dex_file.h revision 576ca0cd692c0b6ae70e776de91015b8ff000a08
1/*
2 * Copyright (C) 2011 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 *      http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#ifndef ART_RUNTIME_DEX_FILE_H_
18#define ART_RUNTIME_DEX_FILE_H_
19
20#include <memory>
21#include <string>
22#include <vector>
23
24#include "base/logging.h"
25#include "base/mutex.h"  // For Locks::mutator_lock_.
26#include "globals.h"
27#include "invoke_type.h"
28#include "jni.h"
29#include "modifiers.h"
30#include "safe_map.h"
31
32namespace art {
33
34// TODO: remove dependencies on mirror classes, primarily by moving
35// EncodedStaticFieldValueIterator to its own file.
36namespace mirror {
37  class ArtField;
38  class ArtMethod;
39  class ClassLoader;
40  class DexCache;
41}  // namespace mirror
42class ClassLinker;
43class MemMap;
44class Signature;
45template<class T> class Handle;
46class StringPiece;
47class ZipArchive;
48
49// TODO: move all of the macro functionality into the DexCache class.
50class DexFile {
51 public:
52  static const byte kDexMagic[];
53  static const byte kDexMagicVersion[];
54  static const size_t kSha1DigestSize = 20;
55  static const uint32_t kDexEndianConstant = 0x12345678;
56
57  // name of the DexFile entry within a zip archive
58  static const char* kClassesDex;
59
60  // The value of an invalid index.
61  static const uint32_t kDexNoIndex = 0xFFFFFFFF;
62
63  // The value of an invalid index.
64  static const uint16_t kDexNoIndex16 = 0xFFFF;
65
66  // Raw header_item.
67  struct Header {
68    uint8_t magic_[8];
69    uint32_t checksum_;  // See also location_checksum_
70    uint8_t signature_[kSha1DigestSize];
71    uint32_t file_size_;  // size of entire file
72    uint32_t header_size_;  // offset to start of next section
73    uint32_t endian_tag_;
74    uint32_t link_size_;  // unused
75    uint32_t link_off_;  // unused
76    uint32_t map_off_;  // unused
77    uint32_t string_ids_size_;  // number of StringIds
78    uint32_t string_ids_off_;  // file offset of StringIds array
79    uint32_t type_ids_size_;  // number of TypeIds, we don't support more than 65535
80    uint32_t type_ids_off_;  // file offset of TypeIds array
81    uint32_t proto_ids_size_;  // number of ProtoIds, we don't support more than 65535
82    uint32_t proto_ids_off_;  // file offset of ProtoIds array
83    uint32_t field_ids_size_;  // number of FieldIds
84    uint32_t field_ids_off_;  // file offset of FieldIds array
85    uint32_t method_ids_size_;  // number of MethodIds
86    uint32_t method_ids_off_;  // file offset of MethodIds array
87    uint32_t class_defs_size_;  // number of ClassDefs
88    uint32_t class_defs_off_;  // file offset of ClassDef array
89    uint32_t data_size_;  // unused
90    uint32_t data_off_;  // unused
91
92   private:
93    DISALLOW_COPY_AND_ASSIGN(Header);
94  };
95
96  // Map item type codes.
97  enum {
98    kDexTypeHeaderItem               = 0x0000,
99    kDexTypeStringIdItem             = 0x0001,
100    kDexTypeTypeIdItem               = 0x0002,
101    kDexTypeProtoIdItem              = 0x0003,
102    kDexTypeFieldIdItem              = 0x0004,
103    kDexTypeMethodIdItem             = 0x0005,
104    kDexTypeClassDefItem             = 0x0006,
105    kDexTypeMapList                  = 0x1000,
106    kDexTypeTypeList                 = 0x1001,
107    kDexTypeAnnotationSetRefList     = 0x1002,
108    kDexTypeAnnotationSetItem        = 0x1003,
109    kDexTypeClassDataItem            = 0x2000,
110    kDexTypeCodeItem                 = 0x2001,
111    kDexTypeStringDataItem           = 0x2002,
112    kDexTypeDebugInfoItem            = 0x2003,
113    kDexTypeAnnotationItem           = 0x2004,
114    kDexTypeEncodedArrayItem         = 0x2005,
115    kDexTypeAnnotationsDirectoryItem = 0x2006,
116  };
117
118  struct MapItem {
119    uint16_t type_;
120    uint16_t unused_;
121    uint32_t size_;
122    uint32_t offset_;
123
124   private:
125    DISALLOW_COPY_AND_ASSIGN(MapItem);
126  };
127
128  struct MapList {
129    uint32_t size_;
130    MapItem list_[1];
131
132   private:
133    DISALLOW_COPY_AND_ASSIGN(MapList);
134  };
135
136  // Raw string_id_item.
137  struct StringId {
138    uint32_t string_data_off_;  // offset in bytes from the base address
139
140   private:
141    DISALLOW_COPY_AND_ASSIGN(StringId);
142  };
143
144  // Raw type_id_item.
145  struct TypeId {
146    uint32_t descriptor_idx_;  // index into string_ids
147
148   private:
149    DISALLOW_COPY_AND_ASSIGN(TypeId);
150  };
151
152  // Raw field_id_item.
153  struct FieldId {
154    uint16_t class_idx_;  // index into type_ids_ array for defining class
155    uint16_t type_idx_;  // index into type_ids_ array for field type
156    uint32_t name_idx_;  // index into string_ids_ array for field name
157
158   private:
159    DISALLOW_COPY_AND_ASSIGN(FieldId);
160  };
161
162  // Raw method_id_item.
163  struct MethodId {
164    uint16_t class_idx_;  // index into type_ids_ array for defining class
165    uint16_t proto_idx_;  // index into proto_ids_ array for method prototype
166    uint32_t name_idx_;  // index into string_ids_ array for method name
167
168   private:
169    DISALLOW_COPY_AND_ASSIGN(MethodId);
170  };
171
172  // Raw proto_id_item.
173  struct ProtoId {
174    uint32_t shorty_idx_;  // index into string_ids array for shorty descriptor
175    uint16_t return_type_idx_;  // index into type_ids array for return type
176    uint16_t pad_;             // padding = 0
177    uint32_t parameters_off_;  // file offset to type_list for parameter types
178
179   private:
180    DISALLOW_COPY_AND_ASSIGN(ProtoId);
181  };
182
183  // Raw class_def_item.
184  struct ClassDef {
185    uint16_t class_idx_;  // index into type_ids_ array for this class
186    uint16_t pad1_;  // padding = 0
187    uint32_t access_flags_;
188    uint16_t superclass_idx_;  // index into type_ids_ array for superclass
189    uint16_t pad2_;  // padding = 0
190    uint32_t interfaces_off_;  // file offset to TypeList
191    uint32_t source_file_idx_;  // index into string_ids_ for source file name
192    uint32_t annotations_off_;  // file offset to annotations_directory_item
193    uint32_t class_data_off_;  // file offset to class_data_item
194    uint32_t static_values_off_;  // file offset to EncodedArray
195
196   private:
197    DISALLOW_COPY_AND_ASSIGN(ClassDef);
198  };
199
200  // Raw type_item.
201  struct TypeItem {
202    uint16_t type_idx_;  // index into type_ids section
203
204   private:
205    DISALLOW_COPY_AND_ASSIGN(TypeItem);
206  };
207
208  // Raw type_list.
209  class TypeList {
210   public:
211    uint32_t Size() const {
212      return size_;
213    }
214
215    const TypeItem& GetTypeItem(uint32_t idx) const {
216      DCHECK_LT(idx, this->size_);
217      return this->list_[idx];
218    }
219
220   private:
221    uint32_t size_;  // size of the list, in entries
222    TypeItem list_[1];  // elements of the list
223    DISALLOW_COPY_AND_ASSIGN(TypeList);
224  };
225
226  // Raw code_item.
227  struct CodeItem {
228    uint16_t registers_size_;
229    uint16_t ins_size_;
230    uint16_t outs_size_;
231    uint16_t tries_size_;
232    uint32_t debug_info_off_;  // file offset to debug info stream
233    uint32_t insns_size_in_code_units_;  // size of the insns array, in 2 byte code units
234    uint16_t insns_[1];
235
236   private:
237    DISALLOW_COPY_AND_ASSIGN(CodeItem);
238  };
239
240  // Raw try_item.
241  struct TryItem {
242    uint32_t start_addr_;
243    uint16_t insn_count_;
244    uint16_t handler_off_;
245
246   private:
247    DISALLOW_COPY_AND_ASSIGN(TryItem);
248  };
249
250  // Annotation constants.
251  enum {
252    kDexVisibilityBuild         = 0x00,     /* annotation visibility */
253    kDexVisibilityRuntime       = 0x01,
254    kDexVisibilitySystem        = 0x02,
255
256    kDexAnnotationByte          = 0x00,
257    kDexAnnotationShort         = 0x02,
258    kDexAnnotationChar          = 0x03,
259    kDexAnnotationInt           = 0x04,
260    kDexAnnotationLong          = 0x06,
261    kDexAnnotationFloat         = 0x10,
262    kDexAnnotationDouble        = 0x11,
263    kDexAnnotationString        = 0x17,
264    kDexAnnotationType          = 0x18,
265    kDexAnnotationField         = 0x19,
266    kDexAnnotationMethod        = 0x1a,
267    kDexAnnotationEnum          = 0x1b,
268    kDexAnnotationArray         = 0x1c,
269    kDexAnnotationAnnotation    = 0x1d,
270    kDexAnnotationNull          = 0x1e,
271    kDexAnnotationBoolean       = 0x1f,
272
273    kDexAnnotationValueTypeMask = 0x1f,     /* low 5 bits */
274    kDexAnnotationValueArgShift = 5,
275  };
276
277  struct AnnotationsDirectoryItem {
278    uint32_t class_annotations_off_;
279    uint32_t fields_size_;
280    uint32_t methods_size_;
281    uint32_t parameters_size_;
282
283   private:
284    DISALLOW_COPY_AND_ASSIGN(AnnotationsDirectoryItem);
285  };
286
287  struct FieldAnnotationsItem {
288    uint32_t field_idx_;
289    uint32_t annotations_off_;
290
291   private:
292    DISALLOW_COPY_AND_ASSIGN(FieldAnnotationsItem);
293  };
294
295  struct MethodAnnotationsItem {
296    uint32_t method_idx_;
297    uint32_t annotations_off_;
298
299   private:
300    DISALLOW_COPY_AND_ASSIGN(MethodAnnotationsItem);
301  };
302
303  struct ParameterAnnotationsItem {
304    uint32_t method_idx_;
305    uint32_t annotations_off_;
306
307   private:
308    DISALLOW_COPY_AND_ASSIGN(ParameterAnnotationsItem);
309  };
310
311  struct AnnotationSetRefItem {
312    uint32_t annotations_off_;
313
314   private:
315    DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefItem);
316  };
317
318  struct AnnotationSetRefList {
319    uint32_t size_;
320    AnnotationSetRefItem list_[1];
321
322   private:
323    DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefList);
324  };
325
326  struct AnnotationSetItem {
327    uint32_t size_;
328    uint32_t entries_[1];
329
330   private:
331    DISALLOW_COPY_AND_ASSIGN(AnnotationSetItem);
332  };
333
334  struct AnnotationItem {
335    uint8_t visibility_;
336    uint8_t annotation_[1];
337
338   private:
339    DISALLOW_COPY_AND_ASSIGN(AnnotationItem);
340  };
341
342  typedef std::pair<const DexFile*, const DexFile::ClassDef*> ClassPathEntry;
343  typedef std::vector<const DexFile*> ClassPath;
344
345  // Search a collection of DexFiles for a descriptor
346  static ClassPathEntry FindInClassPath(const char* descriptor,
347                                        const ClassPath& class_path);
348
349  // Returns the checksum of a file for comparison with GetLocationChecksum().
350  // For .dex files, this is the header checksum.
351  // For zip files, this is the classes.dex zip entry CRC32 checksum.
352  // Return true if the checksum could be found, false otherwise.
353  static bool GetChecksum(const char* filename, uint32_t* checksum, std::string* error_msg);
354
355  // Opens .dex file, guessing the container format based on file extension
356  static const DexFile* Open(const char* filename, const char* location, std::string* error_msg);
357
358  // Opens .dex file, backed by existing memory
359  static const DexFile* Open(const uint8_t* base, size_t size,
360                             const std::string& location,
361                             uint32_t location_checksum,
362                             std::string* error_msg) {
363    return OpenMemory(base, size, location, location_checksum, NULL, error_msg);
364  }
365
366  // Opens .dex file from the classes.dex in a zip archive
367  static const DexFile* Open(const ZipArchive& zip_archive, const std::string& location,
368                             std::string* error_msg);
369
370  // Closes a .dex file.
371  virtual ~DexFile();
372
373  const std::string& GetLocation() const {
374    return location_;
375  }
376
377  // For DexFiles directly from .dex files, this is the checksum from the DexFile::Header.
378  // For DexFiles opened from a zip files, this will be the ZipEntry CRC32 of classes.dex.
379  uint32_t GetLocationChecksum() const {
380    return location_checksum_;
381  }
382
383  const Header& GetHeader() const {
384    DCHECK(header_ != NULL) << GetLocation();
385    return *header_;
386  }
387
388  // Decode the dex magic version
389  uint32_t GetVersion() const;
390
391  // Returns true if the byte string points to the magic value.
392  static bool IsMagicValid(const byte* magic);
393
394  // Returns true if the byte string after the magic is the correct value.
395  static bool IsVersionValid(const byte* magic);
396
397  // Returns the number of string identifiers in the .dex file.
398  size_t NumStringIds() const {
399    DCHECK(header_ != NULL) << GetLocation();
400    return header_->string_ids_size_;
401  }
402
403  // Returns the StringId at the specified index.
404  const StringId& GetStringId(uint32_t idx) const {
405    DCHECK_LT(idx, NumStringIds()) << GetLocation();
406    return string_ids_[idx];
407  }
408
409  uint32_t GetIndexForStringId(const StringId& string_id) const {
410    CHECK_GE(&string_id, string_ids_) << GetLocation();
411    CHECK_LT(&string_id, string_ids_ + header_->string_ids_size_) << GetLocation();
412    return &string_id - string_ids_;
413  }
414
415  int32_t GetStringLength(const StringId& string_id) const;
416
417  // Returns a pointer to the UTF-8 string data referred to by the given string_id as well as the
418  // length of the string when decoded as a UTF-16 string. Note the UTF-16 length is not the same
419  // as the string length of the string data.
420  const char* GetStringDataAndUtf16Length(const StringId& string_id, uint32_t* utf16_length) const;
421
422  const char* GetStringData(const StringId& string_id) const {
423    uint32_t ignored;
424    return GetStringDataAndUtf16Length(string_id, &ignored);
425  }
426
427  // Index version of GetStringDataAndUtf16Length.
428  const char* StringDataAndUtf16LengthByIdx(uint32_t idx, uint32_t* utf16_length) const {
429    if (idx == kDexNoIndex) {
430      *utf16_length = 0;
431      return NULL;
432    }
433    const StringId& string_id = GetStringId(idx);
434    return GetStringDataAndUtf16Length(string_id, utf16_length);
435  }
436
437  const char* StringDataByIdx(uint32_t idx) const {
438    uint32_t unicode_length;
439    return StringDataAndUtf16LengthByIdx(idx, &unicode_length);
440  }
441
442  // Looks up a string id for a given modified utf8 string.
443  const StringId* FindStringId(const char* string) const;
444
445  // Looks up a string id for a given utf16 string.
446  const StringId* FindStringId(const uint16_t* string) const;
447
448  // Returns the number of type identifiers in the .dex file.
449  size_t NumTypeIds() const {
450    DCHECK(header_ != NULL) << GetLocation();
451    return header_->type_ids_size_;
452  }
453
454  // Returns the TypeId at the specified index.
455  const TypeId& GetTypeId(uint32_t idx) const {
456    DCHECK_LT(idx, NumTypeIds()) << GetLocation();
457    return type_ids_[idx];
458  }
459
460  uint16_t GetIndexForTypeId(const TypeId& type_id) const {
461    CHECK_GE(&type_id, type_ids_) << GetLocation();
462    CHECK_LT(&type_id, type_ids_ + header_->type_ids_size_) << GetLocation();
463    size_t result = &type_id - type_ids_;
464    DCHECK_LT(result, 65536U) << GetLocation();
465    return static_cast<uint16_t>(result);
466  }
467
468  // Get the descriptor string associated with a given type index.
469  const char* StringByTypeIdx(uint32_t idx, uint32_t* unicode_length) const {
470    const TypeId& type_id = GetTypeId(idx);
471    return StringDataAndUtf16LengthByIdx(type_id.descriptor_idx_, unicode_length);
472  }
473
474  const char* StringByTypeIdx(uint32_t idx) const {
475    const TypeId& type_id = GetTypeId(idx);
476    return StringDataByIdx(type_id.descriptor_idx_);
477  }
478
479  // Returns the type descriptor string of a type id.
480  const char* GetTypeDescriptor(const TypeId& type_id) const {
481    return StringDataByIdx(type_id.descriptor_idx_);
482  }
483
484  // Looks up a type for the given string index
485  const TypeId* FindTypeId(uint32_t string_idx) const;
486
487  // Returns the number of field identifiers in the .dex file.
488  size_t NumFieldIds() const {
489    DCHECK(header_ != NULL) << GetLocation();
490    return header_->field_ids_size_;
491  }
492
493  // Returns the FieldId at the specified index.
494  const FieldId& GetFieldId(uint32_t idx) const {
495    DCHECK_LT(idx, NumFieldIds()) << GetLocation();
496    return field_ids_[idx];
497  }
498
499  uint32_t GetIndexForFieldId(const FieldId& field_id) const {
500    CHECK_GE(&field_id, field_ids_) << GetLocation();
501    CHECK_LT(&field_id, field_ids_ + header_->field_ids_size_) << GetLocation();
502    return &field_id - field_ids_;
503  }
504
505  // Looks up a field by its declaring class, name and type
506  const FieldId* FindFieldId(const DexFile::TypeId& declaring_klass,
507                             const DexFile::StringId& name,
508                             const DexFile::TypeId& type) const;
509
510  // Returns the declaring class descriptor string of a field id.
511  const char* GetFieldDeclaringClassDescriptor(const FieldId& field_id) const {
512    const DexFile::TypeId& type_id = GetTypeId(field_id.class_idx_);
513    return GetTypeDescriptor(type_id);
514  }
515
516  // Returns the class descriptor string of a field id.
517  const char* GetFieldTypeDescriptor(const FieldId& field_id) const {
518    const DexFile::TypeId& type_id = GetTypeId(field_id.type_idx_);
519    return GetTypeDescriptor(type_id);
520  }
521
522  // Returns the name of a field id.
523  const char* GetFieldName(const FieldId& field_id) const {
524    return StringDataByIdx(field_id.name_idx_);
525  }
526
527  // Returns the number of method identifiers in the .dex file.
528  size_t NumMethodIds() const {
529    DCHECK(header_ != NULL) << GetLocation();
530    return header_->method_ids_size_;
531  }
532
533  // Returns the MethodId at the specified index.
534  const MethodId& GetMethodId(uint32_t idx) const {
535    DCHECK_LT(idx, NumMethodIds()) << GetLocation();
536    return method_ids_[idx];
537  }
538
539  uint32_t GetIndexForMethodId(const MethodId& method_id) const {
540    CHECK_GE(&method_id, method_ids_) << GetLocation();
541    CHECK_LT(&method_id, method_ids_ + header_->method_ids_size_) << GetLocation();
542    return &method_id - method_ids_;
543  }
544
545  // Looks up a method by its declaring class, name and proto_id
546  const MethodId* FindMethodId(const DexFile::TypeId& declaring_klass,
547                               const DexFile::StringId& name,
548                               const DexFile::ProtoId& signature) const;
549
550  // Returns the declaring class descriptor string of a method id.
551  const char* GetMethodDeclaringClassDescriptor(const MethodId& method_id) const {
552    const DexFile::TypeId& type_id = GetTypeId(method_id.class_idx_);
553    return GetTypeDescriptor(type_id);
554  }
555
556  // Returns the prototype of a method id.
557  const ProtoId& GetMethodPrototype(const MethodId& method_id) const {
558    return GetProtoId(method_id.proto_idx_);
559  }
560
561  // Returns a representation of the signature of a method id.
562  const Signature GetMethodSignature(const MethodId& method_id) const;
563
564  // Returns the name of a method id.
565  const char* GetMethodName(const MethodId& method_id) const {
566    return StringDataByIdx(method_id.name_idx_);
567  }
568
569  // Returns the shorty of a method id.
570  const char* GetMethodShorty(const MethodId& method_id) const {
571    return StringDataByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_);
572  }
573  const char* GetMethodShorty(const MethodId& method_id, uint32_t* length) const {
574    // Using the UTF16 length is safe here as shorties are guaranteed to be ASCII characters.
575    return StringDataAndUtf16LengthByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_, length);
576  }
577  // Returns the number of class definitions in the .dex file.
578  size_t NumClassDefs() const {
579    DCHECK(header_ != NULL) << GetLocation();
580    return header_->class_defs_size_;
581  }
582
583  // Returns the ClassDef at the specified index.
584  const ClassDef& GetClassDef(uint16_t idx) const {
585    DCHECK_LT(idx, NumClassDefs()) << GetLocation();
586    return class_defs_[idx];
587  }
588
589  uint16_t GetIndexForClassDef(const ClassDef& class_def) const {
590    CHECK_GE(&class_def, class_defs_) << GetLocation();
591    CHECK_LT(&class_def, class_defs_ + header_->class_defs_size_) << GetLocation();
592    return &class_def - class_defs_;
593  }
594
595  // Returns the class descriptor string of a class definition.
596  const char* GetClassDescriptor(const ClassDef& class_def) const {
597    return StringByTypeIdx(class_def.class_idx_);
598  }
599
600  // Looks up a class definition by its class descriptor.
601  const ClassDef* FindClassDef(const char* descriptor) const;
602
603  // Looks up a class definition by its type index.
604  const ClassDef* FindClassDef(uint16_t type_idx) const;
605
606  const TypeList* GetInterfacesList(const ClassDef& class_def) const {
607    if (class_def.interfaces_off_ == 0) {
608        return NULL;
609    } else {
610      const byte* addr = begin_ + class_def.interfaces_off_;
611      return reinterpret_cast<const TypeList*>(addr);
612    }
613  }
614
615  // Returns a pointer to the raw memory mapped class_data_item
616  const byte* GetClassData(const ClassDef& class_def) const {
617    if (class_def.class_data_off_ == 0) {
618      return NULL;
619    } else {
620      return begin_ + class_def.class_data_off_;
621    }
622  }
623
624  //
625  const CodeItem* GetCodeItem(const uint32_t code_off) const {
626    if (code_off == 0) {
627      return NULL;  // native or abstract method
628    } else {
629      const byte* addr = begin_ + code_off;
630      return reinterpret_cast<const CodeItem*>(addr);
631    }
632  }
633
634  const char* GetReturnTypeDescriptor(const ProtoId& proto_id) const {
635    return StringByTypeIdx(proto_id.return_type_idx_);
636  }
637
638  // Returns the number of prototype identifiers in the .dex file.
639  size_t NumProtoIds() const {
640    DCHECK(header_ != NULL) << GetLocation();
641    return header_->proto_ids_size_;
642  }
643
644  // Returns the ProtoId at the specified index.
645  const ProtoId& GetProtoId(uint32_t idx) const {
646    DCHECK_LT(idx, NumProtoIds()) << GetLocation();
647    return proto_ids_[idx];
648  }
649
650  uint16_t GetIndexForProtoId(const ProtoId& proto_id) const {
651    CHECK_GE(&proto_id, proto_ids_) << GetLocation();
652    CHECK_LT(&proto_id, proto_ids_ + header_->proto_ids_size_) << GetLocation();
653    return &proto_id - proto_ids_;
654  }
655
656  // Looks up a proto id for a given return type and signature type list
657  const ProtoId* FindProtoId(uint16_t return_type_idx,
658                             const uint16_t* signature_type_idxs, uint32_t signature_length) const;
659  const ProtoId* FindProtoId(uint16_t return_type_idx,
660                             const std::vector<uint16_t>& signature_type_idxs) const {
661    return FindProtoId(return_type_idx, &signature_type_idxs[0], signature_type_idxs.size());
662  }
663
664  // Given a signature place the type ids into the given vector, returns true on success
665  bool CreateTypeList(const StringPiece& signature, uint16_t* return_type_idx,
666                      std::vector<uint16_t>* param_type_idxs) const;
667
668  // Create a Signature from the given string signature or return Signature::NoSignature if not
669  // possible.
670  const Signature CreateSignature(const StringPiece& signature) const;
671
672  // Returns the short form method descriptor for the given prototype.
673  const char* GetShorty(uint32_t proto_idx) const {
674    const ProtoId& proto_id = GetProtoId(proto_idx);
675    return StringDataByIdx(proto_id.shorty_idx_);
676  }
677
678  const TypeList* GetProtoParameters(const ProtoId& proto_id) const {
679    if (proto_id.parameters_off_ == 0) {
680      return NULL;
681    } else {
682      const byte* addr = begin_ + proto_id.parameters_off_;
683      return reinterpret_cast<const TypeList*>(addr);
684    }
685  }
686
687  const byte* GetEncodedStaticFieldValuesArray(const ClassDef& class_def) const {
688    if (class_def.static_values_off_ == 0) {
689      return 0;
690    } else {
691      return begin_ + class_def.static_values_off_;
692    }
693  }
694
695  static const TryItem* GetTryItems(const CodeItem& code_item, uint32_t offset);
696
697  // Get the base of the encoded data for the given DexCode.
698  static const byte* GetCatchHandlerData(const CodeItem& code_item, uint32_t offset) {
699    const byte* handler_data =
700        reinterpret_cast<const byte*>(GetTryItems(code_item, code_item.tries_size_));
701    return handler_data + offset;
702  }
703
704  // Find which try region is associated with the given address (ie dex pc). Returns -1 if none.
705  static int32_t FindTryItem(const CodeItem &code_item, uint32_t address);
706
707  // Find the handler offset associated with the given address (ie dex pc). Returns -1 if none.
708  static int32_t FindCatchHandlerOffset(const CodeItem &code_item, uint32_t address);
709
710  // Get the pointer to the start of the debugging data
711  const byte* GetDebugInfoStream(const CodeItem* code_item) const {
712    if (code_item->debug_info_off_ == 0) {
713      return NULL;
714    } else {
715      return begin_ + code_item->debug_info_off_;
716    }
717  }
718
719  // Callback for "new position table entry".
720  // Returning true causes the decoder to stop early.
721  typedef bool (*DexDebugNewPositionCb)(void* context, uint32_t address, uint32_t line_num);
722
723  // Callback for "new locals table entry". "signature" is an empty string
724  // if no signature is available for an entry.
725  typedef void (*DexDebugNewLocalCb)(void* context, uint16_t reg,
726                                     uint32_t start_address,
727                                     uint32_t end_address,
728                                     const char* name,
729                                     const char* descriptor,
730                                     const char* signature);
731
732  static bool LineNumForPcCb(void* context, uint32_t address, uint32_t line_num);
733
734  // Debug info opcodes and constants
735  enum {
736    DBG_END_SEQUENCE         = 0x00,
737    DBG_ADVANCE_PC           = 0x01,
738    DBG_ADVANCE_LINE         = 0x02,
739    DBG_START_LOCAL          = 0x03,
740    DBG_START_LOCAL_EXTENDED = 0x04,
741    DBG_END_LOCAL            = 0x05,
742    DBG_RESTART_LOCAL        = 0x06,
743    DBG_SET_PROLOGUE_END     = 0x07,
744    DBG_SET_EPILOGUE_BEGIN   = 0x08,
745    DBG_SET_FILE             = 0x09,
746    DBG_FIRST_SPECIAL        = 0x0a,
747    DBG_LINE_BASE            = -4,
748    DBG_LINE_RANGE           = 15,
749  };
750
751  struct LocalInfo {
752    LocalInfo()
753        : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {}
754
755    const char* name_;  // E.g., list
756    const char* descriptor_;  // E.g., Ljava/util/LinkedList;
757    const char* signature_;  // E.g., java.util.LinkedList<java.lang.Integer>
758    uint16_t start_address_;  // PC location where the local is first defined.
759    bool is_live_;  // Is the local defined and live.
760
761   private:
762    DISALLOW_COPY_AND_ASSIGN(LocalInfo);
763  };
764
765  struct LineNumFromPcContext {
766    LineNumFromPcContext(uint32_t address, uint32_t line_num)
767        : address_(address), line_num_(line_num) {}
768    uint32_t address_;
769    uint32_t line_num_;
770   private:
771    DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext);
772  };
773
774  void InvokeLocalCbIfLive(void* context, int reg, uint32_t end_address,
775                           LocalInfo* local_in_reg, DexDebugNewLocalCb local_cb) const {
776    if (local_cb != NULL && local_in_reg[reg].is_live_) {
777      local_cb(context, reg, local_in_reg[reg].start_address_, end_address,
778          local_in_reg[reg].name_, local_in_reg[reg].descriptor_,
779          local_in_reg[reg].signature_ != NULL ? local_in_reg[reg].signature_ : "");
780    }
781  }
782
783  // Determine the source file line number based on the program counter.
784  // "pc" is an offset, in 16-bit units, from the start of the method's code.
785  //
786  // Returns -1 if no match was found (possibly because the source files were
787  // compiled without "-g", so no line number information is present).
788  // Returns -2 for native methods (as expected in exception traces).
789  //
790  // This is used by runtime; therefore use art::Method not art::DexFile::Method.
791  int32_t GetLineNumFromPC(mirror::ArtMethod* method, uint32_t rel_pc) const
792      SHARED_LOCKS_REQUIRED(Locks::mutator_lock_);
793
794  void DecodeDebugInfo(const CodeItem* code_item, bool is_static, uint32_t method_idx,
795                       DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb,
796                       void* context) const;
797
798  const char* GetSourceFile(const ClassDef& class_def) const {
799    if (class_def.source_file_idx_ == 0xffffffff) {
800      return NULL;
801    } else {
802      return StringDataByIdx(class_def.source_file_idx_);
803    }
804  }
805
806  int GetPermissions() const;
807
808  bool IsReadOnly() const;
809
810  bool EnableWrite() const;
811
812  bool DisableWrite() const;
813
814  const byte* Begin() const {
815    return begin_;
816  }
817
818  size_t Size() const {
819    return size_;
820  }
821
822 private:
823  // Opens a .dex file
824  static const DexFile* OpenFile(int fd, const char* location, bool verify, std::string* error_msg);
825
826  // Opens a dex file from within a .jar, .zip, or .apk file
827  static const DexFile* OpenZip(int fd, const std::string& location, std::string* error_msg);
828
829  // Opens a .dex file at the given address backed by a MemMap
830  static const DexFile* OpenMemory(const std::string& location,
831                                   uint32_t location_checksum,
832                                   MemMap* mem_map,
833                                   std::string* error_msg);
834
835  // Opens a .dex file at the given address, optionally backed by a MemMap
836  static const DexFile* OpenMemory(const byte* dex_file,
837                                   size_t size,
838                                   const std::string& location,
839                                   uint32_t location_checksum,
840                                   MemMap* mem_map,
841                                   std::string* error_msg);
842
843  DexFile(const byte* base, size_t size,
844          const std::string& location,
845          uint32_t location_checksum,
846          MemMap* mem_map);
847
848  // Top-level initializer that calls other Init methods.
849  bool Init(std::string* error_msg);
850
851  // Returns true if the header magic and version numbers are of the expected values.
852  bool CheckMagicAndVersion(std::string* error_msg) const;
853
854  void DecodeDebugInfo0(const CodeItem* code_item, bool is_static, uint32_t method_idx,
855      DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb,
856      void* context, const byte* stream, LocalInfo* local_in_reg) const;
857
858  // The base address of the memory mapping.
859  const byte* const begin_;
860
861  // The size of the underlying memory allocation in bytes.
862  const size_t size_;
863
864  // Typically the dex file name when available, alternatively some identifying string.
865  //
866  // The ClassLinker will use this to match DexFiles the boot class
867  // path to DexCache::GetLocation when loading from an image.
868  const std::string location_;
869
870  const uint32_t location_checksum_;
871
872  // Manages the underlying memory allocation.
873  std::unique_ptr<MemMap> mem_map_;
874
875  // Points to the header section.
876  const Header* const header_;
877
878  // Points to the base of the string identifier list.
879  const StringId* const string_ids_;
880
881  // Points to the base of the type identifier list.
882  const TypeId* const type_ids_;
883
884  // Points to the base of the field identifier list.
885  const FieldId* const field_ids_;
886
887  // Points to the base of the method identifier list.
888  const MethodId* const method_ids_;
889
890  // Points to the base of the prototype identifier list.
891  const ProtoId* const proto_ids_;
892
893  // Points to the base of the class definition list.
894  const ClassDef* const class_defs_;
895};
896std::ostream& operator<<(std::ostream& os, const DexFile& dex_file);
897
898// Iterate over a dex file's ProtoId's paramters
899class DexFileParameterIterator {
900 public:
901  DexFileParameterIterator(const DexFile& dex_file, const DexFile::ProtoId& proto_id)
902      : dex_file_(dex_file), size_(0), pos_(0) {
903    type_list_ = dex_file_.GetProtoParameters(proto_id);
904    if (type_list_ != NULL) {
905      size_ = type_list_->Size();
906    }
907  }
908  bool HasNext() const { return pos_ < size_; }
909  void Next() { ++pos_; }
910  uint16_t GetTypeIdx() {
911    return type_list_->GetTypeItem(pos_).type_idx_;
912  }
913  const char* GetDescriptor() {
914    return dex_file_.StringByTypeIdx(GetTypeIdx());
915  }
916 private:
917  const DexFile& dex_file_;
918  const DexFile::TypeList* type_list_;
919  uint32_t size_;
920  uint32_t pos_;
921  DISALLOW_IMPLICIT_CONSTRUCTORS(DexFileParameterIterator);
922};
923
924// Abstract the signature of a method.
925class Signature {
926 public:
927  std::string ToString() const;
928
929  static Signature NoSignature() {
930    return Signature();
931  }
932
933  bool operator==(const Signature& rhs) const;
934  bool operator!=(const Signature& rhs) const {
935    return !(*this == rhs);
936  }
937
938  bool operator==(const StringPiece& rhs) const;
939
940 private:
941  Signature(const DexFile* dex, const DexFile::ProtoId& proto) : dex_file_(dex), proto_id_(&proto) {
942  }
943
944  Signature() : dex_file_(nullptr), proto_id_(nullptr) {
945  }
946
947  friend class DexFile;
948
949  const DexFile* const dex_file_;
950  const DexFile::ProtoId* const proto_id_;
951};
952std::ostream& operator<<(std::ostream& os, const Signature& sig);
953
954// Iterate and decode class_data_item
955class ClassDataItemIterator {
956 public:
957  ClassDataItemIterator(const DexFile& dex_file, const byte* raw_class_data_item)
958      : dex_file_(dex_file), pos_(0), ptr_pos_(raw_class_data_item), last_idx_(0) {
959    ReadClassDataHeader();
960    if (EndOfInstanceFieldsPos() > 0) {
961      ReadClassDataField();
962    } else if (EndOfVirtualMethodsPos() > 0) {
963      ReadClassDataMethod();
964    }
965  }
966  uint32_t NumStaticFields() const {
967    return header_.static_fields_size_;
968  }
969  uint32_t NumInstanceFields() const {
970    return header_.instance_fields_size_;
971  }
972  uint32_t NumDirectMethods() const {
973    return header_.direct_methods_size_;
974  }
975  uint32_t NumVirtualMethods() const {
976    return header_.virtual_methods_size_;
977  }
978  bool HasNextStaticField() const {
979    return pos_ < EndOfStaticFieldsPos();
980  }
981  bool HasNextInstanceField() const {
982    return pos_ >= EndOfStaticFieldsPos() && pos_ < EndOfInstanceFieldsPos();
983  }
984  bool HasNextDirectMethod() const {
985    return pos_ >= EndOfInstanceFieldsPos() && pos_ < EndOfDirectMethodsPos();
986  }
987  bool HasNextVirtualMethod() const {
988    return pos_ >= EndOfDirectMethodsPos() && pos_ < EndOfVirtualMethodsPos();
989  }
990  bool HasNext() const {
991    return pos_ < EndOfVirtualMethodsPos();
992  }
993  inline void Next() {
994    pos_++;
995    if (pos_ < EndOfStaticFieldsPos()) {
996      last_idx_ = GetMemberIndex();
997      ReadClassDataField();
998    } else if (pos_ == EndOfStaticFieldsPos() && NumInstanceFields() > 0) {
999      last_idx_ = 0;  // transition to next array, reset last index
1000      ReadClassDataField();
1001    } else if (pos_ < EndOfInstanceFieldsPos()) {
1002      last_idx_ = GetMemberIndex();
1003      ReadClassDataField();
1004    } else if (pos_ == EndOfInstanceFieldsPos() && NumDirectMethods() > 0) {
1005      last_idx_ = 0;  // transition to next array, reset last index
1006      ReadClassDataMethod();
1007    } else if (pos_ < EndOfDirectMethodsPos()) {
1008      last_idx_ = GetMemberIndex();
1009      ReadClassDataMethod();
1010    } else if (pos_ == EndOfDirectMethodsPos() && NumVirtualMethods() > 0) {
1011      last_idx_ = 0;  // transition to next array, reset last index
1012      ReadClassDataMethod();
1013    } else if (pos_ < EndOfVirtualMethodsPos()) {
1014      last_idx_ = GetMemberIndex();
1015      ReadClassDataMethod();
1016    } else {
1017      DCHECK(!HasNext());
1018    }
1019  }
1020  uint32_t GetMemberIndex() const {
1021    if (pos_ < EndOfInstanceFieldsPos()) {
1022      return last_idx_ + field_.field_idx_delta_;
1023    } else {
1024      DCHECK_LT(pos_, EndOfVirtualMethodsPos());
1025      return last_idx_ + method_.method_idx_delta_;
1026    }
1027  }
1028  uint32_t GetMemberAccessFlags() const {
1029    if (pos_ < EndOfInstanceFieldsPos()) {
1030      return field_.access_flags_;
1031    } else {
1032      DCHECK_LT(pos_, EndOfVirtualMethodsPos());
1033      return method_.access_flags_;
1034    }
1035  }
1036  InvokeType GetMethodInvokeType(const DexFile::ClassDef& class_def) const {
1037    if (HasNextDirectMethod()) {
1038      if ((GetMemberAccessFlags() & kAccStatic) != 0) {
1039        return kStatic;
1040      } else {
1041        return kDirect;
1042      }
1043    } else {
1044      DCHECK_EQ(GetMemberAccessFlags() & kAccStatic, 0U);
1045      if ((class_def.access_flags_ & kAccInterface) != 0) {
1046        return kInterface;
1047      } else if ((GetMemberAccessFlags() & kAccConstructor) != 0) {
1048        return kSuper;
1049      } else {
1050        return kVirtual;
1051      }
1052    }
1053  }
1054  const DexFile::CodeItem* GetMethodCodeItem() const {
1055    return dex_file_.GetCodeItem(method_.code_off_);
1056  }
1057  uint32_t GetMethodCodeItemOffset() const {
1058    return method_.code_off_;
1059  }
1060  const byte* EndDataPointer() const {
1061    CHECK(!HasNext());
1062    return ptr_pos_;
1063  }
1064
1065 private:
1066  // A dex file's class_data_item is leb128 encoded, this structure holds a decoded form of the
1067  // header for a class_data_item
1068  struct ClassDataHeader {
1069    uint32_t static_fields_size_;  // the number of static fields
1070    uint32_t instance_fields_size_;  // the number of instance fields
1071    uint32_t direct_methods_size_;  // the number of direct methods
1072    uint32_t virtual_methods_size_;  // the number of virtual methods
1073  } header_;
1074
1075  // Read and decode header from a class_data_item stream into header
1076  void ReadClassDataHeader();
1077
1078  uint32_t EndOfStaticFieldsPos() const {
1079    return header_.static_fields_size_;
1080  }
1081  uint32_t EndOfInstanceFieldsPos() const {
1082    return EndOfStaticFieldsPos() + header_.instance_fields_size_;
1083  }
1084  uint32_t EndOfDirectMethodsPos() const {
1085    return EndOfInstanceFieldsPos() + header_.direct_methods_size_;
1086  }
1087  uint32_t EndOfVirtualMethodsPos() const {
1088    return EndOfDirectMethodsPos() + header_.virtual_methods_size_;
1089  }
1090
1091  // A decoded version of the field of a class_data_item
1092  struct ClassDataField {
1093    uint32_t field_idx_delta_;  // delta of index into the field_ids array for FieldId
1094    uint32_t access_flags_;  // access flags for the field
1095    ClassDataField() :  field_idx_delta_(0), access_flags_(0) {}
1096
1097   private:
1098    DISALLOW_COPY_AND_ASSIGN(ClassDataField);
1099  };
1100  ClassDataField field_;
1101
1102  // Read and decode a field from a class_data_item stream into field
1103  void ReadClassDataField();
1104
1105  // A decoded version of the method of a class_data_item
1106  struct ClassDataMethod {
1107    uint32_t method_idx_delta_;  // delta of index into the method_ids array for MethodId
1108    uint32_t access_flags_;
1109    uint32_t code_off_;
1110    ClassDataMethod() : method_idx_delta_(0), access_flags_(0), code_off_(0) {}
1111
1112   private:
1113    DISALLOW_COPY_AND_ASSIGN(ClassDataMethod);
1114  };
1115  ClassDataMethod method_;
1116
1117  // Read and decode a method from a class_data_item stream into method
1118  void ReadClassDataMethod();
1119
1120  const DexFile& dex_file_;
1121  size_t pos_;  // integral number of items passed
1122  const byte* ptr_pos_;  // pointer into stream of class_data_item
1123  uint32_t last_idx_;  // last read field or method index to apply delta to
1124  DISALLOW_IMPLICIT_CONSTRUCTORS(ClassDataItemIterator);
1125};
1126
1127class EncodedStaticFieldValueIterator {
1128 public:
1129  EncodedStaticFieldValueIterator(const DexFile& dex_file, Handle<mirror::DexCache>* dex_cache,
1130                                  Handle<mirror::ClassLoader>* class_loader,
1131                                  ClassLinker* linker, const DexFile::ClassDef& class_def)
1132      SHARED_LOCKS_REQUIRED(Locks::mutator_lock_);
1133
1134  template<bool kTransactionActive>
1135  void ReadValueToField(mirror::ArtField* field) const SHARED_LOCKS_REQUIRED(Locks::mutator_lock_);
1136
1137  bool HasNext() { return pos_ < array_size_; }
1138
1139  void Next();
1140
1141  enum ValueType {
1142    kByte = 0x00,
1143    kShort = 0x02,
1144    kChar = 0x03,
1145    kInt = 0x04,
1146    kLong = 0x06,
1147    kFloat = 0x10,
1148    kDouble = 0x11,
1149    kString = 0x17,
1150    kType = 0x18,
1151    kField = 0x19,
1152    kMethod = 0x1a,
1153    kEnum = 0x1b,
1154    kArray = 0x1c,
1155    kAnnotation = 0x1d,
1156    kNull = 0x1e,
1157    kBoolean = 0x1f
1158  };
1159
1160 private:
1161  static const byte kEncodedValueTypeMask = 0x1f;  // 0b11111
1162  static const byte kEncodedValueArgShift = 5;
1163
1164  const DexFile& dex_file_;
1165  Handle<mirror::DexCache>* const dex_cache_;  // Dex cache to resolve literal objects.
1166  Handle<mirror::ClassLoader>* const class_loader_;  // ClassLoader to resolve types.
1167  ClassLinker* linker_;  // Linker to resolve literal objects.
1168  size_t array_size_;  // Size of array.
1169  size_t pos_;  // Current position.
1170  const byte* ptr_;  // Pointer into encoded data array.
1171  ValueType type_;  // Type of current encoded value.
1172  jvalue jval_;  // Value of current encoded value.
1173  DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedStaticFieldValueIterator);
1174};
1175std::ostream& operator<<(std::ostream& os, const EncodedStaticFieldValueIterator::ValueType& code);
1176
1177class CatchHandlerIterator {
1178  public:
1179    CatchHandlerIterator(const DexFile::CodeItem& code_item, uint32_t address);
1180
1181    CatchHandlerIterator(const DexFile::CodeItem& code_item,
1182                         const DexFile::TryItem& try_item);
1183
1184    explicit CatchHandlerIterator(const byte* handler_data) {
1185      Init(handler_data);
1186    }
1187
1188    uint16_t GetHandlerTypeIndex() const {
1189      return handler_.type_idx_;
1190    }
1191    uint32_t GetHandlerAddress() const {
1192      return handler_.address_;
1193    }
1194    void Next();
1195    bool HasNext() const {
1196      return remaining_count_ != -1 || catch_all_;
1197    }
1198    // End of this set of catch blocks, convenience method to locate next set of catch blocks
1199    const byte* EndDataPointer() const {
1200      CHECK(!HasNext());
1201      return current_data_;
1202    }
1203
1204  private:
1205    void Init(const DexFile::CodeItem& code_item, int32_t offset);
1206    void Init(const byte* handler_data);
1207
1208    struct CatchHandlerItem {
1209      uint16_t type_idx_;  // type index of the caught exception type
1210      uint32_t address_;  // handler address
1211    } handler_;
1212    const byte *current_data_;  // the current handler in dex file.
1213    int32_t remaining_count_;   // number of handlers not read.
1214    bool catch_all_;            // is there a handler that will catch all exceptions in case
1215                                // that all typed handler does not match.
1216};
1217
1218}  // namespace art
1219
1220#endif  // ART_RUNTIME_DEX_FILE_H_
1221