1 //===-- DWARFFormValue.cpp --------------------------------------*- C++ -*-===//
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 <assert.h>
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 
24 void DWARFFormValue::Clear() {
25   m_unit = nullptr;
26   m_form = 0;
27   m_value = ValueTypeTag();
28 }
29 
30 bool DWARFFormValue::ExtractValue(const DWARFDataExtractor &data,
31                                   lldb::offset_t *offset_ptr) {
32   if (m_form == DW_FORM_implicit_const)
33     return true;
34 
35   bool indirect = false;
36   bool is_block = false;
37   m_value.data = nullptr;
38   uint8_t ref_addr_size;
39   // Read the value for the form into value and follow and DW_FORM_indirect
40   // instances we run into
41   do {
42     indirect = false;
43     switch (m_form) {
44     case DW_FORM_addr:
45       assert(m_unit);
46       m_value.value.uval =
47           data.GetMaxU64(offset_ptr, DWARFUnit::GetAddressByteSize(m_unit));
48       break;
49     case DW_FORM_block1:
50       m_value.value.uval = data.GetU8(offset_ptr);
51       is_block = true;
52       break;
53     case DW_FORM_block2:
54       m_value.value.uval = data.GetU16(offset_ptr);
55       is_block = true;
56       break;
57     case DW_FORM_block4:
58       m_value.value.uval = data.GetU32(offset_ptr);
59       is_block = true;
60       break;
61     case DW_FORM_data16:
62       m_value.value.uval = 16;
63       is_block = true;
64       break;
65     case DW_FORM_exprloc:
66     case DW_FORM_block:
67       m_value.value.uval = data.GetULEB128(offset_ptr);
68       is_block = true;
69       break;
70     case DW_FORM_string:
71       m_value.value.cstr = data.GetCStr(offset_ptr);
72       break;
73     case DW_FORM_sdata:
74       m_value.value.sval = data.GetSLEB128(offset_ptr);
75       break;
76     case DW_FORM_strp:
77     case DW_FORM_line_strp:
78     case DW_FORM_sec_offset:
79       m_value.value.uval = data.GetMaxU64(offset_ptr, 4);
80       break;
81     case DW_FORM_addrx1:
82     case DW_FORM_strx1:
83     case DW_FORM_ref1:
84     case DW_FORM_data1:
85     case DW_FORM_flag:
86       m_value.value.uval = data.GetU8(offset_ptr);
87       break;
88     case DW_FORM_addrx2:
89     case DW_FORM_strx2:
90     case DW_FORM_ref2:
91     case DW_FORM_data2:
92       m_value.value.uval = data.GetU16(offset_ptr);
93       break;
94     case DW_FORM_addrx3:
95     case DW_FORM_strx3:
96       m_value.value.uval = data.GetMaxU64(offset_ptr, 3);
97       break;
98     case DW_FORM_addrx4:
99     case DW_FORM_strx4:
100     case DW_FORM_ref4:
101     case DW_FORM_data4:
102       m_value.value.uval = data.GetU32(offset_ptr);
103       break;
104     case DW_FORM_data8:
105     case DW_FORM_ref8:
106     case DW_FORM_ref_sig8:
107       m_value.value.uval = data.GetU64(offset_ptr);
108       break;
109     case DW_FORM_addrx:
110     case DW_FORM_loclistx:
111     case DW_FORM_rnglistx:
112     case DW_FORM_strx:
113     case DW_FORM_udata:
114     case DW_FORM_ref_udata:
115     case DW_FORM_GNU_str_index:
116     case DW_FORM_GNU_addr_index:
117       m_value.value.uval = data.GetULEB128(offset_ptr);
118       break;
119     case DW_FORM_ref_addr:
120       assert(m_unit);
121       if (m_unit->GetVersion() <= 2)
122         ref_addr_size = m_unit->GetAddressByteSize();
123       else
124         ref_addr_size = 4;
125       m_value.value.uval = data.GetMaxU64(offset_ptr, ref_addr_size);
126       break;
127     case DW_FORM_indirect:
128       m_form = data.GetULEB128(offset_ptr);
129       indirect = true;
130       break;
131     case DW_FORM_flag_present:
132       m_value.value.uval = 1;
133       break;
134     default:
135       return false;
136     }
137   } while (indirect);
138 
139   if (is_block) {
140     m_value.data = data.PeekData(*offset_ptr, m_value.value.uval);
141     if (m_value.data != nullptr) {
142       *offset_ptr += m_value.value.uval;
143     }
144   }
145 
146   return true;
147 }
148 
149 struct FormSize {
150   uint8_t valid:1, size:7;
151 };
152 static FormSize g_form_sizes[] = {
153   {0,0}, // 0x00 unused
154   {0,0}, // 0x01 DW_FORM_addr
155   {0,0}, // 0x02 unused
156   {0,0}, // 0x03 DW_FORM_block2
157   {0,0}, // 0x04 DW_FORM_block4
158   {1,2}, // 0x05 DW_FORM_data2
159   {1,4}, // 0x06 DW_FORM_data4
160   {1,8}, // 0x07 DW_FORM_data8
161   {0,0}, // 0x08 DW_FORM_string
162   {0,0}, // 0x09 DW_FORM_block
163   {0,0}, // 0x0a DW_FORM_block1
164   {1,1}, // 0x0b DW_FORM_data1
165   {1,1}, // 0x0c DW_FORM_flag
166   {0,0}, // 0x0d DW_FORM_sdata
167   {1,4}, // 0x0e DW_FORM_strp
168   {0,0}, // 0x0f DW_FORM_udata
169   {0,0}, // 0x10 DW_FORM_ref_addr (addr size for DWARF2 and earlier, 4 bytes for
170          // DWARF32, 8 bytes for DWARF32 in DWARF 3 and later
171   {1,1}, // 0x11 DW_FORM_ref1
172   {1,2}, // 0x12 DW_FORM_ref2
173   {1,4}, // 0x13 DW_FORM_ref4
174   {1,8}, // 0x14 DW_FORM_ref8
175   {0,0}, // 0x15 DW_FORM_ref_udata
176   {0,0}, // 0x16 DW_FORM_indirect
177   {1,4}, // 0x17 DW_FORM_sec_offset
178   {0,0}, // 0x18 DW_FORM_exprloc
179   {1,0}, // 0x19 DW_FORM_flag_present
180   {0,0}, // 0x1a
181   {0,0}, // 0x1b
182   {0,0}, // 0x1c
183   {0,0}, // 0x1d
184   {0,0}, // 0x1e
185   {0,0}, // 0x1f
186   {1,8}, // 0x20 DW_FORM_ref_sig8
187 };
188 
189 llvm::Optional<uint8_t>
190 DWARFFormValue::GetFixedSize(dw_form_t form, const DWARFUnit *u) {
191   if (form <= DW_FORM_ref_sig8 && g_form_sizes[form].valid)
192     return g_form_sizes[form].size;
193   if (form == DW_FORM_addr && u)
194     return u->GetAddressByteSize();
195   return llvm::None;
196 }
197 
198 llvm::Optional<uint8_t> DWARFFormValue::GetFixedSize() const {
199   return GetFixedSize(m_form, m_unit);
200 }
201 
202 bool DWARFFormValue::SkipValue(const DWARFDataExtractor &debug_info_data,
203                                lldb::offset_t *offset_ptr) const {
204   return DWARFFormValue::SkipValue(m_form, debug_info_data, offset_ptr, m_unit);
205 }
206 
207 bool DWARFFormValue::SkipValue(dw_form_t form,
208                                const DWARFDataExtractor &debug_info_data,
209                                lldb::offset_t *offset_ptr,
210                                const DWARFUnit *unit) {
211   uint8_t ref_addr_size;
212   switch (form) {
213   // Blocks if inlined data that have a length field and the data bytes inlined
214   // in the .debug_info
215   case DW_FORM_exprloc:
216   case DW_FORM_block: {
217     dw_uleb128_t size = debug_info_data.GetULEB128(offset_ptr);
218     *offset_ptr += size;
219   }
220     return true;
221   case DW_FORM_block1: {
222     dw_uleb128_t size = debug_info_data.GetU8(offset_ptr);
223     *offset_ptr += size;
224   }
225     return true;
226   case DW_FORM_block2: {
227     dw_uleb128_t size = debug_info_data.GetU16(offset_ptr);
228     *offset_ptr += size;
229   }
230     return true;
231   case DW_FORM_block4: {
232     dw_uleb128_t size = debug_info_data.GetU32(offset_ptr);
233     *offset_ptr += size;
234   }
235     return true;
236 
237   // Inlined NULL terminated C-strings
238   case DW_FORM_string:
239     debug_info_data.GetCStr(offset_ptr);
240     return true;
241 
242   // Compile unit address sized values
243   case DW_FORM_addr:
244     *offset_ptr += DWARFUnit::GetAddressByteSize(unit);
245     return true;
246 
247   case DW_FORM_ref_addr:
248     ref_addr_size = 4;
249     assert(unit); // Unit must be valid for DW_FORM_ref_addr objects or we will
250                   // get this wrong
251     if (unit->GetVersion() <= 2)
252       ref_addr_size = unit->GetAddressByteSize();
253     else
254       ref_addr_size = 4;
255     *offset_ptr += ref_addr_size;
256     return true;
257 
258   // 0 bytes values (implied from DW_FORM)
259   case DW_FORM_flag_present:
260   case DW_FORM_implicit_const:
261     return true;
262 
263     // 1 byte values
264     case DW_FORM_addrx1:
265     case DW_FORM_data1:
266     case DW_FORM_flag:
267     case DW_FORM_ref1:
268     case DW_FORM_strx1:
269       *offset_ptr += 1;
270       return true;
271 
272     // 2 byte values
273     case DW_FORM_addrx2:
274     case DW_FORM_data2:
275     case DW_FORM_ref2:
276     case DW_FORM_strx2:
277       *offset_ptr += 2;
278       return true;
279 
280     // 3 byte values
281     case DW_FORM_addrx3:
282     case DW_FORM_strx3:
283       *offset_ptr += 3;
284       return true;
285 
286     // 32 bit for DWARF 32, 64 for DWARF 64
287     case DW_FORM_sec_offset:
288     case DW_FORM_strp:
289       *offset_ptr += 4;
290       return true;
291 
292     // 4 byte values
293     case DW_FORM_addrx4:
294     case DW_FORM_data4:
295     case DW_FORM_ref4:
296     case DW_FORM_strx4:
297       *offset_ptr += 4;
298       return true;
299 
300     // 8 byte values
301     case DW_FORM_data8:
302     case DW_FORM_ref8:
303     case DW_FORM_ref_sig8:
304       *offset_ptr += 8;
305       return true;
306 
307     // signed or unsigned LEB 128 values
308     case DW_FORM_addrx:
309     case DW_FORM_loclistx:
310     case DW_FORM_rnglistx:
311     case DW_FORM_sdata:
312     case DW_FORM_udata:
313     case DW_FORM_ref_udata:
314     case DW_FORM_GNU_addr_index:
315     case DW_FORM_GNU_str_index:
316     case DW_FORM_strx:
317       debug_info_data.Skip_LEB128(offset_ptr);
318       return true;
319 
320   case DW_FORM_indirect: {
321     dw_form_t indirect_form = debug_info_data.GetULEB128(offset_ptr);
322     return DWARFFormValue::SkipValue(indirect_form, debug_info_data, offset_ptr,
323                                      unit);
324   }
325 
326   default:
327     break;
328   }
329   return false;
330 }
331 
332 void DWARFFormValue::Dump(Stream &s) const {
333   uint64_t uvalue = Unsigned();
334   bool unit_relative_offset = false;
335 
336   switch (m_form) {
337   case DW_FORM_addr:
338     DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t));
339     break;
340   case DW_FORM_flag:
341   case DW_FORM_data1:
342     s.PutHex8(uvalue);
343     break;
344   case DW_FORM_data2:
345     s.PutHex16(uvalue);
346     break;
347   case DW_FORM_sec_offset:
348   case DW_FORM_data4:
349     s.PutHex32(uvalue);
350     break;
351   case DW_FORM_ref_sig8:
352   case DW_FORM_data8:
353     s.PutHex64(uvalue);
354     break;
355   case DW_FORM_string:
356     s.QuotedCString(AsCString());
357     break;
358   case DW_FORM_exprloc:
359   case DW_FORM_block:
360   case DW_FORM_block1:
361   case DW_FORM_block2:
362   case DW_FORM_block4:
363     if (uvalue > 0) {
364       switch (m_form) {
365       case DW_FORM_exprloc:
366       case DW_FORM_block:
367         s.Printf("<0x%" PRIx64 "> ", uvalue);
368         break;
369       case DW_FORM_block1:
370         s.Printf("<0x%2.2x> ", (uint8_t)uvalue);
371         break;
372       case DW_FORM_block2:
373         s.Printf("<0x%4.4x> ", (uint16_t)uvalue);
374         break;
375       case DW_FORM_block4:
376         s.Printf("<0x%8.8x> ", (uint32_t)uvalue);
377         break;
378       default:
379         break;
380       }
381 
382       const uint8_t *data_ptr = m_value.data;
383       if (data_ptr) {
384         const uint8_t *end_data_ptr =
385             data_ptr + uvalue; // uvalue contains size of block
386         while (data_ptr < end_data_ptr) {
387           s.Printf("%2.2x ", *data_ptr);
388           ++data_ptr;
389         }
390       } else
391         s.PutCString("NULL");
392     }
393     break;
394 
395   case DW_FORM_sdata:
396     s.PutSLEB128(uvalue);
397     break;
398   case DW_FORM_udata:
399     s.PutULEB128(uvalue);
400     break;
401   case DW_FORM_strp: {
402     const char *dbg_str = AsCString();
403     if (dbg_str) {
404       s.QuotedCString(dbg_str);
405     } else {
406       s.PutHex32(uvalue);
407     }
408   } break;
409 
410   case DW_FORM_ref_addr: {
411     assert(m_unit); // Unit must be valid for DW_FORM_ref_addr objects or we
412                     // will get this wrong
413     if (m_unit->GetVersion() <= 2)
414       DumpAddress(s.AsRawOstream(), uvalue, sizeof(uint64_t) * 2);
415     else
416       DumpAddress(s.AsRawOstream(), uvalue,
417                   4 * 2); // 4 for DWARF32, 8 for DWARF64, but we don't
418                           // support DWARF64 yet
419     break;
420   }
421   case DW_FORM_ref1:
422     unit_relative_offset = true;
423     break;
424   case DW_FORM_ref2:
425     unit_relative_offset = true;
426     break;
427   case DW_FORM_ref4:
428     unit_relative_offset = true;
429     break;
430   case DW_FORM_ref8:
431     unit_relative_offset = true;
432     break;
433   case DW_FORM_ref_udata:
434     unit_relative_offset = true;
435     break;
436 
437   // All DW_FORM_indirect attributes should be resolved prior to calling this
438   // function
439   case DW_FORM_indirect:
440     s.PutCString("DW_FORM_indirect");
441     break;
442   case DW_FORM_flag_present:
443     break;
444   default:
445     s.Printf("DW_FORM(0x%4.4x)", m_form);
446     break;
447   }
448 
449   if (unit_relative_offset) {
450     assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
451                     // unit relative or we will get this wrong
452     s.Printf("{0x%8.8" PRIx64 "}", uvalue + m_unit->GetOffset());
453   }
454 }
455 
456 const char *DWARFFormValue::AsCString() const {
457   SymbolFileDWARF &symbol_file = m_unit->GetSymbolFileDWARF();
458 
459   if (m_form == DW_FORM_string) {
460     return m_value.value.cstr;
461   } else if (m_form == DW_FORM_strp) {
462     return symbol_file.GetDWARFContext().getOrLoadStrData().PeekCStr(
463         m_value.value.uval);
464   } else if (m_form == DW_FORM_GNU_str_index) {
465     uint32_t index_size = 4;
466     lldb::offset_t offset = m_value.value.uval * index_size;
467     dw_offset_t str_offset =
468         symbol_file.GetDWARFContext().getOrLoadStrOffsetsData().GetMaxU64(
469             &offset, index_size);
470     return symbol_file.GetDWARFContext().getOrLoadStrData().PeekCStr(
471         str_offset);
472   }
473 
474   if (m_form == DW_FORM_strx || m_form == DW_FORM_strx1 ||
475       m_form == DW_FORM_strx2 || m_form == DW_FORM_strx3 ||
476       m_form == DW_FORM_strx4) {
477 
478     // The same code as above.
479     uint32_t indexSize = 4;
480     lldb::offset_t offset =
481         m_unit->GetStrOffsetsBase() + m_value.value.uval * indexSize;
482     dw_offset_t strOffset =
483         symbol_file.GetDWARFContext().getOrLoadStrOffsetsData().GetMaxU64(
484             &offset, indexSize);
485     return symbol_file.GetDWARFContext().getOrLoadStrData().PeekCStr(strOffset);
486   }
487 
488   if (m_form == DW_FORM_line_strp)
489     return symbol_file.GetDWARFContext().getOrLoadLineStrData().PeekCStr(
490         m_value.value.uval);
491 
492   return nullptr;
493 }
494 
495 dw_addr_t DWARFFormValue::Address() const {
496   SymbolFileDWARF &symbol_file = m_unit->GetSymbolFileDWARF();
497 
498   if (m_form == DW_FORM_addr)
499     return Unsigned();
500 
501   assert(m_unit);
502   assert(m_form == DW_FORM_GNU_addr_index || m_form == DW_FORM_addrx ||
503          m_form == DW_FORM_addrx1 || m_form == DW_FORM_addrx2 ||
504          m_form == DW_FORM_addrx3 || m_form == DW_FORM_addrx4);
505 
506   uint32_t index_size = m_unit->GetAddressByteSize();
507   dw_offset_t addr_base = m_unit->GetAddrBase();
508   lldb::offset_t offset = addr_base + m_value.value.uval * index_size;
509   return symbol_file.GetDWARFContext().getOrLoadAddrData().GetMaxU64(
510       &offset, index_size);
511 }
512 
513 DWARFDIE DWARFFormValue::Reference() const {
514   uint64_t value = m_value.value.uval;
515   switch (m_form) {
516   case DW_FORM_ref1:
517   case DW_FORM_ref2:
518   case DW_FORM_ref4:
519   case DW_FORM_ref8:
520   case DW_FORM_ref_udata:
521     assert(m_unit); // Unit must be valid for DW_FORM_ref forms that are compile
522                     // unit relative or we will get this wrong
523     value += m_unit->GetOffset();
524     if (!m_unit->ContainsDIEOffset(value)) {
525       m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
526           "DW_FORM_ref* DIE reference 0x%" PRIx64 " is outside of its CU",
527           value);
528       return {};
529     }
530     return const_cast<DWARFUnit *>(m_unit)->GetDIE(value);
531 
532   case DW_FORM_ref_addr: {
533     DWARFUnit *ref_cu =
534         m_unit->GetSymbolFileDWARF().DebugInfo()->GetUnitContainingDIEOffset(
535             DIERef::Section::DebugInfo, value);
536     if (!ref_cu) {
537       m_unit->GetSymbolFileDWARF().GetObjectFile()->GetModule()->ReportError(
538           "DW_FORM_ref_addr DIE reference 0x%" PRIx64 " has no matching CU",
539           value);
540       return {};
541     }
542     return ref_cu->GetDIE(value);
543   }
544 
545   case DW_FORM_ref_sig8: {
546     DWARFTypeUnit *tu =
547         m_unit->GetSymbolFileDWARF().DebugInfo()->GetTypeUnitForHash(value);
548     if (!tu)
549       return {};
550     return tu->GetDIE(tu->GetTypeOffset());
551   }
552 
553   default:
554     return {};
555   }
556 }
557 
558 uint64_t DWARFFormValue::Reference(dw_offset_t base_offset) const {
559   uint64_t value = m_value.value.uval;
560   switch (m_form) {
561   case DW_FORM_ref1:
562   case DW_FORM_ref2:
563   case DW_FORM_ref4:
564   case DW_FORM_ref8:
565   case DW_FORM_ref_udata:
566     return value + base_offset;
567 
568   case DW_FORM_ref_addr:
569   case DW_FORM_ref_sig8:
570   case DW_FORM_GNU_ref_alt:
571     return value;
572 
573   default:
574     return DW_INVALID_OFFSET;
575   }
576 }
577 
578 const uint8_t *DWARFFormValue::BlockData() const { return m_value.data; }
579 
580 bool DWARFFormValue::IsBlockForm(const dw_form_t form) {
581   switch (form) {
582   case DW_FORM_exprloc:
583   case DW_FORM_block:
584   case DW_FORM_block1:
585   case DW_FORM_block2:
586   case DW_FORM_block4:
587     return true;
588   }
589   return false;
590 }
591 
592 bool DWARFFormValue::IsDataForm(const dw_form_t form) {
593   switch (form) {
594   case DW_FORM_sdata:
595   case DW_FORM_udata:
596   case DW_FORM_data1:
597   case DW_FORM_data2:
598   case DW_FORM_data4:
599   case DW_FORM_data8:
600     return true;
601   }
602   return false;
603 }
604 
605 bool DWARFFormValue::FormIsSupported(dw_form_t form) {
606   switch (form) {
607     case DW_FORM_addr:
608     case DW_FORM_addrx:
609     case DW_FORM_loclistx:
610     case DW_FORM_rnglistx:
611     case DW_FORM_block2:
612     case DW_FORM_block4:
613     case DW_FORM_data2:
614     case DW_FORM_data4:
615     case DW_FORM_data8:
616     case DW_FORM_string:
617     case DW_FORM_block:
618     case DW_FORM_block1:
619     case DW_FORM_data1:
620     case DW_FORM_flag:
621     case DW_FORM_sdata:
622     case DW_FORM_strp:
623     case DW_FORM_strx:
624     case DW_FORM_strx1:
625     case DW_FORM_strx2:
626     case DW_FORM_strx3:
627     case DW_FORM_strx4:
628     case DW_FORM_udata:
629     case DW_FORM_ref_addr:
630     case DW_FORM_ref1:
631     case DW_FORM_ref2:
632     case DW_FORM_ref4:
633     case DW_FORM_ref8:
634     case DW_FORM_ref_udata:
635     case DW_FORM_indirect:
636     case DW_FORM_sec_offset:
637     case DW_FORM_exprloc:
638     case DW_FORM_flag_present:
639     case DW_FORM_ref_sig8:
640     case DW_FORM_GNU_str_index:
641     case DW_FORM_GNU_addr_index:
642     case DW_FORM_implicit_const:
643       return true;
644     default:
645       break;
646   }
647   return false;
648 }
649