1//===-- DWARFFormValue.cpp ------------------------------------------------===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8
9#include <cassert>
10#include <optional>
11
12#include "lldb/Core/Module.h"
13#include "lldb/Core/dwarf.h"
14#include "lldb/Symbol/ObjectFile.h"
15#include "lldb/Utility/Stream.h"
16
17#include "DWARFDebugInfo.h"
18#include "DWARFFormValue.h"
19#include "DWARFUnit.h"
20
21class DWARFUnit;
22
23using namespace lldb_private;
24using namespace lldb_private::dwarf;
25
26void DWARFFormValue::Clear() {
27  m_unit = nullptr;
28  m_form = 0;
29  m_value = ValueTypeTag();
30}
31
32bool DWARFFormValue::ExtractValue(const DWARFDataExtractor &data,
33                                  lldb::offset_t *offset_ptr) {
34  if (m_form == DW_FORM_implicit_const)
35    return true;
36
37  bool indirect = false;
38  bool is_block = false;
39  m_value.data = nullptr;
40  uint8_t ref_addr_size;
41  // Read the value for the form into value and follow and DW_FORM_indirect
42  // instances we run into
43  do {
44    indirect = false;
45    switch (m_form) {
46    case DW_FORM_addr:
47      assert(m_unit);
48      m_value.value.uval =
49          data.GetMaxU64(offset_ptr, DWARFUnit::GetAddressByteSize(m_unit));
50      break;
51    case DW_FORM_block1:
52      m_value.value.uval = data.GetU8(offset_ptr);
53      is_block = true;
54      break;
55    case DW_FORM_block2:
56      m_value.value.uval = data.GetU16(offset_ptr);
57      is_block = true;
58      break;
59    case DW_FORM_block4:
60      m_value.value.uval = data.GetU32(offset_ptr);
61      is_block = true;
62      break;
63    case DW_FORM_data16:
64      m_value.value.uval = 16;
65      is_block = true;
66      break;
67    case DW_FORM_exprloc:
68    case DW_FORM_block:
69      m_value.value.uval = data.GetULEB128(offset_ptr);
70      is_block = true;
71      break;
72    case DW_FORM_string:
73      m_value.value.cstr = data.GetCStr(offset_ptr);
74      break;
75    case DW_FORM_sdata:
76      m_value.value.sval = data.GetSLEB128(offset_ptr);
77      break;
78    case DW_FORM_strp:
79    case DW_FORM_line_strp:
80    case DW_FORM_sec_offset:
81      m_value.value.uval = data.GetMaxU64(offset_ptr, 4);
82      break;
83    case DW_FORM_addrx1:
84    case DW_FORM_strx1:
85    case DW_FORM_ref1:
86    case DW_FORM_data1:
87    case DW_FORM_flag:
88      m_value.value.uval = data.GetU8(offset_ptr);
89      break;
90    case DW_FORM_addrx2:
91    case DW_FORM_strx2:
92    case DW_FORM_ref2:
93    case DW_FORM_data2:
94      m_value.value.uval = data.GetU16(offset_ptr);
95      break;
96    case DW_FORM_addrx3:
97    case DW_FORM_strx3:
98      m_value.value.uval = data.GetMaxU64(offset_ptr, 3);
99      break;
100    case DW_FORM_addrx4:
101    case DW_FORM_strx4:
102    case DW_FORM_ref4:
103    case DW_FORM_data4:
104      m_value.value.uval = data.GetU32(offset_ptr);
105      break;
106    case DW_FORM_data8:
107    case DW_FORM_ref8:
108    case DW_FORM_ref_sig8:
109      m_value.value.uval = data.GetU64(offset_ptr);
110      break;
111    case DW_FORM_addrx:
112    case DW_FORM_loclistx:
113    case DW_FORM_rnglistx:
114    case DW_FORM_strx:
115    case DW_FORM_udata:
116    case DW_FORM_ref_udata:
117    case DW_FORM_GNU_str_index:
118    case DW_FORM_GNU_addr_index:
119      m_value.value.uval = data.GetULEB128(offset_ptr);
120      break;
121    case DW_FORM_ref_addr:
122      assert(m_unit);
123      if (m_unit->GetVersion() <= 2)
124        ref_addr_size = m_unit->GetAddressByteSize();
125      else
126        ref_addr_size = 4;
127      m_value.value.uval = data.GetMaxU64(offset_ptr, ref_addr_size);
128      break;
129    case DW_FORM_indirect:
130      m_form = data.GetULEB128(offset_ptr);
131      indirect = true;
132      break;
133    case DW_FORM_flag_present:
134      m_value.value.uval = 1;
135      break;
136    default:
137      return false;
138    }
139  } while (indirect);
140
141  if (is_block) {
142    m_value.data = data.PeekData(*offset_ptr, m_value.value.uval);
143    if (m_value.data != nullptr) {
144      *offset_ptr += m_value.value.uval;
145    }
146  }
147
148  return true;
149}
150
151struct FormSize {
152  uint8_t valid:1, size:7;
153};
154static FormSize g_form_sizes[] = {
155    {0, 0}, // 0x00 unused
156    {0, 0}, // 0x01 DW_FORM_addr
157    {0, 0}, // 0x02 unused
158    {0, 0}, // 0x03 DW_FORM_block2
159    {0, 0}, // 0x04 DW_FORM_block4
160    {1, 2}, // 0x05 DW_FORM_data2
161    {1, 4}, // 0x06 DW_FORM_data4
162    {1, 8}, // 0x07 DW_FORM_data8
163    {0, 0}, // 0x08 DW_FORM_string
164    {0, 0}, // 0x09 DW_FORM_block
165    {0, 0}, // 0x0a DW_FORM_block1
166    {1, 1}, // 0x0b DW_FORM_data1
167    {1, 1}, // 0x0c DW_FORM_flag
168    {0, 0}, // 0x0d DW_FORM_sdata
169    {1, 4}, // 0x0e DW_FORM_strp
170    {0, 0}, // 0x0f DW_FORM_udata
171    {0, 0}, // 0x10 DW_FORM_ref_addr (addr size for DWARF2 and earlier, 4 bytes
172            // for DWARF32, 8 bytes for DWARF32 in DWARF 3 and later
173    {1, 1},  // 0x11 DW_FORM_ref1
174    {1, 2},  // 0x12 DW_FORM_ref2
175    {1, 4},  // 0x13 DW_FORM_ref4
176    {1, 8},  // 0x14 DW_FORM_ref8
177    {0, 0},  // 0x15 DW_FORM_ref_udata
178    {0, 0},  // 0x16 DW_FORM_indirect
179    {1, 4},  // 0x17 DW_FORM_sec_offset
180    {0, 0},  // 0x18 DW_FORM_exprloc
181    {1, 0},  // 0x19 DW_FORM_flag_present
182    {0, 0},  // 0x1a DW_FORM_strx (ULEB128)
183    {0, 0},  // 0x1b DW_FORM_addrx (ULEB128)
184    {1, 4},  // 0x1c DW_FORM_ref_sup4
185    {0, 0},  // 0x1d DW_FORM_strp_sup (4 bytes for DWARF32, 8 bytes for DWARF64)
186    {1, 16}, // 0x1e DW_FORM_data16
187    {1, 4},  // 0x1f DW_FORM_line_strp
188    {1, 8},  // 0x20 DW_FORM_ref_sig8
189};
190
191std::optional<uint8_t> DWARFFormValue::GetFixedSize(dw_form_t form,
192                                                    const DWARFUnit *u) {
193  if (form <= DW_FORM_ref_sig8 && g_form_sizes[form].valid)
194    return static_cast<uint8_t>(g_form_sizes[form].size);
195  if (form == DW_FORM_addr && u)
196    return u->GetAddressByteSize();
197  return std::nullopt;
198}
199
200std::optional<uint8_t> DWARFFormValue::GetFixedSize() const {
201  return GetFixedSize(m_form, m_unit);
202}
203
204bool DWARFFormValue::SkipValue(const DWARFDataExtractor &debug_info_data,
205                               lldb::offset_t *offset_ptr) const {
206  return DWARFFormValue::SkipValue(m_form, debug_info_data, offset_ptr, m_unit);
207}
208
209bool DWARFFormValue::SkipValue(dw_form_t form,
210                               const DWARFDataExtractor &debug_info_data,
211                               lldb::offset_t *offset_ptr,
212                               const DWARFUnit *unit) {
213  uint8_t ref_addr_size;
214  switch (form) {
215  // Blocks if inlined data that have a length field and the data bytes inlined
216  // in the .debug_info
217  case DW_FORM_exprloc:
218  case DW_FORM_block: {
219    dw_uleb128_t size = debug_info_data.GetULEB128(offset_ptr);
220    *offset_ptr += size;
221  }
222    return true;
223  case DW_FORM_block1: {
224    dw_uleb128_t size = debug_info_data.GetU8(offset_ptr);
225    *offset_ptr += size;
226  }
227    return true;
228  case DW_FORM_block2: {
229    dw_uleb128_t size = debug_info_data.GetU16(offset_ptr);
230    *offset_ptr += size;
231  }
232    return true;
233  case DW_FORM_block4: {
234    dw_uleb128_t size = debug_info_data.GetU32(offset_ptr);
235    *offset_ptr += size;
236  }
237    return true;
238
239  // Inlined NULL terminated C-strings
240  case DW_FORM_string:
241    debug_info_data.GetCStr(offset_ptr);
242    return true;
243
244  // Compile unit address sized values
245  case DW_FORM_addr:
246    *offset_ptr += DWARFUnit::GetAddressByteSize(unit);
247    return true;
248
249  case DW_FORM_ref_addr:
250    ref_addr_size = 4;
251    assert(unit); // Unit must be valid for DW_FORM_ref_addr objects or we will
252                  // get this wrong
253    if (unit->GetVersion() <= 2)
254      ref_addr_size = unit->GetAddressByteSize();
255    else
256      ref_addr_size = 4;
257    *offset_ptr += ref_addr_size;
258    return true;
259
260  // 0 bytes values (implied from DW_FORM)
261  case DW_FORM_flag_present:
262  case DW_FORM_implicit_const:
263    return true;
264
265    // 1 byte values
266    case DW_FORM_addrx1:
267    case DW_FORM_data1:
268    case DW_FORM_flag:
269    case DW_FORM_ref1:
270    case DW_FORM_strx1:
271      *offset_ptr += 1;
272      return true;
273
274    // 2 byte values
275    case DW_FORM_addrx2:
276    case DW_FORM_data2:
277    case DW_FORM_ref2:
278    case DW_FORM_strx2:
279      *offset_ptr += 2;
280      return true;
281
282    // 3 byte values
283    case DW_FORM_addrx3:
284    case DW_FORM_strx3:
285      *offset_ptr += 3;
286      return true;
287
288    // 32 bit for DWARF 32, 64 for DWARF 64
289    case DW_FORM_sec_offset:
290    case DW_FORM_strp:
291    case DW_FORM_line_strp:
292      *offset_ptr += 4;
293      return true;
294
295    // 4 byte values
296    case DW_FORM_addrx4:
297    case DW_FORM_data4:
298    case DW_FORM_ref4:
299    case DW_FORM_strx4:
300      *offset_ptr += 4;
301      return true;
302
303    // 8 byte values
304    case DW_FORM_data8:
305    case DW_FORM_ref8:
306    case DW_FORM_ref_sig8:
307      *offset_ptr += 8;
308      return true;
309
310    // signed or unsigned LEB 128 values
311    case DW_FORM_addrx:
312    case DW_FORM_loclistx:
313    case DW_FORM_rnglistx:
314    case DW_FORM_sdata:
315    case DW_FORM_udata:
316    case DW_FORM_ref_udata:
317    case DW_FORM_GNU_addr_index:
318    case DW_FORM_GNU_str_index:
319    case DW_FORM_strx:
320      debug_info_data.Skip_LEB128(offset_ptr);
321      return true;
322
323  case DW_FORM_indirect: {
324    dw_form_t indirect_form = debug_info_data.GetULEB128(offset_ptr);
325    return DWARFFormValue::SkipValue(indirect_form, debug_info_data, offset_ptr,
326                                     unit);
327  }
328
329  default:
330    break;
331  }
332  return false;
333}
334
335void DWARFFormValue::Dump(Stream &s) const {
336  uint64_t uvalue = Unsigned();
337  bool unit_relative_offset = false;
338
339  switch (m_form) {
340  case DW_FORM_addr:
341    DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t));
342    break;
343  case DW_FORM_flag:
344  case DW_FORM_data1:
345    s.PutHex8(uvalue);
346    break;
347  case DW_FORM_data2:
348    s.PutHex16(uvalue);
349    break;
350  case DW_FORM_sec_offset:
351  case DW_FORM_data4:
352    s.PutHex32(uvalue);
353    break;
354  case DW_FORM_ref_sig8:
355  case DW_FORM_data8:
356    s.PutHex64(uvalue);
357    break;
358  case DW_FORM_string:
359    s.QuotedCString(AsCString());
360    break;
361  case DW_FORM_exprloc:
362  case DW_FORM_block:
363  case DW_FORM_block1:
364  case DW_FORM_block2:
365  case DW_FORM_block4:
366    if (uvalue > 0) {
367      switch (m_form) {
368      case DW_FORM_exprloc:
369      case DW_FORM_block:
370        s.Printf("<0x%" PRIx64 "> ", uvalue);
371        break;
372      case DW_FORM_block1:
373        s.Printf("<0x%2.2x> ", (uint8_t)uvalue);
374        break;
375      case DW_FORM_block2:
376        s.Printf("<0x%4.4x> ", (uint16_t)uvalue);
377        break;
378      case DW_FORM_block4:
379        s.Printf("<0x%8.8x> ", (uint32_t)uvalue);
380        break;
381      default:
382        break;
383      }
384
385      const uint8_t *data_ptr = m_value.data;
386      if (data_ptr) {
387        const uint8_t *end_data_ptr =
388            data_ptr + uvalue; // uvalue contains size of block
389        while (data_ptr < end_data_ptr) {
390          s.Printf("%2.2x ", *data_ptr);
391          ++data_ptr;
392        }
393      } else
394        s.PutCString("NULL");
395    }
396    break;
397
398  case DW_FORM_sdata:
399    s.PutSLEB128(uvalue);
400    break;
401  case DW_FORM_udata:
402    s.PutULEB128(uvalue);
403    break;
404  case DW_FORM_strp:
405  case DW_FORM_line_strp: {
406    const char *dbg_str = AsCString();
407    if (dbg_str) {
408      s.QuotedCString(dbg_str);
409    } else {
410      s.PutHex32(uvalue);
411    }
412  } break;
413
414  case DW_FORM_ref_addr: {
415    assert(m_unit); // Unit must be valid for DW_FORM_ref_addr objects or we
416                    // will get this wrong
417    if (m_unit->GetVersion() <= 2)
418      DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t) * 2);
419    else
420      DumpAddress(s.AsRawOstream(), uvalue,
421                  4 * 2); // 4 for DWARF32, 8 for DWARF64, but we don't
422                          // support DWARF64 yet
423    break;
424  }
425  case DW_FORM_ref1:
426    unit_relative_offset = true;
427    break;
428  case DW_FORM_ref2:
429    unit_relative_offset = true;
430    break;
431  case DW_FORM_ref4:
432    unit_relative_offset = true;
433    break;
434  case DW_FORM_ref8:
435    unit_relative_offset = true;
436    break;
437  case DW_FORM_ref_udata:
438    unit_relative_offset = true;
439    break;
440
441  // All DW_FORM_indirect attributes should be resolved prior to calling this
442  // function
443  case DW_FORM_indirect:
444    s.PutCString("DW_FORM_indirect");
445    break;
446  case DW_FORM_flag_present:
447    break;
448  default:
449    s.Printf("DW_FORM(0x%4.4x)", m_form);
450    break;
451  }
452
453  if (unit_relative_offset) {
454    assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
455                    // unit relative or we will get this wrong
456    s.Printf("{0x%8.8" PRIx64 "}", uvalue + m_unit->GetOffset());
457  }
458}
459
460const char *DWARFFormValue::AsCString() const {
461  DWARFContext &context = m_unit->GetSymbolFileDWARF().GetDWARFContext();
462
463  if (m_form == DW_FORM_string)
464    return m_value.value.cstr;
465  if (m_form == DW_FORM_strp)
466    return context.getOrLoadStrData().PeekCStr(m_value.value.uval);
467
468  if (m_form == DW_FORM_GNU_str_index || m_form == DW_FORM_strx ||
469      m_form == DW_FORM_strx1 || m_form == DW_FORM_strx2 ||
470      m_form == DW_FORM_strx3 || m_form == DW_FORM_strx4) {
471
472    std::optional<uint64_t> offset =
473        m_unit->GetStringOffsetSectionItem(m_value.value.uval);
474    if (!offset)
475      return nullptr;
476    return context.getOrLoadStrData().PeekCStr(*offset);
477  }
478
479  if (m_form == DW_FORM_line_strp)
480    return context.getOrLoadLineStrData().PeekCStr(m_value.value.uval);
481
482  return nullptr;
483}
484
485dw_addr_t DWARFFormValue::Address() const {
486  SymbolFileDWARF &symbol_file = m_unit->GetSymbolFileDWARF();
487
488  if (m_form == DW_FORM_addr)
489    return Unsigned();
490
491  assert(m_unit);
492  assert(m_form == DW_FORM_GNU_addr_index || m_form == DW_FORM_addrx ||
493         m_form == DW_FORM_addrx1 || m_form == DW_FORM_addrx2 ||
494         m_form == DW_FORM_addrx3 || m_form == DW_FORM_addrx4);
495
496  uint32_t index_size = m_unit->GetAddressByteSize();
497  dw_offset_t addr_base = m_unit->GetAddrBase();
498  lldb::offset_t offset = addr_base + m_value.value.uval * index_size;
499  return symbol_file.GetDWARFContext().getOrLoadAddrData().GetMaxU64(
500      &offset, index_size);
501}
502
503DWARFDIE DWARFFormValue::Reference() const {
504  uint64_t value = m_value.value.uval;
505  switch (m_form) {
506  case DW_FORM_ref1:
507  case DW_FORM_ref2:
508  case DW_FORM_ref4:
509  case DW_FORM_ref8:
510  case DW_FORM_ref_udata:
511    assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
512                    // unit relative or we will get this wrong
513    value += m_unit->GetOffset();
514    if (!m_unit->ContainsDIEOffset(value)) {
515      m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
516          "DW_FORM_ref* DIE reference {0:x16} is outside of its CU", value);
517      return {};
518    }
519    return const_cast<DWARFUnit *>(m_unit)->GetDIE(value);
520
521  case DW_FORM_ref_addr: {
522    DWARFUnit *ref_cu =
523        m_unit->GetSymbolFileDWARF().DebugInfo().GetUnitContainingDIEOffset(
524            DIERef::Section::DebugInfo, value);
525    if (!ref_cu) {
526      m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
527          "DW_FORM_ref_addr DIE reference {0:x16} has no matching CU", value);
528      return {};
529    }
530    return ref_cu->GetDIE(value);
531  }
532
533  case DW_FORM_ref_sig8: {
534    DWARFTypeUnit *tu =
535        m_unit->GetSymbolFileDWARF().DebugInfo().GetTypeUnitForHash(value);
536    if (!tu)
537      return {};
538    return tu->GetDIE(tu->GetTypeOffset());
539  }
540
541  default:
542    return {};
543  }
544}
545
546uint64_t DWARFFormValue::Reference(dw_offset_t base_offset) const {
547  uint64_t value = m_value.value.uval;
548  switch (m_form) {
549  case DW_FORM_ref1:
550  case DW_FORM_ref2:
551  case DW_FORM_ref4:
552  case DW_FORM_ref8:
553  case DW_FORM_ref_udata:
554    return value + base_offset;
555
556  case DW_FORM_ref_addr:
557  case DW_FORM_ref_sig8:
558  case DW_FORM_GNU_ref_alt:
559    return value;
560
561  default:
562    return DW_INVALID_OFFSET;
563  }
564}
565
566const uint8_t *DWARFFormValue::BlockData() const { return m_value.data; }
567
568bool DWARFFormValue::IsBlockForm(const dw_form_t form) {
569  switch (form) {
570  case DW_FORM_exprloc:
571  case DW_FORM_block:
572  case DW_FORM_block1:
573  case DW_FORM_block2:
574  case DW_FORM_block4:
575    return true;
576  }
577  return false;
578}
579
580bool DWARFFormValue::IsDataForm(const dw_form_t form) {
581  switch (form) {
582  case DW_FORM_sdata:
583  case DW_FORM_udata:
584  case DW_FORM_data1:
585  case DW_FORM_data2:
586  case DW_FORM_data4:
587  case DW_FORM_data8:
588    return true;
589  }
590  return false;
591}
592
593bool DWARFFormValue::FormIsSupported(dw_form_t form) {
594  switch (form) {
595    case DW_FORM_addr:
596    case DW_FORM_addrx:
597    case DW_FORM_loclistx:
598    case DW_FORM_rnglistx:
599    case DW_FORM_block2:
600    case DW_FORM_block4:
601    case DW_FORM_data2:
602    case DW_FORM_data4:
603    case DW_FORM_data8:
604    case DW_FORM_string:
605    case DW_FORM_block:
606    case DW_FORM_block1:
607    case DW_FORM_data1:
608    case DW_FORM_flag:
609    case DW_FORM_sdata:
610    case DW_FORM_strp:
611    case DW_FORM_line_strp:
612    case DW_FORM_strx:
613    case DW_FORM_strx1:
614    case DW_FORM_strx2:
615    case DW_FORM_strx3:
616    case DW_FORM_strx4:
617    case DW_FORM_udata:
618    case DW_FORM_ref_addr:
619    case DW_FORM_ref1:
620    case DW_FORM_ref2:
621    case DW_FORM_ref4:
622    case DW_FORM_ref8:
623    case DW_FORM_ref_udata:
624    case DW_FORM_indirect:
625    case DW_FORM_sec_offset:
626    case DW_FORM_exprloc:
627    case DW_FORM_flag_present:
628    case DW_FORM_ref_sig8:
629    case DW_FORM_GNU_str_index:
630    case DW_FORM_GNU_addr_index:
631    case DW_FORM_implicit_const:
632      return true;
633    default:
634      break;
635  }
636  return false;
637}
638