RFR: 8368206: RISC-V: compiler/vectorapi/VectorMaskCompareNotTest.java fails when running without RVV

[Hamlin Li]

On Thu, 25 Sep 2025 04:31:55 GMT, Dingli Zhang <dzhang at openjdk.org> wrote:

>>> Hey, I'm wondering if all the tests under hotspot/jtreg/compiler/vectorapi should `@require` rvv? Otherwise seems they are not really testing anything useful?
>> 
>> @Hamlin-Li Good question! I think almost all IR related tests need RVV. Some non-IR tests can use scalars to implement vectorapi, such as `compiler/vectorapi/TestVectorShuffleIotaByte.java`, which also passes on sg2042 (without RVV).
>
>> Hey, I'm wondering if all the tests under hotspot/jtreg/compiler/vectorapi should `@require` rvv? Otherwise seems they are not really testing anything useful?
> 
> @Hamlin-Li Sorry, I think I misunderstood you earlier.
> Currently, this is the only test case under `hotspot/jtreg/compiler/vectorapi` that fails without RVV, so I plan to merge this PR first and then open a separate issue [JDK-8368602](https://bugs.openjdk.org/browse/JDK-8368602) to discuss the situation you mentioned.
> Does that work for you ?

@DingliZhang Thank you. The current PR looks good.

Not sure what to do with other vectorapi tests, but it's good to have a look if you got time. Thanks!

-------------

PR Comment: https://git.openjdk.org/jdk/pull/27414#issuecomment-3332760654