[llvm] [AMDGPU] Add function attribute to disable TBUFFER combine (PR #156454)

[Harrison Hao via llvm-commits]

================
@@ -195,6 +195,9 @@ SIMachineFunctionInfo::SIMachineFunctionInfo(const Function &F,
     VGPRForAGPRCopy =
         AMDGPU::VGPR_32RegClass.getRegister(ST.getMaxNumVGPRs(F) - 1);
   }
+
+  if (F.hasFnAttribute("amdgpu-disable-tbuffer-combine"))
+    setDisableTBufferCombine(true);
----------------
harrisonGPU wrote:

I think we should enable tbuffer combine by default, because its logic is generally correct, and only certain features disallow it. What do you think?

https://github.com/llvm/llvm-project/pull/156454