dex_file.h revision 833a48501d560c9fa7fc78ef619888138c2d374f
1/* 2 * Copyright (C) 2011 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17#ifndef ART_RUNTIME_DEX_FILE_H_ 18#define ART_RUNTIME_DEX_FILE_H_ 19 20#include <memory> 21#include <string> 22#include <vector> 23 24#include "base/logging.h" 25#include "base/mutex.h" // For Locks::mutator_lock_. 26#include "globals.h" 27#include "invoke_type.h" 28#include "jni.h" 29#include "modifiers.h" 30#include "safe_map.h" 31 32namespace art { 33 34// TODO: remove dependencies on mirror classes, primarily by moving 35// EncodedStaticFieldValueIterator to its own file. 36namespace mirror { 37 class ArtField; 38 class ArtMethod; 39 class ClassLoader; 40 class DexCache; 41} // namespace mirror 42class ClassLinker; 43class MemMap; 44class Signature; 45template<class T> class Handle; 46class StringPiece; 47class ZipArchive; 48 49// TODO: move all of the macro functionality into the DexCache class. 50class DexFile { 51 public: 52 static const byte kDexMagic[]; 53 static const byte kDexMagicVersion[]; 54 static const size_t kSha1DigestSize = 20; 55 static const uint32_t kDexEndianConstant = 0x12345678; 56 57 // name of the DexFile entry within a zip archive 58 static const char* kClassesDex; 59 60 // The value of an invalid index. 61 static const uint32_t kDexNoIndex = 0xFFFFFFFF; 62 63 // The value of an invalid index. 64 static const uint16_t kDexNoIndex16 = 0xFFFF; 65 66 // The separator charactor in MultiDex locations. 67 static constexpr char kMultiDexSeparator = ':'; 68 69 // A string version of the previous. This is a define so that we can merge string literals in the 70 // preprocessor. 71 #define kMultiDexSeparatorString ":" 72 73 // Raw header_item. 74 struct Header { 75 uint8_t magic_[8]; 76 uint32_t checksum_; // See also location_checksum_ 77 uint8_t signature_[kSha1DigestSize]; 78 uint32_t file_size_; // size of entire file 79 uint32_t header_size_; // offset to start of next section 80 uint32_t endian_tag_; 81 uint32_t link_size_; // unused 82 uint32_t link_off_; // unused 83 uint32_t map_off_; // unused 84 uint32_t string_ids_size_; // number of StringIds 85 uint32_t string_ids_off_; // file offset of StringIds array 86 uint32_t type_ids_size_; // number of TypeIds, we don't support more than 65535 87 uint32_t type_ids_off_; // file offset of TypeIds array 88 uint32_t proto_ids_size_; // number of ProtoIds, we don't support more than 65535 89 uint32_t proto_ids_off_; // file offset of ProtoIds array 90 uint32_t field_ids_size_; // number of FieldIds 91 uint32_t field_ids_off_; // file offset of FieldIds array 92 uint32_t method_ids_size_; // number of MethodIds 93 uint32_t method_ids_off_; // file offset of MethodIds array 94 uint32_t class_defs_size_; // number of ClassDefs 95 uint32_t class_defs_off_; // file offset of ClassDef array 96 uint32_t data_size_; // unused 97 uint32_t data_off_; // unused 98 99 private: 100 DISALLOW_COPY_AND_ASSIGN(Header); 101 }; 102 103 // Map item type codes. 104 enum { 105 kDexTypeHeaderItem = 0x0000, 106 kDexTypeStringIdItem = 0x0001, 107 kDexTypeTypeIdItem = 0x0002, 108 kDexTypeProtoIdItem = 0x0003, 109 kDexTypeFieldIdItem = 0x0004, 110 kDexTypeMethodIdItem = 0x0005, 111 kDexTypeClassDefItem = 0x0006, 112 kDexTypeMapList = 0x1000, 113 kDexTypeTypeList = 0x1001, 114 kDexTypeAnnotationSetRefList = 0x1002, 115 kDexTypeAnnotationSetItem = 0x1003, 116 kDexTypeClassDataItem = 0x2000, 117 kDexTypeCodeItem = 0x2001, 118 kDexTypeStringDataItem = 0x2002, 119 kDexTypeDebugInfoItem = 0x2003, 120 kDexTypeAnnotationItem = 0x2004, 121 kDexTypeEncodedArrayItem = 0x2005, 122 kDexTypeAnnotationsDirectoryItem = 0x2006, 123 }; 124 125 struct MapItem { 126 uint16_t type_; 127 uint16_t unused_; 128 uint32_t size_; 129 uint32_t offset_; 130 131 private: 132 DISALLOW_COPY_AND_ASSIGN(MapItem); 133 }; 134 135 struct MapList { 136 uint32_t size_; 137 MapItem list_[1]; 138 139 private: 140 DISALLOW_COPY_AND_ASSIGN(MapList); 141 }; 142 143 // Raw string_id_item. 144 struct StringId { 145 uint32_t string_data_off_; // offset in bytes from the base address 146 147 private: 148 DISALLOW_COPY_AND_ASSIGN(StringId); 149 }; 150 151 // Raw type_id_item. 152 struct TypeId { 153 uint32_t descriptor_idx_; // index into string_ids 154 155 private: 156 DISALLOW_COPY_AND_ASSIGN(TypeId); 157 }; 158 159 // Raw field_id_item. 160 struct FieldId { 161 uint16_t class_idx_; // index into type_ids_ array for defining class 162 uint16_t type_idx_; // index into type_ids_ array for field type 163 uint32_t name_idx_; // index into string_ids_ array for field name 164 165 private: 166 DISALLOW_COPY_AND_ASSIGN(FieldId); 167 }; 168 169 // Raw method_id_item. 170 struct MethodId { 171 uint16_t class_idx_; // index into type_ids_ array for defining class 172 uint16_t proto_idx_; // index into proto_ids_ array for method prototype 173 uint32_t name_idx_; // index into string_ids_ array for method name 174 175 private: 176 DISALLOW_COPY_AND_ASSIGN(MethodId); 177 }; 178 179 // Raw proto_id_item. 180 struct ProtoId { 181 uint32_t shorty_idx_; // index into string_ids array for shorty descriptor 182 uint16_t return_type_idx_; // index into type_ids array for return type 183 uint16_t pad_; // padding = 0 184 uint32_t parameters_off_; // file offset to type_list for parameter types 185 186 private: 187 DISALLOW_COPY_AND_ASSIGN(ProtoId); 188 }; 189 190 // Raw class_def_item. 191 struct ClassDef { 192 uint16_t class_idx_; // index into type_ids_ array for this class 193 uint16_t pad1_; // padding = 0 194 uint32_t access_flags_; 195 uint16_t superclass_idx_; // index into type_ids_ array for superclass 196 uint16_t pad2_; // padding = 0 197 uint32_t interfaces_off_; // file offset to TypeList 198 uint32_t source_file_idx_; // index into string_ids_ for source file name 199 uint32_t annotations_off_; // file offset to annotations_directory_item 200 uint32_t class_data_off_; // file offset to class_data_item 201 uint32_t static_values_off_; // file offset to EncodedArray 202 203 private: 204 DISALLOW_COPY_AND_ASSIGN(ClassDef); 205 }; 206 207 // Raw type_item. 208 struct TypeItem { 209 uint16_t type_idx_; // index into type_ids section 210 211 private: 212 DISALLOW_COPY_AND_ASSIGN(TypeItem); 213 }; 214 215 // Raw type_list. 216 class TypeList { 217 public: 218 uint32_t Size() const { 219 return size_; 220 } 221 222 const TypeItem& GetTypeItem(uint32_t idx) const { 223 DCHECK_LT(idx, this->size_); 224 return this->list_[idx]; 225 } 226 227 private: 228 uint32_t size_; // size of the list, in entries 229 TypeItem list_[1]; // elements of the list 230 DISALLOW_COPY_AND_ASSIGN(TypeList); 231 }; 232 233 // Raw code_item. 234 struct CodeItem { 235 uint16_t registers_size_; 236 uint16_t ins_size_; 237 uint16_t outs_size_; 238 uint16_t tries_size_; 239 uint32_t debug_info_off_; // file offset to debug info stream 240 uint32_t insns_size_in_code_units_; // size of the insns array, in 2 byte code units 241 uint16_t insns_[1]; 242 243 private: 244 DISALLOW_COPY_AND_ASSIGN(CodeItem); 245 }; 246 247 // Raw try_item. 248 struct TryItem { 249 uint32_t start_addr_; 250 uint16_t insn_count_; 251 uint16_t handler_off_; 252 253 private: 254 DISALLOW_COPY_AND_ASSIGN(TryItem); 255 }; 256 257 // Annotation constants. 258 enum { 259 kDexVisibilityBuild = 0x00, /* annotation visibility */ 260 kDexVisibilityRuntime = 0x01, 261 kDexVisibilitySystem = 0x02, 262 263 kDexAnnotationByte = 0x00, 264 kDexAnnotationShort = 0x02, 265 kDexAnnotationChar = 0x03, 266 kDexAnnotationInt = 0x04, 267 kDexAnnotationLong = 0x06, 268 kDexAnnotationFloat = 0x10, 269 kDexAnnotationDouble = 0x11, 270 kDexAnnotationString = 0x17, 271 kDexAnnotationType = 0x18, 272 kDexAnnotationField = 0x19, 273 kDexAnnotationMethod = 0x1a, 274 kDexAnnotationEnum = 0x1b, 275 kDexAnnotationArray = 0x1c, 276 kDexAnnotationAnnotation = 0x1d, 277 kDexAnnotationNull = 0x1e, 278 kDexAnnotationBoolean = 0x1f, 279 280 kDexAnnotationValueTypeMask = 0x1f, /* low 5 bits */ 281 kDexAnnotationValueArgShift = 5, 282 }; 283 284 struct AnnotationsDirectoryItem { 285 uint32_t class_annotations_off_; 286 uint32_t fields_size_; 287 uint32_t methods_size_; 288 uint32_t parameters_size_; 289 290 private: 291 DISALLOW_COPY_AND_ASSIGN(AnnotationsDirectoryItem); 292 }; 293 294 struct FieldAnnotationsItem { 295 uint32_t field_idx_; 296 uint32_t annotations_off_; 297 298 private: 299 DISALLOW_COPY_AND_ASSIGN(FieldAnnotationsItem); 300 }; 301 302 struct MethodAnnotationsItem { 303 uint32_t method_idx_; 304 uint32_t annotations_off_; 305 306 private: 307 DISALLOW_COPY_AND_ASSIGN(MethodAnnotationsItem); 308 }; 309 310 struct ParameterAnnotationsItem { 311 uint32_t method_idx_; 312 uint32_t annotations_off_; 313 314 private: 315 DISALLOW_COPY_AND_ASSIGN(ParameterAnnotationsItem); 316 }; 317 318 struct AnnotationSetRefItem { 319 uint32_t annotations_off_; 320 321 private: 322 DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefItem); 323 }; 324 325 struct AnnotationSetRefList { 326 uint32_t size_; 327 AnnotationSetRefItem list_[1]; 328 329 private: 330 DISALLOW_COPY_AND_ASSIGN(AnnotationSetRefList); 331 }; 332 333 struct AnnotationSetItem { 334 uint32_t size_; 335 uint32_t entries_[1]; 336 337 private: 338 DISALLOW_COPY_AND_ASSIGN(AnnotationSetItem); 339 }; 340 341 struct AnnotationItem { 342 uint8_t visibility_; 343 uint8_t annotation_[1]; 344 345 private: 346 DISALLOW_COPY_AND_ASSIGN(AnnotationItem); 347 }; 348 349 typedef std::pair<const DexFile*, const DexFile::ClassDef*> ClassPathEntry; 350 typedef std::vector<const DexFile*> ClassPath; 351 352 // Search a collection of DexFiles for a descriptor 353 static ClassPathEntry FindInClassPath(const char* descriptor, 354 const ClassPath& class_path); 355 356 // Returns the checksum of a file for comparison with GetLocationChecksum(). 357 // For .dex files, this is the header checksum. 358 // For zip files, this is the classes.dex zip entry CRC32 checksum. 359 // Return true if the checksum could be found, false otherwise. 360 static bool GetChecksum(const char* filename, uint32_t* checksum, std::string* error_msg); 361 362 // Opens .dex files found in the container, guessing the container format based on file extension. 363 static bool Open(const char* filename, const char* location, std::string* error_msg, 364 std::vector<const DexFile*>* dex_files); 365 366 // Opens .dex file, backed by existing memory 367 static const DexFile* Open(const uint8_t* base, size_t size, 368 const std::string& location, 369 uint32_t location_checksum, 370 std::string* error_msg) { 371 return OpenMemory(base, size, location, location_checksum, NULL, error_msg); 372 } 373 374 // Open all classesXXX.dex files from a zip archive. 375 static bool OpenFromZip(const ZipArchive& zip_archive, const std::string& location, 376 std::string* error_msg, std::vector<const DexFile*>* dex_files); 377 378 // Closes a .dex file. 379 virtual ~DexFile(); 380 381 const std::string& GetLocation() const { 382 return location_; 383 } 384 385 // For DexFiles directly from .dex files, this is the checksum from the DexFile::Header. 386 // For DexFiles opened from a zip files, this will be the ZipEntry CRC32 of classes.dex. 387 uint32_t GetLocationChecksum() const { 388 return location_checksum_; 389 } 390 391 const Header& GetHeader() const { 392 DCHECK(header_ != NULL) << GetLocation(); 393 return *header_; 394 } 395 396 // Decode the dex magic version 397 uint32_t GetVersion() const; 398 399 // Returns true if the byte string points to the magic value. 400 static bool IsMagicValid(const byte* magic); 401 402 // Returns true if the byte string after the magic is the correct value. 403 static bool IsVersionValid(const byte* magic); 404 405 // Returns the number of string identifiers in the .dex file. 406 size_t NumStringIds() const { 407 DCHECK(header_ != NULL) << GetLocation(); 408 return header_->string_ids_size_; 409 } 410 411 // Returns the StringId at the specified index. 412 const StringId& GetStringId(uint32_t idx) const { 413 DCHECK_LT(idx, NumStringIds()) << GetLocation(); 414 return string_ids_[idx]; 415 } 416 417 uint32_t GetIndexForStringId(const StringId& string_id) const { 418 CHECK_GE(&string_id, string_ids_) << GetLocation(); 419 CHECK_LT(&string_id, string_ids_ + header_->string_ids_size_) << GetLocation(); 420 return &string_id - string_ids_; 421 } 422 423 int32_t GetStringLength(const StringId& string_id) const; 424 425 // Returns a pointer to the UTF-8 string data referred to by the given string_id as well as the 426 // length of the string when decoded as a UTF-16 string. Note the UTF-16 length is not the same 427 // as the string length of the string data. 428 const char* GetStringDataAndUtf16Length(const StringId& string_id, uint32_t* utf16_length) const; 429 430 const char* GetStringData(const StringId& string_id) const { 431 uint32_t ignored; 432 return GetStringDataAndUtf16Length(string_id, &ignored); 433 } 434 435 // Index version of GetStringDataAndUtf16Length. 436 const char* StringDataAndUtf16LengthByIdx(uint32_t idx, uint32_t* utf16_length) const { 437 if (idx == kDexNoIndex) { 438 *utf16_length = 0; 439 return NULL; 440 } 441 const StringId& string_id = GetStringId(idx); 442 return GetStringDataAndUtf16Length(string_id, utf16_length); 443 } 444 445 const char* StringDataByIdx(uint32_t idx) const { 446 uint32_t unicode_length; 447 return StringDataAndUtf16LengthByIdx(idx, &unicode_length); 448 } 449 450 // Looks up a string id for a given modified utf8 string. 451 const StringId* FindStringId(const char* string) const; 452 453 // Looks up a string id for a given utf16 string. 454 const StringId* FindStringId(const uint16_t* string) const; 455 456 // Returns the number of type identifiers in the .dex file. 457 size_t NumTypeIds() const { 458 DCHECK(header_ != NULL) << GetLocation(); 459 return header_->type_ids_size_; 460 } 461 462 // Returns the TypeId at the specified index. 463 const TypeId& GetTypeId(uint32_t idx) const { 464 DCHECK_LT(idx, NumTypeIds()) << GetLocation(); 465 return type_ids_[idx]; 466 } 467 468 uint16_t GetIndexForTypeId(const TypeId& type_id) const { 469 CHECK_GE(&type_id, type_ids_) << GetLocation(); 470 CHECK_LT(&type_id, type_ids_ + header_->type_ids_size_) << GetLocation(); 471 size_t result = &type_id - type_ids_; 472 DCHECK_LT(result, 65536U) << GetLocation(); 473 return static_cast<uint16_t>(result); 474 } 475 476 // Get the descriptor string associated with a given type index. 477 const char* StringByTypeIdx(uint32_t idx, uint32_t* unicode_length) const { 478 const TypeId& type_id = GetTypeId(idx); 479 return StringDataAndUtf16LengthByIdx(type_id.descriptor_idx_, unicode_length); 480 } 481 482 const char* StringByTypeIdx(uint32_t idx) const { 483 const TypeId& type_id = GetTypeId(idx); 484 return StringDataByIdx(type_id.descriptor_idx_); 485 } 486 487 // Returns the type descriptor string of a type id. 488 const char* GetTypeDescriptor(const TypeId& type_id) const { 489 return StringDataByIdx(type_id.descriptor_idx_); 490 } 491 492 // Looks up a type for the given string index 493 const TypeId* FindTypeId(uint32_t string_idx) const; 494 495 // Returns the number of field identifiers in the .dex file. 496 size_t NumFieldIds() const { 497 DCHECK(header_ != NULL) << GetLocation(); 498 return header_->field_ids_size_; 499 } 500 501 // Returns the FieldId at the specified index. 502 const FieldId& GetFieldId(uint32_t idx) const { 503 DCHECK_LT(idx, NumFieldIds()) << GetLocation(); 504 return field_ids_[idx]; 505 } 506 507 uint32_t GetIndexForFieldId(const FieldId& field_id) const { 508 CHECK_GE(&field_id, field_ids_) << GetLocation(); 509 CHECK_LT(&field_id, field_ids_ + header_->field_ids_size_) << GetLocation(); 510 return &field_id - field_ids_; 511 } 512 513 // Looks up a field by its declaring class, name and type 514 const FieldId* FindFieldId(const DexFile::TypeId& declaring_klass, 515 const DexFile::StringId& name, 516 const DexFile::TypeId& type) const; 517 518 // Returns the declaring class descriptor string of a field id. 519 const char* GetFieldDeclaringClassDescriptor(const FieldId& field_id) const { 520 const DexFile::TypeId& type_id = GetTypeId(field_id.class_idx_); 521 return GetTypeDescriptor(type_id); 522 } 523 524 // Returns the class descriptor string of a field id. 525 const char* GetFieldTypeDescriptor(const FieldId& field_id) const { 526 const DexFile::TypeId& type_id = GetTypeId(field_id.type_idx_); 527 return GetTypeDescriptor(type_id); 528 } 529 530 // Returns the name of a field id. 531 const char* GetFieldName(const FieldId& field_id) const { 532 return StringDataByIdx(field_id.name_idx_); 533 } 534 535 // Returns the number of method identifiers in the .dex file. 536 size_t NumMethodIds() const { 537 DCHECK(header_ != NULL) << GetLocation(); 538 return header_->method_ids_size_; 539 } 540 541 // Returns the MethodId at the specified index. 542 const MethodId& GetMethodId(uint32_t idx) const { 543 DCHECK_LT(idx, NumMethodIds()) << GetLocation(); 544 return method_ids_[idx]; 545 } 546 547 uint32_t GetIndexForMethodId(const MethodId& method_id) const { 548 CHECK_GE(&method_id, method_ids_) << GetLocation(); 549 CHECK_LT(&method_id, method_ids_ + header_->method_ids_size_) << GetLocation(); 550 return &method_id - method_ids_; 551 } 552 553 // Looks up a method by its declaring class, name and proto_id 554 const MethodId* FindMethodId(const DexFile::TypeId& declaring_klass, 555 const DexFile::StringId& name, 556 const DexFile::ProtoId& signature) const; 557 558 // Returns the declaring class descriptor string of a method id. 559 const char* GetMethodDeclaringClassDescriptor(const MethodId& method_id) const { 560 const DexFile::TypeId& type_id = GetTypeId(method_id.class_idx_); 561 return GetTypeDescriptor(type_id); 562 } 563 564 // Returns the prototype of a method id. 565 const ProtoId& GetMethodPrototype(const MethodId& method_id) const { 566 return GetProtoId(method_id.proto_idx_); 567 } 568 569 // Returns a representation of the signature of a method id. 570 const Signature GetMethodSignature(const MethodId& method_id) const; 571 572 // Returns the name of a method id. 573 const char* GetMethodName(const MethodId& method_id) const { 574 return StringDataByIdx(method_id.name_idx_); 575 } 576 577 // Returns the shorty of a method id. 578 const char* GetMethodShorty(const MethodId& method_id) const { 579 return StringDataByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_); 580 } 581 const char* GetMethodShorty(const MethodId& method_id, uint32_t* length) const { 582 // Using the UTF16 length is safe here as shorties are guaranteed to be ASCII characters. 583 return StringDataAndUtf16LengthByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_, length); 584 } 585 // Returns the number of class definitions in the .dex file. 586 size_t NumClassDefs() const { 587 DCHECK(header_ != NULL) << GetLocation(); 588 return header_->class_defs_size_; 589 } 590 591 // Returns the ClassDef at the specified index. 592 const ClassDef& GetClassDef(uint16_t idx) const { 593 DCHECK_LT(idx, NumClassDefs()) << GetLocation(); 594 return class_defs_[idx]; 595 } 596 597 uint16_t GetIndexForClassDef(const ClassDef& class_def) const { 598 CHECK_GE(&class_def, class_defs_) << GetLocation(); 599 CHECK_LT(&class_def, class_defs_ + header_->class_defs_size_) << GetLocation(); 600 return &class_def - class_defs_; 601 } 602 603 // Returns the class descriptor string of a class definition. 604 const char* GetClassDescriptor(const ClassDef& class_def) const { 605 return StringByTypeIdx(class_def.class_idx_); 606 } 607 608 // Looks up a class definition by its class descriptor. 609 const ClassDef* FindClassDef(const char* descriptor) const; 610 611 // Looks up a class definition by its type index. 612 const ClassDef* FindClassDef(uint16_t type_idx) const; 613 614 const TypeList* GetInterfacesList(const ClassDef& class_def) const { 615 if (class_def.interfaces_off_ == 0) { 616 return NULL; 617 } else { 618 const byte* addr = begin_ + class_def.interfaces_off_; 619 return reinterpret_cast<const TypeList*>(addr); 620 } 621 } 622 623 // Returns a pointer to the raw memory mapped class_data_item 624 const byte* GetClassData(const ClassDef& class_def) const { 625 if (class_def.class_data_off_ == 0) { 626 return NULL; 627 } else { 628 return begin_ + class_def.class_data_off_; 629 } 630 } 631 632 // 633 const CodeItem* GetCodeItem(const uint32_t code_off) const { 634 if (code_off == 0) { 635 return NULL; // native or abstract method 636 } else { 637 const byte* addr = begin_ + code_off; 638 return reinterpret_cast<const CodeItem*>(addr); 639 } 640 } 641 642 const char* GetReturnTypeDescriptor(const ProtoId& proto_id) const { 643 return StringByTypeIdx(proto_id.return_type_idx_); 644 } 645 646 // Returns the number of prototype identifiers in the .dex file. 647 size_t NumProtoIds() const { 648 DCHECK(header_ != NULL) << GetLocation(); 649 return header_->proto_ids_size_; 650 } 651 652 // Returns the ProtoId at the specified index. 653 const ProtoId& GetProtoId(uint32_t idx) const { 654 DCHECK_LT(idx, NumProtoIds()) << GetLocation(); 655 return proto_ids_[idx]; 656 } 657 658 uint16_t GetIndexForProtoId(const ProtoId& proto_id) const { 659 CHECK_GE(&proto_id, proto_ids_) << GetLocation(); 660 CHECK_LT(&proto_id, proto_ids_ + header_->proto_ids_size_) << GetLocation(); 661 return &proto_id - proto_ids_; 662 } 663 664 // Looks up a proto id for a given return type and signature type list 665 const ProtoId* FindProtoId(uint16_t return_type_idx, 666 const uint16_t* signature_type_idxs, uint32_t signature_length) const; 667 const ProtoId* FindProtoId(uint16_t return_type_idx, 668 const std::vector<uint16_t>& signature_type_idxs) const { 669 return FindProtoId(return_type_idx, &signature_type_idxs[0], signature_type_idxs.size()); 670 } 671 672 // Given a signature place the type ids into the given vector, returns true on success 673 bool CreateTypeList(const StringPiece& signature, uint16_t* return_type_idx, 674 std::vector<uint16_t>* param_type_idxs) const; 675 676 // Create a Signature from the given string signature or return Signature::NoSignature if not 677 // possible. 678 const Signature CreateSignature(const StringPiece& signature) const; 679 680 // Returns the short form method descriptor for the given prototype. 681 const char* GetShorty(uint32_t proto_idx) const { 682 const ProtoId& proto_id = GetProtoId(proto_idx); 683 return StringDataByIdx(proto_id.shorty_idx_); 684 } 685 686 const TypeList* GetProtoParameters(const ProtoId& proto_id) const { 687 if (proto_id.parameters_off_ == 0) { 688 return NULL; 689 } else { 690 const byte* addr = begin_ + proto_id.parameters_off_; 691 return reinterpret_cast<const TypeList*>(addr); 692 } 693 } 694 695 const byte* GetEncodedStaticFieldValuesArray(const ClassDef& class_def) const { 696 if (class_def.static_values_off_ == 0) { 697 return 0; 698 } else { 699 return begin_ + class_def.static_values_off_; 700 } 701 } 702 703 static const TryItem* GetTryItems(const CodeItem& code_item, uint32_t offset); 704 705 // Get the base of the encoded data for the given DexCode. 706 static const byte* GetCatchHandlerData(const CodeItem& code_item, uint32_t offset) { 707 const byte* handler_data = 708 reinterpret_cast<const byte*>(GetTryItems(code_item, code_item.tries_size_)); 709 return handler_data + offset; 710 } 711 712 // Find which try region is associated with the given address (ie dex pc). Returns -1 if none. 713 static int32_t FindTryItem(const CodeItem &code_item, uint32_t address); 714 715 // Find the handler offset associated with the given address (ie dex pc). Returns -1 if none. 716 static int32_t FindCatchHandlerOffset(const CodeItem &code_item, uint32_t address); 717 718 // Get the pointer to the start of the debugging data 719 const byte* GetDebugInfoStream(const CodeItem* code_item) const { 720 if (code_item->debug_info_off_ == 0) { 721 return NULL; 722 } else { 723 return begin_ + code_item->debug_info_off_; 724 } 725 } 726 727 // Callback for "new position table entry". 728 // Returning true causes the decoder to stop early. 729 typedef bool (*DexDebugNewPositionCb)(void* context, uint32_t address, uint32_t line_num); 730 731 // Callback for "new locals table entry". "signature" is an empty string 732 // if no signature is available for an entry. 733 typedef void (*DexDebugNewLocalCb)(void* context, uint16_t reg, 734 uint32_t start_address, 735 uint32_t end_address, 736 const char* name, 737 const char* descriptor, 738 const char* signature); 739 740 static bool LineNumForPcCb(void* context, uint32_t address, uint32_t line_num); 741 742 // Debug info opcodes and constants 743 enum { 744 DBG_END_SEQUENCE = 0x00, 745 DBG_ADVANCE_PC = 0x01, 746 DBG_ADVANCE_LINE = 0x02, 747 DBG_START_LOCAL = 0x03, 748 DBG_START_LOCAL_EXTENDED = 0x04, 749 DBG_END_LOCAL = 0x05, 750 DBG_RESTART_LOCAL = 0x06, 751 DBG_SET_PROLOGUE_END = 0x07, 752 DBG_SET_EPILOGUE_BEGIN = 0x08, 753 DBG_SET_FILE = 0x09, 754 DBG_FIRST_SPECIAL = 0x0a, 755 DBG_LINE_BASE = -4, 756 DBG_LINE_RANGE = 15, 757 }; 758 759 struct LocalInfo { 760 LocalInfo() 761 : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {} 762 763 const char* name_; // E.g., list 764 const char* descriptor_; // E.g., Ljava/util/LinkedList; 765 const char* signature_; // E.g., java.util.LinkedList<java.lang.Integer> 766 uint16_t start_address_; // PC location where the local is first defined. 767 bool is_live_; // Is the local defined and live. 768 769 private: 770 DISALLOW_COPY_AND_ASSIGN(LocalInfo); 771 }; 772 773 struct LineNumFromPcContext { 774 LineNumFromPcContext(uint32_t address, uint32_t line_num) 775 : address_(address), line_num_(line_num) {} 776 uint32_t address_; 777 uint32_t line_num_; 778 private: 779 DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext); 780 }; 781 782 void InvokeLocalCbIfLive(void* context, int reg, uint32_t end_address, 783 LocalInfo* local_in_reg, DexDebugNewLocalCb local_cb) const { 784 if (local_cb != NULL && local_in_reg[reg].is_live_) { 785 local_cb(context, reg, local_in_reg[reg].start_address_, end_address, 786 local_in_reg[reg].name_, local_in_reg[reg].descriptor_, 787 local_in_reg[reg].signature_ != NULL ? local_in_reg[reg].signature_ : ""); 788 } 789 } 790 791 // Determine the source file line number based on the program counter. 792 // "pc" is an offset, in 16-bit units, from the start of the method's code. 793 // 794 // Returns -1 if no match was found (possibly because the source files were 795 // compiled without "-g", so no line number information is present). 796 // Returns -2 for native methods (as expected in exception traces). 797 // 798 // This is used by runtime; therefore use art::Method not art::DexFile::Method. 799 int32_t GetLineNumFromPC(mirror::ArtMethod* method, uint32_t rel_pc) const 800 SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 801 802 void DecodeDebugInfo(const CodeItem* code_item, bool is_static, uint32_t method_idx, 803 DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb, 804 void* context) const; 805 806 const char* GetSourceFile(const ClassDef& class_def) const { 807 if (class_def.source_file_idx_ == 0xffffffff) { 808 return NULL; 809 } else { 810 return StringDataByIdx(class_def.source_file_idx_); 811 } 812 } 813 814 int GetPermissions() const; 815 816 bool IsReadOnly() const; 817 818 bool EnableWrite() const; 819 820 bool DisableWrite() const; 821 822 const byte* Begin() const { 823 return begin_; 824 } 825 826 size_t Size() const { 827 return size_; 828 } 829 830 private: 831 // Opens a .dex file 832 static const DexFile* OpenFile(int fd, const char* location, bool verify, std::string* error_msg); 833 834 // Opens dex files from within a .jar, .zip, or .apk file 835 static bool OpenZip(int fd, const std::string& location, std::string* error_msg, 836 std::vector<const DexFile*>* dex_files); 837 838 enum class ZipOpenErrorCode { // private 839 kNoError, 840 kEntryNotFound, 841 kExtractToMemoryError, 842 kDexFileError, 843 kMakeReadOnlyError, 844 kVerifyError 845 }; 846 847 // Opens .dex file from the entry_name in a zip archive. error_code is undefined when non-nullptr 848 // return. 849 static const DexFile* Open(const ZipArchive& zip_archive, const char* entry_name, 850 const std::string& location, std::string* error_msg, 851 ZipOpenErrorCode* error_code); 852 853 // Opens a .dex file at the given address backed by a MemMap 854 static const DexFile* OpenMemory(const std::string& location, 855 uint32_t location_checksum, 856 MemMap* mem_map, 857 std::string* error_msg); 858 859 // Opens a .dex file at the given address, optionally backed by a MemMap 860 static const DexFile* OpenMemory(const byte* dex_file, 861 size_t size, 862 const std::string& location, 863 uint32_t location_checksum, 864 MemMap* mem_map, 865 std::string* error_msg); 866 867 DexFile(const byte* base, size_t size, 868 const std::string& location, 869 uint32_t location_checksum, 870 MemMap* mem_map); 871 872 // Top-level initializer that calls other Init methods. 873 bool Init(std::string* error_msg); 874 875 // Returns true if the header magic and version numbers are of the expected values. 876 bool CheckMagicAndVersion(std::string* error_msg) const; 877 878 void DecodeDebugInfo0(const CodeItem* code_item, bool is_static, uint32_t method_idx, 879 DexDebugNewPositionCb position_cb, DexDebugNewLocalCb local_cb, 880 void* context, const byte* stream, LocalInfo* local_in_reg) const; 881 882 // Check whether a location denotes a multidex dex file. This is a very simple check: returns 883 // whether the string contains the separator character. 884 static bool IsMultiDexLocation(const char* location); 885 886 // Splits a multidex location at the last separator character. The second component is a pointer 887 // to the character after the separator. The first is a copy of the substring up to the separator. 888 // 889 // Note: It's the caller's job to free the first component of the returned pair. 890 // Bug 15313523: gcc/libc++ don't allow a unique_ptr for the first component 891 static std::pair<const char*, const char*> SplitMultiDexLocation(const char* location); 892 893 894 // The base address of the memory mapping. 895 const byte* const begin_; 896 897 // The size of the underlying memory allocation in bytes. 898 const size_t size_; 899 900 // Typically the dex file name when available, alternatively some identifying string. 901 // 902 // The ClassLinker will use this to match DexFiles the boot class 903 // path to DexCache::GetLocation when loading from an image. 904 const std::string location_; 905 906 const uint32_t location_checksum_; 907 908 // Manages the underlying memory allocation. 909 std::unique_ptr<MemMap> mem_map_; 910 911 // Points to the header section. 912 const Header* const header_; 913 914 // Points to the base of the string identifier list. 915 const StringId* const string_ids_; 916 917 // Points to the base of the type identifier list. 918 const TypeId* const type_ids_; 919 920 // Points to the base of the field identifier list. 921 const FieldId* const field_ids_; 922 923 // Points to the base of the method identifier list. 924 const MethodId* const method_ids_; 925 926 // Points to the base of the prototype identifier list. 927 const ProtoId* const proto_ids_; 928 929 // Points to the base of the class definition list. 930 const ClassDef* const class_defs_; 931}; 932std::ostream& operator<<(std::ostream& os, const DexFile& dex_file); 933 934// Iterate over a dex file's ProtoId's paramters 935class DexFileParameterIterator { 936 public: 937 DexFileParameterIterator(const DexFile& dex_file, const DexFile::ProtoId& proto_id) 938 : dex_file_(dex_file), size_(0), pos_(0) { 939 type_list_ = dex_file_.GetProtoParameters(proto_id); 940 if (type_list_ != NULL) { 941 size_ = type_list_->Size(); 942 } 943 } 944 bool HasNext() const { return pos_ < size_; } 945 void Next() { ++pos_; } 946 uint16_t GetTypeIdx() { 947 return type_list_->GetTypeItem(pos_).type_idx_; 948 } 949 const char* GetDescriptor() { 950 return dex_file_.StringByTypeIdx(GetTypeIdx()); 951 } 952 private: 953 const DexFile& dex_file_; 954 const DexFile::TypeList* type_list_; 955 uint32_t size_; 956 uint32_t pos_; 957 DISALLOW_IMPLICIT_CONSTRUCTORS(DexFileParameterIterator); 958}; 959 960// Abstract the signature of a method. 961class Signature { 962 public: 963 std::string ToString() const; 964 965 static Signature NoSignature() { 966 return Signature(); 967 } 968 969 bool operator==(const Signature& rhs) const; 970 bool operator!=(const Signature& rhs) const { 971 return !(*this == rhs); 972 } 973 974 bool operator==(const StringPiece& rhs) const; 975 976 private: 977 Signature(const DexFile* dex, const DexFile::ProtoId& proto) : dex_file_(dex), proto_id_(&proto) { 978 } 979 980 Signature() : dex_file_(nullptr), proto_id_(nullptr) { 981 } 982 983 friend class DexFile; 984 985 const DexFile* const dex_file_; 986 const DexFile::ProtoId* const proto_id_; 987}; 988std::ostream& operator<<(std::ostream& os, const Signature& sig); 989 990// Iterate and decode class_data_item 991class ClassDataItemIterator { 992 public: 993 ClassDataItemIterator(const DexFile& dex_file, const byte* raw_class_data_item) 994 : dex_file_(dex_file), pos_(0), ptr_pos_(raw_class_data_item), last_idx_(0) { 995 ReadClassDataHeader(); 996 if (EndOfInstanceFieldsPos() > 0) { 997 ReadClassDataField(); 998 } else if (EndOfVirtualMethodsPos() > 0) { 999 ReadClassDataMethod(); 1000 } 1001 } 1002 uint32_t NumStaticFields() const { 1003 return header_.static_fields_size_; 1004 } 1005 uint32_t NumInstanceFields() const { 1006 return header_.instance_fields_size_; 1007 } 1008 uint32_t NumDirectMethods() const { 1009 return header_.direct_methods_size_; 1010 } 1011 uint32_t NumVirtualMethods() const { 1012 return header_.virtual_methods_size_; 1013 } 1014 bool HasNextStaticField() const { 1015 return pos_ < EndOfStaticFieldsPos(); 1016 } 1017 bool HasNextInstanceField() const { 1018 return pos_ >= EndOfStaticFieldsPos() && pos_ < EndOfInstanceFieldsPos(); 1019 } 1020 bool HasNextDirectMethod() const { 1021 return pos_ >= EndOfInstanceFieldsPos() && pos_ < EndOfDirectMethodsPos(); 1022 } 1023 bool HasNextVirtualMethod() const { 1024 return pos_ >= EndOfDirectMethodsPos() && pos_ < EndOfVirtualMethodsPos(); 1025 } 1026 bool HasNext() const { 1027 return pos_ < EndOfVirtualMethodsPos(); 1028 } 1029 inline void Next() { 1030 pos_++; 1031 if (pos_ < EndOfStaticFieldsPos()) { 1032 last_idx_ = GetMemberIndex(); 1033 ReadClassDataField(); 1034 } else if (pos_ == EndOfStaticFieldsPos() && NumInstanceFields() > 0) { 1035 last_idx_ = 0; // transition to next array, reset last index 1036 ReadClassDataField(); 1037 } else if (pos_ < EndOfInstanceFieldsPos()) { 1038 last_idx_ = GetMemberIndex(); 1039 ReadClassDataField(); 1040 } else if (pos_ == EndOfInstanceFieldsPos() && NumDirectMethods() > 0) { 1041 last_idx_ = 0; // transition to next array, reset last index 1042 ReadClassDataMethod(); 1043 } else if (pos_ < EndOfDirectMethodsPos()) { 1044 last_idx_ = GetMemberIndex(); 1045 ReadClassDataMethod(); 1046 } else if (pos_ == EndOfDirectMethodsPos() && NumVirtualMethods() > 0) { 1047 last_idx_ = 0; // transition to next array, reset last index 1048 ReadClassDataMethod(); 1049 } else if (pos_ < EndOfVirtualMethodsPos()) { 1050 last_idx_ = GetMemberIndex(); 1051 ReadClassDataMethod(); 1052 } else { 1053 DCHECK(!HasNext()); 1054 } 1055 } 1056 uint32_t GetMemberIndex() const { 1057 if (pos_ < EndOfInstanceFieldsPos()) { 1058 return last_idx_ + field_.field_idx_delta_; 1059 } else { 1060 DCHECK_LT(pos_, EndOfVirtualMethodsPos()); 1061 return last_idx_ + method_.method_idx_delta_; 1062 } 1063 } 1064 uint32_t GetMemberAccessFlags() const { 1065 if (pos_ < EndOfInstanceFieldsPos()) { 1066 return field_.access_flags_; 1067 } else { 1068 DCHECK_LT(pos_, EndOfVirtualMethodsPos()); 1069 return method_.access_flags_; 1070 } 1071 } 1072 InvokeType GetMethodInvokeType(const DexFile::ClassDef& class_def) const { 1073 if (HasNextDirectMethod()) { 1074 if ((GetMemberAccessFlags() & kAccStatic) != 0) { 1075 return kStatic; 1076 } else { 1077 return kDirect; 1078 } 1079 } else { 1080 DCHECK_EQ(GetMemberAccessFlags() & kAccStatic, 0U); 1081 if ((class_def.access_flags_ & kAccInterface) != 0) { 1082 return kInterface; 1083 } else if ((GetMemberAccessFlags() & kAccConstructor) != 0) { 1084 return kSuper; 1085 } else { 1086 return kVirtual; 1087 } 1088 } 1089 } 1090 const DexFile::CodeItem* GetMethodCodeItem() const { 1091 return dex_file_.GetCodeItem(method_.code_off_); 1092 } 1093 uint32_t GetMethodCodeItemOffset() const { 1094 return method_.code_off_; 1095 } 1096 const byte* EndDataPointer() const { 1097 CHECK(!HasNext()); 1098 return ptr_pos_; 1099 } 1100 1101 private: 1102 // A dex file's class_data_item is leb128 encoded, this structure holds a decoded form of the 1103 // header for a class_data_item 1104 struct ClassDataHeader { 1105 uint32_t static_fields_size_; // the number of static fields 1106 uint32_t instance_fields_size_; // the number of instance fields 1107 uint32_t direct_methods_size_; // the number of direct methods 1108 uint32_t virtual_methods_size_; // the number of virtual methods 1109 } header_; 1110 1111 // Read and decode header from a class_data_item stream into header 1112 void ReadClassDataHeader(); 1113 1114 uint32_t EndOfStaticFieldsPos() const { 1115 return header_.static_fields_size_; 1116 } 1117 uint32_t EndOfInstanceFieldsPos() const { 1118 return EndOfStaticFieldsPos() + header_.instance_fields_size_; 1119 } 1120 uint32_t EndOfDirectMethodsPos() const { 1121 return EndOfInstanceFieldsPos() + header_.direct_methods_size_; 1122 } 1123 uint32_t EndOfVirtualMethodsPos() const { 1124 return EndOfDirectMethodsPos() + header_.virtual_methods_size_; 1125 } 1126 1127 // A decoded version of the field of a class_data_item 1128 struct ClassDataField { 1129 uint32_t field_idx_delta_; // delta of index into the field_ids array for FieldId 1130 uint32_t access_flags_; // access flags for the field 1131 ClassDataField() : field_idx_delta_(0), access_flags_(0) {} 1132 1133 private: 1134 DISALLOW_COPY_AND_ASSIGN(ClassDataField); 1135 }; 1136 ClassDataField field_; 1137 1138 // Read and decode a field from a class_data_item stream into field 1139 void ReadClassDataField(); 1140 1141 // A decoded version of the method of a class_data_item 1142 struct ClassDataMethod { 1143 uint32_t method_idx_delta_; // delta of index into the method_ids array for MethodId 1144 uint32_t access_flags_; 1145 uint32_t code_off_; 1146 ClassDataMethod() : method_idx_delta_(0), access_flags_(0), code_off_(0) {} 1147 1148 private: 1149 DISALLOW_COPY_AND_ASSIGN(ClassDataMethod); 1150 }; 1151 ClassDataMethod method_; 1152 1153 // Read and decode a method from a class_data_item stream into method 1154 void ReadClassDataMethod(); 1155 1156 const DexFile& dex_file_; 1157 size_t pos_; // integral number of items passed 1158 const byte* ptr_pos_; // pointer into stream of class_data_item 1159 uint32_t last_idx_; // last read field or method index to apply delta to 1160 DISALLOW_IMPLICIT_CONSTRUCTORS(ClassDataItemIterator); 1161}; 1162 1163class EncodedStaticFieldValueIterator { 1164 public: 1165 EncodedStaticFieldValueIterator(const DexFile& dex_file, Handle<mirror::DexCache>* dex_cache, 1166 Handle<mirror::ClassLoader>* class_loader, 1167 ClassLinker* linker, const DexFile::ClassDef& class_def) 1168 SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 1169 1170 template<bool kTransactionActive> 1171 void ReadValueToField(mirror::ArtField* field) const SHARED_LOCKS_REQUIRED(Locks::mutator_lock_); 1172 1173 bool HasNext() { return pos_ < array_size_; } 1174 1175 void Next(); 1176 1177 enum ValueType { 1178 kByte = 0x00, 1179 kShort = 0x02, 1180 kChar = 0x03, 1181 kInt = 0x04, 1182 kLong = 0x06, 1183 kFloat = 0x10, 1184 kDouble = 0x11, 1185 kString = 0x17, 1186 kType = 0x18, 1187 kField = 0x19, 1188 kMethod = 0x1a, 1189 kEnum = 0x1b, 1190 kArray = 0x1c, 1191 kAnnotation = 0x1d, 1192 kNull = 0x1e, 1193 kBoolean = 0x1f 1194 }; 1195 1196 private: 1197 static const byte kEncodedValueTypeMask = 0x1f; // 0b11111 1198 static const byte kEncodedValueArgShift = 5; 1199 1200 const DexFile& dex_file_; 1201 Handle<mirror::DexCache>* const dex_cache_; // Dex cache to resolve literal objects. 1202 Handle<mirror::ClassLoader>* const class_loader_; // ClassLoader to resolve types. 1203 ClassLinker* linker_; // Linker to resolve literal objects. 1204 size_t array_size_; // Size of array. 1205 size_t pos_; // Current position. 1206 const byte* ptr_; // Pointer into encoded data array. 1207 ValueType type_; // Type of current encoded value. 1208 jvalue jval_; // Value of current encoded value. 1209 DISALLOW_IMPLICIT_CONSTRUCTORS(EncodedStaticFieldValueIterator); 1210}; 1211std::ostream& operator<<(std::ostream& os, const EncodedStaticFieldValueIterator::ValueType& code); 1212 1213class CatchHandlerIterator { 1214 public: 1215 CatchHandlerIterator(const DexFile::CodeItem& code_item, uint32_t address); 1216 1217 CatchHandlerIterator(const DexFile::CodeItem& code_item, 1218 const DexFile::TryItem& try_item); 1219 1220 explicit CatchHandlerIterator(const byte* handler_data) { 1221 Init(handler_data); 1222 } 1223 1224 uint16_t GetHandlerTypeIndex() const { 1225 return handler_.type_idx_; 1226 } 1227 uint32_t GetHandlerAddress() const { 1228 return handler_.address_; 1229 } 1230 void Next(); 1231 bool HasNext() const { 1232 return remaining_count_ != -1 || catch_all_; 1233 } 1234 // End of this set of catch blocks, convenience method to locate next set of catch blocks 1235 const byte* EndDataPointer() const { 1236 CHECK(!HasNext()); 1237 return current_data_; 1238 } 1239 1240 private: 1241 void Init(const DexFile::CodeItem& code_item, int32_t offset); 1242 void Init(const byte* handler_data); 1243 1244 struct CatchHandlerItem { 1245 uint16_t type_idx_; // type index of the caught exception type 1246 uint32_t address_; // handler address 1247 } handler_; 1248 const byte *current_data_; // the current handler in dex file. 1249 int32_t remaining_count_; // number of handlers not read. 1250 bool catch_all_; // is there a handler that will catch all exceptions in case 1251 // that all typed handler does not match. 1252}; 1253 1254} // namespace art 1255 1256#endif // ART_RUNTIME_DEX_FILE_H_ 1257