	if (!pte_present(entry)) {
-		if (is_hugetlb_entry_migration(entry))
+		if (huge_pte_migration(entry))
 			need_wait_migration = 1;
-		else if (is_hugetlb_entry_hwpoisoned(entry))
+		else if (huge_pte_hwpoisoned(entry))
 			ret = VM_FAULT_HWPOISON_LARGE |
 				VM_FAULT_SET_HINDEX(hstate_index(h));
 		goto out_mutex;
@@ -3291,8 +3279,8 @@ long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
 		 * (in which case hugetlb_fault waits for the migration,) and
 		 * hwpoisoned hugepages (in which case we need to prevent the
 		 * caller from accessing to them.) In order to do this, we use
-		 * here is_swap_pte instead of is_hugetlb_entry_migration and
-		 * is_hugetlb_entry_hwpoisoned. This is because it simply covers
+		 * here is_swap_pte instead of huge_pte_migration and
+		 * huge_pte_hwpoisoned. This is because it simply covers
 		 * both cases, and because we can't follow correct pages
 		 * directly from any kind of swap entries.
 		 */
@@ -3370,11 +3358,11 @@ unsigned long hugetlb_change_protection(struct vm_area_struct *vma,
 			continue;
 		}
 		pte = huge_ptep_get(ptep);
-		if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) {
+		if (unlikely(huge_pte_hwpoisoned(pte))) {
 			spin_unlock(ptl);
 			continue;
 		}
-		if (unlikely(is_hugetlb_entry_migration(pte))) {
+		if (unlikely(huge_pte_migration(pte))) {
 			swp_entry_t entry = pte_to_swp_entry(pte);
 
 			if (is_write_migration_entry(entry)) {
-- 
1.9.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH 0/2] hugetlb minor improvements

[Naoya Horiguchi]

This patchset does minor improvements and cleanups for mm/hugetlb.c.
It's based on mmotm-2014-09-25-16-28 and shows no regression in libhugetlbfs test.

Tree: git@github.com:Naoya-Horiguchi/linux.git
Branch: mmotm-2014-09-25-16-28/hugetlb_minor_improvements
---
Summary:

Naoya Horiguchi (2):
      mm/hugetlb: improve suboptimal migration/hwpoisoned entry check
      mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)()

 mm/hugetlb.c | 60 ++++++++++++++++++++++--------------------------------------
 1 file changed, 22 insertions(+), 38 deletions(-)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check

[Naoya Horiguchi]

Currently hugetlb_fault() checks at first whether pte of the faulted
address is a migration or hwpoisoned entry. The reason of this approach
is that without the checks, the BUG_ON() in huge_pte_alloc() is triggered,
because it assumes that when pte is not none, it always points to a
normal hugepage, which was correct originally but not after hugetlb
supports page migration or hwpoison.

In order to iron out this weird workaround, this patch changes the
wrongly assumed BUG_ON() in huge_pte_alloc(). This allows us to check
pte_present() case only in proper place, which makes code simpler.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
 mm/hugetlb.c | 28 ++++++++++++----------------
 1 file changed, 12 insertions(+), 16 deletions(-)

diff --git mmotm-2014-09-25-16-28.orig/mm/hugetlb.c mmotm-2014-09-25-16-28/mm/hugetlb.c
index 1ecb625bc498..e6543359be4d 100644
--- mmotm-2014-09-25-16-28.orig/mm/hugetlb.c
+++ mmotm-2014-09-25-16-28/mm/hugetlb.c
@@ -3130,20 +3130,10 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	struct hstate *h = hstate_vma(vma);
 	struct address_space *mapping;
 	int need_wait_lock = 0;
+	int need_wait_migration = 0;
 
 	address &= huge_page_mask(h);
 
-	ptep = huge_pte_offset(mm, address);
-	if (ptep) {
-		entry = huge_ptep_get(ptep);
-		if (unlikely(is_hugetlb_entry_migration(entry))) {
-			migration_entry_wait_huge(vma, mm, ptep);
-			return 0;
-		} else if (unlikely(is_hugetlb_entry_hwpoisoned(entry)))
-			return VM_FAULT_HWPOISON_LARGE |
-				VM_FAULT_SET_HINDEX(hstate_index(h));
-	}
-
 	ptep = huge_pte_alloc(mm, address, huge_page_size(h));
 	if (!ptep)
 		return VM_FAULT_OOM;
@@ -3169,12 +3159,16 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	/*
 	 * entry could be a migration/hwpoison entry at this point, so this
 	 * check prevents the kernel from going below assuming that we have
-	 * a active hugepage in pagecache. This goto expects the 2nd page fault,
-	 * and is_hugetlb_entry_(migration|hwpoisoned) check will properly
-	 * handle it.
+	 * a active hugepage in pagecache.
 	 */
-	if (!pte_present(entry))
+	if (!pte_present(entry)) {
+		if (is_hugetlb_entry_migration(entry))
+			need_wait_migration = 1;
+		else if (is_hugetlb_entry_hwpoisoned(entry))
+			ret = VM_FAULT_HWPOISON_LARGE |
+				VM_FAULT_SET_HINDEX(hstate_index(h));
 		goto out_mutex;
+	}
 
 	/*
 	 * If we are going to COW the mapping later, we examine the pending
@@ -3242,6 +3236,8 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	}
 out_mutex:
 	mutex_unlock(&htlb_fault_mutex_table[hash]);
+	if (need_wait_migration)
+		migration_entry_wait_huge(vma, mm, ptep);
 	if (need_wait_lock)
 		wait_on_page_locked(page);
 	return ret;
@@ -3664,7 +3660,7 @@ pte_t *huge_pte_alloc(struct mm_struct *mm,
 				pte = (pte_t *)pmd_alloc(mm, pud, addr);
 		}
 	}
-	BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte));
+	BUG_ON(pte && !pte_none(*pte) && pte_present(*pte) && !pte_huge(*pte));
 
 	return pte;
 }
-- 
1.9.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>