aboutsummaryrefslogtreecommitdiff
path: root/src/common/linux/dwarf_cu_to_module.cc
blob: 8063d25fea8bf878caa9b3b9546edd8d06df94df (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
// Copyright (c) 2009, Google Inc.
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
//     * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

#include <algorithm>
#include <cassert>

#include "common/linux/dwarf_cu_to_module.h"
#include "common/linux/dwarf_line_to_module.h"

namespace google_breakpad {

using std::map;
using std::vector;

// Data provided by a DWARF specification DIE.
// 
// In DWARF, the DIE for a definition may contain a DW_AT_specification
// attribute giving the offset of the corresponding declaration DIE, and
// the definition DIE may omit information given in the declaration. For
// example, it's common for a function's address range to appear only in
// its definition DIE, but its name to appear only in its declaration
// DIE.
//
// The dumper needs to be able to follow DW_AT_specification links to
// bring all this information together in a FUNC record. Conveniently,
// DIEs that are the target of such links have a DW_AT_declaration flag
// set, so we can identify them when we first see them, and record their
// contents for later reference.
//
// A Specification holds information gathered from a declaration DIE that
// we may need if we find a DW_AT_specification link pointing to it.
struct DwarfCUToModule::Specification {
  // The name of the enclosing scope, or the empty string if there is none.
  string enclosing_name;

  // The name for the specification DIE itself, without any enclosing
  // name components.
  string unqualified_name;
};

// Data global to the DWARF-bearing file that is private to the
// DWARF-to-Module process.
struct DwarfCUToModule::FilePrivate {
  // A map from offsets of DIEs within the .debug_info section to
  // Specifications describing those DIEs. Specification references can
  // cross compilation unit boundaries.
  SpecificationByOffset specifications;
};

DwarfCUToModule::FileContext::FileContext(const string &filename_arg,
                                          Module *module_arg)
    : filename(filename_arg), module(module_arg) {
  file_private = new FilePrivate();
}

DwarfCUToModule::FileContext::~FileContext() {
  delete file_private;
}

// Information global to the particular compilation unit we're
// parsing. This is for data shared across the CU's entire DIE tree,
// and parameters from the code invoking the CU parser.
struct DwarfCUToModule::CUContext {
  CUContext(FileContext *file_context_arg, WarningReporter *reporter_arg)
      : file_context(file_context_arg),
        reporter(reporter_arg),
        language(Language::CPlusPlus) { }
  ~CUContext() {
    for (vector<Module::Function *>::iterator it = functions.begin();
         it != functions.end(); it++)
      delete *it;
  };

  // The DWARF-bearing file into which this CU was incorporated.
  FileContext *file_context;

  // For printing error messages.
  WarningReporter *reporter;

  // The source language of this compilation unit.
  const Language *language;

  // The functions defined in this compilation unit. We accumulate
  // them here during parsing. Then, in DwarfCUToModule::Finish, we
  // assign them lines and add them to file_context->module.
  //
  // Destroying this destroys all the functions this vector points to.
  vector<Module::Function *> functions;
};

// Information about the context of a particular DIE. This is for
// information that changes as we descend the tree towards the leaves:
// the containing classes/namespaces, etc.
struct DwarfCUToModule::DIEContext {
  // The fully-qualified name of the context. For example, for a
  // tree like:
  //
  // DW_TAG_namespace Foo
  //   DW_TAG_class Bar
  //     DW_TAG_subprogram Baz
  //
  // in a C++ compilation unit, the DIEContext's name for the
  // DW_TAG_subprogram DIE would be "Foo::Bar". The DIEContext's
  // name for the DW_TAG_namespace DIE would be "".
  string name;
};

// An abstract base class for all the dumper's DIE handlers.
class DwarfCUToModule::GenericDIEHandler: public dwarf2reader::DIEHandler {
 public:
  // Create a handler for the DIE at OFFSET whose compilation unit is
  // described by CU_CONTEXT, and whose immediate context is described
  // by PARENT_CONTEXT.
  GenericDIEHandler(CUContext *cu_context, DIEContext *parent_context,
                    uint64 offset)
      : cu_context_(cu_context),
        parent_context_(parent_context),
        offset_(offset),
        declaration_(false),
        specification_(NULL) { }

  // Derived classes' ProcessAttributeUnsigned can defer to this to
  // handle DW_AT_declaration, or simply not override it.
  void ProcessAttributeUnsigned(enum DwarfAttribute attr,
                                enum DwarfForm form,
                                uint64 data);

  // Derived classes' ProcessAttributeReference can defer to this to
  // handle DW_AT_specification, or simply not override it.
  void ProcessAttributeReference(enum DwarfAttribute attr,
                                 enum DwarfForm form,
                                 uint64 data);

  // Derived classes' ProcessAttributeReference can defer to this to
  // handle DW_AT_specification, or simply not override it.
  void ProcessAttributeString(enum DwarfAttribute attr,
                              enum DwarfForm form,
                              const string &data);

 protected:
  // Compute and return the fully-qualified name of the DIE. If this
  // DIE is a declaration DIE, to be cited by other DIEs'
  // DW_AT_specification attributes, record its enclosing name and
  // unqualified name in the specification table.
  string ComputeQualifiedName();

  CUContext *cu_context_;
  DIEContext *parent_context_;
  uint64 offset_;

  // If this DIE has a DW_AT_declaration attribute, this is its value.
  // It is false on DIEs with no DW_AT_declaration attribute.
  bool declaration_;

  // If this DIE has a DW_AT_specification attribute, this is the
  // Specification structure for the DIE the attribute refers to.
  // Otherwise, this is NULL.
  Specification *specification_;

  // The value of the DW_AT_name attribute, or the empty string if the
  // DIE has no such attribute.
  string name_attribute_;
};

void DwarfCUToModule::GenericDIEHandler::ProcessAttributeUnsigned(
    enum DwarfAttribute attr,
    enum DwarfForm form,
    uint64 data) {
  switch (attr) {
    case dwarf2reader::DW_AT_declaration: declaration_ = (data != 0); break;
    default: break;
  }
}

void DwarfCUToModule::GenericDIEHandler::ProcessAttributeReference(
    enum DwarfAttribute attr,
    enum DwarfForm form,
    uint64 data) {
  switch (attr) {
    case dwarf2reader::DW_AT_specification: {
      FileContext *file_context = cu_context_->file_context;
      SpecificationByOffset *specifications
          = &file_context->file_private->specifications;
      SpecificationByOffset::iterator spec = specifications->find(data);
      if (spec != specifications->end()) {
        specification_ = &spec->second;
      } else {
        // Technically, there's no reason a DW_AT_specification
        // couldn't be a forward reference, but supporting that would
        // be a lot of work (changing to a two-pass structure), and I
        // don't think any producers we care about ever emit such
        // things.
        cu_context_->reporter->UnknownSpecification(offset_, data);
      }        
      break;
    }
    default: break;
  }
}

void DwarfCUToModule::GenericDIEHandler::ProcessAttributeString(
    enum DwarfAttribute attr,
    enum DwarfForm form,
    const string &data) {
  switch (attr) {
    case dwarf2reader::DW_AT_name: name_attribute_ = data; break;
    default: break;
  }
}

string DwarfCUToModule::GenericDIEHandler::ComputeQualifiedName() {
  // Find our unqualified name. If the DIE has its own DW_AT_name
  // attribute, then use that; otherwise, check our specification.
  const string *unqualified_name;
  if (name_attribute_.empty() && specification_)
    unqualified_name = &specification_->unqualified_name;
  else
    unqualified_name = &name_attribute_;

  // Find the name of our enclosing context. If we have a
  // specification, it's the specification's enclosing context that
  // counts; otherwise, use this DIE's context.
  const string *enclosing_name;
  if (specification_)
    enclosing_name = &specification_->enclosing_name;
  else
    enclosing_name = &parent_context_->name;

  // If this DIE was marked as a declaration, record its names in the
  // specification table.
  if (declaration_) {
    FileContext *file_context = cu_context_->file_context;
    Specification spec;
    spec.enclosing_name = *enclosing_name;
    spec.unqualified_name = *unqualified_name;
    file_context->file_private->specifications[offset_] = spec;
  }

  // Combine the enclosing name and unqualified name to produce our
  // own fully-qualified name.
  return cu_context_->language->MakeQualifiedName(*enclosing_name,
                                                  *unqualified_name);
}

// A handler class for DW_TAG_subprogram DIEs.
class DwarfCUToModule::FuncHandler: public GenericDIEHandler {
 public:
  FuncHandler(CUContext *cu_context, DIEContext *parent_context,
              uint64 offset)
      : GenericDIEHandler(cu_context, parent_context, offset),
        low_pc_(0), high_pc_(0) { }
  void ProcessAttributeUnsigned(enum DwarfAttribute attr,
                                enum DwarfForm form,
                                uint64 data);
  bool EndAttributes();
  void Finish();

 private:
  // The fully-qualified name, as derived from name_attribute_,
  // specification_, parent_context_.  Computed in EndAttributes.
  string name_;
  uint64 low_pc_, high_pc_; // DW_AT_low_pc, DW_AT_high_pc
};

void DwarfCUToModule::FuncHandler::ProcessAttributeUnsigned(
    enum DwarfAttribute attr,
    enum DwarfForm form,
    uint64 data) {
  switch (attr) {
    case dwarf2reader::DW_AT_low_pc:      low_pc_  = data; break;
    case dwarf2reader::DW_AT_high_pc:     high_pc_ = data; break;
    default:
      GenericDIEHandler::ProcessAttributeUnsigned(attr, form, data);
      break;
  }
}

bool DwarfCUToModule::FuncHandler::EndAttributes() {
  // Compute our name, and record a specification, if appropriate.
  name_ = ComputeQualifiedName();
  return true;
}

void DwarfCUToModule::FuncHandler::Finish() {
  // Did we collect the information we need?  Not all DWARF function
  // entries have low and high addresses (for example, inlined
  // functions that were never used), but all the ones we're
  // interested in cover a non-empty range of bytes.
  if (low_pc_ < high_pc_) {
    // Create a Module::Function based on the data we've gathered, and
    // add it to the functions_ list.
    Module::Function *func = new Module::Function;
    func->name_ = name_;
    func->address_ = low_pc_;
    func->size_ = high_pc_ - low_pc_;
    func->parameter_size_ = 0;
    cu_context_->functions.push_back(func);
  }
}

// A handler for DIEs that contain functions and contribute a
// component to their names: namespaces, classes, etc.
class DwarfCUToModule::NamedScopeHandler: public GenericDIEHandler {
 public:
  NamedScopeHandler(CUContext *cu_context, DIEContext *parent_context,
                    uint64 offset)
      : GenericDIEHandler(cu_context, parent_context, offset) { }
  bool EndAttributes();
  DIEHandler *FindChildHandler(uint64 offset, enum DwarfTag tag,
                               const AttributeList &attrs);

 private:
  DIEContext child_context_; // A context for our children.
};

bool DwarfCUToModule::NamedScopeHandler::EndAttributes() {
  child_context_.name = ComputeQualifiedName();
  return true;
}

dwarf2reader::DIEHandler *DwarfCUToModule::NamedScopeHandler::FindChildHandler(
    uint64 offset,
    enum DwarfTag tag,
    const AttributeList &attrs) {
  switch (tag) {
    case dwarf2reader::DW_TAG_subprogram:
      return new FuncHandler(cu_context_, &child_context_, offset);
    case dwarf2reader::DW_TAG_namespace:
    case dwarf2reader::DW_TAG_class_type:
    case dwarf2reader::DW_TAG_structure_type:
    case dwarf2reader::DW_TAG_union_type:
      return new NamedScopeHandler(cu_context_, &child_context_, offset);
    default:
      return NULL;
  }
};

void DwarfCUToModule::WarningReporter::CUHeading() {
  if (printed_cu_header_)
    return;
  fprintf(stderr, "%s: in compilation unit '%s' (offset 0x%llx):\n",
          filename_.c_str(), cu_name_.c_str(), cu_offset_);
  printed_cu_header_ = true;
}

void DwarfCUToModule::WarningReporter::UnknownSpecification(uint64 offset,
                                                            uint64 target) {
  CUHeading();
  fprintf(stderr, "%s: the DIE at offset 0x%llx has a DW_AT_specification"
          " attribute referring to the die at offset 0x%llx, which either"
          " was not marked as a declaration, or comes later in the file",
          filename_.c_str(), offset, target);
}

void DwarfCUToModule::WarningReporter::MissingSection(const string &name) {
  CUHeading();
  fprintf(stderr, "%s: warning: couldn't find DWARF '%s' section\n",
          filename_.c_str(), name.c_str());
}

void DwarfCUToModule::WarningReporter::BadLineInfoOffset(uint64 offset) {
  CUHeading();
  fprintf(stderr, "%s: warning: line number data offset beyond end"
          " of '.debug_line' section\n",
          filename_.c_str());
}

void DwarfCUToModule::WarningReporter::UncoveredHeading() {
  if (printed_unpaired_header_)
    return;
  CUHeading();
  fprintf(stderr, "%s: warning: skipping unpaired lines/functions:\n",
          filename_.c_str());
  printed_unpaired_header_ = true;
}

void DwarfCUToModule::WarningReporter::UncoveredFunction(
    const Module::Function &function) {
  UncoveredHeading();
  fprintf(stderr, "    function%s: %s\n",
          function.size_ == 0 ? " (zero-length)" : "",
          function.name_.c_str());
}

void DwarfCUToModule::WarningReporter::UncoveredLine(const Module::Line &line) {
  UncoveredHeading();
  fprintf(stderr, "    line%s: %s:%d at 0x%llx\n",
          (line.size_ == 0 ? " (zero-length)" : ""),
          line.file_->name_.c_str(), line.number_, line.address_);
}

DwarfCUToModule::DwarfCUToModule(FileContext *file_context,
                                 LineToModuleFunctor *line_reader,
                                 WarningReporter *reporter)
    : line_reader_(line_reader), has_source_line_info_(false) { 
  cu_context_ = new CUContext(file_context, reporter);
  child_context_ = new DIEContext();
}

DwarfCUToModule::~DwarfCUToModule() {
  delete cu_context_;
  delete child_context_;
}

void DwarfCUToModule::ProcessAttributeSigned(enum DwarfAttribute attr,
                                             enum DwarfForm form,
                                             int64 data) {
  switch (attr) {
    case dwarf2reader::DW_AT_language: // source language of this CU
      SetLanguage(static_cast<DwarfLanguage>(data));
      break;
    default:
      break;
  }
}

void DwarfCUToModule::ProcessAttributeUnsigned(enum DwarfAttribute attr,
                                               enum DwarfForm form,
                                               uint64 data) {
  switch (attr) {
    case dwarf2reader::DW_AT_stmt_list: // Line number information.
      has_source_line_info_ = true;
      source_line_offset_ = data;
      break;
    case dwarf2reader::DW_AT_language: // source language of this CU
      SetLanguage(static_cast<DwarfLanguage>(data));
      break;
    default:
      break;
  }
}

void DwarfCUToModule::ProcessAttributeString(enum DwarfAttribute attr,
                                             enum DwarfForm form,
                                             const string &data) {
  if (attr == dwarf2reader::DW_AT_name)
    cu_context_->reporter->SetCUName(data);
}

bool DwarfCUToModule::EndAttributes() {
  return true;
}

dwarf2reader::DIEHandler *DwarfCUToModule::FindChildHandler(
    uint64 offset,
    enum DwarfTag tag,
    const AttributeList &attrs) {
  switch (tag) {
    case dwarf2reader::DW_TAG_subprogram:
      return new FuncHandler(cu_context_, child_context_, offset);
    case dwarf2reader::DW_TAG_namespace:
    case dwarf2reader::DW_TAG_class_type:
    case dwarf2reader::DW_TAG_structure_type:
    case dwarf2reader::DW_TAG_union_type:
      return new NamedScopeHandler(cu_context_, child_context_, offset);
    default:
      return NULL;
  }
}

void DwarfCUToModule::SetLanguage(DwarfLanguage language) {
  switch (language) {
    case dwarf2reader::DW_LANG_Java:
      cu_context_->language = Language::Java;
      break;

    // DWARF has no generic language code for assembly language; this is
    // what the GNU toolchain uses.
    case dwarf2reader::DW_LANG_Mips_Assembler:
      cu_context_->language = Language::Assembler;
      break;

    // C++ covers so many cases that it probably has some way to cope
    // with whatever the other languages throw at us. So make it the
    // default.
    //
    // Objective C and Objective C++ seem to create entries for
    // methods whose DW_AT_name values are already fully-qualified:
    // "-[Classname method:]".  These appear at the top level.
    // 
    // DWARF data for C should never include namespaces or functions
    // nested in struct types, but if it ever does, then C++'s
    // notation is probably not a bad choice for that.
    default:
    case dwarf2reader::DW_LANG_ObjC:
    case dwarf2reader::DW_LANG_ObjC_plus_plus:
    case dwarf2reader::DW_LANG_C:
    case dwarf2reader::DW_LANG_C89:
    case dwarf2reader::DW_LANG_C99:
    case dwarf2reader::DW_LANG_C_plus_plus:
      cu_context_->language = Language::CPlusPlus;
      break;
  }
}

void DwarfCUToModule::ReadSourceLines(uint64 offset) {
  const dwarf2reader::SectionMap &section_map
      = cu_context_->file_context->section_map;
  dwarf2reader::SectionMap::const_iterator map_entry
      = section_map.find(".debug_line");
  if (map_entry == section_map.end()) {
    cu_context_->reporter->MissingSection(".debug_line");
    return;
  }
  const char *section_start = map_entry->second.first;
  uint64 section_length = map_entry->second.second;
  if (offset >= section_length) {
    cu_context_->reporter->BadLineInfoOffset(offset);
    return;
  }
  (*line_reader_)(section_start + offset, section_length - offset,
                  cu_context_->file_context->module, &lines_);
}

namespace {
// Return true if ADDRESS falls within the range of ITEM.
template <class T>
inline bool within(const T &item, Module::Address address) {
  // Because Module::Address is unsigned, and unsigned arithmetic
  // wraps around, this will be false if ADDRESS falls before the
  // start of ITEM, or if it falls after ITEM's end.
  return address - item.address_ < item.size_;
}
}

void DwarfCUToModule::AssignLinesToFunctions() {
  vector<Module::Function *> *functions = &cu_context_->functions;
  WarningReporter *reporter = cu_context_->reporter;

  // This would be simpler if we assumed that source line entries
  // don't cross function boundaries.  However, there's no real reason
  // to assume that (say) a series of function definitions on the same
  // line wouldn't get coalesced into one line number entry.  The
  // DWARF spec certainly makes no such promises.
  //
  // So treat the functions and lines as peers, and take the trouble
  // to compute their ranges' intersections precisely.  In any case,
  // the hair here is a constant factor for performance; the
  // complexity from here on out is linear.

  // Put both our functions and lines in order by address.
  sort(functions->begin(), functions->end(),
       Module::Function::CompareByAddress);
  sort(lines_.begin(), lines_.end(), Module::Line::CompareByAddress);

  // The last line that we used any piece of.  We use this only for
  // generating warnings.
  const Module::Line *last_line_used = NULL;

  // The last function and line we warned about --- so we can avoid
  // doing so more than once.
  const Module::Function *last_function_cited = NULL;
  const Module::Line *last_line_cited = NULL;

  // Make a single pass through both vectors from lower to higher
  // addresses, populating each Function's lines_ vector with lines
  // from our lines_ vector that fall within the function's address
  // range.
  vector<Module::Function *>::iterator func_it = functions->begin();
  vector<Module::Line>::const_iterator line_it = lines_.begin();

  Module::Address current;

  // Pointers to the referents of func_it and line_it, or NULL if the
  // iterator is at the end of the sequence.
  Module::Function *func;
  const Module::Line *line;

  // Start current at the beginning of the first line or function,
  // whichever is earlier.
  if (func_it != functions->end() && line_it != lines_.end()) {
    func = *func_it;
    line = &*line_it;
    current = std::min(func->address_, line->address_);
  } else if (line_it != lines_.end()) {
    func = NULL;
    line = &*line_it;
    current = line->address_;
  } else if (func_it != functions->end()) {
    func = *func_it;
    line = NULL;
    current = (*func_it)->address_;
  } else {
    return;
  }

  while (func || line) {
    // This loop has two invariants that hold at the top.
    //
    // First, at least one of the iterators is not at the end of its
    // sequence, and those that are not refer to the earliest
    // function or line that contains or starts after CURRENT.
    //
    // Note that every byte is in one of four states: it is covered
    // or not covered by a function, and, independently, it is
    // covered or not covered by a line.
    //
    // The second invariant is that CURRENT refers to a byte whose
    // state is different from its predecessor, or it refers to the
    // first byte in the address space. In other words, CURRENT is
    // always the address of a transition.
    //
    // Note that, although each iteration advances CURRENT from one
    // transition address to the next in each iteration, it might
    // not advance the iterators. Suppose we have a function that
    // starts with a line, has a gap, and then a second line, and
    // suppose that we enter an iteration with CURRENT at the end of
    // the first line. The next transition address is the start of
    // the second line, after the gap, so the iteration should
    // advance CURRENT to that point. At the head of that iteration,
    // the invariants require that the line iterator be pointing at
    // the second line. But this is also true at the head of the
    // next. And clearly, the iteration must not change the function
    // iterator. So neither iterator moves.

    // Assert the first invariant (see above).
    assert(!func || current < func->address_ || within(*func, current));
    assert(!line || current < line->address_ || within(*line, current));

    // The next transition after CURRENT.
    Module::Address next_transition;

    // Figure out which state we're in, add lines or warn, and compute
    // the next transition address.
    if (func && current >= func->address_) {
      if (line && current >= line->address_) {
        // Covered by both a line and a function.
        Module::Address func_left = func->size_ - (current - func->address_);
        Module::Address line_left = line->size_ - (current - line->address_);
        // This may overflow, but things work out.
        next_transition = current + std::min(func_left, line_left);
        Module::Line l = *line;
        l.address_ = current;
        l.size_ = next_transition - current;
        func->lines_.push_back(l);
        last_line_used = line;
      } else {
        // Covered by a function, but no line.
        if (func != last_function_cited) {
          reporter->UncoveredFunction(*func);
          last_function_cited = func;
        }
        if (line && within(*func, line->address_))
          next_transition = line->address_;
        else
          // If this overflows, we'll catch it below.
          next_transition = func->address_ + func->size_;
      }
    } else {
      if (line && current >= line->address_) {
        // Covered by a line, but no function.
        //
        // If GCC emits padding after one function to align the start
        // of the next, then it will attribute the padding
        // instructions to the last source line of function (to reduce
        // the size of the line number info), but omit it from the
        // DW_AT_{low,high}_pc range given in .debug_info (since it
        // costs nothing to be precise there). If we did use at least
        // some of the line we're about to skip, and it ends at the
        // start of the next function, then assume this is what
        // happened, and don't warn.
        if (line != last_line_cited
            && !(func
                 && line == last_line_used
                 && func->address_ - line->address_ == line->size_)) {
          reporter->UncoveredLine(*line);
          last_line_cited = line;
        }
        if (func && within(*line, func->address_))
          next_transition = func->address_;
        else
          // If this overflows, we'll catch it below.
          next_transition = line->address_ + line->size_;
      } else {
        // Covered by neither a function nor a line. By the invariant,
        // both func and line begin after CURRENT. The next transition
        // is the start of the next function or next line, whichever
        // is earliest.
        assert (func || line);
        if (func && line)
          next_transition = std::min(func->address_, line->address_);
        else if (func)
          next_transition = func->address_;
        else
          next_transition = line->address_;
      }
    }

    // If a function or line abuts the end of the address space, then
    // next_transition may end up being zero, in which case we've completed
    // our pass. Handle that here, instead of trying to deal with it in
    // each place we compute next_transition.
    if (!next_transition)
      break;

    // Advance iterators as needed. If lines overlap or functions overlap,
    // then we could go around more than once. We don't worry too much
    // about what result we produce in that case, just as long as we don't
    // hang or crash.
    while (func_it != functions->end()
           && current >= (*func_it)->address_
           && !within(**func_it, next_transition))
      func_it++;
    func = (func_it != functions->end()) ? *func_it : NULL;
    while (line_it != lines_.end()
           && current >= line_it->address_
           && !within(*line_it, next_transition))
      line_it++;
    line = (line_it != lines_.end()) ? &*line_it : NULL;

    // We must make progress.
    assert(next_transition > current);
    current = next_transition;
  }
}

void DwarfCUToModule::Finish() {
  // Assembly language files have no function data, and that gives us
  // no place to store our line numbers (even though the GNU toolchain
  // will happily produce source line info for assembly language
  // files).  To avoid spurious warnings about lines we can't assign
  // to functions, skip CUs in languages that lack functions.
  if (!cu_context_->language->HasFunctions())
    return;

  // Read source line info, if we have any.
  if (has_source_line_info_)
    ReadSourceLines(source_line_offset_);

  vector<Module::Function *> *functions = &cu_context_->functions;

  // Dole out lines to the appropriate functions.
  AssignLinesToFunctions();

  // Add our functions, which now have source lines assigned to them,
  // to module_.
  cu_context_->file_context->module->AddFunctions(functions->begin(),
                                                  functions->end());

  // Ownership of the function objects has shifted from cu_context to
  // the Module.
  functions->clear();
}

bool DwarfCUToModule::StartCompilationUnit(uint64 offset,
                                           uint8 address_size,
                                           uint8 offset_size,
                                           uint64 cu_length,
                                           uint8 dwarf_version) {
  return dwarf_version >= 2;
}

bool DwarfCUToModule::StartRootDIE(uint64 offset, enum DwarfTag tag,
                                   const AttributeList& attrs) {
  // We don't deal with partial compilation units (the only other tag
  // likely to be used for root DIE).
  return tag == dwarf2reader::DW_TAG_compile_unit;
}

} // namespace google_breakpad