//===----------------------------------------------------------------------===//
#include "X86RegisterInfo.h"
-#include "X86.h"
+#include "X86FrameLowering.h"
#include "X86InstrBuilder.h"
#include "X86MachineFunctionInfo.h"
#include "X86Subtarget.h"
#include "X86TargetMachine.h"
-#include "llvm/Constants.h"
-#include "llvm/Function.h"
-#include "llvm/Type.h"
-#include "llvm/CodeGen/ValueTypes.h"
-#include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/ADT/BitVector.h"
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
-#include "llvm/CodeGen/MachineFrameInfo.h"
+#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineModuleInfo.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
+#include "llvm/CodeGen/MachineValueType.h"
+#include "llvm/IR/Constants.h"
+#include "llvm/IR/Function.h"
+#include "llvm/IR/Type.h"
#include "llvm/MC/MCAsmInfo.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/ErrorHandling.h"
#include "llvm/Target/TargetFrameLowering.h"
#include "llvm/Target/TargetInstrInfo.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Target/TargetOptions.h"
-#include "llvm/ADT/BitVector.h"
-#include "llvm/ADT/STLExtras.h"
-#include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/CommandLine.h"
+
+using namespace llvm;
#define GET_REGINFO_TARGET_DESC
#include "X86GenRegisterInfo.inc"
-using namespace llvm;
+static cl::opt<bool>
+EnableBasePointer("x86-use-base-pointer", cl::Hidden, cl::init(true),
+ cl::desc("Enable use of a base pointer for complex stack frames"));
-cl::opt<bool>
-ForceStackAlign("force-align-stack",
- cl::desc("Force align the stack to the minimum alignment"
- " needed for the function."),
- cl::init(false), cl::Hidden);
-
-X86RegisterInfo::X86RegisterInfo(X86TargetMachine &tm,
- const TargetInstrInfo &tii)
- : X86GenRegisterInfo(tm.getSubtarget<X86Subtarget>().is64Bit()
- ? X86::RIP : X86::EIP,
- X86_MC::getDwarfRegFlavour(tm.getTargetTriple(), false),
- X86_MC::getDwarfRegFlavour(tm.getTargetTriple(), true)),
- TM(tm), TII(tii) {
+X86RegisterInfo::X86RegisterInfo(const Triple &TT)
+ : X86GenRegisterInfo((TT.isArch64Bit() ? X86::RIP : X86::EIP),
+ X86_MC::getDwarfRegFlavour(TT, false),
+ X86_MC::getDwarfRegFlavour(TT, true),
+ (TT.isArch64Bit() ? X86::RIP : X86::EIP)) {
X86_MC::InitLLVM2SEHRegisterMapping(this);
// Cache some information.
- const X86Subtarget *Subtarget = &TM.getSubtarget<X86Subtarget>();
- Is64Bit = Subtarget->is64Bit();
- IsWin64 = Subtarget->isTargetWin64();
+ Is64Bit = TT.isArch64Bit();
+ IsWin64 = Is64Bit && TT.isOSWindows();
+ // Use a callee-saved register as the base pointer. These registers must
+ // not conflict with any ABI requirements. For example, in 32-bit mode PIC
+ // requires GOT in the EBX register before function calls via PLT GOT pointer.
if (Is64Bit) {
SlotSize = 8;
- StackPtr = X86::RSP;
- FramePtr = X86::RBP;
+ // This matches the simplified 32-bit pointer code in the data layout
+ // computation.
+ // FIXME: Should use the data layout?
+ bool Use64BitReg = TT.getEnvironment() != Triple::GNUX32;
+ StackPtr = Use64BitReg ? X86::RSP : X86::ESP;
+ FramePtr = Use64BitReg ? X86::RBP : X86::EBP;
+ BasePtr = Use64BitReg ? X86::RBX : X86::EBX;
} else {
SlotSize = 4;
StackPtr = X86::ESP;
FramePtr = X86::EBP;
+ BasePtr = X86::ESI;
}
}
-/// getCompactUnwindRegNum - This function maps the register to the number for
-/// compact unwind encoding. Return -1 if the register isn't valid.
-int X86RegisterInfo::getCompactUnwindRegNum(unsigned RegNum, bool isEH) const {
- switch (getLLVMRegNum(RegNum, isEH)) {
- case X86::EBX: case X86::RBX: return 1;
- case X86::ECX: case X86::R12: return 2;
- case X86::EDX: case X86::R13: return 3;
- case X86::EDI: case X86::R14: return 4;
- case X86::ESI: case X86::R15: return 5;
- case X86::EBP: case X86::RBP: return 6;
- }
-
- return -1;
-}
-
bool
X86RegisterInfo::trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
- // Only enable when post-RA scheduling is enabled and this is needed.
- return TM.getSubtargetImpl()->postRAScheduler();
+ // ExeDepsFixer and PostRAScheduler require liveness.
+ return true;
}
int
X86RegisterInfo::getSEHRegNum(unsigned i) const {
- int reg = X86_MC::getX86RegNum(i);
- switch (i) {
- case X86::R8: case X86::R8D: case X86::R8W: case X86::R8B:
- case X86::R9: case X86::R9D: case X86::R9W: case X86::R9B:
- case X86::R10: case X86::R10D: case X86::R10W: case X86::R10B:
- case X86::R11: case X86::R11D: case X86::R11W: case X86::R11B:
- case X86::R12: case X86::R12D: case X86::R12W: case X86::R12B:
- case X86::R13: case X86::R13D: case X86::R13W: case X86::R13B:
- case X86::R14: case X86::R14D: case X86::R14W: case X86::R14B:
- case X86::R15: case X86::R15D: case X86::R15W: case X86::R15B:
- case X86::XMM8: case X86::XMM9: case X86::XMM10: case X86::XMM11:
- case X86::XMM12: case X86::XMM13: case X86::XMM14: case X86::XMM15:
- case X86::YMM8: case X86::YMM9: case X86::YMM10: case X86::YMM11:
- case X86::YMM12: case X86::YMM13: case X86::YMM14: case X86::YMM15:
- reg += 8;
- }
- return reg;
+ return getEncodingValue(i);
}
const TargetRegisterClass *
if (!Is64Bit && SubIdx == X86::sub_8bit) {
A = X86GenRegisterInfo::getSubClassWithSubReg(A, X86::sub_8bit_hi);
if (!A)
- return 0;
+ return nullptr;
}
return X86GenRegisterInfo::getMatchingSuperRegClass(A, B, SubIdx);
}
-const TargetRegisterClass*
-X86RegisterInfo::getLargestLegalSuperClass(const TargetRegisterClass *RC) const{
+const TargetRegisterClass *
+X86RegisterInfo::getLargestLegalSuperClass(const TargetRegisterClass *RC,
+ const MachineFunction &MF) const {
// Don't allow super-classes of GR8_NOREX. This class is only used after
- // extrating sub_8bit_hi sub-registers. The H sub-registers cannot be copied
+ // extracting sub_8bit_hi sub-registers. The H sub-registers cannot be copied
// to the full GR8 register class in 64-bit mode, so we cannot allow the
// reigster class inflation.
//
}
const TargetRegisterClass *
-X86RegisterInfo::getPointerRegClass(const MachineFunction &MF, unsigned Kind)
- const {
+X86RegisterInfo::getPointerRegClass(const MachineFunction &MF,
+ unsigned Kind) const {
+ const X86Subtarget &Subtarget = MF.getSubtarget<X86Subtarget>();
switch (Kind) {
default: llvm_unreachable("Unexpected Kind in getPointerRegClass!");
case 0: // Normal GPRs.
- if (TM.getSubtarget<X86Subtarget>().is64Bit())
+ if (Subtarget.isTarget64BitLP64())
return &X86::GR64RegClass;
return &X86::GR32RegClass;
case 1: // Normal GPRs except the stack pointer (for encoding reasons).
- if (TM.getSubtarget<X86Subtarget>().is64Bit())
+ if (Subtarget.isTarget64BitLP64())
return &X86::GR64_NOSPRegClass;
return &X86::GR32_NOSPRegClass;
- case 2: // Available for tailcall (not callee-saved GPRs).
- if (TM.getSubtarget<X86Subtarget>().isTargetWin64())
+ case 2: // NOREX GPRs.
+ if (Subtarget.isTarget64BitLP64())
+ return &X86::GR64_NOREXRegClass;
+ return &X86::GR32_NOREXRegClass;
+ case 3: // NOREX GPRs except the stack pointer (for encoding reasons).
+ if (Subtarget.isTarget64BitLP64())
+ return &X86::GR64_NOREX_NOSPRegClass;
+ return &X86::GR32_NOREX_NOSPRegClass;
+ case 4: // Available for tailcall (not callee-saved GPRs).
+ const Function *F = MF.getFunction();
+ if (IsWin64 || (F && F->getCallingConv() == CallingConv::X86_64_Win64))
return &X86::GR64_TCW64RegClass;
- if (TM.getSubtarget<X86Subtarget>().is64Bit())
+ else if (Is64Bit)
return &X86::GR64_TCRegClass;
+
+ bool hasHipeCC = (F ? F->getCallingConv() == CallingConv::HiPE : false);
+ if (hasHipeCC)
+ return &X86::GR32RegClass;
return &X86::GR32_TCRegClass;
}
}
unsigned
X86RegisterInfo::getRegPressureLimit(const TargetRegisterClass *RC,
MachineFunction &MF) const {
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
+ const X86FrameLowering *TFI = getFrameLowering(MF);
unsigned FPDiff = TFI->hasFP(MF) ? 1 : 0;
switch (RC->getID()) {
case X86::GR64RegClassID:
return 12 - FPDiff;
case X86::VR128RegClassID:
- return TM.getSubtarget<X86Subtarget>().is64Bit() ? 10 : 4;
+ return Is64Bit ? 10 : 4;
case X86::VR64RegClassID:
return 4;
}
}
-const uint16_t *
+const MCPhysReg *
X86RegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
- bool callsEHReturn = false;
- bool ghcCall = false;
-
- if (MF) {
- callsEHReturn = MF->getMMI().callsEHReturn();
- const Function *F = MF->getFunction();
- ghcCall = (F ? F->getCallingConv() == CallingConv::GHC : false);
+ const X86Subtarget &Subtarget = MF->getSubtarget<X86Subtarget>();
+ bool HasAVX = Subtarget.hasAVX();
+ bool HasAVX512 = Subtarget.hasAVX512();
+ bool CallsEHReturn = MF->getMMI().callsEHReturn();
+
+ assert(MF && "MachineFunction required");
+ switch (MF->getFunction()->getCallingConv()) {
+ case CallingConv::GHC:
+ case CallingConv::HiPE:
+ return CSR_NoRegs_SaveList;
+ case CallingConv::AnyReg:
+ if (HasAVX)
+ return CSR_64_AllRegs_AVX_SaveList;
+ return CSR_64_AllRegs_SaveList;
+ case CallingConv::PreserveMost:
+ return CSR_64_RT_MostRegs_SaveList;
+ case CallingConv::PreserveAll:
+ if (HasAVX)
+ return CSR_64_RT_AllRegs_AVX_SaveList;
+ return CSR_64_RT_AllRegs_SaveList;
+ case CallingConv::Intel_OCL_BI: {
+ if (HasAVX512 && IsWin64)
+ return CSR_Win64_Intel_OCL_BI_AVX512_SaveList;
+ if (HasAVX512 && Is64Bit)
+ return CSR_64_Intel_OCL_BI_AVX512_SaveList;
+ if (HasAVX && IsWin64)
+ return CSR_Win64_Intel_OCL_BI_AVX_SaveList;
+ if (HasAVX && Is64Bit)
+ return CSR_64_Intel_OCL_BI_AVX_SaveList;
+ if (!HasAVX && !IsWin64 && Is64Bit)
+ return CSR_64_Intel_OCL_BI_SaveList;
+ break;
+ }
+ case CallingConv::HHVM:
+ return CSR_64_HHVM_SaveList;
+ case CallingConv::Cold:
+ if (Is64Bit)
+ return CSR_64_MostRegs_SaveList;
+ break;
+ case CallingConv::X86_64_Win64:
+ return CSR_Win64_SaveList;
+ case CallingConv::X86_64_SysV:
+ if (CallsEHReturn)
+ return CSR_64EHRet_SaveList;
+ return CSR_64_SaveList;
+ default:
+ break;
}
- if (ghcCall)
- return CSR_NoRegs_SaveList;
if (Is64Bit) {
if (IsWin64)
return CSR_Win64_SaveList;
- if (callsEHReturn)
+ if (CallsEHReturn)
return CSR_64EHRet_SaveList;
return CSR_64_SaveList;
}
- if (callsEHReturn)
+ if (CallsEHReturn)
return CSR_32EHRet_SaveList;
return CSR_32_SaveList;
}
-const uint32_t*
-X86RegisterInfo::getCallPreservedMask(CallingConv::ID CC) const {
- if (CC == CallingConv::GHC)
+const uint32_t *
+X86RegisterInfo::getCallPreservedMask(const MachineFunction &MF,
+ CallingConv::ID CC) const {
+ const X86Subtarget &Subtarget = MF.getSubtarget<X86Subtarget>();
+ bool HasAVX = Subtarget.hasAVX();
+ bool HasAVX512 = Subtarget.hasAVX512();
+
+ switch (CC) {
+ case CallingConv::GHC:
+ case CallingConv::HiPE:
return CSR_NoRegs_RegMask;
- if (!Is64Bit)
- return CSR_32_RegMask;
- if (IsWin64)
+ case CallingConv::AnyReg:
+ if (HasAVX)
+ return CSR_64_AllRegs_AVX_RegMask;
+ return CSR_64_AllRegs_RegMask;
+ case CallingConv::PreserveMost:
+ return CSR_64_RT_MostRegs_RegMask;
+ case CallingConv::PreserveAll:
+ if (HasAVX)
+ return CSR_64_RT_AllRegs_AVX_RegMask;
+ return CSR_64_RT_AllRegs_RegMask;
+ case CallingConv::Intel_OCL_BI: {
+ if (HasAVX512 && IsWin64)
+ return CSR_Win64_Intel_OCL_BI_AVX512_RegMask;
+ if (HasAVX512 && Is64Bit)
+ return CSR_64_Intel_OCL_BI_AVX512_RegMask;
+ if (HasAVX && IsWin64)
+ return CSR_Win64_Intel_OCL_BI_AVX_RegMask;
+ if (HasAVX && Is64Bit)
+ return CSR_64_Intel_OCL_BI_AVX_RegMask;
+ if (!HasAVX && !IsWin64 && Is64Bit)
+ return CSR_64_Intel_OCL_BI_RegMask;
+ break;
+ }
+ case CallingConv::HHVM:
+ return CSR_64_HHVM_RegMask;
+ case CallingConv::Cold:
+ if (Is64Bit)
+ return CSR_64_MostRegs_RegMask;
+ break;
+ default:
+ break;
+ case CallingConv::X86_64_Win64:
return CSR_Win64_RegMask;
- return CSR_64_RegMask;
+ case CallingConv::X86_64_SysV:
+ return CSR_64_RegMask;
+ }
+
+ // Unlike getCalleeSavedRegs(), we don't have MMI so we can't check
+ // callsEHReturn().
+ if (Is64Bit) {
+ if (IsWin64)
+ return CSR_Win64_RegMask;
+ return CSR_64_RegMask;
+ }
+ return CSR_32_RegMask;
+}
+
+const uint32_t*
+X86RegisterInfo::getNoPreservedMask() const {
+ return CSR_NoRegs_RegMask;
}
BitVector X86RegisterInfo::getReservedRegs(const MachineFunction &MF) const {
BitVector Reserved(getNumRegs());
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
+ const X86FrameLowering *TFI = getFrameLowering(MF);
// Set the stack-pointer register and its aliases as reserved.
- Reserved.set(X86::RSP);
- for (MCSubRegIterator I(X86::RSP, this); I.isValid(); ++I)
+ for (MCSubRegIterator I(X86::RSP, this, /*IncludeSelf=*/true); I.isValid();
+ ++I)
Reserved.set(*I);
// Set the instruction pointer register and its aliases as reserved.
- Reserved.set(X86::RIP);
- for (MCSubRegIterator I(X86::RIP, this); I.isValid(); ++I)
+ for (MCSubRegIterator I(X86::RIP, this, /*IncludeSelf=*/true); I.isValid();
+ ++I)
Reserved.set(*I);
// Set the frame-pointer register and its aliases as reserved if needed.
if (TFI->hasFP(MF)) {
- Reserved.set(X86::RBP);
- for (MCSubRegIterator I(X86::RBP, this); I.isValid(); ++I)
+ for (MCSubRegIterator I(X86::RBP, this, /*IncludeSelf=*/true); I.isValid();
+ ++I)
+ Reserved.set(*I);
+ }
+
+ // Set the base-pointer register and its aliases as reserved if needed.
+ if (hasBasePointer(MF)) {
+ CallingConv::ID CC = MF.getFunction()->getCallingConv();
+ const uint32_t *RegMask = getCallPreservedMask(MF, CC);
+ if (MachineOperand::clobbersPhysReg(RegMask, getBaseRegister()))
+ report_fatal_error(
+ "Stack realignment in presence of dynamic allocas is not supported with"
+ "this calling convention.");
+
+ unsigned BasePtr = getX86SubSuperRegister(getBaseRegister(), MVT::i64,
+ false);
+ for (MCSubRegIterator I(BasePtr, this, /*IncludeSelf=*/true);
+ I.isValid(); ++I)
Reserved.set(*I);
}
Reserved.set(X86::GS);
// Mark the floating point stack registers as reserved.
- Reserved.set(X86::ST0);
- Reserved.set(X86::ST1);
- Reserved.set(X86::ST2);
- Reserved.set(X86::ST3);
- Reserved.set(X86::ST4);
- Reserved.set(X86::ST5);
- Reserved.set(X86::ST6);
- Reserved.set(X86::ST7);
+ for (unsigned n = 0; n != 8; ++n)
+ Reserved.set(X86::ST0 + n);
// Reserve the registers that only exist in 64-bit mode.
if (!Is64Bit) {
for (unsigned n = 0; n != 8; ++n) {
// R8, R9, ...
- static const uint16_t GPR64[] = {
- X86::R8, X86::R9, X86::R10, X86::R11,
- X86::R12, X86::R13, X86::R14, X86::R15
- };
- for (MCRegAliasIterator AI(GPR64[n], this, true); AI.isValid(); ++AI)
+ for (MCRegAliasIterator AI(X86::R8 + n, this, true); AI.isValid(); ++AI)
Reserved.set(*AI);
// XMM8, XMM9, ...
- assert(X86::XMM15 == X86::XMM8+7);
for (MCRegAliasIterator AI(X86::XMM8 + n, this, true); AI.isValid(); ++AI)
Reserved.set(*AI);
}
}
+ if (!Is64Bit || !MF.getSubtarget<X86Subtarget>().hasAVX512()) {
+ for (unsigned n = 16; n != 32; ++n) {
+ for (MCRegAliasIterator AI(X86::XMM0 + n, this, true); AI.isValid(); ++AI)
+ Reserved.set(*AI);
+ }
+ }
return Reserved;
}
+void X86RegisterInfo::adjustStackMapLiveOutMask(uint32_t *Mask) const {
+ // Check if the EFLAGS register is marked as live-out. This shouldn't happen,
+ // because the calling convention defines the EFLAGS register as NOT
+ // preserved.
+ //
+ // Unfortunatelly the EFLAGS show up as live-out after branch folding. Adding
+ // an assert to track this and clear the register afterwards to avoid
+ // unnecessary crashes during release builds.
+ assert(!(Mask[X86::EFLAGS / 32] & (1U << (X86::EFLAGS % 32))) &&
+ "EFLAGS are not live-out from a patchpoint.");
+
+ // Also clean other registers that don't need preserving (IP).
+ for (auto Reg : {X86::EFLAGS, X86::RIP, X86::EIP, X86::IP})
+ Mask[Reg / 32] &= ~(1U << (Reg % 32));
+}
+
//===----------------------------------------------------------------------===//
// Stack Frame Processing methods
//===----------------------------------------------------------------------===//
-bool X86RegisterInfo::canRealignStack(const MachineFunction &MF) const {
- const MachineFrameInfo *MFI = MF.getFrameInfo();
- return (MF.getTarget().Options.RealignStack &&
- !MFI->hasVarSizedObjects());
+static bool CantUseSP(const MachineFrameInfo *MFI) {
+ return MFI->hasVarSizedObjects() || MFI->hasOpaqueSPAdjustment();
}
-bool X86RegisterInfo::needsStackRealignment(const MachineFunction &MF) const {
- const MachineFrameInfo *MFI = MF.getFrameInfo();
- const Function *F = MF.getFunction();
- unsigned StackAlign = TM.getFrameLowering()->getStackAlignment();
- bool requiresRealignment = ((MFI->getMaxAlignment() > StackAlign) ||
- F->hasFnAttr(Attribute::StackAlignment));
-
- // FIXME: Currently we don't support stack realignment for functions with
- // variable-sized allocas.
- // FIXME: It's more complicated than this...
- if (0 && requiresRealignment && MFI->hasVarSizedObjects())
- report_fatal_error(
- "Stack realignment in presence of dynamic allocas is not supported");
-
- // If we've requested that we force align the stack do so now.
- if (ForceStackAlign)
- return canRealignStack(MF);
-
- return requiresRealignment && canRealignStack(MF);
-}
+bool X86RegisterInfo::hasBasePointer(const MachineFunction &MF) const {
+ const MachineFrameInfo *MFI = MF.getFrameInfo();
-bool X86RegisterInfo::hasReservedSpillSlot(const MachineFunction &MF,
- unsigned Reg, int &FrameIdx) const {
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
+ if (!EnableBasePointer)
+ return false;
- if (Reg == FramePtr && TFI->hasFP(MF)) {
- FrameIdx = MF.getFrameInfo()->getObjectIndexBegin();
- return true;
- }
- return false;
+ // When we need stack realignment, we can't address the stack from the frame
+ // pointer. When we have dynamic allocas or stack-adjusting inline asm, we
+ // can't address variables from the stack pointer. MS inline asm can
+ // reference locals while also adjusting the stack pointer. When we can't
+ // use both the SP and the FP, we need a separate base pointer register.
+ bool CantUseFP = needsStackRealignment(MF);
+ return CantUseFP && CantUseSP(MFI);
}
-static unsigned getSUBriOpcode(unsigned is64Bit, int64_t Imm) {
- if (is64Bit) {
- if (isInt<8>(Imm))
- return X86::SUB64ri8;
- return X86::SUB64ri32;
- } else {
- if (isInt<8>(Imm))
- return X86::SUB32ri8;
- return X86::SUB32ri;
- }
-}
+bool X86RegisterInfo::canRealignStack(const MachineFunction &MF) const {
+ if (!TargetRegisterInfo::canRealignStack(MF))
+ return false;
-static unsigned getADDriOpcode(unsigned is64Bit, int64_t Imm) {
- if (is64Bit) {
- if (isInt<8>(Imm))
- return X86::ADD64ri8;
- return X86::ADD64ri32;
- } else {
- if (isInt<8>(Imm))
- return X86::ADD32ri8;
- return X86::ADD32ri;
- }
+ const MachineFrameInfo *MFI = MF.getFrameInfo();
+ const MachineRegisterInfo *MRI = &MF.getRegInfo();
+
+ // Stack realignment requires a frame pointer. If we already started
+ // register allocation with frame pointer elimination, it is too late now.
+ if (!MRI->canReserveReg(FramePtr))
+ return false;
+
+ // If a base pointer is necessary. Check that it isn't too late to reserve
+ // it.
+ if (CantUseSP(MFI))
+ return MRI->canReserveReg(BasePtr);
+ return true;
}
-void X86RegisterInfo::
-eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
- MachineBasicBlock::iterator I) const {
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
- bool reseveCallFrame = TFI->hasReservedCallFrame(MF);
- int Opcode = I->getOpcode();
- bool isDestroy = Opcode == TII.getCallFrameDestroyOpcode();
- DebugLoc DL = I->getDebugLoc();
- uint64_t Amount = !reseveCallFrame ? I->getOperand(0).getImm() : 0;
- uint64_t CalleeAmt = isDestroy ? I->getOperand(1).getImm() : 0;
- I = MBB.erase(I);
-
- if (!reseveCallFrame) {
- // If the stack pointer can be changed after prologue, turn the
- // adjcallstackup instruction into a 'sub ESP, <amt>' and the
- // adjcallstackdown instruction into 'add ESP, <amt>'
- // TODO: consider using push / pop instead of sub + store / add
- if (Amount == 0)
- return;
-
- // We need to keep the stack aligned properly. To do this, we round the
- // amount of space needed for the outgoing arguments up to the next
- // alignment boundary.
- unsigned StackAlign = TM.getFrameLowering()->getStackAlignment();
- Amount = (Amount + StackAlign - 1) / StackAlign * StackAlign;
-
- MachineInstr *New = 0;
- if (Opcode == TII.getCallFrameSetupOpcode()) {
- New = BuildMI(MF, DL, TII.get(getSUBriOpcode(Is64Bit, Amount)),
- StackPtr)
- .addReg(StackPtr)
- .addImm(Amount);
- } else {
- assert(Opcode == TII.getCallFrameDestroyOpcode());
-
- // Factor out the amount the callee already popped.
- Amount -= CalleeAmt;
-
- if (Amount) {
- unsigned Opc = getADDriOpcode(Is64Bit, Amount);
- New = BuildMI(MF, DL, TII.get(Opc), StackPtr)
- .addReg(StackPtr).addImm(Amount);
- }
- }
-
- if (New) {
- // The EFLAGS implicit def is dead.
- New->getOperand(3).setIsDead();
-
- // Replace the pseudo instruction with a new instruction.
- MBB.insert(I, New);
- }
-
- return;
- }
-
- if (Opcode == TII.getCallFrameDestroyOpcode() && CalleeAmt) {
- // If we are performing frame pointer elimination and if the callee pops
- // something off the stack pointer, add it back. We do this until we have
- // more advanced stack pointer tracking ability.
- unsigned Opc = getSUBriOpcode(Is64Bit, CalleeAmt);
- MachineInstr *New = BuildMI(MF, DL, TII.get(Opc), StackPtr)
- .addReg(StackPtr).addImm(CalleeAmt);
-
- // The EFLAGS implicit def is dead.
- New->getOperand(3).setIsDead();
-
- // We are not tracking the stack pointer adjustment by the callee, so make
- // sure we restore the stack pointer immediately after the call, there may
- // be spill code inserted between the CALL and ADJCALLSTACKUP instructions.
- MachineBasicBlock::iterator B = MBB.begin();
- while (I != B && !llvm::prior(I)->isCall())
- --I;
- MBB.insert(I, New);
- }
+bool X86RegisterInfo::hasReservedSpillSlot(const MachineFunction &MF,
+ unsigned Reg, int &FrameIdx) const {
+ // Since X86 defines assignCalleeSavedSpillSlots which always return true
+ // this function neither used nor tested.
+ llvm_unreachable("Unused function on X86. Otherwise need a test case.");
}
void
X86RegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II,
- int SPAdj, RegScavenger *RS) const{
- assert(SPAdj == 0 && "Unexpected");
-
- unsigned i = 0;
+ int SPAdj, unsigned FIOperandNum,
+ RegScavenger *RS) const {
MachineInstr &MI = *II;
MachineFunction &MF = *MI.getParent()->getParent();
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
-
- while (!MI.getOperand(i).isFI()) {
- ++i;
- assert(i < MI.getNumOperands() && "Instr doesn't have FrameIndex operand!");
- }
-
- int FrameIndex = MI.getOperand(i).getIndex();
+ const X86FrameLowering *TFI = getFrameLowering(MF);
+ int FrameIndex = MI.getOperand(FIOperandNum).getIndex();
unsigned BasePtr;
unsigned Opc = MI.getOpcode();
- bool AfterFPPop = Opc == X86::TAILJMPm64 || Opc == X86::TAILJMPm;
- if (needsStackRealignment(MF))
+ bool AfterFPPop = Opc == X86::TAILJMPm64 || Opc == X86::TAILJMPm ||
+ Opc == X86::TCRETURNmi || Opc == X86::TCRETURNmi64;
+
+ if (hasBasePointer(MF))
+ BasePtr = (FrameIndex < 0 ? FramePtr : getBaseRegister());
+ else if (needsStackRealignment(MF))
BasePtr = (FrameIndex < 0 ? FramePtr : StackPtr);
else if (AfterFPPop)
BasePtr = StackPtr;
else
BasePtr = (TFI->hasFP(MF) ? FramePtr : StackPtr);
+ // LOCAL_ESCAPE uses a single offset, with no register. It only works in the
+ // simple FP case, and doesn't work with stack realignment. On 32-bit, the
+ // offset is from the traditional base pointer location. On 64-bit, the
+ // offset is from the SP at the end of the prologue, not the FP location. This
+ // matches the behavior of llvm.frameaddress.
+ unsigned IgnoredFrameReg;
+ if (Opc == TargetOpcode::LOCAL_ESCAPE) {
+ MachineOperand &FI = MI.getOperand(FIOperandNum);
+ int Offset;
+ Offset = TFI->getFrameIndexReference(MF, FrameIndex, IgnoredFrameReg);
+ FI.ChangeToImmediate(Offset);
+ return;
+ }
+
+ // For LEA64_32r when BasePtr is 32-bits (X32) we can use full-size 64-bit
+ // register as source operand, semantic is the same and destination is
+ // 32-bits. It saves one byte per lea in code since 0x67 prefix is avoided.
+ if (Opc == X86::LEA64_32r && X86::GR32RegClass.contains(BasePtr))
+ BasePtr = getX86SubSuperRegister(BasePtr, MVT::i64, false);
+
// This must be part of a four operand memory reference. Replace the
// FrameIndex with base register with EBP. Add an offset to the offset.
- MI.getOperand(i).ChangeToRegister(BasePtr, false);
+ MI.getOperand(FIOperandNum).ChangeToRegister(BasePtr, false);
// Now add the frame object offset to the offset from EBP.
int FIOffset;
const MachineFrameInfo *MFI = MF.getFrameInfo();
FIOffset = MFI->getObjectOffset(FrameIndex) - TFI->getOffsetOfLocalArea();
} else
- FIOffset = TFI->getFrameIndexOffset(MF, FrameIndex);
+ FIOffset = TFI->getFrameIndexReference(MF, FrameIndex, IgnoredFrameReg);
+
+ if (BasePtr == StackPtr)
+ FIOffset += SPAdj;
- if (MI.getOperand(i+3).isImm()) {
+ // The frame index format for stackmaps and patchpoints is different from the
+ // X86 format. It only has a FI and an offset.
+ if (Opc == TargetOpcode::STACKMAP || Opc == TargetOpcode::PATCHPOINT) {
+ assert(BasePtr == FramePtr && "Expected the FP as base register");
+ int64_t Offset = MI.getOperand(FIOperandNum + 1).getImm() + FIOffset;
+ MI.getOperand(FIOperandNum + 1).ChangeToImmediate(Offset);
+ return;
+ }
+
+ if (MI.getOperand(FIOperandNum+3).isImm()) {
// Offset is a 32-bit integer.
- int Imm = (int)(MI.getOperand(i + 3).getImm());
+ int Imm = (int)(MI.getOperand(FIOperandNum + 3).getImm());
int Offset = FIOffset + Imm;
assert((!Is64Bit || isInt<32>((long long)FIOffset + Imm)) &&
"Requesting 64-bit offset in 32-bit immediate!");
- MI.getOperand(i + 3).ChangeToImmediate(Offset);
+ MI.getOperand(FIOperandNum + 3).ChangeToImmediate(Offset);
} else {
// Offset is symbolic. This is extremely rare.
- uint64_t Offset = FIOffset + (uint64_t)MI.getOperand(i+3).getOffset();
- MI.getOperand(i+3).setOffset(Offset);
+ uint64_t Offset = FIOffset +
+ (uint64_t)MI.getOperand(FIOperandNum+3).getOffset();
+ MI.getOperand(FIOperandNum + 3).setOffset(Offset);
}
}
unsigned X86RegisterInfo::getFrameRegister(const MachineFunction &MF) const {
- const TargetFrameLowering *TFI = MF.getTarget().getFrameLowering();
+ const X86FrameLowering *TFI = getFrameLowering(MF);
return TFI->hasFP(MF) ? FramePtr : StackPtr;
}
-unsigned X86RegisterInfo::getEHExceptionRegister() const {
- llvm_unreachable("What is the exception register");
-}
-
-unsigned X86RegisterInfo::getEHHandlerRegister() const {
- llvm_unreachable("What is the exception handler register");
+unsigned
+X86RegisterInfo::getPtrSizedFrameRegister(const MachineFunction &MF) const {
+ const X86Subtarget &Subtarget = MF.getSubtarget<X86Subtarget>();
+ unsigned FrameReg = getFrameRegister(MF);
+ if (Subtarget.isTarget64BitILP32())
+ FrameReg = getX86SubSuperRegister(FrameReg, MVT::i32, false);
+ return FrameReg;
}
namespace llvm {
-unsigned getX86SubSuperRegister(unsigned Reg, EVT VT, bool High) {
- switch (VT.getSimpleVT().SimpleTy) {
- default: return Reg;
+unsigned getX86SubSuperRegisterOrZero(unsigned Reg, MVT::SimpleValueType VT,
+ bool High) {
+ switch (VT) {
+ default: return 0;
case MVT::i8:
if (High) {
switch (Reg) {
- default: return getX86SubSuperRegister(Reg, MVT::i64, High);
+ default: return getX86SubSuperRegister(Reg, MVT::i64);
+ case X86::SIL: case X86::SI: case X86::ESI: case X86::RSI:
+ return X86::SI;
+ case X86::DIL: case X86::DI: case X86::EDI: case X86::RDI:
+ return X86::DI;
+ case X86::BPL: case X86::BP: case X86::EBP: case X86::RBP:
+ return X86::BP;
+ case X86::SPL: case X86::SP: case X86::ESP: case X86::RSP:
+ return X86::SP;
case X86::AH: case X86::AL: case X86::AX: case X86::EAX: case X86::RAX:
return X86::AH;
case X86::DH: case X86::DL: case X86::DX: case X86::EDX: case X86::RDX:
}
case MVT::i16:
switch (Reg) {
- default: return Reg;
+ default: return 0;
case X86::AH: case X86::AL: case X86::AX: case X86::EAX: case X86::RAX:
return X86::AX;
case X86::DH: case X86::DL: case X86::DX: case X86::EDX: case X86::RDX:
}
case MVT::i32:
switch (Reg) {
- default: return Reg;
+ default: return 0;
case X86::AH: case X86::AL: case X86::AX: case X86::EAX: case X86::RAX:
return X86::EAX;
case X86::DH: case X86::DL: case X86::DX: case X86::EDX: case X86::RDX:
return X86::R15D;
}
case MVT::i64:
- // For 64-bit mode if we've requested a "high" register and the
- // Q or r constraints we want one of these high registers or
- // just the register name otherwise.
- if (High) {
- switch (Reg) {
- case X86::SIL: case X86::SI: case X86::ESI: case X86::RSI:
- return X86::SI;
- case X86::DIL: case X86::DI: case X86::EDI: case X86::RDI:
- return X86::DI;
- case X86::BPL: case X86::BP: case X86::EBP: case X86::RBP:
- return X86::BP;
- case X86::SPL: case X86::SP: case X86::ESP: case X86::RSP:
- return X86::SP;
- // Fallthrough.
- }
- }
switch (Reg) {
- default: return Reg;
+ default: return 0;
case X86::AH: case X86::AL: case X86::AX: case X86::EAX: case X86::RAX:
return X86::RAX;
case X86::DH: case X86::DL: case X86::DX: case X86::EDX: case X86::RDX:
}
}
}
-}
-namespace {
- struct MSAH : public MachineFunctionPass {
- static char ID;
- MSAH() : MachineFunctionPass(ID) {}
-
- virtual bool runOnMachineFunction(MachineFunction &MF) {
- const X86TargetMachine *TM =
- static_cast<const X86TargetMachine *>(&MF.getTarget());
- const TargetFrameLowering *TFI = TM->getFrameLowering();
- MachineRegisterInfo &RI = MF.getRegInfo();
- X86MachineFunctionInfo *FuncInfo = MF.getInfo<X86MachineFunctionInfo>();
- unsigned StackAlignment = TFI->getStackAlignment();
-
- // Be over-conservative: scan over all vreg defs and find whether vector
- // registers are used. If yes, there is a possibility that vector register
- // will be spilled and thus require dynamic stack realignment.
- for (unsigned i = 0, e = RI.getNumVirtRegs(); i != e; ++i) {
- unsigned Reg = TargetRegisterInfo::index2VirtReg(i);
- if (RI.getRegClass(Reg)->getAlignment() > StackAlignment) {
- FuncInfo->setForceFramePointer(true);
- return true;
- }
- }
- // Nothing to do
- return false;
- }
-
- virtual const char *getPassName() const {
- return "X86 Maximal Stack Alignment Check";
- }
-
- virtual void getAnalysisUsage(AnalysisUsage &AU) const {
- AU.setPreservesCFG();
- MachineFunctionPass::getAnalysisUsage(AU);
- }
- };
+unsigned getX86SubSuperRegister(unsigned Reg, MVT::SimpleValueType VT,
+ bool High) {
+ unsigned Res = getX86SubSuperRegisterOrZero(Reg, VT, High);
+ if (Res == 0)
+ llvm_unreachable("Unexpected register or VT");
+ return Res;
+}
- char MSAH::ID = 0;
+unsigned get512BitSuperRegister(unsigned Reg) {
+ if (Reg >= X86::XMM0 && Reg <= X86::XMM31)
+ return X86::ZMM0 + (Reg - X86::XMM0);
+ if (Reg >= X86::YMM0 && Reg <= X86::YMM31)
+ return X86::ZMM0 + (Reg - X86::YMM0);
+ if (Reg >= X86::ZMM0 && Reg <= X86::ZMM31)
+ return Reg;
+ llvm_unreachable("Unexpected SIMD register");
}
-FunctionPass*
-llvm::createX86MaxStackAlignmentHeuristicPass() { return new MSAH(); }
+}