[RFC PATCH v2 0/2] Fix storing in XArray check_split tests

[RFC PATCH v2 0/2] Fix storing in XArray check_split tests

[Ackerley Tng]

Hi,

I hit an assertion while making some modifications to
lib/test_xarray.c [1] and I believe this is the fix.

In check_split, the tests split the XArray node and then store values
after the split to verify that splitting worked. While storing and
retrieval works as expected, the node's metadata, specifically
node->nr_values, is not updated correctly.

This led to the assertion being hit in [1], since the storing process
did not increment node->nr_values sufficiently, while the erasing
process assumed the fully-incremented node->nr_values state.

Would like to check my understanding on these:

1. In the multi-index xarray world, is node->nr_values definitely the
   total number of values *and siblings* in the node?

2. IIUC xas_store() has significantly different behavior when entry is
   NULL vs non-NULL: when entry is NULL, xas_store() does not make
   assumptions on the number of siblings and erases all the way till
   the next non-sibling entry. This sounds fair to me, but it's also
   kind of surprising that it is differently handled when entry is
   non-NULL, where xas_store() respects xas->xa_sibs.

3. If xas_store() is dependent on its caller to set up xas correctly
   (also sounds fair), then there are places where xas_store() is
   used, like replace_page_cache_folio() or
   migrate_huge_page_move_mapping(), where xas is set up assuming 0
   order pages. Are those buggy?

Changes in RFC v2:

+ Rebased on Linux 7.0-rc1

[1] https://lore.kernel.org/all/20251028223414.299268-1-ackerleytng@google.com/
[2] RFC v1: https://lore.kernel.org/all/720e32d8e185d5c82659bbdede05e87b3318c413.1769818406.git.ackerleytng@google.com/

Ackerley Tng (2):
  XArray tests: Fix check_split tests to store correctly
  XArray tests: Verify xa_erase behavior in check_split

 lib/test_xarray.c | 28 ++++++++++++++++++++++++----
 1 file changed, 24 insertions(+), 4 deletions(-)


base-commit: 6de23f81a5e08be8fbf5e8d7e9febc72a5b5f27f
--
2.53.0.345.g96ddfc5eaa-goog

[RFC PATCH v1 RESEND 1/2] XArray tests: Fix check_split tests to store correctly

[Ackerley Tng]

__xa_store() does not set up xas->xa_sibs, and when it calls xas_store(),
xas_store() stops prematurely and does not update node->nr_values to count
all values and siblings. Hence, when working with multi-index XArrays,
__xa_store() cannot be used.

Fix tests by calling xas_store() directly with xas->xa_sibs correctly set
up.

Change-Id: I144b96b8724c27ac4a6d53be1e74125bf8d3aebe
Signed-off-by: Ackerley Tng <ackerleytng@google.com>
---
 lib/test_xarray.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/lib/test_xarray.c b/lib/test_xarray.c
index 5ca0aefee9aa5..e71e8ff76900b 100644
--- a/lib/test_xarray.c
+++ b/lib/test_xarray.c
@@ -1846,8 +1846,14 @@ static void check_split_1(struct xarray *xa, unsigned long index,
 	xas_split_alloc(&xas, xa, order, GFP_KERNEL);
 	xas_lock(&xas);
 	xas_split(&xas, xa, order);
-	for (i = 0; i < (1 << order); i += (1 << new_order))
-		__xa_store(xa, index + i, xa_mk_index(index + i), 0);
+	for (i = 0; i < (1 << order); i += (1 << new_order)) {
+		xas_set_order(&xas, index + i, new_order);
+		/*
+		 * Don't worry about -ENOMEM, xas_split_alloc() and
+		 * xas_split() ensures that all nodes are allocated.
+		 */
+		xas_store(&xas, xa_mk_index(index + i));
+	}
 	xas_unlock(&xas);
 
 	for (i = 0; i < (1 << order); i++) {
@@ -1893,8 +1899,14 @@ static void check_split_2(struct xarray *xa, unsigned long index,
 		xas_unlock(&xas);
 		goto out;
 	}
-	for (i = 0; i < (1 << order); i += (1 << new_order))
-		__xa_store(xa, index + i, xa_mk_index(index + i), 0);
+	for (i = 0; i < (1 << order); i += (1 << new_order)) {
+		xas_set_order(&xas, index + i, new_order);
+		/*
+		 * Don't worry about -ENOMEM, xas_split_alloc() and
+		 * xas_split() ensures that all nodes are allocated.
+		 */
+		xas_store(&xas, xa_mk_index(index + i));
+	}
 	xas_unlock(&xas);
 
 	for (i = 0; i < (1 << order); i++) {
-- 
2.53.0.345.g96ddfc5eaa-goog

[RFC PATCH v1 RESEND 2/2] XArray tests: Verify xa_erase behavior in check_split

[Ackerley Tng]

Both __xa_store() and xa_erase() use xas_store() under the hood, but when
the entry being stored is NULL (as in the case of xa_erase()),
xas->xa_sibs (and max) is only checked if the next entry is not a sibling,
hence allowing xas_store() to keep iterating, hence updating
node->nr_values correctly.

Add xa_erase() to check_split tests that verify functionality, with the
added intent to illustrate the usage differences between __xa_store(),
xas_store() and xa_erase() with regard to multi-index XArrays.

Change-Id: Ibbbae5e6339b46203f22658cc84ea9f16025fa14
Signed-off-by: Ackerley Tng <ackerleytng@google.com>
---
 lib/test_xarray.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/lib/test_xarray.c b/lib/test_xarray.c
index e71e8ff76900b..bb9471a3df65c 100644
--- a/lib/test_xarray.c
+++ b/lib/test_xarray.c
@@ -1874,6 +1874,10 @@ static void check_split_1(struct xarray *xa, unsigned long index,
 	rcu_read_unlock();
 	XA_BUG_ON(xa, found != 1 << (order - new_order));
 
+	for (i = 0; i < (1 << order); i += (1 << new_order))
+		xa_erase(xa, index + i);
+	XA_BUG_ON(xa, !xa_empty(xa));
+
 	xa_destroy(xa);
 }
 
@@ -1926,6 +1930,10 @@ static void check_split_2(struct xarray *xa, unsigned long index,
 	}
 	rcu_read_unlock();
 	XA_BUG_ON(xa, found != 1 << (order - new_order));
+
+	for (i = 0; i < (1 << order); i += (1 << new_order))
+		xa_erase(xa, index + i);
+	XA_BUG_ON(xa, !xa_empty(xa));
 out:
 	xas_destroy(&xas);
 	xa_destroy(xa);
-- 
2.53.0.345.g96ddfc5eaa-goog