[mm-unstable PATCH 0/7] Cleanup transhuge_xxx helpers

[mm-unstable PATCH 0/7] Cleanup transhuge_xxx helpers

[Yang Shi]

This series is the follow-up of the discussion about cleaning up transhuge_xxx
helpers at https://lore.kernel.org/linux-mm/627a71f8-e879-69a5-ceb3-fc8d29d2f7f1@suse.cz/.

THP has a bunch of helpers that do VMA sanity check for different paths, they
do the similar checks for the most callsites and have a lot duplicate codes.
And it is confusing what helpers should be used at what conditions.

This series reorganized and cleaned up the code so that we could consolidate
all the checks into hugepage_vma_check().

The transhuge_vma_enabled(), transparent_hugepage_active() and
__transparent_hugepage_enabled() are killed by this series.

Added transhuge_vma_size_ok() helper to remove some duplicate code.


 b/fs/proc/task_mmu.c         |    4 +-
 b/include/linux/huge_mm.h    |  102 ++++++++++++++++++++++++++-----------------------------------------
 b/include/linux/khugepaged.h |   21 +------------
 b/mm/huge_memory.c           |   74 ++++++++++++++++++++++++++++++++++++------------
 b/mm/khugepaged.c            |   92 ++++++++++++++++--------------------------------------------
 b/mm/memory.c                |    6 ++-
 6 files changed, 129 insertions(+), 170 deletions(-)

[PATCH 1/7] mm: khugepaged: check THP flag in hugepage_vma_check()

[Yang Shi]

Currently the THP flag check in hugepage_vma_check() will fallthrough if
the flag is NEVER and VM_HUGEPAGE is set.  This is not a problem for now
since all the callers have the flag checked before or can't be invoked if
the flag is NEVER.

However, the following patch will call hugepage_vma_check() in more
places, for example, page fault, so this flag must be checked in
hugepge_vma_check().

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 mm/khugepaged.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 637bfecd6bf5..92e91c08d96a 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -458,6 +458,9 @@ bool hugepage_vma_check(struct vm_area_struct *vma,
 	if (shmem_file(vma->vm_file))
 		return shmem_huge_enabled(vma);
 
+	if (!khugepaged_enabled())
+		return false;
+
 	/* THP settings require madvise. */
 	if (!(vm_flags & VM_HUGEPAGE) && !khugepaged_always())
 		return false;
-- 
2.26.3

[PATCH 2/7] mm: thp: introduce transhuge_vma_size_ok() helper

[Yang Shi]

There are couple of places that check whether the vma size is ok for
THP or not, they are open coded and duplicate, introduce
transhuge_vma_size_ok() helper to do the job.

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 include/linux/huge_mm.h | 17 +++++++++++++++++
 mm/huge_memory.c        |  5 +----
 mm/khugepaged.c         | 12 ++++++------
 3 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 648cb3ce7099..a8f61db47f2a 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -116,6 +116,18 @@ extern struct kobj_attribute shmem_enabled_attr;
 
 extern unsigned long transparent_hugepage_flags;
 
+/*
+ * The vma size has to be large enough to hold an aligned HPAGE_PMD_SIZE area.
+ */
+static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
+{
+	if (round_up(vma->vm_start, HPAGE_PMD_SIZE) <
+	    (vma->vm_end & HPAGE_PMD_MASK))
+		return true;
+
+	return false;
+}
+
 static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
 		unsigned long addr)
 {
@@ -345,6 +357,11 @@ static inline bool transparent_hugepage_active(struct vm_area_struct *vma)
 	return false;
 }
 
+static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
+{
+	return false;
+}
+
 static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
 		unsigned long addr)
 {
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 80e8b58b4f39..d633f97452c1 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -71,10 +71,7 @@ unsigned long huge_zero_pfn __read_mostly = ~0UL;
 
 bool transparent_hugepage_active(struct vm_area_struct *vma)
 {
-	/* The addr is used to check if the vma size fits */
-	unsigned long addr = (vma->vm_end & HPAGE_PMD_MASK) - HPAGE_PMD_SIZE;
-
-	if (!transhuge_vma_suitable(vma, addr))
+	if (!transhuge_vma_size_ok(vma))
 		return false;
 	if (vma_is_anonymous(vma))
 		return __transparent_hugepage_enabled(vma);
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 92e91c08d96a..5bdb30a76f05 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -454,6 +454,9 @@ bool hugepage_vma_check(struct vm_area_struct *vma,
 				vma->vm_pgoff, HPAGE_PMD_NR))
 		return false;
 
+	if (!transhuge_vma_size_ok(vma))
+		return false;
+
 	/* Enabled via shmem mount options or sysfs settings. */
 	if (shmem_file(vma->vm_file))
 		return shmem_huge_enabled(vma);
@@ -512,9 +515,7 @@ void khugepaged_enter_vma(struct vm_area_struct *vma,
 			  unsigned long vm_flags)
 {
 	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags) &&
-	    khugepaged_enabled() &&
-	    (((vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK) <
-	     (vma->vm_end & HPAGE_PMD_MASK))) {
+	    khugepaged_enabled()) {
 		if (hugepage_vma_check(vma, vm_flags))
 			__khugepaged_enter(vma->vm_mm);
 	}
@@ -2144,10 +2145,9 @@ static unsigned int khugepaged_scan_mm_slot(unsigned int pages,
 			progress++;
 			continue;
 		}
-		hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK;
+
+		hstart = round_up(vma->vm_start, HPAGE_PMD_SIZE);
 		hend = vma->vm_end & HPAGE_PMD_MASK;
-		if (hstart >= hend)
-			goto skip;
 		if (khugepaged_scan.address > hend)
 			goto skip;
 		if (khugepaged_scan.address < hstart)
-- 
2.26.3

[PATCH 3/7] mm: khugepaged: remove the redundant anon vma check

[Yang Shi]

The hugepage_vma_check() already checked it, so remove the redundant
check.

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 mm/khugepaged.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 5bdb30a76f05..4fe1dd3d8ef4 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -966,9 +966,6 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
 		return SCAN_ADDRESS_RANGE;
 	if (!hugepage_vma_check(vma, vma->vm_flags))
 		return SCAN_VMA_CHECK;
-	/* Anon VMA expected */
-	if (!vma->anon_vma || !vma_is_anonymous(vma))
-		return SCAN_VMA_CHECK;
 	return 0;
 }
 
-- 
2.26.3

[PATCH 4/7] mm: khugepaged: use transhuge_vma_suitable replace open-code

[Yang Shi]

The hugepage_vma_revalidate() needs to check if the address is still in
the aligned HPAGE_PMD_SIZE area of the vma when reacquiring mmap_lock,
but it was open-coded, use transhuge_vma_suitable() to do the job.  And
add proper comments for transhuge_vma_suitable().

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 include/linux/huge_mm.h | 6 ++++++
 mm/khugepaged.c         | 5 +----
 2 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index a8f61db47f2a..79d5919beb83 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -128,6 +128,12 @@ static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
 	return false;
 }
 
+/*
+ * Do the below checks:
+ *   - For non-anon vma, check if the vm_pgoff is HPAGE_PMD_NR aligned.
+ *   - For all vmas, check if the haddr is in an aligned HPAGE_PMD_SIZE
+ *     area.
+ */
 static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
 		unsigned long addr)
 {
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 4fe1dd3d8ef4..51f0e6ea3977 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -951,7 +951,6 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
 		struct vm_area_struct **vmap)
 {
 	struct vm_area_struct *vma;
-	unsigned long hstart, hend;
 
 	if (unlikely(khugepaged_test_exit(mm)))
 		return SCAN_ANY_PROCESS;
@@ -960,9 +959,7 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
 	if (!vma)
 		return SCAN_VMA_NULL;
 
-	hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK;
-	hend = vma->vm_end & HPAGE_PMD_MASK;
-	if (address < hstart || address + HPAGE_PMD_SIZE > hend)
+	if (!transhuge_vma_suitable(vma, address))
 		return SCAN_ADDRESS_RANGE;
 	if (!hugepage_vma_check(vma, vma->vm_flags))
 		return SCAN_VMA_CHECK;
-- 
2.26.3

[PATCH 5/7] mm: thp: kill transparent_hugepage_active()

[Yang Shi]

The transparent_hugepage_active() was introduced to show THP eligibility
bit in smaps in proc, smaps is the only user.  But it actually does the
similar check as hugepage_vma_check() which is used by khugepaged.  We
definitely don't have to maintain two similar checks, so kill
transparent_hugepage_active().

Also move hugepage_vma_check() to huge_memory.c and huge_mm.h since it
is not only for khugepaged anymore.

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 fs/proc/task_mmu.c         |  2 +-
 include/linux/huge_mm.h    | 14 +++++++------
 include/linux/khugepaged.h |  2 --
 mm/huge_memory.c           | 41 ++++++++++++++++++++++++++++++------
 mm/khugepaged.c            | 43 --------------------------------------
 5 files changed, 44 insertions(+), 58 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index f9c9abb50bb7..ab4849d07a1f 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -863,7 +863,7 @@ static int show_smap(struct seq_file *m, void *v)
 	__show_smap(m, &mss, false);
 
 	seq_printf(m, "THPeligible:    %d\n",
-		   transparent_hugepage_active(vma));
+		   hugepage_vma_check(vma, vma->vm_flags));
 
 	if (arch_pkeys_enabled())
 		seq_printf(m, "ProtectionKey:  %8u\n", vma_pkey(vma));
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 79d5919beb83..2da52ca8fd52 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -209,7 +209,8 @@ static inline bool file_thp_enabled(struct vm_area_struct *vma)
 	       !inode_is_open_for_write(inode) && S_ISREG(inode->i_mode);
 }
 
-bool transparent_hugepage_active(struct vm_area_struct *vma);
+extern bool hugepage_vma_check(struct vm_area_struct *vma,
+			       unsigned long vm_flags);
 
 #define transparent_hugepage_use_zero_page()				\
 	(transparent_hugepage_flags &					\
@@ -358,11 +359,6 @@ static inline bool __transparent_hugepage_enabled(struct vm_area_struct *vma)
 	return false;
 }
 
-static inline bool transparent_hugepage_active(struct vm_area_struct *vma)
-{
-	return false;
-}
-
 static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
 {
 	return false;
@@ -380,6 +376,12 @@ static inline bool transhuge_vma_enabled(struct vm_area_struct *vma,
 	return false;
 }
 
+static inline bool hugepage_vma_check(struct vm_area_struct *vma,
+				       unsigned long vm_flags)
+{
+	return false;
+}
+
 static inline void prep_transhuge_page(struct page *page) {}
 
 #define transparent_hugepage_flags 0UL
diff --git a/include/linux/khugepaged.h b/include/linux/khugepaged.h
index 392d34c3c59a..795c8d2170b0 100644
--- a/include/linux/khugepaged.h
+++ b/include/linux/khugepaged.h
@@ -10,8 +10,6 @@ extern struct attribute_group khugepaged_attr_group;
 extern int khugepaged_init(void);
 extern void khugepaged_destroy(void);
 extern int start_stop_khugepaged(void);
-extern bool hugepage_vma_check(struct vm_area_struct *vma,
-			       unsigned long vm_flags);
 extern void __khugepaged_enter(struct mm_struct *mm);
 extern void __khugepaged_exit(struct mm_struct *mm);
 extern void khugepaged_enter_vma(struct vm_area_struct *vma,
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index d633f97452c1..d9ec82892b05 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -69,18 +69,47 @@ static atomic_t huge_zero_refcount;
 struct page *huge_zero_page __read_mostly;
 unsigned long huge_zero_pfn __read_mostly = ~0UL;
 
-bool transparent_hugepage_active(struct vm_area_struct *vma)
+bool hugepage_vma_check(struct vm_area_struct *vma,
+			unsigned long vm_flags)
 {
+	if (!transhuge_vma_enabled(vma, vm_flags))
+		return false;
+
+	if (vm_flags & VM_NO_KHUGEPAGED)
+		return false;
+
+	/* Don't run khugepaged against DAX vma */
+	if (vma_is_dax(vma))
+		return false;
+
+	if (vma->vm_file && !IS_ALIGNED((vma->vm_start >> PAGE_SHIFT) -
+				vma->vm_pgoff, HPAGE_PMD_NR))
+		return false;
+
 	if (!transhuge_vma_size_ok(vma))
 		return false;
-	if (vma_is_anonymous(vma))
-		return __transparent_hugepage_enabled(vma);
-	if (vma_is_shmem(vma))
+
+	/* Enabled via shmem mount options or sysfs settings. */
+	if (shmem_file(vma->vm_file))
 		return shmem_huge_enabled(vma);
-	if (transhuge_vma_enabled(vma, vma->vm_flags) && file_thp_enabled(vma))
+
+	if (!khugepaged_enabled())
+		return false;
+
+	/* THP settings require madvise. */
+	if (!(vm_flags & VM_HUGEPAGE) && !khugepaged_always())
+		return false;
+
+	/* Only regular file is valid */
+	if (file_thp_enabled(vma))
 		return true;
 
-	return false;
+	if (!vma->anon_vma || !vma_is_anonymous(vma))
+		return false;
+	if (vma_is_temporary_stack(vma))
+		return false;
+
+	return true;
 }
 
 static bool get_huge_zero_page(void)
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 51f0e6ea3977..a7ac1ae418f8 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -437,49 +437,6 @@ static inline int khugepaged_test_exit(struct mm_struct *mm)
 	return atomic_read(&mm->mm_users) == 0;
 }
 
-bool hugepage_vma_check(struct vm_area_struct *vma,
-			unsigned long vm_flags)
-{
-	if (!transhuge_vma_enabled(vma, vm_flags))
-		return false;
-
-	if (vm_flags & VM_NO_KHUGEPAGED)
-		return false;
-
-	/* Don't run khugepaged against DAX vma */
-	if (vma_is_dax(vma))
-		return false;
-
-	if (vma->vm_file && !IS_ALIGNED((vma->vm_start >> PAGE_SHIFT) -
-				vma->vm_pgoff, HPAGE_PMD_NR))
-		return false;
-
-	if (!transhuge_vma_size_ok(vma))
-		return false;
-
-	/* Enabled via shmem mount options or sysfs settings. */
-	if (shmem_file(vma->vm_file))
-		return shmem_huge_enabled(vma);
-
-	if (!khugepaged_enabled())
-		return false;
-
-	/* THP settings require madvise. */
-	if (!(vm_flags & VM_HUGEPAGE) && !khugepaged_always())
-		return false;
-
-	/* Only regular file is valid */
-	if (file_thp_enabled(vma))
-		return true;
-
-	if (!vma->anon_vma || !vma_is_anonymous(vma))
-		return false;
-	if (vma_is_temporary_stack(vma))
-		return false;
-
-	return true;
-}
-
 void __khugepaged_enter(struct mm_struct *mm)
 {
 	struct mm_slot *mm_slot;
-- 
2.26.3

[PATCH 6/7] mm: thp: kill __transhuge_page_enabled()

[Yang Shi]

The page fault path checks THP eligibility with
__transhuge_page_enabled() which does the similar thing as
hugepage_vma_check(), so use hugepage_vma_check() instead.

However page fault allows DAX and !anon_vma cases, so added a new flag,
in_pf, to hugepage_vma_check() to make page fault work correctly.

The in_pf flag is also used to skip shmem and file THP for page fault
since shmem handles THP in its own shmem_fault() and file THP allocation
on fault is not supported yet.

Also remove hugepage_vma_enabled() since hugepage_vma_check() is the
only caller now, it is not necessary to have a helper function.

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 fs/proc/task_mmu.c         |  2 +-
 include/linux/huge_mm.h    | 57 ++------------------------------------
 include/linux/khugepaged.h |  2 +-
 mm/huge_memory.c           | 24 +++++++++++-----
 mm/khugepaged.c            |  8 +++---
 mm/memory.c                |  6 ++--
 6 files changed, 29 insertions(+), 70 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index ab4849d07a1f..28f472a2eb51 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -863,7 +863,7 @@ static int show_smap(struct seq_file *m, void *v)
 	__show_smap(m, &mss, false);
 
 	seq_printf(m, "THPeligible:    %d\n",
-		   hugepage_vma_check(vma, vma->vm_flags));
+		   hugepage_vma_check(vma, vma->vm_flags, false));
 
 	if (arch_pkeys_enabled())
 		seq_printf(m, "ProtectionKey:  %8u\n", vma_pkey(vma));
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 2da52ca8fd52..67c913e66e60 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -153,48 +153,6 @@ static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
 	return true;
 }
 
-static inline bool transhuge_vma_enabled(struct vm_area_struct *vma,
-					  unsigned long vm_flags)
-{
-	/* Explicitly disabled through madvise. */
-	if ((vm_flags & VM_NOHUGEPAGE) ||
-	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags))
-		return false;
-	return true;
-}
-
-/*
- * to be used on vmas which are known to support THP.
- * Use transparent_hugepage_active otherwise
- */
-static inline bool __transparent_hugepage_enabled(struct vm_area_struct *vma)
-{
-
-	/*
-	 * If the hardware/firmware marked hugepage support disabled.
-	 */
-	if (transparent_hugepage_flags & (1 << TRANSPARENT_HUGEPAGE_NEVER_DAX))
-		return false;
-
-	if (!transhuge_vma_enabled(vma, vma->vm_flags))
-		return false;
-
-	if (vma_is_temporary_stack(vma))
-		return false;
-
-	if (transparent_hugepage_flags & (1 << TRANSPARENT_HUGEPAGE_FLAG))
-		return true;
-
-	if (vma_is_dax(vma))
-		return true;
-
-	if (transparent_hugepage_flags &
-				(1 << TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG))
-		return !!(vma->vm_flags & VM_HUGEPAGE);
-
-	return false;
-}
-
 static inline bool file_thp_enabled(struct vm_area_struct *vma)
 {
 	struct inode *inode;
@@ -210,7 +168,7 @@ static inline bool file_thp_enabled(struct vm_area_struct *vma)
 }
 
 extern bool hugepage_vma_check(struct vm_area_struct *vma,
-			       unsigned long vm_flags);
+			       unsigned long vm_flags, bool in_pf);
 
 #define transparent_hugepage_use_zero_page()				\
 	(transparent_hugepage_flags &					\
@@ -354,11 +312,6 @@ static inline bool folio_test_pmd_mappable(struct folio *folio)
 	return false;
 }
 
-static inline bool __transparent_hugepage_enabled(struct vm_area_struct *vma)
-{
-	return false;
-}
-
 static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
 {
 	return false;
@@ -370,14 +323,8 @@ static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
 	return false;
 }
 
-static inline bool transhuge_vma_enabled(struct vm_area_struct *vma,
-					  unsigned long vm_flags)
-{
-	return false;
-}
-
 static inline bool hugepage_vma_check(struct vm_area_struct *vma,
-				       unsigned long vm_flags)
+				       unsigned long vm_flags, bool in_pf)
 {
 	return false;
 }
diff --git a/include/linux/khugepaged.h b/include/linux/khugepaged.h
index 795c8d2170b0..8a6452e089ca 100644
--- a/include/linux/khugepaged.h
+++ b/include/linux/khugepaged.h
@@ -55,7 +55,7 @@ static inline void khugepaged_enter(struct vm_area_struct *vma,
 {
 	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags) &&
 	    khugepaged_enabled()) {
-		if (hugepage_vma_check(vma, vm_flags))
+		if (hugepage_vma_check(vma, vm_flags, false))
 			__khugepaged_enter(vma->vm_mm);
 	}
 }
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index d9ec82892b05..b4cab3aa2534 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -70,17 +70,26 @@ struct page *huge_zero_page __read_mostly;
 unsigned long huge_zero_pfn __read_mostly = ~0UL;
 
 bool hugepage_vma_check(struct vm_area_struct *vma,
-			unsigned long vm_flags)
+			unsigned long vm_flags, bool in_pf)
 {
-	if (!transhuge_vma_enabled(vma, vm_flags))
+	/* Explicitly disabled through madvise or prctl. */
+	if ((vm_flags & VM_NOHUGEPAGE) ||
+	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags))
 		return false;
 
+	/*
+	 * If the hardware/firmware marked hugepage support disabled.
+	 */
+	if (transparent_hugepage_flags & (1 << TRANSPARENT_HUGEPAGE_NEVER_DAX))
+		return false;
+
+	/* Special VMA and hugetlb VMA */
 	if (vm_flags & VM_NO_KHUGEPAGED)
 		return false;
 
-	/* Don't run khugepaged against DAX vma */
+	/* khugepaged doesn't collapse DAX vma, but page fault is fine. */
 	if (vma_is_dax(vma))
-		return false;
+		return in_pf;
 
 	if (vma->vm_file && !IS_ALIGNED((vma->vm_start >> PAGE_SHIFT) -
 				vma->vm_pgoff, HPAGE_PMD_NR))
@@ -90,7 +99,7 @@ bool hugepage_vma_check(struct vm_area_struct *vma,
 		return false;
 
 	/* Enabled via shmem mount options or sysfs settings. */
-	if (shmem_file(vma->vm_file))
+	if (!in_pf && shmem_file(vma->vm_file))
 		return shmem_huge_enabled(vma);
 
 	if (!khugepaged_enabled())
@@ -101,10 +110,11 @@ bool hugepage_vma_check(struct vm_area_struct *vma,
 		return false;
 
 	/* Only regular file is valid */
-	if (file_thp_enabled(vma))
+	if (!in_pf && file_thp_enabled(vma))
 		return true;
 
-	if (!vma->anon_vma || !vma_is_anonymous(vma))
+	/* anon_vma may be not prepared before the first page fault */
+	if ((!in_pf && !vma->anon_vma) || !vma_is_anonymous(vma))
 		return false;
 	if (vma_is_temporary_stack(vma))
 		return false;
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index a7ac1ae418f8..292ed9c0db1e 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -473,7 +473,7 @@ void khugepaged_enter_vma(struct vm_area_struct *vma,
 {
 	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags) &&
 	    khugepaged_enabled()) {
-		if (hugepage_vma_check(vma, vm_flags))
+		if (hugepage_vma_check(vma, vm_flags, false))
 			__khugepaged_enter(vma->vm_mm);
 	}
 }
@@ -918,7 +918,7 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
 
 	if (!transhuge_vma_suitable(vma, address))
 		return SCAN_ADDRESS_RANGE;
-	if (!hugepage_vma_check(vma, vma->vm_flags))
+	if (!hugepage_vma_check(vma, vma->vm_flags, false))
 		return SCAN_VMA_CHECK;
 	return 0;
 }
@@ -1399,7 +1399,7 @@ void collapse_pte_mapped_thp(struct mm_struct *mm, unsigned long addr)
 	 * the valid THP. Add extra VM_HUGEPAGE so hugepage_vma_check()
 	 * will not fail the vma for missing VM_HUGEPAGE
 	 */
-	if (!hugepage_vma_check(vma, vma->vm_flags | VM_HUGEPAGE))
+	if (!hugepage_vma_check(vma, vma->vm_flags | VM_HUGEPAGE, false))
 		return;
 
 	/* Keep pmd pgtable for uffd-wp; see comment in retract_page_tables() */
@@ -2091,7 +2091,7 @@ static unsigned int khugepaged_scan_mm_slot(unsigned int pages,
 			progress++;
 			break;
 		}
-		if (!hugepage_vma_check(vma, vma->vm_flags)) {
+		if (!hugepage_vma_check(vma, vma->vm_flags, false)) {
 skip:
 			progress++;
 			continue;
diff --git a/mm/memory.c b/mm/memory.c
index 95735bee3876..0850b3e38dbf 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -4979,7 +4979,8 @@ static vm_fault_t __handle_mm_fault(struct vm_area_struct *vma,
 	if (!vmf.pud)
 		return VM_FAULT_OOM;
 retry_pud:
-	if (pud_none(*vmf.pud) && __transparent_hugepage_enabled(vma)) {
+	if (pud_none(*vmf.pud) &&
+	    hugepage_vma_check(vma, vma->vm_flags, true)) {
 		ret = create_huge_pud(&vmf);
 		if (!(ret & VM_FAULT_FALLBACK))
 			return ret;
@@ -5012,7 +5013,8 @@ static vm_fault_t __handle_mm_fault(struct vm_area_struct *vma,
 	if (pud_trans_unstable(vmf.pud))
 		goto retry_pud;
 
-	if (pmd_none(*vmf.pmd) && __transparent_hugepage_enabled(vma)) {
+	if (pmd_none(*vmf.pmd) &&
+	    hugepage_vma_check(vma, vma->vm_flags, true)) {
 		ret = create_huge_pmd(&vmf);
 		if (!(ret & VM_FAULT_FALLBACK))
 			return ret;
-- 
2.26.3

[PATCH 7/7] mm: khugepaged: reorg some khugepaged helpers

[Yang Shi]

The khugepaged_{enabled|always|req_madv} are not khugepaged only
anymore, move them to huge_mm.h and rename to hugepage_flags_xxx, and
remove khugepaged_req_madv due to no users.

Also move khugepaged_defrag to khugepaged.c since its only caller is in
that file, it doesn't have to be in a header file.

Signed-off-by: Yang Shi <shy828301@gmail.com>
---
 include/linux/huge_mm.h    |  8 ++++++++
 include/linux/khugepaged.h | 17 +----------------
 mm/huge_memory.c           |  4 ++--
 mm/khugepaged.c            | 18 +++++++++++-------
 4 files changed, 22 insertions(+), 25 deletions(-)

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 67c913e66e60..e6f89d8ad5f8 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -116,6 +116,14 @@ extern struct kobj_attribute shmem_enabled_attr;
 
 extern unsigned long transparent_hugepage_flags;
 
+#define hugepage_flags_enabled()					       \
+	(transparent_hugepage_flags &				       \
+	 ((1<<TRANSPARENT_HUGEPAGE_FLAG) |		       \
+	  (1<<TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG)))
+#define hugepage_flags_always()				\
+	(transparent_hugepage_flags &			\
+	 (1<<TRANSPARENT_HUGEPAGE_FLAG))
+
 /*
  * The vma size has to be large enough to hold an aligned HPAGE_PMD_SIZE area.
  */
diff --git a/include/linux/khugepaged.h b/include/linux/khugepaged.h
index 8a6452e089ca..ca54e5c42ed0 100644
--- a/include/linux/khugepaged.h
+++ b/include/linux/khugepaged.h
@@ -24,20 +24,6 @@ static inline void collapse_pte_mapped_thp(struct mm_struct *mm,
 }
 #endif
 
-#define khugepaged_enabled()					       \
-	(transparent_hugepage_flags &				       \
-	 ((1<<TRANSPARENT_HUGEPAGE_FLAG) |		       \
-	  (1<<TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG)))
-#define khugepaged_always()				\
-	(transparent_hugepage_flags &			\
-	 (1<<TRANSPARENT_HUGEPAGE_FLAG))
-#define khugepaged_req_madv()					\
-	(transparent_hugepage_flags &				\
-	 (1<<TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG))
-#define khugepaged_defrag()					\
-	(transparent_hugepage_flags &				\
-	 (1<<TRANSPARENT_HUGEPAGE_DEFRAG_KHUGEPAGED_FLAG))
-
 static inline void khugepaged_fork(struct mm_struct *mm, struct mm_struct *oldmm)
 {
 	if (test_bit(MMF_VM_HUGEPAGE, &oldmm->flags))
@@ -53,8 +39,7 @@ static inline void khugepaged_exit(struct mm_struct *mm)
 static inline void khugepaged_enter(struct vm_area_struct *vma,
 				   unsigned long vm_flags)
 {
-	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags) &&
-	    khugepaged_enabled()) {
+	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags)) {
 		if (hugepage_vma_check(vma, vm_flags, false))
 			__khugepaged_enter(vma->vm_mm);
 	}
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index b4cab3aa2534..f4d7b6378b83 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -102,11 +102,11 @@ bool hugepage_vma_check(struct vm_area_struct *vma,
 	if (!in_pf && shmem_file(vma->vm_file))
 		return shmem_huge_enabled(vma);
 
-	if (!khugepaged_enabled())
+	if (!hugepage_flags_enabled())
 		return false;
 
 	/* THP settings require madvise. */
-	if (!(vm_flags & VM_HUGEPAGE) && !khugepaged_always())
+	if (!(vm_flags & VM_HUGEPAGE) && !hugepage_flags_always())
 		return false;
 
 	/* Only regular file is valid */
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 292ed9c0db1e..37ea5a5396cd 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -472,7 +472,7 @@ void khugepaged_enter_vma(struct vm_area_struct *vma,
 			  unsigned long vm_flags)
 {
 	if (!test_bit(MMF_VM_HUGEPAGE, &vma->vm_mm->flags) &&
-	    khugepaged_enabled()) {
+	    hugepage_flags_enabled()) {
 		if (hugepage_vma_check(vma, vm_flags, false))
 			__khugepaged_enter(vma->vm_mm);
 	}
@@ -763,6 +763,10 @@ static bool khugepaged_scan_abort(int nid)
 	return false;
 }
 
+#define khugepaged_defrag()					\
+	(transparent_hugepage_flags &				\
+	 (1<<TRANSPARENT_HUGEPAGE_DEFRAG_KHUGEPAGED_FLAG))
+
 /* Defrag for khugepaged will enter direct reclaim/compaction if necessary */
 static inline gfp_t alloc_hugepage_khugepaged_gfpmask(void)
 {
@@ -860,7 +864,7 @@ static struct page *khugepaged_alloc_hugepage(bool *wait)
 			khugepaged_alloc_sleep();
 		} else
 			count_vm_event(THP_COLLAPSE_ALLOC);
-	} while (unlikely(!hpage) && likely(khugepaged_enabled()));
+	} while (unlikely(!hpage) && likely(hugepage_flags_enabled()));
 
 	return hpage;
 }
@@ -2175,7 +2179,7 @@ static unsigned int khugepaged_scan_mm_slot(unsigned int pages,
 static int khugepaged_has_work(void)
 {
 	return !list_empty(&khugepaged_scan.mm_head) &&
-		khugepaged_enabled();
+		hugepage_flags_enabled();
 }
 
 static int khugepaged_wait_event(void)
@@ -2240,7 +2244,7 @@ static void khugepaged_wait_work(void)
 		return;
 	}
 
-	if (khugepaged_enabled())
+	if (hugepage_flags_enabled())
 		wait_event_freezable(khugepaged_wait, khugepaged_wait_event());
 }
 
@@ -2271,7 +2275,7 @@ static void set_recommended_min_free_kbytes(void)
 	int nr_zones = 0;
 	unsigned long recommended_min;
 
-	if (!khugepaged_enabled()) {
+	if (!hugepage_flags_enabled()) {
 		calculate_min_free_kbytes();
 		goto update_wmarks;
 	}
@@ -2321,7 +2325,7 @@ int start_stop_khugepaged(void)
 	int err = 0;
 
 	mutex_lock(&khugepaged_mutex);
-	if (khugepaged_enabled()) {
+	if (hugepage_flags_enabled()) {
 		if (!khugepaged_thread)
 			khugepaged_thread = kthread_run(khugepaged, NULL,
 							"khugepaged");
@@ -2347,7 +2351,7 @@ int start_stop_khugepaged(void)
 void khugepaged_min_free_kbytes_update(void)
 {
 	mutex_lock(&khugepaged_mutex);
-	if (khugepaged_enabled() && khugepaged_thread)
+	if (hugepage_flags_enabled() && khugepaged_thread)
 		set_recommended_min_free_kbytes();
 	mutex_unlock(&khugepaged_mutex);
 }
-- 
2.26.3

Re: [PATCH 2/7] mm: thp: introduce transhuge_vma_size_ok() helper

[Hillf Danton]

On Fri, 20 May 2022 14:16:00 -0700 Yang Shi wrote:
> There are couple of places that check whether the vma size is ok for
> THP or not, they are open coded and duplicate, introduce
> transhuge_vma_size_ok() helper to do the job.
> 
> Signed-off-by: Yang Shi <shy828301@gmail.com>
> ---
>  include/linux/huge_mm.h | 17 +++++++++++++++++
>  mm/huge_memory.c        |  5 +----
>  mm/khugepaged.c         | 12 ++++++------
>  3 files changed, 24 insertions(+), 10 deletions(-)
> 
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index 648cb3ce7099..a8f61db47f2a 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -116,6 +116,18 @@ extern struct kobj_attribute shmem_enabled_attr;
>  
>  extern unsigned long transparent_hugepage_flags;
>  
> +/*
> + * The vma size has to be large enough to hold an aligned HPAGE_PMD_SIZE area.
> + */
> +static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
> +{
> +	if (round_up(vma->vm_start, HPAGE_PMD_SIZE) <
> +	    (vma->vm_end & HPAGE_PMD_MASK))
> +		return true;
> +
> +	return false;
> +}
> +
>  static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
>  		unsigned long addr)
>  {
> @@ -345,6 +357,11 @@ static inline bool transparent_hugepage_active(struct vm_area_struct *vma)
>  	return false;
>  }
>  
> +static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
> +{
> +	return false;
> +}
> +
>  static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
>  		unsigned long addr)
>  {
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 80e8b58b4f39..d633f97452c1 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -71,10 +71,7 @@ unsigned long huge_zero_pfn __read_mostly = ~0UL;
>  
>  bool transparent_hugepage_active(struct vm_area_struct *vma)
>  {
> -	/* The addr is used to check if the vma size fits */
> -	unsigned long addr = (vma->vm_end & HPAGE_PMD_MASK) - HPAGE_PMD_SIZE;
> -
> -	if (!transhuge_vma_suitable(vma, addr))
> +	if (!transhuge_vma_size_ok(vma))
>  		return false;

Given the comment added in 4/7, this patch adds change in semantic.
If that is intened, add some words in log message for it.

>  	if (vma_is_anonymous(vma))
>  		return __transparent_hugepage_enabled(vma);

Re: [PATCH 2/7] mm: thp: introduce transhuge_vma_size_ok() helper

[Yang Shi]

On Fri, May 20, 2022 at 6:40 PM Hillf Danton <hdanton@sina.com> wrote:
>
> On Fri, 20 May 2022 14:16:00 -0700 Yang Shi wrote:
> > There are couple of places that check whether the vma size is ok for
> > THP or not, they are open coded and duplicate, introduce
> > transhuge_vma_size_ok() helper to do the job.
> >
> > Signed-off-by: Yang Shi <shy828301@gmail.com>
> > ---
> >  include/linux/huge_mm.h | 17 +++++++++++++++++
> >  mm/huge_memory.c        |  5 +----
> >  mm/khugepaged.c         | 12 ++++++------
> >  3 files changed, 24 insertions(+), 10 deletions(-)
> >
> > diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> > index 648cb3ce7099..a8f61db47f2a 100644
> > --- a/include/linux/huge_mm.h
> > +++ b/include/linux/huge_mm.h
> > @@ -116,6 +116,18 @@ extern struct kobj_attribute shmem_enabled_attr;
> >
> >  extern unsigned long transparent_hugepage_flags;
> >
> > +/*
> > + * The vma size has to be large enough to hold an aligned HPAGE_PMD_SIZE area.
> > + */
> > +static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
> > +{
> > +     if (round_up(vma->vm_start, HPAGE_PMD_SIZE) <
> > +         (vma->vm_end & HPAGE_PMD_MASK))
> > +             return true;
> > +
> > +     return false;
> > +}
> > +
> >  static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
> >               unsigned long addr)
> >  {
> > @@ -345,6 +357,11 @@ static inline bool transparent_hugepage_active(struct vm_area_struct *vma)
> >       return false;
> >  }
> >
> > +static inline bool transhuge_vma_size_ok(struct vm_area_struct *vma)
> > +{
> > +     return false;
> > +}
> > +
> >  static inline bool transhuge_vma_suitable(struct vm_area_struct *vma,
> >               unsigned long addr)
> >  {
> > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > index 80e8b58b4f39..d633f97452c1 100644
> > --- a/mm/huge_memory.c
> > +++ b/mm/huge_memory.c
> > @@ -71,10 +71,7 @@ unsigned long huge_zero_pfn __read_mostly = ~0UL;
> >
> >  bool transparent_hugepage_active(struct vm_area_struct *vma)
> >  {
> > -     /* The addr is used to check if the vma size fits */
> > -     unsigned long addr = (vma->vm_end & HPAGE_PMD_MASK) - HPAGE_PMD_SIZE;
> > -
> > -     if (!transhuge_vma_suitable(vma, addr))
> > +     if (!transhuge_vma_size_ok(vma))
> >               return false;
>
> Given the comment added in 4/7, this patch adds change in semantic.
> If that is intened, add some words in log message for it.

It should not change the semantic even with 4/7. The old comment says
"The addr is used to check if the vma size fits", so the purpose of
transhuge_vma_suitable() here is used to check if the vma size is ok
or not.

>
> >       if (vma_is_anonymous(vma))
> >               return __transparent_hugepage_enabled(vma);
>