From 2c3e0051c31c3f5b2328b447eadf1cf9c4427442 Mon Sep 17 00:00:00 2001 From: Pirama Arumuga Nainar Date: Wed, 6 May 2015 11:46:36 -0700 Subject: Update aosp/master LLVM for rebase to r235153 Change-Id: I9bf53792f9fc30570e81a8d80d296c681d005ea7 (cherry picked from commit 0c7f116bb6950ef819323d855415b2f2b0aad987) --- lib/Target/SystemZ/CMakeLists.txt | 1 + .../SystemZ/Disassembler/SystemZDisassembler.cpp | 2 +- .../SystemZ/InstPrinter/SystemZInstPrinter.cpp | 3 +- .../SystemZ/InstPrinter/SystemZInstPrinter.h | 3 +- lib/Target/SystemZ/LLVMBuild.txt | 2 +- .../SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp | 2 +- .../SystemZ/MCTargetDesc/SystemZMCCodeEmitter.cpp | 2 +- .../SystemZ/MCTargetDesc/SystemZMCObjectWriter.cpp | 4 +- .../SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp | 5 +- .../SystemZ/MCTargetDesc/SystemZMCTargetDesc.h | 3 +- lib/Target/SystemZ/SystemZ.h | 12 + lib/Target/SystemZ/SystemZAsmPrinter.cpp | 23 -- lib/Target/SystemZ/SystemZAsmPrinter.h | 1 - lib/Target/SystemZ/SystemZISelDAGToDAG.cpp | 11 +- lib/Target/SystemZ/SystemZISelLowering.cpp | 250 ++++++++++++++++++++- lib/Target/SystemZ/SystemZISelLowering.h | 20 ++ lib/Target/SystemZ/SystemZInstrFormats.td | 11 + lib/Target/SystemZ/SystemZInstrInfo.cpp | 34 +-- lib/Target/SystemZ/SystemZInstrInfo.td | 65 ++++++ lib/Target/SystemZ/SystemZOperators.td | 13 ++ lib/Target/SystemZ/SystemZProcessors.td | 30 ++- lib/Target/SystemZ/SystemZSubtarget.cpp | 9 +- lib/Target/SystemZ/SystemZSubtarget.h | 18 ++ lib/Target/SystemZ/SystemZTargetMachine.cpp | 7 + lib/Target/SystemZ/SystemZTargetMachine.h | 1 + lib/Target/SystemZ/SystemZTargetTransformInfo.cpp | 240 ++++++++++++++++++++ lib/Target/SystemZ/SystemZTargetTransformInfo.h | 70 ++++++ 27 files changed, 778 insertions(+), 64 deletions(-) create mode 100644 lib/Target/SystemZ/SystemZTargetTransformInfo.cpp create mode 100644 lib/Target/SystemZ/SystemZTargetTransformInfo.h (limited to 'lib/Target/SystemZ') diff --git a/lib/Target/SystemZ/CMakeLists.txt b/lib/Target/SystemZ/CMakeLists.txt index 60a3912..336f037 100644 --- a/lib/Target/SystemZ/CMakeLists.txt +++ b/lib/Target/SystemZ/CMakeLists.txt @@ -29,6 +29,7 @@ add_llvm_target(SystemZCodeGen SystemZShortenInst.cpp SystemZSubtarget.cpp SystemZTargetMachine.cpp + SystemZTargetTransformInfo.cpp ) add_subdirectory(AsmParser) diff --git a/lib/Target/SystemZ/Disassembler/SystemZDisassembler.cpp b/lib/Target/SystemZ/Disassembler/SystemZDisassembler.cpp index 23173bf..84400f8 100644 --- a/lib/Target/SystemZ/Disassembler/SystemZDisassembler.cpp +++ b/lib/Target/SystemZ/Disassembler/SystemZDisassembler.cpp @@ -25,7 +25,7 @@ class SystemZDisassembler : public MCDisassembler { public: SystemZDisassembler(const MCSubtargetInfo &STI, MCContext &Ctx) : MCDisassembler(STI, Ctx) {} - virtual ~SystemZDisassembler() {} + ~SystemZDisassembler() override {} DecodeStatus getInstruction(MCInst &instr, uint64_t &Size, ArrayRef Bytes, uint64_t Address, diff --git a/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.cpp b/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.cpp index 996a492..cf1ee54 100644 --- a/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.cpp +++ b/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.cpp @@ -43,7 +43,8 @@ void SystemZInstPrinter::printOperand(const MCOperand &MO, raw_ostream &O) { } void SystemZInstPrinter::printInst(const MCInst *MI, raw_ostream &O, - StringRef Annot) { + StringRef Annot, + const MCSubtargetInfo &STI) { printInstruction(MI, O); printAnnotation(O, Annot); } diff --git a/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.h b/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.h index 732e5fa..6f56c7b 100644 --- a/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.h +++ b/lib/Target/SystemZ/InstPrinter/SystemZInstPrinter.h @@ -39,7 +39,8 @@ public: // Override MCInstPrinter. void printRegName(raw_ostream &O, unsigned RegNo) const override; - void printInst(const MCInst *MI, raw_ostream &O, StringRef Annot) override; + void printInst(const MCInst *MI, raw_ostream &O, StringRef Annot, + const MCSubtargetInfo &STI) override; private: // Print various types of operand. diff --git a/lib/Target/SystemZ/LLVMBuild.txt b/lib/Target/SystemZ/LLVMBuild.txt index 542aaee..6f8431d 100644 --- a/lib/Target/SystemZ/LLVMBuild.txt +++ b/lib/Target/SystemZ/LLVMBuild.txt @@ -31,5 +31,5 @@ has_jit = 1 type = Library name = SystemZCodeGen parent = SystemZ -required_libraries = AsmPrinter CodeGen Core MC SelectionDAG Support SystemZAsmPrinter SystemZDesc SystemZInfo Target +required_libraries = Analysis AsmPrinter CodeGen Core MC SelectionDAG Support SystemZAsmPrinter SystemZDesc SystemZInfo Target add_to_library_groups = SystemZ diff --git a/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp b/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp index b79b1d8..1c3887a 100644 --- a/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp +++ b/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp @@ -62,7 +62,7 @@ public: llvm_unreachable("SystemZ does do not have assembler relaxation"); } bool writeNopData(uint64_t Count, MCObjectWriter *OW) const override; - MCObjectWriter *createObjectWriter(raw_ostream &OS) const override { + MCObjectWriter *createObjectWriter(raw_pwrite_stream &OS) const override { return createSystemZObjectWriter(OS, OSABI); } }; diff --git a/lib/Target/SystemZ/MCTargetDesc/SystemZMCCodeEmitter.cpp b/lib/Target/SystemZ/MCTargetDesc/SystemZMCCodeEmitter.cpp index 40dc48e..8dd70b9 100644 --- a/lib/Target/SystemZ/MCTargetDesc/SystemZMCCodeEmitter.cpp +++ b/lib/Target/SystemZ/MCTargetDesc/SystemZMCCodeEmitter.cpp @@ -32,7 +32,7 @@ public: : MCII(mcii), Ctx(ctx) { } - ~SystemZMCCodeEmitter() {} + ~SystemZMCCodeEmitter() override {} // OVerride MCCodeEmitter. void EncodeInstruction(const MCInst &MI, raw_ostream &OS, diff --git a/lib/Target/SystemZ/MCTargetDesc/SystemZMCObjectWriter.cpp b/lib/Target/SystemZ/MCTargetDesc/SystemZMCObjectWriter.cpp index 2632518..ee1af02 100644 --- a/lib/Target/SystemZ/MCTargetDesc/SystemZMCObjectWriter.cpp +++ b/lib/Target/SystemZ/MCTargetDesc/SystemZMCObjectWriter.cpp @@ -20,7 +20,7 @@ class SystemZObjectWriter : public MCELFObjectTargetWriter { public: SystemZObjectWriter(uint8_t OSABI); - virtual ~SystemZObjectWriter(); + ~SystemZObjectWriter() override; protected: // Override MCELFObjectTargetWriter. @@ -152,7 +152,7 @@ unsigned SystemZObjectWriter::GetRelocType(const MCValue &Target, } } -MCObjectWriter *llvm::createSystemZObjectWriter(raw_ostream &OS, +MCObjectWriter *llvm::createSystemZObjectWriter(raw_pwrite_stream &OS, uint8_t OSABI) { MCELFObjectTargetWriter *MOTW = new SystemZObjectWriter(OSABI); return createELFObjectWriter(MOTW, OS, /*IsLittleEndian=*/false); diff --git a/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp b/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp index ffd05a9..ea56fb1 100644 --- a/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp +++ b/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp @@ -172,12 +172,11 @@ static MCCodeGenInfo *createSystemZMCCodeGenInfo(StringRef TT, Reloc::Model RM, return X; } -static MCInstPrinter *createSystemZMCInstPrinter(const Target &T, +static MCInstPrinter *createSystemZMCInstPrinter(const Triple &T, unsigned SyntaxVariant, const MCAsmInfo &MAI, const MCInstrInfo &MII, - const MCRegisterInfo &MRI, - const MCSubtargetInfo &STI) { + const MCRegisterInfo &MRI) { return new SystemZInstPrinter(MAI, MII, MRI); } diff --git a/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.h b/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.h index 962c950..2b2647b 100644 --- a/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.h +++ b/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.h @@ -23,6 +23,7 @@ class MCRegisterInfo; class MCSubtargetInfo; class StringRef; class Target; +class raw_pwrite_stream; class raw_ostream; extern Target TheSystemZTarget; @@ -77,7 +78,7 @@ MCAsmBackend *createSystemZMCAsmBackend(const Target &T, const MCRegisterInfo &MRI, StringRef TT, StringRef CPU); -MCObjectWriter *createSystemZObjectWriter(raw_ostream &OS, uint8_t OSABI); +MCObjectWriter *createSystemZObjectWriter(raw_pwrite_stream &OS, uint8_t OSABI); } // end namespace llvm // Defines symbolic names for SystemZ registers. diff --git a/lib/Target/SystemZ/SystemZ.h b/lib/Target/SystemZ/SystemZ.h index 5f17edb..b3a7310 100644 --- a/lib/Target/SystemZ/SystemZ.h +++ b/lib/Target/SystemZ/SystemZ.h @@ -68,6 +68,18 @@ const unsigned CCMASK_TM_MSB_0 = CCMASK_0 | CCMASK_1; const unsigned CCMASK_TM_MSB_1 = CCMASK_2 | CCMASK_3; const unsigned CCMASK_TM = CCMASK_ANY; +// Condition-code mask assignments for TRANSACTION_BEGIN. +const unsigned CCMASK_TBEGIN_STARTED = CCMASK_0; +const unsigned CCMASK_TBEGIN_INDETERMINATE = CCMASK_1; +const unsigned CCMASK_TBEGIN_TRANSIENT = CCMASK_2; +const unsigned CCMASK_TBEGIN_PERSISTENT = CCMASK_3; +const unsigned CCMASK_TBEGIN = CCMASK_ANY; + +// Condition-code mask assignments for TRANSACTION_END. +const unsigned CCMASK_TEND_TX = CCMASK_0; +const unsigned CCMASK_TEND_NOTX = CCMASK_2; +const unsigned CCMASK_TEND = CCMASK_TEND_TX | CCMASK_TEND_NOTX; + // The position of the low CC bit in an IPM result. const unsigned IPM_CC = 28; diff --git a/lib/Target/SystemZ/SystemZAsmPrinter.cpp b/lib/Target/SystemZ/SystemZAsmPrinter.cpp index 18e37e3..2524733 100644 --- a/lib/Target/SystemZ/SystemZAsmPrinter.cpp +++ b/lib/Target/SystemZ/SystemZAsmPrinter.cpp @@ -255,29 +255,6 @@ bool SystemZAsmPrinter::PrintAsmMemoryOperand(const MachineInstr *MI, return false; } -void SystemZAsmPrinter::EmitEndOfAsmFile(Module &M) { - if (Triple(TM.getTargetTriple()).isOSBinFormatELF()) { - auto &TLOFELF = - static_cast(getObjFileLowering()); - - MachineModuleInfoELF &MMIELF = MMI->getObjFileInfo(); - - // Output stubs for external and common global variables. - MachineModuleInfoELF::SymbolListTy Stubs = MMIELF.GetGVStubList(); - if (!Stubs.empty()) { - OutStreamer.SwitchSection(TLOFELF.getDataRelSection()); - const DataLayout *TD = TM.getDataLayout(); - - for (unsigned i = 0, e = Stubs.size(); i != e; ++i) { - OutStreamer.EmitLabel(Stubs[i].first); - OutStreamer.EmitSymbolValue(Stubs[i].second.getPointer(), - TD->getPointerSize(0)); - } - Stubs.clear(); - } - } -} - // Force static initialization. extern "C" void LLVMInitializeSystemZAsmPrinter() { RegisterAsmPrinter X(TheSystemZTarget); diff --git a/lib/Target/SystemZ/SystemZAsmPrinter.h b/lib/Target/SystemZ/SystemZAsmPrinter.h index a4d5b78..7f6e823 100644 --- a/lib/Target/SystemZ/SystemZAsmPrinter.h +++ b/lib/Target/SystemZ/SystemZAsmPrinter.h @@ -38,7 +38,6 @@ public: bool PrintAsmMemoryOperand(const MachineInstr *MI, unsigned OpNo, unsigned AsmVariant, const char *ExtraCode, raw_ostream &OS) override; - void EmitEndOfAsmFile(Module &M) override; }; } // end namespace llvm diff --git a/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp b/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp index a52aa25..1a58b53 100644 --- a/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp +++ b/lib/Target/SystemZ/SystemZISelDAGToDAG.cpp @@ -898,6 +898,9 @@ SDNode *SystemZDAGToDAGISel::tryRISBGZero(SDNode *N) { } unsigned Opcode = SystemZ::RISBG; + // Prefer RISBGN if available, since it does not clobber CC. + if (Subtarget->hasMiscellaneousExtensions()) + Opcode = SystemZ::RISBGN; EVT OpcodeVT = MVT::i64; if (VT == MVT::i32 && Subtarget->hasHighWord()) { Opcode = SystemZ::RISBMux; @@ -945,9 +948,13 @@ SDNode *SystemZDAGToDAGISel::tryRxSBG(SDNode *N, unsigned Opcode) { // See whether we can avoid an AND in the first operand by converting // ROSBG to RISBG. - if (Opcode == SystemZ::ROSBG && detectOrAndInsertion(Op0, RxSBG[I].Mask)) + if (Opcode == SystemZ::ROSBG && detectOrAndInsertion(Op0, RxSBG[I].Mask)) { Opcode = SystemZ::RISBG; - + // Prefer RISBGN if available, since it does not clobber CC. + if (Subtarget->hasMiscellaneousExtensions()) + Opcode = SystemZ::RISBGN; + } + EVT VT = N->getValueType(0); SDValue Ops[5] = { convertTo(SDLoc(N), MVT::i64, Op0), diff --git a/lib/Target/SystemZ/SystemZISelLowering.cpp b/lib/Target/SystemZ/SystemZISelLowering.cpp index 0ca8bcd..21882cb 100644 --- a/lib/Target/SystemZ/SystemZISelLowering.cpp +++ b/lib/Target/SystemZ/SystemZISelLowering.cpp @@ -20,6 +20,7 @@ #include "llvm/CodeGen/MachineInstrBuilder.h" #include "llvm/CodeGen/MachineRegisterInfo.h" #include "llvm/CodeGen/TargetLoweringObjectFileImpl.h" +#include "llvm/IR/Intrinsics.h" #include using namespace llvm; @@ -163,8 +164,13 @@ SystemZTargetLowering::SystemZTargetLowering(const TargetMachine &tm, // available, or if the operand is constant. setOperationAction(ISD::ATOMIC_LOAD_SUB, VT, Custom); + // Use POPCNT on z196 and above. + if (Subtarget.hasPopulationCount()) + setOperationAction(ISD::CTPOP, VT, Custom); + else + setOperationAction(ISD::CTPOP, VT, Expand); + // No special instructions for these. - setOperationAction(ISD::CTPOP, VT, Expand); setOperationAction(ISD::CTTZ, VT, Expand); setOperationAction(ISD::CTTZ_ZERO_UNDEF, VT, Expand); setOperationAction(ISD::CTLZ_ZERO_UNDEF, VT, Expand); @@ -299,6 +305,9 @@ SystemZTargetLowering::SystemZTargetLowering(const TargetMachine &tm, // Codes for which we want to perform some z-specific combinations. setTargetDAGCombine(ISD::SIGN_EXTEND); + // Handle intrinsics. + setOperationAction(ISD::INTRINSIC_W_CHAIN, MVT::Other, Custom); + // We want to use MVC in preference to even a single load/store pair. MaxStoresPerMemcpy = 0; MaxStoresPerMemcpyOptSize = 0; @@ -342,6 +351,16 @@ bool SystemZTargetLowering::isFPImmLegal(const APFloat &Imm, EVT VT) const { return Imm.isZero() || Imm.isNegZero(); } +bool SystemZTargetLowering::isLegalICmpImmediate(int64_t Imm) const { + // We can use CGFI or CLGFI. + return isInt<32>(Imm) || isUInt<32>(Imm); +} + +bool SystemZTargetLowering::isLegalAddImmediate(int64_t Imm) const { + // We can use ALGFI or SLGFI. + return isUInt<32>(Imm) || isUInt<32>(-Imm); +} + bool SystemZTargetLowering::allowsMisalignedMemoryAccesses(EVT VT, unsigned, unsigned, @@ -1016,6 +1035,53 @@ prepareVolatileOrAtomicLoad(SDValue Chain, SDLoc DL, SelectionDAG &DAG) const { return DAG.getNode(SystemZISD::SERIALIZE, DL, MVT::Other, Chain); } +// Return true if Op is an intrinsic node with chain that returns the CC value +// as its only (other) argument. Provide the associated SystemZISD opcode and +// the mask of valid CC values if so. +static bool isIntrinsicWithCCAndChain(SDValue Op, unsigned &Opcode, + unsigned &CCValid) { + unsigned Id = cast(Op.getOperand(1))->getZExtValue(); + switch (Id) { + case Intrinsic::s390_tbegin: + Opcode = SystemZISD::TBEGIN; + CCValid = SystemZ::CCMASK_TBEGIN; + return true; + + case Intrinsic::s390_tbegin_nofloat: + Opcode = SystemZISD::TBEGIN_NOFLOAT; + CCValid = SystemZ::CCMASK_TBEGIN; + return true; + + case Intrinsic::s390_tend: + Opcode = SystemZISD::TEND; + CCValid = SystemZ::CCMASK_TEND; + return true; + + default: + return false; + } +} + +// Emit an intrinsic with chain with a glued value instead of its CC result. +static SDValue emitIntrinsicWithChainAndGlue(SelectionDAG &DAG, SDValue Op, + unsigned Opcode) { + // Copy all operands except the intrinsic ID. + unsigned NumOps = Op.getNumOperands(); + SmallVector Ops; + Ops.reserve(NumOps - 1); + Ops.push_back(Op.getOperand(0)); + for (unsigned I = 2; I < NumOps; ++I) + Ops.push_back(Op.getOperand(I)); + + assert(Op->getNumValues() == 2 && "Expected only CC result and chain"); + SDVTList RawVTs = DAG.getVTList(MVT::Other, MVT::Glue); + SDValue Intr = DAG.getNode(Opcode, SDLoc(Op), RawVTs, Ops); + SDValue OldChain = SDValue(Op.getNode(), 1); + SDValue NewChain = SDValue(Intr.getNode(), 0); + DAG.ReplaceAllUsesOfValueWith(OldChain, NewChain); + return Intr; +} + // CC is a comparison that will be implemented using an integer or // floating-point comparison. Return the condition code mask for // a branch on true. In the integer case, CCMASK_CMP_UO is set for @@ -1530,6 +1596,8 @@ static void adjustForTestUnderMask(SelectionDAG &DAG, Comparison &C) { MaskVal = -(CmpVal & -CmpVal); NewC.ICmpType = SystemZICMP::UnsignedOnly; } + if (!MaskVal) + return; // Check whether the combination of mask, comparison value and comparison // type are suitable. @@ -1571,9 +1639,53 @@ static void adjustForTestUnderMask(SelectionDAG &DAG, Comparison &C) { C.CCMask = NewCCMask; } +// Return a Comparison that tests the condition-code result of intrinsic +// node Call against constant integer CC using comparison code Cond. +// Opcode is the opcode of the SystemZISD operation for the intrinsic +// and CCValid is the set of possible condition-code results. +static Comparison getIntrinsicCmp(SelectionDAG &DAG, unsigned Opcode, + SDValue Call, unsigned CCValid, uint64_t CC, + ISD::CondCode Cond) { + Comparison C(Call, SDValue()); + C.Opcode = Opcode; + C.CCValid = CCValid; + if (Cond == ISD::SETEQ) + // bit 3 for CC==0, bit 0 for CC==3, always false for CC>3. + C.CCMask = CC < 4 ? 1 << (3 - CC) : 0; + else if (Cond == ISD::SETNE) + // ...and the inverse of that. + C.CCMask = CC < 4 ? ~(1 << (3 - CC)) : -1; + else if (Cond == ISD::SETLT || Cond == ISD::SETULT) + // bits above bit 3 for CC==0 (always false), bits above bit 0 for CC==3, + // always true for CC>3. + C.CCMask = CC < 4 ? -1 << (4 - CC) : -1; + else if (Cond == ISD::SETGE || Cond == ISD::SETUGE) + // ...and the inverse of that. + C.CCMask = CC < 4 ? ~(-1 << (4 - CC)) : 0; + else if (Cond == ISD::SETLE || Cond == ISD::SETULE) + // bit 3 and above for CC==0, bit 0 and above for CC==3 (always true), + // always true for CC>3. + C.CCMask = CC < 4 ? -1 << (3 - CC) : -1; + else if (Cond == ISD::SETGT || Cond == ISD::SETUGT) + // ...and the inverse of that. + C.CCMask = CC < 4 ? ~(-1 << (3 - CC)) : 0; + else + llvm_unreachable("Unexpected integer comparison type"); + C.CCMask &= CCValid; + return C; +} + // Decide how to implement a comparison of type Cond between CmpOp0 with CmpOp1. static Comparison getCmp(SelectionDAG &DAG, SDValue CmpOp0, SDValue CmpOp1, ISD::CondCode Cond) { + if (CmpOp1.getOpcode() == ISD::Constant) { + uint64_t Constant = cast(CmpOp1)->getZExtValue(); + unsigned Opcode, CCValid; + if (CmpOp0.getOpcode() == ISD::INTRINSIC_W_CHAIN && + CmpOp0.getResNo() == 0 && CmpOp0->hasNUsesOfValue(1, 0) && + isIntrinsicWithCCAndChain(CmpOp0, Opcode, CCValid)) + return getIntrinsicCmp(DAG, Opcode, CmpOp0, CCValid, Constant, Cond); + } Comparison C(CmpOp0, CmpOp1); C.CCMask = CCMaskForCondCode(Cond); if (C.Op0.getValueType().isFloatingPoint()) { @@ -1615,6 +1727,17 @@ static Comparison getCmp(SelectionDAG &DAG, SDValue CmpOp0, SDValue CmpOp1, // Emit the comparison instruction described by C. static SDValue emitCmp(SelectionDAG &DAG, SDLoc DL, Comparison &C) { + if (!C.Op1.getNode()) { + SDValue Op; + switch (C.Op0.getOpcode()) { + case ISD::INTRINSIC_W_CHAIN: + Op = emitIntrinsicWithChainAndGlue(DAG, C.Op0, C.Opcode); + break; + default: + llvm_unreachable("Invalid comparison operands"); + } + return SDValue(Op.getNode(), Op->getNumValues() - 1); + } if (C.Opcode == SystemZISD::ICMP) return DAG.getNode(SystemZISD::ICMP, DL, MVT::Glue, C.Op0, C.Op1, DAG.getConstant(C.ICmpType, MVT::i32)); @@ -1696,7 +1819,6 @@ SDValue SystemZTargetLowering::lowerSETCC(SDValue Op, } SDValue SystemZTargetLowering::lowerBR_CC(SDValue Op, SelectionDAG &DAG) const { - SDValue Chain = Op.getOperand(0); ISD::CondCode CC = cast(Op.getOperand(1))->get(); SDValue CmpOp0 = Op.getOperand(2); SDValue CmpOp1 = Op.getOperand(3); @@ -1706,7 +1828,7 @@ SDValue SystemZTargetLowering::lowerBR_CC(SDValue Op, SelectionDAG &DAG) const { Comparison C(getCmp(DAG, CmpOp0, CmpOp1, CC)); SDValue Glue = emitCmp(DAG, DL, C); return DAG.getNode(SystemZISD::BR_CCMASK, DL, Op.getValueType(), - Chain, DAG.getConstant(C.CCValid, MVT::i32), + Op.getOperand(0), DAG.getConstant(C.CCValid, MVT::i32), DAG.getConstant(C.CCMask, MVT::i32), Dest, Glue); } @@ -2100,6 +2222,7 @@ SDValue SystemZTargetLowering::lowerVACOPY(SDValue Op, return DAG.getMemcpy(Chain, DL, DstPtr, SrcPtr, DAG.getIntPtrConstant(32), /*Align*/8, /*isVolatile*/false, /*AlwaysInline*/false, + /*isTailCall*/false, MachinePointerInfo(DstSV), MachinePointerInfo(SrcSV)); } @@ -2292,6 +2415,46 @@ SDValue SystemZTargetLowering::lowerOR(SDValue Op, SelectionDAG &DAG) const { MVT::i64, HighOp, Low32); } +SDValue SystemZTargetLowering::lowerCTPOP(SDValue Op, + SelectionDAG &DAG) const { + EVT VT = Op.getValueType(); + int64_t OrigBitSize = VT.getSizeInBits(); + SDLoc DL(Op); + + // Get the known-zero mask for the operand. + Op = Op.getOperand(0); + APInt KnownZero, KnownOne; + DAG.computeKnownBits(Op, KnownZero, KnownOne); + unsigned NumSignificantBits = (~KnownZero).getActiveBits(); + if (NumSignificantBits == 0) + return DAG.getConstant(0, VT); + + // Skip known-zero high parts of the operand. + int64_t BitSize = (int64_t)1 << Log2_32_Ceil(NumSignificantBits); + BitSize = std::min(BitSize, OrigBitSize); + + // The POPCNT instruction counts the number of bits in each byte. + Op = DAG.getNode(ISD::ANY_EXTEND, DL, MVT::i64, Op); + Op = DAG.getNode(SystemZISD::POPCNT, DL, MVT::i64, Op); + Op = DAG.getNode(ISD::TRUNCATE, DL, VT, Op); + + // Add up per-byte counts in a binary tree. All bits of Op at + // position larger than BitSize remain zero throughout. + for (int64_t I = BitSize / 2; I >= 8; I = I / 2) { + SDValue Tmp = DAG.getNode(ISD::SHL, DL, VT, Op, DAG.getConstant(I, VT)); + if (BitSize != OrigBitSize) + Tmp = DAG.getNode(ISD::AND, DL, VT, Tmp, + DAG.getConstant(((uint64_t)1 << BitSize) - 1, VT)); + Op = DAG.getNode(ISD::ADD, DL, VT, Op, Tmp); + } + + // Extract overall result from high byte. + if (BitSize > 8) + Op = DAG.getNode(ISD::SRL, DL, VT, Op, DAG.getConstant(BitSize - 8, VT)); + + return Op; +} + // Op is an atomic load. Lower it into a normal volatile load. SDValue SystemZTargetLowering::lowerATOMIC_LOAD(SDValue Op, SelectionDAG &DAG) const { @@ -2505,6 +2668,30 @@ SDValue SystemZTargetLowering::lowerPREFETCH(SDValue Op, Node->getMemoryVT(), Node->getMemOperand()); } +// Return an i32 that contains the value of CC immediately after After, +// whose final operand must be MVT::Glue. +static SDValue getCCResult(SelectionDAG &DAG, SDNode *After) { + SDValue Glue = SDValue(After, After->getNumValues() - 1); + SDValue IPM = DAG.getNode(SystemZISD::IPM, SDLoc(After), MVT::i32, Glue); + return DAG.getNode(ISD::SRL, SDLoc(After), MVT::i32, IPM, + DAG.getConstant(SystemZ::IPM_CC, MVT::i32)); +} + +SDValue +SystemZTargetLowering::lowerINTRINSIC_W_CHAIN(SDValue Op, + SelectionDAG &DAG) const { + unsigned Opcode, CCValid; + if (isIntrinsicWithCCAndChain(Op, Opcode, CCValid)) { + assert(Op->getNumValues() == 2 && "Expected only CC result and chain"); + SDValue Glued = emitIntrinsicWithChainAndGlue(DAG, Op, Opcode); + SDValue CC = getCCResult(DAG, Glued.getNode()); + DAG.ReplaceAllUsesOfValueWith(SDValue(Op.getNode(), 0), CC); + return SDValue(); + } + + return SDValue(); +} + SDValue SystemZTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const { switch (Op.getOpcode()) { @@ -2542,6 +2729,8 @@ SDValue SystemZTargetLowering::LowerOperation(SDValue Op, return lowerUDIVREM(Op, DAG); case ISD::OR: return lowerOR(Op, DAG); + case ISD::CTPOP: + return lowerCTPOP(Op, DAG); case ISD::ATOMIC_SWAP: return lowerATOMIC_LOAD_OP(Op, DAG, SystemZISD::ATOMIC_SWAPW); case ISD::ATOMIC_STORE: @@ -2576,6 +2765,8 @@ SDValue SystemZTargetLowering::LowerOperation(SDValue Op, return lowerSTACKRESTORE(Op, DAG); case ISD::PREFETCH: return lowerPREFETCH(Op, DAG); + case ISD::INTRINSIC_W_CHAIN: + return lowerINTRINSIC_W_CHAIN(Op, DAG); default: llvm_unreachable("Unexpected node to lower"); } @@ -2616,6 +2807,9 @@ const char *SystemZTargetLowering::getTargetNodeName(unsigned Opcode) const { OPCODE(SEARCH_STRING); OPCODE(IPM); OPCODE(SERIALIZE); + OPCODE(TBEGIN); + OPCODE(TBEGIN_NOFLOAT); + OPCODE(TEND); OPCODE(ATOMIC_SWAPW); OPCODE(ATOMIC_LOADW_ADD); OPCODE(ATOMIC_LOADW_SUB); @@ -3443,6 +3637,50 @@ SystemZTargetLowering::emitStringWrapper(MachineInstr *MI, return DoneMBB; } +// Update TBEGIN instruction with final opcode and register clobbers. +MachineBasicBlock * +SystemZTargetLowering::emitTransactionBegin(MachineInstr *MI, + MachineBasicBlock *MBB, + unsigned Opcode, + bool NoFloat) const { + MachineFunction &MF = *MBB->getParent(); + const TargetFrameLowering *TFI = Subtarget.getFrameLowering(); + const SystemZInstrInfo *TII = Subtarget.getInstrInfo(); + + // Update opcode. + MI->setDesc(TII->get(Opcode)); + + // We cannot handle a TBEGIN that clobbers the stack or frame pointer. + // Make sure to add the corresponding GRSM bits if they are missing. + uint64_t Control = MI->getOperand(2).getImm(); + static const unsigned GPRControlBit[16] = { + 0x8000, 0x8000, 0x4000, 0x4000, 0x2000, 0x2000, 0x1000, 0x1000, + 0x0800, 0x0800, 0x0400, 0x0400, 0x0200, 0x0200, 0x0100, 0x0100 + }; + Control |= GPRControlBit[15]; + if (TFI->hasFP(MF)) + Control |= GPRControlBit[11]; + MI->getOperand(2).setImm(Control); + + // Add GPR clobbers. + for (int I = 0; I < 16; I++) { + if ((Control & GPRControlBit[I]) == 0) { + unsigned Reg = SystemZMC::GR64Regs[I]; + MI->addOperand(MachineOperand::CreateReg(Reg, true, true)); + } + } + + // Add FPR clobbers. + if (!NoFloat && (Control & 4) != 0) { + for (int I = 0; I < 16; I++) { + unsigned Reg = SystemZMC::FP64Regs[I]; + MI->addOperand(MachineOperand::CreateReg(Reg, true, true)); + } + } + + return MBB; +} + MachineBasicBlock *SystemZTargetLowering:: EmitInstrWithCustomInserter(MachineInstr *MI, MachineBasicBlock *MBB) const { switch (MI->getOpcode()) { @@ -3684,6 +3922,12 @@ EmitInstrWithCustomInserter(MachineInstr *MI, MachineBasicBlock *MBB) const { return emitStringWrapper(MI, MBB, SystemZ::MVST); case SystemZ::SRSTLoop: return emitStringWrapper(MI, MBB, SystemZ::SRST); + case SystemZ::TBEGIN: + return emitTransactionBegin(MI, MBB, SystemZ::TBEGIN, false); + case SystemZ::TBEGIN_nofloat: + return emitTransactionBegin(MI, MBB, SystemZ::TBEGIN, true); + case SystemZ::TBEGINC: + return emitTransactionBegin(MI, MBB, SystemZ::TBEGINC, true); default: llvm_unreachable("Unexpected instr type to insert"); } diff --git a/lib/Target/SystemZ/SystemZISelLowering.h b/lib/Target/SystemZ/SystemZISelLowering.h index 23c62c9..56d7ef4 100644 --- a/lib/Target/SystemZ/SystemZISelLowering.h +++ b/lib/Target/SystemZ/SystemZISelLowering.h @@ -87,6 +87,9 @@ enum { // the number of the register. EXTRACT_ACCESS, + // Count number of bits set in operand 0 per byte. + POPCNT, + // Wrappers around the ISD opcodes of the same name. The output and // first input operands are GR128s. The trailing numbers are the // widths of the second operand in bits. @@ -143,6 +146,15 @@ enum { // Perform a serialization operation. (BCR 15,0 or BCR 14,0.) SERIALIZE, + // Transaction begin. The first operand is the chain, the second + // the TDB pointer, and the third the immediate control field. + // Returns chain and glue. + TBEGIN, + TBEGIN_NOFLOAT, + + // Transaction end. Just the chain operand. Returns chain and glue. + TEND, + // Wrappers around the inner loop of an 8- or 16-bit ATOMIC_SWAP or // ATOMIC_LOAD_. // @@ -213,6 +225,8 @@ public: EVT getSetCCResultType(LLVMContext &, EVT) const override; bool isFMAFasterThanFMulAndFAdd(EVT VT) const override; bool isFPImmLegal(const APFloat &Imm, EVT VT) const override; + bool isLegalICmpImmediate(int64_t Imm) const override; + bool isLegalAddImmediate(int64_t Imm) const override; bool isLegalAddressingMode(const AddrMode &AM, Type *Ty) const override; bool allowsMisalignedMemoryAccesses(EVT VT, unsigned AS, unsigned Align, @@ -302,6 +316,7 @@ private: SDValue lowerUDIVREM(SDValue Op, SelectionDAG &DAG) const; SDValue lowerBITCAST(SDValue Op, SelectionDAG &DAG) const; SDValue lowerOR(SDValue Op, SelectionDAG &DAG) const; + SDValue lowerCTPOP(SDValue Op, SelectionDAG &DAG) const; SDValue lowerATOMIC_LOAD(SDValue Op, SelectionDAG &DAG) const; SDValue lowerATOMIC_STORE(SDValue Op, SelectionDAG &DAG) const; SDValue lowerATOMIC_LOAD_OP(SDValue Op, SelectionDAG &DAG, @@ -312,6 +327,7 @@ private: SDValue lowerSTACKSAVE(SDValue Op, SelectionDAG &DAG) const; SDValue lowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG) const; SDValue lowerPREFETCH(SDValue Op, SelectionDAG &DAG) const; + SDValue lowerINTRINSIC_W_CHAIN(SDValue Op, SelectionDAG &DAG) const; // If the last instruction before MBBI in MBB was some form of COMPARE, // try to replace it with a COMPARE AND BRANCH just before MBBI. @@ -349,6 +365,10 @@ private: MachineBasicBlock *emitStringWrapper(MachineInstr *MI, MachineBasicBlock *BB, unsigned Opcode) const; + MachineBasicBlock *emitTransactionBegin(MachineInstr *MI, + MachineBasicBlock *MBB, + unsigned Opcode, + bool NoFloat) const; }; } // end namespace llvm diff --git a/lib/Target/SystemZ/SystemZInstrFormats.td b/lib/Target/SystemZ/SystemZInstrFormats.td index 9f59a1c..2d3c9e2 100644 --- a/lib/Target/SystemZ/SystemZInstrFormats.td +++ b/lib/Target/SystemZ/SystemZInstrFormats.td @@ -473,6 +473,17 @@ class InstSS op, dag outs, dag ins, string asmstr, list pattern> let Inst{15-0} = BD2; } +class InstS op, dag outs, dag ins, string asmstr, list pattern> + : InstSystemZ<4, outs, ins, asmstr, pattern> { + field bits<32> Inst; + field bits<32> SoftFail = 0; + + bits<16> BD2; + + let Inst{31-16} = op; + let Inst{15-0} = BD2; +} + //===----------------------------------------------------------------------===// // Instruction definitions with semantics //===----------------------------------------------------------------------===// diff --git a/lib/Target/SystemZ/SystemZInstrInfo.cpp b/lib/Target/SystemZ/SystemZInstrInfo.cpp index 5128993..3a02859 100644 --- a/lib/Target/SystemZ/SystemZInstrInfo.cpp +++ b/lib/Target/SystemZ/SystemZInstrInfo.cpp @@ -723,9 +723,12 @@ SystemZInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI, unsigned Start, End; if (isRxSBGMask(Imm, And.RegSize, Start, End)) { unsigned NewOpcode; - if (And.RegSize == 64) + if (And.RegSize == 64) { NewOpcode = SystemZ::RISBG; - else { + // Prefer RISBGN if available, since it does not clobber CC. + if (STI.hasMiscellaneousExtensions()) + NewOpcode = SystemZ::RISBGN; + } else { NewOpcode = SystemZ::RISBMux; Start &= 31; End &= 31; @@ -1146,17 +1149,22 @@ unsigned SystemZInstrInfo::getOpcodeForOffset(unsigned Opcode, unsigned SystemZInstrInfo::getLoadAndTest(unsigned Opcode) const { switch (Opcode) { - case SystemZ::L: return SystemZ::LT; - case SystemZ::LY: return SystemZ::LT; - case SystemZ::LG: return SystemZ::LTG; - case SystemZ::LGF: return SystemZ::LTGF; - case SystemZ::LR: return SystemZ::LTR; - case SystemZ::LGFR: return SystemZ::LTGFR; - case SystemZ::LGR: return SystemZ::LTGR; - case SystemZ::LER: return SystemZ::LTEBR; - case SystemZ::LDR: return SystemZ::LTDBR; - case SystemZ::LXR: return SystemZ::LTXBR; - default: return 0; + case SystemZ::L: return SystemZ::LT; + case SystemZ::LY: return SystemZ::LT; + case SystemZ::LG: return SystemZ::LTG; + case SystemZ::LGF: return SystemZ::LTGF; + case SystemZ::LR: return SystemZ::LTR; + case SystemZ::LGFR: return SystemZ::LTGFR; + case SystemZ::LGR: return SystemZ::LTGR; + case SystemZ::LER: return SystemZ::LTEBR; + case SystemZ::LDR: return SystemZ::LTDBR; + case SystemZ::LXR: return SystemZ::LTXBR; + // On zEC12 we prefer to use RISBGN. But if there is a chance to + // actually use the condition code, we may turn it back into RISGB. + // Note that RISBG is not really a "load-and-test" instruction, + // but sets the same condition code values, so is OK to use here. + case SystemZ::RISBGN: return SystemZ::RISBG; + default: return 0; } } diff --git a/lib/Target/SystemZ/SystemZInstrInfo.td b/lib/Target/SystemZ/SystemZInstrInfo.td index a7f7747..820f30b 100644 --- a/lib/Target/SystemZ/SystemZInstrInfo.td +++ b/lib/Target/SystemZ/SystemZInstrInfo.td @@ -1061,6 +1061,10 @@ let Defs = [CC] in { def RISBG : RotateSelectRIEf<"risbg", 0xEC55, GR64, GR64>; } +// On zEC12 we have a variant of RISBG that does not set CC. +let Predicates = [FeatureMiscellaneousExtensions] in + def RISBGN : RotateSelectRIEf<"risbgn", 0xEC59, GR64, GR64>; + // Forms of RISBG that only affect one word of the destination register. // They do not set CC. let Predicates = [FeatureHighWord] in { @@ -1358,6 +1362,60 @@ let Defs = [CC] in { } //===----------------------------------------------------------------------===// +// Transactional execution +//===----------------------------------------------------------------------===// + +let Predicates = [FeatureTransactionalExecution] in { + // Transaction Begin + let hasSideEffects = 1, mayStore = 1, + usesCustomInserter = 1, Defs = [CC] in { + def TBEGIN : InstSIL<0xE560, + (outs), (ins bdaddr12only:$BD1, imm32zx16:$I2), + "tbegin\t$BD1, $I2", + [(z_tbegin bdaddr12only:$BD1, imm32zx16:$I2)]>; + def TBEGIN_nofloat : Pseudo<(outs), (ins bdaddr12only:$BD1, imm32zx16:$I2), + [(z_tbegin_nofloat bdaddr12only:$BD1, + imm32zx16:$I2)]>; + def TBEGINC : InstSIL<0xE561, + (outs), (ins bdaddr12only:$BD1, imm32zx16:$I2), + "tbeginc\t$BD1, $I2", + [(int_s390_tbeginc bdaddr12only:$BD1, + imm32zx16:$I2)]>; + } + + // Transaction End + let hasSideEffects = 1, Defs = [CC], BD2 = 0 in + def TEND : InstS<0xB2F8, (outs), (ins), "tend", [(z_tend)]>; + + // Transaction Abort + let hasSideEffects = 1, isTerminator = 1, isBarrier = 1 in + def TABORT : InstS<0xB2FC, (outs), (ins bdaddr12only:$BD2), + "tabort\t$BD2", + [(int_s390_tabort bdaddr12only:$BD2)]>; + + // Nontransactional Store + let hasSideEffects = 1 in + def NTSTG : StoreRXY<"ntstg", 0xE325, int_s390_ntstg, GR64, 8>; + + // Extract Transaction Nesting Depth + let hasSideEffects = 1 in + def ETND : InherentRRE<"etnd", 0xB2EC, GR32, (int_s390_etnd)>; +} + +//===----------------------------------------------------------------------===// +// Processor assist +//===----------------------------------------------------------------------===// + +let Predicates = [FeatureProcessorAssist] in { + let hasSideEffects = 1, R4 = 0 in + def PPA : InstRRF<0xB2E8, (outs), (ins GR64:$R1, GR64:$R2, imm32zx4:$R3), + "ppa\t$R1, $R2, $R3", []>; + def : Pat<(int_s390_ppa_txassist GR32:$src), + (PPA (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GR32:$src, subreg_l32), + 0, 1)>; +} + +//===----------------------------------------------------------------------===// // Miscellaneous Instructions. //===----------------------------------------------------------------------===// @@ -1382,6 +1440,13 @@ let Defs = [CC] in { def : Pat<(ctlz GR64:$src), (EXTRACT_SUBREG (FLOGR GR64:$src), subreg_h64)>; +// Population count. Counts bits set per byte. +let Predicates = [FeaturePopulationCount], Defs = [CC] in { + def POPCNT : InstRRE<0xB9E1, (outs GR64:$R1), (ins GR64:$R2), + "popcnt\t$R1, $R2", + [(set GR64:$R1, (z_popcnt GR64:$R2))]>; +} + // Use subregs to populate the "don't care" bits in a 32-bit to 64-bit anyext. def : Pat<(i64 (anyext GR32:$src)), (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GR32:$src, subreg_l32)>; diff --git a/lib/Target/SystemZ/SystemZOperators.td b/lib/Target/SystemZ/SystemZOperators.td index 51ac5da..3151052 100644 --- a/lib/Target/SystemZ/SystemZOperators.td +++ b/lib/Target/SystemZ/SystemZOperators.td @@ -79,6 +79,9 @@ def SDT_ZI32Intrinsic : SDTypeProfile<1, 0, [SDTCisVT<0, i32>]>; def SDT_ZPrefetch : SDTypeProfile<0, 2, [SDTCisVT<0, i32>, SDTCisPtrTy<1>]>; +def SDT_ZTBegin : SDTypeProfile<0, 2, + [SDTCisPtrTy<0>, + SDTCisVT<1, i32>]>; //===----------------------------------------------------------------------===// // Node definitions @@ -121,6 +124,7 @@ def z_select_ccmask : SDNode<"SystemZISD::SELECT_CCMASK", SDT_ZSelectCCMask, def z_adjdynalloc : SDNode<"SystemZISD::ADJDYNALLOC", SDT_ZAdjDynAlloc>; def z_extract_access : SDNode<"SystemZISD::EXTRACT_ACCESS", SDT_ZExtractAccess>; +def z_popcnt : SDNode<"SystemZISD::POPCNT", SDTIntUnaryOp>; def z_umul_lohi64 : SDNode<"SystemZISD::UMUL_LOHI64", SDT_ZGR128Binary64>; def z_sdivrem32 : SDNode<"SystemZISD::SDIVREM32", SDT_ZGR128Binary32>; def z_sdivrem64 : SDNode<"SystemZISD::SDIVREM64", SDT_ZGR128Binary64>; @@ -179,6 +183,15 @@ def z_prefetch : SDNode<"SystemZISD::PREFETCH", SDT_ZPrefetch, [SDNPHasChain, SDNPMayLoad, SDNPMayStore, SDNPMemOperand]>; +def z_tbegin : SDNode<"SystemZISD::TBEGIN", SDT_ZTBegin, + [SDNPHasChain, SDNPOutGlue, SDNPMayStore, + SDNPSideEffect]>; +def z_tbegin_nofloat : SDNode<"SystemZISD::TBEGIN_NOFLOAT", SDT_ZTBegin, + [SDNPHasChain, SDNPOutGlue, SDNPMayStore, + SDNPSideEffect]>; +def z_tend : SDNode<"SystemZISD::TEND", SDTNone, + [SDNPHasChain, SDNPOutGlue, SDNPSideEffect]>; + //===----------------------------------------------------------------------===// // Pattern fragments //===----------------------------------------------------------------------===// diff --git a/lib/Target/SystemZ/SystemZProcessors.td b/lib/Target/SystemZ/SystemZProcessors.td index 1594854..15614c9 100644 --- a/lib/Target/SystemZ/SystemZProcessors.td +++ b/lib/Target/SystemZ/SystemZProcessors.td @@ -39,6 +39,11 @@ def FeatureFPExtension : SystemZFeature< "Assume that the floating-point extension facility is installed" >; +def FeaturePopulationCount : SystemZFeature< + "population-count", "PopulationCount", + "Assume that the population-count facility is installed" +>; + def FeatureFastSerialization : SystemZFeature< "fast-serialization", "FastSerialization", "Assume that the fast-serialization facility is installed" @@ -50,13 +55,30 @@ def FeatureInterlockedAccess1 : SystemZFeature< >; def FeatureNoInterlockedAccess1 : SystemZMissingFeature<"InterlockedAccess1">; +def FeatureMiscellaneousExtensions : SystemZFeature< + "miscellaneous-extensions", "MiscellaneousExtensions", + "Assume that the miscellaneous-extensions facility is installed" +>; + +def FeatureTransactionalExecution : SystemZFeature< + "transactional-execution", "TransactionalExecution", + "Assume that the transactional-execution facility is installed" +>; + +def FeatureProcessorAssist : SystemZFeature< + "processor-assist", "ProcessorAssist", + "Assume that the processor-assist facility is installed" +>; + def : Processor<"generic", NoItineraries, []>; def : Processor<"z10", NoItineraries, []>; def : Processor<"z196", NoItineraries, [FeatureDistinctOps, FeatureLoadStoreOnCond, FeatureHighWord, - FeatureFPExtension, FeatureFastSerialization, - FeatureInterlockedAccess1]>; + FeatureFPExtension, FeaturePopulationCount, + FeatureFastSerialization, FeatureInterlockedAccess1]>; def : Processor<"zEC12", NoItineraries, [FeatureDistinctOps, FeatureLoadStoreOnCond, FeatureHighWord, - FeatureFPExtension, FeatureFastSerialization, - FeatureInterlockedAccess1]>; + FeatureFPExtension, FeaturePopulationCount, + FeatureFastSerialization, FeatureInterlockedAccess1, + FeatureMiscellaneousExtensions, + FeatureTransactionalExecution, FeatureProcessorAssist]>; diff --git a/lib/Target/SystemZ/SystemZSubtarget.cpp b/lib/Target/SystemZ/SystemZSubtarget.cpp index 31a2bff..de725ae 100644 --- a/lib/Target/SystemZ/SystemZSubtarget.cpp +++ b/lib/Target/SystemZ/SystemZSubtarget.cpp @@ -10,7 +10,6 @@ #include "SystemZSubtarget.h" #include "MCTargetDesc/SystemZMCTargetDesc.h" #include "llvm/IR/GlobalValue.h" -#include "llvm/Support/Host.h" using namespace llvm; @@ -28,10 +27,6 @@ SystemZSubtarget::initializeSubtargetDependencies(StringRef CPU, StringRef FS) { std::string CPUName = CPU; if (CPUName.empty()) CPUName = "generic"; -#if defined(__linux__) && defined(__s390x__) - if (CPUName == "generic") - CPUName = sys::getHostCPUName(); -#endif // Parse features string. ParseSubtargetFeatures(CPUName, FS); return *this; @@ -43,7 +38,9 @@ SystemZSubtarget::SystemZSubtarget(const std::string &TT, const TargetMachine &TM) : SystemZGenSubtargetInfo(TT, CPU, FS), HasDistinctOps(false), HasLoadStoreOnCond(false), HasHighWord(false), HasFPExtension(false), - HasFastSerialization(false), HasInterlockedAccess1(false), + HasPopulationCount(false), HasFastSerialization(false), + HasInterlockedAccess1(false), HasMiscellaneousExtensions(false), + HasTransactionalExecution(false), HasProcessorAssist(false), TargetTriple(TT), InstrInfo(initializeSubtargetDependencies(CPU, FS)), TLInfo(TM, *this), TSInfo(*TM.getDataLayout()), FrameLowering() {} diff --git a/lib/Target/SystemZ/SystemZSubtarget.h b/lib/Target/SystemZ/SystemZSubtarget.h index 99cb1ad..c99e552 100644 --- a/lib/Target/SystemZ/SystemZSubtarget.h +++ b/lib/Target/SystemZ/SystemZSubtarget.h @@ -38,8 +38,12 @@ protected: bool HasLoadStoreOnCond; bool HasHighWord; bool HasFPExtension; + bool HasPopulationCount; bool HasFastSerialization; bool HasInterlockedAccess1; + bool HasMiscellaneousExtensions; + bool HasTransactionalExecution; + bool HasProcessorAssist; private: Triple TargetTriple; @@ -86,12 +90,26 @@ public: // Return true if the target has the floating-point extension facility. bool hasFPExtension() const { return HasFPExtension; } + // Return true if the target has the population-count facility. + bool hasPopulationCount() const { return HasPopulationCount; } + // Return true if the target has the fast-serialization facility. bool hasFastSerialization() const { return HasFastSerialization; } // Return true if the target has interlocked-access facility 1. bool hasInterlockedAccess1() const { return HasInterlockedAccess1; } + // Return true if the target has the miscellaneous-extensions facility. + bool hasMiscellaneousExtensions() const { + return HasMiscellaneousExtensions; + } + + // Return true if the target has the transactional-execution facility. + bool hasTransactionalExecution() const { return HasTransactionalExecution; } + + // Return true if the target has the processor-assist facility. + bool hasProcessorAssist() const { return HasProcessorAssist; } + // Return true if GV can be accessed using LARL for reloc model RM // and code model CM. bool isPC32DBLSymbol(const GlobalValue *GV, Reloc::Model RM, diff --git a/lib/Target/SystemZ/SystemZTargetMachine.cpp b/lib/Target/SystemZ/SystemZTargetMachine.cpp index 86baccb..b2f8175 100644 --- a/lib/Target/SystemZ/SystemZTargetMachine.cpp +++ b/lib/Target/SystemZ/SystemZTargetMachine.cpp @@ -8,6 +8,7 @@ //===----------------------------------------------------------------------===// #include "SystemZTargetMachine.h" +#include "SystemZTargetTransformInfo.h" #include "llvm/CodeGen/Passes.h" #include "llvm/Support/TargetRegistry.h" #include "llvm/Transforms/Scalar.h" @@ -108,3 +109,9 @@ void SystemZPassConfig::addPreEmitPass() { TargetPassConfig *SystemZTargetMachine::createPassConfig(PassManagerBase &PM) { return new SystemZPassConfig(this, PM); } + +TargetIRAnalysis SystemZTargetMachine::getTargetIRAnalysis() { + return TargetIRAnalysis([this](Function &F) { + return TargetTransformInfo(SystemZTTIImpl(this, F)); + }); +} diff --git a/lib/Target/SystemZ/SystemZTargetMachine.h b/lib/Target/SystemZ/SystemZTargetMachine.h index 181b926..5ded07c 100644 --- a/lib/Target/SystemZ/SystemZTargetMachine.h +++ b/lib/Target/SystemZ/SystemZTargetMachine.h @@ -39,6 +39,7 @@ public: } // Override LLVMTargetMachine TargetPassConfig *createPassConfig(PassManagerBase &PM) override; + TargetIRAnalysis getTargetIRAnalysis() override; TargetLoweringObjectFile *getObjFileLowering() const override { return TLOF.get(); } diff --git a/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp b/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp new file mode 100644 index 0000000..3337f63 --- /dev/null +++ b/lib/Target/SystemZ/SystemZTargetTransformInfo.cpp @@ -0,0 +1,240 @@ +//===-- SystemZTargetTransformInfo.cpp - SystemZ-specific TTI -------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements a TargetTransformInfo analysis pass specific to the +// SystemZ target machine. It uses the target's detailed information to provide +// more precise answers to certain TTI queries, while letting the target +// independent and default TTI implementations handle the rest. +// +//===----------------------------------------------------------------------===// + +#include "SystemZTargetTransformInfo.h" +#include "llvm/Analysis/TargetTransformInfo.h" +#include "llvm/CodeGen/BasicTTIImpl.h" +#include "llvm/IR/IntrinsicInst.h" +#include "llvm/Support/Debug.h" +#include "llvm/Target/CostTable.h" +#include "llvm/Target/TargetLowering.h" +using namespace llvm; + +#define DEBUG_TYPE "systemztti" + +//===----------------------------------------------------------------------===// +// +// SystemZ cost model. +// +//===----------------------------------------------------------------------===// + +unsigned SystemZTTIImpl::getIntImmCost(const APInt &Imm, Type *Ty) { + assert(Ty->isIntegerTy()); + + unsigned BitSize = Ty->getPrimitiveSizeInBits(); + // There is no cost model for constants with a bit size of 0. Return TCC_Free + // here, so that constant hoisting will ignore this constant. + if (BitSize == 0) + return TTI::TCC_Free; + // No cost model for operations on integers larger than 64 bit implemented yet. + if (BitSize > 64) + return TTI::TCC_Free; + + if (Imm == 0) + return TTI::TCC_Free; + + if (Imm.getBitWidth() <= 64) { + // Constants loaded via lgfi. + if (isInt<32>(Imm.getSExtValue())) + return TTI::TCC_Basic; + // Constants loaded via llilf. + if (isUInt<32>(Imm.getZExtValue())) + return TTI::TCC_Basic; + // Constants loaded via llihf: + if ((Imm.getZExtValue() & 0xffffffff) == 0) + return TTI::TCC_Basic; + + return 2 * TTI::TCC_Basic; + } + + return 4 * TTI::TCC_Basic; +} + +unsigned SystemZTTIImpl::getIntImmCost(unsigned Opcode, unsigned Idx, + const APInt &Imm, Type *Ty) { + assert(Ty->isIntegerTy()); + + unsigned BitSize = Ty->getPrimitiveSizeInBits(); + // There is no cost model for constants with a bit size of 0. Return TCC_Free + // here, so that constant hoisting will ignore this constant. + if (BitSize == 0) + return TTI::TCC_Free; + // No cost model for operations on integers larger than 64 bit implemented yet. + if (BitSize > 64) + return TTI::TCC_Free; + + switch (Opcode) { + default: + return TTI::TCC_Free; + case Instruction::GetElementPtr: + // Always hoist the base address of a GetElementPtr. This prevents the + // creation of new constants for every base constant that gets constant + // folded with the offset. + if (Idx == 0) + return 2 * TTI::TCC_Basic; + return TTI::TCC_Free; + case Instruction::Store: + if (Idx == 0 && Imm.getBitWidth() <= 64) { + // Any 8-bit immediate store can by implemented via mvi. + if (BitSize == 8) + return TTI::TCC_Free; + // 16-bit immediate values can be stored via mvhhi/mvhi/mvghi. + if (isInt<16>(Imm.getSExtValue())) + return TTI::TCC_Free; + } + break; + case Instruction::ICmp: + if (Idx == 1 && Imm.getBitWidth() <= 64) { + // Comparisons against signed 32-bit immediates implemented via cgfi. + if (isInt<32>(Imm.getSExtValue())) + return TTI::TCC_Free; + // Comparisons against unsigned 32-bit immediates implemented via clgfi. + if (isUInt<32>(Imm.getZExtValue())) + return TTI::TCC_Free; + } + break; + case Instruction::Add: + case Instruction::Sub: + if (Idx == 1 && Imm.getBitWidth() <= 64) { + // We use algfi/slgfi to add/subtract 32-bit unsigned immediates. + if (isUInt<32>(Imm.getZExtValue())) + return TTI::TCC_Free; + // Or their negation, by swapping addition vs. subtraction. + if (isUInt<32>(-Imm.getSExtValue())) + return TTI::TCC_Free; + } + break; + case Instruction::Mul: + if (Idx == 1 && Imm.getBitWidth() <= 64) { + // We use msgfi to multiply by 32-bit signed immediates. + if (isInt<32>(Imm.getSExtValue())) + return TTI::TCC_Free; + } + break; + case Instruction::Or: + case Instruction::Xor: + if (Idx == 1 && Imm.getBitWidth() <= 64) { + // Masks supported by oilf/xilf. + if (isUInt<32>(Imm.getZExtValue())) + return TTI::TCC_Free; + // Masks supported by oihf/xihf. + if ((Imm.getZExtValue() & 0xffffffff) == 0) + return TTI::TCC_Free; + } + break; + case Instruction::And: + if (Idx == 1 && Imm.getBitWidth() <= 64) { + // Any 32-bit AND operation can by implemented via nilf. + if (BitSize <= 32) + return TTI::TCC_Free; + // 64-bit masks supported by nilf. + if (isUInt<32>(~Imm.getZExtValue())) + return TTI::TCC_Free; + // 64-bit masks supported by nilh. + if ((Imm.getZExtValue() & 0xffffffff) == 0xffffffff) + return TTI::TCC_Free; + // Some 64-bit AND operations can be implemented via risbg. + const SystemZInstrInfo *TII = ST->getInstrInfo(); + unsigned Start, End; + if (TII->isRxSBGMask(Imm.getZExtValue(), BitSize, Start, End)) + return TTI::TCC_Free; + } + break; + case Instruction::Shl: + case Instruction::LShr: + case Instruction::AShr: + // Always return TCC_Free for the shift value of a shift instruction. + if (Idx == 1) + return TTI::TCC_Free; + break; + case Instruction::UDiv: + case Instruction::SDiv: + case Instruction::URem: + case Instruction::SRem: + case Instruction::Trunc: + case Instruction::ZExt: + case Instruction::SExt: + case Instruction::IntToPtr: + case Instruction::PtrToInt: + case Instruction::BitCast: + case Instruction::PHI: + case Instruction::Call: + case Instruction::Select: + case Instruction::Ret: + case Instruction::Load: + break; + } + + return SystemZTTIImpl::getIntImmCost(Imm, Ty); +} + +unsigned SystemZTTIImpl::getIntImmCost(Intrinsic::ID IID, unsigned Idx, + const APInt &Imm, Type *Ty) { + assert(Ty->isIntegerTy()); + + unsigned BitSize = Ty->getPrimitiveSizeInBits(); + // There is no cost model for constants with a bit size of 0. Return TCC_Free + // here, so that constant hoisting will ignore this constant. + if (BitSize == 0) + return TTI::TCC_Free; + // No cost model for operations on integers larger than 64 bit implemented yet. + if (BitSize > 64) + return TTI::TCC_Free; + + switch (IID) { + default: + return TTI::TCC_Free; + case Intrinsic::sadd_with_overflow: + case Intrinsic::uadd_with_overflow: + case Intrinsic::ssub_with_overflow: + case Intrinsic::usub_with_overflow: + // These get expanded to include a normal addition/subtraction. + if (Idx == 1 && Imm.getBitWidth() <= 64) { + if (isUInt<32>(Imm.getZExtValue())) + return TTI::TCC_Free; + if (isUInt<32>(-Imm.getSExtValue())) + return TTI::TCC_Free; + } + break; + case Intrinsic::smul_with_overflow: + case Intrinsic::umul_with_overflow: + // These get expanded to include a normal multiplication. + if (Idx == 1 && Imm.getBitWidth() <= 64) { + if (isInt<32>(Imm.getSExtValue())) + return TTI::TCC_Free; + } + break; + case Intrinsic::experimental_stackmap: + if ((Idx < 2) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue()))) + return TTI::TCC_Free; + break; + case Intrinsic::experimental_patchpoint_void: + case Intrinsic::experimental_patchpoint_i64: + if ((Idx < 4) || (Imm.getBitWidth() <= 64 && isInt<64>(Imm.getSExtValue()))) + return TTI::TCC_Free; + break; + } + return SystemZTTIImpl::getIntImmCost(Imm, Ty); +} + +TargetTransformInfo::PopcntSupportKind +SystemZTTIImpl::getPopcntSupport(unsigned TyWidth) { + assert(isPowerOf2_32(TyWidth) && "Type width must be power of 2"); + if (ST->hasPopulationCount() && TyWidth <= 64) + return TTI::PSK_FastHardware; + return TTI::PSK_Software; +} + diff --git a/lib/Target/SystemZ/SystemZTargetTransformInfo.h b/lib/Target/SystemZ/SystemZTargetTransformInfo.h new file mode 100644 index 0000000..d498913 --- /dev/null +++ b/lib/Target/SystemZ/SystemZTargetTransformInfo.h @@ -0,0 +1,70 @@ +//===-- SystemZTargetTransformInfo.h - SystemZ-specific TTI ---------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_LIB_TARGET_SYSTEMZ_SYSTEMZTARGETTRANSFORMINFO_H +#define LLVM_LIB_TARGET_SYSTEMZ_SYSTEMZTARGETTRANSFORMINFO_H + +#include "SystemZTargetMachine.h" +#include "llvm/Analysis/TargetTransformInfo.h" +#include "llvm/CodeGen/BasicTTIImpl.h" + +namespace llvm { + +class SystemZTTIImpl : public BasicTTIImplBase { + typedef BasicTTIImplBase BaseT; + typedef TargetTransformInfo TTI; + friend BaseT; + + const SystemZSubtarget *ST; + const SystemZTargetLowering *TLI; + + const SystemZSubtarget *getST() const { return ST; } + const SystemZTargetLowering *getTLI() const { return TLI; } + +public: + explicit SystemZTTIImpl(const SystemZTargetMachine *TM, Function &F) + : BaseT(TM), ST(TM->getSubtargetImpl(F)), TLI(ST->getTargetLowering()) {} + + // Provide value semantics. MSVC requires that we spell all of these out. + SystemZTTIImpl(const SystemZTTIImpl &Arg) + : BaseT(static_cast(Arg)), ST(Arg.ST), TLI(Arg.TLI) {} + SystemZTTIImpl(SystemZTTIImpl &&Arg) + : BaseT(std::move(static_cast(Arg))), ST(std::move(Arg.ST)), + TLI(std::move(Arg.TLI)) {} + SystemZTTIImpl &operator=(const SystemZTTIImpl &RHS) { + BaseT::operator=(static_cast(RHS)); + ST = RHS.ST; + TLI = RHS.TLI; + return *this; + } + SystemZTTIImpl &operator=(SystemZTTIImpl &&RHS) { + BaseT::operator=(std::move(static_cast(RHS))); + ST = std::move(RHS.ST); + TLI = std::move(RHS.TLI); + return *this; + } + + /// \name Scalar TTI Implementations + /// @{ + + unsigned getIntImmCost(const APInt &Imm, Type *Ty); + + unsigned getIntImmCost(unsigned Opcode, unsigned Idx, const APInt &Imm, + Type *Ty); + unsigned getIntImmCost(Intrinsic::ID IID, unsigned Idx, const APInt &Imm, + Type *Ty); + + TTI::PopcntSupportKind getPopcntSupport(unsigned TyWidth); + + /// @} +}; + +} // end namespace llvm + +#endif -- cgit v1.1