The fix looks good to me.
Thanks,
Alexandr.
On 11/16/2016 4:05 PM, Sergey Bylokhov wrote:
Hello.
Please review the fix for jdk9.
There are a few issues which can cause a hang in some of our menu
stress tests:
- Lack in synchronization in the Menu, we tried to synchronized the
write of fields, but skip synchronization when we read the fields(in
getters for example). In the fix most of the fields were marked as
final(if possible) or volatile.
- When the submenu is removed from Menu/MenuBar we do not reset its
parent field if the Menu/MenuBar have peer==null. So if later we tried
to call MenuBar.setHelpMenu(submenu) we skip this submenu because we
think it was added already.
Bug: https://bugs.openjdk.java.net/browse/JDK-8165769
Webrev can be found at:
http://cr.openjdk.java.net/~serb/8165769/webrev.00