[PATCH] D139275: [SimplifyCFG] `FoldBranchToCommonDest()`: deal with mismatched IV's in PHI's in common successor block

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Transforms/Utils/SimplifyCFG.cpp:3575
+};
+struct SelectCache {
+  using UniqueSelectIndex = unsigned;
----------------
Comment what a SelectCache is?


================
Comment at: llvm/lib/Transforms/Utils/SimplifyCFG.cpp:3740
+          &PhiRecipe = std::get<1>(I);
+
+      Value **MaterializedSelect;
----------------
Comment what this is trying to do


================
Comment at: llvm/lib/Transforms/Utils/SimplifyCFG.cpp:3742-3743
+      Value **MaterializedSelect;
+      if (std::holds_alternative<CompatibleIncomingValue>(PhiRecipe)) {
+        MaterializedSelect = &std::get<CompatibleIncomingValue>(PhiRecipe).V;
+      } else {
----------------
std::get_if?


================
Comment at: llvm/lib/Transforms/Utils/SimplifyCFG.cpp:3754-3755
+                                   SelRecipe.FalseVal, PN.getName() + ".sel");
+          if (isa<Instruction>(*MaterializedSelect))
+            RemapInstruction(cast<Instruction>(*MaterializedSelect), VMap,
+                             RF_NoModuleLevelChanges | RF_IgnoreMissingLocals);
----------------
dyn_cast instead of isa + cast


================
Comment at: llvm/lib/Transforms/Utils/SimplifyCFG.cpp:3887
+      ++NumSelectsNeeded;
+      SawVectorSelect |= PN.getType()->isVectorTy();
+    }
----------------
I'm not really sure what this is doing; should there be a vector variant of the fpclassify test?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D139275/new/

https://reviews.llvm.org/D139275