1 //===- BitcodeAnalyzer.cpp - Internal BitcodeAnalyzer implementation ------===// 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 #include "llvm/Bitcode/BitcodeAnalyzer.h" 10 #include "llvm/Bitcode/BitcodeReader.h" 11 #include "llvm/Bitcode/LLVMBitCodes.h" 12 #include "llvm/Bitstream/BitCodes.h" 13 #include "llvm/Bitstream/BitstreamReader.h" 14 #include "llvm/Support/Format.h" 15 #include "llvm/Support/SHA1.h" 16 17 using namespace llvm; 18 19 static Error reportError(StringRef Message) { 20 return createStringError(std::errc::illegal_byte_sequence, Message.data()); 21 } 22 23 /// Return a symbolic block name if known, otherwise return null. 24 static Optional<const char *> GetBlockName(unsigned BlockID, 25 const BitstreamBlockInfo &BlockInfo, 26 CurStreamTypeType CurStreamType) { 27 // Standard blocks for all bitcode files. 28 if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) { 29 if (BlockID == bitc::BLOCKINFO_BLOCK_ID) 30 return "BLOCKINFO_BLOCK"; 31 return None; 32 } 33 34 // Check to see if we have a blockinfo record for this block, with a name. 35 if (const BitstreamBlockInfo::BlockInfo *Info = 36 BlockInfo.getBlockInfo(BlockID)) { 37 if (!Info->Name.empty()) 38 return Info->Name.c_str(); 39 } 40 41 if (CurStreamType != LLVMIRBitstream) 42 return None; 43 44 switch (BlockID) { 45 default: 46 return None; 47 case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID: 48 return "OPERAND_BUNDLE_TAGS_BLOCK"; 49 case bitc::MODULE_BLOCK_ID: 50 return "MODULE_BLOCK"; 51 case bitc::PARAMATTR_BLOCK_ID: 52 return "PARAMATTR_BLOCK"; 53 case bitc::PARAMATTR_GROUP_BLOCK_ID: 54 return "PARAMATTR_GROUP_BLOCK_ID"; 55 case bitc::TYPE_BLOCK_ID_NEW: 56 return "TYPE_BLOCK_ID"; 57 case bitc::CONSTANTS_BLOCK_ID: 58 return "CONSTANTS_BLOCK"; 59 case bitc::FUNCTION_BLOCK_ID: 60 return "FUNCTION_BLOCK"; 61 case bitc::IDENTIFICATION_BLOCK_ID: 62 return "IDENTIFICATION_BLOCK_ID"; 63 case bitc::VALUE_SYMTAB_BLOCK_ID: 64 return "VALUE_SYMTAB"; 65 case bitc::METADATA_BLOCK_ID: 66 return "METADATA_BLOCK"; 67 case bitc::METADATA_KIND_BLOCK_ID: 68 return "METADATA_KIND_BLOCK"; 69 case bitc::METADATA_ATTACHMENT_ID: 70 return "METADATA_ATTACHMENT_BLOCK"; 71 case bitc::USELIST_BLOCK_ID: 72 return "USELIST_BLOCK_ID"; 73 case bitc::GLOBALVAL_SUMMARY_BLOCK_ID: 74 return "GLOBALVAL_SUMMARY_BLOCK"; 75 case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID: 76 return "FULL_LTO_GLOBALVAL_SUMMARY_BLOCK"; 77 case bitc::MODULE_STRTAB_BLOCK_ID: 78 return "MODULE_STRTAB_BLOCK"; 79 case bitc::STRTAB_BLOCK_ID: 80 return "STRTAB_BLOCK"; 81 case bitc::SYMTAB_BLOCK_ID: 82 return "SYMTAB_BLOCK"; 83 } 84 } 85 86 /// Return a symbolic code name if known, otherwise return null. 87 static Optional<const char *> GetCodeName(unsigned CodeID, unsigned BlockID, 88 const BitstreamBlockInfo &BlockInfo, 89 CurStreamTypeType CurStreamType) { 90 // Standard blocks for all bitcode files. 91 if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) { 92 if (BlockID == bitc::BLOCKINFO_BLOCK_ID) { 93 switch (CodeID) { 94 default: 95 return None; 96 case bitc::BLOCKINFO_CODE_SETBID: 97 return "SETBID"; 98 case bitc::BLOCKINFO_CODE_BLOCKNAME: 99 return "BLOCKNAME"; 100 case bitc::BLOCKINFO_CODE_SETRECORDNAME: 101 return "SETRECORDNAME"; 102 } 103 } 104 return None; 105 } 106 107 // Check to see if we have a blockinfo record for this record, with a name. 108 if (const BitstreamBlockInfo::BlockInfo *Info = 109 BlockInfo.getBlockInfo(BlockID)) { 110 for (unsigned i = 0, e = Info->RecordNames.size(); i != e; ++i) 111 if (Info->RecordNames[i].first == CodeID) 112 return Info->RecordNames[i].second.c_str(); 113 } 114 115 if (CurStreamType != LLVMIRBitstream) 116 return None; 117 118 #define STRINGIFY_CODE(PREFIX, CODE) \ 119 case bitc::PREFIX##_##CODE: \ 120 return #CODE; 121 switch (BlockID) { 122 default: 123 return None; 124 case bitc::MODULE_BLOCK_ID: 125 switch (CodeID) { 126 default: 127 return None; 128 STRINGIFY_CODE(MODULE_CODE, VERSION) 129 STRINGIFY_CODE(MODULE_CODE, TRIPLE) 130 STRINGIFY_CODE(MODULE_CODE, DATALAYOUT) 131 STRINGIFY_CODE(MODULE_CODE, ASM) 132 STRINGIFY_CODE(MODULE_CODE, SECTIONNAME) 133 STRINGIFY_CODE(MODULE_CODE, DEPLIB) // Deprecated, present in old bitcode 134 STRINGIFY_CODE(MODULE_CODE, GLOBALVAR) 135 STRINGIFY_CODE(MODULE_CODE, FUNCTION) 136 STRINGIFY_CODE(MODULE_CODE, ALIAS) 137 STRINGIFY_CODE(MODULE_CODE, GCNAME) 138 STRINGIFY_CODE(MODULE_CODE, COMDAT) 139 STRINGIFY_CODE(MODULE_CODE, VSTOFFSET) 140 STRINGIFY_CODE(MODULE_CODE, METADATA_VALUES_UNUSED) 141 STRINGIFY_CODE(MODULE_CODE, SOURCE_FILENAME) 142 STRINGIFY_CODE(MODULE_CODE, HASH) 143 } 144 case bitc::IDENTIFICATION_BLOCK_ID: 145 switch (CodeID) { 146 default: 147 return None; 148 STRINGIFY_CODE(IDENTIFICATION_CODE, STRING) 149 STRINGIFY_CODE(IDENTIFICATION_CODE, EPOCH) 150 } 151 case bitc::PARAMATTR_BLOCK_ID: 152 switch (CodeID) { 153 default: 154 return None; 155 // FIXME: Should these be different? 156 case bitc::PARAMATTR_CODE_ENTRY_OLD: 157 return "ENTRY"; 158 case bitc::PARAMATTR_CODE_ENTRY: 159 return "ENTRY"; 160 } 161 case bitc::PARAMATTR_GROUP_BLOCK_ID: 162 switch (CodeID) { 163 default: 164 return None; 165 case bitc::PARAMATTR_GRP_CODE_ENTRY: 166 return "ENTRY"; 167 } 168 case bitc::TYPE_BLOCK_ID_NEW: 169 switch (CodeID) { 170 default: 171 return None; 172 STRINGIFY_CODE(TYPE_CODE, NUMENTRY) 173 STRINGIFY_CODE(TYPE_CODE, VOID) 174 STRINGIFY_CODE(TYPE_CODE, FLOAT) 175 STRINGIFY_CODE(TYPE_CODE, DOUBLE) 176 STRINGIFY_CODE(TYPE_CODE, LABEL) 177 STRINGIFY_CODE(TYPE_CODE, OPAQUE) 178 STRINGIFY_CODE(TYPE_CODE, INTEGER) 179 STRINGIFY_CODE(TYPE_CODE, POINTER) 180 STRINGIFY_CODE(TYPE_CODE, HALF) 181 STRINGIFY_CODE(TYPE_CODE, ARRAY) 182 STRINGIFY_CODE(TYPE_CODE, VECTOR) 183 STRINGIFY_CODE(TYPE_CODE, X86_FP80) 184 STRINGIFY_CODE(TYPE_CODE, FP128) 185 STRINGIFY_CODE(TYPE_CODE, PPC_FP128) 186 STRINGIFY_CODE(TYPE_CODE, METADATA) 187 STRINGIFY_CODE(TYPE_CODE, X86_MMX) 188 STRINGIFY_CODE(TYPE_CODE, STRUCT_ANON) 189 STRINGIFY_CODE(TYPE_CODE, STRUCT_NAME) 190 STRINGIFY_CODE(TYPE_CODE, STRUCT_NAMED) 191 STRINGIFY_CODE(TYPE_CODE, FUNCTION) 192 STRINGIFY_CODE(TYPE_CODE, TOKEN) 193 STRINGIFY_CODE(TYPE_CODE, BFLOAT) 194 } 195 196 case bitc::CONSTANTS_BLOCK_ID: 197 switch (CodeID) { 198 default: 199 return None; 200 STRINGIFY_CODE(CST_CODE, SETTYPE) 201 STRINGIFY_CODE(CST_CODE, NULL) 202 STRINGIFY_CODE(CST_CODE, UNDEF) 203 STRINGIFY_CODE(CST_CODE, INTEGER) 204 STRINGIFY_CODE(CST_CODE, WIDE_INTEGER) 205 STRINGIFY_CODE(CST_CODE, FLOAT) 206 STRINGIFY_CODE(CST_CODE, AGGREGATE) 207 STRINGIFY_CODE(CST_CODE, STRING) 208 STRINGIFY_CODE(CST_CODE, CSTRING) 209 STRINGIFY_CODE(CST_CODE, CE_BINOP) 210 STRINGIFY_CODE(CST_CODE, CE_CAST) 211 STRINGIFY_CODE(CST_CODE, CE_GEP) 212 STRINGIFY_CODE(CST_CODE, CE_INBOUNDS_GEP) 213 STRINGIFY_CODE(CST_CODE, CE_SELECT) 214 STRINGIFY_CODE(CST_CODE, CE_EXTRACTELT) 215 STRINGIFY_CODE(CST_CODE, CE_INSERTELT) 216 STRINGIFY_CODE(CST_CODE, CE_SHUFFLEVEC) 217 STRINGIFY_CODE(CST_CODE, CE_CMP) 218 STRINGIFY_CODE(CST_CODE, INLINEASM) 219 STRINGIFY_CODE(CST_CODE, CE_SHUFVEC_EX) 220 STRINGIFY_CODE(CST_CODE, CE_UNOP) 221 STRINGIFY_CODE(CST_CODE, DSO_LOCAL_EQUIVALENT) 222 case bitc::CST_CODE_BLOCKADDRESS: 223 return "CST_CODE_BLOCKADDRESS"; 224 STRINGIFY_CODE(CST_CODE, DATA) 225 } 226 case bitc::FUNCTION_BLOCK_ID: 227 switch (CodeID) { 228 default: 229 return None; 230 STRINGIFY_CODE(FUNC_CODE, DECLAREBLOCKS) 231 STRINGIFY_CODE(FUNC_CODE, INST_BINOP) 232 STRINGIFY_CODE(FUNC_CODE, INST_CAST) 233 STRINGIFY_CODE(FUNC_CODE, INST_GEP_OLD) 234 STRINGIFY_CODE(FUNC_CODE, INST_INBOUNDS_GEP_OLD) 235 STRINGIFY_CODE(FUNC_CODE, INST_SELECT) 236 STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTELT) 237 STRINGIFY_CODE(FUNC_CODE, INST_INSERTELT) 238 STRINGIFY_CODE(FUNC_CODE, INST_SHUFFLEVEC) 239 STRINGIFY_CODE(FUNC_CODE, INST_CMP) 240 STRINGIFY_CODE(FUNC_CODE, INST_RET) 241 STRINGIFY_CODE(FUNC_CODE, INST_BR) 242 STRINGIFY_CODE(FUNC_CODE, INST_SWITCH) 243 STRINGIFY_CODE(FUNC_CODE, INST_INVOKE) 244 STRINGIFY_CODE(FUNC_CODE, INST_UNOP) 245 STRINGIFY_CODE(FUNC_CODE, INST_UNREACHABLE) 246 STRINGIFY_CODE(FUNC_CODE, INST_CLEANUPRET) 247 STRINGIFY_CODE(FUNC_CODE, INST_CATCHRET) 248 STRINGIFY_CODE(FUNC_CODE, INST_CATCHPAD) 249 STRINGIFY_CODE(FUNC_CODE, INST_PHI) 250 STRINGIFY_CODE(FUNC_CODE, INST_ALLOCA) 251 STRINGIFY_CODE(FUNC_CODE, INST_LOAD) 252 STRINGIFY_CODE(FUNC_CODE, INST_VAARG) 253 STRINGIFY_CODE(FUNC_CODE, INST_STORE) 254 STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTVAL) 255 STRINGIFY_CODE(FUNC_CODE, INST_INSERTVAL) 256 STRINGIFY_CODE(FUNC_CODE, INST_CMP2) 257 STRINGIFY_CODE(FUNC_CODE, INST_VSELECT) 258 STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC_AGAIN) 259 STRINGIFY_CODE(FUNC_CODE, INST_CALL) 260 STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC) 261 STRINGIFY_CODE(FUNC_CODE, INST_GEP) 262 STRINGIFY_CODE(FUNC_CODE, OPERAND_BUNDLE) 263 STRINGIFY_CODE(FUNC_CODE, INST_FENCE) 264 STRINGIFY_CODE(FUNC_CODE, INST_ATOMICRMW) 265 STRINGIFY_CODE(FUNC_CODE, INST_LOADATOMIC) 266 STRINGIFY_CODE(FUNC_CODE, INST_STOREATOMIC) 267 STRINGIFY_CODE(FUNC_CODE, INST_CMPXCHG) 268 STRINGIFY_CODE(FUNC_CODE, INST_CALLBR) 269 } 270 case bitc::VALUE_SYMTAB_BLOCK_ID: 271 switch (CodeID) { 272 default: 273 return None; 274 STRINGIFY_CODE(VST_CODE, ENTRY) 275 STRINGIFY_CODE(VST_CODE, BBENTRY) 276 STRINGIFY_CODE(VST_CODE, FNENTRY) 277 STRINGIFY_CODE(VST_CODE, COMBINED_ENTRY) 278 } 279 case bitc::MODULE_STRTAB_BLOCK_ID: 280 switch (CodeID) { 281 default: 282 return None; 283 STRINGIFY_CODE(MST_CODE, ENTRY) 284 STRINGIFY_CODE(MST_CODE, HASH) 285 } 286 case bitc::GLOBALVAL_SUMMARY_BLOCK_ID: 287 case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID: 288 switch (CodeID) { 289 default: 290 return None; 291 STRINGIFY_CODE(FS, PERMODULE) 292 STRINGIFY_CODE(FS, PERMODULE_PROFILE) 293 STRINGIFY_CODE(FS, PERMODULE_RELBF) 294 STRINGIFY_CODE(FS, PERMODULE_GLOBALVAR_INIT_REFS) 295 STRINGIFY_CODE(FS, PERMODULE_VTABLE_GLOBALVAR_INIT_REFS) 296 STRINGIFY_CODE(FS, COMBINED) 297 STRINGIFY_CODE(FS, COMBINED_PROFILE) 298 STRINGIFY_CODE(FS, COMBINED_GLOBALVAR_INIT_REFS) 299 STRINGIFY_CODE(FS, ALIAS) 300 STRINGIFY_CODE(FS, COMBINED_ALIAS) 301 STRINGIFY_CODE(FS, COMBINED_ORIGINAL_NAME) 302 STRINGIFY_CODE(FS, VERSION) 303 STRINGIFY_CODE(FS, FLAGS) 304 STRINGIFY_CODE(FS, TYPE_TESTS) 305 STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_VCALLS) 306 STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_VCALLS) 307 STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_CONST_VCALL) 308 STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_CONST_VCALL) 309 STRINGIFY_CODE(FS, VALUE_GUID) 310 STRINGIFY_CODE(FS, CFI_FUNCTION_DEFS) 311 STRINGIFY_CODE(FS, CFI_FUNCTION_DECLS) 312 STRINGIFY_CODE(FS, TYPE_ID) 313 STRINGIFY_CODE(FS, TYPE_ID_METADATA) 314 STRINGIFY_CODE(FS, BLOCK_COUNT) 315 STRINGIFY_CODE(FS, PARAM_ACCESS) 316 } 317 case bitc::METADATA_ATTACHMENT_ID: 318 switch (CodeID) { 319 default: 320 return None; 321 STRINGIFY_CODE(METADATA, ATTACHMENT) 322 } 323 case bitc::METADATA_BLOCK_ID: 324 switch (CodeID) { 325 default: 326 return None; 327 STRINGIFY_CODE(METADATA, STRING_OLD) 328 STRINGIFY_CODE(METADATA, VALUE) 329 STRINGIFY_CODE(METADATA, NODE) 330 STRINGIFY_CODE(METADATA, NAME) 331 STRINGIFY_CODE(METADATA, DISTINCT_NODE) 332 STRINGIFY_CODE(METADATA, KIND) // Older bitcode has it in a MODULE_BLOCK 333 STRINGIFY_CODE(METADATA, LOCATION) 334 STRINGIFY_CODE(METADATA, OLD_NODE) 335 STRINGIFY_CODE(METADATA, OLD_FN_NODE) 336 STRINGIFY_CODE(METADATA, NAMED_NODE) 337 STRINGIFY_CODE(METADATA, GENERIC_DEBUG) 338 STRINGIFY_CODE(METADATA, SUBRANGE) 339 STRINGIFY_CODE(METADATA, ENUMERATOR) 340 STRINGIFY_CODE(METADATA, BASIC_TYPE) 341 STRINGIFY_CODE(METADATA, FILE) 342 STRINGIFY_CODE(METADATA, DERIVED_TYPE) 343 STRINGIFY_CODE(METADATA, COMPOSITE_TYPE) 344 STRINGIFY_CODE(METADATA, SUBROUTINE_TYPE) 345 STRINGIFY_CODE(METADATA, COMPILE_UNIT) 346 STRINGIFY_CODE(METADATA, SUBPROGRAM) 347 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK) 348 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK_FILE) 349 STRINGIFY_CODE(METADATA, NAMESPACE) 350 STRINGIFY_CODE(METADATA, TEMPLATE_TYPE) 351 STRINGIFY_CODE(METADATA, TEMPLATE_VALUE) 352 STRINGIFY_CODE(METADATA, GLOBAL_VAR) 353 STRINGIFY_CODE(METADATA, LOCAL_VAR) 354 STRINGIFY_CODE(METADATA, EXPRESSION) 355 STRINGIFY_CODE(METADATA, OBJC_PROPERTY) 356 STRINGIFY_CODE(METADATA, IMPORTED_ENTITY) 357 STRINGIFY_CODE(METADATA, MODULE) 358 STRINGIFY_CODE(METADATA, MACRO) 359 STRINGIFY_CODE(METADATA, MACRO_FILE) 360 STRINGIFY_CODE(METADATA, STRINGS) 361 STRINGIFY_CODE(METADATA, GLOBAL_DECL_ATTACHMENT) 362 STRINGIFY_CODE(METADATA, GLOBAL_VAR_EXPR) 363 STRINGIFY_CODE(METADATA, INDEX_OFFSET) 364 STRINGIFY_CODE(METADATA, INDEX) 365 STRINGIFY_CODE(METADATA, ARG_LIST) 366 } 367 case bitc::METADATA_KIND_BLOCK_ID: 368 switch (CodeID) { 369 default: 370 return None; 371 STRINGIFY_CODE(METADATA, KIND) 372 } 373 case bitc::USELIST_BLOCK_ID: 374 switch (CodeID) { 375 default: 376 return None; 377 case bitc::USELIST_CODE_DEFAULT: 378 return "USELIST_CODE_DEFAULT"; 379 case bitc::USELIST_CODE_BB: 380 return "USELIST_CODE_BB"; 381 } 382 383 case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID: 384 switch (CodeID) { 385 default: 386 return None; 387 case bitc::OPERAND_BUNDLE_TAG: 388 return "OPERAND_BUNDLE_TAG"; 389 } 390 case bitc::STRTAB_BLOCK_ID: 391 switch (CodeID) { 392 default: 393 return None; 394 case bitc::STRTAB_BLOB: 395 return "BLOB"; 396 } 397 case bitc::SYMTAB_BLOCK_ID: 398 switch (CodeID) { 399 default: 400 return None; 401 case bitc::SYMTAB_BLOB: 402 return "BLOB"; 403 } 404 } 405 #undef STRINGIFY_CODE 406 } 407 408 static void printSize(raw_ostream &OS, double Bits) { 409 OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32)); 410 } 411 static void printSize(raw_ostream &OS, uint64_t Bits) { 412 OS << format("%lub/%.2fB/%luW", (unsigned long)Bits, (double)Bits / 8, 413 (unsigned long)(Bits / 32)); 414 } 415 416 static Expected<CurStreamTypeType> ReadSignature(BitstreamCursor &Stream) { 417 auto tryRead = [&Stream](char &Dest, size_t size) -> Error { 418 if (Expected<SimpleBitstreamCursor::word_t> MaybeWord = Stream.Read(size)) 419 Dest = MaybeWord.get(); 420 else 421 return MaybeWord.takeError(); 422 return Error::success(); 423 }; 424 425 char Signature[6]; 426 if (Error Err = tryRead(Signature[0], 8)) 427 return std::move(Err); 428 if (Error Err = tryRead(Signature[1], 8)) 429 return std::move(Err); 430 431 // Autodetect the file contents, if it is one we know. 432 if (Signature[0] == 'C' && Signature[1] == 'P') { 433 if (Error Err = tryRead(Signature[2], 8)) 434 return std::move(Err); 435 if (Error Err = tryRead(Signature[3], 8)) 436 return std::move(Err); 437 if (Signature[2] == 'C' && Signature[3] == 'H') 438 return ClangSerializedASTBitstream; 439 } else if (Signature[0] == 'D' && Signature[1] == 'I') { 440 if (Error Err = tryRead(Signature[2], 8)) 441 return std::move(Err); 442 if (Error Err = tryRead(Signature[3], 8)) 443 return std::move(Err); 444 if (Signature[2] == 'A' && Signature[3] == 'G') 445 return ClangSerializedDiagnosticsBitstream; 446 } else if (Signature[0] == 'R' && Signature[1] == 'M') { 447 if (Error Err = tryRead(Signature[2], 8)) 448 return std::move(Err); 449 if (Error Err = tryRead(Signature[3], 8)) 450 return std::move(Err); 451 if (Signature[2] == 'R' && Signature[3] == 'K') 452 return LLVMBitstreamRemarks; 453 } else { 454 if (Error Err = tryRead(Signature[2], 4)) 455 return std::move(Err); 456 if (Error Err = tryRead(Signature[3], 4)) 457 return std::move(Err); 458 if (Error Err = tryRead(Signature[4], 4)) 459 return std::move(Err); 460 if (Error Err = tryRead(Signature[5], 4)) 461 return std::move(Err); 462 if (Signature[0] == 'B' && Signature[1] == 'C' && Signature[2] == 0x0 && 463 Signature[3] == 0xC && Signature[4] == 0xE && Signature[5] == 0xD) 464 return LLVMIRBitstream; 465 } 466 return UnknownBitstream; 467 } 468 469 static Expected<CurStreamTypeType> analyzeHeader(Optional<BCDumpOptions> O, 470 BitstreamCursor &Stream) { 471 ArrayRef<uint8_t> Bytes = Stream.getBitcodeBytes(); 472 const unsigned char *BufPtr = (const unsigned char *)Bytes.data(); 473 const unsigned char *EndBufPtr = BufPtr + Bytes.size(); 474 475 // If we have a wrapper header, parse it and ignore the non-bc file 476 // contents. The magic number is 0x0B17C0DE stored in little endian. 477 if (isBitcodeWrapper(BufPtr, EndBufPtr)) { 478 if (Bytes.size() < BWH_HeaderSize) 479 return reportError("Invalid bitcode wrapper header"); 480 481 if (O) { 482 unsigned Magic = support::endian::read32le(&BufPtr[BWH_MagicField]); 483 unsigned Version = support::endian::read32le(&BufPtr[BWH_VersionField]); 484 unsigned Offset = support::endian::read32le(&BufPtr[BWH_OffsetField]); 485 unsigned Size = support::endian::read32le(&BufPtr[BWH_SizeField]); 486 unsigned CPUType = support::endian::read32le(&BufPtr[BWH_CPUTypeField]); 487 488 O->OS << "<BITCODE_WRAPPER_HEADER" 489 << " Magic=" << format_hex(Magic, 10) 490 << " Version=" << format_hex(Version, 10) 491 << " Offset=" << format_hex(Offset, 10) 492 << " Size=" << format_hex(Size, 10) 493 << " CPUType=" << format_hex(CPUType, 10) << "/>\n"; 494 } 495 496 if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true)) 497 return reportError("Invalid bitcode wrapper header"); 498 } 499 500 // Use the cursor modified by skipping the wrapper header. 501 Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr)); 502 503 return ReadSignature(Stream); 504 } 505 506 static bool canDecodeBlob(unsigned Code, unsigned BlockID) { 507 return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS; 508 } 509 510 Error BitcodeAnalyzer::decodeMetadataStringsBlob(StringRef Indent, 511 ArrayRef<uint64_t> Record, 512 StringRef Blob, 513 raw_ostream &OS) { 514 if (Blob.empty()) 515 return reportError("Cannot decode empty blob."); 516 517 if (Record.size() != 2) 518 return reportError( 519 "Decoding metadata strings blob needs two record entries."); 520 521 unsigned NumStrings = Record[0]; 522 unsigned StringsOffset = Record[1]; 523 OS << " num-strings = " << NumStrings << " {\n"; 524 525 StringRef Lengths = Blob.slice(0, StringsOffset); 526 SimpleBitstreamCursor R(Lengths); 527 StringRef Strings = Blob.drop_front(StringsOffset); 528 do { 529 if (R.AtEndOfStream()) 530 return reportError("bad length"); 531 532 uint32_t Size; 533 if (Error E = R.ReadVBR(6).moveInto(Size)) 534 return E; 535 if (Strings.size() < Size) 536 return reportError("truncated chars"); 537 538 OS << Indent << " '"; 539 OS.write_escaped(Strings.slice(0, Size), /*hex=*/true); 540 OS << "'\n"; 541 Strings = Strings.drop_front(Size); 542 } while (--NumStrings); 543 544 OS << Indent << " }"; 545 return Error::success(); 546 } 547 548 BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer, 549 Optional<StringRef> BlockInfoBuffer) 550 : Stream(Buffer) { 551 if (BlockInfoBuffer) 552 BlockInfoStream.emplace(*BlockInfoBuffer); 553 } 554 555 Error BitcodeAnalyzer::analyze(Optional<BCDumpOptions> O, 556 Optional<StringRef> CheckHash) { 557 if (Error E = analyzeHeader(O, Stream).moveInto(CurStreamType)) 558 return E; 559 560 Stream.setBlockInfo(&BlockInfo); 561 562 // Read block info from BlockInfoStream, if specified. 563 // The block info must be a top-level block. 564 if (BlockInfoStream) { 565 BitstreamCursor BlockInfoCursor(*BlockInfoStream); 566 if (Error E = analyzeHeader(O, BlockInfoCursor).takeError()) 567 return E; 568 569 while (!BlockInfoCursor.AtEndOfStream()) { 570 Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode(); 571 if (!MaybeCode) 572 return MaybeCode.takeError(); 573 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK) 574 return reportError("Invalid record at top-level in block info file"); 575 576 Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID(); 577 if (!MaybeBlockID) 578 return MaybeBlockID.takeError(); 579 if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) { 580 Optional<BitstreamBlockInfo> NewBlockInfo; 581 if (Error E = 582 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true) 583 .moveInto(NewBlockInfo)) 584 return E; 585 if (!NewBlockInfo) 586 return reportError("Malformed BlockInfoBlock in block info file"); 587 BlockInfo = std::move(*NewBlockInfo); 588 break; 589 } 590 591 if (Error Err = BlockInfoCursor.SkipBlock()) 592 return Err; 593 } 594 } 595 596 // Parse the top-level structure. We only allow blocks at the top-level. 597 while (!Stream.AtEndOfStream()) { 598 Expected<unsigned> MaybeCode = Stream.ReadCode(); 599 if (!MaybeCode) 600 return MaybeCode.takeError(); 601 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK) 602 return reportError("Invalid record at top-level"); 603 604 Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID(); 605 if (!MaybeBlockID) 606 return MaybeBlockID.takeError(); 607 608 if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash)) 609 return E; 610 ++NumTopBlocks; 611 } 612 613 return Error::success(); 614 } 615 616 void BitcodeAnalyzer::printStats(BCDumpOptions O, 617 Optional<StringRef> Filename) { 618 uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT; 619 // Print a summary of the read file. 620 O.OS << "Summary "; 621 if (Filename) 622 O.OS << "of " << Filename->data() << ":\n"; 623 O.OS << " Total size: "; 624 printSize(O.OS, BufferSizeBits); 625 O.OS << "\n"; 626 O.OS << " Stream type: "; 627 switch (CurStreamType) { 628 case UnknownBitstream: 629 O.OS << "unknown\n"; 630 break; 631 case LLVMIRBitstream: 632 O.OS << "LLVM IR\n"; 633 break; 634 case ClangSerializedASTBitstream: 635 O.OS << "Clang Serialized AST\n"; 636 break; 637 case ClangSerializedDiagnosticsBitstream: 638 O.OS << "Clang Serialized Diagnostics\n"; 639 break; 640 case LLVMBitstreamRemarks: 641 O.OS << "LLVM Remarks\n"; 642 break; 643 } 644 O.OS << " # Toplevel Blocks: " << NumTopBlocks << "\n"; 645 O.OS << "\n"; 646 647 // Emit per-block stats. 648 O.OS << "Per-block Summary:\n"; 649 for (std::map<unsigned, PerBlockIDStats>::iterator I = BlockIDStats.begin(), 650 E = BlockIDStats.end(); 651 I != E; ++I) { 652 O.OS << " Block ID #" << I->first; 653 if (Optional<const char *> BlockName = 654 GetBlockName(I->first, BlockInfo, CurStreamType)) 655 O.OS << " (" << *BlockName << ")"; 656 O.OS << ":\n"; 657 658 const PerBlockIDStats &Stats = I->second; 659 O.OS << " Num Instances: " << Stats.NumInstances << "\n"; 660 O.OS << " Total Size: "; 661 printSize(O.OS, Stats.NumBits); 662 O.OS << "\n"; 663 double pct = (Stats.NumBits * 100.0) / BufferSizeBits; 664 O.OS << " Percent of file: " << format("%2.4f%%", pct) << "\n"; 665 if (Stats.NumInstances > 1) { 666 O.OS << " Average Size: "; 667 printSize(O.OS, Stats.NumBits / (double)Stats.NumInstances); 668 O.OS << "\n"; 669 O.OS << " Tot/Avg SubBlocks: " << Stats.NumSubBlocks << "/" 670 << Stats.NumSubBlocks / (double)Stats.NumInstances << "\n"; 671 O.OS << " Tot/Avg Abbrevs: " << Stats.NumAbbrevs << "/" 672 << Stats.NumAbbrevs / (double)Stats.NumInstances << "\n"; 673 O.OS << " Tot/Avg Records: " << Stats.NumRecords << "/" 674 << Stats.NumRecords / (double)Stats.NumInstances << "\n"; 675 } else { 676 O.OS << " Num SubBlocks: " << Stats.NumSubBlocks << "\n"; 677 O.OS << " Num Abbrevs: " << Stats.NumAbbrevs << "\n"; 678 O.OS << " Num Records: " << Stats.NumRecords << "\n"; 679 } 680 if (Stats.NumRecords) { 681 double pct = (Stats.NumAbbreviatedRecords * 100.0) / Stats.NumRecords; 682 O.OS << " Percent Abbrevs: " << format("%2.4f%%", pct) << "\n"; 683 } 684 O.OS << "\n"; 685 686 // Print a histogram of the codes we see. 687 if (O.Histogram && !Stats.CodeFreq.empty()) { 688 std::vector<std::pair<unsigned, unsigned>> FreqPairs; // <freq,code> 689 for (unsigned i = 0, e = Stats.CodeFreq.size(); i != e; ++i) 690 if (unsigned Freq = Stats.CodeFreq[i].NumInstances) 691 FreqPairs.push_back(std::make_pair(Freq, i)); 692 llvm::stable_sort(FreqPairs); 693 std::reverse(FreqPairs.begin(), FreqPairs.end()); 694 695 O.OS << "\tRecord Histogram:\n"; 696 O.OS << "\t\t Count # Bits b/Rec % Abv Record Kind\n"; 697 for (unsigned i = 0, e = FreqPairs.size(); i != e; ++i) { 698 const PerRecordStats &RecStats = Stats.CodeFreq[FreqPairs[i].second]; 699 700 O.OS << format("\t\t%7d %9lu", RecStats.NumInstances, 701 (unsigned long)RecStats.TotalBits); 702 703 if (RecStats.NumInstances > 1) 704 O.OS << format(" %9.1f", 705 (double)RecStats.TotalBits / RecStats.NumInstances); 706 else 707 O.OS << " "; 708 709 if (RecStats.NumAbbrev) 710 O.OS << format(" %7.2f", (double)RecStats.NumAbbrev / 711 RecStats.NumInstances * 100); 712 else 713 O.OS << " "; 714 715 O.OS << " "; 716 if (Optional<const char *> CodeName = GetCodeName( 717 FreqPairs[i].second, I->first, BlockInfo, CurStreamType)) 718 O.OS << *CodeName << "\n"; 719 else 720 O.OS << "UnknownCode" << FreqPairs[i].second << "\n"; 721 } 722 O.OS << "\n"; 723 } 724 } 725 } 726 727 Error BitcodeAnalyzer::parseBlock(unsigned BlockID, unsigned IndentLevel, 728 Optional<BCDumpOptions> O, 729 Optional<StringRef> CheckHash) { 730 std::string Indent(IndentLevel * 2, ' '); 731 uint64_t BlockBitStart = Stream.GetCurrentBitNo(); 732 733 // Get the statistics for this BlockID. 734 PerBlockIDStats &BlockStats = BlockIDStats[BlockID]; 735 736 BlockStats.NumInstances++; 737 738 // BLOCKINFO is a special part of the stream. 739 bool DumpRecords = O.hasValue(); 740 if (BlockID == bitc::BLOCKINFO_BLOCK_ID) { 741 if (O && !O->DumpBlockinfo) 742 O->OS << Indent << "<BLOCKINFO_BLOCK/>\n"; 743 Optional<BitstreamBlockInfo> NewBlockInfo; 744 if (Error E = Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true) 745 .moveInto(NewBlockInfo)) 746 return E; 747 if (!NewBlockInfo) 748 return reportError("Malformed BlockInfoBlock"); 749 BlockInfo = std::move(*NewBlockInfo); 750 if (Error Err = Stream.JumpToBit(BlockBitStart)) 751 return Err; 752 // It's not really interesting to dump the contents of the blockinfo 753 // block, so only do it if the user explicitly requests it. 754 DumpRecords = O && O->DumpBlockinfo; 755 } 756 757 unsigned NumWords = 0; 758 if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords)) 759 return Err; 760 761 // Keep it for later, when we see a MODULE_HASH record 762 uint64_t BlockEntryPos = Stream.getCurrentByteNo(); 763 764 Optional<const char *> BlockName = None; 765 if (DumpRecords) { 766 O->OS << Indent << "<"; 767 if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType))) 768 O->OS << *BlockName; 769 else 770 O->OS << "UnknownBlock" << BlockID; 771 772 if (!O->Symbolic && BlockName) 773 O->OS << " BlockID=" << BlockID; 774 775 O->OS << " NumWords=" << NumWords 776 << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n"; 777 } 778 779 SmallVector<uint64_t, 64> Record; 780 781 // Keep the offset to the metadata index if seen. 782 uint64_t MetadataIndexOffset = 0; 783 784 // Read all the records for this block. 785 while (1) { 786 if (Stream.AtEndOfStream()) 787 return reportError("Premature end of bitstream"); 788 789 uint64_t RecordStartBit = Stream.GetCurrentBitNo(); 790 791 BitstreamEntry Entry; 792 if (Error E = Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs) 793 .moveInto(Entry)) 794 return E; 795 796 switch (Entry.Kind) { 797 case BitstreamEntry::Error: 798 return reportError("malformed bitcode file"); 799 case BitstreamEntry::EndBlock: { 800 uint64_t BlockBitEnd = Stream.GetCurrentBitNo(); 801 BlockStats.NumBits += BlockBitEnd - BlockBitStart; 802 if (DumpRecords) { 803 O->OS << Indent << "</"; 804 if (BlockName) 805 O->OS << *BlockName << ">\n"; 806 else 807 O->OS << "UnknownBlock" << BlockID << ">\n"; 808 } 809 return Error::success(); 810 } 811 812 case BitstreamEntry::SubBlock: { 813 uint64_t SubBlockBitStart = Stream.GetCurrentBitNo(); 814 if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash)) 815 return E; 816 ++BlockStats.NumSubBlocks; 817 uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo(); 818 819 // Don't include subblock sizes in the size of this block. 820 BlockBitStart += SubBlockBitEnd - SubBlockBitStart; 821 continue; 822 } 823 case BitstreamEntry::Record: 824 // The interesting case. 825 break; 826 } 827 828 if (Entry.ID == bitc::DEFINE_ABBREV) { 829 if (Error Err = Stream.ReadAbbrevRecord()) 830 return Err; 831 ++BlockStats.NumAbbrevs; 832 continue; 833 } 834 835 Record.clear(); 836 837 ++BlockStats.NumRecords; 838 839 StringRef Blob; 840 uint64_t CurrentRecordPos = Stream.GetCurrentBitNo(); 841 unsigned Code; 842 if (Error E = Stream.readRecord(Entry.ID, Record, &Blob).moveInto(Code)) 843 return E; 844 845 // Increment the # occurrences of this code. 846 if (BlockStats.CodeFreq.size() <= Code) 847 BlockStats.CodeFreq.resize(Code + 1); 848 BlockStats.CodeFreq[Code].NumInstances++; 849 BlockStats.CodeFreq[Code].TotalBits += 850 Stream.GetCurrentBitNo() - RecordStartBit; 851 if (Entry.ID != bitc::UNABBREV_RECORD) { 852 BlockStats.CodeFreq[Code].NumAbbrev++; 853 ++BlockStats.NumAbbreviatedRecords; 854 } 855 856 if (DumpRecords) { 857 O->OS << Indent << " <"; 858 Optional<const char *> CodeName = 859 GetCodeName(Code, BlockID, BlockInfo, CurStreamType); 860 if (CodeName) 861 O->OS << *CodeName; 862 else 863 O->OS << "UnknownCode" << Code; 864 if (!O->Symbolic && CodeName) 865 O->OS << " codeid=" << Code; 866 const BitCodeAbbrev *Abbv = nullptr; 867 if (Entry.ID != bitc::UNABBREV_RECORD) { 868 Abbv = Stream.getAbbrev(Entry.ID); 869 O->OS << " abbrevid=" << Entry.ID; 870 } 871 872 for (unsigned i = 0, e = Record.size(); i != e; ++i) 873 O->OS << " op" << i << "=" << (int64_t)Record[i]; 874 875 // If we found a metadata index, let's verify that we had an offset 876 // before and validate its forward reference offset was correct! 877 if (BlockID == bitc::METADATA_BLOCK_ID) { 878 if (Code == bitc::METADATA_INDEX_OFFSET) { 879 if (Record.size() != 2) 880 O->OS << "(Invalid record)"; 881 else { 882 auto Offset = Record[0] + (Record[1] << 32); 883 MetadataIndexOffset = Stream.GetCurrentBitNo() + Offset; 884 } 885 } 886 if (Code == bitc::METADATA_INDEX) { 887 O->OS << " (offset "; 888 if (MetadataIndexOffset == RecordStartBit) 889 O->OS << "match)"; 890 else 891 O->OS << "mismatch: " << MetadataIndexOffset << " vs " 892 << RecordStartBit << ")"; 893 } 894 } 895 896 // If we found a module hash, let's verify that it matches! 897 if (BlockID == bitc::MODULE_BLOCK_ID && Code == bitc::MODULE_CODE_HASH && 898 CheckHash.hasValue()) { 899 if (Record.size() != 5) 900 O->OS << " (invalid)"; 901 else { 902 // Recompute the hash and compare it to the one in the bitcode 903 SHA1 Hasher; 904 StringRef Hash; 905 Hasher.update(*CheckHash); 906 { 907 int BlockSize = (CurrentRecordPos / 8) - BlockEntryPos; 908 auto Ptr = Stream.getPointerToByte(BlockEntryPos, BlockSize); 909 Hasher.update(ArrayRef<uint8_t>(Ptr, BlockSize)); 910 Hash = Hasher.result(); 911 } 912 std::array<char, 20> RecordedHash; 913 int Pos = 0; 914 for (auto &Val : Record) { 915 assert(!(Val >> 32) && "Unexpected high bits set"); 916 support::endian::write32be(&RecordedHash[Pos], Val); 917 Pos += 4; 918 } 919 if (Hash == StringRef(RecordedHash.data(), RecordedHash.size())) 920 O->OS << " (match)"; 921 else 922 O->OS << " (!mismatch!)"; 923 } 924 } 925 926 O->OS << "/>"; 927 928 if (Abbv) { 929 for (unsigned i = 1, e = Abbv->getNumOperandInfos(); i != e; ++i) { 930 const BitCodeAbbrevOp &Op = Abbv->getOperandInfo(i); 931 if (!Op.isEncoding() || Op.getEncoding() != BitCodeAbbrevOp::Array) 932 continue; 933 assert(i + 2 == e && "Array op not second to last"); 934 std::string Str; 935 bool ArrayIsPrintable = true; 936 for (unsigned j = i - 1, je = Record.size(); j != je; ++j) { 937 if (!isPrint(static_cast<unsigned char>(Record[j]))) { 938 ArrayIsPrintable = false; 939 break; 940 } 941 Str += (char)Record[j]; 942 } 943 if (ArrayIsPrintable) 944 O->OS << " record string = '" << Str << "'"; 945 break; 946 } 947 } 948 949 if (Blob.data()) { 950 if (canDecodeBlob(Code, BlockID)) { 951 if (Error E = decodeMetadataStringsBlob(Indent, Record, Blob, O->OS)) 952 return E; 953 } else { 954 O->OS << " blob data = "; 955 if (O->ShowBinaryBlobs) { 956 O->OS << "'"; 957 O->OS.write_escaped(Blob, /*hex=*/true) << "'"; 958 } else { 959 bool BlobIsPrintable = true; 960 for (unsigned i = 0, e = Blob.size(); i != e; ++i) 961 if (!isPrint(static_cast<unsigned char>(Blob[i]))) { 962 BlobIsPrintable = false; 963 break; 964 } 965 966 if (BlobIsPrintable) 967 O->OS << "'" << Blob << "'"; 968 else 969 O->OS << "unprintable, " << Blob.size() << " bytes."; 970 } 971 } 972 } 973 974 O->OS << "\n"; 975 } 976 977 // Make sure that we can skip the current record. 978 if (Error Err = Stream.JumpToBit(CurrentRecordPos)) 979 return Err; 980 if (Expected<unsigned> Skipped = Stream.skipRecord(Entry.ID)) 981 ; // Do nothing. 982 else 983 return Skipped.takeError(); 984 } 985 } 986 987