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 <string> 21 #include <vector> 22 23 #include "base/logging.h" 24 #include "base/mutex.h" 25 #include "base/stringpiece.h" 26 #include "globals.h" 27 #include "invoke_type.h" 28 #include "jni.h" 29 #include "mem_map.h" 30 #include "modifiers.h" 31 #include "safe_map.h" 32 #include "UniquePtr.h" 33 34 namespace art { 35 36 namespace mirror { 37 class ArtField; 38 class ArtMethod; 39 class ClassLoader; 40 class DexCache; 41 } // namespace mirror 42 class ClassLinker; 43 class ZipArchive; 44 45 // TODO: move all of the macro functionality into the DexCache class. 46 class DexFile { 47 public: 48 static const byte kDexMagic[]; 49 static const byte kDexMagicVersion[]; 50 static const size_t kSha1DigestSize = 20; 51 static const uint32_t kDexEndianConstant = 0x12345678; 52 53 // name of the DexFile entry within a zip archive 54 static const char* kClassesDex; 55 56 // The value of an invalid index. 57 static const uint32_t kDexNoIndex = 0xFFFFFFFF; 58 59 // The value of an invalid index. 60 static const uint16_t kDexNoIndex16 = 0xFFFF; 61 62 // Raw header_item. 63 struct Header { 64 uint8_t magic_[8]; 65 uint32_t checksum_; // See also location_checksum_ 66 uint8_t signature_[kSha1DigestSize]; 67 uint32_t file_size_; // size of entire file 68 uint32_t header_size_; // offset to start of next section 69 uint32_t endian_tag_; 70 uint32_t link_size_; // unused 71 uint32_t link_off_; // unused 72 uint32_t map_off_; // unused 73 uint32_t string_ids_size_; // number of StringIds 74 uint32_t string_ids_off_; // file offset of StringIds array 75 uint32_t type_ids_size_; // number of TypeIds, we don't support more than 65535 76 uint32_t type_ids_off_; // file offset of TypeIds array 77 uint32_t proto_ids_size_; // number of ProtoIds, we don't support more than 65535 78 uint32_t proto_ids_off_; // file offset of ProtoIds array 79 uint32_t field_ids_size_; // number of FieldIds 80 uint32_t field_ids_off_; // file offset of FieldIds array 81 uint32_t method_ids_size_; // number of MethodIds 82 uint32_t method_ids_off_; // file offset of MethodIds array 83 uint32_t class_defs_size_; // number of ClassDefs 84 uint32_t class_defs_off_; // file offset of ClassDef array 85 uint32_t data_size_; // unused 86 uint32_t data_off_; // unused 87 88 private: 89 DISALLOW_COPY_AND_ASSIGN(Header); 90 }; 91 92 // Map item type codes. 93 enum { 94 kDexTypeHeaderItem = 0x0000, 95 kDexTypeStringIdItem = 0x0001, 96 kDexTypeTypeIdItem = 0x0002, 97 kDexTypeProtoIdItem = 0x0003, 98 kDexTypeFieldIdItem = 0x0004, 99 kDexTypeMethodIdItem = 0x0005, 100 kDexTypeClassDefItem = 0x0006, 101 kDexTypeMapList = 0x1000, 102 kDexTypeTypeList = 0x1001, 103 kDexTypeAnnotationSetRefList = 0x1002, 104 kDexTypeAnnotationSetItem = 0x1003, 105 kDexTypeClassDataItem = 0x2000, 106 kDexTypeCodeItem = 0x2001, 107 kDexTypeStringDataItem = 0x2002, 108 kDexTypeDebugInfoItem = 0x2003, 109 kDexTypeAnnotationItem = 0x2004, 110 kDexTypeEncodedArrayItem = 0x2005, 111 kDexTypeAnnotationsDirectoryItem = 0x2006, 112 }; 113 114 struct MapItem { 115 uint16_t type_; 116 uint16_t unused_; 117 uint32_t size_; 118 uint32_t offset_; 119 120 private: 121 DISALLOW_COPY_AND_ASSIGN(MapItem); 122 }; 123 124 struct MapList { 125 uint32_t size_; 126 MapItem list_[1]; 127 128 private: 129 DISALLOW_COPY_AND_ASSIGN(MapList); 130 }; 131 132 // Raw string_id_item. 133 struct StringId { 134 uint32_t string_data_off_; // offset in bytes from the base address 135 136 private: 137 DISALLOW_COPY_AND_ASSIGN(StringId); 138 }; 139 140 // Raw type_id_item. 141 struct TypeId { 142 uint32_t descriptor_idx_; // index into string_ids 143 144 private: 145 DISALLOW_COPY_AND_ASSIGN(TypeId); 146 }; 147 148 // Raw field_id_item. 149 struct FieldId { 150 uint16_t class_idx_; // index into type_ids_ array for defining class 151 uint16_t type_idx_; // index into type_ids_ array for field type 152 uint32_t name_idx_; // index into string_ids_ array for field name 153 154 private: 155 DISALLOW_COPY_AND_ASSIGN(FieldId); 156 }; 157 158 // Raw method_id_item. 159 struct MethodId { 160 uint16_t class_idx_; // index into type_ids_ array for defining class 161 uint16_t proto_idx_; // index into proto_ids_ array for method prototype 162 uint32_t name_idx_; // index into string_ids_ array for method name 163 164 private: 165 DISALLOW_COPY_AND_ASSIGN(MethodId); 166 }; 167 168 // Raw proto_id_item. 169 struct ProtoId { 170 uint32_t shorty_idx_; // index into string_ids array for shorty descriptor 171 uint16_t return_type_idx_; // index into type_ids array for return type 172 uint16_t pad_; // padding = 0 173 uint32_t parameters_off_; // file offset to type_list for parameter types 174 175 private: 176 DISALLOW_COPY_AND_ASSIGN(ProtoId); 177 }; 178 179 // Raw class_def_item. 180 struct ClassDef { 181 uint16_t class_idx_; // index into type_ids_ array for this class 182 uint16_t pad1_; // padding = 0 183 uint32_t access_flags_; 184 uint16_t superclass_idx_; // index into type_ids_ array for superclass 185 uint16_t pad2_; // padding = 0 186 uint32_t interfaces_off_; // file offset to TypeList 187 uint32_t source_file_idx_; // index into string_ids_ for source file name 188 uint32_t annotations_off_; // file offset to annotations_directory_item 189 uint32_t class_data_off_; // file offset to class_data_item 190 uint32_t static_values_off_; // file offset to EncodedArray 191 192 private: 193 DISALLOW_COPY_AND_ASSIGN(ClassDef); 194 }; 195 196 // Raw type_item. 197 struct TypeItem { 198 uint16_t type_idx_; // index into type_ids section 199 200 private: 201 DISALLOW_COPY_AND_ASSIGN(TypeItem); 202 }; 203 204 // Raw type_list. 205 class TypeList { 206 public: Size()207 uint32_t Size() const { 208 return size_; 209 } 210 GetTypeItem(uint32_t idx)211 const TypeItem& GetTypeItem(uint32_t idx) const { 212 DCHECK_LT(idx, this->size_); 213 return this->list_[idx]; 214 } 215 216 private: 217 uint32_t size_; // size of the list, in entries 218 TypeItem list_[1]; // elements of the list 219 DISALLOW_COPY_AND_ASSIGN(TypeList); 220 }; 221 222 // Raw code_item. 223 struct CodeItem { 224 uint16_t registers_size_; 225 uint16_t ins_size_; 226 uint16_t outs_size_; 227 uint16_t tries_size_; 228 uint32_t debug_info_off_; // file offset to debug info stream 229 uint32_t insns_size_in_code_units_; // size of the insns array, in 2 byte code units 230 uint16_t insns_[1]; 231 232 private: 233 DISALLOW_COPY_AND_ASSIGN(CodeItem); 234 }; 235 236 // Raw try_item. 237 struct TryItem { 238 uint32_t start_addr_; 239 uint16_t insn_count_; 240 uint16_t handler_off_; 241 242 private: 243 DISALLOW_COPY_AND_ASSIGN(TryItem); 244 }; 245 246 // Annotation constants. 247 enum { 248 kDexVisibilityBuild = 0x00, /* annotation visibility */ 249 kDexVisibilityRuntime = 0x01, 250 kDexVisibilitySystem = 0x02, 251 252 kDexAnnotationByte = 0x00, 253 kDexAnnotationShort = 0x02, 254 kDexAnnotationChar = 0x03, 255 kDexAnnotationInt = 0x04, 256 kDexAnnotationLong = 0x06, 257 kDexAnnotationFloat = 0x10, 258 kDexAnnotationDouble = 0x11, 259 kDexAnnotationString = 0x17, 260 kDexAnnotationType = 0x18, 261 kDexAnnotationField = 0x19, 262 kDexAnnotationMethod = 0x1a, 263 kDexAnnotationEnum = 0x1b, 264 kDexAnnotationArray = 0x1c, 265 kDexAnnotationAnnotation = 0x1d, 266 kDexAnnotationNull = 0x1e, 267 kDexAnnotationBoolean = 0x1f, 268 269 kDexAnnotationValueTypeMask = 0x1f, /* low 5 bits */ 270 kDexAnnotationValueArgShift = 5, 271 }; 272 273 struct AnnotationsDirectoryItem { 274 uint32_t class_annotations_off_; 275 uint32_t fields_size_; 276 uint32_t methods_size_; 277 uint32_t parameters_size_; 278 279 private: 280 DISALLOW_COPY_AND_ASSIGN(AnnotationsDirectoryItem); 281 }; 282 283 struct FieldAnnotationsItem { 284 uint32_t field_idx_; 285 uint32_t annotations_off_; 286 287 private: 288 DISALLOW_COPY_AND_ASSIGN(FieldAnnotationsItem); 289 }; 290 291 struct MethodAnnotationsItem { 292 uint32_t method_idx_; 293 uint32_t annotations_off_; 294 295 private: 296 DISALLOW_COPY_AND_ASSIGN(MethodAnnotationsItem); 297 }; 298 299 struct ParameterAnnotationsItem { 300 uint32_t method_idx_; 301 uint32_t annotations_off_; 302 303 private: 304 DISALLOW_COPY_AND_ASSIGN(ParameterAnnotationsItem); 305 }; 306 307 struct AnnotationSetRefItem { 308 uint32_t annotations_off_; 309 310 private: 311 DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefItem); 312 }; 313 314 struct AnnotationSetRefList { 315 uint32_t size_; 316 AnnotationSetRefItem list_[1]; 317 318 private: 319 DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefList); 320 }; 321 322 struct AnnotationSetItem { 323 uint32_t size_; 324 uint32_t entries_[1]; 325 326 private: 327 DISALLOW_COPY_AND_ASSIGN(AnnotationSetItem); 328 }; 329 330 struct AnnotationItem { 331 uint8_t visibility_; 332 uint8_t annotation_[1]; 333 334 private: 335 DISALLOW_COPY_AND_ASSIGN(AnnotationItem); 336 }; 337 338 typedef std::pair<const DexFile*, const DexFile::ClassDef*> ClassPathEntry; 339 typedef std::vector<const DexFile*> ClassPath; 340 341 // Search a collection of DexFiles for a descriptor 342 static ClassPathEntry FindInClassPath(const char* descriptor, 343 const ClassPath& class_path); 344 345 // Returns the checksum of a file for comparison with GetLocationChecksum(). 346 // For .dex files, this is the header checksum. 347 // For zip files, this is the classes.dex zip entry CRC32 checksum. 348 // Return true if the checksum could be found, false otherwise. 349 static bool GetChecksum(const std::string& filename, uint32_t* checksum) 350 SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 351 352 // Opens .dex file, guessing the container format based on file extension 353 static const DexFile* Open(const std::string& filename, 354 const std::string& location); 355 356 // Opens .dex file, backed by existing memory Open(const uint8_t * base,size_t size,const std::string & location,uint32_t location_checksum)357 static const DexFile* Open(const uint8_t* base, size_t size, 358 const std::string& location, 359 uint32_t location_checksum) { 360 return OpenMemory(base, size, location, location_checksum, NULL); 361 } 362 363 // Opens .dex file from the classes.dex in a zip archive 364 static const DexFile* Open(const ZipArchive& zip_archive, const std::string& location); 365 366 // Closes a .dex file. 367 virtual ~DexFile(); 368 GetLocation()369 const std::string& GetLocation() const { 370 return location_; 371 } 372 373 // For DexFiles directly from .dex files, this is the checksum from the DexFile::Header. 374 // For DexFiles opened from a zip files, this will be the ZipEntry CRC32 of classes.dex. GetLocationChecksum()375 uint32_t GetLocationChecksum() const { 376 return location_checksum_; 377 } 378 GetHeader()379 const Header& GetHeader() const { 380 DCHECK(header_ != NULL) << GetLocation(); 381 return *header_; 382 } 383 GetModificationLock()384 Mutex& GetModificationLock() { 385 return modification_lock; 386 } 387 388 // Decode the dex magic version 389 uint32_t GetVersion() const; 390 391 // Returns true if the byte string points to the magic value. 392 static bool IsMagicValid(const byte* magic); 393 394 // Returns true if the byte string after the magic is the correct value. 395 static bool IsVersionValid(const byte* magic); 396 397 // Returns the number of string identifiers in the .dex file. NumStringIds()398 size_t NumStringIds() const { 399 DCHECK(header_ != NULL) << GetLocation(); 400 return header_->string_ids_size_; 401 } 402 403 // Returns the StringId at the specified index. GetStringId(uint32_t idx)404 const StringId& GetStringId(uint32_t idx) const { 405 DCHECK_LT(idx, NumStringIds()) << GetLocation(); 406 return string_ids_[idx]; 407 } 408 GetIndexForStringId(const StringId & string_id)409 uint32_t GetIndexForStringId(const StringId& string_id) const { 410 CHECK_GE(&string_id, string_ids_) << GetLocation(); 411 CHECK_LT(&string_id, string_ids_ + header_->string_ids_size_) << GetLocation(); 412 return &string_id - string_ids_; 413 } 414 415 int32_t GetStringLength(const StringId& string_id) const; 416 417 // Returns a pointer to the UTF-8 string data referred to by the given string_id. 418 const char* GetStringDataAndLength(const StringId& string_id, uint32_t* length) const; 419 GetStringData(const StringId & string_id)420 const char* GetStringData(const StringId& string_id) const { 421 uint32_t length; 422 return GetStringDataAndLength(string_id, &length); 423 } 424 425 // return the UTF-8 encoded string with the specified string_id index StringDataAndLengthByIdx(uint32_t idx,uint32_t * unicode_length)426 const char* StringDataAndLengthByIdx(uint32_t idx, uint32_t* unicode_length) const { 427 if (idx == kDexNoIndex) { 428 *unicode_length = 0; 429 return NULL; 430 } 431 const StringId& string_id = GetStringId(idx); 432 return GetStringDataAndLength(string_id, unicode_length); 433 } 434 StringDataByIdx(uint32_t idx)435 const char* StringDataByIdx(uint32_t idx) const { 436 uint32_t unicode_length; 437 return StringDataAndLengthByIdx(idx, &unicode_length); 438 } 439 440 // Looks up a string id for a given modified utf8 string. 441 const StringId* FindStringId(const char* string) const; 442 443 // Looks up a string id for a given utf16 string. 444 const StringId* FindStringId(const uint16_t* string) const; 445 446 // Returns the number of type identifiers in the .dex file. NumTypeIds()447 size_t NumTypeIds() const { 448 DCHECK(header_ != NULL) << GetLocation(); 449 return header_->type_ids_size_; 450 } 451 452 // Returns the TypeId at the specified index. GetTypeId(uint32_t idx)453 const TypeId& GetTypeId(uint32_t idx) const { 454 DCHECK_LT(idx, NumTypeIds()) << GetLocation(); 455 return type_ids_[idx]; 456 } 457 GetIndexForTypeId(const TypeId & type_id)458 uint16_t GetIndexForTypeId(const TypeId& type_id) const { 459 CHECK_GE(&type_id, type_ids_) << GetLocation(); 460 CHECK_LT(&type_id, type_ids_ + header_->type_ids_size_) << GetLocation(); 461 size_t result = &type_id - type_ids_; 462 DCHECK_LT(result, 65536U) << GetLocation(); 463 return static_cast<uint16_t>(result); 464 } 465 466 // Get the descriptor string associated with a given type index. StringByTypeIdx(uint32_t idx,uint32_t * unicode_length)467 const char* StringByTypeIdx(uint32_t idx, uint32_t* unicode_length) const { 468 const TypeId& type_id = GetTypeId(idx); 469 return StringDataAndLengthByIdx(type_id.descriptor_idx_, unicode_length); 470 } 471 StringByTypeIdx(uint32_t idx)472 const char* StringByTypeIdx(uint32_t idx) const { 473 const TypeId& type_id = GetTypeId(idx); 474 return StringDataByIdx(type_id.descriptor_idx_); 475 } 476 477 // Returns the type descriptor string of a type id. GetTypeDescriptor(const TypeId & type_id)478 const char* GetTypeDescriptor(const TypeId& type_id) const { 479 return StringDataByIdx(type_id.descriptor_idx_); 480 } 481 482 // Looks up a type for the given string index 483 const TypeId* FindTypeId(uint32_t string_idx) const; 484 485 // Returns the number of field identifiers in the .dex file. NumFieldIds()486 size_t NumFieldIds() const { 487 DCHECK(header_ != NULL) << GetLocation(); 488 return header_->field_ids_size_; 489 } 490 491 // Returns the FieldId at the specified index. GetFieldId(uint32_t idx)492 const FieldId& GetFieldId(uint32_t idx) const { 493 DCHECK_LT(idx, NumFieldIds()) << GetLocation(); 494 return field_ids_[idx]; 495 } 496 GetIndexForFieldId(const FieldId & field_id)497 uint32_t GetIndexForFieldId(const FieldId& field_id) const { 498 CHECK_GE(&field_id, field_ids_) << GetLocation(); 499 CHECK_LT(&field_id, field_ids_ + header_->field_ids_size_) << GetLocation(); 500 return &field_id - field_ids_; 501 } 502 503 // Looks up a field by its declaring class, name and type 504 const FieldId* FindFieldId(const DexFile::TypeId& declaring_klass, 505 const DexFile::StringId& name, 506 const DexFile::TypeId& type) const; 507 508 // Returns the declaring class descriptor string of a field id. GetFieldDeclaringClassDescriptor(const FieldId & field_id)509 const char* GetFieldDeclaringClassDescriptor(const FieldId& field_id) const { 510 const DexFile::TypeId& type_id = GetTypeId(field_id.class_idx_); 511 return GetTypeDescriptor(type_id); 512 } 513 514 // Returns the class descriptor string of a field id. GetFieldTypeDescriptor(const FieldId & field_id)515 const char* GetFieldTypeDescriptor(const FieldId& field_id) const { 516 const DexFile::TypeId& type_id = GetTypeId(field_id.type_idx_); 517 return GetTypeDescriptor(type_id); 518 } 519 520 // Returns the name of a field id. GetFieldName(const FieldId & field_id)521 const char* GetFieldName(const FieldId& field_id) const { 522 return StringDataByIdx(field_id.name_idx_); 523 } 524 525 // Returns the number of method identifiers in the .dex file. NumMethodIds()526 size_t NumMethodIds() const { 527 DCHECK(header_ != NULL) << GetLocation(); 528 return header_->method_ids_size_; 529 } 530 531 // Returns the MethodId at the specified index. GetMethodId(uint32_t idx)532 const MethodId& GetMethodId(uint32_t idx) const { 533 DCHECK_LT(idx, NumMethodIds()) << GetLocation(); 534 return method_ids_[idx]; 535 } 536 GetIndexForMethodId(const MethodId & method_id)537 uint32_t GetIndexForMethodId(const MethodId& method_id) const { 538 CHECK_GE(&method_id, method_ids_) << GetLocation(); 539 CHECK_LT(&method_id, method_ids_ + header_->method_ids_size_) << GetLocation(); 540 return &method_id - method_ids_; 541 } 542 543 // Looks up a method by its declaring class, name and proto_id 544 const MethodId* FindMethodId(const DexFile::TypeId& declaring_klass, 545 const DexFile::StringId& name, 546 const DexFile::ProtoId& signature) const; 547 548 // Returns the declaring class descriptor string of a method id. GetMethodDeclaringClassDescriptor(const MethodId & method_id)549 const char* GetMethodDeclaringClassDescriptor(const MethodId& method_id) const { 550 const DexFile::TypeId& type_id = GetTypeId(method_id.class_idx_); 551 return GetTypeDescriptor(type_id); 552 } 553 554 // Returns the prototype of a method id. GetMethodPrototype(const MethodId & method_id)555 const ProtoId& GetMethodPrototype(const MethodId& method_id) const { 556 return GetProtoId(method_id.proto_idx_); 557 } 558 559 // Returns the signature of a method id. GetMethodSignature(const MethodId & method_id)560 const std::string GetMethodSignature(const MethodId& method_id) const { 561 return CreateMethodSignature(method_id.proto_idx_, NULL); 562 } 563 564 // Returns the name of a method id. GetMethodName(const MethodId & method_id)565 const char* GetMethodName(const MethodId& method_id) const { 566 return StringDataByIdx(method_id.name_idx_); 567 } 568 569 // Returns the shorty of a method id. GetMethodShorty(const MethodId & method_id)570 const char* GetMethodShorty(const MethodId& method_id) const { 571 return StringDataByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_); 572 } GetMethodShorty(const MethodId & method_id,uint32_t * length)573 const char* GetMethodShorty(const MethodId& method_id, uint32_t* length) const { 574 return StringDataAndLengthByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_, length); 575 } 576 // Returns the number of class definitions in the .dex file. NumClassDefs()577 size_t NumClassDefs() const { 578 DCHECK(header_ != NULL) << GetLocation(); 579 return header_->class_defs_size_; 580 } 581 582 // Returns the ClassDef at the specified index. GetClassDef(uint16_t idx)583 const ClassDef& GetClassDef(uint16_t idx) const { 584 DCHECK_LT(idx, NumClassDefs()) << GetLocation(); 585 return class_defs_[idx]; 586 } 587 GetIndexForClassDef(const ClassDef & class_def)588 uint16_t GetIndexForClassDef(const ClassDef& class_def) const { 589 CHECK_GE(&class_def, class_defs_) << GetLocation(); 590 CHECK_LT(&class_def, class_defs_ + header_->class_defs_size_) << GetLocation(); 591 return &class_def - class_defs_; 592 } 593 594 // Returns the class descriptor string of a class definition. GetClassDescriptor(const ClassDef & class_def)595 const char* GetClassDescriptor(const ClassDef& class_def) const { 596 return StringByTypeIdx(class_def.class_idx_); 597 } 598 599 // Looks up a class definition by its class descriptor. 600 const ClassDef* FindClassDef(const char* descriptor) const; 601 602 // Looks up a class definition by its type index. 603 const ClassDef* FindClassDef(uint16_t type_idx) const; 604 GetInterfacesList(const ClassDef & class_def)605 const TypeList* GetInterfacesList(const ClassDef& class_def) const { 606 if (class_def.interfaces_off_ == 0) { 607 return NULL; 608 } else { 609 const byte* addr = begin_ + class_def.interfaces_off_; 610 return reinterpret_cast<const TypeList*>(addr); 611 } 612 } 613 614 // Returns a pointer to the raw memory mapped class_data_item GetClassData(const ClassDef & class_def)615 const byte* GetClassData(const ClassDef& class_def) const { 616 if (class_def.class_data_off_ == 0) { 617 return NULL; 618 } else { 619 return begin_ + class_def.class_data_off_; 620 } 621 } 622 623 // GetCodeItem(const uint32_t code_off)624 const CodeItem* GetCodeItem(const uint32_t code_off) const { 625 if (code_off == 0) { 626 return NULL; // native or abstract method 627 } else { 628 const byte* addr = begin_ + code_off; 629 return reinterpret_cast<const CodeItem*>(addr); 630 } 631 } 632 GetReturnTypeDescriptor(const ProtoId & proto_id)633 const char* GetReturnTypeDescriptor(const ProtoId& proto_id) const { 634 return StringByTypeIdx(proto_id.return_type_idx_); 635 } 636 637 // Returns the number of prototype identifiers in the .dex file. NumProtoIds()638 size_t NumProtoIds() const { 639 DCHECK(header_ != NULL) << GetLocation(); 640 return header_->proto_ids_size_; 641 } 642 643 // Returns the ProtoId at the specified index. GetProtoId(uint32_t idx)644 const ProtoId& GetProtoId(uint32_t idx) const { 645 DCHECK_LT(idx, NumProtoIds()) << GetLocation(); 646 return proto_ids_[idx]; 647 } 648 GetIndexForProtoId(const ProtoId & proto_id)649 uint16_t GetIndexForProtoId(const ProtoId& proto_id) const { 650 CHECK_GE(&proto_id, proto_ids_) << GetLocation(); 651 CHECK_LT(&proto_id, proto_ids_ + header_->proto_ids_size_) << GetLocation(); 652 return &proto_id - proto_ids_; 653 } 654 655 // Looks up a proto id for a given return type and signature type list 656 const ProtoId* FindProtoId(uint16_t return_type_id, 657 const std::vector<uint16_t>& signature_type_idxs_) const; 658 659 // Given a signature place the type ids into the given vector, returns true on success 660 bool CreateTypeList(uint16_t* return_type_idx, std::vector<uint16_t>* param_type_idxs, 661 const std::string& signature) const; 662 663 // Given a proto_idx decode the type list and return type into a method signature 664 std::string CreateMethodSignature(uint32_t proto_idx, int32_t* unicode_length) const; 665 666 // Returns the short form method descriptor for the given prototype. GetShorty(uint32_t proto_idx)667 const char* GetShorty(uint32_t proto_idx) const { 668 const ProtoId& proto_id = GetProtoId(proto_idx); 669 return StringDataByIdx(proto_id.shorty_idx_); 670 } 671 GetProtoParameters(const ProtoId & proto_id)672 const TypeList* GetProtoParameters(const ProtoId& proto_id) const { 673 if (proto_id.parameters_off_ == 0) { 674 return NULL; 675 } else { 676 const byte* addr = begin_ + proto_id.parameters_off_; 677 return reinterpret_cast<const TypeList*>(addr); 678 } 679 } 680 GetEncodedStaticFieldValuesArray(const ClassDef & class_def)681 const byte* GetEncodedStaticFieldValuesArray(const ClassDef& class_def) const { 682 if (class_def.static_values_off_ == 0) { 683 return 0; 684 } else { 685 return begin_ + class_def.static_values_off_; 686 } 687 } 688 689 static const TryItem* GetTryItems(const CodeItem& code_item, uint32_t offset); 690 691 // Get the base of the encoded data for the given DexCode. GetCatchHandlerData(const CodeItem & code_item,uint32_t offset)692 static const byte* GetCatchHandlerData(const CodeItem& code_item, uint32_t offset) { 693 const byte* handler_data = 694 reinterpret_cast<const byte*>(GetTryItems(code_item, code_item.tries_size_)); 695 return handler_data + offset; 696 } 697 698 // Find which try region is associated with the given address (ie dex pc). Returns -1 if none. 699 static int32_t FindTryItem(const CodeItem &code_item, uint32_t address); 700 701 // Find the handler offset associated with the given address (ie dex pc). Returns -1 if none. 702 static int32_t FindCatchHandlerOffset(const CodeItem &code_item, uint32_t address); 703 704 // Get the pointer to the start of the debugging data GetDebugInfoStream(const CodeItem * code_item)705 const byte* GetDebugInfoStream(const CodeItem* code_item) const { 706 if (code_item->debug_info_off_ == 0) { 707 return NULL; 708 } else { 709 return begin_ + code_item->debug_info_off_; 710 } 711 } 712 713 // Callback for "new position table entry". 714 // Returning true causes the decoder to stop early. 715 typedef bool (*DexDebugNewPositionCb)(void* context, uint32_t address, uint32_t line_num); 716 717 // Callback for "new locals table entry". "signature" is an empty string 718 // if no signature is available for an entry. 719 typedef void (*DexDebugNewLocalCb)(void* context, uint16_t reg, 720 uint32_t start_address, 721 uint32_t end_address, 722 const char* name, 723 const char* descriptor, 724 const char* signature); 725 726 static bool LineNumForPcCb(void* context, uint32_t address, uint32_t line_num); 727 728 // Debug info opcodes and constants 729 enum { 730 DBG_END_SEQUENCE = 0x00, 731 DBG_ADVANCE_PC = 0x01, 732 DBG_ADVANCE_LINE = 0x02, 733 DBG_START_LOCAL = 0x03, 734 DBG_START_LOCAL_EXTENDED = 0x04, 735 DBG_END_LOCAL = 0x05, 736 DBG_RESTART_LOCAL = 0x06, 737 DBG_SET_PROLOGUE_END = 0x07, 738 DBG_SET_EPILOGUE_BEGIN = 0x08, 739 DBG_SET_FILE = 0x09, 740 DBG_FIRST_SPECIAL = 0x0a, 741 DBG_LINE_BASE = -4, 742 DBG_LINE_RANGE = 15, 743 }; 744 745 struct LocalInfo { LocalInfoLocalInfo746 LocalInfo() 747 : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {} 748 749 const char* name_; // E.g., list 750 const char* descriptor_; // E.g., Ljava/util/LinkedList; 751 const char* signature_; // E.g., java.util.LinkedList<java.lang.Integer> 752 uint16_t start_address_; // PC location where the local is first defined. 753 bool is_live_; // Is the local defined and live. 754 755 private: 756 DISALLOW_COPY_AND_ASSIGN(LocalInfo); 757 }; 758 759 struct LineNumFromPcContext { LineNumFromPcContextLineNumFromPcContext760 LineNumFromPcContext(uint32_t address, uint32_t line_num) 761 : address_(address), line_num_(line_num) {} 762 uint32_t address_; 763 uint32_t line_num_; 764 private: 765 DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext); 766 }; 767 InvokeLocalCbIfLive(void * context,int reg,uint32_t end_address,LocalInfo * local_in_reg,DexDebugNewLocalCb local_cb)768 void InvokeLocalCbIfLive(void* context, int reg, uint32_t end_address, 769 LocalInfo* local_in_reg, DexDebugNewLocalCb local_cb) const { 770 if (local_cb != NULL && local_in_reg[reg].is_live_) { 771 local_cb(context, reg, local_in_reg[reg].start_address_, end_address, 772 local_in_reg[reg].name_, local_in_reg[reg].descriptor_, 773 local_in_reg[reg].signature_ != NULL ? local_in_reg[reg].signature_ : ""); 774 } 775 } 776 777 // Determine the source file line number based on the program counter. 778 // "pc" is an offset, in 16-bit units, from the start of the method's code. 779 // 780 // Returns -1 if no match was found (possibly because the source files were 781 // compiled without "-g", so no line number information is present). 782 // Returns -2 for native methods (as expected in exception traces). 783 // 784 // This is used by runtime; therefore use art::Method not art::DexFile::Method. 785 int32_t GetLineNumFromPC(const mirror::ArtMethod* method, uint32_t rel_pc) const 786 SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 787 788 void DecodeDebugInfo(const CodeItem* code_item, bool is_static, uint32_t method_idx, 789 DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb, 790 void* context) const; 791 GetSourceFile(const ClassDef & class_def)792 const char* GetSourceFile(const ClassDef& class_def) const { 793 if (class_def.source_file_idx_ == 0xffffffff) { 794 return NULL; 795 } else { 796 return StringDataByIdx(class_def.source_file_idx_); 797 } 798 } 799 800 int GetPermissions() const; 801 802 bool IsReadOnly() const; 803 804 bool EnableWrite() const; 805 806 bool DisableWrite() const; 807 Begin()808 const byte* Begin() const { 809 return begin_; 810 } 811 Size()812 size_t Size() const { 813 return size_; 814 } 815 816 private: 817 // Opens a .dex file 818 static const DexFile* OpenFile(int fd, 819 const std::string& location, 820 bool verify); 821 822 // Opens a dex file from within a .jar, .zip, or .apk file 823 static const DexFile* OpenZip(int fd, const std::string& location); 824 825 // Opens a .dex file at the given address backed by a MemMap 826 static const DexFile* OpenMemory(const std::string& location, 827 uint32_t location_checksum, 828 MemMap* mem_map); 829 830 // Opens a .dex file at the given address, optionally backed by a MemMap 831 static const DexFile* OpenMemory(const byte* dex_file, 832 size_t size, 833 const std::string& location, 834 uint32_t location_checksum, 835 MemMap* mem_map); 836 DexFile(const byte * base,size_t size,const std::string & location,uint32_t location_checksum,MemMap * mem_map)837 DexFile(const byte* base, size_t size, 838 const std::string& location, 839 uint32_t location_checksum, 840 MemMap* mem_map) 841 : begin_(base), 842 size_(size), 843 location_(location), 844 location_checksum_(location_checksum), 845 mem_map_(mem_map), 846 modification_lock("DEX modification lock"), 847 header_(0), 848 string_ids_(0), 849 type_ids_(0), 850 field_ids_(0), 851 method_ids_(0), 852 proto_ids_(0), 853 class_defs_(0) { 854 CHECK(begin_ != NULL) << GetLocation(); 855 CHECK_GT(size_, 0U) << GetLocation(); 856 } 857 858 // Top-level initializer that calls other Init methods. 859 bool Init(); 860 861 // Caches pointers into to the various file sections. 862 void InitMembers(); 863 864 // Returns true if the header magic and version numbers are of the expected values. 865 bool CheckMagicAndVersion() const; 866 867 void DecodeDebugInfo0(const CodeItem* code_item, bool is_static, uint32_t method_idx, 868 DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb, 869 void* context, const byte* stream, LocalInfo* local_in_reg) const; 870 871 // The base address of the memory mapping. 872 const byte* const begin_; 873 874 // The size of the underlying memory allocation in bytes. 875 const size_t size_; 876 877 // Typically the dex file name when available, alternatively some identifying string. 878 // 879 // The ClassLinker will use this to match DexFiles the boot class 880 // path to DexCache::GetLocation when loading from an image. 881 const std::string location_; 882 883 const uint32_t location_checksum_; 884 885 // Manages the underlying memory allocation. 886 UniquePtr<MemMap> mem_map_; 887 888 // The DEX-to-DEX compiler uses this lock to ensure thread safety when 889 // enabling write access to a read-only DEX file. 890 // TODO: move to Locks::dex_file_modification_lock. 891 Mutex modification_lock; 892 893 // Points to the header section. 894 const Header* header_; 895 896 // Points to the base of the string identifier list. 897 const StringId* string_ids_; 898 899 // Points to the base of the type identifier list. 900 const TypeId* type_ids_; 901 902 // Points to the base of the field identifier list. 903 const FieldId* field_ids_; 904 905 // Points to the base of the method identifier list. 906 const MethodId* method_ids_; 907 908 // Points to the base of the prototype identifier list. 909 const ProtoId* proto_ids_; 910 911 // Points to the base of the class definition list. 912 const ClassDef* class_defs_; 913 }; 914 915 // Iterate over a dex file's ProtoId's paramters 916 class DexFileParameterIterator { 917 public: DexFileParameterIterator(const DexFile & dex_file,const DexFile::ProtoId & proto_id)918 DexFileParameterIterator(const DexFile& dex_file, const DexFile::ProtoId& proto_id) 919 : dex_file_(dex_file), size_(0), pos_(0) { 920 type_list_ = dex_file_.GetProtoParameters(proto_id); 921 if (type_list_ != NULL) { 922 size_ = type_list_->Size(); 923 } 924 } HasNext()925 bool HasNext() const { return pos_ < size_; } Next()926 void Next() { ++pos_; } GetTypeIdx()927 uint16_t GetTypeIdx() { 928 return type_list_->GetTypeItem(pos_).type_idx_; 929 } GetDescriptor()930 const char* GetDescriptor() { 931 return dex_file_.StringByTypeIdx(GetTypeIdx()); 932 } 933 private: 934 const DexFile& dex_file_; 935 const DexFile::TypeList* type_list_; 936 uint32_t size_; 937 uint32_t pos_; 938 DISALLOW_IMPLICIT_CONSTRUCTORS(DexFileParameterIterator); 939 }; 940 941 // Iterate and decode class_data_item 942 class ClassDataItemIterator { 943 public: ClassDataItemIterator(const DexFile & dex_file,const byte * raw_class_data_item)944 ClassDataItemIterator(const DexFile& dex_file, const byte* raw_class_data_item) 945 : dex_file_(dex_file), pos_(0), ptr_pos_(raw_class_data_item), last_idx_(0) { 946 ReadClassDataHeader(); 947 if (EndOfInstanceFieldsPos() > 0) { 948 ReadClassDataField(); 949 } else if (EndOfVirtualMethodsPos() > 0) { 950 ReadClassDataMethod(); 951 } 952 } NumStaticFields()953 uint32_t NumStaticFields() const { 954 return header_.static_fields_size_; 955 } NumInstanceFields()956 uint32_t NumInstanceFields() const { 957 return header_.instance_fields_size_; 958 } NumDirectMethods()959 uint32_t NumDirectMethods() const { 960 return header_.direct_methods_size_; 961 } NumVirtualMethods()962 uint32_t NumVirtualMethods() const { 963 return header_.virtual_methods_size_; 964 } HasNextStaticField()965 bool HasNextStaticField() const { 966 return pos_ < EndOfStaticFieldsPos(); 967 } HasNextInstanceField()968 bool HasNextInstanceField() const { 969 return pos_ >= EndOfStaticFieldsPos() && pos_ < EndOfInstanceFieldsPos(); 970 } HasNextDirectMethod()971 bool HasNextDirectMethod() const { 972 return pos_ >= EndOfInstanceFieldsPos() && pos_ < EndOfDirectMethodsPos(); 973 } HasNextVirtualMethod()974 bool HasNextVirtualMethod() const { 975 return pos_ >= EndOfDirectMethodsPos() && pos_ < EndOfVirtualMethodsPos(); 976 } HasNext()977 bool HasNext() const { 978 return pos_ < EndOfVirtualMethodsPos(); 979 } Next()980 inline void Next() { 981 pos_++; 982 if (pos_ < EndOfStaticFieldsPos()) { 983 last_idx_ = GetMemberIndex(); 984 ReadClassDataField(); 985 } else if (pos_ == EndOfStaticFieldsPos() && NumInstanceFields() > 0) { 986 last_idx_ = 0; // transition to next array, reset last index 987 ReadClassDataField(); 988 } else if (pos_ < EndOfInstanceFieldsPos()) { 989 last_idx_ = GetMemberIndex(); 990 ReadClassDataField(); 991 } else if (pos_ == EndOfInstanceFieldsPos() && NumDirectMethods() > 0) { 992 last_idx_ = 0; // transition to next array, reset last index 993 ReadClassDataMethod(); 994 } else if (pos_ < EndOfDirectMethodsPos()) { 995 last_idx_ = GetMemberIndex(); 996 ReadClassDataMethod(); 997 } else if (pos_ == EndOfDirectMethodsPos() && NumVirtualMethods() > 0) { 998 last_idx_ = 0; // transition to next array, reset last index 999 ReadClassDataMethod(); 1000 } else if (pos_ < EndOfVirtualMethodsPos()) { 1001 last_idx_ = GetMemberIndex(); 1002 ReadClassDataMethod(); 1003 } else { 1004 DCHECK(!HasNext()); 1005 } 1006 } GetMemberIndex()1007 uint32_t GetMemberIndex() const { 1008 if (pos_ < EndOfInstanceFieldsPos()) { 1009 return last_idx_ + field_.field_idx_delta_; 1010 } else { 1011 DCHECK_LT(pos_, EndOfVirtualMethodsPos()); 1012 return last_idx_ + method_.method_idx_delta_; 1013 } 1014 } GetMemberAccessFlags()1015 uint32_t GetMemberAccessFlags() const { 1016 if (pos_ < EndOfInstanceFieldsPos()) { 1017 return field_.access_flags_; 1018 } else { 1019 DCHECK_LT(pos_, EndOfVirtualMethodsPos()); 1020 return method_.access_flags_; 1021 } 1022 } GetMethodInvokeType(const DexFile::ClassDef & class_def)1023 InvokeType GetMethodInvokeType(const DexFile::ClassDef& class_def) const { 1024 if (HasNextDirectMethod()) { 1025 if ((GetMemberAccessFlags() & kAccStatic) != 0) { 1026 return kStatic; 1027 } else { 1028 return kDirect; 1029 } 1030 } else { 1031 DCHECK_EQ(GetMemberAccessFlags() & kAccStatic, 0U); 1032 if ((class_def.access_flags_ & kAccInterface) != 0) { 1033 return kInterface; 1034 } else if ((GetMemberAccessFlags() & kAccConstructor) != 0) { 1035 return kSuper; 1036 } else { 1037 return kVirtual; 1038 } 1039 } 1040 } GetMethodCodeItem()1041 const DexFile::CodeItem* GetMethodCodeItem() const { 1042 return dex_file_.GetCodeItem(method_.code_off_); 1043 } GetMethodCodeItemOffset()1044 uint32_t GetMethodCodeItemOffset() const { 1045 return method_.code_off_; 1046 } EndDataPointer()1047 const byte* EndDataPointer() const { 1048 CHECK(!HasNext()); 1049 return ptr_pos_; 1050 } 1051 1052 private: 1053 // A dex file's class_data_item is leb128 encoded, this structure holds a decoded form of the 1054 // header for a class_data_item 1055 struct ClassDataHeader { 1056 uint32_t static_fields_size_; // the number of static fields 1057 uint32_t instance_fields_size_; // the number of instance fields 1058 uint32_t direct_methods_size_; // the number of direct methods 1059 uint32_t virtual_methods_size_; // the number of virtual methods 1060 } header_; 1061 1062 // Read and decode header from a class_data_item stream into header 1063 void ReadClassDataHeader(); 1064 EndOfStaticFieldsPos()1065 uint32_t EndOfStaticFieldsPos() const { 1066 return header_.static_fields_size_; 1067 } EndOfInstanceFieldsPos()1068 uint32_t EndOfInstanceFieldsPos() const { 1069 return EndOfStaticFieldsPos() + header_.instance_fields_size_; 1070 } EndOfDirectMethodsPos()1071 uint32_t EndOfDirectMethodsPos() const { 1072 return EndOfInstanceFieldsPos() + header_.direct_methods_size_; 1073 } EndOfVirtualMethodsPos()1074 uint32_t EndOfVirtualMethodsPos() const { 1075 return EndOfDirectMethodsPos() + header_.virtual_methods_size_; 1076 } 1077 1078 // A decoded version of the field of a class_data_item 1079 struct ClassDataField { 1080 uint32_t field_idx_delta_; // delta of index into the field_ids array for FieldId 1081 uint32_t access_flags_; // access flags for the field ClassDataFieldClassDataField1082 ClassDataField() : field_idx_delta_(0), access_flags_(0) {} 1083 1084 private: 1085 DISALLOW_COPY_AND_ASSIGN(ClassDataField); 1086 }; 1087 ClassDataField field_; 1088 1089 // Read and decode a field from a class_data_item stream into field 1090 void ReadClassDataField(); 1091 1092 // A decoded version of the method of a class_data_item 1093 struct ClassDataMethod { 1094 uint32_t method_idx_delta_; // delta of index into the method_ids array for MethodId 1095 uint32_t access_flags_; 1096 uint32_t code_off_; ClassDataMethodClassDataMethod1097 ClassDataMethod() : method_idx_delta_(0), access_flags_(0), code_off_(0) {} 1098 1099 private: 1100 DISALLOW_COPY_AND_ASSIGN(ClassDataMethod); 1101 }; 1102 ClassDataMethod method_; 1103 1104 // Read and decode a method from a class_data_item stream into method 1105 void ReadClassDataMethod(); 1106 1107 const DexFile& dex_file_; 1108 size_t pos_; // integral number of items passed 1109 const byte* ptr_pos_; // pointer into stream of class_data_item 1110 uint32_t last_idx_; // last read field or method index to apply delta to 1111 DISALLOW_IMPLICIT_CONSTRUCTORS(ClassDataItemIterator); 1112 }; 1113 1114 class EncodedStaticFieldValueIterator { 1115 public: 1116 EncodedStaticFieldValueIterator(const DexFile& dex_file, mirror::DexCache* dex_cache, 1117 mirror::ClassLoader* class_loader, 1118 ClassLinker* linker, const DexFile::ClassDef& class_def) 1119 SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 1120 1121 void ReadValueToField(mirror::ArtField* field) const SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 1122 HasNext()1123 bool HasNext() { return pos_ < array_size_; } 1124 1125 void Next(); 1126 1127 enum ValueType { 1128 kByte = 0x00, 1129 kShort = 0x02, 1130 kChar = 0x03, 1131 kInt = 0x04, 1132 kLong = 0x06, 1133 kFloat = 0x10, 1134 kDouble = 0x11, 1135 kString = 0x17, 1136 kType = 0x18, 1137 kField = 0x19, 1138 kMethod = 0x1a, 1139 kEnum = 0x1b, 1140 kArray = 0x1c, 1141 kAnnotation = 0x1d, 1142 kNull = 0x1e, 1143 kBoolean = 0x1f 1144 }; 1145 1146 private: 1147 static const byte kEncodedValueTypeMask = 0x1f; // 0b11111 1148 static const byte kEncodedValueArgShift = 5; 1149 1150 const DexFile& dex_file_; 1151 mirror::DexCache* dex_cache_; // Dex cache to resolve literal objects. 1152 mirror::ClassLoader* class_loader_; // ClassLoader to resolve types. 1153 ClassLinker* linker_; // Linker to resolve literal objects. 1154 size_t array_size_; // Size of array. 1155 size_t pos_; // Current position. 1156 const byte* ptr_; // Pointer into encoded data array. 1157 ValueType type_; // Type of current encoded value. 1158 jvalue jval_; // Value of current encoded value. 1159 DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedStaticFieldValueIterator); 1160 }; 1161 std::ostream& operator<<(std::ostream& os, const EncodedStaticFieldValueIterator::ValueType& code); 1162 1163 class CatchHandlerIterator { 1164 public: 1165 CatchHandlerIterator(const DexFile::CodeItem& code_item, uint32_t address); 1166 1167 CatchHandlerIterator(const DexFile::CodeItem& code_item, 1168 const DexFile::TryItem& try_item); 1169 CatchHandlerIterator(const byte * handler_data)1170 explicit CatchHandlerIterator(const byte* handler_data) { 1171 Init(handler_data); 1172 } 1173 GetHandlerTypeIndex()1174 uint16_t GetHandlerTypeIndex() const { 1175 return handler_.type_idx_; 1176 } GetHandlerAddress()1177 uint32_t GetHandlerAddress() const { 1178 return handler_.address_; 1179 } 1180 void Next(); HasNext()1181 bool HasNext() const { 1182 return remaining_count_ != -1 || catch_all_; 1183 } 1184 // End of this set of catch blocks, convenience method to locate next set of catch blocks EndDataPointer()1185 const byte* EndDataPointer() const { 1186 CHECK(!HasNext()); 1187 return current_data_; 1188 } 1189 1190 private: 1191 void Init(const DexFile::CodeItem& code_item, int32_t offset); 1192 void Init(const byte* handler_data); 1193 1194 struct CatchHandlerItem { 1195 uint16_t type_idx_; // type index of the caught exception type 1196 uint32_t address_; // handler address 1197 } handler_; 1198 const byte *current_data_; // the current handler in dex file. 1199 int32_t remaining_count_; // number of handlers not read. 1200 bool catch_all_; // is there a handler that will catch all exceptions in case 1201 // that all typed handler does not match. 1202 }; 1203 1204 } // namespace art 1205 1206 #endif // ART_RUNTIME_DEX_FILE_H_ 1207