#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/ValueTypes.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"
#define GET_REGINFO_TARGET_DESC
#include "X86GenRegisterInfo.inc"
" needed for the function."),
cl::init(false), cl::Hidden);
+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"));
+
X86RegisterInfo::X86RegisterInfo(X86TargetMachine &tm,
const TargetInstrInfo &tii)
- : X86GenRegisterInfo(tm.getSubtarget<X86Subtarget>().is64Bit()
- ? X86::RIP : X86::EIP,
+ : X86GenRegisterInfo((tm.getSubtarget<X86Subtarget>().is64Bit()
+ ? X86::RIP : X86::EIP),
X86_MC::getDwarfRegFlavour(tm.getTargetTriple(), false),
- X86_MC::getDwarfRegFlavour(tm.getTargetTriple(), true)),
+ X86_MC::getDwarfRegFlavour(tm.getTargetTriple(), true),
+ (tm.getSubtarget<X86Subtarget>().is64Bit()
+ ? X86::RIP : X86::EIP)),
TM(tm), TII(tii) {
X86_MC::InitLLVM2SEHRegisterMapping(this);
StackPtr = X86::ESP;
FramePtr = X86::EBP;
}
+ // 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.
+ BasePtr = Is64Bit ? X86::RBX : X86::ESI;
}
/// getCompactUnwindRegNum - This function maps the register to the number for
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 *
return &X86::GR64_TCW64RegClass;
if (TM.getSubtarget<X86Subtarget>().is64Bit())
return &X86::GR64_TCRegClass;
+
+ const Function *F = MF.getFunction();
+ bool hasHipeCC = (F ? F->getCallingConv() == CallingConv::HiPE : false);
+ if (hasHipeCC)
+ return &X86::GR32RegClass;
return &X86::GR32_TCRegClass;
}
}
X86RegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
bool callsEHReturn = false;
bool ghcCall = false;
+ bool oclBiCall = false;
+ bool hipeCall = false;
+ bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
if (MF) {
callsEHReturn = MF->getMMI().callsEHReturn();
const Function *F = MF->getFunction();
ghcCall = (F ? F->getCallingConv() == CallingConv::GHC : false);
+ oclBiCall = (F ? F->getCallingConv() == CallingConv::Intel_OCL_BI : false);
+ hipeCall = (F ? F->getCallingConv() == CallingConv::HiPE : false);
}
- if (ghcCall)
+ if (ghcCall || hipeCall)
return CSR_NoRegs_SaveList;
+ if (oclBiCall) {
+ 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;
+ }
if (Is64Bit) {
if (IsWin64)
return CSR_Win64_SaveList;
const uint32_t*
X86RegisterInfo::getCallPreservedMask(CallingConv::ID CC) const {
- if (CC == CallingConv::GHC)
+ bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
+
+ if (CC == CallingConv::Intel_OCL_BI) {
+ if (IsWin64 && HasAVX)
+ return CSR_Win64_Intel_OCL_BI_AVX_RegMask;
+ if (Is64Bit && HasAVX)
+ return CSR_64_Intel_OCL_BI_AVX_RegMask;
+ if (!HasAVX && !IsWin64 && Is64Bit)
+ return CSR_64_Intel_OCL_BI_RegMask;
+ }
+ if (CC == CallingConv::GHC || CC == CallingConv::HiPE)
return CSR_NoRegs_RegMask;
if (!Is64Bit)
return CSR_32_RegMask;
return CSR_64_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();
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(CC);
+ if (MachineOperand::clobbersPhysReg(RegMask, getBaseRegister()))
+ report_fatal_error(
+ "Stack realignment in presence of dynamic allocas is not supported with"
+ "this calling convention.");
+
+ Reserved.set(getBaseRegister());
+ for (MCSubRegIterator I(getBaseRegister(), this); I.isValid(); ++I)
+ Reserved.set(*I);
+ }
+
// Mark the segment registers as reserved.
Reserved.set(X86::CS);
Reserved.set(X86::SS);
X86::R8, X86::R9, X86::R10, X86::R11,
X86::R12, X86::R13, X86::R14, X86::R15
};
- for (const uint16_t *AI = getOverlaps(GPR64[n]); unsigned Reg = *AI; ++AI)
- Reserved.set(Reg);
+ for (MCRegAliasIterator AI(GPR64[n], this, true); AI.isValid(); ++AI)
+ Reserved.set(*AI);
// XMM8, XMM9, ...
assert(X86::XMM15 == X86::XMM8+7);
- for (const uint16_t *AI = getOverlaps(X86::XMM8 + n); unsigned Reg = *AI;
- ++AI)
- Reserved.set(Reg);
+ for (MCRegAliasIterator AI(X86::XMM8 + n, this, true); AI.isValid(); ++AI)
+ Reserved.set(*AI);
}
}
// Stack Frame Processing methods
//===----------------------------------------------------------------------===//
+bool X86RegisterInfo::hasBasePointer(const MachineFunction &MF) const {
+ const MachineFrameInfo *MFI = MF.getFrameInfo();
+
+ if (!EnableBasePointer)
+ return false;
+
+ // When we need stack realignment and there are dynamic allocas, we can't
+ // reference off of the stack pointer, so we reserve a base pointer.
+ if (needsStackRealignment(MF) && MFI->hasVarSizedObjects())
+ return true;
+
+ return false;
+}
+
bool X86RegisterInfo::canRealignStack(const MachineFunction &MF) const {
const MachineFrameInfo *MFI = MF.getFrameInfo();
- return (MF.getTarget().Options.RealignStack &&
- !MFI->hasVarSizedObjects());
+ const MachineRegisterInfo *MRI = &MF.getRegInfo();
+ if (!MF.getTarget().Options.RealignStack)
+ return false;
+
+ // 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 (MFI->hasVarSizedObjects())
+ return MRI->canReserveReg(BasePtr);
+ return true;
}
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");
+ bool requiresRealignment =
+ ((MFI->getMaxAlignment() > StackAlign) ||
+ F->getAttributes().hasAttribute(AttributeSet::FunctionIndex,
+ Attribute::StackAlignment));
// If we've requested that we force align the stack do so now.
if (ForceStackAlign)
void
X86RegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II,
- int SPAdj, RegScavenger *RS) const{
+ int SPAdj, RegScavenger *RS) const {
assert(SPAdj == 0 && "Unexpected");
unsigned i = 0;
unsigned Opc = MI.getOpcode();
bool AfterFPPop = Opc == X86::TAILJMPm64 || Opc == X86::TAILJMPm;
- if (needsStackRealignment(MF))
+ if (hasBasePointer(MF))
+ BasePtr = (FrameIndex < 0 ? FramePtr : getBaseRegister());
+ else if (needsStackRealignment(MF))
BasePtr = (FrameIndex < 0 ? FramePtr : StackPtr);
else if (AfterFPPop)
BasePtr = StackPtr;
}
namespace llvm {
-unsigned getX86SubSuperRegister(unsigned Reg, EVT VT, bool High) {
- switch (VT.getSimpleVT().SimpleTy) {
- default: return Reg;
+unsigned getX86SubSuperRegister(unsigned Reg, MVT::SimpleValueType VT,
+ bool High) {
+ switch (VT) {
+ default: llvm_unreachable("Unexpected VT");
case MVT::i8:
if (High) {
switch (Reg) {
}
} else {
switch (Reg) {
- default: return 0;
+ default: llvm_unreachable("Unexpected register");
case X86::AH: case X86::AL: case X86::AX: case X86::EAX: case X86::RAX:
return X86::AL;
case X86::DH: case X86::DL: case X86::DX: case X86::EDX: case X86::RDX:
}
case MVT::i16:
switch (Reg) {
- default: return Reg;
+ default: llvm_unreachable("Unexpected register");
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: llvm_unreachable("Unexpected register");
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:
}
}
switch (Reg) {
- default: return Reg;
+ default: llvm_unreachable("Unexpected register");
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);
- }
- };
-
- char MSAH::ID = 0;
-}
-
-FunctionPass*
-llvm::createX86MaxStackAlignmentHeuristicPass() { return new MSAH(); }