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 17using namespace llvm; 18 19static 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. 24static 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. 87static 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, VSTOFFSET) 139 STRINGIFY_CODE(MODULE_CODE, METADATA_VALUES_UNUSED) 140 STRINGIFY_CODE(MODULE_CODE, SOURCE_FILENAME) 141 STRINGIFY_CODE(MODULE_CODE, HASH) 142 } 143 case bitc::IDENTIFICATION_BLOCK_ID: 144 switch (CodeID) { 145 default: 146 return None; 147 STRINGIFY_CODE(IDENTIFICATION_CODE, STRING) 148 STRINGIFY_CODE(IDENTIFICATION_CODE, EPOCH) 149 } 150 case bitc::PARAMATTR_BLOCK_ID: 151 switch (CodeID) { 152 default: 153 return None; 154 // FIXME: Should these be different? 155 case bitc::PARAMATTR_CODE_ENTRY_OLD: 156 return "ENTRY"; 157 case bitc::PARAMATTR_CODE_ENTRY: 158 return "ENTRY"; 159 } 160 case bitc::PARAMATTR_GROUP_BLOCK_ID: 161 switch (CodeID) { 162 default: 163 return None; 164 case bitc::PARAMATTR_GRP_CODE_ENTRY: 165 return "ENTRY"; 166 } 167 case bitc::TYPE_BLOCK_ID_NEW: 168 switch (CodeID) { 169 default: 170 return None; 171 STRINGIFY_CODE(TYPE_CODE, NUMENTRY) 172 STRINGIFY_CODE(TYPE_CODE, VOID) 173 STRINGIFY_CODE(TYPE_CODE, FLOAT) 174 STRINGIFY_CODE(TYPE_CODE, DOUBLE) 175 STRINGIFY_CODE(TYPE_CODE, LABEL) 176 STRINGIFY_CODE(TYPE_CODE, OPAQUE) 177 STRINGIFY_CODE(TYPE_CODE, INTEGER) 178 STRINGIFY_CODE(TYPE_CODE, POINTER) 179 STRINGIFY_CODE(TYPE_CODE, ARRAY) 180 STRINGIFY_CODE(TYPE_CODE, VECTOR) 181 STRINGIFY_CODE(TYPE_CODE, X86_FP80) 182 STRINGIFY_CODE(TYPE_CODE, FP128) 183 STRINGIFY_CODE(TYPE_CODE, PPC_FP128) 184 STRINGIFY_CODE(TYPE_CODE, METADATA) 185 STRINGIFY_CODE(TYPE_CODE, STRUCT_ANON) 186 STRINGIFY_CODE(TYPE_CODE, STRUCT_NAME) 187 STRINGIFY_CODE(TYPE_CODE, STRUCT_NAMED) 188 STRINGIFY_CODE(TYPE_CODE, FUNCTION) 189 } 190 191 case bitc::CONSTANTS_BLOCK_ID: 192 switch (CodeID) { 193 default: 194 return None; 195 STRINGIFY_CODE(CST_CODE, SETTYPE) 196 STRINGIFY_CODE(CST_CODE, NULL) 197 STRINGIFY_CODE(CST_CODE, UNDEF) 198 STRINGIFY_CODE(CST_CODE, INTEGER) 199 STRINGIFY_CODE(CST_CODE, WIDE_INTEGER) 200 STRINGIFY_CODE(CST_CODE, FLOAT) 201 STRINGIFY_CODE(CST_CODE, AGGREGATE) 202 STRINGIFY_CODE(CST_CODE, STRING) 203 STRINGIFY_CODE(CST_CODE, CSTRING) 204 STRINGIFY_CODE(CST_CODE, CE_BINOP) 205 STRINGIFY_CODE(CST_CODE, CE_CAST) 206 STRINGIFY_CODE(CST_CODE, CE_GEP) 207 STRINGIFY_CODE(CST_CODE, CE_INBOUNDS_GEP) 208 STRINGIFY_CODE(CST_CODE, CE_SELECT) 209 STRINGIFY_CODE(CST_CODE, CE_EXTRACTELT) 210 STRINGIFY_CODE(CST_CODE, CE_INSERTELT) 211 STRINGIFY_CODE(CST_CODE, CE_SHUFFLEVEC) 212 STRINGIFY_CODE(CST_CODE, CE_CMP) 213 STRINGIFY_CODE(CST_CODE, INLINEASM) 214 STRINGIFY_CODE(CST_CODE, CE_SHUFVEC_EX) 215 STRINGIFY_CODE(CST_CODE, CE_UNOP) 216 case bitc::CST_CODE_BLOCKADDRESS: 217 return "CST_CODE_BLOCKADDRESS"; 218 STRINGIFY_CODE(CST_CODE, DATA) 219 } 220 case bitc::FUNCTION_BLOCK_ID: 221 switch (CodeID) { 222 default: 223 return None; 224 STRINGIFY_CODE(FUNC_CODE, DECLAREBLOCKS) 225 STRINGIFY_CODE(FUNC_CODE, INST_BINOP) 226 STRINGIFY_CODE(FUNC_CODE, INST_CAST) 227 STRINGIFY_CODE(FUNC_CODE, INST_GEP_OLD) 228 STRINGIFY_CODE(FUNC_CODE, INST_INBOUNDS_GEP_OLD) 229 STRINGIFY_CODE(FUNC_CODE, INST_SELECT) 230 STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTELT) 231 STRINGIFY_CODE(FUNC_CODE, INST_INSERTELT) 232 STRINGIFY_CODE(FUNC_CODE, INST_SHUFFLEVEC) 233 STRINGIFY_CODE(FUNC_CODE, INST_CMP) 234 STRINGIFY_CODE(FUNC_CODE, INST_RET) 235 STRINGIFY_CODE(FUNC_CODE, INST_BR) 236 STRINGIFY_CODE(FUNC_CODE, INST_SWITCH) 237 STRINGIFY_CODE(FUNC_CODE, INST_INVOKE) 238 STRINGIFY_CODE(FUNC_CODE, INST_UNOP) 239 STRINGIFY_CODE(FUNC_CODE, INST_UNREACHABLE) 240 STRINGIFY_CODE(FUNC_CODE, INST_CLEANUPRET) 241 STRINGIFY_CODE(FUNC_CODE, INST_CATCHRET) 242 STRINGIFY_CODE(FUNC_CODE, INST_CATCHPAD) 243 STRINGIFY_CODE(FUNC_CODE, INST_PHI) 244 STRINGIFY_CODE(FUNC_CODE, INST_ALLOCA) 245 STRINGIFY_CODE(FUNC_CODE, INST_LOAD) 246 STRINGIFY_CODE(FUNC_CODE, INST_VAARG) 247 STRINGIFY_CODE(FUNC_CODE, INST_STORE) 248 STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTVAL) 249 STRINGIFY_CODE(FUNC_CODE, INST_INSERTVAL) 250 STRINGIFY_CODE(FUNC_CODE, INST_CMP2) 251 STRINGIFY_CODE(FUNC_CODE, INST_VSELECT) 252 STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC_AGAIN) 253 STRINGIFY_CODE(FUNC_CODE, INST_CALL) 254 STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC) 255 STRINGIFY_CODE(FUNC_CODE, INST_GEP) 256 STRINGIFY_CODE(FUNC_CODE, OPERAND_BUNDLE) 257 STRINGIFY_CODE(FUNC_CODE, INST_FENCE) 258 STRINGIFY_CODE(FUNC_CODE, INST_ATOMICRMW) 259 STRINGIFY_CODE(FUNC_CODE, INST_LOADATOMIC) 260 STRINGIFY_CODE(FUNC_CODE, INST_STOREATOMIC) 261 STRINGIFY_CODE(FUNC_CODE, INST_CMPXCHG) 262 STRINGIFY_CODE(FUNC_CODE, INST_CALLBR) 263 } 264 case bitc::VALUE_SYMTAB_BLOCK_ID: 265 switch (CodeID) { 266 default: 267 return None; 268 STRINGIFY_CODE(VST_CODE, ENTRY) 269 STRINGIFY_CODE(VST_CODE, BBENTRY) 270 STRINGIFY_CODE(VST_CODE, FNENTRY) 271 STRINGIFY_CODE(VST_CODE, COMBINED_ENTRY) 272 } 273 case bitc::MODULE_STRTAB_BLOCK_ID: 274 switch (CodeID) { 275 default: 276 return None; 277 STRINGIFY_CODE(MST_CODE, ENTRY) 278 STRINGIFY_CODE(MST_CODE, HASH) 279 } 280 case bitc::GLOBALVAL_SUMMARY_BLOCK_ID: 281 case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID: 282 switch (CodeID) { 283 default: 284 return None; 285 STRINGIFY_CODE(FS, PERMODULE) 286 STRINGIFY_CODE(FS, PERMODULE_PROFILE) 287 STRINGIFY_CODE(FS, PERMODULE_RELBF) 288 STRINGIFY_CODE(FS, PERMODULE_GLOBALVAR_INIT_REFS) 289 STRINGIFY_CODE(FS, PERMODULE_VTABLE_GLOBALVAR_INIT_REFS) 290 STRINGIFY_CODE(FS, COMBINED) 291 STRINGIFY_CODE(FS, COMBINED_PROFILE) 292 STRINGIFY_CODE(FS, COMBINED_GLOBALVAR_INIT_REFS) 293 STRINGIFY_CODE(FS, ALIAS) 294 STRINGIFY_CODE(FS, COMBINED_ALIAS) 295 STRINGIFY_CODE(FS, COMBINED_ORIGINAL_NAME) 296 STRINGIFY_CODE(FS, VERSION) 297 STRINGIFY_CODE(FS, FLAGS) 298 STRINGIFY_CODE(FS, TYPE_TESTS) 299 STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_VCALLS) 300 STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_VCALLS) 301 STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_CONST_VCALL) 302 STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_CONST_VCALL) 303 STRINGIFY_CODE(FS, VALUE_GUID) 304 STRINGIFY_CODE(FS, CFI_FUNCTION_DEFS) 305 STRINGIFY_CODE(FS, CFI_FUNCTION_DECLS) 306 STRINGIFY_CODE(FS, TYPE_ID) 307 STRINGIFY_CODE(FS, TYPE_ID_METADATA) 308 STRINGIFY_CODE(FS, BLOCK_COUNT) 309 STRINGIFY_CODE(FS, PARAM_ACCESS) 310 } 311 case bitc::METADATA_ATTACHMENT_ID: 312 switch (CodeID) { 313 default: 314 return None; 315 STRINGIFY_CODE(METADATA, ATTACHMENT) 316 } 317 case bitc::METADATA_BLOCK_ID: 318 switch (CodeID) { 319 default: 320 return None; 321 STRINGIFY_CODE(METADATA, STRING_OLD) 322 STRINGIFY_CODE(METADATA, VALUE) 323 STRINGIFY_CODE(METADATA, NODE) 324 STRINGIFY_CODE(METADATA, NAME) 325 STRINGIFY_CODE(METADATA, DISTINCT_NODE) 326 STRINGIFY_CODE(METADATA, KIND) // Older bitcode has it in a MODULE_BLOCK 327 STRINGIFY_CODE(METADATA, LOCATION) 328 STRINGIFY_CODE(METADATA, OLD_NODE) 329 STRINGIFY_CODE(METADATA, OLD_FN_NODE) 330 STRINGIFY_CODE(METADATA, NAMED_NODE) 331 STRINGIFY_CODE(METADATA, GENERIC_DEBUG) 332 STRINGIFY_CODE(METADATA, SUBRANGE) 333 STRINGIFY_CODE(METADATA, ENUMERATOR) 334 STRINGIFY_CODE(METADATA, BASIC_TYPE) 335 STRINGIFY_CODE(METADATA, FILE) 336 STRINGIFY_CODE(METADATA, DERIVED_TYPE) 337 STRINGIFY_CODE(METADATA, COMPOSITE_TYPE) 338 STRINGIFY_CODE(METADATA, SUBROUTINE_TYPE) 339 STRINGIFY_CODE(METADATA, COMPILE_UNIT) 340 STRINGIFY_CODE(METADATA, SUBPROGRAM) 341 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK) 342 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK_FILE) 343 STRINGIFY_CODE(METADATA, NAMESPACE) 344 STRINGIFY_CODE(METADATA, TEMPLATE_TYPE) 345 STRINGIFY_CODE(METADATA, TEMPLATE_VALUE) 346 STRINGIFY_CODE(METADATA, GLOBAL_VAR) 347 STRINGIFY_CODE(METADATA, LOCAL_VAR) 348 STRINGIFY_CODE(METADATA, EXPRESSION) 349 STRINGIFY_CODE(METADATA, OBJC_PROPERTY) 350 STRINGIFY_CODE(METADATA, IMPORTED_ENTITY) 351 STRINGIFY_CODE(METADATA, MODULE) 352 STRINGIFY_CODE(METADATA, MACRO) 353 STRINGIFY_CODE(METADATA, MACRO_FILE) 354 STRINGIFY_CODE(METADATA, STRINGS) 355 STRINGIFY_CODE(METADATA, GLOBAL_DECL_ATTACHMENT) 356 STRINGIFY_CODE(METADATA, GLOBAL_VAR_EXPR) 357 STRINGIFY_CODE(METADATA, INDEX_OFFSET) 358 STRINGIFY_CODE(METADATA, INDEX) 359 } 360 case bitc::METADATA_KIND_BLOCK_ID: 361 switch (CodeID) { 362 default: 363 return None; 364 STRINGIFY_CODE(METADATA, KIND) 365 } 366 case bitc::USELIST_BLOCK_ID: 367 switch (CodeID) { 368 default: 369 return None; 370 case bitc::USELIST_CODE_DEFAULT: 371 return "USELIST_CODE_DEFAULT"; 372 case bitc::USELIST_CODE_BB: 373 return "USELIST_CODE_BB"; 374 } 375 376 case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID: 377 switch (CodeID) { 378 default: 379 return None; 380 case bitc::OPERAND_BUNDLE_TAG: 381 return "OPERAND_BUNDLE_TAG"; 382 } 383 case bitc::STRTAB_BLOCK_ID: 384 switch (CodeID) { 385 default: 386 return None; 387 case bitc::STRTAB_BLOB: 388 return "BLOB"; 389 } 390 case bitc::SYMTAB_BLOCK_ID: 391 switch (CodeID) { 392 default: 393 return None; 394 case bitc::SYMTAB_BLOB: 395 return "BLOB"; 396 } 397 } 398#undef STRINGIFY_CODE 399} 400 401static void printSize(raw_ostream &OS, double Bits) { 402 OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32)); 403} 404static void printSize(raw_ostream &OS, uint64_t Bits) { 405 OS << format("%lub/%.2fB/%luW", (unsigned long)Bits, (double)Bits / 8, 406 (unsigned long)(Bits / 32)); 407} 408 409static Expected<CurStreamTypeType> ReadSignature(BitstreamCursor &Stream) { 410 auto tryRead = [&Stream](char &Dest, size_t size) -> Error { 411 if (Expected<SimpleBitstreamCursor::word_t> MaybeWord = Stream.Read(size)) 412 Dest = MaybeWord.get(); 413 else 414 return MaybeWord.takeError(); 415 return Error::success(); 416 }; 417 418 char Signature[6]; 419 if (Error Err = tryRead(Signature[0], 8)) 420 return std::move(Err); 421 if (Error Err = tryRead(Signature[1], 8)) 422 return std::move(Err); 423 424 // Autodetect the file contents, if it is one we know. 425 if (Signature[0] == 'C' && Signature[1] == 'P') { 426 if (Error Err = tryRead(Signature[2], 8)) 427 return std::move(Err); 428 if (Error Err = tryRead(Signature[3], 8)) 429 return std::move(Err); 430 if (Signature[2] == 'C' && Signature[3] == 'H') 431 return ClangSerializedASTBitstream; 432 } else if (Signature[0] == 'D' && Signature[1] == 'I') { 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] == 'A' && Signature[3] == 'G') 438 return ClangSerializedDiagnosticsBitstream; 439 } else if (Signature[0] == 'R' && Signature[1] == 'M') { 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] == 'R' && Signature[3] == 'K') 445 return LLVMBitstreamRemarks; 446 } else { 447 if (Error Err = tryRead(Signature[2], 4)) 448 return std::move(Err); 449 if (Error Err = tryRead(Signature[3], 4)) 450 return std::move(Err); 451 if (Error Err = tryRead(Signature[4], 4)) 452 return std::move(Err); 453 if (Error Err = tryRead(Signature[5], 4)) 454 return std::move(Err); 455 if (Signature[0] == 'B' && Signature[1] == 'C' && Signature[2] == 0x0 && 456 Signature[3] == 0xC && Signature[4] == 0xE && Signature[5] == 0xD) 457 return LLVMIRBitstream; 458 } 459 return UnknownBitstream; 460} 461 462static Expected<CurStreamTypeType> analyzeHeader(Optional<BCDumpOptions> O, 463 BitstreamCursor &Stream) { 464 ArrayRef<uint8_t> Bytes = Stream.getBitcodeBytes(); 465 const unsigned char *BufPtr = (const unsigned char *)Bytes.data(); 466 const unsigned char *EndBufPtr = BufPtr + Bytes.size(); 467 468 // If we have a wrapper header, parse it and ignore the non-bc file 469 // contents. The magic number is 0x0B17C0DE stored in little endian. 470 if (isBitcodeWrapper(BufPtr, EndBufPtr)) { 471 if (Bytes.size() < BWH_HeaderSize) 472 return reportError("Invalid bitcode wrapper header"); 473 474 if (O) { 475 unsigned Magic = support::endian::read32le(&BufPtr[BWH_MagicField]); 476 unsigned Version = support::endian::read32le(&BufPtr[BWH_VersionField]); 477 unsigned Offset = support::endian::read32le(&BufPtr[BWH_OffsetField]); 478 unsigned Size = support::endian::read32le(&BufPtr[BWH_SizeField]); 479 unsigned CPUType = support::endian::read32le(&BufPtr[BWH_CPUTypeField]); 480 481 O->OS << "<BITCODE_WRAPPER_HEADER" 482 << " Magic=" << format_hex(Magic, 10) 483 << " Version=" << format_hex(Version, 10) 484 << " Offset=" << format_hex(Offset, 10) 485 << " Size=" << format_hex(Size, 10) 486 << " CPUType=" << format_hex(CPUType, 10) << "/>\n"; 487 } 488 489 if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true)) 490 return reportError("Invalid bitcode wrapper header"); 491 } 492 493 // Use the cursor modified by skipping the wrapper header. 494 Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr)); 495 496 return ReadSignature(Stream); 497} 498 499static bool canDecodeBlob(unsigned Code, unsigned BlockID) { 500 return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS; 501} 502 503Error BitcodeAnalyzer::decodeMetadataStringsBlob(StringRef Indent, 504 ArrayRef<uint64_t> Record, 505 StringRef Blob, 506 raw_ostream &OS) { 507 if (Blob.empty()) 508 return reportError("Cannot decode empty blob."); 509 510 if (Record.size() != 2) 511 return reportError( 512 "Decoding metadata strings blob needs two record entries."); 513 514 unsigned NumStrings = Record[0]; 515 unsigned StringsOffset = Record[1]; 516 OS << " num-strings = " << NumStrings << " {\n"; 517 518 StringRef Lengths = Blob.slice(0, StringsOffset); 519 SimpleBitstreamCursor R(Lengths); 520 StringRef Strings = Blob.drop_front(StringsOffset); 521 do { 522 if (R.AtEndOfStream()) 523 return reportError("bad length"); 524 525 Expected<uint32_t> MaybeSize = R.ReadVBR(6); 526 if (!MaybeSize) 527 return MaybeSize.takeError(); 528 uint32_t Size = MaybeSize.get(); 529 if (Strings.size() < Size) 530 return reportError("truncated chars"); 531 532 OS << Indent << " '"; 533 OS.write_escaped(Strings.slice(0, Size), /*hex=*/true); 534 OS << "'\n"; 535 Strings = Strings.drop_front(Size); 536 } while (--NumStrings); 537 538 OS << Indent << " }"; 539 return Error::success(); 540} 541 542BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer, 543 Optional<StringRef> BlockInfoBuffer) 544 : Stream(Buffer) { 545 if (BlockInfoBuffer) 546 BlockInfoStream.emplace(*BlockInfoBuffer); 547} 548 549Error BitcodeAnalyzer::analyze(Optional<BCDumpOptions> O, 550 Optional<StringRef> CheckHash) { 551 Expected<CurStreamTypeType> MaybeType = analyzeHeader(O, Stream); 552 if (!MaybeType) 553 return MaybeType.takeError(); 554 else 555 CurStreamType = *MaybeType; 556 557 Stream.setBlockInfo(&BlockInfo); 558 559 // Read block info from BlockInfoStream, if specified. 560 // The block info must be a top-level block. 561 if (BlockInfoStream) { 562 BitstreamCursor BlockInfoCursor(*BlockInfoStream); 563 Expected<CurStreamTypeType> H = analyzeHeader(O, BlockInfoCursor); 564 if (!H) 565 return H.takeError(); 566 567 while (!BlockInfoCursor.AtEndOfStream()) { 568 Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode(); 569 if (!MaybeCode) 570 return MaybeCode.takeError(); 571 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK) 572 return reportError("Invalid record at top-level in block info file"); 573 574 Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID(); 575 if (!MaybeBlockID) 576 return MaybeBlockID.takeError(); 577 if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) { 578 Expected<Optional<BitstreamBlockInfo>> MaybeNewBlockInfo = 579 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true); 580 if (!MaybeNewBlockInfo) 581 return MaybeNewBlockInfo.takeError(); 582 Optional<BitstreamBlockInfo> NewBlockInfo = 583 std::move(MaybeNewBlockInfo.get()); 584 if (!NewBlockInfo) 585 return reportError("Malformed BlockInfoBlock in block info file"); 586 BlockInfo = std::move(*NewBlockInfo); 587 break; 588 } 589 590 if (Error Err = BlockInfoCursor.SkipBlock()) 591 return Err; 592 } 593 } 594 595 // Parse the top-level structure. We only allow blocks at the top-level. 596 while (!Stream.AtEndOfStream()) { 597 Expected<unsigned> MaybeCode = Stream.ReadCode(); 598 if (!MaybeCode) 599 return MaybeCode.takeError(); 600 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK) 601 return reportError("Invalid record at top-level"); 602 603 Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID(); 604 if (!MaybeBlockID) 605 return MaybeBlockID.takeError(); 606 607 if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash)) 608 return E; 609 ++NumTopBlocks; 610 } 611 612 return Error::success(); 613} 614 615void BitcodeAnalyzer::printStats(BCDumpOptions O, 616 Optional<StringRef> Filename) { 617 uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT; 618 // Print a summary of the read file. 619 O.OS << "Summary "; 620 if (Filename) 621 O.OS << "of " << Filename->data() << ":\n"; 622 O.OS << " Total size: "; 623 printSize(O.OS, BufferSizeBits); 624 O.OS << "\n"; 625 O.OS << " Stream type: "; 626 switch (CurStreamType) { 627 case UnknownBitstream: 628 O.OS << "unknown\n"; 629 break; 630 case LLVMIRBitstream: 631 O.OS << "LLVM IR\n"; 632 break; 633 case ClangSerializedASTBitstream: 634 O.OS << "Clang Serialized AST\n"; 635 break; 636 case ClangSerializedDiagnosticsBitstream: 637 O.OS << "Clang Serialized Diagnostics\n"; 638 break; 639 case LLVMBitstreamRemarks: 640 O.OS << "LLVM Remarks\n"; 641 break; 642 } 643 O.OS << " # Toplevel Blocks: " << NumTopBlocks << "\n"; 644 O.OS << "\n"; 645 646 // Emit per-block stats. 647 O.OS << "Per-block Summary:\n"; 648 for (std::map<unsigned, PerBlockIDStats>::iterator I = BlockIDStats.begin(), 649 E = BlockIDStats.end(); 650 I != E; ++I) { 651 O.OS << " Block ID #" << I->first; 652 if (Optional<const char *> BlockName = 653 GetBlockName(I->first, BlockInfo, CurStreamType)) 654 O.OS << " (" << *BlockName << ")"; 655 O.OS << ":\n"; 656 657 const PerBlockIDStats &Stats = I->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 (unsigned i = 0, e = FreqPairs.size(); i != e; ++i) { 697 const PerRecordStats &RecStats = Stats.CodeFreq[FreqPairs[i].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 FreqPairs[i].second, I->first, BlockInfo, CurStreamType)) 717 O.OS << *CodeName << "\n"; 718 else 719 O.OS << "UnknownCode" << FreqPairs[i].second << "\n"; 720 } 721 O.OS << "\n"; 722 } 723 } 724} 725 726Error 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) 741 O->OS << Indent << "<BLOCKINFO_BLOCK/>\n"; 742 Expected<Optional<BitstreamBlockInfo>> MaybeNewBlockInfo = 743 Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true); 744 if (!MaybeNewBlockInfo) 745 return MaybeNewBlockInfo.takeError(); 746 Optional<BitstreamBlockInfo> NewBlockInfo = 747 std::move(MaybeNewBlockInfo.get()); 748 if (!NewBlockInfo) 749 return reportError("Malformed BlockInfoBlock"); 750 BlockInfo = std::move(*NewBlockInfo); 751 if (Error Err = Stream.JumpToBit(BlockBitStart)) 752 return Err; 753 // It's not really interesting to dump the contents of the blockinfo 754 // block. 755 DumpRecords = false; 756 } 757 758 unsigned NumWords = 0; 759 if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords)) 760 return Err; 761 762 // Keep it for later, when we see a MODULE_HASH record 763 uint64_t BlockEntryPos = Stream.getCurrentByteNo(); 764 765 Optional<const char *> BlockName = None; 766 if (DumpRecords) { 767 O->OS << Indent << "<"; 768 if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType))) 769 O->OS << *BlockName; 770 else 771 O->OS << "UnknownBlock" << BlockID; 772 773 if (!O->Symbolic && BlockName) 774 O->OS << " BlockID=" << BlockID; 775 776 O->OS << " NumWords=" << NumWords 777 << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n"; 778 } 779 780 SmallVector<uint64_t, 64> Record; 781 782 // Keep the offset to the metadata index if seen. 783 uint64_t MetadataIndexOffset = 0; 784 785 // Read all the records for this block. 786 while (1) { 787 if (Stream.AtEndOfStream()) 788 return reportError("Premature end of bitstream"); 789 790 uint64_t RecordStartBit = Stream.GetCurrentBitNo(); 791 792 Expected<BitstreamEntry> MaybeEntry = 793 Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs); 794 if (!MaybeEntry) 795 return MaybeEntry.takeError(); 796 BitstreamEntry Entry = MaybeEntry.get(); 797 798 switch (Entry.Kind) { 799 case BitstreamEntry::Error: 800 return reportError("malformed bitcode file"); 801 case BitstreamEntry::EndBlock: { 802 uint64_t BlockBitEnd = Stream.GetCurrentBitNo(); 803 BlockStats.NumBits += BlockBitEnd - BlockBitStart; 804 if (DumpRecords) { 805 O->OS << Indent << "</"; 806 if (BlockName) 807 O->OS << *BlockName << ">\n"; 808 else 809 O->OS << "UnknownBlock" << BlockID << ">\n"; 810 } 811 return Error::success(); 812 } 813 814 case BitstreamEntry::SubBlock: { 815 uint64_t SubBlockBitStart = Stream.GetCurrentBitNo(); 816 if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash)) 817 return E; 818 ++BlockStats.NumSubBlocks; 819 uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo(); 820 821 // Don't include subblock sizes in the size of this block. 822 BlockBitStart += SubBlockBitEnd - SubBlockBitStart; 823 continue; 824 } 825 case BitstreamEntry::Record: 826 // The interesting case. 827 break; 828 } 829 830 if (Entry.ID == bitc::DEFINE_ABBREV) { 831 if (Error Err = Stream.ReadAbbrevRecord()) 832 return Err; 833 ++BlockStats.NumAbbrevs; 834 continue; 835 } 836 837 Record.clear(); 838 839 ++BlockStats.NumRecords; 840 841 StringRef Blob; 842 uint64_t CurrentRecordPos = Stream.GetCurrentBitNo(); 843 Expected<unsigned> MaybeCode = Stream.readRecord(Entry.ID, Record, &Blob); 844 if (!MaybeCode) 845 return MaybeCode.takeError(); 846 unsigned Code = MaybeCode.get(); 847 848 // Increment the # occurrences of this code. 849 if (BlockStats.CodeFreq.size() <= Code) 850 BlockStats.CodeFreq.resize(Code + 1); 851 BlockStats.CodeFreq[Code].NumInstances++; 852 BlockStats.CodeFreq[Code].TotalBits += 853 Stream.GetCurrentBitNo() - RecordStartBit; 854 if (Entry.ID != bitc::UNABBREV_RECORD) { 855 BlockStats.CodeFreq[Code].NumAbbrev++; 856 ++BlockStats.NumAbbreviatedRecords; 857 } 858 859 if (DumpRecords) { 860 O->OS << Indent << " <"; 861 Optional<const char *> CodeName = 862 GetCodeName(Code, BlockID, BlockInfo, CurStreamType); 863 if (CodeName) 864 O->OS << *CodeName; 865 else 866 O->OS << "UnknownCode" << Code; 867 if (!O->Symbolic && CodeName) 868 O->OS << " codeid=" << Code; 869 const BitCodeAbbrev *Abbv = nullptr; 870 if (Entry.ID != bitc::UNABBREV_RECORD) { 871 Abbv = Stream.getAbbrev(Entry.ID); 872 O->OS << " abbrevid=" << Entry.ID; 873 } 874 875 for (unsigned i = 0, e = Record.size(); i != e; ++i) 876 O->OS << " op" << i << "=" << (int64_t)Record[i]; 877 878 // If we found a metadata index, let's verify that we had an offset 879 // before and validate its forward reference offset was correct! 880 if (BlockID == bitc::METADATA_BLOCK_ID) { 881 if (Code == bitc::METADATA_INDEX_OFFSET) { 882 if (Record.size() != 2) 883 O->OS << "(Invalid record)"; 884 else { 885 auto Offset = Record[0] + (Record[1] << 32); 886 MetadataIndexOffset = Stream.GetCurrentBitNo() + Offset; 887 } 888 } 889 if (Code == bitc::METADATA_INDEX) { 890 O->OS << " (offset "; 891 if (MetadataIndexOffset == RecordStartBit) 892 O->OS << "match)"; 893 else 894 O->OS << "mismatch: " << MetadataIndexOffset << " vs " 895 << RecordStartBit << ")"; 896 } 897 } 898 899 // If we found a module hash, let's verify that it matches! 900 if (BlockID == bitc::MODULE_BLOCK_ID && Code == bitc::MODULE_CODE_HASH && 901 CheckHash.hasValue()) { 902 if (Record.size() != 5) 903 O->OS << " (invalid)"; 904 else { 905 // Recompute the hash and compare it to the one in the bitcode 906 SHA1 Hasher; 907 StringRef Hash; 908 Hasher.update(*CheckHash); 909 { 910 int BlockSize = (CurrentRecordPos / 8) - BlockEntryPos; 911 auto Ptr = Stream.getPointerToByte(BlockEntryPos, BlockSize); 912 Hasher.update(ArrayRef<uint8_t>(Ptr, BlockSize)); 913 Hash = Hasher.result(); 914 } 915 std::array<char, 20> RecordedHash; 916 int Pos = 0; 917 for (auto &Val : Record) { 918 assert(!(Val >> 32) && "Unexpected high bits set"); 919 support::endian::write32be(&RecordedHash[Pos], Val); 920 Pos += 4; 921 } 922 if (Hash == StringRef(RecordedHash.data(), RecordedHash.size())) 923 O->OS << " (match)"; 924 else 925 O->OS << " (!mismatch!)"; 926 } 927 } 928 929 O->OS << "/>"; 930 931 if (Abbv) { 932 for (unsigned i = 1, e = Abbv->getNumOperandInfos(); i != e; ++i) { 933 const BitCodeAbbrevOp &Op = Abbv->getOperandInfo(i); 934 if (!Op.isEncoding() || Op.getEncoding() != BitCodeAbbrevOp::Array) 935 continue; 936 assert(i + 2 == e && "Array op not second to last"); 937 std::string Str; 938 bool ArrayIsPrintable = true; 939 for (unsigned j = i - 1, je = Record.size(); j != je; ++j) { 940 if (!isPrint(static_cast<unsigned char>(Record[j]))) { 941 ArrayIsPrintable = false; 942 break; 943 } 944 Str += (char)Record[j]; 945 } 946 if (ArrayIsPrintable) 947 O->OS << " record string = '" << Str << "'"; 948 break; 949 } 950 } 951 952 if (Blob.data()) { 953 if (canDecodeBlob(Code, BlockID)) { 954 if (Error E = decodeMetadataStringsBlob(Indent, Record, Blob, O->OS)) 955 return E; 956 } else { 957 O->OS << " blob data = "; 958 if (O->ShowBinaryBlobs) { 959 O->OS << "'"; 960 O->OS.write_escaped(Blob, /*hex=*/true) << "'"; 961 } else { 962 bool BlobIsPrintable = true; 963 for (unsigned i = 0, e = Blob.size(); i != e; ++i) 964 if (!isPrint(static_cast<unsigned char>(Blob[i]))) { 965 BlobIsPrintable = false; 966 break; 967 } 968 969 if (BlobIsPrintable) 970 O->OS << "'" << Blob << "'"; 971 else 972 O->OS << "unprintable, " << Blob.size() << " bytes."; 973 } 974 } 975 } 976 977 O->OS << "\n"; 978 } 979 980 // Make sure that we can skip the current record. 981 if (Error Err = Stream.JumpToBit(CurrentRecordPos)) 982 return Err; 983 if (Expected<unsigned> Skipped = Stream.skipRecord(Entry.ID)) 984 ; // Do nothing. 985 else 986 return Skipped.takeError(); 987 } 988} 989 990