[OpenJDK 2D-Dev] [10] RFR: JDK-8188083- NPE in java.awt.image.FilteredImageSource.startProduction

Prahalad Kumar Narayanan prahalad.kumar.narayanan at oracle.com
Tue Nov 28 07:31:38 UTC 2017


Hello Sergey

Thank you for your review response.

I had one version of the test that resembled your test code. 
I uploaded the result (screenshot) of that test case with/ without fix on JBS.

The only point of concern - API documentation mentions that the methods of FilteredImageSource are not supposed to be invoked directly from user code. If the methods are invoked directly from user code, the result is un-defined. Since our JTreg test cases cannot violate the documented guide, I refrained from attaching the test case in the first webrev.

Kindly let me know your views to this.

Thank you once again
Have a good day

Prahalad 

-----Original Message-----
From: Sergey Bylokhov 
Sent: Tuesday, November 28, 2017 12:46 PM
To: Prahalad Kumar Narayanan; 2d-dev
Cc: Philip Race
Subject: Re: [OpenJDK 2D-Dev] [10] RFR: JDK-8188083- NPE in java.awt.image.FilteredImageSource.startProduction

On 27/11/2017 22:12, Prahalad Kumar Narayanan wrote:
> Yes..! As you rightly pointed, the test passes before the fix as well.

Then I suggest to additionally reuse a test which I sent previously, it is fail almost always before the fix. The test which fails one time in
10000 runs will be never fixed, or such bugs will be closed as not reproducible.

> 
> Reason is that, the occurrence of this bug is rare.
>      . Submitter has observed only 15 occurrences of the exception with over 10,000 instances per month.
>      . Though the occurrence is rare, submitter is expecting a fix for this issue.
> 
> So what does this test-case resolve ?
>      . Our change adds "synchronized" contract to the method to fix the Null pointer exception.
>      . To test the changes, we should check for both- Null Pointer Exception and any deadlock that could arise from "synchronized" contract.
>      . The test case helps to address both but in a passive way. Why do I say passive ?
>            . First, the test provides a hope to detect & report NPE though its occurrence is rare.
>            . Second, any deadlock arising from the concerned method, will cause the test to fail upon time-out.
>   
> Thank you once again for your time
> Have a good day
> 
> Prahalad
> 
> -----Original Message-----
> From: Sergey Bylokhov
> Sent: Monday, November 27, 2017 9:11 PM
> To: Prahalad Kumar Narayanan; 2d-dev
> Cc: Philip Race
> Subject: Re: [OpenJDK 2D-Dev] [10] RFR: JDK-8188083- NPE in java.awt.image.FilteredImageSource.startProduction
> 
> Hi,Prahalad.
> On 24/11/2017 21:40, Prahalad Kumar Narayanan wrote:
>> Based on discussions with Sergey, I 've now updated the fix with a test case.
>> The changes are available for review under:
>> http://cr.openjdk.java.net/~pnarayanan/8188083/webrev.01/
> 
> This version of the test is always passed before the fix(I have checked w/ and w/o jtreg).
> 
> --
> Best regards, Sergey.
> 


-- 
Best regards, Sergey.


More information about the 2d-dev mailing list