[PATCH] D112262: [indvars] Rotate zext though icmp to reduce loop varying computation

[Max Kazantsev via Phabricator via llvm-commits]

mkazantsev added inline comments.


================
Comment at: llvm/lib/Transforms/Scalar/IndVarSimplify.cpp:1513
+    Value *LHSOp = nullptr;
+    if (!match(LHS, m_ZExt(m_Value(LHSOp))) || !ICmp->isUnsigned())
+      continue;
----------------
Looks more logical to check `ICmp->isUnsigned()` just where we checked it was icmp.


================
Comment at: llvm/test/Transforms/IndVarSimplify/finite-exit-comparisons.ll:471
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = trunc i16 254 to i8
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
----------------
Please add a negative test where this doesn't work because of precision loss.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112262/new/

https://reviews.llvm.org/D112262