[llvm] [AMDGPU] Support merging 16-bit and 8-bit TBUFFER load/store instruction (PR #145078)

[Harrison Hao via llvm-commits]

================
@@ -2353,6 +2406,19 @@ SILoadStoreOptimizer::collectMergeableInsts(
     if (Swizzled != -1 && MI.getOperand(Swizzled).getImm())
       continue;
 
+    if (InstClass == TBUFFER_LOAD || InstClass == TBUFFER_STORE) {
+      const MachineOperand *Fmt =
+          TII->getNamedOperand(MI, AMDGPU::OpName::format);
+      if (!Fmt) {
+        LLVM_DEBUG(dbgs() << "Skip tbuffer without format operand: " << MI);
----------------
harrisonGPU wrote:

I have already removed it. :-)

https://github.com/llvm/llvm-project/pull/145078