RFR (S): 8230395: Code checks for NULL value returned from NEW_C_HEAP_ARRAY which can not happen

[Harold Seigel]

Hi David,

The change looks good.

Can CPUInformationInterface::initialize() be changed to return void?  It 
looks like it will now always returns TRUE.

Thanks, Harold

On 9/19/2019 1:57 AM, David Holmes wrote:
> Bug: https://bugs.openjdk.java.net/browse/JDK-8230395
> webrev: http://cr.openjdk.java.net/~dholmes/8230395/webrev/
>
> First, thanks to Leo for going through and finding the places that 
> needed to be fixed. I didn't go looking for others.
>
> For the most part I have deleted the NULL checks. A few notes:
>
> - changed to NEW_C_HEAP_ARRAY_RETURN_NULL in a couple of places where 
> it was more consistent with the general recoverability approach of the 
> code in question (and did the opposite change in one place)
> - for the "perf" changes I also noticed that we were checking for NULL 
> after doing "new SomeCHeapObj()" which also cannot return NULL, so 
> removed those NULL checks too
> - the JVMCI thread counter changes became a bit more extensive due to 
> the fact a boolean method could now only every return true. So it was 
> changed to void and that had to filter all the way back up to the 
> JVMCI Java API. The JVMCI/Graal folk need to approve this
>
> Testing: tier 1-3 sanity tests (in progress)
>
> Thanks,
> David