MCAsmStreamer.cpp revision 6559d7688e24e204af273a1e1252639320a7b309 
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  bool IsLittleEndian, IsVerboseAsm;
33  MCInstPrinter *InstPrinter;
34  MCCodeEmitter *Emitter;
35
36  SmallString<128> CommentToEmit;
37  raw_svector_ostream CommentStream;
38public:
39  MCAsmStreamer(MCContext &Context, formatted_raw_ostream &os,
40                const MCAsmInfo &mai,
41                bool isLittleEndian, bool isVerboseAsm, MCInstPrinter *printer,
42                MCCodeEmitter *emitter)
43    : MCStreamer(Context), OS(os), MAI(mai), IsLittleEndian(isLittleEndian),
44      IsVerboseAsm(isVerboseAsm), InstPrinter(printer), Emitter(emitter),
45      CommentStream(CommentToEmit) {}
46  ~MCAsmStreamer() {}
47
48  bool isLittleEndian() const { return IsLittleEndian; }
49
50
51  inline void EmitEOL() {
52    // If we don't have any comments, just emit a \n.
53    if (!IsVerboseAsm) {
54      OS << '\n';
55      return;
56    }
57    EmitCommentsAndEOL();
58  }
59  void EmitCommentsAndEOL();
60
61  /// AddComment - Add a comment that can be emitted to the generated .s
62  /// file if applicable as a QoI issue to make the output of the compiler
63  /// more readable.  This only affects the MCAsmStreamer, and only when
64  /// verbose assembly output is enabled.
65  virtual void AddComment(const Twine &T);
66
67  /// GetCommentOS - Return a raw_ostream that comments can be written to.
68  /// Unlike AddComment, you are required to terminate comments with \n if you
69  /// use this method.
70  virtual raw_ostream &GetCommentOS() {
71    if (!IsVerboseAsm)
72      return nulls();  // Discard comments unless in verbose asm mode.
73    return CommentStream;
74  }
75
76  /// AddBlankLine - Emit a blank line to a .s file to pretty it up.
77  virtual void AddBlankLine() {
78    EmitEOL();
79  }
80
81  /// @name MCStreamer Interface
82  /// @{
83
84  virtual void SwitchSection(const MCSection *Section);
85
86  virtual void EmitLabel(MCSymbol *Symbol);
87
88  virtual void EmitAssemblerFlag(MCAssemblerFlag Flag);
89
90  virtual void EmitAssignment(MCSymbol *Symbol, const MCExpr *Value);
91
92  virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute);
93
94  virtual void EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue);
95
96  virtual void EmitCommonSymbol(MCSymbol *Symbol, uint64_t Size,
97                                unsigned ByteAlignment);
98
99  /// EmitLocalCommonSymbol - Emit a local common (.lcomm) symbol.
100  ///
101  /// @param Symbol - The common symbol to emit.
102  /// @param Size - The size of the common symbol.
103  virtual void EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size);
104
105  virtual void EmitZerofill(const MCSection *Section, MCSymbol *Symbol = 0,
106                            unsigned Size = 0, unsigned ByteAlignment = 0);
107
108  virtual void EmitBytes(StringRef Data, unsigned AddrSpace);
109
110  virtual void EmitValue(const MCExpr *Value, unsigned Size,unsigned AddrSpace);
111  virtual void EmitIntValue(uint64_t Value, unsigned Size, unsigned AddrSpace);
112
113  virtual void EmitFill(uint64_t NumBytes, uint8_t FillValue,
114                        unsigned AddrSpace);
115
116  virtual void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0,
117                                    unsigned ValueSize = 1,
118                                    unsigned MaxBytesToEmit = 0);
119
120  virtual void EmitValueToOffset(const MCExpr *Offset,
121                                 unsigned char Value = 0);
122
123  virtual void EmitInstruction(const MCInst &Inst);
124
125  virtual void Finish();
126
127  /// @}
128};
129
130} // end anonymous namespace.
131
132/// AddComment - Add a comment that can be emitted to the generated .s
133/// file if applicable as a QoI issue to make the output of the compiler
134/// more readable.  This only affects the MCAsmStreamer, and only when
135/// verbose assembly output is enabled.
136void MCAsmStreamer::AddComment(const Twine &T) {
137  if (!IsVerboseAsm) return;
138
139  // Make sure that CommentStream is flushed.
140  CommentStream.flush();
141
142  T.toVector(CommentToEmit);
143  // Each comment goes on its own line.
144  CommentToEmit.push_back('\n');
145
146  // Tell the comment stream that the vector changed underneath it.
147  CommentStream.resync();
148}
149
150void MCAsmStreamer::EmitCommentsAndEOL() {
151  if (CommentToEmit.empty() && CommentStream.GetNumBytesInBuffer() == 0) {
152    OS << '\n';
153    return;
154  }
155
156  CommentStream.flush();
157  StringRef Comments = CommentToEmit.str();
158
159  assert(Comments.back() == '\n' &&
160         "Comment array not newline terminated");
161  do {
162    // Emit a line of comments.
163    OS.PadToColumn(MAI.getCommentColumn());
164    size_t Position = Comments.find('\n');
165    OS << MAI.getCommentString() << ' ' << Comments.substr(0, Position) << '\n';
166
167    Comments = Comments.substr(Position+1);
168  } while (!Comments.empty());
169
170  CommentToEmit.clear();
171  // Tell the comment stream that the vector changed underneath it.
172  CommentStream.resync();
173}
174
175
176static inline int64_t truncateToSize(int64_t Value, unsigned Bytes) {
177  assert(Bytes && "Invalid size!");
178  return Value & ((uint64_t) (int64_t) -1 >> (64 - Bytes * 8));
179}
180
181static inline const MCExpr *truncateToSize(const MCExpr *Value,
182                                           unsigned Bytes) {
183  // FIXME: Do we really need this routine?
184  return Value;
185}
186
187void MCAsmStreamer::SwitchSection(const MCSection *Section) {
188  assert(Section && "Cannot switch to a null section!");
189  if (Section != CurSection) {
190    CurSection = Section;
191    Section->PrintSwitchToSection(MAI, OS);
192  }
193}
194
195void MCAsmStreamer::EmitLabel(MCSymbol *Symbol) {
196  assert(Symbol->isUndefined() && "Cannot define a symbol twice!");
197  assert(CurSection && "Cannot emit before setting section!");
198
199  OS << *Symbol << ":";
200  EmitEOL();
201  Symbol->setSection(*CurSection);
202}
203
204void MCAsmStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) {
205  switch (Flag) {
206  default: assert(0 && "Invalid flag!");
207  case MCAF_SubsectionsViaSymbols: OS << ".subsections_via_symbols"; break;
208  }
209  EmitEOL();
210}
211
212void MCAsmStreamer::EmitAssignment(MCSymbol *Symbol, const MCExpr *Value) {
213  // Only absolute symbols can be redefined.
214  assert((Symbol->isUndefined() || Symbol->isAbsolute()) &&
215         "Cannot define a symbol twice!");
216
217  OS << *Symbol << " = " << *Value;
218  EmitEOL();
219
220  // FIXME: Lift context changes into super class.
221  // FIXME: Set associated section.
222  Symbol->setValue(Value);
223}
224
225void MCAsmStreamer::EmitSymbolAttribute(MCSymbol *Symbol,
226                                        MCSymbolAttr Attribute) {
227  switch (Attribute) {
228  case MCSA_Invalid: assert(0 && "Invalid symbol attribute");
229  case MCSA_Global:         OS << MAI.getGlobalDirective(); break; // .globl
230  case MCSA_Hidden:         OS << ".hidden ";          break;
231  case MCSA_IndirectSymbol: OS << ".indirect_symbol "; break;
232  case MCSA_Internal:       OS << ".internal ";        break;
233  case MCSA_LazyReference:  OS << ".lazy_reference ";  break;
234  case MCSA_Local:          OS << ".local ";           break;
235  case MCSA_NoDeadStrip:    OS << ".no_dead_strip ";   break;
236  case MCSA_PrivateExtern:  OS << ".private_extern ";  break;
237  case MCSA_Protected:      OS << ".protected ";       break;
238  case MCSA_Reference:      OS << ".reference ";       break;
239  case MCSA_Weak:           OS << ".weak ";            break;
240  case MCSA_WeakDefinition: OS << ".weak_definition "; break;
241      // .weak_reference
242  case MCSA_WeakReference:  OS << MAI.getWeakRefDirective(); break;
243  }
244
245  OS << *Symbol;
246  EmitEOL();
247}
248
249void MCAsmStreamer::EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) {
250  OS << ".desc" << ' ' << *Symbol << ',' << DescValue;
251  EmitEOL();
252}
253
254void MCAsmStreamer::EmitCommonSymbol(MCSymbol *Symbol, uint64_t Size,
255                                     unsigned ByteAlignment) {
256  OS << "\t.comm\t" << *Symbol << ',' << Size;
257  if (ByteAlignment != 0) {
258    if (MAI.getAlignmentIsInBytes())
259      OS << ',' << ByteAlignment;
260    else
261      OS << ',' << Log2_32(ByteAlignment);
262  }
263  EmitEOL();
264}
265
266/// EmitLocalCommonSymbol - Emit a local common (.lcomm) symbol.
267///
268/// @param Symbol - The common symbol to emit.
269/// @param Size - The size of the common symbol.
270void MCAsmStreamer::EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size) {
271  assert(MAI.hasLCOMMDirective() && "Doesn't have .lcomm, can't emit it!");
272  OS << "\t.lcomm\t" << *Symbol << ',' << Size;
273  EmitEOL();
274}
275
276void MCAsmStreamer::EmitZerofill(const MCSection *Section, MCSymbol *Symbol,
277                                 unsigned Size, unsigned ByteAlignment) {
278  // Note: a .zerofill directive does not switch sections.
279  OS << ".zerofill ";
280
281  // This is a mach-o specific directive.
282  const MCSectionMachO *MOSection = ((const MCSectionMachO*)Section);
283  OS << MOSection->getSegmentName() << "," << MOSection->getSectionName();
284
285  if (Symbol != NULL) {
286    OS << ',' << *Symbol << ',' << Size;
287    if (ByteAlignment != 0)
288      OS << ',' << Log2_32(ByteAlignment);
289  }
290  EmitEOL();
291}
292
293static inline char toOctal(int X) { return (X&7)+'0'; }
294
295void MCAsmStreamer::EmitBytes(StringRef Data, unsigned AddrSpace) {
296  assert(CurSection && "Cannot emit contents before setting section!");
297  if (Data.empty()) return;
298
299  if (Data.size() == 1) {
300    OS << MAI.getData8bitsDirective(AddrSpace);
301    OS << (unsigned)(unsigned char)Data[0];
302    EmitEOL();
303    return;
304  }
305
306  // If the data ends with 0 and the target supports .asciz, use it, otherwise
307  // use .ascii
308  if (MAI.getAscizDirective() && Data.back() == 0) {
309    OS << MAI.getAscizDirective();
310    Data = Data.substr(0, Data.size()-1);
311  } else {
312    OS << MAI.getAsciiDirective();
313  }
314
315  OS << " \"";
316  for (unsigned i = 0, e = Data.size(); i != e; ++i) {
317    unsigned char C = Data[i];
318    if (C == '"' || C == '\\') {
319      OS << '\\' << (char)C;
320      continue;
321    }
322
323    if (isprint((unsigned char)C)) {
324      OS << (char)C;
325      continue;
326    }
327
328    switch (C) {
329    case '\b': OS << "\\b"; break;
330    case '\f': OS << "\\f"; break;
331    case '\n': OS << "\\n"; break;
332    case '\r': OS << "\\r"; break;
333    case '\t': OS << "\\t"; break;
334    default:
335      OS << '\\';
336      OS << toOctal(C >> 6);
337      OS << toOctal(C >> 3);
338      OS << toOctal(C >> 0);
339      break;
340    }
341  }
342  OS << '"';
343  EmitEOL();
344}
345
346/// EmitIntValue - Special case of EmitValue that avoids the client having
347/// to pass in a MCExpr for constant integers.
348void MCAsmStreamer::EmitIntValue(uint64_t Value, unsigned Size,
349                                 unsigned AddrSpace) {
350  assert(CurSection && "Cannot emit contents before setting section!");
351  const char *Directive = 0;
352  switch (Size) {
353  default: break;
354  case 1: Directive = MAI.getData8bitsDirective(AddrSpace); break;
355  case 2: Directive = MAI.getData16bitsDirective(AddrSpace); break;
356  case 4: Directive = MAI.getData32bitsDirective(AddrSpace); break;
357  case 8:
358    Directive = MAI.getData64bitsDirective(AddrSpace);
359    // If the target doesn't support 64-bit data, emit as two 32-bit halves.
360    if (Directive) break;
361    if (isLittleEndian()) {
362      EmitIntValue((uint32_t)(Value >> 0 ), 4, AddrSpace);
363      EmitIntValue((uint32_t)(Value >> 32), 4, AddrSpace);
364    } else {
365      EmitIntValue((uint32_t)(Value >> 32), 4, AddrSpace);
366      EmitIntValue((uint32_t)(Value >> 0 ), 4, AddrSpace);
367    }
368    return;
369  }
370
371  assert(Directive && "Invalid size for machine code value!");
372  OS << Directive << truncateToSize(Value, Size);
373  EmitEOL();
374}
375
376void MCAsmStreamer::EmitValue(const MCExpr *Value, unsigned Size,
377                              unsigned AddrSpace) {
378  assert(CurSection && "Cannot emit contents before setting section!");
379  const char *Directive = 0;
380  switch (Size) {
381  default: break;
382  case 1: Directive = MAI.getData8bitsDirective(AddrSpace); break;
383  case 2: Directive = MAI.getData16bitsDirective(AddrSpace); break;
384  case 4: Directive = MAI.getData32bitsDirective(AddrSpace); break;
385  case 8: Directive = MAI.getData64bitsDirective(AddrSpace); break;
386  }
387
388  assert(Directive && "Invalid size for machine code value!");
389  OS << Directive << *truncateToSize(Value, Size);
390  EmitEOL();
391}
392
393/// EmitFill - Emit NumBytes bytes worth of the value specified by
394/// FillValue.  This implements directives such as '.space'.
395void MCAsmStreamer::EmitFill(uint64_t NumBytes, uint8_t FillValue,
396                             unsigned AddrSpace) {
397  if (NumBytes == 0) return;
398
399  if (AddrSpace == 0)
400    if (const char *ZeroDirective = MAI.getZeroDirective()) {
401      OS << ZeroDirective << NumBytes;
402      if (FillValue != 0)
403        OS << ',' << (int)FillValue;
404      EmitEOL();
405      return;
406    }
407
408  // Emit a byte at a time.
409  MCStreamer::EmitFill(NumBytes, FillValue, AddrSpace);
410}
411
412void MCAsmStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value,
413                                         unsigned ValueSize,
414                                         unsigned MaxBytesToEmit) {
415  // Some assemblers don't support non-power of two alignments, so we always
416  // emit alignments as a power of two if possible.
417  if (isPowerOf2_32(ByteAlignment)) {
418    switch (ValueSize) {
419    default: llvm_unreachable("Invalid size for machine code value!");
420    case 1: OS << MAI.getAlignDirective(); break;
421    // FIXME: use MAI for this!
422    case 2: OS << ".p2alignw "; break;
423    case 4: OS << ".p2alignl "; break;
424    case 8: llvm_unreachable("Unsupported alignment size!");
425    }
426
427    if (MAI.getAlignmentIsInBytes())
428      OS << ByteAlignment;
429    else
430      OS << Log2_32(ByteAlignment);
431
432    if (Value || MaxBytesToEmit) {
433      OS << ", 0x";
434      OS.write_hex(truncateToSize(Value, ValueSize));
435
436      if (MaxBytesToEmit)
437        OS << ", " << MaxBytesToEmit;
438    }
439    EmitEOL();
440    return;
441  }
442
443  // Non-power of two alignment.  This is not widely supported by assemblers.
444  // FIXME: Parameterize this based on MAI.
445  switch (ValueSize) {
446  default: llvm_unreachable("Invalid size for machine code value!");
447  case 1: OS << ".balign";  break;
448  case 2: OS << ".balignw"; break;
449  case 4: OS << ".balignl"; break;
450  case 8: llvm_unreachable("Unsupported alignment size!");
451  }
452
453  OS << ' ' << ByteAlignment;
454  OS << ", " << truncateToSize(Value, ValueSize);
455  if (MaxBytesToEmit)
456    OS << ", " << MaxBytesToEmit;
457  EmitEOL();
458}
459
460void MCAsmStreamer::EmitValueToOffset(const MCExpr *Offset,
461                                      unsigned char Value) {
462  // FIXME: Verify that Offset is associated with the current section.
463  OS << ".org " << *Offset << ", " << (unsigned) Value;
464  EmitEOL();
465}
466
467void MCAsmStreamer::EmitInstruction(const MCInst &Inst) {
468  assert(CurSection && "Cannot emit contents before setting section!");
469
470  // If we have an AsmPrinter, use that to print.
471  if (InstPrinter) {
472    InstPrinter->printInst(&Inst);
473    EmitEOL();
474
475    // Show the encoding if we have a code emitter.
476    if (Emitter) {
477      SmallString<256> Code;
478      raw_svector_ostream VecOS(Code);
479      Emitter->EncodeInstruction(Inst, VecOS);
480      VecOS.flush();
481
482      OS.indent(20);
483      OS << " # encoding: [";
484      for (unsigned i = 0, e = Code.size(); i != e; ++i) {
485        if (i)
486          OS << ',';
487        OS << format("%#04x", uint8_t(Code[i]));
488      }
489      OS << "]\n";
490    }
491
492    return;
493  }
494
495  // Otherwise fall back to a structural printing for now. Eventually we should
496  // always have access to the target specific printer.
497  Inst.print(OS, &MAI);
498  EmitEOL();
499}
500
501void MCAsmStreamer::Finish() {
502  OS.flush();
503}
504
505MCStreamer *llvm::createAsmStreamer(MCContext &Context,
506                                    formatted_raw_ostream &OS,
507                                    const MCAsmInfo &MAI, bool isLittleEndian,
508                                    bool isVerboseAsm, MCInstPrinter *IP,
509                                    MCCodeEmitter *CE) {
510  return new MCAsmStreamer(Context, OS, MAI, isLittleEndian, isVerboseAsm,
511                           IP, CE);
512}
513