A tool for deriving PKG packet encryption keys for ps4 written in c++
| 1 | #include "DecompilerContext.h" |
| 2 | #include "analysis/MemberAccessAnalysis.h" |
| 3 | #include "common/logging/log.h" |
| 4 | #include <capstone/capstone.h> |
| 5 | #include <chrono> |
| 6 | #include <cstdlib> |
| 7 | #include <cstring> |
| 8 | #include <filesystem> |
| 9 | #include <fstream> |
| 10 | #include <iomanip> |
| 11 | #include <iostream> |
| 12 | #include <queue> |
| 13 | #include <set> |
| 14 | #include <sstream> |
| 15 | #define JSON_ASSERT(x) |
| 16 | #include <nlohmann/json.hpp> |
| 17 | using json = nlohmann::json; |
| 18 | |
| 19 | namespace ShadPKG::Decompiler { |
| 20 | |
| 21 | void DecompilerContext::LoadBinary(const std::vector<uint8_t> &data, |
| 22 | uint64_t baseAddress) { |
| 23 | rawData_ = data; |
| 24 | baseAddress_ = baseAddress; |
| 25 | isAnalyzed_ = false; |
| 26 | functions_.clear(); |
| 27 | // Initialize SymbolDatabase |
| 28 | symbolDatabase_ = std::make_shared<Analysis::SymbolDatabase>(); |
| 29 | } |
| 30 | |
| 31 | bool DecompilerContext::LoadELF(const std::vector<uint8_t> &data) { |
| 32 | if (data.size() < 64) |
| 33 | return false; |
| 34 | |
| 35 | size_t elfOffset = 0; |
| 36 | bool found = false; |
| 37 | |
| 38 | // Check Magic: 7F 45 4C 46 |
| 39 | if (data[0] == 0x7F && data[1] == 'E' && data[2] == 'L' && data[3] == 'F') { |
| 40 | found = true; |
| 41 | } else { |
| 42 | // Search for it (limit search to first 4KB) |
| 43 | for (size_t i = 0; i < std::min(data.size() - 4, (size_t)0x1000); ++i) { |
| 44 | if (data[i] == 0x7F && data[i + 1] == 'E' && data[i + 2] == 'L' && |
| 45 | data[i + 3] == 'F') { |
| 46 | elfOffset = i; |
| 47 | found = true; |
| 48 | break; |
| 49 | } |
| 50 | } |
| 51 | } |
| 52 | |
| 53 | if (!found) { |
| 54 | LOG_ERROR(Common, "No ELF header found in data."); |
| 55 | return false; |
| 56 | } |
| 57 | |
| 58 | elfOffset_ = elfOffset; |
| 59 | const uint8_t *elfBase = data.data() + elfOffset_; |
| 60 | |
| 61 | // Parse ELF Header (64-bit) |
| 62 | uint64_t entryPoint = 0; |
| 63 | uint64_t phOff = 0; |
| 64 | uint16_t phEntSize = 0; |
| 65 | uint16_t phNum = 0; |
| 66 | uint64_t shOff = 0; |
| 67 | uint16_t shEntSize = 0; |
| 68 | uint16_t shNum = 0; |
| 69 | uint16_t shStrNdx = 0; |
| 70 | |
| 71 | if (data.size() < elfOffset + 64) |
| 72 | return false; |
| 73 | |
| 74 | std::memcpy(&entryPoint, elfBase + 0x18, 8); |
| 75 | std::memcpy(&phOff, elfBase + 0x20, 8); |
| 76 | std::memcpy(&phEntSize, elfBase + 0x36, 2); |
| 77 | std::memcpy(&phNum, elfBase + 0x38, 2); |
| 78 | std::memcpy(&shOff, elfBase + 0x28, 8); |
| 79 | std::memcpy(&shEntSize, elfBase + 0x3A, 2); |
| 80 | std::memcpy(&shNum, elfBase + 0x3C, 2); |
| 81 | std::memcpy(&shStrNdx, elfBase + 0x3E, 2); |
| 82 | // Capture entry point for later use in ExportProject |
| 83 | entryPoint_ = entryPoint; |
| 84 | |
| 85 | LOG_INFO(Common, |
| 86 | "ELF Found at offset 0x{:X}. Entry=0x{:X}, PHOff=0x{:X}, PHNum={}", |
| 87 | elfOffset, entryPoint, phOff, phNum); |
| 88 | |
| 89 | segments_.clear(); |
| 90 | uint64_t minVA = UINT64_MAX; |
| 91 | |
| 92 | // Parse Program Headers |
| 93 | for (int i = 0; i < phNum; ++i) { |
| 94 | size_t currentPhOffset = phOff + (i * phEntSize); |
| 95 | if (elfOffset + currentPhOffset + phEntSize > data.size()) |
| 96 | break; |
| 97 | |
| 98 | const uint8_t *phBase = elfBase + currentPhOffset; |
| 99 | |
| 100 | uint32_t p_type; |
| 101 | uint32_t p_flags; |
| 102 | uint64_t p_offset; |
| 103 | uint64_t p_vaddr; |
| 104 | uint64_t p_filesz; |
| 105 | uint64_t p_memsz; |
| 106 | |
| 107 | std::memcpy(&p_type, phBase, 4); |
| 108 | std::memcpy(&p_flags, phBase + 4, 4); |
| 109 | std::memcpy(&p_offset, phBase + 8, 8); |
| 110 | std::memcpy(&p_vaddr, phBase + 0x10, 8); |
| 111 | std::memcpy(&p_filesz, phBase + 0x20, 8); |
| 112 | std::memcpy(&p_memsz, phBase + 0x28, 8); |
| 113 | |
| 114 | if (p_type == 1) { // PT_LOAD |
| 115 | Segment seg; |
| 116 | seg.virtualAddress = p_vaddr; |
| 117 | // Absolute offset in rawData_ |
| 118 | seg.fileOffset = elfOffset + p_offset; |
| 119 | seg.size = p_filesz; |
| 120 | seg.memSize = p_memsz; // Include BSS zero-padding |
| 121 | segments_.push_back(seg); |
| 122 | |
| 123 | if (p_vaddr < minVA) |
| 124 | minVA = p_vaddr; |
| 125 | |
| 126 | LOG_INFO(Common, |
| 127 | "Segment: VA=0x{:X}, Offset=0x{:X} (Abs: 0x{:X}), Size=0x{:X}", |
| 128 | p_vaddr, p_offset, seg.fileOffset, p_filesz); |
| 129 | } else if (p_type == 3) { // PT_DYNAMIC |
| 130 | // Parse .dynamic section to find DT_INIT_ARRAY and DT_INIT_ARRAYSZ |
| 131 | LOG_INFO(Common, "PT_DYNAMIC found at VA=0x{:X}, offset=0x{:X}, size=0x{:X}", |
| 132 | p_vaddr, p_offset, p_filesz); |
| 133 | |
| 134 | if (elfOffset + p_offset + p_filesz <= data.size()) { |
| 135 | const uint8_t *dynBase = elfBase + p_offset; |
| 136 | // Parse dynamic entries (16 bytes each: tag + value) |
| 137 | for (size_t dynOff = 0; dynOff < p_filesz; dynOff += 16) { |
| 138 | uint64_t d_tag = 0; |
| 139 | uint64_t d_val = 0; |
| 140 | std::memcpy(&d_tag, dynBase + dynOff, 8); |
| 141 | std::memcpy(&d_val, dynBase + dynOff + 8, 8); |
| 142 | |
| 143 | if (d_tag == 0) break; // DT_NULL |
| 144 | if (d_tag == 25) { // DT_INIT_ARRAY |
| 145 | initArrayAddr_ = d_val; |
| 146 | LOG_INFO(Common, "DT_INIT_ARRAY found at VA=0x{:X}", initArrayAddr_); |
| 147 | } else if (d_tag == 27) { // DT_INIT_ARRAYSZ |
| 148 | initArraySize_ = d_val; |
| 149 | LOG_INFO(Common, "DT_INIT_ARRAYSZ = 0x{:X}", initArraySize_); |
| 150 | } else if (d_tag == 26) { // DT_FINI_ARRAY |
| 151 | finiArrayAddr_ = d_val; |
| 152 | LOG_INFO(Common, "DT_FINI_ARRAY found at VA=0x{:X}", finiArrayAddr_); |
| 153 | } else if (d_tag == 28) { // DT_FINI_ARRAYSZ |
| 154 | finiArraySize_ = d_val; |
| 155 | LOG_INFO(Common, "DT_FINI_ARRAYSZ = 0x{:X}", finiArraySize_); |
| 156 | } |
| 157 | } |
| 158 | } |
| 159 | } |
| 160 | } |
| 161 | |
| 162 | // Load raw data and set base |
| 163 | LoadBinary(data, minVA != UINT64_MAX ? minVA : 0x400000); |
| 164 | |
| 165 | if (segments_.empty()) { |
| 166 | LOG_WARNING(Common, "No PT_LOAD segments found. Falling back to flat " |
| 167 | "mapping from ELF base."); |
| 168 | Segment seg; |
| 169 | seg.virtualAddress = 0x400000; |
| 170 | seg.fileOffset = elfOffset; |
| 171 | seg.size = data.size() - elfOffset; |
| 172 | segments_.push_back(seg); |
| 173 | baseAddress_ = 0x400000; |
| 174 | } |
| 175 | |
| 176 | // Parse Section Headers to find .init_array and .fini_array |
| 177 | initArrayAddr_ = 0; |
| 178 | initArraySize_ = 0; |
| 179 | finiArrayAddr_ = 0; |
| 180 | finiArraySize_ = 0; |
| 181 | |
| 182 | LOG_INFO(Common, "Section Header Info: shNum={}, shOff=0x{:X}, shEntSize={}, shStrNdx={}", |
| 183 | shNum, shOff, shEntSize, shStrNdx); |
| 184 | |
| 185 | if (shNum > 0 && shOff > 0) { |
| 186 | // Get string table section header first |
| 187 | if (shStrNdx < shNum) { |
| 188 | size_t strShOffset = shOff + (shStrNdx * shEntSize); |
| 189 | if (elfOffset + strShOffset + shEntSize <= data.size()) { |
| 190 | const uint8_t *strShBase = elfBase + strShOffset; |
| 191 | uint64_t strShAddr = 0; |
| 192 | uint64_t strShOffset_file = 0; |
| 193 | uint64_t strShSize = 0; |
| 194 | std::memcpy(&strShAddr, strShBase + 0x10, 8); |
| 195 | std::memcpy(&strShOffset_file, strShBase + 0x18, 8); |
| 196 | std::memcpy(&strShSize, strShBase + 0x20, 8); |
| 197 | |
| 198 | LOG_INFO(Common, "String table: offset=0x{:X}, size=0x{:X}", strShOffset_file, strShSize); |
| 199 | |
| 200 | // Parse all section headers |
| 201 | for (int i = 0; i < shNum; ++i) { |
| 202 | size_t shOffset = shOff + (i * shEntSize); |
| 203 | if (elfOffset + shOffset + shEntSize > data.size()) |
| 204 | break; |
| 205 | |
| 206 | const uint8_t *shBase = elfBase + shOffset; |
| 207 | uint32_t sh_name = 0; |
| 208 | uint32_t sh_type = 0; |
| 209 | uint64_t sh_addr = 0; |
| 210 | uint64_t sh_offset = 0; |
| 211 | uint64_t sh_size = 0; |
| 212 | |
| 213 | std::memcpy(&sh_name, shBase, 4); |
| 214 | std::memcpy(&sh_type, shBase + 4, 4); |
| 215 | std::memcpy(&sh_addr, shBase + 0x10, 8); |
| 216 | std::memcpy(&sh_offset, shBase + 0x18, 8); |
| 217 | std::memcpy(&sh_size, shBase + 0x20, 8); |
| 218 | |
| 219 | // Get section name from string table |
| 220 | if (sh_name < strShSize && elfOffset + strShOffset_file + sh_name < data.size()) { |
| 221 | const char *name = reinterpret_cast<const char *>( |
| 222 | data.data() + elfOffset + strShOffset_file + sh_name); |
| 223 | std::string sectionName(name); |
| 224 | |
| 225 | if (sectionName == ".init_array") { |
| 226 | initArrayAddr_ = sh_addr; |
| 227 | initArraySize_ = sh_size; |
| 228 | LOG_INFO(Common, ".init_array found at VA=0x{:X}, size=0x{:X}", |
| 229 | initArrayAddr_, initArraySize_); |
| 230 | } else if (sectionName == ".fini_array") { |
| 231 | finiArrayAddr_ = sh_addr; |
| 232 | finiArraySize_ = sh_size; |
| 233 | LOG_INFO(Common, ".fini_array found at VA=0x{:X}, size=0x{:X}", |
| 234 | finiArrayAddr_, finiArraySize_); |
| 235 | } |
| 236 | } |
| 237 | } |
| 238 | } |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | return true; |
| 243 | } |
| 244 | |
| 245 | bool DecompilerContext::VirtualAddressToFileOffset(uint64_t va, |
| 246 | uint64_t &offset) const { |
| 247 | for (const auto &seg : segments_) { |
| 248 | if (va >= seg.virtualAddress && va < seg.virtualAddress + seg.size) { |
| 249 | offset = seg.fileOffset + (va - seg.virtualAddress); |
| 250 | return true; |
| 251 | } |
| 252 | } |
| 253 | if (segments_.empty() && va >= baseAddress_) { |
| 254 | offset = va - baseAddress_; |
| 255 | return true; |
| 256 | } |
| 257 | return false; |
| 258 | } |
| 259 | |
| 260 | void DecompilerContext::Analyze() { |
| 261 | if (rawData_.empty()) |
| 262 | return; |
| 263 | |
| 264 | functions_.clear(); |
| 265 | std::set<uint64_t> visitedGlobal; |
| 266 | |
| 267 | uint64_t entryPoint = baseAddress_; |
| 268 | if (rawData_.size() >= elfOffset_ + 0x20 && rawData_[elfOffset_] == 0x7F && |
| 269 | rawData_[elfOffset_ + 1] == 'E') { |
| 270 | std::memcpy(&entryPoint, rawData_.data() + elfOffset_ + 0x18, 8); |
| 271 | } |
| 272 | |
| 273 | std::queue<uint64_t> functionQueue; |
| 274 | functionQueue.push(entryPoint); |
| 275 | visitedGlobal.insert(entryPoint); |
| 276 | |
| 277 | uint64_t epOffset = 0; |
| 278 | if (VirtualAddressToFileOffset(entryPoint, epOffset)) { |
| 279 | if (epOffset + 8 > rawData_.size()) { |
| 280 | LOG_ERROR(Common, "Entry Point 0x{:X} is out of file bounds!", |
| 281 | entryPoint); |
| 282 | return; |
| 283 | } |
| 284 | } else { |
| 285 | LOG_ERROR(Common, "Could not map Entry Point 0x{:X} to file offset!", |
| 286 | entryPoint); |
| 287 | if (entryPoint >= baseAddress_) |
| 288 | epOffset = entryPoint - baseAddress_; |
| 289 | } |
| 290 | |
| 291 | int functionsAnalyzed = 0; |
| 292 | const int MAX_FUNCTIONS = 50000; |
| 293 | |
| 294 | // ┌─────────────────────────────────────────────────────────────────────────┐ |
| 295 | // │ Run Symbol Analysis early to recover function names from ELF │ |
| 296 | // │ This populates symbolDatabase_ with names from .dynsym / .rela.plt │ |
| 297 | // └─────────────────────────────────────────────────────────────────────────┘ |
| 298 | { |
| 299 | auto symbolAnalysis = std::make_shared<Analysis::SymbolAnalysis>( |
| 300 | rawData_, baseAddress_, symbolDatabase_); |
| 301 | symbolAnalysis->analyze(); |
| 302 | LOG_INFO(Common, "Symbol Analysis complete: {} symbols loaded.", |
| 303 | symbolDatabase_ ? symbolDatabase_->getSymbols().size() : 0); |
| 304 | } |
| 305 | |
| 306 | LOG_INFO(Common, "Starting Function Discovery at 0x{:X}...", entryPoint); |
| 307 | |
| 308 | // ┌─────────────────────────────────────────────────────────────────────────┐ |
| 309 | // │ PROLOGUE SCAN FOR ALL FUNCTIONS │ |
| 310 | // │ Many functions are only called via vtables or indirect jumps. │ |
| 311 | // │ Scan .text for common x86-64 function prologues: │ |
| 312 | // │ push rbp; mov rbp, rsp -> 55 48 89 E5 │ |
| 313 | // │ push rbx -> 53 │ |
| 314 | // │ sub rsp, N -> 48 83 EC XX / 48 81 EC XX XX XX XX │ |
| 315 | // └─────────────────────────────────────────────────────────────────────────┘ |
| 316 | LOG_INFO(Common, "Scanning segments for function prologues..."); |
| 317 | progress_.currentPhase = "scanning"; |
| 318 | progress_.prologuesFound = 0; |
| 319 | progress_.functionsAnalyzed = 0; |
| 320 | progress_.isComplete = false; |
| 321 | int prologuesFound = 0; |
| 322 | |
| 323 | for (const auto &seg : segments_) { |
| 324 | if (seg.size < 8) |
| 325 | continue; |
| 326 | uint64_t scanEnd = |
| 327 | std::min(seg.fileOffset + seg.size, (uint64_t)rawData_.size()) - 4; |
| 328 | |
| 329 | for (uint64_t off = seg.fileOffset; off < scanEnd; ++off) { |
| 330 | uint64_t funcVA = seg.virtualAddress + (off - seg.fileOffset); |
| 331 | |
| 332 | if (visitedGlobal.count(funcVA)) |
| 333 | continue; |
| 334 | |
| 335 | // Pattern 1: push rbp; mov rbp, rsp (0x55 0x48 0x89 0xE5) |
| 336 | bool match = (rawData_[off] == 0x55 && rawData_[off + 1] == 0x48 && |
| 337 | rawData_[off + 2] == 0x89 && rawData_[off + 3] == 0xE5); |
| 338 | |
| 339 | // Pattern 2: push rbp; mov rsp variants |
| 340 | if (!match && rawData_[off] == 0x55) { |
| 341 | // Check for sub rsp, imm8 following (48 83 EC XX) |
| 342 | if (off + 5 < scanEnd && rawData_[off + 1] == 0x48 && |
| 343 | rawData_[off + 2] == 0x83 && rawData_[off + 3] == 0xEC) { |
| 344 | match = true; |
| 345 | } |
| 346 | } |
| 347 | |
| 348 | if (match) { |
| 349 | functionQueue.push(funcVA); |
| 350 | visitedGlobal.insert(funcVA); |
| 351 | prologuesFound++; |
| 352 | progress_.prologuesFound = prologuesFound; |
| 353 | } |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | LOG_INFO(Common, "Found {} candidate functions via prologue scan.", |
| 358 | prologuesFound); |
| 359 | |
| 360 | while (!functionQueue.empty() && functionsAnalyzed < MAX_FUNCTIONS) { |
| 361 | uint64_t funcAddr = functionQueue.front(); |
| 362 | functionQueue.pop(); |
| 363 | |
| 364 | auto func = std::make_shared<IR::Function>(); |
| 365 | func->address = funcAddr; |
| 366 | |
| 367 | // ┌─────────────────────────────────────────────────────────────────────────┐ |
| 368 | // │ Try to recover function name from ELF symbols │ |
| 369 | // └─────────────────────────────────────────────────────────────────────────┘ |
| 370 | if (symbolDatabase_) { |
| 371 | auto sym = symbolDatabase_->getSymbol(funcAddr); |
| 372 | if (sym && !sym->name.empty() && |
| 373 | sym->name.find("sub_") == std::string::npos) { |
| 374 | func->name = sym->name; |
| 375 | } |
| 376 | } |
| 377 | if (func->name.empty()) { |
| 378 | std::stringstream nameSS; |
| 379 | nameSS << "sub_" << std::hex << funcAddr; |
| 380 | func->name = nameSS.str(); |
| 381 | } |
| 382 | |
| 383 | func->signature = "void " + func->name + "()"; |
| 384 | func->category = IR::Function::Category::GameLogic; |
| 385 | |
| 386 | csh handle; |
| 387 | if (cs_open(CS_ARCH_X86, CS_MODE_64, &handle) != CS_ERR_OK) |
| 388 | continue; |
| 389 | cs_option(handle, CS_OPT_DETAIL, CS_OPT_ON); |
| 390 | |
| 391 | std::queue<uint64_t> blockQueue; |
| 392 | std::set<uint64_t> visitedBlocks; |
| 393 | |
| 394 | blockQueue.push(funcAddr); |
| 395 | visitedBlocks.insert(funcAddr); |
| 396 | |
| 397 | while (!blockQueue.empty()) { |
| 398 | uint64_t blockAddr = blockQueue.front(); |
| 399 | blockQueue.pop(); |
| 400 | |
| 401 | auto bb = std::make_shared<IR::BasicBlock>(); |
| 402 | bb->id = blockAddr; |
| 403 | bb->startAddress = blockAddr; |
| 404 | |
| 405 | uint64_t currentAddr = blockAddr; |
| 406 | bool blockEnded = false; |
| 407 | int zeroCount = 0; |
| 408 | |
| 409 | for (int i = 0; i < 1000 && !blockEnded; ++i) { |
| 410 | uint64_t fileOffset = 0; |
| 411 | if (!VirtualAddressToFileOffset(currentAddr, fileOffset)) |
| 412 | break; |
| 413 | |
| 414 | if (fileOffset >= rawData_.size()) |
| 415 | break; |
| 416 | |
| 417 | const uint8_t *code = rawData_.data() + fileOffset; |
| 418 | size_t code_size = rawData_.size() - fileOffset; |
| 419 | |
| 420 | if (code[0] == 0x00 && code[1] == 0x00) { |
| 421 | zeroCount++; |
| 422 | if (zeroCount > 8) { |
| 423 | blockEnded = true; |
| 424 | break; |
| 425 | } |
| 426 | } else { |
| 427 | zeroCount = 0; |
| 428 | } |
| 429 | |
| 430 | cs_insn *insn; |
| 431 | size_t count = cs_disasm(handle, code, 15, currentAddr, 1, &insn); |
| 432 | |
| 433 | if (count > 0) { |
| 434 | IR::Instruction instr; |
| 435 | instr.address = insn[0].address; |
| 436 | |
| 437 | instr.opcode = IR::OpCode::NOP; |
| 438 | switch (insn[0].id) { |
| 439 | case X86_INS_MOV: |
| 440 | instr.opcode = IR::OpCode::MOV; |
| 441 | break; |
| 442 | case X86_INS_MOVABS: |
| 443 | instr.opcode = IR::OpCode::MOV; |
| 444 | break; |
| 445 | case X86_INS_MOVAPS: |
| 446 | instr.opcode = IR::OpCode::MOV; |
| 447 | break; |
| 448 | case X86_INS_MOVUPS: |
| 449 | instr.opcode = IR::OpCode::MOV; |
| 450 | break; |
| 451 | case X86_INS_MOVDQA: |
| 452 | instr.opcode = IR::OpCode::MOV; |
| 453 | break; |
| 454 | case X86_INS_MOVDQU: |
| 455 | instr.opcode = IR::OpCode::MOV; |
| 456 | break; |
| 457 | case X86_INS_ADD: |
| 458 | instr.opcode = IR::OpCode::ADD; |
| 459 | break; |
| 460 | case X86_INS_SUB: |
| 461 | instr.opcode = IR::OpCode::SUB; |
| 462 | break; |
| 463 | case X86_INS_RET: |
| 464 | instr.opcode = IR::OpCode::RET; |
| 465 | break; |
| 466 | case X86_INS_CALL: |
| 467 | instr.opcode = IR::OpCode::CALL; |
| 468 | break; |
| 469 | case X86_INS_JMP: |
| 470 | instr.opcode = IR::OpCode::JMP; |
| 471 | break; |
| 472 | case X86_INS_JE: |
| 473 | case X86_INS_JCXZ: |
| 474 | case X86_INS_JECXZ: |
| 475 | case X86_INS_JRCXZ: |
| 476 | instr.opcode = IR::OpCode::JE; |
| 477 | break; |
| 478 | case X86_INS_JNE: |
| 479 | instr.opcode = IR::OpCode::JNE; |
| 480 | break; |
| 481 | case X86_INS_JG: |
| 482 | instr.opcode = IR::OpCode::JG; |
| 483 | break; |
| 484 | case X86_INS_JGE: |
| 485 | instr.opcode = IR::OpCode::JGE; |
| 486 | break; |
| 487 | case X86_INS_JL: |
| 488 | instr.opcode = IR::OpCode::JL; |
| 489 | break; |
| 490 | case X86_INS_JLE: |
| 491 | instr.opcode = IR::OpCode::JLE; |
| 492 | break; |
| 493 | case X86_INS_JA: |
| 494 | instr.opcode = IR::OpCode::JA; |
| 495 | break; |
| 496 | case X86_INS_JAE: |
| 497 | instr.opcode = IR::OpCode::JAE; |
| 498 | break; |
| 499 | case X86_INS_JB: |
| 500 | instr.opcode = IR::OpCode::JB; |
| 501 | break; |
| 502 | case X86_INS_JBE: |
| 503 | instr.opcode = IR::OpCode::JBE; |
| 504 | break; |
| 505 | case X86_INS_JS: |
| 506 | instr.opcode = IR::OpCode::JS; |
| 507 | break; |
| 508 | case X86_INS_JNS: |
| 509 | instr.opcode = IR::OpCode::JNS; |
| 510 | break; |
| 511 | case X86_INS_JO: |
| 512 | instr.opcode = IR::OpCode::JO; |
| 513 | break; |
| 514 | case X86_INS_JNO: |
| 515 | instr.opcode = IR::OpCode::JNO; |
| 516 | break; |
| 517 | case X86_INS_CMP: |
| 518 | instr.opcode = IR::OpCode::CMP; |
| 519 | break; |
| 520 | case X86_INS_LEA: |
| 521 | instr.opcode = IR::OpCode::LEA; |
| 522 | break; |
| 523 | |
| 524 | case X86_INS_MOVSX: |
| 525 | instr.opcode = IR::OpCode::MOVSX; |
| 526 | break; |
| 527 | case X86_INS_MOVSXD: |
| 528 | instr.opcode = IR::OpCode::MOVSX; |
| 529 | break; |
| 530 | case X86_INS_MOVZX: |
| 531 | instr.opcode = IR::OpCode::MOVZX; |
| 532 | break; |
| 533 | case X86_INS_MOVSD: |
| 534 | instr.opcode = IR::OpCode::MOV; |
| 535 | break; // Treat as MOV |
| 536 | case X86_INS_MOVSS: |
| 537 | instr.opcode = IR::OpCode::MOV; |
| 538 | break; // Treat as MOV |
| 539 | |
| 540 | case X86_INS_BSWAP: |
| 541 | instr.opcode = IR::OpCode::BSWAP; |
| 542 | break; |
| 543 | case X86_INS_FISTTP: |
| 544 | instr.opcode = IR::OpCode::FISTTP; |
| 545 | break; |
| 546 | case X86_INS_LEAVE: |
| 547 | instr.opcode = IR::OpCode::LEAVE; |
| 548 | break; |
| 549 | case X86_INS_INT: |
| 550 | instr.opcode = IR::OpCode::INT; |
| 551 | break; // Software interrupt, often for debug/syscall |
| 552 | |
| 553 | // AVX / VEX Support |
| 554 | case X86_INS_VMOVSS: |
| 555 | instr.opcode = IR::OpCode::MOV; |
| 556 | break; |
| 557 | case X86_INS_VMOVSD: |
| 558 | instr.opcode = IR::OpCode::MOV; |
| 559 | break; |
| 560 | case X86_INS_VMOVAPS: |
| 561 | instr.opcode = IR::OpCode::MOV; |
| 562 | break; |
| 563 | case X86_INS_VMOVUPS: |
| 564 | instr.opcode = IR::OpCode::MOV; |
| 565 | break; |
| 566 | case X86_INS_VMOVAPD: |
| 567 | instr.opcode = IR::OpCode::MOV; |
| 568 | break; |
| 569 | case X86_INS_VMOVUPD: |
| 570 | instr.opcode = IR::OpCode::MOV; |
| 571 | break; |
| 572 | case X86_INS_VADDSS: |
| 573 | instr.opcode = IR::OpCode::ADD; |
| 574 | break; |
| 575 | case X86_INS_VADDSD: |
| 576 | instr.opcode = IR::OpCode::ADD; |
| 577 | break; |
| 578 | case X86_INS_VSUBSS: |
| 579 | instr.opcode = IR::OpCode::SUB; |
| 580 | break; |
| 581 | case X86_INS_VSUBSD: |
| 582 | instr.opcode = IR::OpCode::SUB; |
| 583 | break; |
| 584 | case X86_INS_VMULSS: |
| 585 | instr.opcode = IR::OpCode::MUL; |
| 586 | break; |
| 587 | case X86_INS_VMULSD: |
| 588 | instr.opcode = IR::OpCode::MUL; |
| 589 | break; |
| 590 | case X86_INS_VDIVSS: |
| 591 | instr.opcode = IR::OpCode::DIV; |
| 592 | break; |
| 593 | case X86_INS_VDIVSD: |
| 594 | instr.opcode = IR::OpCode::DIV; |
| 595 | break; |
| 596 | case X86_INS_VXORPS: |
| 597 | instr.opcode = IR::OpCode::XOR; |
| 598 | break; |
| 599 | case X86_INS_VXORPD: |
| 600 | instr.opcode = IR::OpCode::XOR; |
| 601 | break; |
| 602 | case X86_INS_VANDPS: |
| 603 | instr.opcode = IR::OpCode::AND; |
| 604 | break; |
| 605 | case X86_INS_VANDPD: |
| 606 | instr.opcode = IR::OpCode::AND; |
| 607 | break; |
| 608 | case X86_INS_VORPS: |
| 609 | instr.opcode = IR::OpCode::OR; |
| 610 | break; |
| 611 | case X86_INS_VORPD: |
| 612 | instr.opcode = IR::OpCode::OR; |
| 613 | break; |
| 614 | case X86_INS_VUCOMISS: |
| 615 | instr.opcode = IR::OpCode::CMP; |
| 616 | break; |
| 617 | case X86_INS_VUCOMISD: |
| 618 | instr.opcode = IR::OpCode::CMP; |
| 619 | break; |
| 620 | case X86_INS_VPCMPGTB: |
| 621 | case X86_INS_VPCMPGTW: |
| 622 | case X86_INS_VPCMPGTD: |
| 623 | case X86_INS_VPCMPGTQ: |
| 624 | instr.opcode = IR::OpCode::CMP; |
| 625 | break; |
| 626 | |
| 627 | // Stack |
| 628 | case X86_INS_PUSH: |
| 629 | instr.opcode = IR::OpCode::PUSH; |
| 630 | break; |
| 631 | case X86_INS_POP: |
| 632 | instr.opcode = IR::OpCode::POP; |
| 633 | break; |
| 634 | |
| 635 | // Logic/Shift |
| 636 | case X86_INS_AND: |
| 637 | instr.opcode = IR::OpCode::AND; |
| 638 | break; |
| 639 | case X86_INS_OR: |
| 640 | instr.opcode = IR::OpCode::OR; |
| 641 | break; |
| 642 | case X86_INS_XOR: |
| 643 | instr.opcode = IR::OpCode::XOR; |
| 644 | break; |
| 645 | case X86_INS_SHL: |
| 646 | instr.opcode = IR::OpCode::SHL; |
| 647 | break; |
| 648 | case X86_INS_SHR: |
| 649 | instr.opcode = IR::OpCode::SHR; |
| 650 | break; |
| 651 | case X86_INS_SAL: |
| 652 | instr.opcode = IR::OpCode::SHL; |
| 653 | break; |
| 654 | case X86_INS_SAR: |
| 655 | instr.opcode = IR::OpCode::SHR; |
| 656 | break; // Treat arithmetic shift as logical for now or add SAR |
| 657 | case X86_INS_TEST: |
| 658 | instr.opcode = IR::OpCode::AND; |
| 659 | break; // TEST is mostly AND for flags |
| 660 | |
| 661 | // Arithmetic |
| 662 | case X86_INS_INC: |
| 663 | instr.opcode = IR::OpCode::ADD; |
| 664 | break; // Handled specially or mapped to ADD 1 |
| 665 | case X86_INS_DEC: |
| 666 | instr.opcode = IR::OpCode::SUB; |
| 667 | break; // Handled specially or mapped to SUB 1 |
| 668 | case X86_INS_NEG: |
| 669 | instr.opcode = IR::OpCode::SUB; |
| 670 | break; // 0 - x? Need proper unary op |
| 671 | case X86_INS_NOT: |
| 672 | instr.opcode = IR::OpCode::XOR; |
| 673 | break; // ~x (Need unary op really) |
| 674 | case X86_INS_MUL: |
| 675 | instr.opcode = IR::OpCode::MUL; |
| 676 | break; |
| 677 | case X86_INS_IMUL: |
| 678 | instr.opcode = IR::OpCode::MUL; |
| 679 | break; |
| 680 | case X86_INS_DIV: |
| 681 | instr.opcode = IR::OpCode::DIV; |
| 682 | break; |
| 683 | case X86_INS_IDIV: |
| 684 | instr.opcode = IR::OpCode::DIV; |
| 685 | break; |
| 686 | |
| 687 | default: |
| 688 | break; |
| 689 | } |
| 690 | instr.disassembly = |
| 691 | std::string(insn[0].mnemonic) + " " + insn[0].op_str; |
| 692 | |
| 693 | if (insn[0].detail) { |
| 694 | for (int j = 0; j < insn[0].detail->x86.op_count; ++j) { |
| 695 | const auto &op = insn[0].detail->x86.operands[j]; |
| 696 | IR::Operand irOp; |
| 697 | if (op.type == X86_OP_REG) { |
| 698 | irOp.type = IR::Operand::Type::Register; |
| 699 | irOp.value = op.reg; |
| 700 | irOp.regName = cs_reg_name(handle, op.reg); |
| 701 | } else if (op.type == X86_OP_IMM) { |
| 702 | irOp.type = IR::Operand::Type::Immediate; |
| 703 | irOp.value = op.imm; |
| 704 | } else if (op.type == X86_OP_MEM) { |
| 705 | irOp.type = IR::Operand::Type::Memory; |
| 706 | // Store raw info for advanced lifting |
| 707 | irOp.memBase = op.mem.base; |
| 708 | irOp.memBaseName = (op.mem.base != X86_REG_INVALID) |
| 709 | ? cs_reg_name(handle, op.mem.base) |
| 710 | : ""; |
| 711 | irOp.memDisp = op.mem.disp; |
| 712 | |
| 713 | if (op.mem.base == X86_REG_RIP) { |
| 714 | irOp.value = insn[0].address + insn[0].size + op.mem.disp; |
| 715 | irOp.name = "RIP"; |
| 716 | } else if (op.mem.base == X86_REG_RSP) { |
| 717 | irOp.name = "RSP"; |
| 718 | irOp.value = (uint64_t)op.mem.disp; |
| 719 | } else if (op.mem.base == X86_REG_RBP) { |
| 720 | irOp.name = "RBP"; |
| 721 | irOp.value = (uint64_t)op.mem.disp; |
| 722 | } else { |
| 723 | irOp.value = 0; |
| 724 | irOp.name = "MEM"; |
| 725 | } |
| 726 | } else { |
| 727 | irOp.type = IR::Operand::Type::Variable; |
| 728 | } |
| 729 | instr.operands.push_back(irOp); |
| 730 | } |
| 731 | } |
| 732 | |
| 733 | bb->instructions.push_back(instr); |
| 734 | |
| 735 | bool isBranch = false; |
| 736 | bool isCall = (insn[0].id == X86_INS_CALL); |
| 737 | bool isRet = (insn[0].id == X86_INS_RET); |
| 738 | |
| 739 | if (insn[0].id == X86_INS_JMP || |
| 740 | (insn[0].id >= X86_INS_JA && insn[0].id <= X86_INS_JS)) { |
| 741 | isBranch = true; |
| 742 | } |
| 743 | |
| 744 | if (isCall) { |
| 745 | if (insn[0].detail->x86.op_count > 0 && |
| 746 | insn[0].detail->x86.operands[0].type == X86_OP_IMM) { |
| 747 | uint64_t target = insn[0].detail->x86.operands[0].imm; |
| 748 | if (visitedGlobal.find(target) == visitedGlobal.end()) { |
| 749 | visitedGlobal.insert(target); |
| 750 | functionQueue.push(target); |
| 751 | } |
| 752 | } |
| 753 | } else if (isBranch) { |
| 754 | blockEnded = true; |
| 755 | if (insn[0].detail->x86.op_count > 0 && |
| 756 | insn[0].detail->x86.operands[0].type == X86_OP_IMM) { |
| 757 | uint64_t target = insn[0].detail->x86.operands[0].imm; |
| 758 | bb->successors.push_back(target); |
| 759 | if (visitedBlocks.find(target) == visitedBlocks.end()) { |
| 760 | visitedBlocks.insert(target); |
| 761 | blockQueue.push(target); |
| 762 | } |
| 763 | } |
| 764 | if (insn[0].id != X86_INS_JMP) { |
| 765 | uint64_t nextAddr = insn[0].address + insn[0].size; |
| 766 | bb->successors.push_back(nextAddr); |
| 767 | if (visitedBlocks.find(nextAddr) == visitedBlocks.end()) { |
| 768 | visitedBlocks.insert(nextAddr); |
| 769 | blockQueue.push(nextAddr); |
| 770 | } |
| 771 | } |
| 772 | } else if (isRet) { |
| 773 | blockEnded = true; |
| 774 | } |
| 775 | |
| 776 | currentAddr += insn[0].size; |
| 777 | cs_free(insn, count); |
| 778 | } else { |
| 779 | blockEnded = true; |
| 780 | } |
| 781 | } |
| 782 | bb->endAddress = currentAddr; |
| 783 | func->basicBlocks.push_back(bb); |
| 784 | } |
| 785 | |
| 786 | cs_close(&handle); |
| 787 | functions_.push_back(func); |
| 788 | functionsAnalyzed++; |
| 789 | progress_.functionsAnalyzed = functionsAnalyzed; |
| 790 | progress_.currentPhase = "analyzing"; |
| 791 | } |
| 792 | |
| 793 | LOG_INFO(Common, "Discovered {} functions.", functions_.size()); |
| 794 | progress_.currentPhase = "complete"; |
| 795 | progress_.isComplete = true; |
| 796 | isAnalyzed_ = true; |
| 797 | } |
| 798 | |
| 799 | void DecompilerContext::AnalyzeFunction(uint64_t startAddress, |
| 800 | std::set<uint64_t> &visitedGlobal) {} |
| 801 | |
| 802 | void DecompilerContext::ExportProject(const std::string &outPath) { |
| 803 | std::filesystem::path root(outPath); |
| 804 | std::filesystem::create_directories(root / "include"); |
| 805 | std::filesystem::create_directories(root / "src"); |
| 806 | |
| 807 | LOG_INFO(Common, "Starting full project analysis and export..."); |
| 808 | |
| 809 | auto symbols = std::make_shared<Analysis::SymbolAnalysis>( |
| 810 | rawData_, baseAddress_, symbolDatabase_); |
| 811 | symbols->analyze(); |
| 812 | |
| 813 | std::vector<std::shared_ptr<AST::FunctionAST>> analyzedASTs; |
| 814 | analyzedASTs.reserve(functions_.size()); |
| 815 | |
| 816 | // ┌─────────────────────────────────────────────────────────────────────────┐ |
| 817 | // │ PASS 1: Full Analysis with progress output and timeout protection │ |
| 818 | // └─────────────────────────────────────────────────────────────────────────┘ |
| 819 | size_t skippedCount = 0; |
| 820 | auto globalStart = std::chrono::steady_clock::now(); |
| 821 | |
| 822 | for (size_t i = 0; i < functions_.size(); ++i) { |
| 823 | auto func = functions_[i]; |
| 824 | |
| 825 | // Progress output every 100 functions |
| 826 | if (i % 100 == 0 || i == functions_.size() - 1) { |
| 827 | auto elapsed = std::chrono::steady_clock::now() - globalStart; |
| 828 | auto secs = |
| 829 | std::chrono::duration_cast<std::chrono::seconds>(elapsed).count(); |
| 830 | std::cout << "\r[" << (i + 1) << "/" << functions_.size() << "] " |
| 831 | << "Analyzing... (" << secs << "s, " << skippedCount |
| 832 | << " skipped)" << std::flush; |
| 833 | } |
| 834 | |
| 835 | size_t funcSize = 0; |
| 836 | for (const auto &bb : func->basicBlocks) { |
| 837 | if (!bb->instructions.empty()) |
| 838 | funcSize += (bb->endAddress - bb->startAddress); |
| 839 | } |
| 840 | |
| 841 | // Skip very large functions (likely data, not code) |
| 842 | if (funcSize > 100000) { |
| 843 | LOG_INFO(Common, "Skipping huge function {} ({}KB)", func->name, |
| 844 | funcSize / 1024); |
| 845 | skippedCount++; |
| 846 | continue; |
| 847 | } |
| 848 | |
| 849 | auto funcStart = std::chrono::steady_clock::now(); |
| 850 | |
| 851 | try { |
| 852 | auto dom = std::make_shared<Analysis::DominatorAnalysis>(); |
| 853 | dom->analyze(func); |
| 854 | |
| 855 | // Timeout check after dominator analysis |
| 856 | auto elapsed = std::chrono::steady_clock::now() - funcStart; |
| 857 | if (elapsed > std::chrono::seconds(5)) { |
| 858 | LOG_INFO(Common, "Skipping {} (dominator timeout)", func->name); |
| 859 | skippedCount++; |
| 860 | continue; |
| 861 | } |
| 862 | |
| 863 | // 3. Variable Lifting |
| 864 | auto lifter = std::make_shared<Lifter::VariableAnalysis>(func); |
| 865 | lifter->analyze(); |
| 866 | |
| 867 | // 4. Structural Analysis |
| 868 | Analysis::StructuralAnalysis structural(func, dom, symbols, lifter); |
| 869 | auto ast = structural.analyze(); |
| 870 | |
| 871 | // Timeout check after structural analysis |
| 872 | elapsed = std::chrono::steady_clock::now() - funcStart; |
| 873 | if (elapsed > std::chrono::seconds(10)) { |
| 874 | LOG_INFO(Common, "Skipping {} (structural timeout)", func->name); |
| 875 | skippedCount++; |
| 876 | continue; |
| 877 | } |
| 878 | |
| 879 | if (!ast->body || ast->body->statements.empty()) { |
| 880 | // Empty function - still add stub |
| 881 | } |
| 882 | |
| 883 | // Apply locals to AST |
| 884 | lifter->applyToAST(ast); |
| 885 | |
| 886 | // 4b. Apply User Types |
| 887 | for (auto &local : ast->locals) { |
| 888 | auto userType = GetUserVarType(func->address, local.stackOffset); |
| 889 | if (userType) { |
| 890 | local.complexType = userType; |
| 891 | } |
| 892 | } |
| 893 | |
| 894 | Analysis::DataFlowAnalysis dataflow(ast); |
| 895 | dataflow.analyze(); |
| 896 | |
| 897 | Analysis::MemberAccessAnalysis memberAccess(typeManager_); |
| 898 | memberAccess.analyze(ast); |
| 899 | |
| 900 | analyzedASTs.push_back(ast); |
| 901 | SetFunctionParamCount(func->address, (int)ast->parameters.size()); |
| 902 | |
| 903 | } catch (const std::exception &e) { |
| 904 | LOG_ERROR(Common, "Exception analyzing {}: {}", func->name, e.what()); |
| 905 | skippedCount++; |
| 906 | continue; |
| 907 | } |
| 908 | } |
| 909 | |
| 910 | std::cout << std::endl; // Newline after progress |
| 911 | LOG_INFO(Common, "Analysis complete: {} functions OK, {} skipped", |
| 912 | analyzedASTs.size(), skippedCount); |
| 913 | |
| 914 | // --- PASS 2: Export Types --- |
| 915 | { |
| 916 | std::ofstream hTypes(root / "include" / "types.h"); |
| 917 | hTypes << "#pragma once\n"; |
| 918 | hTypes << "#include <cstdint>\n\n"; |
| 919 | hTypes << "#if defined(__GNUC__) || defined(__clang__)\n"; |
| 920 | hTypes << "#define _byteswap_uint64 __builtin_bswap64\n"; |
| 921 | hTypes << "#endif\n\n"; |
| 922 | hTypes << "static inline double as_double(uint64_t i) { return " |
| 923 | "*(double*)&i; }\n"; |
| 924 | hTypes |
| 925 | << "static inline float as_float(uint32_t i) { return *(float*)&i; }\n"; |
| 926 | hTypes << "static inline uint64_t as_uint64(double d) { return " |
| 927 | "*(uint64_t*)&d; }\n"; |
| 928 | hTypes << "static inline uint32_t as_uint32(float f) { return " |
| 929 | "*(uint32_t*)&f; }\n\n"; |
| 930 | |
| 931 | for (const auto &[name, type] : typeManager_->getAllTypes()) { |
| 932 | if (type->getKind() == Analysis::Type::Kind::Struct) { |
| 933 | hTypes << "struct " << name << ";\n"; |
| 934 | } |
| 935 | } |
| 936 | hTypes << "\n"; |
| 937 | |
| 938 | for (const auto &[name, type] : typeManager_->getAllTypes()) { |
| 939 | if (type->getKind() == Analysis::Type::Kind::Struct) { |
| 940 | auto structType = std::dynamic_pointer_cast<Analysis::StructType>(type); |
| 941 | hTypes << "struct " << name << " {\n"; |
| 942 | for (const auto &member : structType->getMembers()) { |
| 943 | hTypes << " " << member.type->toString() << " " << member.name |
| 944 | << "; // Offset: " << member.offset << "\n"; |
| 945 | } |
| 946 | hTypes << "};\n\n"; |
| 947 | } |
| 948 | } |
| 949 | } |
| 950 | |
| 951 | // --- PASS 3: Export Functions --- |
| 952 | { |
| 953 | std::map<std::string, std::shared_ptr<AST::FunctionAST>> nameToAst; |
| 954 | for (auto &ast : analyzedASTs) { |
| 955 | nameToAst[ast->name] = ast; |
| 956 | } |
| 957 | |
| 958 | std::ofstream hFuncs(root / "include" / "functions.h"); |
| 959 | hFuncs << "#pragma once\n"; |
| 960 | hFuncs << "#include \"types.h\"\n"; |
| 961 | hFuncs << "#include <cstdint>\n\n"; |
| 962 | |
| 963 | // All functions use uniform int64_t(a1..a6) signature to match CppEmitter output. |
| 964 | // Default args allow call sites with fewer arguments to compile cleanly. |
| 965 | static const char* kFuncSig = |
| 966 | "(int64_t a1 = 0, int64_t a2 = 0, int64_t a3 = 0, " |
| 967 | "int64_t a4 = 0, int64_t a5 = 0, int64_t a6 = 0);\n"; |
| 968 | for (const auto &func : functions_) { |
| 969 | hFuncs << "int64_t " << func->name << kFuncSig; |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | // --- PASS 4: Export Source --- |
| 974 | std::cout << "Exporting source code..." << std::endl; |
| 975 | LOG_INFO(Common, "Starting source export of {} functions", |
| 976 | analyzedASTs.size()); |
| 977 | |
| 978 | int fileCounter = 0; |
| 979 | int funcsPerFile = 50; |
| 980 | |
| 981 | for (size_t i = 0; i < analyzedASTs.size(); i += funcsPerFile) { |
| 982 | if (i % 1000 == 0) { |
| 983 | std::cout << "\rExporting chunks... " << i << "/" << analyzedASTs.size() |
| 984 | << std::flush; |
| 985 | } |
| 986 | |
| 987 | std::string filename = "segment_" + std::to_string(fileCounter++) + ".cpp"; |
| 988 | std::ofstream cppFile(root / "src" / filename); |
| 989 | |
| 990 | cppFile << "#include \"../include/functions.h\"\n"; |
| 991 | cppFile << "#include \"../include/types.h\"\n"; |
| 992 | cppFile << "#include \"../include/globals.h\"\n\n"; |
| 993 | |
| 994 | for (size_t j = i; j < i + funcsPerFile && j < analyzedASTs.size(); ++j) { |
| 995 | auto ast = analyzedASTs[j]; |
| 996 | try { |
| 997 | Codegen::CppEmitter emitter; |
| 998 | cppFile << emitter.generate(ast) << "\n\n"; |
| 999 | } catch (const std::exception &e) { |
| 1000 | LOG_ERROR(Common, "Failed to generate code for {}: {}", ast->name, |
| 1001 | e.what()); |
| 1002 | cppFile << "// Error generating " << ast->name << ": " << e.what() |
| 1003 | << "\n\n"; |
| 1004 | } catch (...) { |
| 1005 | LOG_ERROR(Common, "Unknown failure generating code for {}", ast->name); |
| 1006 | cppFile << "// Unknown error generating " << ast->name << "\n\n"; |
| 1007 | } |
| 1008 | } |
| 1009 | } |
| 1010 | std::cout << std::endl; |
| 1011 | |
| 1012 | // --- PASS 5: Export Globals --- |
| 1013 | { |
| 1014 | std::ofstream hGlobals(root / "include" / "globals.h"); |
| 1015 | std::ofstream cppGlobals(root / "src" / "globals.cpp"); |
| 1016 | |
| 1017 | hGlobals << "#pragma once\n"; |
| 1018 | hGlobals << "#include \"types.h\"\n"; |
| 1019 | hGlobals << "#include <cstdint>\n"; |
| 1020 | hGlobals << "#include \"ps4_stubs.h\"\n\n"; |
| 1021 | // g_ps4_memory is the flat PS4 global address space, backed by ps4MEL. |
| 1022 | // Initialized in main() after runtime_init(). |
| 1023 | hGlobals << "// PS4 flat memory space - index with PS4 virtual address\n"; |
| 1024 | hGlobals << "extern uint8_t* g_ps4_memory;\n\n"; |
| 1025 | hGlobals << "// Translate a value that may be a PS4 virtual address or a host pointer\n"; |
| 1026 | hGlobals << "// into a safe host pointer for dereference.\n"; |
| 1027 | hGlobals << "// - Small values (<= 16MB) are PS4 virtual addresses -> map into g_ps4_memory\n"; |
| 1028 | hGlobals << "// - Values inside the g_ps4_memory allocation are used directly\n"; |
| 1029 | hGlobals << "// - Other values (nullptr or huge) fall back to g_ps4_memory base\n"; |
| 1030 | hGlobals << "inline void* ps4_resolve(int64_t addr) {\n"; |
| 1031 | hGlobals << " if (!g_ps4_memory) return (void*)g_ps4_memory;\n"; |
| 1032 | hGlobals << " const int64_t base = (int64_t)g_ps4_memory;\n"; |
| 1033 | hGlobals << " // Already a host pointer inside g_ps4_memory\n"; |
| 1034 | hGlobals << " if (addr >= base && addr < base + 0x1000000LL) return (void*)addr;\n"; |
| 1035 | hGlobals << " // PS4 virtual address: map into emulated memory (clamp to valid range)\n"; |
| 1036 | hGlobals << " uint64_t va = (uint64_t)addr;\n"; |
| 1037 | hGlobals << " if (va >= 0x1000000ULL) va = va % 0x1000000ULL;\n"; |
| 1038 | hGlobals << " return (void*)&g_ps4_memory[va];\n"; |
| 1039 | hGlobals << "}\n\n"; |
| 1040 | |
| 1041 | cppGlobals << "#include \"../include/globals.h\"\n\n"; |
| 1042 | cppGlobals << "// Initialized in main() via PS4Emu::GetGlobalMemoryBase()\n"; |
| 1043 | cppGlobals << "uint8_t* g_ps4_memory = nullptr;\n\n"; |
| 1044 | |
| 1045 | // We need to iterate the symbol database for globals |
| 1046 | if (symbolDatabase_) { |
| 1047 | const auto &syms = symbolDatabase_->getSymbols(); |
| 1048 | for (const auto &[addr, sym] : syms) { |
| 1049 | if (sym.type == Analysis::SymbolType::GlobalVariable) { |
| 1050 | // Parse type from name: g_TYPE_ADDR |
| 1051 | std::string typeName = "int64_t"; |
| 1052 | |
| 1053 | size_t firstUnderscore = sym.name.find('_'); |
| 1054 | size_t lastUnderscore = sym.name.rfind('_'); |
| 1055 | |
| 1056 | if (firstUnderscore != std::string::npos && |
| 1057 | lastUnderscore != std::string::npos && |
| 1058 | lastUnderscore > firstUnderscore) { |
| 1059 | typeName = sym.name.substr(firstUnderscore + 1, |
| 1060 | lastUnderscore - (firstUnderscore + 1)); |
| 1061 | } |
| 1062 | |
| 1063 | // Read initial value from memory |
| 1064 | std::string initVal = "0"; |
| 1065 | uint64_t offset = addr - baseAddress_; |
| 1066 | |
| 1067 | if (offset < rawData_.size()) { |
| 1068 | std::stringstream ss; |
| 1069 | ss << "0x" << std::hex; |
| 1070 | |
| 1071 | if (typeName.find("int8") != std::string::npos || |
| 1072 | typeName == "char" || typeName == "bool") { |
| 1073 | uint16_t val = |
| 1074 | rawData_[offset]; // Use uint16 for stream formatting of uint8 |
| 1075 | ss << val; |
| 1076 | initVal = ss.str(); |
| 1077 | } else if (typeName.find("int16") != std::string::npos || |
| 1078 | typeName == "short") { |
| 1079 | if (offset + 2 <= rawData_.size()) { |
| 1080 | uint16_t val = |
| 1081 | *reinterpret_cast<const uint16_t *>(&rawData_[offset]); |
| 1082 | ss << val; |
| 1083 | initVal = ss.str(); |
| 1084 | } |
| 1085 | } else if (typeName.find("int32") != std::string::npos || |
| 1086 | typeName == "int" || typeName == "float") { |
| 1087 | if (offset + 4 <= rawData_.size()) { |
| 1088 | uint32_t val = |
| 1089 | *reinterpret_cast<const uint32_t *>(&rawData_[offset]); |
| 1090 | ss << val; |
| 1091 | initVal = ss.str(); |
| 1092 | } |
| 1093 | if (typeName == "float") |
| 1094 | initVal = "0.0f"; // Placeholder |
| 1095 | } else { // int64, long, double, pointer |
| 1096 | if (offset + 8 <= rawData_.size()) { |
| 1097 | uint64_t val = |
| 1098 | *reinterpret_cast<const uint64_t *>(&rawData_[offset]); |
| 1099 | ss << val; |
| 1100 | initVal = ss.str(); |
| 1101 | } |
| 1102 | if (typeName == "double") |
| 1103 | initVal = "0.0"; // Placeholder |
| 1104 | if (typeName == "float") |
| 1105 | initVal = "0.0f"; |
| 1106 | } |
| 1107 | } else if (typeName == "double") { |
| 1108 | initVal = "0.0"; |
| 1109 | } else if (typeName == "float") { |
| 1110 | initVal = "0.0f"; |
| 1111 | } else { |
| 1112 | initVal = "{}"; // Aggregate init/default |
| 1113 | } |
| 1114 | |
| 1115 | hGlobals << "extern " << typeName << " " << sym.name << "; // 0x" |
| 1116 | << std::hex << addr << "\n"; |
| 1117 | cppGlobals << typeName << " " << sym.name << " = " << initVal |
| 1118 | << ";\n"; |
| 1119 | } |
| 1120 | } |
| 1121 | } |
| 1122 | } |
| 1123 | |
| 1124 | // --- PASS 6: Copy ps4MEL Runtime Files --- |
| 1125 | { |
| 1126 | std::filesystem::create_directories(root / "ps4mel"); |
| 1127 | |
| 1128 | // List of ps4MEL files to copy - all headers and implementations |
| 1129 | std::vector<std::string> ps4melFiles = { |
| 1130 | "ps4_memory.cpp", "ps4_memory.h", |
| 1131 | "ps4_kernel.cpp", "ps4_kernel.h", |
| 1132 | "ps4_pthread.cpp", "ps4_pthread.h", |
| 1133 | "ps4_tls.cpp", "ps4_tls.h", |
| 1134 | "ps4_stubs.cpp", "ps4_stubs.h", |
| 1135 | "ps4_safe_memory.h", |
| 1136 | "sdl_backend.cpp", "sdl_backend.h", |
| 1137 | "gnm_driver.h", |
| 1138 | "ps4_graphics.h", |
| 1139 | "runtime.cpp", "runtime.h" |
| 1140 | }; |
| 1141 | |
| 1142 | // Try to find ps4MEL directory relative to executable or source |
| 1143 | std::filesystem::path ps4melSrc; |
| 1144 | std::vector<std::filesystem::path> searchPaths = { |
| 1145 | std::filesystem::current_path() / "ps4MEL", |
| 1146 | std::filesystem::current_path().parent_path() / "ps4MEL", |
| 1147 | std::filesystem::path(__FILE__).parent_path().parent_path().parent_path() / "ps4MEL" |
| 1148 | }; |
| 1149 | |
| 1150 | for (const auto& p : searchPaths) { |
| 1151 | if (std::filesystem::exists(p)) { |
| 1152 | ps4melSrc = p; |
| 1153 | break; |
| 1154 | } |
| 1155 | } |
| 1156 | |
| 1157 | if (!ps4melSrc.empty()) { |
| 1158 | LOG_INFO(Common, "Copying ps4MEL files from: {}", ps4melSrc.string()); |
| 1159 | for (const auto& file : ps4melFiles) { |
| 1160 | std::filesystem::path srcFile = ps4melSrc / file; |
| 1161 | std::filesystem::path dstFile = root / "ps4mel" / file; |
| 1162 | if (std::filesystem::exists(srcFile)) { |
| 1163 | try { |
| 1164 | std::filesystem::copy_file(srcFile, dstFile, |
| 1165 | std::filesystem::copy_options::overwrite_existing); |
| 1166 | } catch (...) { |
| 1167 | LOG_WARNING(Common, "Failed to copy: {}", file); |
| 1168 | } |
| 1169 | } |
| 1170 | } |
| 1171 | } else { |
| 1172 | LOG_WARNING(Common, "ps4MEL directory not found, generating minimal stubs"); |
| 1173 | } |
| 1174 | } |
| 1175 | |
| 1176 | // --- PASS 7: Export Stubs, Runtime and Main --- |
| 1177 | { |
| 1178 | std::filesystem::create_directories(root / "assets"); |
| 1179 | |
| 1180 | std::map<std::string, std::shared_ptr<AST::FunctionAST>> nameToAst; |
| 1181 | for (auto &ast : analyzedASTs) { |
| 1182 | nameToAst[ast->name] = ast; |
| 1183 | } |
| 1184 | |
| 1185 | // 1. Export Stubs |
| 1186 | { |
| 1187 | std::ofstream stubs(root / "src" / "stubs.cpp"); |
| 1188 | stubs << "#include \"../include/functions.h\"\n\n"; |
| 1189 | for (const auto &func : functions_) { |
| 1190 | if (nameToAst.find(func->name) == nameToAst.end()) { |
| 1191 | stubs << "int64_t " << func->name |
| 1192 | << "(int64_t a1, int64_t a2, int64_t a3, " |
| 1193 | "int64_t a4, int64_t a5, int64_t a6) { return 0; }\n"; |
| 1194 | } |
| 1195 | } |
| 1196 | } |
| 1197 | |
| 1198 | // 4. Export Main (using ps4MEL runtime) |
| 1199 | { |
| 1200 | // Find entry point name |
| 1201 | std::string entryName = ""; |
| 1202 | { |
| 1203 | std::stringstream ss; |
| 1204 | ss << "sub_" << std::hex << entryPoint_; |
| 1205 | entryName = ss.str(); |
| 1206 | } |
| 1207 | auto it = std::find_if(functions_.begin(), functions_.end(), |
| 1208 | [this](auto f) { return f->address == entryPoint_; }); |
| 1209 | if (it != functions_.end()) { |
| 1210 | entryName = (*it)->name; |
| 1211 | } |
| 1212 | |
| 1213 | std::ofstream mainCpp(root / "src" / "main.cpp"); |
| 1214 | mainCpp << "/*\n" |
| 1215 | << " * Decompiled Game - Main Entry Point\n" |
| 1216 | << " * Generated by ShadPKG Decompiler\n" |
| 1217 | << " * Entry point: " << entryName |
| 1218 | << " @ 0x" << std::hex << entryPoint_ << std::dec << "\n" |
| 1219 | << " */\n\n"; |
| 1220 | mainCpp << "#include \"../include/functions.h\"\n"; |
| 1221 | mainCpp << "#include \"../include/globals.h\"\n"; |
| 1222 | mainCpp << "#include \"../ps4mel/runtime.h\"\n"; |
| 1223 | mainCpp << "#include \"../ps4mel/ps4_memory.h\"\n"; |
| 1224 | mainCpp << "#include <cstdio>\n"; |
| 1225 | mainCpp << "#include <cstring>\n"; |
| 1226 | mainCpp << "#include <vector>\n"; |
| 1227 | mainCpp << "#include <iostream>\n"; |
| 1228 | mainCpp << "#include <csignal>\n"; |
| 1229 | mainCpp << "#include <cstdlib>\n\n"; |
| 1230 | |
| 1231 | mainCpp << "// ─── Signal handler ───────────────────────────────────────\n"; |
| 1232 | mainCpp << "static volatile bool g_running = true;\n"; |
| 1233 | mainCpp << "static void signal_handler(int sig) {\n"; |
| 1234 | mainCpp << " (void)sig;\n"; |
| 1235 | mainCpp << " std::cout << \"\\n[RUNTIME] Signal caught, shutting down...\" << std::endl;\n"; |
| 1236 | mainCpp << " g_running = false;\n"; |
| 1237 | mainCpp << " std::exit(0);\n"; |
| 1238 | mainCpp << "}\n\n"; |
| 1239 | |
| 1240 | mainCpp << "// ─── Load ELF segments from memory.bin ─────────────────────\n"; |
| 1241 | mainCpp << "static void load_memory_bin() {\n"; |
| 1242 | mainCpp << " FILE* f = std::fopen(\"data/memory.bin\", \"rb\");\n"; |
| 1243 | mainCpp << " if (!f) { std::cout << \"[RUNTIME] No data/memory.bin found, skipping.\" << std::endl; return; }\n"; |
| 1244 | mainCpp << " uint32_t magic = 0, num_segs = 0;\n"; |
| 1245 | mainCpp << " std::fread(&magic, 4, 1, f);\n"; |
| 1246 | mainCpp << " if (magic != 0x34345350u) { std::fclose(f); return; } // 'PS44'\n"; |
| 1247 | mainCpp << " std::fread(&num_segs, 4, 1, f);\n"; |
| 1248 | mainCpp << " for (uint32_t i = 0; i < num_segs; i++) {\n"; |
| 1249 | mainCpp << " uint64_t vaddr = 0, size = 0;\n"; |
| 1250 | mainCpp << " std::fread(&vaddr, 8, 1, f);\n"; |
| 1251 | mainCpp << " std::fread(&size, 8, 1, f);\n"; |
| 1252 | mainCpp << " if (size == 0) continue;\n"; |
| 1253 | mainCpp << " // Translate virtual address to offset in g_ps4_memory\n"; |
| 1254 | mainCpp << " void* dst = PS4Emu::TranslateAddress(vaddr);\n"; |
| 1255 | mainCpp << " if (dst) std::fread(dst, 1, size, f);\n"; |
| 1256 | mainCpp << " else { std::vector<uint8_t> skip(size); std::fread(skip.data(), 1, size, f); }\n"; |
| 1257 | mainCpp << " }\n"; |
| 1258 | mainCpp << " std::fclose(f);\n"; |
| 1259 | mainCpp << " std::cout << \"[RUNTIME] ELF segments loaded from data/memory.bin\" << std::endl;\n"; |
| 1260 | mainCpp << "}\n\n"; |
| 1261 | |
| 1262 | mainCpp << "int main(int argc, char* argv[]) {\n"; |
| 1263 | mainCpp << " (void)argc; (void)argv;\n"; |
| 1264 | mainCpp << " std::signal(SIGINT, signal_handler);\n"; |
| 1265 | mainCpp << " std::signal(SIGTERM, signal_handler);\n\n"; |
| 1266 | mainCpp << " std::cout << \"[RUNTIME] ShadPKG Decompiled Runtime starting...\" << std::endl;\n\n"; |
| 1267 | mainCpp << " // 1. Init ps4MEL (memory, TLS, VFS)\n"; |
| 1268 | mainCpp << " runtime_init();\n\n"; |
| 1269 | mainCpp << " // 2. Point g_ps4_memory at ps4MEL's allocated block\n"; |
| 1270 | mainCpp << " g_ps4_memory = static_cast<uint8_t*>(PS4Emu::GetGlobalMemoryBase());\n"; |
| 1271 | mainCpp << " if (!g_ps4_memory) {\n"; |
| 1272 | mainCpp << " std::cerr << \"[RUNTIME] FATAL: ps4MEL memory not initialized\" << std::endl;\n"; |
| 1273 | mainCpp << " return 1;\n"; |
| 1274 | mainCpp << " }\n\n"; |
| 1275 | mainCpp << " // 3. Load ELF segment data into g_ps4_memory\n"; |
| 1276 | mainCpp << " load_memory_bin();\n\n"; |
| 1277 | |
| 1278 | // Add .init_array execution if present |
| 1279 | if (initArrayAddr_ != 0 && initArraySize_ > 0) { |
| 1280 | mainCpp << " // 4. Execute .init_array constructors\n"; |
| 1281 | mainCpp << " {\n"; |
| 1282 | mainCpp << " typedef int64_t (*init_func_t)(void);\n"; |
| 1283 | mainCpp << " uint64_t init_array_va = 0x" << std::hex << initArrayAddr_ << std::dec << ";\n"; |
| 1284 | mainCpp << " uint64_t init_array_size = 0x" << std::hex << initArraySize_ << std::dec << ";\n"; |
| 1285 | mainCpp << " uint64_t num_funcs = init_array_size / sizeof(uint64_t);\n"; |
| 1286 | mainCpp << " std::cout << \"[RUNTIME] Executing \" << num_funcs << \" .init_array functions...\" << std::endl;\n"; |
| 1287 | mainCpp << " for (uint64_t i = 0; i < num_funcs; i++) {\n"; |
| 1288 | mainCpp << " uint64_t func_ptr_addr = init_array_va + (i * sizeof(uint64_t));\n"; |
| 1289 | mainCpp << " uint64_t func_ptr = *((uint64_t*)(g_ps4_memory + func_ptr_addr));\n"; |
| 1290 | mainCpp << " if (func_ptr != 0) {\n"; |
| 1291 | mainCpp << " init_func_t init_func = reinterpret_cast<init_func_t>(g_ps4_memory + func_ptr);\n"; |
| 1292 | mainCpp << " try {\n"; |
| 1293 | mainCpp << " init_func();\n"; |
| 1294 | mainCpp << " } catch (...) {\n"; |
| 1295 | mainCpp << " std::cerr << \"[RUNTIME] .init_array[\" << i << \"] threw exception\" << std::endl;\n"; |
| 1296 | mainCpp << " }\n"; |
| 1297 | mainCpp << " }\n"; |
| 1298 | mainCpp << " }\n"; |
| 1299 | mainCpp << " }\n\n"; |
| 1300 | } |
| 1301 | |
| 1302 | mainCpp << " std::cout << \"[RUNTIME] Entry: " << entryName << " @ 0x" |
| 1303 | << std::hex << entryPoint_ << std::dec << "\" << std::endl;\n\n"; |
| 1304 | mainCpp << " // 4. Run the decompiled entry point\n"; |
| 1305 | mainCpp << " try {\n"; |
| 1306 | mainCpp << " int64_t result = " << entryName << "();\n"; |
| 1307 | mainCpp << " std::cout << \"[RUNTIME] Entry returned: \" << result << std::endl;\n"; |
| 1308 | mainCpp << " } catch (const std::exception& e) {\n"; |
| 1309 | mainCpp << " std::cerr << \"[RUNTIME] Exception: \" << e.what() << std::endl;\n"; |
| 1310 | mainCpp << " } catch (...) {\n"; |
| 1311 | mainCpp << " std::cerr << \"[RUNTIME] Unknown exception\" << std::endl;\n"; |
| 1312 | mainCpp << " }\n\n"; |
| 1313 | mainCpp << " runtime_shutdown();\n"; |
| 1314 | mainCpp << " std::cout << \"[RUNTIME] Shutdown complete. Frames: \"\n"; |
| 1315 | mainCpp << " << PS4Emu::GetFrameCounter() << std::endl;\n"; |
| 1316 | mainCpp << " return 0;\n"; |
| 1317 | mainCpp << "}\n"; |
| 1318 | } |
| 1319 | } |
| 1320 | |
| 1321 | // --- PASS 8: Export memory.bin (ELF segments for runtime loading) --- |
| 1322 | { |
| 1323 | std::filesystem::create_directories(root / "data"); |
| 1324 | std::ofstream memFile(root / "data" / "memory.bin", std::ios::binary); |
| 1325 | |
| 1326 | uint32_t magic = 0x34345350u; // 'PS44' |
| 1327 | uint32_t numSegs = static_cast<uint32_t>(segments_.size()); |
| 1328 | memFile.write(reinterpret_cast<const char*>(&magic), 4); |
| 1329 | memFile.write(reinterpret_cast<const char*>(&numSegs), 4); |
| 1330 | |
| 1331 | for (const auto& seg : segments_) { |
| 1332 | uint64_t vaddr = seg.virtualAddress; |
| 1333 | uint64_t memSize = seg.memSize > 0 ? seg.memSize : seg.size; // Use memSize for BSS padding |
| 1334 | memFile.write(reinterpret_cast<const char*>(&vaddr), 8); |
| 1335 | memFile.write(reinterpret_cast<const char*>(&memSize), 8); |
| 1336 | |
| 1337 | // Write file data (p_filesz bytes) |
| 1338 | if (seg.size > 0 && seg.fileOffset + seg.size <= rawData_.size()) { |
| 1339 | memFile.write(reinterpret_cast<const char*>(rawData_.data() + seg.fileOffset), |
| 1340 | static_cast<std::streamsize>(seg.size)); |
| 1341 | } else if (seg.size > 0) { |
| 1342 | // Partial write + zero-pad |
| 1343 | uint64_t available = rawData_.size() > seg.fileOffset ? rawData_.size() - seg.fileOffset : 0; |
| 1344 | if (available > 0) { |
| 1345 | memFile.write(reinterpret_cast<const char*>(rawData_.data() + seg.fileOffset), |
| 1346 | static_cast<std::streamsize>(available)); |
| 1347 | } |
| 1348 | std::vector<char> zeros(seg.size - available, 0); |
| 1349 | memFile.write(zeros.data(), static_cast<std::streamsize>(zeros.size())); |
| 1350 | } |
| 1351 | |
| 1352 | // Write BSS zero-padding (memSize - size bytes) |
| 1353 | if (memSize > seg.size) { |
| 1354 | std::vector<char> zeros(memSize - seg.size, 0); |
| 1355 | memFile.write(zeros.data(), static_cast<std::streamsize>(zeros.size())); |
| 1356 | } |
| 1357 | } |
| 1358 | LOG_INFO(Common, "Exported {} ELF segments to data/memory.bin", numSegs); |
| 1359 | } |
| 1360 | |
| 1361 | { |
| 1362 | std::ofstream cmake(root / "CMakeLists.txt"); |
| 1363 | cmake << "cmake_minimum_required(VERSION 3.14)\n"; |
| 1364 | cmake << "project(DecompiledGame CXX)\n\n"; |
| 1365 | cmake << "set(CMAKE_CXX_STANDARD 17)\n"; |
| 1366 | cmake << "set(CMAKE_CXX_STANDARD_REQUIRED ON)\n\n"; |
| 1367 | cmake << "# Compiler flags for large generated files\n"; |
| 1368 | cmake << "add_compile_options(\n"; |
| 1369 | cmake << " -fbracket-depth=1024\n"; |
| 1370 | cmake << " -Wno-unused-variable\n"; |
| 1371 | cmake << " -Wno-unused-but-set-variable\n"; |
| 1372 | cmake << " -Wno-unused-parameter\n"; |
| 1373 | cmake << " -Wno-missing-field-initializers\n"; |
| 1374 | cmake << " -Wno-tautological-compare\n"; |
| 1375 | cmake << " -Wno-tautological-constant-out-of-range-compare\n"; |
| 1376 | cmake << " -Wno-logical-not-parentheses\n"; |
| 1377 | cmake << " -O0\n"; |
| 1378 | cmake << ")\n\n"; |
| 1379 | cmake << "# Include directories\n"; |
| 1380 | cmake << "include_directories(include)\n"; |
| 1381 | cmake << "include_directories(ps4mel)\n\n"; |
| 1382 | cmake << "# SDL2 - required by ps4MEL graphics/stub layer\n"; |
| 1383 | cmake << "find_package(SDL2 QUIET)\n"; |
| 1384 | cmake << "if(SDL2_FOUND)\n"; |
| 1385 | cmake << " message(STATUS \"SDL2 found: ${SDL2_INCLUDE_DIRS}\")\n"; |
| 1386 | cmake << " include_directories(${SDL2_INCLUDE_DIRS})\n"; |
| 1387 | cmake << " set(HAS_SDL2 TRUE)\n"; |
| 1388 | cmake << "else()\n"; |
| 1389 | cmake << " # Try pkg-config fallback\n"; |
| 1390 | cmake << " find_package(PkgConfig QUIET)\n"; |
| 1391 | cmake << " if(PKG_CONFIG_FOUND)\n"; |
| 1392 | cmake << " pkg_check_modules(SDL2 QUIET sdl2)\n"; |
| 1393 | cmake << " if(SDL2_FOUND)\n"; |
| 1394 | cmake << " include_directories(${SDL2_INCLUDE_DIRS})\n"; |
| 1395 | cmake << " set(HAS_SDL2 TRUE)\n"; |
| 1396 | cmake << " endif()\n"; |
| 1397 | cmake << " endif()\n"; |
| 1398 | cmake << " if(NOT HAS_SDL2)\n"; |
| 1399 | cmake << " message(WARNING \"SDL2 not found. Graphics stubs will be disabled.\")\n"; |
| 1400 | cmake << " add_definitions(-DPS4_NO_SDL)\n"; |
| 1401 | cmake << " endif()\n"; |
| 1402 | cmake << "endif()\n\n"; |
| 1403 | cmake << "# Game source files\n"; |
| 1404 | cmake << "file(GLOB GAME_SOURCES \"src/*.cpp\")\n\n"; |
| 1405 | cmake << "# PS4 Runtime Emulation Layer (ps4MEL)\n"; |
| 1406 | cmake << "set(PS4MEL_SOURCES\n"; |
| 1407 | cmake << " ps4mel/ps4_memory.cpp\n"; |
| 1408 | cmake << " ps4mel/ps4_kernel.cpp\n"; |
| 1409 | cmake << " ps4mel/ps4_pthread.cpp\n"; |
| 1410 | cmake << " ps4mel/ps4_tls.cpp\n"; |
| 1411 | cmake << " ps4mel/ps4_stubs.cpp\n"; |
| 1412 | cmake << " ps4mel/runtime.cpp\n"; |
| 1413 | cmake << ")\n"; |
| 1414 | cmake << "if(HAS_SDL2)\n"; |
| 1415 | cmake << " list(APPEND PS4MEL_SOURCES ps4mel/sdl_backend.cpp)\n"; |
| 1416 | cmake << "endif()\n\n"; |
| 1417 | cmake << "add_executable(Game ${GAME_SOURCES} ${PS4MEL_SOURCES})\n\n"; |
| 1418 | cmake << "# Platform-specific linking\n"; |
| 1419 | cmake << "if(APPLE)\n"; |
| 1420 | cmake << " target_link_libraries(Game PRIVATE pthread)\n"; |
| 1421 | cmake << " if(HAS_SDL2)\n"; |
| 1422 | cmake << " target_link_libraries(Game PRIVATE ${SDL2_LIBRARIES})\n"; |
| 1423 | cmake << " endif()\n"; |
| 1424 | cmake << "elseif(UNIX)\n"; |
| 1425 | cmake << " target_link_libraries(Game PRIVATE pthread dl)\n"; |
| 1426 | cmake << " if(HAS_SDL2)\n"; |
| 1427 | cmake << " target_link_libraries(Game PRIVATE ${SDL2_LIBRARIES})\n"; |
| 1428 | cmake << " endif()\n"; |
| 1429 | cmake << "elseif(WIN32)\n"; |
| 1430 | cmake << " if(HAS_SDL2)\n"; |
| 1431 | cmake << " target_link_libraries(Game PRIVATE ${SDL2_LIBRARIES})\n"; |
| 1432 | cmake << " endif()\n"; |
| 1433 | cmake << "endif()\n\n"; |
| 1434 | cmake << "# Copy data folder next to binary\n"; |
| 1435 | cmake << "add_custom_command(TARGET Game POST_BUILD\n"; |
| 1436 | cmake << " COMMAND ${CMAKE_COMMAND} -E copy_directory\n"; |
| 1437 | cmake << " ${CMAKE_SOURCE_DIR}/data $<TARGET_FILE_DIR:Game>/data\n"; |
| 1438 | cmake << " COMMENT \"Copying data/memory.bin to build directory\"\n"; |
| 1439 | cmake << ")\n"; |
| 1440 | } |
| 1441 | |
| 1442 | std::cout << "Project exported to: " << outPath << "\n"; |
| 1443 | } |
| 1444 | |
| 1445 | std::string DecompilerContext::GenerateCode() { |
| 1446 | if (!isAnalyzed_) |
| 1447 | return "// Not analyzed yet"; |
| 1448 | |
| 1449 | std::stringstream ss; |
| 1450 | ss << "// Decompiled by ShadPKG\n"; |
| 1451 | ss << "// Architecture: x86-64\n\n"; |
| 1452 | |
| 1453 | for (const auto &func : functions_) { |
| 1454 | ss << func->signature << " {\n"; |
| 1455 | for (const auto &bb : func->basicBlocks) { |
| 1456 | ss << " // Block: 0x" << std::hex << bb->startAddress << "\n"; |
| 1457 | for (const auto &instr : bb->instructions) { |
| 1458 | ss << " " << instr.disassembly << ";\n"; |
| 1459 | } |
| 1460 | if (!bb->successors.empty()) { |
| 1461 | ss << " goto "; |
| 1462 | for (auto succ : bb->successors) |
| 1463 | ss << "0x" << std::hex << succ << " "; |
| 1464 | ss << ";\n"; |
| 1465 | } |
| 1466 | ss << "\n"; |
| 1467 | } |
| 1468 | ss << "}\n\n"; |
| 1469 | } |
| 1470 | |
| 1471 | return ss.str(); |
| 1472 | } |
| 1473 | |
| 1474 | std::string DecompilerContext::GenerateStructuredCode() { |
| 1475 | if (!isAnalyzed_) |
| 1476 | return "// Not analyzed yet (Run Analysis first)"; |
| 1477 | |
| 1478 | std::stringstream ss; |
| 1479 | ss << "// Decompiled by ShadPKG (Structured Analysis)\n"; |
| 1480 | ss << "// Architecture: x86-64\n\n"; |
| 1481 | |
| 1482 | auto symbols = std::make_shared<Analysis::SymbolAnalysis>( |
| 1483 | rawData_, baseAddress_, symbolDatabase_); |
| 1484 | symbols->analyze(); |
| 1485 | |
| 1486 | for (const auto &func : functions_) { |
| 1487 | auto dom = std::make_shared<Analysis::DominatorAnalysis>(); |
| 1488 | dom->analyze(func); |
| 1489 | |
| 1490 | auto lifter = std::make_shared<Lifter::VariableAnalysis>(func); |
| 1491 | lifter->analyze(); |
| 1492 | |
| 1493 | Analysis::StructuralAnalysis structural(func, dom, symbols, lifter); |
| 1494 | auto ast = structural.analyze(); |
| 1495 | |
| 1496 | lifter->applyToAST(ast); |
| 1497 | |
| 1498 | for (auto &local : ast->locals) { |
| 1499 | auto userType = GetUserVarType(func->address, local.stackOffset); |
| 1500 | if (userType) { |
| 1501 | local.complexType = userType; |
| 1502 | } |
| 1503 | } |
| 1504 | |
| 1505 | Analysis::DataFlowAnalysis dataflow(ast); |
| 1506 | dataflow.analyze(); |
| 1507 | |
| 1508 | Analysis::MemberAccessAnalysis memberAccess(typeManager_); |
| 1509 | memberAccess.analyze(ast); |
| 1510 | |
| 1511 | Codegen::CppEmitter emitter; |
| 1512 | ss << emitter.generate(ast); |
| 1513 | functionTokens_[func->address] = emitter.getTokens(); |
| 1514 | ss << "\n"; |
| 1515 | } |
| 1516 | |
| 1517 | return ss.str(); |
| 1518 | } |
| 1519 | |
| 1520 | const std::vector<Codegen::CppEmitter::Token> & |
| 1521 | DecompilerContext::GetFunctionTokens(uint64_t address) const { |
| 1522 | if (functionTokens_.count(address)) { |
| 1523 | return functionTokens_.at(address); |
| 1524 | } |
| 1525 | static std::vector<Codegen::CppEmitter::Token> empty; |
| 1526 | return empty; |
| 1527 | } |
| 1528 | |
| 1529 | uint64_t DecompilerContext::ReadPointer(uint64_t address) const { |
| 1530 | if (address < baseAddress_) |
| 1531 | return 0; |
| 1532 | uint64_t offset = address - baseAddress_; |
| 1533 | if (offset + 8 > rawData_.size()) |
| 1534 | return 0; |
| 1535 | return *reinterpret_cast<const uint64_t *>(&rawData_[offset]); |
| 1536 | } |
| 1537 | |
| 1538 | int32_t DecompilerContext::ReadInt32(uint64_t address) const { |
| 1539 | if (address < baseAddress_) |
| 1540 | return 0; |
| 1541 | uint64_t offset = address - baseAddress_; |
| 1542 | if (offset + 4 > rawData_.size()) |
| 1543 | return 0; |
| 1544 | return *reinterpret_cast<const int32_t *>(&rawData_[offset]); |
| 1545 | } |
| 1546 | |
| 1547 | std::shared_ptr<IR::Function> |
| 1548 | DecompilerContext::GetFunctionAt(uint64_t address) { |
| 1549 | for (const auto &func : functions_) { |
| 1550 | if (func->address == address) |
| 1551 | return func; |
| 1552 | } |
| 1553 | return nullptr; |
| 1554 | } |
| 1555 | |
| 1556 | bool DecompilerContext::SaveProject(const std::string &path) { |
| 1557 | json j; |
| 1558 | |
| 1559 | // Symbols |
| 1560 | if (symbolDatabase_) { |
| 1561 | for (const auto &[addr, sym] : symbolDatabase_->getSymbols()) { |
| 1562 | j["symbols"].push_back({{"address", addr}, |
| 1563 | {"name", sym.name}, |
| 1564 | {"type", (int)sym.type}, |
| 1565 | {"source", (int)sym.source}}); |
| 1566 | } |
| 1567 | } |
| 1568 | |
| 1569 | // Structs |
| 1570 | for (const auto &[name, type] : typeManager_->getAllTypes()) { |
| 1571 | if (type->getKind() == Analysis::Type::Kind::Struct) { |
| 1572 | auto s = std::dynamic_pointer_cast<Analysis::StructType>(type); |
| 1573 | json sJson; |
| 1574 | sJson["name"] = name; |
| 1575 | for (const auto &m : s->getMembers()) { |
| 1576 | sJson["members"].push_back({{"name", m.name}, |
| 1577 | {"offset", m.offset}, |
| 1578 | {"type", m.type->toString()}}); |
| 1579 | } |
| 1580 | j["structs"].push_back(sJson); |
| 1581 | } |
| 1582 | } |
| 1583 | |
| 1584 | // User Overrides (Stack Types) |
| 1585 | // userVarTypes_: FunctionAddr -> (StackOffset -> Type) |
| 1586 | for (const auto &[funcAddr, offsets] : userVarTypes_) { |
| 1587 | for (const auto &[offset, type] : offsets) { |
| 1588 | j["overrides"].push_back( |
| 1589 | {{"func", funcAddr}, {"offset", offset}, {"type", type->toString()}}); |
| 1590 | } |
| 1591 | } |
| 1592 | |
| 1593 | std::ofstream o(path); |
| 1594 | o << std::setw(4) << j << std::endl; |
| 1595 | return true; |
| 1596 | } |
| 1597 | |
| 1598 | bool DecompilerContext::LoadProject(const std::string &path) { |
| 1599 | std::ifstream i(path); |
| 1600 | if (!i.is_open()) |
| 1601 | return false; |
| 1602 | |
| 1603 | json j; |
| 1604 | i >> j; |
| 1605 | |
| 1606 | // Symbols |
| 1607 | if (j.contains("symbols") && symbolDatabase_) { |
| 1608 | symbolDatabase_->clear(); // Maybe clear existing? |
| 1609 | for (const auto &sym : j["symbols"]) { |
| 1610 | symbolDatabase_->addSymbol(sym["address"], sym["name"], |
| 1611 | (Analysis::SymbolType)sym["type"], |
| 1612 | (Analysis::SymbolSource)sym["source"]); |
| 1613 | } |
| 1614 | } |
| 1615 | |
| 1616 | // Structs |
| 1617 | if (j.contains("structs")) { |
| 1618 | for (const auto &sJson : j["structs"]) { |
| 1619 | std::string name = sJson["name"]; |
| 1620 | // Use createStruct which registers it |
| 1621 | auto s = typeManager_->createStruct(name); |
| 1622 | for (const auto &mJson : sJson["members"]) { |
| 1623 | std::string typeName = mJson["type"]; |
| 1624 | auto type = typeManager_->getType(typeName); |
| 1625 | if (!type) { |
| 1626 | // Basic primitive fallback or implicit creation |
| 1627 | if (typeName.back() == '*') { |
| 1628 | // Assume pointer to something? |
| 1629 | // For now default to int |
| 1630 | type = typeManager_->getType("int"); |
| 1631 | } else { |
| 1632 | type = typeManager_->getType("int"); |
| 1633 | } |
| 1634 | } |
| 1635 | if (type) { |
| 1636 | s->addMember(mJson["name"], type, mJson["offset"]); |
| 1637 | } |
| 1638 | } |
| 1639 | } |
| 1640 | } |
| 1641 | |
| 1642 | // Overrides |
| 1643 | if (j.contains("overrides")) { |
| 1644 | for (const auto &ov : j["overrides"]) { |
| 1645 | uint64_t funcAddr = ov["func"]; |
| 1646 | int offset = ov["offset"]; |
| 1647 | std::string typeName = ov["type"]; |
| 1648 | |
| 1649 | auto type = typeManager_->getType(typeName); |
| 1650 | if (type) { |
| 1651 | SetUserVarType(funcAddr, offset, type); |
| 1652 | } |
| 1653 | } |
| 1654 | } |
| 1655 | |
| 1656 | return true; |
| 1657 | } |
| 1658 | |
| 1659 | } // namespace ShadPKG::Decompiler |