Utilize new SDNode flag functionality to expand current support for fsub

[mcberg2017]

Summary

This patch originated from D46562 and is a proper subset, with some issues addressed for fsub.

[spatel]

LGTM.

lib/CodeGen/SelectionDAG/DAGCombiner.cpp

10411 ↗

(On Diff #150412)

Extra space.

llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

static char isNegatibleForFree(SDValue Op, bool LegalOperations,
                                const TargetLowering &TLI,
                                const TargetOptions *Options,
                                unsigned Depth = 0) {
   // fneg is removable even if it has multiple uses.
   if (Op.getOpcode() == ISD::FNEG) return 2;
 
   // Don't allow anything with multiple uses unless we know it is free.
   EVT VT = Op.getValueType();
+  const SDNodeFlags Flags = Op->getFlags();
   if (!Op.hasOneUse())
     if (!(Op.getOpcode() == ISD::FP_EXTEND &&
           TLI.isFPExtFree(VT, Op.getOperand(0).getValueType())))
       return 0;
 
   // Don't recurse exponentially.
   if (Depth > 6) return 0;
 
if (char V = isNegatibleForFree(Op.getOperand(0), LegalOperations, TLI,
                                     Options, Depth + 1))
       return V;
     // fold (fneg (fadd A, B)) -> (fsub (fneg B), A)
     return isNegatibleForFree(Op.getOperand(1), LegalOperations, TLI, Options,
                               Depth + 1);
   case ISD::FSUB:
     // We can't turn -(A-B) into B-A when we honor signed zeros.
     if (!Options->NoSignedZerosFPMath &&
-        !Op.getNode()->getFlags().hasNoSignedZeros())
+        !Flags.hasNoSignedZeros())
       return 0;
 
     // fold (fneg (fsub A, B)) -> (fsub B, A)
     return 1;
 
   case ISD::FMUL:
   case ISD::FDIV:
     if (Options->HonorSignDependentRoundingFPMath()) return 0;
SDValue DAGCombiner::visitFSUB(SDNode *N) {
 
   // fold (fsub c1, c2) -> c1-c2
   if (N0CFP && N1CFP)
     return DAG.getNode(ISD::FSUB, DL, VT, N0, N1, Flags);
 
   if (SDValue NewSel = foldBinOpIntoSelect(N))
     return NewSel;
 
-  // fold (fsub A, (fneg B)) -> (fadd A, B)
-  if (isNegatibleForFree(N1, LegalOperations, TLI, &Options))
-    return DAG.getNode(ISD::FADD, DL, VT, N0,
-                       GetNegatedExpression(N1, DAG, LegalOperations), Flags);
+  // (fsub A, 0) -> A
+  if (N1CFP && N1CFP->isZero()) {
+    if (!N1CFP->isNegative() || Options.UnsafeFPMath ||
+        Flags.hasNoSignedZeros()) {
+      return N0;
+    }
+  }
+
+  if (N0 == N1) {
+    // (fsub x, x) -> 0.0
+    if (Options.UnsafeFPMath || Flags.hasNoNaNs())
+      return DAG.getConstantFP(0.0f, DL, VT);
+  }
 
-  // FIXME: Auto-upgrade the target/function-level option.
-  if (Options.NoSignedZerosFPMath  || N->getFlags().hasNoSignedZeros()) {
-    // (fsub 0, B) -> -B
-    if (N0CFP && N0CFP->isZero()) {
+    if (Options.NoSignedZerosFPMath || Flags.hasNoSignedZeros()) {
       if (isNegatibleForFree(N1, LegalOperations, TLI, &Options))
         return GetNegatedExpression(N1, DAG, LegalOperations);
       if (!LegalOperations || TLI.isOperationLegal(ISD::FNEG, VT))
         return DAG.getNode(ISD::FNEG, DL, VT, N1, Flags);
     }
   }
 
+  // fold (fsub A, (fneg B)) -> (fadd A, B)
+  if (isNegatibleForFree(N1, LegalOperations, TLI, &Options))
+    return DAG.getNode(ISD::FADD, DL, VT, N0,
+                       GetNegatedExpression(N1, DAG, LegalOperations), Flags);
+
   // If 'unsafe math' is enabled, fold lots of things.
   if (Options.UnsafeFPMath) {
-    // (fsub A, 0) -> A
-    if (N1CFP && N1CFP->isZero())
-      return N0;
-
-    // (fsub x, x) -> 0.0
-    if (N0 == N1)
-      return DAG.getConstantFP(0.0f, DL, VT);
-
     // (fsub x, (fadd x, y)) -> (fneg y)
     // (fsub x, (fadd y, x)) -> (fneg y)
     if (N1.getOpcode() == ISD::FADD) {
       SDValue N10 = N1->getOperand(0);
       SDValue N11 = N1->getOperand(1);
 
       if (N10 == N0 && isNegatibleForFree(N11, LegalOperations, TLI, &Options))
         return GetNegatedExpression(N11, DAG, LegalOperations);

llvm/trunk/test/CodeGen/X86/fp-fold.ll

 ;
 ; UNSAFE-LABEL: fadd_negzero:
 ; UNSAFE:       # %bb.0:
 ; UNSAFE-NEXT:    retq
   %r = fadd float %x, -0.0
   ret float %r
 }
 
+define float @fadd_negzero_nsz(float %x) {
+; ANY-LABEL: fadd_negzero_nsz:
+; ANY:       # %bb.0:
+; ANY-NEXT:    retq
+  %r = fadd nsz float %x, -0.0
+  ret float %r
+}
+
 define float @fadd_zero_nsz(float %x) {
 ; ANY-LABEL: fadd_zero_nsz:
 ; ANY:       # %bb.0:
 ; ANY-NEXT:    retq
   %r = fadd nsz float %x, 0.0
   ret float %r
 }
 
-define float @fadd_negzero_nsz(float %x) {
-; ANY-LABEL: fadd_negzero_nsz:
+define float @fsub_zero(float %x) {
+; ANY-LABEL: fsub_zero:
 ; ANY:       # %bb.0:
 ; ANY-NEXT:    retq
-  %r = fadd nsz float %x, -0.0
+  %r = fsub float %x, 0.0
   ret float %r
 }
 
-define float @fsub_zero(float %x) {
-; STRICT-LABEL: fsub_zero:
-; STRICT:       # %bb.0:
-; STRICT-NEXT:    addss {{.*}}(%rip), %xmm0
-; STRICT-NEXT:    retq
-;
-; UNSAFE-LABEL: fsub_zero:
-; UNSAFE:       # %bb.0:
-; UNSAFE-NEXT:    retq
-  %r = fsub float %x, 0.0
+define float @fsub_self(float %x) {
+; ANY-LABEL: fsub_self:
+; ANY:       # %bb.0:
+; ANY-NEXT:    xorps %xmm0, %xmm0
+; ANY-NEXT:    retq
+  %r = fsub nnan float %x, %x 
   ret float %r
 }
 
 define float @fsub_negzero(float %x) {
 ; STRICT-LABEL: fsub_negzero:
 ; STRICT:       # %bb.0:
 ; STRICT-NEXT:    xorps %xmm1, %xmm1
 ; STRICT-NEXT:    addss %xmm1, %xmm0
 ; STRICT-NEXT:    retq
 ;
 ; UNSAFE-LABEL: fsub_negzero:
 ; UNSAFE:       # %bb.0:
 ; UNSAFE-NEXT:    retq
   %r = fsub float %x, -0.0
   ret float %r
 }
 
-define float @fsub_zero_nsz(float %x) {
-; ANY-LABEL: fsub_zero_nsz:
+define float @fsub_zero_nsz_1(float %x) {
+; ANY-LABEL: fsub_zero_nsz_1:
 ; ANY:       # %bb.0:
 ; ANY-NEXT:    retq
   %r = fsub nsz float %x, 0.0
   ret float %r
 }
 
+define float @fsub_zero_nsz_2(float %x) {
+; ANY-LABEL: fsub_zero_nsz_2:
+; ANY:       # %bb.0:
+; ANY-NEXT:    xorps {{.*}}(%rip), %xmm0
+; ANY-NEXT:    retq
+  %r = fsub nsz float 0.0, %x
+  ret float %r
+}
+
 define float @fsub_negzero_nsz(float %x) {
 ; ANY-LABEL: fsub_negzero_nsz:
 ; ANY:       # %bb.0:
 ; ANY-NEXT:    retq
   %r = fsub nsz float %x, -0.0
   ret float %r
 }