[LV] Use requested calling convention for vector math routines

Some vector math routines, e.g. ArmPL, specify a particular
calling convention on the routines which can help improve
performance by specifying what registers have to be preserved
across the call.

Author: david-arm

llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp

@@ -1206,6 +1206,7 @@ void VPIRPhi::print(raw_ostream &O, const Twine &Indent,
 
 void VPWidenCallRecipe::execute(VPTransformState &State) {
   assert(State.VF.isVector() && "not widening");
+  assert(Variant != nullptr && "Can't create vector function.");
 
   FunctionType *VFTy = Variant->getFunctionType();
   // Add return type if intrinsic is overloaded on it.
@@ -1222,15 +1223,14 @@ void VPWidenCallRecipe::execute(VPTransformState &State) {
     Args.push_back(Arg);
   }
 
-  assert(Variant != nullptr && "Can't create vector function.");
-
   auto *CI = cast_or_null<CallInst>(getUnderlyingValue());
   SmallVector<OperandBundleDef, 1> OpBundles;
   if (CI)
     CI->getOperandBundlesAsDefs(OpBundles);
 
   CallInst *V = State.Builder.CreateCall(Variant, Args, OpBundles);
   setFlags(V);
+  V->setCallingConv(Variant->getCallingConv());
 
   if (!V->getType()->isVoidTy())
     State.set(this, V);