[llvm] [AMDGPU] Implement vop3p complex pattern optmization for gisel (PR #130234)

[Christudasan Devadasan via llvm-commits]

================
@@ -4310,44 +4310,349 @@ AMDGPUInstructionSelector::selectVOP3NoMods(MachineOperand &Root) const {
   }};
 }
 
-std::pair<Register, unsigned>
-AMDGPUInstructionSelector::selectVOP3PModsImpl(
-  Register Src, const MachineRegisterInfo &MRI, bool IsDOT) const {
-  unsigned Mods = 0;
-  MachineInstr *MI = MRI.getVRegDef(Src);
+enum SrcStatus {
----------------
cdevadas wrote:

Better use scoped enumerations.

https://github.com/llvm/llvm-project/pull/130234