OSchip
/
llvm-project
mirror of https://github.com/THU-DSP-LAB/llvm-project.git
13
0
Fork 6
Code Issues Packages Projects Releases Wiki Activity
llvm-project/llvm/lib/Target/PowerPC/PPC64ISelSimple.cpp

2948 lines
107 KiB
C++
Raw Blame History

//===-- PPC64ISelSimple.cpp - A simple instruction selector for PowerPC ---===//
//
// The LLVM Compiler Infrastructure
//
// This file was developed by the LLVM research group and is distributed under
// the University of Illinois Open Source License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#define DEBUG_TYPE "isel"
#include "PowerPC.h"
#include "PowerPCInstrBuilder.h"
#include "PowerPCInstrInfo.h"
#include "PPC64TargetMachine.h"
#include "llvm/Constants.h"
#include "llvm/DerivedTypes.h"
#include "llvm/Function.h"
#include "llvm/Instructions.h"
#include "llvm/Pass.h"
#include "llvm/CodeGen/IntrinsicLowering.h"
#include "llvm/CodeGen/MachineConstantPool.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/SSARegMap.h"
#include "llvm/Target/MRegisterInfo.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Support/GetElementPtrTypeIterator.h"
#include "llvm/Support/InstVisitor.h"
#include "llvm/Support/Debug.h"
#include "llvm/ADT/Statistic.h"
#include <vector>
using namespace llvm;
namespace {
Statistic<> GEPFolds("ppc64-codegen", "Number of GEPs folded");
/// TypeClass - Used by the PowerPC backend to group LLVM types by their basic
/// PPC Representation.
///
enum TypeClass {
cByte, cShort, cInt, cFP32, cFP64, cLong
};
}
/// getClass - Turn a primitive type into a "class" number which is based on the
/// size of the type, and whether or not it is floating point.
///
static inline TypeClass getClass(const Type *Ty) {
switch (Ty->getTypeID()) {
case Type::SByteTyID:
case Type::UByteTyID: return cByte; // Byte operands are class #0
case Type::ShortTyID:
case Type::UShortTyID: return cShort; // Short operands are class #1
case Type::IntTyID:
case Type::UIntTyID: return cInt; // Ints are class #2
case Type::FloatTyID: return cFP32; // Single float is #3
case Type::DoubleTyID: return cFP64; // Double Point is #4
case Type::PointerTyID:
case Type::LongTyID:
case Type::ULongTyID: return cLong; // Longs and pointers are class #5
default:
assert(0 && "Invalid type to getClass!");
return cByte; // not reached
}
}
// getClassB - Just like getClass, but treat boolean values as ints.
static inline TypeClass getClassB(const Type *Ty) {
if (Ty == Type::BoolTy) return cInt;
return getClass(Ty);
}
namespace {
struct PPC64ISel : public FunctionPass, InstVisitor<PPC64ISel> {
PPC64TargetMachine &TM;
MachineFunction *F; // The function we are compiling into
MachineBasicBlock *BB; // The current MBB we are compiling
int VarArgsFrameIndex; // FrameIndex for start of varargs area
std::map<Value*, unsigned> RegMap; // Mapping between Values and SSA Regs
// External functions used in the Module
Function *fmodfFn, *fmodFn, *__cmpdi2Fn, *__fixsfdiFn, *__fixdfdiFn,
*__fixunssfdiFn, *__fixunsdfdiFn, *mallocFn, *freeFn;
// MBBMap - Mapping between LLVM BB -> Machine BB
std::map<const BasicBlock*, MachineBasicBlock*> MBBMap;
// AllocaMap - Mapping from fixed sized alloca instructions to the
// FrameIndex for the alloca.
std::map<AllocaInst*, unsigned> AllocaMap;
// Target configuration data
const unsigned ParameterSaveAreaOffset, MaxArgumentStackSpace;
PPC64ISel(TargetMachine &tm):TM(reinterpret_cast<PPC64TargetMachine&>(tm)),
F(0), BB(0), ParameterSaveAreaOffset(24), MaxArgumentStackSpace(32) {}
bool doInitialization(Module &M) {
// Add external functions that we may call
Type *i = Type::IntTy;
Type *d = Type::DoubleTy;
Type *f = Type::FloatTy;
Type *l = Type::LongTy;
Type *ul = Type::ULongTy;
Type *voidPtr = PointerType::get(Type::SByteTy);
// float fmodf(float, float);
fmodfFn = M.getOrInsertFunction("fmodf", f, f, f, 0);
// double fmod(double, double);
fmodFn = M.getOrInsertFunction("fmod", d, d, d, 0);
// int __cmpdi2(long, long);
__cmpdi2Fn = M.getOrInsertFunction("__cmpdi2", i, l, l, 0);
// long __fixsfdi(float)
__fixsfdiFn = M.getOrInsertFunction("__fixsfdi", l, f, 0);
// long __fixdfdi(double)
__fixdfdiFn = M.getOrInsertFunction("__fixdfdi", l, d, 0);
// unsigned long __fixunssfdi(float)
__fixunssfdiFn = M.getOrInsertFunction("__fixunssfdi", ul, f, 0);
// unsigned long __fixunsdfdi(double)
__fixunsdfdiFn = M.getOrInsertFunction("__fixunsdfdi", ul, d, 0);
// void* malloc(size_t)
mallocFn = M.getOrInsertFunction("malloc", voidPtr, Type::UIntTy, 0);
// void free(void*)
freeFn = M.getOrInsertFunction("free", Type::VoidTy, voidPtr, 0);
return false;
}
/// runOnFunction - Top level implementation of instruction selection for
/// the entire function.
///
bool runOnFunction(Function &Fn) {
// First pass over the function, lower any unknown intrinsic functions
// with the IntrinsicLowering class.
LowerUnknownIntrinsicFunctionCalls(Fn);
F = &MachineFunction::construct(&Fn, TM);
// Create all of the machine basic blocks for the function...
for (Function::iterator I = Fn.begin(), E = Fn.end(); I != E; ++I)
F->getBasicBlockList().push_back(MBBMap[I] = new MachineBasicBlock(I));
BB = &F->front();
// Copy incoming arguments off of the stack...
LoadArgumentsToVirtualRegs(Fn);
// Instruction select everything except PHI nodes
visit(Fn);
// Select the PHI nodes
SelectPHINodes();
RegMap.clear();
MBBMap.clear();
AllocaMap.clear();
F = 0;
// We always build a machine code representation for the function
return true;
}
virtual const char *getPassName() const {
return "PowerPC Simple Instruction Selection";
}
/// visitBasicBlock - This method is called when we are visiting a new basic
/// block. This simply creates a new MachineBasicBlock to emit code into
/// and adds it to the current MachineFunction. Subsequent visit* for
/// instructions will be invoked for all instructions in the basic block.
///
void visitBasicBlock(BasicBlock &LLVM_BB) {
BB = MBBMap[&LLVM_BB];
}
/// LowerUnknownIntrinsicFunctionCalls - This performs a prepass over the
/// function, lowering any calls to unknown intrinsic functions into the
/// equivalent LLVM code.
///
void LowerUnknownIntrinsicFunctionCalls(Function &F);
/// LoadArgumentsToVirtualRegs - Load all of the arguments to this function
/// from the stack into virtual registers.
///
void LoadArgumentsToVirtualRegs(Function &F);
/// SelectPHINodes - Insert machine code to generate phis. This is tricky
/// because we have to generate our sources into the source basic blocks,
/// not the current one.
///
void SelectPHINodes();
// Visitation methods for various instructions. These methods simply emit
// fixed PowerPC code for each instruction.
// Control flow operators
void visitReturnInst(ReturnInst &RI);
void visitBranchInst(BranchInst &BI);
struct ValueRecord {
Value *Val;
unsigned Reg;
const Type *Ty;
ValueRecord(unsigned R, const Type *T) : Val(0), Reg(R), Ty(T) {}
ValueRecord(Value *V) : Val(V), Reg(0), Ty(V->getType()) {}
};
// This struct is for recording the necessary operations to emit the GEP
struct CollapsedGepOp {
bool isMul;
Value *index;
ConstantSInt *size;
CollapsedGepOp(bool mul, Value *i, ConstantSInt *s) :
isMul(mul), index(i), size(s) {}
};
void doCall(const ValueRecord &Ret, MachineInstr *CallMI,
const std::vector<ValueRecord> &Args, bool isVarArg);
void visitCallInst(CallInst &I);
void visitIntrinsicCall(Intrinsic::ID ID, CallInst &I);
// Arithmetic operators
void visitSimpleBinary(BinaryOperator &B, unsigned OpcodeClass);
void visitAdd(BinaryOperator &B) { visitSimpleBinary(B, 0); }
void visitSub(BinaryOperator &B) { visitSimpleBinary(B, 1); }
void visitMul(BinaryOperator &B);
void visitDiv(BinaryOperator &B) { visitDivRem(B); }
void visitRem(BinaryOperator &B) { visitDivRem(B); }
void visitDivRem(BinaryOperator &B);
// Bitwise operators
void visitAnd(BinaryOperator &B) { visitSimpleBinary(B, 2); }
void visitOr (BinaryOperator &B) { visitSimpleBinary(B, 3); }
void visitXor(BinaryOperator &B) { visitSimpleBinary(B, 4); }
// Comparison operators...
void visitSetCondInst(SetCondInst &I);
unsigned EmitComparison(unsigned OpNum, Value *Op0, Value *Op1,
MachineBasicBlock *MBB,
MachineBasicBlock::iterator MBBI);
void visitSelectInst(SelectInst &SI);
// Memory Instructions
void visitLoadInst(LoadInst &I);
void visitStoreInst(StoreInst &I);
void visitGetElementPtrInst(GetElementPtrInst &I);
void visitAllocaInst(AllocaInst &I);
void visitMallocInst(MallocInst &I);
void visitFreeInst(FreeInst &I);
// Other operators
void visitShiftInst(ShiftInst &I);
void visitPHINode(PHINode &I) {} // PHI nodes handled by second pass
void visitCastInst(CastInst &I);
void visitVANextInst(VANextInst &I);
void visitVAArgInst(VAArgInst &I);
void visitInstruction(Instruction &I) {
std::cerr << "Cannot instruction select: " << I;
abort();
}
/// promote32 - Make a value 32-bits wide, and put it somewhere.
///
void promote32(unsigned targetReg, const ValueRecord &VR);
/// emitGEPOperation - Common code shared between visitGetElementPtrInst and
/// constant expression GEP support.
///
void emitGEPOperation(MachineBasicBlock *BB, MachineBasicBlock::iterator IP,
Value *Src, User::op_iterator IdxBegin,
User::op_iterator IdxEnd, unsigned TargetReg,
bool CollapseRemainder, ConstantSInt **Remainder,
unsigned *PendingAddReg);
/// emitCastOperation - Common code shared between visitCastInst and
/// constant expression cast support.
///
void emitCastOperation(MachineBasicBlock *BB,MachineBasicBlock::iterator IP,
Value *Src, const Type *DestTy, unsigned TargetReg);
/// emitSimpleBinaryOperation - Common code shared between visitSimpleBinary
/// and constant expression support.
///
void emitSimpleBinaryOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1,
unsigned OperatorClass, unsigned TargetReg);
/// emitBinaryFPOperation - This method handles emission of floating point
/// Add (0), Sub (1), Mul (2), and Div (3) operations.
void emitBinaryFPOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1,
unsigned OperatorClass, unsigned TargetReg);
void emitMultiply(MachineBasicBlock *BB, MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1, unsigned TargetReg);
void doMultiply(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
unsigned DestReg, Value *Op0, Value *Op1);
/// doMultiplyConst - This method will multiply the value in Op0Reg by the
/// value of the ContantInt *CI
void doMultiplyConst(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
unsigned DestReg, Value *Op0, ConstantInt *CI);
void emitDivRemOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1, bool isDiv,
unsigned TargetReg);
/// emitSetCCOperation - Common code shared between visitSetCondInst and
/// constant expression support.
///
void emitSetCCOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1, unsigned Opcode,
unsigned TargetReg);
/// emitShiftOperation - Common code shared between visitShiftInst and
/// constant expression support.
///
void emitShiftOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Op, Value *ShiftAmount, bool isLeftShift,
const Type *ResultTy, unsigned DestReg);
/// emitSelectOperation - Common code shared between visitSelectInst and the
/// constant expression support.
///
void emitSelectOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Cond, Value *TrueVal, Value *FalseVal,
unsigned DestReg);
/// copyConstantToRegister - Output the instructions required to put the
/// specified constant into the specified register.
///
void copyConstantToRegister(MachineBasicBlock *MBB,
MachineBasicBlock::iterator MBBI,
Constant *C, unsigned Reg);
void emitUCOM(MachineBasicBlock *MBB, MachineBasicBlock::iterator MBBI,
unsigned LHS, unsigned RHS);
/// makeAnotherReg - This method returns the next register number we haven't
/// yet used.
///
unsigned makeAnotherReg(const Type *Ty) {
assert(dynamic_cast<const PPC64RegisterInfo*>(TM.getRegisterInfo()) &&
"Current target doesn't have PPC reg info??");
const PPC64RegisterInfo *PPCRI =
static_cast<const PPC64RegisterInfo*>(TM.getRegisterInfo());
// Add the mapping of regnumber => reg class to MachineFunction
const TargetRegisterClass *RC = PPCRI->getRegClassForType(Ty);
return F->getSSARegMap()->createVirtualRegister(RC);
}
/// getReg - This method turns an LLVM value into a register number.
///
unsigned getReg(Value &V) { return getReg(&V); } // Allow references
unsigned getReg(Value *V) {
// Just append to the end of the current bb.
MachineBasicBlock::iterator It = BB->end();
return getReg(V, BB, It);
}
unsigned getReg(Value *V, MachineBasicBlock *MBB,
MachineBasicBlock::iterator IPt);
/// canUseAsImmediateForOpcode - This method returns whether a ConstantInt
/// is okay to use as an immediate argument to a certain binary operation
bool canUseAsImmediateForOpcode(ConstantInt *CI, unsigned Opcode);
/// getFixedSizedAllocaFI - Return the frame index for a fixed sized alloca
/// that is to be statically allocated with the initial stack frame
/// adjustment.
unsigned getFixedSizedAllocaFI(AllocaInst *AI);
};
}
/// dyn_castFixedAlloca - If the specified value is a fixed size alloca
/// instruction in the entry block, return it. Otherwise, return a null
/// pointer.
static AllocaInst *dyn_castFixedAlloca(Value *V) {
if (AllocaInst *AI = dyn_cast<AllocaInst>(V)) {
BasicBlock *BB = AI->getParent();
if (isa<ConstantUInt>(AI->getArraySize()) && BB ==&BB->getParent()->front())
return AI;
}
return 0;
}
/// getReg - This method turns an LLVM value into a register number.
///
unsigned PPC64ISel::getReg(Value *V, MachineBasicBlock *MBB,
MachineBasicBlock::iterator IPt) {
if (Constant *C = dyn_cast<Constant>(V)) {
unsigned Reg = makeAnotherReg(V->getType());
copyConstantToRegister(MBB, IPt, C, Reg);
return Reg;
} else if (AllocaInst *AI = dyn_castFixedAlloca(V)) {
unsigned Reg = makeAnotherReg(V->getType());
unsigned FI = getFixedSizedAllocaFI(AI);
addFrameReference(BuildMI(*MBB, IPt, PPC::ADDI, 2, Reg), FI, 0, false);
return Reg;
}
unsigned &Reg = RegMap[V];
if (Reg == 0) {
Reg = makeAnotherReg(V->getType());
RegMap[V] = Reg;
}
return Reg;
}
/// canUseAsImmediateForOpcode - This method returns whether a ConstantInt
/// is okay to use as an immediate argument to a certain binary operator.
///
/// Operator is one of: 0 for Add, 1 for Sub, 2 for And, 3 for Or, 4 for Xor.
bool PPC64ISel::canUseAsImmediateForOpcode(ConstantInt *CI, unsigned Operator) {
ConstantSInt *Op1Cs;
ConstantUInt *Op1Cu;
// ADDI, Compare, and non-indexed Load take SIMM
bool cond1 = (Operator == 0)
&& (Op1Cs = dyn_cast<ConstantSInt>(CI))
&& (Op1Cs->getValue() <= 32767)
&& (Op1Cs->getValue() >= -32768);
// SUBI takes -SIMM since it is a mnemonic for ADDI
bool cond2 = (Operator == 1)
&& (Op1Cs = dyn_cast<ConstantSInt>(CI))
&& (Op1Cs->getValue() <= 32768)
&& (Op1Cs->getValue() >= -32767);
// ANDIo, ORI, and XORI take unsigned values
bool cond3 = (Operator >= 2)
&& (Op1Cs = dyn_cast<ConstantSInt>(CI))
&& (Op1Cs->getValue() >= 0)
&& (Op1Cs->getValue() <= 32767);
// ADDI and SUBI take SIMMs, so we have to make sure the UInt would fit
bool cond4 = (Operator < 2)
&& (Op1Cu = dyn_cast<ConstantUInt>(CI))
&& (Op1Cu->getValue() <= 32767);
// ANDIo, ORI, and XORI take UIMMs, so they can be larger
bool cond5 = (Operator >= 2)
&& (Op1Cu = dyn_cast<ConstantUInt>(CI))
&& (Op1Cu->getValue() <= 65535);
if (cond1 || cond2 || cond3 || cond4 || cond5)
return true;
return false;
}
/// getFixedSizedAllocaFI - Return the frame index for a fixed sized alloca
/// that is to be statically allocated with the initial stack frame
/// adjustment.
unsigned PPC64ISel::getFixedSizedAllocaFI(AllocaInst *AI) {
// Already computed this?
std::map<AllocaInst*, unsigned>::iterator I = AllocaMap.lower_bound(AI);
if (I != AllocaMap.end() && I->first == AI) return I->second;
const Type *Ty = AI->getAllocatedType();
ConstantUInt *CUI = cast<ConstantUInt>(AI->getArraySize());
unsigned TySize = TM.getTargetData().getTypeSize(Ty);
TySize *= CUI->getValue(); // Get total allocated size...
unsigned Alignment = TM.getTargetData().getTypeAlignment(Ty);
// Create a new stack object using the frame manager...
int FrameIdx = F->getFrameInfo()->CreateStackObject(TySize, Alignment);
AllocaMap.insert(I, std::make_pair(AI, FrameIdx));
return FrameIdx;
}
/// copyConstantToRegister - Output the instructions required to put the
/// specified constant into the specified register.
///
void PPC64ISel::copyConstantToRegister(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Constant *C, unsigned R) {
if (C->getType()->isIntegral()) {
unsigned Class = getClassB(C->getType());
if (Class == cLong) {
if (ConstantUInt *CUI = dyn_cast<ConstantUInt>(C)) {
uint64_t uval = CUI->getValue();
if (uval < (1LL << 32)) {
ConstantUInt *CU = ConstantUInt::get(Type::UIntTy, uval);
copyConstantToRegister(MBB, IP, CU, R);
return;
}
} else if (ConstantSInt *CSI = dyn_cast<ConstantSInt>(C)) {
int64_t val = CUI->getValue();
if (val < (1LL << 31)) {
ConstantUInt *CU = ConstantUInt::get(Type::UIntTy, val);
copyConstantToRegister(MBB, IP, CU, R);
return;
}
} else {
std::cerr << "Unhandled long constant type!\n";
abort();
}
// Spill long to the constant pool and load it
MachineConstantPool *CP = F->getConstantPool();
unsigned CPI = CP->getConstantPoolIndex(C);
BuildMI(*MBB, IP, PPC::LD, 1, R)
.addReg(PPC::R2).addConstantPoolIndex(CPI);
return;
}
assert(Class <= cInt && "Type not handled yet!");
// Handle bool
if (C->getType() == Type::BoolTy) {
BuildMI(*MBB, IP, PPC::LI, 1, R).addSImm(C == ConstantBool::True);
return;
}
// Handle int
if (ConstantUInt *CUI = dyn_cast<ConstantUInt>(C)) {
unsigned uval = CUI->getValue();
if (uval < 32768) {
BuildMI(*MBB, IP, PPC::LI, 1, R).addSImm(uval);
} else {
unsigned Temp = makeAnotherReg(Type::IntTy);
BuildMI(*MBB, IP, PPC::LIS, 1, Temp).addSImm(uval >> 16);
BuildMI(*MBB, IP, PPC::ORI, 2, R).addReg(Temp).addImm(uval);
}
return;
} else if (ConstantSInt *CSI = dyn_cast<ConstantSInt>(C)) {
int sval = CSI->getValue();
if (sval < 32768 && sval >= -32768) {
BuildMI(*MBB, IP, PPC::LI, 1, R).addSImm(sval);
} else {
unsigned Temp = makeAnotherReg(Type::IntTy);
BuildMI(*MBB, IP, PPC::LIS, 1, Temp).addSImm(sval >> 16);
BuildMI(*MBB, IP, PPC::ORI, 2, R).addReg(Temp).addImm(sval);
}
return;
}
std::cerr << "Unhandled integer constant!\n";
abort();
} else if (ConstantFP *CFP = dyn_cast<ConstantFP>(C)) {
// We need to spill the constant to memory...
MachineConstantPool *CP = F->getConstantPool();
unsigned CPI = CP->getConstantPoolIndex(CFP);
const Type *Ty = CFP->getType();
unsigned LoadOpcode = (Ty == Type::FloatTy) ? PPC::LFS : PPC::LFD;
BuildMI(*MBB,IP,LoadOpcode,2,R).addConstantPoolIndex(CPI).addReg(PPC::R2);
} else if (isa<ConstantPointerNull>(C)) {
// Copy zero (null pointer) to the register.
BuildMI(*MBB, IP, PPC::LI, 1, R).addSImm(0);
} else if (GlobalValue *GV = dyn_cast<GlobalValue>(C)) {
static unsigned OpcodeTable[] = {
PPC::LBZ, PPC::LHZ, PPC::LWZ, PPC::LFS, PPC::LFD, PPC::LD
};
unsigned Opcode = OpcodeTable[getClassB(GV->getType())];
BuildMI(*MBB, IP, Opcode, 2, R).addGlobalAddress(GV).addReg(PPC::R2);
} else {
std::cerr << "Offending constant: " << *C << "\n";
assert(0 && "Type not handled yet!");
}
}
/// LoadArgumentsToVirtualRegs - Load all of the arguments to this function from
/// the stack into virtual registers.
void PPC64ISel::LoadArgumentsToVirtualRegs(Function &Fn) {
unsigned ArgOffset = ParameterSaveAreaOffset;
unsigned GPR_remaining = 8;
unsigned FPR_remaining = 13;
unsigned GPR_idx = 0, FPR_idx = 0;
static const unsigned GPR[] = {
PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
static const unsigned 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
};
MachineFrameInfo *MFI = F->getFrameInfo();
for (Function::aiterator I = Fn.abegin(), E = Fn.aend(); I != E; ++I) {
bool ArgLive = !I->use_empty();
unsigned Reg = ArgLive ? getReg(*I) : 0;
int FI; // Frame object index
switch (getClassB(I->getType())) {
case cByte:
if (ArgLive) {
FI = MFI->CreateFixedObject(4, ArgOffset);
if (GPR_remaining > 0) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, GPR[GPR_idx]);
BuildMI(BB, PPC::OR, 2, Reg).addReg(GPR[GPR_idx])
.addReg(GPR[GPR_idx]);
} else {
addFrameReference(BuildMI(BB, PPC::LBZ, 2, Reg), FI);
}
}
break;
case cShort:
if (ArgLive) {
FI = MFI->CreateFixedObject(4, ArgOffset);
if (GPR_remaining > 0) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, GPR[GPR_idx]);
BuildMI(BB, PPC::OR, 2, Reg).addReg(GPR[GPR_idx])
.addReg(GPR[GPR_idx]);
} else {
addFrameReference(BuildMI(BB, PPC::LHZ, 2, Reg), FI);
}
}
break;
case cInt:
if (ArgLive) {
FI = MFI->CreateFixedObject(4, ArgOffset);
if (GPR_remaining > 0) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, GPR[GPR_idx]);
BuildMI(BB, PPC::OR, 2, Reg).addReg(GPR[GPR_idx])
.addReg(GPR[GPR_idx]);
} else {
addFrameReference(BuildMI(BB, PPC::LWZ, 2, Reg), FI);
}
}
break;
case cLong:
if (ArgLive) {
FI = MFI->CreateFixedObject(8, ArgOffset);
if (GPR_remaining > 1) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, GPR[GPR_idx]);
BuildMI(BB, PPC::OR, 2, Reg).addReg(GPR[GPR_idx])
.addReg(GPR[GPR_idx]);
} else {
addFrameReference(BuildMI(BB, PPC::LD, 2, Reg), FI);
}
}
// longs require 4 additional bytes
ArgOffset += 4;
break;
case cFP32:
if (ArgLive) {
FI = MFI->CreateFixedObject(4, ArgOffset);
if (FPR_remaining > 0) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, FPR[FPR_idx]);
BuildMI(BB, PPC::FMR, 1, Reg).addReg(FPR[FPR_idx]);
FPR_remaining--;
FPR_idx++;
} else {
addFrameReference(BuildMI(BB, PPC::LFS, 2, Reg), FI);
}
}
break;
case cFP64:
if (ArgLive) {
FI = MFI->CreateFixedObject(8, ArgOffset);
if (FPR_remaining > 0) {
BuildMI(BB, PPC::IMPLICIT_DEF, 0, FPR[FPR_idx]);
BuildMI(BB, PPC::FMR, 1, Reg).addReg(FPR[FPR_idx]);
FPR_remaining--;
FPR_idx++;
} else {
addFrameReference(BuildMI(BB, PPC::LFD, 2, Reg), FI);
}
}
// doubles require 4 additional bytes and use 2 GPRs of param space
ArgOffset += 4;
if (GPR_remaining > 0) {
GPR_remaining--;
GPR_idx++;
}
break;
default:
assert(0 && "Unhandled argument type!");
}
ArgOffset += 4; // Each argument takes at least 4 bytes on the stack...
if (GPR_remaining > 0) {
GPR_remaining--; // uses up 2 GPRs
GPR_idx++;
}
}
// If the function takes variable number of arguments, add a frame offset for
// the start of the first vararg value... this is used to expand
// llvm.va_start.
if (Fn.getFunctionType()->isVarArg())
VarArgsFrameIndex = MFI->CreateFixedObject(4, ArgOffset);
}
/// SelectPHINodes - Insert machine code to generate phis. This is tricky
/// because we have to generate our sources into the source basic blocks, not
/// the current one.
///
void PPC64ISel::SelectPHINodes() {
const TargetInstrInfo &TII = *TM.getInstrInfo();
const Function &LF = *F->getFunction(); // The LLVM function...
for (Function::const_iterator I = LF.begin(), E = LF.end(); I != E; ++I) {
const BasicBlock *BB = I;
MachineBasicBlock &MBB = *MBBMap[I];
// Loop over all of the PHI nodes in the LLVM basic block...
MachineBasicBlock::iterator PHIInsertPoint = MBB.begin();
for (BasicBlock::const_iterator I = BB->begin();
PHINode *PN = const_cast<PHINode*>(dyn_cast<PHINode>(I)); ++I) {
// Create a new machine instr PHI node, and insert it.
unsigned PHIReg = getReg(*PN);
MachineInstr *PhiMI = BuildMI(MBB, PHIInsertPoint,
PPC::PHI, PN->getNumOperands(), PHIReg);
// PHIValues - Map of blocks to incoming virtual registers. We use this
// so that we only initialize one incoming value for a particular block,
// even if the block has multiple entries in the PHI node.
//
std::map<MachineBasicBlock*, unsigned> PHIValues;
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
MachineBasicBlock *PredMBB = 0;
for (MachineBasicBlock::pred_iterator PI = MBB.pred_begin (),
PE = MBB.pred_end (); PI != PE; ++PI)
if (PN->getIncomingBlock(i) == (*PI)->getBasicBlock()) {
PredMBB = *PI;
break;
}
assert (PredMBB && "Couldn't find incoming machine-cfg edge for phi");
unsigned ValReg;
std::map<MachineBasicBlock*, unsigned>::iterator EntryIt =
PHIValues.lower_bound(PredMBB);
if (EntryIt != PHIValues.end() && EntryIt->first == PredMBB) {
// We already inserted an initialization of the register for this
// predecessor. Recycle it.
ValReg = EntryIt->second;
} else {
// Get the incoming value into a virtual register.
//
Value *Val = PN->getIncomingValue(i);
// If this is a constant or GlobalValue, we may have to insert code
// into the basic block to compute it into a virtual register.
if ((isa<Constant>(Val) && !isa<ConstantExpr>(Val)) ||
isa<GlobalValue>(Val)) {
// Simple constants get emitted at the end of the basic block,
// before any terminator instructions. We "know" that the code to
// move a constant into a register will never clobber any flags.
ValReg = getReg(Val, PredMBB, PredMBB->getFirstTerminator());
} else {
// Because we don't want to clobber any values which might be in
// physical registers with the computation of this constant (which
// might be arbitrarily complex if it is a constant expression),
// just insert the computation at the top of the basic block.
MachineBasicBlock::iterator PI = PredMBB->begin();
// Skip over any PHI nodes though!
while (PI != PredMBB->end() && PI->getOpcode() == PPC::PHI)
++PI;
ValReg = getReg(Val, PredMBB, PI);
}
// Remember that we inserted a value for this PHI for this predecessor
PHIValues.insert(EntryIt, std::make_pair(PredMBB, ValReg));
}
PhiMI->addRegOperand(ValReg);
PhiMI->addMachineBasicBlockOperand(PredMBB);
}
// Now that we emitted all of the incoming values for the PHI node, make
// sure to reposition the InsertPoint after the PHI that we just added.
// This is needed because we might have inserted a constant into this
// block, right after the PHI's which is before the old insert point!
PHIInsertPoint = PhiMI;
++PHIInsertPoint;
}
}
}
// canFoldSetCCIntoBranchOrSelect - Return the setcc instruction if we can fold
// it into the conditional branch or select instruction which is the only user
// of the cc instruction. This is the case if the conditional branch is the
// only user of the setcc, and if the setcc is in the same basic block as the
// conditional branch.
//
static SetCondInst *canFoldSetCCIntoBranchOrSelect(Value *V) {
if (SetCondInst *SCI = dyn_cast<SetCondInst>(V))
if (SCI->hasOneUse()) {
Instruction *User = cast<Instruction>(SCI->use_back());
if ((isa<BranchInst>(User) || isa<SelectInst>(User)) &&
SCI->getParent() == User->getParent())
return SCI;
}
return 0;
}
// canFoldGEPIntoLoadOrStore - Return the GEP instruction if we can fold it into
// the load or store instruction that is the only user of the GEP.
//
static GetElementPtrInst *canFoldGEPIntoLoadOrStore(Value *V) {
if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(V))
if (GEPI->hasOneUse()) {
Instruction *User = cast<Instruction>(GEPI->use_back());
if (isa<StoreInst>(User) &&
GEPI->getParent() == User->getParent() &&
User->getOperand(0) != GEPI &&
User->getOperand(1) == GEPI) {
++GEPFolds;
return GEPI;
}
if (isa<LoadInst>(User) &&
GEPI->getParent() == User->getParent() &&
User->getOperand(0) == GEPI) {
++GEPFolds;
return GEPI;
}
}
return 0;
}
// Return a fixed numbering for setcc instructions which does not depend on the
// order of the opcodes.
//
static unsigned getSetCCNumber(unsigned Opcode) {
switch (Opcode) {
default: assert(0 && "Unknown setcc instruction!");
case Instruction::SetEQ: return 0;
case Instruction::SetNE: return 1;
case Instruction::SetLT: return 2;
case Instruction::SetGE: return 3;
case Instruction::SetGT: return 4;
case Instruction::SetLE: return 5;
}
}
static unsigned getPPCOpcodeForSetCCNumber(unsigned Opcode) {
switch (Opcode) {
default: assert(0 && "Unknown setcc instruction!");
case Instruction::SetEQ: return PPC::BEQ;
case Instruction::SetNE: return PPC::BNE;
case Instruction::SetLT: return PPC::BLT;
case Instruction::SetGE: return PPC::BGE;
case Instruction::SetGT: return PPC::BGT;
case Instruction::SetLE: return PPC::BLE;
}
}
/// emitUCOM - emits an unordered FP compare.
void PPC64ISel::emitUCOM(MachineBasicBlock *MBB, MachineBasicBlock::iterator IP,
unsigned LHS, unsigned RHS) {
BuildMI(*MBB, IP, PPC::FCMPU, 2, PPC::CR0).addReg(LHS).addReg(RHS);
}
/// EmitComparison - emits a comparison of the two operands, returning the
/// extended setcc code to use. The result is in CR0.
///
unsigned PPC64ISel::EmitComparison(unsigned OpNum, Value *Op0, Value *Op1,
MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP) {
// The arguments are already supposed to be of the same type.
const Type *CompTy = Op0->getType();
unsigned Class = getClassB(CompTy);
unsigned Op0r = getReg(Op0, MBB, IP);
// Before we do a comparison, we have to make sure that we're truncating our
// registers appropriately.
if (Class == cByte) {
unsigned TmpReg = makeAnotherReg(CompTy);
if (CompTy->isSigned())
BuildMI(*MBB, IP, PPC::EXTSB, 1, TmpReg).addReg(Op0r);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, TmpReg).addReg(Op0r).addImm(0)
.addImm(24).addImm(31);
Op0r = TmpReg;
} else if (Class == cShort) {
unsigned TmpReg = makeAnotherReg(CompTy);
if (CompTy->isSigned())
BuildMI(*MBB, IP, PPC::EXTSH, 1, TmpReg).addReg(Op0r);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, TmpReg).addReg(Op0r).addImm(0)
.addImm(16).addImm(31);
Op0r = TmpReg;
}
// Use crand for lt, gt and crandc for le, ge
unsigned CROpcode = (OpNum == 2 || OpNum == 4) ? PPC::CRAND : PPC::CRANDC;
unsigned Opcode = CompTy->isSigned() ? PPC::CMPW : PPC::CMPLW;
unsigned OpcodeImm = CompTy->isSigned() ? PPC::CMPWI : PPC::CMPLWI;
if (Class == cLong) {
Opcode = CompTy->isSigned() ? PPC::CMPD : PPC::CMPLD;
OpcodeImm = CompTy->isSigned() ? PPC::CMPDI : PPC::CMPLDI;
}
// Special case handling of: cmp R, i
if (ConstantInt *CI = dyn_cast<ConstantInt>(Op1)) {
unsigned Op1v = CI->getRawValue() & 0xFFFF;
// Treat compare like ADDI for the purposes of immediate suitability
if (canUseAsImmediateForOpcode(CI, 0)) {
BuildMI(*MBB, IP, OpcodeImm, 2, PPC::CR0).addReg(Op0r).addSImm(Op1v);
} else {
unsigned Op1r = getReg(Op1, MBB, IP);
BuildMI(*MBB, IP, Opcode, 2, PPC::CR0).addReg(Op0r).addReg(Op1r);
}
return OpNum;
}
unsigned Op1r = getReg(Op1, MBB, IP);
switch (Class) {
default: assert(0 && "Unknown type class!");
case cByte:
case cShort:
case cInt:
case cLong:
BuildMI(*MBB, IP, Opcode, 2, PPC::CR0).addReg(Op0r).addReg(Op1r);
break;
case cFP32:
case cFP64:
emitUCOM(MBB, IP, Op0r, Op1r);
break;
}
return OpNum;
}
/// visitSetCondInst - emit code to calculate the condition via
/// EmitComparison(), and possibly store a 0 or 1 to a register as a result
///
void PPC64ISel::visitSetCondInst(SetCondInst &I) {
if (canFoldSetCCIntoBranchOrSelect(&I))
return;
unsigned DestReg = getReg(I);
unsigned OpNum = I.getOpcode();
const Type *Ty = I.getOperand (0)->getType();
EmitComparison(OpNum, I.getOperand(0), I.getOperand(1), BB, BB->end());
unsigned Opcode = getPPCOpcodeForSetCCNumber(OpNum);
MachineBasicBlock *thisMBB = BB;
const BasicBlock *LLVM_BB = BB->getBasicBlock();
ilist<MachineBasicBlock>::iterator It = BB;
++It;
// thisMBB:
// ...
// cmpTY cr0, r1, r2
// bCC copy1MBB
// b copy0MBB
// FIXME: we wouldn't need copy0MBB (we could fold it into thisMBB)
// if we could insert other, non-terminator instructions after the
// bCC. But MBB->getFirstTerminator() can't understand this.
MachineBasicBlock *copy1MBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, copy1MBB);
BuildMI(BB, Opcode, 2).addReg(PPC::CR0).addMBB(copy1MBB);
MachineBasicBlock *copy0MBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, copy0MBB);
BuildMI(BB, PPC::B, 1).addMBB(copy0MBB);
MachineBasicBlock *sinkMBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, sinkMBB);
// Update machine-CFG edges
BB->addSuccessor(copy1MBB);
BB->addSuccessor(copy0MBB);
// copy1MBB:
// %TrueValue = li 1
// b sinkMBB
BB = copy1MBB;
unsigned TrueValue = makeAnotherReg(I.getType());
BuildMI(BB, PPC::LI, 1, TrueValue).addSImm(1);
BuildMI(BB, PPC::B, 1).addMBB(sinkMBB);
// Update machine-CFG edges
BB->addSuccessor(sinkMBB);
// copy0MBB:
// %FalseValue = li 0
// fallthrough
BB = copy0MBB;
unsigned FalseValue = makeAnotherReg(I.getType());
BuildMI(BB, PPC::LI, 1, FalseValue).addSImm(0);
// Update machine-CFG edges
BB->addSuccessor(sinkMBB);
// sinkMBB:
// %Result = phi [ %FalseValue, copy0MBB ], [ %TrueValue, copy1MBB ]
// ...
BB = sinkMBB;
BuildMI(BB, PPC::PHI, 4, DestReg).addReg(FalseValue)
.addMBB(copy0MBB).addReg(TrueValue).addMBB(copy1MBB);
}
void PPC64ISel::visitSelectInst(SelectInst &SI) {
unsigned DestReg = getReg(SI);
MachineBasicBlock::iterator MII = BB->end();
emitSelectOperation(BB, MII, SI.getCondition(), SI.getTrueValue(),
SI.getFalseValue(), DestReg);
}
/// emitSelect - Common code shared between visitSelectInst and the constant
/// expression support.
/// FIXME: this is most likely broken in one or more ways. Namely, PowerPC has
/// no select instruction. FSEL only works for comparisons against zero.
void PPC64ISel::emitSelectOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Cond, Value *TrueVal,
Value *FalseVal, unsigned DestReg) {
unsigned SelectClass = getClassB(TrueVal->getType());
unsigned Opcode;
// See if we can fold the setcc into the select instruction, or if we have
// to get the register of the Cond value
if (SetCondInst *SCI = canFoldSetCCIntoBranchOrSelect(Cond)) {
// We successfully folded the setcc into the select instruction.
unsigned OpNum = getSetCCNumber(SCI->getOpcode());
OpNum = EmitComparison(OpNum, SCI->getOperand(0),SCI->getOperand(1),MBB,IP);
Opcode = getPPCOpcodeForSetCCNumber(SCI->getOpcode());
} else {
unsigned CondReg = getReg(Cond, MBB, IP);
BuildMI(*MBB, IP, PPC::CMPI, 2, PPC::CR0).addReg(CondReg).addSImm(0);
Opcode = getPPCOpcodeForSetCCNumber(Instruction::SetNE);
}
// thisMBB:
// ...
// cmpTY cr0, r1, r2
// bCC copy1MBB
// b copy0MBB
MachineBasicBlock *thisMBB = BB;
const BasicBlock *LLVM_BB = BB->getBasicBlock();
ilist<MachineBasicBlock>::iterator It = BB;
++It;
// FIXME: we wouldn't need copy0MBB (we could fold it into thisMBB)
// if we could insert other, non-terminator instructions after the
// bCC. But MBB->getFirstTerminator() can't understand this.
MachineBasicBlock *copy1MBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, copy1MBB);
BuildMI(BB, Opcode, 2).addReg(PPC::CR0).addMBB(copy1MBB);
MachineBasicBlock *copy0MBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, copy0MBB);
BuildMI(BB, PPC::B, 1).addMBB(copy0MBB);
MachineBasicBlock *sinkMBB = new MachineBasicBlock(LLVM_BB);
F->getBasicBlockList().insert(It, sinkMBB);
// Update machine-CFG edges
BB->addSuccessor(copy1MBB);
BB->addSuccessor(copy0MBB);
// copy1MBB:
// %TrueValue = ...
// b sinkMBB
BB = copy1MBB;
unsigned TrueValue = getReg(TrueVal, BB, BB->begin());
BuildMI(BB, PPC::B, 1).addMBB(sinkMBB);
// Update machine-CFG edges
BB->addSuccessor(sinkMBB);
// copy0MBB:
// %FalseValue = ...
// fallthrough
BB = copy0MBB;
unsigned FalseValue = getReg(FalseVal, BB, BB->begin());
// Update machine-CFG edges
BB->addSuccessor(sinkMBB);
// sinkMBB:
// %Result = phi [ %FalseValue, copy0MBB ], [ %TrueValue, copy1MBB ]
// ...
BB = sinkMBB;
BuildMI(BB, PPC::PHI, 4, DestReg).addReg(FalseValue)
.addMBB(copy0MBB).addReg(TrueValue).addMBB(copy1MBB);
return;
}
/// promote32 - Emit instructions to turn a narrow operand into a 32-bit-wide
/// operand, in the specified target register.
///
void PPC64ISel::promote32(unsigned targetReg, const ValueRecord &VR) {
bool isUnsigned = VR.Ty->isUnsigned() || VR.Ty == Type::BoolTy;
Value *Val = VR.Val;
const Type *Ty = VR.Ty;
if (Val) {
if (Constant *C = dyn_cast<Constant>(Val)) {
Val = ConstantExpr::getCast(C, Type::IntTy);
if (isa<ConstantExpr>(Val)) // Could not fold
Val = C;
else
Ty = Type::IntTy; // Folded!
}
// If this is a simple constant, just emit a load directly to avoid the copy
if (ConstantInt *CI = dyn_cast<ConstantInt>(Val)) {
int TheVal = CI->getRawValue() & 0xFFFFFFFF;
if (TheVal < 32768 && TheVal >= -32768) {
BuildMI(BB, PPC::LI, 1, targetReg).addSImm(TheVal);
} else {
unsigned TmpReg = makeAnotherReg(Type::IntTy);
BuildMI(BB, PPC::LIS, 1, TmpReg).addSImm(TheVal >> 16);
BuildMI(BB, PPC::ORI, 2, targetReg).addReg(TmpReg)
.addImm(TheVal & 0xFFFF);
}
return;
}
}
// Make sure we have the register number for this value...
unsigned Reg = Val ? getReg(Val) : VR.Reg;
switch (getClassB(Ty)) {
case cByte:
// Extend value into target register (8->32)
if (isUnsigned)
BuildMI(BB, PPC::RLWINM, 4, targetReg).addReg(Reg).addZImm(0)
.addZImm(24).addZImm(31);
else
BuildMI(BB, PPC::EXTSB, 1, targetReg).addReg(Reg);
break;
case cShort:
// Extend value into target register (16->32)
if (isUnsigned)
BuildMI(BB, PPC::RLWINM, 4, targetReg).addReg(Reg).addZImm(0)
.addZImm(16).addZImm(31);
else
BuildMI(BB, PPC::EXTSH, 1, targetReg).addReg(Reg);
break;
case cInt:
case cLong:
// Move value into target register (32->32)
BuildMI(BB, PPC::OR, 2, targetReg).addReg(Reg).addReg(Reg);
break;
default:
assert(0 && "Unpromotable operand class in promote32");
}
}
/// visitReturnInst - implemented with BLR
///
void PPC64ISel::visitReturnInst(ReturnInst &I) {
// Only do the processing if this is a non-void return
if (I.getNumOperands() > 0) {
Value *RetVal = I.getOperand(0);
switch (getClassB(RetVal->getType())) {
case cByte: // integral return values: extend or move into r3 and return
case cShort:
case cInt:
case cLong:
promote32(PPC::R3, ValueRecord(RetVal));
break;
case cFP32:
case cFP64: { // Floats & Doubles: Return in f1
unsigned RetReg = getReg(RetVal);
BuildMI(BB, PPC::FMR, 1, PPC::F1).addReg(RetReg);
break;
}
default:
visitInstruction(I);
}
}
BuildMI(BB, PPC::BLR, 1).addImm(1);
}
// getBlockAfter - Return the basic block which occurs lexically after the
// specified one.
static inline BasicBlock *getBlockAfter(BasicBlock *BB) {
Function::iterator I = BB; ++I; // Get iterator to next block
return I != BB->getParent()->end() ? &*I : 0;
}
/// visitBranchInst - Handle conditional and unconditional branches here. Note
/// that since code layout is frozen at this point, that if we are trying to
/// jump to a block that is the immediate successor of the current block, we can
/// just make a fall-through (but we don't currently).
///
void PPC64ISel::visitBranchInst(BranchInst &BI) {
// Update machine-CFG edges
BB->addSuccessor(MBBMap[BI.getSuccessor(0)]);
if (BI.isConditional())
BB->addSuccessor(MBBMap[BI.getSuccessor(1)]);
BasicBlock *NextBB = getBlockAfter(BI.getParent()); // BB after current one
if (!BI.isConditional()) { // Unconditional branch?
if (BI.getSuccessor(0) != NextBB)
BuildMI(BB, PPC::B, 1).addMBB(MBBMap[BI.getSuccessor(0)]);
return;
}
// See if we can fold the setcc into the branch itself...
SetCondInst *SCI = canFoldSetCCIntoBranchOrSelect(BI.getCondition());
if (SCI == 0) {
// Nope, cannot fold setcc into this branch. Emit a branch on a condition
// computed some other way...
unsigned condReg = getReg(BI.getCondition());
BuildMI(BB, PPC::CMPLI, 3, PPC::CR0).addImm(0).addReg(condReg)
.addImm(0);
if (BI.getSuccessor(1) == NextBB) {
if (BI.getSuccessor(0) != NextBB)
BuildMI(BB, PPC::COND_BRANCH, 3).addReg(PPC::CR0).addImm(PPC::BNE)
.addMBB(MBBMap[BI.getSuccessor(0)])
.addMBB(MBBMap[BI.getSuccessor(1)]);
} else {
BuildMI(BB, PPC::COND_BRANCH, 3).addReg(PPC::CR0).addImm(PPC::BEQ)
.addMBB(MBBMap[BI.getSuccessor(1)])
.addMBB(MBBMap[BI.getSuccessor(0)]);
if (BI.getSuccessor(0) != NextBB)
BuildMI(BB, PPC::B, 1).addMBB(MBBMap[BI.getSuccessor(0)]);
}
return;
}
unsigned OpNum = getSetCCNumber(SCI->getOpcode());
unsigned Opcode = getPPCOpcodeForSetCCNumber(SCI->getOpcode());
MachineBasicBlock::iterator MII = BB->end();
OpNum = EmitComparison(OpNum, SCI->getOperand(0), SCI->getOperand(1), BB,MII);
if (BI.getSuccessor(0) != NextBB) {
BuildMI(BB, PPC::COND_BRANCH, 3).addReg(PPC::CR0).addImm(Opcode)
.addMBB(MBBMap[BI.getSuccessor(0)])
.addMBB(MBBMap[BI.getSuccessor(1)]);
if (BI.getSuccessor(1) != NextBB)
BuildMI(BB, PPC::B, 1).addMBB(MBBMap[BI.getSuccessor(1)]);
} else {
// Change to the inverse condition...
if (BI.getSuccessor(1) != NextBB) {
Opcode = PPC64InstrInfo::invertPPCBranchOpcode(Opcode);
BuildMI(BB, PPC::COND_BRANCH, 3).addReg(PPC::CR0).addImm(Opcode)
.addMBB(MBBMap[BI.getSuccessor(1)])
.addMBB(MBBMap[BI.getSuccessor(0)]);
}
}
}
/// doCall - This emits an abstract call instruction, setting up the arguments
/// and the return value as appropriate. For the actual function call itself,
/// it inserts the specified CallMI instruction into the stream.
///
void PPC64ISel::doCall(const ValueRecord &Ret, MachineInstr *CallMI,
const std::vector<ValueRecord> &Args, bool isVarArg) {
// Count how many bytes are to be pushed on the stack, including the linkage
// area, and parameter passing area.
unsigned NumBytes = ParameterSaveAreaOffset;
unsigned ArgOffset = ParameterSaveAreaOffset;
if (!Args.empty()) {
for (unsigned i = 0, e = Args.size(); i != e; ++i)
switch (getClassB(Args[i].Ty)) {
case cByte: case cShort: case cInt:
NumBytes += 4; break;
case cLong:
NumBytes += 8; break;
case cFP32:
NumBytes += 4; break;
case cFP64:
NumBytes += 8; break;
break;
default: assert(0 && "Unknown class!");
}
// Just to be safe, we'll always reserve the full argument passing space in
// case any called code gets funky on us.
if (NumBytes < ParameterSaveAreaOffset + MaxArgumentStackSpace)
NumBytes = ParameterSaveAreaOffset + MaxArgumentStackSpace;
// Adjust the stack pointer for the new arguments...
// These functions are automatically eliminated by the prolog/epilog pass
BuildMI(BB, PPC::ADJCALLSTACKDOWN, 1).addImm(NumBytes);
// Arguments go on the stack in reverse order, as specified by the ABI.
int GPR_remaining = 8, FPR_remaining = 13;
unsigned GPR_idx = 0, FPR_idx = 0;
static const unsigned GPR[] = {
PPC::R3, PPC::R4, PPC::R5, PPC::R6,
PPC::R7, PPC::R8, PPC::R9, PPC::R10,
};
static const unsigned 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
};
for (unsigned i = 0, e = Args.size(); i != e; ++i) {
unsigned ArgReg;
switch (getClassB(Args[i].Ty)) {
case cByte:
case cShort:
// Promote arg to 32 bits wide into a temporary register...
ArgReg = makeAnotherReg(Type::UIntTy);
promote32(ArgReg, Args[i]);
// Reg or stack?
if (GPR_remaining > 0) {
BuildMI(BB, PPC::OR, 2, GPR[GPR_idx]).addReg(ArgReg)
.addReg(ArgReg);
CallMI->addRegOperand(GPR[GPR_idx], MachineOperand::Use);
}
if (GPR_remaining <= 0 || isVarArg) {
BuildMI(BB, PPC::STW, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
}
break;
case cInt:
ArgReg = Args[i].Val ? getReg(Args[i].Val) : Args[i].Reg;
// Reg or stack?
if (GPR_remaining > 0) {
BuildMI(BB, PPC::OR, 2, GPR[GPR_idx]).addReg(ArgReg)
.addReg(ArgReg);
CallMI->addRegOperand(GPR[GPR_idx], MachineOperand::Use);
}
if (GPR_remaining <= 0 || isVarArg) {
BuildMI(BB, PPC::STW, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
}
break;
case cLong:
ArgReg = Args[i].Val ? getReg(Args[i].Val) : Args[i].Reg;
// Reg or stack?
if (GPR_remaining > 0) {
BuildMI(BB, PPC::OR, 2, GPR[GPR_idx]).addReg(ArgReg)
.addReg(ArgReg);
CallMI->addRegOperand(GPR[GPR_idx], MachineOperand::Use);
}
if (GPR_remaining <= 0 || isVarArg) {
BuildMI(BB, PPC::STD, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
}
ArgOffset += 4; // 8 byte entry, not 4.
break;
case cFP32:
ArgReg = Args[i].Val ? getReg(Args[i].Val) : Args[i].Reg;
// Reg or stack?
if (FPR_remaining > 0) {
BuildMI(BB, PPC::FMR, 1, FPR[FPR_idx]).addReg(ArgReg);
CallMI->addRegOperand(FPR[FPR_idx], MachineOperand::Use);
FPR_remaining--;
FPR_idx++;
// If this is a vararg function, and there are GPRs left, also
// pass the float in an int. Otherwise, put it on the stack.
if (isVarArg) {
BuildMI(BB, PPC::STFS, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
if (GPR_remaining > 0) {
BuildMI(BB, PPC::LWZ, 2, GPR[GPR_idx])
.addSImm(ArgOffset).addReg(ArgReg);
CallMI->addRegOperand(GPR[GPR_idx], MachineOperand::Use);
}
}
} else {
BuildMI(BB, PPC::STFS, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
}
break;
case cFP64:
ArgReg = Args[i].Val ? getReg(Args[i].Val) : Args[i].Reg;
// Reg or stack?
if (FPR_remaining > 0) {
BuildMI(BB, PPC::FMR, 1, FPR[FPR_idx]).addReg(ArgReg);
CallMI->addRegOperand(FPR[FPR_idx], MachineOperand::Use);
FPR_remaining--;
FPR_idx++;
// For vararg functions, must pass doubles via int regs as well
if (isVarArg) {
BuildMI(BB, PPC::STFD, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
if (GPR_remaining > 0) {
BuildMI(BB, PPC::LD, 2, GPR[GPR_idx]).addSImm(ArgOffset)
.addReg(PPC::R1);
CallMI->addRegOperand(GPR[GPR_idx], MachineOperand::Use);
}
}
} else {
BuildMI(BB, PPC::STFD, 3).addReg(ArgReg).addSImm(ArgOffset)
.addReg(PPC::R1);
}
// Doubles use 8 bytes
ArgOffset += 4;
break;
default: assert(0 && "Unknown class!");
}
ArgOffset += 4;
GPR_remaining--;
GPR_idx++;
}
} else {
BuildMI(BB, PPC::ADJCALLSTACKDOWN, 1).addImm(0);
}
BuildMI(BB, PPC::IMPLICIT_DEF, 0, PPC::LR);
BB->push_back(CallMI);
BuildMI(BB, PPC::NOP, 0);
// These functions are automatically eliminated by the prolog/epilog pass
BuildMI(BB, PPC::ADJCALLSTACKUP, 1).addImm(NumBytes);
// If there is a return value, scavenge the result from the location the call
// leaves it in...
//
if (Ret.Ty != Type::VoidTy) {
unsigned DestClass = getClassB(Ret.Ty);
switch (DestClass) {
case cByte:
case cShort:
case cInt:
case cLong:
// Integral results are in r3
BuildMI(BB, PPC::OR, 2, Ret.Reg).addReg(PPC::R3).addReg(PPC::R3);
break;
case cFP32: // Floating-point return values live in f1
case cFP64:
BuildMI(BB, PPC::FMR, 1, Ret.Reg).addReg(PPC::F1);
break;
default: assert(0 && "Unknown class!");
}
}
}
/// visitCallInst - Push args on stack and do a procedure call instruction.
void PPC64ISel::visitCallInst(CallInst &CI) {
MachineInstr *TheCall;
Function *F = CI.getCalledFunction();
if (F) {
// Is it an intrinsic function call?
if (Intrinsic::ID ID = (Intrinsic::ID)F->getIntrinsicID()) {
visitIntrinsicCall(ID, CI); // Special intrinsics are not handled here
return;
}
// Emit a CALL instruction with PC-relative displacement.
TheCall = BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(F, true);
} else { // Emit an indirect call through the CTR
unsigned Reg = getReg(CI.getCalledValue());
BuildMI(BB, PPC::MTCTR, 1).addReg(Reg);
TheCall = BuildMI(PPC::CALLindirect, 2).addZImm(20).addZImm(0);
}
std::vector<ValueRecord> Args;
for (unsigned i = 1, e = CI.getNumOperands(); i != e; ++i)
Args.push_back(ValueRecord(CI.getOperand(i)));
unsigned DestReg = CI.getType() != Type::VoidTy ? getReg(CI) : 0;
bool isVarArg = F ? F->getFunctionType()->isVarArg() : true;
doCall(ValueRecord(DestReg, CI.getType()), TheCall, Args, isVarArg);
}
/// dyncastIsNan - Return the operand of an isnan operation if this is an isnan.
///
static Value *dyncastIsNan(Value *V) {
if (CallInst *CI = dyn_cast<CallInst>(V))
if (Function *F = CI->getCalledFunction())
if (F->getIntrinsicID() == Intrinsic::isunordered)
return CI->getOperand(1);
return 0;
}
/// isOnlyUsedByUnorderedComparisons - Return true if this value is only used by
/// or's whos operands are all calls to the isnan predicate.
static bool isOnlyUsedByUnorderedComparisons(Value *V) {
assert(dyncastIsNan(V) && "The value isn't an isnan call!");
// Check all uses, which will be or's of isnans if this predicate is true.
for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E;++UI){
Instruction *I = cast<Instruction>(*UI);
if (I->getOpcode() != Instruction::Or) return false;
if (I->getOperand(0) != V && !dyncastIsNan(I->getOperand(0))) return false;
if (I->getOperand(1) != V && !dyncastIsNan(I->getOperand(1))) return false;
}
return true;
}
/// LowerUnknownIntrinsicFunctionCalls - This performs a prepass over the
/// function, lowering any calls to unknown intrinsic functions into the
/// equivalent LLVM code.
///
void PPC64ISel::LowerUnknownIntrinsicFunctionCalls(Function &F) {
for (Function::iterator BB = F.begin(), E = F.end(); BB != E; ++BB)
for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; )
if (CallInst *CI = dyn_cast<CallInst>(I++))
if (Function *F = CI->getCalledFunction())
switch (F->getIntrinsicID()) {
case Intrinsic::not_intrinsic:
case Intrinsic::vastart:
case Intrinsic::vacopy:
case Intrinsic::vaend:
case Intrinsic::returnaddress:
case Intrinsic::frameaddress:
// FIXME: should lower these ourselves
// case Intrinsic::isunordered:
// case Intrinsic::memcpy: -> doCall(). system memcpy almost
// guaranteed to be faster than anything we generate ourselves
// We directly implement these intrinsics
break;
case Intrinsic::readio: {
// On PPC, memory operations are in-order. Lower this intrinsic
// into a volatile load.
Instruction *Before = CI->getPrev();
LoadInst * LI = new LoadInst(CI->getOperand(1), "", true, CI);
CI->replaceAllUsesWith(LI);
BB->getInstList().erase(CI);
break;
}
case Intrinsic::writeio: {
// On PPC, memory operations are in-order. Lower this intrinsic
// into a volatile store.
Instruction *Before = CI->getPrev();
StoreInst *SI = new StoreInst(CI->getOperand(1),
CI->getOperand(2), true, CI);
CI->replaceAllUsesWith(SI);
BB->getInstList().erase(CI);
break;
}
default:
// All other intrinsic calls we must lower.
Instruction *Before = CI->getPrev();
TM.getIntrinsicLowering().LowerIntrinsicCall(CI);
if (Before) { // Move iterator to instruction after call
I = Before; ++I;
} else {
I = BB->begin();
}
}
}
void PPC64ISel::visitIntrinsicCall(Intrinsic::ID ID, CallInst &CI) {
unsigned TmpReg1, TmpReg2, TmpReg3;
switch (ID) {
case Intrinsic::vastart:
// Get the address of the first vararg value...
TmpReg1 = getReg(CI);
addFrameReference(BuildMI(BB, PPC::ADDI, 2, TmpReg1), VarArgsFrameIndex,
0, false);
return;
case Intrinsic::vacopy:
TmpReg1 = getReg(CI);
TmpReg2 = getReg(CI.getOperand(1));
BuildMI(BB, PPC::OR, 2, TmpReg1).addReg(TmpReg2).addReg(TmpReg2);
return;
case Intrinsic::vaend: return;
case Intrinsic::returnaddress:
TmpReg1 = getReg(CI);
if (cast<Constant>(CI.getOperand(1))->isNullValue()) {
MachineFrameInfo *MFI = F->getFrameInfo();
unsigned NumBytes = MFI->getStackSize();
BuildMI(BB, PPC::LWZ, 2, TmpReg1).addSImm(NumBytes+8)
.addReg(PPC::R1);
} else {
// Values other than zero are not implemented yet.
BuildMI(BB, PPC::LI, 1, TmpReg1).addSImm(0);
}
return;
case Intrinsic::frameaddress:
TmpReg1 = getReg(CI);
if (cast<Constant>(CI.getOperand(1))->isNullValue()) {
BuildMI(BB, PPC::OR, 2, TmpReg1).addReg(PPC::R1).addReg(PPC::R1);
} else {
// Values other than zero are not implemented yet.
BuildMI(BB, PPC::LI, 1, TmpReg1).addSImm(0);
}
return;
#if 0
// This may be useful for supporting isunordered
case Intrinsic::isnan:
// If this is only used by 'isunordered' style comparisons, don't emit it.
if (isOnlyUsedByUnorderedComparisons(&CI)) return;
TmpReg1 = getReg(CI.getOperand(1));
emitUCOM(BB, BB->end(), TmpReg1, TmpReg1);
TmpReg2 = makeAnotherReg(Type::IntTy);
BuildMI(BB, PPC::MFCR, TmpReg2);
TmpReg3 = getReg(CI);
BuildMI(BB, PPC::RLWINM, 4, TmpReg3).addReg(TmpReg2).addImm(4).addImm(31).addImm(31);
return;
#endif
default: assert(0 && "Error: unknown intrinsics should have been lowered!");
}
}
/// visitSimpleBinary - Implement simple binary operators for integral types...
/// OperatorClass is one of: 0 for Add, 1 for Sub, 2 for And, 3 for Or, 4 for
/// Xor.
///
void PPC64ISel::visitSimpleBinary(BinaryOperator &B, unsigned OperatorClass) {
unsigned DestReg = getReg(B);
MachineBasicBlock::iterator MI = BB->end();
Value *Op0 = B.getOperand(0), *Op1 = B.getOperand(1);
unsigned Class = getClassB(B.getType());
emitSimpleBinaryOperation(BB, MI, Op0, Op1, OperatorClass, DestReg);
}
/// emitBinaryFPOperation - This method handles emission of floating point
/// Add (0), Sub (1), Mul (2), and Div (3) operations.
void PPC64ISel::emitBinaryFPOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1,
unsigned OperatorClass, unsigned DestReg){
static const unsigned OpcodeTab[][4] = {
{ PPC::FADDS, PPC::FSUBS, PPC::FMULS, PPC::FDIVS }, // Float
{ PPC::FADD, PPC::FSUB, PPC::FMUL, PPC::FDIV }, // Double
};
// Special case: R1 = op <const fp>, R2
if (ConstantFP *Op0C = dyn_cast<ConstantFP>(Op0))
if (Op0C->isExactlyValue(-0.0) && OperatorClass == 1) {
// -0.0 - X === -X
unsigned op1Reg = getReg(Op1, BB, IP);
BuildMI(*BB, IP, PPC::FNEG, 1, DestReg).addReg(op1Reg);
return;
}
unsigned Opcode = OpcodeTab[Op0->getType() == Type::DoubleTy][OperatorClass];
unsigned Op0r = getReg(Op0, BB, IP);
unsigned Op1r = getReg(Op1, BB, IP);
BuildMI(*BB, IP, Opcode, 2, DestReg).addReg(Op0r).addReg(Op1r);
}
/// emitSimpleBinaryOperation - Implement simple binary operators for integral
/// types... OperatorClass is one of: 0 for Add, 1 for Sub, 2 for And, 3 for
/// Or, 4 for Xor.
///
/// emitSimpleBinaryOperation - Common code shared between visitSimpleBinary
/// and constant expression support.
///
void PPC64ISel::emitSimpleBinaryOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1,
unsigned OperatorClass,
unsigned DestReg) {
unsigned Class = getClassB(Op0->getType());
// Arithmetic and Bitwise operators
static const unsigned OpcodeTab[] = {
PPC::ADD, PPC::SUB, PPC::AND, PPC::OR, PPC::XOR
};
static const unsigned ImmOpcodeTab[] = {
PPC::ADDI, PPC::SUBI, PPC::ANDIo, PPC::ORI, PPC::XORI
};
static const unsigned RImmOpcodeTab[] = {
PPC::ADDI, PPC::SUBFIC, PPC::ANDIo, PPC::ORI, PPC::XORI
};
if (Class == cFP32 || Class == cFP64) {
assert(OperatorClass < 2 && "No logical ops for FP!");
emitBinaryFPOperation(MBB, IP, Op0, Op1, OperatorClass, DestReg);
return;
}
if (Op0->getType() == Type::BoolTy) {
if (OperatorClass == 3)
// If this is an or of two isnan's, emit an FP comparison directly instead
// of or'ing two isnan's together.
if (Value *LHS = dyncastIsNan(Op0))
if (Value *RHS = dyncastIsNan(Op1)) {
unsigned Op0Reg = getReg(RHS, MBB, IP), Op1Reg = getReg(LHS, MBB, IP);
unsigned TmpReg = makeAnotherReg(Type::IntTy);
emitUCOM(MBB, IP, Op0Reg, Op1Reg);
BuildMI(*MBB, IP, PPC::MFCR, TmpReg);
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(TmpReg).addImm(4)
.addImm(31).addImm(31);
return;
}
}
// Special case: op <const int>, Reg
if (ConstantInt *CI = dyn_cast<ConstantInt>(Op0)) {
// sub 0, X -> subfic
if (OperatorClass == 1 && canUseAsImmediateForOpcode(CI, 0)) {
unsigned Op1r = getReg(Op1, MBB, IP);
int imm = CI->getRawValue() & 0xFFFF;
BuildMI(*MBB, IP, PPC::SUBFIC, 2, DestReg).addReg(Op1r).addSImm(imm);
return;
}
// If it is easy to do, swap the operands and emit an immediate op
if (Class != cLong && OperatorClass != 1 &&
canUseAsImmediateForOpcode(CI, OperatorClass)) {
unsigned Op1r = getReg(Op1, MBB, IP);
int imm = CI->getRawValue() & 0xFFFF;
if (OperatorClass < 2)
BuildMI(*MBB, IP, RImmOpcodeTab[OperatorClass], 2, DestReg).addReg(Op1r)
.addSImm(imm);
else
BuildMI(*MBB, IP, RImmOpcodeTab[OperatorClass], 2, DestReg).addReg(Op1r)
.addZImm(imm);
return;
}
}
// Special case: op Reg, <const int>
if (ConstantInt *Op1C = dyn_cast<ConstantInt>(Op1)) {
unsigned Op0r = getReg(Op0, MBB, IP);
// xor X, -1 -> not X
if (OperatorClass == 4 && Op1C->isAllOnesValue()) {
BuildMI(*MBB, IP, PPC::NOR, 2, DestReg).addReg(Op0r).addReg(Op0r);
return;
}
if (canUseAsImmediateForOpcode(Op1C, OperatorClass)) {
int immediate = Op1C->getRawValue() & 0xFFFF;
if (OperatorClass < 2)
BuildMI(*MBB, IP, ImmOpcodeTab[OperatorClass], 2,DestReg).addReg(Op0r)
.addSImm(immediate);
else
BuildMI(*MBB, IP, ImmOpcodeTab[OperatorClass], 2,DestReg).addReg(Op0r)
.addZImm(immediate);
} else {
unsigned Op1r = getReg(Op1, MBB, IP);
BuildMI(*MBB, IP, OpcodeTab[OperatorClass], 2, DestReg).addReg(Op0r)
.addReg(Op1r);
}
return;
}
// We couldn't generate an immediate variant of the op, load both halves into
// registers and emit the appropriate opcode.
unsigned Op0r = getReg(Op0, MBB, IP);
unsigned Op1r = getReg(Op1, MBB, IP);
unsigned Opcode = OpcodeTab[OperatorClass];
BuildMI(*MBB, IP, Opcode, 2, DestReg).addReg(Op0r).addReg(Op1r);
}
// ExactLog2 - This function solves for (Val == 1 << (N-1)) and returns N. It
// returns zero when the input is not exactly a power of two.
static unsigned ExactLog2(unsigned Val) {
if (Val == 0 || (Val & (Val-1))) return 0;
unsigned Count = 0;
while (Val != 1) {
Val >>= 1;
++Count;
}
return Count;
}
/// doMultiply - Emit appropriate instructions to multiply together the
/// Values Op0 and Op1, and put the result in DestReg.
///
void PPC64ISel::doMultiply(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
unsigned DestReg, Value *Op0, Value *Op1) {
unsigned Class0 = getClass(Op0->getType());
unsigned Class1 = getClass(Op1->getType());
unsigned Op0r = getReg(Op0, MBB, IP);
unsigned Op1r = getReg(Op1, MBB, IP);
// 64 x 64 -> 64
if (Class0 == cLong && Class1 == cLong) {
BuildMI(*MBB, IP, PPC::MULLD, 2, DestReg).addReg(Op0r).addReg(Op1r);
return;
}
// 64 x 32 or less, promote 32 to 64 and do a 64 x 64
if (Class0 == cLong && Class1 <= cInt) {
// FIXME: CLEAR or SIGN EXTEND Op1
BuildMI(*MBB, IP, PPC::MULLD, 2, DestReg).addReg(Op0r).addReg(Op1r);
return;
}
// 32 x 32 -> 32
if (Class0 <= cInt && Class1 <= cInt) {
BuildMI(*MBB, IP, PPC::MULLW, 2, DestReg).addReg(Op0r).addReg(Op1r);
return;
}
assert(0 && "doMultiply cannot operate on unknown type!");
}
/// doMultiplyConst - This method will multiply the value in Op0 by the
/// value of the ContantInt *CI
void PPC64ISel::doMultiplyConst(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
unsigned DestReg, Value *Op0, ConstantInt *CI) {
unsigned Class = getClass(Op0->getType());
// Mul op0, 0 ==> 0
if (CI->isNullValue()) {
BuildMI(*MBB, IP, PPC::LI, 1, DestReg).addSImm(0);
return;
}
// Mul op0, 1 ==> op0
if (CI->equalsInt(1)) {
unsigned Op0r = getReg(Op0, MBB, IP);
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(Op0r).addReg(Op0r);
return;
}
// If the element size is exactly a power of 2, use a shift to get it.
if (unsigned Shift = ExactLog2(CI->getRawValue())) {
ConstantUInt *ShiftCI = ConstantUInt::get(Type::UByteTy, Shift);
emitShiftOperation(MBB, IP, Op0, ShiftCI, true, Op0->getType(), DestReg);
return;
}
// If 32 bits or less and immediate is in right range, emit mul by immediate
if (Class == cByte || Class == cShort || Class == cInt) {
if (canUseAsImmediateForOpcode(CI, 0)) {
unsigned Op0r = getReg(Op0, MBB, IP);
unsigned imm = CI->getRawValue() & 0xFFFF;
BuildMI(*MBB, IP, PPC::MULLI, 2, DestReg).addReg(Op0r).addSImm(imm);
return;
}
}
doMultiply(MBB, IP, DestReg, Op0, CI);
}
void PPC64ISel::visitMul(BinaryOperator &I) {
unsigned ResultReg = getReg(I);
Value *Op0 = I.getOperand(0);
Value *Op1 = I.getOperand(1);
MachineBasicBlock::iterator IP = BB->end();
emitMultiply(BB, IP, Op0, Op1, ResultReg);
}
void PPC64ISel::emitMultiply(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1, unsigned DestReg) {
TypeClass Class = getClass(Op0->getType());
switch (Class) {
case cByte:
case cShort:
case cInt:
case cLong:
if (ConstantInt *CI = dyn_cast<ConstantInt>(Op1)) {
doMultiplyConst(MBB, IP, DestReg, Op0, CI);
} else {
doMultiply(MBB, IP, DestReg, Op0, Op1);
}
return;
case cFP32:
case cFP64:
emitBinaryFPOperation(MBB, IP, Op0, Op1, 2, DestReg);
return;
break;
}
}
/// visitDivRem - Handle division and remainder instructions... these
/// instruction both require the same instructions to be generated, they just
/// select the result from a different register. Note that both of these
/// instructions work differently for signed and unsigned operands.
///
void PPC64ISel::visitDivRem(BinaryOperator &I) {
unsigned ResultReg = getReg(I);
Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
MachineBasicBlock::iterator IP = BB->end();
emitDivRemOperation(BB, IP, Op0, Op1, I.getOpcode() == Instruction::Div,
ResultReg);
}
void PPC64ISel::emitDivRemOperation(MachineBasicBlock *BB,
MachineBasicBlock::iterator IP,
Value *Op0, Value *Op1, bool isDiv,
unsigned ResultReg) {
const Type *Ty = Op0->getType();
unsigned Class = getClass(Ty);
switch (Class) {
case cFP32:
if (isDiv) {
// Floating point divide...
emitBinaryFPOperation(BB, IP, Op0, Op1, 3, ResultReg);
return;
} else {
// Floating point remainder via fmodf(float x, float y);
unsigned Op0Reg = getReg(Op0, BB, IP);
unsigned Op1Reg = getReg(Op1, BB, IP);
MachineInstr *TheCall =
BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(fmodfFn, true);
std::vector<ValueRecord> Args;
Args.push_back(ValueRecord(Op0Reg, Type::FloatTy));
Args.push_back(ValueRecord(Op1Reg, Type::FloatTy));
doCall(ValueRecord(ResultReg, Type::FloatTy), TheCall, Args, false);
}
return;
case cFP64:
if (isDiv) {
// Floating point divide...
emitBinaryFPOperation(BB, IP, Op0, Op1, 3, ResultReg);
return;
} else {
// Floating point remainder via fmod(double x, double y);
unsigned Op0Reg = getReg(Op0, BB, IP);
unsigned Op1Reg = getReg(Op1, BB, IP);
MachineInstr *TheCall =
BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(fmodFn, true);
std::vector<ValueRecord> Args;
Args.push_back(ValueRecord(Op0Reg, Type::DoubleTy));
Args.push_back(ValueRecord(Op1Reg, Type::DoubleTy));
doCall(ValueRecord(ResultReg, Type::DoubleTy), TheCall, Args, false);
}
return;
case cLong: case cByte: case cShort: case cInt:
break; // Small integrals, handled below...
default: assert(0 && "Unknown class!");
}
// Special case signed division by power of 2.
if (isDiv)
if (ConstantSInt *CI = dyn_cast<ConstantSInt>(Op1)) {
assert(Class != cLong && "This doesn't handle 64-bit divides!");
int V = CI->getValue();
if (V == 1) { // X /s 1 => X
unsigned Op0Reg = getReg(Op0, BB, IP);
BuildMI(*BB, IP, PPC::OR, 2, ResultReg).addReg(Op0Reg).addReg(Op0Reg);
return;
}
if (V == -1) { // X /s -1 => -X
unsigned Op0Reg = getReg(Op0, BB, IP);
BuildMI(*BB, IP, PPC::NEG, 1, ResultReg).addReg(Op0Reg);
return;
}
unsigned log2V = ExactLog2(V);
if (log2V != 0 && Ty->isSigned()) {
unsigned Op0Reg = getReg(Op0, BB, IP);
unsigned TmpReg = makeAnotherReg(Op0->getType());
unsigned Opcode = Class == cLong ? PPC::SRADI : PPC::SRAWI;
BuildMI(*BB, IP, Opcode, 2, TmpReg).addReg(Op0Reg).addImm(log2V);
BuildMI(*BB, IP, PPC::ADDZE, 1, ResultReg).addReg(TmpReg);
return;
}
}
static const unsigned DivOpcodes[] =
{ PPC::DIVWU, PPC::DIVW, PPC::DIVDU, PPC::DIVD };
unsigned Op0Reg = getReg(Op0, BB, IP);
unsigned Op1Reg = getReg(Op1, BB, IP);
unsigned Opcode = DivOpcodes[2*(Class == cLong) + Ty->isSigned()];
if (isDiv) {
BuildMI(*BB, IP, Opcode, 2, ResultReg).addReg(Op0Reg).addReg(Op1Reg);
} else { // Remainder
unsigned TmpReg1 = makeAnotherReg(Op0->getType());
unsigned TmpReg2 = makeAnotherReg(Op0->getType());
unsigned MulOpcode = Class == cLong ? PPC::MULLD : PPC::MULLW;
BuildMI(*BB, IP, Opcode, 2, TmpReg1).addReg(Op0Reg).addReg(Op1Reg);
BuildMI(*BB, IP, MulOpcode, 2, TmpReg2).addReg(TmpReg1).addReg(Op1Reg);
BuildMI(*BB, IP, PPC::SUBF, 2, ResultReg).addReg(TmpReg2).addReg(Op0Reg);
}
}
/// Shift instructions: 'shl', 'sar', 'shr' - Some special cases here
/// for constant immediate shift values, and for constant immediate
/// shift values equal to 1. Even the general case is sort of special,
/// because the shift amount has to be in CL, not just any old register.
///
void PPC64ISel::visitShiftInst(ShiftInst &I) {
MachineBasicBlock::iterator IP = BB->end();
emitShiftOperation(BB, IP, I.getOperand(0), I.getOperand(1),
I.getOpcode() == Instruction::Shl, I.getType(),
getReg(I));
}
/// emitShiftOperation - Common code shared between visitShiftInst and
/// constant expression support.
///
void PPC64ISel::emitShiftOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Op, Value *ShiftAmount,
bool isLeftShift, const Type *ResultTy,
unsigned DestReg) {
unsigned SrcReg = getReg (Op, MBB, IP);
bool isSigned = ResultTy->isSigned ();
unsigned Class = getClass (ResultTy);
// Longs, as usual, are handled specially...
if (Class == cLong) {
// If we have a constant shift, we can generate much more efficient code
// than otherwise...
//
if (ConstantUInt *CUI = dyn_cast<ConstantUInt>(ShiftAmount)) {
unsigned Amount = CUI->getValue();
assert(Amount < 64 && "Invalid immediate shift amount!");
if (isLeftShift) {
BuildMI(*MBB, IP, PPC::RLDICR, 3, DestReg).addReg(SrcReg).addImm(Amount)
.addImm(63-Amount);
} else {
if (isSigned) {
BuildMI(*MBB, IP, PPC::SRADI, 2, DestReg).addReg(SrcReg)
.addImm(Amount);
} else {
BuildMI(*MBB, IP, PPC::RLDICL, 3, DestReg).addReg(SrcReg)
.addImm(64-Amount).addImm(Amount);
}
}
} else {
unsigned ShiftReg = getReg (ShiftAmount, MBB, IP);
if (isLeftShift) {
BuildMI(*MBB, IP, PPC::SLD, 2, DestReg).addReg(SrcReg).addReg(ShiftReg);
} else {
unsigned Opcode = (isSigned) ? PPC::SRAD : PPC::SRD;
BuildMI(*MBB, IP, Opcode, DestReg).addReg(SrcReg).addReg(ShiftReg);
}
}
return;
}
if (ConstantUInt *CUI = dyn_cast<ConstantUInt>(ShiftAmount)) {
// The shift amount is constant, guaranteed to be a ubyte. Get its value.
assert(CUI->getType() == Type::UByteTy && "Shift amount not a ubyte?");
unsigned Amount = CUI->getValue();
if (isLeftShift) {
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(Amount).addImm(0).addImm(31-Amount);
} else {
if (isSigned) {
BuildMI(*MBB, IP, PPC::SRAWI,2,DestReg).addReg(SrcReg).addImm(Amount);
} else {
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(32-Amount).addImm(Amount).addImm(31);
}
}
} else { // The shift amount is non-constant.
unsigned ShiftAmountReg = getReg(ShiftAmount, MBB, IP);
if (isLeftShift) {
BuildMI(*MBB, IP, PPC::SLW, 2, DestReg).addReg(SrcReg)
.addReg(ShiftAmountReg);
} else {
BuildMI(*MBB, IP, isSigned ? PPC::SRAW : PPC::SRW, 2, DestReg)
.addReg(SrcReg).addReg(ShiftAmountReg);
}
}
}
/// visitLoadInst - Implement LLVM load instructions. Pretty straightforward
/// mapping of LLVM classes to PPC load instructions, with the exception of
/// signed byte loads, which need a sign extension following them.
///
void PPC64ISel::visitLoadInst(LoadInst &I) {
// Immediate opcodes, for reg+imm addressing
static const unsigned ImmOpcodes[] = {
PPC::LBZ, PPC::LHZ, PPC::LWZ,
PPC::LFS, PPC::LFD, PPC::LWZ
};
// Indexed opcodes, for reg+reg addressing
static const unsigned IdxOpcodes[] = {
PPC::LBZX, PPC::LHZX, PPC::LWZX,
PPC::LFSX, PPC::LFDX, PPC::LWZX
};
unsigned Class = getClassB(I.getType());
unsigned ImmOpcode = ImmOpcodes[Class];
unsigned IdxOpcode = IdxOpcodes[Class];
unsigned DestReg = getReg(I);
Value *SourceAddr = I.getOperand(0);
if (Class == cShort && I.getType()->isSigned()) ImmOpcode = PPC::LHA;
if (Class == cShort && I.getType()->isSigned()) IdxOpcode = PPC::LHAX;
if (AllocaInst *AI = dyn_castFixedAlloca(SourceAddr)) {
unsigned FI = getFixedSizedAllocaFI(AI);
if (Class == cByte && I.getType()->isSigned()) {
unsigned TmpReg = makeAnotherReg(I.getType());
addFrameReference(BuildMI(BB, ImmOpcode, 2, TmpReg), FI);
BuildMI(BB, PPC::EXTSB, 1, DestReg).addReg(TmpReg);
} else {
addFrameReference(BuildMI(BB, ImmOpcode, 2, DestReg), FI);
}
return;
}
// If this load is the only use of the GEP instruction that is its address,
// then we can fold the GEP directly into the load instruction.
// emitGEPOperation with a second to last arg of 'true' will place the
// base register for the GEP into baseReg, and the constant offset from that
// into offset. If the offset fits in 16 bits, we can emit a reg+imm store
// otherwise, we copy the offset into another reg, and use reg+reg addressing.
if (GetElementPtrInst *GEPI = canFoldGEPIntoLoadOrStore(SourceAddr)) {
unsigned baseReg = getReg(GEPI);
unsigned pendingAdd;
ConstantSInt *offset;
emitGEPOperation(BB, BB->end(), GEPI->getOperand(0), GEPI->op_begin()+1,
GEPI->op_end(), baseReg, true, &offset, &pendingAdd);
if (pendingAdd == 0 && Class != cLong &&
canUseAsImmediateForOpcode(offset, 0)) {
if (Class == cByte && I.getType()->isSigned()) {
unsigned TmpReg = makeAnotherReg(I.getType());
BuildMI(BB, ImmOpcode, 2, TmpReg).addSImm(offset->getValue())
.addReg(baseReg);
BuildMI(BB, PPC::EXTSB, 1, DestReg).addReg(TmpReg);
} else {
BuildMI(BB, ImmOpcode, 2, DestReg).addSImm(offset->getValue())
.addReg(baseReg);
}
return;
}
unsigned indexReg = (pendingAdd != 0) ? pendingAdd : getReg(offset);
if (Class == cByte && I.getType()->isSigned()) {
unsigned TmpReg = makeAnotherReg(I.getType());
BuildMI(BB, IdxOpcode, 2, TmpReg).addReg(indexReg).addReg(baseReg);
BuildMI(BB, PPC::EXTSB, 1, DestReg).addReg(TmpReg);
} else {
BuildMI(BB, IdxOpcode, 2, DestReg).addReg(indexReg).addReg(baseReg);
}
return;
}
// The fallback case, where the load was from a source that could not be
// folded into the load instruction.
unsigned SrcAddrReg = getReg(SourceAddr);
if (Class == cByte && I.getType()->isSigned()) {
unsigned TmpReg = makeAnotherReg(I.getType());
BuildMI(BB, ImmOpcode, 2, TmpReg).addSImm(0).addReg(SrcAddrReg);
BuildMI(BB, PPC::EXTSB, 1, DestReg).addReg(TmpReg);
} else {
BuildMI(BB, ImmOpcode, 2, DestReg).addSImm(0).addReg(SrcAddrReg);
}
}
/// visitStoreInst - Implement LLVM store instructions
///
void PPC64ISel::visitStoreInst(StoreInst &I) {
// Immediate opcodes, for reg+imm addressing
static const unsigned ImmOpcodes[] = {
PPC::STB, PPC::STH, PPC::STW,
PPC::STFS, PPC::STFD, PPC::STW
};
// Indexed opcodes, for reg+reg addressing
static const unsigned IdxOpcodes[] = {
PPC::STBX, PPC::STHX, PPC::STWX,
PPC::STFSX, PPC::STFDX, PPC::STWX
};
Value *SourceAddr = I.getOperand(1);
const Type *ValTy = I.getOperand(0)->getType();
unsigned Class = getClassB(ValTy);
unsigned ImmOpcode = ImmOpcodes[Class];
unsigned IdxOpcode = IdxOpcodes[Class];
unsigned ValReg = getReg(I.getOperand(0));
// If this store is the only use of the GEP instruction that is its address,
// then we can fold the GEP directly into the store instruction.
// emitGEPOperation with a second to last arg of 'true' will place the
// base register for the GEP into baseReg, and the constant offset from that
// into offset. If the offset fits in 16 bits, we can emit a reg+imm store
// otherwise, we copy the offset into another reg, and use reg+reg addressing.
if (GetElementPtrInst *GEPI = canFoldGEPIntoLoadOrStore(SourceAddr)) {
unsigned baseReg = getReg(GEPI);
unsigned pendingAdd;
ConstantSInt *offset;
emitGEPOperation(BB, BB->end(), GEPI->getOperand(0), GEPI->op_begin()+1,
GEPI->op_end(), baseReg, true, &offset, &pendingAdd);
if (0 == pendingAdd && Class != cLong &&
canUseAsImmediateForOpcode(offset, 0)) {
BuildMI(BB, ImmOpcode, 3).addReg(ValReg).addSImm(offset->getValue())
.addReg(baseReg);
return;
}
unsigned indexReg = (pendingAdd != 0) ? pendingAdd : getReg(offset);
BuildMI(BB, IdxOpcode, 3).addReg(ValReg).addReg(indexReg).addReg(baseReg);
return;
}
// If the store address wasn't the only use of a GEP, we fall back to the
// standard path: store the ValReg at the value in AddressReg.
unsigned AddressReg = getReg(I.getOperand(1));
BuildMI(BB, ImmOpcode, 3).addReg(ValReg).addSImm(0).addReg(AddressReg);
}
/// visitCastInst - Here we have various kinds of copying with or without sign
/// extension going on.
///
void PPC64ISel::visitCastInst(CastInst &CI) {
Value *Op = CI.getOperand(0);
unsigned SrcClass = getClassB(Op->getType());
unsigned DestClass = getClassB(CI.getType());
// If this is a cast from a 32-bit integer to a Long type, and the only uses
// of the case are GEP instructions, then the cast does not need to be
// generated explicitly, it will be folded into the GEP.
if (DestClass == cLong && SrcClass == cInt) {
bool AllUsesAreGEPs = true;
for (Value::use_iterator I = CI.use_begin(), E = CI.use_end(); I != E; ++I)
if (!isa<GetElementPtrInst>(*I)) {
AllUsesAreGEPs = false;
break;
}
// No need to codegen this cast if all users are getelementptr instrs...
if (AllUsesAreGEPs) return;
}
unsigned DestReg = getReg(CI);
MachineBasicBlock::iterator MI = BB->end();
emitCastOperation(BB, MI, Op, CI.getType(), DestReg);
}
/// emitCastOperation - Common code shared between visitCastInst and constant
/// expression cast support.
///
void PPC64ISel::emitCastOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Src, const Type *DestTy,
unsigned DestReg) {
const Type *SrcTy = Src->getType();
unsigned SrcClass = getClassB(SrcTy);
unsigned DestClass = getClassB(DestTy);
unsigned SrcReg = getReg(Src, MBB, IP);
// Implement casts to bool by using compare on the operand followed by set if
// not zero on the result.
if (DestTy == Type::BoolTy) {
switch (SrcClass) {
case cByte:
case cShort:
case cInt:
case cLong: {
unsigned TmpReg = makeAnotherReg(Type::IntTy);
BuildMI(*MBB, IP, PPC::ADDIC, 2, TmpReg).addReg(SrcReg).addSImm(-1);
BuildMI(*MBB, IP, PPC::SUBFE, 2, DestReg).addReg(TmpReg).addReg(SrcReg);
break;
}
case cFP32:
case cFP64:
// FSEL perhaps?
std::cerr << "ERROR: Cast fp-to-bool not implemented!\n";
abort();
}
return;
}
// Handle cast of Float -> Double
if (SrcClass == cFP32 && DestClass == cFP64) {
BuildMI(*MBB, IP, PPC::FMR, 1, DestReg).addReg(SrcReg);
return;
}
// Handle cast of Double -> Float
if (SrcClass == cFP64 && DestClass == cFP32) {
BuildMI(*MBB, IP, PPC::FRSP, 1, DestReg).addReg(SrcReg);
return;
}
// Handle casts from integer to floating point now...
if (DestClass == cFP32 || DestClass == cFP64) {
// Spill the integer to memory and reload it from there.
unsigned TmpReg = makeAnotherReg(Type::DoubleTy);
int ValueFrameIdx =
F->getFrameInfo()->CreateStackObject(Type::DoubleTy, TM.getTargetData());
if (SrcClass == cLong) {
if (SrcTy->isSigned()) {
addFrameReference(BuildMI(*MBB, IP, PPC::STD, 3).addReg(SrcReg),
ValueFrameIdx);
addFrameReference(BuildMI(*MBB, IP, PPC::LFD, 2, TmpReg),
ValueFrameIdx);
BuildMI(*MBB, IP, PPC::FCFID, 1, DestReg).addReg(TmpReg);
} else {
unsigned Scale = getReg(ConstantFP::get(Type::DoubleTy, 0x1p32));
unsigned TmpHi = makeAnotherReg(Type::IntTy);
unsigned TmpLo = makeAnotherReg(Type::IntTy);
unsigned FPLow = makeAnotherReg(Type::DoubleTy);
unsigned FPTmpHi = makeAnotherReg(Type::DoubleTy);
unsigned FPTmpLo = makeAnotherReg(Type::DoubleTy);
int OtherFrameIdx = F->getFrameInfo()->CreateStackObject(Type::DoubleTy,
TM.getTargetData());
BuildMI(*MBB, IP, PPC::RLDICL, 3, TmpHi).addReg(SrcReg).addImm(32)
.addImm(32);
BuildMI(*MBB, IP, PPC::RLDICL, 3, TmpLo).addReg(SrcReg).addImm(0)
.addImm(32);
addFrameReference(BuildMI(*MBB, IP, PPC::STD, 3).addReg(TmpHi),
ValueFrameIdx);
addFrameReference(BuildMI(*MBB, IP, PPC::STD, 3).addReg(TmpLo),
OtherFrameIdx);
addFrameReference(BuildMI(*MBB, IP, PPC::LFD, 2, TmpReg),
ValueFrameIdx);
addFrameReference(BuildMI(*MBB, IP, PPC::LFD, 2, FPLow),
OtherFrameIdx);
BuildMI(*MBB, IP, PPC::FCFID, 1, FPTmpHi).addReg(TmpReg);
BuildMI(*MBB, IP, PPC::FCFID, 1, FPTmpLo).addReg(FPLow);
BuildMI(*MBB, IP, PPC::FMADD, 3, DestReg).addReg(Scale).addReg(FPTmpHi)
.addReg(FPTmpLo);
}
return;
}
// FIXME: really want a promote64
unsigned IntTmp = makeAnotherReg(Type::IntTy);
if (SrcTy->isSigned())
BuildMI(*MBB, IP, PPC::EXTSW, 1, IntTmp).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::RLDICL, 3, IntTmp).addReg(SrcReg).addImm(0)
.addImm(32);
addFrameReference(BuildMI(*MBB, IP, PPC::STD, 3).addReg(IntTmp),
ValueFrameIdx);
addFrameReference(BuildMI(*MBB, IP, PPC::LFD, 2, TmpReg),
ValueFrameIdx);
BuildMI(*MBB, IP, PPC::FCFID, 1, DestReg).addReg(TmpReg);
return;
}
// Handle casts from floating point to integer now...
if (SrcClass == cFP32 || SrcClass == cFP64) {
static Function* const Funcs[] =
{ __fixsfdiFn, __fixdfdiFn, __fixunssfdiFn, __fixunsdfdiFn };
// emit library call
if (DestClass == cLong) {
bool isDouble = SrcClass == cFP64;
unsigned nameIndex = 2 * DestTy->isSigned() + isDouble;
std::vector<ValueRecord> Args;
Args.push_back(ValueRecord(SrcReg, SrcTy));
Function *floatFn = Funcs[nameIndex];
MachineInstr *TheCall =
BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(floatFn, true);
doCall(ValueRecord(DestReg, DestTy), TheCall, Args, false);
return;
}
int ValueFrameIdx =
F->getFrameInfo()->CreateStackObject(SrcTy, TM.getTargetData());
if (DestTy->isSigned()) {
unsigned TempReg = makeAnotherReg(Type::DoubleTy);
// Convert to integer in the FP reg and store it to a stack slot
BuildMI(*BB, IP, PPC::FCTIWZ, 1, TempReg).addReg(SrcReg);
addFrameReference(BuildMI(*BB, IP, PPC::STFD, 3)
.addReg(TempReg), ValueFrameIdx);
// There is no load signed byte opcode, so we must emit a sign extend for
// that particular size. Make sure to source the new integer from the
// correct offset.
if (DestClass == cByte) {
unsigned TempReg2 = makeAnotherReg(DestTy);
addFrameReference(BuildMI(*BB, IP, PPC::LBZ, 2, TempReg2),
ValueFrameIdx, 7);
BuildMI(*MBB, IP, PPC::EXTSB, DestReg).addReg(TempReg2);
} else {
int offset = (DestClass == cShort) ? 6 : 4;
unsigned LoadOp = (DestClass == cShort) ? PPC::LHA : PPC::LWZ;
addFrameReference(BuildMI(*BB, IP, LoadOp, 2, DestReg),
ValueFrameIdx, offset);
}
} else {
unsigned Zero = getReg(ConstantFP::get(Type::DoubleTy, 0.0f));
double maxInt = (1LL << 32) - 1;
unsigned MaxInt = getReg(ConstantFP::get(Type::DoubleTy, maxInt));
double border = 1LL << 31;
unsigned Border = getReg(ConstantFP::get(Type::DoubleTy, border));
unsigned UseZero = makeAnotherReg(Type::DoubleTy);
unsigned UseMaxInt = makeAnotherReg(Type::DoubleTy);
unsigned UseChoice = makeAnotherReg(Type::DoubleTy);
unsigned TmpReg = makeAnotherReg(Type::DoubleTy);
unsigned TmpReg2 = makeAnotherReg(Type::DoubleTy);
unsigned ConvReg = makeAnotherReg(Type::DoubleTy);
unsigned IntTmp = makeAnotherReg(Type::IntTy);
unsigned XorReg = makeAnotherReg(Type::IntTy);
int FrameIdx =
F->getFrameInfo()->CreateStackObject(SrcTy, TM.getTargetData());
// Update machine-CFG edges
MachineBasicBlock *XorMBB = new MachineBasicBlock(BB->getBasicBlock());
MachineBasicBlock *PhiMBB = new MachineBasicBlock(BB->getBasicBlock());
MachineBasicBlock *OldMBB = BB;
ilist<MachineBasicBlock>::iterator It = BB; ++It;
F->getBasicBlockList().insert(It, XorMBB);
F->getBasicBlockList().insert(It, PhiMBB);
BB->addSuccessor(XorMBB);
BB->addSuccessor(PhiMBB);
// Convert from floating point to unsigned 32-bit value
// Use 0 if incoming value is < 0.0
BuildMI(*BB, IP, PPC::FSEL, 3, UseZero).addReg(SrcReg).addReg(SrcReg)
.addReg(Zero);
// Use 2**32 - 1 if incoming value is >= 2**32
BuildMI(*BB, IP, PPC::FSUB, 2, UseMaxInt).addReg(MaxInt).addReg(SrcReg);
BuildMI(*BB, IP, PPC::FSEL, 3, UseChoice).addReg(UseMaxInt)
.addReg(UseZero).addReg(MaxInt);
// Subtract 2**31
BuildMI(*BB, IP, PPC::FSUB, 2, TmpReg).addReg(UseChoice).addReg(Border);
// Use difference if >= 2**31
BuildMI(*BB, IP, PPC::FCMPU, 2, PPC::CR0).addReg(UseChoice)
.addReg(Border);
BuildMI(*BB, IP, PPC::FSEL, 3, TmpReg2).addReg(TmpReg).addReg(TmpReg)
.addReg(UseChoice);
// Convert to integer
BuildMI(*BB, IP, PPC::FCTIWZ, 1, ConvReg).addReg(TmpReg2);
addFrameReference(BuildMI(*BB, IP, PPC::STFD, 3).addReg(ConvReg),
FrameIdx);
if (DestClass == cByte) {
addFrameReference(BuildMI(*BB, IP, PPC::LBZ, 2, DestReg),
FrameIdx, 7);
} else if (DestClass == cShort) {
addFrameReference(BuildMI(*BB, IP, PPC::LHZ, 2, DestReg),
FrameIdx, 6);
} if (DestClass == cInt) {
addFrameReference(BuildMI(*BB, IP, PPC::LWZ, 2, IntTmp),
FrameIdx, 4);
BuildMI(*BB, IP, PPC::BLT, 2).addReg(PPC::CR0).addMBB(PhiMBB);
BuildMI(*BB, IP, PPC::B, 1).addMBB(XorMBB);
// XorMBB:
// add 2**31 if input was >= 2**31
BB = XorMBB;
BuildMI(BB, PPC::XORIS, 2, XorReg).addReg(IntTmp).addImm(0x8000);
XorMBB->addSuccessor(PhiMBB);
// PhiMBB:
// DestReg = phi [ IntTmp, OldMBB ], [ XorReg, XorMBB ]
BB = PhiMBB;
BuildMI(BB, PPC::PHI, 4, DestReg).addReg(IntTmp).addMBB(OldMBB)
.addReg(XorReg).addMBB(XorMBB);
}
}
return;
}
// Check our invariants
assert((SrcClass <= cInt || SrcClass == cLong) &&
"Unhandled source class for cast operation!");
assert((DestClass <= cInt || DestClass == cLong) &&
"Unhandled destination class for cast operation!");
bool sourceUnsigned = SrcTy->isUnsigned() || SrcTy == Type::BoolTy;
bool destUnsigned = DestTy->isUnsigned();
// Unsigned -> Unsigned, clear if larger
if (sourceUnsigned && destUnsigned) {
// handle long dest class now to keep switch clean
if (DestClass == cLong) {
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
return;
}
// handle u{ byte, short, int } x u{ byte, short, int }
unsigned clearBits = (SrcClass == cByte || DestClass == cByte) ? 24 : 16;
switch (SrcClass) {
case cByte:
case cShort:
if (SrcClass == DestClass)
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(0).addImm(clearBits).addImm(31);
break;
case cInt:
case cLong:
if (DestClass == cInt)
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(0).addImm(clearBits).addImm(31);
break;
}
return;
}
// Signed -> Signed
if (!sourceUnsigned && !destUnsigned) {
// handle long dest class now to keep switch clean
if (DestClass == cLong) {
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
return;
}
// handle { byte, short, int } x { byte, short, int }
switch (SrcClass) {
case cByte:
if (DestClass == cByte)
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
break;
case cShort:
if (DestClass == cByte)
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
else if (DestClass == cShort)
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::EXTSH, 1, DestReg).addReg(SrcReg);
break;
case cInt:
case cLong:
if (DestClass == cByte)
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
else if (DestClass == cShort)
BuildMI(*MBB, IP, PPC::EXTSH, 1, DestReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
break;
}
return;
}
// Unsigned -> Signed
if (sourceUnsigned && !destUnsigned) {
// handle long dest class now to keep switch clean
if (DestClass == cLong) {
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
return;
}
// handle u{ byte, short, int } -> { byte, short, int }
switch (SrcClass) {
case cByte:
if (DestClass == cByte)
// uByte 255 -> signed byte == -1
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
else
// uByte 255 -> signed short/int == 255
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg).addImm(0)
.addImm(24).addImm(31);
break;
case cShort:
if (DestClass == cByte)
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
else if (DestClass == cShort)
BuildMI(*MBB, IP, PPC::EXTSH, 1, DestReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg).addImm(0)
.addImm(16).addImm(31);
break;
case cInt:
case cLong:
if (DestClass == cByte)
BuildMI(*MBB, IP, PPC::EXTSB, 1, DestReg).addReg(SrcReg);
else if (DestClass == cShort)
BuildMI(*MBB, IP, PPC::EXTSH, 1, DestReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
break;
}
return;
}
// Signed -> Unsigned
if (!sourceUnsigned && destUnsigned) {
// handle long dest class now to keep switch clean
if (DestClass == cLong) {
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
return;
}
// handle { byte, short, int } -> u{ byte, short, int }
unsigned clearBits = (DestClass == cByte) ? 24 : 16;
switch (SrcClass) {
case cByte:
case cShort:
if (DestClass == cByte || DestClass == cShort)
// sbyte -1 -> ubyte 0x000000FF
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(0).addImm(clearBits).addImm(31);
else
// sbyte -1 -> ubyte 0xFFFFFFFF
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
break;
case cInt:
case cLong:
if (DestClass == cInt)
BuildMI(*MBB, IP, PPC::OR, 2, DestReg).addReg(SrcReg).addReg(SrcReg);
else
BuildMI(*MBB, IP, PPC::RLWINM, 4, DestReg).addReg(SrcReg)
.addImm(0).addImm(clearBits).addImm(31);
break;
}
return;
}
// Anything we haven't handled already, we can't (yet) handle at all.
std::cerr << "Unhandled cast from " << SrcTy->getDescription()
<< "to " << DestTy->getDescription() << '\n';
abort();
}
/// visitVANextInst - Implement the va_next instruction...
///
void PPC64ISel::visitVANextInst(VANextInst &I) {
unsigned VAList = getReg(I.getOperand(0));
unsigned DestReg = getReg(I);
unsigned Size;
switch (I.getArgType()->getTypeID()) {
default:
std::cerr << I;
assert(0 && "Error: bad type for va_next instruction!");
return;
case Type::PointerTyID:
case Type::UIntTyID:
case Type::IntTyID:
Size = 4;
break;
case Type::ULongTyID:
case Type::LongTyID:
case Type::DoubleTyID:
Size = 8;
break;
}
// Increment the VAList pointer...
BuildMI(BB, PPC::ADDI, 2, DestReg).addReg(VAList).addSImm(Size);
}
void PPC64ISel::visitVAArgInst(VAArgInst &I) {
unsigned VAList = getReg(I.getOperand(0));
unsigned DestReg = getReg(I);
switch (I.getType()->getTypeID()) {
default:
std::cerr << I;
assert(0 && "Error: bad type for va_next instruction!");
return;
case Type::PointerTyID:
case Type::UIntTyID:
case Type::IntTyID:
BuildMI(BB, PPC::LWZ, 2, DestReg).addSImm(0).addReg(VAList);
break;
case Type::ULongTyID:
case Type::LongTyID:
BuildMI(BB, PPC::LD, 2, DestReg).addSImm(0).addReg(VAList);
break;
case Type::FloatTyID:
BuildMI(BB, PPC::LFS, 2, DestReg).addSImm(0).addReg(VAList);
break;
case Type::DoubleTyID:
BuildMI(BB, PPC::LFD, 2, DestReg).addSImm(0).addReg(VAList);
break;
}
}
/// visitGetElementPtrInst - instruction-select GEP instructions
///
void PPC64ISel::visitGetElementPtrInst(GetElementPtrInst &I) {
if (canFoldGEPIntoLoadOrStore(&I))
return;
unsigned outputReg = getReg(I);
emitGEPOperation(BB, BB->end(), I.getOperand(0), I.op_begin()+1, I.op_end(),
outputReg, false, 0, 0);
}
/// emitGEPOperation - Common code shared between visitGetElementPtrInst and
/// constant expression GEP support.
///
void PPC64ISel::emitGEPOperation(MachineBasicBlock *MBB,
MachineBasicBlock::iterator IP,
Value *Src, User::op_iterator IdxBegin,
User::op_iterator IdxEnd, unsigned TargetReg,
bool GEPIsFolded, ConstantSInt **RemainderPtr,
unsigned *PendingAddReg) {
const TargetData &TD = TM.getTargetData();
const Type *Ty = Src->getType();
unsigned basePtrReg = getReg(Src, MBB, IP);
int64_t constValue = 0;
// Record the operations to emit the GEP in a vector so that we can emit them
// after having analyzed the entire instruction.
std::vector<CollapsedGepOp> ops;
// GEPs have zero or more indices; we must perform a struct access
// or array access for each one.
for (GetElementPtrInst::op_iterator oi = IdxBegin, oe = IdxEnd; oi != oe;
++oi) {
Value *idx = *oi;
if (const StructType *StTy = dyn_cast<StructType>(Ty)) {
// It's a struct access. idx is the index into the structure,
// which names the field. Use the TargetData structure to
// pick out what the layout of the structure is in memory.
// Use the (constant) structure index's value to find the
// right byte offset from the StructLayout class's list of
// structure member offsets.
unsigned fieldIndex = cast<ConstantUInt>(idx)->getValue();
unsigned memberOffset =
TD.getStructLayout(StTy)->MemberOffsets[fieldIndex];
// StructType member offsets are always constant values. Add it to the
// running total.
constValue += memberOffset;
// The next type is the member of the structure selected by the
// index.
Ty = StTy->getElementType (fieldIndex);
} else if (const SequentialType *SqTy = dyn_cast<SequentialType> (Ty)) {
// Many GEP instructions use a [cast (int/uint) to LongTy] as their
// operand. Handle this case directly now...
if (CastInst *CI = dyn_cast<CastInst>(idx))
if (CI->getOperand(0)->getType() == Type::IntTy ||
CI->getOperand(0)->getType() == Type::UIntTy)
idx = CI->getOperand(0);
// It's an array or pointer access: [ArraySize x ElementType].
// We want to add basePtrReg to (idxReg * sizeof ElementType). First, we
// must find the size of the pointed-to type (Not coincidentally, the next
// type is the type of the elements in the array).
Ty = SqTy->getElementType();
unsigned elementSize = TD.getTypeSize(Ty);
if (ConstantInt *C = dyn_cast<ConstantInt>(idx)) {
if (ConstantSInt *CS = dyn_cast<ConstantSInt>(C))
constValue += CS->getValue() * elementSize;
else if (ConstantUInt *CU = dyn_cast<ConstantUInt>(C))
constValue += CU->getValue() * elementSize;
else
assert(0 && "Invalid ConstantInt GEP index type!");
} else {
// Push current gep state to this point as an add
ops.push_back(CollapsedGepOp(false, 0,
ConstantSInt::get(Type::IntTy,constValue)));
// Push multiply gep op and reset constant value
ops.push_back(CollapsedGepOp(true, idx,
ConstantSInt::get(Type::IntTy, elementSize)));
constValue = 0;
}
}
}
// Emit instructions for all the collapsed ops
bool pendingAdd = false;
unsigned pendingAddReg = 0;
for(std::vector<CollapsedGepOp>::iterator cgo_i = ops.begin(),
cgo_e = ops.end(); cgo_i != cgo_e; ++cgo_i) {
CollapsedGepOp& cgo = *cgo_i;
unsigned nextBasePtrReg = makeAnotherReg(Type::IntTy);
// If we didn't emit an add last time through the loop, we need to now so
// that the base reg is updated appropriately.
if (pendingAdd) {
assert(pendingAddReg != 0 && "Uninitialized register in pending add!");
BuildMI(*MBB, IP, PPC::ADD, 2, nextBasePtrReg).addReg(basePtrReg)
.addReg(pendingAddReg);
basePtrReg = nextBasePtrReg;
nextBasePtrReg = makeAnotherReg(Type::IntTy);
pendingAddReg = 0;
pendingAdd = false;
}
if (cgo.isMul) {
// We know the elementSize is a constant, so we can emit a constant mul
unsigned TmpReg = makeAnotherReg(Type::IntTy);
doMultiplyConst(MBB, IP, nextBasePtrReg, cgo.index, cgo.size);
pendingAddReg = basePtrReg;
pendingAdd = true;
} else {
// Try and generate an immediate addition if possible
if (cgo.size->isNullValue()) {
BuildMI(*MBB, IP, PPC::OR, 2, nextBasePtrReg).addReg(basePtrReg)
.addReg(basePtrReg);
} else if (canUseAsImmediateForOpcode(cgo.size, 0)) {
BuildMI(*MBB, IP, PPC::ADDI, 2, nextBasePtrReg).addReg(basePtrReg)
.addSImm(cgo.size->getValue());
} else {
unsigned Op1r = getReg(cgo.size, MBB, IP);
BuildMI(*MBB, IP, PPC::ADD, 2, nextBasePtrReg).addReg(basePtrReg)
.addReg(Op1r);
}
}
basePtrReg = nextBasePtrReg;
}
// Add the current base register plus any accumulated constant value
ConstantSInt *remainder = ConstantSInt::get(Type::IntTy, constValue);
// If we are emitting this during a fold, copy the current base register to
// the target, and save the current constant offset so the folding load or
// store can try and use it as an immediate.
if (GEPIsFolded) {
// If this is a folded GEP and the last element was an index, then we need
// to do some extra work to turn a shift/add/stw into a shift/stwx
if (pendingAdd && 0 == remainder->getValue()) {
assert(pendingAddReg != 0 && "Uninitialized register in pending add!");
*PendingAddReg = pendingAddReg;
} else {
*PendingAddReg = 0;
if (pendingAdd) {
unsigned nextBasePtrReg = makeAnotherReg(Type::IntTy);
assert(pendingAddReg != 0 && "Uninitialized register in pending add!");
BuildMI(*MBB, IP, PPC::ADD, 2, nextBasePtrReg).addReg(basePtrReg)
.addReg(pendingAddReg);
basePtrReg = nextBasePtrReg;
}
}
BuildMI (*MBB, IP, PPC::OR, 2, TargetReg).addReg(basePtrReg)
.addReg(basePtrReg);
*RemainderPtr = remainder;
return;
}
// If we still have a pending add at this point, emit it now
if (pendingAdd) {
unsigned TmpReg = makeAnotherReg(Type::IntTy);
BuildMI(*MBB, IP, PPC::ADD, 2, TmpReg).addReg(pendingAddReg)
.addReg(basePtrReg);
basePtrReg = TmpReg;
}
// After we have processed all the indices, the result is left in
// basePtrReg. Move it to the register where we were expected to
// put the answer.
if (remainder->isNullValue()) {
BuildMI (*MBB, IP, PPC::OR, 2, TargetReg).addReg(basePtrReg)
.addReg(basePtrReg);
} else if (canUseAsImmediateForOpcode(remainder, 0)) {
BuildMI(*MBB, IP, PPC::ADDI, 2, TargetReg).addReg(basePtrReg)
.addSImm(remainder->getValue());
} else {
unsigned Op1r = getReg(remainder, MBB, IP);
BuildMI(*MBB, IP, PPC::ADD, 2, TargetReg).addReg(basePtrReg).addReg(Op1r);
}
}
/// visitAllocaInst - If this is a fixed size alloca, allocate space from the
/// frame manager, otherwise do it the hard way.
///
void PPC64ISel::visitAllocaInst(AllocaInst &I) {
// If this is a fixed size alloca in the entry block for the function, we
// statically stack allocate the space, so we don't need to do anything here.
//
if (dyn_castFixedAlloca(&I)) return;
// Find the data size of the alloca inst's getAllocatedType.
const Type *Ty = I.getAllocatedType();
unsigned TySize = TM.getTargetData().getTypeSize(Ty);
// Create a register to hold the temporary result of multiplying the type size
// constant by the variable amount.
unsigned TotalSizeReg = makeAnotherReg(Type::UIntTy);
// TotalSizeReg = mul <numelements>, <TypeSize>
MachineBasicBlock::iterator MBBI = BB->end();
ConstantUInt *CUI = ConstantUInt::get(Type::UIntTy, TySize);
doMultiplyConst(BB, MBBI, TotalSizeReg, I.getArraySize(), CUI);
// AddedSize = add <TotalSizeReg>, 15
unsigned AddedSizeReg = makeAnotherReg(Type::UIntTy);
BuildMI(BB, PPC::ADDI, 2, AddedSizeReg).addReg(TotalSizeReg).addSImm(15);
// AlignedSize = and <AddedSize>, ~15
unsigned AlignedSize = makeAnotherReg(Type::UIntTy);
BuildMI(BB, PPC::RLWINM, 4, AlignedSize).addReg(AddedSizeReg).addImm(0)
.addImm(0).addImm(27);
// Subtract size from stack pointer, thereby allocating some space.
BuildMI(BB, PPC::SUB, 2, PPC::R1).addReg(PPC::R1).addReg(AlignedSize);
// Put a pointer to the space into the result register, by copying
// the stack pointer.
BuildMI(BB, PPC::OR, 2, getReg(I)).addReg(PPC::R1).addReg(PPC::R1);
// Inform the Frame Information that we have just allocated a variable-sized
// object.
F->getFrameInfo()->CreateVariableSizedObject();
}
/// visitMallocInst - Malloc instructions are code generated into direct calls
/// to the library malloc.
///
void PPC64ISel::visitMallocInst(MallocInst &I) {
unsigned AllocSize = TM.getTargetData().getTypeSize(I.getAllocatedType());
unsigned Arg;
if (ConstantUInt *C = dyn_cast<ConstantUInt>(I.getOperand(0))) {
Arg = getReg(ConstantUInt::get(Type::UIntTy, C->getValue() * AllocSize));
} else {
Arg = makeAnotherReg(Type::UIntTy);
MachineBasicBlock::iterator MBBI = BB->end();
ConstantUInt *CUI = ConstantUInt::get(Type::UIntTy, AllocSize);
doMultiplyConst(BB, MBBI, Arg, I.getOperand(0), CUI);
}
std::vector<ValueRecord> Args;
Args.push_back(ValueRecord(Arg, Type::UIntTy));
MachineInstr *TheCall =
BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(mallocFn, true);
doCall(ValueRecord(getReg(I), I.getType()), TheCall, Args, false);
}
/// visitFreeInst - Free instructions are code gen'd to call the free libc
/// function.
///
void PPC64ISel::visitFreeInst(FreeInst &I) {
std::vector<ValueRecord> Args;
Args.push_back(ValueRecord(I.getOperand(0)));
MachineInstr *TheCall =
BuildMI(PPC::CALLpcrel, 1).addGlobalAddress(freeFn, true);
doCall(ValueRecord(0, Type::VoidTy), TheCall, Args, false);
}
/// createPPC64ISelSimple - This pass converts an LLVM function into a machine
/// code representation is a very simple peep-hole fashion.
///
FunctionPass *llvm::createPPC64ISelSimple(TargetMachine &TM) {
return new PPC64ISel(TM);
}
Reference in New Issue View Git Blame Copy Permalink