[llvm] [InstCombine] Constant-fold riscv.vmv.v.x (PR #182630)

[Yingwei Zheng via llvm-commits]

================
@@ -4262,6 +4263,44 @@ Instruction *InstCombinerImpl::visitCallInst(CallInst &CI) {
           *II, Builder.CreateZExtOrTrunc(II->getArgOperand(0), II->getType()));
     return nullptr;
   }
+  case Intrinsic::riscv_vmv_v_x: {
----------------
dtcxzyw wrote:

Can we put this under `RISCVTTIImpl::instCombineIntrinsic`? Most of the target-dependent folds are placed under TTIImpl. The only exception is the common transformation shared by both ARM and AArch64 (https://github.com/llvm/llvm-project/pull/169589).

https://github.com/llvm/llvm-project/pull/182630