blob: db7a54b680508f689e6ceef8a9dcc95e95cd5dea [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;
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 Carlstrom7e49dca2011-07-22 18:07:34 -0700319 // Opens a .dex file from the file system.
Brian Carlstromb0460ea2011-07-29 10:08:05 -0700320 static DexFile* OpenFile(const std::string& filename);
321
322 // Opens a .jar, .zip, or .apk file from the file system.
323 static DexFile* OpenZip(const std::string& filename);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700324
Brian Carlstroma663ea52011-08-19 23:33:41 -0700325 // Opens a .dex file from a new allocated pointer. location is used
326 // to identify the source, for example "/system/framework/core.jar"
327 // or "contrived-test-42". When initializing a ClassLinker from an
328 // image, the location is used to match DexCaches the image to their
329 // corresponding DexFiles.N
330 static DexFile* OpenPtr(byte* ptr, size_t length, const std::string& location);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700331
332 // Closes a .dex file.
Brian Carlstromf615a612011-07-23 12:50:34 -0700333 virtual ~DexFile();
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700334
Brian Carlstroma663ea52011-08-19 23:33:41 -0700335 const std::string& GetLocation() const {
336 return location_;
337 }
338
339 const Header& GetHeader() const {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700340 CHECK(header_ != NULL);
341 return *header_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700342 }
343
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700344 // Looks up a class definition by its class descriptor.
345 const ClassDef* FindClassDef(const StringPiece& descriptor) const;
346
347 // Returns the number of string identifiers in the .dex file.
348 size_t NumStringIds() const {
349 CHECK(header_ != NULL);
350 return header_->string_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700351 }
352
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700353 // Returns the number of type identifiers in the .dex file.
354 size_t NumTypeIds() const {
355 CHECK(header_ != NULL);
356 return header_->type_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700357 }
358
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700359 // Returns the number of prototype identifiers in the .dex file.
360 size_t NumProtoIds() const {
361 CHECK(header_ != NULL);
362 return header_->proto_ids_size_;
Carl Shapiro5fafe2b2011-07-09 15:34:41 -0700363 }
364
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700365 // Returns the number of field identifiers in the .dex file.
366 size_t NumFieldIds() const {
367 CHECK(header_ != NULL);
368 return header_->field_ids_size_;
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700369 }
370
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700371 // Returns the number of method identifiers in the .dex file.
372 size_t NumMethodIds() const {
373 CHECK(header_ != NULL);
374 return header_->method_ids_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700375 }
376
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700377 // Returns the number of class definitions in the .dex file.
378 size_t NumClassDefs() const {
379 CHECK(header_ != NULL);
380 return header_->class_defs_size_;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700381 }
382
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700383 // Returns a pointer to the memory mapped class data.
384 // TODO: return a stream
385 const byte* GetClassData(const ClassDef& class_def) const {
386 if (class_def.class_data_off_ == 0) {
387 return NULL;
388 } else {
389 return base_ + class_def.class_data_off_;
390 }
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700391 }
392
Brian Carlstromf615a612011-07-23 12:50:34 -0700393 // Decodes the header section from the class data bytes.
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700394 ClassDataHeader ReadClassDataHeader(const byte** class_data) const {
395 CHECK(class_data != NULL);
396 ClassDataHeader header;
397 memset(&header, 0, sizeof(ClassDataHeader));
398 if (*class_data != NULL) {
399 header.static_fields_size_ = DecodeUnsignedLeb128(class_data);
400 header.instance_fields_size_ = DecodeUnsignedLeb128(class_data);
401 header.direct_methods_size_ = DecodeUnsignedLeb128(class_data);
402 header.virtual_methods_size_ = DecodeUnsignedLeb128(class_data);
403 }
404 return header;
Brian Carlstrom578bbdc2011-07-21 14:07:47 -0700405 }
406
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700407 // Returns the class descriptor string of a class definition.
408 const char* GetClassDescriptor(const ClassDef& class_def) const {
409 return dexStringByTypeIdx(class_def.class_idx_);
410 }
411
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700412 // Returns the type descriptor string of a type id.
413 const char* GetTypeDescriptor(const TypeId& type_id) const {
414 return dexStringById(type_id.descriptor_idx_);
415 }
416
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700417 // Returns the StringId at the specified index.
418 const StringId& GetStringId(uint32_t idx) const {
419 CHECK_LT(idx, NumStringIds());
420 return string_ids_[idx];
421 }
422
423 // Returns the TypeId at the specified index.
424 const TypeId& GetTypeId(uint32_t idx) const {
425 CHECK_LT(idx, NumTypeIds());
426 return type_ids_[idx];
427 }
428
429 // Returns the FieldId at the specified index.
430 const FieldId& GetFieldId(uint32_t idx) const {
431 CHECK_LT(idx, NumFieldIds());
432 return field_ids_[idx];
433 }
434
435 // Returns the MethodId at the specified index.
436 const MethodId& GetMethodId(uint32_t idx) const {
437 CHECK_LT(idx, NumMethodIds());
438 return method_ids_[idx];
439 }
440
441 // Returns the ProtoId at the specified index.
442 const ProtoId& GetProtoId(uint32_t idx) const {
443 CHECK_LT(idx, NumProtoIds());
444 return proto_ids_[idx];
445 }
446
447 // Returns the ClassDef at the specified index.
448 const ClassDef& GetClassDef(uint32_t idx) const {
449 CHECK_LT(idx, NumClassDefs());
450 return class_defs_[idx];
451 }
452
453 const TypeList* GetInterfacesList(const ClassDef& class_def) const {
454 if (class_def.interfaces_off_ == 0) {
455 return NULL;
456 } else {
457 const byte* addr = base_ + class_def.interfaces_off_;
458 return reinterpret_cast<const TypeList*>(addr);
459 }
460 }
461
462 const CodeItem* GetCodeItem(const Method& method) const {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700463 return GetCodeItem(method.code_off_);
464 }
465
466 const CodeItem* GetCodeItem(const uint32_t code_off_) const {
467 if (code_off_ == 0) {
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700468 return NULL; // native or abstract method
469 } else {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700470 const byte* addr = base_ + code_off_;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700471 return reinterpret_cast<const CodeItem*>(addr);
472 }
473 }
474
475 // Returns the short form method descriptor for the given prototype.
476 const char* GetShorty(uint32_t proto_idx) const {
477 const ProtoId& proto_id = GetProtoId(proto_idx);
478 return dexStringById(proto_id.shorty_idx_);
479 }
480
481 const TypeList* GetProtoParameters(const ProtoId& proto_id) const {
482 if (proto_id.parameters_off_ == 0) {
483 return NULL;
484 } else {
485 const byte* addr = base_ + proto_id.parameters_off_;
486 return reinterpret_cast<const TypeList*>(addr);
487 }
488 }
489
Elliott Hughes0c424cb2011-08-26 10:16:25 -0700490 std::string CreateMethodDescriptor(uint32_t proto_idx, int32_t* unicode_length) const;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700491
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700492 const byte* GetEncodedArray(const ClassDef& class_def) const {
493 if (class_def.static_values_off_ == 0) {
494 return 0;
495 } else {
496 return base_ + class_def.static_values_off_;
497 }
498 }
499
500 int32_t GetStringLength(const StringId& string_id) const {
501 const byte* ptr = base_ + string_id.string_data_off_;
502 return DecodeUnsignedLeb128(&ptr);
503 }
504
505 ValueType ReadEncodedValue(const byte** encoded_value, JValue* value) const;
506
507 // From libdex...
508
509 // Returns a pointer to the UTF-8 string data referred to by the
510 // given string_id.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700511 const char* GetStringData(const StringId& string_id, int32_t* length) const {
512 CHECK(length != NULL);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700513 const byte* ptr = base_ + string_id.string_data_off_;
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700514 *length = DecodeUnsignedLeb128(&ptr);
Brian Carlstrom0b138b22011-07-27 15:19:17 -0700515 return reinterpret_cast<const char*>(ptr);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700516 }
517
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700518 const char* GetStringData(const StringId& string_id) const {
519 int32_t length;
520 return GetStringData(string_id, &length);
521 }
522
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700523 // return the UTF-8 encoded string with the specified string_id index
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700524 const char* dexStringById(uint32_t idx, int32_t* unicode_length) const {
Shih-wei Liao195487c2011-08-20 13:29:04 -0700525 if (idx == kDexNoIndex) {
526 *unicode_length = 0;
527 return NULL;
528 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700529 const StringId& string_id = GetStringId(idx);
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700530 return GetStringData(string_id, unicode_length);
531 }
532
533 const char* dexStringById(uint32_t idx) const {
534 int32_t unicode_length;
535 return dexStringById(idx, &unicode_length);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700536 }
537
Shih-wei Liao55df06b2011-08-26 14:39:27 -0700538 String* dexArtStringById(int32_t idx) const;
Shih-wei Liao195487c2011-08-20 13:29:04 -0700539
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700540 // Get the descriptor string associated with a given type index.
Carl Shapiro419ec7b2011-08-03 14:48:33 -0700541 const char* dexStringByTypeIdx(uint32_t idx, int32_t* unicode_length) const {
542 const TypeId& type_id = GetTypeId(idx);
543 return dexStringById(type_id.descriptor_idx_, unicode_length);
544 }
545
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700546 const char* dexStringByTypeIdx(uint32_t idx) const {
547 const TypeId& type_id = GetTypeId(idx);
548 return dexStringById(type_id.descriptor_idx_);
549 }
550
Shih-wei Liao55df06b2011-08-26 14:39:27 -0700551 String* dexArtStringByTypeIdx(int32_t idx) const {
Shih-wei Liao195487c2011-08-20 13:29:04 -0700552 const TypeId& type_id = GetTypeId(idx);
553 return dexArtStringById(type_id.descriptor_idx_);
554 }
555
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700556 // TODO: encoded_field is actually a stream of bytes
557 void dexReadClassDataField(const byte** encoded_field,
Brian Carlstromf615a612011-07-23 12:50:34 -0700558 DexFile::Field* field,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700559 uint32_t* last_idx) const {
560 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_field);
561 field->access_flags_ = DecodeUnsignedLeb128(encoded_field);
562 field->field_idx_ = idx;
563 *last_idx = idx;
564 }
565
566 // TODO: encoded_method is actually a stream of bytes
567 void dexReadClassDataMethod(const byte** encoded_method,
Brian Carlstromf615a612011-07-23 12:50:34 -0700568 DexFile::Method* method,
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700569 uint32_t* last_idx) const {
570 uint32_t idx = *last_idx + DecodeUnsignedLeb128(encoded_method);
571 method->access_flags_ = DecodeUnsignedLeb128(encoded_method);
572 method->code_off_ = DecodeUnsignedLeb128(encoded_method);
573 method->method_idx_ = idx;
574 *last_idx = idx;
575 }
576
jeffhaoba5ebb92011-08-25 17:24:37 -0700577 static const TryItem* dexGetTryItems(const CodeItem& code_item, uint32_t offset) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700578 const uint16_t* insns_end_ = &code_item.insns_[code_item.insns_size_];
579 return reinterpret_cast<const TryItem*>
580 (RoundUp(reinterpret_cast<uint32_t>(insns_end_), 4)) + offset;
581 }
582
583 // Get the base of the encoded data for the given DexCode.
jeffhaoba5ebb92011-08-25 17:24:37 -0700584 static const byte* dexGetCatchHandlerData(const CodeItem& code_item, uint32_t offset) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700585 const byte* handler_data = reinterpret_cast<const byte*>
586 (dexGetTryItems(code_item, code_item.tries_size_));
587 return handler_data + offset;
588 }
589
590 // Find the handler associated with a given address, if any.
591 // Initializes the given iterator and returns true if a match is
592 // found. Returns end if there is no applicable handler.
jeffhaoba5ebb92011-08-25 17:24:37 -0700593 static CatchHandlerIterator dexFindCatchHandler(const CodeItem& code_item, uint32_t address) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700594 CatchHandlerItem handler;
595 handler.address_ = -1;
596 int32_t offset = -1;
597
598 // Short-circuit the overwhelmingly common cases.
599 switch (code_item.tries_size_) {
600 case 0:
601 break;
602 case 1: {
603 const TryItem* tries = dexGetTryItems(code_item, 0);
604 uint32_t start = tries->start_addr_;
605 if (address < start)
606 break;
607
608 uint32_t end = start + tries->insn_count_;
609 if (address >= end)
610 break;
611
612 offset = tries->handler_off_;
613 break;
614 }
615 default:
616 offset = dexFindCatchHandlerOffset0(code_item, code_item.tries_size_, address);
617 }
618
619 if (offset >= 0) {
620 const byte* handler_data = dexGetCatchHandlerData(code_item, offset);
621 return CatchHandlerIterator(handler_data);
622 }
623 return CatchHandlerIterator();
624 }
625
jeffhaoba5ebb92011-08-25 17:24:37 -0700626 static int32_t dexFindCatchHandlerOffset0(const CodeItem &code_item,
627 int32_t tries_size,
628 uint32_t address) {
Shih-wei Liao2fb97532011-08-11 16:17:23 -0700629 // Note: Signed type is important for max and min.
630 int32_t min = 0;
631 int32_t max = tries_size - 1;
632
633 while (max >= min) {
634 int32_t guess = (min + max) >> 1;
635 const TryItem* pTry = dexGetTryItems(code_item, guess);
636 uint32_t start = pTry->start_addr_;
637
638 if (address < start) {
639 max = guess - 1;
640 continue;
641 }
642
643 uint32_t end = start + pTry->insn_count_;
644 if (address >= end) {
645 min = guess + 1;
646 continue;
647 }
648
649 // We have a winner!
650 return (int32_t) pTry->handler_off_;
651 }
652
653 // No match.
654 return -1;
655 }
656
Shih-wei Liao195487c2011-08-20 13:29:04 -0700657 // Get the pointer to the start of the debugging data
658 const byte* dexGetDebugInfoStream(const CodeItem* code_item) const {
659 if (code_item->debug_info_off_ == 0) {
660 return NULL;
661 } else {
662 return base_ + code_item->debug_info_off_;
663 }
664 }
665
666 // Callback for "new position table entry".
667 // Returning true causes the decoder to stop early.
668 typedef bool (*DexDebugNewPositionCb)(void *cnxt, uint32_t address, uint32_t line_num);
669
670 // Callback for "new locals table entry". "signature" is an empty string
671 // if no signature is available for an entry.
672 typedef void (*DexDebugNewLocalCb)(void *cnxt, uint16_t reg,
673 uint32_t startAddress,
674 uint32_t endAddress,
675 const String* name,
676 const String* descriptor,
677 const String* signature);
678
679 static bool LineNumForPcCb(void *cnxt, uint32_t address, uint32_t line_num) {
680 LineNumFromPcContext *context = (LineNumFromPcContext *)cnxt;
681
682 // We know that this callback will be called in
683 // ascending address order, so keep going until we find
684 // a match or we've just gone past it.
685 if (address > context->address_) {
686 // The line number from the previous positions callback
687 // wil be the final result.
688 return true;
689 } else {
690 context->line_num_ = line_num;
691 return address == context->address_;
692 }
693 }
694
695
696 // Debug info opcodes and constants
697 enum {
698 DBG_END_SEQUENCE = 0x00,
699 DBG_ADVANCE_PC = 0x01,
700 DBG_ADVANCE_LINE = 0x02,
701 DBG_START_LOCAL = 0x03,
702 DBG_START_LOCAL_EXTENDED = 0x04,
703 DBG_END_LOCAL = 0x05,
704 DBG_RESTART_LOCAL = 0x06,
705 DBG_SET_PROLOGUE_END = 0x07,
706 DBG_SET_EPILOGUE_BEGIN = 0x08,
707 DBG_SET_FILE = 0x09,
708 DBG_FIRST_SPECIAL = 0x0a,
709 DBG_LINE_BASE = -4,
710 DBG_LINE_RANGE = 15,
711 };
712
713 struct LocalInfo {
714 LocalInfo() : name_(NULL), descriptor_(NULL), signature_(NULL), start_address_(0), is_live_(false) {}
715
716 // E.g., list
717 const String* name_;
718
719 // E.g., Ljava/util/LinkedList;
720 const String* descriptor_;
721
722 // E.g., java.util.LinkedList<java.lang.Integer>
723 const String* signature_;
724
725 // PC location where the local is first defined.
726 uint16_t start_address_;
727
728 // Is the local defined and live.
729 bool is_live_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700730
731 private:
732 DISALLOW_COPY_AND_ASSIGN(LocalInfo);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700733 };
734
735 struct LineNumFromPcContext {
736 LineNumFromPcContext(uint32_t address, uint32_t line_num) :
737 address_(address), line_num_(line_num) {}
738 uint32_t address_;
739 uint32_t line_num_;
Brian Carlstromd2fbb2b2011-08-23 11:57:08 -0700740 private:
741 DISALLOW_COPY_AND_ASSIGN(LineNumFromPcContext);
Shih-wei Liao195487c2011-08-20 13:29:04 -0700742 };
743
744 void InvokeLocalCbIfLive(void *cnxt, int reg, uint32_t end_address,
745 LocalInfo *local_in_reg, DexDebugNewLocalCb local_cb) const {
746 if (local_cb != NULL && local_in_reg[reg].is_live_) {
747 local_cb(cnxt, reg, local_in_reg[reg].start_address_, end_address,
748 local_in_reg[reg].name_, local_in_reg[reg].descriptor_,
749 local_in_reg[reg].signature_);
750 }
751 }
752
753 // Determine the source file line number based on the program counter.
754 // "pc" is an offset, in 16-bit units, from the start of the method's code.
755 //
756 // Returns -1 if no match was found (possibly because the source files were
757 // compiled without "-g", so no line number information is present).
758 // Returns -2 for native methods (as expected in exception traces).
759 //
760 // This is used by runtime; therefore use art::Method not art::DexFile::Method.
761 int32_t GetLineNumFromPC(const art::Method* method, uint32_t rel_pc) const;
762
763 void dexDecodeDebugInfo0(const CodeItem* code_item, const art::Method* method,
764 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
765 void* cnxt, const byte* stream, LocalInfo* local_in_reg) const;
766
767 void dexDecodeDebugInfo(const CodeItem* code_item, const art::Method *method,
768 DexDebugNewPositionCb posCb, DexDebugNewLocalCb local_cb,
769 void* cnxt) const {
770 const byte* stream = dexGetDebugInfoStream(code_item);
771 LocalInfo local_in_reg[code_item->registers_size_];
772
773 if (stream != NULL) {
774 dexDecodeDebugInfo0(code_item, method, posCb, local_cb, cnxt, stream, local_in_reg);
775 }
776 for (int reg = 0; reg < code_item->registers_size_; reg++) {
777 InvokeLocalCbIfLive(cnxt, reg, code_item->insns_size_, local_in_reg, local_cb);
778 }
779 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700780
781 // TODO: const reference
782 uint32_t dexGetIndexForClassDef(const ClassDef* class_def) const {
783 CHECK_GE(class_def, class_defs_);
784 CHECK_LT(class_def, class_defs_ + header_->class_defs_size_);
785 return class_def - class_defs_;
786 }
787
788 const char* dexGetSourceFile(const ClassDef& class_def) const {
789 if (class_def.source_file_idx_ == 0xffffffff) {
790 return NULL;
791 } else {
792 return dexStringById(class_def.source_file_idx_);
793 }
Carl Shapiro0e5d75d2011-07-06 18:28:37 -0700794 }
795
Carl Shapiro1fb86202011-06-27 17:43:13 -0700796 private:
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700797 // Helper class to deallocate underlying storage.
798 class Closer {
799 public:
800 virtual ~Closer();
801 };
802
803 // Helper class to deallocate mmap-backed .dex files.
804 class MmapCloser : public Closer {
805 public:
806 MmapCloser(void* addr, size_t length);
807 virtual ~MmapCloser();
808 private:
809 void* addr_;
810 size_t length_;
811 };
812
813 // Helper class for deallocating new/delete-backed .dex files.
814 class PtrCloser : public Closer {
815 public:
816 PtrCloser(byte* addr);
817 virtual ~PtrCloser();
818 private:
819 byte* addr_;
820 };
821
822 // Opens a .dex file at a the given address.
Brian Carlstroma663ea52011-08-19 23:33:41 -0700823 static DexFile* Open(const byte* dex_file, size_t length, const std::string& location, Closer* closer);
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700824
Brian Carlstroma663ea52011-08-19 23:33:41 -0700825 DexFile(const byte* addr, size_t length, const std::string& location, Closer* closer)
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700826 : base_(addr),
827 length_(length),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700828 location_(location),
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700829 closer_(closer),
830 header_(0),
831 string_ids_(0),
832 type_ids_(0),
833 field_ids_(0),
834 method_ids_(0),
835 proto_ids_(0),
Brian Carlstroma663ea52011-08-19 23:33:41 -0700836 class_defs_(0) {
837 CHECK(addr != NULL);
838 CHECK_GT(length, 0U);
839 CHECK(closer != NULL);
840 }
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700841
842 // Top-level initializer that calls other Init methods.
843 bool Init();
844
845 // Caches pointers into to the various file sections.
846 void InitMembers();
847
848 // Builds the index of descriptors to class definitions.
849 void InitIndex();
850
851 // Returns true if the byte string equals the magic value.
852 bool CheckMagic(const byte* magic);
853
854 // Returns true if the header magic is of the expected value.
855 bool IsMagicValid();
856
857 // The index of descriptors to class definitions.
Brian Carlstromf615a612011-07-23 12:50:34 -0700858 typedef std::map<const StringPiece, const DexFile::ClassDef*> Index;
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700859 Index index_;
860
861 // The base address of the memory mapping.
862 const byte* base_;
863
864 // The size of the underlying memory allocation in bytes.
865 size_t length_;
866
Brian Carlstroma663ea52011-08-19 23:33:41 -0700867 // Typically the dex file name when availble, alternatively some identifying string.
868 //
869 // The ClassLinker will use this to match DexFiles the boot class
870 // path to DexCache::GetLocation when loading from an image.
871 const std::string location_;
872
Brian Carlstrom7e49dca2011-07-22 18:07:34 -0700873 // Helper object to free the underlying allocation.
874 scoped_ptr<Closer> closer_;
875
876 // Points to the header section.
877 const Header* header_;
878
879 // Points to the base of the string identifier list.
880 const StringId* string_ids_;
881
882 // Points to the base of the type identifier list.
883 const TypeId* type_ids_;
884
885 // Points to the base of the field identifier list.
886 const FieldId* field_ids_;
887
888 // Points to the base of the method identifier list.
889 const MethodId* method_ids_;
890
891 // Points to the base of the prototype identifier list.
892 const ProtoId* proto_ids_;
893
894 // Points to the base of the class definition list.
895 const ClassDef* class_defs_;
Carl Shapiro1fb86202011-06-27 17:43:13 -0700896};
897
898} // namespace art
899
900#endif // ART_SRC_DEX_FILE_H_