[InstCombine] canonicalize -0.0 to +0.0 in fcmp

As stated in IEEE-754 and discussed in:
https://bugs.llvm.org/show_bug.cgi?id=38086
...the sign of zero does not affect any FP compare predicate.

Known regressions were fixed with:
rL346097 (D54001)
rL346143

The transform will help reduce pattern-matching complexity to solve:
https://bugs.llvm.org/show_bug.cgi?id=39475
...as well as improve CSE and codegen (a zero constant is almost always
easier to produce than 0x80..00).

llvm-svn: 346147

1 files changed, 7 insertions, 0 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index 9155ad12598..059f7523ff9 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -5397,6 +5397,13 @@ Instruction *InstCombiner::visitFCmpInst(FCmpInst &I) {
         return nullptr;
     }
 
+  // The sign of 0.0 is ignored by fcmp, so canonicalize to +0.0:
+  // fcmp Pred X, -0.0 --> fcmp Pred X, 0.0
+  if (match(Op1, m_AnyZeroFP()) && !match(Op1, m_PosZeroFP())) {
+    I.setOperand(1, ConstantFP::getNullValue(Op1->getType()));
+    return &I;
+  }
+
   // Handle fcmp with instruction LHS and constant RHS.
   Instruction *LHSI;
   Constant *RHSC;