[llvm] [X86] Lower scalar llvm.clmul intrinsics to PCLMULQDQ (#175189) (PR #175216)

[Simon Pilgrim via llvm-commits]

================
@@ -483,6 +483,13 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
 
   setOperationAction(ISD::READCYCLECOUNTER , MVT::i64  , Custom);
 
+  if (Subtarget.hasPCLMUL() && Subtarget.is64Bit()) {
----------------
RKSimon wrote:

Better off putting this inside the hasSSE2 / softfloat / implicit float block?

https://github.com/llvm/llvm-project/pull/175216