xref: /freebsd/contrib/llvm-project/llvm/lib/Target/X86/X86FixupSetCC.cpp (revision 480093f4440d54b30b3025afeac24b48f2ba7a2e)

//===---- X86FixupSetCC.cpp - optimize usage of LEA instructions ----------===//
//
// 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 defines a pass that fixes zero-extension of setcc patterns.
// X86 setcc instructions are modeled to have no input arguments, and a single
// GR8 output argument. This is consistent with other similar instructions
// (e.g. movb), but means it is impossible to directly generate a setcc into
// the lower GR8 of a specified GR32.
// This means that ISel must select (zext (setcc)) into something like
// seta %al; movzbl %al, %eax.
// Unfortunately, this can cause a stall due to the partial register write
// performed by the setcc. Instead, we can use:
// xor %eax, %eax; seta %al
// This both avoids the stall, and encodes shorter.
//===----------------------------------------------------------------------===//

#include "X86.h"
#include "X86InstrInfo.h"
#include "X86Subtarget.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"

using namespace llvm;

#define DEBUG_TYPE "x86-fixup-setcc"

STATISTIC(NumSubstZexts, "Number of setcc + zext pairs substituted");

namespace {
class X86FixupSetCCPass : public MachineFunctionPass {
public:
  X86FixupSetCCPass() : MachineFunctionPass(ID) {}

  StringRef getPassName() const override { return "X86 Fixup SetCC"; }

  bool runOnMachineFunction(MachineFunction &MF) override;

private:
  MachineRegisterInfo *MRI = nullptr;
  const X86InstrInfo *TII = nullptr;

  enum { SearchBound = 16 };

  static char ID;
};

char X86FixupSetCCPass::ID = 0;
}

FunctionPass *llvm::createX86FixupSetCC() { return new X86FixupSetCCPass(); }

bool X86FixupSetCCPass::runOnMachineFunction(MachineFunction &MF) {
  bool Changed = false;
  MRI = &MF.getRegInfo();
  TII = MF.getSubtarget<X86Subtarget>().getInstrInfo();

  SmallVector<MachineInstr*, 4> ToErase;

  for (auto &MBB : MF) {
    MachineInstr *FlagsDefMI = nullptr;
    for (auto &MI : MBB) {
      // Remember the most recent preceding eflags defining instruction.
      if (MI.definesRegister(X86::EFLAGS))
        FlagsDefMI = &MI;

      // Find a setcc that is used by a zext.
      // This doesn't have to be the only use, the transformation is safe
      // regardless.
      if (MI.getOpcode() != X86::SETCCr)
        continue;

      MachineInstr *ZExt = nullptr;
      for (auto &Use : MRI->use_instructions(MI.getOperand(0).getReg()))
        if (Use.getOpcode() == X86::MOVZX32rr8)
          ZExt = &Use;

      if (!ZExt)
        continue;

      if (!FlagsDefMI)
        continue;

      // We'd like to put something that clobbers eflags directly before
      // FlagsDefMI. This can't hurt anything after FlagsDefMI, because
      // it, itself, by definition, clobbers eflags. But it may happen that
      // FlagsDefMI also *uses* eflags, in which case the transformation is
      // invalid.
      if (FlagsDefMI->readsRegister(X86::EFLAGS))
        continue;

      ++NumSubstZexts;
      Changed = true;

      // On 32-bit, we need to be careful to force an ABCD register.
      const TargetRegisterClass *RC = MF.getSubtarget<X86Subtarget>().is64Bit()
                                          ? &X86::GR32RegClass
                                          : &X86::GR32_ABCDRegClass;
      Register ZeroReg = MRI->createVirtualRegister(RC);
      Register InsertReg = MRI->createVirtualRegister(RC);

      // Initialize a register with 0. This must go before the eflags def
      BuildMI(MBB, FlagsDefMI, MI.getDebugLoc(), TII->get(X86::MOV32r0),
              ZeroReg);

      // X86 setcc only takes an output GR8, so fake a GR32 input by inserting
      // the setcc result into the low byte of the zeroed register.
      BuildMI(*ZExt->getParent(), ZExt, ZExt->getDebugLoc(),
              TII->get(X86::INSERT_SUBREG), InsertReg)
          .addReg(ZeroReg)
          .addReg(MI.getOperand(0).getReg())
          .addImm(X86::sub_8bit);
      MRI->replaceRegWith(ZExt->getOperand(0).getReg(), InsertReg);
      ToErase.push_back(ZExt);
    }
  }

  for (auto &I : ToErase)
    I->eraseFromParent();

  return Changed;
}