[PATCH] D110712: [LoopFlatten] Bail if we can't perform flattening after IV widening

[Dave Green via Phabricator via llvm-commits]

dmgreen added inline comments.


================
Comment at: llvm/lib/Transforms/Scalar/LoopFlatten.cpp:748-749
+  // TODO: IV widening without performing the actual flattening transformation
+  // is not ideal. While this codegen change should not matter much, it is an
+  // unnecessary change which is better to avoid. It's unlikely his happens
+  // often, because if it's unprofitibale after widening, it should be
----------------
his -> this


================
Comment at: llvm/lib/Transforms/Scalar/LoopFlatten.cpp:755
+  if (FI.Widened && !CanFlatten)
+    return false;
+
----------------
I think false means "the code didn't change". But it will have had phi's widened so should be returning true?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D110712/new/

https://reviews.llvm.org/D110712