[llvm] [RISCV][GISEL] instruction-select for G_SPLAT_VECTOR (PR #111193)

[Craig Topper via llvm-commits]

topperc wrote:

> ```
> void RISCVInstructionSelector::renderImm(MachineInstrBuilder &MIB,
>                                          const MachineInstr &MI,
>                                          int OpIdx) const {
>   assert(MI.getOpcode() == TargetOpcode::G_CONSTANT && OpIdx == -1 &&
>          "Expected G_CONSTANT");
>   int64_t CstVal = MI.getOperand(1).getCImm()->getSExtValue();
>   MIB.addImm(CstVal);
> }
> ```
> 
> Why must `MI` be a G_CONSTANT? Couldn't it be instead a RISCV::IConstant (sorry for the bad name)? To avoid building G_X in the instruction selector.

So you trust us to build a RISCV::IConstant with legal operands, but you don't trust us to know what G_CONSTANT is legal?

We still need to support G_CONSTANT too for future patches that will create G_VMV_V_X_VL/G_VFMV_F_VL from intrinsics.

https://github.com/llvm/llvm-project/pull/111193