[PATCH mmotm] radix-tree: rewrite radix_tree_locate_item fix

[PATCH mmotm] radix-tree: rewrite radix_tree_locate_item fix

[Hugh Dickins]

radix_tree_locate_item() is often returning the wrong index, causing
swapoff of shmem to hang because it cannot find the swap entry there.
__locate()'s use of base is bogus, it adds an offset twice into index.

Signed-off-by: Hugh Dickins <hughd@google.com>
---
Fix to radix-tree-rewrite-radix_tree_locate_item.patch

 lib/radix-tree.c |    5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

--- 4.6-rc5-mm1/lib/radix-tree.c	2016-04-30 22:55:06.067184898 -0700
+++ linux/lib/radix-tree.c	2016-05-01 18:52:06.668085420 -0700
@@ -1254,15 +1254,14 @@ struct locate_info {
 static unsigned long __locate(struct radix_tree_node *slot, void *item,
 			      unsigned long index, struct locate_info *info)
 {
-	unsigned long base, i;
+	unsigned long i;
 
 	do {
 		unsigned int shift = slot->shift;
-		base = index & ~((1UL << shift) - 1);
 
 		for (i = (index >> shift) & RADIX_TREE_MAP_MASK;
 		     i < RADIX_TREE_MAP_SIZE;
-		     i++, index = base + (i << shift)) {
+		     i++, index += (1UL << shift)) {
 			struct radix_tree_node *node =
 					rcu_dereference_raw(slot->slots[i]);
 			if (node == RADIX_TREE_RETRY)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH] radix-tree: add test for radix_tree_locate_item()

[Ross Zwisler]

Add a unit test that provides coverage for the bug fixed in the commit
entitled "radix-tree: rewrite radix_tree_locate_item fix" from Hugh
Dickins.  I've verified that this test fails before his patch due to
miscalculated 'index' values in __locate() in lib/radix-tree.c, and passes
with his fix.

Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
---
 tools/testing/radix-tree/linux/init.h |  1 +
 tools/testing/radix-tree/main.c       | 15 ++++++++++++++-
 2 files changed, 15 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/radix-tree/linux/init.h

diff --git a/tools/testing/radix-tree/linux/init.h b/tools/testing/radix-tree/linux/init.h
new file mode 100644
index 0000000..360cabb
--- /dev/null
+++ b/tools/testing/radix-tree/linux/init.h
@@ -0,0 +1 @@
+/* An empty file stub that allows radix-tree.c to compile. */
diff --git a/tools/testing/radix-tree/main.c b/tools/testing/radix-tree/main.c
index 65231e9..b7619ff 100644
--- a/tools/testing/radix-tree/main.c
+++ b/tools/testing/radix-tree/main.c
@@ -232,7 +232,7 @@ void copy_tag_check(void)
 	item_kill_tree(&tree);
 }
 
-void __locate_check(struct radix_tree_root *tree, unsigned long index,
+static void __locate_check(struct radix_tree_root *tree, unsigned long index,
 			unsigned order)
 {
 	struct item *item;
@@ -248,12 +248,25 @@ void __locate_check(struct radix_tree_root *tree, unsigned long index,
 	}
 }
 
+static void __order_0_locate_check(void)
+{
+	RADIX_TREE(tree, GFP_KERNEL);
+	int i;
+
+	for (i = 0; i < 50; i++)
+		__locate_check(&tree, rand() % INT_MAX, 0);
+
+	item_kill_tree(&tree);
+}
+
 static void locate_check(void)
 {
 	RADIX_TREE(tree, GFP_KERNEL);
 	unsigned order;
 	unsigned long offset, index;
 
+	__order_0_locate_check();
+
 	for (order = 0; order < 20; order++) {
 		for (offset = 0; offset < (1 << (order + 3));
 		     offset += (1UL << order)) {
-- 
2.5.5

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH mmotm] radix-tree: rewrite radix_tree_locate_item fix

[Ross Zwisler]

On Sun, May 01, 2016 at 09:13:18PM -0700, Hugh Dickins wrote:
> radix_tree_locate_item() is often returning the wrong index, causing
> swapoff of shmem to hang because it cannot find the swap entry there.
> __locate()'s use of base is bogus, it adds an offset twice into index.
> 
> Signed-off-by: Hugh Dickins <hughd@google.com>

Thank you for the fix!

Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>