[LegalizeDAG][X86] Add support for turning STRICT_FADD/SUB/MUL/DIV into libcalls. Use it for fp128 on x86-64.

This requires a minor hack for f32/f64 strict fadd/fsub to avoid
turning those into libcalls.

lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

@@ -4019,6 +4019,7 @@ void SelectionDAGLegalize::ConvertNodeToLibcall(SDNode *Node) {
                        RTLIB::LLRINT_PPCF128, Results);
     break;
   case ISD::FDIV:
+  case ISD::STRICT_FDIV:
     ExpandFPLibCall(Node, RTLIB::DIV_F32, RTLIB::DIV_F64,
                     RTLIB::DIV_F80, RTLIB::DIV_F128,
                     RTLIB::DIV_PPCF128, Results);
@@ -4036,11 +4037,13 @@ void SelectionDAGLegalize::ConvertNodeToLibcall(SDNode *Node) {
                     RTLIB::FMA_PPCF128, Results);
     break;
   case ISD::FADD:
+  case ISD::STRICT_FADD:
     ExpandFPLibCall(Node, RTLIB::ADD_F32, RTLIB::ADD_F64,
                     RTLIB::ADD_F80, RTLIB::ADD_F128,
                     RTLIB::ADD_PPCF128, Results);
     break;
   case ISD::FMUL:
+  case ISD::STRICT_FMUL:
     ExpandFPLibCall(Node, RTLIB::MUL_F32, RTLIB::MUL_F64,
                     RTLIB::MUL_F80, RTLIB::MUL_F128,
                     RTLIB::MUL_PPCF128, Results);
@@ -4058,6 +4061,7 @@ void SelectionDAGLegalize::ConvertNodeToLibcall(SDNode *Node) {
     break;
   }
   case ISD::FSUB:
+  case ISD::STRICT_FSUB:
     ExpandFPLibCall(Node, RTLIB::SUB_F32, RTLIB::SUB_F64,
                     RTLIB::SUB_F80, RTLIB::SUB_F128,
                     RTLIB::SUB_PPCF128, Results);

lib/Target/X86/X86ISelLowering.cpp

@@ -611,6 +611,13 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
       addLegalFPImmediate(APFloat(+0.0)); // xorpd
   }
 
+  // FIXME: Mark these legal to prevent them from being expanded to a
+  // libcall in LegalizeDAG. They'll be mutated by X86ISelDAGToDAG::Select.
+  setOperationAction(ISD::STRICT_FADD, MVT::f32, Legal);
+  setOperationAction(ISD::STRICT_FADD, MVT::f64, Legal);
+  setOperationAction(ISD::STRICT_FSUB, MVT::f32, Legal);
+  setOperationAction(ISD::STRICT_FSUB, MVT::f64, Legal);
+
   // We don't support FMA.
   setOperationAction(ISD::FMA, MVT::f64, Expand);
   setOperationAction(ISD::FMA, MVT::f32, Expand);
@@ -659,11 +666,15 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
 
     addLegalFPImmediate(APFloat::getZero(APFloat::IEEEquad())); // xorps
 
-    setOperationAction(ISD::FADD, MVT::f128, LibCall);
+    setOperationAction(ISD::FADD,        MVT::f128, LibCall);
-    setOperationAction(ISD::FSUB, MVT::f128, LibCall);
+    setOperationAction(ISD::STRICT_FADD, MVT::f128, LibCall);
-    setOperationAction(ISD::FDIV, MVT::f128, LibCall);
+    setOperationAction(ISD::FSUB,        MVT::f128, LibCall);
-    setOperationAction(ISD::FMUL, MVT::f128, LibCall);
+    setOperationAction(ISD::STRICT_FSUB, MVT::f128, LibCall);
-    setOperationAction(ISD::FMA,  MVT::f128, Expand);
+    setOperationAction(ISD::FDIV,        MVT::f128, LibCall);
+    setOperationAction(ISD::STRICT_FDIV, MVT::f128, LibCall);
+    setOperationAction(ISD::FMUL,        MVT::f128, LibCall);
+    setOperationAction(ISD::STRICT_FMUL, MVT::f128, LibCall);
+    setOperationAction(ISD::FMA,         MVT::f128, Expand);
 
     setOperationAction(ISD::FABS, MVT::f128, Custom);
     setOperationAction(ISD::FNEG, MVT::f128, Custom);

test/CodeGen/X86/fp128-libcalls-strict.ll

@@ -0,0 +1,62 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -O2 -mtriple=x86_64-linux-android -mattr=+mmx \
+; RUN:     -enable-legalize-types-checking | FileCheck %s
+; RUN: llc < %s -O2 -mtriple=x86_64-linux-gnu -mattr=+mmx \
+; RUN:     -enable-legalize-types-checking | FileCheck %s
+
+; Check all soft floating point library function calls.
+
+define fp128 @add(fp128 %x, fp128 %y) nounwind strictfp {
+; CHECK-LABEL: add:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    callq __addtf3
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+entry:
+  %add = call fp128 @llvm.experimental.constrained.fadd.f128(fp128 %x, fp128 %y,  metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret fp128 %add
+}
+
+define fp128 @sub(fp128 %x, fp128 %y) nounwind strictfp {
+; CHECK-LABEL: sub:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    callq __subtf3
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+entry:
+  %sub = call fp128 @llvm.experimental.constrained.fsub.f128(fp128 %x, fp128 %y,  metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret fp128 %sub
+}
+
+define fp128 @mul(fp128 %x, fp128 %y) nounwind strictfp {
+; CHECK-LABEL: mul:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    callq __multf3
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+entry:
+  %mul = call fp128 @llvm.experimental.constrained.fmul.f128(fp128 %x, fp128 %y,  metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret fp128 %mul
+}
+
+define fp128 @div(fp128 %x, fp128 %y) nounwind strictfp {
+; CHECK-LABEL: div:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    callq __divtf3
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+entry:
+  %div = call fp128 @llvm.experimental.constrained.fdiv.f128(fp128 %x, fp128 %y,  metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret fp128 %div
+}
+
+attributes #0 = { strictfp }
+
+declare fp128 @llvm.experimental.constrained.fadd.f128(fp128, fp128, metadata, metadata)
+declare fp128 @llvm.experimental.constrained.fsub.f128(fp128, fp128, metadata, metadata)
+declare fp128 @llvm.experimental.constrained.fmul.f128(fp128, fp128, metadata, metadata)
+declare fp128 @llvm.experimental.constrained.fdiv.f128(fp128, fp128, metadata, metadata)