[llvm] [AMDGPU] Extend getMemOperandAACheckLimit (PR #172084)

[Gang Chen via llvm-commits]

================
@@ -440,6 +440,8 @@ class SIInstrInfo final : public AMDGPUGenInstrInfo {
   bool foldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, Register Reg,
                      MachineRegisterInfo *MRI) const final;
 
+  unsigned getMemOperandAACheckLimit() const override { return 32; }
----------------
cmc-rep wrote:

Good point. Doing that, I won't need upstream

https://github.com/llvm/llvm-project/pull/172084