Skip to contentSkip to conversation

Minimize xa_node allocation during xarry split

PATCH Superseded Zi Yan 1 participant 4 messages Feb 13, 2025

Subsystems (6)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Zi Yan

author
#0 lore raw

cover Minimize xa_node allocation during xarry split

Hi all,

When splitting a multi-index entry in XArray from order-n to order-m, existing xas_split_alloc()+xas_split() approach requires
2^(n % XA_CHUNK_SHIFT) xa_node allocations. But its callers, __filemap_add_folio() and shmem_split_large_entry(), use at most 1 xa_node. To minimize xa_node allocation and remove the limitation of no split from order-12 (or above) to order-0 (or anything between 0 and 5)[1], xas_try_split() was added[2], which allocates
(n / XA_CHUNK_SHIFT - m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but can be used by __filemap_add_folio() and shmem_split_large_entry().

xas_split_alloc() and xas_split() split an order-9 to order-0:

         ---------------------------------
         |   |   |   |   |   |   |   |   |
         | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
         |   |   |   |   |   |   |   |   |
         ---------------------------------
           |   |                   |   |
     -------   ---               ---   -------
     |           |     ...       |           |
     V           V               V           V
----------- -----------     ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- -----------     ----------- -----------

xas_try_split() splits an order-9 to order-0:
---------------------------------

   |   |   |   |   |   |   |   |   |
   | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
   |   |   |   |   |   |   |   |   |
   ---------------------------------
     |
     |
     V

-----------
| xa_node |
-----------

xas_try_split() is designed to be called iteratively with n = m + 1. xas_try_split_mini_order() is added to minmize the number of calls to xas_try_split() by telling the caller the next minimal order to split to instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT does not require xa_node allocation and requires 1 xa_node when n=l * XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n > m + 1 when no new xa_node is needed.

xfstests quick group test passed on xfs and tmpfs.

Let me know your comments.

[1] https://lore.kernel.org/linux-mm/[email protected]/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Zi Yan (2):
mm/filemap: use xas_try_split() in __filemap_add_folio(). mm/shmem: use xas_try_split() in shmem_split_large_entry().

 include/linux/xarray.h |  7 +++++++
 lib/xarray.c           | 25 +++++++++++++++++++++++
 mm/filemap.c           | 46 +++++++++++++++++-------------------------
 mm/shmem.c             | 43 +++++++++++++++------------------------
 4 files changed, 67 insertions(+), 54 deletions(-)
Signature · 1 line
2.47.2
author
#1 lore raw

PATCH 1/2 mm/filemap: use xas_try_split() in __filemap_add_folio().

During __filemap_add_folio(), a shadow entry is covering n slots and a folio covers m slots with m < n is to be added. Instead of splitting all n slots, only the m slots covered by the folio need to be split and the remaining n-m shadow entries can be retained with orders ranging from m to n-1. This method only requires (n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT) new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)) new xa_nodes, compared to the original xas_split_alloc() + xas_split() one. For example, to insert an order-0 folio when an order-9 shadow entry is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of 8.

xas_try_split_min_order() is introduced to reduce the number of calls to xas_try_split() during split.

---
 include/linux/xarray.h |  7 +++++++
 lib/xarray.c           | 25 +++++++++++++++++++++++
 mm/filemap.c           | 46 +++++++++++++++++-------------------------
 3 files changed, 51 insertions(+), 27 deletions(-)
include/linux/xarray.h +7 -0
index 9eb8c7425090..6ef3d682b189 100644
@@ -1557,6 +1557,7 @@ void xas_split(struct xa_state *, void *entry, unsigned int order);
 void xas_split_alloc(struct xa_state *, void *entry, unsigned int order, gfp_t);
 void xas_try_split(struct xa_state *xas, void *entry, unsigned int order,
 		gfp_t gfp);
+unsigned int xas_try_split_min_order(unsigned int order);
 #else
 static inline int xa_get_order(struct xarray *xa, unsigned long index)
 {
@@ -1583,6 +1584,12 @@ static inline void xas_try_split(struct xa_state *xas, void *entry,
 		unsigned int order, gfp_t gfp)
 {
 }
+
+static inline unsigned int xas_try_split_min_order(unsigned int order)
+{
+	return 0;
+}
+
 #endif
 
 /**
lib/xarray.c +25 -0
index c38beca77830..1805fde1c361 100644
@@ -1133,6 +1133,28 @@ void xas_split(struct xa_state *xas, void *entry, unsigned int order)
 }
 EXPORT_SYMBOL_GPL(xas_split);
 
+/**
+ * xas_try_split_min_order() - Minimal split order xas_try_split() can accept
+ * @order: Current entry order.
+ *
+ * xas_try_split() can split a multi-index entry to smaller than @order - 1 if
+ * no new xa_node is needed. This function provides the minimal order
+ * xas_try_split() supports.
+ *
+ * Return: the minimal order xas_try_split() supports
+ *
+ * Context: Any context.
+ *
+ */
+unsigned int xas_try_split_min_order(unsigned int order)
+{
+	if (order % XA_CHUNK_SHIFT == 0)
+		return order == 0 ? 0 : order - 1;
+
+	return order - (order % XA_CHUNK_SHIFT);
+}
+EXPORT_SYMBOL_GPL(xas_try_split_min_order);
+
 /**
  * xas_try_split() - Try to split a multi-index entry.
  * @xas: XArray operation state.
@@ -1145,6 +1167,9 @@ EXPORT_SYMBOL_GPL(xas_split);
  * be allocated, the function will use @gfp to get one. If more xa_node are
  * needed, the function gives EINVAL error.
  *
+ * NOTE: use xas_try_split_min_order() to get next split order instead of
+ * @order - 1 if you want to minmize xas_try_split() calls.
+ *
  * Context: Any context.  The caller should hold the xa_lock.
  */
 void xas_try_split(struct xa_state *xas, void *entry, unsigned int order,
mm/filemap.c +19 -27

82 lines not on this page · raw message

Signature · 1 line
2.47.2
author
#2 lore raw

PATCH 2/2 mm/shmem: use xas_try_split() in shmem_split_large_entry().

In reply to Zi Yan

During shmem_split_large_entry(), large swap entries are covering n slots and an order-0 folio needs to be inserted. Instead of splitting all n slots, only the 1 slot covered by the folio need to be split and the remaining n-1 shadow entries can be retained with orders ranging from 0 to n-1. This method only requires (n/XA_CHUNK_SHIFT) new xa_nodes instead of (n % XA_CHUNK_SHIFT) * (n/XA_CHUNK_SHIFT) new xa_nodes, compared to the original xas_split_alloc() + xas_split() one.
For example, to split an order-9 large swap entry (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of 8.

xas_try_split_min_order() is used to reduce the number of calls to xas_try_split() during split.

---
 mm/shmem.c | 43 ++++++++++++++++---------------------------
 1 file changed, 16 insertions(+), 27 deletions(-)
mm/shmem.c +16 -27
index 671f63063fd4..b35ba250c53d 100644
@@ -2162,14 +2162,14 @@ static int shmem_split_large_entry(struct inode *inode, pgoff_t index,
 {
 	struct address_space *mapping = inode->i_mapping;
 	XA_STATE_ORDER(xas, &mapping->i_pages, index, 0);
-	void *alloced_shadow = NULL;
-	int alloced_order = 0, i;
+	int split_order = 0;
+	int i;
 
 	/* Convert user data gfp flags to xarray node gfp flags */
 	gfp &= GFP_RECLAIM_MASK;
 
 	for (;;) {
-		int order = -1, split_order = 0;
+		int order = -1;
 		void *old = NULL;
 
 		xas_lock_irq(&xas);
@@ -2181,20 +2181,21 @@ static int shmem_split_large_entry(struct inode *inode, pgoff_t index,
 
 		order = xas_get_order(&xas);
 
-		/* Swap entry may have changed before we re-acquire the lock */
-		if (alloced_order &&
-		    (old != alloced_shadow || order != alloced_order)) {
-			xas_destroy(&xas);
-			alloced_order = 0;
-		}
-
 		/* Try to split large swap entry in pagecache */
 		if (order > 0) {
-			if (!alloced_order) {
-				split_order = order;
-				goto unlock;
+			int cur_order = order;
+
+			split_order = xas_try_split_min_order(cur_order);
+
+			while (cur_order > 0) {
+				xas_set_order(&xas, index, split_order);
+				xas_try_split(&xas, old, cur_order, GFP_NOWAIT);
+				if (xas_error(&xas))
+					goto unlock;
+				cur_order = split_order;
+				split_order =
+					xas_try_split_min_order(split_order);
 			}
-			xas_split(&xas, old, order);
 
 			/*
 			 * Re-set the swap entry after splitting, and the swap
@@ -2213,26 +2214,14 @@ static int shmem_split_large_entry(struct inode *inode, pgoff_t index,
 unlock:
 		xas_unlock_irq(&xas);
 
-		/* split needed, alloc here and retry. */
-		if (split_order) {
-			xas_split_alloc(&xas, old, split_order, gfp);
-			if (xas_error(&xas))
-				goto error;
-			alloced_shadow = old;
-			alloced_order = split_order;
-			xas_reset(&xas);
-			continue;
-		}
-
 		if (!xas_nomem(&xas, gfp))
 			break;
 	}
 
-error:
 	if (xas_error(&xas))
 		return xas_error(&xas);
 
-	return alloced_order;
+	return split_order;
 }
 
 /*
Signature · 1 line
2.47.2
author
#3 lore raw
In reply to Zi Yan
Quoted 49 lines · Zi Yan

Hi all,

When splitting a multi-index entry in XArray from order-n to order-m, existing xas_split_alloc()+xas_split() approach requires
2^(n % XA_CHUNK_SHIFT) xa_node allocations. But its callers, __filemap_add_folio() and shmem_split_large_entry(), use at most 1 xa_node. To minimize xa_node allocation and remove the limitation of no split from order-12 (or above) to order-0 (or anything between 0 and 5)[1], xas_try_split() was added[2], which allocates
(n / XA_CHUNK_SHIFT - m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but can be used by __filemap_add_folio() and shmem_split_large_entry().

xas_split_alloc() and xas_split() split an order-9 to order-0:

         ---------------------------------
         |   |   |   |   |   |   |   |   |
         | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
         |   |   |   |   |   |   |   |   |
         ---------------------------------
           |   |                   |   |
     -------   ---               ---   -------
     |           |     ...       |           |
     V           V               V           V
----------- -----------     ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- -----------     ----------- -----------

xas_try_split() splits an order-9 to order-0:

   ---------------------------------
   |   |   |   |   |   |   |   |   |
   | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
   |   |   |   |   |   |   |   |   |
   ---------------------------------
     |
     |
     V

-----------
| xa_node |
-----------

xas_try_split() is designed to be called iteratively with n = m + 1. xas_try_split_mini_order() is added to minmize the number of calls to xas_try_split() by telling the caller the next minimal order to split to instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT does not require xa_node allocation and requires 1 xa_node when n=l * XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n > m + 1 when no new xa_node is needed.

Quoted 3 lines

xfstests quick group test passed on xfs and tmpfs.

Let me know your comments.

The patch is on top of mm-everything-2025-02-13-02-18.

Quoted 9 lines

[1] https://lore.kernel.org/linux-mm/[email protected]/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Zi Yan (2):
mm/filemap: use xas_try_split() in __filemap_add_folio(). mm/shmem: use xas_try_split() in shmem_split_large_entry().

Quoted 8 lines
 include/linux/xarray.h |  7 +++++++
 lib/xarray.c           | 25 +++++++++++++++++++++++
 mm/filemap.c           | 46 +++++++++++++++++-------------------------
 mm/shmem.c             | 43 +++++++++++++++------------------------
 4 files changed, 67 insertions(+), 54 deletions(-)
Signature · 1 line
2.47.2

Best Regards,
Yan, Zi