blob: 76b9c2f93e212ed74da84584e704fe9da9aa6704 [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
Brian Carlstrom578bbdc2011-07-21 14:07:47 -070010#include "globals.h"
Brian Carlstrom7e49dca2011-07-22 18:07:34 -070011#include "leb128.h"
12#include "logging.h"
13#include "scoped_ptr.h"
14#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;
232 remaining_count_ = DecodeUnsignedLeb128(&current_data_);
233
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
250 void Next() {
251 if (remaining_count_ > 0) {
252 handler_.type_idx_ = DecodeUnsignedLeb128(&current_data_);
253 handler_.address_ = DecodeUnsignedLeb128(&current_data_);
254 remaining_count_--;
255 return;
256 }
257
258 if (catch_all_) {
259 handler_.type_idx_ = kDexNoIndex;
260 handler_.address_ = DecodeUnsignedLeb128(&current_data_);
261 catch_all_ = false;
262 return;
263 }
264
265 // no more handler
266 remaining_count_ = -1;
267 }
268
Shih-wei Liao1a18c8c2011-08-14 17:47:36 -0700269 bool HasNext() const {
Shih-wei Liao4e5c0b92011-08-11 22:50:08 -0700270 return remaining_count_ == -1 && catch_all_ == false;
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700271 }
272
273 private:
274 CatchHandlerItem handler_;
275 const byte *current_data_; // the current handlder in dex file.
276 int32_t remaining_count_; // number of handler not read.
277 bool catch_all_; // is there a handler that will catch all exceptions in case
278 // that all typed handler does not match.
279 };
280
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700281 // Partially decoded form of class_data_item.
282 struct ClassDataHeader {
283 uint32_t static_fields_size_; // the number of static fields
284 uint32_t instance_fields_size_; // the number of instance fields
285 uint32_t direct_methods_size_; // the number of direct methods
286 uint32_t virtual_methods_size_; // the number of virtual methods
287 };
288
289 // Decoded form of encoded_field.
290 struct Field {
291 uint32_t field_idx_; // index into the field_ids list for the identity of this field
292 uint32_t access_flags_; // access flags for the field
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700293 Field() {};
294 private:
295 DISALLOW_COPY_AND_ASSIGN(Field);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700296 };
297
298 // Decoded form of encoded_method.
299 struct Method {
300 uint32_t method_idx_;
301 uint32_t access_flags_;
302 uint32_t code_off_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700303 Method() {};
304 private:
305 DISALLOW_COPY_AND_ASSIGN(Method);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700306 };
307
Brian Carlstrom74eb46a2011-08-02 20:10:14 -0700308 typedef std::pair<const DexFile*, const DexFile::ClassDef*> ClassPathEntry;
309 typedef std::vector<const DexFile*> ClassPath;
310
311 // Search a collection of DexFiles for a descriptor
312 static ClassPathEntry FindInClassPath(const StringPiece& descriptor,
Brian Carlstrom9ea1cb12011-08-24 23:18:18 -0700313 const ClassPath& class_path);
Brian Carlstrom74eb46a2011-08-02 20:10:14 -0700314
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700315 // Opens a .dex file from the file system.
Brian Carlstromb0460ea2011-07-29 10:08:05 -0700316 static DexFile* OpenFile(const std::string& filename);
317
318 // Opens a .jar, .zip, or .apk file from the file system.
319 static DexFile* OpenZip(const std::string& filename);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700320
Brian Carlstroma663ea52011-08-19 23:33:41 -0700321 // Opens a .dex file from a new allocated pointer. location is used
322 // to identify the source, for example "/system/framework/core.jar"
323 // or "contrived-test-42". When initializing a ClassLinker from an
324 // image, the location is used to match DexCaches the image to their
325 // corresponding DexFiles.N
326 static DexFile* OpenPtr(byte* ptr, size_t length, const std::string& location);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700327
328 // Closes a .dex file.
Brian Carlstromf615a612011-07-23 12:50:34 -0700329 virtual ~DexFile();
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700330
Brian Carlstroma663ea52011-08-19 23:33:41 -0700331 const std::string& GetLocation() const {
332 return location_;
333 }
334
335 const Header& GetHeader() const {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700336 CHECK(header_ != NULL);
337 return *header_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700338 }
339
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700340 // Looks up a class definition by its class descriptor.
341 const ClassDef* FindClassDef(const StringPiece& descriptor) const;
342
343 // Returns the number of string identifiers in the .dex file.
344 size_t NumStringIds() const {
345 CHECK(header_ != NULL);
346 return header_->string_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700347 }
348
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700349 // Returns the number of type identifiers in the .dex file.
350 size_t NumTypeIds() const {
351 CHECK(header_ != NULL);
352 return header_->type_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700353 }
354
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700355 // Returns the number of prototype identifiers in the .dex file.
356 size_t NumProtoIds() const {
357 CHECK(header_ != NULL);
358 return header_->proto_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700359 }
360
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700361 // Returns the number of field identifiers in the .dex file.
362 size_t NumFieldIds() const {
363 CHECK(header_ != NULL);
364 return header_->field_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700365 }
366
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700367 // Returns the number of method identifiers in the .dex file.
368 size_t NumMethodIds() const {
369 CHECK(header_ != NULL);
370 return header_->method_ids_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700371 }
372
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700373 // Returns the number of class definitions in the .dex file.
374 size_t NumClassDefs() const {
375 CHECK(header_ != NULL);
376 return header_->class_defs_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700377 }
378
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700379 // Returns a pointer to the memory mapped class data.
380 // TODO: return a stream
381 const byte* GetClassData(const ClassDef& class_def) const {
382 if (class_def.class_data_off_ == 0) {
383 return NULL;
384 } else {
385 return base_ + class_def.class_data_off_;
386 }
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700387 }
388
Brian Carlstromf615a612011-07-23 12:50:34 -0700389 // Decodes the header section from the class data bytes.
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700390 ClassDataHeader ReadClassDataHeader(const byte** class_data) const {
391 CHECK(class_data != NULL);
392 ClassDataHeader header;
393 memset(&header, 0, sizeof(ClassDataHeader));
394 if (*class_data != NULL) {
395 header.static_fields_size_ = DecodeUnsignedLeb128(class_data);
396 header.instance_fields_size_ = DecodeUnsignedLeb128(class_data);
397 header.direct_methods_size_ = DecodeUnsignedLeb128(class_data);
398 header.virtual_methods_size_ = DecodeUnsignedLeb128(class_data);
399 }
400 return header;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700401 }
402
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700403 // Returns the class descriptor string of a class definition.
404 const char* GetClassDescriptor(const ClassDef& class_def) const {
405 return dexStringByTypeIdx(class_def.class_idx_);
406 }
407
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700408 // Returns the type descriptor string of a type id.
409 const char* GetTypeDescriptor(const TypeId& type_id) const {
410 return dexStringById(type_id.descriptor_idx_);
411 }
412
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700413 // Returns the StringId at the specified index.
414 const StringId& GetStringId(uint32_t idx) const {
415 CHECK_LT(idx, NumStringIds());
416 return string_ids_[idx];
417 }
418
419 // Returns the TypeId at the specified index.
420 const TypeId& GetTypeId(uint32_t idx) const {
421 CHECK_LT(idx, NumTypeIds());
422 return type_ids_[idx];
423 }
424
425 // Returns the FieldId at the specified index.
426 const FieldId& GetFieldId(uint32_t idx) const {
427 CHECK_LT(idx, NumFieldIds());
428 return field_ids_[idx];
429 }
430
431 // Returns the MethodId at the specified index.
432 const MethodId& GetMethodId(uint32_t idx) const {
433 CHECK_LT(idx, NumMethodIds());
434 return method_ids_[idx];
435 }
436
437 // Returns the ProtoId at the specified index.
438 const ProtoId& GetProtoId(uint32_t idx) const {
439 CHECK_LT(idx, NumProtoIds());
440 return proto_ids_[idx];
441 }
442
443 // Returns the ClassDef at the specified index.
444 const ClassDef& GetClassDef(uint32_t idx) const {
445 CHECK_LT(idx, NumClassDefs());
446 return class_defs_[idx];
447 }
448
449 const TypeList* GetInterfacesList(const ClassDef& class_def) const {
450 if (class_def.interfaces_off_ == 0) {
451 return NULL;
452 } else {
453 const byte* addr = base_ + class_def.interfaces_off_;
454 return reinterpret_cast<const TypeList*>(addr);
455 }
456 }
457
458 const CodeItem* GetCodeItem(const Method& method) const {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700459 return GetCodeItem(method.code_off_);
460 }
461
462 const CodeItem* GetCodeItem(const uint32_t code_off_) const {
463 if (code_off_ == 0) {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700464 return NULL; // native or abstract method
465 } else {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700466 const byte* addr = base_ + code_off_;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700467 return reinterpret_cast<const CodeItem*>(addr);
468 }
469 }
470
471 // Returns the short form method descriptor for the given prototype.
472 const char* GetShorty(uint32_t proto_idx) const {
473 const ProtoId& proto_id = GetProtoId(proto_idx);
474 return dexStringById(proto_id.shorty_idx_);
475 }
476
477 const TypeList* GetProtoParameters(const ProtoId& proto_id) const {
478 if (proto_id.parameters_off_ == 0) {
479 return NULL;
480 } else {
481 const byte* addr = base_ + proto_id.parameters_off_;
482 return reinterpret_cast<const TypeList*>(addr);
483 }
484 }
485
Elliott Hughes0c424cb2011-08-26 10:16:25 -0700486 std::string CreateMethodDescriptor(uint32_t proto_idx, int32_t* unicode_length) const;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700487
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700488 const byte* GetEncodedArray(const ClassDef& class_def) const {
489 if (class_def.static_values_off_ == 0) {
490 return 0;
491 } else {
492 return base_ + class_def.static_values_off_;
493 }
494 }
495
496 int32_t GetStringLength(const StringId& string_id) const {
497 const byte* ptr = base_ + string_id.string_data_off_;
498 return DecodeUnsignedLeb128(&ptr);
499 }
500
501 ValueType ReadEncodedValue(const byte** encoded_value, JValue* value) const;
502
503 // From libdex...
504
505 // Returns a pointer to the UTF-8 string data referred to by the
506 // given string_id.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700507 const char* GetStringData(const StringId& string_id, int32_t* length) const {
508 CHECK(length != NULL);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700509 const byte* ptr = base_ + string_id.string_data_off_;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700510 *length = DecodeUnsignedLeb128(&ptr);
Brian Carlstrom0b138b22011-07-27 15:19:17 -0700511 return reinterpret_cast<const char*>(ptr);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700512 }
513
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700514 const char* GetStringData(const StringId& string_id) const {
515 int32_t length;
516 return GetStringData(string_id, &length);
517 }
518
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700519 // return the UTF-8 encoded string with the specified string_id index
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700520 const char* dexStringById(uint32_t idx, int32_t* unicode_length) const {
Shih-wei Liao195487c2011-08-20 13:29:04 -0700521 if (idx == kDexNoIndex) {
522 *unicode_length = 0;
523 return NULL;
524 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700525 const StringId& string_id = GetStringId(idx);
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700526 return GetStringData(string_id, unicode_length);
527 }
528
529 const char* dexStringById(uint32_t idx) const {
530 int32_t unicode_length;
531 return dexStringById(idx, &unicode_length);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700532 }
533
Shih-wei Liao195487c2011-08-20 13:29:04 -0700534 String* dexArtStringById(uint32_t idx) const;
535
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700536 // Get the descriptor string associated with a given type index.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700537 const char* dexStringByTypeIdx(uint32_t idx, int32_t* unicode_length) const {
538 const TypeId& type_id = GetTypeId(idx);
539 return dexStringById(type_id.descriptor_idx_, unicode_length);
540 }
541
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700542 const char* dexStringByTypeIdx(uint32_t idx) const {
543 const TypeId& type_id = GetTypeId(idx);
544 return dexStringById(type_id.descriptor_idx_);
545 }
546
Shih-wei Liao195487c2011-08-20 13:29:04 -0700547 String* dexArtStringByTypeIdx(uint32_t idx) const {
548 const TypeId& type_id = GetTypeId(idx);
549 return dexArtStringById(type_id.descriptor_idx_);
550 }
551
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700552 // TODO: encoded_field is actually a stream of bytes
553 void dexReadClassDataField(const byte** encoded_field,
Brian Carlstromf615a612011-07-23 12:50:34 -0700554 DexFile::Field* field,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700555 uint32_t* last_idx) const {
556 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_field);
557 field->access_flags_ = DecodeUnsignedLeb128(encoded_field);
558 field->field_idx_ = idx;
559 *last_idx = idx;
560 }
561
562 // TODO: encoded_method is actually a stream of bytes
563 void dexReadClassDataMethod(const byte** encoded_method,
Brian Carlstromf615a612011-07-23 12:50:34 -0700564 DexFile::Method* method,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700565 uint32_t* last_idx) const {
566 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_method);
567 method->access_flags_ = DecodeUnsignedLeb128(encoded_method);
568 method->code_off_ = DecodeUnsignedLeb128(encoded_method);
569 method->method_idx_ = idx;
570 *last_idx = idx;
571 }
572
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700573 const TryItem* dexGetTryItems(const CodeItem& code_item, uint32_t offset) const {
574 const uint16_t* insns_end_ = &code_item.insns_[code_item.insns_size_];
575 return reinterpret_cast<const TryItem*>
576 (RoundUp(reinterpret_cast<uint32_t>(insns_end_), 4)) + offset;
577 }
578
579 // Get the base of the encoded data for the given DexCode.
580 const byte* dexGetCatchHandlerData(const CodeItem& code_item, uint32_t offset) const {
581 const byte* handler_data = reinterpret_cast<const byte*>
582 (dexGetTryItems(code_item, code_item.tries_size_));
583 return handler_data + offset;
584 }
585
586 // Find the handler associated with a given address, if any.
587 // Initializes the given iterator and returns true if a match is
588 // found. Returns end if there is no applicable handler.
589 CatchHandlerIterator dexFindCatchHandler(const CodeItem& code_item, uint32_t address) const {
590 CatchHandlerItem handler;
591 handler.address_ = -1;
592 int32_t offset = -1;
593
594 // Short-circuit the overwhelmingly common cases.
595 switch (code_item.tries_size_) {
596 case 0:
597 break;
598 case 1: {
599 const TryItem* tries = dexGetTryItems(code_item, 0);
600 uint32_t start = tries->start_addr_;
601 if (address < start)
602 break;
603
604 uint32_t end = start + tries->insn_count_;
605 if (address >= end)
606 break;
607
608 offset = tries->handler_off_;
609 break;
610 }
611 default:
612 offset = dexFindCatchHandlerOffset0(code_item, code_item.tries_size_, address);
613 }
614
615 if (offset >= 0) {
616 const byte* handler_data = dexGetCatchHandlerData(code_item, offset);
617 return CatchHandlerIterator(handler_data);
618 }
619 return CatchHandlerIterator();
620 }
621
622 int32_t dexFindCatchHandlerOffset0(const CodeItem &code_item,
623 int32_t tries_size,
624 uint32_t address) const {
625 // Note: Signed type is important for max and min.
626 int32_t min = 0;
627 int32_t max = tries_size - 1;
628
629 while (max >= min) {
630 int32_t guess = (min + max) >> 1;
631 const TryItem* pTry = dexGetTryItems(code_item, guess);
632 uint32_t start = pTry->start_addr_;
633
634 if (address < start) {
635 max = guess - 1;
636 continue;
637 }
638
639 uint32_t end = start + pTry->insn_count_;
640 if (address >= end) {
641 min = guess + 1;
642 continue;
643 }
644
645 // We have a winner!
646 return (int32_t) pTry->handler_off_;
647 }
648
649 // No match.
650 return -1;
651 }
652
Shih-wei Liao195487c2011-08-20 13:29:04 -0700653 // Get the pointer to the start of the debugging data
654 const byte* dexGetDebugInfoStream(const CodeItem* code_item) const {
655 if (code_item->debug_info_off_ == 0) {
656 return NULL;
657 } else {
658 return base_ + code_item->debug_info_off_;
659 }
660 }
661
662 // Callback for "new position table entry".
663 // Returning true causes the decoder to stop early.
664 typedef bool (*DexDebugNewPositionCb)(void *cnxt, uint32_t address, uint32_t line_num);
665
666 // Callback for "new locals table entry". "signature" is an empty string
667 // if no signature is available for an entry.
668 typedef void (*DexDebugNewLocalCb)(void *cnxt, uint16_t reg,
669 uint32_t startAddress,
670 uint32_t endAddress,
671 const String* name,
672 const String* descriptor,
673 const String* signature);
674
675 static bool LineNumForPcCb(void *cnxt, uint32_t address, uint32_t line_num) {
676 LineNumFromPcContext *context = (LineNumFromPcContext *)cnxt;
677
678 // We know that this callback will be called in
679 // ascending address order, so keep going until we find
680 // a match or we've just gone past it.
681 if (address > context->address_) {
682 // The line number from the previous positions callback
683 // wil be the final result.
684 return true;
685 } else {
686 context->line_num_ = line_num;
687 return address == context->address_;
688 }
689 }
690
691
692 // Debug info opcodes and constants
693 enum {
694 DBG_END_SEQUENCE = 0x00,
695 DBG_ADVANCE_PC = 0x01,
696 DBG_ADVANCE_LINE = 0x02,
697 DBG_START_LOCAL = 0x03,
698 DBG_START_LOCAL_EXTENDED = 0x04,
699 DBG_END_LOCAL = 0x05,
700 DBG_RESTART_LOCAL = 0x06,
701 DBG_SET_PROLOGUE_END = 0x07,
702 DBG_SET_EPILOGUE_BEGIN = 0x08,
703 DBG_SET_FILE = 0x09,
704 DBG_FIRST_SPECIAL = 0x0a,
705 DBG_LINE_BASE = -4,
706 DBG_LINE_RANGE = 15,
707 };
708
709 struct LocalInfo {
710 LocalInfo() : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {}
711
712 // E.g., list
713 const String* name_;
714
715 // E.g., Ljava/util/LinkedList;
716 const String* descriptor_;
717
718 // E.g., java.util.LinkedList<java.lang.Integer>
719 const String* signature_;
720
721 // PC location where the local is first defined.
722 uint16_t start_address_;
723
724 // Is the local defined and live.
725 bool is_live_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700726
727 private:
728 DISALLOW_COPY_AND_ASSIGN(LocalInfo);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700729 };
730
731 struct LineNumFromPcContext {
732 LineNumFromPcContext(uint32_t address, uint32_t line_num) :
733 address_(address), line_num_(line_num) {}
734 uint32_t address_;
735 uint32_t line_num_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700736 private:
737 DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700738 };
739
740 void InvokeLocalCbIfLive(void *cnxt, int reg, uint32_t end_address,
741 LocalInfo *local_in_reg, DexDebugNewLocalCb local_cb) const {
742 if (local_cb != NULL && local_in_reg[reg].is_live_) {
743 local_cb(cnxt, reg, local_in_reg[reg].start_address_, end_address,
744 local_in_reg[reg].name_, local_in_reg[reg].descriptor_,
745 local_in_reg[reg].signature_);
746 }
747 }
748
749 // Determine the source file line number based on the program counter.
750 // "pc" is an offset, in 16-bit units, from the start of the method's code.
751 //
752 // Returns -1 if no match was found (possibly because the source files were
753 // compiled without "-g", so no line number information is present).
754 // Returns -2 for native methods (as expected in exception traces).
755 //
756 // This is used by runtime; therefore use art::Method not art::DexFile::Method.
757 int32_t GetLineNumFromPC(const art::Method* method, uint32_t rel_pc) const;
758
759 void dexDecodeDebugInfo0(const CodeItem* code_item, const art::Method* method,
760 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
761 void* cnxt, const byte* stream, LocalInfo* local_in_reg) const;
762
763 void dexDecodeDebugInfo(const CodeItem* code_item, const art::Method *method,
764 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
765 void* cnxt) const {
766 const byte* stream = dexGetDebugInfoStream(code_item);
767 LocalInfo local_in_reg[code_item->registers_size_];
768
769 if (stream != NULL) {
770 dexDecodeDebugInfo0(code_item, method, posCb, local_cb, cnxt, stream, local_in_reg);
771 }
772 for (int reg = 0; reg < code_item->registers_size_; reg++) {
773 InvokeLocalCbIfLive(cnxt, reg, code_item->insns_size_, local_in_reg, local_cb);
774 }
775 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700776
777 // TODO: const reference
778 uint32_t dexGetIndexForClassDef(const ClassDef* class_def) const {
779 CHECK_GE(class_def, class_defs_);
780 CHECK_LT(class_def, class_defs_ + header_->class_defs_size_);
781 return class_def - class_defs_;
782 }
783
784 const char* dexGetSourceFile(const ClassDef& class_def) const {
785 if (class_def.source_file_idx_ == 0xffffffff) {
786 return NULL;
787 } else {
788 return dexStringById(class_def.source_file_idx_);
789 }
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700790 }
791
Carl Shapiro1fb86202011-06-27 17:43:13 -0700792 private:
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700793 // Helper class to deallocate underlying storage.
794 class Closer {
795 public:
796 virtual ~Closer();
797 };
798
799 // Helper class to deallocate mmap-backed .dex files.
800 class MmapCloser : public Closer {
801 public:
802 MmapCloser(void* addr, size_t length);
803 virtual ~MmapCloser();
804 private:
805 void* addr_;
806 size_t length_;
807 };
808
809 // Helper class for deallocating new/delete-backed .dex files.
810 class PtrCloser : public Closer {
811 public:
812 PtrCloser(byte* addr);
813 virtual ~PtrCloser();
814 private:
815 byte* addr_;
816 };
817
818 // Opens a .dex file at a the given address.
Brian Carlstroma663ea52011-08-19 23:33:41 -0700819 static DexFile* Open(const byte* dex_file, size_t length, const std::string& location, Closer* closer);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700820
Brian Carlstroma663ea52011-08-19 23:33:41 -0700821 DexFile(const byte* addr, size_t length, const std::string& location, Closer* closer)
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700822 : base_(addr),
823 length_(length),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700824 location_(location),
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700825 closer_(closer),
826 header_(0),
827 string_ids_(0),
828 type_ids_(0),
829 field_ids_(0),
830 method_ids_(0),
831 proto_ids_(0),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700832 class_defs_(0) {
833 CHECK(addr != NULL);
834 CHECK_GT(length, 0U);
835 CHECK(closer != NULL);
836 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700837
838 // Top-level initializer that calls other Init methods.
839 bool Init();
840
841 // Caches pointers into to the various file sections.
842 void InitMembers();
843
844 // Builds the index of descriptors to class definitions.
845 void InitIndex();
846
847 // Returns true if the byte string equals the magic value.
848 bool CheckMagic(const byte* magic);
849
850 // Returns true if the header magic is of the expected value.
851 bool IsMagicValid();
852
853 // The index of descriptors to class definitions.
Brian Carlstromf615a612011-07-23 12:50:34 -0700854 typedef std::map<const StringPiece, const DexFile::ClassDef*> Index;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700855 Index index_;
856
857 // The base address of the memory mapping.
858 const byte* base_;
859
860 // The size of the underlying memory allocation in bytes.
861 size_t length_;
862
Brian Carlstroma663ea52011-08-19 23:33:41 -0700863 // Typically the dex file name when availble, alternatively some identifying string.
864 //
865 // The ClassLinker will use this to match DexFiles the boot class
866 // path to DexCache::GetLocation when loading from an image.
867 const std::string location_;
868
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700869 // Helper object to free the underlying allocation.
870 scoped_ptr<Closer> closer_;
871
872 // Points to the header section.
873 const Header* header_;
874
875 // Points to the base of the string identifier list.
876 const StringId* string_ids_;
877
878 // Points to the base of the type identifier list.
879 const TypeId* type_ids_;
880
881 // Points to the base of the field identifier list.
882 const FieldId* field_ids_;
883
884 // Points to the base of the method identifier list.
885 const MethodId* method_ids_;
886
887 // Points to the base of the prototype identifier list.
888 const ProtoId* proto_ids_;
889
890 // Points to the base of the class definition list.
891 const ClassDef* class_defs_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700892};
893
894} // namespace art
895
896#endif // ART_SRC_DEX_FILE_H_