// This pass expands a branch or jump instruction into a long branch if its
// offset is too large to fit into its immediate field.
//
-// FIXME:
-// 1. Fix pc-region jump instructions which cross 256MB segment boundaries.
-// 2. If program has inline assembly statements whose size cannot be
-// determined accurately, load branch target addresses from the GOT.
+// FIXME: Fix pc-region jump instructions which cross 256MB segment boundaries.
//===----------------------------------------------------------------------===//
-#define DEBUG_TYPE "mips-long-branch"
-
#include "Mips.h"
#include "MCTargetDesc/MipsBaseInfo.h"
+#include "MCTargetDesc/MipsMCNaCl.h"
+#include "MipsMachineFunction.h"
#include "MipsTargetMachine.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
using namespace llvm;
+#define DEBUG_TYPE "mips-long-branch"
+
STATISTIC(LongBranches, "Number of long branches.");
static cl::opt<bool> SkipLongBranch(
bool HasLongBranch;
MachineInstr *Br;
- MBBInfo() : Size(0), HasLongBranch(false), Br(0) {}
+ MBBInfo() : Size(0), HasLongBranch(false), Br(nullptr) {}
};
class MipsLongBranch : public MachineFunctionPass {
public:
static char ID;
MipsLongBranch(TargetMachine &tm)
- : MachineFunctionPass(ID), TM(tm),
- IsPIC(TM.getRelocationModel() == Reloc::PIC_),
- ABI(TM.getSubtarget<MipsSubtarget>().getTargetABI()),
- LongBranchSeqSize(!IsPIC ? 2 : (ABI == MipsSubtarget::N64 ? 13 : 9)) {}
+ : MachineFunctionPass(ID), TM(tm),
+ IsPIC(TM.getRelocationModel() == Reloc::PIC_),
+ ABI(static_cast<const MipsTargetMachine &>(TM).getABI()) {}
- virtual const char *getPassName() const {
+ const char *getPassName() const override {
return "Mips Long Branch";
}
- bool runOnMachineFunction(MachineFunction &F);
+ bool runOnMachineFunction(MachineFunction &F) override;
private:
void splitMBB(MachineBasicBlock *MBB);
MachineFunction *MF;
SmallVector<MBBInfo, 16> MBBInfos;
bool IsPIC;
- unsigned ABI;
+ MipsABIInfo ABI;
unsigned LongBranchSeqSize;
};
return MO.getMBB();
}
- assert(false && "This instruction does not have an MBB operand.");
- return 0;
+ llvm_unreachable("This instruction does not have an MBB operand.");
}
// Traverse the list of instructions backwards until a non-debug instruction is
MBBInfos.resize(MF->size());
const MipsInstrInfo *TII =
- static_cast<const MipsInstrInfo*>(TM.getInstrInfo());
+ static_cast<const MipsInstrInfo *>(MF->getSubtarget().getInstrInfo());
for (unsigned I = 0, E = MBBInfos.size(); I < E; ++I) {
MachineBasicBlock *MBB = MF->getBlockNumbered(I);
// MachineBasicBlock operand MBBOpnd.
void MipsLongBranch::replaceBranch(MachineBasicBlock &MBB, Iter Br,
DebugLoc DL, MachineBasicBlock *MBBOpnd) {
- const MipsInstrInfo *TII =
- static_cast<const MipsInstrInfo*>(TM.getInstrInfo());
+ const MipsInstrInfo *TII = static_cast<const MipsInstrInfo *>(
+ MBB.getParent()->getSubtarget().getInstrInfo());
unsigned NewOpc = TII->getOppositeBranchOpc(Br->getOpcode());
const MCInstrDesc &NewDesc = TII->get(NewOpc);
MIB.addMBB(MBBOpnd);
- // Bundle the instruction in the delay slot to the newly created branch
- // and erase the original branch.
- assert(Br->isBundledWithSucc());
- MachineBasicBlock::instr_iterator II(Br);
- MIBundleBuilder(&*MIB).append((++II)->removeFromBundle());
+ if (Br->hasDelaySlot()) {
+ // Bundle the instruction in the delay slot to the newly created branch
+ // and erase the original branch.
+ assert(Br->isBundledWithSucc());
+ MachineBasicBlock::instr_iterator II(Br);
+ MIBundleBuilder(&*MIB).append((++II)->removeFromBundle());
+ }
Br->eraseFromParent();
}
// Expand branch instructions to long branches.
+// TODO: This function has to be fixed for beqz16 and bnez16, because it
+// currently assumes that all branches have 16-bit offsets, and will produce
+// wrong code if branches whose allowed offsets are [-128, -126, ..., 126]
+// are present.
void MipsLongBranch::expandToLongBranch(MBBInfo &I) {
MachineBasicBlock::iterator Pos;
MachineBasicBlock *MBB = I.Br->getParent(), *TgtMBB = getTargetMBB(*I.Br);
const BasicBlock *BB = MBB->getBasicBlock();
MachineFunction::iterator FallThroughMBB = ++MachineFunction::iterator(MBB);
MachineBasicBlock *LongBrMBB = MF->CreateMachineBasicBlock(BB);
-
+ const MipsSubtarget &Subtarget =
+ static_cast<const MipsSubtarget &>(MF->getSubtarget());
const MipsInstrInfo *TII =
- static_cast<const MipsInstrInfo*>(TM.getInstrInfo());
+ static_cast<const MipsInstrInfo *>(Subtarget.getInstrInfo());
MF->insert(FallThroughMBB, LongBrMBB);
MBB->removeSuccessor(TgtMBB);
LongBrMBB->addSuccessor(BalTgtMBB);
BalTgtMBB->addSuccessor(TgtMBB);
- int64_t TgtAddress = MBBInfos[TgtMBB->getNumber()].Address;
- unsigned BalTgtMBBSize = 5;
- int64_t Offset = TgtAddress - (I.Address + I.Size - BalTgtMBBSize * 4);
- int64_t Lo = SignExtend64<16>(Offset & 0xffff);
- int64_t Hi = SignExtend64<16>(((Offset + 0x8000) >> 16) & 0xffff);
+ // We must select between the MIPS32r6/MIPS64r6 BAL (which is a normal
+ // instruction) and the pre-MIPS32r6/MIPS64r6 definition (which is an
+ // pseudo-instruction wrapping BGEZAL).
+ unsigned BalOp = Subtarget.hasMips32r6() ? Mips::BAL : Mips::BAL_BR;
- if (ABI != MipsSubtarget::N64) {
+ if (!ABI.IsN64()) {
// $longbr:
// addiu $sp, $sp, -8
// sw $ra, 0($sp)
- // bal $baltgt
// lui $at, %hi($tgt - $baltgt)
- // $baltgt:
+ // bal $baltgt
// addiu $at, $at, %lo($tgt - $baltgt)
+ // $baltgt:
// addu $at, $ra, $at
// lw $ra, 0($sp)
// jr $at
BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::SW)).addReg(Mips::RA)
.addReg(Mips::SP).addImm(0);
+ // LUi and ADDiu instructions create 32-bit offset of the target basic
+ // block from the target of BAL instruction. We cannot use immediate
+ // value for this offset because it cannot be determined accurately when
+ // the program has inline assembly statements. We therefore use the
+ // relocation expressions %hi($tgt-$baltgt) and %lo($tgt-$baltgt) which
+ // are resolved during the fixup, so the values will always be correct.
+ //
+ // Since we cannot create %hi($tgt-$baltgt) and %lo($tgt-$baltgt)
+ // expressions at this point (it is possible only at the MC layer),
+ // we replace LUi and ADDiu with pseudo instructions
+ // LONG_BRANCH_LUi and LONG_BRANCH_ADDiu, and add both basic
+ // blocks as operands to these instructions. When lowering these pseudo
+ // instructions to LUi and ADDiu in the MC layer, we will create
+ // %hi($tgt-$baltgt) and %lo($tgt-$baltgt) expressions and add them as
+ // operands to lowered instructions.
+
+ BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::LONG_BRANCH_LUi), Mips::AT)
+ .addMBB(TgtMBB).addMBB(BalTgtMBB);
MIBundleBuilder(*LongBrMBB, Pos)
- .append(BuildMI(*MF, DL, TII->get(Mips::BAL_BR)).addMBB(BalTgtMBB))
- .append(BuildMI(*MF, DL, TII->get(Mips::LUi), Mips::AT).addImm(Hi));
+ .append(BuildMI(*MF, DL, TII->get(BalOp)).addMBB(BalTgtMBB))
+ .append(BuildMI(*MF, DL, TII->get(Mips::LONG_BRANCH_ADDiu), Mips::AT)
+ .addReg(Mips::AT)
+ .addMBB(TgtMBB)
+ .addMBB(BalTgtMBB));
Pos = BalTgtMBB->begin();
- BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::ADDiu), Mips::AT)
- .addReg(Mips::AT).addImm(Lo);
BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::ADDu), Mips::AT)
.addReg(Mips::RA).addReg(Mips::AT);
BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::LW), Mips::RA)
.addReg(Mips::SP).addImm(0);
- MIBundleBuilder(*BalTgtMBB, Pos)
- .append(BuildMI(*MF, DL, TII->get(Mips::JR)).addReg(Mips::AT))
- .append(BuildMI(*MF, DL, TII->get(Mips::ADDiu), Mips::SP)
- .addReg(Mips::SP).addImm(8));
+ if (!Subtarget.isTargetNaCl()) {
+ MIBundleBuilder(*BalTgtMBB, Pos)
+ .append(BuildMI(*MF, DL, TII->get(Mips::JR)).addReg(Mips::AT))
+ .append(BuildMI(*MF, DL, TII->get(Mips::ADDiu), Mips::SP)
+ .addReg(Mips::SP).addImm(8));
+ } else {
+ // In NaCl, modifying the sp is not allowed in branch delay slot.
+ BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::ADDiu), Mips::SP)
+ .addReg(Mips::SP).addImm(8);
+
+ MIBundleBuilder(*BalTgtMBB, Pos)
+ .append(BuildMI(*MF, DL, TII->get(Mips::JR)).addReg(Mips::AT))
+ .append(BuildMI(*MF, DL, TII->get(Mips::NOP)));
+
+ // Bundle-align the target of indirect branch JR.
+ TgtMBB->setAlignment(MIPS_NACL_BUNDLE_ALIGN);
+ }
} else {
// $longbr:
// daddiu $sp, $sp, -16
// sd $ra, 0($sp)
- // lui64 $at, %highest($tgt - $baltgt)
- // daddiu $at, $at, %higher($tgt - $baltgt)
+ // daddiu $at, $zero, %hi($tgt - $baltgt)
// dsll $at, $at, 16
- // daddiu $at, $at, %hi($tgt - $baltgt)
// bal $baltgt
- // dsll $at, $at, 16
- // $baltgt:
// daddiu $at, $at, %lo($tgt - $baltgt)
+ // $baltgt:
// daddu $at, $ra, $at
// ld $ra, 0($sp)
// jr64 $at
// $fallthrough:
//
- int64_t Higher = SignExtend64<16>(((Offset + 0x80008000) >> 32) & 0xffff);
- int64_t Highest =
- SignExtend64<16>(((Offset + 0x800080008000LL) >> 48) & 0xffff);
+ // We assume the branch is within-function, and that offset is within
+ // +/- 2GB. High 32 bits will therefore always be zero.
+
+ // Note that this will work even if the offset is negative, because
+ // of the +1 modification that's added in that case. For example, if the
+ // offset is -1MB (0xFFFFFFFFFFF00000), the computation for %higher is
+ //
+ // 0xFFFFFFFFFFF00000 + 0x80008000 = 0x000000007FF08000
+ //
+ // and the bits [47:32] are zero. For %highest
+ //
+ // 0xFFFFFFFFFFF00000 + 0x800080008000 = 0x000080007FF08000
+ //
+ // and the bits [63:48] are zero.
Pos = LongBrMBB->begin();
.addReg(Mips::SP_64).addImm(-16);
BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::SD)).addReg(Mips::RA_64)
.addReg(Mips::SP_64).addImm(0);
- BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::LUi64), Mips::AT_64)
- .addImm(Highest);
- BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::DADDiu), Mips::AT_64)
- .addReg(Mips::AT_64).addImm(Higher);
+ BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::LONG_BRANCH_DADDiu),
+ Mips::AT_64).addReg(Mips::ZERO_64)
+ .addMBB(TgtMBB, MipsII::MO_ABS_HI).addMBB(BalTgtMBB);
BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::DSLL), Mips::AT_64)
.addReg(Mips::AT_64).addImm(16);
- BuildMI(*LongBrMBB, Pos, DL, TII->get(Mips::DADDiu), Mips::AT_64)
- .addReg(Mips::AT_64).addImm(Hi);
MIBundleBuilder(*LongBrMBB, Pos)
- .append(BuildMI(*MF, DL, TII->get(Mips::BAL_BR)).addMBB(BalTgtMBB))
- .append(BuildMI(*MF, DL, TII->get(Mips::DSLL), Mips::AT_64)
- .addReg(Mips::AT_64).addImm(16));
+ .append(BuildMI(*MF, DL, TII->get(BalOp)).addMBB(BalTgtMBB))
+ .append(
+ BuildMI(*MF, DL, TII->get(Mips::LONG_BRANCH_DADDiu), Mips::AT_64)
+ .addReg(Mips::AT_64)
+ .addMBB(TgtMBB, MipsII::MO_ABS_LO)
+ .addMBB(BalTgtMBB));
Pos = BalTgtMBB->begin();
- BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::DADDiu), Mips::AT_64)
- .addReg(Mips::AT_64).addImm(Lo);
BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::DADDu), Mips::AT_64)
.addReg(Mips::RA_64).addReg(Mips::AT_64);
BuildMI(*BalTgtMBB, Pos, DL, TII->get(Mips::LD), Mips::RA_64)
.addReg(Mips::SP_64).addImm(16));
}
- assert(BalTgtMBBSize == BalTgtMBB->size());
- assert(LongBrMBB->size() + BalTgtMBBSize == LongBranchSeqSize);
+ assert(LongBrMBB->size() + BalTgtMBB->size() == LongBranchSeqSize);
} else {
// $longbr:
// j $tgt
}
bool MipsLongBranch::runOnMachineFunction(MachineFunction &F) {
+ const MipsSubtarget &STI =
+ static_cast<const MipsSubtarget &>(F.getSubtarget());
const MipsInstrInfo *TII =
- static_cast<const MipsInstrInfo*>(TM.getInstrInfo());
+ static_cast<const MipsInstrInfo *>(STI.getInstrInfo());
+ LongBranchSeqSize =
+ !IsPIC ? 2 : (ABI.IsN64() ? 10 : (!STI.isTargetNaCl() ? 9 : 10));
- if (TM.getSubtarget<MipsSubtarget>().inMips16Mode())
+ if (STI.inMips16Mode() || !STI.enableLongBranchPass())
return false;
if ((TM.getRelocationModel() == Reloc::PIC_) &&
- TM.getSubtarget<MipsSubtarget>().isABI_O32() &&
+ static_cast<const MipsTargetMachine &>(TM).getABI().IsO32() &&
F.getInfo<MipsFunctionInfo>()->globalBaseRegSet())
emitGPDisp(F, TII);
if (!I->Br || I->HasLongBranch)
continue;
- int ShVal = TM.getSubtarget<MipsSubtarget>().inMicroMipsMode() ? 2 : 4;
+ int ShVal = STI.inMicroMipsMode() ? 2 : 4;
+ int64_t Offset = computeOffset(I->Br) / ShVal;
+
+ if (STI.isTargetNaCl()) {
+ // The offset calculation does not include sandboxing instructions
+ // that will be added later in the MC layer. Since at this point we
+ // don't know the exact amount of code that "sandboxing" will add, we
+ // conservatively estimate that code will not grow more than 100%.
+ Offset *= 2;
+ }
// Check if offset fits into 16-bit immediate field of branches.
- if (!ForceLongBranch && isInt<16>(computeOffset(I->Br) / ShVal))
+ if (!ForceLongBranch && isInt<16>(Offset))
continue;
I->HasLongBranch = true;