RFR: 8292898: [vectorapi] Unify vector mask cast operation [v2]

[Jatin Bhateja]

On Fri, 23 Sep 2022 07:34:43 GMT, Quan Anh Mai <duke at openjdk.org> wrote:

>> src/hotspot/cpu/x86/c2_MacroAssembler_x86.cpp line 4630:
>> 
>>> 4628:         } else {
>>> 4629:           vpmovsxbw(dst, src, vlen_enc);
>>> 4630:         }
>> 
>> if int/short == 2  and target == AVX512BW:  VPMOVSXBW looks incorrect, expectation is to wide cast each 16 bit lane to 32 bit and not 8 bit to 16 bit.
>
> A mask element of type short can only be 0x0000 or 0xFFFF, in those cases both `vpmovsxbw` and `vpmovsxwd` transform it to 0x00000000 and 0xFFFFFFFF, respectively. The same logic applies to the below circumstances.

Correct, its signed extension, not zero extension.

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

PR: https://git.openjdk.org/jdk/pull/10192