[X86] Don't rely on global contraction flag

[mikolaj-pirog]

As in title. See here for more context: https://discourse.llvm.org/t/allowfpopfusion-vs-sdnodeflags-hasallowcontract/80909

Testing has been updated in previous PRs, so no testing fail should be seen

[llvmbot]

@llvm/pr-subscribers-backend-x86

Author: Mikołaj Piróg (mikolaj-pirog)

Changes

As in title. See here for more context: https://discourse.llvm.org/t/allowfpopfusion-vs-sdnodeflags-hasallowcontract/80909

Testing has been updated in previous PRs, so no testing fail should be seen

---

Full diff: https://github.com/llvm/llvm-project/pull/167252.diff

1 Files Affected:

• (modified) llvm/lib/Target/X86/X86ISelLowering.cpp (+4-9)

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 05a854a0bf3fa..68dca58287c59 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -8434,7 +8434,6 @@ static bool isFMAddSubOrFMSubAdd(const X86Subtarget &Subtarget,
   // or MUL + ADDSUB to FMADDSUB.
   const TargetOptions &Options = DAG.getTarget().Options;
   bool AllowFusion =
-      Options.AllowFPOpFusion == FPOpFusion::Fast ||
       (AllowSubAddOrAddSubContract && Opnd0->getFlags().hasAllowContract());
   if (!AllowFusion)
     return false;
@@ -54160,11 +54159,7 @@ static SDValue combineFMulcFCMulc(SDNode *N, SelectionDAG &DAG,
 //  FADD(A, FMA(B, C, 0)) and FADD(A, FMUL(B, C)) to FMA(B, C, A)
 static SDValue combineFaddCFmul(SDNode *N, SelectionDAG &DAG,
                                 const X86Subtarget &Subtarget) {
-  auto AllowContract = [&DAG](const SDNodeFlags &Flags) {
-    return DAG.getTarget().Options.AllowFPOpFusion == FPOpFusion::Fast ||
-           Flags.hasAllowContract();
-  };
-
+  bool AllowContract = N->getFlags().hasAllowContract();
   auto HasNoSignedZero = [&DAG](const SDNodeFlags &Flags) {
     return DAG.getTarget().Options.NoSignedZerosFPMath ||
            Flags.hasNoSignedZeros();
@@ -54177,7 +54172,7 @@ static SDValue combineFaddCFmul(SDNode *N, SelectionDAG &DAG,
   };
 
   if (N->getOpcode() != ISD::FADD || !Subtarget.hasFP16() ||
-      !AllowContract(N->getFlags()))
+      !AllowContract)
     return SDValue();
 
   EVT VT = N->getValueType(0);
@@ -54188,14 +54183,14 @@ static SDValue combineFaddCFmul(SDNode *N, SelectionDAG &DAG,
   SDValue RHS = N->getOperand(1);
   bool IsConj;
   SDValue FAddOp1, MulOp0, MulOp1;
-  auto GetCFmulFrom = [&MulOp0, &MulOp1, &IsConj, &AllowContract,
+  auto GetCFmulFrom = [&MulOp0, &MulOp1, &IsConj, AllowContract,
                        &IsVectorAllNegativeZero,
                        &HasNoSignedZero](SDValue N) -> bool {
     if (!N.hasOneUse() || N.getOpcode() != ISD::BITCAST)
       return false;
     SDValue Op0 = N.getOperand(0);
     unsigned Opcode = Op0.getOpcode();
-    if (Op0.hasOneUse() && AllowContract(Op0->getFlags())) {
+    if (Op0.hasOneUse() && AllowContract) {
       if ((Opcode == X86ISD::VFMULC || Opcode == X86ISD::VFCMULC)) {
         MulOp0 = Op0.getOperand(0);
         MulOp1 = Op0.getOperand(1);

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[phoebewang]

Should we warn fp-contract is not used on X86?

[mikolaj-pirog]

Should we warn fp-contract is not used on X86?

It's not used by the X86 backend, but target independent code still uses it, so IR without contract on IR but with --fp-contract=fast on CLI may still emit FMA in the end

[phoebewang]

Should we warn fp-contract is not used on X86?

It's not used by the X86 backend, but target independent code still uses it, so IR without contract on IR but with --fp-contract=fast on CLI may still emit FMA in the end

I think that would be a behavioral change for those who used --fp-contract=fast before. So ideal, they should be reminded not to use it any more.

[phoebewang]

LGTM.