[llvm] [AMDGPU][InstCombine] Optimize constant shuffle patterns (PR #192246)

[Matt Arsenault via llvm-commits]

================
@@ -718,6 +657,331 @@ GCNTTIImpl::hoistLaneIntrinsicThroughOperand(InstCombiner &IC,
   return nullptr;
 }
 
+static std::optional<unsigned> evalLaneExpr(Value *V, unsigned Lane,
+                                            const GCNSubtarget &ST,
+                                            const DataLayout &DL,
+                                            unsigned Depth = 0) {
+  if (Depth >= MaxAnalysisRecursionDepth)
+    return std::nullopt;
+
+  if (isThreadID(ST, V))
+    return Lane;
+
+  if (const ConstantInt *CI = dyn_cast<ConstantInt>(V))
+    return CI->getZExtValue();
----------------
arsenm wrote:

```suggestion
  if (const ConstantInt *CI = dyn_cast<ConstantInt>(V))
    return CI->getZExtValue();

  if (isThreadID(ST, V))
    return Lane;
```

https://github.com/llvm/llvm-project/pull/192246