[llvm] AMDGPU/GlobalISelDivergenceLowering: select divergent i1 phis (PR #78482)

[Pierre van Houtryve via llvm-commits]

================
@@ -64,5 +205,14 @@ FunctionPass *llvm::createAMDGPUGlobalISelDivergenceLoweringPass() {
 
 bool AMDGPUGlobalISelDivergenceLowering::runOnMachineFunction(
     MachineFunction &MF) {
-  return false;
+  MachineDominatorTree &DT = getAnalysis<MachineDominatorTree>();
+  MachinePostDominatorTree &PDT = getAnalysis<MachinePostDominatorTree>();
+  MachineUniformityInfo &MUI =
+      getAnalysis<MachineUniformityAnalysisPass>().getUniformityInfo();
+
+  DivergenceLoweringHelper Helper(&MF, &DT, &PDT, &MUI);
+
+  bool Changed = false;
+  Changed |= Helper.lowerPhis();
+  return Changed;
----------------
Pierre-vh wrote:

```suggestion
  return Helper.lowerPhis();
```

https://github.com/llvm/llvm-project/pull/78482