fix TLI's combineRepeatedFPDivisors interface to return the minimum user threshold

This fix was suggested as part of D11345 and is part of fixing PR24141.

With this change, we can avoid walking the uses of a divisor node if the target
doesn't want the combineRepeatedFPDivisors transform in the first place.

There is no NFC-intended other than that.

Differential Revision: http://reviews.llvm.org/D11531

llvm-svn: 243498

llvm/include/llvm/Target/TargetLowering.h

@@ -2732,10 +2732,12 @@ public:
     return SDValue();
   }
 
-  /// Indicate whether this target prefers to combine the given number of FDIVs
-  /// with the same divisor.
-  virtual bool combineRepeatedFPDivisors(unsigned NumUsers) const {
-    return false;
+  /// Indicate whether this target prefers to combine FDIVs with the same
+  /// divisor. If the transform should never be done, return zero. If the
+  /// transform should be done, return the minimum number of divisor uses
+  /// that must exist.
+  virtual unsigned combineRepeatedFPDivisors() const {
+    return 0;
   }
 
   /// Hooks for building estimates in place of slower divisions and square

llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

@@ -8247,23 +8247,29 @@ SDValue DAGCombiner::combineRepeatedFPDivisors(SDNode *N) {
   if (!DAG.getTarget().Options.UnsafeFPMath)
     return SDValue();
   
+  // Skip if current node is a reciprocal.
   SDValue N0 = N->getOperand(0);
   ConstantFPSDNode *N0CFP = dyn_cast<ConstantFPSDNode>(N0);
-
-  // Skip if current node is a reciprocal.
   if (N0CFP && N0CFP->isExactlyValue(1.0))
     return SDValue();
   
+  // Exit early if the target does not want this transform or if there can't
+  // possibly be enough uses of the divisor to make the transform worthwhile.
   SDValue N1 = N->getOperand(1);
-  SmallVector<SDNode *, 4> Users;
+  unsigned MinUses = TLI.combineRepeatedFPDivisors();
+  if (!MinUses || N1->use_size() < MinUses)
+    return SDValue();
 
   // Find all FDIV users of the same divisor.
+  SmallVector<SDNode *, 4> Users;
   for (auto *U : N1->uses()) {
     if (U->getOpcode() == ISD::FDIV && U->getOperand(1) == N1)
       Users.push_back(U);
   }
 
-  if (!TLI.combineRepeatedFPDivisors(Users.size()))
+  // Now that we have the actual number of divisor uses, make sure it meets
+  // the minimum threshold specified by the target.
+  if (Users.size() < MinUses)
     return SDValue();
 
   EVT VT = N->getValueType(0);

llvm/lib/Target/AArch64/AArch64ISelLowering.cpp

@@ -9427,10 +9427,10 @@ bool AArch64TargetLowering::useLoadStackGuardNode() const {
   return true;
 }
 
-bool AArch64TargetLowering::combineRepeatedFPDivisors(unsigned NumUsers) const {
+unsigned AArch64TargetLowering::combineRepeatedFPDivisors() const {
   // Combine multiple FDIVs with the same divisor into multiple FMULs by the
   // reciprocal if there are three or more FDIVs.
-  return NumUsers > 2;
+  return 3;
 }
 
 TargetLoweringBase::LegalizeTypeAction

llvm/lib/Target/AArch64/AArch64ISelLowering.h

@@ -477,7 +477,7 @@ private:
 
   SDValue BuildSDIVPow2(SDNode *N, const APInt &Divisor, SelectionDAG &DAG,
                         std::vector<SDNode *> *Created) const override;
-  bool combineRepeatedFPDivisors(unsigned NumUsers) const override;
+  unsigned combineRepeatedFPDivisors() const override;
 
   ConstraintType getConstraintType(StringRef Constraint) const override;
   unsigned getRegisterByName(const char* RegName, EVT VT,

llvm/lib/Target/PowerPC/PPCISelLowering.cpp

@@ -9131,7 +9131,7 @@ SDValue PPCTargetLowering::getRecipEstimate(SDValue Operand,
   return SDValue();
 }
 
-bool PPCTargetLowering::combineRepeatedFPDivisors(unsigned NumUsers) const {
+unsigned PPCTargetLowering::combineRepeatedFPDivisors() const {
   // Note: This functionality is used only when unsafe-fp-math is enabled, and
   // on cores with reciprocal estimates (which are used when unsafe-fp-math is
   // enabled for division), this functionality is redundant with the default
@@ -9144,12 +9144,12 @@ bool PPCTargetLowering::combineRepeatedFPDivisors(unsigned NumUsers) const {
   // one FP pipeline) for three or more FDIVs (for generic OOO cores).
   switch (Subtarget.getDarwinDirective()) {
   default:
-    return NumUsers > 2;
+    return 3;
   case PPC::DIR_440:
   case PPC::DIR_A2:
   case PPC::DIR_E500mc:
   case PPC::DIR_E5500:
-    return NumUsers > 1;
+    return 2;
   }
 }
 

llvm/lib/Target/PowerPC/PPCISelLowering.h

@@ -853,7 +853,7 @@ namespace llvm {
                              bool &UseOneConstNR) const override;
     SDValue getRecipEstimate(SDValue Operand, DAGCombinerInfo &DCI,
                              unsigned &RefinementSteps) const override;
-    bool combineRepeatedFPDivisors(unsigned NumUsers) const override;
+    unsigned combineRepeatedFPDivisors() const override;
 
     CCAssignFn *useFastISelCCs(unsigned Flag) const;
   };

llvm/lib/Target/X86/X86ISelLowering.cpp

@@ -13308,8 +13308,8 @@ SDValue X86TargetLowering::getRecipEstimate(SDValue Op,
 /// This is because we still need one division to calculate the reciprocal and
 /// then we need two multiplies by that reciprocal as replacements for the
 /// original divisions.
-bool X86TargetLowering::combineRepeatedFPDivisors(unsigned NumUsers) const {
-  return NumUsers > 1;
+unsigned X86TargetLowering::combineRepeatedFPDivisors() const {
+  return 2;
 }
 
 static bool isAllOnes(SDValue V) {

llvm/lib/Target/X86/X86ISelLowering.h

@@ -1124,7 +1124,7 @@ namespace llvm {
                              unsigned &RefinementSteps) const override;
 
     /// Reassociate floating point divisions into multiply by reciprocal.
-    bool combineRepeatedFPDivisors(unsigned NumUsers) const override;
+    unsigned combineRepeatedFPDivisors() const override;
   };
 
   namespace X86 {