On 12/17/2013 01:40 PM, Peter Crosthwaite wrote:
On Tue, Dec 17, 2013 at 7:20 PM, Sergey Fedorov <s.fedo...@samsung.com> wrote:
A single cast cache is used for both an object casting and a class
casting. In case of interface presence a class cast result may be not
the same pointer as opposite to an object casting. So do not cache cast
results for an object casting in a presence of interfaces.
I think this is fixed by my cast cache splitter patch which is
currently enqueued.
http://patchwork.ozlabs.org/patch/294766/
Regards,
Peter
Yes, your cache splitter looks much better. Thank you.
Best regards,
Sergey Fedorov
Signed-off-by: Sergey Fedorov <s.fedo...@samsung.com>
---
qom/object.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/qom/object.c b/qom/object.c
index fc19cf6..f7384de 100644
--- a/qom/object.c
+++ b/qom/object.c
@@ -473,7 +473,7 @@ Object *object_dynamic_cast_assert(Object *obj, const char
*typename,
assert(obj == inst);
- if (obj && obj == inst) {
+ if (obj && obj == inst && !obj->class->interfaces) {
for (i = 1; i < OBJECT_CLASS_CAST_CACHE; i++) {
obj->class->cast_cache[i - 1] = obj->class->cast_cache[i];
}
--
1.7.9.5
