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_LIBDEXFILE_DEX_DEX_FILE_H_ 18 #define ART_LIBDEXFILE_DEX_DEX_FILE_H_ 19 20 #include <memory> 21 #include <optional> 22 #include <string> 23 #include <string_view> 24 #include <vector> 25 26 #include <android-base/logging.h> 27 28 #include "base/globals.h" 29 #include "base/macros.h" 30 #include "base/value_object.h" 31 #include "dex_file_structs.h" 32 #include "dex_file_types.h" 33 #include "jni.h" 34 #include "modifiers.h" 35 36 namespace art { 37 38 class ClassDataItemIterator; 39 class ClassIterator; 40 class CompactDexFile; 41 class DexInstructionIterator; 42 enum InvokeType : uint32_t; 43 template <typename Iter> class IterationRange; 44 class MemMap; 45 class OatDexFile; 46 class Signature; 47 class StandardDexFile; 48 class ZipArchive; 49 50 namespace hiddenapi { 51 enum class Domain : char; 52 } // namespace hiddenapi 53 54 // Some instances of DexFile own the storage referred to by DexFile. Clients who create 55 // such management do so by subclassing Container. 56 class DexFileContainer { 57 public: DexFileContainer()58 DexFileContainer() { } ~DexFileContainer()59 virtual ~DexFileContainer() { } 60 virtual int GetPermissions() = 0; 61 virtual bool IsReadOnly() = 0; 62 virtual bool EnableWrite() = 0; 63 virtual bool DisableWrite() = 0; 64 65 private: 66 DISALLOW_COPY_AND_ASSIGN(DexFileContainer); 67 }; 68 69 // Dex file is the API that exposes native dex files (ordinary dex files) and CompactDex. 70 // Originally, the dex file format used by ART was mostly the same as APKs. The only change was 71 // quickened opcodes and layout optimizations. 72 // Since ART needs to support both native dex files and CompactDex files, the DexFile interface 73 // provides an abstraction to facilitate this. 74 class DexFile { 75 public: 76 // Number of bytes in the dex file magic. 77 static constexpr size_t kDexMagicSize = 4; 78 static constexpr size_t kDexVersionLen = 4; 79 80 // First Dex format version enforcing class definition ordering rules. 81 static constexpr uint32_t kClassDefinitionOrderEnforcedVersion = 37; 82 83 static constexpr size_t kSha1DigestSize = 20; 84 static constexpr uint32_t kDexEndianConstant = 0x12345678; 85 86 // The value of an invalid index. 87 static constexpr uint16_t kDexNoIndex16 = 0xFFFF; 88 static constexpr uint32_t kDexNoIndex32 = 0xFFFFFFFF; 89 90 // Raw header_item. 91 struct Header { 92 uint8_t magic_[8] = {}; 93 uint32_t checksum_ = 0; // See also location_checksum_ 94 uint8_t signature_[kSha1DigestSize] = {}; 95 uint32_t file_size_ = 0; // size of entire file 96 uint32_t header_size_ = 0; // offset to start of next section 97 uint32_t endian_tag_ = 0; 98 uint32_t link_size_ = 0; // unused 99 uint32_t link_off_ = 0; // unused 100 uint32_t map_off_ = 0; // map list offset from data_off_ 101 uint32_t string_ids_size_ = 0; // number of StringIds 102 uint32_t string_ids_off_ = 0; // file offset of StringIds array 103 uint32_t type_ids_size_ = 0; // number of TypeIds, we don't support more than 65535 104 uint32_t type_ids_off_ = 0; // file offset of TypeIds array 105 uint32_t proto_ids_size_ = 0; // number of ProtoIds, we don't support more than 65535 106 uint32_t proto_ids_off_ = 0; // file offset of ProtoIds array 107 uint32_t field_ids_size_ = 0; // number of FieldIds 108 uint32_t field_ids_off_ = 0; // file offset of FieldIds array 109 uint32_t method_ids_size_ = 0; // number of MethodIds 110 uint32_t method_ids_off_ = 0; // file offset of MethodIds array 111 uint32_t class_defs_size_ = 0; // number of ClassDefs 112 uint32_t class_defs_off_ = 0; // file offset of ClassDef array 113 uint32_t data_size_ = 0; // size of data section 114 uint32_t data_off_ = 0; // file offset of data section 115 116 // Decode the dex magic version 117 uint32_t GetVersion() const; 118 }; 119 120 // Map item type codes. 121 enum MapItemType : uint16_t { // private 122 kDexTypeHeaderItem = 0x0000, 123 kDexTypeStringIdItem = 0x0001, 124 kDexTypeTypeIdItem = 0x0002, 125 kDexTypeProtoIdItem = 0x0003, 126 kDexTypeFieldIdItem = 0x0004, 127 kDexTypeMethodIdItem = 0x0005, 128 kDexTypeClassDefItem = 0x0006, 129 kDexTypeCallSiteIdItem = 0x0007, 130 kDexTypeMethodHandleItem = 0x0008, 131 kDexTypeMapList = 0x1000, 132 kDexTypeTypeList = 0x1001, 133 kDexTypeAnnotationSetRefList = 0x1002, 134 kDexTypeAnnotationSetItem = 0x1003, 135 kDexTypeClassDataItem = 0x2000, 136 kDexTypeCodeItem = 0x2001, 137 kDexTypeStringDataItem = 0x2002, 138 kDexTypeDebugInfoItem = 0x2003, 139 kDexTypeAnnotationItem = 0x2004, 140 kDexTypeEncodedArrayItem = 0x2005, 141 kDexTypeAnnotationsDirectoryItem = 0x2006, 142 kDexTypeHiddenapiClassData = 0xF000, 143 }; 144 145 // MethodHandle Types 146 enum class MethodHandleType : uint16_t { // private 147 kStaticPut = 0x0000, // a setter for a given static field. 148 kStaticGet = 0x0001, // a getter for a given static field. 149 kInstancePut = 0x0002, // a setter for a given instance field. 150 kInstanceGet = 0x0003, // a getter for a given instance field. 151 kInvokeStatic = 0x0004, // an invoker for a given static method. 152 kInvokeInstance = 0x0005, // invoke_instance : an invoker for a given instance method. This 153 // can be any non-static method on any class (or interface) except 154 // for “<init>”. 155 kInvokeConstructor = 0x0006, // an invoker for a given constructor. 156 kInvokeDirect = 0x0007, // an invoker for a direct (special) method. 157 kInvokeInterface = 0x0008, // an invoker for an interface method. 158 kLast = kInvokeInterface 159 }; 160 161 // Annotation constants. 162 enum { 163 kDexVisibilityBuild = 0x00, /* annotation visibility */ 164 kDexVisibilityRuntime = 0x01, 165 kDexVisibilitySystem = 0x02, 166 167 kDexAnnotationByte = 0x00, 168 kDexAnnotationShort = 0x02, 169 kDexAnnotationChar = 0x03, 170 kDexAnnotationInt = 0x04, 171 kDexAnnotationLong = 0x06, 172 kDexAnnotationFloat = 0x10, 173 kDexAnnotationDouble = 0x11, 174 kDexAnnotationMethodType = 0x15, 175 kDexAnnotationMethodHandle = 0x16, 176 kDexAnnotationString = 0x17, 177 kDexAnnotationType = 0x18, 178 kDexAnnotationField = 0x19, 179 kDexAnnotationMethod = 0x1a, 180 kDexAnnotationEnum = 0x1b, 181 kDexAnnotationArray = 0x1c, 182 kDexAnnotationAnnotation = 0x1d, 183 kDexAnnotationNull = 0x1e, 184 kDexAnnotationBoolean = 0x1f, 185 186 kDexAnnotationValueTypeMask = 0x1f, /* low 5 bits */ 187 kDexAnnotationValueArgShift = 5, 188 }; 189 190 enum AnnotationResultStyle { // private 191 kAllObjects, 192 kPrimitivesOrObjects, 193 kAllRaw 194 }; 195 196 struct AnnotationValue; 197 198 // Closes a .dex file. 199 virtual ~DexFile(); 200 GetLocation()201 const std::string& GetLocation() const { 202 return location_; 203 } 204 205 // For DexFiles directly from .dex files, this is the checksum from the DexFile::Header. 206 // For DexFiles opened from a zip files, this will be the ZipEntry CRC32 of classes.dex. GetLocationChecksum()207 uint32_t GetLocationChecksum() const { 208 return location_checksum_; 209 } 210 GetHeader()211 const Header& GetHeader() const { 212 DCHECK(header_ != nullptr) << GetLocation(); 213 return *header_; 214 } 215 216 // Decode the dex magic version GetDexVersion()217 uint32_t GetDexVersion() const { 218 return GetHeader().GetVersion(); 219 } 220 221 // Returns true if the byte string points to the magic value. 222 virtual bool IsMagicValid() const = 0; 223 224 // Returns true if the byte string after the magic is the correct value. 225 virtual bool IsVersionValid() const = 0; 226 227 // Returns true if the dex file supports default methods. 228 virtual bool SupportsDefaultMethods() const = 0; 229 230 // Returns the maximum size in bytes needed to store an equivalent dex file strictly conforming to 231 // the dex file specification. That is the size if we wanted to get rid of all the 232 // quickening/compact-dexing/etc. 233 // 234 // TODO This should really be an exact size! b/72402467 235 virtual size_t GetDequickenedSize() const = 0; 236 237 // Returns the number of string identifiers in the .dex file. NumStringIds()238 size_t NumStringIds() const { 239 DCHECK(header_ != nullptr) << GetLocation(); 240 return header_->string_ids_size_; 241 } 242 243 // Returns the StringId at the specified index. GetStringId(dex::StringIndex idx)244 const dex::StringId& GetStringId(dex::StringIndex idx) const { 245 DCHECK_LT(idx.index_, NumStringIds()) << GetLocation(); 246 return string_ids_[idx.index_]; 247 } 248 GetIndexForStringId(const dex::StringId & string_id)249 dex::StringIndex GetIndexForStringId(const dex::StringId& string_id) const { 250 CHECK_GE(&string_id, string_ids_) << GetLocation(); 251 CHECK_LT(&string_id, string_ids_ + header_->string_ids_size_) << GetLocation(); 252 return dex::StringIndex(&string_id - string_ids_); 253 } 254 255 int32_t GetStringLength(const dex::StringId& string_id) const; 256 257 // Returns a pointer to the UTF-8 string data referred to by the given string_id as well as the 258 // length of the string when decoded as a UTF-16 string. Note the UTF-16 length is not the same 259 // as the string length of the string data. 260 const char* GetStringDataAndUtf16Length(const dex::StringId& string_id, 261 uint32_t* utf16_length) const; 262 263 const char* GetStringData(const dex::StringId& string_id) const; 264 265 // Index version of GetStringDataAndUtf16Length. 266 const char* StringDataAndUtf16LengthByIdx(dex::StringIndex idx, uint32_t* utf16_length) const; 267 268 const char* StringDataByIdx(dex::StringIndex idx) const; 269 std::string_view StringViewByIdx(dex::StringIndex idx) const; 270 271 // Looks up a string id for a given modified utf8 string. 272 const dex::StringId* FindStringId(const char* string) const; 273 274 const dex::TypeId* FindTypeId(const char* string) const; FindTypeId(std::string_view string)275 const dex::TypeId* FindTypeId(std::string_view string) const { 276 return FindTypeId(std::string(string).c_str()); 277 } 278 279 // Returns the number of type identifiers in the .dex file. NumTypeIds()280 uint32_t NumTypeIds() const { 281 DCHECK(header_ != nullptr) << GetLocation(); 282 return header_->type_ids_size_; 283 } 284 IsTypeIndexValid(dex::TypeIndex idx)285 bool IsTypeIndexValid(dex::TypeIndex idx) const { 286 return idx.IsValid() && idx.index_ < NumTypeIds(); 287 } 288 289 // Returns the TypeId at the specified index. GetTypeId(dex::TypeIndex idx)290 const dex::TypeId& GetTypeId(dex::TypeIndex idx) const { 291 DCHECK_LT(idx.index_, NumTypeIds()) << GetLocation(); 292 return type_ids_[idx.index_]; 293 } 294 GetIndexForTypeId(const dex::TypeId & type_id)295 dex::TypeIndex GetIndexForTypeId(const dex::TypeId& type_id) const { 296 CHECK_GE(&type_id, type_ids_) << GetLocation(); 297 CHECK_LT(&type_id, type_ids_ + header_->type_ids_size_) << GetLocation(); 298 size_t result = &type_id - type_ids_; 299 DCHECK_LT(result, 65536U) << GetLocation(); 300 return dex::TypeIndex(static_cast<uint16_t>(result)); 301 } 302 303 // Get the descriptor string associated with a given type index. 304 const char* StringByTypeIdx(dex::TypeIndex idx, uint32_t* unicode_length) const; 305 306 const char* StringByTypeIdx(dex::TypeIndex idx) const; 307 308 // Returns the type descriptor string of a type id. 309 const char* GetTypeDescriptor(const dex::TypeId& type_id) const; 310 std::string_view GetTypeDescriptorView(const dex::TypeId& type_id) const; 311 312 // Looks up a type for the given string index 313 const dex::TypeId* FindTypeId(dex::StringIndex string_idx) const; 314 315 // Returns the number of field identifiers in the .dex file. NumFieldIds()316 size_t NumFieldIds() const { 317 DCHECK(header_ != nullptr) << GetLocation(); 318 return header_->field_ids_size_; 319 } 320 321 // Returns the FieldId at the specified index. GetFieldId(uint32_t idx)322 const dex::FieldId& GetFieldId(uint32_t idx) const { 323 DCHECK_LT(idx, NumFieldIds()) << GetLocation(); 324 return field_ids_[idx]; 325 } 326 GetIndexForFieldId(const dex::FieldId & field_id)327 uint32_t GetIndexForFieldId(const dex::FieldId& field_id) const { 328 CHECK_GE(&field_id, field_ids_) << GetLocation(); 329 CHECK_LT(&field_id, field_ids_ + header_->field_ids_size_) << GetLocation(); 330 return &field_id - field_ids_; 331 } 332 333 // Looks up a field by its declaring class, name and type 334 const dex::FieldId* FindFieldId(const dex::TypeId& declaring_klass, 335 const dex::StringId& name, 336 const dex::TypeId& type) const; 337 338 // Return the code-item offset associated with the class and method or nullopt 339 // if the method does not exist or has no code. 340 std::optional<uint32_t> GetCodeItemOffset(const dex::ClassDef& class_def, 341 uint32_t dex_method_idx) const; 342 343 // Return the code-item offset associated with the class and method or 344 // LOG(FATAL) if the method does not exist or has no code. 345 uint32_t FindCodeItemOffset(const dex::ClassDef& class_def, 346 uint32_t dex_method_idx) const; 347 348 virtual uint32_t GetCodeItemSize(const dex::CodeItem& disk_code_item) const = 0; 349 350 // Returns the declaring class descriptor string of a field id. GetFieldDeclaringClassDescriptor(const dex::FieldId & field_id)351 const char* GetFieldDeclaringClassDescriptor(const dex::FieldId& field_id) const { 352 const dex::TypeId& type_id = GetTypeId(field_id.class_idx_); 353 return GetTypeDescriptor(type_id); 354 } 355 356 // Returns the class descriptor string of a field id. 357 const char* GetFieldTypeDescriptor(const dex::FieldId& field_id) const; 358 std::string_view GetFieldTypeDescriptorView(const dex::FieldId& field_id) const; 359 360 // Returns the name of a field id. 361 const char* GetFieldName(const dex::FieldId& field_id) const; 362 std::string_view GetFieldNameView(const dex::FieldId& field_id) const; 363 364 // Returns the number of method identifiers in the .dex file. NumMethodIds()365 size_t NumMethodIds() const { 366 DCHECK(header_ != nullptr) << GetLocation(); 367 return header_->method_ids_size_; 368 } 369 370 // Returns the MethodId at the specified index. GetMethodId(uint32_t idx)371 const dex::MethodId& GetMethodId(uint32_t idx) const { 372 DCHECK_LT(idx, NumMethodIds()) << GetLocation(); 373 return method_ids_[idx]; 374 } 375 GetIndexForMethodId(const dex::MethodId & method_id)376 uint32_t GetIndexForMethodId(const dex::MethodId& method_id) const { 377 CHECK_GE(&method_id, method_ids_) << GetLocation(); 378 CHECK_LT(&method_id, method_ids_ + header_->method_ids_size_) << GetLocation(); 379 return &method_id - method_ids_; 380 } 381 382 // Looks up a method by its declaring class, name and proto_id 383 const dex::MethodId* FindMethodId(const dex::TypeId& declaring_klass, 384 const dex::StringId& name, 385 const dex::ProtoId& signature) const; 386 387 const dex::MethodId* FindMethodIdByIndex(dex::TypeIndex declaring_klass, 388 dex::StringIndex name, 389 dex::ProtoIndex signature) const; 390 391 // Returns the declaring class descriptor string of a method id. 392 const char* GetMethodDeclaringClassDescriptor(const dex::MethodId& method_id) const; 393 394 // Returns the prototype of a method id. GetMethodPrototype(const dex::MethodId & method_id)395 const dex::ProtoId& GetMethodPrototype(const dex::MethodId& method_id) const { 396 return GetProtoId(method_id.proto_idx_); 397 } 398 399 // Returns a representation of the signature of a method id. 400 const Signature GetMethodSignature(const dex::MethodId& method_id) const; 401 402 // Returns a representation of the signature of a proto id. 403 const Signature GetProtoSignature(const dex::ProtoId& proto_id) const; 404 405 // Returns the name of a method id. 406 const char* GetMethodName(const dex::MethodId& method_id) const; 407 const char* GetMethodName(const dex::MethodId& method_id, uint32_t* utf_length) const; 408 const char* GetMethodName(uint32_t idx) const; 409 const char* GetMethodName(uint32_t idx, uint32_t* utf_length) const; 410 std::string_view GetMethodNameView(const dex::MethodId& method_id) const; 411 std::string_view GetMethodNameView(uint32_t idx) const; 412 413 // Returns the shorty of a method by its index. 414 const char* GetMethodShorty(uint32_t idx) const; 415 416 // Returns the shorty of a method id. 417 const char* GetMethodShorty(const dex::MethodId& method_id) const; 418 const char* GetMethodShorty(const dex::MethodId& method_id, uint32_t* length) const; 419 420 // Returns the number of class definitions in the .dex file. NumClassDefs()421 uint32_t NumClassDefs() const { 422 DCHECK(header_ != nullptr) << GetLocation(); 423 return header_->class_defs_size_; 424 } 425 426 // Returns the ClassDef at the specified index. GetClassDef(uint16_t idx)427 const dex::ClassDef& GetClassDef(uint16_t idx) const { 428 DCHECK_LT(idx, NumClassDefs()) << GetLocation(); 429 return class_defs_[idx]; 430 } 431 GetIndexForClassDef(const dex::ClassDef & class_def)432 uint16_t GetIndexForClassDef(const dex::ClassDef& class_def) const { 433 CHECK_GE(&class_def, class_defs_) << GetLocation(); 434 CHECK_LT(&class_def, class_defs_ + header_->class_defs_size_) << GetLocation(); 435 return &class_def - class_defs_; 436 } 437 438 // Returns the class descriptor string of a class definition. 439 const char* GetClassDescriptor(const dex::ClassDef& class_def) const; 440 441 // Looks up a class definition by its type index. 442 const dex::ClassDef* FindClassDef(dex::TypeIndex type_idx) const; 443 GetInterfacesList(const dex::ClassDef & class_def)444 const dex::TypeList* GetInterfacesList(const dex::ClassDef& class_def) const { 445 return DataPointer<dex::TypeList>(class_def.interfaces_off_); 446 } 447 NumMethodHandles()448 uint32_t NumMethodHandles() const { 449 return num_method_handles_; 450 } 451 GetMethodHandle(uint32_t idx)452 const dex::MethodHandleItem& GetMethodHandle(uint32_t idx) const { 453 CHECK_LT(idx, NumMethodHandles()); 454 return method_handles_[idx]; 455 } 456 NumCallSiteIds()457 uint32_t NumCallSiteIds() const { 458 return num_call_site_ids_; 459 } 460 GetCallSiteId(uint32_t idx)461 const dex::CallSiteIdItem& GetCallSiteId(uint32_t idx) const { 462 CHECK_LT(idx, NumCallSiteIds()); 463 return call_site_ids_[idx]; 464 } 465 466 // Returns a pointer to the raw memory mapped class_data_item GetClassData(const dex::ClassDef & class_def)467 const uint8_t* GetClassData(const dex::ClassDef& class_def) const { 468 return DataPointer<uint8_t>(class_def.class_data_off_); 469 } 470 471 // Return the code item for a provided offset. GetCodeItem(const uint32_t code_off)472 const dex::CodeItem* GetCodeItem(const uint32_t code_off) const { 473 // May be null for native or abstract methods. 474 return DataPointer<dex::CodeItem>(code_off); 475 } 476 477 const char* GetReturnTypeDescriptor(const dex::ProtoId& proto_id) const; 478 479 // Returns the number of prototype identifiers in the .dex file. NumProtoIds()480 size_t NumProtoIds() const { 481 DCHECK(header_ != nullptr) << GetLocation(); 482 return header_->proto_ids_size_; 483 } 484 485 // Returns the ProtoId at the specified index. GetProtoId(dex::ProtoIndex idx)486 const dex::ProtoId& GetProtoId(dex::ProtoIndex idx) const { 487 DCHECK_LT(idx.index_, NumProtoIds()) << GetLocation(); 488 return proto_ids_[idx.index_]; 489 } 490 GetIndexForProtoId(const dex::ProtoId & proto_id)491 dex::ProtoIndex GetIndexForProtoId(const dex::ProtoId& proto_id) const { 492 CHECK_GE(&proto_id, proto_ids_) << GetLocation(); 493 CHECK_LT(&proto_id, proto_ids_ + header_->proto_ids_size_) << GetLocation(); 494 return dex::ProtoIndex(&proto_id - proto_ids_); 495 } 496 497 // Looks up a proto id for a given return type and signature type list 498 const dex::ProtoId* FindProtoId(dex::TypeIndex return_type_idx, 499 const dex::TypeIndex* signature_type_idxs, 500 uint32_t signature_length) const; FindProtoId(dex::TypeIndex return_type_idx,const std::vector<dex::TypeIndex> & signature_type_idxs)501 const dex::ProtoId* FindProtoId(dex::TypeIndex return_type_idx, 502 const std::vector<dex::TypeIndex>& signature_type_idxs) const { 503 return FindProtoId(return_type_idx, &signature_type_idxs[0], signature_type_idxs.size()); 504 } 505 506 // Given a signature place the type ids into the given vector, returns true on success 507 bool CreateTypeList(std::string_view signature, 508 dex::TypeIndex* return_type_idx, 509 std::vector<dex::TypeIndex>* param_type_idxs) const; 510 511 // Returns the short form method descriptor for the given prototype. 512 const char* GetShorty(dex::ProtoIndex proto_idx) const; 513 std::string_view GetShortyView(const dex::ProtoId& proto_id) const; 514 GetProtoParameters(const dex::ProtoId & proto_id)515 const dex::TypeList* GetProtoParameters(const dex::ProtoId& proto_id) const { 516 return DataPointer<dex::TypeList>(proto_id.parameters_off_); 517 } 518 GetEncodedStaticFieldValuesArray(const dex::ClassDef & class_def)519 const uint8_t* GetEncodedStaticFieldValuesArray(const dex::ClassDef& class_def) const { 520 return DataPointer<uint8_t>(class_def.static_values_off_); 521 } 522 GetCallSiteEncodedValuesArray(const dex::CallSiteIdItem & call_site_id)523 const uint8_t* GetCallSiteEncodedValuesArray(const dex::CallSiteIdItem& call_site_id) const { 524 return DataBegin() + call_site_id.data_off_; 525 } 526 527 dex::ProtoIndex GetProtoIndexForCallSite(uint32_t call_site_idx) const; 528 529 static const dex::TryItem* GetTryItems(const DexInstructionIterator& code_item_end, 530 uint32_t offset); 531 532 // Get the base of the encoded data for the given DexCode. 533 static const uint8_t* GetCatchHandlerData(const DexInstructionIterator& code_item_end, 534 uint32_t tries_size, 535 uint32_t offset); 536 537 // Find which try region is associated with the given address (ie dex pc). Returns -1 if none. 538 static int32_t FindTryItem(const dex::TryItem* try_items, uint32_t tries_size, uint32_t address); 539 540 // Get the pointer to the start of the debugging data GetDebugInfoStream(uint32_t debug_info_off)541 const uint8_t* GetDebugInfoStream(uint32_t debug_info_off) const { 542 // Check that the offset is in bounds. 543 // Note that although the specification says that 0 should be used if there 544 // is no debug information, some applications incorrectly use 0xFFFFFFFF. 545 return (debug_info_off == 0 || debug_info_off >= data_size_) 546 ? nullptr 547 : DataBegin() + debug_info_off; 548 } 549 550 struct PositionInfo { 551 PositionInfo() = default; 552 553 uint32_t address_ = 0; // In 16-bit code units. 554 uint32_t line_ = 0; // Source code line number starting at 1. 555 const char* source_file_ = nullptr; // nullptr if the file from ClassDef still applies. 556 bool prologue_end_ = false; 557 bool epilogue_begin_ = false; 558 }; 559 560 struct LocalInfo { 561 LocalInfo() = default; 562 563 const char* name_ = nullptr; // E.g., list. It can be nullptr if unknown. 564 const char* descriptor_ = nullptr; // E.g., Ljava/util/LinkedList; 565 const char* signature_ = nullptr; // E.g., java.util.LinkedList<java.lang.Integer> 566 uint32_t start_address_ = 0; // PC location where the local is first defined. 567 uint32_t end_address_ = 0; // PC location where the local is no longer defined. 568 uint16_t reg_ = 0; // Dex register which stores the values. 569 bool is_live_ = false; // Is the local defined and live. 570 }; 571 572 // Callback for "new locals table entry". 573 typedef void (*DexDebugNewLocalCb)(void* context, const LocalInfo& entry); 574 GetAnnotationsDirectory(const dex::ClassDef & class_def)575 const dex::AnnotationsDirectoryItem* GetAnnotationsDirectory(const dex::ClassDef& class_def) 576 const { 577 return DataPointer<dex::AnnotationsDirectoryItem>(class_def.annotations_off_); 578 } 579 GetClassAnnotationSet(const dex::AnnotationsDirectoryItem * anno_dir)580 const dex::AnnotationSetItem* GetClassAnnotationSet(const dex::AnnotationsDirectoryItem* anno_dir) 581 const { 582 return DataPointer<dex::AnnotationSetItem>(anno_dir->class_annotations_off_); 583 } 584 GetFieldAnnotations(const dex::AnnotationsDirectoryItem * anno_dir)585 const dex::FieldAnnotationsItem* GetFieldAnnotations( 586 const dex::AnnotationsDirectoryItem* anno_dir) const { 587 return (anno_dir->fields_size_ == 0) 588 ? nullptr 589 : reinterpret_cast<const dex::FieldAnnotationsItem*>(&anno_dir[1]); 590 } 591 GetMethodAnnotations(const dex::AnnotationsDirectoryItem * anno_dir)592 const dex::MethodAnnotationsItem* GetMethodAnnotations( 593 const dex::AnnotationsDirectoryItem* anno_dir) const { 594 if (anno_dir->methods_size_ == 0) { 595 return nullptr; 596 } 597 // Skip past the header and field annotations. 598 const uint8_t* addr = reinterpret_cast<const uint8_t*>(&anno_dir[1]); 599 addr += anno_dir->fields_size_ * sizeof(dex::FieldAnnotationsItem); 600 return reinterpret_cast<const dex::MethodAnnotationsItem*>(addr); 601 } 602 GetParameterAnnotations(const dex::AnnotationsDirectoryItem * anno_dir)603 const dex::ParameterAnnotationsItem* GetParameterAnnotations( 604 const dex::AnnotationsDirectoryItem* anno_dir) const { 605 if (anno_dir->parameters_size_ == 0) { 606 return nullptr; 607 } 608 // Skip past the header, field annotations, and method annotations. 609 const uint8_t* addr = reinterpret_cast<const uint8_t*>(&anno_dir[1]); 610 addr += anno_dir->fields_size_ * sizeof(dex::FieldAnnotationsItem); 611 addr += anno_dir->methods_size_ * sizeof(dex::MethodAnnotationsItem); 612 return reinterpret_cast<const dex::ParameterAnnotationsItem*>(addr); 613 } 614 GetFieldAnnotationSetItem(const dex::FieldAnnotationsItem & anno_item)615 const dex::AnnotationSetItem* GetFieldAnnotationSetItem( 616 const dex::FieldAnnotationsItem& anno_item) const { 617 return DataPointer<dex::AnnotationSetItem>(anno_item.annotations_off_); 618 } 619 GetMethodAnnotationSetItem(const dex::MethodAnnotationsItem & anno_item)620 const dex::AnnotationSetItem* GetMethodAnnotationSetItem( 621 const dex::MethodAnnotationsItem& anno_item) const { 622 return DataPointer<dex::AnnotationSetItem>(anno_item.annotations_off_); 623 } 624 GetParameterAnnotationSetRefList(const dex::ParameterAnnotationsItem * anno_item)625 const dex::AnnotationSetRefList* GetParameterAnnotationSetRefList( 626 const dex::ParameterAnnotationsItem* anno_item) const { 627 return DataPointer<dex::AnnotationSetRefList>(anno_item->annotations_off_); 628 } 629 GetAnnotationItemAtOffset(uint32_t offset)630 ALWAYS_INLINE const dex::AnnotationItem* GetAnnotationItemAtOffset(uint32_t offset) const { 631 return DataPointer<dex::AnnotationItem>(offset); 632 } 633 GetHiddenapiClassDataAtOffset(uint32_t offset)634 ALWAYS_INLINE const dex::HiddenapiClassData* GetHiddenapiClassDataAtOffset(uint32_t offset) 635 const { 636 return DataPointer<dex::HiddenapiClassData>(offset); 637 } 638 GetHiddenapiClassData()639 ALWAYS_INLINE const dex::HiddenapiClassData* GetHiddenapiClassData() const { 640 return hiddenapi_class_data_; 641 } 642 HasHiddenapiClassData()643 ALWAYS_INLINE bool HasHiddenapiClassData() const { 644 return hiddenapi_class_data_ != nullptr; 645 } 646 GetAnnotationItem(const dex::AnnotationSetItem * set_item,uint32_t index)647 const dex::AnnotationItem* GetAnnotationItem(const dex::AnnotationSetItem* set_item, 648 uint32_t index) const { 649 DCHECK_LE(index, set_item->size_); 650 return GetAnnotationItemAtOffset(set_item->entries_[index]); 651 } 652 GetSetRefItemItem(const dex::AnnotationSetRefItem * anno_item)653 const dex::AnnotationSetItem* GetSetRefItemItem(const dex::AnnotationSetRefItem* anno_item) 654 const { 655 return DataPointer<dex::AnnotationSetItem>(anno_item->annotations_off_); 656 } 657 658 // Debug info opcodes and constants 659 enum { 660 DBG_END_SEQUENCE = 0x00, 661 DBG_ADVANCE_PC = 0x01, 662 DBG_ADVANCE_LINE = 0x02, 663 DBG_START_LOCAL = 0x03, 664 DBG_START_LOCAL_EXTENDED = 0x04, 665 DBG_END_LOCAL = 0x05, 666 DBG_RESTART_LOCAL = 0x06, 667 DBG_SET_PROLOGUE_END = 0x07, 668 DBG_SET_EPILOGUE_BEGIN = 0x08, 669 DBG_SET_FILE = 0x09, 670 DBG_FIRST_SPECIAL = 0x0a, 671 DBG_LINE_BASE = -4, 672 DBG_LINE_RANGE = 15, 673 }; 674 675 // Returns false if there is no debugging information or if it cannot be decoded. 676 template<typename NewLocalCallback, typename IndexToStringData, typename TypeIndexToStringData> 677 static bool DecodeDebugLocalInfo(const uint8_t* stream, 678 const std::string& location, 679 const char* declaring_class_descriptor, 680 const std::vector<const char*>& arg_descriptors, 681 const std::string& method_name, 682 bool is_static, 683 uint16_t registers_size, 684 uint16_t ins_size, 685 uint16_t insns_size_in_code_units, 686 const IndexToStringData& index_to_string_data, 687 const TypeIndexToStringData& type_index_to_string_data, 688 const NewLocalCallback& new_local) NO_THREAD_SAFETY_ANALYSIS; 689 template<typename NewLocalCallback> 690 bool DecodeDebugLocalInfo(uint32_t registers_size, 691 uint32_t ins_size, 692 uint32_t insns_size_in_code_units, 693 uint32_t debug_info_offset, 694 bool is_static, 695 uint32_t method_idx, 696 const NewLocalCallback& new_local) const; 697 698 // Returns false if there is no debugging information or if it cannot be decoded. 699 template<typename DexDebugNewPosition, typename IndexToStringData> 700 static bool DecodeDebugPositionInfo(const uint8_t* stream, 701 const IndexToStringData& index_to_string_data, 702 const DexDebugNewPosition& position_functor); 703 GetSourceFile(const dex::ClassDef & class_def)704 const char* GetSourceFile(const dex::ClassDef& class_def) const { 705 if (!class_def.source_file_idx_.IsValid()) { 706 return nullptr; 707 } else { 708 return StringDataByIdx(class_def.source_file_idx_); 709 } 710 } 711 712 int GetPermissions() const; 713 714 bool IsReadOnly() const; 715 716 bool EnableWrite() const; 717 718 bool DisableWrite() const; 719 Begin()720 const uint8_t* Begin() const { 721 return begin_; 722 } 723 Size()724 size_t Size() const { 725 return size_; 726 } 727 DataBegin()728 const uint8_t* DataBegin() const { 729 return data_begin_; 730 } 731 DataSize()732 size_t DataSize() const { 733 return data_size_; 734 } 735 736 template <typename T> DataPointer(size_t offset)737 const T* DataPointer(size_t offset) const { 738 DCHECK_LT(offset, DataSize()) << "Offset past end of data section"; 739 return (offset != 0u) ? reinterpret_cast<const T*>(DataBegin() + offset) : nullptr; 740 } 741 GetOatDexFile()742 const OatDexFile* GetOatDexFile() const { 743 return oat_dex_file_; 744 } 745 746 // Used by oat writer. SetOatDexFile(const OatDexFile * oat_dex_file)747 void SetOatDexFile(const OatDexFile* oat_dex_file) const { 748 oat_dex_file_ = oat_dex_file; 749 } 750 751 // Read MapItems and validate/set remaining offsets. GetMapList()752 const dex::MapList* GetMapList() const { 753 return reinterpret_cast<const dex::MapList*>(DataBegin() + header_->map_off_); 754 } 755 756 // Utility methods for reading integral values from a buffer. 757 static int32_t ReadSignedInt(const uint8_t* ptr, int zwidth); 758 static uint32_t ReadUnsignedInt(const uint8_t* ptr, int zwidth, bool fill_on_right); 759 static int64_t ReadSignedLong(const uint8_t* ptr, int zwidth); 760 static uint64_t ReadUnsignedLong(const uint8_t* ptr, int zwidth, bool fill_on_right); 761 762 // Recalculates the checksum of the dex file. Does not use the current value in the header. 763 virtual uint32_t CalculateChecksum() const; 764 static uint32_t CalculateChecksum(const uint8_t* begin, size_t size); 765 static uint32_t ChecksumMemoryRange(const uint8_t* begin, size_t size); 766 767 // Number of bytes at the beginning of the dex file header which are skipped 768 // when computing the adler32 checksum of the entire file. 769 static constexpr uint32_t kNumNonChecksumBytes = OFFSETOF_MEMBER(DexFile::Header, signature_); 770 771 // Appends a human-readable form of the method at an index. 772 void AppendPrettyMethod(uint32_t method_idx, bool with_signature, std::string* result) const; 773 // Returns a human-readable form of the field at an index. 774 std::string PrettyField(uint32_t field_idx, bool with_type = true) const; 775 // Returns a human-readable form of the type at an index. 776 std::string PrettyType(dex::TypeIndex type_idx) const; 777 778 ALWAYS_INLINE std::string PrettyMethod(uint32_t method_idx, bool with_signature = true) const { 779 std::string result; 780 AppendPrettyMethod(method_idx, with_signature, &result); 781 return result; 782 } 783 784 // Not virtual for performance reasons. IsCompactDexFile()785 ALWAYS_INLINE bool IsCompactDexFile() const { 786 return is_compact_dex_; 787 } IsStandardDexFile()788 ALWAYS_INLINE bool IsStandardDexFile() const { 789 return !is_compact_dex_; 790 } 791 ALWAYS_INLINE const StandardDexFile* AsStandardDexFile() const; 792 ALWAYS_INLINE const CompactDexFile* AsCompactDexFile() const; 793 GetHiddenapiDomain()794 hiddenapi::Domain GetHiddenapiDomain() const { return hiddenapi_domain_; } SetHiddenapiDomain(hiddenapi::Domain value)795 void SetHiddenapiDomain(hiddenapi::Domain value) const { hiddenapi_domain_ = value; } 796 IsInMainSection(const void * addr)797 bool IsInMainSection(const void* addr) const { 798 return Begin() <= addr && addr < Begin() + Size(); 799 } 800 IsInDataSection(const void * addr)801 bool IsInDataSection(const void* addr) const { 802 return DataBegin() <= addr && addr < DataBegin() + DataSize(); 803 } 804 GetContainer()805 DexFileContainer* GetContainer() const { 806 return container_.get(); 807 } 808 809 IterationRange<ClassIterator> GetClasses() const; 810 811 template <typename Visitor> 812 static uint32_t DecodeDebugInfoParameterNames(const uint8_t** debug_info, 813 const Visitor& visitor); 814 815 static inline bool StringEquals(const DexFile* df1, dex::StringIndex sidx1, 816 const DexFile* df2, dex::StringIndex sidx2); 817 818 protected: 819 // First Dex format version supporting default methods. 820 static constexpr uint32_t kDefaultMethodsVersion = 37; 821 822 DexFile(const uint8_t* base, 823 size_t size, 824 const uint8_t* data_begin, 825 size_t data_size, 826 const std::string& location, 827 uint32_t location_checksum, 828 const OatDexFile* oat_dex_file, 829 std::unique_ptr<DexFileContainer> container, 830 bool is_compact_dex); 831 832 // Top-level initializer that calls other Init methods. 833 bool Init(std::string* error_msg); 834 835 // Returns true if the header magic and version numbers are of the expected values. 836 bool CheckMagicAndVersion(std::string* error_msg) const; 837 838 // Initialize section info for sections only found in map. Returns true on success. 839 void InitializeSectionsFromMapList(); 840 841 // The base address of the memory mapping. 842 const uint8_t* const begin_; 843 844 // The size of the underlying memory allocation in bytes. 845 const size_t size_; 846 847 // The base address of the data section (same as Begin() for standard dex). 848 const uint8_t* const data_begin_; 849 850 // The size of the data section. 851 const size_t data_size_; 852 853 // Typically the dex file name when available, alternatively some identifying string. 854 // 855 // The ClassLinker will use this to match DexFiles the boot class 856 // path to DexCache::GetLocation when loading from an image. 857 const std::string location_; 858 859 const uint32_t location_checksum_; 860 861 // Points to the header section. 862 const Header* const header_; 863 864 // Points to the base of the string identifier list. 865 const dex::StringId* const string_ids_; 866 867 // Points to the base of the type identifier list. 868 const dex::TypeId* const type_ids_; 869 870 // Points to the base of the field identifier list. 871 const dex::FieldId* const field_ids_; 872 873 // Points to the base of the method identifier list. 874 const dex::MethodId* const method_ids_; 875 876 // Points to the base of the prototype identifier list. 877 const dex::ProtoId* const proto_ids_; 878 879 // Points to the base of the class definition list. 880 const dex::ClassDef* const class_defs_; 881 882 // Points to the base of the method handles list. 883 const dex::MethodHandleItem* method_handles_; 884 885 // Number of elements in the method handles list. 886 size_t num_method_handles_; 887 888 // Points to the base of the call sites id list. 889 const dex::CallSiteIdItem* call_site_ids_; 890 891 // Number of elements in the call sites list. 892 size_t num_call_site_ids_; 893 894 // Points to the base of the hiddenapi class data item_, or nullptr if the dex 895 // file does not have one. 896 const dex::HiddenapiClassData* hiddenapi_class_data_; 897 898 // If this dex file was loaded from an oat file, oat_dex_file_ contains a 899 // pointer to the OatDexFile it was loaded from. Otherwise oat_dex_file_ is 900 // null. 901 mutable const OatDexFile* oat_dex_file_; 902 903 // Manages the underlying memory allocation. 904 std::unique_ptr<DexFileContainer> container_; 905 906 // If the dex file is a compact dex file. If false then the dex file is a standard dex file. 907 const bool is_compact_dex_; 908 909 // The domain this dex file belongs to for hidden API access checks. 910 // It is decleared `mutable` because the domain is assigned after the DexFile 911 // has been created and can be changed later by the runtime. 912 mutable hiddenapi::Domain hiddenapi_domain_; 913 914 friend class DexFileLoader; 915 friend class DexFileVerifierTest; 916 friend class OatWriter; 917 }; 918 919 std::ostream& operator<<(std::ostream& os, const DexFile& dex_file); 920 921 // Iterate over a dex file's ProtoId's paramters 922 class DexFileParameterIterator { 923 public: DexFileParameterIterator(const DexFile & dex_file,const dex::ProtoId & proto_id)924 DexFileParameterIterator(const DexFile& dex_file, const dex::ProtoId& proto_id) 925 : dex_file_(dex_file) { 926 type_list_ = dex_file_.GetProtoParameters(proto_id); 927 if (type_list_ != nullptr) { 928 size_ = type_list_->Size(); 929 } 930 } HasNext()931 bool HasNext() const { return pos_ < size_; } Size()932 size_t Size() const { return size_; } Next()933 void Next() { ++pos_; } GetTypeIdx()934 dex::TypeIndex GetTypeIdx() { 935 return type_list_->GetTypeItem(pos_).type_idx_; 936 } GetDescriptor()937 const char* GetDescriptor() { 938 return dex_file_.StringByTypeIdx(dex::TypeIndex(GetTypeIdx())); 939 } 940 private: 941 const DexFile& dex_file_; 942 const dex::TypeList* type_list_ = nullptr; 943 uint32_t size_ = 0; 944 uint32_t pos_ = 0; 945 DISALLOW_IMPLICIT_CONSTRUCTORS(DexFileParameterIterator); 946 }; 947 948 class EncodedArrayValueIterator { 949 public: 950 EncodedArrayValueIterator(const DexFile& dex_file, const uint8_t* array_data); 951 HasNext()952 bool HasNext() const { return pos_ < array_size_; } 953 954 void Next(); 955 956 enum ValueType { 957 kByte = 0x00, 958 kShort = 0x02, 959 kChar = 0x03, 960 kInt = 0x04, 961 kLong = 0x06, 962 kFloat = 0x10, 963 kDouble = 0x11, 964 kMethodType = 0x15, 965 kMethodHandle = 0x16, 966 kString = 0x17, 967 kType = 0x18, 968 kField = 0x19, 969 kMethod = 0x1a, 970 kEnum = 0x1b, 971 kArray = 0x1c, 972 kAnnotation = 0x1d, 973 kNull = 0x1e, 974 kBoolean = 0x1f, 975 }; 976 GetValueType()977 ValueType GetValueType() const { return type_; } GetJavaValue()978 const jvalue& GetJavaValue() const { return jval_; } 979 980 protected: 981 static constexpr uint8_t kEncodedValueTypeMask = 0x1f; // 0b11111 982 static constexpr uint8_t kEncodedValueArgShift = 5; 983 984 const DexFile& dex_file_; 985 size_t array_size_; // Size of array. 986 size_t pos_; // Current position. 987 const uint8_t* ptr_; // Pointer into encoded data array. 988 ValueType type_; // Type of current encoded value. 989 jvalue jval_; // Value of current encoded value. 990 991 private: 992 DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedArrayValueIterator); 993 }; 994 std::ostream& operator<<(std::ostream& os, EncodedArrayValueIterator::ValueType code); 995 996 class EncodedStaticFieldValueIterator : public EncodedArrayValueIterator { 997 public: EncodedStaticFieldValueIterator(const DexFile & dex_file,const dex::ClassDef & class_def)998 EncodedStaticFieldValueIterator(const DexFile& dex_file, 999 const dex::ClassDef& class_def) 1000 : EncodedArrayValueIterator(dex_file, 1001 dex_file.GetEncodedStaticFieldValuesArray(class_def)) 1002 {} 1003 1004 private: 1005 DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedStaticFieldValueIterator); 1006 }; 1007 1008 class CallSiteArrayValueIterator : public EncodedArrayValueIterator { 1009 public: CallSiteArrayValueIterator(const DexFile & dex_file,const dex::CallSiteIdItem & call_site_id)1010 CallSiteArrayValueIterator(const DexFile& dex_file, 1011 const dex::CallSiteIdItem& call_site_id) 1012 : EncodedArrayValueIterator(dex_file, 1013 dex_file.GetCallSiteEncodedValuesArray(call_site_id)) 1014 {} 1015 Size()1016 uint32_t Size() const { return array_size_; } 1017 1018 private: 1019 DISALLOW_IMPLICIT_CONSTRUCTORS(CallSiteArrayValueIterator); 1020 }; 1021 1022 } // namespace art 1023 1024 #endif // ART_LIBDEXFILE_DEX_DEX_FILE_H_ 1025