[PATCH] D35411: [SimplifyCFG] Defer folding unconditional branches to LateSimplifyCFG if it can destroy canonical loop structure.

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

I'm sort of worried this could have unexpected consequences; do you have performance numbers?  (LLVM testsuite or SPEC)



================
Comment at: lib/CodeGen/CodeGenPrepare.cpp:668
 
     eliminateMostlyEmptyBlock(BB);
     MadeChange = true;
----------------
Do you need to update the latch set here?


================
Comment at: test/Transforms/LoopUnroll/pr33605.ll:1
+; RUN: opt < %s -disable-output -stats -simplifycfg -loop-rotate -loop-unroll -info-output-file - | FileCheck %s --check-prefix=STATS
+
----------------
We usually prefer tests for transformation passes which just check the output of the pass itself, so they aren't so sensitive to changes in other passes.


https://reviews.llvm.org/D35411