MCAsmStreamer.cpp revision 523d70ec1f8daa67bb8a9fe8f7b6b3d076a26c99 
1//===- lib/MC/MCAsmStreamer.cpp - Text Assembly Output --------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "llvm/MC/MCStreamer.h"
11#include "llvm/MC/MCAsmInfo.h"
12#include "llvm/MC/MCCodeEmitter.h"
13#include "llvm/MC/MCContext.h"
14#include "llvm/MC/MCExpr.h"
15#include "llvm/MC/MCInst.h"
16#include "llvm/MC/MCInstPrinter.h"
17#include "llvm/MC/MCSectionMachO.h"
18#include "llvm/MC/MCSymbol.h"
19#include "llvm/ADT/SmallString.h"
20#include "llvm/ADT/Twine.h"
21#include "llvm/Support/ErrorHandling.h"
22#include "llvm/Support/MathExtras.h"
23#include "llvm/Support/Format.h"
24#include "llvm/Support/FormattedStream.h"
25using namespace llvm;
26
27namespace {
28
29class MCAsmStreamer : public MCStreamer {
30  formatted_raw_ostream &OS;
31  const MCAsmInfo &MAI;
32  MCInstPrinter *InstPrinter;
33  MCCodeEmitter *Emitter;
34
35  SmallString<128> CommentToEmit;
36  raw_svector_ostream CommentStream;
37
38  unsigned IsLittleEndian : 1;
39  unsigned IsVerboseAsm : 1;
40  unsigned ShowInst : 1;
41
42public:
43  MCAsmStreamer(MCContext &Context, formatted_raw_ostream &os,
44                const MCAsmInfo &mai,
45                bool isLittleEndian, bool isVerboseAsm, MCInstPrinter *printer,
46                MCCodeEmitter *emitter, bool showInst)
47    : MCStreamer(Context), OS(os), MAI(mai), InstPrinter(printer),
48      Emitter(emitter), CommentStream(CommentToEmit),
49      IsLittleEndian(isLittleEndian), IsVerboseAsm(isVerboseAsm),
50      ShowInst(showInst) {}
51  ~MCAsmStreamer() {}
52
53  bool isLittleEndian() const { return IsLittleEndian; }
54
55  inline void EmitEOL() {
56    // If we don't have any comments, just emit a \n.
57    if (!IsVerboseAsm) {
58      OS << '\n';
59      return;
60    }
61    EmitCommentsAndEOL();
62  }
63  void EmitCommentsAndEOL();
64
65  /// isVerboseAsm - Return true if this streamer supports verbose assembly at
66  /// all.
67  virtual bool isVerboseAsm() const { return IsVerboseAsm; }
68
69  /// AddComment - Add a comment that can be emitted to the generated .s
70  /// file if applicable as a QoI issue to make the output of the compiler
71  /// more readable.  This only affects the MCAsmStreamer, and only when
72  /// verbose assembly output is enabled.
73  virtual void AddComment(const Twine &T);
74
75  /// GetCommentOS - Return a raw_ostream that comments can be written to.
76  /// Unlike AddComment, you are required to terminate comments with \n if you
77  /// use this method.
78  virtual raw_ostream &GetCommentOS() {
79    if (!IsVerboseAsm)
80      return nulls();  // Discard comments unless in verbose asm mode.
81    return CommentStream;
82  }
83
84  /// AddBlankLine - Emit a blank line to a .s file to pretty it up.
85  virtual void AddBlankLine() {
86    EmitEOL();
87  }
88
89  /// @name MCStreamer Interface
90  /// @{
91
92  virtual void SwitchSection(const MCSection *Section);
93
94  virtual void EmitLabel(MCSymbol *Symbol);
95
96  virtual void EmitAssemblerFlag(MCAssemblerFlag Flag);
97
98  virtual void EmitAssignment(MCSymbol *Symbol, const MCExpr *Value);
99
100  virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute);
101
102  virtual void EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue);
103
104  virtual void EmitELFSize(MCSymbol *Symbol, const MCExpr *Value);
105  virtual void EmitCommonSymbol(MCSymbol *Symbol, uint64_t Size,
106                                unsigned ByteAlignment);
107
108  /// EmitLocalCommonSymbol - Emit a local common (.lcomm) symbol.
109  ///
110  /// @param Symbol - The common symbol to emit.
111  /// @param Size - The size of the common symbol.
112  virtual void EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size);
113
114  virtual void EmitZerofill(const MCSection *Section, MCSymbol *Symbol = 0,
115                            unsigned Size = 0, unsigned ByteAlignment = 0);
116
117  virtual void EmitBytes(StringRef Data, unsigned AddrSpace);
118
119  virtual void EmitValue(const MCExpr *Value, unsigned Size,unsigned AddrSpace);
120  virtual void EmitIntValue(uint64_t Value, unsigned Size, unsigned AddrSpace);
121  virtual void EmitGPRel32Value(const MCExpr *Value);
122
123
124  virtual void EmitFill(uint64_t NumBytes, uint8_t FillValue,
125                        unsigned AddrSpace);
126
127  virtual void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0,
128                                    unsigned ValueSize = 1,
129                                    unsigned MaxBytesToEmit = 0);
130
131  virtual void EmitValueToOffset(const MCExpr *Offset,
132                                 unsigned char Value = 0);
133
134  virtual void EmitFileDirective(StringRef Filename);
135  virtual void EmitDwarfFileDirective(unsigned FileNo, StringRef Filename);
136
137  virtual void EmitInstruction(const MCInst &Inst);
138
139  virtual void Finish();
140
141  /// @}
142};
143
144} // end anonymous namespace.
145
146/// AddComment - Add a comment that can be emitted to the generated .s
147/// file if applicable as a QoI issue to make the output of the compiler
148/// more readable.  This only affects the MCAsmStreamer, and only when
149/// verbose assembly output is enabled.
150void MCAsmStreamer::AddComment(const Twine &T) {
151  if (!IsVerboseAsm) return;
152
153  // Make sure that CommentStream is flushed.
154  CommentStream.flush();
155
156  T.toVector(CommentToEmit);
157  // Each comment goes on its own line.
158  CommentToEmit.push_back('\n');
159
160  // Tell the comment stream that the vector changed underneath it.
161  CommentStream.resync();
162}
163
164void MCAsmStreamer::EmitCommentsAndEOL() {
165  if (CommentToEmit.empty() && CommentStream.GetNumBytesInBuffer() == 0) {
166    OS << '\n';
167    return;
168  }
169
170  CommentStream.flush();
171  StringRef Comments = CommentToEmit.str();
172
173  assert(Comments.back() == '\n' &&
174         "Comment array not newline terminated");
175  do {
176    // Emit a line of comments.
177    OS.PadToColumn(MAI.getCommentColumn());
178    size_t Position = Comments.find('\n');
179    OS << MAI.getCommentString() << ' ' << Comments.substr(0, Position) << '\n';
180
181    Comments = Comments.substr(Position+1);
182  } while (!Comments.empty());
183
184  CommentToEmit.clear();
185  // Tell the comment stream that the vector changed underneath it.
186  CommentStream.resync();
187}
188
189
190static inline int64_t truncateToSize(int64_t Value, unsigned Bytes) {
191  assert(Bytes && "Invalid size!");
192  return Value & ((uint64_t) (int64_t) -1 >> (64 - Bytes * 8));
193}
194
195void MCAsmStreamer::SwitchSection(const MCSection *Section) {
196  assert(Section && "Cannot switch to a null section!");
197  if (Section != CurSection) {
198    CurSection = Section;
199    Section->PrintSwitchToSection(MAI, OS);
200  }
201}
202
203void MCAsmStreamer::EmitLabel(MCSymbol *Symbol) {
204  assert(Symbol->isUndefined() && "Cannot define a symbol twice!");
205  assert(CurSection && "Cannot emit before setting section!");
206
207  OS << *Symbol << ":";
208  EmitEOL();
209  Symbol->setSection(*CurSection);
210}
211
212void MCAsmStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) {
213  switch (Flag) {
214  default: assert(0 && "Invalid flag!");
215  case MCAF_SubsectionsViaSymbols: OS << ".subsections_via_symbols"; break;
216  }
217  EmitEOL();
218}
219
220void MCAsmStreamer::EmitAssignment(MCSymbol *Symbol, const MCExpr *Value) {
221  // Only absolute symbols can be redefined.
222  assert((Symbol->isUndefined() || Symbol->isAbsolute()) &&
223         "Cannot define a symbol twice!");
224
225  OS << *Symbol << " = " << *Value;
226  EmitEOL();
227
228  // FIXME: Lift context changes into super class.
229  // FIXME: Set associated section.
230  Symbol->setValue(Value);
231}
232
233void MCAsmStreamer::EmitSymbolAttribute(MCSymbol *Symbol,
234                                        MCSymbolAttr Attribute) {
235  switch (Attribute) {
236  case MCSA_Invalid: assert(0 && "Invalid symbol attribute");
237  case MCSA_ELF_TypeFunction:    /// .type _foo, STT_FUNC  # aka @function
238  case MCSA_ELF_TypeIndFunction: /// .type _foo, STT_GNU_IFUNC
239  case MCSA_ELF_TypeObject:      /// .type _foo, STT_OBJECT  # aka @object
240  case MCSA_ELF_TypeTLS:         /// .type _foo, STT_TLS     # aka @tls_object
241  case MCSA_ELF_TypeCommon:      /// .type _foo, STT_COMMON  # aka @common
242  case MCSA_ELF_TypeNoType:      /// .type _foo, STT_NOTYPE  # aka @notype
243    assert(MAI.hasDotTypeDotSizeDirective() && "Symbol Attr not supported");
244    OS << "\t.type\t" << *Symbol << ','
245       << ((MAI.getCommentString()[0] != '@') ? '@' : '%');
246    switch (Attribute) {
247    default: assert(0 && "Unknown ELF .type");
248    case MCSA_ELF_TypeFunction:    OS << "function"; break;
249    case MCSA_ELF_TypeIndFunction: OS << "gnu_indirect_function"; break;
250    case MCSA_ELF_TypeObject:      OS << "object"; break;
251    case MCSA_ELF_TypeTLS:         OS << "tls_object"; break;
252    case MCSA_ELF_TypeCommon:      OS << "common"; break;
253    case MCSA_ELF_TypeNoType:      OS << "no_type"; break;
254    }
255    EmitEOL();
256    return;
257  case MCSA_Global: // .globl/.global
258    OS << MAI.getGlobalDirective();
259    break;
260  case MCSA_Hidden:         OS << ".hidden ";          break;
261  case MCSA_IndirectSymbol: OS << ".indirect_symbol "; break;
262  case MCSA_Internal:       OS << ".internal ";        break;
263  case MCSA_LazyReference:  OS << ".lazy_reference ";  break;
264  case MCSA_Local:          OS << ".local ";           break;
265  case MCSA_NoDeadStrip:    OS << ".no_dead_strip ";   break;
266  case MCSA_PrivateExtern:  OS << ".private_extern ";  break;
267  case MCSA_Protected:      OS << ".protected ";       break;
268  case MCSA_Reference:      OS << ".reference ";       break;
269  case MCSA_Weak:           OS << ".weak ";            break;
270  case MCSA_WeakDefinition: OS << ".weak_definition "; break;
271      // .weak_reference
272  case MCSA_WeakReference:  OS << MAI.getWeakRefDirective(); break;
273  }
274
275  OS << *Symbol;
276  EmitEOL();
277}
278
279void MCAsmStreamer::EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) {
280  OS << ".desc" << ' ' << *Symbol << ',' << DescValue;
281  EmitEOL();
282}
283
284void MCAsmStreamer::EmitELFSize(MCSymbol *Symbol, const MCExpr *Value) {
285  assert(MAI.hasDotTypeDotSizeDirective());
286  OS << "\t.size\t" << *Symbol << ", " << *Value << '\n';
287}
288
289void MCAsmStreamer::EmitCommonSymbol(MCSymbol *Symbol, uint64_t Size,
290                                     unsigned ByteAlignment) {
291  OS << "\t.comm\t" << *Symbol << ',' << Size;
292  if (ByteAlignment != 0) {
293    if (MAI.getCOMMDirectiveAlignmentIsInBytes())
294      OS << ',' << ByteAlignment;
295    else
296      OS << ',' << Log2_32(ByteAlignment);
297  }
298  EmitEOL();
299}
300
301/// EmitLocalCommonSymbol - Emit a local common (.lcomm) symbol.
302///
303/// @param Symbol - The common symbol to emit.
304/// @param Size - The size of the common symbol.
305void MCAsmStreamer::EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size) {
306  assert(MAI.hasLCOMMDirective() && "Doesn't have .lcomm, can't emit it!");
307  OS << "\t.lcomm\t" << *Symbol << ',' << Size;
308  EmitEOL();
309}
310
311void MCAsmStreamer::EmitZerofill(const MCSection *Section, MCSymbol *Symbol,
312                                 unsigned Size, unsigned ByteAlignment) {
313  // Note: a .zerofill directive does not switch sections.
314  OS << ".zerofill ";
315
316  // This is a mach-o specific directive.
317  const MCSectionMachO *MOSection = ((const MCSectionMachO*)Section);
318  OS << MOSection->getSegmentName() << "," << MOSection->getSectionName();
319
320  if (Symbol != NULL) {
321    OS << ',' << *Symbol << ',' << Size;
322    if (ByteAlignment != 0)
323      OS << ',' << Log2_32(ByteAlignment);
324  }
325  EmitEOL();
326}
327
328static inline char toOctal(int X) { return (X&7)+'0'; }
329
330static void PrintQuotedString(StringRef Data, raw_ostream &OS) {
331  OS << '"';
332
333  for (unsigned i = 0, e = Data.size(); i != e; ++i) {
334    unsigned char C = Data[i];
335    if (C == '"' || C == '\\') {
336      OS << '\\' << (char)C;
337      continue;
338    }
339
340    if (isprint((unsigned char)C)) {
341      OS << (char)C;
342      continue;
343    }
344
345    switch (C) {
346      case '\b': OS << "\\b"; break;
347      case '\f': OS << "\\f"; break;
348      case '\n': OS << "\\n"; break;
349      case '\r': OS << "\\r"; break;
350      case '\t': OS << "\\t"; break;
351      default:
352        OS << '\\';
353        OS << toOctal(C >> 6);
354        OS << toOctal(C >> 3);
355        OS << toOctal(C >> 0);
356        break;
357    }
358  }
359
360  OS << '"';
361}
362
363
364void MCAsmStreamer::EmitBytes(StringRef Data, unsigned AddrSpace) {
365  assert(CurSection && "Cannot emit contents before setting section!");
366  if (Data.empty()) return;
367
368  if (Data.size() == 1) {
369    OS << MAI.getData8bitsDirective(AddrSpace);
370    OS << (unsigned)(unsigned char)Data[0];
371    EmitEOL();
372    return;
373  }
374
375  // If the data ends with 0 and the target supports .asciz, use it, otherwise
376  // use .ascii
377  if (MAI.getAscizDirective() && Data.back() == 0) {
378    OS << MAI.getAscizDirective();
379    Data = Data.substr(0, Data.size()-1);
380  } else {
381    OS << MAI.getAsciiDirective();
382  }
383
384  OS << ' ';
385  PrintQuotedString(Data, OS);
386  EmitEOL();
387}
388
389/// EmitIntValue - Special case of EmitValue that avoids the client having
390/// to pass in a MCExpr for constant integers.
391void MCAsmStreamer::EmitIntValue(uint64_t Value, unsigned Size,
392                                 unsigned AddrSpace) {
393  assert(CurSection && "Cannot emit contents before setting section!");
394  const char *Directive = 0;
395  switch (Size) {
396  default: break;
397  case 1: Directive = MAI.getData8bitsDirective(AddrSpace); break;
398  case 2: Directive = MAI.getData16bitsDirective(AddrSpace); break;
399  case 4: Directive = MAI.getData32bitsDirective(AddrSpace); break;
400  case 8:
401    Directive = MAI.getData64bitsDirective(AddrSpace);
402    // If the target doesn't support 64-bit data, emit as two 32-bit halves.
403    if (Directive) break;
404    if (isLittleEndian()) {
405      EmitIntValue((uint32_t)(Value >> 0 ), 4, AddrSpace);
406      EmitIntValue((uint32_t)(Value >> 32), 4, AddrSpace);
407    } else {
408      EmitIntValue((uint32_t)(Value >> 32), 4, AddrSpace);
409      EmitIntValue((uint32_t)(Value >> 0 ), 4, AddrSpace);
410    }
411    return;
412  }
413
414  assert(Directive && "Invalid size for machine code value!");
415  OS << Directive << truncateToSize(Value, Size);
416  EmitEOL();
417}
418
419void MCAsmStreamer::EmitValue(const MCExpr *Value, unsigned Size,
420                              unsigned AddrSpace) {
421  assert(CurSection && "Cannot emit contents before setting section!");
422  const char *Directive = 0;
423  switch (Size) {
424  default: break;
425  case 1: Directive = MAI.getData8bitsDirective(AddrSpace); break;
426  case 2: Directive = MAI.getData16bitsDirective(AddrSpace); break;
427  case 4: Directive = MAI.getData32bitsDirective(AddrSpace); break;
428  case 8: Directive = MAI.getData64bitsDirective(AddrSpace); break;
429  }
430
431  assert(Directive && "Invalid size for machine code value!");
432  OS << Directive << *Value;
433  EmitEOL();
434}
435
436void MCAsmStreamer::EmitGPRel32Value(const MCExpr *Value) {
437  assert(MAI.getGPRel32Directive() != 0);
438  OS << MAI.getGPRel32Directive() << *Value;
439  EmitEOL();
440}
441
442
443/// EmitFill - Emit NumBytes bytes worth of the value specified by
444/// FillValue.  This implements directives such as '.space'.
445void MCAsmStreamer::EmitFill(uint64_t NumBytes, uint8_t FillValue,
446                             unsigned AddrSpace) {
447  if (NumBytes == 0) return;
448
449  if (AddrSpace == 0)
450    if (const char *ZeroDirective = MAI.getZeroDirective()) {
451      OS << ZeroDirective << NumBytes;
452      if (FillValue != 0)
453        OS << ',' << (int)FillValue;
454      EmitEOL();
455      return;
456    }
457
458  // Emit a byte at a time.
459  MCStreamer::EmitFill(NumBytes, FillValue, AddrSpace);
460}
461
462void MCAsmStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value,
463                                         unsigned ValueSize,
464                                         unsigned MaxBytesToEmit) {
465  // Some assemblers don't support non-power of two alignments, so we always
466  // emit alignments as a power of two if possible.
467  if (isPowerOf2_32(ByteAlignment)) {
468    switch (ValueSize) {
469    default: llvm_unreachable("Invalid size for machine code value!");
470    case 1: OS << MAI.getAlignDirective(); break;
471    // FIXME: use MAI for this!
472    case 2: OS << ".p2alignw "; break;
473    case 4: OS << ".p2alignl "; break;
474    case 8: llvm_unreachable("Unsupported alignment size!");
475    }
476
477    if (MAI.getAlignmentIsInBytes())
478      OS << ByteAlignment;
479    else
480      OS << Log2_32(ByteAlignment);
481
482    if (Value || MaxBytesToEmit) {
483      OS << ", 0x";
484      OS.write_hex(truncateToSize(Value, ValueSize));
485
486      if (MaxBytesToEmit)
487        OS << ", " << MaxBytesToEmit;
488    }
489    EmitEOL();
490    return;
491  }
492
493  // Non-power of two alignment.  This is not widely supported by assemblers.
494  // FIXME: Parameterize this based on MAI.
495  switch (ValueSize) {
496  default: llvm_unreachable("Invalid size for machine code value!");
497  case 1: OS << ".balign";  break;
498  case 2: OS << ".balignw"; break;
499  case 4: OS << ".balignl"; break;
500  case 8: llvm_unreachable("Unsupported alignment size!");
501  }
502
503  OS << ' ' << ByteAlignment;
504  OS << ", " << truncateToSize(Value, ValueSize);
505  if (MaxBytesToEmit)
506    OS << ", " << MaxBytesToEmit;
507  EmitEOL();
508}
509
510void MCAsmStreamer::EmitValueToOffset(const MCExpr *Offset,
511                                      unsigned char Value) {
512  // FIXME: Verify that Offset is associated with the current section.
513  OS << ".org " << *Offset << ", " << (unsigned) Value;
514  EmitEOL();
515}
516
517
518void MCAsmStreamer::EmitFileDirective(StringRef Filename) {
519  assert(MAI.hasSingleParameterDotFile());
520  OS << "\t.file\t";
521  PrintQuotedString(Filename, OS);
522  EmitEOL();
523}
524
525void MCAsmStreamer::EmitDwarfFileDirective(unsigned FileNo, StringRef Filename){
526  OS << "\t.file\t" << FileNo << ' ';
527  PrintQuotedString(Filename, OS);
528  EmitEOL();
529}
530
531
532void MCAsmStreamer::EmitInstruction(const MCInst &Inst) {
533  assert(CurSection && "Cannot emit contents before setting section!");
534
535  // Show the MCInst if enabled.
536  if (ShowInst) {
537    raw_ostream &OS = GetCommentOS();
538    OS << "inst: ";
539    Inst.print(OS, &MAI);
540    OS << "\n";
541  }
542
543  // Show the encoding in a comment if we have a code emitter.
544  if (Emitter) {
545    SmallString<256> Code;
546    raw_svector_ostream VecOS(Code);
547    Emitter->EncodeInstruction(Inst, VecOS);
548    VecOS.flush();
549
550    raw_ostream &OS = GetCommentOS();
551    OS << "encoding: [";
552    for (unsigned i = 0, e = Code.size(); i != e; ++i) {
553      if (i)
554        OS << ',';
555      OS << format("%#04x", uint8_t(Code[i]));
556    }
557    OS << "]\n";
558  }
559
560  // If we have an AsmPrinter, use that to print, otherwise dump the MCInst.
561  if (InstPrinter)
562    InstPrinter->printInst(&Inst);
563  else
564    Inst.print(OS, &MAI);
565  EmitEOL();
566}
567
568void MCAsmStreamer::Finish() {
569  OS.flush();
570}
571
572MCStreamer *llvm::createAsmStreamer(MCContext &Context,
573                                    formatted_raw_ostream &OS,
574                                    const MCAsmInfo &MAI, bool isLittleEndian,
575                                    bool isVerboseAsm, MCInstPrinter *IP,
576                                    MCCodeEmitter *CE, bool ShowInst) {
577  return new MCAsmStreamer(Context, OS, MAI, isLittleEndian, isVerboseAsm,
578                           IP, CE, ShowInst);
579}
580