• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/value_object.h"
26 #include "dex_file_types.h"
27 #include "globals.h"
28 #include "invoke_type.h"
29 #include "jni.h"
30 #include "modifiers.h"
31 
32 namespace art {
33 
34 class MemMap;
35 class OatDexFile;
36 class Signature;
37 class StringPiece;
38 class ZipArchive;
39 
40 class DexFile {
41  public:
42   // First Dex format version supporting default methods.
43   static const uint32_t kDefaultMethodsVersion = 37;
44   // First Dex format version enforcing class definition ordering rules.
45   static const uint32_t kClassDefinitionOrderEnforcedVersion = 37;
46 
47   static const uint8_t kDexMagic[];
48   static constexpr size_t kNumDexVersions = 3;
49   static constexpr size_t kDexVersionLen = 4;
50   static const uint8_t kDexMagicVersions[kNumDexVersions][kDexVersionLen];
51 
52   static constexpr size_t kSha1DigestSize = 20;
53   static constexpr uint32_t kDexEndianConstant = 0x12345678;
54 
55   // name of the DexFile entry within a zip archive
56   static const char* kClassesDex;
57 
58   // The value of an invalid index.
59   static const uint32_t kDexNoIndex = 0xFFFFFFFF;
60 
61   // The value of an invalid index.
62   static const uint16_t kDexNoIndex16 = 0xFFFF;
63 
64   // The separator character in MultiDex locations.
65   static constexpr char kMultiDexSeparator = '!';
66 
67   // A string version of the previous. This is a define so that we can merge string literals in the
68   // preprocessor.
69   #define kMultiDexSeparatorString "!"
70 
71   // Raw header_item.
72   struct Header {
73     uint8_t magic_[8];
74     uint32_t checksum_;  // See also location_checksum_
75     uint8_t signature_[kSha1DigestSize];
76     uint32_t file_size_;  // size of entire file
77     uint32_t header_size_;  // offset to start of next section
78     uint32_t endian_tag_;
79     uint32_t link_size_;  // unused
80     uint32_t link_off_;  // unused
81     uint32_t map_off_;  // unused
82     uint32_t string_ids_size_;  // number of StringIds
83     uint32_t string_ids_off_;  // file offset of StringIds array
84     uint32_t type_ids_size_;  // number of TypeIds, we don't support more than 65535
85     uint32_t type_ids_off_;  // file offset of TypeIds array
86     uint32_t proto_ids_size_;  // number of ProtoIds, we don't support more than 65535
87     uint32_t proto_ids_off_;  // file offset of ProtoIds array
88     uint32_t field_ids_size_;  // number of FieldIds
89     uint32_t field_ids_off_;  // file offset of FieldIds array
90     uint32_t method_ids_size_;  // number of MethodIds
91     uint32_t method_ids_off_;  // file offset of MethodIds array
92     uint32_t class_defs_size_;  // number of ClassDefs
93     uint32_t class_defs_off_;  // file offset of ClassDef array
94     uint32_t data_size_;  // size of data section
95     uint32_t data_off_;  // file offset of data section
96 
97     // Decode the dex magic version
98     uint32_t GetVersion() const;
99 
100    private:
101     DISALLOW_COPY_AND_ASSIGN(Header);
102   };
103 
104   // Map item type codes.
105   enum MapItemType : uint16_t {  // private
106     kDexTypeHeaderItem               = 0x0000,
107     kDexTypeStringIdItem             = 0x0001,
108     kDexTypeTypeIdItem               = 0x0002,
109     kDexTypeProtoIdItem              = 0x0003,
110     kDexTypeFieldIdItem              = 0x0004,
111     kDexTypeMethodIdItem             = 0x0005,
112     kDexTypeClassDefItem             = 0x0006,
113     kDexTypeCallSiteIdItem           = 0x0007,
114     kDexTypeMethodHandleItem         = 0x0008,
115     kDexTypeMapList                  = 0x1000,
116     kDexTypeTypeList                 = 0x1001,
117     kDexTypeAnnotationSetRefList     = 0x1002,
118     kDexTypeAnnotationSetItem        = 0x1003,
119     kDexTypeClassDataItem            = 0x2000,
120     kDexTypeCodeItem                 = 0x2001,
121     kDexTypeStringDataItem           = 0x2002,
122     kDexTypeDebugInfoItem            = 0x2003,
123     kDexTypeAnnotationItem           = 0x2004,
124     kDexTypeEncodedArrayItem         = 0x2005,
125     kDexTypeAnnotationsDirectoryItem = 0x2006,
126   };
127 
128   struct MapItem {
129     uint16_t type_;
130     uint16_t unused_;
131     uint32_t size_;
132     uint32_t offset_;
133 
134    private:
135     DISALLOW_COPY_AND_ASSIGN(MapItem);
136   };
137 
138   struct MapList {
139     uint32_t size_;
140     MapItem list_[1];
141 
142    private:
143     DISALLOW_COPY_AND_ASSIGN(MapList);
144   };
145 
146   // Raw string_id_item.
147   struct StringId {
148     uint32_t string_data_off_;  // offset in bytes from the base address
149 
150    private:
151     DISALLOW_COPY_AND_ASSIGN(StringId);
152   };
153 
154   // Raw type_id_item.
155   struct TypeId {
156     dex::StringIndex descriptor_idx_;  // index into string_ids
157 
158    private:
159     DISALLOW_COPY_AND_ASSIGN(TypeId);
160   };
161 
162   // Raw field_id_item.
163   struct FieldId {
164     dex::TypeIndex class_idx_;   // index into type_ids_ array for defining class
165     dex::TypeIndex type_idx_;    // index into type_ids_ array for field type
166     dex::StringIndex name_idx_;  // index into string_ids_ array for field name
167 
168    private:
169     DISALLOW_COPY_AND_ASSIGN(FieldId);
170   };
171 
172   // Raw proto_id_item.
173   struct ProtoId {
174     dex::StringIndex shorty_idx_;     // index into string_ids array for shorty descriptor
175     dex::TypeIndex 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 method_id_item.
184   struct MethodId {
185     dex::TypeIndex class_idx_;   // index into type_ids_ array for defining class
186     uint16_t proto_idx_;         // index into proto_ids_ array for method prototype
187     dex::StringIndex name_idx_;  // index into string_ids_ array for method name
188 
189    private:
190     DISALLOW_COPY_AND_ASSIGN(MethodId);
191   };
192 
193   // Raw class_def_item.
194   struct ClassDef {
195     dex::TypeIndex class_idx_;  // index into type_ids_ array for this class
196     uint16_t pad1_;  // padding = 0
197     uint32_t access_flags_;
198     dex::TypeIndex superclass_idx_;  // index into type_ids_ array for superclass
199     uint16_t pad2_;  // padding = 0
200     uint32_t interfaces_off_;  // file offset to TypeList
201     dex::StringIndex source_file_idx_;  // index into string_ids_ for source file name
202     uint32_t annotations_off_;  // file offset to annotations_directory_item
203     uint32_t class_data_off_;  // file offset to class_data_item
204     uint32_t static_values_off_;  // file offset to EncodedArray
205 
206     // Returns the valid access flags, that is, Java modifier bits relevant to the ClassDef type
207     // (class or interface). These are all in the lower 16b and do not contain runtime flags.
GetJavaAccessFlagsClassDef208     uint32_t GetJavaAccessFlags() const {
209       // Make sure that none of our runtime-only flags are set.
210       static_assert((kAccValidClassFlags & kAccJavaFlagsMask) == kAccValidClassFlags,
211                     "Valid class flags not a subset of Java flags");
212       static_assert((kAccValidInterfaceFlags & kAccJavaFlagsMask) == kAccValidInterfaceFlags,
213                     "Valid interface flags not a subset of Java flags");
214 
215       if ((access_flags_ & kAccInterface) != 0) {
216         // Interface.
217         return access_flags_ & kAccValidInterfaceFlags;
218       } else {
219         // Class.
220         return access_flags_ & kAccValidClassFlags;
221       }
222     }
223 
224    private:
225     DISALLOW_COPY_AND_ASSIGN(ClassDef);
226   };
227 
228   // Raw type_item.
229   struct TypeItem {
230     dex::TypeIndex type_idx_;  // index into type_ids section
231 
232    private:
233     DISALLOW_COPY_AND_ASSIGN(TypeItem);
234   };
235 
236   // Raw type_list.
237   class TypeList {
238    public:
Size()239     uint32_t Size() const {
240       return size_;
241     }
242 
GetTypeItem(uint32_t idx)243     const TypeItem& GetTypeItem(uint32_t idx) const {
244       DCHECK_LT(idx, this->size_);
245       return this->list_[idx];
246     }
247 
248     // Size in bytes of the part of the list that is common.
GetHeaderSize()249     static constexpr size_t GetHeaderSize() {
250       return 4U;
251     }
252 
253     // Size in bytes of the whole type list including all the stored elements.
GetListSize(size_t count)254     static constexpr size_t GetListSize(size_t count) {
255       return GetHeaderSize() + sizeof(TypeItem) * count;
256     }
257 
258    private:
259     uint32_t size_;  // size of the list, in entries
260     TypeItem list_[1];  // elements of the list
261     DISALLOW_COPY_AND_ASSIGN(TypeList);
262   };
263 
264   // MethodHandle Types
265   enum class MethodHandleType : uint16_t {  // private
266     kStaticPut         = 0x0000,  // a setter for a given static field.
267     kStaticGet         = 0x0001,  // a getter for a given static field.
268     kInstancePut       = 0x0002,  // a setter for a given instance field.
269     kInstanceGet       = 0x0003,  // a getter for a given instance field.
270     kInvokeStatic      = 0x0004,  // an invoker for a given static method.
271     kInvokeInstance    = 0x0005,  // invoke_instance : an invoker for a given instance method. This
272                                   // can be any non-static method on any class (or interface) except
273                                   // for “<init>”.
274     kInvokeConstructor = 0x0006,  // an invoker for a given constructor.
275     kInvokeDirect      = 0x0007,  // an invoker for a direct (special) method.
276     kInvokeInterface   = 0x0008,  // an invoker for an interface method.
277     kLast = kInvokeInterface
278   };
279 
280   // raw method_handle_item
281   struct MethodHandleItem {
282     uint16_t method_handle_type_;
283     uint16_t reserved1_;            // Reserved for future use.
284     uint16_t field_or_method_idx_;  // Field index for accessors, method index otherwise.
285     uint16_t reserved2_;            // Reserved for future use.
286    private:
287     DISALLOW_COPY_AND_ASSIGN(MethodHandleItem);
288   };
289 
290   // raw call_site_id_item
291   struct CallSiteIdItem {
292     uint32_t data_off_;  // Offset into data section pointing to encoded array items.
293    private:
294     DISALLOW_COPY_AND_ASSIGN(CallSiteIdItem);
295   };
296 
297   // Raw code_item.
298   struct CodeItem {
299     uint16_t registers_size_;            // the number of registers used by this code
300                                          //   (locals + parameters)
301     uint16_t ins_size_;                  // the number of words of incoming arguments to the method
302                                          //   that this code is for
303     uint16_t outs_size_;                 // the number of words of outgoing argument space required
304                                          //   by this code for method invocation
305     uint16_t tries_size_;                // the number of try_items for this instance. If non-zero,
306                                          //   then these appear as the tries array just after the
307                                          //   insns in this instance.
308     uint32_t debug_info_off_;            // file offset to debug info stream
309     uint32_t insns_size_in_code_units_;  // size of the insns array, in 2 byte code units
310     uint16_t insns_[1];                  // actual array of bytecode.
311 
312    private:
313     DISALLOW_COPY_AND_ASSIGN(CodeItem);
314   };
315 
316   // Raw try_item.
317   struct TryItem {
318     uint32_t start_addr_;
319     uint16_t insn_count_;
320     uint16_t handler_off_;
321 
322    private:
323     DISALLOW_COPY_AND_ASSIGN(TryItem);
324   };
325 
326   // Annotation constants.
327   enum {
328     kDexVisibilityBuild         = 0x00,     /* annotation visibility */
329     kDexVisibilityRuntime       = 0x01,
330     kDexVisibilitySystem        = 0x02,
331 
332     kDexAnnotationByte          = 0x00,
333     kDexAnnotationShort         = 0x02,
334     kDexAnnotationChar          = 0x03,
335     kDexAnnotationInt           = 0x04,
336     kDexAnnotationLong          = 0x06,
337     kDexAnnotationFloat         = 0x10,
338     kDexAnnotationDouble        = 0x11,
339     kDexAnnotationMethodType    = 0x15,
340     kDexAnnotationMethodHandle  = 0x16,
341     kDexAnnotationString        = 0x17,
342     kDexAnnotationType          = 0x18,
343     kDexAnnotationField         = 0x19,
344     kDexAnnotationMethod        = 0x1a,
345     kDexAnnotationEnum          = 0x1b,
346     kDexAnnotationArray         = 0x1c,
347     kDexAnnotationAnnotation    = 0x1d,
348     kDexAnnotationNull          = 0x1e,
349     kDexAnnotationBoolean       = 0x1f,
350 
351     kDexAnnotationValueTypeMask = 0x1f,     /* low 5 bits */
352     kDexAnnotationValueArgShift = 5,
353   };
354 
355   struct AnnotationsDirectoryItem {
356     uint32_t class_annotations_off_;
357     uint32_t fields_size_;
358     uint32_t methods_size_;
359     uint32_t parameters_size_;
360 
361    private:
362     DISALLOW_COPY_AND_ASSIGN(AnnotationsDirectoryItem);
363   };
364 
365   struct FieldAnnotationsItem {
366     uint32_t field_idx_;
367     uint32_t annotations_off_;
368 
369    private:
370     DISALLOW_COPY_AND_ASSIGN(FieldAnnotationsItem);
371   };
372 
373   struct MethodAnnotationsItem {
374     uint32_t method_idx_;
375     uint32_t annotations_off_;
376 
377    private:
378     DISALLOW_COPY_AND_ASSIGN(MethodAnnotationsItem);
379   };
380 
381   struct ParameterAnnotationsItem {
382     uint32_t method_idx_;
383     uint32_t annotations_off_;
384 
385    private:
386     DISALLOW_COPY_AND_ASSIGN(ParameterAnnotationsItem);
387   };
388 
389   struct AnnotationSetRefItem {
390     uint32_t annotations_off_;
391 
392    private:
393     DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefItem);
394   };
395 
396   struct AnnotationSetRefList {
397     uint32_t size_;
398     AnnotationSetRefItem list_[1];
399 
400    private:
401     DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefList);
402   };
403 
404   struct AnnotationSetItem {
405     uint32_t size_;
406     uint32_t entries_[1];
407 
408    private:
409     DISALLOW_COPY_AND_ASSIGN(AnnotationSetItem);
410   };
411 
412   struct AnnotationItem {
413     uint8_t visibility_;
414     uint8_t annotation_[1];
415 
416    private:
417     DISALLOW_COPY_AND_ASSIGN(AnnotationItem);
418   };
419 
420   enum AnnotationResultStyle {  // private
421     kAllObjects,
422     kPrimitivesOrObjects,
423     kAllRaw
424   };
425 
426   struct AnnotationValue;
427 
428   // Returns the checksums of a file for comparison with GetLocationChecksum().
429   // For .dex files, this is the single header checksum.
430   // For zip files, this is the zip entry CRC32 checksum for classes.dex and
431   // each additional multidex entry classes2.dex, classes3.dex, etc.
432   // Return true if the checksums could be found, false otherwise.
433   static bool GetMultiDexChecksums(const char* filename,
434                                    std::vector<uint32_t>* checksums,
435                                    std::string* error_msg);
436 
437   // Check whether a location denotes a multidex dex file. This is a very simple check: returns
438   // whether the string contains the separator character.
439   static bool IsMultiDexLocation(const char* location);
440 
441   // Opens .dex file, backed by existing memory
442   static std::unique_ptr<const DexFile> Open(const uint8_t* base,
443                                              size_t size,
444                                              const std::string& location,
445                                              uint32_t location_checksum,
446                                              const OatDexFile* oat_dex_file,
447                                              bool verify,
448                                              bool verify_checksum,
449                                              std::string* error_msg);
450 
451   // Opens .dex file that has been memory-mapped by the caller.
452   static std::unique_ptr<const DexFile> Open(const std::string& location,
453                                              uint32_t location_checkum,
454                                              std::unique_ptr<MemMap> mem_map,
455                                              bool verify,
456                                              bool verify_checksum,
457                                              std::string* error_msg);
458 
459   // Opens all .dex files found in the file, guessing the container format based on file extension.
460   static bool Open(const char* filename,
461                    const std::string& location,
462                    bool verify_checksum,
463                    std::string* error_msg,
464                    std::vector<std::unique_ptr<const DexFile>>* dex_files);
465 
466   // Open a single dex file from an fd.
467   static std::unique_ptr<const DexFile> OpenDex(int fd,
468                                                 const std::string& location,
469                                                 bool verify_checksum,
470                                                 std::string* error_msg);
471 
472   // Opens dex files from within a .jar, .zip, or .apk file
473   static bool OpenZip(int fd,
474                       const std::string& location,
475                       bool verify_checksum,
476                       std::string* error_msg,
477                       std::vector<std::unique_ptr<const DexFile>>* dex_files);
478 
479   // Closes a .dex file.
480   virtual ~DexFile();
481 
GetLocation()482   const std::string& GetLocation() const {
483     return location_;
484   }
485 
486   // For normal dex files, location and base location coincide. If a dex file is part of a multidex
487   // archive, the base location is the name of the originating jar/apk, stripped of any internal
488   // classes*.dex path.
GetBaseLocation(const char * location)489   static std::string GetBaseLocation(const char* location) {
490     const char* pos = strrchr(location, kMultiDexSeparator);
491     if (pos == nullptr) {
492       return location;
493     } else {
494       return std::string(location, pos - location);
495     }
496   }
497 
GetBaseLocation(const std::string & location)498   static std::string GetBaseLocation(const std::string& location) {
499     return GetBaseLocation(location.c_str());
500   }
501 
502   // Returns the '!classes*.dex' part of the dex location. Returns an empty
503   // string if there is no multidex suffix for the given location.
504   // The kMultiDexSeparator is included in the returned suffix.
GetMultiDexSuffix(const std::string & location)505   static std::string GetMultiDexSuffix(const std::string& location) {
506     size_t pos = location.rfind(kMultiDexSeparator);
507     if (pos == std::string::npos) {
508       return "";
509     } else {
510       return location.substr(pos);
511     }
512   }
513 
GetBaseLocation()514   std::string GetBaseLocation() const {
515     return GetBaseLocation(location_);
516   }
517 
518   // For DexFiles directly from .dex files, this is the checksum from the DexFile::Header.
519   // For DexFiles opened from a zip files, this will be the ZipEntry CRC32 of classes.dex.
GetLocationChecksum()520   uint32_t GetLocationChecksum() const {
521     return location_checksum_;
522   }
523 
GetHeader()524   const Header& GetHeader() const {
525     DCHECK(header_ != nullptr) << GetLocation();
526     return *header_;
527   }
528 
529   // Decode the dex magic version
GetVersion()530   uint32_t GetVersion() const {
531     return GetHeader().GetVersion();
532   }
533 
534   // Returns true if the byte string points to the magic value.
535   static bool IsMagicValid(const uint8_t* magic);
536 
537   // Returns true if the byte string after the magic is the correct value.
538   static bool IsVersionValid(const uint8_t* magic);
539 
540   // Returns the number of string identifiers in the .dex file.
NumStringIds()541   size_t NumStringIds() const {
542     DCHECK(header_ != nullptr) << GetLocation();
543     return header_->string_ids_size_;
544   }
545 
546   // Returns the StringId at the specified index.
GetStringId(dex::StringIndex idx)547   const StringId& GetStringId(dex::StringIndex idx) const {
548     DCHECK_LT(idx.index_, NumStringIds()) << GetLocation();
549     return string_ids_[idx.index_];
550   }
551 
GetIndexForStringId(const StringId & string_id)552   dex::StringIndex GetIndexForStringId(const StringId& string_id) const {
553     CHECK_GE(&string_id, string_ids_) << GetLocation();
554     CHECK_LT(&string_id, string_ids_ + header_->string_ids_size_) << GetLocation();
555     return dex::StringIndex(&string_id - string_ids_);
556   }
557 
558   int32_t GetStringLength(const StringId& string_id) const;
559 
560   // Returns a pointer to the UTF-8 string data referred to by the given string_id as well as the
561   // length of the string when decoded as a UTF-16 string. Note the UTF-16 length is not the same
562   // as the string length of the string data.
563   const char* GetStringDataAndUtf16Length(const StringId& string_id, uint32_t* utf16_length) const;
564 
565   const char* GetStringData(const StringId& string_id) const;
566 
567   // Index version of GetStringDataAndUtf16Length.
568   const char* StringDataAndUtf16LengthByIdx(dex::StringIndex idx, uint32_t* utf16_length) const;
569 
570   const char* StringDataByIdx(dex::StringIndex idx) const;
571 
572   // Looks up a string id for a given modified utf8 string.
573   const StringId* FindStringId(const char* string) const;
574 
575   const TypeId* FindTypeId(const char* string) const;
576 
577   // Looks up a string id for a given utf16 string.
578   const StringId* FindStringId(const uint16_t* string, size_t length) const;
579 
580   // Returns the number of type identifiers in the .dex file.
NumTypeIds()581   uint32_t NumTypeIds() const {
582     DCHECK(header_ != nullptr) << GetLocation();
583     return header_->type_ids_size_;
584   }
585 
IsTypeIndexValid(dex::TypeIndex idx)586   bool IsTypeIndexValid(dex::TypeIndex idx) const {
587     return idx.IsValid() && idx.index_ < NumTypeIds();
588   }
589 
590   // Returns the TypeId at the specified index.
GetTypeId(dex::TypeIndex idx)591   const TypeId& GetTypeId(dex::TypeIndex idx) const {
592     DCHECK_LT(idx.index_, NumTypeIds()) << GetLocation();
593     return type_ids_[idx.index_];
594   }
595 
GetIndexForTypeId(const TypeId & type_id)596   dex::TypeIndex GetIndexForTypeId(const TypeId& type_id) const {
597     CHECK_GE(&type_id, type_ids_) << GetLocation();
598     CHECK_LT(&type_id, type_ids_ + header_->type_ids_size_) << GetLocation();
599     size_t result = &type_id - type_ids_;
600     DCHECK_LT(result, 65536U) << GetLocation();
601     return dex::TypeIndex(static_cast<uint16_t>(result));
602   }
603 
604   // Get the descriptor string associated with a given type index.
605   const char* StringByTypeIdx(dex::TypeIndex idx, uint32_t* unicode_length) const;
606 
607   const char* StringByTypeIdx(dex::TypeIndex idx) const;
608 
609   // Returns the type descriptor string of a type id.
610   const char* GetTypeDescriptor(const TypeId& type_id) const;
611 
612   // Looks up a type for the given string index
613   const TypeId* FindTypeId(dex::StringIndex string_idx) const;
614 
615   // Returns the number of field identifiers in the .dex file.
NumFieldIds()616   size_t NumFieldIds() const {
617     DCHECK(header_ != nullptr) << GetLocation();
618     return header_->field_ids_size_;
619   }
620 
621   // Returns the FieldId at the specified index.
GetFieldId(uint32_t idx)622   const FieldId& GetFieldId(uint32_t idx) const {
623     DCHECK_LT(idx, NumFieldIds()) << GetLocation();
624     return field_ids_[idx];
625   }
626 
GetIndexForFieldId(const FieldId & field_id)627   uint32_t GetIndexForFieldId(const FieldId& field_id) const {
628     CHECK_GE(&field_id, field_ids_) << GetLocation();
629     CHECK_LT(&field_id, field_ids_ + header_->field_ids_size_) << GetLocation();
630     return &field_id - field_ids_;
631   }
632 
633   // Looks up a field by its declaring class, name and type
634   const FieldId* FindFieldId(const DexFile::TypeId& declaring_klass,
635                              const DexFile::StringId& name,
636                              const DexFile::TypeId& type) const;
637 
638   uint32_t FindCodeItemOffset(const DexFile::ClassDef& class_def,
639                               uint32_t dex_method_idx) const;
640 
641   static uint32_t GetCodeItemSize(const DexFile::CodeItem& disk_code_item);
642 
643   // Returns the declaring class descriptor string of a field id.
GetFieldDeclaringClassDescriptor(const FieldId & field_id)644   const char* GetFieldDeclaringClassDescriptor(const FieldId& field_id) const {
645     const DexFile::TypeId& type_id = GetTypeId(field_id.class_idx_);
646     return GetTypeDescriptor(type_id);
647   }
648 
649   // Returns the class descriptor string of a field id.
650   const char* GetFieldTypeDescriptor(const FieldId& field_id) const;
651 
652   // Returns the name of a field id.
653   const char* GetFieldName(const FieldId& field_id) const;
654 
655   // Returns the number of method identifiers in the .dex file.
NumMethodIds()656   size_t NumMethodIds() const {
657     DCHECK(header_ != nullptr) << GetLocation();
658     return header_->method_ids_size_;
659   }
660 
661   // Returns the MethodId at the specified index.
GetMethodId(uint32_t idx)662   const MethodId& GetMethodId(uint32_t idx) const {
663     DCHECK_LT(idx, NumMethodIds()) << GetLocation();
664     return method_ids_[idx];
665   }
666 
GetIndexForMethodId(const MethodId & method_id)667   uint32_t GetIndexForMethodId(const MethodId& method_id) const {
668     CHECK_GE(&method_id, method_ids_) << GetLocation();
669     CHECK_LT(&method_id, method_ids_ + header_->method_ids_size_) << GetLocation();
670     return &method_id - method_ids_;
671   }
672 
673   // Looks up a method by its declaring class, name and proto_id
674   const MethodId* FindMethodId(const DexFile::TypeId& declaring_klass,
675                                const DexFile::StringId& name,
676                                const DexFile::ProtoId& signature) const;
677 
678   // Returns the declaring class descriptor string of a method id.
679   const char* GetMethodDeclaringClassDescriptor(const MethodId& method_id) const;
680 
681   // Returns the prototype of a method id.
GetMethodPrototype(const MethodId & method_id)682   const ProtoId& GetMethodPrototype(const MethodId& method_id) const {
683     return GetProtoId(method_id.proto_idx_);
684   }
685 
686   // Returns a representation of the signature of a method id.
687   const Signature GetMethodSignature(const MethodId& method_id) const;
688 
689   // Returns a representation of the signature of a proto id.
690   const Signature GetProtoSignature(const ProtoId& proto_id) const;
691 
692   // Returns the name of a method id.
693   const char* GetMethodName(const MethodId& method_id) const;
694 
695   // Returns the shorty of a method by its index.
696   const char* GetMethodShorty(uint32_t idx) const;
697 
698   // Returns the shorty of a method id.
699   const char* GetMethodShorty(const MethodId& method_id) const;
700   const char* GetMethodShorty(const MethodId& method_id, uint32_t* length) const;
701 
702   // Returns the number of class definitions in the .dex file.
NumClassDefs()703   uint32_t NumClassDefs() const {
704     DCHECK(header_ != nullptr) << GetLocation();
705     return header_->class_defs_size_;
706   }
707 
708   // Returns the ClassDef at the specified index.
GetClassDef(uint16_t idx)709   const ClassDef& GetClassDef(uint16_t idx) const {
710     DCHECK_LT(idx, NumClassDefs()) << GetLocation();
711     return class_defs_[idx];
712   }
713 
GetIndexForClassDef(const ClassDef & class_def)714   uint16_t GetIndexForClassDef(const ClassDef& class_def) const {
715     CHECK_GE(&class_def, class_defs_) << GetLocation();
716     CHECK_LT(&class_def, class_defs_ + header_->class_defs_size_) << GetLocation();
717     return &class_def - class_defs_;
718   }
719 
720   // Returns the class descriptor string of a class definition.
721   const char* GetClassDescriptor(const ClassDef& class_def) const;
722 
723   // Looks up a class definition by its type index.
724   const ClassDef* FindClassDef(dex::TypeIndex type_idx) const;
725 
GetInterfacesList(const ClassDef & class_def)726   const TypeList* GetInterfacesList(const ClassDef& class_def) const {
727     if (class_def.interfaces_off_ == 0) {
728         return nullptr;
729     } else {
730       const uint8_t* addr = begin_ + class_def.interfaces_off_;
731       return reinterpret_cast<const TypeList*>(addr);
732     }
733   }
734 
NumMethodHandles()735   uint32_t NumMethodHandles() const {
736     return num_method_handles_;
737   }
738 
GetMethodHandle(uint32_t idx)739   const MethodHandleItem& GetMethodHandle(uint32_t idx) const {
740     CHECK_LT(idx, NumMethodHandles());
741     return method_handles_[idx];
742   }
743 
NumCallSiteIds()744   uint32_t NumCallSiteIds() const {
745     return num_call_site_ids_;
746   }
747 
GetCallSiteId(uint32_t idx)748   const CallSiteIdItem& GetCallSiteId(uint32_t idx) const {
749     CHECK_LT(idx, NumCallSiteIds());
750     return call_site_ids_[idx];
751   }
752 
753   // Returns a pointer to the raw memory mapped class_data_item
GetClassData(const ClassDef & class_def)754   const uint8_t* GetClassData(const ClassDef& class_def) const {
755     if (class_def.class_data_off_ == 0) {
756       return nullptr;
757     } else {
758       return begin_ + class_def.class_data_off_;
759     }
760   }
761 
762   //
GetCodeItem(const uint32_t code_off)763   const CodeItem* GetCodeItem(const uint32_t code_off) const {
764     DCHECK_LT(code_off, size_) << "Code item offset larger then maximum allowed offset";
765     if (code_off == 0) {
766       return nullptr;  // native or abstract method
767     } else {
768       const uint8_t* addr = begin_ + code_off;
769       return reinterpret_cast<const CodeItem*>(addr);
770     }
771   }
772 
773   const char* GetReturnTypeDescriptor(const ProtoId& proto_id) const;
774 
775   // Returns the number of prototype identifiers in the .dex file.
NumProtoIds()776   size_t NumProtoIds() const {
777     DCHECK(header_ != nullptr) << GetLocation();
778     return header_->proto_ids_size_;
779   }
780 
781   // Returns the ProtoId at the specified index.
GetProtoId(uint16_t idx)782   const ProtoId& GetProtoId(uint16_t idx) const {
783     DCHECK_LT(idx, NumProtoIds()) << GetLocation();
784     return proto_ids_[idx];
785   }
786 
GetIndexForProtoId(const ProtoId & proto_id)787   uint16_t GetIndexForProtoId(const ProtoId& proto_id) const {
788     CHECK_GE(&proto_id, proto_ids_) << GetLocation();
789     CHECK_LT(&proto_id, proto_ids_ + header_->proto_ids_size_) << GetLocation();
790     return &proto_id - proto_ids_;
791   }
792 
793   // Looks up a proto id for a given return type and signature type list
794   const ProtoId* FindProtoId(dex::TypeIndex return_type_idx,
795                              const dex::TypeIndex* signature_type_idxs,
796                              uint32_t signature_length) const;
FindProtoId(dex::TypeIndex return_type_idx,const std::vector<dex::TypeIndex> & signature_type_idxs)797   const ProtoId* FindProtoId(dex::TypeIndex return_type_idx,
798                              const std::vector<dex::TypeIndex>& signature_type_idxs) const {
799     return FindProtoId(return_type_idx, &signature_type_idxs[0], signature_type_idxs.size());
800   }
801 
802   // Given a signature place the type ids into the given vector, returns true on success
803   bool CreateTypeList(const StringPiece& signature,
804                       dex::TypeIndex* return_type_idx,
805                       std::vector<dex::TypeIndex>* param_type_idxs) const;
806 
807   // Create a Signature from the given string signature or return Signature::NoSignature if not
808   // possible.
809   const Signature CreateSignature(const StringPiece& signature) const;
810 
811   // Returns the short form method descriptor for the given prototype.
812   const char* GetShorty(uint32_t proto_idx) const;
813 
GetProtoParameters(const ProtoId & proto_id)814   const TypeList* GetProtoParameters(const ProtoId& proto_id) const {
815     if (proto_id.parameters_off_ == 0) {
816       return nullptr;
817     } else {
818       const uint8_t* addr = begin_ + proto_id.parameters_off_;
819       return reinterpret_cast<const TypeList*>(addr);
820     }
821   }
822 
GetEncodedStaticFieldValuesArray(const ClassDef & class_def)823   const uint8_t* GetEncodedStaticFieldValuesArray(const ClassDef& class_def) const {
824     if (class_def.static_values_off_ == 0) {
825       return 0;
826     } else {
827       return begin_ + class_def.static_values_off_;
828     }
829   }
830 
GetCallSiteEncodedValuesArray(const CallSiteIdItem & call_site_id)831   const uint8_t* GetCallSiteEncodedValuesArray(const CallSiteIdItem& call_site_id) const {
832     return begin_ + call_site_id.data_off_;
833   }
834 
835   static const TryItem* GetTryItems(const CodeItem& code_item, uint32_t offset);
836 
837   // Get the base of the encoded data for the given DexCode.
GetCatchHandlerData(const CodeItem & code_item,uint32_t offset)838   static const uint8_t* GetCatchHandlerData(const CodeItem& code_item, uint32_t offset) {
839     const uint8_t* handler_data =
840         reinterpret_cast<const uint8_t*>(GetTryItems(code_item, code_item.tries_size_));
841     return handler_data + offset;
842   }
843 
844   // Find which try region is associated with the given address (ie dex pc). Returns -1 if none.
845   static int32_t FindTryItem(const CodeItem &code_item, uint32_t address);
846 
847   // Find the handler offset associated with the given address (ie dex pc). Returns -1 if none.
848   static int32_t FindCatchHandlerOffset(const CodeItem &code_item, uint32_t address);
849 
850   // Get the pointer to the start of the debugging data
GetDebugInfoStream(const CodeItem * code_item)851   const uint8_t* GetDebugInfoStream(const CodeItem* code_item) const {
852     // Check that the offset is in bounds.
853     // Note that although the specification says that 0 should be used if there
854     // is no debug information, some applications incorrectly use 0xFFFFFFFF.
855     if (code_item->debug_info_off_ == 0 || code_item->debug_info_off_ >= size_) {
856       return nullptr;
857     } else {
858       return begin_ + code_item->debug_info_off_;
859     }
860   }
861 
862   struct PositionInfo {
PositionInfoPositionInfo863     PositionInfo()
864         : address_(0),
865           line_(0),
866           source_file_(nullptr),
867           prologue_end_(false),
868           epilogue_begin_(false) {
869     }
870 
871     uint32_t address_;  // In 16-bit code units.
872     uint32_t line_;  // Source code line number starting at 1.
873     const char* source_file_;  // nullptr if the file from ClassDef still applies.
874     bool prologue_end_;
875     bool epilogue_begin_;
876   };
877 
878   // Callback for "new position table entry".
879   // Returning true causes the decoder to stop early.
880   typedef bool (*DexDebugNewPositionCb)(void* context, const PositionInfo& entry);
881 
882   struct LocalInfo {
LocalInfoLocalInfo883     LocalInfo()
884         : name_(nullptr),
885           descriptor_(nullptr),
886           signature_(nullptr),
887           start_address_(0),
888           end_address_(0),
889           reg_(0),
890           is_live_(false) {
891     }
892 
893     const char* name_;  // E.g., list.  It can be nullptr if unknown.
894     const char* descriptor_;  // E.g., Ljava/util/LinkedList;
895     const char* signature_;  // E.g., java.util.LinkedList<java.lang.Integer>
896     uint32_t start_address_;  // PC location where the local is first defined.
897     uint32_t end_address_;  // PC location where the local is no longer defined.
898     uint16_t reg_;  // Dex register which stores the values.
899     bool is_live_;  // Is the local defined and live.
900   };
901 
902   // Callback for "new locals table entry".
903   typedef void (*DexDebugNewLocalCb)(void* context, const LocalInfo& entry);
904 
905   static bool LineNumForPcCb(void* context, const PositionInfo& entry);
906 
GetAnnotationsDirectory(const ClassDef & class_def)907   const AnnotationsDirectoryItem* GetAnnotationsDirectory(const ClassDef& class_def) const {
908     if (class_def.annotations_off_ == 0) {
909       return nullptr;
910     } else {
911       return reinterpret_cast<const AnnotationsDirectoryItem*>(begin_ + class_def.annotations_off_);
912     }
913   }
914 
GetClassAnnotationSet(const AnnotationsDirectoryItem * anno_dir)915   const AnnotationSetItem* GetClassAnnotationSet(const AnnotationsDirectoryItem* anno_dir) const {
916     if (anno_dir->class_annotations_off_ == 0) {
917       return nullptr;
918     } else {
919       return reinterpret_cast<const AnnotationSetItem*>(begin_ + anno_dir->class_annotations_off_);
920     }
921   }
922 
GetFieldAnnotations(const AnnotationsDirectoryItem * anno_dir)923   const FieldAnnotationsItem* GetFieldAnnotations(const AnnotationsDirectoryItem* anno_dir) const {
924     if (anno_dir->fields_size_ == 0) {
925       return nullptr;
926     } else {
927       return reinterpret_cast<const FieldAnnotationsItem*>(&anno_dir[1]);
928     }
929   }
930 
GetMethodAnnotations(const AnnotationsDirectoryItem * anno_dir)931   const MethodAnnotationsItem* GetMethodAnnotations(const AnnotationsDirectoryItem* anno_dir)
932       const {
933     if (anno_dir->methods_size_ == 0) {
934       return nullptr;
935     } else {
936       // Skip past the header and field annotations.
937       const uint8_t* addr = reinterpret_cast<const uint8_t*>(&anno_dir[1]);
938       addr += anno_dir->fields_size_ * sizeof(FieldAnnotationsItem);
939       return reinterpret_cast<const MethodAnnotationsItem*>(addr);
940     }
941   }
942 
GetParameterAnnotations(const AnnotationsDirectoryItem * anno_dir)943   const ParameterAnnotationsItem* GetParameterAnnotations(const AnnotationsDirectoryItem* anno_dir)
944       const {
945     if (anno_dir->parameters_size_ == 0) {
946       return nullptr;
947     } else {
948       // Skip past the header, field annotations, and method annotations.
949       const uint8_t* addr = reinterpret_cast<const uint8_t*>(&anno_dir[1]);
950       addr += anno_dir->fields_size_ * sizeof(FieldAnnotationsItem);
951       addr += anno_dir->methods_size_ * sizeof(MethodAnnotationsItem);
952       return reinterpret_cast<const ParameterAnnotationsItem*>(addr);
953     }
954   }
955 
GetFieldAnnotationSetItem(const FieldAnnotationsItem & anno_item)956   const AnnotationSetItem* GetFieldAnnotationSetItem(const FieldAnnotationsItem& anno_item) const {
957     uint32_t offset = anno_item.annotations_off_;
958     if (offset == 0) {
959       return nullptr;
960     } else {
961       return reinterpret_cast<const AnnotationSetItem*>(begin_ + offset);
962     }
963   }
964 
GetMethodAnnotationSetItem(const MethodAnnotationsItem & anno_item)965   const AnnotationSetItem* GetMethodAnnotationSetItem(const MethodAnnotationsItem& anno_item)
966       const {
967     uint32_t offset = anno_item.annotations_off_;
968     if (offset == 0) {
969       return nullptr;
970     } else {
971       return reinterpret_cast<const AnnotationSetItem*>(begin_ + offset);
972     }
973   }
974 
GetParameterAnnotationSetRefList(const ParameterAnnotationsItem * anno_item)975   const AnnotationSetRefList* GetParameterAnnotationSetRefList(
976       const ParameterAnnotationsItem* anno_item) const {
977     uint32_t offset = anno_item->annotations_off_;
978     if (offset == 0) {
979       return nullptr;
980     }
981     return reinterpret_cast<const AnnotationSetRefList*>(begin_ + offset);
982   }
983 
GetAnnotationItem(const AnnotationSetItem * set_item,uint32_t index)984   const AnnotationItem* GetAnnotationItem(const AnnotationSetItem* set_item, uint32_t index) const {
985     DCHECK_LE(index, set_item->size_);
986     uint32_t offset = set_item->entries_[index];
987     if (offset == 0) {
988       return nullptr;
989     } else {
990       return reinterpret_cast<const AnnotationItem*>(begin_ + offset);
991     }
992   }
993 
GetSetRefItemItem(const AnnotationSetRefItem * anno_item)994   const AnnotationSetItem* GetSetRefItemItem(const AnnotationSetRefItem* anno_item) const {
995     uint32_t offset = anno_item->annotations_off_;
996     if (offset == 0) {
997       return nullptr;
998     }
999     return reinterpret_cast<const AnnotationSetItem*>(begin_ + offset);
1000   }
1001 
1002   // Debug info opcodes and constants
1003   enum {
1004     DBG_END_SEQUENCE         = 0x00,
1005     DBG_ADVANCE_PC           = 0x01,
1006     DBG_ADVANCE_LINE         = 0x02,
1007     DBG_START_LOCAL          = 0x03,
1008     DBG_START_LOCAL_EXTENDED = 0x04,
1009     DBG_END_LOCAL            = 0x05,
1010     DBG_RESTART_LOCAL        = 0x06,
1011     DBG_SET_PROLOGUE_END     = 0x07,
1012     DBG_SET_EPILOGUE_BEGIN   = 0x08,
1013     DBG_SET_FILE             = 0x09,
1014     DBG_FIRST_SPECIAL        = 0x0a,
1015     DBG_LINE_BASE            = -4,
1016     DBG_LINE_RANGE           = 15,
1017   };
1018 
1019   struct LineNumFromPcContext {
LineNumFromPcContextLineNumFromPcContext1020     LineNumFromPcContext(uint32_t address, uint32_t line_num)
1021         : address_(address), line_num_(line_num) {}
1022     uint32_t address_;
1023     uint32_t line_num_;
1024    private:
1025     DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext);
1026   };
1027 
1028   // Returns false if there is no debugging information or if it cannot be decoded.
1029   bool DecodeDebugLocalInfo(const CodeItem* code_item, bool is_static, uint32_t method_idx,
1030                             DexDebugNewLocalCb local_cb, void* context) const;
1031 
1032   // Returns false if there is no debugging information or if it cannot be decoded.
1033   bool DecodeDebugPositionInfo(const CodeItem* code_item, DexDebugNewPositionCb position_cb,
1034                                void* context) const;
1035 
GetSourceFile(const ClassDef & class_def)1036   const char* GetSourceFile(const ClassDef& class_def) const {
1037     if (!class_def.source_file_idx_.IsValid()) {
1038       return nullptr;
1039     } else {
1040       return StringDataByIdx(class_def.source_file_idx_);
1041     }
1042   }
1043 
1044   int GetPermissions() const;
1045 
1046   bool IsReadOnly() const;
1047 
1048   bool EnableWrite() const;
1049 
1050   bool DisableWrite() const;
1051 
Begin()1052   const uint8_t* Begin() const {
1053     return begin_;
1054   }
1055 
Size()1056   size_t Size() const {
1057     return size_;
1058   }
1059 
1060   // Return the name of the index-th classes.dex in a multidex zip file. This is classes.dex for
1061   // index == 0, and classes{index + 1}.dex else.
1062   static std::string GetMultiDexClassesDexName(size_t index);
1063 
1064   // Return the (possibly synthetic) dex location for a multidex entry. This is dex_location for
1065   // index == 0, and dex_location + multi-dex-separator + GetMultiDexClassesDexName(index) else.
1066   static std::string GetMultiDexLocation(size_t index, const char* dex_location);
1067 
1068   // Returns the canonical form of the given dex location.
1069   //
1070   // There are different flavors of "dex locations" as follows:
1071   // the file name of a dex file:
1072   //     The actual file path that the dex file has on disk.
1073   // dex_location:
1074   //     This acts as a key for the class linker to know which dex file to load.
1075   //     It may correspond to either an old odex file or a particular dex file
1076   //     inside an oat file. In the first case it will also match the file name
1077   //     of the dex file. In the second case (oat) it will include the file name
1078   //     and possibly some multidex annotation to uniquely identify it.
1079   // canonical_dex_location:
1080   //     the dex_location where it's file name part has been made canonical.
1081   static std::string GetDexCanonicalLocation(const char* dex_location);
1082 
GetOatDexFile()1083   const OatDexFile* GetOatDexFile() const {
1084     return oat_dex_file_;
1085   }
1086 
1087   // Used by oat writer.
SetOatDexFile(OatDexFile * oat_dex_file)1088   void SetOatDexFile(OatDexFile* oat_dex_file) const {
1089     oat_dex_file_ = oat_dex_file;
1090   }
1091 
1092   // Utility methods for reading integral values from a buffer.
1093   static int32_t ReadSignedInt(const uint8_t* ptr, int zwidth);
1094   static uint32_t ReadUnsignedInt(const uint8_t* ptr, int zwidth, bool fill_on_right);
1095   static int64_t ReadSignedLong(const uint8_t* ptr, int zwidth);
1096   static uint64_t ReadUnsignedLong(const uint8_t* ptr, int zwidth, bool fill_on_right);
1097 
1098   // Recalculates the checksum of the dex file. Does not use the current value in the header.
1099   uint32_t CalculateChecksum() const;
1100 
1101   // Returns a human-readable form of the method at an index.
1102   std::string PrettyMethod(uint32_t method_idx, bool with_signature = true) const;
1103   // Returns a human-readable form of the field at an index.
1104   std::string PrettyField(uint32_t field_idx, bool with_type = true) const;
1105   // Returns a human-readable form of the type at an index.
1106   std::string PrettyType(dex::TypeIndex type_idx) const;
1107 
1108  private:
1109   static std::unique_ptr<const DexFile> OpenFile(int fd,
1110                                                  const std::string& location,
1111                                                  bool verify,
1112                                                  bool verify_checksum,
1113                                                  std::string* error_msg);
1114 
1115   enum class ZipOpenErrorCode {  // private
1116     kNoError,
1117     kEntryNotFound,
1118     kExtractToMemoryError,
1119     kDexFileError,
1120     kMakeReadOnlyError,
1121     kVerifyError
1122   };
1123 
1124   // Open all classesXXX.dex files from a zip archive.
1125   static bool OpenAllDexFilesFromZip(const ZipArchive& zip_archive,
1126                                      const std::string& location,
1127                                      bool verify_checksum,
1128                                      std::string* error_msg,
1129                                      std::vector<std::unique_ptr<const DexFile>>* dex_files);
1130 
1131   // Opens .dex file from the entry_name in a zip archive. error_code is undefined when non-null
1132   // return.
1133   static std::unique_ptr<const DexFile> OpenOneDexFileFromZip(const ZipArchive& zip_archive,
1134                                                               const char* entry_name,
1135                                                               const std::string& location,
1136                                                               bool verify_checksum,
1137                                                               std::string* error_msg,
1138                                                               ZipOpenErrorCode* error_code);
1139 
1140   enum class VerifyResult {  // private
1141     kVerifyNotAttempted,
1142     kVerifySucceeded,
1143     kVerifyFailed
1144   };
1145 
1146   static std::unique_ptr<DexFile> OpenCommon(const uint8_t* base,
1147                                              size_t size,
1148                                              const std::string& location,
1149                                              uint32_t location_checksum,
1150                                              const OatDexFile* oat_dex_file,
1151                                              bool verify,
1152                                              bool verify_checksum,
1153                                              std::string* error_msg,
1154                                              VerifyResult* verify_result = nullptr);
1155 
1156 
1157   // Opens a .dex file at the given address, optionally backed by a MemMap
1158   static std::unique_ptr<const DexFile> OpenMemory(const uint8_t* dex_file,
1159                                                    size_t size,
1160                                                    const std::string& location,
1161                                                    uint32_t location_checksum,
1162                                                    std::unique_ptr<MemMap> mem_map,
1163                                                    const OatDexFile* oat_dex_file,
1164                                                    std::string* error_msg);
1165 
1166   DexFile(const uint8_t* base,
1167           size_t size,
1168           const std::string& location,
1169           uint32_t location_checksum,
1170           const OatDexFile* oat_dex_file);
1171 
1172   // Top-level initializer that calls other Init methods.
1173   bool Init(std::string* error_msg);
1174 
1175   // Returns true if the header magic and version numbers are of the expected values.
1176   bool CheckMagicAndVersion(std::string* error_msg) const;
1177 
1178   // Initialize section info for sections only found in map. Returns true on success.
1179   void InitializeSectionsFromMapList();
1180 
1181   // The base address of the memory mapping.
1182   const uint8_t* const begin_;
1183 
1184   // The size of the underlying memory allocation in bytes.
1185   const size_t size_;
1186 
1187   // Typically the dex file name when available, alternatively some identifying string.
1188   //
1189   // The ClassLinker will use this to match DexFiles the boot class
1190   // path to DexCache::GetLocation when loading from an image.
1191   const std::string location_;
1192 
1193   const uint32_t location_checksum_;
1194 
1195   // Manages the underlying memory allocation.
1196   std::unique_ptr<MemMap> mem_map_;
1197 
1198   // Points to the header section.
1199   const Header* const header_;
1200 
1201   // Points to the base of the string identifier list.
1202   const StringId* const string_ids_;
1203 
1204   // Points to the base of the type identifier list.
1205   const TypeId* const type_ids_;
1206 
1207   // Points to the base of the field identifier list.
1208   const FieldId* const field_ids_;
1209 
1210   // Points to the base of the method identifier list.
1211   const MethodId* const method_ids_;
1212 
1213   // Points to the base of the prototype identifier list.
1214   const ProtoId* const proto_ids_;
1215 
1216   // Points to the base of the class definition list.
1217   const ClassDef* const class_defs_;
1218 
1219   // Points to the base of the method handles list.
1220   const MethodHandleItem* method_handles_;
1221 
1222   // Number of elements in the method handles list.
1223   size_t num_method_handles_;
1224 
1225   // Points to the base of the call sites id list.
1226   const CallSiteIdItem* call_site_ids_;
1227 
1228   // Number of elements in the call sites list.
1229   size_t num_call_site_ids_;
1230 
1231   // If this dex file was loaded from an oat file, oat_dex_file_ contains a
1232   // pointer to the OatDexFile it was loaded from. Otherwise oat_dex_file_ is
1233   // null.
1234   mutable const OatDexFile* oat_dex_file_;
1235 
1236   friend class DexFileVerifierTest;
1237   friend class OatWriter;
1238   ART_FRIEND_TEST(ClassLinkerTest, RegisterDexFileName);  // for constructor
1239 };
1240 
1241 struct DexFileReference {
DexFileReferenceDexFileReference1242   DexFileReference(const DexFile* file, uint32_t idx) : dex_file(file), index(idx) { }
1243   const DexFile* dex_file;
1244   uint32_t index;
1245 };
1246 
1247 std::ostream& operator<<(std::ostream& os, const DexFile& dex_file);
1248 
1249 // Iterate over a dex file's ProtoId's paramters
1250 class DexFileParameterIterator {
1251  public:
DexFileParameterIterator(const DexFile & dex_file,const DexFile::ProtoId & proto_id)1252   DexFileParameterIterator(const DexFile& dex_file, const DexFile::ProtoId& proto_id)
1253       : dex_file_(dex_file), size_(0), pos_(0) {
1254     type_list_ = dex_file_.GetProtoParameters(proto_id);
1255     if (type_list_ != nullptr) {
1256       size_ = type_list_->Size();
1257     }
1258   }
HasNext()1259   bool HasNext() const { return pos_ < size_; }
Size()1260   size_t Size() const { return size_; }
Next()1261   void Next() { ++pos_; }
GetTypeIdx()1262   dex::TypeIndex GetTypeIdx() {
1263     return type_list_->GetTypeItem(pos_).type_idx_;
1264   }
GetDescriptor()1265   const char* GetDescriptor() {
1266     return dex_file_.StringByTypeIdx(dex::TypeIndex(GetTypeIdx()));
1267   }
1268  private:
1269   const DexFile& dex_file_;
1270   const DexFile::TypeList* type_list_;
1271   uint32_t size_;
1272   uint32_t pos_;
1273   DISALLOW_IMPLICIT_CONSTRUCTORS(DexFileParameterIterator);
1274 };
1275 
1276 // Abstract the signature of a method.
1277 class Signature : public ValueObject {
1278  public:
1279   std::string ToString() const;
1280 
NoSignature()1281   static Signature NoSignature() {
1282     return Signature();
1283   }
1284 
1285   bool IsVoid() const;
1286   uint32_t GetNumberOfParameters() const;
1287 
1288   bool operator==(const Signature& rhs) const;
1289   bool operator!=(const Signature& rhs) const {
1290     return !(*this == rhs);
1291   }
1292 
1293   bool operator==(const StringPiece& rhs) const;
1294 
1295  private:
Signature(const DexFile * dex,const DexFile::ProtoId & proto)1296   Signature(const DexFile* dex, const DexFile::ProtoId& proto) : dex_file_(dex), proto_id_(&proto) {
1297   }
1298 
Signature()1299   Signature() : dex_file_(nullptr), proto_id_(nullptr) {
1300   }
1301 
1302   friend class DexFile;
1303 
1304   const DexFile* const dex_file_;
1305   const DexFile::ProtoId* const proto_id_;
1306 };
1307 std::ostream& operator<<(std::ostream& os, const Signature& sig);
1308 
1309 // Iterate and decode class_data_item
1310 class ClassDataItemIterator {
1311  public:
ClassDataItemIterator(const DexFile & dex_file,const uint8_t * raw_class_data_item)1312   ClassDataItemIterator(const DexFile& dex_file, const uint8_t* raw_class_data_item)
1313       : dex_file_(dex_file), pos_(0), ptr_pos_(raw_class_data_item), last_idx_(0) {
1314     ReadClassDataHeader();
1315     if (EndOfInstanceFieldsPos() > 0) {
1316       ReadClassDataField();
1317     } else if (EndOfVirtualMethodsPos() > 0) {
1318       ReadClassDataMethod();
1319     }
1320   }
NumStaticFields()1321   uint32_t NumStaticFields() const {
1322     return header_.static_fields_size_;
1323   }
NumInstanceFields()1324   uint32_t NumInstanceFields() const {
1325     return header_.instance_fields_size_;
1326   }
NumDirectMethods()1327   uint32_t NumDirectMethods() const {
1328     return header_.direct_methods_size_;
1329   }
NumVirtualMethods()1330   uint32_t NumVirtualMethods() const {
1331     return header_.virtual_methods_size_;
1332   }
IsAtMethod()1333   bool IsAtMethod() const {
1334     return pos_ >= EndOfInstanceFieldsPos();
1335   }
HasNextStaticField()1336   bool HasNextStaticField() const {
1337     return pos_ < EndOfStaticFieldsPos();
1338   }
HasNextInstanceField()1339   bool HasNextInstanceField() const {
1340     return pos_ >= EndOfStaticFieldsPos() && pos_ < EndOfInstanceFieldsPos();
1341   }
HasNextDirectMethod()1342   bool HasNextDirectMethod() const {
1343     return pos_ >= EndOfInstanceFieldsPos() && pos_ < EndOfDirectMethodsPos();
1344   }
HasNextVirtualMethod()1345   bool HasNextVirtualMethod() const {
1346     return pos_ >= EndOfDirectMethodsPos() && pos_ < EndOfVirtualMethodsPos();
1347   }
SkipStaticFields()1348   void SkipStaticFields() {
1349     while (HasNextStaticField()) {
1350       Next();
1351     }
1352   }
SkipInstanceFields()1353   void SkipInstanceFields() {
1354     while (HasNextInstanceField()) {
1355       Next();
1356     }
1357   }
SkipAllFields()1358   void SkipAllFields() {
1359     SkipStaticFields();
1360     SkipInstanceFields();
1361   }
SkipDirectMethods()1362   void SkipDirectMethods() {
1363     while (HasNextDirectMethod()) {
1364       Next();
1365     }
1366   }
SkipVirtualMethods()1367   void SkipVirtualMethods() {
1368     while (HasNextVirtualMethod()) {
1369       Next();
1370     }
1371   }
HasNext()1372   bool HasNext() const {
1373     return pos_ < EndOfVirtualMethodsPos();
1374   }
Next()1375   inline void Next() {
1376     pos_++;
1377     if (pos_ < EndOfStaticFieldsPos()) {
1378       last_idx_ = GetMemberIndex();
1379       ReadClassDataField();
1380     } else if (pos_ == EndOfStaticFieldsPos() && NumInstanceFields() > 0) {
1381       last_idx_ = 0;  // transition to next array, reset last index
1382       ReadClassDataField();
1383     } else if (pos_ < EndOfInstanceFieldsPos()) {
1384       last_idx_ = GetMemberIndex();
1385       ReadClassDataField();
1386     } else if (pos_ == EndOfInstanceFieldsPos() && NumDirectMethods() > 0) {
1387       last_idx_ = 0;  // transition to next array, reset last index
1388       ReadClassDataMethod();
1389     } else if (pos_ < EndOfDirectMethodsPos()) {
1390       last_idx_ = GetMemberIndex();
1391       ReadClassDataMethod();
1392     } else if (pos_ == EndOfDirectMethodsPos() && NumVirtualMethods() > 0) {
1393       last_idx_ = 0;  // transition to next array, reset last index
1394       ReadClassDataMethod();
1395     } else if (pos_ < EndOfVirtualMethodsPos()) {
1396       last_idx_ = GetMemberIndex();
1397       ReadClassDataMethod();
1398     } else {
1399       DCHECK(!HasNext());
1400     }
1401   }
GetMemberIndex()1402   uint32_t GetMemberIndex() const {
1403     if (pos_ < EndOfInstanceFieldsPos()) {
1404       return last_idx_ + field_.field_idx_delta_;
1405     } else {
1406       DCHECK_LT(pos_, EndOfVirtualMethodsPos());
1407       return last_idx_ + method_.method_idx_delta_;
1408     }
1409   }
GetRawMemberAccessFlags()1410   uint32_t GetRawMemberAccessFlags() const {
1411     if (pos_ < EndOfInstanceFieldsPos()) {
1412       return field_.access_flags_;
1413     } else {
1414       DCHECK_LT(pos_, EndOfVirtualMethodsPos());
1415       return method_.access_flags_;
1416     }
1417   }
GetFieldAccessFlags()1418   uint32_t GetFieldAccessFlags() const {
1419     return GetRawMemberAccessFlags() & kAccValidFieldFlags;
1420   }
GetMethodAccessFlags()1421   uint32_t GetMethodAccessFlags() const {
1422     return GetRawMemberAccessFlags() & kAccValidMethodFlags;
1423   }
MemberIsNative()1424   bool MemberIsNative() const {
1425     return GetRawMemberAccessFlags() & kAccNative;
1426   }
MemberIsFinal()1427   bool MemberIsFinal() const {
1428     return GetRawMemberAccessFlags() & kAccFinal;
1429   }
GetMethodInvokeType(const DexFile::ClassDef & class_def)1430   InvokeType GetMethodInvokeType(const DexFile::ClassDef& class_def) const {
1431     if (HasNextDirectMethod()) {
1432       if ((GetRawMemberAccessFlags() & kAccStatic) != 0) {
1433         return kStatic;
1434       } else {
1435         return kDirect;
1436       }
1437     } else {
1438       DCHECK_EQ(GetRawMemberAccessFlags() & kAccStatic, 0U);
1439       if ((class_def.access_flags_ & kAccInterface) != 0) {
1440         return kInterface;
1441       } else if ((GetRawMemberAccessFlags() & kAccConstructor) != 0) {
1442         return kSuper;
1443       } else {
1444         return kVirtual;
1445       }
1446     }
1447   }
GetMethodCodeItem()1448   const DexFile::CodeItem* GetMethodCodeItem() const {
1449     return dex_file_.GetCodeItem(method_.code_off_);
1450   }
GetMethodCodeItemOffset()1451   uint32_t GetMethodCodeItemOffset() const {
1452     return method_.code_off_;
1453   }
DataPointer()1454   const uint8_t* DataPointer() const {
1455     return ptr_pos_;
1456   }
EndDataPointer()1457   const uint8_t* EndDataPointer() const {
1458     CHECK(!HasNext());
1459     return ptr_pos_;
1460   }
1461 
1462  private:
1463   // A dex file's class_data_item is leb128 encoded, this structure holds a decoded form of the
1464   // header for a class_data_item
1465   struct ClassDataHeader {
1466     uint32_t static_fields_size_;  // the number of static fields
1467     uint32_t instance_fields_size_;  // the number of instance fields
1468     uint32_t direct_methods_size_;  // the number of direct methods
1469     uint32_t virtual_methods_size_;  // the number of virtual methods
1470   } header_;
1471 
1472   // Read and decode header from a class_data_item stream into header
1473   void ReadClassDataHeader();
1474 
EndOfStaticFieldsPos()1475   uint32_t EndOfStaticFieldsPos() const {
1476     return header_.static_fields_size_;
1477   }
EndOfInstanceFieldsPos()1478   uint32_t EndOfInstanceFieldsPos() const {
1479     return EndOfStaticFieldsPos() + header_.instance_fields_size_;
1480   }
EndOfDirectMethodsPos()1481   uint32_t EndOfDirectMethodsPos() const {
1482     return EndOfInstanceFieldsPos() + header_.direct_methods_size_;
1483   }
EndOfVirtualMethodsPos()1484   uint32_t EndOfVirtualMethodsPos() const {
1485     return EndOfDirectMethodsPos() + header_.virtual_methods_size_;
1486   }
1487 
1488   // A decoded version of the field of a class_data_item
1489   struct ClassDataField {
1490     uint32_t field_idx_delta_;  // delta of index into the field_ids array for FieldId
1491     uint32_t access_flags_;  // access flags for the field
ClassDataFieldClassDataField1492     ClassDataField() :  field_idx_delta_(0), access_flags_(0) {}
1493 
1494    private:
1495     DISALLOW_COPY_AND_ASSIGN(ClassDataField);
1496   };
1497   ClassDataField field_;
1498 
1499   // Read and decode a field from a class_data_item stream into field
1500   void ReadClassDataField();
1501 
1502   // A decoded version of the method of a class_data_item
1503   struct ClassDataMethod {
1504     uint32_t method_idx_delta_;  // delta of index into the method_ids array for MethodId
1505     uint32_t access_flags_;
1506     uint32_t code_off_;
ClassDataMethodClassDataMethod1507     ClassDataMethod() : method_idx_delta_(0), access_flags_(0), code_off_(0) {}
1508 
1509    private:
1510     DISALLOW_COPY_AND_ASSIGN(ClassDataMethod);
1511   };
1512   ClassDataMethod method_;
1513 
1514   // Read and decode a method from a class_data_item stream into method
1515   void ReadClassDataMethod();
1516 
1517   const DexFile& dex_file_;
1518   size_t pos_;  // integral number of items passed
1519   const uint8_t* ptr_pos_;  // pointer into stream of class_data_item
1520   uint32_t last_idx_;  // last read field or method index to apply delta to
1521   DISALLOW_IMPLICIT_CONSTRUCTORS(ClassDataItemIterator);
1522 };
1523 
1524 class EncodedArrayValueIterator {
1525  public:
1526   EncodedArrayValueIterator(const DexFile& dex_file, const uint8_t* array_data);
1527 
HasNext()1528   bool HasNext() const { return pos_ < array_size_; }
1529 
1530   void Next();
1531 
1532   enum ValueType {
1533     kByte         = 0x00,
1534     kShort        = 0x02,
1535     kChar         = 0x03,
1536     kInt          = 0x04,
1537     kLong         = 0x06,
1538     kFloat        = 0x10,
1539     kDouble       = 0x11,
1540     kMethodType   = 0x15,
1541     kMethodHandle = 0x16,
1542     kString       = 0x17,
1543     kType         = 0x18,
1544     kField        = 0x19,
1545     kMethod       = 0x1a,
1546     kEnum         = 0x1b,
1547     kArray        = 0x1c,
1548     kAnnotation   = 0x1d,
1549     kNull         = 0x1e,
1550     kBoolean      = 0x1f,
1551   };
1552 
GetValueType()1553   ValueType GetValueType() const { return type_; }
GetJavaValue()1554   const jvalue& GetJavaValue() const { return jval_; }
1555 
1556  protected:
1557   static constexpr uint8_t kEncodedValueTypeMask = 0x1f;  // 0b11111
1558   static constexpr uint8_t kEncodedValueArgShift = 5;
1559 
1560   const DexFile& dex_file_;
1561   size_t array_size_;  // Size of array.
1562   size_t pos_;  // Current position.
1563   const uint8_t* ptr_;  // Pointer into encoded data array.
1564   ValueType type_;  // Type of current encoded value.
1565   jvalue jval_;  // Value of current encoded value.
1566 
1567  private:
1568   DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedArrayValueIterator);
1569 };
1570 std::ostream& operator<<(std::ostream& os, const EncodedArrayValueIterator::ValueType& code);
1571 
1572 class EncodedStaticFieldValueIterator : public EncodedArrayValueIterator {
1573  public:
EncodedStaticFieldValueIterator(const DexFile & dex_file,const DexFile::ClassDef & class_def)1574   EncodedStaticFieldValueIterator(const DexFile& dex_file,
1575                                   const DexFile::ClassDef& class_def)
1576       : EncodedArrayValueIterator(dex_file,
1577                                   dex_file.GetEncodedStaticFieldValuesArray(class_def))
1578   {}
1579 
1580  private:
1581   DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedStaticFieldValueIterator);
1582 };
1583 std::ostream& operator<<(std::ostream& os, const EncodedStaticFieldValueIterator::ValueType& code);
1584 
1585 class CallSiteArrayValueIterator : public EncodedArrayValueIterator {
1586  public:
CallSiteArrayValueIterator(const DexFile & dex_file,const DexFile::CallSiteIdItem & call_site_id)1587   CallSiteArrayValueIterator(const DexFile& dex_file,
1588                              const DexFile::CallSiteIdItem& call_site_id)
1589       : EncodedArrayValueIterator(dex_file,
1590                                   dex_file.GetCallSiteEncodedValuesArray(call_site_id))
1591   {}
1592 
Size()1593   uint32_t Size() const { return array_size_; }
1594 
1595  private:
1596   DISALLOW_IMPLICIT_CONSTRUCTORS(CallSiteArrayValueIterator);
1597 };
1598 std::ostream& operator<<(std::ostream& os, const CallSiteArrayValueIterator::ValueType& code);
1599 
1600 class CatchHandlerIterator {
1601   public:
1602     CatchHandlerIterator(const DexFile::CodeItem& code_item, uint32_t address);
1603 
1604     CatchHandlerIterator(const DexFile::CodeItem& code_item,
1605                          const DexFile::TryItem& try_item);
1606 
CatchHandlerIterator(const uint8_t * handler_data)1607     explicit CatchHandlerIterator(const uint8_t* handler_data) {
1608       Init(handler_data);
1609     }
1610 
GetHandlerTypeIndex()1611     dex::TypeIndex GetHandlerTypeIndex() const {
1612       return handler_.type_idx_;
1613     }
GetHandlerAddress()1614     uint32_t GetHandlerAddress() const {
1615       return handler_.address_;
1616     }
1617     void Next();
HasNext()1618     bool HasNext() const {
1619       return remaining_count_ != -1 || catch_all_;
1620     }
1621     // End of this set of catch blocks, convenience method to locate next set of catch blocks
EndDataPointer()1622     const uint8_t* EndDataPointer() const {
1623       CHECK(!HasNext());
1624       return current_data_;
1625     }
1626 
1627   private:
1628     void Init(const DexFile::CodeItem& code_item, int32_t offset);
1629     void Init(const uint8_t* handler_data);
1630 
1631     struct CatchHandlerItem {
1632       dex::TypeIndex type_idx_;  // type index of the caught exception type
1633       uint32_t address_;  // handler address
1634     } handler_;
1635     const uint8_t* current_data_;  // the current handler in dex file.
1636     int32_t remaining_count_;   // number of handlers not read.
1637     bool catch_all_;            // is there a handler that will catch all exceptions in case
1638                                 // that all typed handler does not match.
1639 };
1640 
1641 }  // namespace art
1642 
1643 #endif  // ART_RUNTIME_DEX_FILE_H_
1644