1 /*
2  * Copyright (C) 2011 The Android Open Source Project
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  *      http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 
17 #ifndef ART_LIBDEXFILE_DEX_DEX_FILE_INL_H_
18 #define ART_LIBDEXFILE_DEX_DEX_FILE_INL_H_
19 
20 #include "dex_file.h"
21 
22 #include "base/casts.h"
23 #include "base/iteration_range.h"
24 #include "base/leb128.h"
25 #include "base/utils.h"
26 #include "class_iterator.h"
27 #include "compact_dex_file.h"
28 #include "dex_instruction_iterator.h"
29 #include "invoke_type.h"
30 #include "standard_dex_file.h"
31 
32 namespace art {
33 
StringViewFromUtf16Length(const char * utf8_data,size_t utf16_length)34 inline std::string_view StringViewFromUtf16Length(const char* utf8_data, size_t utf16_length) {
35   size_t utf8_length = LIKELY(utf8_data[utf16_length] == 0)  // Is ASCII?
36                            ? utf16_length
37                            : utf16_length + strlen(utf8_data + utf16_length);
38   return std::string_view(utf8_data, utf8_length);
39 }
40 
GetStringLength(const dex::StringId & string_id)41 inline int32_t DexFile::GetStringLength(const dex::StringId& string_id) const {
42   const uint8_t* ptr = DataBegin() + string_id.string_data_off_;
43   return DecodeUnsignedLeb128(&ptr);
44 }
45 
GetStringDataAndUtf16Length(const dex::StringId & string_id,uint32_t * utf16_length)46 inline const char* DexFile::GetStringDataAndUtf16Length(const dex::StringId& string_id,
47                                                         uint32_t* utf16_length) const {
48   DCHECK(utf16_length != nullptr) << GetLocation();
49   const uint8_t* ptr = DataBegin() + string_id.string_data_off_;
50   *utf16_length = DecodeUnsignedLeb128(&ptr);
51   return reinterpret_cast<const char*>(ptr);
52 }
53 
GetStringData(const dex::StringId & string_id)54 inline const char* DexFile::GetStringData(const dex::StringId& string_id) const {
55   uint32_t ignored;
56   return GetStringDataAndUtf16Length(string_id, &ignored);
57 }
58 
StringDataAndUtf16LengthByIdx(dex::StringIndex idx,uint32_t * utf16_length)59 inline const char* DexFile::StringDataAndUtf16LengthByIdx(dex::StringIndex idx,
60                                                           uint32_t* utf16_length) const {
61   if (!idx.IsValid()) {
62     *utf16_length = 0;
63     return nullptr;
64   }
65   const dex::StringId& string_id = GetStringId(idx);
66   return GetStringDataAndUtf16Length(string_id, utf16_length);
67 }
68 
StringDataByIdx(dex::StringIndex idx)69 inline const char* DexFile::StringDataByIdx(dex::StringIndex idx) const {
70   uint32_t unicode_length;
71   return StringDataAndUtf16LengthByIdx(idx, &unicode_length);
72 }
73 
StringViewByIdx(dex::StringIndex idx)74 inline std::string_view DexFile::StringViewByIdx(dex::StringIndex idx) const {
75   uint32_t unicode_length;
76   const char* data = StringDataAndUtf16LengthByIdx(idx, &unicode_length);
77   return data != nullptr ? StringViewFromUtf16Length(data, unicode_length) : std::string_view("");
78 }
79 
StringByTypeIdx(dex::TypeIndex idx,uint32_t * unicode_length)80 inline const char* DexFile::StringByTypeIdx(dex::TypeIndex idx, uint32_t* unicode_length) const {
81   if (!idx.IsValid()) {
82     return nullptr;
83   }
84   const dex::TypeId& type_id = GetTypeId(idx);
85   return StringDataAndUtf16LengthByIdx(type_id.descriptor_idx_, unicode_length);
86 }
87 
StringByTypeIdx(dex::TypeIndex idx)88 inline const char* DexFile::StringByTypeIdx(dex::TypeIndex idx) const {
89   if (!idx.IsValid()) {
90     return nullptr;
91   }
92   const dex::TypeId& type_id = GetTypeId(idx);
93   return StringDataByIdx(type_id.descriptor_idx_);
94 }
95 
GetTypeDescriptor(const dex::TypeId & type_id)96 inline const char* DexFile::GetTypeDescriptor(const dex::TypeId& type_id) const {
97   return StringDataByIdx(type_id.descriptor_idx_);
98 }
99 
GetFieldTypeDescriptor(const dex::FieldId & field_id)100 inline const char* DexFile::GetFieldTypeDescriptor(const dex::FieldId& field_id) const {
101   const dex::TypeId& type_id = GetTypeId(field_id.type_idx_);
102   return GetTypeDescriptor(type_id);
103 }
104 
GetFieldName(const dex::FieldId & field_id)105 inline const char* DexFile::GetFieldName(const dex::FieldId& field_id) const {
106   return StringDataByIdx(field_id.name_idx_);
107 }
108 
GetMethodDeclaringClassDescriptor(const dex::MethodId & method_id)109 inline const char* DexFile::GetMethodDeclaringClassDescriptor(const dex::MethodId& method_id)
110     const {
111   const dex::TypeId& type_id = GetTypeId(method_id.class_idx_);
112   return GetTypeDescriptor(type_id);
113 }
114 
GetMethodSignature(const dex::MethodId & method_id)115 inline const Signature DexFile::GetMethodSignature(const dex::MethodId& method_id) const {
116   return Signature(this, GetProtoId(method_id.proto_idx_));
117 }
118 
GetProtoSignature(const dex::ProtoId & proto_id)119 inline const Signature DexFile::GetProtoSignature(const dex::ProtoId& proto_id) const {
120   return Signature(this, proto_id);
121 }
122 
GetMethodName(const dex::MethodId & method_id)123 inline const char* DexFile::GetMethodName(const dex::MethodId& method_id) const {
124   return StringDataByIdx(method_id.name_idx_);
125 }
126 
GetMethodName(const dex::MethodId & method_id,uint32_t * utf_length)127 inline const char* DexFile::GetMethodName(const dex::MethodId& method_id, uint32_t* utf_length)
128     const {
129   return StringDataAndUtf16LengthByIdx(method_id.name_idx_, utf_length);
130 }
131 
GetMethodName(uint32_t idx,uint32_t * utf_length)132 inline const char* DexFile::GetMethodName(uint32_t idx, uint32_t* utf_length) const {
133   return StringDataAndUtf16LengthByIdx(GetMethodId(idx).name_idx_, utf_length);
134 }
135 
GetMethodShorty(uint32_t idx)136 inline const char* DexFile::GetMethodShorty(uint32_t idx) const {
137   return StringDataByIdx(GetProtoId(GetMethodId(idx).proto_idx_).shorty_idx_);
138 }
139 
GetMethodShorty(const dex::MethodId & method_id)140 inline const char* DexFile::GetMethodShorty(const dex::MethodId& method_id) const {
141   return StringDataByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_);
142 }
143 
GetMethodShorty(const dex::MethodId & method_id,uint32_t * length)144 inline const char* DexFile::GetMethodShorty(const dex::MethodId& method_id, uint32_t* length)
145     const {
146   // Using the UTF16 length is safe here as shorties are guaranteed to be ASCII characters.
147   return StringDataAndUtf16LengthByIdx(GetProtoId(method_id.proto_idx_).shorty_idx_, length);
148 }
149 
GetClassDescriptor(const dex::ClassDef & class_def)150 inline const char* DexFile::GetClassDescriptor(const dex::ClassDef& class_def) const {
151   return StringByTypeIdx(class_def.class_idx_);
152 }
153 
GetReturnTypeDescriptor(const dex::ProtoId & proto_id)154 inline const char* DexFile::GetReturnTypeDescriptor(const dex::ProtoId& proto_id) const {
155   return StringByTypeIdx(proto_id.return_type_idx_);
156 }
157 
GetShorty(dex::ProtoIndex proto_idx)158 inline const char* DexFile::GetShorty(dex::ProtoIndex proto_idx) const {
159   const dex::ProtoId& proto_id = GetProtoId(proto_idx);
160   return StringDataByIdx(proto_id.shorty_idx_);
161 }
162 
GetTryItems(const DexInstructionIterator & code_item_end,uint32_t offset)163 inline const dex::TryItem* DexFile::GetTryItems(const DexInstructionIterator& code_item_end,
164                                                 uint32_t offset) {
165   return reinterpret_cast<const dex::TryItem*>
166       (RoundUp(reinterpret_cast<uintptr_t>(&code_item_end.Inst()), dex::TryItem::kAlignment)) +
167           offset;
168 }
169 
StringEquals(const DexFile * df1,dex::StringIndex sidx1,const DexFile * df2,dex::StringIndex sidx2)170 inline bool DexFile::StringEquals(const DexFile* df1, dex::StringIndex sidx1,
171                                   const DexFile* df2, dex::StringIndex sidx2) {
172   uint32_t s1_len;  // Note: utf16 length != mutf8 length.
173   const char* s1_data = df1->StringDataAndUtf16LengthByIdx(sidx1, &s1_len);
174   uint32_t s2_len;
175   const char* s2_data = df2->StringDataAndUtf16LengthByIdx(sidx2, &s2_len);
176   return (s1_len == s2_len) && (strcmp(s1_data, s2_data) == 0);
177 }
178 
179 template<typename NewLocalCallback, typename IndexToStringData, typename TypeIndexToStringData>
DecodeDebugLocalInfo(const uint8_t * stream,const std::string & location,const char * declaring_class_descriptor,const std::vector<const char * > & arg_descriptors,const std::string & method_name,bool is_static,uint16_t registers_size,uint16_t ins_size,uint16_t insns_size_in_code_units,const IndexToStringData & index_to_string_data,const TypeIndexToStringData & type_index_to_string_data,const NewLocalCallback & new_local_callback)180 bool DexFile::DecodeDebugLocalInfo(const uint8_t* stream,
181                                    const std::string& location,
182                                    const char* declaring_class_descriptor,
183                                    const std::vector<const char*>& arg_descriptors,
184                                    const std::string& method_name,
185                                    bool is_static,
186                                    uint16_t registers_size,
187                                    uint16_t ins_size,
188                                    uint16_t insns_size_in_code_units,
189                                    const IndexToStringData& index_to_string_data,
190                                    const TypeIndexToStringData& type_index_to_string_data,
191                                    const NewLocalCallback& new_local_callback) {
192   if (stream == nullptr) {
193     return false;
194   }
195   std::vector<LocalInfo> local_in_reg(registers_size);
196 
197   uint16_t arg_reg = registers_size - ins_size;
198   if (!is_static) {
199     const char* descriptor = declaring_class_descriptor;
200     local_in_reg[arg_reg].name_ = "this";
201     local_in_reg[arg_reg].descriptor_ = descriptor;
202     local_in_reg[arg_reg].signature_ = nullptr;
203     local_in_reg[arg_reg].start_address_ = 0;
204     local_in_reg[arg_reg].reg_ = arg_reg;
205     local_in_reg[arg_reg].is_live_ = true;
206     arg_reg++;
207   }
208 
209   DecodeUnsignedLeb128(&stream);  // Line.
210   uint32_t parameters_size = DecodeUnsignedLeb128(&stream);
211   uint32_t i;
212   if (parameters_size != arg_descriptors.size()) {
213     LOG(ERROR) << "invalid stream - problem with parameter iterator in " << location
214                << " for method " << method_name;
215     return false;
216   }
217   for (i = 0; i < parameters_size && i < arg_descriptors.size(); ++i) {
218     if (arg_reg >= registers_size) {
219       LOG(ERROR) << "invalid stream - arg reg >= reg size (" << arg_reg
220                  << " >= " << registers_size << ") in " << location;
221       return false;
222     }
223     uint32_t name_idx = DecodeUnsignedLeb128P1(&stream);
224     const char* descriptor = arg_descriptors[i];
225     local_in_reg[arg_reg].name_ = index_to_string_data(name_idx);
226     local_in_reg[arg_reg].descriptor_ = descriptor;
227     local_in_reg[arg_reg].signature_ = nullptr;
228     local_in_reg[arg_reg].start_address_ = 0;
229     local_in_reg[arg_reg].reg_ = arg_reg;
230     local_in_reg[arg_reg].is_live_ = true;
231     switch (*descriptor) {
232       case 'D':
233       case 'J':
234         arg_reg += 2;
235         break;
236       default:
237         arg_reg += 1;
238         break;
239     }
240   }
241 
242   uint32_t address = 0;
243   for (;;)  {
244     uint8_t opcode = *stream++;
245     switch (opcode) {
246       case DBG_END_SEQUENCE:
247         // Emit all variables which are still alive at the end of the method.
248         for (uint16_t reg = 0; reg < registers_size; reg++) {
249           if (local_in_reg[reg].is_live_) {
250             local_in_reg[reg].end_address_ = insns_size_in_code_units;
251             new_local_callback(local_in_reg[reg]);
252           }
253         }
254         return true;
255       case DBG_ADVANCE_PC:
256         address += DecodeUnsignedLeb128(&stream);
257         break;
258       case DBG_ADVANCE_LINE:
259         DecodeSignedLeb128(&stream);  // Line.
260         break;
261       case DBG_START_LOCAL:
262       case DBG_START_LOCAL_EXTENDED: {
263         uint16_t reg = DecodeUnsignedLeb128(&stream);
264         if (reg >= registers_size) {
265           LOG(ERROR) << "invalid stream - reg >= reg size (" << reg << " >= "
266                      << registers_size << ") in " << location;
267           return false;
268         }
269 
270         uint32_t name_idx = DecodeUnsignedLeb128P1(&stream);
271         uint16_t descriptor_idx = DecodeUnsignedLeb128P1(&stream);
272         uint32_t signature_idx = dex::kDexNoIndex;
273         if (opcode == DBG_START_LOCAL_EXTENDED) {
274           signature_idx = DecodeUnsignedLeb128P1(&stream);
275         }
276 
277         // Emit what was previously there, if anything
278         if (local_in_reg[reg].is_live_) {
279           local_in_reg[reg].end_address_ = address;
280           new_local_callback(local_in_reg[reg]);
281         }
282 
283         local_in_reg[reg].name_ = index_to_string_data(name_idx);
284         local_in_reg[reg].descriptor_ = type_index_to_string_data(descriptor_idx);;
285         local_in_reg[reg].signature_ = index_to_string_data(signature_idx);
286         local_in_reg[reg].start_address_ = address;
287         local_in_reg[reg].reg_ = reg;
288         local_in_reg[reg].is_live_ = true;
289         break;
290       }
291       case DBG_END_LOCAL: {
292         uint16_t reg = DecodeUnsignedLeb128(&stream);
293         if (reg >= registers_size) {
294           LOG(ERROR) << "invalid stream - reg >= reg size (" << reg << " >= "
295                      << registers_size << ") in " << location;
296           return false;
297         }
298         // If the register is live, close it properly. Otherwise, closing an already
299         // closed register is sloppy, but harmless if no further action is taken.
300         if (local_in_reg[reg].is_live_) {
301           local_in_reg[reg].end_address_ = address;
302           new_local_callback(local_in_reg[reg]);
303           local_in_reg[reg].is_live_ = false;
304         }
305         break;
306       }
307       case DBG_RESTART_LOCAL: {
308         uint16_t reg = DecodeUnsignedLeb128(&stream);
309         if (reg >= registers_size) {
310           LOG(ERROR) << "invalid stream - reg >= reg size (" << reg << " >= "
311                      << registers_size << ") in " << location;
312           return false;
313         }
314         // If the register is live, the "restart" is superfluous,
315         // and we don't want to mess with the existing start address.
316         if (!local_in_reg[reg].is_live_) {
317           local_in_reg[reg].start_address_ = address;
318           local_in_reg[reg].is_live_ = true;
319         }
320         break;
321       }
322       case DBG_SET_PROLOGUE_END:
323       case DBG_SET_EPILOGUE_BEGIN:
324         break;
325       case DBG_SET_FILE:
326         DecodeUnsignedLeb128P1(&stream);  // name.
327         break;
328       default:
329         address += (opcode - DBG_FIRST_SPECIAL) / DBG_LINE_RANGE;
330         break;
331     }
332   }
333 }
334 
335 template<typename NewLocalCallback>
DecodeDebugLocalInfo(uint32_t registers_size,uint32_t ins_size,uint32_t insns_size_in_code_units,uint32_t debug_info_offset,bool is_static,uint32_t method_idx,const NewLocalCallback & new_local_callback)336 bool DexFile::DecodeDebugLocalInfo(uint32_t registers_size,
337                                    uint32_t ins_size,
338                                    uint32_t insns_size_in_code_units,
339                                    uint32_t debug_info_offset,
340                                    bool is_static,
341                                    uint32_t method_idx,
342                                    const NewLocalCallback& new_local_callback) const {
343   const uint8_t* const stream = GetDebugInfoStream(debug_info_offset);
344   if (stream == nullptr) {
345     return false;
346   }
347   std::vector<const char*> arg_descriptors;
348   DexFileParameterIterator it(*this, GetMethodPrototype(GetMethodId(method_idx)));
349   for (; it.HasNext(); it.Next()) {
350     arg_descriptors.push_back(it.GetDescriptor());
351   }
352   return DecodeDebugLocalInfo(stream,
353                               GetLocation(),
354                               GetMethodDeclaringClassDescriptor(GetMethodId(method_idx)),
355                               arg_descriptors,
356                               this->PrettyMethod(method_idx),
357                               is_static,
358                               registers_size,
359                               ins_size,
360                               insns_size_in_code_units,
361                               [this](uint32_t idx) {
362                                 return StringDataByIdx(dex::StringIndex(idx));
363                               },
364                               [this](uint32_t idx) {
365                                 return StringByTypeIdx(dex::TypeIndex(
366                                     dchecked_integral_cast<uint16_t>(idx)));
367                               },
368                               new_local_callback);
369 }
370 
371 template<typename DexDebugNewPosition, typename IndexToStringData>
DecodeDebugPositionInfo(const uint8_t * stream,const IndexToStringData & index_to_string_data,const DexDebugNewPosition & position_functor)372 bool DexFile::DecodeDebugPositionInfo(const uint8_t* stream,
373                                       const IndexToStringData& index_to_string_data,
374                                       const DexDebugNewPosition& position_functor) {
375   if (stream == nullptr) {
376     return false;
377   }
378 
379   PositionInfo entry;
380   entry.line_ = DecodeDebugInfoParameterNames(&stream, VoidFunctor());
381 
382   for (;;)  {
383     uint8_t opcode = *stream++;
384     switch (opcode) {
385       case DBG_END_SEQUENCE:
386         return true;  // end of stream.
387       case DBG_ADVANCE_PC:
388         entry.address_ += DecodeUnsignedLeb128(&stream);
389         break;
390       case DBG_ADVANCE_LINE:
391         entry.line_ += DecodeSignedLeb128(&stream);
392         break;
393       case DBG_START_LOCAL:
394         DecodeUnsignedLeb128(&stream);  // reg.
395         DecodeUnsignedLeb128P1(&stream);  // name.
396         DecodeUnsignedLeb128P1(&stream);  // descriptor.
397         break;
398       case DBG_START_LOCAL_EXTENDED:
399         DecodeUnsignedLeb128(&stream);  // reg.
400         DecodeUnsignedLeb128P1(&stream);  // name.
401         DecodeUnsignedLeb128P1(&stream);  // descriptor.
402         DecodeUnsignedLeb128P1(&stream);  // signature.
403         break;
404       case DBG_END_LOCAL:
405       case DBG_RESTART_LOCAL:
406         DecodeUnsignedLeb128(&stream);  // reg.
407         break;
408       case DBG_SET_PROLOGUE_END:
409         entry.prologue_end_ = true;
410         break;
411       case DBG_SET_EPILOGUE_BEGIN:
412         entry.epilogue_begin_ = true;
413         break;
414       case DBG_SET_FILE: {
415         uint32_t name_idx = DecodeUnsignedLeb128P1(&stream);
416         entry.source_file_ = index_to_string_data(name_idx);
417         break;
418       }
419       default: {
420         int adjopcode = opcode - DBG_FIRST_SPECIAL;
421         entry.address_ += adjopcode / DBG_LINE_RANGE;
422         entry.line_ += DBG_LINE_BASE + (adjopcode % DBG_LINE_RANGE);
423         if (position_functor(entry)) {
424           return true;  // early exit.
425         }
426         entry.prologue_end_ = false;
427         entry.epilogue_begin_ = false;
428         break;
429       }
430     }
431   }
432 }
433 
AsCompactDexFile()434 inline const CompactDexFile* DexFile::AsCompactDexFile() const {
435   DCHECK(IsCompactDexFile());
436   return down_cast<const CompactDexFile*>(this);
437 }
438 
AsStandardDexFile()439 inline const StandardDexFile* DexFile::AsStandardDexFile() const {
440   DCHECK(IsStandardDexFile());
441   return down_cast<const StandardDexFile*>(this);
442 }
443 
444 // Get the base of the encoded data for the given DexCode.
GetCatchHandlerData(const DexInstructionIterator & code_item_end,uint32_t tries_size,uint32_t offset)445 inline const uint8_t* DexFile::GetCatchHandlerData(const DexInstructionIterator& code_item_end,
446                                                    uint32_t tries_size,
447                                                    uint32_t offset) {
448   const uint8_t* handler_data =
449       reinterpret_cast<const uint8_t*>(GetTryItems(code_item_end, tries_size));
450   return handler_data + offset;
451 }
452 
GetClasses()453 inline IterationRange<ClassIterator> DexFile::GetClasses() const {
454   return { ClassIterator(*this, 0u), ClassIterator(*this, NumClassDefs()) };
455 }
456 
457 // Returns the line number
458 template <typename Visitor>
DecodeDebugInfoParameterNames(const uint8_t ** debug_info,const Visitor & visitor)459 inline uint32_t DexFile::DecodeDebugInfoParameterNames(const uint8_t** debug_info,
460                                                        const Visitor& visitor) {
461   uint32_t line = DecodeUnsignedLeb128(debug_info);
462   const uint32_t parameters_size = DecodeUnsignedLeb128(debug_info);
463   for (uint32_t i = 0; i < parameters_size; ++i) {
464     visitor(dex::StringIndex(DecodeUnsignedLeb128P1(debug_info)));
465   }
466   return line;
467 }
468 
469 }  // namespace art
470 
471 #endif  // ART_LIBDEXFILE_DEX_DEX_FILE_INL_H_
472