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