[llvm] [AMDGPU] Add option to bias SGPR allocation to reduce read hazards (PR #129869)

[Matt Arsenault via llvm-commits]

================
@@ -3904,3 +3909,166 @@ SIRegisterInfo::getVRegFlagsOfReg(Register Reg,
     RegFlags.push_back("WWM_REG");
   return RegFlags;
 }
+
+unsigned SIRegisterInfo::getSGPRHazardAvoidanceStrategy(
+    const MachineFunction &MF) const {
+  if (SGPRHazardAvoidanceStrategy.getNumOccurrences()) {
+    return SGPRHazardAvoidanceStrategy;
+  } else {
----------------
arsenm wrote:

no else after return 

https://github.com/llvm/llvm-project/pull/129869