[ORC] Add std::tuple support to SimplePackedSerialization.
[llvm-project.git] / llvm / lib / Bitcode / Reader / BitcodeAnalyzer.cpp
blobf577d3886e015ac28a9e6b6a2c646bcfa2f912d8
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 //===----------------------------------------------------------------------===//
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"
17 using namespace llvm;
19 static Error reportError(StringRef Message) {
20 return createStringError(std::errc::illegal_byte_sequence, Message.data());
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;
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();
41 if (CurStreamType != LLVMIRBitstream)
42 return None;
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";
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";
104 return None;
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();
115 if (CurStreamType != LLVMIRBitstream)
116 return None;
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)
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)
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";
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";
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)
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)
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)
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)
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)
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)
317 case bitc::METADATA_ATTACHMENT_ID:
318 switch (CodeID) {
319 default:
320 return None;
321 STRINGIFY_CODE(METADATA, ATTACHMENT)
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)
367 case bitc::METADATA_KIND_BLOCK_ID:
368 switch (CodeID) {
369 default:
370 return None;
371 STRINGIFY_CODE(METADATA, KIND)
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";
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";
390 case bitc::STRTAB_BLOCK_ID:
391 switch (CodeID) {
392 default:
393 return None;
394 case bitc::STRTAB_BLOB:
395 return "BLOB";
397 case bitc::SYMTAB_BLOCK_ID:
398 switch (CodeID) {
399 default:
400 return None;
401 case bitc::SYMTAB_BLOB:
402 return "BLOB";
405 #undef STRINGIFY_CODE
408 static void printSize(raw_ostream &OS, double Bits) {
409 OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32));
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));
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();
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);
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;
466 return UnknownBitstream;
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();
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");
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]);
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";
496 if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true))
497 return reportError("Invalid bitcode wrapper header");
500 // Use the cursor modified by skipping the wrapper header.
501 Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
503 return ReadSignature(Stream);
506 static bool canDecodeBlob(unsigned Code, unsigned BlockID) {
507 return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS;
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.");
517 if (Record.size() != 2)
518 return reportError(
519 "Decoding metadata strings blob needs two record entries.");
521 unsigned NumStrings = Record[0];
522 unsigned StringsOffset = Record[1];
523 OS << " num-strings = " << NumStrings << " {\n";
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");
532 Expected<uint32_t> MaybeSize = R.ReadVBR(6);
533 if (!MaybeSize)
534 return MaybeSize.takeError();
535 uint32_t Size = MaybeSize.get();
536 if (Strings.size() < Size)
537 return reportError("truncated chars");
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);
545 OS << Indent << " }";
546 return Error::success();
549 BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer,
550 Optional<StringRef> BlockInfoBuffer)
551 : Stream(Buffer) {
552 if (BlockInfoBuffer)
553 BlockInfoStream.emplace(*BlockInfoBuffer);
556 Error BitcodeAnalyzer::analyze(Optional<BCDumpOptions> O,
557 Optional<StringRef> CheckHash) {
558 Expected<CurStreamTypeType> MaybeType = analyzeHeader(O, Stream);
559 if (!MaybeType)
560 return MaybeType.takeError();
561 else
562 CurStreamType = *MaybeType;
564 Stream.setBlockInfo(&BlockInfo);
566 // Read block info from BlockInfoStream, if specified.
567 // The block info must be a top-level block.
568 if (BlockInfoStream) {
569 BitstreamCursor BlockInfoCursor(*BlockInfoStream);
570 Expected<CurStreamTypeType> H = analyzeHeader(O, BlockInfoCursor);
571 if (!H)
572 return H.takeError();
574 while (!BlockInfoCursor.AtEndOfStream()) {
575 Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode();
576 if (!MaybeCode)
577 return MaybeCode.takeError();
578 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
579 return reportError("Invalid record at top-level in block info file");
581 Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID();
582 if (!MaybeBlockID)
583 return MaybeBlockID.takeError();
584 if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) {
585 Expected<Optional<BitstreamBlockInfo>> MaybeNewBlockInfo =
586 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true);
587 if (!MaybeNewBlockInfo)
588 return MaybeNewBlockInfo.takeError();
589 Optional<BitstreamBlockInfo> NewBlockInfo =
590 std::move(MaybeNewBlockInfo.get());
591 if (!NewBlockInfo)
592 return reportError("Malformed BlockInfoBlock in block info file");
593 BlockInfo = std::move(*NewBlockInfo);
594 break;
597 if (Error Err = BlockInfoCursor.SkipBlock())
598 return Err;
602 // Parse the top-level structure. We only allow blocks at the top-level.
603 while (!Stream.AtEndOfStream()) {
604 Expected<unsigned> MaybeCode = Stream.ReadCode();
605 if (!MaybeCode)
606 return MaybeCode.takeError();
607 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
608 return reportError("Invalid record at top-level");
610 Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID();
611 if (!MaybeBlockID)
612 return MaybeBlockID.takeError();
614 if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash))
615 return E;
616 ++NumTopBlocks;
619 return Error::success();
622 void BitcodeAnalyzer::printStats(BCDumpOptions O,
623 Optional<StringRef> Filename) {
624 uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT;
625 // Print a summary of the read file.
626 O.OS << "Summary ";
627 if (Filename)
628 O.OS << "of " << Filename->data() << ":\n";
629 O.OS << " Total size: ";
630 printSize(O.OS, BufferSizeBits);
631 O.OS << "\n";
632 O.OS << " Stream type: ";
633 switch (CurStreamType) {
634 case UnknownBitstream:
635 O.OS << "unknown\n";
636 break;
637 case LLVMIRBitstream:
638 O.OS << "LLVM IR\n";
639 break;
640 case ClangSerializedASTBitstream:
641 O.OS << "Clang Serialized AST\n";
642 break;
643 case ClangSerializedDiagnosticsBitstream:
644 O.OS << "Clang Serialized Diagnostics\n";
645 break;
646 case LLVMBitstreamRemarks:
647 O.OS << "LLVM Remarks\n";
648 break;
650 O.OS << " # Toplevel Blocks: " << NumTopBlocks << "\n";
651 O.OS << "\n";
653 // Emit per-block stats.
654 O.OS << "Per-block Summary:\n";
655 for (std::map<unsigned, PerBlockIDStats>::iterator I = BlockIDStats.begin(),
656 E = BlockIDStats.end();
657 I != E; ++I) {
658 O.OS << " Block ID #" << I->first;
659 if (Optional<const char *> BlockName =
660 GetBlockName(I->first, BlockInfo, CurStreamType))
661 O.OS << " (" << *BlockName << ")";
662 O.OS << ":\n";
664 const PerBlockIDStats &Stats = I->second;
665 O.OS << " Num Instances: " << Stats.NumInstances << "\n";
666 O.OS << " Total Size: ";
667 printSize(O.OS, Stats.NumBits);
668 O.OS << "\n";
669 double pct = (Stats.NumBits * 100.0) / BufferSizeBits;
670 O.OS << " Percent of file: " << format("%2.4f%%", pct) << "\n";
671 if (Stats.NumInstances > 1) {
672 O.OS << " Average Size: ";
673 printSize(O.OS, Stats.NumBits / (double)Stats.NumInstances);
674 O.OS << "\n";
675 O.OS << " Tot/Avg SubBlocks: " << Stats.NumSubBlocks << "/"
676 << Stats.NumSubBlocks / (double)Stats.NumInstances << "\n";
677 O.OS << " Tot/Avg Abbrevs: " << Stats.NumAbbrevs << "/"
678 << Stats.NumAbbrevs / (double)Stats.NumInstances << "\n";
679 O.OS << " Tot/Avg Records: " << Stats.NumRecords << "/"
680 << Stats.NumRecords / (double)Stats.NumInstances << "\n";
681 } else {
682 O.OS << " Num SubBlocks: " << Stats.NumSubBlocks << "\n";
683 O.OS << " Num Abbrevs: " << Stats.NumAbbrevs << "\n";
684 O.OS << " Num Records: " << Stats.NumRecords << "\n";
686 if (Stats.NumRecords) {
687 double pct = (Stats.NumAbbreviatedRecords * 100.0) / Stats.NumRecords;
688 O.OS << " Percent Abbrevs: " << format("%2.4f%%", pct) << "\n";
690 O.OS << "\n";
692 // Print a histogram of the codes we see.
693 if (O.Histogram && !Stats.CodeFreq.empty()) {
694 std::vector<std::pair<unsigned, unsigned>> FreqPairs; // <freq,code>
695 for (unsigned i = 0, e = Stats.CodeFreq.size(); i != e; ++i)
696 if (unsigned Freq = Stats.CodeFreq[i].NumInstances)
697 FreqPairs.push_back(std::make_pair(Freq, i));
698 llvm::stable_sort(FreqPairs);
699 std::reverse(FreqPairs.begin(), FreqPairs.end());
701 O.OS << "\tRecord Histogram:\n";
702 O.OS << "\t\t Count # Bits b/Rec % Abv Record Kind\n";
703 for (unsigned i = 0, e = FreqPairs.size(); i != e; ++i) {
704 const PerRecordStats &RecStats = Stats.CodeFreq[FreqPairs[i].second];
706 O.OS << format("\t\t%7d %9lu", RecStats.NumInstances,
707 (unsigned long)RecStats.TotalBits);
709 if (RecStats.NumInstances > 1)
710 O.OS << format(" %9.1f",
711 (double)RecStats.TotalBits / RecStats.NumInstances);
712 else
713 O.OS << " ";
715 if (RecStats.NumAbbrev)
716 O.OS << format(" %7.2f", (double)RecStats.NumAbbrev /
717 RecStats.NumInstances * 100);
718 else
719 O.OS << " ";
721 O.OS << " ";
722 if (Optional<const char *> CodeName = GetCodeName(
723 FreqPairs[i].second, I->first, BlockInfo, CurStreamType))
724 O.OS << *CodeName << "\n";
725 else
726 O.OS << "UnknownCode" << FreqPairs[i].second << "\n";
728 O.OS << "\n";
733 Error BitcodeAnalyzer::parseBlock(unsigned BlockID, unsigned IndentLevel,
734 Optional<BCDumpOptions> O,
735 Optional<StringRef> CheckHash) {
736 std::string Indent(IndentLevel * 2, ' ');
737 uint64_t BlockBitStart = Stream.GetCurrentBitNo();
739 // Get the statistics for this BlockID.
740 PerBlockIDStats &BlockStats = BlockIDStats[BlockID];
742 BlockStats.NumInstances++;
744 // BLOCKINFO is a special part of the stream.
745 bool DumpRecords = O.hasValue();
746 if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
747 if (O)
748 O->OS << Indent << "<BLOCKINFO_BLOCK/>\n";
749 Expected<Optional<BitstreamBlockInfo>> MaybeNewBlockInfo =
750 Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true);
751 if (!MaybeNewBlockInfo)
752 return MaybeNewBlockInfo.takeError();
753 Optional<BitstreamBlockInfo> NewBlockInfo =
754 std::move(MaybeNewBlockInfo.get());
755 if (!NewBlockInfo)
756 return reportError("Malformed BlockInfoBlock");
757 BlockInfo = std::move(*NewBlockInfo);
758 if (Error Err = Stream.JumpToBit(BlockBitStart))
759 return Err;
760 // It's not really interesting to dump the contents of the blockinfo
761 // block.
762 DumpRecords = false;
765 unsigned NumWords = 0;
766 if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords))
767 return Err;
769 // Keep it for later, when we see a MODULE_HASH record
770 uint64_t BlockEntryPos = Stream.getCurrentByteNo();
772 Optional<const char *> BlockName = None;
773 if (DumpRecords) {
774 O->OS << Indent << "<";
775 if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType)))
776 O->OS << *BlockName;
777 else
778 O->OS << "UnknownBlock" << BlockID;
780 if (!O->Symbolic && BlockName)
781 O->OS << " BlockID=" << BlockID;
783 O->OS << " NumWords=" << NumWords
784 << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n";
787 SmallVector<uint64_t, 64> Record;
789 // Keep the offset to the metadata index if seen.
790 uint64_t MetadataIndexOffset = 0;
792 // Read all the records for this block.
793 while (1) {
794 if (Stream.AtEndOfStream())
795 return reportError("Premature end of bitstream");
797 uint64_t RecordStartBit = Stream.GetCurrentBitNo();
799 Expected<BitstreamEntry> MaybeEntry =
800 Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs);
801 if (!MaybeEntry)
802 return MaybeEntry.takeError();
803 BitstreamEntry Entry = MaybeEntry.get();
805 switch (Entry.Kind) {
806 case BitstreamEntry::Error:
807 return reportError("malformed bitcode file");
808 case BitstreamEntry::EndBlock: {
809 uint64_t BlockBitEnd = Stream.GetCurrentBitNo();
810 BlockStats.NumBits += BlockBitEnd - BlockBitStart;
811 if (DumpRecords) {
812 O->OS << Indent << "</";
813 if (BlockName)
814 O->OS << *BlockName << ">\n";
815 else
816 O->OS << "UnknownBlock" << BlockID << ">\n";
818 return Error::success();
821 case BitstreamEntry::SubBlock: {
822 uint64_t SubBlockBitStart = Stream.GetCurrentBitNo();
823 if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash))
824 return E;
825 ++BlockStats.NumSubBlocks;
826 uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo();
828 // Don't include subblock sizes in the size of this block.
829 BlockBitStart += SubBlockBitEnd - SubBlockBitStart;
830 continue;
832 case BitstreamEntry::Record:
833 // The interesting case.
834 break;
837 if (Entry.ID == bitc::DEFINE_ABBREV) {
838 if (Error Err = Stream.ReadAbbrevRecord())
839 return Err;
840 ++BlockStats.NumAbbrevs;
841 continue;
844 Record.clear();
846 ++BlockStats.NumRecords;
848 StringRef Blob;
849 uint64_t CurrentRecordPos = Stream.GetCurrentBitNo();
850 Expected<unsigned> MaybeCode = Stream.readRecord(Entry.ID, Record, &Blob);
851 if (!MaybeCode)
852 return MaybeCode.takeError();
853 unsigned Code = MaybeCode.get();
855 // Increment the # occurrences of this code.
856 if (BlockStats.CodeFreq.size() <= Code)
857 BlockStats.CodeFreq.resize(Code + 1);
858 BlockStats.CodeFreq[Code].NumInstances++;
859 BlockStats.CodeFreq[Code].TotalBits +=
860 Stream.GetCurrentBitNo() - RecordStartBit;
861 if (Entry.ID != bitc::UNABBREV_RECORD) {
862 BlockStats.CodeFreq[Code].NumAbbrev++;
863 ++BlockStats.NumAbbreviatedRecords;
866 if (DumpRecords) {
867 O->OS << Indent << " <";
868 Optional<const char *> CodeName =
869 GetCodeName(Code, BlockID, BlockInfo, CurStreamType);
870 if (CodeName)
871 O->OS << *CodeName;
872 else
873 O->OS << "UnknownCode" << Code;
874 if (!O->Symbolic && CodeName)
875 O->OS << " codeid=" << Code;
876 const BitCodeAbbrev *Abbv = nullptr;
877 if (Entry.ID != bitc::UNABBREV_RECORD) {
878 Abbv = Stream.getAbbrev(Entry.ID);
879 O->OS << " abbrevid=" << Entry.ID;
882 for (unsigned i = 0, e = Record.size(); i != e; ++i)
883 O->OS << " op" << i << "=" << (int64_t)Record[i];
885 // If we found a metadata index, let's verify that we had an offset
886 // before and validate its forward reference offset was correct!
887 if (BlockID == bitc::METADATA_BLOCK_ID) {
888 if (Code == bitc::METADATA_INDEX_OFFSET) {
889 if (Record.size() != 2)
890 O->OS << "(Invalid record)";
891 else {
892 auto Offset = Record[0] + (Record[1] << 32);
893 MetadataIndexOffset = Stream.GetCurrentBitNo() + Offset;
896 if (Code == bitc::METADATA_INDEX) {
897 O->OS << " (offset ";
898 if (MetadataIndexOffset == RecordStartBit)
899 O->OS << "match)";
900 else
901 O->OS << "mismatch: " << MetadataIndexOffset << " vs "
902 << RecordStartBit << ")";
906 // If we found a module hash, let's verify that it matches!
907 if (BlockID == bitc::MODULE_BLOCK_ID && Code == bitc::MODULE_CODE_HASH &&
908 CheckHash.hasValue()) {
909 if (Record.size() != 5)
910 O->OS << " (invalid)";
911 else {
912 // Recompute the hash and compare it to the one in the bitcode
913 SHA1 Hasher;
914 StringRef Hash;
915 Hasher.update(*CheckHash);
917 int BlockSize = (CurrentRecordPos / 8) - BlockEntryPos;
918 auto Ptr = Stream.getPointerToByte(BlockEntryPos, BlockSize);
919 Hasher.update(ArrayRef<uint8_t>(Ptr, BlockSize));
920 Hash = Hasher.result();
922 std::array<char, 20> RecordedHash;
923 int Pos = 0;
924 for (auto &Val : Record) {
925 assert(!(Val >> 32) && "Unexpected high bits set");
926 support::endian::write32be(&RecordedHash[Pos], Val);
927 Pos += 4;
929 if (Hash == StringRef(RecordedHash.data(), RecordedHash.size()))
930 O->OS << " (match)";
931 else
932 O->OS << " (!mismatch!)";
936 O->OS << "/>";
938 if (Abbv) {
939 for (unsigned i = 1, e = Abbv->getNumOperandInfos(); i != e; ++i) {
940 const BitCodeAbbrevOp &Op = Abbv->getOperandInfo(i);
941 if (!Op.isEncoding() || Op.getEncoding() != BitCodeAbbrevOp::Array)
942 continue;
943 assert(i + 2 == e && "Array op not second to last");
944 std::string Str;
945 bool ArrayIsPrintable = true;
946 for (unsigned j = i - 1, je = Record.size(); j != je; ++j) {
947 if (!isPrint(static_cast<unsigned char>(Record[j]))) {
948 ArrayIsPrintable = false;
949 break;
951 Str += (char)Record[j];
953 if (ArrayIsPrintable)
954 O->OS << " record string = '" << Str << "'";
955 break;
959 if (Blob.data()) {
960 if (canDecodeBlob(Code, BlockID)) {
961 if (Error E = decodeMetadataStringsBlob(Indent, Record, Blob, O->OS))
962 return E;
963 } else {
964 O->OS << " blob data = ";
965 if (O->ShowBinaryBlobs) {
966 O->OS << "'";
967 O->OS.write_escaped(Blob, /*hex=*/true) << "'";
968 } else {
969 bool BlobIsPrintable = true;
970 for (unsigned i = 0, e = Blob.size(); i != e; ++i)
971 if (!isPrint(static_cast<unsigned char>(Blob[i]))) {
972 BlobIsPrintable = false;
973 break;
976 if (BlobIsPrintable)
977 O->OS << "'" << Blob << "'";
978 else
979 O->OS << "unprintable, " << Blob.size() << " bytes.";
984 O->OS << "\n";
987 // Make sure that we can skip the current record.
988 if (Error Err = Stream.JumpToBit(CurrentRecordPos))
989 return Err;
990 if (Expected<unsigned> Skipped = Stream.skipRecord(Entry.ID))
991 ; // Do nothing.
992 else
993 return Skipped.takeError();