SparcFrameLowering.cpp

Go to the documentation of this file.

//===-- SparcFrameLowering.cpp - Sparc Frame Information ------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file contains the Sparc implementation of TargetFrameLowering class.
//
//===----------------------------------------------------------------------===//
 
#include "SparcFrameLowering.h"
#include "SparcInstrInfo.h"
#include "SparcMachineFunctionInfo.h"
#include "SparcSubtarget.h"
#include "llvm/CodeGen/CFIInstBuilder.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineModuleInfo.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/RegisterScavenging.h"
 
using namespace llvm;
 
#define DEBUG_TYPE "sparc-frame-lowering"
 
SparcFrameLowering::SparcFrameLowering(const SparcSubtarget &ST)
    : TargetFrameLowering(TargetFrameLowering::StackGrowsDown,
                          ST.is64Bit() ? Align(16) : Align(8), 0,
                          ST.is64Bit() ? Align(16) : Align(8),
                          /*StackRealignable=*/false),
      STI(ST) {}
 
void SparcFrameLowering::emitSPAdjustment(MachineFunction &MF,
                                          MachineBasicBlock &MBB,
                                          MachineBasicBlock::iterator MBBI,
                                          const DebugLoc &dl, int NumBytes,
                                          unsigned ADDrr,
                                          unsigned ADDri) const {
 
  const SparcInstrInfo &TII =
      *static_cast<const SparcInstrInfo *>(MF.getSubtarget().getInstrInfo());
 
  if (NumBytes >= -4096 && NumBytes < 4096) {
    BuildMI(MBB, MBBI, dl, TII.get(ADDri), SP::O6)
      .addReg(SP::O6).addImm(NumBytes);
    return;
  }
 
  // Emit this the hard way.  This clobbers G1 which we always know is
  // available here.
  if (NumBytes >= 0) {
    // Emit nonnegative numbers with sethi + or.
    // sethi %hi(NumBytes), %g1
    // or %g1, %lo(NumBytes), %g1
    // add %sp, %g1, %sp
    BuildMI(MBB, MBBI, dl, TII.get(SP::SETHIi), SP::G1)
      .addImm(HI22(NumBytes));
    BuildMI(MBB, MBBI, dl, TII.get(SP::ORri), SP::G1)
      .addReg(SP::G1).addImm(LO10(NumBytes));
    BuildMI(MBB, MBBI, dl, TII.get(ADDrr), SP::O6)
      .addReg(SP::O6).addReg(SP::G1);
    return ;
  }
 
  // Emit negative numbers with sethi + xor.
  // sethi %hix(NumBytes), %g1
  // xor %g1, %lox(NumBytes), %g1
  // add %sp, %g1, %sp
  BuildMI(MBB, MBBI, dl, TII.get(SP::SETHIi), SP::G1)
    .addImm(HIX22(NumBytes));
  BuildMI(MBB, MBBI, dl, TII.get(SP::XORri), SP::G1)
    .addReg(SP::G1).addImm(LOX10(NumBytes));
  BuildMI(MBB, MBBI, dl, TII.get(ADDrr), SP::O6)
    .addReg(SP::O6).addReg(SP::G1);
}
 
void SparcFrameLowering::emitPrologue(MachineFunction &MF,
                                      MachineBasicBlock &MBB) const {
  SparcMachineFunctionInfo *FuncInfo = MF.getInfo<SparcMachineFunctionInfo>();
 
  assert(&MF.front() == &MBB && "Shrink-wrapping not yet supported");
  MachineFrameInfo &MFI = MF.getFrameInfo();
  const SparcSubtarget &Subtarget = MF.getSubtarget<SparcSubtarget>();
  MachineBasicBlock::iterator MBBI = MBB.begin();
 
  // Get the number of bytes to allocate from the FrameInfo
  int NumBytes = (int) MFI.getStackSize();
 
  unsigned SAVEri = SP::SAVEri;
  unsigned SAVErr = SP::SAVErr;
  if (FuncInfo->isLeafProc()) {
    if (NumBytes == 0)
      return;
    SAVEri = SP::ADDri;
    SAVErr = SP::ADDrr;
  }
 
  // The SPARC ABI is a bit odd in that it requires a reserved 92-byte
  // (128 in v9) area in the user's stack, starting at %sp. Thus, the
  // first part of the stack that can actually be used is located at
  // %sp + 92.
  //
  // We therefore need to add that offset to the total stack size
  // after all the stack objects are placed by
  // PrologEpilogInserter calculateFrameObjectOffsets. However, since the stack needs to be
  // aligned *after* the extra size is added, we need to disable
  // calculateFrameObjectOffsets's built-in stack alignment, by having
  // targetHandlesStackFrameRounding return true.
 
 
  // Add the extra call frame stack size, if needed. (This is the same
  // code as in PrologEpilogInserter, but also gets disabled by
  // targetHandlesStackFrameRounding)
  if (MFI.adjustsStack() && hasReservedCallFrame(MF))
    NumBytes += MFI.getMaxCallFrameSize();
 
  // Adds the SPARC subtarget-specific spill area to the stack
  // size. Also ensures target-required alignment.
  NumBytes = Subtarget.getAdjustedFrameSize(NumBytes);
 
  // Finally, ensure that the size is sufficiently aligned for the
  // data on the stack.
  NumBytes = alignTo(NumBytes, MFI.getMaxAlign());
 
  // Update stack size with corrected value.
  MFI.setStackSize(NumBytes);
 
  emitSPAdjustment(MF, MBB, MBBI, DebugLoc(), -NumBytes, SAVErr, SAVEri);
 
  if (MF.needsFrameMoves()) {
    CFIInstBuilder CFIBuilder(MBB, MBBI, MachineInstr::NoFlags);
    if (FuncInfo->isLeafProc()) {
      CFIBuilder.buildDefCFAOffset(NumBytes + Subtarget.getStackPointerBias());
    } else {
      CFIBuilder.buildDefCFARegister(SP::I6);
      CFIBuilder.buildWindowSave();
      CFIBuilder.buildRegister(SP::O7, SP::I7);
    }
  }
}
 
MachineBasicBlock::iterator SparcFrameLowering::
eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
                              MachineBasicBlock::iterator I) const {
  if (!hasReservedCallFrame(MF)) {
    MachineInstr &MI = *I;
    int Size = MI.getOperand(0).getImm();
    if (MI.getOpcode() == SP::ADJCALLSTACKDOWN)
      Size = -Size;
 
    if (Size)
      emitSPAdjustment(MF, MBB, I, DebugLoc(), Size, SP::ADDrr, SP::ADDri);
  }
  return MBB.erase(I);
}
 
 
void SparcFrameLowering::emitEpilogue(MachineFunction &MF,
                                  MachineBasicBlock &MBB) const {
  SparcMachineFunctionInfo *FuncInfo = MF.getInfo<SparcMachineFunctionInfo>();
  MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
  const SparcInstrInfo &TII =
      *static_cast<const SparcInstrInfo *>(MF.getSubtarget().getInstrInfo());
  DebugLoc dl = MBBI->getDebugLoc();
  assert((MBBI->getOpcode() == SP::RETL || MBBI->getOpcode() == SP::TAIL_CALL ||
          MBBI->getOpcode() == SP::TAIL_CALLri) &&
         "Can only put epilog before 'retl' or 'tail_call' instruction!");
  if (!FuncInfo->isLeafProc()) {
    BuildMI(MBB, MBBI, dl, TII.get(SP::RESTORErr), SP::G0)
        .addReg(SP::G0)
        .addReg(SP::G0)
        .setMIFlag(MachineInstr::FrameDestroy);
    return;
  }
  MachineFrameInfo &MFI = MF.getFrameInfo();
 
  int NumBytes = (int) MFI.getStackSize();
  if (NumBytes != 0)
    emitSPAdjustment(MF, MBB, MBBI, dl, NumBytes, SP::ADDrr, SP::ADDri);
 
  // Preserve return address in %o7
  if (MBBI->getOpcode() == SP::TAIL_CALL) {
    MBB.addLiveIn(SP::O7);
    BuildMI(MBB, MBBI, dl, TII.get(SP::ORrr), SP::G1)
        .addReg(SP::G0)
        .addReg(SP::O7);
    BuildMI(MBB, MBBI, dl, TII.get(SP::ORrr), SP::O7)
        .addReg(SP::G0)
        .addReg(SP::G1)
        .setMIFlag(MachineInstr::FrameDestroy);
  }
}
 
bool SparcFrameLowering::hasReservedCallFrame(const MachineFunction &MF) const {
  // Reserve call frame if there are no variable sized objects on the stack.
  return !MF.getFrameInfo().hasVarSizedObjects();
}
 
// hasFPImpl - Return true if the specified function should have a dedicated
// frame pointer register.  This is true if the function has variable sized
// allocas or if frame pointer elimination is disabled.
bool SparcFrameLowering::hasFPImpl(const MachineFunction &MF) const {
  const MachineFrameInfo &MFI = MF.getFrameInfo();
  return MF.disableFramePointerElim() || MFI.hasVarSizedObjects() ||
         MFI.isFrameAddressTaken();
}
 
StackOffset
SparcFrameLowering::getFrameIndexReference(const MachineFunction &MF, int FI,
                                           Register &FrameReg) const {
  const SparcSubtarget &Subtarget = MF.getSubtarget<SparcSubtarget>();
  const MachineFrameInfo &MFI = MF.getFrameInfo();
  const SparcRegisterInfo *RegInfo = Subtarget.getRegisterInfo();
  const SparcMachineFunctionInfo *FuncInfo = MF.getInfo<SparcMachineFunctionInfo>();
  bool isFixed = MFI.isFixedObjectIndex(FI);
 
  // Addressable stack objects are accessed using neg. offsets from
  // %fp, or positive offsets from %sp.
  bool UseFP;
 
  // Sparc uses FP-based references in general, even when "hasFP" is
  // false. That function is rather a misnomer, because %fp is
  // actually always available, unless isLeafProc.
  if (FuncInfo->isLeafProc()) {
    // If there's a leaf proc, all offsets need to be %sp-based,
    // because we haven't caused %fp to actually point to our frame.
    UseFP = false;
  } else if (isFixed) {
    // Otherwise, argument access should always use %fp.
    UseFP = true;
  } else {
    // Finally, default to using %fp.
    UseFP = true;
  }
 
  int64_t FrameOffset = MF.getFrameInfo().getObjectOffset(FI) +
      Subtarget.getStackPointerBias();
 
  if (UseFP) {
    FrameReg = RegInfo->getFrameRegister(MF);
    return StackOffset::getFixed(FrameOffset);
  } else {
    FrameReg = SP::O6; // %sp
    return StackOffset::getFixed(FrameOffset + MF.getFrameInfo().getStackSize());
  }
}
 
[[maybe_unused]] static bool verifyLeafProcRegUse(MachineRegisterInfo *MRI) {
 
  for (unsigned reg = SP::I0; reg <= SP::I7; ++reg)
    if (MRI->isPhysRegUsed(reg))
      return false;
 
  for (unsigned reg = SP::L0; reg <= SP::L7; ++reg)
    if (MRI->isPhysRegUsed(reg))
      return false;
 
  return true;
}
 
bool SparcFrameLowering::isLeafProc(MachineFunction &MF) const
{
 
  MachineRegisterInfo &MRI = MF.getRegInfo();
  MachineFrameInfo    &MFI = MF.getFrameInfo();
  SparcMachineFunctionInfo *SMFI = MF.getInfo<SparcMachineFunctionInfo>();
 
  return !(MFI.hasCalls()               // has calls
           || MRI.isPhysRegUsed(SP::L0) // Too many registers needed
           || MRI.isPhysRegUsed(SP::O6) // %sp is used
           || hasFP(MF)                 // need %fp
           || MF.hasInlineAsm()         // has inline assembly
           || SMFI->isLargeProc());     // is a large procedure
}
 
void SparcFrameLowering::remapRegsForLeafProc(MachineFunction &MF) const {
  MachineRegisterInfo &MRI = MF.getRegInfo();
  SmallVector<MCRegister, 2> UntrackedLiveIns;
 
  // 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;
 
    // Reserved registers are not tracked by liveness
    if (MRI.isReserved(reg) && !MRI.isReserved(mapped_reg))
      UntrackedLiveIns.push_back(mapped_reg);
 
    // Replace I register with O register.
    MRI.replaceRegWith(reg, mapped_reg);
 
    // Also replace register pair super-registers.
    if ((reg - SP::I0) % 2 == 0) {
      unsigned preg = (reg - SP::I0) / 2 + SP::I0_I1;
      unsigned mapped_preg = preg - SP::I0_I1 + SP::O0_O1;
      MRI.replaceRegWith(preg, mapped_preg);
    }
  }
 
  // Rewrite MBB's Live-ins.
  for (MachineBasicBlock &MBB : MF) {
    for (unsigned reg = SP::I0_I1; reg <= SP::I6_I7; ++reg) {
      if (!MBB.isLiveIn(reg))
        continue;
      MBB.removeLiveIn(reg);
      MBB.addLiveIn(reg - SP::I0_I1 + SP::O0_O1);
    }
    for (unsigned reg = SP::I0; reg <= SP::I7; ++reg) {
      if (!MBB.isLiveIn(reg))
        continue;
      MBB.removeLiveIn(reg);
      MBB.addLiveIn(reg - SP::I0 + SP::O0);
    }
    for (MCRegister Reg : UntrackedLiveIns)
      if (!MBB.isLiveIn(Reg))
        MBB.addLiveIn(Reg);
  }
  for (MCRegister Reg : UntrackedLiveIns)
    MRI.clearKillFlags(Reg);
 
  assert(verifyLeafProcRegUse(&MRI));
#ifdef EXPENSIVE_CHECKS
  MF.verify(0, "After LeafProc Remapping");
#endif
}
 
static uint64_t estimateFunctionSizeInBytes(const SparcInstrInfo *TII,
                                            const MachineFunction &MF) {
  uint64_t FuncSize = 0;
  for (auto &MBB : MF)
    for (auto &MI : MBB) {
      // Be conservative here and count the DS too.
      uint64_t Multiplier = MI.hasDelaySlot() ? 2 : 1;
      FuncSize += (TII->getInstSizeInBytes(MI) * Multiplier);
    }
  return FuncSize;
}
 
void SparcFrameLowering::determineCalleeSaves(MachineFunction &MF,
                                              BitVector &SavedRegs,
                                              RegScavenger *RS) const {
  TargetFrameLowering::determineCalleeSaves(MF, SavedRegs, RS);
  SparcMachineFunctionInfo *MFI = MF.getInfo<SparcMachineFunctionInfo>();
  const SparcInstrInfo *TII = STI.getInstrInfo();
  const SparcSubtarget &Subtarget = MF.getSubtarget<SparcSubtarget>();
 
  bool IsLargeFunction = !isInt<23>(estimateFunctionSizeInBytes(TII, MF));
  MFI->setLargeProc(IsLargeFunction);
 
  if (!Subtarget.getCLOpts().disable_sparc_leaf_proc && isLeafProc(MF)) {
    MFI->setLeafProc(true);
 
    remapRegsForLeafProc(MF);
  }
}
 
void SparcFrameLowering::processFunctionBeforeFrameFinalized(
    MachineFunction &MF, RegScavenger *RS) const {
  const SparcRegisterInfo *RI = STI.getRegisterInfo();
  const TargetRegisterClass &RC =
      STI.is64Bit() ? SP::I64RegsRegClass : SP::IntRegsRegClass;
  MachineFrameInfo &MFI = MF.getFrameInfo();
  SparcMachineFunctionInfo *SMFI = MF.getInfo<SparcMachineFunctionInfo>();
 
  // Far branches beyond 23-bit offset require a spill slot for scratch
  // register.
  if (SMFI->isLargeProc()) {
    int FI =
        MFI.CreateSpillStackObject(RI->getSpillSize(RC), RI->getSpillAlign(RC));
    RS->addScavengingFrameIndex(FI);
    if (SMFI->getBranchRelaxationSpillFrameIndex() == -1)
      SMFI->setBranchRelaxationSpillFrameIndex(FI);
    LLVM_DEBUG(dbgs() << "Allocated FI(" << FI
                      << ") as the emergency spill slot.\n");
  }
}
 
void SparcFrameLowering::processFunctionBeforeFrameIndicesReplaced(
    MachineFunction &MF, RegScavenger *RS) const {
  if (!MF.getInfo<SparcMachineFunctionInfo>()->isLeafProc())
    return;
 
  // PEI takes the entry block's parameter DBG_VALUEs out of the function
  // while determineCalleeSaves runs, so remapRegsForLeafProc missed them.
  MachineRegisterInfo &MRI = MF.getRegInfo();
  auto Remap = [&MRI](unsigned Reg, unsigned MappedReg) {
    for (MachineOperand &MO : make_early_inc_range(MRI.reg_operands(Reg)))
      if (MO.isDebug())
        MO.setReg(MappedReg);
  };
  for (unsigned Reg = SP::I0; Reg <= SP::I7; ++Reg)
    Remap(Reg, Reg - SP::I0 + SP::O0);
  for (unsigned Reg = SP::I0_I1; Reg <= SP::I6_I7; ++Reg)
    Remap(Reg, Reg - SP::I0_I1 + SP::O0_O1);
}