From: Daniel Gomez <da.gomez@samsung.com>

Add large folio support for shmem write and fallocate paths matching the
same high order preference mechanism used in the iomap buffered IO path
as used in __filemap_get_folio().

Add shmem_mapping_size_order() to get a hint for the order of the folio
based on the file size which takes care of the mapping requirements.

If the top level huge page (controlled by '/sys/kernel/mm/transparent_hugepage/shmem_enabled')
is enabled, we just allow PMD sized THP to keep interface backward
compatibility.

Co-developed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Signed-off-by: Daniel Gomez <da.gomez@samsung.com>
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 mm/shmem.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 48 insertions(+), 3 deletions(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index 0613421e09e7..6dece90ff421 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1672,6 +1672,36 @@ bool shmem_hpage_pmd_enabled(void)
 	return false;
 }
 
+/**
+ * shmem_mapping_size_order - Get maximum folio order for the given file size.
+ * @mapping: Target address_space.
+ * @index: The page index.
+ * @size: The suggested size of the folio to create.
+ *
+ * This returns a high order for folios (when supported) based on the file size
+ * which the mapping currently allows at the given index. The index is relevant
+ * due to alignment considerations the mapping might have. The returned order
+ * may be less than the size passed.
+ *
+ * Like __filemap_get_folio order calculation.
+ *
+ * Return: The order.
+ */
+static inline unsigned int
+shmem_mapping_size_order(struct address_space *mapping, pgoff_t index, size_t size)
+{
+	unsigned int order = get_order(max_t(size_t, size, PAGE_SIZE));
+
+	if (!mapping_large_folio_support(mapping))
+		return 0;
+
+	/* If we're not aligned, allocate a smaller folio */
+	if (index & ((1UL << order) - 1))
+		order = __ffs(index);
+
+	return min_t(size_t, order, MAX_PAGECACHE_ORDER);
+}
+
 unsigned long shmem_allowable_huge_orders(struct inode *inode,
 				struct vm_area_struct *vma, pgoff_t index,
 				loff_t write_end, bool shmem_huge_force)
@@ -1694,11 +1724,26 @@ unsigned long shmem_allowable_huge_orders(struct inode *inode,
 	global_huge = shmem_huge_global_enabled(inode, index, write_end,
 					shmem_huge_force, vma, vm_flags);
 	if (!vma || !vma_is_anon_shmem(vma)) {
+		size_t len;
+
+		/*
+		 * For tmpfs, if top level huge page is enabled, we just allow
+		 * PMD sized THP to keep interface backward compatibility.
+		 */
+		if (global_huge)
+			return BIT(HPAGE_PMD_ORDER);
+
+		if (!write_end)
+			return 0;
+
 		/*
-		 * For tmpfs, we now only support PMD sized THP if huge page
-		 * is enabled, otherwise fallback to order 0.
+		 * Otherwise, get a highest order hint based on the size of
+		 * write and fallocate paths, then will try each allowable
+		 * huge orders.
 		 */
-		return global_huge ? BIT(HPAGE_PMD_ORDER) : 0;
+		len = write_end - (index << PAGE_SHIFT);
+		order = shmem_mapping_size_order(inode->i_mapping, index, len);
+		return order > 0 ? BIT(order + 1) - 1 : 0;
 	}
 
 	/*
-- 
2.39.3

On Thu, Sep 26, 2024 at 04:27:26PM +0800, Baolin Wang wrote:
> +static inline unsigned int
> +shmem_mapping_size_order(struct address_space *mapping, pgoff_t index, size_t size)
> +{
> +	unsigned int order = get_order(max_t(size_t, size, PAGE_SIZE));

Why introduce the max_t() call here?  Did nobody read the documentation
or implementation for get_order() before writing this patch?

Besides, get_order() is wrong (at least relative to other filesystems).
get_order() rounds up instead of down, so what should we do for a write()
of size 512 * 1024 + 1 byte?  Other filesystems allocate an order-8 folio
plus an order-0 folio.  This code would have us allocate an order-9 folio.
I think that's a bad idea.

On 9/26/2024 2:16 PM, Matthew Wilcox wrote:
> On Thu, Sep 26, 2024 at 04:27:26PM +0800, Baolin Wang wrote:
>> +static inline unsigned int
>> +shmem_mapping_size_order(struct address_space *mapping, pgoff_t index, size_t size)
>> +{
>> +	unsigned int order = get_order(max_t(size_t, size, PAGE_SIZE));
> 
> Why introduce the max_t() call here?  Did nobody read the documentation
> or implementation for get_order() before writing this patch?

get_order() result is undefined if the size is 0. I've used max_t() here 
to avoid that case. Perhaps should we prevent that case before getting here?

> 
> Besides, get_order() is wrong (at least relative to other filesystems).
> get_order() rounds up instead of down, so what should we do for a write()
> of size 512 * 1024 + 1 byte?  Other filesystems allocate an order-8 folio
> plus an order-0 folio.  This code would have us allocate an order-9 folio.
> I think that's a bad idea.
> 

I think one of my earlier attemps was to use fgf_set_order + 
FGF_GET_ORDER() as in iomap. But the solution taken there was to share 
code between shmem and filemap and that wasn't considered a good idea. 
Shall we just replicate iomap_get_folio()? Or else, what do you suggest 
here?

Daniel

On Thu, Sep 26, 2024 at 02:58:31PM +0200, Daniel Gomez wrote:
> On 9/26/2024 2:16 PM, Matthew Wilcox wrote:
> > On Thu, Sep 26, 2024 at 04:27:26PM +0800, Baolin Wang wrote:
> > > +static inline unsigned int
> > > +shmem_mapping_size_order(struct address_space *mapping, pgoff_t index, size_t size)
> > > +{
> > > +	unsigned int order = get_order(max_t(size_t, size, PAGE_SIZE));
> > 
> > Why introduce the max_t() call here?  Did nobody read the documentation
> > or implementation for get_order() before writing this patch?
> 
> get_order() result is undefined if the size is 0. I've used max_t() here to
> avoid that case. Perhaps should we prevent that case before getting here?

Surely we've handled a length-0 write before we get here?

> I think one of my earlier attemps was to use fgf_set_order + FGF_GET_ORDER()
> as in iomap. But the solution taken there was to share code between shmem
> and filemap and that wasn't considered a good idea. Shall we just replicate
> iomap_get_folio()? Or else, what do you suggest here?

We could move three of the four lines from fgf_set_order() into a
new function and call it from both fgf_set_order() and shmem?
>

On 2024/9/26 21:40, Matthew Wilcox wrote:
> On Thu, Sep 26, 2024 at 02:58:31PM +0200, Daniel Gomez wrote:
>> On 9/26/2024 2:16 PM, Matthew Wilcox wrote:
>>> On Thu, Sep 26, 2024 at 04:27:26PM +0800, Baolin Wang wrote:
>>>> +static inline unsigned int
>>>> +shmem_mapping_size_order(struct address_space *mapping, pgoff_t index, size_t size)
>>>> +{
>>>> +	unsigned int order = get_order(max_t(size_t, size, PAGE_SIZE));
>>>
>>> Why introduce the max_t() call here?  Did nobody read the documentation
>>> or implementation for get_order() before writing this patch?
>>
>> get_order() result is undefined if the size is 0. I've used max_t() here to
>> avoid that case. Perhaps should we prevent that case before getting here?
> 
> Surely we've handled a length-0 write before we get here?
> 
>> I think one of my earlier attemps was to use fgf_set_order + FGF_GET_ORDER()
>> as in iomap. But the solution taken there was to share code between shmem
>> and filemap and that wasn't considered a good idea. Shall we just replicate
>> iomap_get_folio()? Or else, what do you suggest here?
> 
> We could move three of the four lines from fgf_set_order() into a
> new function and call it from both fgf_set_order() and shmem?

Sounds good. How about the following changes? Do you have a perferred 
name for the new helper? Thanks.

diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index d9c7edb6422b..ce418acd2737 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -629,6 +629,16 @@ typedef unsigned int __bitwise fgf_t;

  #define FGP_WRITEBEGIN         (FGP_LOCK | FGP_WRITE | FGP_CREAT | 
FGP_STABLE)

+static inline unsigned int filemap_get_order(size_t size)
+{
+       unsigned int shift = ilog2(size);
+
+       if (shift <= PAGE_SHIFT)
+               return 0;
+
+       return shift - PAGE_SHIFT;
+}
+
  /**
   * fgf_set_order - Encode a length in the fgf_t flags.
   * @size: The suggested size of the folio to create.
@@ -642,11 +652,11 @@ typedef unsigned int __bitwise fgf_t;
   */
  static inline fgf_t fgf_set_order(size_t size)
  {
-       unsigned int shift = ilog2(size);
+       unsigned int order = filemap_get_order(size);

-       if (shift <= PAGE_SHIFT)
+       if (!order)
                 return 0;
-       return (__force fgf_t)((shift - PAGE_SHIFT) << 26);
+       return (__force fgf_t)(order << 26);
  }

  void *filemap_get_entry(struct address_space *mapping, pgoff_t index);