<AWT Dev> [8] Request for review: 8003169: [macosx] JVM crash after disconnecting from projector

Anthony Petrov anthony.petrov at oracle.com
Tue Feb 26 07:45:37 PST 2013


best regards,

On 2/26/2013 19:38, Sergey Bylokhov wrote:
> Hi, Anthony.
> Even if this is possible I don't want to delete this code, otherwise I 
> should synchronize operation of this method with the native 
> callbacks(change of parameters of the display, etc)
> 26.02.2013 18:28, Anthony Petrov wrote:
>> Hi Sergey,
>> The fix looks good. However, is there a need to dispatch the 
>> nativeGetScreenInsets operation on the AppKit thread? Form a recent 
>> discussion on threading in FX we found out that NSScreen doesn't imply 
>> any threading restrictions on its usage, i.e. you can access NSScreen 
>> APIs from any thread. Would anything break if you comment out the 
>> performOnMainThreadWaiting call and execute the block directly on the 
>> current thread?
>> -- 
>> best regards,
>> Anthony
>> On 2/26/2013 17:45, Sergey Bylokhov wrote:
>>> Hello,
>>> Please review the fix for jdk 8. Fix will be ported to jdk7 as well.
>>> In the fix  getScreenInsets() implementation was moved to the one 
>>> native method. I assume that it doesn't lead to incorrect nsscreen in 
>>> the middle of operations, if we run all appropriate operation in one 
>>> step on appkit thread. CFRetain/CFRelease are unnecessary in this case.
>>> nativeGetDisplay** methods were changed to static, because in the 
>>> native they implemented as a static.
>>> Bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8003169
>>> Webrev can be found at: 
>>> http://cr.openjdk.java.net/~serb/8003169/webrev.00

More information about the awt-dev mailing list