SIMachineFunctionInfo::SpilledReg SIMachineFunctionInfo::getSpilledReg(
MachineFunction *MF,
unsigned FrameIndex,
unsigned SubIdx) {
const MachineFrameInfo *FrameInfo = MF->getFrameInfo();
const SIRegisterInfo *TRI = static_cast<const SIRegisterInfo *>(
MF->getSubtarget<AMDGPUSubtarget>().getRegisterInfo());
MachineRegisterInfo &MRI = MF->getRegInfo();
int64_t Offset = FrameInfo->getObjectOffset(FrameIndex);
Offset += SubIdx * 4;
unsigned LaneVGPRIdx = Offset / (64 * 4);
unsigned Lane = (Offset / 4) % 64;
struct SpilledReg Spill;
if (!LaneVGPRs.count(LaneVGPRIdx)) {
unsigned LaneVGPR = TRI->findUnusedRegister(MRI, &AMDGPU::VGPR_32RegClass);
LaneVGPRs[LaneVGPRIdx] = LaneVGPR;
// Add this register as live-in to all blocks to avoid machine verifer
// complaining about use of an undefined physical register.
for (MachineFunction::iterator BI = MF->begin(), BE = MF->end();
BI != BE; ++BI) {
BI->addLiveIn(LaneVGPR);
}
}
Spill.VGPR = LaneVGPRs[LaneVGPRIdx];
Spill.Lane = Lane;
return Spill;
}
// Add newly allocated physical registers to the MBB live in sets.
void RegAllocBase::addMBBLiveIns(MachineFunction *MF) {
NamedRegionTimer T("MBB Live Ins", TimerGroupName, TimePassesIsEnabled);
SlotIndexes *Indexes = LIS->getSlotIndexes();
if (MF->size() <= 1)
return;
LiveIntervalUnion::SegmentIter SI;
for (unsigned PhysReg = 0; PhysReg < PhysReg2LiveUnion.numRegs(); ++PhysReg) {
LiveIntervalUnion &LiveUnion = PhysReg2LiveUnion[PhysReg];
if (LiveUnion.empty())
continue;
MachineFunction::iterator MBB = llvm::next(MF->begin());
MachineFunction::iterator MFE = MF->end();
SlotIndex Start, Stop;
tie(Start, Stop) = Indexes->getMBBRange(MBB);
SI.setMap(LiveUnion.getMap());
SI.find(Start);
while (SI.valid()) {
if (SI.start() <= Start) {
if (!MBB->isLiveIn(PhysReg))
MBB->addLiveIn(PhysReg);
} else if (SI.start() > Stop)
MBB = Indexes->getMBBFromIndex(SI.start().getPrevIndex());
if (++MBB == MFE)
break;
tie(Start, Stop) = Indexes->getMBBRange(MBB);
SI.advanceTo(Start);
}
}
}
void SparcFrameLowering::remapRegsForLeafProc(MachineFunction &MF) const {
MachineRegisterInfo &MRI = MF.getRegInfo();
// Remap %i[0-7] to %o[0-7].
for (unsigned reg = SP::I0; reg <= SP::I7; ++reg) {
if (!MRI.isPhysRegUsed(reg))
continue;
unsigned mapped_reg = (reg - SP::I0 + SP::O0);
assert(!MRI.isPhysRegUsed(mapped_reg));
// Replace I register with O register.
MRI.replaceRegWith(reg, mapped_reg);
// Mark the reg unused.
MRI.setPhysRegUnused(reg);
}
// Rewrite MBB's Live-ins.
for (MachineFunction::iterator MBB = MF.begin(), E = MF.end();
MBB != E; ++MBB) {
for (unsigned reg = SP::I0; reg <= SP::I7; ++reg) {
if (!MBB->isLiveIn(reg))
continue;
MBB->removeLiveIn(reg);
MBB->addLiveIn(reg - SP::I0 + SP::O0);
}
}
assert(verifyLeafProcRegUse(&MRI));
#ifdef XDEBUG
MF.verify(0, "After LeafProc Remapping");
#endif
}
void SystemZRegisterInfo::emitPrologue(MachineFunction &MF) const {
MachineBasicBlock &MBB = MF.front(); // Prolog goes in entry BB
const TargetFrameInfo &TFI = *MF.getTarget().getFrameInfo();
MachineFrameInfo *MFI = MF.getFrameInfo();
SystemZMachineFunctionInfo *SystemZMFI =
MF.getInfo<SystemZMachineFunctionInfo>();
MachineBasicBlock::iterator MBBI = MBB.begin();
DebugLoc DL = (MBBI != MBB.end() ? MBBI->getDebugLoc() :
DebugLoc::getUnknownLoc());
// Get the number of bytes to allocate from the FrameInfo.
// Note that area for callee-saved stuff is already allocated, thus we need to
// 'undo' the stack movement.
uint64_t StackSize = MFI->getStackSize();
StackSize -= SystemZMFI->getCalleeSavedFrameSize();
uint64_t NumBytes = StackSize - TFI.getOffsetOfLocalArea();
// Skip the callee-saved push instructions.
while (MBBI != MBB.end() &&
(MBBI->getOpcode() == SystemZ::MOV64mr ||
MBBI->getOpcode() == SystemZ::MOV64mrm))
++MBBI;
if (MBBI != MBB.end())
DL = MBBI->getDebugLoc();
// adjust stack pointer: R15 -= numbytes
if (StackSize || MFI->hasCalls()) {
assert(MF.getRegInfo().isPhysRegUsed(SystemZ::R15D) &&
"Invalid stack frame calculation!");
emitSPUpdate(MBB, MBBI, -(int64_t)NumBytes, TII);
}
if (hasFP(MF)) {
// Update R11 with the new base value...
BuildMI(MBB, MBBI, DL, TII.get(SystemZ::MOV64rr), SystemZ::R11D)
.addReg(SystemZ::R15D);
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = llvm::next(MF.begin()), E = MF.end();
I != E; ++I)
I->addLiveIn(SystemZ::R11D);
}
}
void XTCFrameLowering::emitEpilogue(MachineFunction &MF,
MachineBasicBlock &MBB) const {
MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
MachineFrameInfo *MFI = MF.getFrameInfo();
XTCFunctionInfo *XTCFI = MF.getInfo<XTCFunctionInfo>();
const XTCInstrInfo &TII =
*static_cast<const XTCInstrInfo*>(MF.getTarget().getInstrInfo());
DebugLoc dl = MBBI->getDebugLoc();
CallingConv::ID CallConv = MF.getFunction()->getCallingConv();
// Get the FI's where RA and FP are saved.
int FPOffset = XTCFI->getFPStackOffset();
int RAOffset = XTCFI->getRAStackOffset();
if (hasFP(MF)) {
/* Save current FP into stack */
// BuildMI(MBB, MBBI, dl, TII.get(XTC::STWPREI)).addReg(XTC::r14, RegState::Kill).addReg(XTC::r15).addImm(-4);
/* Copy from current SP */
BuildMI(MBB, MBBI, dl, TII.get(XTC::COPY), XTC::r15).addReg(XTC::r14);
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = llvm::next(MF.begin()), E = MF.end();
I != E; ++I)
I->addLiveIn(XTC::r14);
}
/*
// lwi R15, R1, stack_loc
if (MFI->adjustsStack() || requiresRA) {
BuildMI(MBB, MBBI, dl, TII.get(XTC::LWI), XTC::R15)
.addReg(XTC::R1).addImm(RAOffset);
*/
// Get the number of bytes from FrameInfo
int StackSize = (int) MFI->getStackSize();
if (StackSize) {
BuildMI(MBB, MBBI, dl, TII.get(XTC::ADDI), XTC::r15)
.addReg(XTC::r15).addImm(StackSize);
}
}
SIMachineFunctionInfo::SpilledReg SIMachineFunctionInfo::getSpilledReg(
MachineFunction *MF,
unsigned FrameIndex,
unsigned SubIdx) {
const MachineFrameInfo *FrameInfo = MF->getFrameInfo();
const SIRegisterInfo *TRI = static_cast<const SIRegisterInfo *>(
MF->getSubtarget<AMDGPUSubtarget>().getRegisterInfo());
MachineRegisterInfo &MRI = MF->getRegInfo();
int64_t Offset = FrameInfo->getObjectOffset(FrameIndex);
Offset += SubIdx * 4;
unsigned LaneVGPRIdx = Offset / (64 * 4);
unsigned Lane = (Offset / 4) % 64;
struct SpilledReg Spill;
if (!LaneVGPRs.count(LaneVGPRIdx)) {
unsigned LaneVGPR = TRI->findUnusedRegister(MRI, &AMDGPU::VGPR_32RegClass);
if (LaneVGPR == AMDGPU::NoRegister) {
LLVMContext &Ctx = MF->getFunction()->getContext();
Ctx.emitError("Ran out of VGPRs for spilling SGPR");
// When compiling from inside Mesa, the compilation continues.
// Select an arbitrary register to avoid triggering assertions
// during subsequent passes.
LaneVGPR = AMDGPU::VGPR0;
}
LaneVGPRs[LaneVGPRIdx] = LaneVGPR;
// Add this register as live-in to all blocks to avoid machine verifer
// complaining about use of an undefined physical register.
for (MachineFunction::iterator BI = MF->begin(), BE = MF->end();
BI != BE; ++BI) {
BI->addLiveIn(LaneVGPR);
}
}
Spill.VGPR = LaneVGPRs[LaneVGPRIdx];
Spill.Lane = Lane;
return Spill;
}
void MSP430RegisterInfo::emitPrologue(MachineFunction &MF) const {
MachineBasicBlock &MBB = MF.front(); // Prolog goes in entry BB
MachineFrameInfo *MFI = MF.getFrameInfo();
MSP430MachineFunctionInfo *MSP430FI = MF.getInfo<MSP430MachineFunctionInfo>();
MachineBasicBlock::iterator MBBI = MBB.begin();
DebugLoc DL = (MBBI != MBB.end() ? MBBI->getDebugLoc() :
DebugLoc::getUnknownLoc());
// Get the number of bytes to allocate from the FrameInfo.
uint64_t StackSize = MFI->getStackSize();
uint64_t NumBytes = 0;
if (hasFP(MF)) {
// Calculate required stack adjustment
uint64_t FrameSize = StackSize - 2;
NumBytes = FrameSize - MSP430FI->getCalleeSavedFrameSize();
// Get the offset of the stack slot for the EBP register... which is
// guaranteed to be the last slot by processFunctionBeforeFrameFinalized.
// Update the frame offset adjustment.
MFI->setOffsetAdjustment(-NumBytes);
// Save FPW into the appropriate stack slot...
BuildMI(MBB, MBBI, DL, TII.get(MSP430::PUSH16r))
.addReg(MSP430::FPW, RegState::Kill);
// Update FPW with the new base value...
BuildMI(MBB, MBBI, DL, TII.get(MSP430::MOV16rr), MSP430::FPW)
.addReg(MSP430::SPW);
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = next(MF.begin()), E = MF.end();
I != E; ++I)
I->addLiveIn(MSP430::FPW);
} else
NumBytes = StackSize - MSP430FI->getCalleeSavedFrameSize();
// Skip the callee-saved push instructions.
while (MBBI != MBB.end() && (MBBI->getOpcode() == MSP430::PUSH16r))
++MBBI;
if (MBBI != MBB.end())
DL = MBBI->getDebugLoc();
if (NumBytes) { // adjust stack pointer: SPW -= numbytes
// If there is an SUB16ri of SPW immediately before this instruction, merge
// the two.
//NumBytes -= mergeSPUpdates(MBB, MBBI, true);
// If there is an ADD16ri or SUB16ri of SPW immediately after this
// instruction, merge the two instructions.
// mergeSPUpdatesDown(MBB, MBBI, &NumBytes);
if (NumBytes) {
MachineInstr *MI =
BuildMI(MBB, MBBI, DL, TII.get(MSP430::SUB16ri), MSP430::SPW)
.addReg(MSP430::SPW).addImm(NumBytes);
// The SRW implicit def is dead.
MI->getOperand(3).setIsDead();
}
}
}
void AVRFrameLowering::emitPrologue(MachineFunction &MF,
MachineBasicBlock &MBB) const {
MachineBasicBlock::iterator MBBI = MBB.begin();
CallingConv::ID CallConv = MF.getFunction()->getCallingConv();
DebugLoc DL = (MBBI != MBB.end()) ? MBBI->getDebugLoc() : DebugLoc();
const AVRSubtarget &STI = MF.getSubtarget<AVRSubtarget>();
const AVRInstrInfo &TII = *STI.getInstrInfo();
bool HasFP = hasFP(MF);
// Interrupt handlers re-enable interrupts in function entry.
if (CallConv == CallingConv::AVR_INTR) {
BuildMI(MBB, MBBI, DL, TII.get(AVR::BSETs))
.addImm(0x07)
.setMIFlag(MachineInstr::FrameSetup);
}
// Save the frame pointer if we have one.
if (HasFP) {
BuildMI(MBB, MBBI, DL, TII.get(AVR::PUSHWRr))
.addReg(AVR::R29R28, RegState::Kill)
.setMIFlag(MachineInstr::FrameSetup);
}
// Emit special prologue code to save R1, R0 and SREG in interrupt/signal
// handlers before saving any other registers.
if (CallConv == CallingConv::AVR_INTR ||
CallConv == CallingConv::AVR_SIGNAL) {
BuildMI(MBB, MBBI, DL, TII.get(AVR::PUSHWRr))
.addReg(AVR::R1R0, RegState::Kill)
.setMIFlag(MachineInstr::FrameSetup);
BuildMI(MBB, MBBI, DL, TII.get(AVR::INRdA), AVR::R0)
.addImm(0x3f)
.setMIFlag(MachineInstr::FrameSetup);
BuildMI(MBB, MBBI, DL, TII.get(AVR::PUSHRr))
.addReg(AVR::R0, RegState::Kill)
.setMIFlag(MachineInstr::FrameSetup);
BuildMI(MBB, MBBI, DL, TII.get(AVR::EORRdRr))
.addReg(AVR::R0, RegState::Define)
.addReg(AVR::R0, RegState::Kill)
.addReg(AVR::R0, RegState::Kill)
.setMIFlag(MachineInstr::FrameSetup);
}
// Early exit if the frame pointer is not needed in this function.
if (!HasFP) {
return;
}
const MachineFrameInfo &MFI = MF.getFrameInfo();
const AVRMachineFunctionInfo *AFI = MF.getInfo<AVRMachineFunctionInfo>();
unsigned FrameSize = MFI.getStackSize() - AFI->getCalleeSavedFrameSize();
// Skip the callee-saved push instructions.
while (
(MBBI != MBB.end()) && MBBI->getFlag(MachineInstr::FrameSetup) &&
(MBBI->getOpcode() == AVR::PUSHRr || MBBI->getOpcode() == AVR::PUSHWRr)) {
++MBBI;
}
// Update Y with the new base value.
BuildMI(MBB, MBBI, DL, TII.get(AVR::SPREAD), AVR::R29R28)
.addReg(AVR::SP)
.setMIFlag(MachineInstr::FrameSetup);
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = std::next(MF.begin()), E = MF.end();
I != E; ++I) {
I->addLiveIn(AVR::R29R28);
}
if (!FrameSize) {
return;
}
// Reserve the necessary frame memory by doing FP -= <size>.
unsigned Opcode = (isUInt<6>(FrameSize)) ? AVR::SBIWRdK : AVR::SUBIWRdK;
MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(Opcode), AVR::R29R28)
.addReg(AVR::R29R28, RegState::Kill)
.addImm(FrameSize)
.setMIFlag(MachineInstr::FrameSetup);
// The SREG implicit def is dead.
MI->getOperand(3).setIsDead();
// Write back R29R28 to SP and temporarily disable interrupts.
BuildMI(MBB, MBBI, DL, TII.get(AVR::SPWRITE), AVR::SP)
.addReg(AVR::R29R28)
.setMIFlag(MachineInstr::FrameSetup);
}
void SystemZFrameLowering::emitPrologue(MachineFunction &MF) const {
MachineBasicBlock &MBB = MF.front();
MachineFrameInfo *MFFrame = MF.getFrameInfo();
const SystemZInstrInfo *ZII =
static_cast<const SystemZInstrInfo*>(MF.getTarget().getInstrInfo());
SystemZMachineFunctionInfo *ZFI = MF.getInfo<SystemZMachineFunctionInfo>();
MachineBasicBlock::iterator MBBI = MBB.begin();
MachineModuleInfo &MMI = MF.getMMI();
const std::vector<CalleeSavedInfo> &CSI = MFFrame->getCalleeSavedInfo();
bool HasFP = hasFP(MF);
DebugLoc DL = MBBI != MBB.end() ? MBBI->getDebugLoc() : DebugLoc();
// The current offset of the stack pointer from the CFA.
int64_t SPOffsetFromCFA = -SystemZMC::CFAOffsetFromInitialSP;
if (ZFI->getLowSavedGPR()) {
// Skip over the GPR saves.
if (MBBI != MBB.end() && MBBI->getOpcode() == SystemZ::STMG)
++MBBI;
else
llvm_unreachable("Couldn't skip over GPR saves");
// Add CFI for the GPR saves.
MCSymbol *GPRSaveLabel = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL,
ZII->get(TargetOpcode::PROLOG_LABEL)).addSym(GPRSaveLabel);
for (std::vector<CalleeSavedInfo>::const_iterator
I = CSI.begin(), E = CSI.end(); I != E; ++I) {
unsigned Reg = I->getReg();
if (SystemZ::GR64BitRegClass.contains(Reg)) {
int64_t Offset = SPOffsetFromCFA + RegSpillOffsets[Reg];
MachineLocation StackSlot(MachineLocation::VirtualFP, Offset);
MachineLocation RegValue(Reg);
MMI.addFrameMove(GPRSaveLabel, StackSlot, RegValue);
}
}
}
uint64_t StackSize = getAllocatedStackSize(MF);
if (StackSize) {
// Allocate StackSize bytes.
int64_t Delta = -int64_t(StackSize);
emitIncrement(MBB, MBBI, DL, SystemZ::R15D, Delta, ZII);
// Add CFI for the allocation.
MCSymbol *AdjustSPLabel = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, ZII->get(TargetOpcode::PROLOG_LABEL))
.addSym(AdjustSPLabel);
MachineLocation FPDest(MachineLocation::VirtualFP);
MachineLocation FPSrc(MachineLocation::VirtualFP, SPOffsetFromCFA + Delta);
MMI.addFrameMove(AdjustSPLabel, FPDest, FPSrc);
SPOffsetFromCFA += Delta;
}
if (HasFP) {
// Copy the base of the frame to R11.
BuildMI(MBB, MBBI, DL, ZII->get(SystemZ::LGR), SystemZ::R11D)
.addReg(SystemZ::R15D);
// Add CFI for the new frame location.
MCSymbol *SetFPLabel = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, ZII->get(TargetOpcode::PROLOG_LABEL))
.addSym(SetFPLabel);
MachineLocation HardFP(SystemZ::R11D);
MachineLocation VirtualFP(MachineLocation::VirtualFP);
MMI.addFrameMove(SetFPLabel, HardFP, VirtualFP);
// Mark the FramePtr as live at the beginning of every block except
// the entry block. (We'll have marked R11 as live on entry when
// saving the GPRs.)
for (MachineFunction::iterator
I = llvm::next(MF.begin()), E = MF.end(); I != E; ++I)
I->addLiveIn(SystemZ::R11D);
}
// Skip over the FPR saves.
MCSymbol *FPRSaveLabel = 0;
for (std::vector<CalleeSavedInfo>::const_iterator
I = CSI.begin(), E = CSI.end(); I != E; ++I) {
unsigned Reg = I->getReg();
if (SystemZ::FP64BitRegClass.contains(Reg)) {
if (MBBI != MBB.end() &&
(MBBI->getOpcode() == SystemZ::STD ||
MBBI->getOpcode() == SystemZ::STDY))
++MBBI;
else
llvm_unreachable("Couldn't skip over FPR save");
// Add CFI for the this save.
if (!FPRSaveLabel)
FPRSaveLabel = MMI.getContext().CreateTempSymbol();
unsigned Reg = I->getReg();
int64_t Offset = getFrameIndexOffset(MF, I->getFrameIdx());
MachineLocation Slot(MachineLocation::VirtualFP,
SPOffsetFromCFA + Offset);
MachineLocation RegValue(Reg);
MMI.addFrameMove(FPRSaveLabel, Slot, RegValue);
}
}
// Complete the CFI for the FPR saves, modelling them as taking effect
// after the last save.
if (FPRSaveLabel)
BuildMI(MBB, MBBI, DL, ZII->get(TargetOpcode::PROLOG_LABEL))
.addSym(FPRSaveLabel);
}
/// emitPrologue - Push callee-saved registers onto the stack, which
/// automatically adjust the stack pointer. Adjust the stack pointer to allocate
/// space for local variables. Also emit labels used by the exception handler to
/// generate the exception handling frames.
void X86FrameInfo::emitPrologue(MachineFunction &MF) const {
MachineBasicBlock &MBB = MF.front(); // Prologue goes in entry BB.
MachineBasicBlock::iterator MBBI = MBB.begin();
MachineFrameInfo *MFI = MF.getFrameInfo();
const Function *Fn = MF.getFunction();
const X86RegisterInfo *RegInfo = TM.getRegisterInfo();
const X86InstrInfo &TII = *TM.getInstrInfo();
MachineModuleInfo &MMI = MF.getMMI();
X86MachineFunctionInfo *X86FI = MF.getInfo<X86MachineFunctionInfo>();
bool needsFrameMoves = MMI.hasDebugInfo() ||
!Fn->doesNotThrow() || UnwindTablesMandatory;
uint64_t MaxAlign = MFI->getMaxAlignment(); // Desired stack alignment.
uint64_t StackSize = MFI->getStackSize(); // Number of bytes to allocate.
bool HasFP = hasFP(MF);
bool Is64Bit = STI.is64Bit();
bool IsWin64 = STI.isTargetWin64();
unsigned StackAlign = getStackAlignment();
unsigned SlotSize = RegInfo->getSlotSize();
unsigned FramePtr = RegInfo->getFrameRegister(MF);
unsigned StackPtr = RegInfo->getStackRegister();
DebugLoc DL;
// If we're forcing a stack realignment we can't rely on just the frame
// info, we need to know the ABI stack alignment as well in case we
// have a call out. Otherwise just make sure we have some alignment - we'll
// go with the minimum SlotSize.
if (ForceStackAlign) {
if (MFI->hasCalls())
MaxAlign = (StackAlign > MaxAlign) ? StackAlign : MaxAlign;
else if (MaxAlign < SlotSize)
MaxAlign = SlotSize;
}
// Add RETADDR move area to callee saved frame size.
int TailCallReturnAddrDelta = X86FI->getTCReturnAddrDelta();
if (TailCallReturnAddrDelta < 0)
X86FI->setCalleeSavedFrameSize(
X86FI->getCalleeSavedFrameSize() - TailCallReturnAddrDelta);
// If this is x86-64 and the Red Zone is not disabled, if we are a leaf
// function, and use up to 128 bytes of stack space, don't have a frame
// pointer, calls, or dynamic alloca then we do not need to adjust the
// stack pointer (we fit in the Red Zone).
if (Is64Bit && !Fn->hasFnAttr(Attribute::NoRedZone) &&
!RegInfo->needsStackRealignment(MF) &&
!MFI->hasVarSizedObjects() && // No dynamic alloca.
!MFI->adjustsStack() && // No calls.
!IsWin64) { // Win64 has no Red Zone
uint64_t MinSize = X86FI->getCalleeSavedFrameSize();
if (HasFP) MinSize += SlotSize;
StackSize = std::max(MinSize, StackSize > 128 ? StackSize - 128 : 0);
MFI->setStackSize(StackSize);
} else if (IsWin64) {
// We need to always allocate 32 bytes as register spill area.
// FIXME: We might reuse these 32 bytes for leaf functions.
StackSize += 32;
MFI->setStackSize(StackSize);
}
// Insert stack pointer adjustment for later moving of return addr. Only
// applies to tail call optimized functions where the callee argument stack
// size is bigger than the callers.
if (TailCallReturnAddrDelta < 0) {
MachineInstr *MI =
BuildMI(MBB, MBBI, DL,
TII.get(getSUBriOpcode(Is64Bit, -TailCallReturnAddrDelta)),
StackPtr)
.addReg(StackPtr)
.addImm(-TailCallReturnAddrDelta);
MI->getOperand(3).setIsDead(); // The EFLAGS implicit def is dead.
}
// Mapping for machine moves:
//
// DST: VirtualFP AND
// SRC: VirtualFP => DW_CFA_def_cfa_offset
// ELSE => DW_CFA_def_cfa
//
// SRC: VirtualFP AND
// DST: Register => DW_CFA_def_cfa_register
//
// ELSE
// OFFSET < 0 => DW_CFA_offset_extended_sf
// REG < 64 => DW_CFA_offset + Reg
// ELSE => DW_CFA_offset_extended
std::vector<MachineMove> &Moves = MMI.getFrameMoves();
const TargetData *TD = MF.getTarget().getTargetData();
uint64_t NumBytes = 0;
int stackGrowth = -TD->getPointerSize();
if (HasFP) {
// Calculate required stack adjustment.
uint64_t FrameSize = StackSize - SlotSize;
if (RegInfo->needsStackRealignment(MF))
FrameSize = (FrameSize + MaxAlign - 1) / MaxAlign * MaxAlign;
NumBytes = FrameSize - X86FI->getCalleeSavedFrameSize();
// Get the offset of the stack slot for the EBP register, which is
// guaranteed to be the last slot by processFunctionBeforeFrameFinalized.
// Update the frame offset adjustment.
MFI->setOffsetAdjustment(-NumBytes);
// Save EBP/RBP into the appropriate stack slot.
BuildMI(MBB, MBBI, DL, TII.get(Is64Bit ? X86::PUSH64r : X86::PUSH32r))
.addReg(FramePtr, RegState::Kill);
if (needsFrameMoves) {
// Mark the place where EBP/RBP was saved.
MCSymbol *FrameLabel = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, TII.get(X86::PROLOG_LABEL)).addSym(FrameLabel);
// Define the current CFA rule to use the provided offset.
if (StackSize) {
MachineLocation SPDst(MachineLocation::VirtualFP);
MachineLocation SPSrc(MachineLocation::VirtualFP, 2 * stackGrowth);
Moves.push_back(MachineMove(FrameLabel, SPDst, SPSrc));
} else {
// FIXME: Verify & implement for FP
MachineLocation SPDst(StackPtr);
MachineLocation SPSrc(StackPtr, stackGrowth);
Moves.push_back(MachineMove(FrameLabel, SPDst, SPSrc));
}
// Change the rule for the FramePtr to be an "offset" rule.
MachineLocation FPDst(MachineLocation::VirtualFP, 2 * stackGrowth);
MachineLocation FPSrc(FramePtr);
Moves.push_back(MachineMove(FrameLabel, FPDst, FPSrc));
}
// Update EBP with the new base value...
BuildMI(MBB, MBBI, DL,
TII.get(Is64Bit ? X86::MOV64rr : X86::MOV32rr), FramePtr)
.addReg(StackPtr);
if (needsFrameMoves) {
// Mark effective beginning of when frame pointer becomes valid.
MCSymbol *FrameLabel = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, TII.get(X86::PROLOG_LABEL)).addSym(FrameLabel);
// Define the current CFA to use the EBP/RBP register.
MachineLocation FPDst(FramePtr);
MachineLocation FPSrc(MachineLocation::VirtualFP);
Moves.push_back(MachineMove(FrameLabel, FPDst, FPSrc));
}
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = llvm::next(MF.begin()), E = MF.end();
I != E; ++I)
I->addLiveIn(FramePtr);
// Realign stack
if (RegInfo->needsStackRealignment(MF)) {
MachineInstr *MI =
BuildMI(MBB, MBBI, DL,
TII.get(Is64Bit ? X86::AND64ri32 : X86::AND32ri),
StackPtr).addReg(StackPtr).addImm(-MaxAlign);
// The EFLAGS implicit def is dead.
MI->getOperand(3).setIsDead();
}
} else {
NumBytes = StackSize - X86FI->getCalleeSavedFrameSize();
}
// Skip the callee-saved push instructions.
bool PushedRegs = false;
int StackOffset = 2 * stackGrowth;
while (MBBI != MBB.end() &&
(MBBI->getOpcode() == X86::PUSH32r ||
MBBI->getOpcode() == X86::PUSH64r)) {
PushedRegs = true;
++MBBI;
if (!HasFP && needsFrameMoves) {
// Mark callee-saved push instruction.
MCSymbol *Label = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, TII.get(X86::PROLOG_LABEL)).addSym(Label);
// Define the current CFA rule to use the provided offset.
unsigned Ptr = StackSize ?
MachineLocation::VirtualFP : StackPtr;
MachineLocation SPDst(Ptr);
MachineLocation SPSrc(Ptr, StackOffset);
Moves.push_back(MachineMove(Label, SPDst, SPSrc));
StackOffset += stackGrowth;
}
}
DL = MBB.findDebugLoc(MBBI);
// If there is an SUB32ri of ESP immediately before this instruction, merge
// the two. This can be the case when tail call elimination is enabled and
// the callee has more arguments then the caller.
NumBytes -= mergeSPUpdates(MBB, MBBI, StackPtr, true);
// If there is an ADD32ri or SUB32ri of ESP immediately after this
// instruction, merge the two instructions.
mergeSPUpdatesDown(MBB, MBBI, StackPtr, &NumBytes);
// Adjust stack pointer: ESP -= numbytes.
// Windows and cygwin/mingw require a prologue helper routine when allocating
// more than 4K bytes on the stack. Windows uses __chkstk and cygwin/mingw
// uses __alloca. __alloca and the 32-bit version of __chkstk will probe the
// stack and adjust the stack pointer in one go. The 64-bit version of
// __chkstk is only responsible for probing the stack. The 64-bit prologue is
// responsible for adjusting the stack pointer. Touching the stack at 4K
// increments is necessary to ensure that the guard pages used by the OS
// virtual memory manager are allocated in correct sequence.
if (NumBytes >= 4096 && (STI.isTargetCygMing() || STI.isTargetWin32())) {
// Check whether EAX is livein for this function.
bool isEAXAlive = isEAXLiveIn(MF);
const char *StackProbeSymbol =
STI.isTargetWindows() ? "_chkstk" : "_alloca";
unsigned CallOp = Is64Bit ? X86::CALL64pcrel32 : X86::CALLpcrel32;
if (!isEAXAlive) {
BuildMI(MBB, MBBI, DL, TII.get(X86::MOV32ri), X86::EAX)
.addImm(NumBytes);
BuildMI(MBB, MBBI, DL, TII.get(CallOp))
.addExternalSymbol(StackProbeSymbol)
.addReg(StackPtr, RegState::Define | RegState::Implicit)
.addReg(X86::EFLAGS, RegState::Define | RegState::Implicit);
} else {
// Save EAX
BuildMI(MBB, MBBI, DL, TII.get(X86::PUSH32r))
.addReg(X86::EAX, RegState::Kill);
// Allocate NumBytes-4 bytes on stack. We'll also use 4 already
// allocated bytes for EAX.
BuildMI(MBB, MBBI, DL, TII.get(X86::MOV32ri), X86::EAX)
.addImm(NumBytes - 4);
BuildMI(MBB, MBBI, DL, TII.get(CallOp))
.addExternalSymbol(StackProbeSymbol)
.addReg(StackPtr, RegState::Define | RegState::Implicit)
.addReg(X86::EFLAGS, RegState::Define | RegState::Implicit);
// Restore EAX
MachineInstr *MI = addRegOffset(BuildMI(MF, DL, TII.get(X86::MOV32rm),
X86::EAX),
StackPtr, false, NumBytes - 4);
MBB.insert(MBBI, MI);
}
} else if (NumBytes >= 4096 && STI.isTargetWin64()) {
// Sanity check that EAX is not livein for this function. It should
// should not be, so throw an assert.
assert(!isEAXLiveIn(MF) && "EAX is livein in the Win64 case!");
// Handle the 64-bit Windows ABI case where we need to call __chkstk.
// Function prologue is responsible for adjusting the stack pointer.
BuildMI(MBB, MBBI, DL, TII.get(X86::MOV32ri), X86::EAX)
.addImm(NumBytes);
BuildMI(MBB, MBBI, DL, TII.get(X86::WINCALL64pcrel32))
.addExternalSymbol("__chkstk")
.addReg(StackPtr, RegState::Define | RegState::Implicit);
emitSPUpdate(MBB, MBBI, StackPtr, -(int64_t)NumBytes, Is64Bit, TII);
} else if (NumBytes)
emitSPUpdate(MBB, MBBI, StackPtr, -(int64_t)NumBytes, Is64Bit, TII);
if ((NumBytes || PushedRegs) && needsFrameMoves) {
// Mark end of stack pointer adjustment.
MCSymbol *Label = MMI.getContext().CreateTempSymbol();
BuildMI(MBB, MBBI, DL, TII.get(X86::PROLOG_LABEL)).addSym(Label);
if (!HasFP && NumBytes) {
// Define the current CFA rule to use the provided offset.
if (StackSize) {
MachineLocation SPDst(MachineLocation::VirtualFP);
MachineLocation SPSrc(MachineLocation::VirtualFP,
-StackSize + stackGrowth);
Moves.push_back(MachineMove(Label, SPDst, SPSrc));
} else {
// FIXME: Verify & implement for FP
MachineLocation SPDst(StackPtr);
MachineLocation SPSrc(StackPtr, stackGrowth);
Moves.push_back(MachineMove(Label, SPDst, SPSrc));
}
}
// Emit DWARF info specifying the offsets of the callee-saved registers.
if (PushedRegs)
emitCalleeSavedFrameMoves(MF, Label, HasFP ? FramePtr : StackPtr);
}
}
void X86RegisterInfo::emitPrologue(MachineFunction &MF) const {
MachineBasicBlock &MBB = MF.front(); // Prolog goes in entry BB
MachineFrameInfo *MFI = MF.getFrameInfo();
const Function* Fn = MF.getFunction();
const X86Subtarget* Subtarget = &MF.getTarget().getSubtarget<X86Subtarget>();
MachineModuleInfo *MMI = MFI->getMachineModuleInfo();
X86MachineFunctionInfo *X86FI = MF.getInfo<X86MachineFunctionInfo>();
MachineBasicBlock::iterator MBBI = MBB.begin();
bool needsFrameMoves = (MMI && MMI->hasDebugInfo()) ||
!Fn->doesNotThrow() ||
UnwindTablesMandatory;
// Prepare for frame info.
unsigned FrameLabelId = 0;
// Get the number of bytes to allocate from the FrameInfo.
uint64_t StackSize = MFI->getStackSize();
// Get desired stack alignment
uint64_t MaxAlign = MFI->getMaxAlignment();
// Add RETADDR move area to callee saved frame size.
int TailCallReturnAddrDelta = X86FI->getTCReturnAddrDelta();
if (TailCallReturnAddrDelta < 0)
X86FI->setCalleeSavedFrameSize(
X86FI->getCalleeSavedFrameSize() +(-TailCallReturnAddrDelta));
// If this is x86-64 and the Red Zone is not disabled, if we are a leaf
// function, and use up to 128 bytes of stack space, don't have a frame
// pointer, calls, or dynamic alloca then we do not need to adjust the
// stack pointer (we fit in the Red Zone).
if (Is64Bit && !DisableRedZone &&
!needsStackRealignment(MF) &&
!MFI->hasVarSizedObjects() && // No dynamic alloca.
!MFI->hasCalls()) { // No calls.
uint64_t MinSize = X86FI->getCalleeSavedFrameSize();
if (hasFP(MF)) MinSize += SlotSize;
StackSize = std::max(MinSize,
StackSize > 128 ? StackSize - 128 : 0);
MFI->setStackSize(StackSize);
}
// Insert stack pointer adjustment for later moving of return addr. Only
// applies to tail call optimized functions where the callee argument stack
// size is bigger than the callers.
if (TailCallReturnAddrDelta < 0) {
MachineInstr *MI =
BuildMI(MBB, MBBI, TII.get(Is64Bit? X86::SUB64ri32 : X86::SUB32ri),
StackPtr).addReg(StackPtr).addImm(-TailCallReturnAddrDelta);
// The EFLAGS implicit def is dead.
MI->getOperand(3).setIsDead();
}
uint64_t NumBytes = 0;
if (hasFP(MF)) {
// Calculate required stack adjustment
uint64_t FrameSize = StackSize - SlotSize;
if (needsStackRealignment(MF))
FrameSize = (FrameSize + MaxAlign - 1)/MaxAlign*MaxAlign;
NumBytes = FrameSize - X86FI->getCalleeSavedFrameSize();
// Get the offset of the stack slot for the EBP register... which is
// guaranteed to be the last slot by processFunctionBeforeFrameFinalized.
// Update the frame offset adjustment.
MFI->setOffsetAdjustment(-NumBytes);
// Save EBP into the appropriate stack slot...
BuildMI(MBB, MBBI, TII.get(Is64Bit ? X86::PUSH64r : X86::PUSH32r))
.addReg(FramePtr, /*isDef=*/false, /*isImp=*/false, /*isKill=*/true);
if (needsFrameMoves) {
// Mark effective beginning of when frame pointer becomes valid.
FrameLabelId = MMI->NextLabelID();
BuildMI(MBB, MBBI, TII.get(X86::DBG_LABEL)).addImm(FrameLabelId);
}
// Update EBP with the new base value...
BuildMI(MBB, MBBI, TII.get(Is64Bit ? X86::MOV64rr : X86::MOV32rr), FramePtr)
.addReg(StackPtr);
// Mark the FramePtr as live-in in every block except the entry.
for (MachineFunction::iterator I = next(MF.begin()), E = MF.end();
I != E; ++I)
I->addLiveIn(FramePtr);
// Realign stack
if (needsStackRealignment(MF)) {
MachineInstr *MI =
BuildMI(MBB, MBBI,
TII.get(Is64Bit ? X86::AND64ri32 : X86::AND32ri),
StackPtr).addReg(StackPtr).addImm(-MaxAlign);
// The EFLAGS implicit def is dead.
MI->getOperand(3).setIsDead();
}
} else
NumBytes = StackSize - X86FI->getCalleeSavedFrameSize();
unsigned ReadyLabelId = 0;
if (needsFrameMoves) {
// Mark effective beginning of when frame pointer is ready.
ReadyLabelId = MMI->NextLabelID();
BuildMI(MBB, MBBI, TII.get(X86::DBG_LABEL)).addImm(ReadyLabelId);
}
// Skip the callee-saved push instructions.
while (MBBI != MBB.end() &&
(MBBI->getOpcode() == X86::PUSH32r ||
MBBI->getOpcode() == X86::PUSH64r))
++MBBI;
if (NumBytes) { // adjust stack pointer: ESP -= numbytes
if (NumBytes >= 4096 && Subtarget->isTargetCygMing()) {
// Check, whether EAX is livein for this function
bool isEAXAlive = false;
for (MachineRegisterInfo::livein_iterator
II = MF.getRegInfo().livein_begin(),
EE = MF.getRegInfo().livein_end(); (II != EE) && !isEAXAlive; ++II) {
unsigned Reg = II->first;
isEAXAlive = (Reg == X86::EAX || Reg == X86::AX ||
Reg == X86::AH || Reg == X86::AL);
}
// Function prologue calls _alloca to probe the stack when allocating
// more than 4k bytes in one go. Touching the stack at 4K increments is
// necessary to ensure that the guard pages used by the OS virtual memory
// manager are allocated in correct sequence.
if (!isEAXAlive) {
BuildMI(MBB, MBBI, TII.get(X86::MOV32ri), X86::EAX).addImm(NumBytes);
BuildMI(MBB, MBBI, TII.get(X86::CALLpcrel32))
.addExternalSymbol("_alloca");
} else {
// Save EAX
BuildMI(MBB, MBBI, TII.get(X86::PUSH32r))
.addReg(X86::EAX, /*isDef=*/false, /*isImp=*/false, /*isKill=*/true);
// Allocate NumBytes-4 bytes on stack. We'll also use 4 already
// allocated bytes for EAX.
BuildMI(MBB, MBBI, TII.get(X86::MOV32ri), X86::EAX).addImm(NumBytes-4);
BuildMI(MBB, MBBI, TII.get(X86::CALLpcrel32))
.addExternalSymbol("_alloca");
// Restore EAX
MachineInstr *MI = addRegOffset(BuildMI(MF, TII.get(X86::MOV32rm),X86::EAX),
StackPtr, false, NumBytes-4);
MBB.insert(MBBI, MI);
}
} else {
// If there is an SUB32ri of ESP immediately before this instruction,
// merge the two. This can be the case when tail call elimination is
// enabled and the callee has more arguments then the caller.
NumBytes -= mergeSPUpdates(MBB, MBBI, StackPtr, true);
// If there is an ADD32ri or SUB32ri of ESP immediately after this
// instruction, merge the two instructions.
mergeSPUpdatesDown(MBB, MBBI, StackPtr, &NumBytes);
if (NumBytes)
emitSPUpdate(MBB, MBBI, StackPtr, -(int64_t)NumBytes, Is64Bit, TII);
}
}
if (needsFrameMoves)
emitFrameMoves(MF, FrameLabelId, ReadyLabelId);
}