1 //===-- ObjectContainerBSDArchive.cpp ---------------------------*- C++ -*-===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 10 #include "ObjectContainerBSDArchive.h" 11 12 #if defined(_WIN32) || defined(__ANDROID__) 13 // Defines from ar, missing on Windows 14 #define ARMAG "!<arch>\n" 15 #define SARMAG 8 16 #define ARFMAG "`\n" 17 18 typedef struct ar_hdr { 19 char ar_name[16]; 20 char ar_date[12]; 21 char ar_uid[6], ar_gid[6]; 22 char ar_mode[8]; 23 char ar_size[10]; 24 char ar_fmag[2]; 25 } ar_hdr; 26 #else 27 #include <ar.h> 28 #endif 29 30 #include "lldb/Core/ArchSpec.h" 31 #include "lldb/Core/Module.h" 32 #include "lldb/Core/ModuleSpec.h" 33 #include "lldb/Core/PluginManager.h" 34 #include "lldb/Core/Timer.h" 35 #include "lldb/Host/FileSystem.h" 36 #include "lldb/Symbol/ObjectFile.h" 37 #include "lldb/Utility/DataBufferLLVM.h" 38 #include "lldb/Utility/Stream.h" 39 40 #include "llvm/Support/MemoryBuffer.h" 41 42 using namespace lldb; 43 using namespace lldb_private; 44 45 ObjectContainerBSDArchive::Object::Object() 46 : ar_name(), ar_date(0), ar_uid(0), ar_gid(0), ar_mode(0), ar_size(0), 47 ar_file_offset(0), ar_file_size(0) {} 48 49 void ObjectContainerBSDArchive::Object::Clear() { 50 ar_name.Clear(); 51 ar_date = 0; 52 ar_uid = 0; 53 ar_gid = 0; 54 ar_mode = 0; 55 ar_size = 0; 56 ar_file_offset = 0; 57 ar_file_size = 0; 58 } 59 60 lldb::offset_t 61 ObjectContainerBSDArchive::Object::Extract(const DataExtractor &data, 62 lldb::offset_t offset) { 63 size_t ar_name_len = 0; 64 std::string str; 65 char *err; 66 67 // File header 68 // 69 // The common format is as follows. 70 // 71 // Offset Length Name Format 72 // 0 16 File name ASCII right padded with spaces (no spaces 73 // allowed in file name) 74 // 16 12 File mod Decimal as cstring right padded with 75 // spaces 76 // 28 6 Owner ID Decimal as cstring right padded with 77 // spaces 78 // 34 6 Group ID Decimal as cstring right padded with 79 // spaces 80 // 40 8 File mode Octal as cstring right padded with 81 // spaces 82 // 48 10 File byte size Decimal as cstring right padded with 83 // spaces 84 // 58 2 File magic 0x60 0x0A 85 86 // Make sure there is enough data for the file header and bail if not 87 if (!data.ValidOffsetForDataOfSize(offset, 60)) 88 return LLDB_INVALID_OFFSET; 89 90 str.assign((const char *)data.GetData(&offset, 16), 16); 91 if (str.find("#1/") == 0) { 92 // If the name is longer than 16 bytes, or contains an embedded space 93 // then it will use this format where the length of the name is 94 // here and the name characters are after this header. 95 ar_name_len = strtoul(str.c_str() + 3, &err, 10); 96 } else { 97 // Strip off any trailing spaces. 98 const size_t last_pos = str.find_last_not_of(' '); 99 if (last_pos != std::string::npos) { 100 if (last_pos + 1 < 16) 101 str.erase(last_pos + 1); 102 } 103 ar_name.SetCString(str.c_str()); 104 } 105 106 str.assign((const char *)data.GetData(&offset, 12), 12); 107 ar_date = strtoul(str.c_str(), &err, 10); 108 109 str.assign((const char *)data.GetData(&offset, 6), 6); 110 ar_uid = strtoul(str.c_str(), &err, 10); 111 112 str.assign((const char *)data.GetData(&offset, 6), 6); 113 ar_gid = strtoul(str.c_str(), &err, 10); 114 115 str.assign((const char *)data.GetData(&offset, 8), 8); 116 ar_mode = strtoul(str.c_str(), &err, 8); 117 118 str.assign((const char *)data.GetData(&offset, 10), 10); 119 ar_size = strtoul(str.c_str(), &err, 10); 120 121 str.assign((const char *)data.GetData(&offset, 2), 2); 122 if (str == ARFMAG) { 123 if (ar_name_len > 0) { 124 const void *ar_name_ptr = data.GetData(&offset, ar_name_len); 125 // Make sure there was enough data for the string value and bail if not 126 if (ar_name_ptr == NULL) 127 return LLDB_INVALID_OFFSET; 128 str.assign((const char *)ar_name_ptr, ar_name_len); 129 ar_name.SetCString(str.c_str()); 130 } 131 ar_file_offset = offset; 132 ar_file_size = ar_size - ar_name_len; 133 return offset; 134 } 135 return LLDB_INVALID_OFFSET; 136 } 137 138 ObjectContainerBSDArchive::Archive::Archive(const lldb_private::ArchSpec &arch, 139 const llvm::sys::TimePoint<> &time, 140 lldb::offset_t file_offset, 141 lldb_private::DataExtractor &data) 142 : m_arch(arch), m_time(time), m_file_offset(file_offset), m_objects(), 143 m_data(data) {} 144 145 ObjectContainerBSDArchive::Archive::~Archive() {} 146 147 size_t ObjectContainerBSDArchive::Archive::ParseObjects() { 148 DataExtractor &data = m_data; 149 std::string str; 150 lldb::offset_t offset = 0; 151 str.assign((const char *)data.GetData(&offset, SARMAG), SARMAG); 152 if (str == ARMAG) { 153 Object obj; 154 do { 155 offset = obj.Extract(data, offset); 156 if (offset == LLDB_INVALID_OFFSET) 157 break; 158 size_t obj_idx = m_objects.size(); 159 m_objects.push_back(obj); 160 // Insert all of the C strings out of order for now... 161 m_object_name_to_index_map.Append(obj.ar_name, obj_idx); 162 offset += obj.ar_file_size; 163 obj.Clear(); 164 } while (data.ValidOffset(offset)); 165 166 // Now sort all of the object name pointers 167 m_object_name_to_index_map.Sort(); 168 } 169 return m_objects.size(); 170 } 171 172 ObjectContainerBSDArchive::Object * 173 ObjectContainerBSDArchive::Archive::FindObject( 174 const ConstString &object_name, 175 const llvm::sys::TimePoint<> &object_mod_time) { 176 const ObjectNameToIndexMap::Entry *match = 177 m_object_name_to_index_map.FindFirstValueForName(object_name); 178 if (match) { 179 if (object_mod_time != llvm::sys::TimePoint<>()) { 180 const uint64_t object_date = llvm::sys::toTimeT(object_mod_time); 181 if (m_objects[match->value].ar_date == object_date) 182 return &m_objects[match->value]; 183 const ObjectNameToIndexMap::Entry *next_match = 184 m_object_name_to_index_map.FindNextValueForName(match); 185 while (next_match) { 186 if (m_objects[next_match->value].ar_date == object_date) 187 return &m_objects[next_match->value]; 188 next_match = 189 m_object_name_to_index_map.FindNextValueForName(next_match); 190 } 191 } else { 192 return &m_objects[match->value]; 193 } 194 } 195 return NULL; 196 } 197 198 ObjectContainerBSDArchive::Archive::shared_ptr 199 ObjectContainerBSDArchive::Archive::FindCachedArchive( 200 const FileSpec &file, const ArchSpec &arch, 201 const llvm::sys::TimePoint<> &time, lldb::offset_t file_offset) { 202 std::lock_guard<std::recursive_mutex> guard(Archive::GetArchiveCacheMutex()); 203 shared_ptr archive_sp; 204 Archive::Map &archive_map = Archive::GetArchiveCache(); 205 Archive::Map::iterator pos = archive_map.find(file); 206 // Don't cache a value for "archive_map.end()" below since we might 207 // delete an archive entry... 208 while (pos != archive_map.end() && pos->first == file) { 209 bool match = true; 210 if (arch.IsValid() && 211 pos->second->GetArchitecture().IsCompatibleMatch(arch) == false) 212 match = false; 213 else if (file_offset != LLDB_INVALID_OFFSET && 214 pos->second->GetFileOffset() != file_offset) 215 match = false; 216 if (match) { 217 if (pos->second->GetModificationTime() == time) { 218 return pos->second; 219 } else { 220 // We have a file at the same path with the same architecture 221 // whose modification time doesn't match. It doesn't make sense 222 // for us to continue to use this BSD archive since we cache only 223 // the object info which consists of file time info and also the 224 // file offset and file size of any contained objects. Since 225 // this information is now out of date, we won't get the correct 226 // information if we go and extract the file data, so we should 227 // remove the old and outdated entry. 228 archive_map.erase(pos); 229 pos = archive_map.find(file); 230 continue; // Continue to next iteration so we don't increment pos 231 // below... 232 } 233 } 234 ++pos; 235 } 236 return archive_sp; 237 } 238 239 ObjectContainerBSDArchive::Archive::shared_ptr 240 ObjectContainerBSDArchive::Archive::ParseAndCacheArchiveForFile( 241 const FileSpec &file, const ArchSpec &arch, 242 const llvm::sys::TimePoint<> &time, lldb::offset_t file_offset, 243 DataExtractor &data) { 244 shared_ptr archive_sp(new Archive(arch, time, file_offset, data)); 245 if (archive_sp) { 246 const size_t num_objects = archive_sp->ParseObjects(); 247 if (num_objects > 0) { 248 std::lock_guard<std::recursive_mutex> guard( 249 Archive::GetArchiveCacheMutex()); 250 Archive::GetArchiveCache().insert(std::make_pair(file, archive_sp)); 251 } else { 252 archive_sp.reset(); 253 } 254 } 255 return archive_sp; 256 } 257 258 ObjectContainerBSDArchive::Archive::Map & 259 ObjectContainerBSDArchive::Archive::GetArchiveCache() { 260 static Archive::Map g_archive_map; 261 return g_archive_map; 262 } 263 264 std::recursive_mutex & 265 ObjectContainerBSDArchive::Archive::GetArchiveCacheMutex() { 266 static std::recursive_mutex g_archive_map_mutex; 267 return g_archive_map_mutex; 268 } 269 270 void ObjectContainerBSDArchive::Initialize() { 271 PluginManager::RegisterPlugin(GetPluginNameStatic(), 272 GetPluginDescriptionStatic(), CreateInstance, 273 GetModuleSpecifications); 274 } 275 276 void ObjectContainerBSDArchive::Terminate() { 277 PluginManager::UnregisterPlugin(CreateInstance); 278 } 279 280 lldb_private::ConstString ObjectContainerBSDArchive::GetPluginNameStatic() { 281 static ConstString g_name("bsd-archive"); 282 return g_name; 283 } 284 285 const char *ObjectContainerBSDArchive::GetPluginDescriptionStatic() { 286 return "BSD Archive object container reader."; 287 } 288 289 ObjectContainer *ObjectContainerBSDArchive::CreateInstance( 290 const lldb::ModuleSP &module_sp, DataBufferSP &data_sp, 291 lldb::offset_t data_offset, const FileSpec *file, 292 lldb::offset_t file_offset, lldb::offset_t length) { 293 ConstString object_name(module_sp->GetObjectName()); 294 if (!object_name) 295 return nullptr; 296 297 if (data_sp) { 298 // We have data, which means this is the first 512 bytes of the file 299 // Check to see if the magic bytes match and if they do, read the entire 300 // table of contents for the archive and cache it 301 DataExtractor data; 302 data.SetData(data_sp, data_offset, length); 303 if (file && data_sp && ObjectContainerBSDArchive::MagicBytesMatch(data)) { 304 static Timer::Category func_cat(LLVM_PRETTY_FUNCTION); 305 Timer scoped_timer( 306 func_cat, 307 "ObjectContainerBSDArchive::CreateInstance (module = %s, file = " 308 "%p, file_offset = 0x%8.8" PRIx64 ", file_size = 0x%8.8" PRIx64 ")", 309 module_sp->GetFileSpec().GetPath().c_str(), 310 static_cast<const void *>(file), static_cast<uint64_t>(file_offset), 311 static_cast<uint64_t>(length)); 312 313 // Map the entire .a file to be sure that we don't lose any data if the 314 // file gets updated by a new build while this .a file is being used for 315 // debugging 316 DataBufferSP archive_data_sp = 317 DataBufferLLVM::CreateSliceFromPath(file->GetPath(), length, file_offset); 318 if (!archive_data_sp) 319 return nullptr; 320 321 lldb::offset_t archive_data_offset = 0; 322 323 Archive::shared_ptr archive_sp(Archive::FindCachedArchive( 324 *file, module_sp->GetArchitecture(), module_sp->GetModificationTime(), 325 file_offset)); 326 std::unique_ptr<ObjectContainerBSDArchive> container_ap( 327 new ObjectContainerBSDArchive(module_sp, archive_data_sp, 328 archive_data_offset, file, file_offset, 329 length)); 330 331 if (container_ap.get()) { 332 if (archive_sp) { 333 // We already have this archive in our cache, use it 334 container_ap->SetArchive(archive_sp); 335 return container_ap.release(); 336 } else if (container_ap->ParseHeader()) 337 return container_ap.release(); 338 } 339 } 340 } else { 341 // No data, just check for a cached archive 342 Archive::shared_ptr archive_sp(Archive::FindCachedArchive( 343 *file, module_sp->GetArchitecture(), module_sp->GetModificationTime(), 344 file_offset)); 345 if (archive_sp) { 346 std::unique_ptr<ObjectContainerBSDArchive> container_ap( 347 new ObjectContainerBSDArchive(module_sp, data_sp, data_offset, file, 348 file_offset, length)); 349 350 if (container_ap.get()) { 351 // We already have this archive in our cache, use it 352 container_ap->SetArchive(archive_sp); 353 return container_ap.release(); 354 } 355 } 356 } 357 return NULL; 358 } 359 360 bool ObjectContainerBSDArchive::MagicBytesMatch(const DataExtractor &data) { 361 uint32_t offset = 0; 362 const char *armag = (const char *)data.PeekData(offset, sizeof(ar_hdr)); 363 if (armag && ::strncmp(armag, ARMAG, SARMAG) == 0) { 364 armag += offsetof(struct ar_hdr, ar_fmag) + SARMAG; 365 if (strncmp(armag, ARFMAG, 2) == 0) 366 return true; 367 } 368 return false; 369 } 370 371 ObjectContainerBSDArchive::ObjectContainerBSDArchive( 372 const lldb::ModuleSP &module_sp, DataBufferSP &data_sp, 373 lldb::offset_t data_offset, const lldb_private::FileSpec *file, 374 lldb::offset_t file_offset, lldb::offset_t size) 375 : ObjectContainer(module_sp, file, file_offset, size, data_sp, data_offset), 376 m_archive_sp() {} 377 void ObjectContainerBSDArchive::SetArchive(Archive::shared_ptr &archive_sp) { 378 m_archive_sp = archive_sp; 379 } 380 381 ObjectContainerBSDArchive::~ObjectContainerBSDArchive() {} 382 383 bool ObjectContainerBSDArchive::ParseHeader() { 384 if (m_archive_sp.get() == NULL) { 385 if (m_data.GetByteSize() > 0) { 386 ModuleSP module_sp(GetModule()); 387 if (module_sp) { 388 m_archive_sp = Archive::ParseAndCacheArchiveForFile( 389 m_file, module_sp->GetArchitecture(), 390 module_sp->GetModificationTime(), m_offset, m_data); 391 } 392 // Clear the m_data that contains the entire archive 393 // data and let our m_archive_sp hold onto the data. 394 m_data.Clear(); 395 } 396 } 397 return m_archive_sp.get() != NULL; 398 } 399 400 void ObjectContainerBSDArchive::Dump(Stream *s) const { 401 s->Printf("%p: ", static_cast<const void *>(this)); 402 s->Indent(); 403 const size_t num_archs = GetNumArchitectures(); 404 const size_t num_objects = GetNumObjects(); 405 s->Printf("ObjectContainerBSDArchive, num_archs = %" PRIu64 406 ", num_objects = %" PRIu64 "", 407 (uint64_t)num_archs, (uint64_t)num_objects); 408 uint32_t i; 409 ArchSpec arch; 410 s->IndentMore(); 411 for (i = 0; i < num_archs; i++) { 412 s->Indent(); 413 GetArchitectureAtIndex(i, arch); 414 s->Printf("arch[%u] = %s\n", i, arch.GetArchitectureName()); 415 } 416 for (i = 0; i < num_objects; i++) { 417 s->Indent(); 418 s->Printf("object[%u] = %s\n", i, GetObjectNameAtIndex(i)); 419 } 420 s->IndentLess(); 421 s->EOL(); 422 } 423 424 ObjectFileSP ObjectContainerBSDArchive::GetObjectFile(const FileSpec *file) { 425 ModuleSP module_sp(GetModule()); 426 if (module_sp) { 427 if (module_sp->GetObjectName() && m_archive_sp) { 428 Object *object = m_archive_sp->FindObject( 429 module_sp->GetObjectName(), module_sp->GetObjectModificationTime()); 430 if (object) { 431 lldb::offset_t data_offset = object->ar_file_offset; 432 return ObjectFile::FindPlugin( 433 module_sp, file, m_offset + object->ar_file_offset, 434 object->ar_file_size, m_archive_sp->GetData().GetSharedDataBuffer(), 435 data_offset); 436 } 437 } 438 } 439 return ObjectFileSP(); 440 } 441 442 //------------------------------------------------------------------ 443 // PluginInterface protocol 444 //------------------------------------------------------------------ 445 lldb_private::ConstString ObjectContainerBSDArchive::GetPluginName() { 446 return GetPluginNameStatic(); 447 } 448 449 uint32_t ObjectContainerBSDArchive::GetPluginVersion() { return 1; } 450 451 size_t ObjectContainerBSDArchive::GetModuleSpecifications( 452 const lldb_private::FileSpec &file, lldb::DataBufferSP &data_sp, 453 lldb::offset_t data_offset, lldb::offset_t file_offset, 454 lldb::offset_t file_size, lldb_private::ModuleSpecList &specs) { 455 456 // We have data, which means this is the first 512 bytes of the file 457 // Check to see if the magic bytes match and if they do, read the entire 458 // table of contents for the archive and cache it 459 DataExtractor data; 460 data.SetData(data_sp, data_offset, data_sp->GetByteSize()); 461 if (!file || !data_sp || !ObjectContainerBSDArchive::MagicBytesMatch(data)) 462 return 0; 463 464 const size_t initial_count = specs.GetSize(); 465 llvm::sys::TimePoint<> file_mod_time = FileSystem::GetModificationTime(file); 466 Archive::shared_ptr archive_sp( 467 Archive::FindCachedArchive(file, ArchSpec(), file_mod_time, file_offset)); 468 bool set_archive_arch = false; 469 if (!archive_sp) { 470 set_archive_arch = true; 471 data_sp = 472 DataBufferLLVM::CreateSliceFromPath(file.GetPath(), file_size, file_offset); 473 if (data_sp) { 474 data.SetData(data_sp, 0, data_sp->GetByteSize()); 475 archive_sp = Archive::ParseAndCacheArchiveForFile( 476 file, ArchSpec(), file_mod_time, file_offset, data); 477 } 478 } 479 480 if (archive_sp) { 481 const size_t num_objects = archive_sp->GetNumObjects(); 482 for (size_t idx = 0; idx < num_objects; ++idx) { 483 const Object *object = archive_sp->GetObjectAtIndex(idx); 484 if (object) { 485 const lldb::offset_t object_file_offset = 486 file_offset + object->ar_file_offset; 487 if (object->ar_file_offset < file_size && 488 file_size > object_file_offset) { 489 if (ObjectFile::GetModuleSpecifications( 490 file, object_file_offset, file_size - object_file_offset, 491 specs)) { 492 ModuleSpec &spec = 493 specs.GetModuleSpecRefAtIndex(specs.GetSize() - 1); 494 llvm::sys::TimePoint<> object_mod_time( 495 std::chrono::seconds(object->ar_date)); 496 spec.GetObjectName() = object->ar_name; 497 spec.SetObjectOffset(object_file_offset); 498 spec.SetObjectSize(file_size - object_file_offset); 499 spec.GetObjectModificationTime() = object_mod_time; 500 } 501 } 502 } 503 } 504 } 505 const size_t end_count = specs.GetSize(); 506 size_t num_specs_added = end_count - initial_count; 507 if (set_archive_arch && num_specs_added > 0) { 508 // The archive was created but we didn't have an architecture 509 // so we need to set it 510 for (size_t i = initial_count; i < end_count; ++i) { 511 ModuleSpec module_spec; 512 if (specs.GetModuleSpecAtIndex(i, module_spec)) { 513 if (module_spec.GetArchitecture().IsValid()) { 514 archive_sp->SetArchitecture(module_spec.GetArchitecture()); 515 break; 516 } 517 } 518 } 519 } 520 return num_specs_added; 521 } 522