DWARFFormValue.cpp revision 288943 
1//===-- DWARFFormValue.cpp ------------------------------------------------===//
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 "SyntaxHighlighting.h"
11#include "llvm/ADT/ArrayRef.h"
12#include "llvm/ADT/StringRef.h"
13#include "llvm/DebugInfo/DWARF/DWARFCompileUnit.h"
14#include "llvm/DebugInfo/DWARF/DWARFContext.h"
15#include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
16#include "llvm/Support/Debug.h"
17#include "llvm/Support/Dwarf.h"
18#include "llvm/Support/Format.h"
19#include "llvm/Support/raw_ostream.h"
20#include <cassert>
21#include <climits>
22using namespace llvm;
23using namespace dwarf;
24using namespace syntax;
25
26namespace {
27uint8_t getRefAddrSize(uint8_t AddrSize, uint16_t Version) {
28  // FIXME: Support DWARF64.
29  return (Version == 2) ? AddrSize : 4;
30}
31
32template <uint8_t AddrSize, uint8_t RefAddrSize>
33ArrayRef<uint8_t> makeFixedFormSizesArrayRef() {
34  static const uint8_t sizes[] = {
35    0,           // 0x00 unused
36    AddrSize,    // 0x01 DW_FORM_addr
37    0,           // 0x02 unused
38    0,           // 0x03 DW_FORM_block2
39    0,           // 0x04 DW_FORM_block4
40    2,           // 0x05 DW_FORM_data2
41    4,           // 0x06 DW_FORM_data4
42    8,           // 0x07 DW_FORM_data8
43    0,           // 0x08 DW_FORM_string
44    0,           // 0x09 DW_FORM_block
45    0,           // 0x0a DW_FORM_block1
46    1,           // 0x0b DW_FORM_data1
47    1,           // 0x0c DW_FORM_flag
48    0,           // 0x0d DW_FORM_sdata
49    4,           // 0x0e DW_FORM_strp
50    0,           // 0x0f DW_FORM_udata
51    RefAddrSize, // 0x10 DW_FORM_ref_addr
52    1,           // 0x11 DW_FORM_ref1
53    2,           // 0x12 DW_FORM_ref2
54    4,           // 0x13 DW_FORM_ref4
55    8,           // 0x14 DW_FORM_ref8
56    0,           // 0x15 DW_FORM_ref_udata
57    0,           // 0x16 DW_FORM_indirect
58    4,           // 0x17 DW_FORM_sec_offset
59    0,           // 0x18 DW_FORM_exprloc
60    0,           // 0x19 DW_FORM_flag_present
61  };
62  return makeArrayRef(sizes);
63}
64}
65
66ArrayRef<uint8_t> DWARFFormValue::getFixedFormSizes(uint8_t AddrSize,
67                                                    uint16_t Version) {
68  uint8_t RefAddrSize = getRefAddrSize(AddrSize, Version);
69  if (AddrSize == 4 && RefAddrSize == 4)
70    return makeFixedFormSizesArrayRef<4, 4>();
71  if (AddrSize == 4 && RefAddrSize == 8)
72    return makeFixedFormSizesArrayRef<4, 8>();
73  if (AddrSize == 8 && RefAddrSize == 4)
74    return makeFixedFormSizesArrayRef<8, 4>();
75  if (AddrSize == 8 && RefAddrSize == 8)
76    return makeFixedFormSizesArrayRef<8, 8>();
77  return None;
78}
79
80static const DWARFFormValue::FormClass DWARF4FormClasses[] = {
81  DWARFFormValue::FC_Unknown,       // 0x0
82  DWARFFormValue::FC_Address,       // 0x01 DW_FORM_addr
83  DWARFFormValue::FC_Unknown,       // 0x02 unused
84  DWARFFormValue::FC_Block,         // 0x03 DW_FORM_block2
85  DWARFFormValue::FC_Block,         // 0x04 DW_FORM_block4
86  DWARFFormValue::FC_Constant,      // 0x05 DW_FORM_data2
87  // --- These can be FC_SectionOffset in DWARF3 and below:
88  DWARFFormValue::FC_Constant,      // 0x06 DW_FORM_data4
89  DWARFFormValue::FC_Constant,      // 0x07 DW_FORM_data8
90  // ---
91  DWARFFormValue::FC_String,        // 0x08 DW_FORM_string
92  DWARFFormValue::FC_Block,         // 0x09 DW_FORM_block
93  DWARFFormValue::FC_Block,         // 0x0a DW_FORM_block1
94  DWARFFormValue::FC_Constant,      // 0x0b DW_FORM_data1
95  DWARFFormValue::FC_Flag,          // 0x0c DW_FORM_flag
96  DWARFFormValue::FC_Constant,      // 0x0d DW_FORM_sdata
97  DWARFFormValue::FC_String,        // 0x0e DW_FORM_strp
98  DWARFFormValue::FC_Constant,      // 0x0f DW_FORM_udata
99  DWARFFormValue::FC_Reference,     // 0x10 DW_FORM_ref_addr
100  DWARFFormValue::FC_Reference,     // 0x11 DW_FORM_ref1
101  DWARFFormValue::FC_Reference,     // 0x12 DW_FORM_ref2
102  DWARFFormValue::FC_Reference,     // 0x13 DW_FORM_ref4
103  DWARFFormValue::FC_Reference,     // 0x14 DW_FORM_ref8
104  DWARFFormValue::FC_Reference,     // 0x15 DW_FORM_ref_udata
105  DWARFFormValue::FC_Indirect,      // 0x16 DW_FORM_indirect
106  DWARFFormValue::FC_SectionOffset, // 0x17 DW_FORM_sec_offset
107  DWARFFormValue::FC_Exprloc,       // 0x18 DW_FORM_exprloc
108  DWARFFormValue::FC_Flag,          // 0x19 DW_FORM_flag_present
109};
110
111bool DWARFFormValue::isFormClass(DWARFFormValue::FormClass FC) const {
112  // First, check DWARF4 form classes.
113  if (Form < ArrayRef<FormClass>(DWARF4FormClasses).size() &&
114      DWARF4FormClasses[Form] == FC)
115    return true;
116  // Check more forms from DWARF4 and DWARF5 proposals.
117  switch (Form) {
118  case DW_FORM_ref_sig8:
119  case DW_FORM_GNU_ref_alt:
120    return (FC == FC_Reference);
121  case DW_FORM_GNU_addr_index:
122    return (FC == FC_Address);
123  case DW_FORM_GNU_str_index:
124  case DW_FORM_GNU_strp_alt:
125    return (FC == FC_String);
126  }
127  // In DWARF3 DW_FORM_data4 and DW_FORM_data8 served also as a section offset.
128  // Don't check for DWARF version here, as some producers may still do this
129  // by mistake.
130  return (Form == DW_FORM_data4 || Form == DW_FORM_data8) &&
131         FC == FC_SectionOffset;
132}
133
134bool DWARFFormValue::extractValue(DataExtractor data, uint32_t *offset_ptr,
135                                  const DWARFUnit *cu) {
136  bool indirect = false;
137  bool is_block = false;
138  Value.data = nullptr;
139  // Read the value for the form into value and follow and DW_FORM_indirect
140  // instances we run into
141  do {
142    indirect = false;
143    switch (Form) {
144    case DW_FORM_addr:
145    case DW_FORM_ref_addr: {
146      if (!cu)
147        return false;
148      uint16_t AddrSize =
149          (Form == DW_FORM_addr)
150              ? cu->getAddressByteSize()
151              : getRefAddrSize(cu->getAddressByteSize(), cu->getVersion());
152      RelocAddrMap::const_iterator AI = cu->getRelocMap()->find(*offset_ptr);
153      if (AI != cu->getRelocMap()->end()) {
154        const std::pair<uint8_t, int64_t> &R = AI->second;
155        Value.uval = data.getUnsigned(offset_ptr, AddrSize) + R.second;
156      } else
157        Value.uval = data.getUnsigned(offset_ptr, AddrSize);
158      break;
159    }
160    case DW_FORM_exprloc:
161    case DW_FORM_block:
162      Value.uval = data.getULEB128(offset_ptr);
163      is_block = true;
164      break;
165    case DW_FORM_block1:
166      Value.uval = data.getU8(offset_ptr);
167      is_block = true;
168      break;
169    case DW_FORM_block2:
170      Value.uval = data.getU16(offset_ptr);
171      is_block = true;
172      break;
173    case DW_FORM_block4:
174      Value.uval = data.getU32(offset_ptr);
175      is_block = true;
176      break;
177    case DW_FORM_data1:
178    case DW_FORM_ref1:
179    case DW_FORM_flag:
180      Value.uval = data.getU8(offset_ptr);
181      break;
182    case DW_FORM_data2:
183    case DW_FORM_ref2:
184      Value.uval = data.getU16(offset_ptr);
185      break;
186    case DW_FORM_data4:
187    case DW_FORM_ref4: {
188      Value.uval = data.getU32(offset_ptr);
189      if (!cu)
190        break;
191      RelocAddrMap::const_iterator AI = cu->getRelocMap()->find(*offset_ptr-4);
192      if (AI != cu->getRelocMap()->end())
193        Value.uval += AI->second.second;
194      break;
195    }
196    case DW_FORM_data8:
197    case DW_FORM_ref8:
198      Value.uval = data.getU64(offset_ptr);
199      break;
200    case DW_FORM_sdata:
201      Value.sval = data.getSLEB128(offset_ptr);
202      break;
203    case DW_FORM_udata:
204    case DW_FORM_ref_udata:
205      Value.uval = data.getULEB128(offset_ptr);
206      break;
207    case DW_FORM_string:
208      Value.cstr = data.getCStr(offset_ptr);
209      break;
210    case DW_FORM_indirect:
211      Form = data.getULEB128(offset_ptr);
212      indirect = true;
213      break;
214    case DW_FORM_sec_offset:
215    case DW_FORM_strp:
216    case DW_FORM_GNU_ref_alt:
217    case DW_FORM_GNU_strp_alt: {
218      // FIXME: This is 64-bit for DWARF64.
219      Value.uval = data.getU32(offset_ptr);
220      if (!cu)
221        break;
222      RelocAddrMap::const_iterator AI =
223          cu->getRelocMap()->find(*offset_ptr - 4);
224      if (AI != cu->getRelocMap()->end())
225        Value.uval += AI->second.second;
226      break;
227    }
228    case DW_FORM_flag_present:
229      Value.uval = 1;
230      break;
231    case DW_FORM_ref_sig8:
232      Value.uval = data.getU64(offset_ptr);
233      break;
234    case DW_FORM_GNU_addr_index:
235    case DW_FORM_GNU_str_index:
236      Value.uval = data.getULEB128(offset_ptr);
237      break;
238    default:
239      return false;
240    }
241  } while (indirect);
242
243  if (is_block) {
244    StringRef str = data.getData().substr(*offset_ptr, Value.uval);
245    Value.data = nullptr;
246    if (!str.empty()) {
247      Value.data = reinterpret_cast<const uint8_t *>(str.data());
248      *offset_ptr += Value.uval;
249    }
250  }
251
252  return true;
253}
254
255bool
256DWARFFormValue::skipValue(DataExtractor debug_info_data, uint32_t* offset_ptr,
257                          const DWARFUnit *cu) const {
258  return DWARFFormValue::skipValue(Form, debug_info_data, offset_ptr, cu);
259}
260
261bool
262DWARFFormValue::skipValue(uint16_t form, DataExtractor debug_info_data,
263                          uint32_t *offset_ptr, const DWARFUnit *cu) {
264  bool indirect = false;
265  do {
266    switch (form) {
267    // Blocks if inlined data that have a length field and the data bytes
268    // inlined in the .debug_info
269    case DW_FORM_exprloc:
270    case DW_FORM_block: {
271      uint64_t size = debug_info_data.getULEB128(offset_ptr);
272      *offset_ptr += size;
273      return true;
274    }
275    case DW_FORM_block1: {
276      uint8_t size = debug_info_data.getU8(offset_ptr);
277      *offset_ptr += size;
278      return true;
279    }
280    case DW_FORM_block2: {
281      uint16_t size = debug_info_data.getU16(offset_ptr);
282      *offset_ptr += size;
283      return true;
284    }
285    case DW_FORM_block4: {
286      uint32_t size = debug_info_data.getU32(offset_ptr);
287      *offset_ptr += size;
288      return true;
289    }
290
291    // Inlined NULL terminated C-strings
292    case DW_FORM_string:
293      debug_info_data.getCStr(offset_ptr);
294      return true;
295
296    // Compile unit address sized values
297    case DW_FORM_addr:
298      *offset_ptr += cu->getAddressByteSize();
299      return true;
300    case DW_FORM_ref_addr:
301      *offset_ptr += getRefAddrSize(cu->getAddressByteSize(), cu->getVersion());
302      return true;
303
304    // 0 byte values - implied from the form.
305    case DW_FORM_flag_present:
306      return true;
307
308    // 1 byte values
309    case DW_FORM_data1:
310    case DW_FORM_flag:
311    case DW_FORM_ref1:
312      *offset_ptr += 1;
313      return true;
314
315    // 2 byte values
316    case DW_FORM_data2:
317    case DW_FORM_ref2:
318      *offset_ptr += 2;
319      return true;
320
321    // 4 byte values
322    case DW_FORM_data4:
323    case DW_FORM_ref4:
324      *offset_ptr += 4;
325      return true;
326
327    // 8 byte values
328    case DW_FORM_data8:
329    case DW_FORM_ref8:
330    case DW_FORM_ref_sig8:
331      *offset_ptr += 8;
332      return true;
333
334    // signed or unsigned LEB 128 values
335    //  case DW_FORM_APPLE_db_str:
336    case DW_FORM_sdata:
337    case DW_FORM_udata:
338    case DW_FORM_ref_udata:
339    case DW_FORM_GNU_str_index:
340    case DW_FORM_GNU_addr_index:
341      debug_info_data.getULEB128(offset_ptr);
342      return true;
343
344    case DW_FORM_indirect:
345      indirect = true;
346      form = debug_info_data.getULEB128(offset_ptr);
347      break;
348
349    // FIXME: 4 for DWARF32, 8 for DWARF64.
350    case DW_FORM_sec_offset:
351    case DW_FORM_strp:
352    case DW_FORM_GNU_ref_alt:
353    case DW_FORM_GNU_strp_alt:
354      *offset_ptr += 4;
355      return true;
356
357    default:
358      return false;
359    }
360  } while (indirect);
361  return true;
362}
363
364void
365DWARFFormValue::dump(raw_ostream &OS, const DWARFUnit *cu) const {
366  uint64_t uvalue = Value.uval;
367  bool cu_relative_offset = false;
368
369  switch (Form) {
370  case DW_FORM_addr:      OS << format("0x%016" PRIx64, uvalue); break;
371  case DW_FORM_GNU_addr_index: {
372    OS << format(" indexed (%8.8x) address = ", (uint32_t)uvalue);
373    uint64_t Address;
374    if (cu->getAddrOffsetSectionItem(uvalue, Address))
375      OS << format("0x%016" PRIx64, Address);
376    else
377      OS << "<no .debug_addr section>";
378    break;
379  }
380  case DW_FORM_flag_present: OS << "true"; break;
381  case DW_FORM_flag:
382  case DW_FORM_data1:     OS << format("0x%02x", (uint8_t)uvalue); break;
383  case DW_FORM_data2:     OS << format("0x%04x", (uint16_t)uvalue); break;
384  case DW_FORM_data4:     OS << format("0x%08x", (uint32_t)uvalue); break;
385  case DW_FORM_ref_sig8:
386  case DW_FORM_data8:     OS << format("0x%016" PRIx64, uvalue); break;
387  case DW_FORM_string:
388    OS << '"';
389    OS.write_escaped(Value.cstr);
390    OS << '"';
391    break;
392  case DW_FORM_exprloc:
393  case DW_FORM_block:
394  case DW_FORM_block1:
395  case DW_FORM_block2:
396  case DW_FORM_block4:
397    if (uvalue > 0) {
398      switch (Form) {
399      case DW_FORM_exprloc:
400      case DW_FORM_block:  OS << format("<0x%" PRIx64 "> ", uvalue);     break;
401      case DW_FORM_block1: OS << format("<0x%2.2x> ", (uint8_t)uvalue);  break;
402      case DW_FORM_block2: OS << format("<0x%4.4x> ", (uint16_t)uvalue); break;
403      case DW_FORM_block4: OS << format("<0x%8.8x> ", (uint32_t)uvalue); break;
404      default: break;
405      }
406
407      const uint8_t* data_ptr = Value.data;
408      if (data_ptr) {
409        // uvalue contains size of block
410        const uint8_t* end_data_ptr = data_ptr + uvalue;
411        while (data_ptr < end_data_ptr) {
412          OS << format("%2.2x ", *data_ptr);
413          ++data_ptr;
414        }
415      }
416      else
417        OS << "NULL";
418    }
419    break;
420
421  case DW_FORM_sdata:     OS << Value.sval; break;
422  case DW_FORM_udata:     OS << Value.uval; break;
423  case DW_FORM_strp: {
424    OS << format(" .debug_str[0x%8.8x] = ", (uint32_t)uvalue);
425    dumpString(OS, cu);
426    break;
427  }
428  case DW_FORM_GNU_str_index: {
429    OS << format(" indexed (%8.8x) string = ", (uint32_t)uvalue);
430    dumpString(OS, cu);
431    break;
432  }
433  case DW_FORM_GNU_strp_alt: {
434    OS << format("alt indirect string, offset: 0x%" PRIx64 "", uvalue);
435    dumpString(OS, cu);
436    break;
437  }
438  case DW_FORM_ref_addr:
439    OS << format("0x%016" PRIx64, uvalue);
440    break;
441  case DW_FORM_ref1:
442    cu_relative_offset = true;
443    OS << format("cu + 0x%2.2x", (uint8_t)uvalue);
444    break;
445  case DW_FORM_ref2:
446    cu_relative_offset = true;
447    OS << format("cu + 0x%4.4x", (uint16_t)uvalue);
448    break;
449  case DW_FORM_ref4:
450    cu_relative_offset = true;
451    OS << format("cu + 0x%4.4x", (uint32_t)uvalue);
452    break;
453  case DW_FORM_ref8:
454    cu_relative_offset = true;
455    OS << format("cu + 0x%8.8" PRIx64, uvalue);
456    break;
457  case DW_FORM_ref_udata:
458    cu_relative_offset = true;
459    OS << format("cu + 0x%" PRIx64, uvalue);
460    break;
461  case DW_FORM_GNU_ref_alt:
462    OS << format("<alt 0x%" PRIx64 ">", uvalue);
463    break;
464
465    // All DW_FORM_indirect attributes should be resolved prior to calling
466    // this function
467  case DW_FORM_indirect:
468    OS << "DW_FORM_indirect";
469    break;
470
471    // Should be formatted to 64-bit for DWARF64.
472  case DW_FORM_sec_offset:
473    OS << format("0x%08x", (uint32_t)uvalue);
474    break;
475
476  default:
477    OS << format("DW_FORM(0x%4.4x)", Form);
478    break;
479  }
480
481  if (cu_relative_offset) {
482    OS << " => {";
483    WithColor(OS, syntax::Address).get()
484      << format("0x%8.8" PRIx64, uvalue + (cu ? cu->getOffset() : 0));
485    OS << "}";
486  }
487}
488
489void DWARFFormValue::dumpString(raw_ostream &OS, const DWARFUnit *U) const {
490  Optional<const char *> DbgStr = getAsCString(U);
491  if (DbgStr.hasValue()) {
492    raw_ostream &COS = WithColor(OS, syntax::String);
493    COS << '"';
494    COS.write_escaped(DbgStr.getValue());
495    COS << '"';
496  }
497}
498
499Optional<const char *> DWARFFormValue::getAsCString(const DWARFUnit *U) const {
500  if (!isFormClass(FC_String))
501    return None;
502  if (Form == DW_FORM_string)
503    return Value.cstr;
504  // FIXME: Add support for DW_FORM_GNU_strp_alt
505  if (Form == DW_FORM_GNU_strp_alt || U == nullptr)
506    return None;
507  uint32_t Offset = Value.uval;
508  if (Form == DW_FORM_GNU_str_index) {
509    uint32_t StrOffset;
510    if (!U->getStringOffsetSectionItem(Offset, StrOffset))
511      return None;
512    Offset = StrOffset;
513  }
514  if (const char *Str = U->getStringExtractor().getCStr(&Offset)) {
515    return Str;
516  }
517  return None;
518}
519
520Optional<uint64_t> DWARFFormValue::getAsAddress(const DWARFUnit *U) const {
521  if (!isFormClass(FC_Address))
522    return None;
523  if (Form == DW_FORM_GNU_addr_index) {
524    uint32_t Index = Value.uval;
525    uint64_t Result;
526    if (!U || !U->getAddrOffsetSectionItem(Index, Result))
527      return None;
528    return Result;
529  }
530  return Value.uval;
531}
532
533Optional<uint64_t> DWARFFormValue::getAsReference(const DWARFUnit *U) const {
534  if (!isFormClass(FC_Reference))
535    return None;
536  switch (Form) {
537  case DW_FORM_ref1:
538  case DW_FORM_ref2:
539  case DW_FORM_ref4:
540  case DW_FORM_ref8:
541  case DW_FORM_ref_udata:
542    if (!U)
543      return None;
544    return Value.uval + U->getOffset();
545  case DW_FORM_ref_addr:
546    return Value.uval;
547  // FIXME: Add proper support for DW_FORM_ref_sig8 and DW_FORM_GNU_ref_alt.
548  default:
549    return None;
550  }
551}
552
553Optional<uint64_t> DWARFFormValue::getAsSectionOffset() const {
554  if (!isFormClass(FC_SectionOffset))
555    return None;
556  return Value.uval;
557}
558
559Optional<uint64_t> DWARFFormValue::getAsUnsignedConstant() const {
560  if ((!isFormClass(FC_Constant) && !isFormClass(FC_Flag))
561      || Form == DW_FORM_sdata)
562    return None;
563  return Value.uval;
564}
565
566Optional<int64_t> DWARFFormValue::getAsSignedConstant() const {
567  if ((!isFormClass(FC_Constant) && !isFormClass(FC_Flag)) ||
568      (Form == DW_FORM_udata && uint64_t(LLONG_MAX) < Value.uval))
569    return None;
570  switch (Form) {
571  case DW_FORM_data4:
572    return int32_t(Value.uval);
573  case DW_FORM_data2:
574    return int16_t(Value.uval);
575  case DW_FORM_data1:
576    return int8_t(Value.uval);
577  case DW_FORM_sdata:
578  case DW_FORM_data8:
579  default:
580    return Value.sval;
581  }
582}
583
584Optional<ArrayRef<uint8_t>> DWARFFormValue::getAsBlock() const {
585  if (!isFormClass(FC_Block) && !isFormClass(FC_Exprloc))
586    return None;
587  return ArrayRef<uint8_t>(Value.data, Value.uval);
588}
589
590