[vectorIntrinsics+mask] RFR: 8272479: Java API and IR changes for masked rearrange operation [v2]

[Xiaohong Gong]

On Mon, 16 Aug 2021 08:13:32 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>> - Re-organized the Java API implementation to use one common entry point of masked/non-masked vector compare operation.
>>  - Intrinsic side changes to absorb newly passed mask if target support predicated instructions else handled appropriately by blending full vector rearrange vector with broadcasted ZERO vector.
>
> Jatin Bhateja has updated the pull request incrementally with one additional commit since the last revision:
> 
>   8272479: Review comments resolution.

The compiler part looks better for me! Just several trivial comments.

src/hotspot/share/opto/vectorIntrinsics.cpp line 1761:

> 1759:       shuffle_klass == NULL ||
> 1760:       elem_klass == NULL    ||
> 1761:       vlen == NULL) {

Revert all these changes?

src/hotspot/share/opto/vectorIntrinsics.cpp line 1813:

> 1811:     }
> 1812:   }
> 1813:   if (!arch_supports_vector(Op_VectorRearrange, num_elem, elem_bt, is_masked_op ? VecMaskUsePred : VecMaskNotUsed)) {

For masked op, it's better to use the mask type `VecMaskUsePred | VecMaskUseLoad` here?

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

PR: https://git.openjdk.java.net/panama-vector/pull/110