Arm64 SVE: Fix conditionalselect with constant arguments

[a74nh]

Fixes #116847

When folding, allow arg1 to be a constant mask

[a74nh]

@dotnet/arm64-contrib @kunalspathak

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

[kunalspathak]

/azp run Antigen, Fuzzlyn

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[kunalspathak]

wait, we should be check isSet with content of input0, right?

[a74nh]

wait, we should be check isSet with content of input0, right?

Yes, I should be checking significant bit of input0

[kunalspathak]

There is a bug in EvaluateSimdVectorToPattern where we are not validating it against the contents of arg0.

[a74nh]

The simd.h changes in this PR are out of date. We should wait for #116854 to be merged first.

[kunalspathak]

i will hold off reviewing this until #116854 is merged.

[a74nh]

Fuzzlyn was showing some errors with

Assert failure(PID 1845356 [0x001c286c], Thread: 1845356 [0x1c286c]): Assertion failed 'OperIs(GT_CNS_VEC)' in 'Program:M1(byref,int,System.Runtime.Intrinsics.Vector64`1[ushort]):ushort' during 'Morph - Global' (IL size 56; hash 0xc2f1cf02; FullOpts)

    File: /mnt/sdb/home/alahay01/dotnet/runtime_madv/src/coreclr/jit/gtstructs.h:64
    Image: /mnt/sdb/home/alahay01/dotnet/runtime_madv/artifacts/tests/coreclr/linux.arm64.Checked/Tests/Core_Root/corerun

This is due to EvalHWIntrinsicFunTernary() not treating arg1 of NI_Sve_ConditionalSelect` as a mask.

The fix is closely related to the fix in this PR, so I've put it into this PR with a test.

I've also updated simd.h to the latest from #116854

[kunalspathak]

LGTM

[a74nh]

I've put another closely related Fuzzlyn failure into this PR.

Assert failure(PID 3199223 [0x0030d0f7], Thread: 3199223 [0x30d0f7]): Assertion failed 'unreached' in 'ConditionalSelectConstants:ConditionalSelectConstsNoMaskPattern():bool' during 'Generate code' (IL size 74; hash 0xca97730c; FullOpts)

    File: /mnt/sdb/home/alahay01/dotnet/runtime_sve/src/coreclr/jit/codegenarm64.cpp:2379
    Image: /mnt/sdb/home/alahay01/dotnet/runtime_sve/artifacts/tests/coreclr/linux.arm64.Checked/Tests/Core_Root/corerun

What happens here is all the inputs to the conditionalselect are constants, so it is optimised away into a single vector. It is then used as a mask in the next function call, so is optimised to a cns mask. The mask cannot be converted into a valid pattern during codegen (for use in a ptrue), so it asserts. The fix is to convert it to a constant vector with conversion to mask. Best way to do that is during lowering (see LowerCnsMask()), this ensures we have the registers etc required for the conversion.

[a74nh]

With #116991 applied on top I don't see any fuzzlyn errors over 50mins. I'm happy with this PR now.

[a74nh]

Updated as per review comments. Fuzzlyn still looks good when #116991 is applied.