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
21using namespace lldb_private;
22using namespace lldb_private::dwarf;
23using namespace lldb_private::plugin::dwarf;
24
25void DWARFFormValue::Clear() {
26  m_unit = nullptr;
27  m_form = dw_form_t(0);
28  m_value = ValueTypeTag();
29}
30
31bool DWARFFormValue::ExtractValue(const DWARFDataExtractor &data,
32                                  lldb::offset_t *offset_ptr) {
33  if (m_form == DW_FORM_implicit_const)
34    return true;
35
36  bool indirect = false;
37  bool is_block = false;
38  m_value.data = nullptr;
39  uint8_t ref_addr_size;
40  // Read the value for the form into value and follow and DW_FORM_indirect
41  // instances we run into
42  do {
43    indirect = false;
44    switch (m_form) {
45    case DW_FORM_addr:
46      assert(m_unit);
47      m_value.value.uval =
48          data.GetMaxU64(offset_ptr, DWARFUnit::GetAddressByteSize(m_unit));
49      break;
50    case DW_FORM_block1:
51      m_value.value.uval = data.GetU8(offset_ptr);
52      is_block = true;
53      break;
54    case DW_FORM_block2:
55      m_value.value.uval = data.GetU16(offset_ptr);
56      is_block = true;
57      break;
58    case DW_FORM_block4:
59      m_value.value.uval = data.GetU32(offset_ptr);
60      is_block = true;
61      break;
62    case DW_FORM_data16:
63      m_value.value.uval = 16;
64      is_block = true;
65      break;
66    case DW_FORM_exprloc:
67    case DW_FORM_block:
68      m_value.value.uval = data.GetULEB128(offset_ptr);
69      is_block = true;
70      break;
71    case DW_FORM_string:
72      m_value.value.cstr = data.GetCStr(offset_ptr);
73      break;
74    case DW_FORM_sdata:
75      m_value.value.sval = data.GetSLEB128(offset_ptr);
76      break;
77    case DW_FORM_strp:
78    case DW_FORM_line_strp:
79    case DW_FORM_sec_offset:
80      m_value.value.uval = data.GetMaxU64(offset_ptr, 4);
81      break;
82    case DW_FORM_addrx1:
83    case DW_FORM_strx1:
84    case DW_FORM_ref1:
85    case DW_FORM_data1:
86    case DW_FORM_flag:
87      m_value.value.uval = data.GetU8(offset_ptr);
88      break;
89    case DW_FORM_addrx2:
90    case DW_FORM_strx2:
91    case DW_FORM_ref2:
92    case DW_FORM_data2:
93      m_value.value.uval = data.GetU16(offset_ptr);
94      break;
95    case DW_FORM_addrx3:
96    case DW_FORM_strx3:
97      m_value.value.uval = data.GetMaxU64(offset_ptr, 3);
98      break;
99    case DW_FORM_addrx4:
100    case DW_FORM_strx4:
101    case DW_FORM_ref4:
102    case DW_FORM_data4:
103      m_value.value.uval = data.GetU32(offset_ptr);
104      break;
105    case DW_FORM_data8:
106    case DW_FORM_ref8:
107    case DW_FORM_ref_sig8:
108      m_value.value.uval = data.GetU64(offset_ptr);
109      break;
110    case DW_FORM_addrx:
111    case DW_FORM_loclistx:
112    case DW_FORM_rnglistx:
113    case DW_FORM_strx:
114    case DW_FORM_udata:
115    case DW_FORM_ref_udata:
116    case DW_FORM_GNU_str_index:
117    case DW_FORM_GNU_addr_index:
118      m_value.value.uval = data.GetULEB128(offset_ptr);
119      break;
120    case DW_FORM_ref_addr:
121      assert(m_unit);
122      if (m_unit->GetVersion() <= 2)
123        ref_addr_size = m_unit->GetAddressByteSize();
124      else
125        ref_addr_size = 4;
126      m_value.value.uval = data.GetMaxU64(offset_ptr, ref_addr_size);
127      break;
128    case DW_FORM_indirect:
129      m_form = static_cast<dw_form_t>(data.GetULEB128(offset_ptr));
130      indirect = true;
131      break;
132    case DW_FORM_flag_present:
133      m_value.value.uval = 1;
134      break;
135    default:
136      return false;
137    }
138  } while (indirect);
139
140  if (is_block) {
141    m_value.data = data.PeekData(*offset_ptr, m_value.value.uval);
142    if (m_value.data != nullptr) {
143      *offset_ptr += m_value.value.uval;
144    }
145  }
146
147  return true;
148}
149
150struct FormSize {
151  uint8_t valid:1, size:7;
152};
153static FormSize g_form_sizes[] = {
154    {0, 0}, // 0x00 unused
155    {0, 0}, // 0x01 DW_FORM_addr
156    {0, 0}, // 0x02 unused
157    {0, 0}, // 0x03 DW_FORM_block2
158    {0, 0}, // 0x04 DW_FORM_block4
159    {1, 2}, // 0x05 DW_FORM_data2
160    {1, 4}, // 0x06 DW_FORM_data4
161    {1, 8}, // 0x07 DW_FORM_data8
162    {0, 0}, // 0x08 DW_FORM_string
163    {0, 0}, // 0x09 DW_FORM_block
164    {0, 0}, // 0x0a DW_FORM_block1
165    {1, 1}, // 0x0b DW_FORM_data1
166    {1, 1}, // 0x0c DW_FORM_flag
167    {0, 0}, // 0x0d DW_FORM_sdata
168    {1, 4}, // 0x0e DW_FORM_strp
169    {0, 0}, // 0x0f DW_FORM_udata
170    {0, 0}, // 0x10 DW_FORM_ref_addr (addr size for DWARF2 and earlier, 4 bytes
171            // for DWARF32, 8 bytes for DWARF32 in DWARF 3 and later
172    {1, 1},  // 0x11 DW_FORM_ref1
173    {1, 2},  // 0x12 DW_FORM_ref2
174    {1, 4},  // 0x13 DW_FORM_ref4
175    {1, 8},  // 0x14 DW_FORM_ref8
176    {0, 0},  // 0x15 DW_FORM_ref_udata
177    {0, 0},  // 0x16 DW_FORM_indirect
178    {1, 4},  // 0x17 DW_FORM_sec_offset
179    {0, 0},  // 0x18 DW_FORM_exprloc
180    {1, 0},  // 0x19 DW_FORM_flag_present
181    {0, 0},  // 0x1a DW_FORM_strx (ULEB128)
182    {0, 0},  // 0x1b DW_FORM_addrx (ULEB128)
183    {1, 4},  // 0x1c DW_FORM_ref_sup4
184    {0, 0},  // 0x1d DW_FORM_strp_sup (4 bytes for DWARF32, 8 bytes for DWARF64)
185    {1, 16}, // 0x1e DW_FORM_data16
186    {1, 4},  // 0x1f DW_FORM_line_strp
187    {1, 8},  // 0x20 DW_FORM_ref_sig8
188};
189
190std::optional<uint8_t> DWARFFormValue::GetFixedSize(dw_form_t form,
191                                                    const DWARFUnit *u) {
192  if (form <= DW_FORM_ref_sig8 && g_form_sizes[form].valid)
193    return static_cast<uint8_t>(g_form_sizes[form].size);
194  if (form == DW_FORM_addr && u)
195    return u->GetAddressByteSize();
196  return std::nullopt;
197}
198
199std::optional<uint8_t> DWARFFormValue::GetFixedSize() const {
200  return GetFixedSize(m_form, m_unit);
201}
202
203bool DWARFFormValue::SkipValue(const DWARFDataExtractor &debug_info_data,
204                               lldb::offset_t *offset_ptr) const {
205  return DWARFFormValue::SkipValue(m_form, debug_info_data, offset_ptr, m_unit);
206}
207
208bool DWARFFormValue::SkipValue(dw_form_t form,
209                               const DWARFDataExtractor &debug_info_data,
210                               lldb::offset_t *offset_ptr,
211                               const DWARFUnit *unit) {
212  uint8_t ref_addr_size;
213  switch (form) {
214  // Blocks if inlined data that have a length field and the data bytes inlined
215  // in the .debug_info
216  case DW_FORM_exprloc:
217  case DW_FORM_block: {
218    uint64_t size = debug_info_data.GetULEB128(offset_ptr);
219    *offset_ptr += size;
220  }
221    return true;
222  case DW_FORM_block1: {
223    uint8_t size = debug_info_data.GetU8(offset_ptr);
224    *offset_ptr += size;
225  }
226    return true;
227  case DW_FORM_block2: {
228    uint16_t size = debug_info_data.GetU16(offset_ptr);
229    *offset_ptr += size;
230  }
231    return true;
232  case DW_FORM_block4: {
233    uint32_t size = debug_info_data.GetU32(offset_ptr);
234    *offset_ptr += size;
235  }
236    return true;
237
238  // Inlined NULL terminated C-strings
239  case DW_FORM_string:
240    debug_info_data.GetCStr(offset_ptr);
241    return true;
242
243  // Compile unit address sized values
244  case DW_FORM_addr:
245    *offset_ptr += DWARFUnit::GetAddressByteSize(unit);
246    return true;
247
248  case DW_FORM_ref_addr:
249    ref_addr_size = 4;
250    assert(unit); // Unit must be valid for DW_FORM_ref_addr objects or we will
251                  // get this wrong
252    if (unit->GetVersion() <= 2)
253      ref_addr_size = unit->GetAddressByteSize();
254    else
255      ref_addr_size = 4;
256    *offset_ptr += ref_addr_size;
257    return true;
258
259  // 0 bytes values (implied from DW_FORM)
260  case DW_FORM_flag_present:
261  case DW_FORM_implicit_const:
262    return true;
263
264    // 1 byte values
265    case DW_FORM_addrx1:
266    case DW_FORM_data1:
267    case DW_FORM_flag:
268    case DW_FORM_ref1:
269    case DW_FORM_strx1:
270      *offset_ptr += 1;
271      return true;
272
273    // 2 byte values
274    case DW_FORM_addrx2:
275    case DW_FORM_data2:
276    case DW_FORM_ref2:
277    case DW_FORM_strx2:
278      *offset_ptr += 2;
279      return true;
280
281    // 3 byte values
282    case DW_FORM_addrx3:
283    case DW_FORM_strx3:
284      *offset_ptr += 3;
285      return true;
286
287    // 32 bit for DWARF 32, 64 for DWARF 64
288    case DW_FORM_sec_offset:
289    case DW_FORM_strp:
290    case DW_FORM_line_strp:
291      *offset_ptr += 4;
292      return true;
293
294    // 4 byte values
295    case DW_FORM_addrx4:
296    case DW_FORM_data4:
297    case DW_FORM_ref4:
298    case DW_FORM_strx4:
299      *offset_ptr += 4;
300      return true;
301
302    // 8 byte values
303    case DW_FORM_data8:
304    case DW_FORM_ref8:
305    case DW_FORM_ref_sig8:
306      *offset_ptr += 8;
307      return true;
308
309    // signed or unsigned LEB 128 values
310    case DW_FORM_addrx:
311    case DW_FORM_loclistx:
312    case DW_FORM_rnglistx:
313    case DW_FORM_sdata:
314    case DW_FORM_udata:
315    case DW_FORM_ref_udata:
316    case DW_FORM_GNU_addr_index:
317    case DW_FORM_GNU_str_index:
318    case DW_FORM_strx:
319      debug_info_data.Skip_LEB128(offset_ptr);
320      return true;
321
322  case DW_FORM_indirect: {
323      auto indirect_form =
324          static_cast<dw_form_t>(debug_info_data.GetULEB128(offset_ptr));
325      return DWARFFormValue::SkipValue(indirect_form, debug_info_data,
326                                       offset_ptr, 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
503std::pair<DWARFUnit *, uint64_t>
504DWARFFormValue::ReferencedUnitAndOffset() const {
505  uint64_t value = m_value.value.uval;
506  switch (m_form) {
507  case DW_FORM_ref1:
508  case DW_FORM_ref2:
509  case DW_FORM_ref4:
510  case DW_FORM_ref8:
511  case DW_FORM_ref_udata:
512    assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
513                    // unit relative or we will get this wrong
514    value += m_unit->GetOffset();
515    if (!m_unit->ContainsDIEOffset(value)) {
516      m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
517          "DW_FORM_ref* DIE reference {0:x16} is outside of its CU", value);
518      return {nullptr, 0};
519    }
520    return {const_cast<DWARFUnit *>(m_unit), value};
521
522  case DW_FORM_ref_addr: {
523    DWARFUnit *ref_cu =
524        m_unit->GetSymbolFileDWARF().DebugInfo().GetUnitContainingDIEOffset(
525            DIERef::Section::DebugInfo, value);
526    if (!ref_cu) {
527      m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
528          "DW_FORM_ref_addr DIE reference {0:x16} has no matching CU", value);
529      return {nullptr, 0};
530    }
531    return {ref_cu, value};
532  }
533
534  case DW_FORM_ref_sig8: {
535    DWARFTypeUnit *tu =
536        m_unit->GetSymbolFileDWARF().DebugInfo().GetTypeUnitForHash(value);
537    if (!tu)
538      return {nullptr, 0};
539    return {tu, tu->GetTypeOffset()};
540  }
541
542  default:
543    return {nullptr, 0};
544  }
545}
546
547DWARFDIE DWARFFormValue::Reference() const {
548  auto [unit, offset] = ReferencedUnitAndOffset();
549  return unit ? unit->GetDIE(offset) : DWARFDIE();
550}
551
552uint64_t DWARFFormValue::Reference(dw_offset_t base_offset) const {
553  uint64_t value = m_value.value.uval;
554  switch (m_form) {
555  case DW_FORM_ref1:
556  case DW_FORM_ref2:
557  case DW_FORM_ref4:
558  case DW_FORM_ref8:
559  case DW_FORM_ref_udata:
560    return value + base_offset;
561
562  case DW_FORM_ref_addr:
563  case DW_FORM_ref_sig8:
564  case DW_FORM_GNU_ref_alt:
565    return value;
566
567  default:
568    return DW_INVALID_OFFSET;
569  }
570}
571
572const uint8_t *DWARFFormValue::BlockData() const { return m_value.data; }
573
574bool DWARFFormValue::IsBlockForm(const dw_form_t form) {
575  switch (form) {
576  case DW_FORM_exprloc:
577  case DW_FORM_block:
578  case DW_FORM_block1:
579  case DW_FORM_block2:
580  case DW_FORM_block4:
581    return true;
582  default:
583    return false;
584  }
585  llvm_unreachable("All cases handled above!");
586}
587
588bool DWARFFormValue::IsDataForm(const dw_form_t form) {
589  switch (form) {
590  case DW_FORM_sdata:
591  case DW_FORM_udata:
592  case DW_FORM_data1:
593  case DW_FORM_data2:
594  case DW_FORM_data4:
595  case DW_FORM_data8:
596    return true;
597  default:
598    return false;
599  }
600  llvm_unreachable("All cases handled above!");
601}
602
603bool DWARFFormValue::FormIsSupported(dw_form_t form) {
604  switch (form) {
605    case DW_FORM_addr:
606    case DW_FORM_addrx:
607    case DW_FORM_loclistx:
608    case DW_FORM_rnglistx:
609    case DW_FORM_block2:
610    case DW_FORM_block4:
611    case DW_FORM_data2:
612    case DW_FORM_data4:
613    case DW_FORM_data8:
614    case DW_FORM_string:
615    case DW_FORM_block:
616    case DW_FORM_block1:
617    case DW_FORM_data1:
618    case DW_FORM_flag:
619    case DW_FORM_sdata:
620    case DW_FORM_strp:
621    case DW_FORM_line_strp:
622    case DW_FORM_strx:
623    case DW_FORM_strx1:
624    case DW_FORM_strx2:
625    case DW_FORM_strx3:
626    case DW_FORM_strx4:
627    case DW_FORM_udata:
628    case DW_FORM_ref_addr:
629    case DW_FORM_ref1:
630    case DW_FORM_ref2:
631    case DW_FORM_ref4:
632    case DW_FORM_ref8:
633    case DW_FORM_ref_udata:
634    case DW_FORM_indirect:
635    case DW_FORM_sec_offset:
636    case DW_FORM_exprloc:
637    case DW_FORM_flag_present:
638    case DW_FORM_ref_sig8:
639    case DW_FORM_GNU_str_index:
640    case DW_FORM_GNU_addr_index:
641    case DW_FORM_implicit_const:
642      return true;
643    default:
644      break;
645  }
646  return false;
647}
648