[ValueTracking] determine sign of 0.0 from select when matching min/max FP
In PR39475:
https://bugs.llvm.org/show_bug.cgi?id=39475
..we may fail to recognize/simplify fabs() in some cases because we do not
canonicalize fcmp with a -0.0 operand.
Adding that canonicalization can cause regressions on min/max FP tests, so
that's this patch: for the purpose of determining whether something is min/max,
let the value returned by the select determine how we treat a 0.0 operand in the fcmp.
This patch doesn't actually change the -0.0 to +0.0. It just changes the analysis, so
we don't fail to recognize equivalent min/max patterns that only differ in the
signbit of 0.0.
Differential Revision: https://reviews.llvm.org/D54001
llvm-svn: 346097
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
index 6e08272..ed17441 100644
--- a/llvm/lib/Analysis/ValueTracking.cpp
+++ b/llvm/lib/Analysis/ValueTracking.cpp
@@ -4760,6 +4760,27 @@
Value *TrueVal, Value *FalseVal,
Value *&LHS, Value *&RHS,
unsigned Depth) {
+ if (CmpInst::isFPPredicate(Pred)) {
+ // IEEE-754 ignores the sign of 0.0 in comparisons. So if the select has one
+ // 0.0 operand, set the compare's 0.0 operands to that same value for the
+ // purpose of identifying min/max. Disregard vector constants with undefined
+ // elements because those can not be back-propagated for analysis.
+ Value *OutputZeroVal = nullptr;
+ if (match(TrueVal, m_AnyZeroFP()) && !match(FalseVal, m_AnyZeroFP()) &&
+ !cast<Constant>(TrueVal)->containsUndefElement())
+ OutputZeroVal = TrueVal;
+ else if (match(FalseVal, m_AnyZeroFP()) && !match(TrueVal, m_AnyZeroFP()) &&
+ !cast<Constant>(FalseVal)->containsUndefElement())
+ OutputZeroVal = FalseVal;
+
+ if (OutputZeroVal) {
+ if (match(CmpLHS, m_AnyZeroFP()))
+ CmpLHS = OutputZeroVal;
+ if (match(CmpRHS, m_AnyZeroFP()))
+ CmpRHS = OutputZeroVal;
+ }
+ }
+
LHS = CmpLHS;
RHS = CmpRHS;