libsanitizer/sanitizer_common/sanitizer_symbolizer_libcdep.cc

   1 //===-- sanitizer_symbolizer_libcdep.cc -----------------------------------===//
   2 //
   3 // This file is distributed under the University of Illinois Open Source
   4 // License. See LICENSE.TXT for details.
   5 //
   6 //===----------------------------------------------------------------------===//
   7 //
   8 // This file is shared between AddressSanitizer and ThreadSanitizer
   9 // run-time libraries.
  10 //===----------------------------------------------------------------------===//
  11
  12 #include "sanitizer_allocator_internal.h"
  13 #include "sanitizer_internal_defs.h"
  14 #include "sanitizer_symbolizer_internal.h"
  15
  16 namespace __sanitizer {
  17
  18 const char *ExtractToken(const char *str, const char *delims, char **result) {
  19   uptr prefix_len = internal_strcspn(str, delims);
  20   *result = (char*)InternalAlloc(prefix_len + 1);
  21   internal_memcpy(*result, str, prefix_len);
  22   (*result)[prefix_len] = '\0';
  23   const char *prefix_end = str + prefix_len;
  24   if (*prefix_end != '\0') prefix_end++;
  25   return prefix_end;
  26 }
  27
  28 const char *ExtractInt(const char *str, const char *delims, int *result) {
  29   char *buff;
  30   const char *ret = ExtractToken(str, delims, &buff);
  31   if (buff != 0) {
  32     *result = (int)internal_atoll(buff);
  33   }
  34   InternalFree(buff);
  35   return ret;
  36 }
  37
  38 const char *ExtractUptr(const char *str, const char *delims, uptr *result) {
  39   char *buff;
  40   const char *ret = ExtractToken(str, delims, &buff);
  41   if (buff != 0) {
  42     *result = (uptr)internal_atoll(buff);
  43   }
  44   InternalFree(buff);
  45   return ret;
  46 }
  47
  48 const char *ExtractTokenUpToDelimiter(const char *str, const char *delimiter,
  49                                       char **result) {
  50   const char *found_delimiter = internal_strstr(str, delimiter);
  51   uptr prefix_len =
  52       found_delimiter ? found_delimiter - str : internal_strlen(str);
  53   *result = (char *)InternalAlloc(prefix_len + 1);
  54   internal_memcpy(*result, str, prefix_len);
  55   (*result)[prefix_len] = '\0';
  56   const char *prefix_end = str + prefix_len;
  57   if (*prefix_end != '\0') prefix_end += internal_strlen(delimiter);
  58   return prefix_end;
  59 }
  60
  61 SymbolizedStack *Symbolizer::SymbolizePC(uptr addr) {
  62   BlockingMutexLock l(&mu_);
  63   const char *module_name;
  64   uptr module_offset;
  65   SymbolizedStack *res = SymbolizedStack::New(addr);
  66   if (!FindModuleNameAndOffsetForAddress(addr, &module_name, &module_offset))
  67     return res;
  68   // Always fill data about module name and offset.
  69   res->info.FillModuleInfo(module_name, module_offset);
  70   for (auto iter = Iterator(&tools_); iter.hasNext();) {
  71     auto *tool = iter.next();
  72     SymbolizerScope sym_scope(this);
  73     if (tool->SymbolizePC(addr, res)) {
  74       return res;
  75     }
  76   }
  77   return res;
  78 }
  79
  80 bool Symbolizer::SymbolizeData(uptr addr, DataInfo *info) {
  81   BlockingMutexLock l(&mu_);
  82   const char *module_name;
  83   uptr module_offset;
  84   if (!FindModuleNameAndOffsetForAddress(addr, &module_name, &module_offset))
  85     return false;
  86   info->Clear();
  87   info->module = internal_strdup(module_name);
  88   info->module_offset = module_offset;
  89   for (auto iter = Iterator(&tools_); iter.hasNext();) {
  90     auto *tool = iter.next();
  91     SymbolizerScope sym_scope(this);
  92     if (tool->SymbolizeData(addr, info)) {
  93       return true;
  94     }
  95   }
  96   return true;
  97 }
  98
  99 bool Symbolizer::GetModuleNameAndOffsetForPC(uptr pc, const char **module_name,
 100                                              uptr *module_address) {
 101   BlockingMutexLock l(&mu_);
 102   const char *internal_module_name = nullptr;
 103   if (!FindModuleNameAndOffsetForAddress(pc, &internal_module_name,
 104                                          module_address))
 105     return false;
 106
 107   if (module_name)
 108     *module_name = module_names_.GetOwnedCopy(internal_module_name);
 109   return true;
 110 }
 111
 112 void Symbolizer::Flush() {
 113   BlockingMutexLock l(&mu_);
 114   for (auto iter = Iterator(&tools_); iter.hasNext();) {
 115     auto *tool = iter.next();
 116     SymbolizerScope sym_scope(this);
 117     tool->Flush();
 118   }
 119 }
 120
 121 const char *Symbolizer::Demangle(const char *name) {
 122   BlockingMutexLock l(&mu_);
 123   for (auto iter = Iterator(&tools_); iter.hasNext();) {
 124     auto *tool = iter.next();
 125     SymbolizerScope sym_scope(this);
 126     if (const char *demangled = tool->Demangle(name))
 127       return demangled;
 128   }
 129   return PlatformDemangle(name);
 130 }
 131
 132 void Symbolizer::PrepareForSandboxing() {
 133   BlockingMutexLock l(&mu_);
 134   PlatformPrepareForSandboxing();
 135 }
 136
 137 bool Symbolizer::FindModuleNameAndOffsetForAddress(uptr address,
 138                                                    const char **module_name,
 139                                                    uptr *module_offset) {
 140   LoadedModule *module = FindModuleForAddress(address);
 141   if (module == 0)
 142     return false;
 143   *module_name = module->full_name();
 144   *module_offset = address - module->base_address();
 145   return true;
 146 }
 147
 148 LoadedModule *Symbolizer::FindModuleForAddress(uptr address) {
 149   bool modules_were_reloaded = false;
 150   if (!modules_fresh_) {
 151     for (uptr i = 0; i < n_modules_; i++)
 152       modules_[i].clear();
 153     n_modules_ =
 154         GetListOfModules(modules_, kMaxNumberOfModules, /* filter */ nullptr);
 155     CHECK_GT(n_modules_, 0);
 156     CHECK_LT(n_modules_, kMaxNumberOfModules);
 157     modules_fresh_ = true;
 158     modules_were_reloaded = true;
 159   }
 160   for (uptr i = 0; i < n_modules_; i++) {
 161     if (modules_[i].containsAddress(address)) {
 162       return &modules_[i];
 163     }
 164   }
 165   // Reload the modules and look up again, if we haven't tried it yet.
 166   if (!modules_were_reloaded) {
 167     // FIXME: set modules_fresh_ from dlopen()/dlclose() interceptors.
 168     // It's too aggressive to reload the list of modules each time we fail
 169     // to find a module for a given address.
 170     modules_fresh_ = false;
 171     return FindModuleForAddress(address);
 172   }
 173   return 0;
 174 }
 175
 176 Symbolizer *Symbolizer::GetOrInit() {
 177   SpinMutexLock l(&init_mu_);
 178   if (symbolizer_)
 179     return symbolizer_;
 180   symbolizer_ = PlatformInit();
 181   CHECK(symbolizer_);
 182   return symbolizer_;
 183 }
 184
 185 // For now we assume the following protocol:
 186 // For each request of the form
 187 //   <module_name> <module_offset>
 188 // passed to STDIN, external symbolizer prints to STDOUT response:
 189 //   <function_name>
 190 //   <file_name>:<line_number>:<column_number>
 191 //   <function_name>
 192 //   <file_name>:<line_number>:<column_number>
 193 //   ...
 194 //   <empty line>
 195 class LLVMSymbolizerProcess : public SymbolizerProcess {
 196  public:
 197   explicit LLVMSymbolizerProcess(const char *path) : SymbolizerProcess(path) {}
 198
 199  private:
 200   bool ReachedEndOfOutput(const char *buffer, uptr length) const override {
 201     // Empty line marks the end of llvm-symbolizer output.
 202     return length >= 2 && buffer[length - 1] == '\n' &&
 203            buffer[length - 2] == '\n';
 204   }
 205
 206   void GetArgV(const char *path_to_binary,
 207                const char *(&argv)[kArgVMax]) const override {
 208 #if defined(__x86_64h__)
 209     const char* const kSymbolizerArch = "--default-arch=x86_64h";
 210 #elif defined(__x86_64__)
 211     const char* const kSymbolizerArch = "--default-arch=x86_64";
 212 #elif defined(__i386__)
 213     const char* const kSymbolizerArch = "--default-arch=i386";
 214 #elif defined(__powerpc64__) && defined(__BIG_ENDIAN__)
 215     const char* const kSymbolizerArch = "--default-arch=powerpc64";
 216 #elif defined(__powerpc64__) && defined(__LITTLE_ENDIAN__)
 217     const char* const kSymbolizerArch = "--default-arch=powerpc64le";
 218 #else
 219     const char* const kSymbolizerArch = "--default-arch=unknown";
 220 #endif
 221
 222     const char *const inline_flag = common_flags()->symbolize_inline_frames
 223                                         ? "--inlining=true"
 224                                         : "--inlining=false";
 225     int i = 0;
 226     argv[i++] = path_to_binary;
 227     argv[i++] = inline_flag;
 228     argv[i++] = kSymbolizerArch;
 229     argv[i++] = nullptr;
 230   }
 231 };
 232
 233 LLVMSymbolizer::LLVMSymbolizer(const char *path, LowLevelAllocator *allocator)
 234     : symbolizer_process_(new(*allocator) LLVMSymbolizerProcess(path)) {}
 235
 236 // Parse a <file>:<line>[:<column>] buffer. The file path may contain colons on
 237 // Windows, so extract tokens from the right hand side first. The column info is
 238 // also optional.
 239 static const char *ParseFileLineInfo(AddressInfo *info, const char *str) {
 240   char *file_line_info = 0;
 241   str = ExtractToken(str, "\n", &file_line_info);
 242   CHECK(file_line_info);
 243   // Parse the last :<int>, which must be there.
 244   char *last_colon = internal_strrchr(file_line_info, ':');
 245   CHECK(last_colon);
 246   int line_or_column = internal_atoll(last_colon + 1);
 247   // Truncate the string at the last colon and find the next-to-last colon.
 248   *last_colon = '\0';
 249   last_colon = internal_strrchr(file_line_info, ':');
 250   if (last_colon && IsDigit(last_colon[1])) {
 251     // If the second-to-last colon is followed by a digit, it must be the line
 252     // number, and the previous parsed number was a column.
 253     info->line = internal_atoll(last_colon + 1);
 254     info->column = line_or_column;
 255     *last_colon = '\0';
 256   } else {
 257     // Otherwise, we have line info but no column info.
 258     info->line = line_or_column;
 259     info->column = 0;
 260   }
 261   ExtractToken(file_line_info, "", &info->file);
 262   InternalFree(file_line_info);
 263   return str;
 264 }
 265
 266 // Parses one or more two-line strings in the following format:
 267 //   <function_name>
 268 //   <file_name>:<line_number>[:<column_number>]
 269 // Used by LLVMSymbolizer, Addr2LinePool and InternalSymbolizer, since all of
 270 // them use the same output format.
 271 void ParseSymbolizePCOutput(const char *str, SymbolizedStack *res) {
 272   bool top_frame = true;
 273   SymbolizedStack *last = res;
 274   while (true) {
 275     char *function_name = 0;
 276     str = ExtractToken(str, "\n", &function_name);
 277     CHECK(function_name);
 278     if (function_name[0] == '\0') {
 279       // There are no more frames.
 280       InternalFree(function_name);
 281       break;
 282     }
 283     SymbolizedStack *cur;
 284     if (top_frame) {
 285       cur = res;
 286       top_frame = false;
 287     } else {
 288       cur = SymbolizedStack::New(res->info.address);
 289       cur->info.FillModuleInfo(res->info.module, res->info.module_offset);
 290       last->next = cur;
 291       last = cur;
 292     }
 293
 294     AddressInfo *info = &cur->info;
 295     info->function = function_name;
 296     str = ParseFileLineInfo(info, str);
 297
 298     // Functions and filenames can be "??", in which case we write 0
 299     // to address info to mark that names are unknown.
 300     if (0 == internal_strcmp(info->function, "??")) {
 301       InternalFree(info->function);
 302       info->function = 0;
 303     }
 304     if (0 == internal_strcmp(info->file, "??")) {
 305       InternalFree(info->file);
 306       info->file = 0;
 307     }
 308   }
 309 }
 310
 311 // Parses a two-line string in the following format:
 312 //   <symbol_name>
 313 //   <start_address> <size>
 314 // Used by LLVMSymbolizer and InternalSymbolizer.
 315 void ParseSymbolizeDataOutput(const char *str, DataInfo *info) {
 316   str = ExtractToken(str, "\n", &info->name);
 317   str = ExtractUptr(str, " ", &info->start);
 318   str = ExtractUptr(str, "\n", &info->size);
 319 }
 320
 321 bool LLVMSymbolizer::SymbolizePC(uptr addr, SymbolizedStack *stack) {
 322   if (const char *buf = SendCommand(/*is_data*/ false, stack->info.module,
 323                                     stack->info.module_offset)) {
 324     ParseSymbolizePCOutput(buf, stack);
 325     return true;
 326   }
 327   return false;
 328 }
 329
 330 bool LLVMSymbolizer::SymbolizeData(uptr addr, DataInfo *info) {
 331   if (const char *buf =
 332           SendCommand(/*is_data*/ true, info->module, info->module_offset)) {
 333     ParseSymbolizeDataOutput(buf, info);
 334     info->start += (addr - info->module_offset); // Add the base address.
 335     return true;
 336   }
 337   return false;
 338 }
 339
 340 const char *LLVMSymbolizer::SendCommand(bool is_data, const char *module_name,
 341                                         uptr module_offset) {
 342   CHECK(module_name);
 343   internal_snprintf(buffer_, kBufferSize, "%s\"%s\" 0x%zx\n",
 344                     is_data ? "DATA " : "", module_name, module_offset);
 345   return symbolizer_process_->SendCommand(buffer_);
 346 }
 347
 348 SymbolizerProcess::SymbolizerProcess(const char *path, bool use_forkpty)
 349     : path_(path),
 350       input_fd_(kInvalidFd),
 351       output_fd_(kInvalidFd),
 352       times_restarted_(0),
 353       failed_to_start_(false),
 354       reported_invalid_path_(false),
 355       use_forkpty_(use_forkpty) {
 356   CHECK(path_);
 357   CHECK_NE(path_[0], '\0');
 358 }
 359
 360 const char *SymbolizerProcess::SendCommand(const char *command) {
 361   for (; times_restarted_ < kMaxTimesRestarted; times_restarted_++) {
 362     // Start or restart symbolizer if we failed to send command to it.
 363     if (const char *res = SendCommandImpl(command))
 364       return res;
 365     Restart();
 366   }
 367   if (!failed_to_start_) {
 368     Report("WARNING: Failed to use and restart external symbolizer!\n");
 369     failed_to_start_ = true;
 370   }
 371   return 0;
 372 }
 373
 374 const char *SymbolizerProcess::SendCommandImpl(const char *command) {
 375   if (input_fd_ == kInvalidFd || output_fd_ == kInvalidFd)
 376       return 0;
 377   if (!WriteToSymbolizer(command, internal_strlen(command)))
 378       return 0;
 379   if (!ReadFromSymbolizer(buffer_, kBufferSize))
 380       return 0;
 381   return buffer_;
 382 }
 383
 384 bool SymbolizerProcess::Restart() {
 385   if (input_fd_ != kInvalidFd)
 386     CloseFile(input_fd_);
 387   if (output_fd_ != kInvalidFd)
 388     CloseFile(output_fd_);
 389   return StartSymbolizerSubprocess();
 390 }
 391
 392 bool SymbolizerProcess::ReadFromSymbolizer(char *buffer, uptr max_length) {
 393   if (max_length == 0)
 394     return true;
 395   uptr read_len = 0;
 396   while (true) {
 397     uptr just_read = 0;
 398     bool success = ReadFromFile(input_fd_, buffer + read_len,
 399                                 max_length - read_len - 1, &just_read);
 400     // We can't read 0 bytes, as we don't expect external symbolizer to close
 401     // its stdout.
 402     if (!success || just_read == 0) {
 403       Report("WARNING: Can't read from symbolizer at fd %d\n", input_fd_);
 404       return false;
 405     }
 406     read_len += just_read;
 407     if (ReachedEndOfOutput(buffer, read_len))
 408       break;
 409   }
 410   buffer[read_len] = '\0';
 411   return true;
 412 }
 413
 414 bool SymbolizerProcess::WriteToSymbolizer(const char *buffer, uptr length) {
 415   if (length == 0)
 416     return true;
 417   uptr write_len = 0;
 418   bool success = WriteToFile(output_fd_, buffer, length, &write_len);
 419   if (!success || write_len != length) {
 420     Report("WARNING: Can't write to symbolizer at fd %d\n", output_fd_);
 421     return false;
 422   }
 423   return true;
 424 }
 425
 426 }  // namespace __sanitizer