[vectorIntrinsics] RFR: 8271005: AArch64: Add SVE codegen for VectorMask reduction nodes

[Eric Liu]

On Thu, 22 Jul 2021 10:19:35 GMT, Xiaohong Gong <xgong at openjdk.org> wrote:

> This patch adds the SVE backend implementation for the following VectorMask reduction nodes:
> 
>  - VectorMaskTrueCountNode
>  - VectorMaskFirstTrueNode
>  - VectorMaskLastTrueNode
> 
> It also adds the optimized rules when the mask inputs of these nodes are kind of `"StoreVectorMaskNode"`, which can optimize out the needless codegen for` "StoreVectorMaskNode"`.
> 
> Also change defined temp governing predicate registers to all valid predicate registers for some rules.
> 
> Note that this is the SVE vector implementation version that the VectorMask is represented like the normal vector while not the predicate.

Marked as reviewed by eliu (no project role).

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

PR: https://git.openjdk.java.net/panama-vector/pull/100