blob: 98e79a7ed5a7cc8791738dec03aa2ff7fa2b254a [file] [log] [blame]
Carl Shapiro1fb86202011-06-27 17:43:13 -07001// Copyright 2011 Google Inc. All Rights Reserved.
2
3#ifndef ART_SRC_DEX_FILE_H_
4#define ART_SRC_DEX_FILE_H_
5
Brian Carlstrom7e49dca2011-07-22 18:07:34 -07006#include <map>
Elliott Hughes0c424cb2011-08-26 10:16:25 -07007#include <string>
Brian Carlstrom74eb46a2011-08-02 20:10:14 -07008#include <vector>
Brian Carlstrom7e49dca2011-07-22 18:07:34 -07009
Elliott Hughes90a33692011-08-30 13:27:07 -070010#include "UniquePtr.h"
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070011#include "globals.h"
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070012#include "leb128.h"
13#include "logging.h"
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070014#include "stringpiece.h"
15#include "strutil.h"
Shih-wei Liao2fb97532011-08-11 16:17:23 -070016#include "utils.h"
Carl Shapiro1fb86202011-06-27 17:43:13 -070017
18namespace art {
19
Carl Shapiro5fafe2b2011-07-09 15:34:41 -070020union JValue;
Shih-wei Liao195487c2011-08-20 13:29:04 -070021class String;
22class Method;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -070023
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070024// TODO: move all of the macro functionality into the DexCache class.
Brian Carlstromf615a612011-07-23 12:50:34 -070025class DexFile {
Carl Shapiro1fb86202011-06-27 17:43:13 -070026 public:
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070027 static const byte kDexMagic[];
28 static const byte kDexMagicVersion[];
29 static const size_t kSha1DigestSize = 20;
Carl Shapiro80d4dde2011-06-28 16:24:07 -070030
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070031 static const byte kEncodedValueTypeMask = 0x1f; // 0b11111
32 static const byte kEncodedValueArgShift = 5;
33
34 // The value of an invalid index.
35 static const uint32_t kDexNoIndex = 0xFFFFFFFF;
36
37 enum ValueType {
38 kByte = 0x00,
39 kShort = 0x02,
40 kChar = 0x03,
41 kInt = 0x04,
42 kLong = 0x06,
43 kFloat = 0x10,
44 kDouble = 0x11,
45 kString = 0x17,
46 kType = 0x18,
47 kField = 0x19,
48 kMethod = 0x1a,
49 kEnum = 0x1b,
50 kArray = 0x1c,
51 kAnnotation = 0x1d,
52 kNull = 0x1e,
53 kBoolean = 0x1f
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070054 };
Carl Shapiro1fb86202011-06-27 17:43:13 -070055
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070056 // Raw header_item.
57 struct Header {
58 uint8_t magic_[8];
59 uint32_t checksum_;
60 uint8_t signature_[kSha1DigestSize];
61 uint32_t file_size_; // length of entire file
62 uint32_t header_size_; // offset to start of next section
63 uint32_t endian_tag_;
64 uint32_t link_size_;
65 uint32_t link_off_;
66 uint32_t map_off_;
67 uint32_t string_ids_size_;
68 uint32_t string_ids_off_;
69 uint32_t type_ids_size_;
70 uint32_t type_ids_off_;
71 uint32_t proto_ids_size_;
72 uint32_t proto_ids_off_;
73 uint32_t field_ids_size_;
74 uint32_t field_ids_off_;
75 uint32_t method_ids_size_;
76 uint32_t method_ids_off_;
77 uint32_t class_defs_size_;
78 uint32_t class_defs_off_;
79 uint32_t data_size_;
80 uint32_t data_off_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -070081 private:
82 DISALLOW_COPY_AND_ASSIGN(Header);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070083 };
Carl Shapiro1fb86202011-06-27 17:43:13 -070084
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070085 // Raw string_id_item.
86 struct StringId {
87 uint32_t string_data_off_; // offset in bytes from the base address
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -070088 private:
89 DISALLOW_COPY_AND_ASSIGN(StringId);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070090 };
91
92 // Raw type_id_item.
93 struct TypeId {
94 uint32_t descriptor_idx_; // index into string_ids
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -070095 private:
96 DISALLOW_COPY_AND_ASSIGN(TypeId);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070097 };
98
99 // Raw field_id_item.
100 struct FieldId {
Brian Carlstrom4a96b602011-07-26 16:40:23 -0700101 uint16_t class_idx_; // index into type_ids_ list for defining class
102 uint16_t type_idx_; // index into type_ids_ for field type
103 uint32_t name_idx_; // index into string_ids_ for field name
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700104 private:
105 DISALLOW_COPY_AND_ASSIGN(FieldId);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700106 };
107
108 // Raw method_id_item.
109 struct MethodId {
Brian Carlstrom4a96b602011-07-26 16:40:23 -0700110 uint16_t class_idx_; // index into type_ids_ list for defining class
111 uint16_t proto_idx_; // index into proto_ids_ for method prototype
112 uint32_t name_idx_; // index into string_ids_ for method name
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700113 private:
114 DISALLOW_COPY_AND_ASSIGN(MethodId);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700115 };
116
117 // Raw proto_id_item.
118 struct ProtoId {
119 uint32_t shorty_idx_; // index into string_ids for shorty descriptor
120 uint32_t return_type_idx_; // index into type_ids list for return type
121 uint32_t parameters_off_; // file offset to type_list for parameter types
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700122 private:
123 DISALLOW_COPY_AND_ASSIGN(ProtoId);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700124 };
125
126 // Raw class_def_item.
127 struct ClassDef {
Brian Carlstrom4a96b602011-07-26 16:40:23 -0700128 uint32_t class_idx_; // index into type_ids_ for this class
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700129 uint32_t access_flags_;
Brian Carlstrom4a96b602011-07-26 16:40:23 -0700130 uint32_t superclass_idx_; // index into type_ids_ for superclass
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700131 uint32_t interfaces_off_; // file offset to TypeList
Brian Carlstrom4a96b602011-07-26 16:40:23 -0700132 uint32_t source_file_idx_; // index into string_ids_ for source file name
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700133 uint32_t annotations_off_; // file offset to annotations_directory_item
134 uint32_t class_data_off_; // file offset to class_data_item
135 uint32_t static_values_off_; // file offset to EncodedArray
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700136 private:
137 DISALLOW_COPY_AND_ASSIGN(ClassDef);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700138 };
139
140 // Raw type_item.
141 struct TypeItem {
142 uint16_t type_idx_; // index into type_ids section
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700143 private:
144 DISALLOW_COPY_AND_ASSIGN(TypeItem);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700145 };
146
147 // Raw type_list.
148 class TypeList {
149 public:
150 uint32_t Size() const {
151 return size_;
152 }
153
154 const TypeItem& GetTypeItem(uint32_t idx) const {
155 CHECK_LT(idx, this->size_);
156 return this->list_[idx];
157 }
158
159 private:
160 uint32_t size_; // size of the list, in entries
161 TypeItem list_[1]; // elements of the list
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700162 DISALLOW_COPY_AND_ASSIGN(TypeList);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700163 };
164
165 class ParameterIterator { // TODO: stream
166 public:
Brian Carlstromf615a612011-07-23 12:50:34 -0700167 ParameterIterator(const DexFile& dex_file, const ProtoId& proto_id)
168 : dex_file_(dex_file), size_(0), pos_(0) {
169 type_list_ = dex_file_.GetProtoParameters(proto_id);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700170 if (type_list_ != NULL) {
171 size_ = type_list_->Size();
172 }
173 }
174 bool HasNext() const { return pos_ != size_; }
175 void Next() { ++pos_; }
176 const char* GetDescriptor() {
177 uint32_t type_idx = type_list_->GetTypeItem(pos_).type_idx_;
Brian Carlstromf615a612011-07-23 12:50:34 -0700178 return dex_file_.dexStringByTypeIdx(type_idx);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700179 }
180 private:
Brian Carlstromf615a612011-07-23 12:50:34 -0700181 const DexFile& dex_file_;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700182 const TypeList* type_list_;
183 uint32_t size_;
184 uint32_t pos_;
185 DISALLOW_IMPLICIT_CONSTRUCTORS(ParameterIterator);
186 };
187
188 ParameterIterator* GetParameterIterator(const ProtoId& proto_id) const {
189 return new ParameterIterator(*this, proto_id);
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700190 }
191
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700192 const char* GetReturnTypeDescriptor(const ProtoId& proto_id) const {
193 return dexStringByTypeIdx(proto_id.return_type_idx_);
Carl Shapiro1fb86202011-06-27 17:43:13 -0700194 }
195
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700196 // Raw code_item.
197 struct CodeItem {
198 uint16_t registers_size_;
199 uint16_t ins_size_;
200 uint16_t outs_size_;
201 uint16_t tries_size_;
202 uint32_t debug_info_off_; // file offset to debug info stream
203 uint32_t insns_size_; // size of the insns array, in 2 byte code units
204 uint16_t insns_[1];
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700205 private:
206 DISALLOW_COPY_AND_ASSIGN(CodeItem);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700207 };
208
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700209 struct CatchHandlerItem {
210 uint32_t type_idx_; // type index of the caught exception type
211 uint32_t address_; // handler address
212 };
213
Carl Shapiro2eaa9682011-08-04 19:26:11 -0700214 // Raw try_item.
215 struct TryItem {
216 uint32_t start_addr_;
217 uint16_t insn_count_;
218 uint16_t handler_off_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700219 private:
220 DISALLOW_COPY_AND_ASSIGN(TryItem);
Carl Shapiro2eaa9682011-08-04 19:26:11 -0700221 };
222
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700223 class CatchHandlerIterator {
224 public:
225 CatchHandlerIterator() {
226 remaining_count_ = -1;
227 catch_all_ = false;
228 }
229
230 CatchHandlerIterator(const byte* handler_data) {
231 current_data_ = handler_data;
jeffhaoba5ebb92011-08-25 17:24:37 -0700232 remaining_count_ = DecodeSignedLeb128(&current_data_);
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700233
234 // If remaining_count_ is non-positive, then it is the negative of
235 // the number of catch types, and the catches are followed by a
236 // catch-all handler.
237 if (remaining_count_ <= 0) {
238 catch_all_ = true;
239 remaining_count_ = -remaining_count_;
240 } else {
241 catch_all_ = false;
242 }
243 Next();
244 }
245
Shih-wei Liaofe909f22011-08-12 19:20:26 -0700246 const CatchHandlerItem& Get() const {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700247 return handler_;
248 }
249
jeffhaoba5ebb92011-08-25 17:24:37 -0700250 const byte* GetData() const {
251 return current_data_;
252 }
253
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700254 void Next() {
255 if (remaining_count_ > 0) {
256 handler_.type_idx_ = DecodeUnsignedLeb128(&current_data_);
257 handler_.address_ = DecodeUnsignedLeb128(&current_data_);
258 remaining_count_--;
259 return;
260 }
261
262 if (catch_all_) {
263 handler_.type_idx_ = kDexNoIndex;
264 handler_.address_ = DecodeUnsignedLeb128(&current_data_);
265 catch_all_ = false;
266 return;
267 }
268
269 // no more handler
270 remaining_count_ = -1;
271 }
272
Shih-wei Liao1a18c8c2011-08-14 17:47:36 -0700273 bool HasNext() const {
Shih-wei Liao4e5c0b92011-08-11 22:50:08 -0700274 return remaining_count_ == -1 && catch_all_ == false;
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700275 }
276
277 private:
278 CatchHandlerItem handler_;
279 const byte *current_data_; // the current handlder in dex file.
280 int32_t remaining_count_; // number of handler not read.
281 bool catch_all_; // is there a handler that will catch all exceptions in case
282 // that all typed handler does not match.
283 };
284
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700285 // Partially decoded form of class_data_item.
286 struct ClassDataHeader {
287 uint32_t static_fields_size_; // the number of static fields
288 uint32_t instance_fields_size_; // the number of instance fields
289 uint32_t direct_methods_size_; // the number of direct methods
290 uint32_t virtual_methods_size_; // the number of virtual methods
291 };
292
293 // Decoded form of encoded_field.
294 struct Field {
295 uint32_t field_idx_; // index into the field_ids list for the identity of this field
296 uint32_t access_flags_; // access flags for the field
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700297 Field() {};
298 private:
299 DISALLOW_COPY_AND_ASSIGN(Field);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700300 };
301
302 // Decoded form of encoded_method.
303 struct Method {
304 uint32_t method_idx_;
305 uint32_t access_flags_;
306 uint32_t code_off_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700307 Method() {};
308 private:
309 DISALLOW_COPY_AND_ASSIGN(Method);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700310 };
311
Brian Carlstrom74eb46a2011-08-02 20:10:14 -0700312 typedef std::pair<const DexFile*, const DexFile::ClassDef*> ClassPathEntry;
313 typedef std::vector<const DexFile*> ClassPath;
314
315 // Search a collection of DexFiles for a descriptor
316 static ClassPathEntry FindInClassPath(const StringPiece& descriptor,
Brian Carlstrom9ea1cb12011-08-24 23:18:18 -0700317 const ClassPath& class_path);
Brian Carlstrom74eb46a2011-08-02 20:10:14 -0700318
Brian Carlstrom69b15fb2011-09-03 12:25:21 -0700319 // Opens .dex file, guessing the format based on file extension
320 static const DexFile* Open(const std::string& filename);
321
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700322 // Opens a .dex file from the file system.
Brian Carlstrom9f30b382011-08-28 22:41:38 -0700323 static const DexFile* OpenFile(const std::string& filename);
Brian Carlstromb0460ea2011-07-29 10:08:05 -0700324
325 // Opens a .jar, .zip, or .apk file from the file system.
Brian Carlstrom9f30b382011-08-28 22:41:38 -0700326 static const DexFile* OpenZip(const std::string& filename);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700327
Brian Carlstroma663ea52011-08-19 23:33:41 -0700328 // Opens a .dex file from a new allocated pointer. location is used
329 // to identify the source, for example "/system/framework/core.jar"
330 // or "contrived-test-42". When initializing a ClassLinker from an
331 // image, the location is used to match DexCaches the image to their
332 // corresponding DexFiles.N
Brian Carlstrom9f30b382011-08-28 22:41:38 -0700333 static const DexFile* OpenPtr(byte* ptr, size_t length, const std::string& location);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700334
335 // Closes a .dex file.
Brian Carlstromf615a612011-07-23 12:50:34 -0700336 virtual ~DexFile();
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700337
Brian Carlstroma663ea52011-08-19 23:33:41 -0700338 const std::string& GetLocation() const {
339 return location_;
340 }
341
342 const Header& GetHeader() const {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700343 CHECK(header_ != NULL);
344 return *header_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700345 }
346
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700347 // Looks up a class definition by its class descriptor.
348 const ClassDef* FindClassDef(const StringPiece& descriptor) const;
349
350 // Returns the number of string identifiers in the .dex file.
351 size_t NumStringIds() const {
352 CHECK(header_ != NULL);
353 return header_->string_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700354 }
355
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700356 // Returns the number of type identifiers in the .dex file.
357 size_t NumTypeIds() const {
358 CHECK(header_ != NULL);
359 return header_->type_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700360 }
361
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700362 // Returns the number of prototype identifiers in the .dex file.
363 size_t NumProtoIds() const {
364 CHECK(header_ != NULL);
365 return header_->proto_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700366 }
367
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700368 // Returns the number of field identifiers in the .dex file.
369 size_t NumFieldIds() const {
370 CHECK(header_ != NULL);
371 return header_->field_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700372 }
373
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700374 // Returns the number of method identifiers in the .dex file.
375 size_t NumMethodIds() const {
376 CHECK(header_ != NULL);
377 return header_->method_ids_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700378 }
379
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700380 // Returns the number of class definitions in the .dex file.
381 size_t NumClassDefs() const {
382 CHECK(header_ != NULL);
383 return header_->class_defs_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700384 }
385
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700386 // Returns a pointer to the memory mapped class data.
387 // TODO: return a stream
388 const byte* GetClassData(const ClassDef& class_def) const {
389 if (class_def.class_data_off_ == 0) {
390 return NULL;
391 } else {
392 return base_ + class_def.class_data_off_;
393 }
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700394 }
395
Brian Carlstromf615a612011-07-23 12:50:34 -0700396 // Decodes the header section from the class data bytes.
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700397 ClassDataHeader ReadClassDataHeader(const byte** class_data) const {
398 CHECK(class_data != NULL);
399 ClassDataHeader header;
400 memset(&header, 0, sizeof(ClassDataHeader));
401 if (*class_data != NULL) {
402 header.static_fields_size_ = DecodeUnsignedLeb128(class_data);
403 header.instance_fields_size_ = DecodeUnsignedLeb128(class_data);
404 header.direct_methods_size_ = DecodeUnsignedLeb128(class_data);
405 header.virtual_methods_size_ = DecodeUnsignedLeb128(class_data);
406 }
407 return header;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700408 }
409
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700410 // Returns the class descriptor string of a class definition.
411 const char* GetClassDescriptor(const ClassDef& class_def) const {
412 return dexStringByTypeIdx(class_def.class_idx_);
413 }
414
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700415 // Returns the type descriptor string of a type id.
416 const char* GetTypeDescriptor(const TypeId& type_id) const {
417 return dexStringById(type_id.descriptor_idx_);
418 }
419
Brian Carlstromb9edb842011-08-28 16:31:06 -0700420 // Returns the class descriptor string of a field id.
421 const char* GetFieldClassDescriptor(const FieldId& field_id) const {
422 const DexFile::TypeId& type_id = GetTypeId(field_id.class_idx_);
423 return GetTypeDescriptor(type_id);
424 }
425
426 // Returns the name of a field id.
427 const char* GetFieldName(const FieldId& field_id) const {
428 return dexStringById(field_id.name_idx_);
429 }
430
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700431 // Returns the StringId at the specified index.
432 const StringId& GetStringId(uint32_t idx) const {
433 CHECK_LT(idx, NumStringIds());
434 return string_ids_[idx];
435 }
436
437 // Returns the TypeId at the specified index.
438 const TypeId& GetTypeId(uint32_t idx) const {
439 CHECK_LT(idx, NumTypeIds());
440 return type_ids_[idx];
441 }
442
443 // Returns the FieldId at the specified index.
444 const FieldId& GetFieldId(uint32_t idx) const {
445 CHECK_LT(idx, NumFieldIds());
446 return field_ids_[idx];
447 }
448
449 // Returns the MethodId at the specified index.
450 const MethodId& GetMethodId(uint32_t idx) const {
451 CHECK_LT(idx, NumMethodIds());
452 return method_ids_[idx];
453 }
454
455 // Returns the ProtoId at the specified index.
456 const ProtoId& GetProtoId(uint32_t idx) const {
457 CHECK_LT(idx, NumProtoIds());
458 return proto_ids_[idx];
459 }
460
461 // Returns the ClassDef at the specified index.
462 const ClassDef& GetClassDef(uint32_t idx) const {
463 CHECK_LT(idx, NumClassDefs());
464 return class_defs_[idx];
465 }
466
467 const TypeList* GetInterfacesList(const ClassDef& class_def) const {
468 if (class_def.interfaces_off_ == 0) {
469 return NULL;
470 } else {
471 const byte* addr = base_ + class_def.interfaces_off_;
472 return reinterpret_cast<const TypeList*>(addr);
473 }
474 }
475
476 const CodeItem* GetCodeItem(const Method& method) const {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700477 return GetCodeItem(method.code_off_);
478 }
479
480 const CodeItem* GetCodeItem(const uint32_t code_off_) const {
481 if (code_off_ == 0) {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700482 return NULL; // native or abstract method
483 } else {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700484 const byte* addr = base_ + code_off_;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700485 return reinterpret_cast<const CodeItem*>(addr);
486 }
487 }
488
489 // Returns the short form method descriptor for the given prototype.
490 const char* GetShorty(uint32_t proto_idx) const {
491 const ProtoId& proto_id = GetProtoId(proto_idx);
492 return dexStringById(proto_id.shorty_idx_);
493 }
494
495 const TypeList* GetProtoParameters(const ProtoId& proto_id) const {
496 if (proto_id.parameters_off_ == 0) {
497 return NULL;
498 } else {
499 const byte* addr = base_ + proto_id.parameters_off_;
500 return reinterpret_cast<const TypeList*>(addr);
501 }
502 }
503
Elliott Hughes0c424cb2011-08-26 10:16:25 -0700504 std::string CreateMethodDescriptor(uint32_t proto_idx, int32_t* unicode_length) const;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700505
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700506 const byte* GetEncodedArray(const ClassDef& class_def) const {
507 if (class_def.static_values_off_ == 0) {
508 return 0;
509 } else {
510 return base_ + class_def.static_values_off_;
511 }
512 }
513
514 int32_t GetStringLength(const StringId& string_id) const {
515 const byte* ptr = base_ + string_id.string_data_off_;
516 return DecodeUnsignedLeb128(&ptr);
517 }
518
519 ValueType ReadEncodedValue(const byte** encoded_value, JValue* value) const;
520
521 // From libdex...
522
523 // Returns a pointer to the UTF-8 string data referred to by the
524 // given string_id.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700525 const char* GetStringData(const StringId& string_id, int32_t* length) const {
526 CHECK(length != NULL);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700527 const byte* ptr = base_ + string_id.string_data_off_;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700528 *length = DecodeUnsignedLeb128(&ptr);
Brian Carlstrom0b138b22011-07-27 15:19:17 -0700529 return reinterpret_cast<const char*>(ptr);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700530 }
531
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700532 const char* GetStringData(const StringId& string_id) const {
533 int32_t length;
534 return GetStringData(string_id, &length);
535 }
536
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700537 // return the UTF-8 encoded string with the specified string_id index
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700538 const char* dexStringById(uint32_t idx, int32_t* unicode_length) const {
Shih-wei Liao195487c2011-08-20 13:29:04 -0700539 if (idx == kDexNoIndex) {
540 *unicode_length = 0;
541 return NULL;
542 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700543 const StringId& string_id = GetStringId(idx);
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700544 return GetStringData(string_id, unicode_length);
545 }
546
547 const char* dexStringById(uint32_t idx) const {
548 int32_t unicode_length;
549 return dexStringById(idx, &unicode_length);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700550 }
551
Shih-wei Liao55df06b2011-08-26 14:39:27 -0700552 String* dexArtStringById(int32_t idx) const;
Shih-wei Liao195487c2011-08-20 13:29:04 -0700553
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700554 // Get the descriptor string associated with a given type index.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700555 const char* dexStringByTypeIdx(uint32_t idx, int32_t* unicode_length) const {
556 const TypeId& type_id = GetTypeId(idx);
557 return dexStringById(type_id.descriptor_idx_, unicode_length);
558 }
559
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700560 const char* dexStringByTypeIdx(uint32_t idx) const {
561 const TypeId& type_id = GetTypeId(idx);
562 return dexStringById(type_id.descriptor_idx_);
563 }
564
Shih-wei Liao55df06b2011-08-26 14:39:27 -0700565 String* dexArtStringByTypeIdx(int32_t idx) const {
Shih-wei Liao195487c2011-08-20 13:29:04 -0700566 const TypeId& type_id = GetTypeId(idx);
567 return dexArtStringById(type_id.descriptor_idx_);
568 }
569
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700570 // TODO: encoded_field is actually a stream of bytes
571 void dexReadClassDataField(const byte** encoded_field,
Brian Carlstromf615a612011-07-23 12:50:34 -0700572 DexFile::Field* field,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700573 uint32_t* last_idx) const {
574 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_field);
575 field->access_flags_ = DecodeUnsignedLeb128(encoded_field);
576 field->field_idx_ = idx;
577 *last_idx = idx;
578 }
579
580 // TODO: encoded_method is actually a stream of bytes
581 void dexReadClassDataMethod(const byte** encoded_method,
Brian Carlstromf615a612011-07-23 12:50:34 -0700582 DexFile::Method* method,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700583 uint32_t* last_idx) const {
584 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_method);
585 method->access_flags_ = DecodeUnsignedLeb128(encoded_method);
586 method->code_off_ = DecodeUnsignedLeb128(encoded_method);
587 method->method_idx_ = idx;
588 *last_idx = idx;
589 }
590
jeffhaoba5ebb92011-08-25 17:24:37 -0700591 static const TryItem* dexGetTryItems(const CodeItem& code_item, uint32_t offset) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700592 const uint16_t* insns_end_ = &code_item.insns_[code_item.insns_size_];
593 return reinterpret_cast<const TryItem*>
594 (RoundUp(reinterpret_cast<uint32_t>(insns_end_), 4)) + offset;
595 }
596
597 // Get the base of the encoded data for the given DexCode.
jeffhaoba5ebb92011-08-25 17:24:37 -0700598 static const byte* dexGetCatchHandlerData(const CodeItem& code_item, uint32_t offset) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700599 const byte* handler_data = reinterpret_cast<const byte*>
600 (dexGetTryItems(code_item, code_item.tries_size_));
601 return handler_data + offset;
602 }
603
604 // Find the handler associated with a given address, if any.
605 // Initializes the given iterator and returns true if a match is
606 // found. Returns end if there is no applicable handler.
jeffhaoba5ebb92011-08-25 17:24:37 -0700607 static CatchHandlerIterator dexFindCatchHandler(const CodeItem& code_item, uint32_t address) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700608 CatchHandlerItem handler;
609 handler.address_ = -1;
610 int32_t offset = -1;
611
612 // Short-circuit the overwhelmingly common cases.
613 switch (code_item.tries_size_) {
614 case 0:
615 break;
616 case 1: {
617 const TryItem* tries = dexGetTryItems(code_item, 0);
618 uint32_t start = tries->start_addr_;
619 if (address < start)
620 break;
621
622 uint32_t end = start + tries->insn_count_;
623 if (address >= end)
624 break;
625
626 offset = tries->handler_off_;
627 break;
628 }
629 default:
630 offset = dexFindCatchHandlerOffset0(code_item, code_item.tries_size_, address);
631 }
632
633 if (offset >= 0) {
634 const byte* handler_data = dexGetCatchHandlerData(code_item, offset);
635 return CatchHandlerIterator(handler_data);
636 }
637 return CatchHandlerIterator();
638 }
639
jeffhaoba5ebb92011-08-25 17:24:37 -0700640 static int32_t dexFindCatchHandlerOffset0(const CodeItem &code_item,
641 int32_t tries_size,
642 uint32_t address) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700643 // Note: Signed type is important for max and min.
644 int32_t min = 0;
645 int32_t max = tries_size - 1;
646
647 while (max >= min) {
648 int32_t guess = (min + max) >> 1;
649 const TryItem* pTry = dexGetTryItems(code_item, guess);
650 uint32_t start = pTry->start_addr_;
651
652 if (address < start) {
653 max = guess - 1;
654 continue;
655 }
656
657 uint32_t end = start + pTry->insn_count_;
658 if (address >= end) {
659 min = guess + 1;
660 continue;
661 }
662
663 // We have a winner!
664 return (int32_t) pTry->handler_off_;
665 }
666
667 // No match.
668 return -1;
669 }
670
Shih-wei Liao195487c2011-08-20 13:29:04 -0700671 // Get the pointer to the start of the debugging data
672 const byte* dexGetDebugInfoStream(const CodeItem* code_item) const {
673 if (code_item->debug_info_off_ == 0) {
674 return NULL;
675 } else {
676 return base_ + code_item->debug_info_off_;
677 }
678 }
679
680 // Callback for "new position table entry".
681 // Returning true causes the decoder to stop early.
682 typedef bool (*DexDebugNewPositionCb)(void *cnxt, uint32_t address, uint32_t line_num);
683
684 // Callback for "new locals table entry". "signature" is an empty string
685 // if no signature is available for an entry.
686 typedef void (*DexDebugNewLocalCb)(void *cnxt, uint16_t reg,
687 uint32_t startAddress,
688 uint32_t endAddress,
689 const String* name,
690 const String* descriptor,
691 const String* signature);
692
693 static bool LineNumForPcCb(void *cnxt, uint32_t address, uint32_t line_num) {
694 LineNumFromPcContext *context = (LineNumFromPcContext *)cnxt;
695
696 // We know that this callback will be called in
697 // ascending address order, so keep going until we find
698 // a match or we've just gone past it.
699 if (address > context->address_) {
700 // The line number from the previous positions callback
701 // wil be the final result.
702 return true;
703 } else {
704 context->line_num_ = line_num;
705 return address == context->address_;
706 }
707 }
708
709
710 // Debug info opcodes and constants
711 enum {
712 DBG_END_SEQUENCE = 0x00,
713 DBG_ADVANCE_PC = 0x01,
714 DBG_ADVANCE_LINE = 0x02,
715 DBG_START_LOCAL = 0x03,
716 DBG_START_LOCAL_EXTENDED = 0x04,
717 DBG_END_LOCAL = 0x05,
718 DBG_RESTART_LOCAL = 0x06,
719 DBG_SET_PROLOGUE_END = 0x07,
720 DBG_SET_EPILOGUE_BEGIN = 0x08,
721 DBG_SET_FILE = 0x09,
722 DBG_FIRST_SPECIAL = 0x0a,
723 DBG_LINE_BASE = -4,
724 DBG_LINE_RANGE = 15,
725 };
726
727 struct LocalInfo {
728 LocalInfo() : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {}
729
730 // E.g., list
731 const String* name_;
732
733 // E.g., Ljava/util/LinkedList;
734 const String* descriptor_;
735
736 // E.g., java.util.LinkedList<java.lang.Integer>
737 const String* signature_;
738
739 // PC location where the local is first defined.
740 uint16_t start_address_;
741
742 // Is the local defined and live.
743 bool is_live_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700744
745 private:
746 DISALLOW_COPY_AND_ASSIGN(LocalInfo);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700747 };
748
749 struct LineNumFromPcContext {
750 LineNumFromPcContext(uint32_t address, uint32_t line_num) :
751 address_(address), line_num_(line_num) {}
752 uint32_t address_;
753 uint32_t line_num_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700754 private:
755 DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700756 };
757
758 void InvokeLocalCbIfLive(void *cnxt, int reg, uint32_t end_address,
759 LocalInfo *local_in_reg, DexDebugNewLocalCb local_cb) const {
760 if (local_cb != NULL && local_in_reg[reg].is_live_) {
761 local_cb(cnxt, reg, local_in_reg[reg].start_address_, end_address,
762 local_in_reg[reg].name_, local_in_reg[reg].descriptor_,
763 local_in_reg[reg].signature_);
764 }
765 }
766
767 // Determine the source file line number based on the program counter.
768 // "pc" is an offset, in 16-bit units, from the start of the method's code.
769 //
770 // Returns -1 if no match was found (possibly because the source files were
771 // compiled without "-g", so no line number information is present).
772 // Returns -2 for native methods (as expected in exception traces).
773 //
774 // This is used by runtime; therefore use art::Method not art::DexFile::Method.
775 int32_t GetLineNumFromPC(const art::Method* method, uint32_t rel_pc) const;
776
777 void dexDecodeDebugInfo0(const CodeItem* code_item, const art::Method* method,
778 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
779 void* cnxt, const byte* stream, LocalInfo* local_in_reg) const;
780
781 void dexDecodeDebugInfo(const CodeItem* code_item, const art::Method *method,
782 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
783 void* cnxt) const {
784 const byte* stream = dexGetDebugInfoStream(code_item);
785 LocalInfo local_in_reg[code_item->registers_size_];
786
787 if (stream != NULL) {
788 dexDecodeDebugInfo0(code_item, method, posCb, local_cb, cnxt, stream, local_in_reg);
789 }
790 for (int reg = 0; reg < code_item->registers_size_; reg++) {
791 InvokeLocalCbIfLive(cnxt, reg, code_item->insns_size_, local_in_reg, local_cb);
792 }
793 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700794
795 // TODO: const reference
796 uint32_t dexGetIndexForClassDef(const ClassDef* class_def) const {
797 CHECK_GE(class_def, class_defs_);
798 CHECK_LT(class_def, class_defs_ + header_->class_defs_size_);
799 return class_def - class_defs_;
800 }
801
802 const char* dexGetSourceFile(const ClassDef& class_def) const {
803 if (class_def.source_file_idx_ == 0xffffffff) {
804 return NULL;
805 } else {
806 return dexStringById(class_def.source_file_idx_);
807 }
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700808 }
809
Carl Shapiro1fb86202011-06-27 17:43:13 -0700810 private:
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700811 // Helper class to deallocate underlying storage.
812 class Closer {
813 public:
814 virtual ~Closer();
815 };
816
817 // Helper class to deallocate mmap-backed .dex files.
818 class MmapCloser : public Closer {
819 public:
820 MmapCloser(void* addr, size_t length);
821 virtual ~MmapCloser();
822 private:
823 void* addr_;
824 size_t length_;
825 };
826
827 // Helper class for deallocating new/delete-backed .dex files.
828 class PtrCloser : public Closer {
829 public:
830 PtrCloser(byte* addr);
831 virtual ~PtrCloser();
832 private:
833 byte* addr_;
834 };
835
Brian Carlstrom9f30b382011-08-28 22:41:38 -0700836 // Opens a .dex file at the given address.
837 static const DexFile* Open(const byte* dex_file,
838 size_t length,
839 const std::string& location,
840 Closer* closer);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700841
Brian Carlstroma663ea52011-08-19 23:33:41 -0700842 DexFile(const byte* addr, size_t length, const std::string& location, Closer* closer)
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700843 : base_(addr),
844 length_(length),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700845 location_(location),
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700846 closer_(closer),
847 header_(0),
848 string_ids_(0),
849 type_ids_(0),
850 field_ids_(0),
851 method_ids_(0),
852 proto_ids_(0),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700853 class_defs_(0) {
854 CHECK(addr != NULL);
855 CHECK_GT(length, 0U);
856 CHECK(closer != NULL);
857 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700858
859 // Top-level initializer that calls other Init methods.
860 bool Init();
861
862 // Caches pointers into to the various file sections.
863 void InitMembers();
864
865 // Builds the index of descriptors to class definitions.
866 void InitIndex();
867
868 // Returns true if the byte string equals the magic value.
869 bool CheckMagic(const byte* magic);
870
871 // Returns true if the header magic is of the expected value.
872 bool IsMagicValid();
873
874 // The index of descriptors to class definitions.
Brian Carlstromf615a612011-07-23 12:50:34 -0700875 typedef std::map<const StringPiece, const DexFile::ClassDef*> Index;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700876 Index index_;
877
878 // The base address of the memory mapping.
879 const byte* base_;
880
881 // The size of the underlying memory allocation in bytes.
882 size_t length_;
883
Brian Carlstroma663ea52011-08-19 23:33:41 -0700884 // Typically the dex file name when availble, alternatively some identifying string.
885 //
886 // The ClassLinker will use this to match DexFiles the boot class
887 // path to DexCache::GetLocation when loading from an image.
888 const std::string location_;
889
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700890 // Helper object to free the underlying allocation.
Elliott Hughes90a33692011-08-30 13:27:07 -0700891 UniquePtr<Closer> closer_;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700892
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_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700913};
914
915} // namespace art
916
917#endif // ART_SRC_DEX_FILE_H_