X86AsmPrinter.cpp revision 055acae18c4ba5f149237927f88a842cd77285cf 
1//===-- X86AsmPrinter.cpp - Convert X86 LLVM code to Intel assembly -------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file was developed by the LLVM research group and is distributed under
6// the University of Illinois Open Source License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file contains a printer that converts from our internal representation
11// of machine-dependent LLVM code to Intel-format assembly language. This
12// printer is the output mechanism used by `llc' and `lli -print-machineinstrs'
13// on X86.
14//
15//===----------------------------------------------------------------------===//
16
17#include "X86.h"
18#include "X86InstrInfo.h"
19#include "X86TargetMachine.h"
20#include "llvm/Constants.h"
21#include "llvm/DerivedTypes.h"
22#include "llvm/Module.h"
23#include "llvm/Assembly/Writer.h"
24#include "llvm/CodeGen/AsmPrinter.h"
25#include "llvm/CodeGen/MachineCodeEmitter.h"
26#include "llvm/CodeGen/MachineConstantPool.h"
27#include "llvm/CodeGen/MachineFunctionPass.h"
28#include "llvm/CodeGen/MachineInstr.h"
29#include "llvm/CodeGen/ValueTypes.h"
30#include "llvm/Target/TargetMachine.h"
31#include "llvm/Support/Mangler.h"
32#include "Support/Statistic.h"
33#include "Support/StringExtras.h"
34#include "Support/CommandLine.h"
35using namespace llvm;
36
37namespace {
38  Statistic<> EmittedInsts("asm-printer", "Number of machine instrs printed");
39
40  struct GasBugWorkaroundEmitter : public MachineCodeEmitter {
41    GasBugWorkaroundEmitter(std::ostream& o)
42      : O(o), OldFlags(O.flags()), firstByte(true) {
43      O << std::hex;
44    }
45
46    ~GasBugWorkaroundEmitter() {
47      O.flags(OldFlags);
48    }
49
50    virtual void emitByte(unsigned char B) {
51      if (!firstByte) O << "\n\t";
52      firstByte = false;
53      O << ".byte 0x" << (unsigned) B;
54    }
55
56    // These should never be called
57    virtual void emitWord(unsigned W) { assert(0); }
58    virtual uint64_t getGlobalValueAddress(GlobalValue *V) { abort(); }
59    virtual uint64_t getGlobalValueAddress(const std::string &Name) { abort(); }
60    virtual uint64_t getConstantPoolEntryAddress(unsigned Index) { abort(); }
61    virtual uint64_t getCurrentPCValue() { abort(); }
62    virtual uint64_t forceCompilationOf(Function *F) { abort(); }
63
64  private:
65    std::ostream& O;
66    std::ios::fmtflags OldFlags;
67    bool firstByte;
68  };
69
70  struct X86AsmPrinter : public AsmPrinter {
71    X86AsmPrinter(std::ostream &O, TargetMachine &TM) : AsmPrinter(O, TM) { }
72
73    virtual const char *getPassName() const {
74      return "X86 Assembly Printer";
75    }
76
77    /// printInstruction - This method is automatically generated by tablegen
78    /// from the instruction set description.  This method returns true if the
79    /// machine instruction was sufficiently described to print it, otherwise it
80    /// returns false.
81    bool printInstruction(const MachineInstr *MI);
82
83    // This method is used by the tablegen'erated instruction printer.
84    void printOperand(const MachineInstr *MI, unsigned OpNo, MVT::ValueType VT){
85      const MachineOperand &MO = MI->getOperand(OpNo);
86      if (MO.getType() == MachineOperand::MO_MachineRegister) {
87        assert(MRegisterInfo::isPhysicalRegister(MO.getReg())&&"Not physref??");
88        // Bug Workaround: See note in Printer::doInitialization about %.
89        O << "%" << TM.getRegisterInfo()->get(MO.getReg()).Name;
90      } else {
91        printOp(MO);
92      }
93    }
94
95    void printCallOperand(const MachineInstr *MI, unsigned OpNo,
96                          MVT::ValueType VT) {
97      printOp(MI->getOperand(OpNo), true); // Don't print "OFFSET".
98    }
99
100    void printMemoryOperand(const MachineInstr *MI, unsigned OpNo,
101                            MVT::ValueType VT) {
102      switch (VT) {
103      default: assert(0 && "Unknown arg size!");
104      case MVT::i8:   O << "BYTE PTR "; break;
105      case MVT::i16:  O << "WORD PTR "; break;
106      case MVT::i32:
107      case MVT::f32:  O << "DWORD PTR "; break;
108      case MVT::i64:
109      case MVT::f64:  O << "QWORD PTR "; break;
110      case MVT::f80:  O << "XWORD PTR "; break;
111      }
112      printMemReference(MI, OpNo);
113    }
114
115    void printMachineInstruction(const MachineInstr *MI);
116    void printOp(const MachineOperand &MO, bool elideOffsetKeyword = false);
117    void printMemReference(const MachineInstr *MI, unsigned Op);
118    void printConstantPool(MachineConstantPool *MCP);
119    bool runOnMachineFunction(MachineFunction &F);
120    bool doInitialization(Module &M);
121    bool doFinalization(Module &M);
122    void emitGlobalConstant(const Constant* CV);
123  };
124} // end of anonymous namespace
125
126/// createX86CodePrinterPass - Returns a pass that prints the X86
127/// assembly code for a MachineFunction to the given output stream,
128/// using the given target machine description.  This should work
129/// regardless of whether the function is in SSA form.
130///
131FunctionPass *llvm::createX86CodePrinterPass(std::ostream &o,TargetMachine &tm){
132  return new X86AsmPrinter(o, tm);
133}
134
135
136// Include the auto-generated portion of the assembly writer.
137#include "X86GenAsmWriter.inc"
138
139
140/// toOctal - Convert the low order bits of X into an octal digit.
141///
142static inline char toOctal(int X) {
143  return (X&7)+'0';
144}
145
146/// getAsCString - Return the specified array as a C compatible
147/// string, only if the predicate isStringCompatible is true.
148///
149static void printAsCString(std::ostream &O, const ConstantArray *CVA) {
150  assert(CVA->isString() && "Array is not string compatible!");
151
152  O << "\"";
153  for (unsigned i = 0; i != CVA->getNumOperands(); ++i) {
154    unsigned char C = cast<ConstantInt>(CVA->getOperand(i))->getRawValue();
155
156    if (C == '"') {
157      O << "\\\"";
158    } else if (C == '\\') {
159      O << "\\\\";
160    } else if (isprint(C)) {
161      O << C;
162    } else {
163      switch(C) {
164      case '\b': O << "\\b"; break;
165      case '\f': O << "\\f"; break;
166      case '\n': O << "\\n"; break;
167      case '\r': O << "\\r"; break;
168      case '\t': O << "\\t"; break;
169      default:
170        O << '\\';
171        O << toOctal(C >> 6);
172        O << toOctal(C >> 3);
173        O << toOctal(C >> 0);
174        break;
175      }
176    }
177  }
178  O << "\"";
179}
180
181// Print a constant value or values, with the appropriate storage class as a
182// prefix.
183void X86AsmPrinter::emitGlobalConstant(const Constant *CV) {
184  const TargetData &TD = TM.getTargetData();
185
186  if (CV->isNullValue()) {
187    O << "\t.zero\t " << TD.getTypeSize(CV->getType()) << "\n";
188    return;
189  } else if (const ConstantArray *CVA = dyn_cast<ConstantArray>(CV)) {
190    if (CVA->isString()) {
191      O << "\t.ascii\t";
192      printAsCString(O, CVA);
193      O << "\n";
194    } else { // Not a string.  Print the values in successive locations
195      for (unsigned i = 0, e = CVA->getNumOperands(); i != e; ++i)
196        emitGlobalConstant(CVA->getOperand(i));
197    }
198    return;
199  } else if (const ConstantStruct *CVS = dyn_cast<ConstantStruct>(CV)) {
200    // Print the fields in successive locations. Pad to align if needed!
201    const StructLayout *cvsLayout = TD.getStructLayout(CVS->getType());
202    unsigned sizeSoFar = 0;
203    for (unsigned i = 0, e = CVS->getNumOperands(); i != e; ++i) {
204      const Constant* field = CVS->getOperand(i);
205
206      // Check if padding is needed and insert one or more 0s.
207      unsigned fieldSize = TD.getTypeSize(field->getType());
208      unsigned padSize = ((i == e-1? cvsLayout->StructSize
209                           : cvsLayout->MemberOffsets[i+1])
210                          - cvsLayout->MemberOffsets[i]) - fieldSize;
211      sizeSoFar += fieldSize + padSize;
212
213      // Now print the actual field value
214      emitGlobalConstant(field);
215
216      // Insert the field padding unless it's zero bytes...
217      if (padSize)
218        O << "\t.zero\t " << padSize << "\n";
219    }
220    assert(sizeSoFar == cvsLayout->StructSize &&
221           "Layout of constant struct may be incorrect!");
222    return;
223  } else if (const ConstantFP *CFP = dyn_cast<ConstantFP>(CV)) {
224    // FP Constants are printed as integer constants to avoid losing
225    // precision...
226    double Val = CFP->getValue();
227    switch (CFP->getType()->getTypeID()) {
228    default: assert(0 && "Unknown floating point type!");
229    case Type::FloatTyID: {
230      union FU {                            // Abide by C TBAA rules
231        float FVal;
232        unsigned UVal;
233      } U;
234      U.FVal = Val;
235      O << ".long\t" << U.UVal << "\t# float " << Val << "\n";
236      return;
237    }
238    case Type::DoubleTyID: {
239      union DU {                            // Abide by C TBAA rules
240        double FVal;
241        uint64_t UVal;
242      } U;
243      U.FVal = Val;
244      O << ".quad\t" << U.UVal << "\t# double " << Val << "\n";
245      return;
246    }
247    }
248  }
249
250  const Type *type = CV->getType();
251  O << "\t";
252  switch (type->getTypeID()) {
253  case Type::BoolTyID: case Type::UByteTyID: case Type::SByteTyID:
254    O << ".byte";
255    break;
256  case Type::UShortTyID: case Type::ShortTyID:
257    O << ".word";
258    break;
259  case Type::FloatTyID: case Type::PointerTyID:
260  case Type::UIntTyID: case Type::IntTyID:
261    O << ".long";
262    break;
263  case Type::DoubleTyID:
264  case Type::ULongTyID: case Type::LongTyID:
265    O << ".quad";
266    break;
267  default:
268    assert (0 && "Can't handle printing this type of thing");
269    break;
270  }
271  O << "\t";
272  emitConstantValueOnly(CV);
273  O << "\n";
274}
275
276/// printConstantPool - Print to the current output stream assembly
277/// representations of the constants in the constant pool MCP. This is
278/// used to print out constants which have been "spilled to memory" by
279/// the code generator.
280///
281void X86AsmPrinter::printConstantPool(MachineConstantPool *MCP) {
282  const std::vector<Constant*> &CP = MCP->getConstants();
283  const TargetData &TD = TM.getTargetData();
284
285  if (CP.empty()) return;
286
287  for (unsigned i = 0, e = CP.size(); i != e; ++i) {
288    O << "\t.section .rodata\n";
289    O << "\t.align " << (unsigned)TD.getTypeAlignment(CP[i]->getType())
290      << "\n";
291    O << ".CPI" << CurrentFnName << "_" << i << ":\t\t\t\t\t#"
292      << *CP[i] << "\n";
293    emitGlobalConstant(CP[i]);
294  }
295}
296
297/// runOnMachineFunction - This uses the printMachineInstruction()
298/// method to print assembly for each instruction.
299///
300bool X86AsmPrinter::runOnMachineFunction(MachineFunction &MF) {
301  setupMachineFunction(MF);
302  O << "\n\n";
303
304  // Print out constants referenced by the function
305  printConstantPool(MF.getConstantPool());
306
307  // Print out labels for the function.
308  O << "\t.text\n";
309  O << "\t.align 16\n";
310  O << "\t.globl\t" << CurrentFnName << "\n";
311  O << "\t.type\t" << CurrentFnName << ", @function\n";
312  O << CurrentFnName << ":\n";
313
314  // Print out code for the function.
315  for (MachineFunction::const_iterator I = MF.begin(), E = MF.end();
316       I != E; ++I) {
317    // Print a label for the basic block.
318    O << ".LBB" << CurrentFnName << "_" << I->getNumber() << ":\t# "
319      << I->getBasicBlock()->getName() << "\n";
320    for (MachineBasicBlock::const_iterator II = I->begin(), E = I->end();
321         II != E; ++II) {
322      // Print the assembly for the instruction.
323      O << "\t";
324      printMachineInstruction(II);
325    }
326  }
327
328  // We didn't modify anything.
329  return false;
330}
331
332static bool isScale(const MachineOperand &MO) {
333  return MO.isImmediate() &&
334    (MO.getImmedValue() == 1 || MO.getImmedValue() == 2 ||
335     MO.getImmedValue() == 4 || MO.getImmedValue() == 8);
336}
337
338static bool isMem(const MachineInstr *MI, unsigned Op) {
339  if (MI->getOperand(Op).isFrameIndex()) return true;
340  if (MI->getOperand(Op).isConstantPoolIndex()) return true;
341  return Op+4 <= MI->getNumOperands() &&
342    MI->getOperand(Op  ).isRegister() && isScale(MI->getOperand(Op+1)) &&
343    MI->getOperand(Op+2).isRegister() && MI->getOperand(Op+3).isImmediate();
344}
345
346
347
348void X86AsmPrinter::printOp(const MachineOperand &MO,
349                            bool elideOffsetKeyword /* = false */) {
350  const MRegisterInfo &RI = *TM.getRegisterInfo();
351  switch (MO.getType()) {
352  case MachineOperand::MO_VirtualRegister:
353    if (Value *V = MO.getVRegValueOrNull()) {
354      O << "<" << V->getName() << ">";
355      return;
356    }
357    // FALLTHROUGH
358  case MachineOperand::MO_MachineRegister:
359    if (MRegisterInfo::isPhysicalRegister(MO.getReg()))
360      // Bug Workaround: See note in Printer::doInitialization about %.
361      O << "%" << RI.get(MO.getReg()).Name;
362    else
363      O << "%reg" << MO.getReg();
364    return;
365
366  case MachineOperand::MO_SignExtendedImmed:
367  case MachineOperand::MO_UnextendedImmed:
368    O << (int)MO.getImmedValue();
369    return;
370  case MachineOperand::MO_MachineBasicBlock: {
371    MachineBasicBlock *MBBOp = MO.getMachineBasicBlock();
372    O << ".LBB" << Mang->getValueName(MBBOp->getParent()->getFunction())
373      << "_" << MBBOp->getNumber () << "\t# "
374      << MBBOp->getBasicBlock ()->getName ();
375    return;
376  }
377  case MachineOperand::MO_PCRelativeDisp:
378    std::cerr << "Shouldn't use addPCDisp() when building X86 MachineInstrs";
379    abort ();
380    return;
381  case MachineOperand::MO_GlobalAddress:
382    if (!elideOffsetKeyword)
383      O << "OFFSET ";
384    O << Mang->getValueName(MO.getGlobal());
385    return;
386  case MachineOperand::MO_ExternalSymbol:
387    O << MO.getSymbolName();
388    return;
389  default:
390    O << "<unknown operand type>"; return;
391  }
392}
393
394void X86AsmPrinter::printMemReference(const MachineInstr *MI, unsigned Op) {
395  assert(isMem(MI, Op) && "Invalid memory reference!");
396
397  if (MI->getOperand(Op).isFrameIndex()) {
398    O << "[frame slot #" << MI->getOperand(Op).getFrameIndex();
399    if (MI->getOperand(Op+3).getImmedValue())
400      O << " + " << MI->getOperand(Op+3).getImmedValue();
401    O << "]";
402    return;
403  } else if (MI->getOperand(Op).isConstantPoolIndex()) {
404    O << "[.CPI" << CurrentFnName << "_"
405      << MI->getOperand(Op).getConstantPoolIndex();
406    if (MI->getOperand(Op+3).getImmedValue())
407      O << " + " << MI->getOperand(Op+3).getImmedValue();
408    O << "]";
409    return;
410  }
411
412  const MachineOperand &BaseReg  = MI->getOperand(Op);
413  int ScaleVal                   = MI->getOperand(Op+1).getImmedValue();
414  const MachineOperand &IndexReg = MI->getOperand(Op+2);
415  int DispVal                    = MI->getOperand(Op+3).getImmedValue();
416
417  O << "[";
418  bool NeedPlus = false;
419  if (BaseReg.getReg()) {
420    printOp(BaseReg);
421    NeedPlus = true;
422  }
423
424  if (IndexReg.getReg()) {
425    if (NeedPlus) O << " + ";
426    if (ScaleVal != 1)
427      O << ScaleVal << "*";
428    printOp(IndexReg);
429    NeedPlus = true;
430  }
431
432  if (DispVal) {
433    if (NeedPlus)
434      if (DispVal > 0)
435        O << " + ";
436      else {
437        O << " - ";
438        DispVal = -DispVal;
439      }
440    O << DispVal;
441  }
442  O << "]";
443}
444
445
446/// printMachineInstruction -- Print out a single X86 LLVM instruction
447/// MI in Intel syntax to the current output stream.
448///
449void X86AsmPrinter::printMachineInstruction(const MachineInstr *MI) {
450  ++EmittedInsts;
451
452  // gas bugs:
453  //
454  // The 80-bit FP store-pop instruction "fstp XWORD PTR [...]"  is misassembled
455  // by gas in intel_syntax mode as its 32-bit equivalent "fstp DWORD PTR
456  // [...]". Workaround: Output the raw opcode bytes instead of the instruction.
457  //
458  // The 80-bit FP load instruction "fld XWORD PTR [...]" is misassembled by gas
459  // in intel_syntax mode as its 32-bit equivalent "fld DWORD PTR
460  // [...]". Workaround: Output the raw opcode bytes instead of the instruction.
461  //
462  // gas intel_syntax mode treats "fild QWORD PTR [...]" as an invalid opcode,
463  // saying "64 bit operations are only supported in 64 bit modes." libopcodes
464  // disassembles it as "fild DWORD PTR [...]", which is wrong. Workaround:
465  // Output the raw opcode bytes instead of the instruction.
466  //
467  // gas intel_syntax mode treats "fistp QWORD PTR [...]" as an invalid opcode,
468  // saying "64 bit operations are only supported in 64 bit modes." libopcodes
469  // disassembles it as "fistpll DWORD PTR [...]", which is wrong. Workaround:
470  // Output the raw opcode bytes instead of the instruction.
471  switch (MI->getOpcode()) {
472  case X86::FSTP80m:
473  case X86::FLD80m:
474  case X86::FILD64m:
475  case X86::FISTP64m:
476    GasBugWorkaroundEmitter gwe(O);
477    X86::emitInstruction(gwe, (X86InstrInfo&)*TM.getInstrInfo(), *MI);
478    O << "\t# ";
479  }
480
481  // Call the autogenerated instruction printer routines.
482  bool Handled = printInstruction(MI);
483  if (!Handled) {
484    MI->dump();
485    assert(0 && "Do not know how to print this instruction!");
486    abort();
487  }
488}
489
490bool X86AsmPrinter::doInitialization(Module &M) {
491  AsmPrinter::doInitialization(M);
492  // Tell gas we are outputting Intel syntax (not AT&T syntax) assembly.
493  //
494  // Bug: gas in `intel_syntax noprefix' mode interprets the symbol `Sp' in an
495  // instruction as a reference to the register named sp, and if you try to
496  // reference a symbol `Sp' (e.g. `mov ECX, OFFSET Sp') then it gets lowercased
497  // before being looked up in the symbol table. This creates spurious
498  // `undefined symbol' errors when linking. Workaround: Do not use `noprefix'
499  // mode, and decorate all register names with percent signs.
500  O << "\t.intel_syntax\n";
501  return false;
502}
503
504// SwitchSection - Switch to the specified section of the executable if we are
505// not already in it!
506//
507static void SwitchSection(std::ostream &OS, std::string &CurSection,
508                          const char *NewSection) {
509  if (CurSection != NewSection) {
510    CurSection = NewSection;
511    if (!CurSection.empty())
512      OS << "\t" << NewSection << "\n";
513  }
514}
515
516bool X86AsmPrinter::doFinalization(Module &M) {
517  const TargetData &TD = TM.getTargetData();
518  std::string CurSection;
519
520  // Print out module-level global variables here.
521  for (Module::const_giterator I = M.gbegin(), E = M.gend(); I != E; ++I)
522    if (I->hasInitializer()) {   // External global require no code
523      O << "\n\n";
524      std::string name = Mang->getValueName(I);
525      Constant *C = I->getInitializer();
526      unsigned Size = TD.getTypeSize(C->getType());
527      unsigned Align = TD.getTypeAlignment(C->getType());
528
529      if (C->isNullValue() &&
530          (I->hasLinkOnceLinkage() || I->hasInternalLinkage() ||
531           I->hasWeakLinkage() /* FIXME: Verify correct */)) {
532        SwitchSection(O, CurSection, ".data");
533        if (I->hasInternalLinkage())
534          O << "\t.local " << name << "\n";
535
536        O << "\t.comm " << name << "," << TD.getTypeSize(C->getType())
537          << "," << (unsigned)TD.getTypeAlignment(C->getType());
538        O << "\t\t# ";
539        WriteAsOperand(O, I, true, true, &M);
540        O << "\n";
541      } else {
542        switch (I->getLinkage()) {
543        case GlobalValue::LinkOnceLinkage:
544        case GlobalValue::WeakLinkage:   // FIXME: Verify correct for weak.
545          // Nonnull linkonce -> weak
546          O << "\t.weak " << name << "\n";
547          SwitchSection(O, CurSection, "");
548          O << "\t.section\t.llvm.linkonce.d." << name << ",\"aw\",@progbits\n";
549          break;
550
551        case GlobalValue::AppendingLinkage:
552          // FIXME: appending linkage variables should go into a section of
553          // their name or something.  For now, just emit them as external.
554        case GlobalValue::ExternalLinkage:
555          // If external or appending, declare as a global symbol
556          O << "\t.globl " << name << "\n";
557          // FALL THROUGH
558        case GlobalValue::InternalLinkage:
559          if (C->isNullValue())
560            SwitchSection(O, CurSection, ".bss");
561          else
562            SwitchSection(O, CurSection, ".data");
563          break;
564        }
565
566        O << "\t.align " << Align << "\n";
567        O << "\t.type " << name << ",@object\n";
568        O << "\t.size " << name << "," << Size << "\n";
569        O << name << ":\t\t\t\t# ";
570        WriteAsOperand(O, I, true, true, &M);
571        O << " = ";
572        WriteAsOperand(O, C, false, false, &M);
573        O << "\n";
574        emitGlobalConstant(C);
575      }
576    }
577
578  AsmPrinter::doFinalization(M);
579  return false; // success
580}
581