Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1 | // Copyright 2018 The Chromium Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | #include "components/zucchini/disassembler_dex.h" |
| 6 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 7 | #include <stddef.h> |
| 8 | #include <stdlib.h> |
| 9 | |
| 10 | #include <algorithm> |
| 11 | #include <cctype> |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 12 | #include <cmath> |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 13 | #include <iterator> |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 14 | #include <set> |
| 15 | #include <utility> |
| 16 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 17 | #include "base/bind.h" |
| 18 | #include "base/callback.h" |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 19 | #include "base/logging.h" |
Peter Kasting | ead42e9 | 2021-07-08 15:04:58 +0000 | [diff] [blame] | 20 | #include "base/numerics/checked_math.h" |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 21 | #include "base/numerics/safe_conversions.h" |
| 22 | #include "base/strings/stringprintf.h" |
| 23 | #include "components/zucchini/buffer_source.h" |
| 24 | #include "components/zucchini/buffer_view.h" |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 25 | #include "components/zucchini/io_utils.h" |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 26 | #include "third_party/abseil-cpp/absl/types/optional.h" |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 27 | |
| 28 | namespace zucchini { |
| 29 | |
| 30 | namespace { |
| 31 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 32 | // A DEX item specified by an offset, if absent, has a sentinel value of 0 since |
| 33 | // 0 is never a valid item offset (it points to magic at start of DEX). |
| 34 | constexpr offset_t kDexSentinelOffset = 0U; |
| 35 | |
| 36 | // A DEX item specified by an index, if absent, has a sentinel value of |
| 37 | // NO_INDEX = 0xFFFFFFFF. This is represented as an offset_t for uniformity. |
| 38 | constexpr offset_t kDexSentinelIndexAsOffset = 0xFFFFFFFFU; |
| 39 | |
| 40 | static_assert(kDexSentinelIndexAsOffset != kInvalidOffset, |
| 41 | "Sentinel should not be confused with invalid offset."); |
| 42 | |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 43 | // Size of a Dalvik instruction unit. Need to cast to signed int because |
| 44 | // sizeof() gives size_t, which dominates when operated on ptrdiff_t, then |
| 45 | // wrecks havoc for base::checked_cast<int16_t>(). |
| 46 | constexpr int kInstrUnitSize = static_cast<int>(sizeof(uint16_t)); |
| 47 | |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 48 | // Checks if |offset| is byte aligned to 32 bits or 4 bytes. |
| 49 | bool Is32BitAligned(offset_t offset) { |
| 50 | return offset % 4 == 0; |
| 51 | } |
| 52 | |
Samuel Huang | 06a9be1 | 2018-07-11 20:02:23 +0000 | [diff] [blame] | 53 | // Returns a lower bound for the size of an item of type |type_item_code|. |
| 54 | // - For fixed-length items (e.g., kTypeFieldIdItem) this is the exact size. |
| 55 | // - For variant-length items (e.g., kTypeCodeItem), returns a value that is |
| 56 | // known to be less than the item length (e.g., header size). |
| 57 | // - For items not handled by this function, returns 1 for sanity check. |
| 58 | size_t GetItemBaseSize(uint16_t type_item_code) { |
| 59 | switch (type_item_code) { |
| 60 | case dex::kTypeStringIdItem: |
| 61 | return sizeof(dex::StringIdItem); |
| 62 | case dex::kTypeTypeIdItem: |
| 63 | return sizeof(dex::TypeIdItem); |
| 64 | case dex::kTypeProtoIdItem: |
| 65 | return sizeof(dex::ProtoIdItem); |
| 66 | case dex::kTypeFieldIdItem: |
| 67 | return sizeof(dex::FieldIdItem); |
| 68 | case dex::kTypeMethodIdItem: |
| 69 | return sizeof(dex::MethodIdItem); |
| 70 | case dex::kTypeClassDefItem: |
| 71 | return sizeof(dex::ClassDefItem); |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 72 | case dex::kTypeCallSiteIdItem: |
| 73 | return sizeof(dex::CallSiteIdItem); |
| 74 | case dex::kTypeMethodHandleItem: |
| 75 | return sizeof(dex::MethodHandleItem); |
Samuel Huang | 06a9be1 | 2018-07-11 20:02:23 +0000 | [diff] [blame] | 76 | // No need to handle dex::kTypeMapList. |
| 77 | case dex::kTypeTypeList: |
| 78 | return sizeof(uint32_t); // Variable-length. |
| 79 | case dex::kTypeAnnotationSetRefList: |
| 80 | return sizeof(uint32_t); // Variable-length. |
| 81 | case dex::kTypeAnnotationSetItem: |
| 82 | return sizeof(uint32_t); // Variable-length. |
| 83 | case dex::kTypeCodeItem: |
| 84 | return sizeof(dex::CodeItem); // Variable-length. |
| 85 | case dex::kTypeAnnotationsDirectoryItem: |
| 86 | return sizeof(dex::AnnotationsDirectoryItem); // Variable-length. |
| 87 | default: |
| 88 | return 1U; // Unhandled item. For sanity check assume size >= 1. |
| 89 | } |
| 90 | } |
| 91 | |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 92 | /******** CodeItemParser ********/ |
| 93 | |
| 94 | // A parser to extract successive code items from a DEX image whose header has |
| 95 | // been parsed. |
| 96 | class CodeItemParser { |
| 97 | public: |
| 98 | using size_type = BufferSource::size_type; |
| 99 | |
| 100 | explicit CodeItemParser(ConstBufferView image) : image_(image) {} |
| 101 | |
| 102 | // Initializes the parser, returns true on success and false on error. |
| 103 | bool Init(const dex::MapItem& code_map_item) { |
| 104 | // Sanity check to quickly fail if |code_map_item.offset| or |
| 105 | // |code_map_item.size| is too large. This is a heuristic because code item |
| 106 | // sizes need to be parsed (sizeof(dex::CodeItem) is a lower bound). |
| 107 | if (!image_.covers_array(code_map_item.offset, code_map_item.size, |
| 108 | sizeof(dex::CodeItem))) { |
| 109 | return false; |
| 110 | } |
| 111 | source_ = std::move(BufferSource(image_).Skip(code_map_item.offset)); |
| 112 | return true; |
| 113 | } |
| 114 | |
| 115 | // Extracts the header of the next code item, and skips the variable-length |
| 116 | // data. Returns the offset of the code item if successful. Otherwise returns |
| 117 | // kInvalidOffset, and thereafter the parser becomes valid. For reference, |
| 118 | // here's a pseudo-struct of a complete code item: |
| 119 | // |
| 120 | // struct code_item { |
| 121 | // // 4-byte aligned here. |
| 122 | // // 16-byte header defined (dex::CodeItem). |
| 123 | // uint16_t registers_size; |
| 124 | // uint16_t ins_size; |
| 125 | // uint16_t outs_size; |
| 126 | // uint16_t tries_size; |
| 127 | // uint32_t debug_info_off; |
| 128 | // uint32_t insns_size; |
| 129 | // |
| 130 | // // Variable-length data follow. |
| 131 | // uint16_t insns[insns_size]; // Instruction bytes. |
| 132 | // uint16_t padding[(tries_size > 0 && insns_size % 2 == 1) ? 1 : 0]; |
| 133 | // |
| 134 | // if (tries_size > 0) { |
| 135 | // // 4-byte aligned here. |
| 136 | // struct try_item { // dex::TryItem. |
| 137 | // uint32_t start_addr; |
| 138 | // uint16_t insn_count; |
| 139 | // uint16_t handler_off; |
| 140 | // } tries[tries_size]; |
| 141 | // |
| 142 | // struct encoded_catch_handler_list { |
| 143 | // uleb128 handlers_size; |
| 144 | // struct encoded_catch_handler { |
| 145 | // sleb128 encoded_catch_handler_size; |
| 146 | // struct encoded_type_addr_pair { |
| 147 | // uleb128 type_idx; |
| 148 | // uleb128 addr; |
| 149 | // } handlers[abs(encoded_catch_handler_size)]; |
| 150 | // if (encoded_catch_handler_size <= 0) { |
| 151 | // uleb128 catch_all_addr; |
| 152 | // } |
| 153 | // } handlers_list[handlers_size]; |
| 154 | // } handlers_group; // Confusingly called "handlers" in DEX doc. |
| 155 | // } |
| 156 | // |
| 157 | // // Padding to 4-bytes align next code_item *only if more exist*. |
| 158 | // } |
| 159 | offset_t GetNext() { |
| 160 | // Read header CodeItem. |
| 161 | if (!source_.AlignOn(image_, 4U)) |
| 162 | return kInvalidOffset; |
| 163 | const offset_t code_item_offset = |
| 164 | base::checked_cast<offset_t>(source_.begin() - image_.begin()); |
| 165 | const auto* code_item = source_.GetPointer<const dex::CodeItem>(); |
| 166 | if (!code_item) |
| 167 | return kInvalidOffset; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 168 | DCHECK(Is32BitAligned(code_item_offset)); |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 169 | |
Samuel Huang | d758cb7 | 2018-07-19 14:32:28 +0000 | [diff] [blame] | 170 | // TODO(huangs): Fail if |code_item->insns_size == 0| (Constraint A1). |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 171 | // Skip instruction bytes. |
| 172 | if (!source_.GetArray<uint16_t>(code_item->insns_size)) |
| 173 | return kInvalidOffset; |
| 174 | // Skip padding if present. |
| 175 | if (code_item->tries_size > 0 && !source_.AlignOn(image_, 4U)) |
| 176 | return kInvalidOffset; |
| 177 | |
| 178 | // Skip tries[] and handlers_group to arrive at the next code item. Parsing |
| 179 | // is nontrivial due to use of uleb128 / sleb128. |
| 180 | if (code_item->tries_size > 0) { |
| 181 | // Skip (try_item) tries[]. |
| 182 | if (!source_.GetArray<dex::TryItem>(code_item->tries_size)) |
| 183 | return kInvalidOffset; |
| 184 | |
| 185 | // Skip handlers_group. |
| 186 | uint32_t handlers_size = 0; |
| 187 | if (!source_.GetUleb128(&handlers_size)) |
| 188 | return kInvalidOffset; |
| 189 | // Sanity check to quickly reject excessively large |handlers_size|. |
| 190 | if (source_.Remaining() < static_cast<size_type>(handlers_size)) |
| 191 | return kInvalidOffset; |
| 192 | |
| 193 | // Skip (encoded_catch_handler) handlers_list[]. |
| 194 | for (uint32_t k = 0; k < handlers_size; ++k) { |
| 195 | int32_t encoded_catch_handler_size = 0; |
| 196 | if (!source_.GetSleb128(&encoded_catch_handler_size)) |
| 197 | return kInvalidOffset; |
| 198 | const size_type abs_size = std::abs(encoded_catch_handler_size); |
| 199 | if (source_.Remaining() < abs_size) // Sanity check. |
| 200 | return kInvalidOffset; |
| 201 | // Skip (encoded_type_addr_pair) handlers[]. |
| 202 | for (size_type j = 0; j < abs_size; ++j) { |
| 203 | if (!source_.SkipLeb128() || !source_.SkipLeb128()) |
| 204 | return kInvalidOffset; |
| 205 | } |
| 206 | // Skip catch_all_addr. |
| 207 | if (encoded_catch_handler_size <= 0) { |
| 208 | if (!source_.SkipLeb128()) |
| 209 | return kInvalidOffset; |
| 210 | } |
| 211 | } |
| 212 | } |
| 213 | // Success! |code_item->insns_size| is validated, but its content is still |
| 214 | // considered unsafe and requires validation. |
| 215 | return code_item_offset; |
| 216 | } |
| 217 | |
| 218 | // Given |code_item_offset| that points to the start of a valid code item in |
| 219 | // |image|, returns |insns| bytes as ConstBufferView. |
| 220 | static ConstBufferView GetCodeItemInsns(ConstBufferView image, |
| 221 | offset_t code_item_offset) { |
| 222 | BufferSource source(BufferSource(image).Skip(code_item_offset)); |
| 223 | const auto* code_item = source.GetPointer<const dex::CodeItem>(); |
| 224 | DCHECK(code_item); |
| 225 | BufferRegion insns{0, code_item->insns_size * kInstrUnitSize}; |
| 226 | DCHECK(source.covers(insns)); |
| 227 | return source[insns]; |
| 228 | } |
| 229 | |
| 230 | private: |
| 231 | ConstBufferView image_; |
| 232 | BufferSource source_; |
| 233 | }; |
| 234 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 235 | /******** InstructionParser ********/ |
| 236 | |
| 237 | // A class that successively reads |code_item| for Dalvik instructions, which |
| 238 | // are found at |insns|, spanning |insns_size| uint16_t "units". These units |
| 239 | // store instructions followed by optional non-instruction "payload". Finding |
| 240 | // payload boundary requires parsing: On finding an instruction that uses (and |
| 241 | // points to) payload, the boundary is updated. |
| 242 | class InstructionParser { |
| 243 | public: |
| 244 | struct Value { |
| 245 | offset_t instr_offset; |
| 246 | const dex::Instruction* instr = nullptr; // null for unknown instructions. |
| 247 | }; |
| 248 | |
| 249 | // Returns pointer to DEX Instruction data for |opcode|, or null if |opcode| |
| 250 | // is unknown. An internal initialize-on-first-use table is used for fast |
| 251 | // lookup. |
| 252 | const dex::Instruction* FindDalvikInstruction(uint8_t opcode) { |
| 253 | static bool is_init = false; |
| 254 | static const dex::Instruction* instruction_table[256]; |
| 255 | if (!is_init) { |
| 256 | is_init = true; |
| 257 | std::fill(std::begin(instruction_table), std::end(instruction_table), |
| 258 | nullptr); |
| 259 | for (const dex::Instruction& instr : dex::kByteCode) { |
| 260 | std::fill(instruction_table + instr.opcode, |
| 261 | instruction_table + instr.opcode + instr.variant, &instr); |
| 262 | } |
| 263 | } |
| 264 | return instruction_table[opcode]; |
| 265 | } |
| 266 | |
| 267 | InstructionParser() = default; |
| 268 | |
| 269 | InstructionParser(ConstBufferView image, offset_t base_offset) |
| 270 | : image_begin_(image.begin()), |
| 271 | insns_(CodeItemParser::GetCodeItemInsns(image, base_offset)), |
| 272 | payload_boundary_(insns_.end()) {} |
| 273 | |
| 274 | // Reads the next instruction. On success, makes the data read available via |
| 275 | // value() and returns true. Otherwise (done or found error) returns false. |
| 276 | bool ReadNext() { |
| 277 | // Do not scan past payload boundary. |
| 278 | if (insns_.begin() >= payload_boundary_) |
| 279 | return false; |
| 280 | |
| 281 | const offset_t instr_offset = |
| 282 | base::checked_cast<offset_t>(insns_.begin() - image_begin_); |
| 283 | const uint8_t op = insns_.read<uint8_t>(0); |
| 284 | const dex::Instruction* instr = FindDalvikInstruction(op); |
| 285 | |
| 286 | // Stop on finding unknown instructions. ODEX files might trigger this. |
| 287 | if (!instr) { |
| 288 | LOG(WARNING) << "Unknown Dalvik instruction detected at " |
| 289 | << AsHex<8>(instr_offset) << "."; |
| 290 | return false; |
| 291 | } |
| 292 | |
| 293 | const int instr_length_units = instr->layout; |
| 294 | const size_t instr_length_bytes = instr_length_units * kInstrUnitSize; |
| 295 | if (insns_.size() < instr_length_bytes) |
| 296 | return false; |
| 297 | |
| 298 | // Handle instructions with variable-length data payload (31t). |
| 299 | if (instr->opcode == 0x26 || // fill-array-data |
| 300 | instr->opcode == 0x2B || // packed-switch |
| 301 | instr->opcode == 0x2C) { // sparse-switch |
| 302 | const int32_t unsafe_payload_rel_units = insns_.read<int32_t>(2); |
| 303 | // Payload must be in current code item, after current instruction. |
| 304 | if (unsafe_payload_rel_units < instr_length_units || |
| 305 | static_cast<uint32_t>(unsafe_payload_rel_units) >= |
| 306 | insns_.size() / kInstrUnitSize) { |
| 307 | LOG(WARNING) << "Invalid payload found."; |
| 308 | return false; |
| 309 | } |
| 310 | // Update boundary between instructions and payload. |
| 311 | const ConstBufferView::const_iterator payload_it = |
| 312 | insns_.begin() + unsafe_payload_rel_units * kInstrUnitSize; |
| 313 | payload_boundary_ = std::min(payload_boundary_, payload_it); |
| 314 | } |
| 315 | |
| 316 | insns_.remove_prefix(instr_length_bytes); |
| 317 | value_ = {instr_offset, instr}; |
| 318 | return true; |
| 319 | } |
| 320 | |
| 321 | const Value& value() const { return value_; } |
| 322 | |
| 323 | private: |
| 324 | ConstBufferView::const_iterator image_begin_; |
| 325 | ConstBufferView insns_; |
| 326 | ConstBufferView::const_iterator payload_boundary_; |
| 327 | Value value_; |
| 328 | }; |
| 329 | |
| 330 | /******** InstructionReferenceReader ********/ |
| 331 | |
| 332 | // A class to visit |code_items|, parse instructions, and emit embedded |
| 333 | // References of a type determined by |filter_| and |mapper_|. Only References |
| 334 | // located in |[lo, hi)| are emitted. |lo| and |hi| are assumed to never |
| 335 | // straddle the body of a Reference. |
| 336 | class InstructionReferenceReader : public ReferenceReader { |
| 337 | public: |
| 338 | // A function that takes a parsed Dalvik instruction and decides whether it |
| 339 | // contains a specific type of Reference. If true, then returns the Reference |
| 340 | // location. Otherwise returns kInvalidOffset. |
| 341 | using Filter = |
| 342 | base::RepeatingCallback<offset_t(const InstructionParser::Value&)>; |
| 343 | // A function that takes Reference location from |filter_| to extract the |
| 344 | // stored target. If valid, returns it. Otherwise returns kInvalidOffset. |
| 345 | using Mapper = base::RepeatingCallback<offset_t(offset_t)>; |
| 346 | |
| 347 | InstructionReferenceReader(ConstBufferView image, |
| 348 | offset_t lo, |
| 349 | offset_t hi, |
| 350 | const std::vector<offset_t>& code_item_offsets, |
| 351 | Filter&& filter, |
| 352 | Mapper&& mapper) |
| 353 | : image_(image), |
| 354 | lo_(lo), |
| 355 | hi_(hi), |
| 356 | end_it_(code_item_offsets.end()), |
| 357 | filter_(std::move(filter)), |
| 358 | mapper_(std::move(mapper)) { |
| 359 | const auto begin_it = code_item_offsets.begin(); |
| 360 | // Use binary search to find the code item that contains |lo_|. |
| 361 | auto comp = [](offset_t test_offset, offset_t code_item_offset) { |
| 362 | return test_offset < code_item_offset; |
| 363 | }; |
| 364 | cur_it_ = std::upper_bound(begin_it, end_it_, lo_, comp); |
| 365 | if (cur_it_ != begin_it) |
| 366 | --cur_it_; |
| 367 | parser_ = InstructionParser(image_, *cur_it_); |
| 368 | } |
| 369 | |
| 370 | // ReferenceReader: |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 371 | absl::optional<Reference> GetNext() override { |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 372 | while (true) { |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 373 | while (parser_.ReadNext()) { |
| 374 | const auto& v = parser_.value(); |
| 375 | DCHECK_NE(v.instr, nullptr); |
| 376 | if (v.instr_offset >= hi_) |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 377 | return absl::nullopt; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 378 | const offset_t location = filter_.Run(v); |
| 379 | if (location == kInvalidOffset || location < lo_) |
| 380 | continue; |
| 381 | // The general check is |location + reference_width > hi_|. However, by |
| 382 | // assumption |hi_| and |lo_| do not straddle the body of a Reference. |
| 383 | // So |reference_width| is unneeded. |
| 384 | if (location >= hi_) |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 385 | return absl::nullopt; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 386 | offset_t target = mapper_.Run(location); |
| 387 | if (target != kInvalidOffset) |
| 388 | return Reference{location, target}; |
| 389 | else |
| 390 | LOG(WARNING) << "Invalid target at " << AsHex<8>(location) << "."; |
| 391 | } |
| 392 | ++cur_it_; |
| 393 | if (cur_it_ == end_it_) |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 394 | return absl::nullopt; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 395 | parser_ = InstructionParser(image_, *cur_it_); |
| 396 | } |
| 397 | } |
| 398 | |
| 399 | private: |
| 400 | const ConstBufferView image_; |
| 401 | const offset_t lo_; |
| 402 | const offset_t hi_; |
| 403 | const std::vector<offset_t>::const_iterator end_it_; |
| 404 | const Filter filter_; |
| 405 | const Mapper mapper_; |
| 406 | std::vector<offset_t>::const_iterator cur_it_; |
| 407 | InstructionParser parser_; |
| 408 | }; |
| 409 | |
| 410 | /******** ItemReferenceReader ********/ |
| 411 | |
| 412 | // A class to visit fixed-size item elements (determined by |item_size|) and |
| 413 | // emit a "member variable of interest" (MVI, determined by |rel_location| and |
| 414 | // |mapper|) as Reference. Only MVIs lying in |[lo, hi)| are emitted. |lo| and |
| 415 | // |hi| are assumed to never straddle the body of a Reference. |
| 416 | class ItemReferenceReader : public ReferenceReader { |
| 417 | public: |
| 418 | // A function that takes an MVI's location and emit its target offset. |
| 419 | using Mapper = base::RepeatingCallback<offset_t(offset_t)>; |
| 420 | |
| 421 | // |item_size| is the size of a fixed-size item. |rel_location| is the |
| 422 | // relative location of MVI from the start of the item containing it. |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 423 | // |rel_item_offset| is the offset to use relative to |item_offset| in cases |
| 424 | // where a value other than |rel_location| is required. For an example of this |
| 425 | // see ReadMethodHandleFieldOrMethodId. |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 426 | ItemReferenceReader(offset_t lo, |
| 427 | offset_t hi, |
| 428 | const dex::MapItem& map_item, |
| 429 | size_t item_size, |
| 430 | size_t rel_location, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 431 | Mapper&& mapper, |
| 432 | bool mapper_wants_item = false) |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 433 | : hi_(hi), |
| 434 | item_base_offset_(base::checked_cast<offset_t>(map_item.offset)), |
| 435 | num_items_(base::checked_cast<uint32_t>(map_item.size)), |
| 436 | item_size_(base::checked_cast<uint32_t>(item_size)), |
| 437 | rel_location_(base::checked_cast<uint32_t>(rel_location)), |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 438 | mapper_input_delta_( |
| 439 | mapper_wants_item ? 0 : base::checked_cast<uint32_t>(rel_location)), |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 440 | mapper_(std::move(mapper)) { |
| 441 | static_assert(sizeof(decltype(map_item.offset)) <= sizeof(offset_t), |
| 442 | "map_item.offset too large."); |
| 443 | static_assert(sizeof(decltype(map_item.size)) <= sizeof(offset_t), |
| 444 | "map_item.size too large."); |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 445 | if (!item_base_offset_) { |
| 446 | // Empty item: Assign |cur_idx| to |num_items_| to skip everything. |
| 447 | cur_idx_ = num_items_; |
| 448 | } else if (lo < item_base_offset_) { |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 449 | cur_idx_ = 0; |
| 450 | } else if (lo < OffsetOfIndex(num_items_)) { |
| 451 | cur_idx_ = (lo - item_base_offset_) / item_size_; |
| 452 | // Fine-tune: Advance if |lo| lies beyond the MVI. |
| 453 | if (lo > OffsetOfIndex(cur_idx_) + rel_location_) |
| 454 | ++cur_idx_; |
| 455 | } else { |
| 456 | cur_idx_ = num_items_; |
| 457 | } |
| 458 | } |
| 459 | |
| 460 | // ReferenceReader: |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 461 | absl::optional<Reference> GetNext() override { |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 462 | while (cur_idx_ < num_items_) { |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 463 | const offset_t item_offset = OffsetOfIndex(cur_idx_); |
| 464 | const offset_t location = item_offset + rel_location_; |
| 465 | // The general check is |location + reference_width > hi_|. However, by |
| 466 | // assumption |hi_| and |lo_| do not straddle the body of a Reference. So |
| 467 | // |reference_width| is unneeded. |
| 468 | if (location >= hi_) |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 469 | break; |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 470 | |
| 471 | // |location == item_offset + mapper_input_delta_| in the majority of |
| 472 | // cases. The exception is when |mapper_| wants an item aligned location |
| 473 | // instead e.g. ReadMethodHandleFieldOrMethodId. |
| 474 | const offset_t target = mapper_.Run(item_offset + mapper_input_delta_); |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 475 | |
| 476 | // kDexSentinelOffset (0) may appear for the following: |
| 477 | // - ProtoIdItem: parameters_off. |
| 478 | // - ClassDefItem: interfaces_off, annotations_off, class_data_off, |
| 479 | // static_values_off. |
| 480 | // - AnnotationsDirectoryItem: class_annotations_off. |
| 481 | // - AnnotationSetRefItem: annotations_off. |
| 482 | // kDexSentinelIndexAsOffset (0xFFFFFFFF) may appear for the following: |
| 483 | // - ClassDefItem: superclass_idx, source_file_idx. |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 484 | // - MethodHandleItem: |mapper_| uses ReadMethodHandleFieldOrMethodId and |
| 485 | // determines the item at |cur_idx_| is not of the required reference |
| 486 | // type. |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 487 | if (target == kDexSentinelOffset || target == kDexSentinelIndexAsOffset) { |
| 488 | ++cur_idx_; |
| 489 | continue; |
| 490 | } |
| 491 | |
| 492 | if (target == kInvalidOffset) { |
| 493 | LOG(WARNING) << "Invalid item target at " << AsHex<8>(location) << "."; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 494 | break; |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 495 | } |
| 496 | ++cur_idx_; |
| 497 | return Reference{location, target}; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 498 | } |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 499 | return absl::nullopt; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 500 | } |
| 501 | |
| 502 | private: |
| 503 | offset_t OffsetOfIndex(uint32_t idx) { |
| 504 | return base::checked_cast<uint32_t>(item_base_offset_ + idx * item_size_); |
| 505 | } |
| 506 | |
| 507 | const offset_t hi_; |
| 508 | const offset_t item_base_offset_; |
| 509 | const uint32_t num_items_; |
| 510 | const uint32_t item_size_; |
| 511 | const uint32_t rel_location_; |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 512 | const uint32_t mapper_input_delta_; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 513 | const Mapper mapper_; |
| 514 | offset_t cur_idx_ = 0; |
| 515 | }; |
| 516 | |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 517 | // Parses a flattened jagged list of lists of items that looks like: |
| 518 | // NTTT|NTT|NTTTT|N|NTT... |
| 519 | // where |N| is an uint32_t representing the number of items in each sub-list, |
| 520 | // and "T" is a fixed-size item (|item_width|) of type "T". On success, stores |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 521 | // the offset of each |T| into |item_offsets|, and returns true. Otherwise |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 522 | // (e.g., on finding any structural problem) returns false. |
| 523 | bool ParseItemOffsets(ConstBufferView image, |
| 524 | const dex::MapItem& map_item, |
| 525 | size_t item_width, |
| 526 | std::vector<offset_t>* item_offsets) { |
| 527 | // Sanity check: |image| should at least fit |map_item.size| copies of "N". |
| 528 | if (!image.covers_array(map_item.offset, map_item.size, sizeof(uint32_t))) |
| 529 | return false; |
| 530 | BufferSource source = std::move(BufferSource(image).Skip(map_item.offset)); |
| 531 | item_offsets->clear(); |
| 532 | for (uint32_t i = 0; i < map_item.size; ++i) { |
| 533 | if (!source.AlignOn(image, 4U)) |
| 534 | return false; |
| 535 | uint32_t unsafe_size; |
| 536 | if (!source.GetValue<uint32_t>(&unsafe_size)) |
| 537 | return false; |
| 538 | DCHECK(Is32BitAligned( |
| 539 | base::checked_cast<offset_t>(source.begin() - image.begin()))); |
| 540 | if (!source.covers_array(0, unsafe_size, item_width)) |
| 541 | return false; |
| 542 | for (uint32_t j = 0; j < unsafe_size; ++j) { |
| 543 | item_offsets->push_back( |
| 544 | base::checked_cast<offset_t>(source.begin() - image.begin())); |
| 545 | source.Skip(item_width); |
| 546 | } |
| 547 | } |
| 548 | return true; |
| 549 | } |
| 550 | |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 551 | // Parses AnnotationDirectoryItems of the format (using RegEx) "(AF*M*P*)*", |
| 552 | // where: |
| 553 | // A = AnnotationsDirectoryItem (contains class annotation), |
| 554 | // F = FieldAnnotation, |
| 555 | // M = MethodAnnotation, |
| 556 | // P = ParameterAnnotation. |
| 557 | // On success, stores the offsets of each class, field, method and parameter |
| 558 | // annotation for each item into |*_annotation_offsets|. Otherwise on finding |
| 559 | // structural issues returns false. |
| 560 | bool ParseAnnotationsDirectoryItems( |
| 561 | ConstBufferView image, |
| 562 | const dex::MapItem& annotations_directory_map_item, |
| 563 | std::vector<offset_t>* annotations_directory_item_offsets, |
| 564 | std::vector<offset_t>* field_annotation_offsets, |
| 565 | std::vector<offset_t>* method_annotation_offsets, |
| 566 | std::vector<offset_t>* parameter_annotation_offsets) { |
| 567 | // Sanity check: |image| should at least fit |
| 568 | // |annotations_directory_map_item.size| copies of "A". |
| 569 | if (!image.covers_array(annotations_directory_map_item.offset, |
| 570 | annotations_directory_map_item.size, |
| 571 | sizeof(dex::AnnotationsDirectoryItem))) { |
| 572 | return false; |
| 573 | } |
| 574 | BufferSource source = std::move( |
| 575 | BufferSource(image).Skip(annotations_directory_map_item.offset)); |
| 576 | annotations_directory_item_offsets->clear(); |
| 577 | field_annotation_offsets->clear(); |
| 578 | method_annotation_offsets->clear(); |
| 579 | parameter_annotation_offsets->clear(); |
| 580 | |
| 581 | // Helper to process sublists. |
| 582 | auto parse_list = [&source, image](uint32_t unsafe_size, size_t item_width, |
| 583 | std::vector<offset_t>* item_offsets) { |
| 584 | DCHECK(Is32BitAligned( |
| 585 | base::checked_cast<offset_t>(source.begin() - image.begin()))); |
| 586 | if (!source.covers_array(0, unsafe_size, item_width)) |
| 587 | return false; |
| 588 | item_offsets->reserve(item_offsets->size() + unsafe_size); |
| 589 | for (uint32_t i = 0; i < unsafe_size; ++i) { |
| 590 | item_offsets->push_back( |
| 591 | base::checked_cast<offset_t>(source.begin() - image.begin())); |
| 592 | source.Skip(item_width); |
| 593 | } |
| 594 | return true; |
| 595 | }; |
| 596 | |
| 597 | annotations_directory_item_offsets->reserve( |
| 598 | annotations_directory_map_item.size); |
| 599 | for (uint32_t i = 0; i < annotations_directory_map_item.size; ++i) { |
| 600 | if (!source.AlignOn(image, 4U)) |
| 601 | return false; |
| 602 | // Parse header. |
| 603 | annotations_directory_item_offsets->push_back( |
| 604 | base::checked_cast<offset_t>(source.begin() - image.begin())); |
| 605 | dex::AnnotationsDirectoryItem unsafe_annotations_directory_item; |
| 606 | if (!source.GetValue(&unsafe_annotations_directory_item)) |
| 607 | return false; |
| 608 | // Parse sublists. |
| 609 | if (!(parse_list(unsafe_annotations_directory_item.fields_size, |
| 610 | sizeof(dex::FieldAnnotation), field_annotation_offsets) && |
| 611 | parse_list(unsafe_annotations_directory_item.annotated_methods_size, |
| 612 | sizeof(dex::MethodAnnotation), |
| 613 | method_annotation_offsets) && |
| 614 | parse_list( |
| 615 | unsafe_annotations_directory_item.annotated_parameters_size, |
| 616 | sizeof(dex::ParameterAnnotation), |
| 617 | parameter_annotation_offsets))) { |
| 618 | return false; |
| 619 | } |
| 620 | } |
| 621 | return true; |
| 622 | } |
| 623 | |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 624 | /******** CachedItemListReferenceReader ********/ |
| 625 | |
| 626 | // A class that takes sorted |item_offsets|, and emits all member variable of |
| 627 | // interest (MVIs) that fall inside |[lo, hi)|. The MVI of each item has |
| 628 | // location of |rel_location| from item offset, and has target extracted with |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 629 | // |mapper| (which performs validation). By the "atomicity assumption", |
| 630 | // [|lo, hi)| never cut across an MVI. |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 631 | class CachedItemListReferenceReader : public ReferenceReader { |
| 632 | public: |
| 633 | // A function that takes an MVI's location and emit its target offset. |
| 634 | using Mapper = base::RepeatingCallback<offset_t(offset_t)>; |
| 635 | |
| 636 | CachedItemListReferenceReader(offset_t lo, |
| 637 | offset_t hi, |
| 638 | uint32_t rel_location, |
| 639 | const std::vector<offset_t>& item_offsets, |
| 640 | Mapper&& mapper) |
| 641 | : hi_(hi), |
| 642 | rel_location_(rel_location), |
| 643 | end_it_(item_offsets.cend()), |
| 644 | mapper_(mapper) { |
| 645 | cur_it_ = std::upper_bound(item_offsets.cbegin(), item_offsets.cend(), lo); |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 646 | // Adding |rel_location_| is necessary as references can be offset from the |
| 647 | // start of the item. |
| 648 | if (cur_it_ != item_offsets.begin() && *(cur_it_ - 1) + rel_location_ >= lo) |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 649 | --cur_it_; |
| 650 | } |
Samuel Huang | f137bf4 | 2021-08-13 15:42:26 +0000 | [diff] [blame] | 651 | CachedItemListReferenceReader(const CachedItemListReferenceReader&) = delete; |
| 652 | const CachedItemListReferenceReader& operator=( |
| 653 | const CachedItemListReferenceReader&) = delete; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 654 | |
| 655 | // ReferenceReader: |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 656 | absl::optional<Reference> GetNext() override { |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 657 | while (cur_it_ < end_it_) { |
| 658 | const offset_t location = *cur_it_ + rel_location_; |
| 659 | if (location >= hi_) // Check is simplified by atomicity assumption. |
| 660 | break; |
| 661 | const offset_t target = mapper_.Run(location); |
| 662 | if (target == kInvalidOffset) { |
| 663 | LOG(WARNING) << "Invalid item target at " << AsHex<8>(location) << "."; |
| 664 | break; |
| 665 | } |
| 666 | ++cur_it_; |
| 667 | |
| 668 | // kDexSentinelOffset is a sentinel for; |
| 669 | // - AnnotationsDirectoryItem: class_annotations_off |
| 670 | if (target == kDexSentinelOffset) |
| 671 | continue; |
| 672 | return Reference{location, target}; |
| 673 | } |
Anton Bikineev | 1a96551 | 2021-05-15 22:35:36 +0000 | [diff] [blame] | 674 | return absl::nullopt; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 675 | } |
| 676 | |
| 677 | private: |
| 678 | const offset_t hi_; |
| 679 | const uint32_t rel_location_; |
| 680 | const std::vector<offset_t>::const_iterator end_it_; |
| 681 | const Mapper mapper_; |
| 682 | std::vector<offset_t>::const_iterator cur_it_; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 683 | }; |
| 684 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 685 | // Reads an INT index at |location| in |image| and translates the index to the |
| 686 | // offset of a fixed-size item specified by |target_map_item| and |
| 687 | // |target_item_size|. Returns the target offset if valid, or kInvalidOffset |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 688 | // otherwise. This is compatible with |
| 689 | // CachedReferenceListReferenceReader::Mapper, |
| 690 | // InstructionReferenceReader::Mapper, and ItemReferenceReader::Mapper. |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 691 | template <typename INT> |
| 692 | static offset_t ReadTargetIndex(ConstBufferView image, |
| 693 | const dex::MapItem& target_map_item, |
| 694 | size_t target_item_size, |
| 695 | offset_t location) { |
| 696 | static_assert(sizeof(INT) <= sizeof(offset_t), |
| 697 | "INT may not fit into offset_t."); |
| 698 | const offset_t unsafe_idx = image.read<INT>(location); |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 699 | // kDexSentinalIndexAsOffset (0xFFFFFFFF) is a sentinel for |
| 700 | // - ClassDefItem: superclass_idx, source_file_idx. |
| 701 | if (unsafe_idx == kDexSentinelIndexAsOffset) |
| 702 | return unsafe_idx; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 703 | if (unsafe_idx >= target_map_item.size) |
| 704 | return kInvalidOffset; |
| 705 | return target_map_item.offset + |
| 706 | base::checked_cast<offset_t>(unsafe_idx * target_item_size); |
| 707 | } |
| 708 | |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 709 | // Reads a field or method index of the MethodHandleItem located at |location| |
| 710 | // in |image| and translates |method_handle_item.field_or_method_id| to the |
| 711 | // offset of a fixed-size item specified by |target_map_item| and |
| 712 | // |target_item_size|. The index is deemed to be of the correct target type if |
| 713 | // |method_handle_item.method_handle_type| falls within the range [|min_type|, |
| 714 | // |max_type|]. If the target type is correct ReadTargetIndex is called. |
| 715 | // Returns the target offset if valid, or kDexSentinelIndexAsOffset if |
| 716 | // |method_handle_item.method_handle_type| is of the wrong type, or |
| 717 | // kInvalidOffset otherwise. |
| 718 | // |
| 719 | // As of DEX version 39 MethodHandleType values for FieldId and MethodId each |
| 720 | // form one consecutive block of values. If this changes, then the interface to |
| 721 | // this function will need to be redesigned. |
| 722 | static offset_t ReadMethodHandleFieldOrMethodId( |
| 723 | ConstBufferView image, |
| 724 | const dex::MapItem& target_map_item, |
| 725 | size_t target_item_size, |
| 726 | dex::MethodHandleType min_type, |
| 727 | dex::MethodHandleType max_type, |
| 728 | offset_t location) { |
| 729 | dex::MethodHandleItem method_handle_item = |
| 730 | image.read<dex::MethodHandleItem>(location); |
| 731 | |
| 732 | // Cannot use base::checked_cast as dex::MethodHandleType is an enum class so |
| 733 | // static_assert on the size instead. |
| 734 | static_assert(sizeof(decltype(dex::MethodHandleItem::method_handle_type)) <= |
| 735 | sizeof(dex::MethodHandleType), |
| 736 | "dex::MethodHandleItem::method_handle_type may not fit into " |
| 737 | "dex::MethodHandleType."); |
| 738 | dex::MethodHandleType method_handle_type = |
| 739 | static_cast<dex::MethodHandleType>(method_handle_item.method_handle_type); |
| 740 | |
| 741 | if (method_handle_type >= dex::MethodHandleType::kMaxMethodHandleType) { |
| 742 | return kInvalidOffset; |
| 743 | } |
| 744 | |
| 745 | // Use DexSentinelIndexAsOffset to skip the item as it isn't of the |
| 746 | // corresponding method handle type. |
| 747 | if (method_handle_type < min_type || method_handle_type > max_type) { |
| 748 | return kDexSentinelIndexAsOffset; |
| 749 | } |
| 750 | |
| 751 | return ReadTargetIndex<decltype(dex::MethodHandleItem::field_or_method_id)>( |
| 752 | image, target_map_item, target_item_size, |
| 753 | location + offsetof(dex::MethodHandleItem, field_or_method_id)); |
| 754 | } |
| 755 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 756 | // Reads uint32_t value in |image| at (valid) |location| and checks whether it |
| 757 | // is a safe offset of a fixed-size item. Returns the target offset (possibly a |
| 758 | // sentinel) if valid, or kInvalidOffset otherwise. This is compatible with |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 759 | // CachedReferenceListReferenceReader::Mapper, |
| 760 | // InstructionReferenceReader::Mapper, and ItemReferenceReader::Mapper. |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 761 | static offset_t ReadTargetOffset32(ConstBufferView image, offset_t location) { |
| 762 | const offset_t unsafe_target = |
| 763 | static_cast<offset_t>(image.read<uint32_t>(location)); |
| 764 | // Skip and don't validate kDexSentinelOffset as it is indicative of an |
| 765 | // empty reference. |
| 766 | if (unsafe_target == kDexSentinelOffset) |
| 767 | return unsafe_target; |
| 768 | |
| 769 | // TODO(huangs): Check that |unsafe_target| is within the correct data |
| 770 | // section. |
| 771 | if (unsafe_target >= image.size()) |
| 772 | return kInvalidOffset; |
| 773 | return unsafe_target; |
| 774 | } |
| 775 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 776 | /******** ReferenceWriterAdaptor ********/ |
| 777 | |
| 778 | // A ReferenceWriter that adapts a callback that performs type-specific |
| 779 | // Reference writes. |
| 780 | class ReferenceWriterAdaptor : public ReferenceWriter { |
| 781 | public: |
| 782 | using Writer = base::RepeatingCallback<void(Reference, MutableBufferView)>; |
| 783 | |
| 784 | ReferenceWriterAdaptor(MutableBufferView image, Writer&& writer) |
| 785 | : image_(image), writer_(std::move(writer)) {} |
| 786 | |
| 787 | // ReferenceWriter: |
| 788 | void PutNext(Reference ref) override { writer_.Run(ref, image_); } |
| 789 | |
| 790 | private: |
| 791 | MutableBufferView image_; |
| 792 | Writer writer_; |
| 793 | }; |
| 794 | |
| 795 | // Helper that's compatible with ReferenceWriterAdaptor::Writer. |
| 796 | // Given that |ref.target| points to the start of a fixed size DEX item (e.g., |
| 797 | // FieldIdItem), translates |ref.target| to item index, and writes the result to |
| 798 | // |ref.location| as |INT|. |
| 799 | template <typename INT> |
| 800 | static void WriteTargetIndex(const dex::MapItem& target_map_item, |
| 801 | size_t target_item_size, |
| 802 | Reference ref, |
| 803 | MutableBufferView image) { |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 804 | const size_t unsafe_idx = |
| 805 | (ref.target - target_map_item.offset) / target_item_size; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 806 | // Verify that index is within bound. |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 807 | if (unsafe_idx >= target_map_item.size) { |
| 808 | LOG(ERROR) << "Target index out of bounds at: " << AsHex<8>(ref.location) |
| 809 | << "."; |
| 810 | return; |
| 811 | } |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 812 | // Verify that |ref.target| points to start of item. |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 813 | DCHECK_EQ(ref.target, target_map_item.offset + unsafe_idx * target_item_size); |
| 814 | image.write<INT>(ref.location, base::checked_cast<INT>(unsafe_idx)); |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 815 | } |
| 816 | |
| 817 | // Buffer for ReadDexHeader() to optionally return results. |
| 818 | struct ReadDexHeaderResults { |
| 819 | BufferSource source; |
| 820 | const dex::HeaderItem* header; |
| 821 | int dex_version; |
| 822 | }; |
| 823 | |
| 824 | // Returns whether |image| points to a DEX file. If this is a possibility and |
| 825 | // |opt_results| is not null, then uses it to pass extracted data to enable |
| 826 | // further parsing. |
| 827 | bool ReadDexHeader(ConstBufferView image, ReadDexHeaderResults* opt_results) { |
| 828 | // This part needs to be fairly efficient since it may be called many times. |
| 829 | BufferSource source(image); |
| 830 | const dex::HeaderItem* header = source.GetPointer<dex::HeaderItem>(); |
| 831 | if (!header) |
| 832 | return false; |
| 833 | if (header->magic[0] != 'd' || header->magic[1] != 'e' || |
| 834 | header->magic[2] != 'x' || header->magic[3] != '\n' || |
| 835 | header->magic[7] != '\0') { |
| 836 | return false; |
| 837 | } |
| 838 | |
| 839 | // Magic matches: More detailed tests can be conducted. |
| 840 | int dex_version = 0; |
| 841 | for (int i = 4; i < 7; ++i) { |
| 842 | if (!isdigit(header->magic[i])) |
| 843 | return false; |
| 844 | dex_version = dex_version * 10 + (header->magic[i] - '0'); |
| 845 | } |
| 846 | |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 847 | // Only support DEX versions 35, 37, 38, and 39 |
| 848 | if (dex_version != 35 && dex_version != 37 && dex_version != 38 && |
| 849 | dex_version != 39) { |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 850 | return false; |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 851 | } |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 852 | |
| 853 | if (header->file_size > image.size() || |
| 854 | header->file_size < sizeof(dex::HeaderItem) || |
| 855 | header->map_off < sizeof(dex::HeaderItem)) { |
| 856 | return false; |
| 857 | } |
| 858 | |
| 859 | if (opt_results) |
| 860 | *opt_results = {source, header, dex_version}; |
| 861 | return true; |
| 862 | } |
| 863 | |
| 864 | } // namespace |
| 865 | |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 866 | /******** DisassemblerDex ********/ |
| 867 | |
| 868 | DisassemblerDex::DisassemblerDex() : Disassembler(4) {} |
| 869 | |
| 870 | DisassemblerDex::~DisassemblerDex() = default; |
| 871 | |
| 872 | // static. |
| 873 | bool DisassemblerDex::QuickDetect(ConstBufferView image) { |
| 874 | return ReadDexHeader(image, nullptr); |
| 875 | } |
| 876 | |
| 877 | ExecutableType DisassemblerDex::GetExeType() const { |
| 878 | return kExeTypeDex; |
| 879 | } |
| 880 | |
| 881 | std::string DisassemblerDex::GetExeTypeString() const { |
| 882 | return base::StringPrintf("DEX (version %d)", dex_version_); |
| 883 | } |
| 884 | |
| 885 | std::vector<ReferenceGroup> DisassemblerDex::MakeReferenceGroups() const { |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 886 | // Must follow DisassemblerDex::ReferenceType order. Initialized on first use. |
| 887 | return { |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 888 | {{4, TypeTag(kTypeIdToDescriptorStringId), PoolTag(kStringId)}, |
| 889 | &DisassemblerDex::MakeReadTypeIdToDescriptorStringId32, |
| 890 | &DisassemblerDex::MakeWriteStringId32}, |
| 891 | {{4, TypeTag(kProtoIdToShortyStringId), PoolTag(kStringId)}, |
| 892 | &DisassemblerDex::MakeReadProtoIdToShortyStringId32, |
| 893 | &DisassemblerDex::MakeWriteStringId32}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 894 | {{4, TypeTag(kFieldIdToNameStringId), PoolTag(kStringId)}, |
| 895 | &DisassemblerDex::MakeReadFieldToNameStringId32, |
| 896 | &DisassemblerDex::MakeWriteStringId32}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 897 | {{4, TypeTag(kMethodIdToNameStringId), PoolTag(kStringId)}, |
| 898 | &DisassemblerDex::MakeReadMethodIdToNameStringId32, |
| 899 | &DisassemblerDex::MakeWriteStringId32}, |
| 900 | {{4, TypeTag(kClassDefToSourceFileStringId), PoolTag(kStringId)}, |
| 901 | &DisassemblerDex::MakeReadClassDefToSourceFileStringId32, |
| 902 | &DisassemblerDex::MakeWriteStringId32}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 903 | {{2, TypeTag(kCodeToStringId16), PoolTag(kStringId)}, |
| 904 | &DisassemblerDex::MakeReadCodeToStringId16, |
| 905 | &DisassemblerDex::MakeWriteStringId16}, |
| 906 | {{4, TypeTag(kCodeToStringId32), PoolTag(kStringId)}, |
| 907 | &DisassemblerDex::MakeReadCodeToStringId32, |
| 908 | &DisassemblerDex::MakeWriteStringId32}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 909 | {{4, TypeTag(kProtoIdToReturnTypeId), PoolTag(kTypeId)}, |
| 910 | &DisassemblerDex::MakeReadProtoIdToReturnTypeId32, |
| 911 | &DisassemblerDex::MakeWriteTypeId32}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 912 | {{2, TypeTag(kFieldIdToClassTypeId), PoolTag(kTypeId)}, |
| 913 | &DisassemblerDex::MakeReadFieldToClassTypeId16, |
| 914 | &DisassemblerDex::MakeWriteTypeId16}, |
| 915 | {{2, TypeTag(kFieldIdToTypeId), PoolTag(kTypeId)}, |
| 916 | &DisassemblerDex::MakeReadFieldToTypeId16, |
| 917 | &DisassemblerDex::MakeWriteTypeId16}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 918 | {{2, TypeTag(kMethodIdToClassTypeId), PoolTag(kTypeId)}, |
| 919 | &DisassemblerDex::MakeReadMethodIdToClassTypeId16, |
| 920 | &DisassemblerDex::MakeWriteTypeId16}, |
| 921 | {{4, TypeTag(kClassDefToClassTypeId), PoolTag(kTypeId)}, |
| 922 | &DisassemblerDex::MakeReadClassDefToClassTypeId32, |
| 923 | &DisassemblerDex::MakeWriteTypeId32}, |
| 924 | {{4, TypeTag(kClassDefToSuperClassTypeId), PoolTag(kTypeId)}, |
| 925 | &DisassemblerDex::MakeReadClassDefToSuperClassTypeId32, |
| 926 | &DisassemblerDex::MakeWriteTypeId32}, |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 927 | {{2, TypeTag(kTypeListToTypeId), PoolTag(kTypeId)}, |
| 928 | &DisassemblerDex::MakeReadTypeListToTypeId16, |
| 929 | &DisassemblerDex::MakeWriteTypeId16}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 930 | {{2, TypeTag(kCodeToTypeId), PoolTag(kTypeId)}, |
| 931 | &DisassemblerDex::MakeReadCodeToTypeId16, |
| 932 | &DisassemblerDex::MakeWriteTypeId16}, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 933 | {{2, TypeTag(kCodeToProtoId), PoolTag(kProtoId)}, |
| 934 | &DisassemblerDex::MakeReadCodeToProtoId16, |
| 935 | &DisassemblerDex::MakeWriteProtoId16}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 936 | {{2, TypeTag(kMethodIdToProtoId), PoolTag(kProtoId)}, |
| 937 | &DisassemblerDex::MakeReadMethodIdToProtoId16, |
| 938 | &DisassemblerDex::MakeWriteProtoId16}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 939 | {{2, TypeTag(kCodeToFieldId), PoolTag(kFieldId)}, |
| 940 | &DisassemblerDex::MakeReadCodeToFieldId16, |
| 941 | &DisassemblerDex::MakeWriteFieldId16}, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 942 | {{2, TypeTag(kMethodHandleToFieldId), PoolTag(kFieldId)}, |
| 943 | &DisassemblerDex::MakeReadMethodHandleToFieldId16, |
| 944 | &DisassemblerDex::MakeWriteFieldId16}, |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 945 | {{4, TypeTag(kAnnotationsDirectoryToFieldId), PoolTag(kFieldId)}, |
| 946 | &DisassemblerDex::MakeReadAnnotationsDirectoryToFieldId32, |
| 947 | &DisassemblerDex::MakeWriteFieldId32}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 948 | {{2, TypeTag(kCodeToMethodId), PoolTag(kMethodId)}, |
| 949 | &DisassemblerDex::MakeReadCodeToMethodId16, |
| 950 | &DisassemblerDex::MakeWriteMethodId16}, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 951 | {{2, TypeTag(kMethodHandleToMethodId), PoolTag(kMethodId)}, |
| 952 | &DisassemblerDex::MakeReadMethodHandleToMethodId16, |
| 953 | &DisassemblerDex::MakeWriteMethodId16}, |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 954 | {{4, TypeTag(kAnnotationsDirectoryToMethodId), PoolTag(kMethodId)}, |
| 955 | &DisassemblerDex::MakeReadAnnotationsDirectoryToMethodId32, |
| 956 | &DisassemblerDex::MakeWriteMethodId32}, |
| 957 | {{4, TypeTag(kAnnotationsDirectoryToParameterMethodId), |
| 958 | PoolTag(kMethodId)}, |
| 959 | &DisassemblerDex::MakeReadAnnotationsDirectoryToParameterMethodId32, |
| 960 | &DisassemblerDex::MakeWriteMethodId32}, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 961 | {{2, TypeTag(kCodeToCallSiteId), PoolTag(kCallSiteId)}, |
| 962 | &DisassemblerDex::MakeReadCodeToCallSiteId16, |
| 963 | &DisassemblerDex::MakeWriteCallSiteId16}, |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 964 | {{2, TypeTag(kCodeToMethodHandle), PoolTag(kMethodHandle)}, |
| 965 | &DisassemblerDex::MakeReadCodeToMethodHandle16, |
| 966 | &DisassemblerDex::MakeWriteMethodHandle16}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 967 | {{4, TypeTag(kProtoIdToParametersTypeList), PoolTag(kTypeList)}, |
| 968 | &DisassemblerDex::MakeReadProtoIdToParametersTypeList, |
| 969 | &DisassemblerDex::MakeWriteAbs32}, |
| 970 | {{4, TypeTag(kClassDefToInterfacesTypeList), PoolTag(kTypeList)}, |
| 971 | &DisassemblerDex::MakeReadClassDefToInterfacesTypeList, |
| 972 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 973 | {{4, TypeTag(kAnnotationsDirectoryToParameterAnnotationSetRef), |
| 974 | PoolTag(kAnnotationSetRefList)}, |
| 975 | &DisassemblerDex:: |
| 976 | MakeReadAnnotationsDirectoryToParameterAnnotationSetRef, |
| 977 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 978 | {{4, TypeTag(kAnnotationSetRefListToAnnotationSet), |
| 979 | PoolTag(kAnnotionSet)}, |
| 980 | &DisassemblerDex::MakeReadAnnotationSetRefListToAnnotationSet, |
| 981 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 982 | {{4, TypeTag(kAnnotationsDirectoryToClassAnnotationSet), |
| 983 | PoolTag(kAnnotionSet)}, |
| 984 | &DisassemblerDex::MakeReadAnnotationsDirectoryToClassAnnotationSet, |
| 985 | &DisassemblerDex::MakeWriteAbs32}, |
| 986 | {{4, TypeTag(kAnnotationsDirectoryToFieldAnnotationSet), |
| 987 | PoolTag(kAnnotionSet)}, |
| 988 | &DisassemblerDex::MakeReadAnnotationsDirectoryToFieldAnnotationSet, |
| 989 | &DisassemblerDex::MakeWriteAbs32}, |
| 990 | {{4, TypeTag(kAnnotationsDirectoryToMethodAnnotationSet), |
| 991 | PoolTag(kAnnotionSet)}, |
| 992 | &DisassemblerDex::MakeReadAnnotationsDirectoryToMethodAnnotationSet, |
| 993 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 994 | {{4, TypeTag(kClassDefToClassData), PoolTag(kClassData)}, |
| 995 | &DisassemblerDex::MakeReadClassDefToClassData, |
| 996 | &DisassemblerDex::MakeWriteAbs32}, |
| 997 | {{1, TypeTag(kCodeToRelCode8), PoolTag(kCode)}, |
| 998 | &DisassemblerDex::MakeReadCodeToRelCode8, |
| 999 | &DisassemblerDex::MakeWriteRelCode8}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1000 | {{2, TypeTag(kCodeToRelCode16), PoolTag(kCode)}, |
| 1001 | &DisassemblerDex::MakeReadCodeToRelCode16, |
| 1002 | &DisassemblerDex::MakeWriteRelCode16}, |
| 1003 | {{4, TypeTag(kCodeToRelCode32), PoolTag(kCode)}, |
| 1004 | &DisassemblerDex::MakeReadCodeToRelCode32, |
| 1005 | &DisassemblerDex::MakeWriteRelCode32}, |
| 1006 | {{4, TypeTag(kStringIdToStringData), PoolTag(kStringData)}, |
| 1007 | &DisassemblerDex::MakeReadStringIdToStringData, |
| 1008 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 1009 | {{4, TypeTag(kAnnotationSetToAnnotation), PoolTag(kAnnotation)}, |
| 1010 | &DisassemblerDex::MakeReadAnnotationSetToAnnotation, |
| 1011 | &DisassemblerDex::MakeWriteAbs32}, |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1012 | {{4, TypeTag(kClassDefToStaticValuesEncodedArray), |
| 1013 | PoolTag(kEncodedArray)}, |
| 1014 | &DisassemblerDex::MakeReadClassDefToStaticValuesEncodedArray, |
| 1015 | &DisassemblerDex::MakeWriteAbs32}, |
| 1016 | {{4, TypeTag(kClassDefToAnnotationDirectory), |
| 1017 | PoolTag(kAnnotationsDirectory)}, |
| 1018 | &DisassemblerDex::MakeReadClassDefToAnnotationDirectory, |
| 1019 | &DisassemblerDex::MakeWriteAbs32}, |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1020 | {{4, TypeTag(kCallSiteIdToCallSite), PoolTag(kCallSite)}, |
| 1021 | &DisassemblerDex::MakeReadCallSiteIdToCallSite32, |
| 1022 | &DisassemblerDex::MakeWriteAbs32}, |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1023 | }; |
| 1024 | } |
| 1025 | |
| 1026 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadStringIdToStringData( |
| 1027 | offset_t lo, |
| 1028 | offset_t hi) { |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1029 | // dex::StringIdItem::string_data_off mapper. |
| 1030 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1031 | return std::make_unique<ItemReferenceReader>( |
| 1032 | lo, hi, string_map_item_, sizeof(dex::StringIdItem), |
| 1033 | offsetof(dex::StringIdItem, string_data_off), std::move(mapper)); |
| 1034 | } |
| 1035 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1036 | std::unique_ptr<ReferenceReader> |
| 1037 | DisassemblerDex::MakeReadTypeIdToDescriptorStringId32(offset_t lo, |
| 1038 | offset_t hi) { |
| 1039 | auto mapper = base::BindRepeating( |
| 1040 | ReadTargetIndex<decltype(dex::TypeIdItem::descriptor_idx)>, image_, |
| 1041 | string_map_item_, sizeof(dex::StringIdItem)); |
| 1042 | return std::make_unique<ItemReferenceReader>( |
| 1043 | lo, hi, type_map_item_, sizeof(dex::TypeIdItem), |
| 1044 | offsetof(dex::TypeIdItem, descriptor_idx), std::move(mapper)); |
| 1045 | } |
| 1046 | |
| 1047 | std::unique_ptr<ReferenceReader> |
| 1048 | DisassemblerDex::MakeReadProtoIdToShortyStringId32(offset_t lo, offset_t hi) { |
| 1049 | auto mapper = base::BindRepeating( |
| 1050 | ReadTargetIndex<decltype(dex::ProtoIdItem::shorty_idx)>, image_, |
| 1051 | string_map_item_, sizeof(dex::StringIdItem)); |
| 1052 | return std::make_unique<ItemReferenceReader>( |
| 1053 | lo, hi, proto_map_item_, sizeof(dex::ProtoIdItem), |
| 1054 | offsetof(dex::ProtoIdItem, shorty_idx), std::move(mapper)); |
| 1055 | } |
| 1056 | |
| 1057 | std::unique_ptr<ReferenceReader> |
| 1058 | DisassemblerDex::MakeReadProtoIdToReturnTypeId32(offset_t lo, offset_t hi) { |
| 1059 | auto mapper = base::BindRepeating( |
| 1060 | ReadTargetIndex<decltype(dex::ProtoIdItem::return_type_idx)>, image_, |
| 1061 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1062 | return std::make_unique<ItemReferenceReader>( |
| 1063 | lo, hi, proto_map_item_, sizeof(dex::ProtoIdItem), |
| 1064 | offsetof(dex::ProtoIdItem, return_type_idx), std::move(mapper)); |
| 1065 | } |
| 1066 | |
| 1067 | std::unique_ptr<ReferenceReader> |
| 1068 | DisassemblerDex::MakeReadProtoIdToParametersTypeList(offset_t lo, offset_t hi) { |
| 1069 | // dex::ProtoIdItem::parameters_off mapper. |
| 1070 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1071 | return std::make_unique<ItemReferenceReader>( |
| 1072 | lo, hi, proto_map_item_, sizeof(dex::ProtoIdItem), |
| 1073 | offsetof(dex::ProtoIdItem, parameters_off), std::move(mapper)); |
| 1074 | } |
| 1075 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1076 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadFieldToClassTypeId16( |
| 1077 | offset_t lo, |
| 1078 | offset_t hi) { |
| 1079 | auto mapper = base::BindRepeating( |
| 1080 | ReadTargetIndex<decltype(dex::FieldIdItem::class_idx)>, image_, |
| 1081 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1082 | return std::make_unique<ItemReferenceReader>( |
| 1083 | lo, hi, field_map_item_, sizeof(dex::FieldIdItem), |
| 1084 | offsetof(dex::FieldIdItem, class_idx), std::move(mapper)); |
| 1085 | } |
| 1086 | |
| 1087 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadFieldToTypeId16( |
| 1088 | offset_t lo, |
| 1089 | offset_t hi) { |
| 1090 | auto mapper = |
| 1091 | base::BindRepeating(ReadTargetIndex<decltype(dex::FieldIdItem::type_idx)>, |
| 1092 | image_, type_map_item_, sizeof(dex::TypeIdItem)); |
| 1093 | return std::make_unique<ItemReferenceReader>( |
| 1094 | lo, hi, field_map_item_, sizeof(dex::FieldIdItem), |
| 1095 | offsetof(dex::FieldIdItem, type_idx), std::move(mapper)); |
| 1096 | } |
| 1097 | |
| 1098 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadFieldToNameStringId32( |
| 1099 | offset_t lo, |
| 1100 | offset_t hi) { |
| 1101 | auto mapper = |
| 1102 | base::BindRepeating(ReadTargetIndex<decltype(dex::FieldIdItem::name_idx)>, |
| 1103 | image_, string_map_item_, sizeof(dex::StringIdItem)); |
| 1104 | return std::make_unique<ItemReferenceReader>( |
| 1105 | lo, hi, field_map_item_, sizeof(dex::FieldIdItem), |
| 1106 | offsetof(dex::FieldIdItem, name_idx), std::move(mapper)); |
| 1107 | } |
| 1108 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1109 | std::unique_ptr<ReferenceReader> |
| 1110 | DisassemblerDex::MakeReadMethodIdToClassTypeId16(offset_t lo, offset_t hi) { |
| 1111 | auto mapper = base::BindRepeating( |
| 1112 | ReadTargetIndex<decltype(dex::MethodIdItem::class_idx)>, image_, |
| 1113 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1114 | return std::make_unique<ItemReferenceReader>( |
| 1115 | lo, hi, method_map_item_, sizeof(dex::MethodIdItem), |
| 1116 | offsetof(dex::MethodIdItem, class_idx), std::move(mapper)); |
| 1117 | } |
| 1118 | |
| 1119 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadMethodIdToProtoId16( |
| 1120 | offset_t lo, |
| 1121 | offset_t hi) { |
| 1122 | auto mapper = base::BindRepeating( |
| 1123 | ReadTargetIndex<decltype(dex::MethodIdItem::proto_idx)>, image_, |
| 1124 | proto_map_item_, sizeof(dex::ProtoIdItem)); |
| 1125 | return std::make_unique<ItemReferenceReader>( |
| 1126 | lo, hi, method_map_item_, sizeof(dex::MethodIdItem), |
| 1127 | offsetof(dex::MethodIdItem, proto_idx), std::move(mapper)); |
| 1128 | } |
| 1129 | |
| 1130 | std::unique_ptr<ReferenceReader> |
| 1131 | DisassemblerDex::MakeReadMethodIdToNameStringId32(offset_t lo, offset_t hi) { |
| 1132 | auto mapper = base::BindRepeating( |
| 1133 | ReadTargetIndex<decltype(dex::MethodIdItem::name_idx)>, image_, |
| 1134 | string_map_item_, sizeof(dex::StringIdItem)); |
| 1135 | return std::make_unique<ItemReferenceReader>( |
| 1136 | lo, hi, method_map_item_, sizeof(dex::MethodIdItem), |
| 1137 | offsetof(dex::MethodIdItem, name_idx), std::move(mapper)); |
| 1138 | } |
| 1139 | |
| 1140 | std::unique_ptr<ReferenceReader> |
| 1141 | DisassemblerDex::MakeReadClassDefToClassTypeId32(offset_t lo, offset_t hi) { |
| 1142 | auto mapper = base::BindRepeating( |
| 1143 | ReadTargetIndex<decltype(dex::ClassDefItem::superclass_idx)>, image_, |
| 1144 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1145 | return std::make_unique<ItemReferenceReader>( |
| 1146 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1147 | offsetof(dex::ClassDefItem, class_idx), std::move(mapper)); |
| 1148 | } |
| 1149 | |
| 1150 | std::unique_ptr<ReferenceReader> |
| 1151 | DisassemblerDex::MakeReadClassDefToSuperClassTypeId32(offset_t lo, |
| 1152 | offset_t hi) { |
| 1153 | auto mapper = base::BindRepeating( |
| 1154 | ReadTargetIndex<decltype(dex::ClassDefItem::superclass_idx)>, image_, |
| 1155 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1156 | return std::make_unique<ItemReferenceReader>( |
| 1157 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1158 | offsetof(dex::ClassDefItem, superclass_idx), std::move(mapper)); |
| 1159 | } |
| 1160 | |
| 1161 | std::unique_ptr<ReferenceReader> |
| 1162 | DisassemblerDex::MakeReadClassDefToInterfacesTypeList(offset_t lo, |
| 1163 | offset_t hi) { |
| 1164 | // dex::ClassDefItem::interfaces_off mapper. |
| 1165 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1166 | return std::make_unique<ItemReferenceReader>( |
| 1167 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1168 | offsetof(dex::ClassDefItem, interfaces_off), std::move(mapper)); |
| 1169 | } |
| 1170 | |
| 1171 | std::unique_ptr<ReferenceReader> |
| 1172 | DisassemblerDex::MakeReadClassDefToSourceFileStringId32(offset_t lo, |
| 1173 | offset_t hi) { |
| 1174 | auto mapper = base::BindRepeating( |
| 1175 | ReadTargetIndex<decltype(dex::ClassDefItem::source_file_idx)>, image_, |
| 1176 | string_map_item_, sizeof(dex::StringIdItem)); |
| 1177 | return std::make_unique<ItemReferenceReader>( |
| 1178 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1179 | offsetof(dex::ClassDefItem, source_file_idx), std::move(mapper)); |
| 1180 | } |
| 1181 | |
| 1182 | std::unique_ptr<ReferenceReader> |
| 1183 | DisassemblerDex::MakeReadClassDefToAnnotationDirectory(offset_t lo, |
| 1184 | offset_t hi) { |
| 1185 | // dex::ClassDefItem::annotations_off mapper. |
| 1186 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1187 | return std::make_unique<ItemReferenceReader>( |
| 1188 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1189 | offsetof(dex::ClassDefItem, annotations_off), std::move(mapper)); |
| 1190 | } |
| 1191 | |
| 1192 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadClassDefToClassData( |
| 1193 | offset_t lo, |
| 1194 | offset_t hi) { |
| 1195 | // dex::ClassDefItem::class_data_off mapper. |
| 1196 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1197 | return std::make_unique<ItemReferenceReader>( |
| 1198 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1199 | offsetof(dex::ClassDefItem, class_data_off), std::move(mapper)); |
| 1200 | } |
| 1201 | |
| 1202 | std::unique_ptr<ReferenceReader> |
| 1203 | DisassemblerDex::MakeReadClassDefToStaticValuesEncodedArray(offset_t lo, |
| 1204 | offset_t hi) { |
| 1205 | // dex::ClassDefItem::static_values_off mapper. |
| 1206 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1207 | return std::make_unique<ItemReferenceReader>( |
| 1208 | lo, hi, class_def_map_item_, sizeof(dex::ClassDefItem), |
| 1209 | offsetof(dex::ClassDefItem, static_values_off), std::move(mapper)); |
| 1210 | } |
| 1211 | |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1212 | std::unique_ptr<ReferenceReader> |
| 1213 | DisassemblerDex::MakeReadCallSiteIdToCallSite32(offset_t lo, offset_t hi) { |
| 1214 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1215 | return std::make_unique<ItemReferenceReader>( |
| 1216 | lo, hi, call_site_map_item_, sizeof(dex::CallSiteIdItem), |
| 1217 | offsetof(dex::CallSiteIdItem, call_site_off), std::move(mapper)); |
| 1218 | } |
| 1219 | |
| 1220 | std::unique_ptr<ReferenceReader> |
| 1221 | DisassemblerDex::MakeReadMethodHandleToFieldId16(offset_t lo, offset_t hi) { |
| 1222 | auto mapper = base::BindRepeating(ReadMethodHandleFieldOrMethodId, image_, |
| 1223 | field_map_item_, sizeof(dex::FieldIdItem), |
| 1224 | dex::MethodHandleType::kStaticPut, |
| 1225 | dex::MethodHandleType::kInstanceGet); |
| 1226 | // Use |mapper_wants_item == true| for ItemReferenceReader such that |
| 1227 | // |location| is aligned with MethodHandleItem when passed to |mapper|. This |
| 1228 | // allows ReadMethodHandleFieldOrMethodId to safely determine whether the |
| 1229 | // reference in the MethodHandleItem is of the correct type to be emitted. |
| 1230 | return std::make_unique<ItemReferenceReader>( |
| 1231 | lo, hi, method_handle_map_item_, sizeof(dex::MethodHandleItem), |
| 1232 | offsetof(dex::MethodHandleItem, field_or_method_id), std::move(mapper), |
| 1233 | /*mapper_wants_item=*/true); |
| 1234 | } |
| 1235 | |
| 1236 | std::unique_ptr<ReferenceReader> |
| 1237 | DisassemblerDex::MakeReadMethodHandleToMethodId16(offset_t lo, offset_t hi) { |
| 1238 | auto mapper = base::BindRepeating(ReadMethodHandleFieldOrMethodId, image_, |
| 1239 | method_map_item_, sizeof(dex::MethodIdItem), |
| 1240 | dex::MethodHandleType::kInvokeStatic, |
| 1241 | dex::MethodHandleType::kInvokeInterface); |
| 1242 | // Use |mapper_wants_item == true| for ItemReferenceReader such that |
| 1243 | // |location| is aligned with MethodHandleItem when passed to |mapper|. This |
| 1244 | // allows ReadMethodHandleFieldOrMethodId to safely determine whether the |
| 1245 | // reference in the MethodHandleItem is of the correct type to be emitted. |
| 1246 | return std::make_unique<ItemReferenceReader>( |
| 1247 | lo, hi, method_handle_map_item_, sizeof(dex::MethodHandleItem), |
| 1248 | offsetof(dex::MethodHandleItem, field_or_method_id), std::move(mapper), |
| 1249 | /*mapper_wants_item=*/true); |
| 1250 | } |
| 1251 | |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 1252 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadTypeListToTypeId16( |
| 1253 | offset_t lo, |
| 1254 | offset_t hi) { |
| 1255 | auto mapper = |
| 1256 | base::BindRepeating(ReadTargetIndex<decltype(dex::TypeItem::type_idx)>, |
| 1257 | image_, type_map_item_, sizeof(dex::TypeIdItem)); |
| 1258 | return std::make_unique<CachedItemListReferenceReader>( |
| 1259 | lo, hi, offsetof(dex::TypeItem, type_idx), type_list_offsets_, |
| 1260 | std::move(mapper)); |
| 1261 | } |
| 1262 | |
| 1263 | std::unique_ptr<ReferenceReader> |
| 1264 | DisassemblerDex::MakeReadAnnotationSetToAnnotation(offset_t lo, offset_t hi) { |
| 1265 | // dex::AnnotationOffItem::annotation_off mapper. |
| 1266 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1267 | return std::make_unique<CachedItemListReferenceReader>( |
| 1268 | lo, hi, offsetof(dex::AnnotationOffItem, annotation_off), |
| 1269 | annotation_set_offsets_, std::move(mapper)); |
| 1270 | } |
| 1271 | |
| 1272 | std::unique_ptr<ReferenceReader> |
| 1273 | DisassemblerDex::MakeReadAnnotationSetRefListToAnnotationSet(offset_t lo, |
| 1274 | offset_t hi) { |
| 1275 | // dex::AnnotationSetRefItem::annotations_off mapper. |
| 1276 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1277 | return std::make_unique<CachedItemListReferenceReader>( |
| 1278 | lo, hi, offsetof(dex::AnnotationSetRefItem, annotations_off), |
| 1279 | annotation_set_ref_list_offsets_, std::move(mapper)); |
| 1280 | } |
| 1281 | |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 1282 | std::unique_ptr<ReferenceReader> |
| 1283 | DisassemblerDex::MakeReadAnnotationsDirectoryToClassAnnotationSet(offset_t lo, |
| 1284 | offset_t hi) { |
| 1285 | // dex::AnnotationsDirectoryItem::class_annotations_off mapper. |
| 1286 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1287 | return std::make_unique<CachedItemListReferenceReader>( |
| 1288 | lo, hi, offsetof(dex::AnnotationsDirectoryItem, class_annotations_off), |
| 1289 | annotations_directory_item_offsets_, std::move(mapper)); |
| 1290 | } |
| 1291 | |
| 1292 | std::unique_ptr<ReferenceReader> |
| 1293 | DisassemblerDex::MakeReadAnnotationsDirectoryToFieldId32(offset_t lo, |
| 1294 | offset_t hi) { |
| 1295 | auto mapper = base::BindRepeating( |
| 1296 | ReadTargetIndex<decltype(dex::FieldAnnotation::field_idx)>, image_, |
| 1297 | field_map_item_, sizeof(dex::FieldIdItem)); |
| 1298 | return std::make_unique<CachedItemListReferenceReader>( |
| 1299 | lo, hi, offsetof(dex::FieldAnnotation, field_idx), |
| 1300 | annotations_directory_item_field_annotation_offsets_, std::move(mapper)); |
| 1301 | } |
| 1302 | |
| 1303 | std::unique_ptr<ReferenceReader> |
| 1304 | DisassemblerDex::MakeReadAnnotationsDirectoryToFieldAnnotationSet(offset_t lo, |
| 1305 | offset_t hi) { |
| 1306 | // dex::FieldAnnotation::annotations_off mapper. |
| 1307 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1308 | return std::make_unique<CachedItemListReferenceReader>( |
| 1309 | lo, hi, offsetof(dex::FieldAnnotation, annotations_off), |
| 1310 | annotations_directory_item_field_annotation_offsets_, std::move(mapper)); |
| 1311 | } |
| 1312 | |
| 1313 | std::unique_ptr<ReferenceReader> |
| 1314 | DisassemblerDex::MakeReadAnnotationsDirectoryToMethodId32(offset_t lo, |
| 1315 | offset_t hi) { |
| 1316 | auto mapper = base::BindRepeating( |
| 1317 | ReadTargetIndex<decltype(dex::MethodAnnotation::method_idx)>, image_, |
| 1318 | method_map_item_, sizeof(dex::MethodIdItem)); |
| 1319 | return std::make_unique<CachedItemListReferenceReader>( |
| 1320 | lo, hi, offsetof(dex::MethodAnnotation, method_idx), |
| 1321 | annotations_directory_item_method_annotation_offsets_, std::move(mapper)); |
| 1322 | } |
| 1323 | |
| 1324 | std::unique_ptr<ReferenceReader> |
| 1325 | DisassemblerDex::MakeReadAnnotationsDirectoryToMethodAnnotationSet( |
| 1326 | offset_t lo, |
| 1327 | offset_t hi) { |
| 1328 | // dex::MethodAnnotation::annotations_off mapper. |
| 1329 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1330 | return std::make_unique<CachedItemListReferenceReader>( |
| 1331 | lo, hi, offsetof(dex::MethodAnnotation, annotations_off), |
| 1332 | annotations_directory_item_method_annotation_offsets_, std::move(mapper)); |
| 1333 | } |
| 1334 | |
| 1335 | std::unique_ptr<ReferenceReader> |
| 1336 | DisassemblerDex::MakeReadAnnotationsDirectoryToParameterMethodId32( |
| 1337 | offset_t lo, |
| 1338 | offset_t hi) { |
| 1339 | auto mapper = base::BindRepeating( |
| 1340 | ReadTargetIndex<decltype(dex::ParameterAnnotation::method_idx)>, image_, |
| 1341 | method_map_item_, sizeof(dex::MethodIdItem)); |
| 1342 | return std::make_unique<CachedItemListReferenceReader>( |
| 1343 | lo, hi, offsetof(dex::ParameterAnnotation, method_idx), |
| 1344 | annotations_directory_item_parameter_annotation_offsets_, |
| 1345 | std::move(mapper)); |
| 1346 | } |
| 1347 | |
| 1348 | std::unique_ptr<ReferenceReader> |
| 1349 | DisassemblerDex::MakeReadAnnotationsDirectoryToParameterAnnotationSetRef( |
| 1350 | offset_t lo, |
| 1351 | offset_t hi) { |
| 1352 | // dex::ParameterAnnotation::annotations_off mapper. |
| 1353 | auto mapper = base::BindRepeating(ReadTargetOffset32, image_); |
| 1354 | return std::make_unique<CachedItemListReferenceReader>( |
| 1355 | lo, hi, offsetof(dex::ParameterAnnotation, annotations_off), |
| 1356 | annotations_directory_item_parameter_annotation_offsets_, |
| 1357 | std::move(mapper)); |
| 1358 | } |
| 1359 | |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1360 | // MakeReadCode* readers use offset relative to the instruction beginning based |
| 1361 | // on the instruction format ID. |
| 1362 | // See https://source.android.com/devices/tech/dalvik/instruction-formats |
| 1363 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1364 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToStringId16( |
| 1365 | offset_t lo, |
| 1366 | offset_t hi) { |
| 1367 | auto filter = base::BindRepeating( |
| 1368 | [](const InstructionParser::Value& value) -> offset_t { |
| 1369 | if (value.instr->format == dex::FormatId::c && |
| 1370 | (value.instr->opcode == 0x1A)) { // const-string |
| 1371 | // BBBB from e.g., const-string vAA, string@BBBB. |
| 1372 | return value.instr_offset + 2; |
| 1373 | } |
| 1374 | return kInvalidOffset; |
| 1375 | }); |
| 1376 | auto mapper = |
| 1377 | base::BindRepeating(ReadTargetIndex<uint16_t>, image_, string_map_item_, |
| 1378 | sizeof(dex::StringIdItem)); |
| 1379 | return std::make_unique<InstructionReferenceReader>( |
| 1380 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1381 | } |
| 1382 | |
| 1383 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToStringId32( |
| 1384 | offset_t lo, |
| 1385 | offset_t hi) { |
| 1386 | auto filter = base::BindRepeating( |
| 1387 | [](const InstructionParser::Value& value) -> offset_t { |
| 1388 | if (value.instr->format == dex::FormatId::c && |
| 1389 | (value.instr->opcode == 0x1B)) { // const-string/jumbo |
| 1390 | // BBBBBBBB from e.g., const-string/jumbo vAA, string@BBBBBBBB. |
| 1391 | return value.instr_offset + 2; |
| 1392 | } |
| 1393 | return kInvalidOffset; |
| 1394 | }); |
| 1395 | auto mapper = |
| 1396 | base::BindRepeating(ReadTargetIndex<uint32_t>, image_, string_map_item_, |
| 1397 | sizeof(dex::StringIdItem)); |
| 1398 | return std::make_unique<InstructionReferenceReader>( |
| 1399 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1400 | } |
| 1401 | |
| 1402 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToTypeId16( |
| 1403 | offset_t lo, |
| 1404 | offset_t hi) { |
| 1405 | auto filter = base::BindRepeating( |
| 1406 | [](const InstructionParser::Value& value) -> offset_t { |
| 1407 | if (value.instr->format == dex::FormatId::c && |
| 1408 | (value.instr->opcode == 0x1C || // const-class |
| 1409 | value.instr->opcode == 0x1F || // check-cast |
| 1410 | value.instr->opcode == 0x20 || // instance-of |
| 1411 | value.instr->opcode == 0x22 || // new-instance |
| 1412 | value.instr->opcode == 0x23 || // new-array |
| 1413 | value.instr->opcode == 0x24 || // filled-new-array |
| 1414 | value.instr->opcode == 0x25)) { // filled-new-array/range |
| 1415 | // BBBB from e.g., const-class vAA, type@BBBB. |
| 1416 | return value.instr_offset + 2; |
| 1417 | } |
| 1418 | return kInvalidOffset; |
| 1419 | }); |
| 1420 | auto mapper = base::BindRepeating(ReadTargetIndex<uint16_t>, image_, |
| 1421 | type_map_item_, sizeof(dex::TypeIdItem)); |
| 1422 | return std::make_unique<InstructionReferenceReader>( |
| 1423 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1424 | } |
| 1425 | |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1426 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToProtoId16( |
| 1427 | offset_t lo, |
| 1428 | offset_t hi) { |
| 1429 | auto filter = base::BindRepeating( |
| 1430 | [](const InstructionParser::Value& value) -> offset_t { |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 1431 | if (value.instr->format == dex::FormatId::c) { |
| 1432 | if (value.instr->opcode == 0xFA || // invoke-polymorphic |
| 1433 | value.instr->opcode == 0xFB) { // invoke-polymorphic/range |
| 1434 | // HHHH from e.g, invoke-polymorphic {vC, vD, vE, vF, vG}, |
| 1435 | // meth@BBBB, proto@HHHH |
| 1436 | return value.instr_offset + 6; |
| 1437 | } |
| 1438 | if (value.instr->opcode == 0xFF) { // const-method-type |
| 1439 | // BBBB from e.g., const-method-type vAA, proto@BBBB |
| 1440 | return value.instr_offset + 2; |
| 1441 | } |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1442 | } |
| 1443 | return kInvalidOffset; |
| 1444 | }); |
| 1445 | auto mapper = base::BindRepeating(ReadTargetIndex<uint16_t>, image_, |
| 1446 | proto_map_item_, sizeof(dex::ProtoIdItem)); |
| 1447 | return std::make_unique<InstructionReferenceReader>( |
| 1448 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1449 | } |
| 1450 | |
| 1451 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToCallSiteId16( |
| 1452 | offset_t lo, |
| 1453 | offset_t hi) { |
| 1454 | auto filter = base::BindRepeating( |
| 1455 | [](const InstructionParser::Value& value) -> offset_t { |
| 1456 | if (value.instr->format == dex::FormatId::c && |
| 1457 | (value.instr->opcode == 0xFC || // invoke-custom |
| 1458 | value.instr->opcode == 0xFD)) { // invoke-custom/range |
| 1459 | // BBBB from e.g, invoke-custom {vC, vD, vE, vF, vG}, |
| 1460 | // call_site@BBBB |
| 1461 | return value.instr_offset + 2; |
| 1462 | } |
| 1463 | return kInvalidOffset; |
| 1464 | }); |
| 1465 | auto mapper = |
| 1466 | base::BindRepeating(ReadTargetIndex<uint16_t>, image_, |
| 1467 | call_site_map_item_, sizeof(dex::CallSiteIdItem)); |
| 1468 | return std::make_unique<InstructionReferenceReader>( |
| 1469 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1470 | } |
| 1471 | |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 1472 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToMethodHandle16( |
| 1473 | offset_t lo, |
| 1474 | offset_t hi) { |
| 1475 | auto filter = base::BindRepeating( |
| 1476 | [](const InstructionParser::Value& value) -> offset_t { |
| 1477 | if (value.instr->format == dex::FormatId::c && |
| 1478 | value.instr->opcode == 0xFE) { // const-method-handle |
| 1479 | // BBBB from e.g, const-method-handle vAA, method_handle@BBBB |
| 1480 | return value.instr_offset + 2; |
| 1481 | } |
| 1482 | return kInvalidOffset; |
| 1483 | }); |
| 1484 | auto mapper = base::BindRepeating(ReadTargetIndex<uint16_t>, image_, |
| 1485 | method_handle_map_item_, |
| 1486 | sizeof(dex::MethodHandleItem)); |
| 1487 | return std::make_unique<InstructionReferenceReader>( |
| 1488 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1489 | } |
| 1490 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1491 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToFieldId16( |
| 1492 | offset_t lo, |
| 1493 | offset_t hi) { |
| 1494 | auto filter = base::BindRepeating( |
| 1495 | [](const InstructionParser::Value& value) -> offset_t { |
| 1496 | if (value.instr->format == dex::FormatId::c && |
| 1497 | (value.instr->opcode == 0x52 || // iinstanceop (iget-*, iput-*) |
| 1498 | value.instr->opcode == 0x60)) { // sstaticop (sget-*, sput-*) |
| 1499 | // CCCC from e.g., iget vA, vB, field@CCCC. |
| 1500 | return value.instr_offset + 2; |
| 1501 | } |
| 1502 | return kInvalidOffset; |
| 1503 | }); |
| 1504 | auto mapper = base::BindRepeating(ReadTargetIndex<uint16_t>, image_, |
| 1505 | field_map_item_, sizeof(dex::FieldIdItem)); |
| 1506 | return std::make_unique<InstructionReferenceReader>( |
| 1507 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1508 | } |
| 1509 | |
| 1510 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToMethodId16( |
| 1511 | offset_t lo, |
| 1512 | offset_t hi) { |
| 1513 | auto filter = base::BindRepeating( |
| 1514 | [](const InstructionParser::Value& value) -> offset_t { |
| 1515 | if (value.instr->format == dex::FormatId::c && |
| 1516 | (value.instr->opcode == 0x6E || // invoke-kind |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1517 | value.instr->opcode == 0x74 || // invoke-kind/range |
| 1518 | value.instr->opcode == 0xFA || // invoke-polymorphic |
| 1519 | value.instr->opcode == 0xFB)) { // invoke-polymorphic/range |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1520 | // BBBB from e.g., invoke-virtual {vC, vD, vE, vF, vG}, meth@BBBB. |
| 1521 | return value.instr_offset + 2; |
| 1522 | } |
| 1523 | return kInvalidOffset; |
| 1524 | }); |
| 1525 | auto mapper = |
| 1526 | base::BindRepeating(ReadTargetIndex<uint16_t>, image_, method_map_item_, |
| 1527 | sizeof(dex::MethodIdItem)); |
| 1528 | return std::make_unique<InstructionReferenceReader>( |
| 1529 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1530 | } |
| 1531 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1532 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToRelCode8( |
| 1533 | offset_t lo, |
| 1534 | offset_t hi) { |
| 1535 | auto filter = base::BindRepeating( |
| 1536 | [](const InstructionParser::Value& value) -> offset_t { |
| 1537 | if (value.instr->format == dex::FormatId::t && |
| 1538 | value.instr->opcode == 0x28) { // goto |
| 1539 | // +AA from e.g., goto +AA. |
| 1540 | return value.instr_offset + 1; |
| 1541 | } |
| 1542 | return kInvalidOffset; |
| 1543 | }); |
| 1544 | auto mapper = base::BindRepeating( |
| 1545 | [](DisassemblerDex* dis, offset_t location) { |
| 1546 | // Address is relative to the current instruction, which begins 1 unit |
| 1547 | // before |location|. This needs to be subtracted out. Also, store as |
| 1548 | // int32_t so |unsafe_delta - 1| won't underflow! |
| 1549 | int32_t unsafe_delta = dis->image_.read<int8_t>(location); |
| 1550 | offset_t unsafe_target = static_cast<offset_t>( |
| 1551 | location + (unsafe_delta - 1) * kInstrUnitSize); |
| 1552 | // TODO(huangs): Check that |unsafe_target| stays within code item. |
| 1553 | return unsafe_target; |
| 1554 | }, |
| 1555 | base::Unretained(this)); |
| 1556 | return std::make_unique<InstructionReferenceReader>( |
| 1557 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1558 | } |
| 1559 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1560 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToRelCode16( |
| 1561 | offset_t lo, |
| 1562 | offset_t hi) { |
| 1563 | auto filter = base::BindRepeating( |
| 1564 | [](const InstructionParser::Value& value) -> offset_t { |
| 1565 | if (value.instr->format == dex::FormatId::t && |
| 1566 | (value.instr->opcode == 0x29 || // goto/16 |
| 1567 | value.instr->opcode == 0x32 || // if-test |
| 1568 | value.instr->opcode == 0x38)) { // if-testz |
| 1569 | // +AAAA from e.g., goto/16 +AAAA. |
| 1570 | return value.instr_offset + 2; |
| 1571 | } |
| 1572 | return kInvalidOffset; |
| 1573 | }); |
| 1574 | auto mapper = base::BindRepeating( |
| 1575 | [](DisassemblerDex* dis, offset_t location) { |
| 1576 | // Address is relative to the current instruction, which begins 1 unit |
| 1577 | // before |location|. This needs to be subtracted out. Also, store as |
| 1578 | // int32_t so |unsafe_delta - 1| won't underflow! |
| 1579 | int32_t unsafe_delta = dis->image_.read<int16_t>(location); |
| 1580 | offset_t unsafe_target = static_cast<offset_t>( |
| 1581 | location + (unsafe_delta - 1) * kInstrUnitSize); |
| 1582 | // TODO(huangs): Check that |unsafe_target| stays within code item. |
| 1583 | return unsafe_target; |
| 1584 | }, |
| 1585 | base::Unretained(this)); |
| 1586 | return std::make_unique<InstructionReferenceReader>( |
| 1587 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1588 | } |
| 1589 | |
| 1590 | std::unique_ptr<ReferenceReader> DisassemblerDex::MakeReadCodeToRelCode32( |
| 1591 | offset_t lo, |
| 1592 | offset_t hi) { |
| 1593 | auto filter = base::BindRepeating( |
| 1594 | [](const InstructionParser::Value& value) -> offset_t { |
| 1595 | if (value.instr->format == dex::FormatId::t && |
| 1596 | (value.instr->opcode == 0x26 || // fill-array-data |
| 1597 | value.instr->opcode == 0x2A || // goto/32 |
| 1598 | value.instr->opcode == 0x2B || // packed-switch |
| 1599 | value.instr->opcode == 0x2C)) { // sparse-switch |
| 1600 | // +BBBBBBBB from e.g., fill-array-data vAA, +BBBBBBBB. |
| 1601 | // +AAAAAAAA from e.g., goto/32 +AAAAAAAA. |
| 1602 | return value.instr_offset + 2; |
| 1603 | } |
| 1604 | return kInvalidOffset; |
| 1605 | }); |
| 1606 | auto mapper = base::BindRepeating( |
| 1607 | [](DisassemblerDex* dis, offset_t location) { |
| 1608 | // Address is relative to the current instruction, which begins 1 unit |
Calder Kitagawa | 620015b | 2018-07-11 14:55:44 +0000 | [diff] [blame] | 1609 | // before |location|. This needs to be subtracted out. Use int64_t to |
| 1610 | // avoid underflow and overflow. |
| 1611 | int64_t unsafe_delta = dis->image_.read<int32_t>(location); |
| 1612 | int64_t unsafe_target = location + (unsafe_delta - 1) * kInstrUnitSize; |
| 1613 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1614 | // TODO(huangs): Check that |unsafe_target| stays within code item. |
Calder Kitagawa | 620015b | 2018-07-11 14:55:44 +0000 | [diff] [blame] | 1615 | offset_t checked_unsafe_target = |
| 1616 | static_cast<offset_t>(base::CheckedNumeric<offset_t>(unsafe_target) |
| 1617 | .ValueOrDefault(kInvalidOffset)); |
| 1618 | return checked_unsafe_target < kOffsetBound ? checked_unsafe_target |
| 1619 | : kInvalidOffset; |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1620 | }, |
| 1621 | base::Unretained(this)); |
| 1622 | return std::make_unique<InstructionReferenceReader>( |
| 1623 | image_, lo, hi, code_item_offsets_, std::move(filter), std::move(mapper)); |
| 1624 | } |
| 1625 | |
| 1626 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteStringId16( |
| 1627 | MutableBufferView image) { |
| 1628 | auto writer = base::BindRepeating( |
| 1629 | WriteTargetIndex<uint16_t>, string_map_item_, sizeof(dex::StringIdItem)); |
| 1630 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1631 | } |
| 1632 | |
| 1633 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteStringId32( |
| 1634 | MutableBufferView image) { |
| 1635 | auto writer = base::BindRepeating( |
| 1636 | WriteTargetIndex<uint32_t>, string_map_item_, sizeof(dex::StringIdItem)); |
| 1637 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1638 | } |
| 1639 | |
| 1640 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteTypeId16( |
| 1641 | MutableBufferView image) { |
| 1642 | auto writer = base::BindRepeating(WriteTargetIndex<uint16_t>, type_map_item_, |
| 1643 | sizeof(dex::TypeIdItem)); |
| 1644 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1645 | } |
| 1646 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1647 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteTypeId32( |
| 1648 | MutableBufferView image) { |
| 1649 | auto writer = base::BindRepeating(WriteTargetIndex<uint32_t>, type_map_item_, |
| 1650 | sizeof(dex::TypeIdItem)); |
| 1651 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1652 | } |
| 1653 | |
| 1654 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteProtoId16( |
| 1655 | MutableBufferView image) { |
| 1656 | auto writer = base::BindRepeating(WriteTargetIndex<uint16_t>, proto_map_item_, |
| 1657 | sizeof(dex::ProtoIdItem)); |
| 1658 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1659 | } |
| 1660 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1661 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteFieldId16( |
| 1662 | MutableBufferView image) { |
| 1663 | auto writer = base::BindRepeating(WriteTargetIndex<uint16_t>, field_map_item_, |
| 1664 | sizeof(dex::FieldIdItem)); |
| 1665 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1666 | } |
| 1667 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1668 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteFieldId32( |
| 1669 | MutableBufferView image) { |
| 1670 | auto writer = base::BindRepeating(WriteTargetIndex<uint32_t>, field_map_item_, |
| 1671 | sizeof(dex::FieldIdItem)); |
| 1672 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1673 | } |
| 1674 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1675 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteMethodId16( |
| 1676 | MutableBufferView image) { |
| 1677 | auto writer = base::BindRepeating( |
| 1678 | WriteTargetIndex<uint16_t>, method_map_item_, sizeof(dex::MethodIdItem)); |
| 1679 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1680 | } |
| 1681 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1682 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteMethodId32( |
| 1683 | MutableBufferView image) { |
| 1684 | auto writer = base::BindRepeating( |
| 1685 | WriteTargetIndex<uint32_t>, method_map_item_, sizeof(dex::MethodIdItem)); |
| 1686 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1687 | } |
| 1688 | |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1689 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteCallSiteId16( |
| 1690 | MutableBufferView image) { |
| 1691 | auto writer = |
| 1692 | base::BindRepeating(WriteTargetIndex<uint16_t>, call_site_map_item_, |
| 1693 | sizeof(dex::CallSiteIdItem)); |
| 1694 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1695 | } |
| 1696 | |
ckitagawa | 1269b5c | 2021-09-07 21:12:21 +0000 | [diff] [blame] | 1697 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteMethodHandle16( |
| 1698 | MutableBufferView image) { |
| 1699 | auto writer = |
| 1700 | base::BindRepeating(WriteTargetIndex<uint16_t>, method_handle_map_item_, |
| 1701 | sizeof(dex::MethodHandleItem)); |
| 1702 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1703 | } |
| 1704 | |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1705 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteRelCode8( |
| 1706 | MutableBufferView image) { |
| 1707 | auto writer = base::BindRepeating([](Reference ref, MutableBufferView image) { |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1708 | ptrdiff_t unsafe_byte_diff = |
| 1709 | static_cast<ptrdiff_t>(ref.target) - ref.location; |
| 1710 | DCHECK_EQ(0, unsafe_byte_diff % kInstrUnitSize); |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1711 | // |delta| is relative to start of instruction, which is 1 unit before |
| 1712 | // |ref.location|. The subtraction above removed too much, so +1 to fix. |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1713 | base::CheckedNumeric<int8_t> delta((unsafe_byte_diff / kInstrUnitSize) + 1); |
| 1714 | if (!delta.IsValid()) { |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 1715 | LOG(ERROR) << "Invalid reference at: " << AsHex<8>(ref.location) << "."; |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1716 | return; |
| 1717 | } |
| 1718 | image.write<int8_t>(ref.location, delta.ValueOrDie()); |
Calder Kitagawa | 1837991 | 2018-06-05 14:10:29 +0000 | [diff] [blame] | 1719 | }); |
| 1720 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1721 | } |
| 1722 | |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1723 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteRelCode16( |
| 1724 | MutableBufferView image) { |
| 1725 | auto writer = base::BindRepeating([](Reference ref, MutableBufferView image) { |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1726 | ptrdiff_t unsafe_byte_diff = |
| 1727 | static_cast<ptrdiff_t>(ref.target) - ref.location; |
| 1728 | DCHECK_EQ(0, unsafe_byte_diff % kInstrUnitSize); |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1729 | // |delta| is relative to start of instruction, which is 1 unit before |
| 1730 | // |ref.location|. The subtraction above removed too much, so +1 to fix. |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1731 | base::CheckedNumeric<int16_t> delta((unsafe_byte_diff / kInstrUnitSize) + |
| 1732 | 1); |
| 1733 | if (!delta.IsValid()) { |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 1734 | LOG(ERROR) << "Invalid reference at: " << AsHex<8>(ref.location) << "."; |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1735 | return; |
| 1736 | } |
| 1737 | image.write<int16_t>(ref.location, delta.ValueOrDie()); |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1738 | }); |
| 1739 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1740 | } |
| 1741 | |
| 1742 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteRelCode32( |
| 1743 | MutableBufferView image) { |
| 1744 | auto writer = base::BindRepeating([](Reference ref, MutableBufferView image) { |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1745 | ptrdiff_t unsafe_byte_diff = |
| 1746 | static_cast<ptrdiff_t>(ref.target) - ref.location; |
| 1747 | DCHECK_EQ(0, unsafe_byte_diff % kInstrUnitSize); |
| 1748 | // |delta| is relative to start of instruction, which is 1 unit before |
| 1749 | // |ref.location|. The subtraction above removed too much, so +1 to fix. |
| 1750 | base::CheckedNumeric<int32_t> delta((unsafe_byte_diff / kInstrUnitSize) + |
| 1751 | 1); |
| 1752 | if (!delta.IsValid()) { |
Calder Kitagawa | 3a5639a | 2018-07-09 18:53:48 +0000 | [diff] [blame] | 1753 | LOG(ERROR) << "Invalid reference at: " << AsHex<8>(ref.location) << "."; |
Calder Kitagawa | 40a74cc | 2018-07-04 20:44:27 +0000 | [diff] [blame] | 1754 | return; |
| 1755 | } |
| 1756 | image.write<int32_t>(ref.location, delta.ValueOrDie()); |
Samuel Huang | a8a2a94 | 2018-04-09 15:18:17 +0000 | [diff] [blame] | 1757 | }); |
| 1758 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
| 1759 | } |
| 1760 | |
| 1761 | std::unique_ptr<ReferenceWriter> DisassemblerDex::MakeWriteAbs32( |
| 1762 | MutableBufferView image) { |
| 1763 | auto writer = base::BindRepeating([](Reference ref, MutableBufferView image) { |
| 1764 | image.write<uint32_t>(ref.location, ref.target); |
| 1765 | }); |
| 1766 | return std::make_unique<ReferenceWriterAdaptor>(image, std::move(writer)); |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1767 | } |
| 1768 | |
| 1769 | bool DisassemblerDex::Parse(ConstBufferView image) { |
| 1770 | image_ = image; |
| 1771 | return ParseHeader(); |
| 1772 | } |
| 1773 | |
| 1774 | bool DisassemblerDex::ParseHeader() { |
| 1775 | ReadDexHeaderResults results; |
| 1776 | if (!ReadDexHeader(image_, &results)) |
| 1777 | return false; |
| 1778 | |
| 1779 | header_ = results.header; |
| 1780 | dex_version_ = results.dex_version; |
| 1781 | BufferSource source = results.source; |
| 1782 | |
| 1783 | // DEX header contains file size, so use it to resize |image_| right away. |
| 1784 | image_.shrink(header_->file_size); |
| 1785 | |
| 1786 | // Read map list. This is not a fixed-size array, so instead of reading |
| 1787 | // MapList directly, read |MapList::size| first, then visit elements in |
| 1788 | // |MapList::list|. |
| 1789 | static_assert( |
| 1790 | offsetof(dex::MapList, list) == sizeof(decltype(dex::MapList::size)), |
| 1791 | "MapList size error."); |
| 1792 | source = std::move(BufferSource(image_).Skip(header_->map_off)); |
| 1793 | decltype(dex::MapList::size) list_size = 0; |
| 1794 | if (!source.GetValue(&list_size) || list_size > dex::kMaxItemListSize) |
| 1795 | return false; |
| 1796 | const auto* item_list = source.GetArray<const dex::MapItem>(list_size); |
| 1797 | if (!item_list) |
| 1798 | return false; |
| 1799 | |
| 1800 | // Read and validate map list, ensuring that required item types are present. |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1801 | // GetItemBaseSize() should have an entry for each item. |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1802 | for (offset_t i = 0; i < list_size; ++i) { |
| 1803 | const dex::MapItem* item = &item_list[i]; |
Samuel Huang | 06a9be1 | 2018-07-11 20:02:23 +0000 | [diff] [blame] | 1804 | // Reject unreasonably large |item->size|. |
| 1805 | size_t item_size = GetItemBaseSize(item->type); |
| 1806 | // Confusing name: |item->size| is actually the number of items. |
| 1807 | if (!image_.covers_array(item->offset, item->size, item_size)) |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1808 | return false; |
| 1809 | if (!map_item_map_.insert(std::make_pair(item->type, item)).second) |
| 1810 | return false; // A given type must appear at most once. |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1811 | } |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1812 | |
| 1813 | // Make local copies of main map items. |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1814 | if (map_item_map_.count(dex::kTypeStringIdItem)) { |
| 1815 | string_map_item_ = *map_item_map_[dex::kTypeStringIdItem]; |
| 1816 | } |
| 1817 | if (map_item_map_.count(dex::kTypeTypeIdItem)) { |
| 1818 | type_map_item_ = *map_item_map_[dex::kTypeTypeIdItem]; |
| 1819 | } |
| 1820 | if (map_item_map_.count(dex::kTypeProtoIdItem)) { |
| 1821 | proto_map_item_ = *map_item_map_[dex::kTypeProtoIdItem]; |
| 1822 | } |
| 1823 | if (map_item_map_.count(dex::kTypeFieldIdItem)) { |
| 1824 | field_map_item_ = *map_item_map_[dex::kTypeFieldIdItem]; |
| 1825 | } |
| 1826 | if (map_item_map_.count(dex::kTypeMethodIdItem)) { |
| 1827 | method_map_item_ = *map_item_map_[dex::kTypeMethodIdItem]; |
| 1828 | } |
| 1829 | if (map_item_map_.count(dex::kTypeClassDefItem)) { |
| 1830 | class_def_map_item_ = *map_item_map_[dex::kTypeClassDefItem]; |
| 1831 | } |
| 1832 | if (map_item_map_.count(dex::kTypeCallSiteIdItem)) { |
| 1833 | call_site_map_item_ = *map_item_map_[dex::kTypeCallSiteIdItem]; |
| 1834 | } |
| 1835 | if (map_item_map_.count(dex::kTypeMethodHandleItem)) { |
| 1836 | method_handle_map_item_ = *map_item_map_[dex::kTypeMethodHandleItem]; |
| 1837 | } |
| 1838 | if (map_item_map_.count(dex::kTypeTypeList)) { |
| 1839 | type_list_map_item_ = *map_item_map_[dex::kTypeTypeList]; |
| 1840 | } |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 1841 | if (map_item_map_.count(dex::kTypeAnnotationSetRefList)) { |
| 1842 | annotation_set_ref_list_map_item_ = |
| 1843 | *map_item_map_[dex::kTypeAnnotationSetRefList]; |
| 1844 | } |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1845 | if (map_item_map_.count(dex::kTypeAnnotationSetItem)) { |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 1846 | annotation_set_map_item_ = *map_item_map_[dex::kTypeAnnotationSetItem]; |
ckitagawa | 26518ff | 2021-09-03 15:48:28 +0000 | [diff] [blame] | 1847 | } |
| 1848 | if (map_item_map_.count(dex::kTypeCodeItem)) { |
| 1849 | code_map_item_ = *map_item_map_[dex::kTypeCodeItem]; |
| 1850 | } |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 1851 | if (map_item_map_.count(dex::kTypeAnnotationsDirectoryItem)) { |
| 1852 | annotations_directory_map_item_ = |
| 1853 | *map_item_map_[dex::kTypeAnnotationsDirectoryItem]; |
Calder Kitagawa | 6b80ac7 | 2018-06-11 13:54:24 +0000 | [diff] [blame] | 1854 | } |
| 1855 | |
Calder Kitagawa | 1b25709 | 2018-06-12 14:37:39 +0000 | [diff] [blame] | 1856 | // Iteratively parse variable length lists, annotations directory items, and |
| 1857 | // code items blocks. Any failure would indicate invalid DEX. Success |
| 1858 | // indicates that no structural problem is found. However, contained |
| 1859 | // references data read from parsed items still require validation. |
| 1860 | if (!(ParseItemOffsets(image_, type_list_map_item_, sizeof(dex::TypeItem), |
| 1861 | &type_list_offsets_) && |
| 1862 | ParseItemOffsets(image_, annotation_set_ref_list_map_item_, |
| 1863 | sizeof(dex::AnnotationSetRefItem), |
| 1864 | &annotation_set_ref_list_offsets_) && |
| 1865 | ParseItemOffsets(image_, annotation_set_map_item_, |
| 1866 | sizeof(dex::AnnotationOffItem), |
| 1867 | &annotation_set_offsets_) && |
| 1868 | ParseAnnotationsDirectoryItems( |
| 1869 | image_, annotations_directory_map_item_, |
| 1870 | &annotations_directory_item_offsets_, |
| 1871 | &annotations_directory_item_field_annotation_offsets_, |
| 1872 | &annotations_directory_item_method_annotation_offsets_, |
| 1873 | &annotations_directory_item_parameter_annotation_offsets_))) { |
| 1874 | return false; |
| 1875 | } |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1876 | CodeItemParser code_item_parser(image_); |
| 1877 | if (!code_item_parser.Init(code_map_item_)) |
| 1878 | return false; |
| 1879 | code_item_offsets_.resize(code_map_item_.size); |
| 1880 | for (size_t i = 0; i < code_map_item_.size; ++i) { |
| 1881 | const offset_t code_item_offset = code_item_parser.GetNext(); |
| 1882 | if (code_item_offset == kInvalidOffset) |
| 1883 | return false; |
| 1884 | code_item_offsets_[i] = code_item_offset; |
| 1885 | } |
Calder Kitagawa | 82e8472 | 2018-06-28 21:48:40 +0000 | [diff] [blame] | 1886 | // DEX files are required to have parsable code items. |
| 1887 | return !code_item_offsets_.empty(); |
Etienne Pierre-Doray | fff1ca3 | 2018-03-29 13:33:46 +0000 | [diff] [blame] | 1888 | } |
| 1889 | |
| 1890 | } // namespace zucchini |