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 
11 #include "lldb/Core/Module.h"
12 #include "lldb/Core/dwarf.h"
13 #include "lldb/Symbol/ObjectFile.h"
14 #include "lldb/Utility/Stream.h"
15 
16 #include "DWARFDebugInfo.h"
17 #include "DWARFFormValue.h"
18 #include "DWARFUnit.h"
19 
20 class DWARFUnit;
21 
22 using namespace lldb_private;
23 using namespace lldb_private::dwarf;
24 
25 void DWARFFormValue::Clear() {
26   m_unit = nullptr;
27   m_form = 0;
28   m_value = ValueTypeTag();
29 }
30 
31 bool 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 = 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 
150 struct FormSize {
151   uint8_t valid:1, size:7;
152 };
153 static 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 
190 llvm::Optional<uint8_t>
191 DWARFFormValue::GetFixedSize(dw_form_t form, const DWARFUnit *u) {
192   if (form <= DW_FORM_ref_sig8 && g_form_sizes[form].valid)
193     return g_form_sizes[form].size;
194   if (form == DW_FORM_addr && u)
195     return u->GetAddressByteSize();
196   return llvm::None;
197 }
198 
199 llvm::Optional<uint8_t> DWARFFormValue::GetFixedSize() const {
200   return GetFixedSize(m_form, m_unit);
201 }
202 
203 bool 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 
208 bool 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     dw_uleb128_t size = debug_info_data.GetULEB128(offset_ptr);
219     *offset_ptr += size;
220   }
221     return true;
222   case DW_FORM_block1: {
223     dw_uleb128_t size = debug_info_data.GetU8(offset_ptr);
224     *offset_ptr += size;
225   }
226     return true;
227   case DW_FORM_block2: {
228     dw_uleb128_t size = debug_info_data.GetU16(offset_ptr);
229     *offset_ptr += size;
230   }
231     return true;
232   case DW_FORM_block4: {
233     dw_uleb128_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     dw_form_t indirect_form = debug_info_data.GetULEB128(offset_ptr);
324     return DWARFFormValue::SkipValue(indirect_form, debug_info_data, offset_ptr,
325                                      unit);
326   }
327 
328   default:
329     break;
330   }
331   return false;
332 }
333 
334 void DWARFFormValue::Dump(Stream &s) const {
335   uint64_t uvalue = Unsigned();
336   bool unit_relative_offset = false;
337 
338   switch (m_form) {
339   case DW_FORM_addr:
340     DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t));
341     break;
342   case DW_FORM_flag:
343   case DW_FORM_data1:
344     s.PutHex8(uvalue);
345     break;
346   case DW_FORM_data2:
347     s.PutHex16(uvalue);
348     break;
349   case DW_FORM_sec_offset:
350   case DW_FORM_data4:
351     s.PutHex32(uvalue);
352     break;
353   case DW_FORM_ref_sig8:
354   case DW_FORM_data8:
355     s.PutHex64(uvalue);
356     break;
357   case DW_FORM_string:
358     s.QuotedCString(AsCString());
359     break;
360   case DW_FORM_exprloc:
361   case DW_FORM_block:
362   case DW_FORM_block1:
363   case DW_FORM_block2:
364   case DW_FORM_block4:
365     if (uvalue > 0) {
366       switch (m_form) {
367       case DW_FORM_exprloc:
368       case DW_FORM_block:
369         s.Printf("<0x%" PRIx64 "> ", uvalue);
370         break;
371       case DW_FORM_block1:
372         s.Printf("<0x%2.2x> ", (uint8_t)uvalue);
373         break;
374       case DW_FORM_block2:
375         s.Printf("<0x%4.4x> ", (uint16_t)uvalue);
376         break;
377       case DW_FORM_block4:
378         s.Printf("<0x%8.8x> ", (uint32_t)uvalue);
379         break;
380       default:
381         break;
382       }
383 
384       const uint8_t *data_ptr = m_value.data;
385       if (data_ptr) {
386         const uint8_t *end_data_ptr =
387             data_ptr + uvalue; // uvalue contains size of block
388         while (data_ptr < end_data_ptr) {
389           s.Printf("%2.2x ", *data_ptr);
390           ++data_ptr;
391         }
392       } else
393         s.PutCString("NULL");
394     }
395     break;
396 
397   case DW_FORM_sdata:
398     s.PutSLEB128(uvalue);
399     break;
400   case DW_FORM_udata:
401     s.PutULEB128(uvalue);
402     break;
403   case DW_FORM_strp:
404   case DW_FORM_line_strp: {
405     const char *dbg_str = AsCString();
406     if (dbg_str) {
407       s.QuotedCString(dbg_str);
408     } else {
409       s.PutHex32(uvalue);
410     }
411   } break;
412 
413   case DW_FORM_ref_addr: {
414     assert(m_unit); // Unit must be valid for DW_FORM_ref_addr objects or we
415                     // will get this wrong
416     if (m_unit->GetVersion() <= 2)
417       DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t) * 2);
418     else
419       DumpAddress(s.AsRawOstream(), uvalue,
420                   4 * 2); // 4 for DWARF32, 8 for DWARF64, but we don't
421                           // support DWARF64 yet
422     break;
423   }
424   case DW_FORM_ref1:
425     unit_relative_offset = true;
426     break;
427   case DW_FORM_ref2:
428     unit_relative_offset = true;
429     break;
430   case DW_FORM_ref4:
431     unit_relative_offset = true;
432     break;
433   case DW_FORM_ref8:
434     unit_relative_offset = true;
435     break;
436   case DW_FORM_ref_udata:
437     unit_relative_offset = true;
438     break;
439 
440   // All DW_FORM_indirect attributes should be resolved prior to calling this
441   // function
442   case DW_FORM_indirect:
443     s.PutCString("DW_FORM_indirect");
444     break;
445   case DW_FORM_flag_present:
446     break;
447   default:
448     s.Printf("DW_FORM(0x%4.4x)", m_form);
449     break;
450   }
451 
452   if (unit_relative_offset) {
453     assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
454                     // unit relative or we will get this wrong
455     s.Printf("{0x%8.8" PRIx64 "}", uvalue + m_unit->GetOffset());
456   }
457 }
458 
459 const char *DWARFFormValue::AsCString() const {
460   DWARFContext &context = m_unit->GetSymbolFileDWARF().GetDWARFContext();
461 
462   if (m_form == DW_FORM_string)
463     return m_value.value.cstr;
464   if (m_form == DW_FORM_strp)
465     return context.getOrLoadStrData().PeekCStr(m_value.value.uval);
466 
467   if (m_form == DW_FORM_GNU_str_index || m_form == DW_FORM_strx ||
468       m_form == DW_FORM_strx1 || m_form == DW_FORM_strx2 ||
469       m_form == DW_FORM_strx3 || m_form == DW_FORM_strx4) {
470 
471     llvm::Optional<uint64_t> offset =
472         m_unit->GetStringOffsetSectionItem(m_value.value.uval);
473     if (!offset)
474       return nullptr;
475     return context.getOrLoadStrData().PeekCStr(*offset);
476   }
477 
478   if (m_form == DW_FORM_line_strp)
479     return context.getOrLoadLineStrData().PeekCStr(m_value.value.uval);
480 
481   return nullptr;
482 }
483 
484 dw_addr_t DWARFFormValue::Address() const {
485   SymbolFileDWARF &symbol_file = m_unit->GetSymbolFileDWARF();
486 
487   if (m_form == DW_FORM_addr)
488     return Unsigned();
489 
490   assert(m_unit);
491   assert(m_form == DW_FORM_GNU_addr_index || m_form == DW_FORM_addrx ||
492          m_form == DW_FORM_addrx1 || m_form == DW_FORM_addrx2 ||
493          m_form == DW_FORM_addrx3 || m_form == DW_FORM_addrx4);
494 
495   uint32_t index_size = m_unit->GetAddressByteSize();
496   dw_offset_t addr_base = m_unit->GetAddrBase();
497   lldb::offset_t offset = addr_base + m_value.value.uval * index_size;
498   return symbol_file.GetDWARFContext().getOrLoadAddrData().GetMaxU64(
499       &offset, index_size);
500 }
501 
502 DWARFDIE DWARFFormValue::Reference() const {
503   uint64_t value = m_value.value.uval;
504   switch (m_form) {
505   case DW_FORM_ref1:
506   case DW_FORM_ref2:
507   case DW_FORM_ref4:
508   case DW_FORM_ref8:
509   case DW_FORM_ref_udata:
510     assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
511                     // unit relative or we will get this wrong
512     value += m_unit->GetOffset();
513     if (!m_unit->ContainsDIEOffset(value)) {
514       m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
515           "DW_FORM_ref* DIE reference 0x%" PRIx64 " is outside of its CU",
516           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 0x%" PRIx64 " has no matching CU",
528           value);
529       return {};
530     }
531     return ref_cu->GetDIE(value);
532   }
533 
534   case DW_FORM_ref_sig8: {
535     DWARFTypeUnit *tu =
536         m_unit->GetSymbolFileDWARF().DebugInfo().GetTypeUnitForHash(value);
537     if (!tu)
538       return {};
539     return tu->GetDIE(tu->GetTypeOffset());
540   }
541 
542   default:
543     return {};
544   }
545 }
546 
547 uint64_t DWARFFormValue::Reference(dw_offset_t base_offset) const {
548   uint64_t value = m_value.value.uval;
549   switch (m_form) {
550   case DW_FORM_ref1:
551   case DW_FORM_ref2:
552   case DW_FORM_ref4:
553   case DW_FORM_ref8:
554   case DW_FORM_ref_udata:
555     return value + base_offset;
556 
557   case DW_FORM_ref_addr:
558   case DW_FORM_ref_sig8:
559   case DW_FORM_GNU_ref_alt:
560     return value;
561 
562   default:
563     return DW_INVALID_OFFSET;
564   }
565 }
566 
567 const uint8_t *DWARFFormValue::BlockData() const { return m_value.data; }
568 
569 bool DWARFFormValue::IsBlockForm(const dw_form_t form) {
570   switch (form) {
571   case DW_FORM_exprloc:
572   case DW_FORM_block:
573   case DW_FORM_block1:
574   case DW_FORM_block2:
575   case DW_FORM_block4:
576     return true;
577   }
578   return false;
579 }
580 
581 bool DWARFFormValue::IsDataForm(const dw_form_t form) {
582   switch (form) {
583   case DW_FORM_sdata:
584   case DW_FORM_udata:
585   case DW_FORM_data1:
586   case DW_FORM_data2:
587   case DW_FORM_data4:
588   case DW_FORM_data8:
589     return true;
590   }
591   return false;
592 }
593 
594 bool DWARFFormValue::FormIsSupported(dw_form_t form) {
595   switch (form) {
596     case DW_FORM_addr:
597     case DW_FORM_addrx:
598     case DW_FORM_loclistx:
599     case DW_FORM_rnglistx:
600     case DW_FORM_block2:
601     case DW_FORM_block4:
602     case DW_FORM_data2:
603     case DW_FORM_data4:
604     case DW_FORM_data8:
605     case DW_FORM_string:
606     case DW_FORM_block:
607     case DW_FORM_block1:
608     case DW_FORM_data1:
609     case DW_FORM_flag:
610     case DW_FORM_sdata:
611     case DW_FORM_strp:
612     case DW_FORM_line_strp:
613     case DW_FORM_strx:
614     case DW_FORM_strx1:
615     case DW_FORM_strx2:
616     case DW_FORM_strx3:
617     case DW_FORM_strx4:
618     case DW_FORM_udata:
619     case DW_FORM_ref_addr:
620     case DW_FORM_ref1:
621     case DW_FORM_ref2:
622     case DW_FORM_ref4:
623     case DW_FORM_ref8:
624     case DW_FORM_ref_udata:
625     case DW_FORM_indirect:
626     case DW_FORM_sec_offset:
627     case DW_FORM_exprloc:
628     case DW_FORM_flag_present:
629     case DW_FORM_ref_sig8:
630     case DW_FORM_GNU_str_index:
631     case DW_FORM_GNU_addr_index:
632     case DW_FORM_implicit_const:
633       return true;
634     default:
635       break;
636   }
637   return false;
638 }
639