[clang] [llvm] [AMDGPU][SIInsertWaitcnt] Implement Waitcnt Expansion for Profiling (PR #169345)

[Jay Foad via cfe-commits]

================
@@ -375,7 +401,10 @@ class WaitcntGenerator {
 
 class WaitcntGeneratorPreGFX12 : public WaitcntGenerator {
 public:
-  using WaitcntGenerator::WaitcntGenerator;
+  WaitcntGeneratorPreGFX12() = default;
+  WaitcntGeneratorPreGFX12(const MachineFunction &MF,
+                           const AMDGPU::HardwareLimits *Limits)
+      : WaitcntGenerator(MF, NUM_NORMAL_INST_CNTS, Limits) {}
----------------
jayfoad wrote:

Was there any particular reason to change this? It seems like WaitcntGeneratorPreGFX12 could still just inherit constructors from the base class instead of redefining them here.

https://github.com/llvm/llvm-project/pull/169345