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 false;
 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 false;
 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   CHECK(name);
 163   Lock l(&mu_);
 164   for (auto &tool : tools_) {
 165     SymbolizerScope sym_scope(this);
 166     if (const char *demangled = tool.Demangle(name))
 167       return demangled;
 168   }
 169   if (const char *demangled = PlatformDemangle(name))
 170     return demangled;
 171   return name;
 172 }
 173
 174 bool Symbolizer::FindModuleNameAndOffsetForAddress(uptr address,
 175                                                    const char **module_name,
 176                                                    uptr *module_offset,
 177                                                    ModuleArch *module_arch) {
 178   const LoadedModule *module = FindModuleForAddress(address);
 179   if (!module)
 180     return false;
 181   *module_name = module->full_name();
 182   *module_offset = address - module->base_address();
 183   *module_arch = module->arch();
 184   return true;
 185 }
 186
 187 void Symbolizer::RefreshModules() {
 188   modules_.init();
 189   fallback_modules_.fallbackInit();
 190   RAW_CHECK(modules_.size() > 0);
 191   modules_fresh_ = true;
 192 }
 193
 194 static const LoadedModule *SearchForModule(const ListOfModules &modules,
 195                                            uptr address) {
 196   for (uptr i = 0; i < modules.size(); i++) {
 197     if (modules[i].containsAddress(address)) {
 198       return &modules[i];
 199     }
 200   }
 201   return nullptr;
 202 }
 203
 204 const LoadedModule *Symbolizer::FindModuleForAddress(uptr address) {
 205   bool modules_were_reloaded = false;
 206   if (!modules_fresh_) {
 207     RefreshModules();
 208     modules_were_reloaded = true;
 209   }
 210   const LoadedModule *module = SearchForModule(modules_, address);
 211   if (module) return module;
 212
 213   // dlopen/dlclose interceptors invalidate the module list, but when
 214   // interception is disabled, we need to retry if the lookup fails in
 215   // case the module list changed.
 216 #if !SANITIZER_INTERCEPT_DLOPEN_DLCLOSE
 217   if (!modules_were_reloaded) {
 218     RefreshModules();
 219     module = SearchForModule(modules_, address);
 220     if (module) return module;
 221   }
 222 #endif
 223
 224   if (fallback_modules_.size()) {
 225     module = SearchForModule(fallback_modules_, address);
 226   }
 227   return module;
 228 }
 229
 230 // For now we assume the following protocol:
 231 // For each request of the form
 232 //   <module_name> <module_offset>
 233 // passed to STDIN, external symbolizer prints to STDOUT response:
 234 //   <function_name>
 235 //   <file_name>:<line_number>:<column_number>
 236 //   <function_name>
 237 //   <file_name>:<line_number>:<column_number>
 238 //   ...
 239 //   <empty line>
 240 class LLVMSymbolizerProcess final : public SymbolizerProcess {
 241  public:
 242   explicit LLVMSymbolizerProcess(const char *path)
 243       : SymbolizerProcess(path, /*use_posix_spawn=*/SANITIZER_APPLE) {}
 244
 245  private:
 246   bool ReachedEndOfOutput(const char *buffer, uptr length) const override {
 247     // Empty line marks the end of llvm-symbolizer output.
 248     return length >= 2 && buffer[length - 1] == '\n' &&
 249            buffer[length - 2] == '\n';
 250   }
 251
 252   // When adding a new architecture, don't forget to also update
 253   // script/asan_symbolize.py and sanitizer_common.h.
 254   void GetArgV(const char *path_to_binary,
 255                const char *(&argv)[kArgVMax]) const override {
 256 #if defined(__x86_64h__)
 257     const char* const kSymbolizerArch = "--default-arch=x86_64h";
 258 #elif defined(__x86_64__)
 259     const char* const kSymbolizerArch = "--default-arch=x86_64";
 260 #elif defined(__i386__)
 261     const char* const kSymbolizerArch = "--default-arch=i386";
 262 #elif SANITIZER_LOONGARCH64
 263     const char *const kSymbolizerArch = "--default-arch=loongarch64";
 264 #elif SANITIZER_RISCV64
 265     const char *const kSymbolizerArch = "--default-arch=riscv64";
 266 #elif defined(__aarch64__)
 267     const char* const kSymbolizerArch = "--default-arch=arm64";
 268 #elif defined(__arm__)
 269     const char* const kSymbolizerArch = "--default-arch=arm";
 270 #elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
 271     const char* const kSymbolizerArch = "--default-arch=powerpc64";
 272 #elif defined(__powerpc64__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
 273     const char* const kSymbolizerArch = "--default-arch=powerpc64le";
 274 #elif defined(__s390x__)
 275     const char* const kSymbolizerArch = "--default-arch=s390x";
 276 #elif defined(__s390__)
 277     const char* const kSymbolizerArch = "--default-arch=s390";
 278 #else
 279     const char* const kSymbolizerArch = "--default-arch=unknown";
 280 #endif
 281
 282     const char *const demangle_flag =
 283         common_flags()->demangle ? "--demangle" : "--no-demangle";
 284     const char *const inline_flag =
 285         common_flags()->symbolize_inline_frames ? "--inlines" : "--no-inlines";
 286     int i = 0;
 287     argv[i++] = path_to_binary;
 288     argv[i++] = demangle_flag;
 289     argv[i++] = inline_flag;
 290     argv[i++] = kSymbolizerArch;
 291     argv[i++] = nullptr;
 292     CHECK_LE(i, kArgVMax);
 293   }
 294 };
 295
 296 LLVMSymbolizer::LLVMSymbolizer(const char *path, LowLevelAllocator *allocator)
 297     : symbolizer_process_(new(*allocator) LLVMSymbolizerProcess(path)) {}
 298
 299 // Parse a <file>:<line>[:<column>] buffer. The file path may contain colons on
 300 // Windows, so extract tokens from the right hand side first. The column info is
 301 // also optional.
 302 static const char *ParseFileLineInfo(AddressInfo *info, const char *str) {
 303   char *file_line_info = nullptr;
 304   str = ExtractToken(str, "\n", &file_line_info);
 305   CHECK(file_line_info);
 306
 307   if (uptr size = internal_strlen(file_line_info)) {
 308     char *back = file_line_info + size - 1;
 309     for (int i = 0; i < 2; ++i) {
 310       while (back > file_line_info && IsDigit(*back)) --back;
 311       if (*back != ':' || !IsDigit(back[1])) break;
 312       info->column = info->line;
 313       info->line = internal_atoll(back + 1);
 314       // Truncate the string at the colon to keep only filename.
 315       *back = '\0';
 316       --back;
 317     }
 318     ExtractToken(file_line_info, "", &info->file);
 319   }
 320
 321   InternalFree(file_line_info);
 322   return str;
 323 }
 324
 325 // Parses one or more two-line strings in the following format:
 326 //   <function_name>
 327 //   <file_name>:<line_number>[:<column_number>]
 328 // Used by LLVMSymbolizer, Addr2LinePool and InternalSymbolizer, since all of
 329 // them use the same output format.
 330 void ParseSymbolizePCOutput(const char *str, SymbolizedStack *res) {
 331   bool top_frame = true;
 332   SymbolizedStack *last = res;
 333   while (true) {
 334     char *function_name = nullptr;
 335     str = ExtractToken(str, "\n", &function_name);
 336     CHECK(function_name);
 337     if (function_name[0] == '\0') {
 338       // There are no more frames.
 339       InternalFree(function_name);
 340       break;
 341     }
 342     SymbolizedStack *cur;
 343     if (top_frame) {
 344       cur = res;
 345       top_frame = false;
 346     } else {
 347       cur = SymbolizedStack::New(res->info.address);
 348       cur->info.FillModuleInfo(res->info.module, res->info.module_offset,
 349                                res->info.module_arch);
 350       last->next = cur;
 351       last = cur;
 352     }
 353
 354     AddressInfo *info = &cur->info;
 355     info->function = function_name;
 356     str = ParseFileLineInfo(info, str);
 357
 358     // Functions and filenames can be "??", in which case we write 0
 359     // to address info to mark that names are unknown.
 360     if (0 == internal_strcmp(info->function, "??")) {
 361       InternalFree(info->function);
 362       info->function = 0;
 363     }
 364     if (info->file && 0 == internal_strcmp(info->file, "??")) {
 365       InternalFree(info->file);
 366       info->file = 0;
 367     }
 368   }
 369 }
 370
 371 // Parses a two- or three-line string in the following format:
 372 //   <symbol_name>
 373 //   <start_address> <size>
 374 //   <filename>:<column>
 375 // Used by LLVMSymbolizer and InternalSymbolizer. LLVMSymbolizer added support
 376 // for symbolizing the third line in D123538, but we support the older two-line
 377 // information as well.
 378 void ParseSymbolizeDataOutput(const char *str, DataInfo *info) {
 379   str = ExtractToken(str, "\n", &info->name);
 380   str = ExtractUptr(str, " ", &info->start);
 381   str = ExtractUptr(str, "\n", &info->size);
 382   // Note: If the third line isn't present, these calls will set info.{file,
 383   // line} to empty strings.
 384   str = ExtractToken(str, ":", &info->file);
 385   str = ExtractUptr(str, "\n", &info->line);
 386 }
 387
 388 void ParseSymbolizeFrameOutput(const char *str,
 389                                InternalMmapVector<LocalInfo> *locals) {
 390   if (internal_strncmp(str, "??", 2) == 0)
 391     return;
 392
 393   while (*str) {
 394     LocalInfo local;
 395     str = ExtractToken(str, "\n", &local.function_name);
 396     str = ExtractToken(str, "\n", &local.name);
 397
 398     AddressInfo addr;
 399     str = ParseFileLineInfo(&addr, str);
 400     local.decl_file = addr.file;
 401     local.decl_line = addr.line;
 402
 403     local.has_frame_offset = internal_strncmp(str, "??", 2) != 0;
 404     str = ExtractSptr(str, " ", &local.frame_offset);
 405
 406     local.has_size = internal_strncmp(str, "??", 2) != 0;
 407     str = ExtractUptr(str, " ", &local.size);
 408
 409     local.has_tag_offset = internal_strncmp(str, "??", 2) != 0;
 410     str = ExtractUptr(str, "\n", &local.tag_offset);
 411
 412     locals->push_back(local);
 413   }
 414 }
 415
 416 bool LLVMSymbolizer::SymbolizePC(uptr addr, SymbolizedStack *stack) {
 417   AddressInfo *info = &stack->info;
 418   const char *buf = FormatAndSendCommand(
 419       "CODE", info->module, info->module_offset, info->module_arch);
 420   if (!buf)
 421     return false;
 422   ParseSymbolizePCOutput(buf, stack);
 423   return true;
 424 }
 425
 426 bool LLVMSymbolizer::SymbolizeData(uptr addr, DataInfo *info) {
 427   const char *buf = FormatAndSendCommand(
 428       "DATA", info->module, info->module_offset, info->module_arch);
 429   if (!buf)
 430     return false;
 431   ParseSymbolizeDataOutput(buf, info);
 432   info->start += (addr - info->module_offset); // Add the base address.
 433   return true;
 434 }
 435
 436 bool LLVMSymbolizer::SymbolizeFrame(uptr addr, FrameInfo *info) {
 437   const char *buf = FormatAndSendCommand(
 438       "FRAME", info->module, info->module_offset, info->module_arch);
 439   if (!buf)
 440     return false;
 441   ParseSymbolizeFrameOutput(buf, &info->locals);
 442   return true;
 443 }
 444
 445 const char *LLVMSymbolizer::FormatAndSendCommand(const char *command_prefix,
 446                                                  const char *module_name,
 447                                                  uptr module_offset,
 448                                                  ModuleArch arch) {
 449   CHECK(module_name);
 450   int size_needed = 0;
 451   if (arch == kModuleArchUnknown)
 452     size_needed = internal_snprintf(buffer_, kBufferSize, "%s \"%s\" 0x%zx\n",
 453                                     command_prefix, module_name, module_offset);
 454   else
 455     size_needed = internal_snprintf(buffer_, kBufferSize,
 456                                     "%s \"%s:%s\" 0x%zx\n", command_prefix,
 457                                     module_name, ModuleArchToString(arch),
 458                                     module_offset);
 459
 460   if (size_needed >= static_cast<int>(kBufferSize)) {
 461     Report("WARNING: Command buffer too small");
 462     return nullptr;
 463   }
 464
 465   return symbolizer_process_->SendCommand(buffer_);
 466 }
 467
 468 SymbolizerProcess::SymbolizerProcess(const char *path, bool use_posix_spawn)
 469     : path_(path),
 470       input_fd_(kInvalidFd),
 471       output_fd_(kInvalidFd),
 472       times_restarted_(0),
 473       failed_to_start_(false),
 474       reported_invalid_path_(false),
 475       use_posix_spawn_(use_posix_spawn) {
 476   CHECK(path_);
 477   CHECK_NE(path_[0], '\0');
 478 }
 479
 480 static bool IsSameModule(const char* path) {
 481   if (const char* ProcessName = GetProcessName()) {
 482     if (const char* SymbolizerName = StripModuleName(path)) {
 483       return !internal_strcmp(ProcessName, SymbolizerName);
 484     }
 485   }
 486   return false;
 487 }
 488
 489 const char *SymbolizerProcess::SendCommand(const char *command) {
 490   if (failed_to_start_)
 491     return nullptr;
 492   if (IsSameModule(path_)) {
 493     Report("WARNING: Symbolizer was blocked from starting itself!\n");
 494     failed_to_start_ = true;
 495     return nullptr;
 496   }
 497   for (; times_restarted_ < kMaxTimesRestarted; times_restarted_++) {
 498     // Start or restart symbolizer if we failed to send command to it.
 499     if (const char *res = SendCommandImpl(command))
 500       return res;
 501     Restart();
 502   }
 503   if (!failed_to_start_) {
 504     Report("WARNING: Failed to use and restart external symbolizer!\n");
 505     failed_to_start_ = true;
 506   }
 507   return nullptr;
 508 }
 509
 510 const char *SymbolizerProcess::SendCommandImpl(const char *command) {
 511   if (input_fd_ == kInvalidFd || output_fd_ == kInvalidFd)
 512       return nullptr;
 513   if (!WriteToSymbolizer(command, internal_strlen(command)))
 514       return nullptr;
 515   if (!ReadFromSymbolizer())
 516     return nullptr;
 517   return buffer_.data();
 518 }
 519
 520 bool SymbolizerProcess::Restart() {
 521   if (input_fd_ != kInvalidFd)
 522     CloseFile(input_fd_);
 523   if (output_fd_ != kInvalidFd)
 524     CloseFile(output_fd_);
 525   return StartSymbolizerSubprocess();
 526 }
 527
 528 bool SymbolizerProcess::ReadFromSymbolizer() {
 529   buffer_.clear();
 530   constexpr uptr max_length = 1024;
 531   bool ret = true;
 532   do {
 533     uptr just_read = 0;
 534     uptr size_before = buffer_.size();
 535     buffer_.resize(size_before + max_length);
 536     buffer_.resize(buffer_.capacity());
 537     bool ret = ReadFromFile(input_fd_, &buffer_[size_before],
 538                             buffer_.size() - size_before, &just_read);
 539
 540     if (!ret)
 541       just_read = 0;
 542
 543     buffer_.resize(size_before + just_read);
 544
 545     // We can't read 0 bytes, as we don't expect external symbolizer to close
 546     // its stdout.
 547     if (just_read == 0) {
 548       Report("WARNING: Can't read from symbolizer at fd %d\n", input_fd_);
 549       ret = false;
 550       break;
 551     }
 552   } while (!ReachedEndOfOutput(buffer_.data(), buffer_.size()));
 553   buffer_.push_back('\0');
 554   return ret;
 555 }
 556
 557 bool SymbolizerProcess::WriteToSymbolizer(const char *buffer, uptr length) {
 558   if (length == 0)
 559     return true;
 560   uptr write_len = 0;
 561   bool success = WriteToFile(output_fd_, buffer, length, &write_len);
 562   if (!success || write_len != length) {
 563     Report("WARNING: Can't write to symbolizer at fd %d\n", output_fd_);
 564     return false;
 565   }
 566   return true;
 567 }
 568
 569 #endif  // !SANITIZER_SYMBOLIZER_MARKUP
 570
 571 }  // namespace __sanitizer