[llvm] [X86][GlobalIsel] support G_FABS (PR #136718)

[Simon Pilgrim via llvm-commits]

================
@@ -835,6 +841,43 @@ bool X86LegalizerInfo::legalizeNarrowingStore(MachineInstr &MI,
   return true;
 }
 
+bool X86LegalizerInfo::legalizeFAbs(MachineInstr &MI,
+                                         MachineRegisterInfo &MRI,
+                                         LegalizerHelper &Helper) const {
+
+  MachineIRBuilder &MIRBuilder = Helper.MIRBuilder;
+  Register SrcReg = MI.getOperand(1).getReg();
+  Register DstReg = MI.getOperand(0).getReg();
+  LLT Ty = MRI.getType(DstReg);
+  if (Subtarget.is32Bit()) {
----------------
RKSimon wrote:

I don't understand - why are you making x86_64 use a constant pool for all float types?

https://github.com/llvm/llvm-project/pull/136718