PPCISelLowering.cpp [plain text]
#include "PPCISelLowering.h"
#include "MCTargetDesc/PPCPredicates.h"
#include "PPCMachineFunctionInfo.h"
#include "PPCPerfectShuffle.h"
#include "PPCTargetMachine.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/CodeGen/CallingConvLower.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/SelectionDAG.h"
#include "llvm/CodeGen/TargetLoweringObjectFileImpl.h"
#include "llvm/IR/CallingConv.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetOptions.h"
using namespace llvm;
static bool CC_PPC_SVR4_Custom_Dummy(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State);
static bool CC_PPC_SVR4_Custom_AlignArgRegs(unsigned &ValNo, MVT &ValVT,
MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State);
static bool CC_PPC_SVR4_Custom_AlignFPArgRegs(unsigned &ValNo, MVT &ValVT,
MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State);
static cl::opt<bool> DisablePPCPreinc("disable-ppc-preinc",
cl::desc("disable preincrement load/store generation on PPC"), cl::Hidden);
static cl::opt<bool> DisableILPPref("disable-ppc-ilp-pref",
cl::desc("disable setting the node scheduling preference to ILP on PPC"), cl::Hidden);
static TargetLoweringObjectFile *CreateTLOF(const PPCTargetMachine &TM) {
if (TM.getSubtargetImpl()->isDarwin())
return new TargetLoweringObjectFileMachO();
return new TargetLoweringObjectFileELF();
}
PPCTargetLowering::PPCTargetLowering(PPCTargetMachine &TM)
: TargetLowering(TM, CreateTLOF(TM)), PPCSubTarget(*TM.getSubtargetImpl()) {
const PPCSubtarget *Subtarget = &TM.getSubtarget<PPCSubtarget>();
setPow2DivIsCheap();
setUseUnderscoreSetJmp(true);
setUseUnderscoreLongJmp(true);
bool isPPC64 = Subtarget->isPPC64();
setMinStackArgumentAlignment(isPPC64 ? 8:4);
addRegisterClass(MVT::i32, &PPC::GPRCRegClass);
addRegisterClass(MVT::f32, &PPC::F4RCRegClass);
addRegisterClass(MVT::f64, &PPC::F8RCRegClass);
setLoadExtAction(ISD::SEXTLOAD, MVT::i1, Promote);
setLoadExtAction(ISD::SEXTLOAD, MVT::i8, Expand);
setTruncStoreAction(MVT::f64, MVT::f32, Expand);
setIndexedLoadAction(ISD::PRE_INC, MVT::i1, Legal);
setIndexedLoadAction(ISD::PRE_INC, MVT::i8, Legal);
setIndexedLoadAction(ISD::PRE_INC, MVT::i16, Legal);
setIndexedLoadAction(ISD::PRE_INC, MVT::i32, Legal);
setIndexedLoadAction(ISD::PRE_INC, MVT::i64, Legal);
setIndexedStoreAction(ISD::PRE_INC, MVT::i1, Legal);
setIndexedStoreAction(ISD::PRE_INC, MVT::i8, Legal);
setIndexedStoreAction(ISD::PRE_INC, MVT::i16, Legal);
setIndexedStoreAction(ISD::PRE_INC, MVT::i32, Legal);
setIndexedStoreAction(ISD::PRE_INC, MVT::i64, Legal);
setOperationAction(ISD::FP_ROUND_INREG, MVT::ppcf128, Custom);
setOperationAction(ISD::FFLOOR, MVT::ppcf128, Expand);
setOperationAction(ISD::FCEIL, MVT::ppcf128, Expand);
setOperationAction(ISD::FTRUNC, MVT::ppcf128, Expand);
setOperationAction(ISD::FRINT, MVT::ppcf128, Expand);
setOperationAction(ISD::FNEARBYINT, MVT::ppcf128, Expand);
setOperationAction(ISD::SREM, MVT::i32, Expand);
setOperationAction(ISD::UREM, MVT::i32, Expand);
setOperationAction(ISD::SREM, MVT::i64, Expand);
setOperationAction(ISD::UREM, MVT::i64, Expand);
setOperationAction(ISD::UMUL_LOHI, MVT::i32, Expand);
setOperationAction(ISD::SMUL_LOHI, MVT::i32, Expand);
setOperationAction(ISD::UMUL_LOHI, MVT::i64, Expand);
setOperationAction(ISD::SMUL_LOHI, MVT::i64, Expand);
setOperationAction(ISD::UDIVREM, MVT::i32, Expand);
setOperationAction(ISD::SDIVREM, MVT::i32, Expand);
setOperationAction(ISD::UDIVREM, MVT::i64, Expand);
setOperationAction(ISD::SDIVREM, MVT::i64, Expand);
setOperationAction(ISD::FSIN , MVT::f64, Expand);
setOperationAction(ISD::FCOS , MVT::f64, Expand);
setOperationAction(ISD::FSINCOS, MVT::f64, Expand);
setOperationAction(ISD::FREM , MVT::f64, Expand);
setOperationAction(ISD::FPOW , MVT::f64, Expand);
setOperationAction(ISD::FMA , MVT::f64, Legal);
setOperationAction(ISD::FSIN , MVT::f32, Expand);
setOperationAction(ISD::FCOS , MVT::f32, Expand);
setOperationAction(ISD::FSINCOS, MVT::f32, Expand);
setOperationAction(ISD::FREM , MVT::f32, Expand);
setOperationAction(ISD::FPOW , MVT::f32, Expand);
setOperationAction(ISD::FMA , MVT::f32, Legal);
setOperationAction(ISD::FLT_ROUNDS_, MVT::i32, Custom);
if (!Subtarget->hasFSQRT()) {
setOperationAction(ISD::FSQRT, MVT::f64, Expand);
setOperationAction(ISD::FSQRT, MVT::f32, Expand);
}
setOperationAction(ISD::FCOPYSIGN, MVT::f64, Expand);
setOperationAction(ISD::FCOPYSIGN, MVT::f32, Expand);
setOperationAction(ISD::BSWAP, MVT::i32 , Expand);
setOperationAction(ISD::CTPOP, MVT::i32 , Expand);
setOperationAction(ISD::CTTZ , MVT::i32 , Expand);
setOperationAction(ISD::CTTZ_ZERO_UNDEF, MVT::i32, Expand);
setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i32, Expand);
setOperationAction(ISD::BSWAP, MVT::i64 , Expand);
setOperationAction(ISD::CTPOP, MVT::i64 , Expand);
setOperationAction(ISD::CTTZ , MVT::i64 , Expand);
setOperationAction(ISD::CTTZ_ZERO_UNDEF, MVT::i64, Expand);
setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, Expand);
setOperationAction(ISD::ROTR, MVT::i32 , Expand);
setOperationAction(ISD::ROTR, MVT::i64 , Expand);
setOperationAction(ISD::SELECT, MVT::i32, Expand);
setOperationAction(ISD::SELECT, MVT::i64, Expand);
setOperationAction(ISD::SELECT, MVT::f32, Expand);
setOperationAction(ISD::SELECT, MVT::f64, Expand);
setOperationAction(ISD::SELECT_CC, MVT::f32, Custom);
setOperationAction(ISD::SELECT_CC, MVT::f64, Custom);
setOperationAction(ISD::SETCC, MVT::i32, Custom);
setOperationAction(ISD::BRCOND, MVT::Other, Expand);
setOperationAction(ISD::BR_JT, MVT::Other, Expand);
setOperationAction(ISD::FP_TO_SINT, MVT::i32, Custom);
setOperationAction(ISD::SINT_TO_FP, MVT::i32, Expand);
setOperationAction(ISD::UINT_TO_FP, MVT::i32, Expand);
setOperationAction(ISD::BITCAST, MVT::f32, Expand);
setOperationAction(ISD::BITCAST, MVT::i32, Expand);
setOperationAction(ISD::BITCAST, MVT::i64, Expand);
setOperationAction(ISD::BITCAST, MVT::f64, Expand);
setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i1, Expand);
setOperationAction(ISD::EXCEPTIONADDR, MVT::i64, Expand);
setOperationAction(ISD::EHSELECTION, MVT::i64, Expand);
setOperationAction(ISD::EXCEPTIONADDR, MVT::i32, Expand);
setOperationAction(ISD::EHSELECTION, MVT::i32, Expand);
setOperationAction(ISD::GlobalAddress, MVT::i32, Custom);
setOperationAction(ISD::GlobalTLSAddress, MVT::i32, Custom);
setOperationAction(ISD::BlockAddress, MVT::i32, Custom);
setOperationAction(ISD::ConstantPool, MVT::i32, Custom);
setOperationAction(ISD::JumpTable, MVT::i32, Custom);
setOperationAction(ISD::GlobalAddress, MVT::i64, Custom);
setOperationAction(ISD::GlobalTLSAddress, MVT::i64, Custom);
setOperationAction(ISD::BlockAddress, MVT::i64, Custom);
setOperationAction(ISD::ConstantPool, MVT::i64, Custom);
setOperationAction(ISD::JumpTable, MVT::i64, Custom);
setOperationAction(ISD::TRAP, MVT::Other, Legal);
setOperationAction(ISD::INIT_TRAMPOLINE, MVT::Other, Custom);
setOperationAction(ISD::ADJUST_TRAMPOLINE, MVT::Other, Custom);
setOperationAction(ISD::VASTART , MVT::Other, Custom);
if (Subtarget->isSVR4ABI()) {
if (isPPC64) {
setOperationAction(ISD::VAARG, MVT::i1, Promote);
AddPromotedToType (ISD::VAARG, MVT::i1, MVT::i64);
setOperationAction(ISD::VAARG, MVT::i8, Promote);
AddPromotedToType (ISD::VAARG, MVT::i8, MVT::i64);
setOperationAction(ISD::VAARG, MVT::i16, Promote);
AddPromotedToType (ISD::VAARG, MVT::i16, MVT::i64);
setOperationAction(ISD::VAARG, MVT::i32, Promote);
AddPromotedToType (ISD::VAARG, MVT::i32, MVT::i64);
setOperationAction(ISD::VAARG, MVT::Other, Expand);
} else {
setOperationAction(ISD::VAARG, MVT::Other, Custom);
setOperationAction(ISD::VAARG, MVT::i64, Custom);
}
} else
setOperationAction(ISD::VAARG, MVT::Other, Expand);
setOperationAction(ISD::VACOPY , MVT::Other, Expand);
setOperationAction(ISD::VAEND , MVT::Other, Expand);
setOperationAction(ISD::STACKSAVE , MVT::Other, Expand);
setOperationAction(ISD::STACKRESTORE , MVT::Other, Custom);
setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i32 , Custom);
setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i64 , Custom);
setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::Other, Custom);
setCondCodeAction(ISD::SETULT, MVT::f32, Expand);
setCondCodeAction(ISD::SETULT, MVT::f64, Expand);
setCondCodeAction(ISD::SETUGT, MVT::f32, Expand);
setCondCodeAction(ISD::SETUGT, MVT::f64, Expand);
setCondCodeAction(ISD::SETUEQ, MVT::f32, Expand);
setCondCodeAction(ISD::SETUEQ, MVT::f64, Expand);
setCondCodeAction(ISD::SETOGE, MVT::f32, Expand);
setCondCodeAction(ISD::SETOGE, MVT::f64, Expand);
setCondCodeAction(ISD::SETOLE, MVT::f32, Expand);
setCondCodeAction(ISD::SETOLE, MVT::f64, Expand);
setCondCodeAction(ISD::SETONE, MVT::f32, Expand);
setCondCodeAction(ISD::SETONE, MVT::f64, Expand);
if (Subtarget->has64BitSupport()) {
setOperationAction(ISD::FP_TO_SINT, MVT::i64, Custom);
setOperationAction(ISD::FP_TO_UINT, MVT::i64, Expand);
setOperationAction(ISD::SINT_TO_FP, MVT::i64, Custom);
setOperationAction(ISD::UINT_TO_FP, MVT::i64, Expand);
setOperationAction(ISD::FP_TO_UINT, MVT::i32, Custom);
} else {
setOperationAction(ISD::FP_TO_UINT, MVT::i32, Expand);
}
if (Subtarget->use64BitRegs()) {
addRegisterClass(MVT::i64, &PPC::G8RCRegClass);
setOperationAction(ISD::BUILD_PAIR, MVT::i64, Expand);
setOperationAction(ISD::SHL_PARTS, MVT::i64, Custom);
setOperationAction(ISD::SRA_PARTS, MVT::i64, Custom);
setOperationAction(ISD::SRL_PARTS, MVT::i64, Custom);
} else {
setOperationAction(ISD::SHL_PARTS, MVT::i32, Custom);
setOperationAction(ISD::SRA_PARTS, MVT::i32, Custom);
setOperationAction(ISD::SRL_PARTS, MVT::i32, Custom);
}
if (Subtarget->hasAltivec()) {
for (unsigned i = (unsigned)MVT::FIRST_VECTOR_VALUETYPE;
i <= (unsigned)MVT::LAST_VECTOR_VALUETYPE; ++i) {
MVT::SimpleValueType VT = (MVT::SimpleValueType)i;
setOperationAction(ISD::ADD , VT, Legal);
setOperationAction(ISD::SUB , VT, Legal);
setOperationAction(ISD::VECTOR_SHUFFLE, VT, Promote);
AddPromotedToType (ISD::VECTOR_SHUFFLE, VT, MVT::v16i8);
setOperationAction(ISD::AND , VT, Promote);
AddPromotedToType (ISD::AND , VT, MVT::v4i32);
setOperationAction(ISD::OR , VT, Promote);
AddPromotedToType (ISD::OR , VT, MVT::v4i32);
setOperationAction(ISD::XOR , VT, Promote);
AddPromotedToType (ISD::XOR , VT, MVT::v4i32);
setOperationAction(ISD::LOAD , VT, Promote);
AddPromotedToType (ISD::LOAD , VT, MVT::v4i32);
setOperationAction(ISD::SELECT, VT, Promote);
AddPromotedToType (ISD::SELECT, VT, MVT::v4i32);
setOperationAction(ISD::STORE, VT, Promote);
AddPromotedToType (ISD::STORE, VT, MVT::v4i32);
setOperationAction(ISD::MUL , VT, Expand);
setOperationAction(ISD::SDIV, VT, Expand);
setOperationAction(ISD::SREM, VT, Expand);
setOperationAction(ISD::UDIV, VT, Expand);
setOperationAction(ISD::UREM, VT, Expand);
setOperationAction(ISD::FDIV, VT, Expand);
setOperationAction(ISD::FNEG, VT, Expand);
setOperationAction(ISD::FSQRT, VT, Expand);
setOperationAction(ISD::FLOG, VT, Expand);
setOperationAction(ISD::FLOG10, VT, Expand);
setOperationAction(ISD::FLOG2, VT, Expand);
setOperationAction(ISD::FEXP, VT, Expand);
setOperationAction(ISD::FEXP2, VT, Expand);
setOperationAction(ISD::FSIN, VT, Expand);
setOperationAction(ISD::FCOS, VT, Expand);
setOperationAction(ISD::FABS, VT, Expand);
setOperationAction(ISD::FPOWI, VT, Expand);
setOperationAction(ISD::FFLOOR, VT, Expand);
setOperationAction(ISD::FCEIL, VT, Expand);
setOperationAction(ISD::FTRUNC, VT, Expand);
setOperationAction(ISD::FRINT, VT, Expand);
setOperationAction(ISD::FNEARBYINT, VT, Expand);
setOperationAction(ISD::EXTRACT_VECTOR_ELT, VT, Expand);
setOperationAction(ISD::INSERT_VECTOR_ELT, VT, Expand);
setOperationAction(ISD::BUILD_VECTOR, VT, Expand);
setOperationAction(ISD::UMUL_LOHI, VT, Expand);
setOperationAction(ISD::SMUL_LOHI, VT, Expand);
setOperationAction(ISD::UDIVREM, VT, Expand);
setOperationAction(ISD::SDIVREM, VT, Expand);
setOperationAction(ISD::SCALAR_TO_VECTOR, VT, Expand);
setOperationAction(ISD::FPOW, VT, Expand);
setOperationAction(ISD::CTPOP, VT, Expand);
setOperationAction(ISD::CTLZ, VT, Expand);
setOperationAction(ISD::CTLZ_ZERO_UNDEF, VT, Expand);
setOperationAction(ISD::CTTZ, VT, Expand);
setOperationAction(ISD::CTTZ_ZERO_UNDEF, VT, Expand);
setOperationAction(ISD::VSELECT, VT, Expand);
setOperationAction(ISD::SIGN_EXTEND_INREG, VT, Expand);
for (unsigned j = (unsigned)MVT::FIRST_VECTOR_VALUETYPE;
j <= (unsigned)MVT::LAST_VECTOR_VALUETYPE; ++j) {
MVT::SimpleValueType InnerVT = (MVT::SimpleValueType)j;
setTruncStoreAction(VT, InnerVT, Expand);
}
setLoadExtAction(ISD::SEXTLOAD, VT, Expand);
setLoadExtAction(ISD::ZEXTLOAD, VT, Expand);
setLoadExtAction(ISD::EXTLOAD, VT, Expand);
}
setOperationAction(ISD::VECTOR_SHUFFLE, MVT::v16i8, Custom);
setOperationAction(ISD::AND , MVT::v4i32, Legal);
setOperationAction(ISD::OR , MVT::v4i32, Legal);
setOperationAction(ISD::XOR , MVT::v4i32, Legal);
setOperationAction(ISD::LOAD , MVT::v4i32, Legal);
setOperationAction(ISD::SELECT, MVT::v4i32, Expand);
setOperationAction(ISD::STORE , MVT::v4i32, Legal);
setOperationAction(ISD::FP_TO_SINT, MVT::v4i32, Legal);
setOperationAction(ISD::FP_TO_UINT, MVT::v4i32, Legal);
setOperationAction(ISD::SINT_TO_FP, MVT::v4i32, Legal);
setOperationAction(ISD::UINT_TO_FP, MVT::v4i32, Legal);
setOperationAction(ISD::FFLOOR, MVT::v4f32, Legal);
setOperationAction(ISD::FCEIL, MVT::v4f32, Legal);
setOperationAction(ISD::FTRUNC, MVT::v4f32, Legal);
setOperationAction(ISD::FNEARBYINT, MVT::v4f32, Legal);
addRegisterClass(MVT::v4f32, &PPC::VRRCRegClass);
addRegisterClass(MVT::v4i32, &PPC::VRRCRegClass);
addRegisterClass(MVT::v8i16, &PPC::VRRCRegClass);
addRegisterClass(MVT::v16i8, &PPC::VRRCRegClass);
setOperationAction(ISD::MUL, MVT::v4f32, Legal);
setOperationAction(ISD::FMA, MVT::v4f32, Legal);
setOperationAction(ISD::MUL, MVT::v4i32, Custom);
setOperationAction(ISD::MUL, MVT::v8i16, Custom);
setOperationAction(ISD::MUL, MVT::v16i8, Custom);
setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v4f32, Custom);
setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v4i32, Custom);
setOperationAction(ISD::BUILD_VECTOR, MVT::v16i8, Custom);
setOperationAction(ISD::BUILD_VECTOR, MVT::v8i16, Custom);
setOperationAction(ISD::BUILD_VECTOR, MVT::v4i32, Custom);
setOperationAction(ISD::BUILD_VECTOR, MVT::v4f32, Custom);
setCondCodeAction(ISD::SETUO, MVT::v4f32, Expand);
setCondCodeAction(ISD::SETUEQ, MVT::v4f32, Expand);
setCondCodeAction(ISD::SETUGT, MVT::v4f32, Expand);
setCondCodeAction(ISD::SETUGE, MVT::v4f32, Expand);
setCondCodeAction(ISD::SETULT, MVT::v4f32, Expand);
setCondCodeAction(ISD::SETULE, MVT::v4f32, Expand);
}
if (Subtarget->has64BitSupport()) {
setOperationAction(ISD::PREFETCH, MVT::Other, Legal);
setOperationAction(ISD::READCYCLECOUNTER, MVT::i64, Legal);
}
setOperationAction(ISD::ATOMIC_LOAD, MVT::i32, Expand);
setOperationAction(ISD::ATOMIC_STORE, MVT::i32, Expand);
setOperationAction(ISD::ATOMIC_LOAD, MVT::i64, Expand);
setOperationAction(ISD::ATOMIC_STORE, MVT::i64, Expand);
setBooleanContents(ZeroOrOneBooleanContent);
setBooleanVectorContents(ZeroOrOneBooleanContent);
if (isPPC64) {
setStackPointerRegisterToSaveRestore(PPC::X1);
setExceptionPointerRegister(PPC::X3);
setExceptionSelectorRegister(PPC::X4);
} else {
setStackPointerRegisterToSaveRestore(PPC::R1);
setExceptionPointerRegister(PPC::R3);
setExceptionSelectorRegister(PPC::R4);
}
setTargetDAGCombine(ISD::SINT_TO_FP);
setTargetDAGCombine(ISD::STORE);
setTargetDAGCombine(ISD::BR_CC);
setTargetDAGCombine(ISD::BSWAP);
if (Subtarget->isDarwin()) {
setLibcallName(RTLIB::COS_PPCF128, "cosl$LDBL128");
setLibcallName(RTLIB::POW_PPCF128, "powl$LDBL128");
setLibcallName(RTLIB::REM_PPCF128, "fmodl$LDBL128");
setLibcallName(RTLIB::SIN_PPCF128, "sinl$LDBL128");
setLibcallName(RTLIB::SQRT_PPCF128, "sqrtl$LDBL128");
setLibcallName(RTLIB::LOG_PPCF128, "logl$LDBL128");
setLibcallName(RTLIB::LOG2_PPCF128, "log2l$LDBL128");
setLibcallName(RTLIB::LOG10_PPCF128, "log10l$LDBL128");
setLibcallName(RTLIB::EXP_PPCF128, "expl$LDBL128");
setLibcallName(RTLIB::EXP2_PPCF128, "exp2l$LDBL128");
}
setMinFunctionAlignment(2);
if (PPCSubTarget.isDarwin())
setPrefFunctionAlignment(4);
if (isPPC64 && Subtarget->isJITCodeModel())
setSupportJumpTables(false);
setInsertFencesForAtomic(true);
setSchedulingPreference(Sched::Hybrid);
computeRegisterProperties();
if (Subtarget->getDarwinDirective() == PPC::DIR_E500mc ||
Subtarget->getDarwinDirective() == PPC::DIR_E5500) {
maxStoresPerMemset = 32;
maxStoresPerMemsetOptSize = 16;
maxStoresPerMemcpy = 32;
maxStoresPerMemcpyOptSize = 8;
maxStoresPerMemmove = 32;
maxStoresPerMemmoveOptSize = 8;
setPrefFunctionAlignment(4);
benefitFromCodePlacementOpt = true;
}
}
unsigned PPCTargetLowering::getByValTypeAlignment(Type *Ty) const {
const TargetMachine &TM = getTargetMachine();
if (TM.getSubtarget<PPCSubtarget>().isDarwin())
return 4;
if (VectorType *VTy = dyn_cast<VectorType>(Ty))
if (VTy->getBitWidth() >= 128)
return 16;
if (PPCSubTarget.isPPC64())
return 8;
return 4;
}
const char *PPCTargetLowering::getTargetNodeName(unsigned Opcode) const {
switch (Opcode) {
default: return 0;
case PPCISD::FSEL: return "PPCISD::FSEL";
case PPCISD::FCFID: return "PPCISD::FCFID";
case PPCISD::FCTIDZ: return "PPCISD::FCTIDZ";
case PPCISD::FCTIWZ: return "PPCISD::FCTIWZ";
case PPCISD::STFIWX: return "PPCISD::STFIWX";
case PPCISD::VMADDFP: return "PPCISD::VMADDFP";
case PPCISD::VNMSUBFP: return "PPCISD::VNMSUBFP";
case PPCISD::VPERM: return "PPCISD::VPERM";
case PPCISD::Hi: return "PPCISD::Hi";
case PPCISD::Lo: return "PPCISD::Lo";
case PPCISD::TOC_ENTRY: return "PPCISD::TOC_ENTRY";
case PPCISD::TOC_RESTORE: return "PPCISD::TOC_RESTORE";
case PPCISD::LOAD: return "PPCISD::LOAD";
case PPCISD::LOAD_TOC: return "PPCISD::LOAD_TOC";
case PPCISD::DYNALLOC: return "PPCISD::DYNALLOC";
case PPCISD::GlobalBaseReg: return "PPCISD::GlobalBaseReg";
case PPCISD::SRL: return "PPCISD::SRL";
case PPCISD::SRA: return "PPCISD::SRA";
case PPCISD::SHL: return "PPCISD::SHL";
case PPCISD::EXTSW_32: return "PPCISD::EXTSW_32";
case PPCISD::STD_32: return "PPCISD::STD_32";
case PPCISD::CALL_SVR4: return "PPCISD::CALL_SVR4";
case PPCISD::CALL_NOP_SVR4: return "PPCISD::CALL_NOP_SVR4";
case PPCISD::CALL_Darwin: return "PPCISD::CALL_Darwin";
case PPCISD::NOP: return "PPCISD::NOP";
case PPCISD::MTCTR: return "PPCISD::MTCTR";
case PPCISD::BCTRL_Darwin: return "PPCISD::BCTRL_Darwin";
case PPCISD::BCTRL_SVR4: return "PPCISD::BCTRL_SVR4";
case PPCISD::RET_FLAG: return "PPCISD::RET_FLAG";
case PPCISD::MFCR: return "PPCISD::MFCR";
case PPCISD::VCMP: return "PPCISD::VCMP";
case PPCISD::VCMPo: return "PPCISD::VCMPo";
case PPCISD::LBRX: return "PPCISD::LBRX";
case PPCISD::STBRX: return "PPCISD::STBRX";
case PPCISD::LARX: return "PPCISD::LARX";
case PPCISD::STCX: return "PPCISD::STCX";
case PPCISD::COND_BRANCH: return "PPCISD::COND_BRANCH";
case PPCISD::MFFS: return "PPCISD::MFFS";
case PPCISD::MTFSB0: return "PPCISD::MTFSB0";
case PPCISD::MTFSB1: return "PPCISD::MTFSB1";
case PPCISD::FADDRTZ: return "PPCISD::FADDRTZ";
case PPCISD::MTFSF: return "PPCISD::MTFSF";
case PPCISD::TC_RETURN: return "PPCISD::TC_RETURN";
case PPCISD::CR6SET: return "PPCISD::CR6SET";
case PPCISD::CR6UNSET: return "PPCISD::CR6UNSET";
case PPCISD::ADDIS_TOC_HA: return "PPCISD::ADDIS_TOC_HA";
case PPCISD::LD_TOC_L: return "PPCISD::LD_TOC_L";
case PPCISD::ADDI_TOC_L: return "PPCISD::ADDI_TOC_L";
case PPCISD::ADDIS_GOT_TPREL_HA: return "PPCISD::ADDIS_GOT_TPREL_HA";
case PPCISD::LD_GOT_TPREL_L: return "PPCISD::LD_GOT_TPREL_L";
case PPCISD::ADD_TLS: return "PPCISD::ADD_TLS";
case PPCISD::ADDIS_TLSGD_HA: return "PPCISD::ADDIS_TLSGD_HA";
case PPCISD::ADDI_TLSGD_L: return "PPCISD::ADDI_TLSGD_L";
case PPCISD::GET_TLS_ADDR: return "PPCISD::GET_TLS_ADDR";
case PPCISD::ADDIS_TLSLD_HA: return "PPCISD::ADDIS_TLSLD_HA";
case PPCISD::ADDI_TLSLD_L: return "PPCISD::ADDI_TLSLD_L";
case PPCISD::GET_TLSLD_ADDR: return "PPCISD::GET_TLSLD_ADDR";
case PPCISD::ADDIS_DTPREL_HA: return "PPCISD::ADDIS_DTPREL_HA";
case PPCISD::ADDI_DTPREL_L: return "PPCISD::ADDI_DTPREL_L";
}
}
EVT PPCTargetLowering::getSetCCResultType(EVT VT) const {
if (!VT.isVector())
return MVT::i32;
return VT.changeVectorElementTypeToInteger();
}
static bool isFloatingPointZero(SDValue Op) {
if (ConstantFPSDNode *CFP = dyn_cast<ConstantFPSDNode>(Op))
return CFP->getValueAPF().isZero();
else if (ISD::isEXTLoad(Op.getNode()) || ISD::isNON_EXTLoad(Op.getNode())) {
if (ConstantPoolSDNode *CP = dyn_cast<ConstantPoolSDNode>(Op.getOperand(1)))
if (const ConstantFP *CFP = dyn_cast<ConstantFP>(CP->getConstVal()))
return CFP->getValueAPF().isZero();
}
return false;
}
static bool isConstantOrUndef(int Op, int Val) {
return Op < 0 || Op == Val;
}
bool PPC::isVPKUHUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary) {
if (!isUnary) {
for (unsigned i = 0; i != 16; ++i)
if (!isConstantOrUndef(N->getMaskElt(i), i*2+1))
return false;
} else {
for (unsigned i = 0; i != 8; ++i)
if (!isConstantOrUndef(N->getMaskElt(i), i*2+1) ||
!isConstantOrUndef(N->getMaskElt(i+8), i*2+1))
return false;
}
return true;
}
bool PPC::isVPKUWUMShuffleMask(ShuffleVectorSDNode *N, bool isUnary) {
if (!isUnary) {
for (unsigned i = 0; i != 16; i += 2)
if (!isConstantOrUndef(N->getMaskElt(i ), i*2+2) ||
!isConstantOrUndef(N->getMaskElt(i+1), i*2+3))
return false;
} else {
for (unsigned i = 0; i != 8; i += 2)
if (!isConstantOrUndef(N->getMaskElt(i ), i*2+2) ||
!isConstantOrUndef(N->getMaskElt(i+1), i*2+3) ||
!isConstantOrUndef(N->getMaskElt(i+8), i*2+2) ||
!isConstantOrUndef(N->getMaskElt(i+9), i*2+3))
return false;
}
return true;
}
static bool isVMerge(ShuffleVectorSDNode *N, unsigned UnitSize,
unsigned LHSStart, unsigned RHSStart) {
assert(N->getValueType(0) == MVT::v16i8 &&
"PPC only supports shuffles by bytes!");
assert((UnitSize == 1 || UnitSize == 2 || UnitSize == 4) &&
"Unsupported merge size!");
for (unsigned i = 0; i != 8/UnitSize; ++i) for (unsigned j = 0; j != UnitSize; ++j) { if (!isConstantOrUndef(N->getMaskElt(i*UnitSize*2+j),
LHSStart+j+i*UnitSize) ||
!isConstantOrUndef(N->getMaskElt(i*UnitSize*2+UnitSize+j),
RHSStart+j+i*UnitSize))
return false;
}
return true;
}
bool PPC::isVMRGLShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
bool isUnary) {
if (!isUnary)
return isVMerge(N, UnitSize, 8, 24);
return isVMerge(N, UnitSize, 8, 8);
}
bool PPC::isVMRGHShuffleMask(ShuffleVectorSDNode *N, unsigned UnitSize,
bool isUnary) {
if (!isUnary)
return isVMerge(N, UnitSize, 0, 16);
return isVMerge(N, UnitSize, 0, 0);
}
int PPC::isVSLDOIShuffleMask(SDNode *N, bool isUnary) {
assert(N->getValueType(0) == MVT::v16i8 &&
"PPC only supports shuffles by bytes!");
ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
unsigned i;
for (i = 0; i != 16 && SVOp->getMaskElt(i) < 0; ++i)
;
if (i == 16) return -1;
unsigned ShiftAmt = SVOp->getMaskElt(i);
if (ShiftAmt < i) return -1;
ShiftAmt -= i;
if (!isUnary) {
for (++i; i != 16; ++i)
if (!isConstantOrUndef(SVOp->getMaskElt(i), ShiftAmt+i))
return -1;
} else {
for (++i; i != 16; ++i)
if (!isConstantOrUndef(SVOp->getMaskElt(i), (ShiftAmt+i) & 15))
return -1;
}
return ShiftAmt;
}
bool PPC::isSplatShuffleMask(ShuffleVectorSDNode *N, unsigned EltSize) {
assert(N->getValueType(0) == MVT::v16i8 &&
(EltSize == 1 || EltSize == 2 || EltSize == 4));
unsigned ElementBase = N->getMaskElt(0);
if (ElementBase >= 16)
return false;
for (unsigned i = 1; i != EltSize; ++i)
if (N->getMaskElt(i) < 0 || N->getMaskElt(i) != (int)(i+ElementBase))
return false;
for (unsigned i = EltSize, e = 16; i != e; i += EltSize) {
if (N->getMaskElt(i) < 0) continue;
for (unsigned j = 0; j != EltSize; ++j)
if (N->getMaskElt(i+j) != N->getMaskElt(j))
return false;
}
return true;
}
bool PPC::isAllNegativeZeroVector(SDNode *N) {
BuildVectorSDNode *BV = cast<BuildVectorSDNode>(N);
APInt APVal, APUndef;
unsigned BitSize;
bool HasAnyUndefs;
if (BV->isConstantSplat(APVal, APUndef, BitSize, HasAnyUndefs, 32, true))
if (ConstantFPSDNode *CFP = dyn_cast<ConstantFPSDNode>(N->getOperand(0)))
return CFP->getValueAPF().isNegZero();
return false;
}
unsigned PPC::getVSPLTImmediate(SDNode *N, unsigned EltSize) {
ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(N);
assert(isSplatShuffleMask(SVOp, EltSize));
return SVOp->getMaskElt(0) / EltSize;
}
SDValue PPC::get_VSPLTI_elt(SDNode *N, unsigned ByteSize, SelectionDAG &DAG) {
SDValue OpVal(0, 0);
unsigned EltSize = 16/N->getNumOperands();
if (EltSize < ByteSize) {
unsigned Multiple = ByteSize/EltSize; SDValue UniquedVals[4];
assert(Multiple > 1 && Multiple <= 4 && "How can this happen?");
for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
if (N->getOperand(i).getOpcode() == ISD::UNDEF) continue;
if (!isa<ConstantSDNode>(N->getOperand(i))) return SDValue();
if (UniquedVals[i&(Multiple-1)].getNode() == 0)
UniquedVals[i&(Multiple-1)] = N->getOperand(i);
else if (UniquedVals[i&(Multiple-1)] != N->getOperand(i))
return SDValue(); }
bool LeadingZero = true;
bool LeadingOnes = true;
for (unsigned i = 0; i != Multiple-1; ++i) {
if (UniquedVals[i].getNode() == 0) continue;
LeadingZero &= cast<ConstantSDNode>(UniquedVals[i])->isNullValue();
LeadingOnes &= cast<ConstantSDNode>(UniquedVals[i])->isAllOnesValue();
}
if (LeadingZero) {
if (UniquedVals[Multiple-1].getNode() == 0)
return DAG.getTargetConstant(0, MVT::i32); int Val = cast<ConstantSDNode>(UniquedVals[Multiple-1])->getZExtValue();
if (Val < 16)
return DAG.getTargetConstant(Val, MVT::i32); }
if (LeadingOnes) {
if (UniquedVals[Multiple-1].getNode() == 0)
return DAG.getTargetConstant(~0U, MVT::i32); int Val =cast<ConstantSDNode>(UniquedVals[Multiple-1])->getSExtValue();
if (Val >= -16) return DAG.getTargetConstant(Val, MVT::i32);
}
return SDValue();
}
for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
if (N->getOperand(i).getOpcode() == ISD::UNDEF) continue;
if (OpVal.getNode() == 0)
OpVal = N->getOperand(i);
else if (OpVal != N->getOperand(i))
return SDValue();
}
if (OpVal.getNode() == 0) return SDValue();
unsigned ValSizeInBytes = EltSize;
uint64_t Value = 0;
if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(OpVal)) {
Value = CN->getZExtValue();
} else if (ConstantFPSDNode *CN = dyn_cast<ConstantFPSDNode>(OpVal)) {
assert(CN->getValueType(0) == MVT::f32 && "Only one legal FP vector type!");
Value = FloatToBits(CN->getValueAPF().convertToFloat());
}
if (ValSizeInBytes < ByteSize) return SDValue();
while (ValSizeInBytes > ByteSize) {
ValSizeInBytes >>= 1;
if (((Value >> (ValSizeInBytes*8)) & ((1 << (8*ValSizeInBytes))-1)) !=
(Value & ((1 << (8*ValSizeInBytes))-1)))
return SDValue();
}
int MaskVal = SignExtend32(Value, ByteSize * 8);
if (MaskVal == 0) return SDValue();
if (SignExtend32<5>(MaskVal) == MaskVal)
return DAG.getTargetConstant(MaskVal, MVT::i32);
return SDValue();
}
static bool isIntS16Immediate(SDNode *N, short &Imm) {
if (N->getOpcode() != ISD::Constant)
return false;
Imm = (short)cast<ConstantSDNode>(N)->getZExtValue();
if (N->getValueType(0) == MVT::i32)
return Imm == (int32_t)cast<ConstantSDNode>(N)->getZExtValue();
else
return Imm == (int64_t)cast<ConstantSDNode>(N)->getZExtValue();
}
static bool isIntS16Immediate(SDValue Op, short &Imm) {
return isIntS16Immediate(Op.getNode(), Imm);
}
bool PPCTargetLowering::SelectAddressRegReg(SDValue N, SDValue &Base,
SDValue &Index,
SelectionDAG &DAG) const {
short imm = 0;
if (N.getOpcode() == ISD::ADD) {
if (isIntS16Immediate(N.getOperand(1), imm))
return false; if (N.getOperand(1).getOpcode() == PPCISD::Lo)
return false;
Base = N.getOperand(0);
Index = N.getOperand(1);
return true;
} else if (N.getOpcode() == ISD::OR) {
if (isIntS16Immediate(N.getOperand(1), imm))
return false;
APInt LHSKnownZero, LHSKnownOne;
APInt RHSKnownZero, RHSKnownOne;
DAG.ComputeMaskedBits(N.getOperand(0),
LHSKnownZero, LHSKnownOne);
if (LHSKnownZero.getBoolValue()) {
DAG.ComputeMaskedBits(N.getOperand(1),
RHSKnownZero, RHSKnownOne);
if (~(LHSKnownZero | RHSKnownZero) == 0) {
Base = N.getOperand(0);
Index = N.getOperand(1);
return true;
}
}
}
return false;
}
bool PPCTargetLowering::SelectAddressRegImm(SDValue N, SDValue &Disp,
SDValue &Base,
SelectionDAG &DAG) const {
DebugLoc dl = N.getDebugLoc();
if (SelectAddressRegReg(N, Disp, Base, DAG))
return false;
if (N.getOpcode() == ISD::ADD) {
short imm = 0;
if (isIntS16Immediate(N.getOperand(1), imm)) {
Disp = DAG.getTargetConstant((int)imm & 0xFFFF, MVT::i32);
if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N.getOperand(0))) {
Base = DAG.getTargetFrameIndex(FI->getIndex(), N.getValueType());
} else {
Base = N.getOperand(0);
}
return true; } else if (N.getOperand(1).getOpcode() == PPCISD::Lo) {
assert(!cast<ConstantSDNode>(N.getOperand(1).getOperand(1))->getZExtValue()
&& "Cannot handle constant offsets yet!");
Disp = N.getOperand(1).getOperand(0); assert(Disp.getOpcode() == ISD::TargetGlobalAddress ||
Disp.getOpcode() == ISD::TargetGlobalTLSAddress ||
Disp.getOpcode() == ISD::TargetConstantPool ||
Disp.getOpcode() == ISD::TargetJumpTable);
Base = N.getOperand(0);
return true; }
} else if (N.getOpcode() == ISD::OR) {
short imm = 0;
if (isIntS16Immediate(N.getOperand(1), imm)) {
APInt LHSKnownZero, LHSKnownOne;
DAG.ComputeMaskedBits(N.getOperand(0), LHSKnownZero, LHSKnownOne);
if ((LHSKnownZero.getZExtValue()|~(uint64_t)imm) == ~0ULL) {
Base = N.getOperand(0);
Disp = DAG.getTargetConstant((int)imm & 0xFFFF, MVT::i32);
return true;
}
}
} else if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(N)) {
short Imm;
if (isIntS16Immediate(CN, Imm)) {
Disp = DAG.getTargetConstant(Imm, CN->getValueType(0));
Base = DAG.getRegister(PPCSubTarget.isPPC64() ? PPC::X0 : PPC::R0,
CN->getValueType(0));
return true;
}
if (CN->getValueType(0) == MVT::i32 ||
(int64_t)CN->getZExtValue() == (int)CN->getZExtValue()) {
int Addr = (int)CN->getZExtValue();
Disp = DAG.getTargetConstant((short)Addr, MVT::i32);
Base = DAG.getTargetConstant((Addr - (signed short)Addr) >> 16, MVT::i32);
unsigned Opc = CN->getValueType(0) == MVT::i32 ? PPC::LIS : PPC::LIS8;
Base = SDValue(DAG.getMachineNode(Opc, dl, CN->getValueType(0), Base), 0);
return true;
}
}
Disp = DAG.getTargetConstant(0, getPointerTy());
if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N))
Base = DAG.getTargetFrameIndex(FI->getIndex(), N.getValueType());
else
Base = N;
return true; }
bool PPCTargetLowering::SelectAddressRegRegOnly(SDValue N, SDValue &Base,
SDValue &Index,
SelectionDAG &DAG) const {
if (SelectAddressRegReg(N, Base, Index, DAG))
return true;
if (N.getOpcode() == ISD::ADD) {
Base = N.getOperand(0);
Index = N.getOperand(1);
return true;
}
Base = DAG.getRegister(PPCSubTarget.isPPC64() ? PPC::X0 : PPC::R0,
N.getValueType());
Index = N;
return true;
}
bool PPCTargetLowering::SelectAddressRegImmShift(SDValue N, SDValue &Disp,
SDValue &Base,
SelectionDAG &DAG) const {
DebugLoc dl = N.getDebugLoc();
if (SelectAddressRegReg(N, Disp, Base, DAG))
return false;
if (N.getOpcode() == ISD::ADD) {
short imm = 0;
if (isIntS16Immediate(N.getOperand(1), imm) && (imm & 3) == 0) {
Disp = DAG.getTargetConstant(((int)imm & 0xFFFF) >> 2, MVT::i32);
if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N.getOperand(0))) {
Base = DAG.getTargetFrameIndex(FI->getIndex(), N.getValueType());
} else {
Base = N.getOperand(0);
}
return true; } else if (N.getOperand(1).getOpcode() == PPCISD::Lo) {
assert(!cast<ConstantSDNode>(N.getOperand(1).getOperand(1))->getZExtValue()
&& "Cannot handle constant offsets yet!");
Disp = N.getOperand(1).getOperand(0); assert(Disp.getOpcode() == ISD::TargetGlobalAddress ||
Disp.getOpcode() == ISD::TargetConstantPool ||
Disp.getOpcode() == ISD::TargetJumpTable);
Base = N.getOperand(0);
return true; }
} else if (N.getOpcode() == ISD::OR) {
short imm = 0;
if (isIntS16Immediate(N.getOperand(1), imm) && (imm & 3) == 0) {
APInt LHSKnownZero, LHSKnownOne;
DAG.ComputeMaskedBits(N.getOperand(0), LHSKnownZero, LHSKnownOne);
if ((LHSKnownZero.getZExtValue()|~(uint64_t)imm) == ~0ULL) {
Base = N.getOperand(0);
Disp = DAG.getTargetConstant(((int)imm & 0xFFFF) >> 2, MVT::i32);
return true;
}
}
} else if (ConstantSDNode *CN = dyn_cast<ConstantSDNode>(N)) {
if ((CN->getZExtValue() & 3) == 0) {
short Imm;
if (isIntS16Immediate(CN, Imm)) {
Disp = DAG.getTargetConstant((unsigned short)Imm >> 2, getPointerTy());
Base = DAG.getRegister(PPCSubTarget.isPPC64() ? PPC::X0 : PPC::R0,
CN->getValueType(0));
return true;
}
if (CN->getValueType(0) == MVT::i32 ||
(int64_t)CN->getZExtValue() == (int)CN->getZExtValue()) {
int Addr = (int)CN->getZExtValue();
Disp = DAG.getTargetConstant((short)Addr >> 2, MVT::i32);
Base = DAG.getTargetConstant((Addr-(signed short)Addr) >> 16, MVT::i32);
unsigned Opc = CN->getValueType(0) == MVT::i32 ? PPC::LIS : PPC::LIS8;
Base = SDValue(DAG.getMachineNode(Opc, dl, CN->getValueType(0), Base),0);
return true;
}
}
}
Disp = DAG.getTargetConstant(0, getPointerTy());
if (FrameIndexSDNode *FI = dyn_cast<FrameIndexSDNode>(N))
Base = DAG.getTargetFrameIndex(FI->getIndex(), N.getValueType());
else
Base = N;
return true; }
bool PPCTargetLowering::getPreIndexedAddressParts(SDNode *N, SDValue &Base,
SDValue &Offset,
ISD::MemIndexedMode &AM,
SelectionDAG &DAG) const {
if (DisablePPCPreinc) return false;
SDValue Ptr;
EVT VT;
if (LoadSDNode *LD = dyn_cast<LoadSDNode>(N)) {
Ptr = LD->getBasePtr();
VT = LD->getMemoryVT();
} else if (StoreSDNode *ST = dyn_cast<StoreSDNode>(N)) {
Ptr = ST->getBasePtr();
VT = ST->getMemoryVT();
} else
return false;
if (VT.isVector())
return false;
if (SelectAddressRegReg(Ptr, Offset, Base, DAG)) {
AM = ISD::PRE_INC;
return true;
}
if (VT != MVT::i64) {
if (!SelectAddressRegImm(Ptr, Offset, Base, DAG))
return false;
} else {
if (!SelectAddressRegImmShift(Ptr, Offset, Base, DAG))
return false;
}
if (LoadSDNode *LD = dyn_cast<LoadSDNode>(N)) {
if (LD->getValueType(0) == MVT::i64 && LD->getMemoryVT() == MVT::i32 &&
LD->getExtensionType() == ISD::SEXTLOAD &&
isa<ConstantSDNode>(Offset))
return false;
}
AM = ISD::PRE_INC;
return true;
}
static bool GetLabelAccessInfo(const TargetMachine &TM, unsigned &HiOpFlags,
unsigned &LoOpFlags, const GlobalValue *GV = 0) {
HiOpFlags = PPCII::MO_HA16;
LoOpFlags = PPCII::MO_LO16;
bool isPIC = TM.getRelocationModel() == Reloc::PIC_ &&
TM.getSubtarget<PPCSubtarget>().isDarwin();
if (isPIC) {
HiOpFlags |= PPCII::MO_PIC_FLAG;
LoOpFlags |= PPCII::MO_PIC_FLAG;
}
if (GV && TM.getSubtarget<PPCSubtarget>().hasLazyResolverStub(GV, TM)) {
HiOpFlags |= PPCII::MO_NLP_FLAG;
LoOpFlags |= PPCII::MO_NLP_FLAG;
if (GV->hasHiddenVisibility()) {
HiOpFlags |= PPCII::MO_NLP_HIDDEN_FLAG;
LoOpFlags |= PPCII::MO_NLP_HIDDEN_FLAG;
}
}
return isPIC;
}
static SDValue LowerLabelRef(SDValue HiPart, SDValue LoPart, bool isPIC,
SelectionDAG &DAG) {
EVT PtrVT = HiPart.getValueType();
SDValue Zero = DAG.getConstant(0, PtrVT);
DebugLoc DL = HiPart.getDebugLoc();
SDValue Hi = DAG.getNode(PPCISD::Hi, DL, PtrVT, HiPart, Zero);
SDValue Lo = DAG.getNode(PPCISD::Lo, DL, PtrVT, LoPart, Zero);
if (isPIC)
Hi = DAG.getNode(ISD::ADD, DL, PtrVT,
DAG.getNode(PPCISD::GlobalBaseReg, DL, PtrVT), Hi);
return DAG.getNode(ISD::ADD, DL, PtrVT, Hi, Lo);
}
SDValue PPCTargetLowering::LowerConstantPool(SDValue Op,
SelectionDAG &DAG) const {
EVT PtrVT = Op.getValueType();
ConstantPoolSDNode *CP = cast<ConstantPoolSDNode>(Op);
const Constant *C = CP->getConstVal();
if (PPCSubTarget.isSVR4ABI() && PPCSubTarget.isPPC64()) {
SDValue GA = DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(), 0);
return DAG.getNode(PPCISD::TOC_ENTRY, CP->getDebugLoc(), MVT::i64, GA,
DAG.getRegister(PPC::X2, MVT::i64));
}
unsigned MOHiFlag, MOLoFlag;
bool isPIC = GetLabelAccessInfo(DAG.getTarget(), MOHiFlag, MOLoFlag);
SDValue CPIHi =
DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(), 0, MOHiFlag);
SDValue CPILo =
DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(), 0, MOLoFlag);
return LowerLabelRef(CPIHi, CPILo, isPIC, DAG);
}
SDValue PPCTargetLowering::LowerJumpTable(SDValue Op, SelectionDAG &DAG) const {
EVT PtrVT = Op.getValueType();
JumpTableSDNode *JT = cast<JumpTableSDNode>(Op);
if (PPCSubTarget.isSVR4ABI() && PPCSubTarget.isPPC64()) {
SDValue GA = DAG.getTargetJumpTable(JT->getIndex(), PtrVT);
return DAG.getNode(PPCISD::TOC_ENTRY, JT->getDebugLoc(), MVT::i64, GA,
DAG.getRegister(PPC::X2, MVT::i64));
}
unsigned MOHiFlag, MOLoFlag;
bool isPIC = GetLabelAccessInfo(DAG.getTarget(), MOHiFlag, MOLoFlag);
SDValue JTIHi = DAG.getTargetJumpTable(JT->getIndex(), PtrVT, MOHiFlag);
SDValue JTILo = DAG.getTargetJumpTable(JT->getIndex(), PtrVT, MOLoFlag);
return LowerLabelRef(JTIHi, JTILo, isPIC, DAG);
}
SDValue PPCTargetLowering::LowerBlockAddress(SDValue Op,
SelectionDAG &DAG) const {
EVT PtrVT = Op.getValueType();
const BlockAddress *BA = cast<BlockAddressSDNode>(Op)->getBlockAddress();
unsigned MOHiFlag, MOLoFlag;
bool isPIC = GetLabelAccessInfo(DAG.getTarget(), MOHiFlag, MOLoFlag);
SDValue TgtBAHi = DAG.getTargetBlockAddress(BA, PtrVT, 0, MOHiFlag);
SDValue TgtBALo = DAG.getTargetBlockAddress(BA, PtrVT, 0, MOLoFlag);
return LowerLabelRef(TgtBAHi, TgtBALo, isPIC, DAG);
}
SDValue PPCTargetLowering::LowerGlobalTLSAddress(SDValue Op,
SelectionDAG &DAG) const {
GlobalAddressSDNode *GA = cast<GlobalAddressSDNode>(Op);
DebugLoc dl = GA->getDebugLoc();
const GlobalValue *GV = GA->getGlobal();
EVT PtrVT = getPointerTy();
bool is64bit = PPCSubTarget.isPPC64();
TLSModel::Model Model = getTargetMachine().getTLSModel(GV);
if (Model == TLSModel::LocalExec) {
SDValue TGAHi = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0,
PPCII::MO_TPREL16_HA);
SDValue TGALo = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0,
PPCII::MO_TPREL16_LO);
SDValue TLSReg = DAG.getRegister(is64bit ? PPC::X13 : PPC::R2,
is64bit ? MVT::i64 : MVT::i32);
SDValue Hi = DAG.getNode(PPCISD::Hi, dl, PtrVT, TGAHi, TLSReg);
return DAG.getNode(PPCISD::Lo, dl, PtrVT, TGALo, Hi);
}
if (!is64bit)
llvm_unreachable("only local-exec is currently supported for ppc32");
if (Model == TLSModel::InitialExec) {
SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0);
SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64);
SDValue TPOffsetHi = DAG.getNode(PPCISD::ADDIS_GOT_TPREL_HA, dl,
PtrVT, GOTReg, TGA);
SDValue TPOffset = DAG.getNode(PPCISD::LD_GOT_TPREL_L, dl,
PtrVT, TGA, TPOffsetHi);
return DAG.getNode(PPCISD::ADD_TLS, dl, PtrVT, TPOffset, TGA);
}
if (Model == TLSModel::GeneralDynamic) {
SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0);
SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64);
SDValue GOTEntryHi = DAG.getNode(PPCISD::ADDIS_TLSGD_HA, dl, PtrVT,
GOTReg, TGA);
SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSGD_L, dl, PtrVT,
GOTEntryHi, TGA);
SDValue Chain = DAG.getEntryNode();
Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, GOTEntry);
SDValue ParmReg = DAG.getRegister(PPC::X3, MVT::i64);
SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLS_ADDR, dl,
PtrVT, ParmReg, TGA);
Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, TLSAddr);
return DAG.getCopyFromReg(Chain, dl, PPC::X3, PtrVT);
}
if (Model == TLSModel::LocalDynamic) {
SDValue TGA = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, 0);
SDValue GOTReg = DAG.getRegister(PPC::X2, MVT::i64);
SDValue GOTEntryHi = DAG.getNode(PPCISD::ADDIS_TLSLD_HA, dl, PtrVT,
GOTReg, TGA);
SDValue GOTEntry = DAG.getNode(PPCISD::ADDI_TLSLD_L, dl, PtrVT,
GOTEntryHi, TGA);
SDValue Chain = DAG.getEntryNode();
Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, GOTEntry);
SDValue ParmReg = DAG.getRegister(PPC::X3, MVT::i64);
SDValue TLSAddr = DAG.getNode(PPCISD::GET_TLSLD_ADDR, dl,
PtrVT, ParmReg, TGA);
Chain = DAG.getCopyToReg(Chain, dl, PPC::X3, TLSAddr);
SDValue DtvOffsetHi = DAG.getNode(PPCISD::ADDIS_DTPREL_HA, dl, PtrVT,
Chain, ParmReg, TGA);
return DAG.getNode(PPCISD::ADDI_DTPREL_L, dl, PtrVT, DtvOffsetHi, TGA);
}
llvm_unreachable("Unknown TLS model!");
}
SDValue PPCTargetLowering::LowerGlobalAddress(SDValue Op,
SelectionDAG &DAG) const {
EVT PtrVT = Op.getValueType();
GlobalAddressSDNode *GSDN = cast<GlobalAddressSDNode>(Op);
DebugLoc DL = GSDN->getDebugLoc();
const GlobalValue *GV = GSDN->getGlobal();
if (PPCSubTarget.isSVR4ABI() && PPCSubTarget.isPPC64()) {
SDValue GA = DAG.getTargetGlobalAddress(GV, DL, PtrVT, GSDN->getOffset());
return DAG.getNode(PPCISD::TOC_ENTRY, DL, MVT::i64, GA,
DAG.getRegister(PPC::X2, MVT::i64));
}
unsigned MOHiFlag, MOLoFlag;
bool isPIC = GetLabelAccessInfo(DAG.getTarget(), MOHiFlag, MOLoFlag, GV);
SDValue GAHi =
DAG.getTargetGlobalAddress(GV, DL, PtrVT, GSDN->getOffset(), MOHiFlag);
SDValue GALo =
DAG.getTargetGlobalAddress(GV, DL, PtrVT, GSDN->getOffset(), MOLoFlag);
SDValue Ptr = LowerLabelRef(GAHi, GALo, isPIC, DAG);
if (MOHiFlag & PPCII::MO_NLP_FLAG)
Ptr = DAG.getLoad(PtrVT, DL, DAG.getEntryNode(), Ptr, MachinePointerInfo(),
false, false, false, 0);
return Ptr;
}
SDValue PPCTargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) const {
ISD::CondCode CC = cast<CondCodeSDNode>(Op.getOperand(2))->get();
DebugLoc dl = Op.getDebugLoc();
if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Op.getOperand(1))) {
if (C->isNullValue() && CC == ISD::SETEQ) {
EVT VT = Op.getOperand(0).getValueType();
SDValue Zext = Op.getOperand(0);
if (VT.bitsLT(MVT::i32)) {
VT = MVT::i32;
Zext = DAG.getNode(ISD::ZERO_EXTEND, dl, VT, Op.getOperand(0));
}
unsigned Log2b = Log2_32(VT.getSizeInBits());
SDValue Clz = DAG.getNode(ISD::CTLZ, dl, VT, Zext);
SDValue Scc = DAG.getNode(ISD::SRL, dl, VT, Clz,
DAG.getConstant(Log2b, MVT::i32));
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, Scc);
}
if (C->isAllOnesValue() || C->isNullValue())
return SDValue();
}
EVT LHSVT = Op.getOperand(0).getValueType();
if (LHSVT.isInteger() && (CC == ISD::SETEQ || CC == ISD::SETNE)) {
EVT VT = Op.getValueType();
SDValue Sub = DAG.getNode(ISD::XOR, dl, LHSVT, Op.getOperand(0),
Op.getOperand(1));
return DAG.getSetCC(dl, VT, Sub, DAG.getConstant(0, LHSVT), CC);
}
return SDValue();
}
SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
const PPCSubtarget &Subtarget) const {
SDNode *Node = Op.getNode();
EVT VT = Node->getValueType(0);
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue InChain = Node->getOperand(0);
SDValue VAListPtr = Node->getOperand(1);
const Value *SV = cast<SrcValueSDNode>(Node->getOperand(2))->getValue();
DebugLoc dl = Node->getDebugLoc();
assert(!Subtarget.isPPC64() && "LowerVAARG is PPC32 only");
SDValue GprIndex = DAG.getExtLoad(ISD::ZEXTLOAD, dl, MVT::i32, InChain,
VAListPtr, MachinePointerInfo(SV), MVT::i8,
false, false, 0);
InChain = GprIndex.getValue(1);
if (VT == MVT::i64) {
SDValue GprAnd = DAG.getNode(ISD::AND, dl, MVT::i32, GprIndex,
DAG.getConstant(1, MVT::i32));
SDValue CC64 = DAG.getSetCC(dl, MVT::i32, GprAnd,
DAG.getConstant(0, MVT::i32), ISD::SETNE);
SDValue GprIndexPlusOne = DAG.getNode(ISD::ADD, dl, MVT::i32, GprIndex,
DAG.getConstant(1, MVT::i32));
GprIndex = DAG.getNode(ISD::SELECT, dl, MVT::i32, CC64, GprIndexPlusOne,
GprIndex);
}
SDValue FprPtr = DAG.getNode(ISD::ADD, dl, PtrVT, VAListPtr,
DAG.getConstant(1, MVT::i32));
SDValue FprIndex = DAG.getExtLoad(ISD::ZEXTLOAD, dl, MVT::i32, InChain,
FprPtr, MachinePointerInfo(SV), MVT::i8,
false, false, 0);
InChain = FprIndex.getValue(1);
SDValue RegSaveAreaPtr = DAG.getNode(ISD::ADD, dl, PtrVT, VAListPtr,
DAG.getConstant(8, MVT::i32));
SDValue OverflowAreaPtr = DAG.getNode(ISD::ADD, dl, PtrVT, VAListPtr,
DAG.getConstant(4, MVT::i32));
SDValue OverflowArea = DAG.getLoad(MVT::i32, dl, InChain, OverflowAreaPtr,
MachinePointerInfo(), false, false,
false, 0);
InChain = OverflowArea.getValue(1);
SDValue RegSaveArea = DAG.getLoad(MVT::i32, dl, InChain, RegSaveAreaPtr,
MachinePointerInfo(), false, false,
false, 0);
InChain = RegSaveArea.getValue(1);
SDValue CC = DAG.getSetCC(dl, MVT::i32, VT.isInteger() ? GprIndex : FprIndex,
DAG.getConstant(8, MVT::i32), ISD::SETLT);
SDValue RegConstant = DAG.getNode(ISD::MUL, dl, MVT::i32,
VT.isInteger() ? GprIndex : FprIndex,
DAG.getConstant(VT.isInteger() ? 4 : 8,
MVT::i32));
SDValue OurReg = DAG.getNode(ISD::ADD, dl, PtrVT, RegSaveArea,
RegConstant);
if (VT.isFloatingPoint())
OurReg = DAG.getNode(ISD::ADD, dl, PtrVT, OurReg,
DAG.getConstant(32, MVT::i32));
SDValue IndexPlus1 = DAG.getNode(ISD::ADD, dl, MVT::i32,
VT.isInteger() ? GprIndex : FprIndex,
DAG.getConstant(VT == MVT::i64 ? 2 : 1,
MVT::i32));
InChain = DAG.getTruncStore(InChain, dl, IndexPlus1,
VT.isInteger() ? VAListPtr : FprPtr,
MachinePointerInfo(SV),
MVT::i8, false, false, 0);
SDValue Result = DAG.getNode(ISD::SELECT, dl, PtrVT, CC, OurReg, OverflowArea);
SDValue OverflowAreaPlusN = DAG.getNode(ISD::ADD, dl, PtrVT, OverflowArea,
DAG.getConstant(VT.isInteger() ? 4 : 8,
MVT::i32));
OverflowArea = DAG.getNode(ISD::SELECT, dl, MVT::i32, CC, OverflowArea,
OverflowAreaPlusN);
InChain = DAG.getTruncStore(InChain, dl, OverflowArea,
OverflowAreaPtr,
MachinePointerInfo(),
MVT::i32, false, false, 0);
return DAG.getLoad(VT, dl, InChain, Result, MachinePointerInfo(),
false, false, false, 0);
}
SDValue PPCTargetLowering::LowerADJUST_TRAMPOLINE(SDValue Op,
SelectionDAG &DAG) const {
return Op.getOperand(0);
}
SDValue PPCTargetLowering::LowerINIT_TRAMPOLINE(SDValue Op,
SelectionDAG &DAG) const {
SDValue Chain = Op.getOperand(0);
SDValue Trmp = Op.getOperand(1); SDValue FPtr = Op.getOperand(2); SDValue Nest = Op.getOperand(3); DebugLoc dl = Op.getDebugLoc();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isPPC64 = (PtrVT == MVT::i64);
Type *IntPtrTy =
DAG.getTargetLoweringInfo().getDataLayout()->getIntPtrType(
*DAG.getContext());
TargetLowering::ArgListTy Args;
TargetLowering::ArgListEntry Entry;
Entry.Ty = IntPtrTy;
Entry.Node = Trmp; Args.push_back(Entry);
Entry.Node = DAG.getConstant(isPPC64 ? 48 : 40,
isPPC64 ? MVT::i64 : MVT::i32);
Args.push_back(Entry);
Entry.Node = FPtr; Args.push_back(Entry);
Entry.Node = Nest; Args.push_back(Entry);
TargetLowering::CallLoweringInfo CLI(Chain,
Type::getVoidTy(*DAG.getContext()),
false, false, false, false, 0,
CallingConv::C,
false,
false,
true,
DAG.getExternalSymbol("__trampoline_setup", PtrVT),
Args, DAG, dl);
std::pair<SDValue, SDValue> CallResult = LowerCallTo(CLI);
return CallResult.second;
}
SDValue PPCTargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG,
const PPCSubtarget &Subtarget) const {
MachineFunction &MF = DAG.getMachineFunction();
PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
DebugLoc dl = Op.getDebugLoc();
if (Subtarget.isDarwinABI() || Subtarget.isPPC64()) {
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
return DAG.getStore(Op.getOperand(0), dl, FR, Op.getOperand(1),
MachinePointerInfo(SV),
false, false, 0);
}
SDValue ArgGPR = DAG.getConstant(FuncInfo->getVarArgsNumGPR(), MVT::i32);
SDValue ArgFPR = DAG.getConstant(FuncInfo->getVarArgsNumFPR(), MVT::i32);
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue StackOffsetFI = DAG.getFrameIndex(FuncInfo->getVarArgsStackOffset(),
PtrVT);
SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(),
PtrVT);
uint64_t FrameOffset = PtrVT.getSizeInBits()/8;
SDValue ConstFrameOffset = DAG.getConstant(FrameOffset, PtrVT);
uint64_t StackOffset = PtrVT.getSizeInBits()/8 - 1;
SDValue ConstStackOffset = DAG.getConstant(StackOffset, PtrVT);
uint64_t FPROffset = 1;
SDValue ConstFPROffset = DAG.getConstant(FPROffset, PtrVT);
const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
SDValue firstStore = DAG.getTruncStore(Op.getOperand(0), dl, ArgGPR,
Op.getOperand(1),
MachinePointerInfo(SV),
MVT::i8, false, false, 0);
uint64_t nextOffset = FPROffset;
SDValue nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, Op.getOperand(1),
ConstFPROffset);
SDValue secondStore =
DAG.getTruncStore(firstStore, dl, ArgFPR, nextPtr,
MachinePointerInfo(SV, nextOffset), MVT::i8,
false, false, 0);
nextOffset += StackOffset;
nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, nextPtr, ConstStackOffset);
SDValue thirdStore =
DAG.getStore(secondStore, dl, StackOffsetFI, nextPtr,
MachinePointerInfo(SV, nextOffset),
false, false, 0);
nextOffset += FrameOffset;
nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, nextPtr, ConstFrameOffset);
return DAG.getStore(thirdStore, dl, FR, nextPtr,
MachinePointerInfo(SV, nextOffset),
false, false, 0);
}
#include "PPCGenCallingConv.inc"
static bool CC_PPC_SVR4_Custom_Dummy(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State) {
return true;
}
static bool CC_PPC_SVR4_Custom_AlignArgRegs(unsigned &ValNo, MVT &ValVT,
MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State) {
static const uint16_t ArgRegs[] = {
PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
const unsigned NumArgRegs = array_lengthof(ArgRegs);
unsigned RegNum = State.getFirstUnallocated(ArgRegs, NumArgRegs);
if (RegNum != NumArgRegs && RegNum % 2 == 1) {
State.AllocateReg(ArgRegs[RegNum]);
}
return false;
}
static bool CC_PPC_SVR4_Custom_AlignFPArgRegs(unsigned &ValNo, MVT &ValVT,
MVT &LocVT,
CCValAssign::LocInfo &LocInfo,
ISD::ArgFlagsTy &ArgFlags,
CCState &State) {
static const uint16_t ArgRegs[] = {
PPC::F1, PPC::F2, PPC::F3, PPC::F4, PPC::F5, PPC::F6, PPC::F7,
PPC::F8
};
const unsigned NumArgRegs = array_lengthof(ArgRegs);
unsigned RegNum = State.getFirstUnallocated(ArgRegs, NumArgRegs);
if (RegNum != NumArgRegs && ArgRegs[RegNum] == PPC::F8) {
State.AllocateReg(ArgRegs[RegNum]);
}
return false;
}
static const uint16_t *GetFPR() {
static const uint16_t FPR[] = {
PPC::F1, PPC::F2, PPC::F3, PPC::F4, PPC::F5, PPC::F6, PPC::F7,
PPC::F8, PPC::F9, PPC::F10, PPC::F11, PPC::F12, PPC::F13
};
return FPR;
}
static unsigned CalculateStackSlotSize(EVT ArgVT, ISD::ArgFlagsTy Flags,
unsigned PtrByteSize) {
unsigned ArgSize = ArgVT.getSizeInBits()/8;
if (Flags.isByVal())
ArgSize = Flags.getByValSize();
ArgSize = ((ArgSize + PtrByteSize - 1)/PtrByteSize) * PtrByteSize;
return ArgSize;
}
SDValue
PPCTargetLowering::LowerFormalArguments(SDValue Chain,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::InputArg>
&Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals)
const {
if (PPCSubTarget.isSVR4ABI()) {
if (PPCSubTarget.isPPC64())
return LowerFormalArguments_64SVR4(Chain, CallConv, isVarArg, Ins,
dl, DAG, InVals);
else
return LowerFormalArguments_32SVR4(Chain, CallConv, isVarArg, Ins,
dl, DAG, InVals);
} else {
return LowerFormalArguments_Darwin(Chain, CallConv, isVarArg, Ins,
dl, DAG, InVals);
}
}
SDValue
PPCTargetLowering::LowerFormalArguments_32SVR4(
SDValue Chain,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::InputArg>
&Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *MFI = MF.getFrameInfo();
PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
(CallConv == CallingConv::Fast));
unsigned PtrByteSize = 4;
SmallVector<CCValAssign, 16> ArgLocs;
CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), ArgLocs, *DAG.getContext());
CCInfo.AllocateStack(PPCFrameLowering::getLinkageSize(false, false), PtrByteSize);
CCInfo.AnalyzeFormalArguments(Ins, CC_PPC_SVR4);
for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
CCValAssign &VA = ArgLocs[i];
if (VA.isRegLoc()) {
const TargetRegisterClass *RC;
EVT ValVT = VA.getValVT();
switch (ValVT.getSimpleVT().SimpleTy) {
default:
llvm_unreachable("ValVT not supported by formal arguments Lowering");
case MVT::i32:
RC = &PPC::GPRCRegClass;
break;
case MVT::f32:
RC = &PPC::F4RCRegClass;
break;
case MVT::f64:
RC = &PPC::F8RCRegClass;
break;
case MVT::v16i8:
case MVT::v8i16:
case MVT::v4i32:
case MVT::v4f32:
RC = &PPC::VRRCRegClass;
break;
}
unsigned Reg = MF.addLiveIn(VA.getLocReg(), RC);
SDValue ArgValue = DAG.getCopyFromReg(Chain, dl, Reg, ValVT);
InVals.push_back(ArgValue);
} else {
assert(VA.isMemLoc());
unsigned ArgSize = VA.getLocVT().getSizeInBits() / 8;
int FI = MFI->CreateFixedObject(ArgSize, VA.getLocMemOffset(),
isImmutable);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
InVals.push_back(DAG.getLoad(VA.getValVT(), dl, Chain, FIN,
MachinePointerInfo(),
false, false, false, 0));
}
}
SmallVector<CCValAssign, 16> ByValArgLocs;
CCState CCByValInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), ByValArgLocs, *DAG.getContext());
CCByValInfo.AllocateStack(CCInfo.getNextStackOffset(), PtrByteSize);
CCByValInfo.AnalyzeFormalArguments(Ins, CC_PPC_SVR4_ByVal);
unsigned MinReservedArea = CCByValInfo.getNextStackOffset();
PPCFunctionInfo *FI = MF.getInfo<PPCFunctionInfo>();
MinReservedArea =
std::max(MinReservedArea,
PPCFrameLowering::getMinCallFrameSize(false, false));
unsigned TargetAlign = DAG.getMachineFunction().getTarget().getFrameLowering()->
getStackAlignment();
unsigned AlignMask = TargetAlign-1;
MinReservedArea = (MinReservedArea + AlignMask) & ~AlignMask;
FI->setMinReservedArea(MinReservedArea);
SmallVector<SDValue, 8> MemOps;
if (isVarArg) {
static const uint16_t GPArgRegs[] = {
PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
const unsigned NumGPArgRegs = array_lengthof(GPArgRegs);
static const uint16_t FPArgRegs[] = {
PPC::F1, PPC::F2, PPC::F3, PPC::F4, PPC::F5, PPC::F6, PPC::F7,
PPC::F8
};
const unsigned NumFPArgRegs = array_lengthof(FPArgRegs);
FuncInfo->setVarArgsNumGPR(CCInfo.getFirstUnallocated(GPArgRegs,
NumGPArgRegs));
FuncInfo->setVarArgsNumFPR(CCInfo.getFirstUnallocated(FPArgRegs,
NumFPArgRegs));
int Depth = NumGPArgRegs * PtrVT.getSizeInBits()/8 +
NumFPArgRegs * EVT(MVT::f64).getSizeInBits()/8;
FuncInfo->setVarArgsStackOffset(
MFI->CreateFixedObject(PtrVT.getSizeInBits()/8,
CCInfo.getNextStackOffset(), true));
FuncInfo->setVarArgsFrameIndex(MFI->CreateStackObject(Depth, 8, false));
SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
for (unsigned GPRIndex = 0; GPRIndex != NumGPArgRegs; ++GPRIndex) {
unsigned VReg = MF.getRegInfo().getLiveInVirtReg(GPArgRegs[GPRIndex]);
if (!VReg)
VReg = MF.addLiveIn(GPArgRegs[GPRIndex], &PPC::GPRCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(), false, false, 0);
MemOps.push_back(Store);
SDValue PtrOff = DAG.getConstant(PtrVT.getSizeInBits()/8, PtrVT);
FIN = DAG.getNode(ISD::ADD, dl, PtrOff.getValueType(), FIN, PtrOff);
}
for (unsigned FPRIndex = 0; FPRIndex != NumFPArgRegs; ++FPRIndex) {
unsigned VReg = MF.getRegInfo().getLiveInVirtReg(FPArgRegs[FPRIndex]);
if (!VReg)
VReg = MF.addLiveIn(FPArgRegs[FPRIndex], &PPC::F8RCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, MVT::f64);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(), false, false, 0);
MemOps.push_back(Store);
SDValue PtrOff = DAG.getConstant(EVT(MVT::f64).getSizeInBits()/8,
PtrVT);
FIN = DAG.getNode(ISD::ADD, dl, PtrOff.getValueType(), FIN, PtrOff);
}
}
if (!MemOps.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl,
MVT::Other, &MemOps[0], MemOps.size());
return Chain;
}
SDValue
PPCTargetLowering::extendArgForPPC64(ISD::ArgFlagsTy Flags, EVT ObjectVT,
SelectionDAG &DAG, SDValue ArgVal,
DebugLoc dl) const {
if (Flags.isSExt())
ArgVal = DAG.getNode(ISD::AssertSext, dl, MVT::i64, ArgVal,
DAG.getValueType(ObjectVT));
else if (Flags.isZExt())
ArgVal = DAG.getNode(ISD::AssertZext, dl, MVT::i64, ArgVal,
DAG.getValueType(ObjectVT));
return DAG.getNode(ISD::TRUNCATE, dl, MVT::i32, ArgVal);
}
void
PPCTargetLowering::setMinReservedArea(MachineFunction &MF, SelectionDAG &DAG,
unsigned nAltivecParamsAtEnd,
unsigned MinReservedArea,
bool isPPC64) const {
PPCFunctionInfo *FI = MF.getInfo<PPCFunctionInfo>();
if (nAltivecParamsAtEnd) {
MinReservedArea = ((MinReservedArea+15)/16)*16;
MinReservedArea += 16*nAltivecParamsAtEnd;
}
MinReservedArea =
std::max(MinReservedArea,
PPCFrameLowering::getMinCallFrameSize(isPPC64, true));
unsigned TargetAlign
= DAG.getMachineFunction().getTarget().getFrameLowering()->
getStackAlignment();
unsigned AlignMask = TargetAlign-1;
MinReservedArea = (MinReservedArea + AlignMask) & ~AlignMask;
FI->setMinReservedArea(MinReservedArea);
}
SDValue
PPCTargetLowering::LowerFormalArguments_64SVR4(
SDValue Chain,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::InputArg>
&Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *MFI = MF.getFrameInfo();
PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
(CallConv == CallingConv::Fast));
unsigned PtrByteSize = 8;
unsigned ArgOffset = PPCFrameLowering::getLinkageSize(true, true);
unsigned MinReservedArea = ArgOffset;
static const uint16_t GPR[] = {
PPC::X3, PPC::X4, PPC::X5, PPC::X6,
PPC::X7, PPC::X8, PPC::X9, PPC::X10,
};
static const uint16_t *FPR = GetFPR();
static const uint16_t VR[] = {
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
const unsigned Num_GPR_Regs = array_lengthof(GPR);
const unsigned Num_FPR_Regs = 13;
const unsigned Num_VR_Regs = array_lengthof(VR);
unsigned GPR_idx = 0, FPR_idx = 0, VR_idx = 0;
SmallVector<SDValue, 8> MemOps;
unsigned nAltivecParamsAtEnd = 0;
Function::const_arg_iterator FuncArg = MF.getFunction()->arg_begin();
for (unsigned ArgNo = 0, e = Ins.size(); ArgNo != e; ++ArgNo, ++FuncArg) {
SDValue ArgVal;
bool needsLoad = false;
EVT ObjectVT = Ins[ArgNo].VT;
unsigned ObjSize = ObjectVT.getSizeInBits()/8;
unsigned ArgSize = ObjSize;
ISD::ArgFlagsTy Flags = Ins[ArgNo].Flags;
unsigned CurArgOffset = ArgOffset;
if (ObjectVT==MVT::v4f32 || ObjectVT==MVT::v4i32 ||
ObjectVT==MVT::v8i16 || ObjectVT==MVT::v16i8) {
if (isVarArg) {
MinReservedArea = ((MinReservedArea+15)/16)*16;
MinReservedArea += CalculateStackSlotSize(ObjectVT,
Flags,
PtrByteSize);
} else
nAltivecParamsAtEnd++;
} else
MinReservedArea += CalculateStackSlotSize(Ins[ArgNo].VT,
Flags,
PtrByteSize);
if (Flags.isByVal()) {
ObjSize = Flags.getByValSize();
ArgSize = ((ObjSize + PtrByteSize - 1)/PtrByteSize) * PtrByteSize;
if (!ObjSize) {
int FI = MFI->CreateFixedObject(PtrByteSize, ArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
InVals.push_back(FIN);
continue;
}
if (ObjSize < PtrByteSize)
CurArgOffset = CurArgOffset + (PtrByteSize - ObjSize);
int FI = MFI->CreateFixedObject(ObjSize, CurArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
InVals.push_back(FIN);
if (ObjSize < 8) {
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store;
if (ObjSize==1 || ObjSize==2 || ObjSize==4) {
EVT ObjType = (ObjSize == 1 ? MVT::i8 :
(ObjSize == 2 ? MVT::i16 : MVT::i32));
Store = DAG.getTruncStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(FuncArg, CurArgOffset),
ObjType, false, false, 0);
} else {
int FI = MFI->CreateFixedObject(PtrByteSize, ArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(FuncArg, ArgOffset),
false, false, 0);
}
MemOps.push_back(Store);
++GPR_idx;
}
ArgOffset += PtrByteSize;
continue;
}
for (unsigned j = 0; j < ArgSize; j += PtrByteSize) {
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg;
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
int FI = MFI->CreateFixedObject(PtrByteSize, ArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(FuncArg, ArgOffset),
false, false, 0);
MemOps.push_back(Store);
++GPR_idx;
ArgOffset += PtrByteSize;
} else {
ArgOffset += ArgSize - j;
break;
}
}
continue;
}
switch (ObjectVT.getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unhandled argument type!");
case MVT::i32:
case MVT::i64:
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, MVT::i64);
if (ObjectVT == MVT::i32)
ArgVal = extendArgForPPC64(Flags, ObjectVT, DAG, ArgVal, dl);
++GPR_idx;
} else {
needsLoad = true;
ArgSize = PtrByteSize;
}
ArgOffset += 8;
break;
case MVT::f32:
case MVT::f64:
if (GPR_idx != Num_GPR_Regs) {
++GPR_idx;
}
if (FPR_idx != Num_FPR_Regs) {
unsigned VReg;
if (ObjectVT == MVT::f32)
VReg = MF.addLiveIn(FPR[FPR_idx], &PPC::F4RCRegClass);
else
VReg = MF.addLiveIn(FPR[FPR_idx], &PPC::F8RCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, ObjectVT);
++FPR_idx;
} else {
needsLoad = true;
ArgSize = PtrByteSize;
}
ArgOffset += 8;
break;
case MVT::v4f32:
case MVT::v4i32:
case MVT::v8i16:
case MVT::v16i8:
if (VR_idx != Num_VR_Regs) {
unsigned VReg = MF.addLiveIn(VR[VR_idx], &PPC::VRRCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, ObjectVT);
if (isVarArg) {
while ((ArgOffset % 16) != 0) {
ArgOffset += PtrByteSize;
if (GPR_idx != Num_GPR_Regs)
GPR_idx++;
}
ArgOffset += 16;
GPR_idx = std::min(GPR_idx+4, Num_GPR_Regs); }
++VR_idx;
} else {
ArgOffset = ((ArgOffset+15)/16)*16;
CurArgOffset = ArgOffset;
ArgOffset += 16;
needsLoad = true;
}
break;
}
if (needsLoad) {
int FI = MFI->CreateFixedObject(ObjSize,
CurArgOffset + (ArgSize - ObjSize),
isImmutable);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo(),
false, false, false, 0);
}
InVals.push_back(ArgVal);
}
setMinReservedArea(MF, DAG, nAltivecParamsAtEnd, MinReservedArea, true);
if (isVarArg) {
int Depth = ArgOffset;
FuncInfo->setVarArgsFrameIndex(
MFI->CreateFixedObject(PtrByteSize, Depth, true));
SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
for (; GPR_idx != Num_GPR_Regs; ++GPR_idx) {
unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(), false, false, 0);
MemOps.push_back(Store);
SDValue PtrOff = DAG.getConstant(PtrByteSize, PtrVT);
FIN = DAG.getNode(ISD::ADD, dl, PtrOff.getValueType(), FIN, PtrOff);
}
}
if (!MemOps.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl,
MVT::Other, &MemOps[0], MemOps.size());
return Chain;
}
SDValue
PPCTargetLowering::LowerFormalArguments_Darwin(
SDValue Chain,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::InputArg>
&Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *MFI = MF.getFrameInfo();
PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isPPC64 = PtrVT == MVT::i64;
bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
(CallConv == CallingConv::Fast));
unsigned PtrByteSize = isPPC64 ? 8 : 4;
unsigned ArgOffset = PPCFrameLowering::getLinkageSize(isPPC64, true);
unsigned MinReservedArea = ArgOffset;
static const uint16_t GPR_32[] = { PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
static const uint16_t GPR_64[] = { PPC::X3, PPC::X4, PPC::X5, PPC::X6,
PPC::X7, PPC::X8, PPC::X9, PPC::X10,
};
static const uint16_t *FPR = GetFPR();
static const uint16_t VR[] = {
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
const unsigned Num_GPR_Regs = array_lengthof(GPR_32);
const unsigned Num_FPR_Regs = 13;
const unsigned Num_VR_Regs = array_lengthof( VR);
unsigned GPR_idx = 0, FPR_idx = 0, VR_idx = 0;
const uint16_t *GPR = isPPC64 ? GPR_64 : GPR_32;
unsigned VecArgOffset = ArgOffset;
if (!isVarArg && !isPPC64) {
for (unsigned ArgNo = 0, e = Ins.size(); ArgNo != e;
++ArgNo) {
EVT ObjectVT = Ins[ArgNo].VT;
ISD::ArgFlagsTy Flags = Ins[ArgNo].Flags;
if (Flags.isByVal()) {
unsigned ObjSize = Flags.getByValSize();
unsigned ArgSize =
((ObjSize + PtrByteSize - 1)/PtrByteSize) * PtrByteSize;
VecArgOffset += ArgSize;
continue;
}
switch(ObjectVT.getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unhandled argument type!");
case MVT::i32:
case MVT::f32:
VecArgOffset += 4;
break;
case MVT::i64: case MVT::f64:
VecArgOffset += 8;
break;
case MVT::v4f32:
case MVT::v4i32:
case MVT::v8i16:
case MVT::v16i8:
break;
}
}
}
VecArgOffset = ((VecArgOffset+15)/16)*16;
VecArgOffset += 12*16;
SmallVector<SDValue, 8> MemOps;
unsigned nAltivecParamsAtEnd = 0;
Function::const_arg_iterator FuncArg = MF.getFunction()->arg_begin();
for (unsigned ArgNo = 0, e = Ins.size(); ArgNo != e; ++ArgNo, ++FuncArg) {
SDValue ArgVal;
bool needsLoad = false;
EVT ObjectVT = Ins[ArgNo].VT;
unsigned ObjSize = ObjectVT.getSizeInBits()/8;
unsigned ArgSize = ObjSize;
ISD::ArgFlagsTy Flags = Ins[ArgNo].Flags;
unsigned CurArgOffset = ArgOffset;
if (ObjectVT==MVT::v4f32 || ObjectVT==MVT::v4i32 ||
ObjectVT==MVT::v8i16 || ObjectVT==MVT::v16i8) {
if (isVarArg || isPPC64) {
MinReservedArea = ((MinReservedArea+15)/16)*16;
MinReservedArea += CalculateStackSlotSize(ObjectVT,
Flags,
PtrByteSize);
} else nAltivecParamsAtEnd++;
} else
MinReservedArea += CalculateStackSlotSize(Ins[ArgNo].VT,
Flags,
PtrByteSize);
if (Flags.isByVal()) {
ObjSize = Flags.getByValSize();
ArgSize = ((ObjSize + PtrByteSize - 1)/PtrByteSize) * PtrByteSize;
if (ObjSize==1 || ObjSize==2) {
CurArgOffset = CurArgOffset + (4 - ObjSize);
}
int FI = MFI->CreateFixedObject(ObjSize, CurArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
InVals.push_back(FIN);
if (ObjSize==1 || ObjSize==2) {
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg;
if (isPPC64)
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
else
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
EVT ObjType = ObjSize == 1 ? MVT::i8 : MVT::i16;
SDValue Store = DAG.getTruncStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(FuncArg,
CurArgOffset),
ObjType, false, false, 0);
MemOps.push_back(Store);
++GPR_idx;
}
ArgOffset += PtrByteSize;
continue;
}
for (unsigned j = 0; j < ArgSize; j += PtrByteSize) {
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg;
if (isPPC64)
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
else
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
int FI = MFI->CreateFixedObject(PtrByteSize, ArgOffset, true);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(FuncArg, ArgOffset),
false, false, 0);
MemOps.push_back(Store);
++GPR_idx;
ArgOffset += PtrByteSize;
} else {
ArgOffset += ArgSize - (ArgOffset-CurArgOffset);
break;
}
}
continue;
}
switch (ObjectVT.getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unhandled argument type!");
case MVT::i32:
if (!isPPC64) {
if (GPR_idx != Num_GPR_Regs) {
unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, MVT::i32);
++GPR_idx;
} else {
needsLoad = true;
ArgSize = PtrByteSize;
}
ArgOffset += PtrByteSize;
break;
}
case MVT::i64: if (GPR_idx != Num_GPR_Regs) {
unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, MVT::i64);
if (ObjectVT == MVT::i32)
ArgVal = extendArgForPPC64(Flags, ObjectVT, DAG, ArgVal, dl);
++GPR_idx;
} else {
needsLoad = true;
ArgSize = PtrByteSize;
}
ArgOffset += 8;
break;
case MVT::f32:
case MVT::f64:
if (GPR_idx != Num_GPR_Regs) {
++GPR_idx;
if (ObjSize == 8 && GPR_idx != Num_GPR_Regs && !isPPC64)
++GPR_idx;
}
if (FPR_idx != Num_FPR_Regs) {
unsigned VReg;
if (ObjectVT == MVT::f32)
VReg = MF.addLiveIn(FPR[FPR_idx], &PPC::F4RCRegClass);
else
VReg = MF.addLiveIn(FPR[FPR_idx], &PPC::F8RCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, ObjectVT);
++FPR_idx;
} else {
needsLoad = true;
}
ArgOffset += isPPC64 ? 8 : ObjSize;
break;
case MVT::v4f32:
case MVT::v4i32:
case MVT::v8i16:
case MVT::v16i8:
if (VR_idx != Num_VR_Regs) {
unsigned VReg = MF.addLiveIn(VR[VR_idx], &PPC::VRRCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, ObjectVT);
if (isVarArg) {
while ((ArgOffset % 16) != 0) {
ArgOffset += PtrByteSize;
if (GPR_idx != Num_GPR_Regs)
GPR_idx++;
}
ArgOffset += 16;
GPR_idx = std::min(GPR_idx+4, Num_GPR_Regs); }
++VR_idx;
} else {
if (!isVarArg && !isPPC64) {
CurArgOffset = VecArgOffset;
VecArgOffset += 16;
} else {
ArgOffset = ((ArgOffset+15)/16)*16;
CurArgOffset = ArgOffset;
ArgOffset += 16;
}
needsLoad = true;
}
break;
}
if (needsLoad) {
int FI = MFI->CreateFixedObject(ObjSize,
CurArgOffset + (ArgSize - ObjSize),
isImmutable);
SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo(),
false, false, false, 0);
}
InVals.push_back(ArgVal);
}
setMinReservedArea(MF, DAG, nAltivecParamsAtEnd, MinReservedArea, isPPC64);
if (isVarArg) {
int Depth = ArgOffset;
FuncInfo->setVarArgsFrameIndex(
MFI->CreateFixedObject(PtrVT.getSizeInBits()/8,
Depth, true));
SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
for (; GPR_idx != Num_GPR_Regs; ++GPR_idx) {
unsigned VReg;
if (isPPC64)
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
else
VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
MachinePointerInfo(), false, false, 0);
MemOps.push_back(Store);
SDValue PtrOff = DAG.getConstant(PtrVT.getSizeInBits()/8, PtrVT);
FIN = DAG.getNode(ISD::ADD, dl, PtrOff.getValueType(), FIN, PtrOff);
}
}
if (!MemOps.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl,
MVT::Other, &MemOps[0], MemOps.size());
return Chain;
}
static unsigned
CalculateParameterAndLinkageAreaSize(SelectionDAG &DAG,
bool isPPC64,
bool isVarArg,
unsigned CC,
const SmallVectorImpl<ISD::OutputArg>
&Outs,
const SmallVectorImpl<SDValue> &OutVals,
unsigned &nAltivecParamsAtEnd) {
unsigned NumBytes = PPCFrameLowering::getLinkageSize(isPPC64, true);
unsigned NumOps = Outs.size();
unsigned PtrByteSize = isPPC64 ? 8 : 4;
nAltivecParamsAtEnd = 0;
for (unsigned i = 0; i != NumOps; ++i) {
ISD::ArgFlagsTy Flags = Outs[i].Flags;
EVT ArgVT = Outs[i].VT;
if (ArgVT==MVT::v4f32 || ArgVT==MVT::v4i32 ||
ArgVT==MVT::v8i16 || ArgVT==MVT::v16i8) {
if (!isVarArg && !isPPC64) {
nAltivecParamsAtEnd++;
continue;
}
NumBytes = ((NumBytes+15)/16)*16;
}
NumBytes += CalculateStackSlotSize(ArgVT, Flags, PtrByteSize);
}
if (nAltivecParamsAtEnd) {
NumBytes = ((NumBytes+15)/16)*16;
NumBytes += 16*nAltivecParamsAtEnd;
}
NumBytes = std::max(NumBytes,
PPCFrameLowering::getMinCallFrameSize(isPPC64, true));
if (CC == CallingConv::Fast && DAG.getTarget().Options.GuaranteedTailCallOpt){
unsigned TargetAlign = DAG.getMachineFunction().getTarget().
getFrameLowering()->getStackAlignment();
unsigned AlignMask = TargetAlign-1;
NumBytes = (NumBytes + AlignMask) & ~AlignMask;
}
return NumBytes;
}
static int CalculateTailCallSPDiff(SelectionDAG& DAG, bool isTailCall,
unsigned ParamSize) {
if (!isTailCall) return 0;
PPCFunctionInfo *FI = DAG.getMachineFunction().getInfo<PPCFunctionInfo>();
unsigned CallerMinReservedArea = FI->getMinReservedArea();
int SPDiff = (int)CallerMinReservedArea - (int)ParamSize;
if (SPDiff < FI->getTailCallSPDelta())
FI->setTailCallSPDelta(SPDiff);
return SPDiff;
}
bool
PPCTargetLowering::IsEligibleForTailCallOptimization(SDValue Callee,
CallingConv::ID CalleeCC,
bool isVarArg,
const SmallVectorImpl<ISD::InputArg> &Ins,
SelectionDAG& DAG) const {
if (!getTargetMachine().Options.GuaranteedTailCallOpt)
return false;
if (isVarArg)
return false;
MachineFunction &MF = DAG.getMachineFunction();
CallingConv::ID CallerCC = MF.getFunction()->getCallingConv();
if (CalleeCC == CallingConv::Fast && CallerCC == CalleeCC) {
for (unsigned i = 0; i != Ins.size(); i++) {
ISD::ArgFlagsTy Flags = Ins[i].Flags;
if (Flags.isByVal()) return false;
}
if (getTargetMachine().getRelocationModel() != Reloc::PIC_)
return true;
if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee))
return G->getGlobal()->hasHiddenVisibility()
|| G->getGlobal()->hasProtectedVisibility();
}
return false;
}
static SDNode *isBLACompatibleAddress(SDValue Op, SelectionDAG &DAG) {
ConstantSDNode *C = dyn_cast<ConstantSDNode>(Op);
if (!C) return 0;
int Addr = C->getZExtValue();
if ((Addr & 3) != 0 || SignExtend32<26>(Addr) != Addr)
return 0;
return DAG.getConstant((int)C->getZExtValue() >> 2,
DAG.getTargetLoweringInfo().getPointerTy()).getNode();
}
namespace {
struct TailCallArgumentInfo {
SDValue Arg;
SDValue FrameIdxOp;
int FrameIdx;
TailCallArgumentInfo() : FrameIdx(0) {}
};
}
static void
StoreTailCallArgumentsToStackSlot(SelectionDAG &DAG,
SDValue Chain,
const SmallVector<TailCallArgumentInfo, 8> &TailCallArgs,
SmallVector<SDValue, 8> &MemOpChains,
DebugLoc dl) {
for (unsigned i = 0, e = TailCallArgs.size(); i != e; ++i) {
SDValue Arg = TailCallArgs[i].Arg;
SDValue FIN = TailCallArgs[i].FrameIdxOp;
int FI = TailCallArgs[i].FrameIdx;
MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, FIN,
MachinePointerInfo::getFixedStack(FI),
false, false, 0));
}
}
static SDValue EmitTailCallStoreFPAndRetAddr(SelectionDAG &DAG,
MachineFunction &MF,
SDValue Chain,
SDValue OldRetAddr,
SDValue OldFP,
int SPDiff,
bool isPPC64,
bool isDarwinABI,
DebugLoc dl) {
if (SPDiff) {
int SlotSize = isPPC64 ? 8 : 4;
int NewRetAddrLoc = SPDiff + PPCFrameLowering::getReturnSaveOffset(isPPC64,
isDarwinABI);
int NewRetAddr = MF.getFrameInfo()->CreateFixedObject(SlotSize,
NewRetAddrLoc, true);
EVT VT = isPPC64 ? MVT::i64 : MVT::i32;
SDValue NewRetAddrFrIdx = DAG.getFrameIndex(NewRetAddr, VT);
Chain = DAG.getStore(Chain, dl, OldRetAddr, NewRetAddrFrIdx,
MachinePointerInfo::getFixedStack(NewRetAddr),
false, false, 0);
if (isDarwinABI) {
int NewFPLoc =
SPDiff + PPCFrameLowering::getFramePointerSaveOffset(isPPC64, isDarwinABI);
int NewFPIdx = MF.getFrameInfo()->CreateFixedObject(SlotSize, NewFPLoc,
true);
SDValue NewFramePtrIdx = DAG.getFrameIndex(NewFPIdx, VT);
Chain = DAG.getStore(Chain, dl, OldFP, NewFramePtrIdx,
MachinePointerInfo::getFixedStack(NewFPIdx),
false, false, 0);
}
}
return Chain;
}
static void
CalculateTailCallArgDest(SelectionDAG &DAG, MachineFunction &MF, bool isPPC64,
SDValue Arg, int SPDiff, unsigned ArgOffset,
SmallVector<TailCallArgumentInfo, 8>& TailCallArguments) {
int Offset = ArgOffset + SPDiff;
uint32_t OpSize = (Arg.getValueType().getSizeInBits()+7)/8;
int FI = MF.getFrameInfo()->CreateFixedObject(OpSize, Offset, true);
EVT VT = isPPC64 ? MVT::i64 : MVT::i32;
SDValue FIN = DAG.getFrameIndex(FI, VT);
TailCallArgumentInfo Info;
Info.Arg = Arg;
Info.FrameIdxOp = FIN;
Info.FrameIdx = FI;
TailCallArguments.push_back(Info);
}
SDValue PPCTargetLowering::EmitTailCallLoadFPAndRetAddr(SelectionDAG & DAG,
int SPDiff,
SDValue Chain,
SDValue &LROpOut,
SDValue &FPOpOut,
bool isDarwinABI,
DebugLoc dl) const {
if (SPDiff) {
EVT VT = PPCSubTarget.isPPC64() ? MVT::i64 : MVT::i32;
LROpOut = getReturnAddrFrameIndex(DAG);
LROpOut = DAG.getLoad(VT, dl, Chain, LROpOut, MachinePointerInfo(),
false, false, false, 0);
Chain = SDValue(LROpOut.getNode(), 1);
if (isDarwinABI) {
FPOpOut = getFramePointerFrameIndex(DAG);
FPOpOut = DAG.getLoad(VT, dl, Chain, FPOpOut, MachinePointerInfo(),
false, false, false, 0);
Chain = SDValue(FPOpOut.getNode(), 1);
}
}
return Chain;
}
static SDValue
CreateCopyOfByValArgument(SDValue Src, SDValue Dst, SDValue Chain,
ISD::ArgFlagsTy Flags, SelectionDAG &DAG,
DebugLoc dl) {
SDValue SizeNode = DAG.getConstant(Flags.getByValSize(), MVT::i32);
return DAG.getMemcpy(Chain, dl, Dst, Src, SizeNode, Flags.getByValAlign(),
false, false, MachinePointerInfo(0),
MachinePointerInfo(0));
}
static void
LowerMemOpCallTo(SelectionDAG &DAG, MachineFunction &MF, SDValue Chain,
SDValue Arg, SDValue PtrOff, int SPDiff,
unsigned ArgOffset, bool isPPC64, bool isTailCall,
bool isVector, SmallVector<SDValue, 8> &MemOpChains,
SmallVector<TailCallArgumentInfo, 8> &TailCallArguments,
DebugLoc dl) {
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
if (!isTailCall) {
if (isVector) {
SDValue StackPtr;
if (isPPC64)
StackPtr = DAG.getRegister(PPC::X1, MVT::i64);
else
StackPtr = DAG.getRegister(PPC::R1, MVT::i32);
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr,
DAG.getConstant(ArgOffset, PtrVT));
}
MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff,
MachinePointerInfo(), false, false, 0));
} else CalculateTailCallArgDest(DAG, MF, isPPC64, Arg, SPDiff, ArgOffset,
TailCallArguments);
}
static
void PrepareTailCall(SelectionDAG &DAG, SDValue &InFlag, SDValue &Chain,
DebugLoc dl, bool isPPC64, int SPDiff, unsigned NumBytes,
SDValue LROp, SDValue FPOp, bool isDarwinABI,
SmallVector<TailCallArgumentInfo, 8> &TailCallArguments) {
MachineFunction &MF = DAG.getMachineFunction();
SmallVector<SDValue, 8> MemOpChains2;
InFlag = SDValue();
StoreTailCallArgumentsToStackSlot(DAG, Chain, TailCallArguments,
MemOpChains2, dl);
if (!MemOpChains2.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
&MemOpChains2[0], MemOpChains2.size());
Chain = EmitTailCallStoreFPAndRetAddr(DAG, MF, Chain, LROp, FPOp, SPDiff,
isPPC64, isDarwinABI, dl);
Chain = DAG.getCALLSEQ_END(Chain, DAG.getIntPtrConstant(NumBytes, true),
DAG.getIntPtrConstant(0, true), InFlag);
InFlag = Chain.getValue(1);
}
static
unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
SDValue &Chain, DebugLoc dl, int SPDiff, bool isTailCall,
SmallVector<std::pair<unsigned, SDValue>, 8> &RegsToPass,
SmallVector<SDValue, 8> &Ops, std::vector<EVT> &NodeTys,
const PPCSubtarget &PPCSubTarget) {
bool isPPC64 = PPCSubTarget.isPPC64();
bool isSVR4ABI = PPCSubTarget.isSVR4ABI();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
NodeTys.push_back(MVT::Other); NodeTys.push_back(MVT::Glue);
unsigned CallOpc = isSVR4ABI ? PPCISD::CALL_SVR4 : PPCISD::CALL_Darwin;
bool needIndirectCall = true;
if (SDNode *Dest = isBLACompatibleAddress(Callee, DAG)) {
Callee = SDValue(Dest, 0);
needIndirectCall = false;
}
if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee)) {
if (!DAG.getTarget().getSubtarget<PPCSubtarget>().isJITCodeModel()) {
unsigned OpFlags = 0;
if (DAG.getTarget().getRelocationModel() != Reloc::Static &&
(PPCSubTarget.getTargetTriple().isMacOSX() &&
PPCSubTarget.getTargetTriple().isMacOSXVersionLT(10, 5)) &&
(G->getGlobal()->isDeclaration() ||
G->getGlobal()->isWeakForLinker())) {
OpFlags = PPCII::MO_DARWIN_STUB;
}
Callee = DAG.getTargetGlobalAddress(G->getGlobal(), dl,
Callee.getValueType(),
0, OpFlags);
needIndirectCall = false;
}
}
if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) {
unsigned char OpFlags = 0;
if (DAG.getTarget().getRelocationModel() != Reloc::Static &&
(PPCSubTarget.getTargetTriple().isMacOSX() &&
PPCSubTarget.getTargetTriple().isMacOSXVersionLT(10, 5))) {
OpFlags = PPCII::MO_DARWIN_STUB;
}
Callee = DAG.getTargetExternalSymbol(S->getSymbol(), Callee.getValueType(),
OpFlags);
needIndirectCall = false;
}
if (needIndirectCall) {
SDValue MTCTROps[] = {Chain, Callee, InFlag};
if (isSVR4ABI && isPPC64) {
SDVTList VTs = DAG.getVTList(MVT::i64, MVT::Other, MVT::Glue);
SDValue LoadFuncPtr = DAG.getNode(PPCISD::LOAD, dl, VTs, MTCTROps,
InFlag.getNode() ? 3 : 2);
Chain = LoadFuncPtr.getValue(1);
InFlag = LoadFuncPtr.getValue(2);
SDValue PtrOff = DAG.getIntPtrConstant(16);
SDValue AddPtr = DAG.getNode(ISD::ADD, dl, MVT::i64, Callee, PtrOff);
SDValue LoadEnvPtr = DAG.getNode(PPCISD::LOAD, dl, VTs, Chain, AddPtr,
InFlag);
Chain = LoadEnvPtr.getValue(1);
InFlag = LoadEnvPtr.getValue(2);
SDValue EnvVal = DAG.getCopyToReg(Chain, dl, PPC::X11, LoadEnvPtr,
InFlag);
Chain = EnvVal.getValue(0);
InFlag = EnvVal.getValue(1);
VTs = DAG.getVTList(MVT::Other, MVT::Glue);
SDValue LoadTOCPtr = DAG.getNode(PPCISD::LOAD_TOC, dl, VTs, Chain,
Callee, InFlag);
Chain = LoadTOCPtr.getValue(0);
InFlag = LoadTOCPtr.getValue(1);
MTCTROps[0] = Chain;
MTCTROps[1] = LoadFuncPtr;
MTCTROps[2] = InFlag;
}
Chain = DAG.getNode(PPCISD::MTCTR, dl, NodeTys, MTCTROps,
2 + (InFlag.getNode() != 0));
InFlag = Chain.getValue(1);
NodeTys.clear();
NodeTys.push_back(MVT::Other);
NodeTys.push_back(MVT::Glue);
Ops.push_back(Chain);
CallOpc = isSVR4ABI ? PPCISD::BCTRL_SVR4 : PPCISD::BCTRL_Darwin;
Callee.setNode(0);
if (isTailCall)
Ops.push_back(DAG.getRegister(isPPC64 ? PPC::CTR8 : PPC::CTR, PtrVT));
}
if (Callee.getNode()) {
Ops.push_back(Chain);
Ops.push_back(Callee);
}
if (isTailCall)
Ops.push_back(DAG.getConstant(SPDiff, MVT::i32));
for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i)
Ops.push_back(DAG.getRegister(RegsToPass[i].first,
RegsToPass[i].second.getValueType()));
return CallOpc;
}
static
bool isLocalCall(const SDValue &Callee)
{
if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee))
return !G->getGlobal()->isDeclaration() &&
!G->getGlobal()->isWeakForLinker();
return false;
}
SDValue
PPCTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::InputArg> &Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
SmallVector<CCValAssign, 16> RVLocs;
CCState CCRetInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), RVLocs, *DAG.getContext());
CCRetInfo.AnalyzeCallResult(Ins, RetCC_PPC);
for (unsigned i = 0, e = RVLocs.size(); i != e; ++i) {
CCValAssign &VA = RVLocs[i];
assert(VA.isRegLoc() && "Can only return in registers!");
SDValue Val = DAG.getCopyFromReg(Chain, dl,
VA.getLocReg(), VA.getLocVT(), InFlag);
Chain = Val.getValue(1);
InFlag = Val.getValue(2);
switch (VA.getLocInfo()) {
default: llvm_unreachable("Unknown loc info!");
case CCValAssign::Full: break;
case CCValAssign::AExt:
Val = DAG.getNode(ISD::TRUNCATE, dl, VA.getValVT(), Val);
break;
case CCValAssign::ZExt:
Val = DAG.getNode(ISD::AssertZext, dl, VA.getLocVT(), Val,
DAG.getValueType(VA.getValVT()));
Val = DAG.getNode(ISD::TRUNCATE, dl, VA.getValVT(), Val);
break;
case CCValAssign::SExt:
Val = DAG.getNode(ISD::AssertSext, dl, VA.getLocVT(), Val,
DAG.getValueType(VA.getValVT()));
Val = DAG.getNode(ISD::TRUNCATE, dl, VA.getValVT(), Val);
break;
}
InVals.push_back(Val);
}
return Chain;
}
SDValue
PPCTargetLowering::FinishCall(CallingConv::ID CallConv, DebugLoc dl,
bool isTailCall, bool isVarArg,
SelectionDAG &DAG,
SmallVector<std::pair<unsigned, SDValue>, 8>
&RegsToPass,
SDValue InFlag, SDValue Chain,
SDValue &Callee,
int SPDiff, unsigned NumBytes,
const SmallVectorImpl<ISD::InputArg> &Ins,
SmallVectorImpl<SDValue> &InVals) const {
std::vector<EVT> NodeTys;
SmallVector<SDValue, 8> Ops;
unsigned CallOpc = PrepareCall(DAG, Callee, InFlag, Chain, dl, SPDiff,
isTailCall, RegsToPass, Ops, NodeTys,
PPCSubTarget);
if (isVarArg && PPCSubTarget.isSVR4ABI() && !PPCSubTarget.isPPC64())
Ops.push_back(DAG.getRegister(PPC::CR1EQ, MVT::i32));
int BytesCalleePops =
(CallConv == CallingConv::Fast &&
getTargetMachine().Options.GuaranteedTailCallOpt) ? NumBytes : 0;
const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo();
const uint32_t *Mask = TRI->getCallPreservedMask(CallConv);
assert(Mask && "Missing call preserved mask for calling convention");
Ops.push_back(DAG.getRegisterMask(Mask));
if (InFlag.getNode())
Ops.push_back(InFlag);
if (isTailCall) {
if (DAG.getMachineFunction().getRegInfo().liveout_empty()) {
SmallVector<CCValAssign, 16> RVLocs;
CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), RVLocs, *DAG.getContext());
CCInfo.AnalyzeCallResult(Ins, RetCC_PPC);
for (unsigned i = 0; i != RVLocs.size(); ++i)
DAG.getMachineFunction().getRegInfo().addLiveOut(RVLocs[i].getLocReg());
}
assert(((Callee.getOpcode() == ISD::Register &&
cast<RegisterSDNode>(Callee)->getReg() == PPC::CTR) ||
Callee.getOpcode() == ISD::TargetExternalSymbol ||
Callee.getOpcode() == ISD::TargetGlobalAddress ||
isa<ConstantSDNode>(Callee)) &&
"Expecting an global address, external symbol, absolute value or register");
return DAG.getNode(PPCISD::TC_RETURN, dl, MVT::Other, &Ops[0], Ops.size());
}
bool needsTOCRestore = false;
if (!isTailCall && PPCSubTarget.isSVR4ABI()&& PPCSubTarget.isPPC64()) {
if (CallOpc == PPCISD::BCTRL_SVR4) {
needsTOCRestore = true;
} else if ((CallOpc == PPCISD::CALL_SVR4) && !isLocalCall(Callee)) {
CallOpc = PPCISD::CALL_NOP_SVR4;
}
}
Chain = DAG.getNode(CallOpc, dl, NodeTys, &Ops[0], Ops.size());
InFlag = Chain.getValue(1);
if (needsTOCRestore) {
SDVTList VTs = DAG.getVTList(MVT::Other, MVT::Glue);
Chain = DAG.getNode(PPCISD::TOC_RESTORE, dl, VTs, Chain, InFlag);
InFlag = Chain.getValue(1);
}
Chain = DAG.getCALLSEQ_END(Chain, DAG.getIntPtrConstant(NumBytes, true),
DAG.getIntPtrConstant(BytesCalleePops, true),
InFlag);
if (!Ins.empty())
InFlag = Chain.getValue(1);
return LowerCallResult(Chain, InFlag, CallConv, isVarArg,
Ins, dl, DAG, InVals);
}
SDValue
PPCTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
SmallVectorImpl<SDValue> &InVals) const {
SelectionDAG &DAG = CLI.DAG;
DebugLoc &dl = CLI.DL;
SmallVector<ISD::OutputArg, 32> &Outs = CLI.Outs;
SmallVector<SDValue, 32> &OutVals = CLI.OutVals;
SmallVector<ISD::InputArg, 32> &Ins = CLI.Ins;
SDValue Chain = CLI.Chain;
SDValue Callee = CLI.Callee;
bool &isTailCall = CLI.IsTailCall;
CallingConv::ID CallConv = CLI.CallConv;
bool isVarArg = CLI.IsVarArg;
if (isTailCall)
isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv, isVarArg,
Ins, DAG);
if (PPCSubTarget.isSVR4ABI()) {
if (PPCSubTarget.isPPC64())
return LowerCall_64SVR4(Chain, Callee, CallConv, isVarArg,
isTailCall, Outs, OutVals, Ins,
dl, DAG, InVals);
else
return LowerCall_32SVR4(Chain, Callee, CallConv, isVarArg,
isTailCall, Outs, OutVals, Ins,
dl, DAG, InVals);
}
return LowerCall_Darwin(Chain, Callee, CallConv, isVarArg,
isTailCall, Outs, OutVals, Ins,
dl, DAG, InVals);
}
SDValue
PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
CallingConv::ID CallConv, bool isVarArg,
bool isTailCall,
const SmallVectorImpl<ISD::OutputArg> &Outs,
const SmallVectorImpl<SDValue> &OutVals,
const SmallVectorImpl<ISD::InputArg> &Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
assert((CallConv == CallingConv::C ||
CallConv == CallingConv::Fast) && "Unknown calling convention!");
unsigned PtrByteSize = 4;
MachineFunction &MF = DAG.getMachineFunction();
if (getTargetMachine().Options.GuaranteedTailCallOpt &&
CallConv == CallingConv::Fast)
MF.getInfo<PPCFunctionInfo>()->setHasFastCall();
SmallVector<CCValAssign, 16> ArgLocs;
CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), ArgLocs, *DAG.getContext());
CCInfo.AllocateStack(PPCFrameLowering::getLinkageSize(false, false), PtrByteSize);
if (isVarArg) {
unsigned NumArgs = Outs.size();
for (unsigned i = 0; i != NumArgs; ++i) {
MVT ArgVT = Outs[i].VT;
ISD::ArgFlagsTy ArgFlags = Outs[i].Flags;
bool Result;
if (Outs[i].IsFixed) {
Result = CC_PPC_SVR4(i, ArgVT, ArgVT, CCValAssign::Full, ArgFlags,
CCInfo);
} else {
Result = CC_PPC_SVR4_VarArg(i, ArgVT, ArgVT, CCValAssign::Full,
ArgFlags, CCInfo);
}
if (Result) {
#ifndef NDEBUG
errs() << "Call operand #" << i << " has unhandled type "
<< EVT(ArgVT).getEVTString() << "\n";
#endif
llvm_unreachable(0);
}
}
} else {
CCInfo.AnalyzeCallOperands(Outs, CC_PPC_SVR4);
}
SmallVector<CCValAssign, 16> ByValArgLocs;
CCState CCByValInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), ByValArgLocs, *DAG.getContext());
CCByValInfo.AllocateStack(CCInfo.getNextStackOffset(), PtrByteSize);
CCByValInfo.AnalyzeCallOperands(Outs, CC_PPC_SVR4_ByVal);
unsigned NumBytes = CCByValInfo.getNextStackOffset();
int SPDiff = CalculateTailCallSPDiff(DAG, isTailCall, NumBytes);
Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, true));
SDValue CallSeqStart = Chain;
SDValue LROp, FPOp;
Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, false,
dl);
SDValue StackPtr = DAG.getRegister(PPC::R1, MVT::i32);
SmallVector<std::pair<unsigned, SDValue>, 8> RegsToPass;
SmallVector<TailCallArgumentInfo, 8> TailCallArguments;
SmallVector<SDValue, 8> MemOpChains;
bool seenFloatArg = false;
for (unsigned i = 0, j = 0, e = ArgLocs.size();
i != e;
++i) {
CCValAssign &VA = ArgLocs[i];
SDValue Arg = OutVals[i];
ISD::ArgFlagsTy Flags = Outs[i].Flags;
if (Flags.isByVal()) {
assert((j < ByValArgLocs.size()) && "Index out of bounds!");
CCValAssign &ByValVA = ByValArgLocs[j++];
assert((VA.getValNo() == ByValVA.getValNo()) && "ValNo mismatch!");
unsigned LocMemOffset = ByValVA.getLocMemOffset();
SDValue PtrOff = DAG.getIntPtrConstant(LocMemOffset);
PtrOff = DAG.getNode(ISD::ADD, dl, getPointerTy(), StackPtr, PtrOff);
SDValue MemcpyCall =
CreateCopyOfByValArgument(Arg, PtrOff,
CallSeqStart.getNode()->getOperand(0),
Flags, DAG, dl);
SDValue NewCallSeqStart = DAG.getCALLSEQ_START(MemcpyCall,
CallSeqStart.getNode()->getOperand(1));
DAG.ReplaceAllUsesWith(CallSeqStart.getNode(),
NewCallSeqStart.getNode());
Chain = CallSeqStart = NewCallSeqStart;
Arg = PtrOff;
}
if (VA.isRegLoc()) {
seenFloatArg |= VA.getLocVT().isFloatingPoint();
RegsToPass.push_back(std::make_pair(VA.getLocReg(), Arg));
} else {
assert(VA.isMemLoc());
unsigned LocMemOffset = VA.getLocMemOffset();
if (!isTailCall) {
SDValue PtrOff = DAG.getIntPtrConstant(LocMemOffset);
PtrOff = DAG.getNode(ISD::ADD, dl, getPointerTy(), StackPtr, PtrOff);
MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff,
MachinePointerInfo(),
false, false, 0));
} else {
CalculateTailCallArgDest(DAG, MF, false, Arg, SPDiff, LocMemOffset,
TailCallArguments);
}
}
}
if (!MemOpChains.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
&MemOpChains[0], MemOpChains.size());
SDValue InFlag;
for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) {
Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
RegsToPass[i].second, InFlag);
InFlag = Chain.getValue(1);
}
if (isVarArg) {
SDVTList VTs = DAG.getVTList(MVT::Other, MVT::Glue);
SDValue Ops[] = { Chain, InFlag };
Chain = DAG.getNode(seenFloatArg ? PPCISD::CR6SET : PPCISD::CR6UNSET,
dl, VTs, Ops, InFlag.getNode() ? 2 : 1);
InFlag = Chain.getValue(1);
}
if (isTailCall)
PrepareTailCall(DAG, InFlag, Chain, dl, false, SPDiff, NumBytes, LROp, FPOp,
false, TailCallArguments);
return FinishCall(CallConv, dl, isTailCall, isVarArg, DAG,
RegsToPass, InFlag, Chain, Callee, SPDiff, NumBytes,
Ins, InVals);
}
SDValue
PPCTargetLowering::createMemcpyOutsideCallSeq(SDValue Arg, SDValue PtrOff,
SDValue CallSeqStart,
ISD::ArgFlagsTy Flags,
SelectionDAG &DAG,
DebugLoc dl) const {
SDValue MemcpyCall = CreateCopyOfByValArgument(Arg, PtrOff,
CallSeqStart.getNode()->getOperand(0),
Flags, DAG, dl);
SDValue NewCallSeqStart = DAG.getCALLSEQ_START(MemcpyCall,
CallSeqStart.getNode()->getOperand(1));
DAG.ReplaceAllUsesWith(CallSeqStart.getNode(),
NewCallSeqStart.getNode());
return NewCallSeqStart;
}
SDValue
PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
CallingConv::ID CallConv, bool isVarArg,
bool isTailCall,
const SmallVectorImpl<ISD::OutputArg> &Outs,
const SmallVectorImpl<SDValue> &OutVals,
const SmallVectorImpl<ISD::InputArg> &Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
unsigned NumOps = Outs.size();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
unsigned PtrByteSize = 8;
MachineFunction &MF = DAG.getMachineFunction();
if (getTargetMachine().Options.GuaranteedTailCallOpt &&
CallConv == CallingConv::Fast)
MF.getInfo<PPCFunctionInfo>()->setHasFastCall();
unsigned nAltivecParamsAtEnd = 0;
unsigned NumBytes =
CalculateParameterAndLinkageAreaSize(DAG, true, isVarArg, CallConv,
Outs, OutVals, nAltivecParamsAtEnd);
int SPDiff = CalculateTailCallSPDiff(DAG, isTailCall, NumBytes);
if (isTailCall)
Chain = DAG.getStackArgumentTokenFactor(Chain);
Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, true));
SDValue CallSeqStart = Chain;
SDValue LROp, FPOp;
Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, true,
dl);
SDValue StackPtr = DAG.getRegister(PPC::X1, MVT::i64);
unsigned ArgOffset = PPCFrameLowering::getLinkageSize(true, true);
unsigned GPR_idx = 0, FPR_idx = 0, VR_idx = 0;
static const uint16_t GPR[] = {
PPC::X3, PPC::X4, PPC::X5, PPC::X6,
PPC::X7, PPC::X8, PPC::X9, PPC::X10,
};
static const uint16_t *FPR = GetFPR();
static const uint16_t VR[] = {
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
const unsigned NumGPRs = array_lengthof(GPR);
const unsigned NumFPRs = 13;
const unsigned NumVRs = array_lengthof(VR);
SmallVector<std::pair<unsigned, SDValue>, 8> RegsToPass;
SmallVector<TailCallArgumentInfo, 8> TailCallArguments;
SmallVector<SDValue, 8> MemOpChains;
for (unsigned i = 0; i != NumOps; ++i) {
SDValue Arg = OutVals[i];
ISD::ArgFlagsTy Flags = Outs[i].Flags;
SDValue PtrOff;
PtrOff = DAG.getConstant(ArgOffset, StackPtr.getValueType());
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr, PtrOff);
if (Arg.getValueType() == MVT::i32) {
unsigned ExtOp = Flags.isSExt() ? ISD::SIGN_EXTEND : ISD::ZERO_EXTEND;
Arg = DAG.getNode(ExtOp, dl, MVT::i64, Arg);
}
if (Flags.isByVal()) {
unsigned Size = Flags.getByValSize();
if (Size == 0)
continue;
if (Size==1 || Size==2 || Size==4) {
EVT VT = (Size==1) ? MVT::i8 : ((Size==2) ? MVT::i16 : MVT::i32);
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getExtLoad(ISD::EXTLOAD, dl, PtrVT, Chain, Arg,
MachinePointerInfo(), VT,
false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
ArgOffset += PtrByteSize;
continue;
}
}
if (GPR_idx == NumGPRs && Size < 8) {
SDValue Const = DAG.getConstant(PtrByteSize - Size,
PtrOff.getValueType());
SDValue AddPtr = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, Const);
Chain = CallSeqStart = createMemcpyOutsideCallSeq(Arg, AddPtr,
CallSeqStart,
Flags, DAG, dl);
ArgOffset += PtrByteSize;
continue;
}
if (Size >= 8)
Chain = CallSeqStart = createMemcpyOutsideCallSeq(Arg, PtrOff,
CallSeqStart,
Flags, DAG, dl);
if (Size < 8 && GPR_idx != NumGPRs) {
SDValue Const = DAG.getConstant(8 - Size, PtrOff.getValueType());
SDValue AddPtr = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, Const);
Chain = CallSeqStart = createMemcpyOutsideCallSeq(Arg, AddPtr,
CallSeqStart,
Flags, DAG, dl);
SDValue Load = DAG.getLoad(PtrVT, dl, Chain, PtrOff,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
ArgOffset += PtrByteSize;
continue;
}
for (unsigned j=0; j<Size; j+=PtrByteSize) {
SDValue Const = DAG.getConstant(j, PtrOff.getValueType());
SDValue AddArg = DAG.getNode(ISD::ADD, dl, PtrVT, Arg, Const);
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getLoad(PtrVT, dl, Chain, AddArg,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
ArgOffset += PtrByteSize;
} else {
ArgOffset += ((Size - j + PtrByteSize-1)/PtrByteSize)*PtrByteSize;
break;
}
}
continue;
}
switch (Arg.getValueType().getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unexpected ValueType for argument!");
case MVT::i32:
case MVT::i64:
if (GPR_idx != NumGPRs) {
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Arg));
} else {
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
true, isTailCall, false, MemOpChains,
TailCallArguments, dl);
}
ArgOffset += PtrByteSize;
break;
case MVT::f32:
case MVT::f64:
if (FPR_idx != NumFPRs) {
RegsToPass.push_back(std::make_pair(FPR[FPR_idx++], Arg));
if (isVarArg) {
SDValue StoreOff;
if (Arg.getValueType().getSimpleVT().SimpleTy == MVT::f32) {
SDValue ConstFour = DAG.getConstant(4, PtrOff.getValueType());
StoreOff = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, ConstFour);
} else
StoreOff = PtrOff;
SDValue Store = DAG.getStore(Chain, dl, Arg, StoreOff,
MachinePointerInfo(), false, false, 0);
MemOpChains.push_back(Store);
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getLoad(PtrVT, dl, Store, PtrOff,
MachinePointerInfo(), false, false,
false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
}
} else if (GPR_idx != NumGPRs)
++GPR_idx;
} else {
if (Arg.getValueType() == MVT::f32) {
SDValue ConstFour = DAG.getConstant(4, PtrOff.getValueType());
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, ConstFour);
}
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
true, isTailCall, false, MemOpChains,
TailCallArguments, dl);
}
ArgOffset += 8;
break;
case MVT::v4f32:
case MVT::v4i32:
case MVT::v8i16:
case MVT::v16i8:
if (isVarArg) {
while (ArgOffset % 16 !=0) {
ArgOffset += PtrByteSize;
if (GPR_idx != NumGPRs)
GPR_idx++;
}
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr,
DAG.getConstant(ArgOffset, PtrVT));
SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
MachinePointerInfo(), false, false, 0);
MemOpChains.push_back(Store);
if (VR_idx != NumVRs) {
SDValue Load = DAG.getLoad(MVT::v4f32, dl, Store, PtrOff,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(VR[VR_idx++], Load));
}
ArgOffset += 16;
for (unsigned i=0; i<16; i+=PtrByteSize) {
if (GPR_idx == NumGPRs)
break;
SDValue Ix = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff,
DAG.getConstant(i, PtrVT));
SDValue Load = DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
}
break;
}
if (VR_idx != NumVRs) {
RegsToPass.push_back(std::make_pair(VR[VR_idx++], Arg));
} else {
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
true, isTailCall, true, MemOpChains,
TailCallArguments, dl);
ArgOffset += 16;
}
break;
}
}
if (!MemOpChains.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
&MemOpChains[0], MemOpChains.size());
if (!isTailCall &&
!dyn_cast<GlobalAddressSDNode>(Callee) &&
!dyn_cast<ExternalSymbolSDNode>(Callee) &&
!isBLACompatibleAddress(Callee, DAG)) {
SDValue Val = DAG.getCopyFromReg(Chain, dl, PPC::X2, MVT::i64);
SDValue PtrOff = DAG.getIntPtrConstant(40);
SDValue AddPtr = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr, PtrOff);
Chain = DAG.getStore(Val.getValue(1), dl, Val, AddPtr, MachinePointerInfo(),
false, false, 0);
RegsToPass.push_back(std::make_pair((unsigned)PPC::X12, Callee));
}
SDValue InFlag;
for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) {
Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
RegsToPass[i].second, InFlag);
InFlag = Chain.getValue(1);
}
if (isTailCall)
PrepareTailCall(DAG, InFlag, Chain, dl, true, SPDiff, NumBytes, LROp,
FPOp, true, TailCallArguments);
return FinishCall(CallConv, dl, isTailCall, isVarArg, DAG,
RegsToPass, InFlag, Chain, Callee, SPDiff, NumBytes,
Ins, InVals);
}
SDValue
PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
CallingConv::ID CallConv, bool isVarArg,
bool isTailCall,
const SmallVectorImpl<ISD::OutputArg> &Outs,
const SmallVectorImpl<SDValue> &OutVals,
const SmallVectorImpl<ISD::InputArg> &Ins,
DebugLoc dl, SelectionDAG &DAG,
SmallVectorImpl<SDValue> &InVals) const {
unsigned NumOps = Outs.size();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isPPC64 = PtrVT == MVT::i64;
unsigned PtrByteSize = isPPC64 ? 8 : 4;
MachineFunction &MF = DAG.getMachineFunction();
if (getTargetMachine().Options.GuaranteedTailCallOpt &&
CallConv == CallingConv::Fast)
MF.getInfo<PPCFunctionInfo>()->setHasFastCall();
unsigned nAltivecParamsAtEnd = 0;
unsigned NumBytes =
CalculateParameterAndLinkageAreaSize(DAG, isPPC64, isVarArg, CallConv,
Outs, OutVals,
nAltivecParamsAtEnd);
int SPDiff = CalculateTailCallSPDiff(DAG, isTailCall, NumBytes);
if (isTailCall)
Chain = DAG.getStackArgumentTokenFactor(Chain);
Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, true));
SDValue CallSeqStart = Chain;
SDValue LROp, FPOp;
Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, true,
dl);
SDValue StackPtr;
if (isPPC64)
StackPtr = DAG.getRegister(PPC::X1, MVT::i64);
else
StackPtr = DAG.getRegister(PPC::R1, MVT::i32);
unsigned ArgOffset = PPCFrameLowering::getLinkageSize(isPPC64, true);
unsigned GPR_idx = 0, FPR_idx = 0, VR_idx = 0;
static const uint16_t GPR_32[] = { PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
static const uint16_t GPR_64[] = { PPC::X3, PPC::X4, PPC::X5, PPC::X6,
PPC::X7, PPC::X8, PPC::X9, PPC::X10,
};
static const uint16_t *FPR = GetFPR();
static const uint16_t VR[] = {
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
const unsigned NumGPRs = array_lengthof(GPR_32);
const unsigned NumFPRs = 13;
const unsigned NumVRs = array_lengthof(VR);
const uint16_t *GPR = isPPC64 ? GPR_64 : GPR_32;
SmallVector<std::pair<unsigned, SDValue>, 8> RegsToPass;
SmallVector<TailCallArgumentInfo, 8> TailCallArguments;
SmallVector<SDValue, 8> MemOpChains;
for (unsigned i = 0; i != NumOps; ++i) {
SDValue Arg = OutVals[i];
ISD::ArgFlagsTy Flags = Outs[i].Flags;
SDValue PtrOff;
PtrOff = DAG.getConstant(ArgOffset, StackPtr.getValueType());
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr, PtrOff);
if (isPPC64 && Arg.getValueType() == MVT::i32) {
unsigned ExtOp = Flags.isSExt() ? ISD::SIGN_EXTEND : ISD::ZERO_EXTEND;
Arg = DAG.getNode(ExtOp, dl, MVT::i64, Arg);
}
if (Flags.isByVal()) {
unsigned Size = Flags.getByValSize();
if (Size==1 || Size==2) {
EVT VT = (Size==1) ? MVT::i8 : MVT::i16;
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getExtLoad(ISD::EXTLOAD, dl, PtrVT, Chain, Arg,
MachinePointerInfo(), VT,
false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
ArgOffset += PtrByteSize;
} else {
SDValue Const = DAG.getConstant(PtrByteSize - Size,
PtrOff.getValueType());
SDValue AddPtr = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, Const);
Chain = CallSeqStart = createMemcpyOutsideCallSeq(Arg, AddPtr,
CallSeqStart,
Flags, DAG, dl);
ArgOffset += PtrByteSize;
}
continue;
}
Chain = CallSeqStart = createMemcpyOutsideCallSeq(Arg, PtrOff,
CallSeqStart,
Flags, DAG, dl);
for (unsigned j=0; j<Size; j+=PtrByteSize) {
SDValue Const = DAG.getConstant(j, PtrOff.getValueType());
SDValue AddArg = DAG.getNode(ISD::ADD, dl, PtrVT, Arg, Const);
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getLoad(PtrVT, dl, Chain, AddArg,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
ArgOffset += PtrByteSize;
} else {
ArgOffset += ((Size - j + PtrByteSize-1)/PtrByteSize)*PtrByteSize;
break;
}
}
continue;
}
switch (Arg.getValueType().getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unexpected ValueType for argument!");
case MVT::i32:
case MVT::i64:
if (GPR_idx != NumGPRs) {
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Arg));
} else {
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
isPPC64, isTailCall, false, MemOpChains,
TailCallArguments, dl);
}
ArgOffset += PtrByteSize;
break;
case MVT::f32:
case MVT::f64:
if (FPR_idx != NumFPRs) {
RegsToPass.push_back(std::make_pair(FPR[FPR_idx++], Arg));
if (isVarArg) {
SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
MachinePointerInfo(), false, false, 0);
MemOpChains.push_back(Store);
if (GPR_idx != NumGPRs) {
SDValue Load = DAG.getLoad(PtrVT, dl, Store, PtrOff,
MachinePointerInfo(), false, false,
false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
}
if (GPR_idx != NumGPRs && Arg.getValueType() == MVT::f64 && !isPPC64){
SDValue ConstFour = DAG.getConstant(4, PtrOff.getValueType());
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, ConstFour);
SDValue Load = DAG.getLoad(PtrVT, dl, Store, PtrOff,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
}
} else {
if (GPR_idx != NumGPRs)
++GPR_idx;
if (GPR_idx != NumGPRs && Arg.getValueType() == MVT::f64 &&
!isPPC64) ++GPR_idx;
}
} else
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
isPPC64, isTailCall, false, MemOpChains,
TailCallArguments, dl);
if (isPPC64)
ArgOffset += 8;
else
ArgOffset += Arg.getValueType() == MVT::f32 ? 4 : 8;
break;
case MVT::v4f32:
case MVT::v4i32:
case MVT::v8i16:
case MVT::v16i8:
if (isVarArg) {
while (ArgOffset % 16 !=0) {
ArgOffset += PtrByteSize;
if (GPR_idx != NumGPRs)
GPR_idx++;
}
PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr,
DAG.getConstant(ArgOffset, PtrVT));
SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
MachinePointerInfo(), false, false, 0);
MemOpChains.push_back(Store);
if (VR_idx != NumVRs) {
SDValue Load = DAG.getLoad(MVT::v4f32, dl, Store, PtrOff,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(VR[VR_idx++], Load));
}
ArgOffset += 16;
for (unsigned i=0; i<16; i+=PtrByteSize) {
if (GPR_idx == NumGPRs)
break;
SDValue Ix = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff,
DAG.getConstant(i, PtrVT));
SDValue Load = DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
}
break;
}
if (VR_idx != NumVRs) {
RegsToPass.push_back(std::make_pair(VR[VR_idx++], Arg));
} else if (nAltivecParamsAtEnd==0) {
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
isPPC64, isTailCall, true, MemOpChains,
TailCallArguments, dl);
ArgOffset += 16;
}
break;
}
}
if (!isVarArg && nAltivecParamsAtEnd > NumVRs) {
unsigned j = 0;
ArgOffset = ((ArgOffset+15)/16)*16;
ArgOffset += 12*16;
for (unsigned i = 0; i != NumOps; ++i) {
SDValue Arg = OutVals[i];
EVT ArgType = Outs[i].VT;
if (ArgType==MVT::v4f32 || ArgType==MVT::v4i32 ||
ArgType==MVT::v8i16 || ArgType==MVT::v16i8) {
if (++j > NumVRs) {
SDValue PtrOff;
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
isPPC64, isTailCall, true, MemOpChains,
TailCallArguments, dl);
ArgOffset += 16;
}
}
}
}
if (!MemOpChains.empty())
Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
&MemOpChains[0], MemOpChains.size());
if (!isTailCall &&
!dyn_cast<GlobalAddressSDNode>(Callee) &&
!dyn_cast<ExternalSymbolSDNode>(Callee) &&
!isBLACompatibleAddress(Callee, DAG))
RegsToPass.push_back(std::make_pair((unsigned)(isPPC64 ? PPC::X12 :
PPC::R12), Callee));
SDValue InFlag;
for (unsigned i = 0, e = RegsToPass.size(); i != e; ++i) {
Chain = DAG.getCopyToReg(Chain, dl, RegsToPass[i].first,
RegsToPass[i].second, InFlag);
InFlag = Chain.getValue(1);
}
if (isTailCall)
PrepareTailCall(DAG, InFlag, Chain, dl, isPPC64, SPDiff, NumBytes, LROp,
FPOp, true, TailCallArguments);
return FinishCall(CallConv, dl, isTailCall, isVarArg, DAG,
RegsToPass, InFlag, Chain, Callee, SPDiff, NumBytes,
Ins, InVals);
}
bool
PPCTargetLowering::CanLowerReturn(CallingConv::ID CallConv,
MachineFunction &MF, bool isVarArg,
const SmallVectorImpl<ISD::OutputArg> &Outs,
LLVMContext &Context) const {
SmallVector<CCValAssign, 16> RVLocs;
CCState CCInfo(CallConv, isVarArg, MF, getTargetMachine(),
RVLocs, Context);
return CCInfo.CheckReturn(Outs, RetCC_PPC);
}
SDValue
PPCTargetLowering::LowerReturn(SDValue Chain,
CallingConv::ID CallConv, bool isVarArg,
const SmallVectorImpl<ISD::OutputArg> &Outs,
const SmallVectorImpl<SDValue> &OutVals,
DebugLoc dl, SelectionDAG &DAG) const {
SmallVector<CCValAssign, 16> RVLocs;
CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(),
getTargetMachine(), RVLocs, *DAG.getContext());
CCInfo.AnalyzeReturn(Outs, RetCC_PPC);
if (DAG.getMachineFunction().getRegInfo().liveout_empty()) {
for (unsigned i = 0; i != RVLocs.size(); ++i)
DAG.getMachineFunction().getRegInfo().addLiveOut(RVLocs[i].getLocReg());
}
SDValue Flag;
for (unsigned i = 0; i != RVLocs.size(); ++i) {
CCValAssign &VA = RVLocs[i];
assert(VA.isRegLoc() && "Can only return in registers!");
SDValue Arg = OutVals[i];
switch (VA.getLocInfo()) {
default: llvm_unreachable("Unknown loc info!");
case CCValAssign::Full: break;
case CCValAssign::AExt:
Arg = DAG.getNode(ISD::ANY_EXTEND, dl, VA.getLocVT(), Arg);
break;
case CCValAssign::ZExt:
Arg = DAG.getNode(ISD::ZERO_EXTEND, dl, VA.getLocVT(), Arg);
break;
case CCValAssign::SExt:
Arg = DAG.getNode(ISD::SIGN_EXTEND, dl, VA.getLocVT(), Arg);
break;
}
Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), Arg, Flag);
Flag = Chain.getValue(1);
}
if (Flag.getNode())
return DAG.getNode(PPCISD::RET_FLAG, dl, MVT::Other, Chain, Flag);
else
return DAG.getNode(PPCISD::RET_FLAG, dl, MVT::Other, Chain);
}
SDValue PPCTargetLowering::LowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG,
const PPCSubtarget &Subtarget) const {
DebugLoc dl = Op.getDebugLoc();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isPPC64 = Subtarget.isPPC64();
unsigned SP = isPPC64 ? PPC::X1 : PPC::R1;
SDValue StackPtr = DAG.getRegister(SP, PtrVT);
SDValue Chain = Op.getOperand(0);
SDValue SaveSP = Op.getOperand(1);
SDValue LoadLinkSP = DAG.getLoad(PtrVT, dl, Chain, StackPtr,
MachinePointerInfo(),
false, false, false, 0);
Chain = DAG.getCopyToReg(LoadLinkSP.getValue(1), dl, SP, SaveSP);
return DAG.getStore(Chain, dl, LoadLinkSP, StackPtr, MachinePointerInfo(),
false, false, 0);
}
SDValue
PPCTargetLowering::getReturnAddrFrameIndex(SelectionDAG & DAG) const {
MachineFunction &MF = DAG.getMachineFunction();
bool isPPC64 = PPCSubTarget.isPPC64();
bool isDarwinABI = PPCSubTarget.isDarwinABI();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
PPCFunctionInfo *FI = MF.getInfo<PPCFunctionInfo>();
int RASI = FI->getReturnAddrSaveIndex();
if (!RASI) {
int LROffset = PPCFrameLowering::getReturnSaveOffset(isPPC64, isDarwinABI);
RASI = MF.getFrameInfo()->CreateFixedObject(isPPC64? 8 : 4, LROffset, true);
FI->setReturnAddrSaveIndex(RASI);
}
return DAG.getFrameIndex(RASI, PtrVT);
}
SDValue
PPCTargetLowering::getFramePointerFrameIndex(SelectionDAG & DAG) const {
MachineFunction &MF = DAG.getMachineFunction();
bool isPPC64 = PPCSubTarget.isPPC64();
bool isDarwinABI = PPCSubTarget.isDarwinABI();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
PPCFunctionInfo *FI = MF.getInfo<PPCFunctionInfo>();
int FPSI = FI->getFramePointerSaveIndex();
if (!FPSI) {
int FPOffset = PPCFrameLowering::getFramePointerSaveOffset(isPPC64,
isDarwinABI);
FPSI = MF.getFrameInfo()->CreateFixedObject(isPPC64? 8 : 4, FPOffset, true);
FI->setFramePointerSaveIndex(FPSI);
}
return DAG.getFrameIndex(FPSI, PtrVT);
}
SDValue PPCTargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op,
SelectionDAG &DAG,
const PPCSubtarget &Subtarget) const {
SDValue Chain = Op.getOperand(0);
SDValue Size = Op.getOperand(1);
DebugLoc dl = Op.getDebugLoc();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue NegSize = DAG.getNode(ISD::SUB, dl, PtrVT,
DAG.getConstant(0, PtrVT), Size);
SDValue FPSIdx = getFramePointerFrameIndex(DAG);
SDValue Ops[3] = { Chain, NegSize, FPSIdx };
SDVTList VTs = DAG.getVTList(PtrVT, MVT::Other);
return DAG.getNode(PPCISD::DYNALLOC, dl, VTs, Ops, 3);
}
SDValue PPCTargetLowering::LowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const {
if (!Op.getOperand(0).getValueType().isFloatingPoint() ||
!Op.getOperand(2).getValueType().isFloatingPoint())
return Op;
ISD::CondCode CC = cast<CondCodeSDNode>(Op.getOperand(4))->get();
if (CC == ISD::SETEQ || CC == ISD::SETNE) return Op;
EVT ResVT = Op.getValueType();
EVT CmpVT = Op.getOperand(0).getValueType();
SDValue LHS = Op.getOperand(0), RHS = Op.getOperand(1);
SDValue TV = Op.getOperand(2), FV = Op.getOperand(3);
DebugLoc dl = Op.getDebugLoc();
if (isFloatingPointZero(RHS))
switch (CC) {
default: break; case ISD::SETULT:
case ISD::SETLT:
std::swap(TV, FV); case ISD::SETOGE:
case ISD::SETGE:
if (LHS.getValueType() == MVT::f32) LHS = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, LHS);
return DAG.getNode(PPCISD::FSEL, dl, ResVT, LHS, TV, FV);
case ISD::SETUGT:
case ISD::SETGT:
std::swap(TV, FV); case ISD::SETOLE:
case ISD::SETLE:
if (LHS.getValueType() == MVT::f32) LHS = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, LHS);
return DAG.getNode(PPCISD::FSEL, dl, ResVT,
DAG.getNode(ISD::FNEG, dl, MVT::f64, LHS), TV, FV);
}
SDValue Cmp;
switch (CC) {
default: break; case ISD::SETULT:
case ISD::SETLT:
Cmp = DAG.getNode(ISD::FSUB, dl, CmpVT, LHS, RHS);
if (Cmp.getValueType() == MVT::f32) Cmp = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Cmp);
return DAG.getNode(PPCISD::FSEL, dl, ResVT, Cmp, FV, TV);
case ISD::SETOGE:
case ISD::SETGE:
Cmp = DAG.getNode(ISD::FSUB, dl, CmpVT, LHS, RHS);
if (Cmp.getValueType() == MVT::f32) Cmp = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Cmp);
return DAG.getNode(PPCISD::FSEL, dl, ResVT, Cmp, TV, FV);
case ISD::SETUGT:
case ISD::SETGT:
Cmp = DAG.getNode(ISD::FSUB, dl, CmpVT, RHS, LHS);
if (Cmp.getValueType() == MVT::f32) Cmp = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Cmp);
return DAG.getNode(PPCISD::FSEL, dl, ResVT, Cmp, FV, TV);
case ISD::SETOLE:
case ISD::SETLE:
Cmp = DAG.getNode(ISD::FSUB, dl, CmpVT, RHS, LHS);
if (Cmp.getValueType() == MVT::f32) Cmp = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Cmp);
return DAG.getNode(PPCISD::FSEL, dl, ResVT, Cmp, TV, FV);
}
return Op;
}
SDValue PPCTargetLowering::LowerFP_TO_INT(SDValue Op, SelectionDAG &DAG,
DebugLoc dl) const {
assert(Op.getOperand(0).getValueType().isFloatingPoint());
SDValue Src = Op.getOperand(0);
if (Src.getValueType() == MVT::f32)
Src = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Src);
SDValue Tmp;
switch (Op.getValueType().getSimpleVT().SimpleTy) {
default: llvm_unreachable("Unhandled FP_TO_INT type in custom expander!");
case MVT::i32:
Tmp = DAG.getNode(Op.getOpcode()==ISD::FP_TO_SINT ? PPCISD::FCTIWZ :
PPCISD::FCTIDZ,
dl, MVT::f64, Src);
break;
case MVT::i64:
Tmp = DAG.getNode(PPCISD::FCTIDZ, dl, MVT::f64, Src);
break;
}
SDValue FIPtr = DAG.CreateStackTemporary(MVT::f64);
SDValue Chain = DAG.getStore(DAG.getEntryNode(), dl, Tmp, FIPtr,
MachinePointerInfo(), false, false, 0);
if (Op.getValueType() == MVT::i32)
FIPtr = DAG.getNode(ISD::ADD, dl, FIPtr.getValueType(), FIPtr,
DAG.getConstant(4, FIPtr.getValueType()));
return DAG.getLoad(Op.getValueType(), dl, Chain, FIPtr, MachinePointerInfo(),
false, false, false, 0);
}
SDValue PPCTargetLowering::LowerSINT_TO_FP(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
if (Op.getValueType() != MVT::f32 && Op.getValueType() != MVT::f64)
return SDValue();
if (Op.getOperand(0).getValueType() == MVT::i64) {
SDValue SINT = Op.getOperand(0);
if (Op.getValueType() == MVT::f32 &&
!DAG.getTarget().Options.UnsafeFPMath) {
SDValue Round = DAG.getNode(ISD::AND, dl, MVT::i64,
SINT, DAG.getConstant(2047, MVT::i64));
Round = DAG.getNode(ISD::ADD, dl, MVT::i64,
Round, DAG.getConstant(2047, MVT::i64));
Round = DAG.getNode(ISD::OR, dl, MVT::i64, Round, SINT);
Round = DAG.getNode(ISD::AND, dl, MVT::i64,
Round, DAG.getConstant(-2048, MVT::i64));
SDValue Cond = DAG.getNode(ISD::SRA, dl, MVT::i64,
SINT, DAG.getConstant(53, MVT::i32));
Cond = DAG.getNode(ISD::ADD, dl, MVT::i64,
Cond, DAG.getConstant(1, MVT::i64));
Cond = DAG.getSetCC(dl, MVT::i32,
Cond, DAG.getConstant(1, MVT::i64), ISD::SETUGT);
SINT = DAG.getNode(ISD::SELECT, dl, MVT::i64, Cond, Round, SINT);
}
SDValue Bits = DAG.getNode(ISD::BITCAST, dl, MVT::f64, SINT);
SDValue FP = DAG.getNode(PPCISD::FCFID, dl, MVT::f64, Bits);
if (Op.getValueType() == MVT::f32)
FP = DAG.getNode(ISD::FP_ROUND, dl,
MVT::f32, FP, DAG.getIntPtrConstant(0));
return FP;
}
assert(Op.getOperand(0).getValueType() == MVT::i32 &&
"Unhandled SINT_TO_FP type in custom expander!");
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *FrameInfo = MF.getFrameInfo();
int FrameIdx = FrameInfo->CreateStackObject(8, 8, false);
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
SDValue Ext64 = DAG.getNode(PPCISD::EXTSW_32, dl, MVT::i32,
Op.getOperand(0));
MachineMemOperand *MMO =
MF.getMachineMemOperand(MachinePointerInfo::getFixedStack(FrameIdx),
MachineMemOperand::MOStore, 8, 8);
SDValue Ops[] = { DAG.getEntryNode(), Ext64, FIdx };
SDValue Store =
DAG.getMemIntrinsicNode(PPCISD::STD_32, dl, DAG.getVTList(MVT::Other),
Ops, 4, MVT::i64, MMO);
SDValue Ld = DAG.getLoad(MVT::f64, dl, Store, FIdx, MachinePointerInfo(),
false, false, false, 0);
SDValue FP = DAG.getNode(PPCISD::FCFID, dl, MVT::f64, Ld);
if (Op.getValueType() == MVT::f32)
FP = DAG.getNode(ISD::FP_ROUND, dl, MVT::f32, FP, DAG.getIntPtrConstant(0));
return FP;
}
SDValue PPCTargetLowering::LowerFLT_ROUNDS_(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
MachineFunction &MF = DAG.getMachineFunction();
EVT VT = Op.getValueType();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
SDValue MFFSreg, InFlag;
EVT NodeTys[] = {
MVT::f64, MVT::Glue };
SDValue Chain = DAG.getNode(PPCISD::MFFS, dl, NodeTys, &InFlag, 0);
int SSFI = MF.getFrameInfo()->CreateStackObject(8, 8, false);
SDValue StackSlot = DAG.getFrameIndex(SSFI, PtrVT);
SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Chain,
StackSlot, MachinePointerInfo(), false, false,0);
SDValue Four = DAG.getConstant(4, PtrVT);
SDValue Addr = DAG.getNode(ISD::ADD, dl, PtrVT, StackSlot, Four);
SDValue CWD = DAG.getLoad(MVT::i32, dl, Store, Addr, MachinePointerInfo(),
false, false, false, 0);
SDValue CWD1 =
DAG.getNode(ISD::AND, dl, MVT::i32,
CWD, DAG.getConstant(3, MVT::i32));
SDValue CWD2 =
DAG.getNode(ISD::SRL, dl, MVT::i32,
DAG.getNode(ISD::AND, dl, MVT::i32,
DAG.getNode(ISD::XOR, dl, MVT::i32,
CWD, DAG.getConstant(3, MVT::i32)),
DAG.getConstant(3, MVT::i32)),
DAG.getConstant(1, MVT::i32));
SDValue RetVal =
DAG.getNode(ISD::XOR, dl, MVT::i32, CWD1, CWD2);
return DAG.getNode((VT.getSizeInBits() < 16 ?
ISD::TRUNCATE : ISD::ZERO_EXTEND), dl, VT, RetVal);
}
SDValue PPCTargetLowering::LowerSHL_PARTS(SDValue Op, SelectionDAG &DAG) const {
EVT VT = Op.getValueType();
unsigned BitWidth = VT.getSizeInBits();
DebugLoc dl = Op.getDebugLoc();
assert(Op.getNumOperands() == 3 &&
VT == Op.getOperand(1).getValueType() &&
"Unexpected SHL!");
SDValue Lo = Op.getOperand(0);
SDValue Hi = Op.getOperand(1);
SDValue Amt = Op.getOperand(2);
EVT AmtVT = Amt.getValueType();
SDValue Tmp1 = DAG.getNode(ISD::SUB, dl, AmtVT,
DAG.getConstant(BitWidth, AmtVT), Amt);
SDValue Tmp2 = DAG.getNode(PPCISD::SHL, dl, VT, Hi, Amt);
SDValue Tmp3 = DAG.getNode(PPCISD::SRL, dl, VT, Lo, Tmp1);
SDValue Tmp4 = DAG.getNode(ISD::OR , dl, VT, Tmp2, Tmp3);
SDValue Tmp5 = DAG.getNode(ISD::ADD, dl, AmtVT, Amt,
DAG.getConstant(-BitWidth, AmtVT));
SDValue Tmp6 = DAG.getNode(PPCISD::SHL, dl, VT, Lo, Tmp5);
SDValue OutHi = DAG.getNode(ISD::OR, dl, VT, Tmp4, Tmp6);
SDValue OutLo = DAG.getNode(PPCISD::SHL, dl, VT, Lo, Amt);
SDValue OutOps[] = { OutLo, OutHi };
return DAG.getMergeValues(OutOps, 2, dl);
}
SDValue PPCTargetLowering::LowerSRL_PARTS(SDValue Op, SelectionDAG &DAG) const {
EVT VT = Op.getValueType();
DebugLoc dl = Op.getDebugLoc();
unsigned BitWidth = VT.getSizeInBits();
assert(Op.getNumOperands() == 3 &&
VT == Op.getOperand(1).getValueType() &&
"Unexpected SRL!");
SDValue Lo = Op.getOperand(0);
SDValue Hi = Op.getOperand(1);
SDValue Amt = Op.getOperand(2);
EVT AmtVT = Amt.getValueType();
SDValue Tmp1 = DAG.getNode(ISD::SUB, dl, AmtVT,
DAG.getConstant(BitWidth, AmtVT), Amt);
SDValue Tmp2 = DAG.getNode(PPCISD::SRL, dl, VT, Lo, Amt);
SDValue Tmp3 = DAG.getNode(PPCISD::SHL, dl, VT, Hi, Tmp1);
SDValue Tmp4 = DAG.getNode(ISD::OR, dl, VT, Tmp2, Tmp3);
SDValue Tmp5 = DAG.getNode(ISD::ADD, dl, AmtVT, Amt,
DAG.getConstant(-BitWidth, AmtVT));
SDValue Tmp6 = DAG.getNode(PPCISD::SRL, dl, VT, Hi, Tmp5);
SDValue OutLo = DAG.getNode(ISD::OR, dl, VT, Tmp4, Tmp6);
SDValue OutHi = DAG.getNode(PPCISD::SRL, dl, VT, Hi, Amt);
SDValue OutOps[] = { OutLo, OutHi };
return DAG.getMergeValues(OutOps, 2, dl);
}
SDValue PPCTargetLowering::LowerSRA_PARTS(SDValue Op, SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
EVT VT = Op.getValueType();
unsigned BitWidth = VT.getSizeInBits();
assert(Op.getNumOperands() == 3 &&
VT == Op.getOperand(1).getValueType() &&
"Unexpected SRA!");
SDValue Lo = Op.getOperand(0);
SDValue Hi = Op.getOperand(1);
SDValue Amt = Op.getOperand(2);
EVT AmtVT = Amt.getValueType();
SDValue Tmp1 = DAG.getNode(ISD::SUB, dl, AmtVT,
DAG.getConstant(BitWidth, AmtVT), Amt);
SDValue Tmp2 = DAG.getNode(PPCISD::SRL, dl, VT, Lo, Amt);
SDValue Tmp3 = DAG.getNode(PPCISD::SHL, dl, VT, Hi, Tmp1);
SDValue Tmp4 = DAG.getNode(ISD::OR, dl, VT, Tmp2, Tmp3);
SDValue Tmp5 = DAG.getNode(ISD::ADD, dl, AmtVT, Amt,
DAG.getConstant(-BitWidth, AmtVT));
SDValue Tmp6 = DAG.getNode(PPCISD::SRA, dl, VT, Hi, Tmp5);
SDValue OutHi = DAG.getNode(PPCISD::SRA, dl, VT, Hi, Amt);
SDValue OutLo = DAG.getSelectCC(dl, Tmp5, DAG.getConstant(0, AmtVT),
Tmp4, Tmp6, ISD::SETLE);
SDValue OutOps[] = { OutLo, OutHi };
return DAG.getMergeValues(OutOps, 2, dl);
}
static SDValue BuildSplatI(int Val, unsigned SplatSize, EVT VT,
SelectionDAG &DAG, DebugLoc dl) {
assert(Val >= -16 && Val <= 15 && "vsplti is out of range!");
static const EVT VTys[] = { MVT::v16i8, MVT::v8i16, MVT::Other, MVT::v4i32
};
EVT ReqVT = VT != MVT::Other ? VT : VTys[SplatSize-1];
if (Val == -1)
SplatSize = 1;
EVT CanonicalVT = VTys[SplatSize-1];
SDValue Elt = DAG.getConstant(Val, MVT::i32);
SmallVector<SDValue, 8> Ops;
Ops.assign(CanonicalVT.getVectorNumElements(), Elt);
SDValue Res = DAG.getNode(ISD::BUILD_VECTOR, dl, CanonicalVT,
&Ops[0], Ops.size());
return DAG.getNode(ISD::BITCAST, dl, ReqVT, Res);
}
static SDValue BuildIntrinsicOp(unsigned IID, SDValue LHS, SDValue RHS,
SelectionDAG &DAG, DebugLoc dl,
EVT DestVT = MVT::Other) {
if (DestVT == MVT::Other) DestVT = LHS.getValueType();
return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, DestVT,
DAG.getConstant(IID, MVT::i32), LHS, RHS);
}
static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op0, SDValue Op1,
SDValue Op2, SelectionDAG &DAG,
DebugLoc dl, EVT DestVT = MVT::Other) {
if (DestVT == MVT::Other) DestVT = Op0.getValueType();
return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, DestVT,
DAG.getConstant(IID, MVT::i32), Op0, Op1, Op2);
}
static SDValue BuildVSLDOI(SDValue LHS, SDValue RHS, unsigned Amt,
EVT VT, SelectionDAG &DAG, DebugLoc dl) {
LHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, LHS);
RHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, RHS);
int Ops[16];
for (unsigned i = 0; i != 16; ++i)
Ops[i] = i + Amt;
SDValue T = DAG.getVectorShuffle(MVT::v16i8, dl, LHS, RHS, Ops);
return DAG.getNode(ISD::BITCAST, dl, VT, T);
}
SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
BuildVectorSDNode *BVN = dyn_cast<BuildVectorSDNode>(Op.getNode());
assert(BVN != 0 && "Expected a BuildVectorSDNode in LowerBUILD_VECTOR");
APInt APSplatBits, APSplatUndef;
unsigned SplatBitSize;
bool HasAnyUndefs;
if (! BVN->isConstantSplat(APSplatBits, APSplatUndef, SplatBitSize,
HasAnyUndefs, 0, true) || SplatBitSize > 32)
return SDValue();
unsigned SplatBits = APSplatBits.getZExtValue();
unsigned SplatUndef = APSplatUndef.getZExtValue();
unsigned SplatSize = SplatBitSize / 8;
if (SplatBits == 0) {
if (Op.getValueType() != MVT::v4i32 || HasAnyUndefs) {
SDValue Z = DAG.getConstant(0, MVT::i32);
Z = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Z, Z, Z, Z);
Op = DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Z);
}
return Op;
}
int32_t SextVal= (int32_t(SplatBits << (32-SplatBitSize)) >>
(32-SplatBitSize));
if (SextVal >= -16 && SextVal <= 15)
return BuildSplatI(SextVal, SplatSize, Op.getValueType(), DAG, dl);
if (SextVal >= -32 && SextVal <= 30 && (SextVal & 1) == 0) {
SDValue Res = BuildSplatI(SextVal >> 1, SplatSize, MVT::Other, DAG, dl);
Res = DAG.getNode(ISD::ADD, dl, Res.getValueType(), Res, Res);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
if (SplatSize == 4 && SplatBits == (0x7FFFFFFF&~SplatUndef)) {
SDValue OnesV = BuildSplatI(-1, 4, MVT::v4i32, DAG, dl);
SDValue Res = BuildIntrinsicOp(Intrinsic::ppc_altivec_vslw, OnesV,
OnesV, DAG, dl);
Res = DAG.getNode(ISD::XOR, dl, MVT::v4i32, Res, OnesV);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
static const signed char SplatCsts[] = {
-1, 1, -2, 2, -3, 3, -4, 4, -5, 5, -6, 6, -7, 7,
-8, 8, -9, 9, -10, 10, -11, 11, -12, 12, -13, 13, 14, -14, 15, -15, -16
};
for (unsigned idx = 0; idx < array_lengthof(SplatCsts); ++idx) {
int i = SplatCsts[idx];
unsigned TypeShiftAmt = i & (SplatBitSize-1);
if (SextVal == (int)((unsigned)i << TypeShiftAmt)) {
SDValue Res = BuildSplatI(i, SplatSize, MVT::Other, DAG, dl);
static const unsigned IIDs[] = { Intrinsic::ppc_altivec_vslb, Intrinsic::ppc_altivec_vslh, 0,
Intrinsic::ppc_altivec_vslw
};
Res = BuildIntrinsicOp(IIDs[SplatSize-1], Res, Res, DAG, dl);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
if (SextVal == (int)((unsigned)i >> TypeShiftAmt)) {
SDValue Res = BuildSplatI(i, SplatSize, MVT::Other, DAG, dl);
static const unsigned IIDs[] = { Intrinsic::ppc_altivec_vsrb, Intrinsic::ppc_altivec_vsrh, 0,
Intrinsic::ppc_altivec_vsrw
};
Res = BuildIntrinsicOp(IIDs[SplatSize-1], Res, Res, DAG, dl);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
if (SextVal == (int)((unsigned)i >> TypeShiftAmt)) {
SDValue Res = BuildSplatI(i, SplatSize, MVT::Other, DAG, dl);
static const unsigned IIDs[] = { Intrinsic::ppc_altivec_vsrab, Intrinsic::ppc_altivec_vsrah, 0,
Intrinsic::ppc_altivec_vsraw
};
Res = BuildIntrinsicOp(IIDs[SplatSize-1], Res, Res, DAG, dl);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
if (SextVal == (int)(((unsigned)i << TypeShiftAmt) |
((unsigned)i >> (SplatBitSize-TypeShiftAmt)))) {
SDValue Res = BuildSplatI(i, SplatSize, MVT::Other, DAG, dl);
static const unsigned IIDs[] = { Intrinsic::ppc_altivec_vrlb, Intrinsic::ppc_altivec_vrlh, 0,
Intrinsic::ppc_altivec_vrlw
};
Res = BuildIntrinsicOp(IIDs[SplatSize-1], Res, Res, DAG, dl);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Res);
}
if (SextVal == (int)(((unsigned)i << 8) | (i < 0 ? 0xFF : 0))) {
SDValue T = BuildSplatI(i, SplatSize, MVT::v16i8, DAG, dl);
return BuildVSLDOI(T, T, 1, Op.getValueType(), DAG, dl);
}
if (SextVal == (int)(((unsigned)i << 16) | (i < 0 ? 0xFFFF : 0))) {
SDValue T = BuildSplatI(i, SplatSize, MVT::v16i8, DAG, dl);
return BuildVSLDOI(T, T, 2, Op.getValueType(), DAG, dl);
}
if (SextVal == (int)(((unsigned)i << 24) | (i < 0 ? 0xFFFFFF : 0))) {
SDValue T = BuildSplatI(i, SplatSize, MVT::v16i8, DAG, dl);
return BuildVSLDOI(T, T, 3, Op.getValueType(), DAG, dl);
}
}
if (SextVal >= 0 && SextVal <= 31) {
SDValue LHS = BuildSplatI(SextVal-16, SplatSize, MVT::Other, DAG, dl);
SDValue RHS = BuildSplatI(-16, SplatSize, MVT::Other, DAG, dl);
LHS = DAG.getNode(ISD::SUB, dl, LHS.getValueType(), LHS, RHS);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), LHS);
}
if (SextVal >= -31 && SextVal <= 0) {
SDValue LHS = BuildSplatI(SextVal+16, SplatSize, MVT::Other, DAG, dl);
SDValue RHS = BuildSplatI(-16, SplatSize, MVT::Other, DAG, dl);
LHS = DAG.getNode(ISD::ADD, dl, LHS.getValueType(), LHS, RHS);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), LHS);
}
return SDValue();
}
static SDValue GeneratePerfectShuffle(unsigned PFEntry, SDValue LHS,
SDValue RHS, SelectionDAG &DAG,
DebugLoc dl) {
unsigned OpNum = (PFEntry >> 26) & 0x0F;
unsigned LHSID = (PFEntry >> 13) & ((1 << 13)-1);
unsigned RHSID = (PFEntry >> 0) & ((1 << 13)-1);
enum {
OP_COPY = 0, OP_VMRGHW,
OP_VMRGLW,
OP_VSPLTISW0,
OP_VSPLTISW1,
OP_VSPLTISW2,
OP_VSPLTISW3,
OP_VSLDOI4,
OP_VSLDOI8,
OP_VSLDOI12
};
if (OpNum == OP_COPY) {
if (LHSID == (1*9+2)*9+3) return LHS;
assert(LHSID == ((4*9+5)*9+6)*9+7 && "Illegal OP_COPY!");
return RHS;
}
SDValue OpLHS, OpRHS;
OpLHS = GeneratePerfectShuffle(PerfectShuffleTable[LHSID], LHS, RHS, DAG, dl);
OpRHS = GeneratePerfectShuffle(PerfectShuffleTable[RHSID], LHS, RHS, DAG, dl);
int ShufIdxs[16];
switch (OpNum) {
default: llvm_unreachable("Unknown i32 permute!");
case OP_VMRGHW:
ShufIdxs[ 0] = 0; ShufIdxs[ 1] = 1; ShufIdxs[ 2] = 2; ShufIdxs[ 3] = 3;
ShufIdxs[ 4] = 16; ShufIdxs[ 5] = 17; ShufIdxs[ 6] = 18; ShufIdxs[ 7] = 19;
ShufIdxs[ 8] = 4; ShufIdxs[ 9] = 5; ShufIdxs[10] = 6; ShufIdxs[11] = 7;
ShufIdxs[12] = 20; ShufIdxs[13] = 21; ShufIdxs[14] = 22; ShufIdxs[15] = 23;
break;
case OP_VMRGLW:
ShufIdxs[ 0] = 8; ShufIdxs[ 1] = 9; ShufIdxs[ 2] = 10; ShufIdxs[ 3] = 11;
ShufIdxs[ 4] = 24; ShufIdxs[ 5] = 25; ShufIdxs[ 6] = 26; ShufIdxs[ 7] = 27;
ShufIdxs[ 8] = 12; ShufIdxs[ 9] = 13; ShufIdxs[10] = 14; ShufIdxs[11] = 15;
ShufIdxs[12] = 28; ShufIdxs[13] = 29; ShufIdxs[14] = 30; ShufIdxs[15] = 31;
break;
case OP_VSPLTISW0:
for (unsigned i = 0; i != 16; ++i)
ShufIdxs[i] = (i&3)+0;
break;
case OP_VSPLTISW1:
for (unsigned i = 0; i != 16; ++i)
ShufIdxs[i] = (i&3)+4;
break;
case OP_VSPLTISW2:
for (unsigned i = 0; i != 16; ++i)
ShufIdxs[i] = (i&3)+8;
break;
case OP_VSPLTISW3:
for (unsigned i = 0; i != 16; ++i)
ShufIdxs[i] = (i&3)+12;
break;
case OP_VSLDOI4:
return BuildVSLDOI(OpLHS, OpRHS, 4, OpLHS.getValueType(), DAG, dl);
case OP_VSLDOI8:
return BuildVSLDOI(OpLHS, OpRHS, 8, OpLHS.getValueType(), DAG, dl);
case OP_VSLDOI12:
return BuildVSLDOI(OpLHS, OpRHS, 12, OpLHS.getValueType(), DAG, dl);
}
EVT VT = OpLHS.getValueType();
OpLHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, OpLHS);
OpRHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, OpRHS);
SDValue T = DAG.getVectorShuffle(MVT::v16i8, dl, OpLHS, OpRHS, ShufIdxs);
return DAG.getNode(ISD::BITCAST, dl, VT, T);
}
SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
SDValue V1 = Op.getOperand(0);
SDValue V2 = Op.getOperand(1);
ShuffleVectorSDNode *SVOp = cast<ShuffleVectorSDNode>(Op);
EVT VT = Op.getValueType();
if (V2.getOpcode() == ISD::UNDEF) {
if (PPC::isSplatShuffleMask(SVOp, 1) ||
PPC::isSplatShuffleMask(SVOp, 2) ||
PPC::isSplatShuffleMask(SVOp, 4) ||
PPC::isVPKUWUMShuffleMask(SVOp, true) ||
PPC::isVPKUHUMShuffleMask(SVOp, true) ||
PPC::isVSLDOIShuffleMask(SVOp, true) != -1 ||
PPC::isVMRGLShuffleMask(SVOp, 1, true) ||
PPC::isVMRGLShuffleMask(SVOp, 2, true) ||
PPC::isVMRGLShuffleMask(SVOp, 4, true) ||
PPC::isVMRGHShuffleMask(SVOp, 1, true) ||
PPC::isVMRGHShuffleMask(SVOp, 2, true) ||
PPC::isVMRGHShuffleMask(SVOp, 4, true)) {
return Op;
}
}
if (PPC::isVPKUWUMShuffleMask(SVOp, false) ||
PPC::isVPKUHUMShuffleMask(SVOp, false) ||
PPC::isVSLDOIShuffleMask(SVOp, false) != -1 ||
PPC::isVMRGLShuffleMask(SVOp, 1, false) ||
PPC::isVMRGLShuffleMask(SVOp, 2, false) ||
PPC::isVMRGLShuffleMask(SVOp, 4, false) ||
PPC::isVMRGHShuffleMask(SVOp, 1, false) ||
PPC::isVMRGHShuffleMask(SVOp, 2, false) ||
PPC::isVMRGHShuffleMask(SVOp, 4, false))
return Op;
ArrayRef<int> PermMask = SVOp->getMask();
unsigned PFIndexes[4];
bool isFourElementShuffle = true;
for (unsigned i = 0; i != 4 && isFourElementShuffle; ++i) { unsigned EltNo = 8; for (unsigned j = 0; j != 4; ++j) { if (PermMask[i*4+j] < 0)
continue;
unsigned ByteSource = PermMask[i*4+j];
if ((ByteSource & 3) != j) {
isFourElementShuffle = false;
break;
}
if (EltNo == 8) {
EltNo = ByteSource/4;
} else if (EltNo != ByteSource/4) {
isFourElementShuffle = false;
break;
}
}
PFIndexes[i] = EltNo;
}
if (isFourElementShuffle) {
unsigned PFTableIndex =
PFIndexes[0]*9*9*9+PFIndexes[1]*9*9+PFIndexes[2]*9+PFIndexes[3];
unsigned PFEntry = PerfectShuffleTable[PFTableIndex];
unsigned Cost = (PFEntry >> 30);
if (Cost < 3)
return GeneratePerfectShuffle(PFEntry, V1, V2, DAG, dl);
}
if (V2.getOpcode() == ISD::UNDEF) V2 = V1;
EVT EltVT = V1.getValueType().getVectorElementType();
unsigned BytesPerElement = EltVT.getSizeInBits()/8;
SmallVector<SDValue, 16> ResultMask;
for (unsigned i = 0, e = VT.getVectorNumElements(); i != e; ++i) {
unsigned SrcElt = PermMask[i] < 0 ? 0 : PermMask[i];
for (unsigned j = 0; j != BytesPerElement; ++j)
ResultMask.push_back(DAG.getConstant(SrcElt*BytesPerElement+j,
MVT::i32));
}
SDValue VPermMask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v16i8,
&ResultMask[0], ResultMask.size());
return DAG.getNode(PPCISD::VPERM, dl, V1.getValueType(), V1, V2, VPermMask);
}
static bool getAltivecCompareInfo(SDValue Intrin, int &CompareOpc,
bool &isDot) {
unsigned IntrinsicID =
cast<ConstantSDNode>(Intrin.getOperand(0))->getZExtValue();
CompareOpc = -1;
isDot = false;
switch (IntrinsicID) {
default: return false;
case Intrinsic::ppc_altivec_vcmpbfp_p: CompareOpc = 966; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpeqfp_p: CompareOpc = 198; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpequb_p: CompareOpc = 6; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpequh_p: CompareOpc = 70; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpequw_p: CompareOpc = 134; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgefp_p: CompareOpc = 454; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtfp_p: CompareOpc = 710; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtsb_p: CompareOpc = 774; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtsh_p: CompareOpc = 838; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtsw_p: CompareOpc = 902; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtub_p: CompareOpc = 518; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtuh_p: CompareOpc = 582; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpgtuw_p: CompareOpc = 646; isDot = 1; break;
case Intrinsic::ppc_altivec_vcmpbfp: CompareOpc = 966; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpeqfp: CompareOpc = 198; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpequb: CompareOpc = 6; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpequh: CompareOpc = 70; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpequw: CompareOpc = 134; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgefp: CompareOpc = 454; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtfp: CompareOpc = 710; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtsb: CompareOpc = 774; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtsh: CompareOpc = 838; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtsw: CompareOpc = 902; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtub: CompareOpc = 518; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtuh: CompareOpc = 582; isDot = 0; break;
case Intrinsic::ppc_altivec_vcmpgtuw: CompareOpc = 646; isDot = 0; break;
}
return true;
}
SDValue PPCTargetLowering::LowerINTRINSIC_WO_CHAIN(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
int CompareOpc;
bool isDot;
if (!getAltivecCompareInfo(Op, CompareOpc, isDot))
return SDValue();
if (!isDot) {
SDValue Tmp = DAG.getNode(PPCISD::VCMP, dl, Op.getOperand(2).getValueType(),
Op.getOperand(1), Op.getOperand(2),
DAG.getConstant(CompareOpc, MVT::i32));
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Tmp);
}
SDValue Ops[] = {
Op.getOperand(2), Op.getOperand(3), DAG.getConstant(CompareOpc, MVT::i32)
};
EVT VTs[] = { Op.getOperand(2).getValueType(), MVT::Glue };
SDValue CompNode = DAG.getNode(PPCISD::VCMPo, dl, VTs, Ops, 3);
SDValue Flags = DAG.getNode(PPCISD::MFCR, dl, MVT::i32,
DAG.getRegister(PPC::CR6, MVT::i32),
CompNode.getValue(1));
unsigned BitNo; bool InvertBit; switch (cast<ConstantSDNode>(Op.getOperand(1))->getZExtValue()) {
default: case 0: BitNo = 0; InvertBit = false;
break;
case 1: BitNo = 0; InvertBit = true;
break;
case 2: BitNo = 2; InvertBit = false;
break;
case 3: BitNo = 2; InvertBit = true;
break;
}
Flags = DAG.getNode(ISD::SRL, dl, MVT::i32, Flags,
DAG.getConstant(8-(3-BitNo), MVT::i32));
Flags = DAG.getNode(ISD::AND, dl, MVT::i32, Flags,
DAG.getConstant(1, MVT::i32));
if (InvertBit)
Flags = DAG.getNode(ISD::XOR, dl, MVT::i32, Flags,
DAG.getConstant(1, MVT::i32));
return Flags;
}
SDValue PPCTargetLowering::LowerSCALAR_TO_VECTOR(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
MachineFrameInfo *FrameInfo = DAG.getMachineFunction().getFrameInfo();
int FrameIdx = FrameInfo->CreateStackObject(16, 16, false);
EVT PtrVT = getPointerTy();
SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
SDValue Store = DAG.getStore(DAG.getEntryNode(), dl,
Op.getOperand(0), FIdx, MachinePointerInfo(),
false, false, 0);
return DAG.getLoad(Op.getValueType(), dl, Store, FIdx, MachinePointerInfo(),
false, false, false, 0);
}
SDValue PPCTargetLowering::LowerMUL(SDValue Op, SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
if (Op.getValueType() == MVT::v4i32) {
SDValue LHS = Op.getOperand(0), RHS = Op.getOperand(1);
SDValue Zero = BuildSplatI( 0, 1, MVT::v4i32, DAG, dl);
SDValue Neg16 = BuildSplatI(-16, 4, MVT::v4i32, DAG, dl);
SDValue RHSSwap = BuildIntrinsicOp(Intrinsic::ppc_altivec_vrlw, RHS, Neg16, DAG, dl);
LHS = DAG.getNode(ISD::BITCAST, dl, MVT::v8i16, LHS);
RHS = DAG.getNode(ISD::BITCAST, dl, MVT::v8i16, RHS);
RHSSwap = DAG.getNode(ISD::BITCAST, dl, MVT::v8i16, RHSSwap);
SDValue LoProd = BuildIntrinsicOp(Intrinsic::ppc_altivec_vmulouh,
LHS, RHS, DAG, dl, MVT::v4i32);
SDValue HiProd = BuildIntrinsicOp(Intrinsic::ppc_altivec_vmsumuhm,
LHS, RHSSwap, Zero, DAG, dl, MVT::v4i32);
HiProd = BuildIntrinsicOp(Intrinsic::ppc_altivec_vslw, HiProd,
Neg16, DAG, dl);
return DAG.getNode(ISD::ADD, dl, MVT::v4i32, LoProd, HiProd);
} else if (Op.getValueType() == MVT::v8i16) {
SDValue LHS = Op.getOperand(0), RHS = Op.getOperand(1);
SDValue Zero = BuildSplatI(0, 1, MVT::v8i16, DAG, dl);
return BuildIntrinsicOp(Intrinsic::ppc_altivec_vmladduhm,
LHS, RHS, Zero, DAG, dl);
} else if (Op.getValueType() == MVT::v16i8) {
SDValue LHS = Op.getOperand(0), RHS = Op.getOperand(1);
SDValue EvenParts = BuildIntrinsicOp(Intrinsic::ppc_altivec_vmuleub,
LHS, RHS, DAG, dl, MVT::v8i16);
EvenParts = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, EvenParts);
SDValue OddParts = BuildIntrinsicOp(Intrinsic::ppc_altivec_vmuloub,
LHS, RHS, DAG, dl, MVT::v8i16);
OddParts = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, OddParts);
int Ops[16];
for (unsigned i = 0; i != 8; ++i) {
Ops[i*2 ] = 2*i+1;
Ops[i*2+1] = 2*i+1+16;
}
return DAG.getVectorShuffle(MVT::v16i8, dl, EvenParts, OddParts, Ops);
} else {
llvm_unreachable("Unknown mul to lower!");
}
}
SDValue PPCTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
switch (Op.getOpcode()) {
default: llvm_unreachable("Wasn't expecting to be able to lower this!");
case ISD::ConstantPool: return LowerConstantPool(Op, DAG);
case ISD::BlockAddress: return LowerBlockAddress(Op, DAG);
case ISD::GlobalAddress: return LowerGlobalAddress(Op, DAG);
case ISD::GlobalTLSAddress: return LowerGlobalTLSAddress(Op, DAG);
case ISD::JumpTable: return LowerJumpTable(Op, DAG);
case ISD::SETCC: return LowerSETCC(Op, DAG);
case ISD::INIT_TRAMPOLINE: return LowerINIT_TRAMPOLINE(Op, DAG);
case ISD::ADJUST_TRAMPOLINE: return LowerADJUST_TRAMPOLINE(Op, DAG);
case ISD::VASTART:
return LowerVASTART(Op, DAG, PPCSubTarget);
case ISD::VAARG:
return LowerVAARG(Op, DAG, PPCSubTarget);
case ISD::STACKRESTORE: return LowerSTACKRESTORE(Op, DAG, PPCSubTarget);
case ISD::DYNAMIC_STACKALLOC:
return LowerDYNAMIC_STACKALLOC(Op, DAG, PPCSubTarget);
case ISD::SELECT_CC: return LowerSELECT_CC(Op, DAG);
case ISD::FP_TO_UINT:
case ISD::FP_TO_SINT: return LowerFP_TO_INT(Op, DAG,
Op.getDebugLoc());
case ISD::SINT_TO_FP: return LowerSINT_TO_FP(Op, DAG);
case ISD::FLT_ROUNDS_: return LowerFLT_ROUNDS_(Op, DAG);
case ISD::SHL_PARTS: return LowerSHL_PARTS(Op, DAG);
case ISD::SRL_PARTS: return LowerSRL_PARTS(Op, DAG);
case ISD::SRA_PARTS: return LowerSRA_PARTS(Op, DAG);
case ISD::BUILD_VECTOR: return LowerBUILD_VECTOR(Op, DAG);
case ISD::VECTOR_SHUFFLE: return LowerVECTOR_SHUFFLE(Op, DAG);
case ISD::INTRINSIC_WO_CHAIN: return LowerINTRINSIC_WO_CHAIN(Op, DAG);
case ISD::SCALAR_TO_VECTOR: return LowerSCALAR_TO_VECTOR(Op, DAG);
case ISD::MUL: return LowerMUL(Op, DAG);
case ISD::RETURNADDR: return LowerRETURNADDR(Op, DAG);
case ISD::FRAMEADDR: return LowerFRAMEADDR(Op, DAG);
}
}
void PPCTargetLowering::ReplaceNodeResults(SDNode *N,
SmallVectorImpl<SDValue>&Results,
SelectionDAG &DAG) const {
const TargetMachine &TM = getTargetMachine();
DebugLoc dl = N->getDebugLoc();
switch (N->getOpcode()) {
default:
llvm_unreachable("Do not know how to custom type legalize this operation!");
case ISD::VAARG: {
if (!TM.getSubtarget<PPCSubtarget>().isSVR4ABI()
|| TM.getSubtarget<PPCSubtarget>().isPPC64())
return;
EVT VT = N->getValueType(0);
if (VT == MVT::i64) {
SDValue NewNode = LowerVAARG(SDValue(N, 1), DAG, PPCSubTarget);
Results.push_back(NewNode);
Results.push_back(NewNode.getValue(1));
}
return;
}
case ISD::FP_ROUND_INREG: {
assert(N->getValueType(0) == MVT::ppcf128);
assert(N->getOperand(0).getValueType() == MVT::ppcf128);
SDValue Lo = DAG.getNode(ISD::EXTRACT_ELEMENT, dl,
MVT::f64, N->getOperand(0),
DAG.getIntPtrConstant(0));
SDValue Hi = DAG.getNode(ISD::EXTRACT_ELEMENT, dl,
MVT::f64, N->getOperand(0),
DAG.getIntPtrConstant(1));
std::vector<EVT> NodeTys;
SDValue Ops[4], Result, MFFSreg, InFlag, FPreg;
NodeTys.push_back(MVT::f64); NodeTys.push_back(MVT::Glue); Result = DAG.getNode(PPCISD::MFFS, dl, NodeTys, &InFlag, 0);
MFFSreg = Result.getValue(0);
InFlag = Result.getValue(1);
NodeTys.clear();
NodeTys.push_back(MVT::Glue); Ops[0] = DAG.getConstant(31, MVT::i32);
Ops[1] = InFlag;
Result = DAG.getNode(PPCISD::MTFSB1, dl, NodeTys, Ops, 2);
InFlag = Result.getValue(0);
NodeTys.clear();
NodeTys.push_back(MVT::Glue); Ops[0] = DAG.getConstant(30, MVT::i32);
Ops[1] = InFlag;
Result = DAG.getNode(PPCISD::MTFSB0, dl, NodeTys, Ops, 2);
InFlag = Result.getValue(0);
NodeTys.clear();
NodeTys.push_back(MVT::f64); NodeTys.push_back(MVT::Glue); Ops[0] = Lo;
Ops[1] = Hi;
Ops[2] = InFlag;
Result = DAG.getNode(PPCISD::FADDRTZ, dl, NodeTys, Ops, 3);
FPreg = Result.getValue(0);
InFlag = Result.getValue(1);
NodeTys.clear();
NodeTys.push_back(MVT::f64);
Ops[0] = DAG.getConstant(1, MVT::i32);
Ops[1] = MFFSreg;
Ops[2] = FPreg;
Ops[3] = InFlag;
Result = DAG.getNode(PPCISD::MTFSF, dl, NodeTys, Ops, 4);
FPreg = Result.getValue(0);
Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, MVT::ppcf128,
FPreg, FPreg));
return;
}
case ISD::FP_TO_SINT:
Results.push_back(LowerFP_TO_INT(SDValue(N, 0), DAG, dl));
return;
}
}
MachineBasicBlock *
PPCTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
bool is64bit, unsigned BinOpcode) const {
const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
const BasicBlock *LLVM_BB = BB->getBasicBlock();
MachineFunction *F = BB->getParent();
MachineFunction::iterator It = BB;
++It;
unsigned dest = MI->getOperand(0).getReg();
unsigned ptrA = MI->getOperand(1).getReg();
unsigned ptrB = MI->getOperand(2).getReg();
unsigned incr = MI->getOperand(3).getReg();
DebugLoc dl = MI->getDebugLoc();
MachineBasicBlock *loopMBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
F->insert(It, loopMBB);
F->insert(It, exitMBB);
exitMBB->splice(exitMBB->begin(), BB,
llvm::next(MachineBasicBlock::iterator(MI)),
BB->end());
exitMBB->transferSuccessorsAndUpdatePHIs(BB);
MachineRegisterInfo &RegInfo = F->getRegInfo();
unsigned TmpReg = (!BinOpcode) ? incr :
RegInfo.createVirtualRegister(
is64bit ? (const TargetRegisterClass *) &PPC::G8RCRegClass :
(const TargetRegisterClass *) &PPC::GPRCRegClass);
BB->addSuccessor(loopMBB);
BB = loopMBB;
BuildMI(BB, dl, TII->get(is64bit ? PPC::LDARX : PPC::LWARX), dest)
.addReg(ptrA).addReg(ptrB);
if (BinOpcode)
BuildMI(BB, dl, TII->get(BinOpcode), TmpReg).addReg(incr).addReg(dest);
BuildMI(BB, dl, TII->get(is64bit ? PPC::STDCX : PPC::STWCX))
.addReg(TmpReg).addReg(ptrA).addReg(ptrB);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(loopMBB);
BB->addSuccessor(loopMBB);
BB->addSuccessor(exitMBB);
BB = exitMBB;
return BB;
}
MachineBasicBlock *
PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr *MI,
MachineBasicBlock *BB,
bool is8bit, unsigned BinOpcode) const {
const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
bool is64bit = PPCSubTarget.isPPC64();
unsigned ZeroReg = is64bit ? PPC::X0 : PPC::R0;
const BasicBlock *LLVM_BB = BB->getBasicBlock();
MachineFunction *F = BB->getParent();
MachineFunction::iterator It = BB;
++It;
unsigned dest = MI->getOperand(0).getReg();
unsigned ptrA = MI->getOperand(1).getReg();
unsigned ptrB = MI->getOperand(2).getReg();
unsigned incr = MI->getOperand(3).getReg();
DebugLoc dl = MI->getDebugLoc();
MachineBasicBlock *loopMBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
F->insert(It, loopMBB);
F->insert(It, exitMBB);
exitMBB->splice(exitMBB->begin(), BB,
llvm::next(MachineBasicBlock::iterator(MI)),
BB->end());
exitMBB->transferSuccessorsAndUpdatePHIs(BB);
MachineRegisterInfo &RegInfo = F->getRegInfo();
const TargetRegisterClass *RC =
is64bit ? (const TargetRegisterClass *) &PPC::G8RCRegClass :
(const TargetRegisterClass *) &PPC::GPRCRegClass;
unsigned PtrReg = RegInfo.createVirtualRegister(RC);
unsigned Shift1Reg = RegInfo.createVirtualRegister(RC);
unsigned ShiftReg = RegInfo.createVirtualRegister(RC);
unsigned Incr2Reg = RegInfo.createVirtualRegister(RC);
unsigned MaskReg = RegInfo.createVirtualRegister(RC);
unsigned Mask2Reg = RegInfo.createVirtualRegister(RC);
unsigned Mask3Reg = RegInfo.createVirtualRegister(RC);
unsigned Tmp2Reg = RegInfo.createVirtualRegister(RC);
unsigned Tmp3Reg = RegInfo.createVirtualRegister(RC);
unsigned Tmp4Reg = RegInfo.createVirtualRegister(RC);
unsigned TmpDestReg = RegInfo.createVirtualRegister(RC);
unsigned Ptr1Reg;
unsigned TmpReg = (!BinOpcode) ? Incr2Reg : RegInfo.createVirtualRegister(RC);
BB->addSuccessor(loopMBB);
if (ptrA != ZeroReg) {
Ptr1Reg = RegInfo.createVirtualRegister(RC);
BuildMI(BB, dl, TII->get(is64bit ? PPC::ADD8 : PPC::ADD4), Ptr1Reg)
.addReg(ptrA).addReg(ptrB);
} else {
Ptr1Reg = ptrB;
}
BuildMI(BB, dl, TII->get(PPC::RLWINM), Shift1Reg).addReg(Ptr1Reg)
.addImm(3).addImm(27).addImm(is8bit ? 28 : 27);
BuildMI(BB, dl, TII->get(is64bit ? PPC::XORI8 : PPC::XORI), ShiftReg)
.addReg(Shift1Reg).addImm(is8bit ? 24 : 16);
if (is64bit)
BuildMI(BB, dl, TII->get(PPC::RLDICR), PtrReg)
.addReg(Ptr1Reg).addImm(0).addImm(61);
else
BuildMI(BB, dl, TII->get(PPC::RLWINM), PtrReg)
.addReg(Ptr1Reg).addImm(0).addImm(0).addImm(29);
BuildMI(BB, dl, TII->get(PPC::SLW), Incr2Reg)
.addReg(incr).addReg(ShiftReg);
if (is8bit)
BuildMI(BB, dl, TII->get(PPC::LI), Mask2Reg).addImm(255);
else {
BuildMI(BB, dl, TII->get(PPC::LI), Mask3Reg).addImm(0);
BuildMI(BB, dl, TII->get(PPC::ORI),Mask2Reg).addReg(Mask3Reg).addImm(65535);
}
BuildMI(BB, dl, TII->get(PPC::SLW), MaskReg)
.addReg(Mask2Reg).addReg(ShiftReg);
BB = loopMBB;
BuildMI(BB, dl, TII->get(PPC::LWARX), TmpDestReg)
.addReg(ZeroReg).addReg(PtrReg);
if (BinOpcode)
BuildMI(BB, dl, TII->get(BinOpcode), TmpReg)
.addReg(Incr2Reg).addReg(TmpDestReg);
BuildMI(BB, dl, TII->get(is64bit ? PPC::ANDC8 : PPC::ANDC), Tmp2Reg)
.addReg(TmpDestReg).addReg(MaskReg);
BuildMI(BB, dl, TII->get(is64bit ? PPC::AND8 : PPC::AND), Tmp3Reg)
.addReg(TmpReg).addReg(MaskReg);
BuildMI(BB, dl, TII->get(is64bit ? PPC::OR8 : PPC::OR), Tmp4Reg)
.addReg(Tmp3Reg).addReg(Tmp2Reg);
BuildMI(BB, dl, TII->get(is64bit ? PPC::STDCX : PPC::STWCX))
.addReg(Tmp4Reg).addReg(ZeroReg).addReg(PtrReg);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(loopMBB);
BB->addSuccessor(loopMBB);
BB->addSuccessor(exitMBB);
BB = exitMBB;
BuildMI(*BB, BB->begin(), dl, TII->get(PPC::SRW), dest).addReg(TmpDestReg)
.addReg(ShiftReg);
return BB;
}
MachineBasicBlock *
PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
MachineBasicBlock *BB) const {
const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
const BasicBlock *LLVM_BB = BB->getBasicBlock();
MachineFunction::iterator It = BB;
++It;
MachineFunction *F = BB->getParent();
if (PPCSubTarget.hasISEL() && (MI->getOpcode() == PPC::SELECT_CC_I4 ||
MI->getOpcode() == PPC::SELECT_CC_I8)) {
unsigned OpCode = MI->getOpcode() == PPC::SELECT_CC_I8 ?
PPC::ISEL8 : PPC::ISEL;
unsigned SelectPred = MI->getOperand(4).getImm();
DebugLoc dl = MI->getDebugLoc();
unsigned BO = SelectPred & 0xF;
assert((BO == 12 || BO == 4) && "invalid predicate BO field for isel");
unsigned TrueOpNo, FalseOpNo;
if (BO == 12) {
TrueOpNo = 2;
FalseOpNo = 3;
} else {
TrueOpNo = 3;
FalseOpNo = 2;
SelectPred = PPC::InvertPredicate((PPC::Predicate)SelectPred);
}
BuildMI(*BB, MI, dl, TII->get(OpCode), MI->getOperand(0).getReg())
.addReg(MI->getOperand(TrueOpNo).getReg())
.addReg(MI->getOperand(FalseOpNo).getReg())
.addImm(SelectPred).addReg(MI->getOperand(1).getReg());
} else if (MI->getOpcode() == PPC::SELECT_CC_I4 ||
MI->getOpcode() == PPC::SELECT_CC_I8 ||
MI->getOpcode() == PPC::SELECT_CC_F4 ||
MI->getOpcode() == PPC::SELECT_CC_F8 ||
MI->getOpcode() == PPC::SELECT_CC_VRRC) {
MachineBasicBlock *thisMBB = BB;
MachineBasicBlock *copy0MBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *sinkMBB = F->CreateMachineBasicBlock(LLVM_BB);
unsigned SelectPred = MI->getOperand(4).getImm();
DebugLoc dl = MI->getDebugLoc();
F->insert(It, copy0MBB);
F->insert(It, sinkMBB);
sinkMBB->splice(sinkMBB->begin(), BB,
llvm::next(MachineBasicBlock::iterator(MI)),
BB->end());
sinkMBB->transferSuccessorsAndUpdatePHIs(BB);
BB->addSuccessor(copy0MBB);
BB->addSuccessor(sinkMBB);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(SelectPred).addReg(MI->getOperand(1).getReg()).addMBB(sinkMBB);
BB = copy0MBB;
BB->addSuccessor(sinkMBB);
BB = sinkMBB;
BuildMI(*BB, BB->begin(), dl,
TII->get(PPC::PHI), MI->getOperand(0).getReg())
.addReg(MI->getOperand(3).getReg()).addMBB(copy0MBB)
.addReg(MI->getOperand(2).getReg()).addMBB(thisMBB);
}
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::ADD4);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::ADD4);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::ADD4);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::ADD8);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::AND);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::AND);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::AND);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::AND8);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::OR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::OR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::OR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::OR8);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::XOR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::XOR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::XOR);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::XOR8);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::ANDC);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::ANDC);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::ANDC);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::ANDC8);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::SUBF);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::SUBF);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I32)
BB = EmitAtomicBinary(MI, BB, false, PPC::SUBF);
else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I64)
BB = EmitAtomicBinary(MI, BB, true, PPC::SUBF8);
else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I8)
BB = EmitPartwordAtomicBinary(MI, BB, true, 0);
else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I16)
BB = EmitPartwordAtomicBinary(MI, BB, false, 0);
else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I32)
BB = EmitAtomicBinary(MI, BB, false, 0);
else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I64)
BB = EmitAtomicBinary(MI, BB, true, 0);
else if (MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I32 ||
MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I64) {
bool is64bit = MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I64;
unsigned dest = MI->getOperand(0).getReg();
unsigned ptrA = MI->getOperand(1).getReg();
unsigned ptrB = MI->getOperand(2).getReg();
unsigned oldval = MI->getOperand(3).getReg();
unsigned newval = MI->getOperand(4).getReg();
DebugLoc dl = MI->getDebugLoc();
MachineBasicBlock *loop1MBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *loop2MBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *midMBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
F->insert(It, loop1MBB);
F->insert(It, loop2MBB);
F->insert(It, midMBB);
F->insert(It, exitMBB);
exitMBB->splice(exitMBB->begin(), BB,
llvm::next(MachineBasicBlock::iterator(MI)),
BB->end());
exitMBB->transferSuccessorsAndUpdatePHIs(BB);
BB->addSuccessor(loop1MBB);
BB = loop1MBB;
BuildMI(BB, dl, TII->get(is64bit ? PPC::LDARX : PPC::LWARX), dest)
.addReg(ptrA).addReg(ptrB);
BuildMI(BB, dl, TII->get(is64bit ? PPC::CMPD : PPC::CMPW), PPC::CR0)
.addReg(oldval).addReg(dest);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(midMBB);
BB->addSuccessor(loop2MBB);
BB->addSuccessor(midMBB);
BB = loop2MBB;
BuildMI(BB, dl, TII->get(is64bit ? PPC::STDCX : PPC::STWCX))
.addReg(newval).addReg(ptrA).addReg(ptrB);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(loop1MBB);
BuildMI(BB, dl, TII->get(PPC::B)).addMBB(exitMBB);
BB->addSuccessor(loop1MBB);
BB->addSuccessor(exitMBB);
BB = midMBB;
BuildMI(BB, dl, TII->get(is64bit ? PPC::STDCX : PPC::STWCX))
.addReg(dest).addReg(ptrA).addReg(ptrB);
BB->addSuccessor(exitMBB);
BB = exitMBB;
} else if (MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I8 ||
MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I16) {
bool is64bit = PPCSubTarget.isPPC64();
bool is8bit = MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I8;
unsigned dest = MI->getOperand(0).getReg();
unsigned ptrA = MI->getOperand(1).getReg();
unsigned ptrB = MI->getOperand(2).getReg();
unsigned oldval = MI->getOperand(3).getReg();
unsigned newval = MI->getOperand(4).getReg();
DebugLoc dl = MI->getDebugLoc();
MachineBasicBlock *loop1MBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *loop2MBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *midMBB = F->CreateMachineBasicBlock(LLVM_BB);
MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
F->insert(It, loop1MBB);
F->insert(It, loop2MBB);
F->insert(It, midMBB);
F->insert(It, exitMBB);
exitMBB->splice(exitMBB->begin(), BB,
llvm::next(MachineBasicBlock::iterator(MI)),
BB->end());
exitMBB->transferSuccessorsAndUpdatePHIs(BB);
MachineRegisterInfo &RegInfo = F->getRegInfo();
const TargetRegisterClass *RC =
is64bit ? (const TargetRegisterClass *) &PPC::G8RCRegClass :
(const TargetRegisterClass *) &PPC::GPRCRegClass;
unsigned PtrReg = RegInfo.createVirtualRegister(RC);
unsigned Shift1Reg = RegInfo.createVirtualRegister(RC);
unsigned ShiftReg = RegInfo.createVirtualRegister(RC);
unsigned NewVal2Reg = RegInfo.createVirtualRegister(RC);
unsigned NewVal3Reg = RegInfo.createVirtualRegister(RC);
unsigned OldVal2Reg = RegInfo.createVirtualRegister(RC);
unsigned OldVal3Reg = RegInfo.createVirtualRegister(RC);
unsigned MaskReg = RegInfo.createVirtualRegister(RC);
unsigned Mask2Reg = RegInfo.createVirtualRegister(RC);
unsigned Mask3Reg = RegInfo.createVirtualRegister(RC);
unsigned Tmp2Reg = RegInfo.createVirtualRegister(RC);
unsigned Tmp4Reg = RegInfo.createVirtualRegister(RC);
unsigned TmpDestReg = RegInfo.createVirtualRegister(RC);
unsigned Ptr1Reg;
unsigned TmpReg = RegInfo.createVirtualRegister(RC);
unsigned ZeroReg = is64bit ? PPC::X0 : PPC::R0;
BB->addSuccessor(loop1MBB);
if (ptrA != ZeroReg) {
Ptr1Reg = RegInfo.createVirtualRegister(RC);
BuildMI(BB, dl, TII->get(is64bit ? PPC::ADD8 : PPC::ADD4), Ptr1Reg)
.addReg(ptrA).addReg(ptrB);
} else {
Ptr1Reg = ptrB;
}
BuildMI(BB, dl, TII->get(PPC::RLWINM), Shift1Reg).addReg(Ptr1Reg)
.addImm(3).addImm(27).addImm(is8bit ? 28 : 27);
BuildMI(BB, dl, TII->get(is64bit ? PPC::XORI8 : PPC::XORI), ShiftReg)
.addReg(Shift1Reg).addImm(is8bit ? 24 : 16);
if (is64bit)
BuildMI(BB, dl, TII->get(PPC::RLDICR), PtrReg)
.addReg(Ptr1Reg).addImm(0).addImm(61);
else
BuildMI(BB, dl, TII->get(PPC::RLWINM), PtrReg)
.addReg(Ptr1Reg).addImm(0).addImm(0).addImm(29);
BuildMI(BB, dl, TII->get(PPC::SLW), NewVal2Reg)
.addReg(newval).addReg(ShiftReg);
BuildMI(BB, dl, TII->get(PPC::SLW), OldVal2Reg)
.addReg(oldval).addReg(ShiftReg);
if (is8bit)
BuildMI(BB, dl, TII->get(PPC::LI), Mask2Reg).addImm(255);
else {
BuildMI(BB, dl, TII->get(PPC::LI), Mask3Reg).addImm(0);
BuildMI(BB, dl, TII->get(PPC::ORI), Mask2Reg)
.addReg(Mask3Reg).addImm(65535);
}
BuildMI(BB, dl, TII->get(PPC::SLW), MaskReg)
.addReg(Mask2Reg).addReg(ShiftReg);
BuildMI(BB, dl, TII->get(PPC::AND), NewVal3Reg)
.addReg(NewVal2Reg).addReg(MaskReg);
BuildMI(BB, dl, TII->get(PPC::AND), OldVal3Reg)
.addReg(OldVal2Reg).addReg(MaskReg);
BB = loop1MBB;
BuildMI(BB, dl, TII->get(PPC::LWARX), TmpDestReg)
.addReg(ZeroReg).addReg(PtrReg);
BuildMI(BB, dl, TII->get(PPC::AND),TmpReg)
.addReg(TmpDestReg).addReg(MaskReg);
BuildMI(BB, dl, TII->get(PPC::CMPW), PPC::CR0)
.addReg(TmpReg).addReg(OldVal3Reg);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(midMBB);
BB->addSuccessor(loop2MBB);
BB->addSuccessor(midMBB);
BB = loop2MBB;
BuildMI(BB, dl, TII->get(PPC::ANDC),Tmp2Reg)
.addReg(TmpDestReg).addReg(MaskReg);
BuildMI(BB, dl, TII->get(PPC::OR),Tmp4Reg)
.addReg(Tmp2Reg).addReg(NewVal3Reg);
BuildMI(BB, dl, TII->get(PPC::STWCX)).addReg(Tmp4Reg)
.addReg(ZeroReg).addReg(PtrReg);
BuildMI(BB, dl, TII->get(PPC::BCC))
.addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(loop1MBB);
BuildMI(BB, dl, TII->get(PPC::B)).addMBB(exitMBB);
BB->addSuccessor(loop1MBB);
BB->addSuccessor(exitMBB);
BB = midMBB;
BuildMI(BB, dl, TII->get(PPC::STWCX)).addReg(TmpDestReg)
.addReg(ZeroReg).addReg(PtrReg);
BB->addSuccessor(exitMBB);
BB = exitMBB;
BuildMI(*BB, BB->begin(), dl, TII->get(PPC::SRW),dest).addReg(TmpReg)
.addReg(ShiftReg);
} else {
llvm_unreachable("Unexpected instr type to insert");
}
MI->eraseFromParent(); return BB;
}
SDValue PPCTargetLowering::PerformDAGCombine(SDNode *N,
DAGCombinerInfo &DCI) const {
const TargetMachine &TM = getTargetMachine();
SelectionDAG &DAG = DCI.DAG;
DebugLoc dl = N->getDebugLoc();
switch (N->getOpcode()) {
default: break;
case PPCISD::SHL:
if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(N->getOperand(0))) {
if (C->isNullValue()) return N->getOperand(0);
}
break;
case PPCISD::SRL:
if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(N->getOperand(0))) {
if (C->isNullValue()) return N->getOperand(0);
}
break;
case PPCISD::SRA:
if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(N->getOperand(0))) {
if (C->isNullValue() || C->isAllOnesValue()) return N->getOperand(0);
}
break;
case ISD::SINT_TO_FP:
if (TM.getSubtarget<PPCSubtarget>().has64BitSupport()) {
if (N->getOperand(0).getOpcode() == ISD::FP_TO_SINT) {
if (N->getOperand(0).getValueType() == MVT::i64 &&
N->getOperand(0).getOperand(0).getValueType() != MVT::ppcf128) {
SDValue Val = N->getOperand(0).getOperand(0);
if (Val.getValueType() == MVT::f32) {
Val = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Val);
DCI.AddToWorklist(Val.getNode());
}
Val = DAG.getNode(PPCISD::FCTIDZ, dl, MVT::f64, Val);
DCI.AddToWorklist(Val.getNode());
Val = DAG.getNode(PPCISD::FCFID, dl, MVT::f64, Val);
DCI.AddToWorklist(Val.getNode());
if (N->getValueType(0) == MVT::f32) {
Val = DAG.getNode(ISD::FP_ROUND, dl, MVT::f32, Val,
DAG.getIntPtrConstant(0));
DCI.AddToWorklist(Val.getNode());
}
return Val;
} else if (N->getOperand(0).getValueType() == MVT::i32) {
}
}
}
break;
case ISD::STORE:
if (TM.getSubtarget<PPCSubtarget>().hasSTFIWX() &&
!cast<StoreSDNode>(N)->isTruncatingStore() &&
N->getOperand(1).getOpcode() == ISD::FP_TO_SINT &&
N->getOperand(1).getValueType() == MVT::i32 &&
N->getOperand(1).getOperand(0).getValueType() != MVT::ppcf128) {
SDValue Val = N->getOperand(1).getOperand(0);
if (Val.getValueType() == MVT::f32) {
Val = DAG.getNode(ISD::FP_EXTEND, dl, MVT::f64, Val);
DCI.AddToWorklist(Val.getNode());
}
Val = DAG.getNode(PPCISD::FCTIWZ, dl, MVT::f64, Val);
DCI.AddToWorklist(Val.getNode());
Val = DAG.getNode(PPCISD::STFIWX, dl, MVT::Other, N->getOperand(0), Val,
N->getOperand(2), N->getOperand(3));
DCI.AddToWorklist(Val.getNode());
return Val;
}
if (cast<StoreSDNode>(N)->isUnindexed() &&
N->getOperand(1).getOpcode() == ISD::BSWAP &&
N->getOperand(1).getNode()->hasOneUse() &&
(N->getOperand(1).getValueType() == MVT::i32 ||
N->getOperand(1).getValueType() == MVT::i16)) {
SDValue BSwapOp = N->getOperand(1).getOperand(0);
if (BSwapOp.getValueType() == MVT::i16)
BSwapOp = DAG.getNode(ISD::ANY_EXTEND, dl, MVT::i32, BSwapOp);
SDValue Ops[] = {
N->getOperand(0), BSwapOp, N->getOperand(2),
DAG.getValueType(N->getOperand(1).getValueType())
};
return
DAG.getMemIntrinsicNode(PPCISD::STBRX, dl, DAG.getVTList(MVT::Other),
Ops, array_lengthof(Ops),
cast<StoreSDNode>(N)->getMemoryVT(),
cast<StoreSDNode>(N)->getMemOperand());
}
break;
case ISD::BSWAP:
if (ISD::isNON_EXTLoad(N->getOperand(0).getNode()) &&
N->getOperand(0).hasOneUse() &&
(N->getValueType(0) == MVT::i32 || N->getValueType(0) == MVT::i16)) {
SDValue Load = N->getOperand(0);
LoadSDNode *LD = cast<LoadSDNode>(Load);
SDValue Ops[] = {
LD->getChain(), LD->getBasePtr(), DAG.getValueType(N->getValueType(0)) };
SDValue BSLoad =
DAG.getMemIntrinsicNode(PPCISD::LBRX, dl,
DAG.getVTList(MVT::i32, MVT::Other), Ops, 3,
LD->getMemoryVT(), LD->getMemOperand());
SDValue ResVal = BSLoad;
if (N->getValueType(0) == MVT::i16)
ResVal = DAG.getNode(ISD::TRUNCATE, dl, MVT::i16, BSLoad);
DCI.CombineTo(N, ResVal);
DCI.CombineTo(Load.getNode(), ResVal, BSLoad.getValue(1));
return SDValue(N, 0);
}
break;
case PPCISD::VCMP: {
if (!N->getOperand(0).hasOneUse() &&
!N->getOperand(1).hasOneUse() &&
!N->getOperand(2).hasOneUse()) {
SDNode *VCMPoNode = 0;
SDNode *LHSN = N->getOperand(0).getNode();
for (SDNode::use_iterator UI = LHSN->use_begin(), E = LHSN->use_end();
UI != E; ++UI)
if (UI->getOpcode() == PPCISD::VCMPo &&
UI->getOperand(1) == N->getOperand(1) &&
UI->getOperand(2) == N->getOperand(2) &&
UI->getOperand(0) == N->getOperand(0)) {
VCMPoNode = *UI;
break;
}
if (!VCMPoNode || VCMPoNode->hasNUsesOfValue(0, 1))
break;
SDNode *FlagUser = 0;
for (SDNode::use_iterator UI = VCMPoNode->use_begin();
FlagUser == 0; ++UI) {
assert(UI != VCMPoNode->use_end() && "Didn't find user!");
SDNode *User = *UI;
for (unsigned i = 0, e = User->getNumOperands(); i != e; ++i) {
if (User->getOperand(i) == SDValue(VCMPoNode, 1)) {
FlagUser = User;
break;
}
}
}
if (FlagUser->getOpcode() == PPCISD::MFCR)
return SDValue(VCMPoNode, 0);
}
break;
}
case ISD::BR_CC: {
ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(1))->get();
SDValue LHS = N->getOperand(2), RHS = N->getOperand(3);
int CompareOpc;
bool isDot;
if (LHS.getOpcode() == ISD::INTRINSIC_WO_CHAIN &&
isa<ConstantSDNode>(RHS) && (CC == ISD::SETEQ || CC == ISD::SETNE) &&
getAltivecCompareInfo(LHS, CompareOpc, isDot)) {
assert(isDot && "Can't compare against a vector result!");
unsigned Val = cast<ConstantSDNode>(RHS)->getZExtValue();
if (Val != 0 && Val != 1) {
if (CC == ISD::SETEQ) return N->getOperand(0);
return DAG.getNode(ISD::BR, dl, MVT::Other,
N->getOperand(0), N->getOperand(4));
}
bool BranchOnWhenPredTrue = (CC == ISD::SETEQ) ^ (Val == 0);
SDValue Ops[] = {
LHS.getOperand(2), LHS.getOperand(3), DAG.getConstant(CompareOpc, MVT::i32)
};
EVT VTs[] = { LHS.getOperand(2).getValueType(), MVT::Glue };
SDValue CompNode = DAG.getNode(PPCISD::VCMPo, dl, VTs, Ops, 3);
PPC::Predicate CompOpc;
switch (cast<ConstantSDNode>(LHS.getOperand(1))->getZExtValue()) {
default: case 0: CompOpc = BranchOnWhenPredTrue ? PPC::PRED_EQ : PPC::PRED_NE;
break;
case 1: CompOpc = BranchOnWhenPredTrue ? PPC::PRED_NE : PPC::PRED_EQ;
break;
case 2: CompOpc = BranchOnWhenPredTrue ? PPC::PRED_LT : PPC::PRED_GE;
break;
case 3: CompOpc = BranchOnWhenPredTrue ? PPC::PRED_GE : PPC::PRED_LT;
break;
}
return DAG.getNode(PPCISD::COND_BRANCH, dl, MVT::Other, N->getOperand(0),
DAG.getConstant(CompOpc, MVT::i32),
DAG.getRegister(PPC::CR6, MVT::i32),
N->getOperand(4), CompNode.getValue(1));
}
break;
}
}
return SDValue();
}
void PPCTargetLowering::computeMaskedBitsForTargetNode(const SDValue Op,
APInt &KnownZero,
APInt &KnownOne,
const SelectionDAG &DAG,
unsigned Depth) const {
KnownZero = KnownOne = APInt(KnownZero.getBitWidth(), 0);
switch (Op.getOpcode()) {
default: break;
case PPCISD::LBRX: {
if (cast<VTSDNode>(Op.getOperand(2))->getVT() == MVT::i16)
KnownZero = 0xFFFF0000;
break;
}
case ISD::INTRINSIC_WO_CHAIN: {
switch (cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue()) {
default: break;
case Intrinsic::ppc_altivec_vcmpbfp_p:
case Intrinsic::ppc_altivec_vcmpeqfp_p:
case Intrinsic::ppc_altivec_vcmpequb_p:
case Intrinsic::ppc_altivec_vcmpequh_p:
case Intrinsic::ppc_altivec_vcmpequw_p:
case Intrinsic::ppc_altivec_vcmpgefp_p:
case Intrinsic::ppc_altivec_vcmpgtfp_p:
case Intrinsic::ppc_altivec_vcmpgtsb_p:
case Intrinsic::ppc_altivec_vcmpgtsh_p:
case Intrinsic::ppc_altivec_vcmpgtsw_p:
case Intrinsic::ppc_altivec_vcmpgtub_p:
case Intrinsic::ppc_altivec_vcmpgtuh_p:
case Intrinsic::ppc_altivec_vcmpgtuw_p:
KnownZero = ~1U; break;
}
}
}
}
PPCTargetLowering::ConstraintType
PPCTargetLowering::getConstraintType(const std::string &Constraint) const {
if (Constraint.size() == 1) {
switch (Constraint[0]) {
default: break;
case 'b':
case 'r':
case 'f':
case 'v':
case 'y':
return C_RegisterClass;
case 'Z':
return C_Memory;
}
}
return TargetLowering::getConstraintType(Constraint);
}
TargetLowering::ConstraintWeight
PPCTargetLowering::getSingleConstraintMatchWeight(
AsmOperandInfo &info, const char *constraint) const {
ConstraintWeight weight = CW_Invalid;
Value *CallOperandVal = info.CallOperandVal;
if (CallOperandVal == NULL)
return CW_Default;
Type *type = CallOperandVal->getType();
switch (*constraint) {
default:
weight = TargetLowering::getSingleConstraintMatchWeight(info, constraint);
break;
case 'b':
if (type->isIntegerTy())
weight = CW_Register;
break;
case 'f':
if (type->isFloatTy())
weight = CW_Register;
break;
case 'd':
if (type->isDoubleTy())
weight = CW_Register;
break;
case 'v':
if (type->isVectorTy())
weight = CW_Register;
break;
case 'y':
weight = CW_Register;
break;
case 'Z':
weight = CW_Memory;
break;
}
return weight;
}
std::pair<unsigned, const TargetRegisterClass*>
PPCTargetLowering::getRegForInlineAsmConstraint(const std::string &Constraint,
EVT VT) const {
if (Constraint.size() == 1) {
switch (Constraint[0]) {
case 'b': case 'r': if (VT == MVT::i64 && PPCSubTarget.isPPC64())
return std::make_pair(0U, &PPC::G8RCRegClass);
return std::make_pair(0U, &PPC::GPRCRegClass);
case 'f':
if (VT == MVT::f32 || VT == MVT::i32)
return std::make_pair(0U, &PPC::F4RCRegClass);
if (VT == MVT::f64 || VT == MVT::i64)
return std::make_pair(0U, &PPC::F8RCRegClass);
break;
case 'v':
return std::make_pair(0U, &PPC::VRRCRegClass);
case 'y': return std::make_pair(0U, &PPC::CRRCRegClass);
}
}
return TargetLowering::getRegForInlineAsmConstraint(Constraint, VT);
}
void PPCTargetLowering::LowerAsmOperandForConstraint(SDValue Op,
std::string &Constraint,
std::vector<SDValue>&Ops,
SelectionDAG &DAG) const {
SDValue Result(0,0);
if (Constraint.length() > 1) return;
char Letter = Constraint[0];
switch (Letter) {
default: break;
case 'I':
case 'J':
case 'K':
case 'L':
case 'M':
case 'N':
case 'O':
case 'P': {
ConstantSDNode *CST = dyn_cast<ConstantSDNode>(Op);
if (!CST) return; unsigned Value = CST->getZExtValue();
switch (Letter) {
default: llvm_unreachable("Unknown constraint letter!");
case 'I': if ((short)Value == (int)Value)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'J': case 'L': if ((short)Value == 0)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'K': if ((Value >> 16) == 0)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'M': if (Value > 31)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'N': if ((int)Value > 0 && isPowerOf2_32(Value))
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'O': if (Value == 0)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
case 'P': if ((short)-Value == (int)-Value)
Result = DAG.getTargetConstant(Value, Op.getValueType());
break;
}
break;
}
}
if (Result.getNode()) {
Ops.push_back(Result);
return;
}
TargetLowering::LowerAsmOperandForConstraint(Op, Constraint, Ops, DAG);
}
bool PPCTargetLowering::isLegalAddressingMode(const AddrMode &AM,
Type *Ty) const {
if (AM.BaseOffs <= -(1LL << 16) || AM.BaseOffs >= (1LL << 16)-1)
return false;
if (AM.BaseGV)
return false;
switch (AM.Scale) {
case 0: break;
case 1:
if (AM.HasBaseReg && AM.BaseOffs) return false;
break;
case 2:
if (AM.HasBaseReg || AM.BaseOffs) return false;
break;
default:
return false;
}
return true;
}
bool PPCTargetLowering::isLegalAddressImmediate(int64_t V,Type *Ty) const{
return (V > -(1 << 16) && V < (1 << 16)-1);
}
bool PPCTargetLowering::isLegalAddressImmediate(GlobalValue* GV) const {
return false;
}
SDValue PPCTargetLowering::LowerRETURNADDR(SDValue Op,
SelectionDAG &DAG) const {
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *MFI = MF.getFrameInfo();
MFI->setReturnAddressIsTaken(true);
DebugLoc dl = Op.getDebugLoc();
unsigned Depth = cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue();
PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
FuncInfo->setLRStoreRequired();
bool isPPC64 = PPCSubTarget.isPPC64();
bool isDarwinABI = PPCSubTarget.isDarwinABI();
if (Depth > 0) {
SDValue FrameAddr = LowerFRAMEADDR(Op, DAG);
SDValue Offset =
DAG.getConstant(PPCFrameLowering::getReturnSaveOffset(isPPC64, isDarwinABI),
isPPC64? MVT::i64 : MVT::i32);
return DAG.getLoad(getPointerTy(), dl, DAG.getEntryNode(),
DAG.getNode(ISD::ADD, dl, getPointerTy(),
FrameAddr, Offset),
MachinePointerInfo(), false, false, false, 0);
}
SDValue RetAddrFI = getReturnAddrFrameIndex(DAG);
return DAG.getLoad(getPointerTy(), dl, DAG.getEntryNode(),
RetAddrFI, MachinePointerInfo(), false, false, false, 0);
}
SDValue PPCTargetLowering::LowerFRAMEADDR(SDValue Op,
SelectionDAG &DAG) const {
DebugLoc dl = Op.getDebugLoc();
unsigned Depth = cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue();
EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
bool isPPC64 = PtrVT == MVT::i64;
MachineFunction &MF = DAG.getMachineFunction();
MachineFrameInfo *MFI = MF.getFrameInfo();
MFI->setFrameAddressIsTaken(true);
bool is31 = (getTargetMachine().Options.DisableFramePointerElim(MF) ||
MFI->hasVarSizedObjects()) &&
MFI->getStackSize() &&
!MF.getFunction()->getAttributes().
hasAttribute(AttributeSet::FunctionIndex, Attribute::Naked);
unsigned FrameReg = isPPC64 ? (is31 ? PPC::X31 : PPC::X1) :
(is31 ? PPC::R31 : PPC::R1);
SDValue FrameAddr = DAG.getCopyFromReg(DAG.getEntryNode(), dl, FrameReg,
PtrVT);
while (Depth--)
FrameAddr = DAG.getLoad(Op.getValueType(), dl, DAG.getEntryNode(),
FrameAddr, MachinePointerInfo(), false, false,
false, 0);
return FrameAddr;
}
bool
PPCTargetLowering::isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const {
return false;
}
EVT PPCTargetLowering::getOptimalMemOpType(uint64_t Size,
unsigned DstAlign, unsigned SrcAlign,
bool IsMemset, bool ZeroMemset,
bool MemcpyStrSrc,
MachineFunction &MF) const {
if (this->PPCSubTarget.isPPC64()) {
return MVT::i64;
} else {
return MVT::i32;
}
}
bool PPCTargetLowering::isFMAFasterThanMulAndAdd(EVT VT) const {
if (!VT.isSimple())
return false;
switch (VT.getSimpleVT().SimpleTy) {
case MVT::f32:
case MVT::f64:
case MVT::v4f32:
return true;
default:
break;
}
return false;
}
Sched::Preference PPCTargetLowering::getSchedulingPreference(SDNode *N) const {
if (DisableILPPref)
return TargetLowering::getSchedulingPreference(N);
return Sched::ILP;
}