[Mlir-commits] [flang] [mlir] [Flang] Add new ConvertComplexPow pass for Flang (PR #158642)

[Akash Banerjee]

================
@@ -1698,15 +1698,15 @@ static constexpr MathOperation mathOperations[] = {
      genFuncType<Ty::Complex<8>, Ty::Complex<8>, Ty::Integer<4>>,
      genComplexPow},
     {"pow", RTNAME_STRING(cqpowi), FuncTypeComplex16Complex16Integer4,
-     genLibF128Call},
+     genComplexPow},
----------------
TIFitis wrote:

`genComplexMathOp` would lower to `libCall` `if (!forceMlirComplex && !canUseApprox && !isAMDGPU)`
which means we would restrict lowering to `complex.pow` for some cases. Is that okay?

https://github.com/llvm/llvm-project/pull/158642