[llvm] r237079 - Migrate existing backends that care about software floating point

Eric Christopher echristo at gmail.com
Wed May 13 13:59:10 PDT 2015


Yep, next on my list. I hadn't just because I wanted to get the existing
format down in a way that we can deal with and see if there's a need for
additional command line options etc. The backends can adapt to the changed
and existing codegen, etc right now and I don't think there's any
additional need for command line options, or upgrading the IR to a new
format.

-eric

On Tue, May 12, 2015 at 11:04 AM Akira Hatanaka <ahatanak at gmail.com> wrote:

> Is clang going to stop emitting attribute "use-soft-float" now that
> "+soft-float" is embedded in the feature string?
>
> On Mon, May 11, 2015 at 6:26 PM, Eric Christopher <echristo at gmail.com>
> wrote:
>
>> Author: echristo
>> Date: Mon May 11 20:26:05 2015
>> New Revision: 237079
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=237079&view=rev
>> Log:
>> Migrate existing backends that care about software floating point
>> to use the information in the module rather than TargetOptions.
>>
>> We've had and clang has used the use-soft-float attribute for some
>> time now so have the backends set a subtarget feature based on
>> a particular function now that subtargets are created based on
>> functions and function attributes.
>>
>> For the one middle end soft float check go ahead and create
>> an overloadable TargetLowering::useSoftFloat function that
>> just checks the TargetSubtargetInfo in all cases.
>>
>> Also remove the command line option that hard codes whether or
>> not soft-float is set by using the attribute for all of the
>> target specific test cases - for the generic just go ahead and
>> add the attribute in the one case that showed up.
>>
>> Modified:
>>     llvm/trunk/include/llvm/CodeGen/CommandFlags.h
>>     llvm/trunk/include/llvm/Target/TargetLowering.h
>>     llvm/trunk/include/llvm/Target/TargetOptions.h
>>     llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
>>     llvm/trunk/lib/Target/ARM/ARM.td
>>     llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
>>     llvm/trunk/lib/Target/ARM/ARMISelLowering.h
>>     llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp
>>     llvm/trunk/lib/Target/ARM/ARMSubtarget.h
>>     llvm/trunk/lib/Target/ARM/ARMTargetMachine.cpp
>>     llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp
>>     llvm/trunk/lib/Target/Mips/MipsISelLowering.h
>>     llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp
>>     llvm/trunk/lib/Target/TargetMachine.cpp
>>     llvm/trunk/lib/Target/X86/X86.td
>>     llvm/trunk/lib/Target/X86/X86FastISel.cpp
>>     llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>>     llvm/trunk/lib/Target/X86/X86ISelLowering.h
>>     llvm/trunk/lib/Target/X86/X86Subtarget.cpp
>>     llvm/trunk/lib/Target/X86/X86Subtarget.h
>>     llvm/trunk/lib/Target/X86/X86TargetMachine.cpp
>>     llvm/trunk/test/CodeGen/ARM/vfp-libcalls.ll
>>     llvm/trunk/test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll
>>     llvm/trunk/test/CodeGen/X86/cvt16.ll
>>     llvm/trunk/test/CodeGen/X86/floor-soft-float.ll
>>     llvm/trunk/test/CodeGen/X86/soft-fp.ll
>>     llvm/trunk/tools/llc/llc.cpp
>>     llvm/trunk/tools/lli/lli.cpp
>>
>> Modified: llvm/trunk/include/llvm/CodeGen/CommandFlags.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/CommandFlags.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/include/llvm/CodeGen/CommandFlags.h (original)
>> +++ llvm/trunk/include/llvm/CodeGen/CommandFlags.h Mon May 11 20:26:05
>> 2015
>> @@ -126,11 +126,6 @@ EnableHonorSignDependentRoundingFPMath("
>>        cl::desc("Force codegen to assume rounding mode can change
>> dynamically"),
>>        cl::init(false));
>>
>> -cl::opt<bool>
>> -GenerateSoftFloatCalls("soft-float",
>> -                    cl::desc("Generate software floating point library
>> calls"),
>> -                    cl::init(false));
>> -
>>  cl::opt<llvm::FloatABI::ABIType>
>>  FloatABIForCalls("float-abi",
>>                   cl::desc("Choose float ABI type"),
>> @@ -241,7 +236,6 @@ static inline TargetOptions InitTargetOp
>>    Options.NoNaNsFPMath = EnableNoNaNsFPMath;
>>    Options.HonorSignDependentRoundingFPMathOption =
>>        EnableHonorSignDependentRoundingFPMath;
>> -  Options.UseSoftFloat = GenerateSoftFloatCalls;
>>    if (FloatABIForCalls != FloatABI::Default)
>>      Options.FloatABIType = FloatABIForCalls;
>>    Options.NoZerosInBSS = DontPlaceZerosInBSS;
>>
>> Modified: llvm/trunk/include/llvm/Target/TargetLowering.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetLowering.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/include/llvm/Target/TargetLowering.h (original)
>> +++ llvm/trunk/include/llvm/Target/TargetLowering.h Mon May 11 20:26:05
>> 2015
>> @@ -165,6 +165,7 @@ public:
>>
>>    bool isBigEndian() const { return !IsLittleEndian; }
>>    bool isLittleEndian() const { return IsLittleEndian; }
>> +  virtual bool useSoftFloat() const { return false; }
>>
>>    /// Return the pointer type for the given address space, defaults to
>>    /// the pointer type from the data layout.
>>
>> Modified: llvm/trunk/include/llvm/Target/TargetOptions.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetOptions.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/include/llvm/Target/TargetOptions.h (original)
>> +++ llvm/trunk/include/llvm/Target/TargetOptions.h Mon May 11 20:26:05
>> 2015
>> @@ -63,7 +63,7 @@ namespace llvm {
>>          : PrintMachineCode(false), NoFramePointerElim(false),
>>            LessPreciseFPMADOption(false), UnsafeFPMath(false),
>>            NoInfsFPMath(false), NoNaNsFPMath(false),
>> -          HonorSignDependentRoundingFPMathOption(false),
>> UseSoftFloat(false),
>> +          HonorSignDependentRoundingFPMathOption(false),
>>            NoZerosInBSS(false),
>>            GuaranteedTailCallOpt(false),
>>            DisableTailCalls(false), StackAlignmentOverride(0),
>> @@ -127,12 +127,6 @@ namespace llvm {
>>      unsigned HonorSignDependentRoundingFPMathOption : 1;
>>      bool HonorSignDependentRoundingFPMath() const;
>>
>> -    /// UseSoftFloat - This flag is enabled when the -soft-float flag is
>> -    /// specified on the command line.  When this flag is on, the code
>> generator
>> -    /// will generate libcalls to the software floating point library
>> instead of
>> -    /// target FP instructions.
>> -    unsigned UseSoftFloat : 1;
>> -
>>      /// NoZerosInBSS - By default some codegens place zero-initialized
>> data to
>>      /// .bss section. This flag disables such behaviour (necessary, e.g.
>> for
>>      /// crt*.o compiling).
>> @@ -240,7 +234,6 @@ inline bool operator==(const TargetOptio
>>      ARE_EQUAL(NoInfsFPMath) &&
>>      ARE_EQUAL(NoNaNsFPMath) &&
>>      ARE_EQUAL(HonorSignDependentRoundingFPMathOption) &&
>> -    ARE_EQUAL(UseSoftFloat) &&
>>      ARE_EQUAL(NoZerosInBSS) &&
>>      ARE_EQUAL(GuaranteedTailCallOpt) &&
>>      ARE_EQUAL(DisableTailCalls) &&
>>
>> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp (original)
>> +++ llvm/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp Mon May 11
>> 20:26:05 2015
>> @@ -3456,7 +3456,7 @@ void SelectionDAGLegalize::ExpandNode(SD
>>      break;
>>    }
>>    case ISD::FP_TO_FP16: {
>> -    if (!TM.Options.UseSoftFloat && TM.Options.UnsafeFPMath) {
>> +    if (!TLI.useSoftFloat() && TM.Options.UnsafeFPMath) {
>>        SDValue Op = Node->getOperand(0);
>>        MVT SVT = Op.getSimpleValueType();
>>        if ((SVT == MVT::f64 || SVT == MVT::f80) &&
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARM.td
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARM.td?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARM.td (original)
>> +++ llvm/trunk/lib/Target/ARM/ARM.td Mon May 11 20:26:05 2015
>> @@ -23,6 +23,9 @@ include "llvm/Target/Target.td"
>>  def ModeThumb  : SubtargetFeature<"thumb-mode", "InThumbMode", "true",
>>                                    "Thumb mode">;
>>
>> +def ModeSoftFloat : SubtargetFeature<"soft-float", "UseSoftFloat",
>> "true",
>> +                                     "Use software floating point
>> features.">;
>> +
>>
>>  //===----------------------------------------------------------------------===//
>>  // ARM Subtarget features.
>>  //
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp Mon May 11 20:26:05 2015
>> @@ -170,7 +170,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>    if (Subtarget->isTargetMachO()) {
>>      // Uses VFP for Thumb libfuncs if available.
>>      if (Subtarget->isThumb() && Subtarget->hasVFP2() &&
>> -        Subtarget->hasARMOps() && !TM.Options.UseSoftFloat) {
>> +        Subtarget->hasARMOps() && !Subtarget->useSoftFloat()) {
>>        // Single-precision floating-point arithmetic.
>>        setLibcallName(RTLIB::ADD_F32, "__addsf3vfp");
>>        setLibcallName(RTLIB::SUB_F32, "__subsf3vfp");
>> @@ -401,7 +401,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>      addRegisterClass(MVT::i32, &ARM::tGPRRegClass);
>>    else
>>      addRegisterClass(MVT::i32, &ARM::GPRRegClass);
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasVFP2() &&
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasVFP2() &&
>>        !Subtarget->isThumb1Only()) {
>>      addRegisterClass(MVT::f32, &ARM::SPRRegClass);
>>      addRegisterClass(MVT::f64, &ARM::DPRRegClass);
>> @@ -820,7 +820,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>    }
>>    setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i1, Expand);
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasVFP2() &&
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasVFP2() &&
>>        !Subtarget->isThumb1Only()) {
>>      // Turn f64->i64 into VMOVRRD, i64 -> f64 to VMOVDRR
>>      // iff target supports vfp2.
>> @@ -861,7 +861,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>    setOperationAction(ISD::FSINCOS,   MVT::f32, Expand);
>>    setOperationAction(ISD::FREM,      MVT::f64, Expand);
>>    setOperationAction(ISD::FREM,      MVT::f32, Expand);
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasVFP2() &&
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasVFP2() &&
>>        !Subtarget->isThumb1Only()) {
>>      setOperationAction(ISD::FCOPYSIGN, MVT::f64, Custom);
>>      setOperationAction(ISD::FCOPYSIGN, MVT::f32, Custom);
>> @@ -875,7 +875,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>    }
>>
>>    // Various VFP goodness
>> -  if (!TM.Options.UseSoftFloat && !Subtarget->isThumb1Only()) {
>> +  if (!Subtarget->useSoftFloat() && !Subtarget->isThumb1Only()) {
>>      // FP-ARMv8 adds f64 <-> f16 conversion. Before that it should be
>> expanded.
>>      if (!Subtarget->hasFPARMv8() || Subtarget->isFPOnlySP()) {
>>        setOperationAction(ISD::FP16_TO_FP, MVT::f64, Expand);
>> @@ -932,7 +932,7 @@ ARMTargetLowering::ARMTargetLowering(con
>>
>>    setStackPointerRegisterToSaveRestore(ARM::SP);
>>
>> -  if (TM.Options.UseSoftFloat || Subtarget->isThumb1Only() ||
>> +  if (Subtarget->useSoftFloat() || Subtarget->isThumb1Only() ||
>>        !Subtarget->hasVFP2())
>>      setSchedulingPreference(Sched::RegPressure);
>>    else
>> @@ -956,6 +956,10 @@ ARMTargetLowering::ARMTargetLowering(con
>>    setMinFunctionAlignment(Subtarget->isThumb() ? 1 : 2);
>>  }
>>
>> +bool ARMTargetLowering::useSoftFloat() const {
>> +  return Subtarget->useSoftFloat();
>> +}
>> +
>>  // FIXME: It might make sense to define the representative register
>> class as the
>>  // nearest super-register that has a non-null superset. For example,
>> DPR_VFP2 is
>>  // a super-register of SPR, and DPR is a superset if DPR_VFP2.
>> Consequently,
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMISelLowering.h (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMISelLowering.h Mon May 11 20:26:05 2015
>> @@ -231,6 +231,7 @@ namespace llvm {
>>                                 const ARMSubtarget &STI);
>>
>>      unsigned getJumpTableEncoding() const override;
>> +    bool useSoftFloat() const override;
>>
>>      SDValue LowerOperation(SDValue Op, SelectionDAG &DAG) const override;
>>
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp Mon May 11 20:26:05 2015
>> @@ -145,6 +145,7 @@ void ARMSubtarget::initializeEnvironment
>>    HasVMLxForwarding = false;
>>    SlowFPBrcc = false;
>>    InThumbMode = false;
>> +  UseSoftFloat = false;
>>    HasThumb2 = false;
>>    NoARM = false;
>>    IsR9Reserved = ReserveR9;
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMSubtarget.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMSubtarget.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMSubtarget.h (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMSubtarget.h Mon May 11 20:26:05 2015
>> @@ -100,6 +100,9 @@ protected:
>>    /// InThumbMode - True if compiling for Thumb, false for ARM.
>>    bool InThumbMode;
>>
>> +  /// UseSoftFloat - True if we're using software floating point
>> features.
>> +  bool UseSoftFloat;
>> +
>>    /// HasThumb2 - True if Thumb2 instructions are supported.
>>    bool HasThumb2;
>>
>> @@ -393,6 +396,7 @@ public:
>>    bool isAPCS_ABI() const;
>>    bool isAAPCS_ABI() const;
>>
>> +  bool useSoftFloat() const { return UseSoftFloat; }
>>    bool isThumb() const { return InThumbMode; }
>>    bool isThumb1Only() const { return InThumbMode && !HasThumb2; }
>>    bool isThumb2() const { return InThumbMode && HasThumb2; }
>>
>> Modified: llvm/trunk/lib/Target/ARM/ARMTargetMachine.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMTargetMachine.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/ARM/ARMTargetMachine.cpp (original)
>> +++ llvm/trunk/lib/Target/ARM/ARMTargetMachine.cpp Mon May 11 20:26:05
>> 2015
>> @@ -207,13 +207,15 @@ ARMBaseTargetMachine::getSubtargetImpl(c
>>    // function before we can generate a subtarget. We also need to use
>>    // it as a key for the subtarget since that can be the only difference
>>    // between two functions.
>> -  Attribute SFAttr = F.getFnAttribute("use-soft-float");
>> -  bool SoftFloat = !SFAttr.hasAttribute(Attribute::None)
>> -                       ? SFAttr.getValueAsString() == "true"
>> -                       : Options.UseSoftFloat;
>> +  bool SoftFloat =
>> +      F.hasFnAttribute("use-soft-float") &&
>> +      F.getFnAttribute("use-soft-float").getValueAsString() == "true";
>> +  // If the soft float attribute is set on the function turn on the soft
>> float
>> +  // subtarget feature.
>> +  if (SoftFloat)
>> +    FS += FS.empty() ? "+soft-float" : ",+soft-float";
>>
>> -  auto &I = SubtargetMap[CPU + FS + (SoftFloat ? "use-soft-float=true"
>> -                                               :
>> "use-soft-float=false")];
>> +  auto &I = SubtargetMap[CPU + FS];
>>    if (!I) {
>>      // This needs to be done before we create a new subtarget since any
>>      // creation will depend on the TM and the code generation flags on
>> the
>>
>> Modified: llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp Mon May 11 20:26:05
>> 2015
>> @@ -3598,6 +3598,10 @@ unsigned MipsTargetLowering::getJumpTabl
>>    return TargetLowering::getJumpTableEncoding();
>>  }
>>
>> +bool MipsTargetLowering::useSoftFloat() const {
>> +  return Subtarget.useSoftFloat();
>> +}
>> +
>>  void MipsTargetLowering::copyByValRegs(
>>      SDValue Chain, SDLoc DL, std::vector<SDValue> &OutChains,
>> SelectionDAG &DAG,
>>      const ISD::ArgFlagsTy &Flags, SmallVectorImpl<SDValue> &InVals,
>>
>> Modified: llvm/trunk/lib/Target/Mips/MipsISelLowering.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsISelLowering.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/Mips/MipsISelLowering.h (original)
>> +++ llvm/trunk/lib/Target/Mips/MipsISelLowering.h Mon May 11 20:26:05 2015
>> @@ -530,6 +530,7 @@ namespace llvm {
>>      bool isFPImmLegal(const APFloat &Imm, EVT VT) const override;
>>
>>      unsigned getJumpTableEncoding() const override;
>> +    bool useSoftFloat() const override;
>>
>>      /// Emit a sign-extension using sll/sra, seb, or seh appropriately.
>>      MachineBasicBlock *emitSignExtendToI32InReg(MachineInstr *MI,
>>
>> Modified: llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp (original)
>> +++ llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp Mon May 11 20:26:05
>> 2015
>> @@ -140,10 +140,9 @@ MipsTargetMachine::getSubtargetImpl(cons
>>    // FIXME: This is related to the code below to reset the target
>> options,
>>    // we need to know whether or not the soft float flag is set on the
>>    // function, so we can enable it as a subtarget feature.
>> -  Attribute SFAttr = F.getFnAttribute("use-soft-float");
>> -  bool softFloat = !SFAttr.hasAttribute(Attribute::None)
>> -                       ? SFAttr.getValueAsString() == "true"
>> -                       : Options.UseSoftFloat;
>> +  bool softFloat =
>> +      F.hasFnAttribute("use-soft-float") &&
>> +      F.getFnAttribute("use-soft-float").getValueAsString() == "true";
>>
>>    if (hasMips16Attr)
>>      FS += FS.empty() ? "+mips16" : ",+mips16";
>>
>> Modified: llvm/trunk/lib/Target/TargetMachine.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/TargetMachine.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/TargetMachine.cpp (original)
>> +++ llvm/trunk/lib/Target/TargetMachine.cpp Mon May 11 20:26:05 2015
>> @@ -71,7 +71,6 @@ void TargetMachine::resetTargetOptions(c
>>    RESET_OPTION(UnsafeFPMath, "unsafe-fp-math");
>>    RESET_OPTION(NoInfsFPMath, "no-infs-fp-math");
>>    RESET_OPTION(NoNaNsFPMath, "no-nans-fp-math");
>> -  RESET_OPTION(UseSoftFloat, "use-soft-float");
>>    RESET_OPTION(DisableTailCalls, "disable-tail-calls");
>>
>>    Options.MCOptions.SanitizeAddress =
>> F.hasFnAttribute(Attribute::SanitizeAddress);
>>
>> Modified: llvm/trunk/lib/Target/X86/X86.td
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86.td?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86.td (original)
>> +++ llvm/trunk/lib/Target/X86/X86.td Mon May 11 20:26:05 2015
>> @@ -192,6 +192,9 @@ def FeatureUseSqrtEst : SubtargetFeature
>>                              "Use RSQRT* to optimize square root
>> calculations">;
>>  def FeatureUseRecipEst : SubtargetFeature<"use-recip-est",
>> "UseReciprocalEst",
>>                            "true", "Use RCP* to optimize division
>> calculations">;
>> +def FeatureSoftFloat
>> +    : SubtargetFeature<"soft-float", "UseSoftFloat", "true",
>> +                       "Use software floating point features.">;
>>
>>
>>  //===----------------------------------------------------------------------===//
>>  // X86 processors supported.
>>
>> Modified: llvm/trunk/lib/Target/X86/X86FastISel.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86FastISel.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86FastISel.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86FastISel.cpp Mon May 11 20:26:05 2015
>> @@ -2254,7 +2254,7 @@ bool X86FastISel::fastLowerIntrinsicCall
>>    default: return false;
>>    case Intrinsic::convert_from_fp16:
>>    case Intrinsic::convert_to_fp16: {
>> -    if (TM.Options.UseSoftFloat || !Subtarget->hasF16C())
>> +    if (Subtarget->useSoftFloat() || !Subtarget->hasF16C())
>>        return false;
>>
>>      const Value *Op = II->getArgOperand(0);
>>
>> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Mon May 11 20:26:05 2015
>> @@ -183,7 +183,7 @@ X86TargetLowering::X86TargetLowering(con
>>    if (Subtarget->is64Bit()) {
>>      setOperationAction(ISD::UINT_TO_FP     , MVT::i32  , Promote);
>>      setOperationAction(ISD::UINT_TO_FP     , MVT::i64  , Custom);
>> -  } else if (!TM.Options.UseSoftFloat) {
>> +  } else if (!Subtarget->useSoftFloat()) {
>>      // We have an algorithm for SSE2->double, and we turn this into a
>>      // 64-bit FILD followed by conditional FADD for other targets.
>>      setOperationAction(ISD::UINT_TO_FP     , MVT::i64  , Custom);
>> @@ -197,7 +197,7 @@ X86TargetLowering::X86TargetLowering(con
>>    setOperationAction(ISD::SINT_TO_FP       , MVT::i1   , Promote);
>>    setOperationAction(ISD::SINT_TO_FP       , MVT::i8   , Promote);
>>
>> -  if (!TM.Options.UseSoftFloat) {
>> +  if (!Subtarget->useSoftFloat()) {
>>      // SSE has no i16 to fp conversion, only i32
>>      if (X86ScalarSSEf32) {
>>        setOperationAction(ISD::SINT_TO_FP     , MVT::i16  , Promote);
>> @@ -240,7 +240,7 @@ X86TargetLowering::X86TargetLowering(con
>>    if (Subtarget->is64Bit()) {
>>      setOperationAction(ISD::FP_TO_UINT     , MVT::i64  , Expand);
>>      setOperationAction(ISD::FP_TO_UINT     , MVT::i32  , Promote);
>> -  } else if (!TM.Options.UseSoftFloat) {
>> +  } else if (!Subtarget->useSoftFloat()) {
>>      // Since AVX is a superset of SSE3, only check for SSE here.
>>      if (Subtarget->hasSSE1() && !Subtarget->hasSSE3())
>>        // Expand FP_TO_UINT into a select.
>> @@ -368,7 +368,7 @@ X86TargetLowering::X86TargetLowering(con
>>    // Special handling for half-precision floating point conversions.
>>    // If we don't have F16C support, then lower half float conversions
>>    // into library calls.
>> -  if (TM.Options.UseSoftFloat || !Subtarget->hasF16C()) {
>> +  if (Subtarget->useSoftFloat() || !Subtarget->hasF16C()) {
>>      setOperationAction(ISD::FP16_TO_FP, MVT::f32, Expand);
>>      setOperationAction(ISD::FP_TO_FP16, MVT::f32, Expand);
>>    }
>> @@ -517,7 +517,7 @@ X86TargetLowering::X86TargetLowering(con
>>    setOperationAction(ISD::GC_TRANSITION_START, MVT::Other, Custom);
>>    setOperationAction(ISD::GC_TRANSITION_END, MVT::Other, Custom);
>>
>> -  if (!TM.Options.UseSoftFloat && X86ScalarSSEf64) {
>> +  if (!Subtarget->useSoftFloat() && X86ScalarSSEf64) {
>>      // f32 and f64 use SSE.
>>      // Set up the FP register classes.
>>      addRegisterClass(MVT::f32, &X86::FR32RegClass);
>> @@ -551,7 +551,7 @@ X86TargetLowering::X86TargetLowering(con
>>      // cases we handle.
>>      addLegalFPImmediate(APFloat(+0.0)); // xorpd
>>      addLegalFPImmediate(APFloat(+0.0f)); // xorps
>> -  } else if (!TM.Options.UseSoftFloat && X86ScalarSSEf32) {
>> +  } else if (!Subtarget->useSoftFloat() && X86ScalarSSEf32) {
>>      // Use SSE for f32, x87 for f64.
>>      // Set up the FP register classes.
>>      addRegisterClass(MVT::f32, &X86::FR32RegClass);
>> @@ -586,7 +586,7 @@ X86TargetLowering::X86TargetLowering(con
>>        setOperationAction(ISD::FCOS   , MVT::f64, Expand);
>>        setOperationAction(ISD::FSINCOS, MVT::f64, Expand);
>>      }
>> -  } else if (!TM.Options.UseSoftFloat) {
>> +  } else if (!Subtarget->useSoftFloat()) {
>>      // f32 and f64 in x87.
>>      // Set up the FP register classes.
>>      addRegisterClass(MVT::f64, &X86::RFP64RegClass);
>> @@ -620,7 +620,7 @@ X86TargetLowering::X86TargetLowering(con
>>    setOperationAction(ISD::FMA, MVT::f32, Expand);
>>
>>    // Long double always uses X87.
>> -  if (!TM.Options.UseSoftFloat) {
>> +  if (!Subtarget->useSoftFloat()) {
>>      addRegisterClass(MVT::f80, &X86::RFP80RegClass);
>>      setOperationAction(ISD::UNDEF,     MVT::f80, Expand);
>>      setOperationAction(ISD::FCOPYSIGN, MVT::f80, Expand);
>> @@ -760,7 +760,7 @@ X86TargetLowering::X86TargetLowering(con
>>
>>    // FIXME: In order to prevent SSE instructions being expanded to MMX
>> ones
>>    // with -msoft-float, disable use of MMX as well.
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasMMX()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasMMX()) {
>>      addRegisterClass(MVT::x86mmx, &X86::VR64RegClass);
>>      // No operations on x86mmx supported, everything uses intrinsics.
>>    }
>> @@ -778,7 +778,7 @@ X86TargetLowering::X86TargetLowering(con
>>    }
>>    setOperationAction(ISD::INSERT_VECTOR_ELT,  MVT::v1i64, Expand);
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasSSE1()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasSSE1()) {
>>      addRegisterClass(MVT::v4f32, &X86::VR128RegClass);
>>
>>      setOperationAction(ISD::FADD,               MVT::v4f32, Legal);
>> @@ -797,7 +797,7 @@ X86TargetLowering::X86TargetLowering(con
>>      setOperationAction(ISD::UINT_TO_FP,         MVT::v4i32, Custom);
>>    }
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasSSE2()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasSSE2()) {
>>      addRegisterClass(MVT::v2f64, &X86::VR128RegClass);
>>
>>      // FIXME: Unfortunately, -soft-float and -no-implicit-float mean XMM
>> @@ -942,7 +942,7 @@ X86TargetLowering::X86TargetLowering(con
>>      setOperationAction(ISD::BITCAST,            MVT::v8i8,  Custom);
>>    }
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasSSE41()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasSSE41()) {
>>      for (MVT RoundedTy : {MVT::f32, MVT::f64, MVT::v4f32, MVT::v2f64}) {
>>        setOperationAction(ISD::FFLOOR,           RoundedTy,  Legal);
>>        setOperationAction(ISD::FCEIL,            RoundedTy,  Legal);
>> @@ -1024,7 +1024,7 @@ X86TargetLowering::X86TargetLowering(con
>>      setOperationAction(ISD::SRA,               MVT::v4i32, Custom);
>>    }
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasFp256()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasFp256()) {
>>      addRegisterClass(MVT::v32i8,  &X86::VR256RegClass);
>>      addRegisterClass(MVT::v16i16, &X86::VR256RegClass);
>>      addRegisterClass(MVT::v8i32,  &X86::VR256RegClass);
>> @@ -1244,7 +1244,7 @@ X86TargetLowering::X86TargetLowering(con
>>      }
>>    }
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasAVX512()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasAVX512()) {
>>      addRegisterClass(MVT::v16i32, &X86::VR512RegClass);
>>      addRegisterClass(MVT::v16f32, &X86::VR512RegClass);
>>      addRegisterClass(MVT::v8i64,  &X86::VR512RegClass);
>> @@ -1447,7 +1447,7 @@ X86TargetLowering::X86TargetLowering(con
>>      }
>>    }// has  AVX-512
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasBWI()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasBWI()) {
>>      addRegisterClass(MVT::v32i16, &X86::VR512RegClass);
>>      addRegisterClass(MVT::v64i8,  &X86::VR512RegClass);
>>
>> @@ -1484,7 +1484,7 @@ X86TargetLowering::X86TargetLowering(con
>>      }
>>    }
>>
>> -  if (!TM.Options.UseSoftFloat && Subtarget->hasVLX()) {
>> +  if (!Subtarget->useSoftFloat() && Subtarget->hasVLX()) {
>>      addRegisterClass(MVT::v4i1,   &X86::VK4RegClass);
>>      addRegisterClass(MVT::v2i1,   &X86::VK2RegClass);
>>
>> @@ -1789,6 +1789,10 @@ unsigned X86TargetLowering::getJumpTable
>>    return TargetLowering::getJumpTableEncoding();
>>  }
>>
>> +bool X86TargetLowering::useSoftFloat() const {
>> +  return Subtarget->useSoftFloat();
>> +}
>> +
>>  const MCExpr *
>>  X86TargetLowering::LowerCustomJumpTableEntry(const MachineJumpTableInfo
>> *MJTI,
>>                                               const MachineBasicBlock
>> *MBB,
>> @@ -2294,7 +2298,7 @@ static ArrayRef<MCPhysReg> get64BitArgum
>>
>>    const Function *Fn = MF.getFunction();
>>    bool NoImplicitFloatOps =
>> Fn->hasFnAttribute(Attribute::NoImplicitFloat);
>> -  bool isSoftFloat = MF.getTarget().Options.UseSoftFloat;
>> +  bool isSoftFloat = Subtarget->useSoftFloat();
>>    assert(!(isSoftFloat && NoImplicitFloatOps) &&
>>           "SSE register cannot be used when SSE is disabled!");
>>    if (isSoftFloat || NoImplicitFloatOps || !Subtarget->hasSSE1())
>> @@ -2468,7 +2472,7 @@ X86TargetLowering::LowerFormalArguments(
>>    bool IsWinEHParent = WinEHParent && WinEHParent == Fn;
>>
>>    // Figure out if XMM registers are in use.
>> -  assert(!(MF.getTarget().Options.UseSoftFloat &&
>> +  assert(!(Subtarget->useSoftFloat() &&
>>             Fn->hasFnAttribute(Attribute::NoImplicitFloat)) &&
>>           "SSE register cannot be used when SSE is disabled!");
>>
>> @@ -14600,7 +14604,7 @@ SDValue X86TargetLowering::LowerVAARG(SD
>>
>>    if (ArgMode == 2) {
>>      // Sanity Check: Make sure using fp_offset makes sense.
>> -    assert(!DAG.getTarget().Options.UseSoftFloat &&
>> +    assert(!Subtarget->useSoftFloat() &&
>>             !(DAG.getMachineFunction().getFunction()->hasFnAttribute(
>>                 Attribute::NoImplicitFloat)) &&
>>             Subtarget->hasSSE1());
>> @@ -23203,8 +23207,8 @@ static SDValue PerformSTORECombine(SDNod
>>
>>    const Function *F = DAG.getMachineFunction().getFunction();
>>    bool NoImplicitFloatOps =
>> F->hasFnAttribute(Attribute::NoImplicitFloat);
>> -  bool F64IsLegal = !DAG.getTarget().Options.UseSoftFloat &&
>> !NoImplicitFloatOps
>> -                     && Subtarget->hasSSE2();
>> +  bool F64IsLegal =
>> +      !Subtarget->useSoftFloat() && !NoImplicitFloatOps &&
>> Subtarget->hasSSE2();
>>    if ((VT.isVector() ||
>>         (VT == MVT::i64 && F64IsLegal && !Subtarget->is64Bit())) &&
>>        isa<LoadSDNode>(St->getValue()) &&
>>
>> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86ISelLowering.h (original)
>> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.h Mon May 11 20:26:05 2015
>> @@ -572,6 +572,7 @@ namespace llvm {
>>                                 const X86Subtarget &STI);
>>
>>      unsigned getJumpTableEncoding() const override;
>> +    bool useSoftFloat() const override;
>>
>>      MVT getScalarShiftAmountTy(EVT LHSTy) const override { return
>> MVT::i8; }
>>
>>
>> Modified: llvm/trunk/lib/Target/X86/X86Subtarget.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Subtarget.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86Subtarget.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86Subtarget.cpp Mon May 11 20:26:05 2015
>> @@ -278,6 +278,7 @@ void X86Subtarget::initializeEnvironment
>>    stackAlignment = 4;
>>    // FIXME: this is a known good value for Yonah. How about others?
>>    MaxInlineSizeThreshold = 128;
>> +  UseSoftFloat = false;
>>  }
>>
>>  X86Subtarget &X86Subtarget::initializeSubtargetDependencies(StringRef
>> CPU,
>>
>> Modified: llvm/trunk/lib/Target/X86/X86Subtarget.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Subtarget.h?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86Subtarget.h (original)
>> +++ llvm/trunk/lib/Target/X86/X86Subtarget.h Mon May 11 20:26:05 2015
>> @@ -218,6 +218,9 @@ protected:
>>    /// Processor has AVX-512 Vector Length eXtenstions
>>    bool HasVLX;
>>
>> +  /// Use software floating point for code generation.
>> +  bool UseSoftFloat;
>> +
>>    /// The minimum alignment known to hold of the stack frame on
>>    /// entry to the function and which must be maintained by every
>> function.
>>    unsigned stackAlignment;
>> @@ -385,6 +388,7 @@ public:
>>
>>    bool isAtom() const { return X86ProcFamily == IntelAtom; }
>>    bool isSLM() const { return X86ProcFamily == IntelSLM; }
>> +  bool useSoftFloat() const { return UseSoftFloat; }
>>
>>    const Triple &getTargetTriple() const { return TargetTriple; }
>>
>>
>> Modified: llvm/trunk/lib/Target/X86/X86TargetMachine.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86TargetMachine.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86TargetMachine.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86TargetMachine.cpp Mon May 11 20:26:05
>> 2015
>> @@ -131,13 +131,15 @@ X86TargetMachine::getSubtargetImpl(const
>>    // function before we can generate a subtarget. We also need to use
>>    // it as a key for the subtarget since that can be the only difference
>>    // between two functions.
>> -  Attribute SFAttr = F.getFnAttribute("use-soft-float");
>> -  bool SoftFloat = !SFAttr.hasAttribute(Attribute::None)
>> -                       ? SFAttr.getValueAsString() == "true"
>> -                       : Options.UseSoftFloat;
>> +  bool SoftFloat =
>> +      F.hasFnAttribute("use-soft-float") &&
>> +      F.getFnAttribute("use-soft-float").getValueAsString() == "true";
>> +  // If the soft float attribute is set on the function turn on the soft
>> float
>> +  // subtarget feature.
>> +  if (SoftFloat)
>> +    FS += FS.empty() ? "+soft-float" : ",+soft-float";
>>
>> -  auto &I = SubtargetMap[CPU + FS + (SoftFloat ? "use-soft-float=true"
>> -                                               :
>> "use-soft-float=false")];
>> +  auto &I = SubtargetMap[CPU + FS];
>>    if (!I) {
>>      // This needs to be done before we create a new subtarget since any
>>      // creation will depend on the TM and the code generation flags on
>> the
>>
>> Modified: llvm/trunk/test/CodeGen/ARM/vfp-libcalls.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/vfp-libcalls.ll?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/ARM/vfp-libcalls.ll (original)
>> +++ llvm/trunk/test/CodeGen/ARM/vfp-libcalls.ll Mon May 11 20:26:05 2015
>> @@ -1,6 +1,6 @@
>>  ; RUN: llc -mtriple=armv6-apple-ios -mcpu=arm1136jf-s -o - %s |
>> FileCheck %s --check-prefix=CHECK-HARD
>>  ; RUN: llc -mtriple=thumbv6-apple-ios -mcpu=arm1136jf-s -o - %s |
>> FileCheck %s --check-prefix=CHECK-SOFTISH
>> -; RUN: llc -mtriple=armv7s-apple-ios -soft-float -mcpu=arm1136jf-s -o -
>> %s | FileCheck %s --check-prefix=CHECK-SOFT
>> +; RUN: llc -mtriple=armv7s-apple-ios -mattr=+soft-float
>> -mcpu=arm1136jf-s -o - %s | FileCheck %s --check-prefix=CHECK-SOFT
>>
>>  define float @test_call(float %a, float %b) {
>>  ; CHECK-HARD: vadd.f32 {{s[0-9]+}}, {{s[0-9]+}}, {{s[0-9]+}}
>> @@ -8,4 +8,4 @@ define float @test_call(float %a, float
>>  ; CHECK-SOFT: bl ___addsf3{{$}}
>>    %sum = fadd float %a, %b
>>    ret float %sum
>> -}
>> \ No newline at end of file
>> +}
>>
>> Modified:
>> llvm/trunk/test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll
>> (original)
>> +++ llvm/trunk/test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll
>> Mon May 11 20:26:05 2015
>> @@ -1,10 +1,14 @@
>> -; RUN: llc < %s -soft-float
>> +; RUN: llc < %s
>>  ; PR3899
>>
>>  @m = external global <2 x double>
>>
>> -define double @vector_ex() nounwind {
>> +define double @vector_ex() nounwind #0 {
>>         %v = load <2 x double>, <2 x double>* @m
>>         %x = extractelement <2 x double> %v, i32 1
>>         ret double %x
>>  }
>> +
>> +; Soft-float attribute so that targets that pay attention to soft float
>> will
>> +; make floating point types illegal and we'll exercise the legalizer
>> code.
>> +attributes #0 = { "use-soft-float" = "true" }
>>
>> Modified: llvm/trunk/test/CodeGen/X86/cvt16.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cvt16.ll?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/X86/cvt16.ll (original)
>> +++ llvm/trunk/test/CodeGen/X86/cvt16.ll Mon May 11 20:26:05 2015
>> @@ -1,7 +1,7 @@
>>  ; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -mattr=-f16c | FileCheck %s -check-prefix=CHECK
>> -check-prefix=LIBCALL
>>  ; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -mattr=+f16c | FileCheck %s -check-prefix=CHECK
>> -check-prefix=F16C
>> -; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -soft-float=1 -mattr=-f16c | FileCheck %s -check-prefix=CHECK
>> -check-prefix=SOFTFLOAT
>> -; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -soft-float=1 -mattr=+f16c | FileCheck %s -check-prefix=CHECK
>> -check-prefix=SOFTFLOAT
>> +; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -mattr=-f16c,+soft-float | FileCheck %s -check-prefix=CHECK
>> -check-prefix=SOFTFLOAT
>> +; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu
>> -mcpu=corei7 -mattr=+f16c,+soft-float | FileCheck %s -check-prefix=CHECK
>> -check-prefix=SOFTFLOAT
>>
>>  ; This is a test for float to half float conversions on x86-64.
>>  ;
>>
>> Modified: llvm/trunk/test/CodeGen/X86/floor-soft-float.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/floor-soft-float.ll?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/X86/floor-soft-float.ll (original)
>> +++ llvm/trunk/test/CodeGen/X86/floor-soft-float.ll Mon May 11 20:26:05
>> 2015
>> @@ -1,5 +1,5 @@
>> -; RUN: llc < %s -march=x86-64 -mattr=+sse4.1,-avx -soft-float=0 |
>> FileCheck %s --check-prefix=CHECK-HARD-FLOAT
>> -; RUN: llc < %s -march=x86-64 -mattr=+sse4.1,-avx -soft-float=1 |
>> FileCheck %s --check-prefix=CHECK-SOFT-FLOAT
>> +; RUN: llc < %s -march=x86-64 -mattr=+sse4.1,-avx | FileCheck %s
>> --check-prefix=CHECK-HARD-FLOAT
>> +; RUN: llc < %s -march=x86-64 -mattr=+sse4.1,-avx,+soft-float |
>> FileCheck %s --check-prefix=CHECK-SOFT-FLOAT
>>
>>  target triple = "x86_64-unknown-linux-gnu"
>>
>>
>> Modified: llvm/trunk/test/CodeGen/X86/soft-fp.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/soft-fp.ll?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/X86/soft-fp.ll (original)
>> +++ llvm/trunk/test/CodeGen/X86/soft-fp.ll Mon May 11 20:26:05 2015
>> @@ -1,5 +1,5 @@
>> -; RUN: llc < %s -march=x86    -mattr=+sse2 -soft-float | FileCheck %s
>> -; RUN: llc < %s -march=x86-64 -mattr=+sse2 -soft-float | FileCheck %s
>> +; RUN: llc < %s -march=x86    -mattr=+sse2,+soft-float | FileCheck %s
>> +; RUN: llc < %s -march=x86-64 -mattr=+sse2,+soft-float | FileCheck %s
>>
>>  ; CHECK-NOT: xmm{[0-9]+}
>>
>>
>> Modified: llvm/trunk/tools/llc/llc.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/llc/llc.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/tools/llc/llc.cpp (original)
>> +++ llvm/trunk/tools/llc/llc.cpp Mon May 11 20:26:05 2015
>> @@ -281,9 +281,8 @@ static int compileModule(char **argv, LL
>>      return 0;
>>
>>    assert(M && "Should have exited if we didn't have a module!");
>> -
>> -  if (GenerateSoftFloatCalls)
>> -    FloatABIForCalls = FloatABI::Soft;
>> +  if (FloatABIForCalls != FloatABI::Default)
>> +    Options.FloatABIType = FloatABIForCalls;
>>
>>    // Figure out where we are going to send the output.
>>    std::unique_ptr<tool_output_file> Out =
>>
>> Modified: llvm/trunk/tools/lli/lli.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/lli/lli.cpp?rev=237079&r1=237078&r2=237079&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/tools/lli/lli.cpp (original)
>> +++ llvm/trunk/tools/lli/lli.cpp Mon May 11 20:26:05 2015
>> @@ -465,11 +465,8 @@ int main(int argc, char **argv, char * c
>>    builder.setOptLevel(OLvl);
>>
>>    TargetOptions Options;
>> -  Options.UseSoftFloat = GenerateSoftFloatCalls;
>>    if (FloatABIForCalls != FloatABI::Default)
>>      Options.FloatABIType = FloatABIForCalls;
>> -  if (GenerateSoftFloatCalls)
>> -    FloatABIForCalls = FloatABI::Soft;
>>
>>    builder.setTargetOptions(Options);
>>
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150513/d6866ebc/attachment.html>


More information about the llvm-commits mailing list