<AWT Dev> [14] Review Request: 8235638 NPE in LWWindowPeer.getOnscreenGraphics()

[Sergey Bylokhov]

I am sorry, I have uploaded it from one system, and then accidentally removed
it by synchronization from another system. webrev is accessible now.

On 12/11/19 9:58 am, Phil Race wrote:
> Maybe Sergey forgot to post it.
> 
> -phil.
> 
> On 12/11/19 1:53 AM, Prasanta Sadhukhan wrote:
>>
>> Hi Dmitry,
>>
>> It seems that webrev is not accessible anymore, but I guess it will be good if both the fix can be clubbed together.
>>
>> Regards
>>
>> Prasanta
>>
>> On 11-Dec-19 3:15 PM, Dmitry Markov wrote:
>>> Hi Prasanta,
>>>
>>> I guess the NPE, you observed, is already addressed by https://bugs.openjdk.java.net/browse/JDK-8235739 (which is on review now, see http://mail.openjdk.java.net/pipermail/awt-dev/2019-December/015621.html).
>>> Can you try with fix for 8235739, please?
>>>
>>> Thanks,
>>> Dmitry
>>>
>>>> On 11 Dec 2019, at 09:16, Prasanta Sadhukhan <prasanta.sadhukhan at oracle.com <mailto:prasanta.sadhukhan at oracle.com>> wrote:
>>>>
>>>> Hi Sergey,
>>>>
>>>> The regression test fails for me on windows with NPE
>>>>
>>>> java.lang.NullPointerException
>>>> 	at java.desktop/sun.awt.windows.WComponentPeer.getGraphics(WComponentPeer.java:582)
>>>> 	at java.desktop/java.awt.Component.getGraphics(Component.java:3124)
>>>> 	at GetGraphicsStressTest.lambda$test$2(GetGraphicsStressTest.java:63)
>>>> 	at java.base/java.lang.Thread.run(Thread.java:833)
>>>> Probably we need to do a null check there in windows too.
>>>>
>>>> Regards
>>>> Prasanta
>>>> On 11-Dec-19 1:02 AM, Sergey Bylokhov wrote:
>>>>> Hello.
>>>>> Please review the fix for JDK 14.
>>>>>
>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8235638
>>>>> Fix: http://cr.openjdk.java.net/~serb/8235638/webrev.00
>>>>>
>>>>> I have found a root cause of intermittent failures of some stress tests in the JDK on macOS.
>>>>> Such tests usually show/hide a lot of frames, and fails because of NPE in the
>>>>> LWWindowPeer.getOnscreenGraphics()
>>>>> The reason is incorrect null check. We should read the surfaceData to the local var apply a
>>>>> null check and then use it, otherwise, the data may be changed to null after the check.
>>>>>
>>>>>
>>>
> 


-- 
Best regards, Sergey.