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