xref: /llvm-project/llvm/lib/Bitcode/Reader/BitcodeAnalyzer.cpp (revision 9513f2fdf2ad50f55726154a6b6a4aa463bc457f)
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 #include <optional>
17 
18 using namespace llvm;
19 
20 static Error reportError(StringRef Message) {
21   return createStringError(std::errc::illegal_byte_sequence, Message.data());
22 }
23 
24 /// Return a symbolic block name if known, otherwise return null.
25 static std::optional<const char *>
26 GetBlockName(unsigned BlockID, const BitstreamBlockInfo &BlockInfo,
27              CurStreamTypeType CurStreamType) {
28   // Standard blocks for all bitcode files.
29   if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) {
30     if (BlockID == bitc::BLOCKINFO_BLOCK_ID)
31       return "BLOCKINFO_BLOCK";
32     return std::nullopt;
33   }
34 
35   // Check to see if we have a blockinfo record for this block, with a name.
36   if (const BitstreamBlockInfo::BlockInfo *Info =
37           BlockInfo.getBlockInfo(BlockID)) {
38     if (!Info->Name.empty())
39       return Info->Name.c_str();
40   }
41 
42   if (CurStreamType != LLVMIRBitstream)
43     return std::nullopt;
44 
45   switch (BlockID) {
46   default:
47     return std::nullopt;
48   case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID:
49     return "OPERAND_BUNDLE_TAGS_BLOCK";
50   case bitc::MODULE_BLOCK_ID:
51     return "MODULE_BLOCK";
52   case bitc::PARAMATTR_BLOCK_ID:
53     return "PARAMATTR_BLOCK";
54   case bitc::PARAMATTR_GROUP_BLOCK_ID:
55     return "PARAMATTR_GROUP_BLOCK_ID";
56   case bitc::TYPE_BLOCK_ID_NEW:
57     return "TYPE_BLOCK_ID";
58   case bitc::CONSTANTS_BLOCK_ID:
59     return "CONSTANTS_BLOCK";
60   case bitc::FUNCTION_BLOCK_ID:
61     return "FUNCTION_BLOCK";
62   case bitc::IDENTIFICATION_BLOCK_ID:
63     return "IDENTIFICATION_BLOCK_ID";
64   case bitc::VALUE_SYMTAB_BLOCK_ID:
65     return "VALUE_SYMTAB";
66   case bitc::METADATA_BLOCK_ID:
67     return "METADATA_BLOCK";
68   case bitc::METADATA_KIND_BLOCK_ID:
69     return "METADATA_KIND_BLOCK";
70   case bitc::METADATA_ATTACHMENT_ID:
71     return "METADATA_ATTACHMENT_BLOCK";
72   case bitc::USELIST_BLOCK_ID:
73     return "USELIST_BLOCK_ID";
74   case bitc::GLOBALVAL_SUMMARY_BLOCK_ID:
75     return "GLOBALVAL_SUMMARY_BLOCK";
76   case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID:
77     return "FULL_LTO_GLOBALVAL_SUMMARY_BLOCK";
78   case bitc::MODULE_STRTAB_BLOCK_ID:
79     return "MODULE_STRTAB_BLOCK";
80   case bitc::STRTAB_BLOCK_ID:
81     return "STRTAB_BLOCK";
82   case bitc::SYMTAB_BLOCK_ID:
83     return "SYMTAB_BLOCK";
84   }
85 }
86 
87 /// Return a symbolic code name if known, otherwise return null.
88 static std::optional<const char *>
89 GetCodeName(unsigned CodeID, unsigned BlockID,
90             const BitstreamBlockInfo &BlockInfo,
91             CurStreamTypeType CurStreamType) {
92   // Standard blocks for all bitcode files.
93   if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) {
94     if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
95       switch (CodeID) {
96       default:
97         return std::nullopt;
98       case bitc::BLOCKINFO_CODE_SETBID:
99         return "SETBID";
100       case bitc::BLOCKINFO_CODE_BLOCKNAME:
101         return "BLOCKNAME";
102       case bitc::BLOCKINFO_CODE_SETRECORDNAME:
103         return "SETRECORDNAME";
104       }
105     }
106     return std::nullopt;
107   }
108 
109   // Check to see if we have a blockinfo record for this record, with a name.
110   if (const BitstreamBlockInfo::BlockInfo *Info =
111           BlockInfo.getBlockInfo(BlockID)) {
112     for (const std::pair<unsigned, std::string> &RN : Info->RecordNames)
113       if (RN.first == CodeID)
114         return RN.second.c_str();
115   }
116 
117   if (CurStreamType != LLVMIRBitstream)
118     return std::nullopt;
119 
120 #define STRINGIFY_CODE(PREFIX, CODE)                                           \
121   case bitc::PREFIX##_##CODE:                                                  \
122     return #CODE;
123   switch (BlockID) {
124   default:
125     return std::nullopt;
126   case bitc::MODULE_BLOCK_ID:
127     switch (CodeID) {
128     default:
129       return std::nullopt;
130       STRINGIFY_CODE(MODULE_CODE, VERSION)
131       STRINGIFY_CODE(MODULE_CODE, TRIPLE)
132       STRINGIFY_CODE(MODULE_CODE, DATALAYOUT)
133       STRINGIFY_CODE(MODULE_CODE, ASM)
134       STRINGIFY_CODE(MODULE_CODE, SECTIONNAME)
135       STRINGIFY_CODE(MODULE_CODE, DEPLIB) // Deprecated, present in old bitcode
136       STRINGIFY_CODE(MODULE_CODE, GLOBALVAR)
137       STRINGIFY_CODE(MODULE_CODE, FUNCTION)
138       STRINGIFY_CODE(MODULE_CODE, ALIAS)
139       STRINGIFY_CODE(MODULE_CODE, GCNAME)
140       STRINGIFY_CODE(MODULE_CODE, COMDAT)
141       STRINGIFY_CODE(MODULE_CODE, VSTOFFSET)
142       STRINGIFY_CODE(MODULE_CODE, METADATA_VALUES_UNUSED)
143       STRINGIFY_CODE(MODULE_CODE, SOURCE_FILENAME)
144       STRINGIFY_CODE(MODULE_CODE, HASH)
145     }
146   case bitc::IDENTIFICATION_BLOCK_ID:
147     switch (CodeID) {
148     default:
149       return std::nullopt;
150       STRINGIFY_CODE(IDENTIFICATION_CODE, STRING)
151       STRINGIFY_CODE(IDENTIFICATION_CODE, EPOCH)
152     }
153   case bitc::PARAMATTR_BLOCK_ID:
154     switch (CodeID) {
155     default:
156       return std::nullopt;
157     // FIXME: Should these be different?
158     case bitc::PARAMATTR_CODE_ENTRY_OLD:
159       return "ENTRY";
160     case bitc::PARAMATTR_CODE_ENTRY:
161       return "ENTRY";
162     }
163   case bitc::PARAMATTR_GROUP_BLOCK_ID:
164     switch (CodeID) {
165     default:
166       return std::nullopt;
167     case bitc::PARAMATTR_GRP_CODE_ENTRY:
168       return "ENTRY";
169     }
170   case bitc::TYPE_BLOCK_ID_NEW:
171     switch (CodeID) {
172     default:
173       return std::nullopt;
174       STRINGIFY_CODE(TYPE_CODE, NUMENTRY)
175       STRINGIFY_CODE(TYPE_CODE, VOID)
176       STRINGIFY_CODE(TYPE_CODE, FLOAT)
177       STRINGIFY_CODE(TYPE_CODE, DOUBLE)
178       STRINGIFY_CODE(TYPE_CODE, LABEL)
179       STRINGIFY_CODE(TYPE_CODE, OPAQUE)
180       STRINGIFY_CODE(TYPE_CODE, INTEGER)
181       STRINGIFY_CODE(TYPE_CODE, POINTER)
182       STRINGIFY_CODE(TYPE_CODE, HALF)
183       STRINGIFY_CODE(TYPE_CODE, ARRAY)
184       STRINGIFY_CODE(TYPE_CODE, VECTOR)
185       STRINGIFY_CODE(TYPE_CODE, X86_FP80)
186       STRINGIFY_CODE(TYPE_CODE, FP128)
187       STRINGIFY_CODE(TYPE_CODE, PPC_FP128)
188       STRINGIFY_CODE(TYPE_CODE, METADATA)
189       STRINGIFY_CODE(TYPE_CODE, X86_MMX)
190       STRINGIFY_CODE(TYPE_CODE, STRUCT_ANON)
191       STRINGIFY_CODE(TYPE_CODE, STRUCT_NAME)
192       STRINGIFY_CODE(TYPE_CODE, STRUCT_NAMED)
193       STRINGIFY_CODE(TYPE_CODE, FUNCTION)
194       STRINGIFY_CODE(TYPE_CODE, TOKEN)
195       STRINGIFY_CODE(TYPE_CODE, BFLOAT)
196     }
197 
198   case bitc::CONSTANTS_BLOCK_ID:
199     switch (CodeID) {
200     default:
201       return std::nullopt;
202       STRINGIFY_CODE(CST_CODE, SETTYPE)
203       STRINGIFY_CODE(CST_CODE, NULL)
204       STRINGIFY_CODE(CST_CODE, UNDEF)
205       STRINGIFY_CODE(CST_CODE, INTEGER)
206       STRINGIFY_CODE(CST_CODE, WIDE_INTEGER)
207       STRINGIFY_CODE(CST_CODE, FLOAT)
208       STRINGIFY_CODE(CST_CODE, AGGREGATE)
209       STRINGIFY_CODE(CST_CODE, STRING)
210       STRINGIFY_CODE(CST_CODE, CSTRING)
211       STRINGIFY_CODE(CST_CODE, CE_BINOP)
212       STRINGIFY_CODE(CST_CODE, CE_CAST)
213       STRINGIFY_CODE(CST_CODE, CE_GEP)
214       STRINGIFY_CODE(CST_CODE, CE_INBOUNDS_GEP)
215       STRINGIFY_CODE(CST_CODE, CE_SELECT)
216       STRINGIFY_CODE(CST_CODE, CE_EXTRACTELT)
217       STRINGIFY_CODE(CST_CODE, CE_INSERTELT)
218       STRINGIFY_CODE(CST_CODE, CE_SHUFFLEVEC)
219       STRINGIFY_CODE(CST_CODE, CE_CMP)
220       STRINGIFY_CODE(CST_CODE, INLINEASM)
221       STRINGIFY_CODE(CST_CODE, CE_SHUFVEC_EX)
222       STRINGIFY_CODE(CST_CODE, CE_UNOP)
223       STRINGIFY_CODE(CST_CODE, DSO_LOCAL_EQUIVALENT)
224       STRINGIFY_CODE(CST_CODE, NO_CFI_VALUE)
225       STRINGIFY_CODE(CST_CODE, PTRAUTH)
226     case bitc::CST_CODE_BLOCKADDRESS:
227       return "CST_CODE_BLOCKADDRESS";
228       STRINGIFY_CODE(CST_CODE, DATA)
229     }
230   case bitc::FUNCTION_BLOCK_ID:
231     switch (CodeID) {
232     default:
233       return std::nullopt;
234       STRINGIFY_CODE(FUNC_CODE, DECLAREBLOCKS)
235       STRINGIFY_CODE(FUNC_CODE, INST_BINOP)
236       STRINGIFY_CODE(FUNC_CODE, INST_CAST)
237       STRINGIFY_CODE(FUNC_CODE, INST_GEP_OLD)
238       STRINGIFY_CODE(FUNC_CODE, INST_INBOUNDS_GEP_OLD)
239       STRINGIFY_CODE(FUNC_CODE, INST_SELECT)
240       STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTELT)
241       STRINGIFY_CODE(FUNC_CODE, INST_INSERTELT)
242       STRINGIFY_CODE(FUNC_CODE, INST_SHUFFLEVEC)
243       STRINGIFY_CODE(FUNC_CODE, INST_CMP)
244       STRINGIFY_CODE(FUNC_CODE, INST_RET)
245       STRINGIFY_CODE(FUNC_CODE, INST_BR)
246       STRINGIFY_CODE(FUNC_CODE, INST_SWITCH)
247       STRINGIFY_CODE(FUNC_CODE, INST_INVOKE)
248       STRINGIFY_CODE(FUNC_CODE, INST_UNOP)
249       STRINGIFY_CODE(FUNC_CODE, INST_UNREACHABLE)
250       STRINGIFY_CODE(FUNC_CODE, INST_CLEANUPRET)
251       STRINGIFY_CODE(FUNC_CODE, INST_CATCHRET)
252       STRINGIFY_CODE(FUNC_CODE, INST_CATCHPAD)
253       STRINGIFY_CODE(FUNC_CODE, INST_PHI)
254       STRINGIFY_CODE(FUNC_CODE, INST_ALLOCA)
255       STRINGIFY_CODE(FUNC_CODE, INST_LOAD)
256       STRINGIFY_CODE(FUNC_CODE, INST_VAARG)
257       STRINGIFY_CODE(FUNC_CODE, INST_STORE)
258       STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTVAL)
259       STRINGIFY_CODE(FUNC_CODE, INST_INSERTVAL)
260       STRINGIFY_CODE(FUNC_CODE, INST_CMP2)
261       STRINGIFY_CODE(FUNC_CODE, INST_VSELECT)
262       STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC_AGAIN)
263       STRINGIFY_CODE(FUNC_CODE, INST_CALL)
264       STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC)
265       STRINGIFY_CODE(FUNC_CODE, INST_GEP)
266       STRINGIFY_CODE(FUNC_CODE, OPERAND_BUNDLE)
267       STRINGIFY_CODE(FUNC_CODE, INST_FENCE)
268       STRINGIFY_CODE(FUNC_CODE, INST_ATOMICRMW)
269       STRINGIFY_CODE(FUNC_CODE, INST_LOADATOMIC)
270       STRINGIFY_CODE(FUNC_CODE, INST_STOREATOMIC)
271       STRINGIFY_CODE(FUNC_CODE, INST_CMPXCHG)
272       STRINGIFY_CODE(FUNC_CODE, INST_CALLBR)
273       STRINGIFY_CODE(FUNC_CODE, BLOCKADDR_USERS)
274       STRINGIFY_CODE(FUNC_CODE, DEBUG_RECORD_DECLARE)
275       STRINGIFY_CODE(FUNC_CODE, DEBUG_RECORD_VALUE)
276       STRINGIFY_CODE(FUNC_CODE, DEBUG_RECORD_ASSIGN)
277       STRINGIFY_CODE(FUNC_CODE, DEBUG_RECORD_VALUE_SIMPLE)
278       STRINGIFY_CODE(FUNC_CODE, DEBUG_RECORD_LABEL)
279     }
280   case bitc::VALUE_SYMTAB_BLOCK_ID:
281     switch (CodeID) {
282     default:
283       return std::nullopt;
284       STRINGIFY_CODE(VST_CODE, ENTRY)
285       STRINGIFY_CODE(VST_CODE, BBENTRY)
286       STRINGIFY_CODE(VST_CODE, FNENTRY)
287       STRINGIFY_CODE(VST_CODE, COMBINED_ENTRY)
288     }
289   case bitc::MODULE_STRTAB_BLOCK_ID:
290     switch (CodeID) {
291     default:
292       return std::nullopt;
293       STRINGIFY_CODE(MST_CODE, ENTRY)
294       STRINGIFY_CODE(MST_CODE, HASH)
295     }
296   case bitc::GLOBALVAL_SUMMARY_BLOCK_ID:
297   case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID:
298     switch (CodeID) {
299     default:
300       return std::nullopt;
301       STRINGIFY_CODE(FS, PERMODULE)
302       STRINGIFY_CODE(FS, PERMODULE_PROFILE)
303       STRINGIFY_CODE(FS, PERMODULE_RELBF)
304       STRINGIFY_CODE(FS, PERMODULE_GLOBALVAR_INIT_REFS)
305       STRINGIFY_CODE(FS, PERMODULE_VTABLE_GLOBALVAR_INIT_REFS)
306       STRINGIFY_CODE(FS, COMBINED)
307       STRINGIFY_CODE(FS, COMBINED_PROFILE)
308       STRINGIFY_CODE(FS, COMBINED_GLOBALVAR_INIT_REFS)
309       STRINGIFY_CODE(FS, ALIAS)
310       STRINGIFY_CODE(FS, COMBINED_ALIAS)
311       STRINGIFY_CODE(FS, COMBINED_ORIGINAL_NAME)
312       STRINGIFY_CODE(FS, VERSION)
313       STRINGIFY_CODE(FS, FLAGS)
314       STRINGIFY_CODE(FS, TYPE_TESTS)
315       STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_VCALLS)
316       STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_VCALLS)
317       STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_CONST_VCALL)
318       STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_CONST_VCALL)
319       STRINGIFY_CODE(FS, VALUE_GUID)
320       STRINGIFY_CODE(FS, CFI_FUNCTION_DEFS)
321       STRINGIFY_CODE(FS, CFI_FUNCTION_DECLS)
322       STRINGIFY_CODE(FS, TYPE_ID)
323       STRINGIFY_CODE(FS, TYPE_ID_METADATA)
324       STRINGIFY_CODE(FS, BLOCK_COUNT)
325       STRINGIFY_CODE(FS, PARAM_ACCESS)
326       STRINGIFY_CODE(FS, PERMODULE_CALLSITE_INFO)
327       STRINGIFY_CODE(FS, PERMODULE_ALLOC_INFO)
328       STRINGIFY_CODE(FS, COMBINED_CALLSITE_INFO)
329       STRINGIFY_CODE(FS, COMBINED_ALLOC_INFO)
330       STRINGIFY_CODE(FS, STACK_IDS)
331       STRINGIFY_CODE(FS, ALLOC_CONTEXT_IDS)
332     }
333   case bitc::METADATA_ATTACHMENT_ID:
334     switch (CodeID) {
335     default:
336       return std::nullopt;
337       STRINGIFY_CODE(METADATA, ATTACHMENT)
338     }
339   case bitc::METADATA_BLOCK_ID:
340     switch (CodeID) {
341     default:
342       return std::nullopt;
343       STRINGIFY_CODE(METADATA, STRING_OLD)
344       STRINGIFY_CODE(METADATA, VALUE)
345       STRINGIFY_CODE(METADATA, NODE)
346       STRINGIFY_CODE(METADATA, NAME)
347       STRINGIFY_CODE(METADATA, DISTINCT_NODE)
348       STRINGIFY_CODE(METADATA, KIND) // Older bitcode has it in a MODULE_BLOCK
349       STRINGIFY_CODE(METADATA, LOCATION)
350       STRINGIFY_CODE(METADATA, OLD_NODE)
351       STRINGIFY_CODE(METADATA, OLD_FN_NODE)
352       STRINGIFY_CODE(METADATA, NAMED_NODE)
353       STRINGIFY_CODE(METADATA, GENERIC_DEBUG)
354       STRINGIFY_CODE(METADATA, SUBRANGE)
355       STRINGIFY_CODE(METADATA, ENUMERATOR)
356       STRINGIFY_CODE(METADATA, BASIC_TYPE)
357       STRINGIFY_CODE(METADATA, FILE)
358       STRINGIFY_CODE(METADATA, DERIVED_TYPE)
359       STRINGIFY_CODE(METADATA, COMPOSITE_TYPE)
360       STRINGIFY_CODE(METADATA, SUBROUTINE_TYPE)
361       STRINGIFY_CODE(METADATA, COMPILE_UNIT)
362       STRINGIFY_CODE(METADATA, SUBPROGRAM)
363       STRINGIFY_CODE(METADATA, LEXICAL_BLOCK)
364       STRINGIFY_CODE(METADATA, LEXICAL_BLOCK_FILE)
365       STRINGIFY_CODE(METADATA, NAMESPACE)
366       STRINGIFY_CODE(METADATA, TEMPLATE_TYPE)
367       STRINGIFY_CODE(METADATA, TEMPLATE_VALUE)
368       STRINGIFY_CODE(METADATA, GLOBAL_VAR)
369       STRINGIFY_CODE(METADATA, LOCAL_VAR)
370       STRINGIFY_CODE(METADATA, EXPRESSION)
371       STRINGIFY_CODE(METADATA, OBJC_PROPERTY)
372       STRINGIFY_CODE(METADATA, IMPORTED_ENTITY)
373       STRINGIFY_CODE(METADATA, MODULE)
374       STRINGIFY_CODE(METADATA, MACRO)
375       STRINGIFY_CODE(METADATA, MACRO_FILE)
376       STRINGIFY_CODE(METADATA, STRINGS)
377       STRINGIFY_CODE(METADATA, GLOBAL_DECL_ATTACHMENT)
378       STRINGIFY_CODE(METADATA, GLOBAL_VAR_EXPR)
379       STRINGIFY_CODE(METADATA, INDEX_OFFSET)
380       STRINGIFY_CODE(METADATA, INDEX)
381       STRINGIFY_CODE(METADATA, ARG_LIST)
382     }
383   case bitc::METADATA_KIND_BLOCK_ID:
384     switch (CodeID) {
385     default:
386       return std::nullopt;
387       STRINGIFY_CODE(METADATA, KIND)
388     }
389   case bitc::USELIST_BLOCK_ID:
390     switch (CodeID) {
391     default:
392       return std::nullopt;
393     case bitc::USELIST_CODE_DEFAULT:
394       return "USELIST_CODE_DEFAULT";
395     case bitc::USELIST_CODE_BB:
396       return "USELIST_CODE_BB";
397     }
398 
399   case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID:
400     switch (CodeID) {
401     default:
402       return std::nullopt;
403     case bitc::OPERAND_BUNDLE_TAG:
404       return "OPERAND_BUNDLE_TAG";
405     }
406   case bitc::STRTAB_BLOCK_ID:
407     switch (CodeID) {
408     default:
409       return std::nullopt;
410     case bitc::STRTAB_BLOB:
411       return "BLOB";
412     }
413   case bitc::SYMTAB_BLOCK_ID:
414     switch (CodeID) {
415     default:
416       return std::nullopt;
417     case bitc::SYMTAB_BLOB:
418       return "BLOB";
419     }
420   }
421 #undef STRINGIFY_CODE
422 }
423 
424 static void printSize(raw_ostream &OS, double Bits) {
425   OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32));
426 }
427 static void printSize(raw_ostream &OS, uint64_t Bits) {
428   OS << format("%lub/%.2fB/%luW", (unsigned long)Bits, (double)Bits / 8,
429                (unsigned long)(Bits / 32));
430 }
431 
432 static Expected<CurStreamTypeType> ReadSignature(BitstreamCursor &Stream) {
433   auto tryRead = [&Stream](char &Dest, size_t size) -> Error {
434     if (Expected<SimpleBitstreamCursor::word_t> MaybeWord = Stream.Read(size))
435       Dest = MaybeWord.get();
436     else
437       return MaybeWord.takeError();
438     return Error::success();
439   };
440 
441   char Signature[6];
442   if (Error Err = tryRead(Signature[0], 8))
443     return std::move(Err);
444   if (Error Err = tryRead(Signature[1], 8))
445     return std::move(Err);
446 
447   // Autodetect the file contents, if it is one we know.
448   if (Signature[0] == 'C' && Signature[1] == 'P') {
449     if (Error Err = tryRead(Signature[2], 8))
450       return std::move(Err);
451     if (Error Err = tryRead(Signature[3], 8))
452       return std::move(Err);
453     if (Signature[2] == 'C' && Signature[3] == 'H')
454       return ClangSerializedASTBitstream;
455   } else if (Signature[0] == 'D' && Signature[1] == 'I') {
456     if (Error Err = tryRead(Signature[2], 8))
457       return std::move(Err);
458     if (Error Err = tryRead(Signature[3], 8))
459       return std::move(Err);
460     if (Signature[2] == 'A' && Signature[3] == 'G')
461       return ClangSerializedDiagnosticsBitstream;
462   } else if (Signature[0] == 'R' && Signature[1] == 'M') {
463     if (Error Err = tryRead(Signature[2], 8))
464       return std::move(Err);
465     if (Error Err = tryRead(Signature[3], 8))
466       return std::move(Err);
467     if (Signature[2] == 'R' && Signature[3] == 'K')
468       return LLVMBitstreamRemarks;
469   } else {
470     if (Error Err = tryRead(Signature[2], 4))
471       return std::move(Err);
472     if (Error Err = tryRead(Signature[3], 4))
473       return std::move(Err);
474     if (Error Err = tryRead(Signature[4], 4))
475       return std::move(Err);
476     if (Error Err = tryRead(Signature[5], 4))
477       return std::move(Err);
478     if (Signature[0] == 'B' && Signature[1] == 'C' && Signature[2] == 0x0 &&
479         Signature[3] == 0xC && Signature[4] == 0xE && Signature[5] == 0xD)
480       return LLVMIRBitstream;
481   }
482   return UnknownBitstream;
483 }
484 
485 static Expected<CurStreamTypeType> analyzeHeader(std::optional<BCDumpOptions> O,
486                                                  BitstreamCursor &Stream) {
487   ArrayRef<uint8_t> Bytes = Stream.getBitcodeBytes();
488   const unsigned char *BufPtr = (const unsigned char *)Bytes.data();
489   const unsigned char *EndBufPtr = BufPtr + Bytes.size();
490 
491   // If we have a wrapper header, parse it and ignore the non-bc file
492   // contents. The magic number is 0x0B17C0DE stored in little endian.
493   if (isBitcodeWrapper(BufPtr, EndBufPtr)) {
494     if (Bytes.size() < BWH_HeaderSize)
495       return reportError("Invalid bitcode wrapper header");
496 
497     if (O) {
498       unsigned Magic = support::endian::read32le(&BufPtr[BWH_MagicField]);
499       unsigned Version = support::endian::read32le(&BufPtr[BWH_VersionField]);
500       unsigned Offset = support::endian::read32le(&BufPtr[BWH_OffsetField]);
501       unsigned Size = support::endian::read32le(&BufPtr[BWH_SizeField]);
502       unsigned CPUType = support::endian::read32le(&BufPtr[BWH_CPUTypeField]);
503 
504       O->OS << "<BITCODE_WRAPPER_HEADER"
505             << " Magic=" << format_hex(Magic, 10)
506             << " Version=" << format_hex(Version, 10)
507             << " Offset=" << format_hex(Offset, 10)
508             << " Size=" << format_hex(Size, 10)
509             << " CPUType=" << format_hex(CPUType, 10) << "/>\n";
510     }
511 
512     if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true))
513       return reportError("Invalid bitcode wrapper header");
514   }
515 
516   // Use the cursor modified by skipping the wrapper header.
517   Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
518 
519   return ReadSignature(Stream);
520 }
521 
522 static bool canDecodeBlob(unsigned Code, unsigned BlockID) {
523   return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS;
524 }
525 
526 Error BitcodeAnalyzer::decodeMetadataStringsBlob(StringRef Indent,
527                                                  ArrayRef<uint64_t> Record,
528                                                  StringRef Blob,
529                                                  raw_ostream &OS) {
530   if (Blob.empty())
531     return reportError("Cannot decode empty blob.");
532 
533   if (Record.size() != 2)
534     return reportError(
535         "Decoding metadata strings blob needs two record entries.");
536 
537   unsigned NumStrings = Record[0];
538   unsigned StringsOffset = Record[1];
539   OS << " num-strings = " << NumStrings << " {\n";
540 
541   StringRef Lengths = Blob.slice(0, StringsOffset);
542   SimpleBitstreamCursor R(Lengths);
543   StringRef Strings = Blob.drop_front(StringsOffset);
544   do {
545     if (R.AtEndOfStream())
546       return reportError("bad length");
547 
548     uint32_t Size;
549     if (Error E = R.ReadVBR(6).moveInto(Size))
550       return E;
551     if (Strings.size() < Size)
552       return reportError("truncated chars");
553 
554     OS << Indent << "    '";
555     OS.write_escaped(Strings.slice(0, Size), /*hex=*/true);
556     OS << "'\n";
557     Strings = Strings.drop_front(Size);
558   } while (--NumStrings);
559 
560   OS << Indent << "  }";
561   return Error::success();
562 }
563 
564 BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer,
565                                  std::optional<StringRef> BlockInfoBuffer)
566     : Stream(Buffer) {
567   if (BlockInfoBuffer)
568     BlockInfoStream.emplace(*BlockInfoBuffer);
569 }
570 
571 Error BitcodeAnalyzer::analyze(std::optional<BCDumpOptions> O,
572                                std::optional<StringRef> CheckHash) {
573   if (Error E = analyzeHeader(O, Stream).moveInto(CurStreamType))
574     return E;
575 
576   Stream.setBlockInfo(&BlockInfo);
577 
578   // Read block info from BlockInfoStream, if specified.
579   // The block info must be a top-level block.
580   if (BlockInfoStream) {
581     BitstreamCursor BlockInfoCursor(*BlockInfoStream);
582     if (Error E = analyzeHeader(O, BlockInfoCursor).takeError())
583       return E;
584 
585     while (!BlockInfoCursor.AtEndOfStream()) {
586       Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode();
587       if (!MaybeCode)
588         return MaybeCode.takeError();
589       if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
590         return reportError("Invalid record at top-level in block info file");
591 
592       Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID();
593       if (!MaybeBlockID)
594         return MaybeBlockID.takeError();
595       if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) {
596         std::optional<BitstreamBlockInfo> NewBlockInfo;
597         if (Error E =
598                 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
599                     .moveInto(NewBlockInfo))
600           return E;
601         if (!NewBlockInfo)
602           return reportError("Malformed BlockInfoBlock in block info file");
603         BlockInfo = std::move(*NewBlockInfo);
604         break;
605       }
606 
607       if (Error Err = BlockInfoCursor.SkipBlock())
608         return Err;
609     }
610   }
611 
612   // Parse the top-level structure.  We only allow blocks at the top-level.
613   while (!Stream.AtEndOfStream()) {
614     Expected<unsigned> MaybeCode = Stream.ReadCode();
615     if (!MaybeCode)
616       return MaybeCode.takeError();
617     if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
618       return reportError("Invalid record at top-level");
619 
620     Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID();
621     if (!MaybeBlockID)
622       return MaybeBlockID.takeError();
623 
624     if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash))
625       return E;
626     ++NumTopBlocks;
627   }
628 
629   return Error::success();
630 }
631 
632 void BitcodeAnalyzer::printStats(BCDumpOptions O,
633                                  std::optional<StringRef> Filename) {
634   uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT;
635   // Print a summary of the read file.
636   O.OS << "Summary ";
637   if (Filename)
638     O.OS << "of " << Filename->data() << ":\n";
639   O.OS << "         Total size: ";
640   printSize(O.OS, BufferSizeBits);
641   O.OS << "\n";
642   O.OS << "        Stream type: ";
643   switch (CurStreamType) {
644   case UnknownBitstream:
645     O.OS << "unknown\n";
646     break;
647   case LLVMIRBitstream:
648     O.OS << "LLVM IR\n";
649     break;
650   case ClangSerializedASTBitstream:
651     O.OS << "Clang Serialized AST\n";
652     break;
653   case ClangSerializedDiagnosticsBitstream:
654     O.OS << "Clang Serialized Diagnostics\n";
655     break;
656   case LLVMBitstreamRemarks:
657     O.OS << "LLVM Remarks\n";
658     break;
659   }
660   O.OS << "  # Toplevel Blocks: " << NumTopBlocks << "\n";
661   O.OS << "\n";
662 
663   // Emit per-block stats.
664   O.OS << "Per-block Summary:\n";
665   for (const auto &Stat : BlockIDStats) {
666     O.OS << "  Block ID #" << Stat.first;
667     if (std::optional<const char *> BlockName =
668             GetBlockName(Stat.first, BlockInfo, CurStreamType))
669       O.OS << " (" << *BlockName << ")";
670     O.OS << ":\n";
671 
672     const PerBlockIDStats &Stats = Stat.second;
673     O.OS << "      Num Instances: " << Stats.NumInstances << "\n";
674     O.OS << "         Total Size: ";
675     printSize(O.OS, Stats.NumBits);
676     O.OS << "\n";
677     double pct = (Stats.NumBits * 100.0) / BufferSizeBits;
678     O.OS << "    Percent of file: " << format("%2.4f%%", pct) << "\n";
679     if (Stats.NumInstances > 1) {
680       O.OS << "       Average Size: ";
681       printSize(O.OS, Stats.NumBits / (double)Stats.NumInstances);
682       O.OS << "\n";
683       O.OS << "  Tot/Avg SubBlocks: " << Stats.NumSubBlocks << "/"
684            << Stats.NumSubBlocks / (double)Stats.NumInstances << "\n";
685       O.OS << "    Tot/Avg Abbrevs: " << Stats.NumAbbrevs << "/"
686            << Stats.NumAbbrevs / (double)Stats.NumInstances << "\n";
687       O.OS << "    Tot/Avg Records: " << Stats.NumRecords << "/"
688            << Stats.NumRecords / (double)Stats.NumInstances << "\n";
689     } else {
690       O.OS << "      Num SubBlocks: " << Stats.NumSubBlocks << "\n";
691       O.OS << "        Num Abbrevs: " << Stats.NumAbbrevs << "\n";
692       O.OS << "        Num Records: " << Stats.NumRecords << "\n";
693     }
694     if (Stats.NumRecords) {
695       double pct = (Stats.NumAbbreviatedRecords * 100.0) / Stats.NumRecords;
696       O.OS << "    Percent Abbrevs: " << format("%2.4f%%", pct) << "\n";
697     }
698     O.OS << "\n";
699 
700     // Print a histogram of the codes we see.
701     if (O.Histogram && !Stats.CodeFreq.empty()) {
702       std::vector<std::pair<unsigned, unsigned>> FreqPairs; // <freq,code>
703       for (unsigned i = 0, e = Stats.CodeFreq.size(); i != e; ++i)
704         if (unsigned Freq = Stats.CodeFreq[i].NumInstances)
705           FreqPairs.push_back(std::make_pair(Freq, i));
706       llvm::stable_sort(FreqPairs);
707       std::reverse(FreqPairs.begin(), FreqPairs.end());
708 
709       O.OS << "\tRecord Histogram:\n";
710       O.OS << "\t\t  Count    # Bits     b/Rec   % Abv  Record Kind\n";
711       for (const auto &FreqPair : FreqPairs) {
712         const PerRecordStats &RecStats = Stats.CodeFreq[FreqPair.second];
713 
714         O.OS << format("\t\t%7d %9lu", RecStats.NumInstances,
715                        (unsigned long)RecStats.TotalBits);
716 
717         if (RecStats.NumInstances > 1)
718           O.OS << format(" %9.1f",
719                          (double)RecStats.TotalBits / RecStats.NumInstances);
720         else
721           O.OS << "          ";
722 
723         if (RecStats.NumAbbrev)
724           O.OS << format(" %7.2f", (double)RecStats.NumAbbrev /
725                                        RecStats.NumInstances * 100);
726         else
727           O.OS << "        ";
728 
729         O.OS << "  ";
730         if (std::optional<const char *> CodeName = GetCodeName(
731                 FreqPair.second, Stat.first, BlockInfo, CurStreamType))
732           O.OS << *CodeName << "\n";
733         else
734           O.OS << "UnknownCode" << FreqPair.second << "\n";
735       }
736       O.OS << "\n";
737     }
738   }
739 }
740 
741 Error BitcodeAnalyzer::parseBlock(unsigned BlockID, unsigned IndentLevel,
742                                   std::optional<BCDumpOptions> O,
743                                   std::optional<StringRef> CheckHash) {
744   std::string Indent(IndentLevel * 2, ' ');
745   uint64_t BlockBitStart = Stream.GetCurrentBitNo();
746 
747   // Get the statistics for this BlockID.
748   PerBlockIDStats &BlockStats = BlockIDStats[BlockID];
749 
750   BlockStats.NumInstances++;
751 
752   // BLOCKINFO is a special part of the stream.
753   bool DumpRecords = O.has_value();
754   if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
755     if (O && !O->DumpBlockinfo)
756       O->OS << Indent << "<BLOCKINFO_BLOCK/>\n";
757     std::optional<BitstreamBlockInfo> NewBlockInfo;
758     if (Error E = Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
759                       .moveInto(NewBlockInfo))
760       return E;
761     if (!NewBlockInfo)
762       return reportError("Malformed BlockInfoBlock");
763     BlockInfo = std::move(*NewBlockInfo);
764     if (Error Err = Stream.JumpToBit(BlockBitStart))
765       return Err;
766     // It's not really interesting to dump the contents of the blockinfo
767     // block, so only do it if the user explicitly requests it.
768     DumpRecords = O && O->DumpBlockinfo;
769   }
770 
771   unsigned NumWords = 0;
772   if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords))
773     return Err;
774 
775   // Keep it for later, when we see a MODULE_HASH record
776   uint64_t BlockEntryPos = Stream.getCurrentByteNo();
777 
778   std::optional<const char *> BlockName;
779   if (DumpRecords) {
780     O->OS << Indent << "<";
781     if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType)))
782       O->OS << *BlockName;
783     else
784       O->OS << "UnknownBlock" << BlockID;
785 
786     if (!O->Symbolic && BlockName)
787       O->OS << " BlockID=" << BlockID;
788 
789     O->OS << " NumWords=" << NumWords
790           << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n";
791   }
792 
793   SmallVector<uint64_t, 64> Record;
794 
795   // Keep the offset to the metadata index if seen.
796   uint64_t MetadataIndexOffset = 0;
797 
798   // Read all the records for this block.
799   while (true) {
800     if (Stream.AtEndOfStream())
801       return reportError("Premature end of bitstream");
802 
803     uint64_t RecordStartBit = Stream.GetCurrentBitNo();
804 
805     BitstreamEntry Entry;
806     if (Error E = Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs)
807                       .moveInto(Entry))
808       return E;
809 
810     switch (Entry.Kind) {
811     case BitstreamEntry::Error:
812       return reportError("malformed bitcode file");
813     case BitstreamEntry::EndBlock: {
814       uint64_t BlockBitEnd = Stream.GetCurrentBitNo();
815       BlockStats.NumBits += BlockBitEnd - BlockBitStart;
816       if (DumpRecords) {
817         O->OS << Indent << "</";
818         if (BlockName)
819           O->OS << *BlockName << ">\n";
820         else
821           O->OS << "UnknownBlock" << BlockID << ">\n";
822       }
823       return Error::success();
824     }
825 
826     case BitstreamEntry::SubBlock: {
827       uint64_t SubBlockBitStart = Stream.GetCurrentBitNo();
828       if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash))
829         return E;
830       ++BlockStats.NumSubBlocks;
831       uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo();
832 
833       // Don't include subblock sizes in the size of this block.
834       BlockBitStart += SubBlockBitEnd - SubBlockBitStart;
835       continue;
836     }
837     case BitstreamEntry::Record:
838       // The interesting case.
839       break;
840     }
841 
842     if (Entry.ID == bitc::DEFINE_ABBREV) {
843       if (Error Err = Stream.ReadAbbrevRecord())
844         return Err;
845       ++BlockStats.NumAbbrevs;
846       continue;
847     }
848 
849     Record.clear();
850 
851     ++BlockStats.NumRecords;
852 
853     StringRef Blob;
854     uint64_t CurrentRecordPos = Stream.GetCurrentBitNo();
855     unsigned Code;
856     if (Error E = Stream.readRecord(Entry.ID, Record, &Blob).moveInto(Code))
857       return E;
858 
859     // Increment the # occurrences of this code.
860     if (BlockStats.CodeFreq.size() <= Code)
861       BlockStats.CodeFreq.resize(Code + 1);
862     BlockStats.CodeFreq[Code].NumInstances++;
863     BlockStats.CodeFreq[Code].TotalBits +=
864         Stream.GetCurrentBitNo() - RecordStartBit;
865     if (Entry.ID != bitc::UNABBREV_RECORD) {
866       BlockStats.CodeFreq[Code].NumAbbrev++;
867       ++BlockStats.NumAbbreviatedRecords;
868     }
869 
870     if (DumpRecords) {
871       O->OS << Indent << "  <";
872       std::optional<const char *> CodeName =
873           GetCodeName(Code, BlockID, BlockInfo, CurStreamType);
874       if (CodeName)
875         O->OS << *CodeName;
876       else
877         O->OS << "UnknownCode" << Code;
878       if (!O->Symbolic && CodeName)
879         O->OS << " codeid=" << Code;
880       const BitCodeAbbrev *Abbv = nullptr;
881       if (Entry.ID != bitc::UNABBREV_RECORD) {
882         Expected<const BitCodeAbbrev *> MaybeAbbv = Stream.getAbbrev(Entry.ID);
883         if (!MaybeAbbv)
884           return MaybeAbbv.takeError();
885         Abbv = MaybeAbbv.get();
886         O->OS << " abbrevid=" << Entry.ID;
887       }
888 
889       for (unsigned i = 0, e = Record.size(); i != e; ++i)
890         O->OS << " op" << i << "=" << (int64_t)Record[i];
891 
892       // If we found a metadata index, let's verify that we had an offset
893       // before and validate its forward reference offset was correct!
894       if (BlockID == bitc::METADATA_BLOCK_ID) {
895         if (Code == bitc::METADATA_INDEX_OFFSET) {
896           if (Record.size() != 2)
897             O->OS << "(Invalid record)";
898           else {
899             auto Offset = Record[0] + (Record[1] << 32);
900             MetadataIndexOffset = Stream.GetCurrentBitNo() + Offset;
901           }
902         }
903         if (Code == bitc::METADATA_INDEX) {
904           O->OS << " (offset ";
905           if (MetadataIndexOffset == RecordStartBit)
906             O->OS << "match)";
907           else
908             O->OS << "mismatch: " << MetadataIndexOffset << " vs "
909                   << RecordStartBit << ")";
910         }
911       }
912 
913       // If we found a module hash, let's verify that it matches!
914       if (BlockID == bitc::MODULE_BLOCK_ID && Code == bitc::MODULE_CODE_HASH &&
915           CheckHash) {
916         if (Record.size() != 5)
917           O->OS << " (invalid)";
918         else {
919           // Recompute the hash and compare it to the one in the bitcode
920           SHA1 Hasher;
921           std::array<uint8_t, 20> Hash;
922           Hasher.update(*CheckHash);
923           {
924             int BlockSize = (CurrentRecordPos / 8) - BlockEntryPos;
925             auto Ptr = Stream.getPointerToByte(BlockEntryPos, BlockSize);
926             Hasher.update(ArrayRef<uint8_t>(Ptr, BlockSize));
927             Hash = Hasher.result();
928           }
929           std::array<uint8_t, 20> RecordedHash;
930           int Pos = 0;
931           for (auto &Val : Record) {
932             assert(!(Val >> 32) && "Unexpected high bits set");
933             support::endian::write32be(&RecordedHash[Pos], Val);
934             Pos += 4;
935           }
936           if (Hash == RecordedHash)
937             O->OS << " (match)";
938           else
939             O->OS << " (!mismatch!)";
940         }
941       }
942 
943       O->OS << "/>";
944 
945       if (Abbv) {
946         for (unsigned i = 1, e = Abbv->getNumOperandInfos(); i != e; ++i) {
947           const BitCodeAbbrevOp &Op = Abbv->getOperandInfo(i);
948           if (!Op.isEncoding() || Op.getEncoding() != BitCodeAbbrevOp::Array)
949             continue;
950           assert(i + 2 == e && "Array op not second to last");
951           std::string Str;
952           bool ArrayIsPrintable = true;
953           for (unsigned j = i - 1, je = Record.size(); j != je; ++j) {
954             if (!isPrint(static_cast<unsigned char>(Record[j]))) {
955               ArrayIsPrintable = false;
956               break;
957             }
958             Str += (char)Record[j];
959           }
960           if (ArrayIsPrintable)
961             O->OS << " record string = '" << Str << "'";
962           break;
963         }
964       }
965 
966       if (Blob.data()) {
967         if (canDecodeBlob(Code, BlockID)) {
968           if (Error E = decodeMetadataStringsBlob(Indent, Record, Blob, O->OS))
969             return E;
970         } else {
971           O->OS << " blob data = ";
972           if (O->ShowBinaryBlobs) {
973             O->OS << "'";
974             O->OS.write_escaped(Blob, /*hex=*/true) << "'";
975           } else {
976             bool BlobIsPrintable = true;
977             for (char C : Blob)
978               if (!isPrint(static_cast<unsigned char>(C))) {
979                 BlobIsPrintable = false;
980                 break;
981               }
982 
983             if (BlobIsPrintable)
984               O->OS << "'" << Blob << "'";
985             else
986               O->OS << "unprintable, " << Blob.size() << " bytes.";
987           }
988         }
989       }
990 
991       O->OS << "\n";
992     }
993 
994     // Make sure that we can skip the current record.
995     if (Error Err = Stream.JumpToBit(CurrentRecordPos))
996       return Err;
997     if (Expected<unsigned> Skipped = Stream.skipRecord(Entry.ID))
998       ; // Do nothing.
999     else
1000       return Skipped.takeError();
1001   }
1002 }
1003 
1004