libsanitizer/sanitizer_common/sanitizer_symbolizer_libcdep.cpp

   1 //===-- sanitizer_symbolizer_libcdep.cpp ----------------------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file is shared between AddressSanitizer and ThreadSanitizer
  10 // run-time libraries.
  11 //===----------------------------------------------------------------------===//
  12
  13 #include "sanitizer_allocator_internal.h"
  14 #include "sanitizer_internal_defs.h"
  15 #include "sanitizer_platform.h"
  16 #include "sanitizer_symbolizer_internal.h"
  17
  18 namespace __sanitizer {
  19
  20 Symbolizer *Symbolizer::GetOrInit() {
  21   SpinMutexLock l(&init_mu_);
  22   if (symbolizer_)
  23     return symbolizer_;
  24   symbolizer_ = PlatformInit();
  25   CHECK(symbolizer_);
  26   return symbolizer_;
  27 }
  28
  29 // See sanitizer_symbolizer_markup.cpp.
  30 #if !SANITIZER_SYMBOLIZER_MARKUP
  31
  32 const char *ExtractToken(const char *str, const char *delims, char **result) {
  33   uptr prefix_len = internal_strcspn(str, delims);
  34   *result = (char*)InternalAlloc(prefix_len + 1);
  35   internal_memcpy(*result, str, prefix_len);
  36   (*result)[prefix_len] = '\0';
  37   const char *prefix_end = str + prefix_len;
  38   if (*prefix_end != '\0') prefix_end++;
  39   return prefix_end;
  40 }
  41
  42 const char *ExtractInt(const char *str, const char *delims, int *result) {
  43   char *buff = nullptr;
  44   const char *ret = ExtractToken(str, delims, &buff);
  45   if (buff) {
  46     *result = (int)internal_atoll(buff);
  47   }
  48   InternalFree(buff);
  49   return ret;
  50 }
  51
  52 const char *ExtractUptr(const char *str, const char *delims, uptr *result) {
  53   char *buff = nullptr;
  54   const char *ret = ExtractToken(str, delims, &buff);
  55   if (buff) {
  56     *result = (uptr)internal_atoll(buff);
  57   }
  58   InternalFree(buff);
  59   return ret;
  60 }
  61
  62 const char *ExtractSptr(const char *str, const char *delims, sptr *result) {
  63   char *buff = nullptr;
  64   const char *ret = ExtractToken(str, delims, &buff);
  65   if (buff) {
  66     *result = (sptr)internal_atoll(buff);
  67   }
  68   InternalFree(buff);
  69   return ret;
  70 }
  71
  72 const char *ExtractTokenUpToDelimiter(const char *str, const char *delimiter,
  73                                       char **result) {
  74   const char *found_delimiter = internal_strstr(str, delimiter);
  75   uptr prefix_len =
  76       found_delimiter ? found_delimiter - str : internal_strlen(str);
  77   *result = (char *)InternalAlloc(prefix_len + 1);
  78   internal_memcpy(*result, str, prefix_len);
  79   (*result)[prefix_len] = '\0';
  80   const char *prefix_end = str + prefix_len;
  81   if (*prefix_end != '\0') prefix_end += internal_strlen(delimiter);
  82   return prefix_end;
  83 }
  84
  85 SymbolizedStack *Symbolizer::SymbolizePC(uptr addr) {
  86   Lock l(&mu_);
  87   SymbolizedStack *res = SymbolizedStack::New(addr);
  88   auto *mod = FindModuleForAddress(addr);
  89   if (!mod)
  90     return res;
  91   // Always fill data about module name and offset.
  92   res->info.FillModuleInfo(*mod);
  93   for (auto &tool : tools_) {
  94     SymbolizerScope sym_scope(this);
  95     if (tool.SymbolizePC(addr, res)) {
  96       return res;
  97     }
  98   }
  99   return res;
 100 }
 101
 102 bool Symbolizer::SymbolizeData(uptr addr, DataInfo *info) {
 103   Lock l(&mu_);
 104   const char *module_name = nullptr;
 105   uptr module_offset;
 106   ModuleArch arch;
 107   if (!FindModuleNameAndOffsetForAddress(addr, &module_name, &module_offset,
 108                                          &arch))
 109     return false;
 110   info->Clear();
 111   info->module = internal_strdup(module_name);
 112   info->module_offset = module_offset;
 113   info->module_arch = arch;
 114   for (auto &tool : tools_) {
 115     SymbolizerScope sym_scope(this);
 116     if (tool.SymbolizeData(addr, info)) {
 117       return true;
 118     }
 119   }
 120   return true;
 121 }
 122
 123 bool Symbolizer::SymbolizeFrame(uptr addr, FrameInfo *info) {
 124   Lock l(&mu_);
 125   const char *module_name = nullptr;
 126   if (!FindModuleNameAndOffsetForAddress(
 127           addr, &module_name, &info->module_offset, &info->module_arch))
 128     return false;
 129   info->module = internal_strdup(module_name);
 130   for (auto &tool : tools_) {
 131     SymbolizerScope sym_scope(this);
 132     if (tool.SymbolizeFrame(addr, info)) {
 133       return true;
 134     }
 135   }
 136   return true;
 137 }
 138
 139 bool Symbolizer::GetModuleNameAndOffsetForPC(uptr pc, const char **module_name,
 140                                              uptr *module_address) {
 141   Lock l(&mu_);
 142   const char *internal_module_name = nullptr;
 143   ModuleArch arch;
 144   if (!FindModuleNameAndOffsetForAddress(pc, &internal_module_name,
 145                                          module_address, &arch))
 146     return false;
 147
 148   if (module_name)
 149     *module_name = module_names_.GetOwnedCopy(internal_module_name);
 150   return true;
 151 }
 152
 153 void Symbolizer::Flush() {
 154   Lock l(&mu_);
 155   for (auto &tool : tools_) {
 156     SymbolizerScope sym_scope(this);
 157     tool.Flush();
 158   }
 159 }
 160
 161 const char *Symbolizer::Demangle(const char *name) {
 162   Lock l(&mu_);
 163   for (auto &tool : tools_) {
 164     SymbolizerScope sym_scope(this);
 165     if (const char *demangled = tool.Demangle(name))
 166       return demangled;
 167   }
 168   return PlatformDemangle(name);
 169 }
 170
 171 bool Symbolizer::FindModuleNameAndOffsetForAddress(uptr address,
 172                                                    const char **module_name,
 173                                                    uptr *module_offset,
 174                                                    ModuleArch *module_arch) {
 175   const LoadedModule *module = FindModuleForAddress(address);
 176   if (!module)
 177     return false;
 178   *module_name = module->full_name();
 179   *module_offset = address - module->base_address();
 180   *module_arch = module->arch();
 181   return true;
 182 }
 183
 184 void Symbolizer::RefreshModules() {
 185   modules_.init();
 186   fallback_modules_.fallbackInit();
 187   RAW_CHECK(modules_.size() > 0);
 188   modules_fresh_ = true;
 189 }
 190
 191 static const LoadedModule *SearchForModule(const ListOfModules &modules,
 192                                            uptr address) {
 193   for (uptr i = 0; i < modules.size(); i++) {
 194     if (modules[i].containsAddress(address)) {
 195       return &modules[i];
 196     }
 197   }
 198   return nullptr;
 199 }
 200
 201 const LoadedModule *Symbolizer::FindModuleForAddress(uptr address) {
 202   bool modules_were_reloaded = false;
 203   if (!modules_fresh_) {
 204     RefreshModules();
 205     modules_were_reloaded = true;
 206   }
 207   const LoadedModule *module = SearchForModule(modules_, address);
 208   if (module) return module;
 209
 210   // dlopen/dlclose interceptors invalidate the module list, but when
 211   // interception is disabled, we need to retry if the lookup fails in
 212   // case the module list changed.
 213 #if !SANITIZER_INTERCEPT_DLOPEN_DLCLOSE
 214   if (!modules_were_reloaded) {
 215     RefreshModules();
 216     module = SearchForModule(modules_, address);
 217     if (module) return module;
 218   }
 219 #endif
 220
 221   if (fallback_modules_.size()) {
 222     module = SearchForModule(fallback_modules_, address);
 223   }
 224   return module;
 225 }
 226
 227 // For now we assume the following protocol:
 228 // For each request of the form
 229 //   <module_name> <module_offset>
 230 // passed to STDIN, external symbolizer prints to STDOUT response:
 231 //   <function_name>
 232 //   <file_name>:<line_number>:<column_number>
 233 //   <function_name>
 234 //   <file_name>:<line_number>:<column_number>
 235 //   ...
 236 //   <empty line>
 237 class LLVMSymbolizerProcess final : public SymbolizerProcess {
 238  public:
 239   explicit LLVMSymbolizerProcess(const char *path)
 240       : SymbolizerProcess(path, /*use_posix_spawn=*/SANITIZER_APPLE) {}
 241
 242  private:
 243   bool ReachedEndOfOutput(const char *buffer, uptr length) const override {
 244     // Empty line marks the end of llvm-symbolizer output.
 245     return length >= 2 && buffer[length - 1] == '\n' &&
 246            buffer[length - 2] == '\n';
 247   }
 248
 249   // When adding a new architecture, don't forget to also update
 250   // script/asan_symbolize.py and sanitizer_common.h.
 251   void GetArgV(const char *path_to_binary,
 252                const char *(&argv)[kArgVMax]) const override {
 253 #if defined(__x86_64h__)
 254     const char* const kSymbolizerArch = "--default-arch=x86_64h";
 255 #elif defined(__x86_64__)
 256     const char* const kSymbolizerArch = "--default-arch=x86_64";
 257 #elif defined(__i386__)
 258     const char* const kSymbolizerArch = "--default-arch=i386";
 259 #elif SANITIZER_LOONGARCH64
 260     const char *const kSymbolizerArch = "--default-arch=loongarch64";
 261 #elif SANITIZER_RISCV64
 262     const char *const kSymbolizerArch = "--default-arch=riscv64";
 263 #elif defined(__aarch64__)
 264     const char* const kSymbolizerArch = "--default-arch=arm64";
 265 #elif defined(__arm__)
 266     const char* const kSymbolizerArch = "--default-arch=arm";
 267 #elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
 268     const char* const kSymbolizerArch = "--default-arch=powerpc64";
 269 #elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
 270     const char* const kSymbolizerArch = "--default-arch=powerpc64le";
 271 #elif defined(__s390x__)
 272     const char* const kSymbolizerArch = "--default-arch=s390x";
 273 #elif defined(__s390__)
 274     const char* const kSymbolizerArch = "--default-arch=s390";
 275 #else
 276     const char* const kSymbolizerArch = "--default-arch=unknown";
 277 #endif
 278
 279     const char *const demangle_flag =
 280         common_flags()->demangle ? "--demangle" : "--no-demangle";
 281     const char *const inline_flag =
 282         common_flags()->symbolize_inline_frames ? "--inlines" : "--no-inlines";
 283     int i = 0;
 284     argv[i++] = path_to_binary;
 285     argv[i++] = demangle_flag;
 286     argv[i++] = inline_flag;
 287     argv[i++] = kSymbolizerArch;
 288     argv[i++] = nullptr;
 289     CHECK_LE(i, kArgVMax);
 290   }
 291 };
 292
 293 LLVMSymbolizer::LLVMSymbolizer(const char *path, LowLevelAllocator *allocator)
 294     : symbolizer_process_(new(*allocator) LLVMSymbolizerProcess(path)) {}
 295
 296 // Parse a <file>:<line>[:<column>] buffer. The file path may contain colons on
 297 // Windows, so extract tokens from the right hand side first. The column info is
 298 // also optional.
 299 static const char *ParseFileLineInfo(AddressInfo *info, const char *str) {
 300   char *file_line_info = nullptr;
 301   str = ExtractToken(str, "\n", &file_line_info);
 302   CHECK(file_line_info);
 303
 304   if (uptr size = internal_strlen(file_line_info)) {
 305     char *back = file_line_info + size - 1;
 306     for (int i = 0; i < 2; ++i) {
 307       while (back > file_line_info && IsDigit(*back)) --back;
 308       if (*back != ':' || !IsDigit(back[1])) break;
 309       info->column = info->line;
 310       info->line = internal_atoll(back + 1);
 311       // Truncate the string at the colon to keep only filename.
 312       *back = '\0';
 313       --back;
 314     }
 315     ExtractToken(file_line_info, "", &info->file);
 316   }
 317
 318   InternalFree(file_line_info);
 319   return str;
 320 }
 321
 322 // Parses one or more two-line strings in the following format:
 323 //   <function_name>
 324 //   <file_name>:<line_number>[:<column_number>]
 325 // Used by LLVMSymbolizer, Addr2LinePool and InternalSymbolizer, since all of
 326 // them use the same output format.
 327 void ParseSymbolizePCOutput(const char *str, SymbolizedStack *res) {
 328   bool top_frame = true;
 329   SymbolizedStack *last = res;
 330   while (true) {
 331     char *function_name = nullptr;
 332     str = ExtractToken(str, "\n", &function_name);
 333     CHECK(function_name);
 334     if (function_name[0] == '\0') {
 335       // There are no more frames.
 336       InternalFree(function_name);
 337       break;
 338     }
 339     SymbolizedStack *cur;
 340     if (top_frame) {
 341       cur = res;
 342       top_frame = false;
 343     } else {
 344       cur = SymbolizedStack::New(res->info.address);
 345       cur->info.FillModuleInfo(res->info.module, res->info.module_offset,
 346                                res->info.module_arch);
 347       last->next = cur;
 348       last = cur;
 349     }
 350
 351     AddressInfo *info = &cur->info;
 352     info->function = function_name;
 353     str = ParseFileLineInfo(info, str);
 354
 355     // Functions and filenames can be "??", in which case we write 0
 356     // to address info to mark that names are unknown.
 357     if (0 == internal_strcmp(info->function, "??")) {
 358       InternalFree(info->function);
 359       info->function = 0;
 360     }
 361     if (info->file && 0 == internal_strcmp(info->file, "??")) {
 362       InternalFree(info->file);
 363       info->file = 0;
 364     }
 365   }
 366 }
 367
 368 // Parses a two- or three-line string in the following format:
 369 //   <symbol_name>
 370 //   <start_address> <size>
 371 //   <filename>:<column>
 372 // Used by LLVMSymbolizer and InternalSymbolizer. LLVMSymbolizer added support
 373 // for symbolizing the third line in D123538, but we support the older two-line
 374 // information as well.
 375 void ParseSymbolizeDataOutput(const char *str, DataInfo *info) {
 376   str = ExtractToken(str, "\n", &info->name);
 377   str = ExtractUptr(str, " ", &info->start);
 378   str = ExtractUptr(str, "\n", &info->size);
 379   // Note: If the third line isn't present, these calls will set info.{file,
 380   // line} to empty strings.
 381   str = ExtractToken(str, ":", &info->file);
 382   str = ExtractUptr(str, "\n", &info->line);
 383 }
 384
 385 static void ParseSymbolizeFrameOutput(const char *str,
 386                                       InternalMmapVector<LocalInfo> *locals) {
 387   if (internal_strncmp(str, "??", 2) == 0)
 388     return;
 389
 390   while (*str) {
 391     LocalInfo local;
 392     str = ExtractToken(str, "\n", &local.function_name);
 393     str = ExtractToken(str, "\n", &local.name);
 394
 395     AddressInfo addr;
 396     str = ParseFileLineInfo(&addr, str);
 397     local.decl_file = addr.file;
 398     local.decl_line = addr.line;
 399
 400     local.has_frame_offset = internal_strncmp(str, "??", 2) != 0;
 401     str = ExtractSptr(str, " ", &local.frame_offset);
 402
 403     local.has_size = internal_strncmp(str, "??", 2) != 0;
 404     str = ExtractUptr(str, " ", &local.size);
 405
 406     local.has_tag_offset = internal_strncmp(str, "??", 2) != 0;
 407     str = ExtractUptr(str, "\n", &local.tag_offset);
 408
 409     locals->push_back(local);
 410   }
 411 }
 412
 413 bool LLVMSymbolizer::SymbolizePC(uptr addr, SymbolizedStack *stack) {
 414   AddressInfo *info = &stack->info;
 415   const char *buf = FormatAndSendCommand(
 416       "CODE", info->module, info->module_offset, info->module_arch);
 417   if (!buf)
 418     return false;
 419   ParseSymbolizePCOutput(buf, stack);
 420   return true;
 421 }
 422
 423 bool LLVMSymbolizer::SymbolizeData(uptr addr, DataInfo *info) {
 424   const char *buf = FormatAndSendCommand(
 425       "DATA", info->module, info->module_offset, info->module_arch);
 426   if (!buf)
 427     return false;
 428   ParseSymbolizeDataOutput(buf, info);
 429   info->start += (addr - info->module_offset); // Add the base address.
 430   return true;
 431 }
 432
 433 bool LLVMSymbolizer::SymbolizeFrame(uptr addr, FrameInfo *info) {
 434   const char *buf = FormatAndSendCommand(
 435       "FRAME", info->module, info->module_offset, info->module_arch);
 436   if (!buf)
 437     return false;
 438   ParseSymbolizeFrameOutput(buf, &info->locals);
 439   return true;
 440 }
 441
 442 const char *LLVMSymbolizer::FormatAndSendCommand(const char *command_prefix,
 443                                                  const char *module_name,
 444                                                  uptr module_offset,
 445                                                  ModuleArch arch) {
 446   CHECK(module_name);
 447   int size_needed = 0;
 448   if (arch == kModuleArchUnknown)
 449     size_needed = internal_snprintf(buffer_, kBufferSize, "%s \"%s\" 0x%zx\n",
 450                                     command_prefix, module_name, module_offset);
 451   else
 452     size_needed = internal_snprintf(buffer_, kBufferSize,
 453                                     "%s \"%s:%s\" 0x%zx\n", command_prefix,
 454                                     module_name, ModuleArchToString(arch),
 455                                     module_offset);
 456
 457   if (size_needed >= static_cast<int>(kBufferSize)) {
 458     Report("WARNING: Command buffer too small");
 459     return nullptr;
 460   }
 461
 462   return symbolizer_process_->SendCommand(buffer_);
 463 }
 464
 465 SymbolizerProcess::SymbolizerProcess(const char *path, bool use_posix_spawn)
 466     : path_(path),
 467       input_fd_(kInvalidFd),
 468       output_fd_(kInvalidFd),
 469       times_restarted_(0),
 470       failed_to_start_(false),
 471       reported_invalid_path_(false),
 472       use_posix_spawn_(use_posix_spawn) {
 473   CHECK(path_);
 474   CHECK_NE(path_[0], '\0');
 475 }
 476
 477 static bool IsSameModule(const char* path) {
 478   if (const char* ProcessName = GetProcessName()) {
 479     if (const char* SymbolizerName = StripModuleName(path)) {
 480       return !internal_strcmp(ProcessName, SymbolizerName);
 481     }
 482   }
 483   return false;
 484 }
 485
 486 const char *SymbolizerProcess::SendCommand(const char *command) {
 487   if (failed_to_start_)
 488     return nullptr;
 489   if (IsSameModule(path_)) {
 490     Report("WARNING: Symbolizer was blocked from starting itself!\n");
 491     failed_to_start_ = true;
 492     return nullptr;
 493   }
 494   for (; times_restarted_ < kMaxTimesRestarted; times_restarted_++) {
 495     // Start or restart symbolizer if we failed to send command to it.
 496     if (const char *res = SendCommandImpl(command))
 497       return res;
 498     Restart();
 499   }
 500   if (!failed_to_start_) {
 501     Report("WARNING: Failed to use and restart external symbolizer!\n");
 502     failed_to_start_ = true;
 503   }
 504   return nullptr;
 505 }
 506
 507 const char *SymbolizerProcess::SendCommandImpl(const char *command) {
 508   if (input_fd_ == kInvalidFd || output_fd_ == kInvalidFd)
 509       return nullptr;
 510   if (!WriteToSymbolizer(command, internal_strlen(command)))
 511       return nullptr;
 512   if (!ReadFromSymbolizer())
 513     return nullptr;
 514   return buffer_.data();
 515 }
 516
 517 bool SymbolizerProcess::Restart() {
 518   if (input_fd_ != kInvalidFd)
 519     CloseFile(input_fd_);
 520   if (output_fd_ != kInvalidFd)
 521     CloseFile(output_fd_);
 522   return StartSymbolizerSubprocess();
 523 }
 524
 525 bool SymbolizerProcess::ReadFromSymbolizer() {
 526   buffer_.clear();
 527   constexpr uptr max_length = 1024;
 528   bool ret = true;
 529   do {
 530     uptr just_read = 0;
 531     uptr size_before = buffer_.size();
 532     buffer_.resize(size_before + max_length);
 533     buffer_.resize(buffer_.capacity());
 534     bool ret = ReadFromFile(input_fd_, &buffer_[size_before],
 535                             buffer_.size() - size_before, &just_read);
 536
 537     if (!ret)
 538       just_read = 0;
 539
 540     buffer_.resize(size_before + just_read);
 541
 542     // We can't read 0 bytes, as we don't expect external symbolizer to close
 543     // its stdout.
 544     if (just_read == 0) {
 545       Report("WARNING: Can't read from symbolizer at fd %d\n", input_fd_);
 546       ret = false;
 547       break;
 548     }
 549   } while (!ReachedEndOfOutput(buffer_.data(), buffer_.size()));
 550   buffer_.push_back('\0');
 551   return ret;
 552 }
 553
 554 bool SymbolizerProcess::WriteToSymbolizer(const char *buffer, uptr length) {
 555   if (length == 0)
 556     return true;
 557   uptr write_len = 0;
 558   bool success = WriteToFile(output_fd_, buffer, length, &write_len);
 559   if (!success || write_len != length) {
 560     Report("WARNING: Can't write to symbolizer at fd %d\n", output_fd_);
 561     return false;
 562   }
 563   return true;
 564 }
 565
 566 #endif  // !SANITIZER_SYMBOLIZER_MARKUP
 567
 568 }  // namespace __sanitizer