<AWT Dev> [7u6] Review request for 7144542: [macosx] Crash in liblwawt.dylib setBusy() when exiting an FX app run with j2d pipeline

[Artem Ananiev]

Looks fine. Thanks for additional testing!

Artem

On 5/15/2012 8:46 PM, Anthony Petrov wrote:
> Hi Artem,
>
> On 5/10/2012 2:33 PM, Artem Ananiev wrote:
>> On 5/4/2012 6:53 PM, Anthony Petrov wrote:
>>> Please review a fix for
>>> http://bugs.sun.com/view_bug.do?bug_id=7144542 at:
>>>
>>> http://cr.openjdk.java.net/~anthony/7u6-6-crashInSetBusy-7144542.0/
>>
>> So what you do is to change headful CToolkit to headless XToolkit,
>> correct? Note that this change will also trigger Java2D pipeline
>> change, which might not be what we want.
>
> Actually, while performing more testing I've found one issue with the
> fix: we must also choose the correct font manager class. Otherwise the
> MACOSX_PORT-705 is resurrected. An updated fix is here:
>
> http://cr.openjdk.java.net/~anthony/7u6-6-crashInSetBusy-7144542.1/
>
> Meanwhile I've performed some performance measurement using the GUIMark2
> for FX, and here are some results:
>
> BitmapTest:
> w/o the fix: ~4.22 FPS
> with the fix: ~4.19 FPS
>
> VectorTest:
> w/o the fix: ~21.0 FPS
> with the fix: ~21.3 FPS
>
> So it looks like there's no any major performance difference between the
> CToolkit and XToolkit Java2D pipelines.
>
>> Have you tried to find out what exactly leads to the crash when
>> CToolkit is used?
>
> Since the crash is actually happening in the HotSpot code, it's being
> investigated by HotSpot team under 7166725.
>
> --
> best regards,
> Anthony
>
>
>>
>> Thanks,
>>
>> Artem
>>
>>> Note: in JDK 8 we already load the headless version of the AWT library,
>>> and this prevents the crash. Now I'm applying the same logic to JDK 7u6.
>>>
>>> --
>>> best regards,
>>> Anthony