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 
reportError(StringRef Message)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 *>
GetBlockName(unsigned BlockID,const BitstreamBlockInfo & BlockInfo,CurStreamTypeType CurStreamType)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 *>
GetCodeName(unsigned CodeID,unsigned BlockID,const BitstreamBlockInfo & BlockInfo,CurStreamTypeType CurStreamType)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     case bitc::CST_CODE_BLOCKADDRESS:
226       return "CST_CODE_BLOCKADDRESS";
227       STRINGIFY_CODE(CST_CODE, DATA)
228     }
229   case bitc::FUNCTION_BLOCK_ID:
230     switch (CodeID) {
231     default:
232       return std::nullopt;
233       STRINGIFY_CODE(FUNC_CODE, DECLAREBLOCKS)
234       STRINGIFY_CODE(FUNC_CODE, INST_BINOP)
235       STRINGIFY_CODE(FUNC_CODE, INST_CAST)
236       STRINGIFY_CODE(FUNC_CODE, INST_GEP_OLD)
237       STRINGIFY_CODE(FUNC_CODE, INST_INBOUNDS_GEP_OLD)
238       STRINGIFY_CODE(FUNC_CODE, INST_SELECT)
239       STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTELT)
240       STRINGIFY_CODE(FUNC_CODE, INST_INSERTELT)
241       STRINGIFY_CODE(FUNC_CODE, INST_SHUFFLEVEC)
242       STRINGIFY_CODE(FUNC_CODE, INST_CMP)
243       STRINGIFY_CODE(FUNC_CODE, INST_RET)
244       STRINGIFY_CODE(FUNC_CODE, INST_BR)
245       STRINGIFY_CODE(FUNC_CODE, INST_SWITCH)
246       STRINGIFY_CODE(FUNC_CODE, INST_INVOKE)
247       STRINGIFY_CODE(FUNC_CODE, INST_UNOP)
248       STRINGIFY_CODE(FUNC_CODE, INST_UNREACHABLE)
249       STRINGIFY_CODE(FUNC_CODE, INST_CLEANUPRET)
250       STRINGIFY_CODE(FUNC_CODE, INST_CATCHRET)
251       STRINGIFY_CODE(FUNC_CODE, INST_CATCHPAD)
252       STRINGIFY_CODE(FUNC_CODE, INST_PHI)
253       STRINGIFY_CODE(FUNC_CODE, INST_ALLOCA)
254       STRINGIFY_CODE(FUNC_CODE, INST_LOAD)
255       STRINGIFY_CODE(FUNC_CODE, INST_VAARG)
256       STRINGIFY_CODE(FUNC_CODE, INST_STORE)
257       STRINGIFY_CODE(FUNC_CODE, INST_EXTRACTVAL)
258       STRINGIFY_CODE(FUNC_CODE, INST_INSERTVAL)
259       STRINGIFY_CODE(FUNC_CODE, INST_CMP2)
260       STRINGIFY_CODE(FUNC_CODE, INST_VSELECT)
261       STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC_AGAIN)
262       STRINGIFY_CODE(FUNC_CODE, INST_CALL)
263       STRINGIFY_CODE(FUNC_CODE, DEBUG_LOC)
264       STRINGIFY_CODE(FUNC_CODE, INST_GEP)
265       STRINGIFY_CODE(FUNC_CODE, OPERAND_BUNDLE)
266       STRINGIFY_CODE(FUNC_CODE, INST_FENCE)
267       STRINGIFY_CODE(FUNC_CODE, INST_ATOMICRMW)
268       STRINGIFY_CODE(FUNC_CODE, INST_LOADATOMIC)
269       STRINGIFY_CODE(FUNC_CODE, INST_STOREATOMIC)
270       STRINGIFY_CODE(FUNC_CODE, INST_CMPXCHG)
271       STRINGIFY_CODE(FUNC_CODE, INST_CALLBR)
272       STRINGIFY_CODE(FUNC_CODE, BLOCKADDR_USERS)
273     }
274   case bitc::VALUE_SYMTAB_BLOCK_ID:
275     switch (CodeID) {
276     default:
277       return std::nullopt;
278       STRINGIFY_CODE(VST_CODE, ENTRY)
279       STRINGIFY_CODE(VST_CODE, BBENTRY)
280       STRINGIFY_CODE(VST_CODE, FNENTRY)
281       STRINGIFY_CODE(VST_CODE, COMBINED_ENTRY)
282     }
283   case bitc::MODULE_STRTAB_BLOCK_ID:
284     switch (CodeID) {
285     default:
286       return std::nullopt;
287       STRINGIFY_CODE(MST_CODE, ENTRY)
288       STRINGIFY_CODE(MST_CODE, HASH)
289     }
290   case bitc::GLOBALVAL_SUMMARY_BLOCK_ID:
291   case bitc::FULL_LTO_GLOBALVAL_SUMMARY_BLOCK_ID:
292     switch (CodeID) {
293     default:
294       return std::nullopt;
295       STRINGIFY_CODE(FS, PERMODULE)
296       STRINGIFY_CODE(FS, PERMODULE_PROFILE)
297       STRINGIFY_CODE(FS, PERMODULE_RELBF)
298       STRINGIFY_CODE(FS, PERMODULE_GLOBALVAR_INIT_REFS)
299       STRINGIFY_CODE(FS, PERMODULE_VTABLE_GLOBALVAR_INIT_REFS)
300       STRINGIFY_CODE(FS, COMBINED)
301       STRINGIFY_CODE(FS, COMBINED_PROFILE)
302       STRINGIFY_CODE(FS, COMBINED_GLOBALVAR_INIT_REFS)
303       STRINGIFY_CODE(FS, ALIAS)
304       STRINGIFY_CODE(FS, COMBINED_ALIAS)
305       STRINGIFY_CODE(FS, COMBINED_ORIGINAL_NAME)
306       STRINGIFY_CODE(FS, VERSION)
307       STRINGIFY_CODE(FS, FLAGS)
308       STRINGIFY_CODE(FS, TYPE_TESTS)
309       STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_VCALLS)
310       STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_VCALLS)
311       STRINGIFY_CODE(FS, TYPE_TEST_ASSUME_CONST_VCALL)
312       STRINGIFY_CODE(FS, TYPE_CHECKED_LOAD_CONST_VCALL)
313       STRINGIFY_CODE(FS, VALUE_GUID)
314       STRINGIFY_CODE(FS, CFI_FUNCTION_DEFS)
315       STRINGIFY_CODE(FS, CFI_FUNCTION_DECLS)
316       STRINGIFY_CODE(FS, TYPE_ID)
317       STRINGIFY_CODE(FS, TYPE_ID_METADATA)
318       STRINGIFY_CODE(FS, BLOCK_COUNT)
319       STRINGIFY_CODE(FS, PARAM_ACCESS)
320       STRINGIFY_CODE(FS, PERMODULE_CALLSITE_INFO)
321       STRINGIFY_CODE(FS, PERMODULE_ALLOC_INFO)
322       STRINGIFY_CODE(FS, COMBINED_CALLSITE_INFO)
323       STRINGIFY_CODE(FS, COMBINED_ALLOC_INFO)
324       STRINGIFY_CODE(FS, STACK_IDS)
325     }
326   case bitc::METADATA_ATTACHMENT_ID:
327     switch (CodeID) {
328     default:
329       return std::nullopt;
330       STRINGIFY_CODE(METADATA, ATTACHMENT)
331     }
332   case bitc::METADATA_BLOCK_ID:
333     switch (CodeID) {
334     default:
335       return std::nullopt;
336       STRINGIFY_CODE(METADATA, STRING_OLD)
337       STRINGIFY_CODE(METADATA, VALUE)
338       STRINGIFY_CODE(METADATA, NODE)
339       STRINGIFY_CODE(METADATA, NAME)
340       STRINGIFY_CODE(METADATA, DISTINCT_NODE)
341       STRINGIFY_CODE(METADATA, KIND) // Older bitcode has it in a MODULE_BLOCK
342       STRINGIFY_CODE(METADATA, LOCATION)
343       STRINGIFY_CODE(METADATA, OLD_NODE)
344       STRINGIFY_CODE(METADATA, OLD_FN_NODE)
345       STRINGIFY_CODE(METADATA, NAMED_NODE)
346       STRINGIFY_CODE(METADATA, GENERIC_DEBUG)
347       STRINGIFY_CODE(METADATA, SUBRANGE)
348       STRINGIFY_CODE(METADATA, ENUMERATOR)
349       STRINGIFY_CODE(METADATA, BASIC_TYPE)
350       STRINGIFY_CODE(METADATA, FILE)
351       STRINGIFY_CODE(METADATA, DERIVED_TYPE)
352       STRINGIFY_CODE(METADATA, COMPOSITE_TYPE)
353       STRINGIFY_CODE(METADATA, SUBROUTINE_TYPE)
354       STRINGIFY_CODE(METADATA, COMPILE_UNIT)
355       STRINGIFY_CODE(METADATA, SUBPROGRAM)
356       STRINGIFY_CODE(METADATA, LEXICAL_BLOCK)
357       STRINGIFY_CODE(METADATA, LEXICAL_BLOCK_FILE)
358       STRINGIFY_CODE(METADATA, NAMESPACE)
359       STRINGIFY_CODE(METADATA, TEMPLATE_TYPE)
360       STRINGIFY_CODE(METADATA, TEMPLATE_VALUE)
361       STRINGIFY_CODE(METADATA, GLOBAL_VAR)
362       STRINGIFY_CODE(METADATA, LOCAL_VAR)
363       STRINGIFY_CODE(METADATA, EXPRESSION)
364       STRINGIFY_CODE(METADATA, OBJC_PROPERTY)
365       STRINGIFY_CODE(METADATA, IMPORTED_ENTITY)
366       STRINGIFY_CODE(METADATA, MODULE)
367       STRINGIFY_CODE(METADATA, MACRO)
368       STRINGIFY_CODE(METADATA, MACRO_FILE)
369       STRINGIFY_CODE(METADATA, STRINGS)
370       STRINGIFY_CODE(METADATA, GLOBAL_DECL_ATTACHMENT)
371       STRINGIFY_CODE(METADATA, GLOBAL_VAR_EXPR)
372       STRINGIFY_CODE(METADATA, INDEX_OFFSET)
373       STRINGIFY_CODE(METADATA, INDEX)
374       STRINGIFY_CODE(METADATA, ARG_LIST)
375     }
376   case bitc::METADATA_KIND_BLOCK_ID:
377     switch (CodeID) {
378     default:
379       return std::nullopt;
380       STRINGIFY_CODE(METADATA, KIND)
381     }
382   case bitc::USELIST_BLOCK_ID:
383     switch (CodeID) {
384     default:
385       return std::nullopt;
386     case bitc::USELIST_CODE_DEFAULT:
387       return "USELIST_CODE_DEFAULT";
388     case bitc::USELIST_CODE_BB:
389       return "USELIST_CODE_BB";
390     }
391 
392   case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID:
393     switch (CodeID) {
394     default:
395       return std::nullopt;
396     case bitc::OPERAND_BUNDLE_TAG:
397       return "OPERAND_BUNDLE_TAG";
398     }
399   case bitc::STRTAB_BLOCK_ID:
400     switch (CodeID) {
401     default:
402       return std::nullopt;
403     case bitc::STRTAB_BLOB:
404       return "BLOB";
405     }
406   case bitc::SYMTAB_BLOCK_ID:
407     switch (CodeID) {
408     default:
409       return std::nullopt;
410     case bitc::SYMTAB_BLOB:
411       return "BLOB";
412     }
413   }
414 #undef STRINGIFY_CODE
415 }
416 
printSize(raw_ostream & OS,double Bits)417 static void printSize(raw_ostream &OS, double Bits) {
418   OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32));
419 }
printSize(raw_ostream & OS,uint64_t Bits)420 static void printSize(raw_ostream &OS, uint64_t Bits) {
421   OS << format("%lub/%.2fB/%luW", (unsigned long)Bits, (double)Bits / 8,
422                (unsigned long)(Bits / 32));
423 }
424 
ReadSignature(BitstreamCursor & Stream)425 static Expected<CurStreamTypeType> ReadSignature(BitstreamCursor &Stream) {
426   auto tryRead = [&Stream](char &Dest, size_t size) -> Error {
427     if (Expected<SimpleBitstreamCursor::word_t> MaybeWord = Stream.Read(size))
428       Dest = MaybeWord.get();
429     else
430       return MaybeWord.takeError();
431     return Error::success();
432   };
433 
434   char Signature[6];
435   if (Error Err = tryRead(Signature[0], 8))
436     return std::move(Err);
437   if (Error Err = tryRead(Signature[1], 8))
438     return std::move(Err);
439 
440   // Autodetect the file contents, if it is one we know.
441   if (Signature[0] == 'C' && Signature[1] == 'P') {
442     if (Error Err = tryRead(Signature[2], 8))
443       return std::move(Err);
444     if (Error Err = tryRead(Signature[3], 8))
445       return std::move(Err);
446     if (Signature[2] == 'C' && Signature[3] == 'H')
447       return ClangSerializedASTBitstream;
448   } else if (Signature[0] == 'D' && Signature[1] == 'I') {
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] == 'A' && Signature[3] == 'G')
454       return ClangSerializedDiagnosticsBitstream;
455   } else if (Signature[0] == 'R' && Signature[1] == 'M') {
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] == 'R' && Signature[3] == 'K')
461       return LLVMBitstreamRemarks;
462   } else {
463     if (Error Err = tryRead(Signature[2], 4))
464       return std::move(Err);
465     if (Error Err = tryRead(Signature[3], 4))
466       return std::move(Err);
467     if (Error Err = tryRead(Signature[4], 4))
468       return std::move(Err);
469     if (Error Err = tryRead(Signature[5], 4))
470       return std::move(Err);
471     if (Signature[0] == 'B' && Signature[1] == 'C' && Signature[2] == 0x0 &&
472         Signature[3] == 0xC && Signature[4] == 0xE && Signature[5] == 0xD)
473       return LLVMIRBitstream;
474   }
475   return UnknownBitstream;
476 }
477 
analyzeHeader(std::optional<BCDumpOptions> O,BitstreamCursor & Stream)478 static Expected<CurStreamTypeType> analyzeHeader(std::optional<BCDumpOptions> O,
479                                                  BitstreamCursor &Stream) {
480   ArrayRef<uint8_t> Bytes = Stream.getBitcodeBytes();
481   const unsigned char *BufPtr = (const unsigned char *)Bytes.data();
482   const unsigned char *EndBufPtr = BufPtr + Bytes.size();
483 
484   // If we have a wrapper header, parse it and ignore the non-bc file
485   // contents. The magic number is 0x0B17C0DE stored in little endian.
486   if (isBitcodeWrapper(BufPtr, EndBufPtr)) {
487     if (Bytes.size() < BWH_HeaderSize)
488       return reportError("Invalid bitcode wrapper header");
489 
490     if (O) {
491       unsigned Magic = support::endian::read32le(&BufPtr[BWH_MagicField]);
492       unsigned Version = support::endian::read32le(&BufPtr[BWH_VersionField]);
493       unsigned Offset = support::endian::read32le(&BufPtr[BWH_OffsetField]);
494       unsigned Size = support::endian::read32le(&BufPtr[BWH_SizeField]);
495       unsigned CPUType = support::endian::read32le(&BufPtr[BWH_CPUTypeField]);
496 
497       O->OS << "<BITCODE_WRAPPER_HEADER"
498             << " Magic=" << format_hex(Magic, 10)
499             << " Version=" << format_hex(Version, 10)
500             << " Offset=" << format_hex(Offset, 10)
501             << " Size=" << format_hex(Size, 10)
502             << " CPUType=" << format_hex(CPUType, 10) << "/>\n";
503     }
504 
505     if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true))
506       return reportError("Invalid bitcode wrapper header");
507   }
508 
509   // Use the cursor modified by skipping the wrapper header.
510   Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
511 
512   return ReadSignature(Stream);
513 }
514 
canDecodeBlob(unsigned Code,unsigned BlockID)515 static bool canDecodeBlob(unsigned Code, unsigned BlockID) {
516   return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS;
517 }
518 
decodeMetadataStringsBlob(StringRef Indent,ArrayRef<uint64_t> Record,StringRef Blob,raw_ostream & OS)519 Error BitcodeAnalyzer::decodeMetadataStringsBlob(StringRef Indent,
520                                                  ArrayRef<uint64_t> Record,
521                                                  StringRef Blob,
522                                                  raw_ostream &OS) {
523   if (Blob.empty())
524     return reportError("Cannot decode empty blob.");
525 
526   if (Record.size() != 2)
527     return reportError(
528         "Decoding metadata strings blob needs two record entries.");
529 
530   unsigned NumStrings = Record[0];
531   unsigned StringsOffset = Record[1];
532   OS << " num-strings = " << NumStrings << " {\n";
533 
534   StringRef Lengths = Blob.slice(0, StringsOffset);
535   SimpleBitstreamCursor R(Lengths);
536   StringRef Strings = Blob.drop_front(StringsOffset);
537   do {
538     if (R.AtEndOfStream())
539       return reportError("bad length");
540 
541     uint32_t Size;
542     if (Error E = R.ReadVBR(6).moveInto(Size))
543       return E;
544     if (Strings.size() < Size)
545       return reportError("truncated chars");
546 
547     OS << Indent << "    '";
548     OS.write_escaped(Strings.slice(0, Size), /*hex=*/true);
549     OS << "'\n";
550     Strings = Strings.drop_front(Size);
551   } while (--NumStrings);
552 
553   OS << Indent << "  }";
554   return Error::success();
555 }
556 
BitcodeAnalyzer(StringRef Buffer,std::optional<StringRef> BlockInfoBuffer)557 BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer,
558                                  std::optional<StringRef> BlockInfoBuffer)
559     : Stream(Buffer) {
560   if (BlockInfoBuffer)
561     BlockInfoStream.emplace(*BlockInfoBuffer);
562 }
563 
analyze(std::optional<BCDumpOptions> O,std::optional<StringRef> CheckHash)564 Error BitcodeAnalyzer::analyze(std::optional<BCDumpOptions> O,
565                                std::optional<StringRef> CheckHash) {
566   if (Error E = analyzeHeader(O, Stream).moveInto(CurStreamType))
567     return E;
568 
569   Stream.setBlockInfo(&BlockInfo);
570 
571   // Read block info from BlockInfoStream, if specified.
572   // The block info must be a top-level block.
573   if (BlockInfoStream) {
574     BitstreamCursor BlockInfoCursor(*BlockInfoStream);
575     if (Error E = analyzeHeader(O, BlockInfoCursor).takeError())
576       return E;
577 
578     while (!BlockInfoCursor.AtEndOfStream()) {
579       Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode();
580       if (!MaybeCode)
581         return MaybeCode.takeError();
582       if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
583         return reportError("Invalid record at top-level in block info file");
584 
585       Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID();
586       if (!MaybeBlockID)
587         return MaybeBlockID.takeError();
588       if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) {
589         std::optional<BitstreamBlockInfo> NewBlockInfo;
590         if (Error E =
591                 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
592                     .moveInto(NewBlockInfo))
593           return E;
594         if (!NewBlockInfo)
595           return reportError("Malformed BlockInfoBlock in block info file");
596         BlockInfo = std::move(*NewBlockInfo);
597         break;
598       }
599 
600       if (Error Err = BlockInfoCursor.SkipBlock())
601         return Err;
602     }
603   }
604 
605   // Parse the top-level structure.  We only allow blocks at the top-level.
606   while (!Stream.AtEndOfStream()) {
607     Expected<unsigned> MaybeCode = Stream.ReadCode();
608     if (!MaybeCode)
609       return MaybeCode.takeError();
610     if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
611       return reportError("Invalid record at top-level");
612 
613     Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID();
614     if (!MaybeBlockID)
615       return MaybeBlockID.takeError();
616 
617     if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash))
618       return E;
619     ++NumTopBlocks;
620   }
621 
622   return Error::success();
623 }
624 
printStats(BCDumpOptions O,std::optional<StringRef> Filename)625 void BitcodeAnalyzer::printStats(BCDumpOptions O,
626                                  std::optional<StringRef> Filename) {
627   uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT;
628   // Print a summary of the read file.
629   O.OS << "Summary ";
630   if (Filename)
631     O.OS << "of " << Filename->data() << ":\n";
632   O.OS << "         Total size: ";
633   printSize(O.OS, BufferSizeBits);
634   O.OS << "\n";
635   O.OS << "        Stream type: ";
636   switch (CurStreamType) {
637   case UnknownBitstream:
638     O.OS << "unknown\n";
639     break;
640   case LLVMIRBitstream:
641     O.OS << "LLVM IR\n";
642     break;
643   case ClangSerializedASTBitstream:
644     O.OS << "Clang Serialized AST\n";
645     break;
646   case ClangSerializedDiagnosticsBitstream:
647     O.OS << "Clang Serialized Diagnostics\n";
648     break;
649   case LLVMBitstreamRemarks:
650     O.OS << "LLVM Remarks\n";
651     break;
652   }
653   O.OS << "  # Toplevel Blocks: " << NumTopBlocks << "\n";
654   O.OS << "\n";
655 
656   // Emit per-block stats.
657   O.OS << "Per-block Summary:\n";
658   for (const auto &Stat : BlockIDStats) {
659     O.OS << "  Block ID #" << Stat.first;
660     if (std::optional<const char *> BlockName =
661             GetBlockName(Stat.first, BlockInfo, CurStreamType))
662       O.OS << " (" << *BlockName << ")";
663     O.OS << ":\n";
664 
665     const PerBlockIDStats &Stats = Stat.second;
666     O.OS << "      Num Instances: " << Stats.NumInstances << "\n";
667     O.OS << "         Total Size: ";
668     printSize(O.OS, Stats.NumBits);
669     O.OS << "\n";
670     double pct = (Stats.NumBits * 100.0) / BufferSizeBits;
671     O.OS << "    Percent of file: " << format("%2.4f%%", pct) << "\n";
672     if (Stats.NumInstances > 1) {
673       O.OS << "       Average Size: ";
674       printSize(O.OS, Stats.NumBits / (double)Stats.NumInstances);
675       O.OS << "\n";
676       O.OS << "  Tot/Avg SubBlocks: " << Stats.NumSubBlocks << "/"
677            << Stats.NumSubBlocks / (double)Stats.NumInstances << "\n";
678       O.OS << "    Tot/Avg Abbrevs: " << Stats.NumAbbrevs << "/"
679            << Stats.NumAbbrevs / (double)Stats.NumInstances << "\n";
680       O.OS << "    Tot/Avg Records: " << Stats.NumRecords << "/"
681            << Stats.NumRecords / (double)Stats.NumInstances << "\n";
682     } else {
683       O.OS << "      Num SubBlocks: " << Stats.NumSubBlocks << "\n";
684       O.OS << "        Num Abbrevs: " << Stats.NumAbbrevs << "\n";
685       O.OS << "        Num Records: " << Stats.NumRecords << "\n";
686     }
687     if (Stats.NumRecords) {
688       double pct = (Stats.NumAbbreviatedRecords * 100.0) / Stats.NumRecords;
689       O.OS << "    Percent Abbrevs: " << format("%2.4f%%", pct) << "\n";
690     }
691     O.OS << "\n";
692 
693     // Print a histogram of the codes we see.
694     if (O.Histogram && !Stats.CodeFreq.empty()) {
695       std::vector<std::pair<unsigned, unsigned>> FreqPairs; // <freq,code>
696       for (unsigned i = 0, e = Stats.CodeFreq.size(); i != e; ++i)
697         if (unsigned Freq = Stats.CodeFreq[i].NumInstances)
698           FreqPairs.push_back(std::make_pair(Freq, i));
699       llvm::stable_sort(FreqPairs);
700       std::reverse(FreqPairs.begin(), FreqPairs.end());
701 
702       O.OS << "\tRecord Histogram:\n";
703       O.OS << "\t\t  Count    # Bits     b/Rec   % Abv  Record Kind\n";
704       for (const auto &FreqPair : FreqPairs) {
705         const PerRecordStats &RecStats = Stats.CodeFreq[FreqPair.second];
706 
707         O.OS << format("\t\t%7d %9lu", RecStats.NumInstances,
708                        (unsigned long)RecStats.TotalBits);
709 
710         if (RecStats.NumInstances > 1)
711           O.OS << format(" %9.1f",
712                          (double)RecStats.TotalBits / RecStats.NumInstances);
713         else
714           O.OS << "          ";
715 
716         if (RecStats.NumAbbrev)
717           O.OS << format(" %7.2f", (double)RecStats.NumAbbrev /
718                                        RecStats.NumInstances * 100);
719         else
720           O.OS << "        ";
721 
722         O.OS << "  ";
723         if (std::optional<const char *> CodeName = GetCodeName(
724                 FreqPair.second, Stat.first, BlockInfo, CurStreamType))
725           O.OS << *CodeName << "\n";
726         else
727           O.OS << "UnknownCode" << FreqPair.second << "\n";
728       }
729       O.OS << "\n";
730     }
731   }
732 }
733 
parseBlock(unsigned BlockID,unsigned IndentLevel,std::optional<BCDumpOptions> O,std::optional<StringRef> CheckHash)734 Error BitcodeAnalyzer::parseBlock(unsigned BlockID, unsigned IndentLevel,
735                                   std::optional<BCDumpOptions> O,
736                                   std::optional<StringRef> CheckHash) {
737   std::string Indent(IndentLevel * 2, ' ');
738   uint64_t BlockBitStart = Stream.GetCurrentBitNo();
739 
740   // Get the statistics for this BlockID.
741   PerBlockIDStats &BlockStats = BlockIDStats[BlockID];
742 
743   BlockStats.NumInstances++;
744 
745   // BLOCKINFO is a special part of the stream.
746   bool DumpRecords = O.has_value();
747   if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
748     if (O && !O->DumpBlockinfo)
749       O->OS << Indent << "<BLOCKINFO_BLOCK/>\n";
750     std::optional<BitstreamBlockInfo> NewBlockInfo;
751     if (Error E = Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
752                       .moveInto(NewBlockInfo))
753       return E;
754     if (!NewBlockInfo)
755       return reportError("Malformed BlockInfoBlock");
756     BlockInfo = std::move(*NewBlockInfo);
757     if (Error Err = Stream.JumpToBit(BlockBitStart))
758       return Err;
759     // It's not really interesting to dump the contents of the blockinfo
760     // block, so only do it if the user explicitly requests it.
761     DumpRecords = O && O->DumpBlockinfo;
762   }
763 
764   unsigned NumWords = 0;
765   if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords))
766     return Err;
767 
768   // Keep it for later, when we see a MODULE_HASH record
769   uint64_t BlockEntryPos = Stream.getCurrentByteNo();
770 
771   std::optional<const char *> BlockName;
772   if (DumpRecords) {
773     O->OS << Indent << "<";
774     if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType)))
775       O->OS << *BlockName;
776     else
777       O->OS << "UnknownBlock" << BlockID;
778 
779     if (!O->Symbolic && BlockName)
780       O->OS << " BlockID=" << BlockID;
781 
782     O->OS << " NumWords=" << NumWords
783           << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n";
784   }
785 
786   SmallVector<uint64_t, 64> Record;
787 
788   // Keep the offset to the metadata index if seen.
789   uint64_t MetadataIndexOffset = 0;
790 
791   // Read all the records for this block.
792   while (true) {
793     if (Stream.AtEndOfStream())
794       return reportError("Premature end of bitstream");
795 
796     uint64_t RecordStartBit = Stream.GetCurrentBitNo();
797 
798     BitstreamEntry Entry;
799     if (Error E = Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs)
800                       .moveInto(Entry))
801       return E;
802 
803     switch (Entry.Kind) {
804     case BitstreamEntry::Error:
805       return reportError("malformed bitcode file");
806     case BitstreamEntry::EndBlock: {
807       uint64_t BlockBitEnd = Stream.GetCurrentBitNo();
808       BlockStats.NumBits += BlockBitEnd - BlockBitStart;
809       if (DumpRecords) {
810         O->OS << Indent << "</";
811         if (BlockName)
812           O->OS << *BlockName << ">\n";
813         else
814           O->OS << "UnknownBlock" << BlockID << ">\n";
815       }
816       return Error::success();
817     }
818 
819     case BitstreamEntry::SubBlock: {
820       uint64_t SubBlockBitStart = Stream.GetCurrentBitNo();
821       if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash))
822         return E;
823       ++BlockStats.NumSubBlocks;
824       uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo();
825 
826       // Don't include subblock sizes in the size of this block.
827       BlockBitStart += SubBlockBitEnd - SubBlockBitStart;
828       continue;
829     }
830     case BitstreamEntry::Record:
831       // The interesting case.
832       break;
833     }
834 
835     if (Entry.ID == bitc::DEFINE_ABBREV) {
836       if (Error Err = Stream.ReadAbbrevRecord())
837         return Err;
838       ++BlockStats.NumAbbrevs;
839       continue;
840     }
841 
842     Record.clear();
843 
844     ++BlockStats.NumRecords;
845 
846     StringRef Blob;
847     uint64_t CurrentRecordPos = Stream.GetCurrentBitNo();
848     unsigned Code;
849     if (Error E = Stream.readRecord(Entry.ID, Record, &Blob).moveInto(Code))
850       return E;
851 
852     // Increment the # occurrences of this code.
853     if (BlockStats.CodeFreq.size() <= Code)
854       BlockStats.CodeFreq.resize(Code + 1);
855     BlockStats.CodeFreq[Code].NumInstances++;
856     BlockStats.CodeFreq[Code].TotalBits +=
857         Stream.GetCurrentBitNo() - RecordStartBit;
858     if (Entry.ID != bitc::UNABBREV_RECORD) {
859       BlockStats.CodeFreq[Code].NumAbbrev++;
860       ++BlockStats.NumAbbreviatedRecords;
861     }
862 
863     if (DumpRecords) {
864       O->OS << Indent << "  <";
865       std::optional<const char *> CodeName =
866           GetCodeName(Code, BlockID, BlockInfo, CurStreamType);
867       if (CodeName)
868         O->OS << *CodeName;
869       else
870         O->OS << "UnknownCode" << Code;
871       if (!O->Symbolic && CodeName)
872         O->OS << " codeid=" << Code;
873       const BitCodeAbbrev *Abbv = nullptr;
874       if (Entry.ID != bitc::UNABBREV_RECORD) {
875         Expected<const BitCodeAbbrev *> MaybeAbbv = Stream.getAbbrev(Entry.ID);
876         if (!MaybeAbbv)
877           return MaybeAbbv.takeError();
878         Abbv = MaybeAbbv.get();
879         O->OS << " abbrevid=" << Entry.ID;
880       }
881 
882       for (unsigned i = 0, e = Record.size(); i != e; ++i)
883         O->OS << " op" << i << "=" << (int64_t)Record[i];
884 
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;
894           }
895         }
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 << ")";
903         }
904       }
905 
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) {
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           std::array<uint8_t, 20> Hash;
915           Hasher.update(*CheckHash);
916           {
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();
921           }
922           std::array<uint8_t, 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;
928           }
929           if (Hash == RecordedHash)
930             O->OS << " (match)";
931           else
932             O->OS << " (!mismatch!)";
933         }
934       }
935 
936       O->OS << "/>";
937 
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;
950             }
951             Str += (char)Record[j];
952           }
953           if (ArrayIsPrintable)
954             O->OS << " record string = '" << Str << "'";
955           break;
956         }
957       }
958 
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 (char C : Blob)
971               if (!isPrint(static_cast<unsigned char>(C))) {
972                 BlobIsPrintable = false;
973                 break;
974               }
975 
976             if (BlobIsPrintable)
977               O->OS << "'" << Blob << "'";
978             else
979               O->OS << "unprintable, " << Blob.size() << " bytes.";
980           }
981         }
982       }
983 
984       O->OS << "\n";
985     }
986 
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();
994   }
995 }
996 
997