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 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 }
332 case bitc::METADATA_ATTACHMENT_ID:
333 switch (CodeID) {
334 default:
335 return std::nullopt;
336 STRINGIFY_CODE(METADATA, ATTACHMENT)
337 }
338 case bitc::METADATA_BLOCK_ID:
339 switch (CodeID) {
340 default:
341 return std::nullopt;
342 STRINGIFY_CODE(METADATA, STRING_OLD)
343 STRINGIFY_CODE(METADATA, VALUE)
344 STRINGIFY_CODE(METADATA, NODE)
345 STRINGIFY_CODE(METADATA, NAME)
346 STRINGIFY_CODE(METADATA, DISTINCT_NODE)
347 STRINGIFY_CODE(METADATA, KIND) // Older bitcode has it in a MODULE_BLOCK
348 STRINGIFY_CODE(METADATA, LOCATION)
349 STRINGIFY_CODE(METADATA, OLD_NODE)
350 STRINGIFY_CODE(METADATA, OLD_FN_NODE)
351 STRINGIFY_CODE(METADATA, NAMED_NODE)
352 STRINGIFY_CODE(METADATA, GENERIC_DEBUG)
353 STRINGIFY_CODE(METADATA, SUBRANGE)
354 STRINGIFY_CODE(METADATA, ENUMERATOR)
355 STRINGIFY_CODE(METADATA, BASIC_TYPE)
356 STRINGIFY_CODE(METADATA, FILE)
357 STRINGIFY_CODE(METADATA, DERIVED_TYPE)
358 STRINGIFY_CODE(METADATA, COMPOSITE_TYPE)
359 STRINGIFY_CODE(METADATA, SUBROUTINE_TYPE)
360 STRINGIFY_CODE(METADATA, COMPILE_UNIT)
361 STRINGIFY_CODE(METADATA, SUBPROGRAM)
362 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK)
363 STRINGIFY_CODE(METADATA, LEXICAL_BLOCK_FILE)
364 STRINGIFY_CODE(METADATA, NAMESPACE)
365 STRINGIFY_CODE(METADATA, TEMPLATE_TYPE)
366 STRINGIFY_CODE(METADATA, TEMPLATE_VALUE)
367 STRINGIFY_CODE(METADATA, GLOBAL_VAR)
368 STRINGIFY_CODE(METADATA, LOCAL_VAR)
369 STRINGIFY_CODE(METADATA, EXPRESSION)
370 STRINGIFY_CODE(METADATA, OBJC_PROPERTY)
371 STRINGIFY_CODE(METADATA, IMPORTED_ENTITY)
372 STRINGIFY_CODE(METADATA, MODULE)
373 STRINGIFY_CODE(METADATA, MACRO)
374 STRINGIFY_CODE(METADATA, MACRO_FILE)
375 STRINGIFY_CODE(METADATA, STRINGS)
376 STRINGIFY_CODE(METADATA, GLOBAL_DECL_ATTACHMENT)
377 STRINGIFY_CODE(METADATA, GLOBAL_VAR_EXPR)
378 STRINGIFY_CODE(METADATA, INDEX_OFFSET)
379 STRINGIFY_CODE(METADATA, INDEX)
380 STRINGIFY_CODE(METADATA, ARG_LIST)
381 }
382 case bitc::METADATA_KIND_BLOCK_ID:
383 switch (CodeID) {
384 default:
385 return std::nullopt;
386 STRINGIFY_CODE(METADATA, KIND)
387 }
388 case bitc::USELIST_BLOCK_ID:
389 switch (CodeID) {
390 default:
391 return std::nullopt;
392 case bitc::USELIST_CODE_DEFAULT:
393 return "USELIST_CODE_DEFAULT";
394 case bitc::USELIST_CODE_BB:
395 return "USELIST_CODE_BB";
396 }
397
398 case bitc::OPERAND_BUNDLE_TAGS_BLOCK_ID:
399 switch (CodeID) {
400 default:
401 return std::nullopt;
402 case bitc::OPERAND_BUNDLE_TAG:
403 return "OPERAND_BUNDLE_TAG";
404 }
405 case bitc::STRTAB_BLOCK_ID:
406 switch (CodeID) {
407 default:
408 return std::nullopt;
409 case bitc::STRTAB_BLOB:
410 return "BLOB";
411 }
412 case bitc::SYMTAB_BLOCK_ID:
413 switch (CodeID) {
414 default:
415 return std::nullopt;
416 case bitc::SYMTAB_BLOB:
417 return "BLOB";
418 }
419 }
420 #undef STRINGIFY_CODE
421 }
422
printSize(raw_ostream & OS,double Bits)423 static void printSize(raw_ostream &OS, double Bits) {
424 OS << format("%.2f/%.2fB/%luW", Bits, Bits / 8, (unsigned long)(Bits / 32));
425 }
printSize(raw_ostream & OS,uint64_t Bits)426 static void printSize(raw_ostream &OS, uint64_t Bits) {
427 OS << format("%lub/%.2fB/%luW", (unsigned long)Bits, (double)Bits / 8,
428 (unsigned long)(Bits / 32));
429 }
430
ReadSignature(BitstreamCursor & Stream)431 static Expected<CurStreamTypeType> ReadSignature(BitstreamCursor &Stream) {
432 auto tryRead = [&Stream](char &Dest, size_t size) -> Error {
433 if (Expected<SimpleBitstreamCursor::word_t> MaybeWord = Stream.Read(size))
434 Dest = MaybeWord.get();
435 else
436 return MaybeWord.takeError();
437 return Error::success();
438 };
439
440 char Signature[6];
441 if (Error Err = tryRead(Signature[0], 8))
442 return std::move(Err);
443 if (Error Err = tryRead(Signature[1], 8))
444 return std::move(Err);
445
446 // Autodetect the file contents, if it is one we know.
447 if (Signature[0] == 'C' && Signature[1] == 'P') {
448 if (Error Err = tryRead(Signature[2], 8))
449 return std::move(Err);
450 if (Error Err = tryRead(Signature[3], 8))
451 return std::move(Err);
452 if (Signature[2] == 'C' && Signature[3] == 'H')
453 return ClangSerializedASTBitstream;
454 } else if (Signature[0] == 'D' && Signature[1] == 'I') {
455 if (Error Err = tryRead(Signature[2], 8))
456 return std::move(Err);
457 if (Error Err = tryRead(Signature[3], 8))
458 return std::move(Err);
459 if (Signature[2] == 'A' && Signature[3] == 'G')
460 return ClangSerializedDiagnosticsBitstream;
461 } else if (Signature[0] == 'R' && Signature[1] == 'M') {
462 if (Error Err = tryRead(Signature[2], 8))
463 return std::move(Err);
464 if (Error Err = tryRead(Signature[3], 8))
465 return std::move(Err);
466 if (Signature[2] == 'R' && Signature[3] == 'K')
467 return LLVMBitstreamRemarks;
468 } else {
469 if (Error Err = tryRead(Signature[2], 4))
470 return std::move(Err);
471 if (Error Err = tryRead(Signature[3], 4))
472 return std::move(Err);
473 if (Error Err = tryRead(Signature[4], 4))
474 return std::move(Err);
475 if (Error Err = tryRead(Signature[5], 4))
476 return std::move(Err);
477 if (Signature[0] == 'B' && Signature[1] == 'C' && Signature[2] == 0x0 &&
478 Signature[3] == 0xC && Signature[4] == 0xE && Signature[5] == 0xD)
479 return LLVMIRBitstream;
480 }
481 return UnknownBitstream;
482 }
483
analyzeHeader(std::optional<BCDumpOptions> O,BitstreamCursor & Stream)484 static Expected<CurStreamTypeType> analyzeHeader(std::optional<BCDumpOptions> O,
485 BitstreamCursor &Stream) {
486 ArrayRef<uint8_t> Bytes = Stream.getBitcodeBytes();
487 const unsigned char *BufPtr = (const unsigned char *)Bytes.data();
488 const unsigned char *EndBufPtr = BufPtr + Bytes.size();
489
490 // If we have a wrapper header, parse it and ignore the non-bc file
491 // contents. The magic number is 0x0B17C0DE stored in little endian.
492 if (isBitcodeWrapper(BufPtr, EndBufPtr)) {
493 if (Bytes.size() < BWH_HeaderSize)
494 return reportError("Invalid bitcode wrapper header");
495
496 if (O) {
497 unsigned Magic = support::endian::read32le(&BufPtr[BWH_MagicField]);
498 unsigned Version = support::endian::read32le(&BufPtr[BWH_VersionField]);
499 unsigned Offset = support::endian::read32le(&BufPtr[BWH_OffsetField]);
500 unsigned Size = support::endian::read32le(&BufPtr[BWH_SizeField]);
501 unsigned CPUType = support::endian::read32le(&BufPtr[BWH_CPUTypeField]);
502
503 O->OS << "<BITCODE_WRAPPER_HEADER"
504 << " Magic=" << format_hex(Magic, 10)
505 << " Version=" << format_hex(Version, 10)
506 << " Offset=" << format_hex(Offset, 10)
507 << " Size=" << format_hex(Size, 10)
508 << " CPUType=" << format_hex(CPUType, 10) << "/>\n";
509 }
510
511 if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true))
512 return reportError("Invalid bitcode wrapper header");
513 }
514
515 // Use the cursor modified by skipping the wrapper header.
516 Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
517
518 return ReadSignature(Stream);
519 }
520
canDecodeBlob(unsigned Code,unsigned BlockID)521 static bool canDecodeBlob(unsigned Code, unsigned BlockID) {
522 return BlockID == bitc::METADATA_BLOCK_ID && Code == bitc::METADATA_STRINGS;
523 }
524
decodeMetadataStringsBlob(StringRef Indent,ArrayRef<uint64_t> Record,StringRef Blob,raw_ostream & OS)525 Error BitcodeAnalyzer::decodeMetadataStringsBlob(StringRef Indent,
526 ArrayRef<uint64_t> Record,
527 StringRef Blob,
528 raw_ostream &OS) {
529 if (Blob.empty())
530 return reportError("Cannot decode empty blob.");
531
532 if (Record.size() != 2)
533 return reportError(
534 "Decoding metadata strings blob needs two record entries.");
535
536 unsigned NumStrings = Record[0];
537 unsigned StringsOffset = Record[1];
538 OS << " num-strings = " << NumStrings << " {\n";
539
540 StringRef Lengths = Blob.slice(0, StringsOffset);
541 SimpleBitstreamCursor R(Lengths);
542 StringRef Strings = Blob.drop_front(StringsOffset);
543 do {
544 if (R.AtEndOfStream())
545 return reportError("bad length");
546
547 uint32_t Size;
548 if (Error E = R.ReadVBR(6).moveInto(Size))
549 return E;
550 if (Strings.size() < Size)
551 return reportError("truncated chars");
552
553 OS << Indent << " '";
554 OS.write_escaped(Strings.slice(0, Size), /*hex=*/true);
555 OS << "'\n";
556 Strings = Strings.drop_front(Size);
557 } while (--NumStrings);
558
559 OS << Indent << " }";
560 return Error::success();
561 }
562
BitcodeAnalyzer(StringRef Buffer,std::optional<StringRef> BlockInfoBuffer)563 BitcodeAnalyzer::BitcodeAnalyzer(StringRef Buffer,
564 std::optional<StringRef> BlockInfoBuffer)
565 : Stream(Buffer) {
566 if (BlockInfoBuffer)
567 BlockInfoStream.emplace(*BlockInfoBuffer);
568 }
569
analyze(std::optional<BCDumpOptions> O,std::optional<StringRef> CheckHash)570 Error BitcodeAnalyzer::analyze(std::optional<BCDumpOptions> O,
571 std::optional<StringRef> CheckHash) {
572 if (Error E = analyzeHeader(O, Stream).moveInto(CurStreamType))
573 return E;
574
575 Stream.setBlockInfo(&BlockInfo);
576
577 // Read block info from BlockInfoStream, if specified.
578 // The block info must be a top-level block.
579 if (BlockInfoStream) {
580 BitstreamCursor BlockInfoCursor(*BlockInfoStream);
581 if (Error E = analyzeHeader(O, BlockInfoCursor).takeError())
582 return E;
583
584 while (!BlockInfoCursor.AtEndOfStream()) {
585 Expected<unsigned> MaybeCode = BlockInfoCursor.ReadCode();
586 if (!MaybeCode)
587 return MaybeCode.takeError();
588 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
589 return reportError("Invalid record at top-level in block info file");
590
591 Expected<unsigned> MaybeBlockID = BlockInfoCursor.ReadSubBlockID();
592 if (!MaybeBlockID)
593 return MaybeBlockID.takeError();
594 if (MaybeBlockID.get() == bitc::BLOCKINFO_BLOCK_ID) {
595 std::optional<BitstreamBlockInfo> NewBlockInfo;
596 if (Error E =
597 BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
598 .moveInto(NewBlockInfo))
599 return E;
600 if (!NewBlockInfo)
601 return reportError("Malformed BlockInfoBlock in block info file");
602 BlockInfo = std::move(*NewBlockInfo);
603 break;
604 }
605
606 if (Error Err = BlockInfoCursor.SkipBlock())
607 return Err;
608 }
609 }
610
611 // Parse the top-level structure. We only allow blocks at the top-level.
612 while (!Stream.AtEndOfStream()) {
613 Expected<unsigned> MaybeCode = Stream.ReadCode();
614 if (!MaybeCode)
615 return MaybeCode.takeError();
616 if (MaybeCode.get() != bitc::ENTER_SUBBLOCK)
617 return reportError("Invalid record at top-level");
618
619 Expected<unsigned> MaybeBlockID = Stream.ReadSubBlockID();
620 if (!MaybeBlockID)
621 return MaybeBlockID.takeError();
622
623 if (Error E = parseBlock(MaybeBlockID.get(), 0, O, CheckHash))
624 return E;
625 ++NumTopBlocks;
626 }
627
628 return Error::success();
629 }
630
printStats(BCDumpOptions O,std::optional<StringRef> Filename)631 void BitcodeAnalyzer::printStats(BCDumpOptions O,
632 std::optional<StringRef> Filename) {
633 uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT;
634 // Print a summary of the read file.
635 O.OS << "Summary ";
636 if (Filename)
637 O.OS << "of " << Filename->data() << ":\n";
638 O.OS << " Total size: ";
639 printSize(O.OS, BufferSizeBits);
640 O.OS << "\n";
641 O.OS << " Stream type: ";
642 switch (CurStreamType) {
643 case UnknownBitstream:
644 O.OS << "unknown\n";
645 break;
646 case LLVMIRBitstream:
647 O.OS << "LLVM IR\n";
648 break;
649 case ClangSerializedASTBitstream:
650 O.OS << "Clang Serialized AST\n";
651 break;
652 case ClangSerializedDiagnosticsBitstream:
653 O.OS << "Clang Serialized Diagnostics\n";
654 break;
655 case LLVMBitstreamRemarks:
656 O.OS << "LLVM Remarks\n";
657 break;
658 }
659 O.OS << " # Toplevel Blocks: " << NumTopBlocks << "\n";
660 O.OS << "\n";
661
662 // Emit per-block stats.
663 O.OS << "Per-block Summary:\n";
664 for (const auto &Stat : BlockIDStats) {
665 O.OS << " Block ID #" << Stat.first;
666 if (std::optional<const char *> BlockName =
667 GetBlockName(Stat.first, BlockInfo, CurStreamType))
668 O.OS << " (" << *BlockName << ")";
669 O.OS << ":\n";
670
671 const PerBlockIDStats &Stats = Stat.second;
672 O.OS << " Num Instances: " << Stats.NumInstances << "\n";
673 O.OS << " Total Size: ";
674 printSize(O.OS, Stats.NumBits);
675 O.OS << "\n";
676 double pct = (Stats.NumBits * 100.0) / BufferSizeBits;
677 O.OS << " Percent of file: " << format("%2.4f%%", pct) << "\n";
678 if (Stats.NumInstances > 1) {
679 O.OS << " Average Size: ";
680 printSize(O.OS, Stats.NumBits / (double)Stats.NumInstances);
681 O.OS << "\n";
682 O.OS << " Tot/Avg SubBlocks: " << Stats.NumSubBlocks << "/"
683 << Stats.NumSubBlocks / (double)Stats.NumInstances << "\n";
684 O.OS << " Tot/Avg Abbrevs: " << Stats.NumAbbrevs << "/"
685 << Stats.NumAbbrevs / (double)Stats.NumInstances << "\n";
686 O.OS << " Tot/Avg Records: " << Stats.NumRecords << "/"
687 << Stats.NumRecords / (double)Stats.NumInstances << "\n";
688 } else {
689 O.OS << " Num SubBlocks: " << Stats.NumSubBlocks << "\n";
690 O.OS << " Num Abbrevs: " << Stats.NumAbbrevs << "\n";
691 O.OS << " Num Records: " << Stats.NumRecords << "\n";
692 }
693 if (Stats.NumRecords) {
694 double pct = (Stats.NumAbbreviatedRecords * 100.0) / Stats.NumRecords;
695 O.OS << " Percent Abbrevs: " << format("%2.4f%%", pct) << "\n";
696 }
697 O.OS << "\n";
698
699 // Print a histogram of the codes we see.
700 if (O.Histogram && !Stats.CodeFreq.empty()) {
701 std::vector<std::pair<unsigned, unsigned>> FreqPairs; // <freq,code>
702 for (unsigned i = 0, e = Stats.CodeFreq.size(); i != e; ++i)
703 if (unsigned Freq = Stats.CodeFreq[i].NumInstances)
704 FreqPairs.push_back(std::make_pair(Freq, i));
705 llvm::stable_sort(FreqPairs);
706 std::reverse(FreqPairs.begin(), FreqPairs.end());
707
708 O.OS << "\tRecord Histogram:\n";
709 O.OS << "\t\t Count # Bits b/Rec % Abv Record Kind\n";
710 for (const auto &FreqPair : FreqPairs) {
711 const PerRecordStats &RecStats = Stats.CodeFreq[FreqPair.second];
712
713 O.OS << format("\t\t%7d %9lu", RecStats.NumInstances,
714 (unsigned long)RecStats.TotalBits);
715
716 if (RecStats.NumInstances > 1)
717 O.OS << format(" %9.1f",
718 (double)RecStats.TotalBits / RecStats.NumInstances);
719 else
720 O.OS << " ";
721
722 if (RecStats.NumAbbrev)
723 O.OS << format(" %7.2f", (double)RecStats.NumAbbrev /
724 RecStats.NumInstances * 100);
725 else
726 O.OS << " ";
727
728 O.OS << " ";
729 if (std::optional<const char *> CodeName = GetCodeName(
730 FreqPair.second, Stat.first, BlockInfo, CurStreamType))
731 O.OS << *CodeName << "\n";
732 else
733 O.OS << "UnknownCode" << FreqPair.second << "\n";
734 }
735 O.OS << "\n";
736 }
737 }
738 }
739
parseBlock(unsigned BlockID,unsigned IndentLevel,std::optional<BCDumpOptions> O,std::optional<StringRef> CheckHash)740 Error BitcodeAnalyzer::parseBlock(unsigned BlockID, unsigned IndentLevel,
741 std::optional<BCDumpOptions> O,
742 std::optional<StringRef> CheckHash) {
743 std::string Indent(IndentLevel * 2, ' ');
744 uint64_t BlockBitStart = Stream.GetCurrentBitNo();
745
746 // Get the statistics for this BlockID.
747 PerBlockIDStats &BlockStats = BlockIDStats[BlockID];
748
749 BlockStats.NumInstances++;
750
751 // BLOCKINFO is a special part of the stream.
752 bool DumpRecords = O.has_value();
753 if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
754 if (O && !O->DumpBlockinfo)
755 O->OS << Indent << "<BLOCKINFO_BLOCK/>\n";
756 std::optional<BitstreamBlockInfo> NewBlockInfo;
757 if (Error E = Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true)
758 .moveInto(NewBlockInfo))
759 return E;
760 if (!NewBlockInfo)
761 return reportError("Malformed BlockInfoBlock");
762 BlockInfo = std::move(*NewBlockInfo);
763 if (Error Err = Stream.JumpToBit(BlockBitStart))
764 return Err;
765 // It's not really interesting to dump the contents of the blockinfo
766 // block, so only do it if the user explicitly requests it.
767 DumpRecords = O && O->DumpBlockinfo;
768 }
769
770 unsigned NumWords = 0;
771 if (Error Err = Stream.EnterSubBlock(BlockID, &NumWords))
772 return Err;
773
774 // Keep it for later, when we see a MODULE_HASH record
775 uint64_t BlockEntryPos = Stream.getCurrentByteNo();
776
777 std::optional<const char *> BlockName;
778 if (DumpRecords) {
779 O->OS << Indent << "<";
780 if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType)))
781 O->OS << *BlockName;
782 else
783 O->OS << "UnknownBlock" << BlockID;
784
785 if (!O->Symbolic && BlockName)
786 O->OS << " BlockID=" << BlockID;
787
788 O->OS << " NumWords=" << NumWords
789 << " BlockCodeSize=" << Stream.getAbbrevIDWidth() << ">\n";
790 }
791
792 SmallVector<uint64_t, 64> Record;
793
794 // Keep the offset to the metadata index if seen.
795 uint64_t MetadataIndexOffset = 0;
796
797 // Read all the records for this block.
798 while (true) {
799 if (Stream.AtEndOfStream())
800 return reportError("Premature end of bitstream");
801
802 uint64_t RecordStartBit = Stream.GetCurrentBitNo();
803
804 BitstreamEntry Entry;
805 if (Error E = Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs)
806 .moveInto(Entry))
807 return E;
808
809 switch (Entry.Kind) {
810 case BitstreamEntry::Error:
811 return reportError("malformed bitcode file");
812 case BitstreamEntry::EndBlock: {
813 uint64_t BlockBitEnd = Stream.GetCurrentBitNo();
814 BlockStats.NumBits += BlockBitEnd - BlockBitStart;
815 if (DumpRecords) {
816 O->OS << Indent << "</";
817 if (BlockName)
818 O->OS << *BlockName << ">\n";
819 else
820 O->OS << "UnknownBlock" << BlockID << ">\n";
821 }
822 return Error::success();
823 }
824
825 case BitstreamEntry::SubBlock: {
826 uint64_t SubBlockBitStart = Stream.GetCurrentBitNo();
827 if (Error E = parseBlock(Entry.ID, IndentLevel + 1, O, CheckHash))
828 return E;
829 ++BlockStats.NumSubBlocks;
830 uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo();
831
832 // Don't include subblock sizes in the size of this block.
833 BlockBitStart += SubBlockBitEnd - SubBlockBitStart;
834 continue;
835 }
836 case BitstreamEntry::Record:
837 // The interesting case.
838 break;
839 }
840
841 if (Entry.ID == bitc::DEFINE_ABBREV) {
842 if (Error Err = Stream.ReadAbbrevRecord())
843 return Err;
844 ++BlockStats.NumAbbrevs;
845 continue;
846 }
847
848 Record.clear();
849
850 ++BlockStats.NumRecords;
851
852 StringRef Blob;
853 uint64_t CurrentRecordPos = Stream.GetCurrentBitNo();
854 unsigned Code;
855 if (Error E = Stream.readRecord(Entry.ID, Record, &Blob).moveInto(Code))
856 return E;
857
858 // Increment the # occurrences of this code.
859 if (BlockStats.CodeFreq.size() <= Code)
860 BlockStats.CodeFreq.resize(Code + 1);
861 BlockStats.CodeFreq[Code].NumInstances++;
862 BlockStats.CodeFreq[Code].TotalBits +=
863 Stream.GetCurrentBitNo() - RecordStartBit;
864 if (Entry.ID != bitc::UNABBREV_RECORD) {
865 BlockStats.CodeFreq[Code].NumAbbrev++;
866 ++BlockStats.NumAbbreviatedRecords;
867 }
868
869 if (DumpRecords) {
870 O->OS << Indent << " <";
871 std::optional<const char *> CodeName =
872 GetCodeName(Code, BlockID, BlockInfo, CurStreamType);
873 if (CodeName)
874 O->OS << *CodeName;
875 else
876 O->OS << "UnknownCode" << Code;
877 if (!O->Symbolic && CodeName)
878 O->OS << " codeid=" << Code;
879 const BitCodeAbbrev *Abbv = nullptr;
880 if (Entry.ID != bitc::UNABBREV_RECORD) {
881 Expected<const BitCodeAbbrev *> MaybeAbbv = Stream.getAbbrev(Entry.ID);
882 if (!MaybeAbbv)
883 return MaybeAbbv.takeError();
884 Abbv = MaybeAbbv.get();
885 O->OS << " abbrevid=" << Entry.ID;
886 }
887
888 for (unsigned i = 0, e = Record.size(); i != e; ++i)
889 O->OS << " op" << i << "=" << (int64_t)Record[i];
890
891 // If we found a metadata index, let's verify that we had an offset
892 // before and validate its forward reference offset was correct!
893 if (BlockID == bitc::METADATA_BLOCK_ID) {
894 if (Code == bitc::METADATA_INDEX_OFFSET) {
895 if (Record.size() != 2)
896 O->OS << "(Invalid record)";
897 else {
898 auto Offset = Record[0] + (Record[1] << 32);
899 MetadataIndexOffset = Stream.GetCurrentBitNo() + Offset;
900 }
901 }
902 if (Code == bitc::METADATA_INDEX) {
903 O->OS << " (offset ";
904 if (MetadataIndexOffset == RecordStartBit)
905 O->OS << "match)";
906 else
907 O->OS << "mismatch: " << MetadataIndexOffset << " vs "
908 << RecordStartBit << ")";
909 }
910 }
911
912 // If we found a module hash, let's verify that it matches!
913 if (BlockID == bitc::MODULE_BLOCK_ID && Code == bitc::MODULE_CODE_HASH &&
914 CheckHash) {
915 if (Record.size() != 5)
916 O->OS << " (invalid)";
917 else {
918 // Recompute the hash and compare it to the one in the bitcode
919 SHA1 Hasher;
920 std::array<uint8_t, 20> Hash;
921 Hasher.update(*CheckHash);
922 {
923 int BlockSize = (CurrentRecordPos / 8) - BlockEntryPos;
924 auto Ptr = Stream.getPointerToByte(BlockEntryPos, BlockSize);
925 Hasher.update(ArrayRef<uint8_t>(Ptr, BlockSize));
926 Hash = Hasher.result();
927 }
928 std::array<uint8_t, 20> RecordedHash;
929 int Pos = 0;
930 for (auto &Val : Record) {
931 assert(!(Val >> 32) && "Unexpected high bits set");
932 support::endian::write32be(&RecordedHash[Pos], Val);
933 Pos += 4;
934 }
935 if (Hash == RecordedHash)
936 O->OS << " (match)";
937 else
938 O->OS << " (!mismatch!)";
939 }
940 }
941
942 O->OS << "/>";
943
944 if (Abbv) {
945 for (unsigned i = 1, e = Abbv->getNumOperandInfos(); i != e; ++i) {
946 const BitCodeAbbrevOp &Op = Abbv->getOperandInfo(i);
947 if (!Op.isEncoding() || Op.getEncoding() != BitCodeAbbrevOp::Array)
948 continue;
949 assert(i + 2 == e && "Array op not second to last");
950 std::string Str;
951 bool ArrayIsPrintable = true;
952 for (unsigned j = i - 1, je = Record.size(); j != je; ++j) {
953 if (!isPrint(static_cast<unsigned char>(Record[j]))) {
954 ArrayIsPrintable = false;
955 break;
956 }
957 Str += (char)Record[j];
958 }
959 if (ArrayIsPrintable)
960 O->OS << " record string = '" << Str << "'";
961 break;
962 }
963 }
964
965 if (Blob.data()) {
966 if (canDecodeBlob(Code, BlockID)) {
967 if (Error E = decodeMetadataStringsBlob(Indent, Record, Blob, O->OS))
968 return E;
969 } else {
970 O->OS << " blob data = ";
971 if (O->ShowBinaryBlobs) {
972 O->OS << "'";
973 O->OS.write_escaped(Blob, /*hex=*/true) << "'";
974 } else {
975 bool BlobIsPrintable = true;
976 for (char C : Blob)
977 if (!isPrint(static_cast<unsigned char>(C))) {
978 BlobIsPrintable = false;
979 break;
980 }
981
982 if (BlobIsPrintable)
983 O->OS << "'" << Blob << "'";
984 else
985 O->OS << "unprintable, " << Blob.size() << " bytes.";
986 }
987 }
988 }
989
990 O->OS << "\n";
991 }
992
993 // Make sure that we can skip the current record.
994 if (Error Err = Stream.JumpToBit(CurrentRecordPos))
995 return Err;
996 if (Expected<unsigned> Skipped = Stream.skipRecord(Entry.ID))
997 ; // Do nothing.
998 else
999 return Skipped.takeError();
1000 }
1001 }
1002
1003