[clang] [llvm] [Clang]: Enable speculative devirtualization (PR #159685)

[Teresa Johnson via llvm-commits]

================
@@ -1641,6 +1642,24 @@ PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level,
   if (!LTOPreLink)
     MPM.addPass(RelLookupTableConverterPass());
 
+  if (PTO.DevirtualizeSpeculatively && LTOPhase == ThinOrFullLTOPhase::None) {
+    MPM.addPass(WholeProgramDevirtPass(
+        /*ExportSummary*/ nullptr,
+        /*ImportSummary*/ nullptr,
+        /*DevirtSpeculatively*/ PTO.DevirtualizeSpeculatively));
+    MPM.addPass(LowerTypeTestsPass(nullptr, nullptr,
+                                   lowertypetests::DropTestKind::Assume));
+    if (EnableModuleInliner) {
----------------
teresajohnson wrote:

Can you remind me why you need another round of inlining vs doing this before the earlier inline pass? Is this an optimization for a specific use case that needs 2 rounds of inlining? I'm concerned about the potential side effects.

https://github.com/llvm/llvm-project/pull/159685