[PATCH v2 0/3] Some cleanups for shmem

[PATCH v2 0/3] Some cleanups for shmem

[Baolin Wang]

Hi,

This series does some cleanups to reuse code, rename functions and simplify
logic to make code more clear. No functional changes are expected.

Changes from v1:
 - Add a dummy function in case CONFIG_TRANSPARENT_HUGEPAGE is not
 enabled, which fixes a building error reported by kernel test robot.

Baolin Wang (3):
  mm: shmem: simplify the suitable huge orders validation for tmpfs
  mm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()
  mm: shmem: move shmem_huge_global_enabled() into
    shmem_allowable_huge_orders()

 include/linux/shmem_fs.h | 11 +----
 mm/huge_memory.c         | 11 ++---
 mm/shmem.c               | 91 +++++++++++++++++++++-------------------
 3 files changed, 53 insertions(+), 60 deletions(-)

-- 
2.39.3

[PATCH v2 1/3] mm: shmem: simplify the suitable huge orders validation for tmpfs

[Baolin Wang]

Move the suitable huge orders validation into shmem_suitable_orders() for
tmpfs, which can reuse some code to simplify the logic.

In addition, we don't have special handling for the error code -E2BIG when
checking for conflicts with PMD sized THP in the pagecache for tmpfs, instead,
it will just fallback to order-0 allocations like this patch does, so this
simplification will not add functional changes.

Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 mm/shmem.c | 39 +++++++++++++++------------------------
 1 file changed, 15 insertions(+), 24 deletions(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index f24dfbd387ba..db7e9808830f 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1685,19 +1685,29 @@ static unsigned long shmem_suitable_orders(struct inode *inode, struct vm_fault
 					   struct address_space *mapping, pgoff_t index,
 					   unsigned long orders)
 {
-	struct vm_area_struct *vma = vmf->vma;
+	struct vm_area_struct *vma = vmf ? vmf->vma : NULL;
 	unsigned long pages;
 	int order;
 
-	orders = thp_vma_suitable_orders(vma, vmf->address, orders);
-	if (!orders)
-		return 0;
+	if (vma) {
+		orders = thp_vma_suitable_orders(vma, vmf->address, orders);
+		if (!orders)
+			return 0;
+	}
 
 	/* Find the highest order that can add into the page cache */
 	order = highest_order(orders);
 	while (orders) {
 		pages = 1UL << order;
 		index = round_down(index, pages);
+		/*
+		 * Check for conflict before waiting on a huge allocation.
+		 * Conflict might be that a huge page has just been allocated
+		 * and added to page cache by a racing thread, or that there
+		 * is already at least one small page in the huge extent.
+		 * Be careful to retry when appropriate, but not forever!
+		 * Elsewhere -EEXIST would be the right code, but not here.
+		 */
 		if (!xa_find(&mapping->i_pages, &index,
 			     index + pages - 1, XA_PRESENT))
 			break;
@@ -1735,7 +1745,6 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf,
 {
 	struct address_space *mapping = inode->i_mapping;
 	struct shmem_inode_info *info = SHMEM_I(inode);
-	struct vm_area_struct *vma = vmf ? vmf->vma : NULL;
 	unsigned long suitable_orders = 0;
 	struct folio *folio = NULL;
 	long pages;
@@ -1745,26 +1754,8 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf,
 		orders = 0;
 
 	if (orders > 0) {
-		if (vma && vma_is_anon_shmem(vma)) {
-			suitable_orders = shmem_suitable_orders(inode, vmf,
+		suitable_orders = shmem_suitable_orders(inode, vmf,
 							mapping, index, orders);
-		} else if (orders & BIT(HPAGE_PMD_ORDER)) {
-			pages = HPAGE_PMD_NR;
-			suitable_orders = BIT(HPAGE_PMD_ORDER);
-			index = round_down(index, HPAGE_PMD_NR);
-
-			/*
-			 * Check for conflict before waiting on a huge allocation.
-			 * Conflict might be that a huge page has just been allocated
-			 * and added to page cache by a racing thread, or that there
-			 * is already at least one small page in the huge extent.
-			 * Be careful to retry when appropriate, but not forever!
-			 * Elsewhere -EEXIST would be the right code, but not here.
-			 */
-			if (xa_find(&mapping->i_pages, &index,
-				    index + HPAGE_PMD_NR - 1, XA_PRESENT))
-				return ERR_PTR(-E2BIG);
-		}
 
 		order = highest_order(suitable_orders);
 		while (suitable_orders) {
-- 
2.39.3

[PATCH v2 2/3] mm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()

[Baolin Wang]

The shmem_is_huge() is now used to check if the top-level huge page is enabled,
thus rename it to reflect its usage.

Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 include/linux/shmem_fs.h |  9 +++++----
 mm/huge_memory.c         |  5 +++--
 mm/shmem.c               | 15 ++++++++-------
 3 files changed, 16 insertions(+), 13 deletions(-)

diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
index 1d06b1e5408a..405ee8d3589a 100644
--- a/include/linux/shmem_fs.h
+++ b/include/linux/shmem_fs.h
@@ -111,14 +111,15 @@ extern void shmem_truncate_range(struct inode *inode, loff_t start, loff_t end);
 int shmem_unuse(unsigned int type);
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-extern bool shmem_is_huge(struct inode *inode, pgoff_t index, bool shmem_huge_force,
-			  struct mm_struct *mm, unsigned long vm_flags);
+extern bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index, bool shmem_huge_force,
+				      struct mm_struct *mm, unsigned long vm_flags);
 unsigned long shmem_allowable_huge_orders(struct inode *inode,
 				struct vm_area_struct *vma, pgoff_t index,
 				bool global_huge);
 #else
-static __always_inline bool shmem_is_huge(struct inode *inode, pgoff_t index, bool shmem_huge_force,
-					  struct mm_struct *mm, unsigned long vm_flags)
+static __always_inline bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
+						      bool shmem_huge_force, struct mm_struct *mm,
+						      unsigned long vm_flags)
 {
 	return false;
 }
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index f9696c94e211..cc9bad12be75 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -152,8 +152,9 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
 	 * own flags.
 	 */
 	if (!in_pf && shmem_file(vma->vm_file)) {
-		bool global_huge = shmem_is_huge(file_inode(vma->vm_file), vma->vm_pgoff,
-							!enforce_sysfs, vma->vm_mm, vm_flags);
+		bool global_huge = shmem_huge_global_enabled(file_inode(vma->vm_file),
+							     vma->vm_pgoff, !enforce_sysfs,
+							     vma->vm_mm, vm_flags);
 
 		if (!vma_is_anon_shmem(vma))
 			return global_huge ? orders : 0;
diff --git a/mm/shmem.c b/mm/shmem.c
index db7e9808830f..1445dcd39b6f 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -548,9 +548,9 @@ static bool shmem_confirm_swap(struct address_space *mapping,
 
 static int shmem_huge __read_mostly = SHMEM_HUGE_NEVER;
 
-static bool __shmem_is_huge(struct inode *inode, pgoff_t index,
-			    bool shmem_huge_force, struct mm_struct *mm,
-			    unsigned long vm_flags)
+static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
+					bool shmem_huge_force, struct mm_struct *mm,
+					unsigned long vm_flags)
 {
 	loff_t i_size;
 
@@ -581,14 +581,15 @@ static bool __shmem_is_huge(struct inode *inode, pgoff_t index,
 	}
 }
 
-bool shmem_is_huge(struct inode *inode, pgoff_t index,
+bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
 		   bool shmem_huge_force, struct mm_struct *mm,
 		   unsigned long vm_flags)
 {
 	if (HPAGE_PMD_ORDER > MAX_PAGECACHE_ORDER)
 		return false;
 
-	return __shmem_is_huge(inode, index, shmem_huge_force, mm, vm_flags);
+	return __shmem_huge_global_enabled(inode, index, shmem_huge_force,
+					   mm, vm_flags);
 }
 
 #if defined(CONFIG_SYSFS)
@@ -1156,7 +1157,7 @@ static int shmem_getattr(struct mnt_idmap *idmap,
 			STATX_ATTR_NODUMP);
 	generic_fillattr(idmap, request_mask, inode, stat);
 
-	if (shmem_is_huge(inode, 0, false, NULL, 0))
+	if (shmem_huge_global_enabled(inode, 0, false, NULL, 0))
 		stat->blksize = HPAGE_PMD_SIZE;
 
 	if (request_mask & STATX_BTIME) {
@@ -2153,7 +2154,7 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
 		return 0;
 	}
 
-	huge = shmem_is_huge(inode, index, false, fault_mm,
+	huge = shmem_huge_global_enabled(inode, index, false, fault_mm,
 			     vma ? vma->vm_flags : 0);
 	/* Find hugepage orders that are allowed for anonymous shmem. */
 	if (vma && vma_is_anon_shmem(vma))
-- 
2.39.3

[PATCH v2 3/3] mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

[Baolin Wang]

Move shmem_huge_global_enabled() into the shmem_allowable_huge_orders() function,
so that shmem_allowable_huge_orders() can also help to find the allowable huge
orders for tmpfs. Moreover the shmem_huge_global_enabled() can become static.

No functional changes.

Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 include/linux/shmem_fs.h | 12 ++----------
 mm/huge_memory.c         | 12 +++---------
 mm/shmem.c               | 41 ++++++++++++++++++++++++++--------------
 3 files changed, 32 insertions(+), 33 deletions(-)

diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
index 405ee8d3589a..1564d7d3ca61 100644
--- a/include/linux/shmem_fs.h
+++ b/include/linux/shmem_fs.h
@@ -111,21 +111,13 @@ extern void shmem_truncate_range(struct inode *inode, loff_t start, loff_t end);
 int shmem_unuse(unsigned int type);
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-extern bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index, bool shmem_huge_force,
-				      struct mm_struct *mm, unsigned long vm_flags);
 unsigned long shmem_allowable_huge_orders(struct inode *inode,
 				struct vm_area_struct *vma, pgoff_t index,
-				bool global_huge);
+				bool shmem_huge_force);
 #else
-static __always_inline bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
-						      bool shmem_huge_force, struct mm_struct *mm,
-						      unsigned long vm_flags)
-{
-	return false;
-}
 static inline unsigned long shmem_allowable_huge_orders(struct inode *inode,
 				struct vm_area_struct *vma, pgoff_t index,
-				bool global_huge)
+				bool shmem_huge_force)
 {
 	return 0;
 }
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index cc9bad12be75..f69980b5b5fc 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -151,16 +151,10 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
 	 * Must be done before hugepage flags check since shmem has its
 	 * own flags.
 	 */
-	if (!in_pf && shmem_file(vma->vm_file)) {
-		bool global_huge = shmem_huge_global_enabled(file_inode(vma->vm_file),
-							     vma->vm_pgoff, !enforce_sysfs,
-							     vma->vm_mm, vm_flags);
-
-		if (!vma_is_anon_shmem(vma))
-			return global_huge ? orders : 0;
+	if (!in_pf && shmem_file(vma->vm_file))
 		return shmem_allowable_huge_orders(file_inode(vma->vm_file),
-							vma, vma->vm_pgoff, global_huge);
-	}
+						   vma, vma->vm_pgoff,
+						   !enforce_sysfs);
 
 	if (!vma_is_anonymous(vma)) {
 		/*
diff --git a/mm/shmem.c b/mm/shmem.c
index 1445dcd39b6f..4d274f5a17d9 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -581,7 +581,7 @@ static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
 	}
 }
 
-bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
+static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
 		   bool shmem_huge_force, struct mm_struct *mm,
 		   unsigned long vm_flags)
 {
@@ -772,6 +772,13 @@ static unsigned long shmem_unused_huge_shrink(struct shmem_sb_info *sbinfo,
 {
 	return 0;
 }
+
+static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
+		bool shmem_huge_force, struct mm_struct *mm,
+		unsigned long vm_flags)
+{
+	return false;
+}
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 
 /*
@@ -1625,27 +1632,39 @@ static gfp_t limit_gfp_mask(gfp_t huge_gfp, gfp_t limit_gfp)
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 unsigned long shmem_allowable_huge_orders(struct inode *inode,
 				struct vm_area_struct *vma, pgoff_t index,
-				bool global_huge)
+				bool shmem_huge_force)
 {
 	unsigned long mask = READ_ONCE(huge_shmem_orders_always);
 	unsigned long within_size_orders = READ_ONCE(huge_shmem_orders_within_size);
-	unsigned long vm_flags = vma->vm_flags;
+	unsigned long vm_flags = vma ? vma->vm_flags : 0;
+	struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;
 	/*
 	 * Check all the (large) orders below HPAGE_PMD_ORDER + 1 that
 	 * are enabled for this vma.
 	 */
 	unsigned long orders = BIT(PMD_ORDER + 1) - 1;
+	bool global_huge;
 	loff_t i_size;
 	int order;
 
-	if ((vm_flags & VM_NOHUGEPAGE) ||
-	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags))
+	if (vma && ((vm_flags & VM_NOHUGEPAGE) ||
+	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags)))
 		return 0;
 
 	/* If the hardware/firmware marked hugepage support disabled. */
 	if (transparent_hugepage_flags & (1 << TRANSPARENT_HUGEPAGE_UNSUPPORTED))
 		return 0;
 
+	global_huge = shmem_huge_global_enabled(inode, index, shmem_huge_force,
+						fault_mm, vm_flags);
+	if (!vma || !vma_is_anon_shmem(vma)) {
+		/*
+		 * For tmpfs, we now only support PMD sized THP if huge page
+		 * is enabled, otherwise fallback to order 0.
+		 */
+		return global_huge ? BIT(HPAGE_PMD_ORDER) : 0;
+	}
+
 	/*
 	 * Following the 'deny' semantics of the top level, force the huge
 	 * option off from all mounts.
@@ -2081,7 +2100,7 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
 	struct mm_struct *fault_mm;
 	struct folio *folio;
 	int error;
-	bool alloced, huge;
+	bool alloced;
 	unsigned long orders = 0;
 
 	if (WARN_ON_ONCE(!shmem_mapping(inode->i_mapping)))
@@ -2154,14 +2173,8 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
 		return 0;
 	}
 
-	huge = shmem_huge_global_enabled(inode, index, false, fault_mm,
-			     vma ? vma->vm_flags : 0);
-	/* Find hugepage orders that are allowed for anonymous shmem. */
-	if (vma && vma_is_anon_shmem(vma))
-		orders = shmem_allowable_huge_orders(inode, vma, index, huge);
-	else if (huge)
-		orders = BIT(HPAGE_PMD_ORDER);
-
+	/* Find hugepage orders that are allowed for anonymous shmem and tmpfs. */
+	orders = shmem_allowable_huge_orders(inode, vma, index, false);
 	if (orders > 0) {
 		gfp_t huge_gfp;
 
-- 
2.39.3

Re: [PATCH v2 1/3] mm: shmem: simplify the suitable huge orders validation for tmpfs

[Ryan Roberts]

On 13/07/2024 14:24, Baolin Wang wrote:
> Move the suitable huge orders validation into shmem_suitable_orders() for
> tmpfs, which can reuse some code to simplify the logic.
> 
> In addition, we don't have special handling for the error code -E2BIG when
> checking for conflicts with PMD sized THP in the pagecache for tmpfs, instead,
> it will just fallback to order-0 allocations like this patch does, so this
> simplification will not add functional changes.
> 
> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>

Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>

> ---
>  mm/shmem.c | 39 +++++++++++++++------------------------
>  1 file changed, 15 insertions(+), 24 deletions(-)
> 
> diff --git a/mm/shmem.c b/mm/shmem.c
> index f24dfbd387ba..db7e9808830f 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -1685,19 +1685,29 @@ static unsigned long shmem_suitable_orders(struct inode *inode, struct vm_fault
>  					   struct address_space *mapping, pgoff_t index,
>  					   unsigned long orders)
>  {
> -	struct vm_area_struct *vma = vmf->vma;
> +	struct vm_area_struct *vma = vmf ? vmf->vma : NULL;
>  	unsigned long pages;
>  	int order;
>  
> -	orders = thp_vma_suitable_orders(vma, vmf->address, orders);
> -	if (!orders)
> -		return 0;
> +	if (vma) {
> +		orders = thp_vma_suitable_orders(vma, vmf->address, orders);
> +		if (!orders)
> +			return 0;
> +	}
>  
>  	/* Find the highest order that can add into the page cache */
>  	order = highest_order(orders);
>  	while (orders) {
>  		pages = 1UL << order;
>  		index = round_down(index, pages);
> +		/*
> +		 * Check for conflict before waiting on a huge allocation.
> +		 * Conflict might be that a huge page has just been allocated
> +		 * and added to page cache by a racing thread, or that there
> +		 * is already at least one small page in the huge extent.
> +		 * Be careful to retry when appropriate, but not forever!
> +		 * Elsewhere -EEXIST would be the right code, but not here.
> +		 */
>  		if (!xa_find(&mapping->i_pages, &index,
>  			     index + pages - 1, XA_PRESENT))
>  			break;
> @@ -1735,7 +1745,6 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf,
>  {
>  	struct address_space *mapping = inode->i_mapping;
>  	struct shmem_inode_info *info = SHMEM_I(inode);
> -	struct vm_area_struct *vma = vmf ? vmf->vma : NULL;
>  	unsigned long suitable_orders = 0;
>  	struct folio *folio = NULL;
>  	long pages;
> @@ -1745,26 +1754,8 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf,
>  		orders = 0;
>  
>  	if (orders > 0) {
> -		if (vma && vma_is_anon_shmem(vma)) {
> -			suitable_orders = shmem_suitable_orders(inode, vmf,
> +		suitable_orders = shmem_suitable_orders(inode, vmf,
>  							mapping, index, orders);
> -		} else if (orders & BIT(HPAGE_PMD_ORDER)) {
> -			pages = HPAGE_PMD_NR;
> -			suitable_orders = BIT(HPAGE_PMD_ORDER);
> -			index = round_down(index, HPAGE_PMD_NR);
> -
> -			/*
> -			 * Check for conflict before waiting on a huge allocation.
> -			 * Conflict might be that a huge page has just been allocated
> -			 * and added to page cache by a racing thread, or that there
> -			 * is already at least one small page in the huge extent.
> -			 * Be careful to retry when appropriate, but not forever!
> -			 * Elsewhere -EEXIST would be the right code, but not here.
> -			 */
> -			if (xa_find(&mapping->i_pages, &index,
> -				    index + HPAGE_PMD_NR - 1, XA_PRESENT))
> -				return ERR_PTR(-E2BIG);
> -		}
>  
>  		order = highest_order(suitable_orders);
>  		while (suitable_orders) {

Re: [PATCH v2 2/3] mm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()

[Ryan Roberts]

On 13/07/2024 14:24, Baolin Wang wrote:
> The shmem_is_huge() is now used to check if the top-level huge page is enabled,
> thus rename it to reflect its usage.
> 
> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>

Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>

> ---
>  include/linux/shmem_fs.h |  9 +++++----
>  mm/huge_memory.c         |  5 +++--
>  mm/shmem.c               | 15 ++++++++-------
>  3 files changed, 16 insertions(+), 13 deletions(-)
> 
> diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
> index 1d06b1e5408a..405ee8d3589a 100644
> --- a/include/linux/shmem_fs.h
> +++ b/include/linux/shmem_fs.h
> @@ -111,14 +111,15 @@ extern void shmem_truncate_range(struct inode *inode, loff_t start, loff_t end);
>  int shmem_unuse(unsigned int type);
>  
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> -extern bool shmem_is_huge(struct inode *inode, pgoff_t index, bool shmem_huge_force,
> -			  struct mm_struct *mm, unsigned long vm_flags);
> +extern bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index, bool shmem_huge_force,
> +				      struct mm_struct *mm, unsigned long vm_flags);
>  unsigned long shmem_allowable_huge_orders(struct inode *inode,
>  				struct vm_area_struct *vma, pgoff_t index,
>  				bool global_huge);
>  #else
> -static __always_inline bool shmem_is_huge(struct inode *inode, pgoff_t index, bool shmem_huge_force,
> -					  struct mm_struct *mm, unsigned long vm_flags)
> +static __always_inline bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
> +						      bool shmem_huge_force, struct mm_struct *mm,
> +						      unsigned long vm_flags)
>  {
>  	return false;
>  }
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index f9696c94e211..cc9bad12be75 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -152,8 +152,9 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
>  	 * own flags.
>  	 */
>  	if (!in_pf && shmem_file(vma->vm_file)) {
> -		bool global_huge = shmem_is_huge(file_inode(vma->vm_file), vma->vm_pgoff,
> -							!enforce_sysfs, vma->vm_mm, vm_flags);
> +		bool global_huge = shmem_huge_global_enabled(file_inode(vma->vm_file),
> +							     vma->vm_pgoff, !enforce_sysfs,
> +							     vma->vm_mm, vm_flags);
>  
>  		if (!vma_is_anon_shmem(vma))
>  			return global_huge ? orders : 0;
> diff --git a/mm/shmem.c b/mm/shmem.c
> index db7e9808830f..1445dcd39b6f 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -548,9 +548,9 @@ static bool shmem_confirm_swap(struct address_space *mapping,
>  
>  static int shmem_huge __read_mostly = SHMEM_HUGE_NEVER;
>  
> -static bool __shmem_is_huge(struct inode *inode, pgoff_t index,
> -			    bool shmem_huge_force, struct mm_struct *mm,
> -			    unsigned long vm_flags)
> +static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
> +					bool shmem_huge_force, struct mm_struct *mm,
> +					unsigned long vm_flags)
>  {
>  	loff_t i_size;
>  
> @@ -581,14 +581,15 @@ static bool __shmem_is_huge(struct inode *inode, pgoff_t index,
>  	}
>  }
>  
> -bool shmem_is_huge(struct inode *inode, pgoff_t index,
> +bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>  		   bool shmem_huge_force, struct mm_struct *mm,
>  		   unsigned long vm_flags)
>  {
>  	if (HPAGE_PMD_ORDER > MAX_PAGECACHE_ORDER)
>  		return false;
>  
> -	return __shmem_is_huge(inode, index, shmem_huge_force, mm, vm_flags);
> +	return __shmem_huge_global_enabled(inode, index, shmem_huge_force,
> +					   mm, vm_flags);
>  }
>  
>  #if defined(CONFIG_SYSFS)
> @@ -1156,7 +1157,7 @@ static int shmem_getattr(struct mnt_idmap *idmap,
>  			STATX_ATTR_NODUMP);
>  	generic_fillattr(idmap, request_mask, inode, stat);
>  
> -	if (shmem_is_huge(inode, 0, false, NULL, 0))
> +	if (shmem_huge_global_enabled(inode, 0, false, NULL, 0))
>  		stat->blksize = HPAGE_PMD_SIZE;
>  
>  	if (request_mask & STATX_BTIME) {
> @@ -2153,7 +2154,7 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
>  		return 0;
>  	}
>  
> -	huge = shmem_is_huge(inode, index, false, fault_mm,
> +	huge = shmem_huge_global_enabled(inode, index, false, fault_mm,
>  			     vma ? vma->vm_flags : 0);
>  	/* Find hugepage orders that are allowed for anonymous shmem. */
>  	if (vma && vma_is_anon_shmem(vma))

Re: [PATCH v2 3/3] mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

[Ryan Roberts]

On 13/07/2024 14:24, Baolin Wang wrote:
> Move shmem_huge_global_enabled() into the shmem_allowable_huge_orders() function,
> so that shmem_allowable_huge_orders() can also help to find the allowable huge
> orders for tmpfs. Moreover the shmem_huge_global_enabled() can become static.
> 
> No functional changes.
> 
> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>

one nit below, but either way:

Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>

> ---
>  include/linux/shmem_fs.h | 12 ++----------
>  mm/huge_memory.c         | 12 +++---------
>  mm/shmem.c               | 41 ++++++++++++++++++++++++++--------------
>  3 files changed, 32 insertions(+), 33 deletions(-)
> 
> diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
> index 405ee8d3589a..1564d7d3ca61 100644
> --- a/include/linux/shmem_fs.h
> +++ b/include/linux/shmem_fs.h
> @@ -111,21 +111,13 @@ extern void shmem_truncate_range(struct inode *inode, loff_t start, loff_t end);
>  int shmem_unuse(unsigned int type);
>  
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> -extern bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index, bool shmem_huge_force,
> -				      struct mm_struct *mm, unsigned long vm_flags);
>  unsigned long shmem_allowable_huge_orders(struct inode *inode,
>  				struct vm_area_struct *vma, pgoff_t index,
> -				bool global_huge);
> +				bool shmem_huge_force);
>  #else
> -static __always_inline bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
> -						      bool shmem_huge_force, struct mm_struct *mm,
> -						      unsigned long vm_flags)
> -{
> -	return false;
> -}
>  static inline unsigned long shmem_allowable_huge_orders(struct inode *inode,
>  				struct vm_area_struct *vma, pgoff_t index,
> -				bool global_huge)
> +				bool shmem_huge_force)
>  {
>  	return 0;
>  }
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index cc9bad12be75..f69980b5b5fc 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -151,16 +151,10 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
>  	 * Must be done before hugepage flags check since shmem has its
>  	 * own flags.
>  	 */
> -	if (!in_pf && shmem_file(vma->vm_file)) {
> -		bool global_huge = shmem_huge_global_enabled(file_inode(vma->vm_file),
> -							     vma->vm_pgoff, !enforce_sysfs,
> -							     vma->vm_mm, vm_flags);
> -
> -		if (!vma_is_anon_shmem(vma))
> -			return global_huge ? orders : 0;
> +	if (!in_pf && shmem_file(vma->vm_file))
>  		return shmem_allowable_huge_orders(file_inode(vma->vm_file),
> -							vma, vma->vm_pgoff, global_huge);
> -	}
> +						   vma, vma->vm_pgoff,
> +						   !enforce_sysfs);
>  
>  	if (!vma_is_anonymous(vma)) {
>  		/*
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 1445dcd39b6f..4d274f5a17d9 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -581,7 +581,7 @@ static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>  	}
>  }
>  
> -bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
> +static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>  		   bool shmem_huge_force, struct mm_struct *mm,
>  		   unsigned long vm_flags)
>  {
> @@ -772,6 +772,13 @@ static unsigned long shmem_unused_huge_shrink(struct shmem_sb_info *sbinfo,
>  {
>  	return 0;
>  }
> +
> +static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
> +		bool shmem_huge_force, struct mm_struct *mm,
> +		unsigned long vm_flags)
> +{
> +	return false;
> +}
>  #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
>  
>  /*
> @@ -1625,27 +1632,39 @@ static gfp_t limit_gfp_mask(gfp_t huge_gfp, gfp_t limit_gfp)
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>  unsigned long shmem_allowable_huge_orders(struct inode *inode,
>  				struct vm_area_struct *vma, pgoff_t index,
> -				bool global_huge)
> +				bool shmem_huge_force)
>  {
>  	unsigned long mask = READ_ONCE(huge_shmem_orders_always);
>  	unsigned long within_size_orders = READ_ONCE(huge_shmem_orders_within_size);
> -	unsigned long vm_flags = vma->vm_flags;
> +	unsigned long vm_flags = vma ? vma->vm_flags : 0;
> +	struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;

nit: rather than deriving the fault_mm here, I wonder if its cleaner to just
pass vma to shmem_huge_global_enabled()? shmem_huge_global_enabled() is just
using it as a guard to access vm_flags, which you can just as easily do by
testing the vma for non-NULL. And you can access mm flags with vma->vm_mm->flags
after testing the vma too.

>  	/*
>  	 * Check all the (large) orders below HPAGE_PMD_ORDER + 1 that
>  	 * are enabled for this vma.
>  	 */
>  	unsigned long orders = BIT(PMD_ORDER + 1) - 1;
> +	bool global_huge;
>  	loff_t i_size;
>  	int order;
>  
> -	if ((vm_flags & VM_NOHUGEPAGE) ||
> -	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags))
> +	if (vma && ((vm_flags & VM_NOHUGEPAGE) ||
> +	    test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags)))
>  		return 0;
>  
>  	/* If the hardware/firmware marked hugepage support disabled. */
>  	if (transparent_hugepage_flags & (1 << TRANSPARENT_HUGEPAGE_UNSUPPORTED))
>  		return 0;
>  
> +	global_huge = shmem_huge_global_enabled(inode, index, shmem_huge_force,
> +						fault_mm, vm_flags);
> +	if (!vma || !vma_is_anon_shmem(vma)) {
> +		/*
> +		 * For tmpfs, we now only support PMD sized THP if huge page
> +		 * is enabled, otherwise fallback to order 0.
> +		 */
> +		return global_huge ? BIT(HPAGE_PMD_ORDER) : 0;
> +	}
> +
>  	/*
>  	 * Following the 'deny' semantics of the top level, force the huge
>  	 * option off from all mounts.
> @@ -2081,7 +2100,7 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
>  	struct mm_struct *fault_mm;
>  	struct folio *folio;
>  	int error;
> -	bool alloced, huge;
> +	bool alloced;
>  	unsigned long orders = 0;
>  
>  	if (WARN_ON_ONCE(!shmem_mapping(inode->i_mapping)))
> @@ -2154,14 +2173,8 @@ static int shmem_get_folio_gfp(struct inode *inode, pgoff_t index,
>  		return 0;
>  	}
>  
> -	huge = shmem_huge_global_enabled(inode, index, false, fault_mm,
> -			     vma ? vma->vm_flags : 0);
> -	/* Find hugepage orders that are allowed for anonymous shmem. */
> -	if (vma && vma_is_anon_shmem(vma))
> -		orders = shmem_allowable_huge_orders(inode, vma, index, huge);
> -	else if (huge)
> -		orders = BIT(HPAGE_PMD_ORDER);
> -
> +	/* Find hugepage orders that are allowed for anonymous shmem and tmpfs. */
> +	orders = shmem_allowable_huge_orders(inode, vma, index, false);
>  	if (orders > 0) {
>  		gfp_t huge_gfp;
>  

Re: [PATCH v2 3/3] mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

[Baolin Wang]

(Sorry for the late reply due to my vacation.)

On 2024/7/15 21:36, Ryan Roberts wrote:
> On 13/07/2024 14:24, Baolin Wang wrote:
>> Move shmem_huge_global_enabled() into the shmem_allowable_huge_orders() function,
>> so that shmem_allowable_huge_orders() can also help to find the allowable huge
>> orders for tmpfs. Moreover the shmem_huge_global_enabled() can become static.
>>
>> No functional changes.
>>
>> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
> 
> one nit below, but either way:
> 
> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> 
>> ---
>>   include/linux/shmem_fs.h | 12 ++----------
>>   mm/huge_memory.c         | 12 +++---------
>>   mm/shmem.c               | 41 ++++++++++++++++++++++++++--------------
>>   3 files changed, 32 insertions(+), 33 deletions(-)
>>
>> diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
>> index 405ee8d3589a..1564d7d3ca61 100644
>> --- a/include/linux/shmem_fs.h
>> +++ b/include/linux/shmem_fs.h
>> @@ -111,21 +111,13 @@ extern void shmem_truncate_range(struct inode *inode, loff_t start, loff_t end);
>>   int shmem_unuse(unsigned int type);
>>   
>>   #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>> -extern bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index, bool shmem_huge_force,
>> -				      struct mm_struct *mm, unsigned long vm_flags);
>>   unsigned long shmem_allowable_huge_orders(struct inode *inode,
>>   				struct vm_area_struct *vma, pgoff_t index,
>> -				bool global_huge);
>> +				bool shmem_huge_force);
>>   #else
>> -static __always_inline bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>> -						      bool shmem_huge_force, struct mm_struct *mm,
>> -						      unsigned long vm_flags)
>> -{
>> -	return false;
>> -}
>>   static inline unsigned long shmem_allowable_huge_orders(struct inode *inode,
>>   				struct vm_area_struct *vma, pgoff_t index,
>> -				bool global_huge)
>> +				bool shmem_huge_force)
>>   {
>>   	return 0;
>>   }
>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>> index cc9bad12be75..f69980b5b5fc 100644
>> --- a/mm/huge_memory.c
>> +++ b/mm/huge_memory.c
>> @@ -151,16 +151,10 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
>>   	 * Must be done before hugepage flags check since shmem has its
>>   	 * own flags.
>>   	 */
>> -	if (!in_pf && shmem_file(vma->vm_file)) {
>> -		bool global_huge = shmem_huge_global_enabled(file_inode(vma->vm_file),
>> -							     vma->vm_pgoff, !enforce_sysfs,
>> -							     vma->vm_mm, vm_flags);
>> -
>> -		if (!vma_is_anon_shmem(vma))
>> -			return global_huge ? orders : 0;
>> +	if (!in_pf && shmem_file(vma->vm_file))
>>   		return shmem_allowable_huge_orders(file_inode(vma->vm_file),
>> -							vma, vma->vm_pgoff, global_huge);
>> -	}
>> +						   vma, vma->vm_pgoff,
>> +						   !enforce_sysfs);
>>   
>>   	if (!vma_is_anonymous(vma)) {
>>   		/*
>> diff --git a/mm/shmem.c b/mm/shmem.c
>> index 1445dcd39b6f..4d274f5a17d9 100644
>> --- a/mm/shmem.c
>> +++ b/mm/shmem.c
>> @@ -581,7 +581,7 @@ static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>>   	}
>>   }
>>   
>> -bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>> +static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>>   		   bool shmem_huge_force, struct mm_struct *mm,
>>   		   unsigned long vm_flags)
>>   {
>> @@ -772,6 +772,13 @@ static unsigned long shmem_unused_huge_shrink(struct shmem_sb_info *sbinfo,
>>   {
>>   	return 0;
>>   }
>> +
>> +static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
>> +		bool shmem_huge_force, struct mm_struct *mm,
>> +		unsigned long vm_flags)
>> +{
>> +	return false;
>> +}
>>   #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
>>   
>>   /*
>> @@ -1625,27 +1632,39 @@ static gfp_t limit_gfp_mask(gfp_t huge_gfp, gfp_t limit_gfp)
>>   #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>   unsigned long shmem_allowable_huge_orders(struct inode *inode,
>>   				struct vm_area_struct *vma, pgoff_t index,
>> -				bool global_huge)
>> +				bool shmem_huge_force)
>>   {
>>   	unsigned long mask = READ_ONCE(huge_shmem_orders_always);
>>   	unsigned long within_size_orders = READ_ONCE(huge_shmem_orders_within_size);
>> -	unsigned long vm_flags = vma->vm_flags;
>> +	unsigned long vm_flags = vma ? vma->vm_flags : 0;
>> +	struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;
> 
> nit: rather than deriving the fault_mm here, I wonder if its cleaner to just
> pass vma to shmem_huge_global_enabled()? shmem_huge_global_enabled() is just
> using it as a guard to access vm_flags, which you can just as easily do by
> testing the vma for non-NULL. And you can access mm flags with vma->vm_mm->flags
> after testing the vma too.

Make sense to me, and will do in next version.

Thanks for reviewing.

Re: [PATCH v2 0/3] Some cleanups for shmem

[Andrew Morton]

On Sat, 13 Jul 2024 21:24:19 +0800 Baolin Wang <baolin.wang@linux.alibaba.com> wrote:

> Changes from v1:
>  - Add a dummy function in case CONFIG_TRANSPARENT_HUGEPAGE is not
>  enabled, which fixes a building error reported by kernel test robot.

The only difference I'm seeing from the v1 series is the below update
to [3/3]:

--- a/mm/shmem.c~mm-shmem-move-shmem_huge_global_enabled-into-shmem_allowable_huge_orders-v2
+++ a/mm/shmem.c
@@ -549,10 +549,9 @@ static bool shmem_confirm_swap(struct ad
 static int shmem_huge __read_mostly = SHMEM_HUGE_NEVER;
 
 static bool __shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
-					bool shmem_huge_force, struct vm_area_struct *vma,
+					bool shmem_huge_force, struct mm_struct *mm,
 					unsigned long vm_flags)
 {
-	struct mm_struct *mm = vma ? vma->vm_mm : NULL;
 	loff_t i_size;
 
 	if (!S_ISREG(inode->i_mode))
@@ -583,14 +582,14 @@ static bool __shmem_huge_global_enabled(
 }
 
 static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
-		   bool shmem_huge_force, struct vm_area_struct *vma,
+		   bool shmem_huge_force, struct mm_struct *mm,
 		   unsigned long vm_flags)
 {
 	if (HPAGE_PMD_ORDER > MAX_PAGECACHE_ORDER)
 		return false;
 
 	return __shmem_huge_global_enabled(inode, index, shmem_huge_force,
-					   vma, vm_flags);
+					   mm, vm_flags);
 }
 
 #if defined(CONFIG_SYSFS)
@@ -775,7 +774,7 @@ static unsigned long shmem_unused_huge_s
 }
 
 static bool shmem_huge_global_enabled(struct inode *inode, pgoff_t index,
-		bool shmem_huge_force, struct vm_area_struct *vma,
+		bool shmem_huge_force, struct mm_struct *mm,
 		unsigned long vm_flags)
 {
 	return false;
@@ -1638,6 +1637,7 @@ unsigned long shmem_allowable_huge_order
 	unsigned long mask = READ_ONCE(huge_shmem_orders_always);
 	unsigned long within_size_orders = READ_ONCE(huge_shmem_orders_within_size);
 	unsigned long vm_flags = vma ? vma->vm_flags : 0;
+	struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;
 	/*
 	 * Check all the (large) orders below HPAGE_PMD_ORDER + 1 that
 	 * are enabled for this vma.
@@ -1656,7 +1656,7 @@ unsigned long shmem_allowable_huge_order
 		return 0;
 
 	global_huge = shmem_huge_global_enabled(inode, index, shmem_huge_force,
-						vma, vm_flags);
+						fault_mm, vm_flags);
 	if (!vma || !vma_is_anon_shmem(vma)) {
 		/*
 		 * For tmpfs, we now only support PMD sized THP if huge page
_

Re: [PATCH v2 0/3] Some cleanups for shmem

[Andrew Morton]

On Wed, 24 Jul 2024 12:14:07 -0700 Andrew Morton <akpm@linux-foundation.org> wrote:

> The only difference I'm seeing from the v1 series is the below update
> to [3/3]:

oop. sorry, never mind.

Re: [PATCH v2 1/3] mm: shmem: simplify the suitable huge orders validation for tmpfs

[David Hildenbrand]

On 13.07.24 15:24, Baolin Wang wrote:
> Move the suitable huge orders validation into shmem_suitable_orders() for
> tmpfs, which can reuse some code to simplify the logic.
> 
> In addition, we don't have special handling for the error code -E2BIG when
> checking for conflicts with PMD sized THP in the pagecache for tmpfs, instead,
> it will just fallback to order-0 allocations like this patch does, so this
> simplification will not add functional changes.
> 
> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
> ---

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 2/3] mm: shmem: rename shmem_is_huge() to shmem_huge_global_enabled()

[David Hildenbrand]

On 13.07.24 15:24, Baolin Wang wrote:
> The shmem_is_huge() is now used to check if the top-level huge page is enabled,
> thus rename it to reflect its usage.
> 
> Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
> ---

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 3/3] mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

[David Hildenbrand]

>>>    /*
>>> @@ -1625,27 +1632,39 @@ static gfp_t limit_gfp_mask(gfp_t huge_gfp, gfp_t limit_gfp)
>>>    #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>>    unsigned long shmem_allowable_huge_orders(struct inode *inode,
>>>    				struct vm_area_struct *vma, pgoff_t index,
>>> -				bool global_huge)
>>> +				bool shmem_huge_force)
>>>    {
>>>    	unsigned long mask = READ_ONCE(huge_shmem_orders_always);
>>>    	unsigned long within_size_orders = READ_ONCE(huge_shmem_orders_within_size);
>>> -	unsigned long vm_flags = vma->vm_flags;
>>> +	unsigned long vm_flags = vma ? vma->vm_flags : 0;
>>> +	struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;
>>
>> nit: rather than deriving the fault_mm here, I wonder if its cleaner to just
>> pass vma to shmem_huge_global_enabled()? shmem_huge_global_enabled() is just
>> using it as a guard to access vm_flags, which you can just as easily do by
>> testing the vma for non-NULL. And you can access mm flags with vma->vm_mm->flags
>> after testing the vma too.
> 
> Make sense to me, and will do in next version.

Feel free to add my

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH v2 3/3] mm: shmem: move shmem_huge_global_enabled() into shmem_allowable_huge_orders()

[Baolin Wang]

On 2024/7/25 21:09, David Hildenbrand wrote:
>>>>    /*
>>>> @@ -1625,27 +1632,39 @@ static gfp_t limit_gfp_mask(gfp_t huge_gfp, 
>>>> gfp_t limit_gfp)
>>>>    #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>>>    unsigned long shmem_allowable_huge_orders(struct inode *inode,
>>>>                    struct vm_area_struct *vma, pgoff_t index,
>>>> -                bool global_huge)
>>>> +                bool shmem_huge_force)
>>>>    {
>>>>        unsigned long mask = READ_ONCE(huge_shmem_orders_always);
>>>>        unsigned long within_size_orders = 
>>>> READ_ONCE(huge_shmem_orders_within_size);
>>>> -    unsigned long vm_flags = vma->vm_flags;
>>>> +    unsigned long vm_flags = vma ? vma->vm_flags : 0;
>>>> +    struct mm_struct *fault_mm = vma ? vma->vm_mm : NULL;
>>>
>>> nit: rather than deriving the fault_mm here, I wonder if its cleaner 
>>> to just
>>> pass vma to shmem_huge_global_enabled()? shmem_huge_global_enabled() 
>>> is just
>>> using it as a guard to access vm_flags, which you can just as easily 
>>> do by
>>> testing the vma for non-NULL. And you can access mm flags with 
>>> vma->vm_mm->flags
>>> after testing the vma too.
>>
>> Make sense to me, and will do in next version.
> 
> Feel free to add my
> 
> Acked-by: David Hildenbrand <david@redhat.com>

Thanks David.

Andrew has already queued my v3 patchset into the mm-unstable branch.

Andrew, please help to add David's acked tag for v3 series. Thanks.