LLVM 19.0.0git
MicrosoftDemangle.cpp
Go to the documentation of this file.
1//===- MicrosoftDemangle.cpp ----------------------------------------------===//
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// This file defines a demangler for MSVC-style mangled symbols.
10//
11// This file has no dependencies on the rest of LLVM so that it can be
12// easily reused in other programs such as libcxxabi.
13//
14//===----------------------------------------------------------------------===//
15
17
23
24#include <array>
25#include <cctype>
26#include <cstdio>
27#include <string_view>
28#include <tuple>
29
30using namespace llvm;
31using namespace ms_demangle;
32
33static bool startsWithDigit(std::string_view S) {
34 return !S.empty() && std::isdigit(S.front());
35}
36
37struct NodeList {
38 Node *N = nullptr;
39 NodeList *Next = nullptr;
40};
41
42static bool consumeFront(std::string_view &S, char C) {
43 if (!llvm::itanium_demangle::starts_with(S, C))
44 return false;
45 S.remove_prefix(1);
46 return true;
47}
48
49static bool consumeFront(std::string_view &S, std::string_view C) {
50 if (!llvm::itanium_demangle::starts_with(S, C))
51 return false;
52 S.remove_prefix(C.size());
53 return true;
54}
55
56static bool consumeFront(std::string_view &S, std::string_view PrefixA,
57 std::string_view PrefixB, bool A) {
58 const std::string_view &Prefix = A ? PrefixA : PrefixB;
59 return consumeFront(S, Prefix);
60}
61
62static bool startsWith(std::string_view S, std::string_view PrefixA,
63 std::string_view PrefixB, bool A) {
64 const std::string_view &Prefix = A ? PrefixA : PrefixB;
65 return llvm::itanium_demangle::starts_with(S, Prefix);
66}
67
68static bool isMemberPointer(std::string_view MangledName, bool &Error) {
69 Error = false;
70 const char F = MangledName.front();
71 MangledName.remove_prefix(1);
72 switch (F) {
73 case '$':
74 // This is probably an rvalue reference (e.g. $$Q), and you cannot have an
75 // rvalue reference to a member.
76 return false;
77 case 'A':
78 // 'A' indicates a reference, and you cannot have a reference to a member
79 // function or member.
80 return false;
81 case 'P':
82 case 'Q':
83 case 'R':
84 case 'S':
85 // These 4 values indicate some kind of pointer, but we still don't know
86 // what.
87 break;
88 default:
89 // isMemberPointer() is called only if isPointerType() returns true,
90 // and it rejects other prefixes.
92 }
93
94 // If it starts with a number, then 6 indicates a non-member function
95 // pointer, and 8 indicates a member function pointer.
96 if (startsWithDigit(MangledName)) {
97 if (MangledName[0] != '6' && MangledName[0] != '8') {
98 Error = true;
99 return false;
100 }
101 return (MangledName[0] == '8');
102 }
103
104 // Remove ext qualifiers since those can appear on either type and are
105 // therefore not indicative.
106 consumeFront(MangledName, 'E'); // 64-bit
107 consumeFront(MangledName, 'I'); // restrict
108 consumeFront(MangledName, 'F'); // unaligned
109
110 if (MangledName.empty()) {
111 Error = true;
112 return false;
113 }
114
115 // The next value should be either ABCD (non-member) or QRST (member).
116 switch (MangledName.front()) {
117 case 'A':
118 case 'B':
119 case 'C':
120 case 'D':
121 return false;
122 case 'Q':
123 case 'R':
124 case 'S':
125 case 'T':
126 return true;
127 default:
128 Error = true;
129 return false;
130 }
131}
132
134consumeSpecialIntrinsicKind(std::string_view &MangledName) {
135 if (consumeFront(MangledName, "?_7"))
136 return SpecialIntrinsicKind::Vftable;
137 if (consumeFront(MangledName, "?_8"))
138 return SpecialIntrinsicKind::Vbtable;
139 if (consumeFront(MangledName, "?_9"))
140 return SpecialIntrinsicKind::VcallThunk;
141 if (consumeFront(MangledName, "?_A"))
142 return SpecialIntrinsicKind::Typeof;
143 if (consumeFront(MangledName, "?_B"))
144 return SpecialIntrinsicKind::LocalStaticGuard;
145 if (consumeFront(MangledName, "?_C"))
146 return SpecialIntrinsicKind::StringLiteralSymbol;
147 if (consumeFront(MangledName, "?_P"))
148 return SpecialIntrinsicKind::UdtReturning;
149 if (consumeFront(MangledName, "?_R0"))
150 return SpecialIntrinsicKind::RttiTypeDescriptor;
151 if (consumeFront(MangledName, "?_R1"))
152 return SpecialIntrinsicKind::RttiBaseClassDescriptor;
153 if (consumeFront(MangledName, "?_R2"))
154 return SpecialIntrinsicKind::RttiBaseClassArray;
155 if (consumeFront(MangledName, "?_R3"))
156 return SpecialIntrinsicKind::RttiClassHierarchyDescriptor;
157 if (consumeFront(MangledName, "?_R4"))
158 return SpecialIntrinsicKind::RttiCompleteObjLocator;
159 if (consumeFront(MangledName, "?_S"))
160 return SpecialIntrinsicKind::LocalVftable;
161 if (consumeFront(MangledName, "?__E"))
162 return SpecialIntrinsicKind::DynamicInitializer;
163 if (consumeFront(MangledName, "?__F"))
164 return SpecialIntrinsicKind::DynamicAtexitDestructor;
165 if (consumeFront(MangledName, "?__J"))
166 return SpecialIntrinsicKind::LocalStaticThreadGuard;
167 return SpecialIntrinsicKind::None;
168}
169
170static bool startsWithLocalScopePattern(std::string_view S) {
171 if (!consumeFront(S, '?'))
172 return false;
173
174 size_t End = S.find('?');
175 if (End == std::string_view::npos)
176 return false;
177 std::string_view Candidate = S.substr(0, End);
178 if (Candidate.empty())
179 return false;
180
181 // \?[0-9]\?
182 // ?@? is the discriminator 0.
183 if (Candidate.size() == 1)
184 return Candidate[0] == '@' || (Candidate[0] >= '0' && Candidate[0] <= '9');
185
186 // If it's not 0-9, then it's an encoded number terminated with an @
187 if (Candidate.back() != '@')
188 return false;
189 Candidate.remove_suffix(1);
190
191 // An encoded number starts with B-P and all subsequent digits are in A-P.
192 // Note that the reason the first digit cannot be A is two fold. First, it
193 // would create an ambiguity with ?A which delimits the beginning of an
194 // anonymous namespace. Second, A represents 0, and you don't start a multi
195 // digit number with a leading 0. Presumably the anonymous namespace
196 // ambiguity is also why single digit encoded numbers use 0-9 rather than A-J.
197 if (Candidate[0] < 'B' || Candidate[0] > 'P')
198 return false;
199 Candidate.remove_prefix(1);
200 while (!Candidate.empty()) {
201 if (Candidate[0] < 'A' || Candidate[0] > 'P')
202 return false;
203 Candidate.remove_prefix(1);
204 }
205
206 return true;
207}
208
209static bool isTagType(std::string_view S) {
210 switch (S.front()) {
211 case 'T': // union
212 case 'U': // struct
213 case 'V': // class
214 case 'W': // enum
215 return true;
216 }
217 return false;
218}
219
220static bool isCustomType(std::string_view S) { return S[0] == '?'; }
221
222static bool isPointerType(std::string_view S) {
223 if (llvm::itanium_demangle::starts_with(S, "$$Q")) // foo &&
224 return true;
225
226 switch (S.front()) {
227 case 'A': // foo &
228 case 'P': // foo *
229 case 'Q': // foo *const
230 case 'R': // foo *volatile
231 case 'S': // foo *const volatile
232 return true;
233 }
234 return false;
235}
236
237static bool isArrayType(std::string_view S) { return S[0] == 'Y'; }
238
239static bool isFunctionType(std::string_view S) {
240 return llvm::itanium_demangle::starts_with(S, "$$A8@@") ||
241 llvm::itanium_demangle::starts_with(S, "$$A6");
242}
243
245demangleFunctionRefQualifier(std::string_view &MangledName) {
246 if (consumeFront(MangledName, 'G'))
247 return FunctionRefQualifier::Reference;
248 else if (consumeFront(MangledName, 'H'))
249 return FunctionRefQualifier::RValueReference;
250 return FunctionRefQualifier::None;
251}
252
253static std::pair<Qualifiers, PointerAffinity>
254demanglePointerCVQualifiers(std::string_view &MangledName) {
255 if (consumeFront(MangledName, "$$Q"))
256 return std::make_pair(Q_None, PointerAffinity::RValueReference);
257
258 const char F = MangledName.front();
259 MangledName.remove_prefix(1);
260 switch (F) {
261 case 'A':
262 return std::make_pair(Q_None, PointerAffinity::Reference);
263 case 'P':
264 return std::make_pair(Q_None, PointerAffinity::Pointer);
265 case 'Q':
266 return std::make_pair(Q_Const, PointerAffinity::Pointer);
267 case 'R':
268 return std::make_pair(Q_Volatile, PointerAffinity::Pointer);
269 case 'S':
270 return std::make_pair(Qualifiers(Q_Const | Q_Volatile),
271 PointerAffinity::Pointer);
272 }
273 // This function is only called if isPointerType() returns true,
274 // and it only returns true for the six cases listed above.
276}
277
278std::string_view Demangler::copyString(std::string_view Borrowed) {
279 char *Stable = Arena.allocUnalignedBuffer(Borrowed.size());
280 // This is not a micro-optimization, it avoids UB, should Borrowed be an null
281 // buffer.
282 if (Borrowed.size())
283 std::memcpy(Stable, Borrowed.data(), Borrowed.size());
284
285 return {Stable, Borrowed.size()};
286}
287
289Demangler::demangleSpecialTableSymbolNode(std::string_view &MangledName,
292 switch (K) {
294 NI->Name = "`vftable'";
295 break;
297 NI->Name = "`vbtable'";
298 break;
300 NI->Name = "`local vftable'";
301 break;
303 NI->Name = "`RTTI Complete Object Locator'";
304 break;
305 default:
307 }
308 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
310 STSN->Name = QN;
311 bool IsMember = false;
312 if (MangledName.empty()) {
313 Error = true;
314 return nullptr;
315 }
316 char Front = MangledName.front();
317 MangledName.remove_prefix(1);
318 if (Front != '6' && Front != '7') {
319 Error = true;
320 return nullptr;
321 }
322
323 std::tie(STSN->Quals, IsMember) = demangleQualifiers(MangledName);
324 if (!consumeFront(MangledName, '@'))
325 STSN->TargetName = demangleFullyQualifiedTypeName(MangledName);
326 return STSN;
327}
328
330Demangler::demangleLocalStaticGuard(std::string_view &MangledName,
331 bool IsThread) {
334 LSGI->IsThread = IsThread;
335 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, LSGI);
338 LSGVN->Name = QN;
339
340 if (consumeFront(MangledName, "4IA"))
341 LSGVN->IsVisible = false;
342 else if (consumeFront(MangledName, "5"))
343 LSGVN->IsVisible = true;
344 else {
345 Error = true;
346 return nullptr;
347 }
348
349 if (!MangledName.empty())
350 LSGI->ScopeIndex = demangleUnsigned(MangledName);
351 return LSGVN;
352}
353
355 std::string_view Name) {
357 Id->Name = Name;
358 return Id;
359}
360
364 QN->Components = Arena.alloc<NodeArrayNode>();
365 QN->Components->Count = 1;
366 QN->Components->Nodes = Arena.allocArray<Node *>(1);
367 QN->Components->Nodes[0] = Identifier;
368 return QN;
369}
370
372 std::string_view Name) {
374 return synthesizeQualifiedName(Arena, Id);
375}
376
378 TypeNode *Type,
379 std::string_view VariableName) {
381 VSN->Type = Type;
382 VSN->Name = synthesizeQualifiedName(Arena, VariableName);
383 return VSN;
384}
385
387Demangler::demangleUntypedVariable(ArenaAllocator &Arena,
388 std::string_view &MangledName,
389 std::string_view VariableName) {
390 NamedIdentifierNode *NI = synthesizeNamedIdentifier(Arena, VariableName);
391 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
393 VSN->Name = QN;
394 if (consumeFront(MangledName, "8"))
395 return VSN;
396
397 Error = true;
398 return nullptr;
399}
400
402Demangler::demangleRttiBaseClassDescriptorNode(ArenaAllocator &Arena,
403 std::string_view &MangledName) {
406 RBCDN->NVOffset = demangleUnsigned(MangledName);
407 RBCDN->VBPtrOffset = demangleSigned(MangledName);
408 RBCDN->VBTableOffset = demangleUnsigned(MangledName);
409 RBCDN->Flags = demangleUnsigned(MangledName);
410 if (Error)
411 return nullptr;
412
414 VSN->Name = demangleNameScopeChain(MangledName, RBCDN);
415 consumeFront(MangledName, '8');
416 return VSN;
417}
418
420Demangler::demangleInitFiniStub(std::string_view &MangledName,
421 bool IsDestructor) {
424 DSIN->IsDestructor = IsDestructor;
425
426 bool IsKnownStaticDataMember = false;
427 if (consumeFront(MangledName, '?'))
428 IsKnownStaticDataMember = true;
429
430 SymbolNode *Symbol = demangleDeclarator(MangledName);
431 if (Error)
432 return nullptr;
433
434 FunctionSymbolNode *FSN = nullptr;
435
436 if (Symbol->kind() == NodeKind::VariableSymbol) {
437 DSIN->Variable = static_cast<VariableSymbolNode *>(Symbol);
438
439 // Older versions of clang mangled this type of symbol incorrectly. They
440 // would omit the leading ? and they would only emit a single @ at the end.
441 // The correct mangling is a leading ? and 2 trailing @ signs. Handle
442 // both cases.
443 int AtCount = IsKnownStaticDataMember ? 2 : 1;
444 for (int I = 0; I < AtCount; ++I) {
445 if (consumeFront(MangledName, '@'))
446 continue;
447 Error = true;
448 return nullptr;
449 }
450
451 FSN = demangleFunctionEncoding(MangledName);
452 if (FSN)
453 FSN->Name = synthesizeQualifiedName(Arena, DSIN);
454 } else {
455 if (IsKnownStaticDataMember) {
456 // This was supposed to be a static data member, but we got a function.
457 Error = true;
458 return nullptr;
459 }
460
461 FSN = static_cast<FunctionSymbolNode *>(Symbol);
462 DSIN->Name = Symbol->Name;
463 FSN->Name = synthesizeQualifiedName(Arena, DSIN);
464 }
465
466 return FSN;
467}
468
469SymbolNode *Demangler::demangleSpecialIntrinsic(std::string_view &MangledName) {
471
472 switch (SIK) {
474 return nullptr;
476 return demangleStringLiteral(MangledName);
481 return demangleSpecialTableSymbolNode(MangledName, SIK);
483 return demangleVcallThunkNode(MangledName);
485 return demangleLocalStaticGuard(MangledName, /*IsThread=*/false);
487 return demangleLocalStaticGuard(MangledName, /*IsThread=*/true);
489 TypeNode *T = demangleType(MangledName, QualifierMangleMode::Result);
490 if (Error)
491 break;
492 if (!consumeFront(MangledName, "@8"))
493 break;
494 if (!MangledName.empty())
495 break;
496 return synthesizeVariable(Arena, T, "`RTTI Type Descriptor'");
497 }
499 return demangleUntypedVariable(Arena, MangledName,
500 "`RTTI Base Class Array'");
502 return demangleUntypedVariable(Arena, MangledName,
503 "`RTTI Class Hierarchy Descriptor'");
505 return demangleRttiBaseClassDescriptorNode(Arena, MangledName);
507 return demangleInitFiniStub(MangledName, /*IsDestructor=*/false);
509 return demangleInitFiniStub(MangledName, /*IsDestructor=*/true);
512 // It's unclear which tools produces these manglings, so demangling
513 // support is not (yet?) implemented.
514 break;
516 DEMANGLE_UNREACHABLE; // Never returned by consumeSpecialIntrinsicKind.
517 }
518 Error = true;
519 return nullptr;
520}
521
523Demangler::demangleFunctionIdentifierCode(std::string_view &MangledName) {
524 assert(llvm::itanium_demangle::starts_with(MangledName, '?'));
525 MangledName.remove_prefix(1);
526 if (MangledName.empty()) {
527 Error = true;
528 return nullptr;
529 }
530
531 if (consumeFront(MangledName, "__"))
532 return demangleFunctionIdentifierCode(
534 if (consumeFront(MangledName, "_"))
535 return demangleFunctionIdentifierCode(MangledName,
537 return demangleFunctionIdentifierCode(MangledName,
539}
540
542Demangler::demangleStructorIdentifier(std::string_view &MangledName,
543 bool IsDestructor) {
545 N->IsDestructor = IsDestructor;
546 return N;
547}
548
550Demangler::demangleConversionOperatorIdentifier(std::string_view &MangledName) {
553 return N;
554}
555
557Demangler::demangleLiteralOperatorIdentifier(std::string_view &MangledName) {
560 N->Name = demangleSimpleString(MangledName, /*Memorize=*/false);
561 return N;
562}
563
565Demangler::translateIntrinsicFunctionCode(char CH,
567 using IFK = IntrinsicFunctionKind;
568 if (!(CH >= '0' && CH <= '9') && !(CH >= 'A' && CH <= 'Z')) {
569 Error = true;
570 return IFK::None;
571 }
572
573 // Not all ? identifiers are intrinsics *functions*. This function only maps
574 // operator codes for the special functions, all others are handled elsewhere,
575 // hence the IFK::None entries in the table.
576 static IFK Basic[36] = {
577 IFK::None, // ?0 # Foo::Foo()
578 IFK::None, // ?1 # Foo::~Foo()
579 IFK::New, // ?2 # operator new
580 IFK::Delete, // ?3 # operator delete
581 IFK::Assign, // ?4 # operator=
582 IFK::RightShift, // ?5 # operator>>
583 IFK::LeftShift, // ?6 # operator<<
584 IFK::LogicalNot, // ?7 # operator!
585 IFK::Equals, // ?8 # operator==
586 IFK::NotEquals, // ?9 # operator!=
587 IFK::ArraySubscript, // ?A # operator[]
588 IFK::None, // ?B # Foo::operator <type>()
589 IFK::Pointer, // ?C # operator->
590 IFK::Dereference, // ?D # operator*
591 IFK::Increment, // ?E # operator++
592 IFK::Decrement, // ?F # operator--
593 IFK::Minus, // ?G # operator-
594 IFK::Plus, // ?H # operator+
595 IFK::BitwiseAnd, // ?I # operator&
596 IFK::MemberPointer, // ?J # operator->*
597 IFK::Divide, // ?K # operator/
598 IFK::Modulus, // ?L # operator%
599 IFK::LessThan, // ?M operator<
600 IFK::LessThanEqual, // ?N operator<=
601 IFK::GreaterThan, // ?O operator>
602 IFK::GreaterThanEqual, // ?P operator>=
603 IFK::Comma, // ?Q operator,
604 IFK::Parens, // ?R operator()
605 IFK::BitwiseNot, // ?S operator~
606 IFK::BitwiseXor, // ?T operator^
607 IFK::BitwiseOr, // ?U operator|
608 IFK::LogicalAnd, // ?V operator&&
609 IFK::LogicalOr, // ?W operator||
610 IFK::TimesEqual, // ?X operator*=
611 IFK::PlusEqual, // ?Y operator+=
612 IFK::MinusEqual, // ?Z operator-=
613 };
614 static IFK Under[36] = {
615 IFK::DivEqual, // ?_0 operator/=
616 IFK::ModEqual, // ?_1 operator%=
617 IFK::RshEqual, // ?_2 operator>>=
618 IFK::LshEqual, // ?_3 operator<<=
619 IFK::BitwiseAndEqual, // ?_4 operator&=
620 IFK::BitwiseOrEqual, // ?_5 operator|=
621 IFK::BitwiseXorEqual, // ?_6 operator^=
622 IFK::None, // ?_7 # vftable
623 IFK::None, // ?_8 # vbtable
624 IFK::None, // ?_9 # vcall
625 IFK::None, // ?_A # typeof
626 IFK::None, // ?_B # local static guard
627 IFK::None, // ?_C # string literal
628 IFK::VbaseDtor, // ?_D # vbase destructor
629 IFK::VecDelDtor, // ?_E # vector deleting destructor
630 IFK::DefaultCtorClosure, // ?_F # default constructor closure
631 IFK::ScalarDelDtor, // ?_G # scalar deleting destructor
632 IFK::VecCtorIter, // ?_H # vector constructor iterator
633 IFK::VecDtorIter, // ?_I # vector destructor iterator
634 IFK::VecVbaseCtorIter, // ?_J # vector vbase constructor iterator
635 IFK::VdispMap, // ?_K # virtual displacement map
636 IFK::EHVecCtorIter, // ?_L # eh vector constructor iterator
637 IFK::EHVecDtorIter, // ?_M # eh vector destructor iterator
638 IFK::EHVecVbaseCtorIter, // ?_N # eh vector vbase constructor iterator
639 IFK::CopyCtorClosure, // ?_O # copy constructor closure
640 IFK::None, // ?_P<name> # udt returning <name>
641 IFK::None, // ?_Q # <unknown>
642 IFK::None, // ?_R0 - ?_R4 # RTTI Codes
643 IFK::None, // ?_S # local vftable
644 IFK::LocalVftableCtorClosure, // ?_T # local vftable constructor closure
645 IFK::ArrayNew, // ?_U operator new[]
646 IFK::ArrayDelete, // ?_V operator delete[]
647 IFK::None, // ?_W <unused>
648 IFK::None, // ?_X <unused>
649 IFK::None, // ?_Y <unused>
650 IFK::None, // ?_Z <unused>
651 };
652 static IFK DoubleUnder[36] = {
653 IFK::None, // ?__0 <unused>
654 IFK::None, // ?__1 <unused>
655 IFK::None, // ?__2 <unused>
656 IFK::None, // ?__3 <unused>
657 IFK::None, // ?__4 <unused>
658 IFK::None, // ?__5 <unused>
659 IFK::None, // ?__6 <unused>
660 IFK::None, // ?__7 <unused>
661 IFK::None, // ?__8 <unused>
662 IFK::None, // ?__9 <unused>
663 IFK::ManVectorCtorIter, // ?__A managed vector ctor iterator
664 IFK::ManVectorDtorIter, // ?__B managed vector dtor iterator
665 IFK::EHVectorCopyCtorIter, // ?__C EH vector copy ctor iterator
666 IFK::EHVectorVbaseCopyCtorIter, // ?__D EH vector vbase copy ctor iter
667 IFK::None, // ?__E dynamic initializer for `T'
668 IFK::None, // ?__F dynamic atexit destructor for `T'
669 IFK::VectorCopyCtorIter, // ?__G vector copy constructor iter
670 IFK::VectorVbaseCopyCtorIter, // ?__H vector vbase copy ctor iter
671 IFK::ManVectorVbaseCopyCtorIter, // ?__I managed vector vbase copy ctor
672 // iter
673 IFK::None, // ?__J local static thread guard
674 IFK::None, // ?__K operator ""_name
675 IFK::CoAwait, // ?__L operator co_await
676 IFK::Spaceship, // ?__M operator<=>
677 IFK::None, // ?__N <unused>
678 IFK::None, // ?__O <unused>
679 IFK::None, // ?__P <unused>
680 IFK::None, // ?__Q <unused>
681 IFK::None, // ?__R <unused>
682 IFK::None, // ?__S <unused>
683 IFK::None, // ?__T <unused>
684 IFK::None, // ?__U <unused>
685 IFK::None, // ?__V <unused>
686 IFK::None, // ?__W <unused>
687 IFK::None, // ?__X <unused>
688 IFK::None, // ?__Y <unused>
689 IFK::None, // ?__Z <unused>
690 };
691
692 int Index = (CH >= '0' && CH <= '9') ? (CH - '0') : (CH - 'A' + 10);
693 switch (Group) {
695 return Basic[Index];
697 return Under[Index];
699 return DoubleUnder[Index];
700 }
702}
703
705Demangler::demangleFunctionIdentifierCode(std::string_view &MangledName,
707 if (MangledName.empty()) {
708 Error = true;
709 return nullptr;
710 }
711 const char CH = MangledName.front();
712 switch (Group) {
714 MangledName.remove_prefix(1);
715 switch (CH) {
716 case '0':
717 case '1':
718 return demangleStructorIdentifier(MangledName, CH == '1');
719 case 'B':
720 return demangleConversionOperatorIdentifier(MangledName);
721 default:
723 translateIntrinsicFunctionCode(CH, Group));
724 }
726 MangledName.remove_prefix(1);
728 translateIntrinsicFunctionCode(CH, Group));
730 MangledName.remove_prefix(1);
731 switch (CH) {
732 case 'K':
733 return demangleLiteralOperatorIdentifier(MangledName);
734 default:
736 translateIntrinsicFunctionCode(CH, Group));
737 }
738 }
739
741}
742
743SymbolNode *Demangler::demangleEncodedSymbol(std::string_view &MangledName,
745 if (MangledName.empty()) {
746 Error = true;
747 return nullptr;
748 }
749
750 // Read a variable.
751 switch (MangledName.front()) {
752 case '0':
753 case '1':
754 case '2':
755 case '3':
756 case '4': {
757 StorageClass SC = demangleVariableStorageClass(MangledName);
758 return demangleVariableEncoding(MangledName, SC);
759 }
760 }
761 FunctionSymbolNode *FSN = demangleFunctionEncoding(MangledName);
762
763 IdentifierNode *UQN = Name->getUnqualifiedIdentifier();
766 static_cast<ConversionOperatorIdentifierNode *>(UQN);
767 if (FSN)
768 COIN->TargetType = FSN->Signature->ReturnType;
769 }
770 return FSN;
771}
772
773SymbolNode *Demangler::demangleDeclarator(std::string_view &MangledName) {
774 // What follows is a main symbol name. This may include namespaces or class
775 // back references.
776 QualifiedNameNode *QN = demangleFullyQualifiedSymbolName(MangledName);
777 if (Error)
778 return nullptr;
779
780 SymbolNode *Symbol = demangleEncodedSymbol(MangledName, QN);
781 if (Error)
782 return nullptr;
783 Symbol->Name = QN;
784
788 static_cast<ConversionOperatorIdentifierNode *>(UQN);
789 if (!COIN->TargetType) {
790 Error = true;
791 return nullptr;
792 }
793 }
794 return Symbol;
795}
796
797SymbolNode *Demangler::demangleMD5Name(std::string_view &MangledName) {
798 assert(llvm::itanium_demangle::starts_with(MangledName, "??@"));
799 // This is an MD5 mangled name. We can't demangle it, just return the
800 // mangled name.
801 // An MD5 mangled name is ??@ followed by 32 characters and a terminating @.
802 size_t MD5Last = MangledName.find('@', strlen("??@"));
803 if (MD5Last == std::string_view::npos) {
804 Error = true;
805 return nullptr;
806 }
807 const char *Start = MangledName.data();
808 const size_t StartSize = MangledName.size();
809 MangledName.remove_prefix(MD5Last + 1);
810
811 // There are two additional special cases for MD5 names:
812 // 1. For complete object locators where the object name is long enough
813 // for the object to have an MD5 name, the complete object locator is
814 // called ??@...@??_R4@ (with a trailing "??_R4@" instead of the usual
815 // leading "??_R4". This is handled here.
816 // 2. For catchable types, in versions of MSVC before 2015 (<1900) or after
817 // 2017.2 (>= 1914), the catchable type mangling is _CT??@...@??@...@8
818 // instead of_CT??@...@8 with just one MD5 name. Since we don't yet
819 // demangle catchable types anywhere, this isn't handled for MD5 names
820 // either.
821 consumeFront(MangledName, "??_R4@");
822
823 assert(MangledName.size() < StartSize);
824 const size_t Count = StartSize - MangledName.size();
825 std::string_view MD5(Start, Count);
827 S->Name = synthesizeQualifiedName(Arena, MD5);
828
829 return S;
830}
831
832SymbolNode *Demangler::demangleTypeinfoName(std::string_view &MangledName) {
833 assert(llvm::itanium_demangle::starts_with(MangledName, '.'));
834 consumeFront(MangledName, '.');
835
836 TypeNode *T = demangleType(MangledName, QualifierMangleMode::Result);
837 if (Error || !MangledName.empty()) {
838 Error = true;
839 return nullptr;
840 }
841 return synthesizeVariable(Arena, T, "`RTTI Type Descriptor Name'");
842}
843
844// Parser entry point.
845SymbolNode *Demangler::parse(std::string_view &MangledName) {
846 // Typeinfo names are strings stored in RTTI data. They're not symbol names.
847 // It's still useful to demangle them. They're the only demangled entity
848 // that doesn't start with a "?" but a ".".
849 if (llvm::itanium_demangle::starts_with(MangledName, '.'))
850 return demangleTypeinfoName(MangledName);
851
852 if (llvm::itanium_demangle::starts_with(MangledName, "??@"))
853 return demangleMD5Name(MangledName);
854
855 // MSVC-style mangled symbols must start with '?'.
856 if (!llvm::itanium_demangle::starts_with(MangledName, '?')) {
857 Error = true;
858 return nullptr;
859 }
860
861 consumeFront(MangledName, '?');
862
863 // ?$ is a template instantiation, but all other names that start with ? are
864 // operators / special names.
865 if (SymbolNode *SI = demangleSpecialIntrinsic(MangledName))
866 return SI;
867
868 return demangleDeclarator(MangledName);
869}
870
871TagTypeNode *Demangler::parseTagUniqueName(std::string_view &MangledName) {
872 if (!consumeFront(MangledName, ".?A")) {
873 Error = true;
874 return nullptr;
875 }
876 consumeFront(MangledName, ".?A");
877 if (MangledName.empty()) {
878 Error = true;
879 return nullptr;
880 }
881
882 return demangleClassType(MangledName);
883}
884
885// <type-encoding> ::= <storage-class> <variable-type>
886// <storage-class> ::= 0 # private static member
887// ::= 1 # protected static member
888// ::= 2 # public static member
889// ::= 3 # global
890// ::= 4 # static local
891
893Demangler::demangleVariableEncoding(std::string_view &MangledName,
894 StorageClass SC) {
896
897 VSN->Type = demangleType(MangledName, QualifierMangleMode::Drop);
898 VSN->SC = SC;
899
900 if (Error)
901 return nullptr;
902
903 // <variable-type> ::= <type> <cvr-qualifiers>
904 // ::= <type> <pointee-cvr-qualifiers> # pointers, references
905 switch (VSN->Type->kind()) {
907 PointerTypeNode *PTN = static_cast<PointerTypeNode *>(VSN->Type);
908
909 Qualifiers ExtraChildQuals = Q_None;
910 PTN->Quals = Qualifiers(VSN->Type->Quals |
911 demanglePointerExtQualifiers(MangledName));
912
913 bool IsMember = false;
914 std::tie(ExtraChildQuals, IsMember) = demangleQualifiers(MangledName);
915
916 if (PTN->ClassParent) {
917 QualifiedNameNode *BackRefName =
918 demangleFullyQualifiedTypeName(MangledName);
919 (void)BackRefName;
920 }
921 PTN->Pointee->Quals = Qualifiers(PTN->Pointee->Quals | ExtraChildQuals);
922
923 break;
924 }
925 default:
926 VSN->Type->Quals = demangleQualifiers(MangledName).first;
927 break;
928 }
929
930 return VSN;
931}
932
933// Sometimes numbers are encoded in mangled symbols. For example,
934// "int (*x)[20]" is a valid C type (x is a pointer to an array of
935// length 20), so we need some way to embed numbers as part of symbols.
936// This function parses it.
937//
938// <number> ::= [?] <non-negative integer>
939//
940// <non-negative integer> ::= <decimal digit> # when 1 <= Number <= 10
941// ::= <hex digit>+ @ # when Number == 0 or >= 10
942//
943// <hex-digit> ::= [A-P] # A = 0, B = 1, ...
944std::pair<uint64_t, bool>
945Demangler::demangleNumber(std::string_view &MangledName) {
946 bool IsNegative = consumeFront(MangledName, '?');
947
948 if (startsWithDigit(MangledName)) {
949 uint64_t Ret = MangledName[0] - '0' + 1;
950 MangledName.remove_prefix(1);
951 return {Ret, IsNegative};
952 }
953
954 uint64_t Ret = 0;
955 for (size_t i = 0; i < MangledName.size(); ++i) {
956 char C = MangledName[i];
957 if (C == '@') {
958 MangledName.remove_prefix(i + 1);
959 return {Ret, IsNegative};
960 }
961 if ('A' <= C && C <= 'P') {
962 Ret = (Ret << 4) + (C - 'A');
963 continue;
964 }
965 break;
966 }
967
968 Error = true;
969 return {0ULL, false};
970}
971
972uint64_t Demangler::demangleUnsigned(std::string_view &MangledName) {
973 bool IsNegative = false;
974 uint64_t Number = 0;
975 std::tie(Number, IsNegative) = demangleNumber(MangledName);
976 if (IsNegative)
977 Error = true;
978 return Number;
979}
980
981int64_t Demangler::demangleSigned(std::string_view &MangledName) {
982 bool IsNegative = false;
983 uint64_t Number = 0;
984 std::tie(Number, IsNegative) = demangleNumber(MangledName);
985 if (Number > INT64_MAX)
986 Error = true;
987 int64_t I = static_cast<int64_t>(Number);
988 return IsNegative ? -I : I;
989}
990
991// First 10 strings can be referenced by special BackReferences ?0, ?1, ..., ?9.
992// Memorize it.
993void Demangler::memorizeString(std::string_view S) {
994 if (Backrefs.NamesCount >= BackrefContext::Max)
995 return;
996 for (size_t i = 0; i < Backrefs.NamesCount; ++i)
997 if (S == Backrefs.Names[i]->Name)
998 return;
1000 N->Name = S;
1001 Backrefs.Names[Backrefs.NamesCount++] = N;
1002}
1003
1005Demangler::demangleBackRefName(std::string_view &MangledName) {
1006 assert(startsWithDigit(MangledName));
1007
1008 size_t I = MangledName[0] - '0';
1009 if (I >= Backrefs.NamesCount) {
1010 Error = true;
1011 return nullptr;
1012 }
1013
1014 MangledName.remove_prefix(1);
1015 return Backrefs.Names[I];
1016}
1017
1018void Demangler::memorizeIdentifier(IdentifierNode *Identifier) {
1019 // Render this class template name into a string buffer so that we can
1020 // memorize it for the purpose of back-referencing.
1022 Identifier->output(OB, OF_Default);
1023 std::string_view Owned = copyString(OB);
1024 memorizeString(Owned);
1025 std::free(OB.getBuffer());
1026}
1027
1029Demangler::demangleTemplateInstantiationName(std::string_view &MangledName,
1030 NameBackrefBehavior NBB) {
1031 assert(llvm::itanium_demangle::starts_with(MangledName, "?$"));
1032 consumeFront(MangledName, "?$");
1033
1034 BackrefContext OuterContext;
1035 std::swap(OuterContext, Backrefs);
1036
1038 demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
1039 if (!Error)
1040 Identifier->TemplateParams = demangleTemplateParameterList(MangledName);
1041
1042 std::swap(OuterContext, Backrefs);
1043 if (Error)
1044 return nullptr;
1045
1046 if (NBB & NBB_Template) {
1047 // NBB_Template is only set for types and non-leaf names ("a::" in "a::b").
1048 // Structors and conversion operators only makes sense in a leaf name, so
1049 // reject them in NBB_Template contexts.
1052 Error = true;
1053 return nullptr;
1054 }
1055
1056 memorizeIdentifier(Identifier);
1057 }
1058
1059 return Identifier;
1060}
1061
1063Demangler::demangleSimpleName(std::string_view &MangledName, bool Memorize) {
1064 std::string_view S = demangleSimpleString(MangledName, Memorize);
1065 if (Error)
1066 return nullptr;
1067
1069 Name->Name = S;
1070 return Name;
1071}
1072
1073static bool isRebasedHexDigit(char C) { return (C >= 'A' && C <= 'P'); }
1074
1075static uint8_t rebasedHexDigitToNumber(char C) {
1077 return (C <= 'J') ? (C - 'A') : (10 + C - 'K');
1078}
1079
1080uint8_t Demangler::demangleCharLiteral(std::string_view &MangledName) {
1081 assert(!MangledName.empty());
1082 if (!llvm::itanium_demangle::starts_with(MangledName, '?')) {
1083 const uint8_t F = MangledName.front();
1084 MangledName.remove_prefix(1);
1085 return F;
1086 }
1087
1088 MangledName.remove_prefix(1);
1089 if (MangledName.empty())
1090 goto CharLiteralError;
1091
1092 if (consumeFront(MangledName, '$')) {
1093 // Two hex digits
1094 if (MangledName.size() < 2)
1095 goto CharLiteralError;
1096 std::string_view Nibbles = MangledName.substr(0, 2);
1097 if (!isRebasedHexDigit(Nibbles[0]) || !isRebasedHexDigit(Nibbles[1]))
1098 goto CharLiteralError;
1099 // Don't append the null terminator.
1100 uint8_t C1 = rebasedHexDigitToNumber(Nibbles[0]);
1101 uint8_t C2 = rebasedHexDigitToNumber(Nibbles[1]);
1102 MangledName.remove_prefix(2);
1103 return (C1 << 4) | C2;
1104 }
1105
1106 if (startsWithDigit(MangledName)) {
1107 const char *Lookup = ",/\\:. \n\t'-";
1108 char C = Lookup[MangledName[0] - '0'];
1109 MangledName.remove_prefix(1);
1110 return C;
1111 }
1112
1113 if (MangledName[0] >= 'a' && MangledName[0] <= 'z') {
1114 char Lookup[26] = {'\xE1', '\xE2', '\xE3', '\xE4', '\xE5', '\xE6', '\xE7',
1115 '\xE8', '\xE9', '\xEA', '\xEB', '\xEC', '\xED', '\xEE',
1116 '\xEF', '\xF0', '\xF1', '\xF2', '\xF3', '\xF4', '\xF5',
1117 '\xF6', '\xF7', '\xF8', '\xF9', '\xFA'};
1118 char C = Lookup[MangledName[0] - 'a'];
1119 MangledName.remove_prefix(1);
1120 return C;
1121 }
1122
1123 if (MangledName[0] >= 'A' && MangledName[0] <= 'Z') {
1124 char Lookup[26] = {'\xC1', '\xC2', '\xC3', '\xC4', '\xC5', '\xC6', '\xC7',
1125 '\xC8', '\xC9', '\xCA', '\xCB', '\xCC', '\xCD', '\xCE',
1126 '\xCF', '\xD0', '\xD1', '\xD2', '\xD3', '\xD4', '\xD5',
1127 '\xD6', '\xD7', '\xD8', '\xD9', '\xDA'};
1128 char C = Lookup[MangledName[0] - 'A'];
1129 MangledName.remove_prefix(1);
1130 return C;
1131 }
1132
1133CharLiteralError:
1134 Error = true;
1135 return '\0';
1136}
1137
1138wchar_t Demangler::demangleWcharLiteral(std::string_view &MangledName) {
1139 uint8_t C1, C2;
1140
1141 C1 = demangleCharLiteral(MangledName);
1142 if (Error || MangledName.empty())
1143 goto WCharLiteralError;
1144 C2 = demangleCharLiteral(MangledName);
1145 if (Error)
1146 goto WCharLiteralError;
1147
1148 return ((wchar_t)C1 << 8) | (wchar_t)C2;
1149
1150WCharLiteralError:
1151 Error = true;
1152 return L'\0';
1153}
1154
1155static void writeHexDigit(char *Buffer, uint8_t Digit) {
1156 assert(Digit <= 15);
1157 *Buffer = (Digit < 10) ? ('0' + Digit) : ('A' + Digit - 10);
1158}
1159
1160static void outputHex(OutputBuffer &OB, unsigned C) {
1161 assert (C != 0);
1162
1163 // It's easier to do the math if we can work from right to left, but we need
1164 // to print the numbers from left to right. So render this into a temporary
1165 // buffer first, then output the temporary buffer. Each byte is of the form
1166 // \xAB, which means that each byte needs 4 characters. Since there are at
1167 // most 4 bytes, we need a 4*4+1 = 17 character temporary buffer.
1168 char TempBuffer[17];
1169
1170 ::memset(TempBuffer, 0, sizeof(TempBuffer));
1171 constexpr int MaxPos = sizeof(TempBuffer) - 1;
1172
1173 int Pos = MaxPos - 1; // TempBuffer[MaxPos] is the terminating \0.
1174 while (C != 0) {
1175 for (int I = 0; I < 2; ++I) {
1176 writeHexDigit(&TempBuffer[Pos--], C % 16);
1177 C /= 16;
1178 }
1179 }
1180 TempBuffer[Pos--] = 'x';
1181 assert(Pos >= 0);
1182 TempBuffer[Pos--] = '\\';
1183 OB << std::string_view(&TempBuffer[Pos + 1]);
1184}
1185
1186static void outputEscapedChar(OutputBuffer &OB, unsigned C) {
1187 switch (C) {
1188 case '\0': // nul
1189 OB << "\\0";
1190 return;
1191 case '\'': // single quote
1192 OB << "\\\'";
1193 return;
1194 case '\"': // double quote
1195 OB << "\\\"";
1196 return;
1197 case '\\': // backslash
1198 OB << "\\\\";
1199 return;
1200 case '\a': // bell
1201 OB << "\\a";
1202 return;
1203 case '\b': // backspace
1204 OB << "\\b";
1205 return;
1206 case '\f': // form feed
1207 OB << "\\f";
1208 return;
1209 case '\n': // new line
1210 OB << "\\n";
1211 return;
1212 case '\r': // carriage return
1213 OB << "\\r";
1214 return;
1215 case '\t': // tab
1216 OB << "\\t";
1217 return;
1218 case '\v': // vertical tab
1219 OB << "\\v";
1220 return;
1221 default:
1222 break;
1223 }
1224
1225 if (C > 0x1F && C < 0x7F) {
1226 // Standard ascii char.
1227 OB << (char)C;
1228 return;
1229 }
1230
1231 outputHex(OB, C);
1232}
1233
1234static unsigned countTrailingNullBytes(const uint8_t *StringBytes, int Length) {
1235 const uint8_t *End = StringBytes + Length - 1;
1236 unsigned Count = 0;
1237 while (Length > 0 && *End == 0) {
1238 --Length;
1239 --End;
1240 ++Count;
1241 }
1242 return Count;
1243}
1244
1245static unsigned countEmbeddedNulls(const uint8_t *StringBytes,
1246 unsigned Length) {
1247 unsigned Result = 0;
1248 for (unsigned I = 0; I < Length; ++I) {
1249 if (*StringBytes++ == 0)
1250 ++Result;
1251 }
1252 return Result;
1253}
1254
1255// A mangled (non-wide) string literal stores the total length of the string it
1256// refers to (passed in NumBytes), and it contains up to 32 bytes of actual text
1257// (passed in StringBytes, NumChars).
1258static unsigned guessCharByteSize(const uint8_t *StringBytes, unsigned NumChars,
1259 uint64_t NumBytes) {
1260 assert(NumBytes > 0);
1261
1262 // If the number of bytes is odd, this is guaranteed to be a char string.
1263 if (NumBytes % 2 == 1)
1264 return 1;
1265
1266 // All strings can encode at most 32 bytes of data. If it's less than that,
1267 // then we encoded the entire string. In this case we check for a 1-byte,
1268 // 2-byte, or 4-byte null terminator.
1269 if (NumBytes < 32) {
1270 unsigned TrailingNulls = countTrailingNullBytes(StringBytes, NumChars);
1271 if (TrailingNulls >= 4 && NumBytes % 4 == 0)
1272 return 4;
1273 if (TrailingNulls >= 2)
1274 return 2;
1275 return 1;
1276 }
1277
1278 // The whole string was not able to be encoded. Try to look at embedded null
1279 // terminators to guess. The heuristic is that we count all embedded null
1280 // terminators. If more than 2/3 are null, it's a char32. If more than 1/3
1281 // are null, it's a char16. Otherwise it's a char8. This obviously isn't
1282 // perfect and is biased towards languages that have ascii alphabets, but this
1283 // was always going to be best effort since the encoding is lossy.
1284 unsigned Nulls = countEmbeddedNulls(StringBytes, NumChars);
1285 if (Nulls >= 2 * NumChars / 3 && NumBytes % 4 == 0)
1286 return 4;
1287 if (Nulls >= NumChars / 3)
1288 return 2;
1289 return 1;
1290}
1291
1292static unsigned decodeMultiByteChar(const uint8_t *StringBytes,
1293 unsigned CharIndex, unsigned CharBytes) {
1294 assert(CharBytes == 1 || CharBytes == 2 || CharBytes == 4);
1295 unsigned Offset = CharIndex * CharBytes;
1296 unsigned Result = 0;
1297 StringBytes = StringBytes + Offset;
1298 for (unsigned I = 0; I < CharBytes; ++I) {
1299 unsigned C = static_cast<unsigned>(StringBytes[I]);
1300 Result |= C << (8 * I);
1301 }
1302 return Result;
1303}
1304
1306Demangler::demangleVcallThunkNode(std::string_view &MangledName) {
1309 FSN->Signature = Arena.alloc<ThunkSignatureNode>();
1311
1312 FSN->Name = demangleNameScopeChain(MangledName, VTIN);
1313 if (!Error)
1314 Error = !consumeFront(MangledName, "$B");
1315 if (!Error)
1316 VTIN->OffsetInVTable = demangleUnsigned(MangledName);
1317 if (!Error)
1318 Error = !consumeFront(MangledName, 'A');
1319 if (!Error)
1320 FSN->Signature->CallConvention = demangleCallingConvention(MangledName);
1321 return (Error) ? nullptr : FSN;
1322}
1323
1325Demangler::demangleStringLiteral(std::string_view &MangledName) {
1326 // This function uses goto, so declare all variables up front.
1328 std::string_view CRC;
1329 uint64_t StringByteSize;
1330 bool IsWcharT = false;
1331 bool IsNegative = false;
1332 size_t CrcEndPos = 0;
1333 char F;
1334
1336
1337 // Prefix indicating the beginning of a string literal
1338 if (!consumeFront(MangledName, "@_"))
1339 goto StringLiteralError;
1340 if (MangledName.empty())
1341 goto StringLiteralError;
1342
1343 // Char Type (regular or wchar_t)
1344 F = MangledName.front();
1345 MangledName.remove_prefix(1);
1346 switch (F) {
1347 case '1':
1348 IsWcharT = true;
1350 case '0':
1351 break;
1352 default:
1353 goto StringLiteralError;
1354 }
1355
1356 // Encoded Length
1357 std::tie(StringByteSize, IsNegative) = demangleNumber(MangledName);
1358 if (Error || IsNegative || StringByteSize < (IsWcharT ? 2 : 1))
1359 goto StringLiteralError;
1360
1361 // CRC 32 (always 8 characters plus a terminator)
1362 CrcEndPos = MangledName.find('@');
1363 if (CrcEndPos == std::string_view::npos)
1364 goto StringLiteralError;
1365 CRC = MangledName.substr(0, CrcEndPos);
1366 MangledName.remove_prefix(CrcEndPos + 1);
1367 if (MangledName.empty())
1368 goto StringLiteralError;
1369
1370 if (IsWcharT) {
1371 Result->Char = CharKind::Wchar;
1372 if (StringByteSize > 64)
1373 Result->IsTruncated = true;
1374
1375 while (!consumeFront(MangledName, '@')) {
1376 if (MangledName.size() < 2)
1377 goto StringLiteralError;
1378 wchar_t W = demangleWcharLiteral(MangledName);
1379 if (StringByteSize != 2 || Result->IsTruncated)
1380 outputEscapedChar(OB, W);
1381 StringByteSize -= 2;
1382 if (Error)
1383 goto StringLiteralError;
1384 }
1385 } else {
1386 // The max byte length is actually 32, but some compilers mangled strings
1387 // incorrectly, so we have to assume it can go higher.
1388 constexpr unsigned MaxStringByteLength = 32 * 4;
1389 uint8_t StringBytes[MaxStringByteLength];
1390
1391 unsigned BytesDecoded = 0;
1392 while (!consumeFront(MangledName, '@')) {
1393 if (MangledName.size() < 1 || BytesDecoded >= MaxStringByteLength)
1394 goto StringLiteralError;
1395 StringBytes[BytesDecoded++] = demangleCharLiteral(MangledName);
1396 }
1397
1398 if (StringByteSize > BytesDecoded)
1399 Result->IsTruncated = true;
1400
1401 unsigned CharBytes =
1402 guessCharByteSize(StringBytes, BytesDecoded, StringByteSize);
1403 assert(StringByteSize % CharBytes == 0);
1404 switch (CharBytes) {
1405 case 1:
1406 Result->Char = CharKind::Char;
1407 break;
1408 case 2:
1409 Result->Char = CharKind::Char16;
1410 break;
1411 case 4:
1412 Result->Char = CharKind::Char32;
1413 break;
1414 default:
1416 }
1417 const unsigned NumChars = BytesDecoded / CharBytes;
1418 for (unsigned CharIndex = 0; CharIndex < NumChars; ++CharIndex) {
1419 unsigned NextChar =
1420 decodeMultiByteChar(StringBytes, CharIndex, CharBytes);
1421 if (CharIndex + 1 < NumChars || Result->IsTruncated)
1422 outputEscapedChar(OB, NextChar);
1423 }
1424 }
1425
1426 Result->DecodedString = copyString(OB);
1427 std::free(OB.getBuffer());
1428 return Result;
1429
1430StringLiteralError:
1431 Error = true;
1432 std::free(OB.getBuffer());
1433 return nullptr;
1434}
1435
1436// Returns MangledName's prefix before the first '@', or an error if
1437// MangledName contains no '@' or the prefix has length 0.
1438std::string_view Demangler::demangleSimpleString(std::string_view &MangledName,
1439 bool Memorize) {
1440 std::string_view S;
1441 for (size_t i = 0; i < MangledName.size(); ++i) {
1442 if (MangledName[i] != '@')
1443 continue;
1444 if (i == 0)
1445 break;
1446 S = MangledName.substr(0, i);
1447 MangledName.remove_prefix(i + 1);
1448
1449 if (Memorize)
1450 memorizeString(S);
1451 return S;
1452 }
1453
1454 Error = true;
1455 return {};
1456}
1457
1459Demangler::demangleAnonymousNamespaceName(std::string_view &MangledName) {
1460 assert(llvm::itanium_demangle::starts_with(MangledName, "?A"));
1461 consumeFront(MangledName, "?A");
1462
1464 Node->Name = "`anonymous namespace'";
1465 size_t EndPos = MangledName.find('@');
1466 if (EndPos == std::string_view::npos) {
1467 Error = true;
1468 return nullptr;
1469 }
1470 std::string_view NamespaceKey = MangledName.substr(0, EndPos);
1471 memorizeString(NamespaceKey);
1472 MangledName = MangledName.substr(EndPos + 1);
1473 return Node;
1474}
1475
1477Demangler::demangleLocallyScopedNamePiece(std::string_view &MangledName) {
1478 assert(startsWithLocalScopePattern(MangledName));
1479
1481 consumeFront(MangledName, '?');
1482 uint64_t Number = 0;
1483 bool IsNegative = false;
1484 std::tie(Number, IsNegative) = demangleNumber(MangledName);
1485 assert(!IsNegative);
1486
1487 // One ? to terminate the number
1488 consumeFront(MangledName, '?');
1489
1490 assert(!Error);
1491 Node *Scope = parse(MangledName);
1492 if (Error)
1493 return nullptr;
1494
1495 // Render the parent symbol's name into a buffer.
1497 OB << '`';
1498 Scope->output(OB, OF_Default);
1499 OB << '\'';
1500 OB << "::`" << Number << "'";
1501
1502 Identifier->Name = copyString(OB);
1503 std::free(OB.getBuffer());
1504 return Identifier;
1505}
1506
1507// Parses a type name in the form of A@B@C@@ which represents C::B::A.
1509Demangler::demangleFullyQualifiedTypeName(std::string_view &MangledName) {
1511 demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1512 if (Error)
1513 return nullptr;
1515
1516 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1517 if (Error)
1518 return nullptr;
1519 assert(QN);
1520 return QN;
1521}
1522
1523// Parses a symbol name in the form of A@B@C@@ which represents C::B::A.
1524// Symbol names have slightly different rules regarding what can appear
1525// so we separate out the implementations for flexibility.
1527Demangler::demangleFullyQualifiedSymbolName(std::string_view &MangledName) {
1528 // This is the final component of a symbol name (i.e. the leftmost component
1529 // of a mangled name. Since the only possible template instantiation that
1530 // can appear in this context is a function template, and since those are
1531 // not saved for the purposes of name backreferences, only backref simple
1532 // names.
1534 demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
1535 if (Error)
1536 return nullptr;
1537
1538 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1539 if (Error)
1540 return nullptr;
1541
1543 if (QN->Components->Count < 2) {
1544 Error = true;
1545 return nullptr;
1546 }
1548 static_cast<StructorIdentifierNode *>(Identifier);
1549 Node *ClassNode = QN->Components->Nodes[QN->Components->Count - 2];
1550 SIN->Class = static_cast<IdentifierNode *>(ClassNode);
1551 }
1552 assert(QN);
1553 return QN;
1554}
1555
1557Demangler::demangleUnqualifiedTypeName(std::string_view &MangledName,
1558 bool Memorize) {
1559 // An inner-most name can be a back-reference, because a fully-qualified name
1560 // (e.g. Scope + Inner) can contain other fully qualified names inside of
1561 // them (for example template parameters), and these nested parameters can
1562 // refer to previously mangled types.
1563 if (startsWithDigit(MangledName))
1564 return demangleBackRefName(MangledName);
1565
1566 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1567 return demangleTemplateInstantiationName(MangledName, NBB_Template);
1568
1569 return demangleSimpleName(MangledName, Memorize);
1570}
1571
1573Demangler::demangleUnqualifiedSymbolName(std::string_view &MangledName,
1574 NameBackrefBehavior NBB) {
1575 if (startsWithDigit(MangledName))
1576 return demangleBackRefName(MangledName);
1577 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1578 return demangleTemplateInstantiationName(MangledName, NBB);
1579 if (llvm::itanium_demangle::starts_with(MangledName, '?'))
1580 return demangleFunctionIdentifierCode(MangledName);
1581 return demangleSimpleName(MangledName, /*Memorize=*/(NBB & NBB_Simple) != 0);
1582}
1583
1585Demangler::demangleNameScopePiece(std::string_view &MangledName) {
1586 if (startsWithDigit(MangledName))
1587 return demangleBackRefName(MangledName);
1588
1589 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1590 return demangleTemplateInstantiationName(MangledName, NBB_Template);
1591
1592 if (llvm::itanium_demangle::starts_with(MangledName, "?A"))
1593 return demangleAnonymousNamespaceName(MangledName);
1594
1595 if (startsWithLocalScopePattern(MangledName))
1596 return demangleLocallyScopedNamePiece(MangledName);
1597
1598 return demangleSimpleName(MangledName, /*Memorize=*/true);
1599}
1600
1602 size_t Count) {
1603 NodeArrayNode *N = Arena.alloc<NodeArrayNode>();
1604 N->Count = Count;
1605 N->Nodes = Arena.allocArray<Node *>(Count);
1606 for (size_t I = 0; I < Count; ++I) {
1607 N->Nodes[I] = Head->N;
1608 Head = Head->Next;
1609 }
1610 return N;
1611}
1612
1614Demangler::demangleNameScopeChain(std::string_view &MangledName,
1615 IdentifierNode *UnqualifiedName) {
1616 NodeList *Head = Arena.alloc<NodeList>();
1617
1618 Head->N = UnqualifiedName;
1619
1620 size_t Count = 1;
1621 while (!consumeFront(MangledName, "@")) {
1622 ++Count;
1623 NodeList *NewHead = Arena.alloc<NodeList>();
1624 NewHead->Next = Head;
1625 Head = NewHead;
1626
1627 if (MangledName.empty()) {
1628 Error = true;
1629 return nullptr;
1630 }
1631
1632 assert(!Error);
1633 IdentifierNode *Elem = demangleNameScopePiece(MangledName);
1634 if (Error)
1635 return nullptr;
1636
1637 Head->N = Elem;
1638 }
1639
1641 QN->Components = nodeListToNodeArray(Arena, Head, Count);
1642 return QN;
1643}
1644
1645FuncClass Demangler::demangleFunctionClass(std::string_view &MangledName) {
1646 const char F = MangledName.front();
1647 MangledName.remove_prefix(1);
1648 switch (F) {
1649 case '9':
1651 case 'A':
1652 return FC_Private;
1653 case 'B':
1654 return FuncClass(FC_Private | FC_Far);
1655 case 'C':
1656 return FuncClass(FC_Private | FC_Static);
1657 case 'D':
1659 case 'E':
1661 case 'F':
1663 case 'G':
1665 case 'H':
1667 case 'I':
1668 return FuncClass(FC_Protected);
1669 case 'J':
1670 return FuncClass(FC_Protected | FC_Far);
1671 case 'K':
1673 case 'L':
1675 case 'M':
1677 case 'N':
1679 case 'O':
1681 case 'P':
1683 case 'Q':
1684 return FuncClass(FC_Public);
1685 case 'R':
1686 return FuncClass(FC_Public | FC_Far);
1687 case 'S':
1688 return FuncClass(FC_Public | FC_Static);
1689 case 'T':
1690 return FuncClass(FC_Public | FC_Static | FC_Far);
1691 case 'U':
1692 return FuncClass(FC_Public | FC_Virtual);
1693 case 'V':
1695 case 'W':
1697 case 'X':
1699 case 'Y':
1700 return FuncClass(FC_Global);
1701 case 'Z':
1702 return FuncClass(FC_Global | FC_Far);
1703 case '$': {
1705 if (consumeFront(MangledName, 'R'))
1706 VFlag = FuncClass(VFlag | FC_VirtualThisAdjustEx);
1707 if (MangledName.empty())
1708 break;
1709 const char F = MangledName.front();
1710 MangledName.remove_prefix(1);
1711 switch (F) {
1712 case '0':
1713 return FuncClass(FC_Private | FC_Virtual | VFlag);
1714 case '1':
1715 return FuncClass(FC_Private | FC_Virtual | VFlag | FC_Far);
1716 case '2':
1717 return FuncClass(FC_Protected | FC_Virtual | VFlag);
1718 case '3':
1719 return FuncClass(FC_Protected | FC_Virtual | VFlag | FC_Far);
1720 case '4':
1721 return FuncClass(FC_Public | FC_Virtual | VFlag);
1722 case '5':
1723 return FuncClass(FC_Public | FC_Virtual | VFlag | FC_Far);
1724 }
1725 }
1726 }
1727
1728 Error = true;
1729 return FC_Public;
1730}
1731
1733Demangler::demangleCallingConvention(std::string_view &MangledName) {
1734 if (MangledName.empty()) {
1735 Error = true;
1736 return CallingConv::None;
1737 }
1738
1739 const char F = MangledName.front();
1740 MangledName.remove_prefix(1);
1741 switch (F) {
1742 case 'A':
1743 case 'B':
1744 return CallingConv::Cdecl;
1745 case 'C':
1746 case 'D':
1747 return CallingConv::Pascal;
1748 case 'E':
1749 case 'F':
1750 return CallingConv::Thiscall;
1751 case 'G':
1752 case 'H':
1753 return CallingConv::Stdcall;
1754 case 'I':
1755 case 'J':
1756 return CallingConv::Fastcall;
1757 case 'M':
1758 case 'N':
1759 return CallingConv::Clrcall;
1760 case 'O':
1761 case 'P':
1762 return CallingConv::Eabi;
1763 case 'Q':
1765 case 'S':
1766 return CallingConv::Swift;
1767 case 'W':
1769 }
1770
1771 return CallingConv::None;
1772}
1773
1775Demangler::demangleVariableStorageClass(std::string_view &MangledName) {
1776 assert(MangledName.front() >= '0' && MangledName.front() <= '4');
1777
1778 const char F = MangledName.front();
1779 MangledName.remove_prefix(1);
1780 switch (F) {
1781 case '0':
1783 case '1':
1785 case '2':
1787 case '3':
1788 return StorageClass::Global;
1789 case '4':
1791 }
1793}
1794
1795std::pair<Qualifiers, bool>
1796Demangler::demangleQualifiers(std::string_view &MangledName) {
1797 if (MangledName.empty()) {
1798 Error = true;
1799 return std::make_pair(Q_None, false);
1800 }
1801
1802 const char F = MangledName.front();
1803 MangledName.remove_prefix(1);
1804 switch (F) {
1805 // Member qualifiers
1806 case 'Q':
1807 return std::make_pair(Q_None, true);
1808 case 'R':
1809 return std::make_pair(Q_Const, true);
1810 case 'S':
1811 return std::make_pair(Q_Volatile, true);
1812 case 'T':
1813 return std::make_pair(Qualifiers(Q_Const | Q_Volatile), true);
1814 // Non-Member qualifiers
1815 case 'A':
1816 return std::make_pair(Q_None, false);
1817 case 'B':
1818 return std::make_pair(Q_Const, false);
1819 case 'C':
1820 return std::make_pair(Q_Volatile, false);
1821 case 'D':
1822 return std::make_pair(Qualifiers(Q_Const | Q_Volatile), false);
1823 }
1824 Error = true;
1825 return std::make_pair(Q_None, false);
1826}
1827
1828// <variable-type> ::= <type> <cvr-qualifiers>
1829// ::= <type> <pointee-cvr-qualifiers> # pointers, references
1830TypeNode *Demangler::demangleType(std::string_view &MangledName,
1831 QualifierMangleMode QMM) {
1832 Qualifiers Quals = Q_None;
1833 bool IsMember = false;
1834 if (QMM == QualifierMangleMode::Mangle) {
1835 std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1836 } else if (QMM == QualifierMangleMode::Result) {
1837 if (consumeFront(MangledName, '?'))
1838 std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1839 }
1840
1841 if (MangledName.empty()) {
1842 Error = true;
1843 return nullptr;
1844 }
1845
1846 TypeNode *Ty = nullptr;
1847 if (isTagType(MangledName))
1848 Ty = demangleClassType(MangledName);
1849 else if (isPointerType(MangledName)) {
1850 if (isMemberPointer(MangledName, Error))
1851 Ty = demangleMemberPointerType(MangledName);
1852 else if (!Error)
1853 Ty = demanglePointerType(MangledName);
1854 else
1855 return nullptr;
1856 } else if (isArrayType(MangledName))
1857 Ty = demangleArrayType(MangledName);
1858 else if (isFunctionType(MangledName)) {
1859 if (consumeFront(MangledName, "$$A8@@"))
1860 Ty = demangleFunctionType(MangledName, true);
1861 else {
1862 assert(llvm::itanium_demangle::starts_with(MangledName, "$$A6"));
1863 consumeFront(MangledName, "$$A6");
1864 Ty = demangleFunctionType(MangledName, false);
1865 }
1866 } else if (isCustomType(MangledName)) {
1867 Ty = demangleCustomType(MangledName);
1868 } else {
1869 Ty = demanglePrimitiveType(MangledName);
1870 }
1871
1872 if (!Ty || Error)
1873 return Ty;
1874 Ty->Quals = Qualifiers(Ty->Quals | Quals);
1875 return Ty;
1876}
1877
1878bool Demangler::demangleThrowSpecification(std::string_view &MangledName) {
1879 if (consumeFront(MangledName, "_E"))
1880 return true;
1881 if (consumeFront(MangledName, 'Z'))
1882 return false;
1883
1884 Error = true;
1885 return false;
1886}
1887
1889Demangler::demangleFunctionType(std::string_view &MangledName,
1890 bool HasThisQuals) {
1892
1893 if (HasThisQuals) {
1894 FTy->Quals = demanglePointerExtQualifiers(MangledName);
1895 FTy->RefQualifier = demangleFunctionRefQualifier(MangledName);
1896 FTy->Quals = Qualifiers(FTy->Quals | demangleQualifiers(MangledName).first);
1897 }
1898
1899 // Fields that appear on both member and non-member functions.
1900 FTy->CallConvention = demangleCallingConvention(MangledName);
1901
1902 // <return-type> ::= <type>
1903 // ::= @ # structors (they have no declared return type)
1904 bool IsStructor = consumeFront(MangledName, '@');
1905 if (!IsStructor)
1906 FTy->ReturnType = demangleType(MangledName, QualifierMangleMode::Result);
1907
1908 FTy->Params = demangleFunctionParameterList(MangledName, FTy->IsVariadic);
1909
1910 FTy->IsNoexcept = demangleThrowSpecification(MangledName);
1911
1912 return FTy;
1913}
1914
1916Demangler::demangleFunctionEncoding(std::string_view &MangledName) {
1917 FuncClass ExtraFlags = FC_None;
1918 if (consumeFront(MangledName, "$$J0"))
1919 ExtraFlags = FC_ExternC;
1920
1921 if (MangledName.empty()) {
1922 Error = true;
1923 return nullptr;
1924 }
1925
1926 FuncClass FC = demangleFunctionClass(MangledName);
1927 FC = FuncClass(ExtraFlags | FC);
1928
1929 FunctionSignatureNode *FSN = nullptr;
1930 ThunkSignatureNode *TTN = nullptr;
1931 if (FC & FC_StaticThisAdjust) {
1932 TTN = Arena.alloc<ThunkSignatureNode>();
1933 TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1934 } else if (FC & FC_VirtualThisAdjust) {
1935 TTN = Arena.alloc<ThunkSignatureNode>();
1936 if (FC & FC_VirtualThisAdjustEx) {
1937 TTN->ThisAdjust.VBPtrOffset = demangleSigned(MangledName);
1938 TTN->ThisAdjust.VBOffsetOffset = demangleSigned(MangledName);
1939 }
1940 TTN->ThisAdjust.VtordispOffset = demangleSigned(MangledName);
1941 TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1942 }
1943
1944 if (FC & FC_NoParameterList) {
1945 // This is an extern "C" function whose full signature hasn't been mangled.
1946 // This happens when we need to mangle a local symbol inside of an extern
1947 // "C" function.
1948 FSN = Arena.alloc<FunctionSignatureNode>();
1949 } else {
1950 bool HasThisQuals = !(FC & (FC_Global | FC_Static));
1951 FSN = demangleFunctionType(MangledName, HasThisQuals);
1952 }
1953
1954 if (Error)
1955 return nullptr;
1956
1957 if (TTN) {
1958 *static_cast<FunctionSignatureNode *>(TTN) = *FSN;
1959 FSN = TTN;
1960 }
1961 FSN->FunctionClass = FC;
1962
1964 Symbol->Signature = FSN;
1965 return Symbol;
1966}
1967
1968CustomTypeNode *Demangler::demangleCustomType(std::string_view &MangledName) {
1969 assert(llvm::itanium_demangle::starts_with(MangledName, '?'));
1970 MangledName.remove_prefix(1);
1971
1972 CustomTypeNode *CTN = Arena.alloc<CustomTypeNode>();
1973 CTN->Identifier = demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1974 if (!consumeFront(MangledName, '@'))
1975 Error = true;
1976 if (Error)
1977 return nullptr;
1978 return CTN;
1979}
1980
1981// Reads a primitive type.
1983Demangler::demanglePrimitiveType(std::string_view &MangledName) {
1984 if (consumeFront(MangledName, "$$T"))
1986
1987 const char F = MangledName.front();
1988 MangledName.remove_prefix(1);
1989 switch (F) {
1990 case 'X':
1992 case 'D':
1994 case 'C':
1996 case 'E':
1998 case 'F':
2000 case 'G':
2002 case 'H':
2004 case 'I':
2006 case 'J':
2008 case 'K':
2010 case 'M':
2012 case 'N':
2014 case 'O':
2016 case '_': {
2017 if (MangledName.empty()) {
2018 Error = true;
2019 return nullptr;
2020 }
2021 const char F = MangledName.front();
2022 MangledName.remove_prefix(1);
2023 switch (F) {
2024 case 'N':
2026 case 'J':
2028 case 'K':
2030 case 'W':
2032 case 'Q':
2034 case 'S':
2036 case 'U':
2038 }
2039 break;
2040 }
2041 }
2042 Error = true;
2043 return nullptr;
2044}
2045
2046TagTypeNode *Demangler::demangleClassType(std::string_view &MangledName) {
2047 TagTypeNode *TT = nullptr;
2048
2049 const char F = MangledName.front();
2050 MangledName.remove_prefix(1);
2051 switch (F) {
2052 case 'T':
2054 break;
2055 case 'U':
2057 break;
2058 case 'V':
2060 break;
2061 case 'W':
2062 if (!consumeFront(MangledName, '4')) {
2063 Error = true;
2064 return nullptr;
2065 }
2067 break;
2068 default:
2069 assert(false);
2070 }
2071
2072 TT->QualifiedName = demangleFullyQualifiedTypeName(MangledName);
2073 return TT;
2074}
2075
2076// <pointer-type> ::= E? <pointer-cvr-qualifiers> <ext-qualifiers> <type>
2077// # the E is required for 64-bit non-static pointers
2078PointerTypeNode *Demangler::demanglePointerType(std::string_view &MangledName) {
2080
2081 std::tie(Pointer->Quals, Pointer->Affinity) =
2082 demanglePointerCVQualifiers(MangledName);
2083
2084 if (consumeFront(MangledName, "6")) {
2085 Pointer->Pointee = demangleFunctionType(MangledName, false);
2086 return Pointer;
2087 }
2088
2089 Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
2090 Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
2091
2092 Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Mangle);
2093 return Pointer;
2094}
2095
2097Demangler::demangleMemberPointerType(std::string_view &MangledName) {
2099
2100 std::tie(Pointer->Quals, Pointer->Affinity) =
2101 demanglePointerCVQualifiers(MangledName);
2103
2104 Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
2105 Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
2106
2107 // isMemberPointer() only returns true if there is at least one character
2108 // after the qualifiers.
2109 if (consumeFront(MangledName, "8")) {
2110 Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
2111 Pointer->Pointee = demangleFunctionType(MangledName, true);
2112 } else {
2113 Qualifiers PointeeQuals = Q_None;
2114 bool IsMember = false;
2115 std::tie(PointeeQuals, IsMember) = demangleQualifiers(MangledName);
2116 assert(IsMember || Error);
2117 Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
2118
2119 Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Drop);
2120 if (Pointer->Pointee)
2121 Pointer->Pointee->Quals = PointeeQuals;
2122 }
2123
2124 return Pointer;
2125}
2126
2128Demangler::demanglePointerExtQualifiers(std::string_view &MangledName) {
2129 Qualifiers Quals = Q_None;
2130 if (consumeFront(MangledName, 'E'))
2131 Quals = Qualifiers(Quals | Q_Pointer64);
2132 if (consumeFront(MangledName, 'I'))
2133 Quals = Qualifiers(Quals | Q_Restrict);
2134 if (consumeFront(MangledName, 'F'))
2135 Quals = Qualifiers(Quals | Q_Unaligned);
2136
2137 return Quals;
2138}
2139
2140ArrayTypeNode *Demangler::demangleArrayType(std::string_view &MangledName) {
2141 assert(MangledName.front() == 'Y');
2142 MangledName.remove_prefix(1);
2143
2144 uint64_t Rank = 0;
2145 bool IsNegative = false;
2146 std::tie(Rank, IsNegative) = demangleNumber(MangledName);
2147 if (IsNegative || Rank == 0) {
2148 Error = true;
2149 return nullptr;
2150 }
2151
2152 ArrayTypeNode *ATy = Arena.alloc<ArrayTypeNode>();
2153 NodeList *Head = Arena.alloc<NodeList>();
2154 NodeList *Tail = Head;
2155
2156 for (uint64_t I = 0; I < Rank; ++I) {
2157 uint64_t D = 0;
2158 std::tie(D, IsNegative) = demangleNumber(MangledName);
2159 if (Error || IsNegative) {
2160 Error = true;
2161 return nullptr;
2162 }
2163 Tail->N = Arena.alloc<IntegerLiteralNode>(D, IsNegative);
2164 if (I + 1 < Rank) {
2165 Tail->Next = Arena.alloc<NodeList>();
2166 Tail = Tail->Next;
2167 }
2168 }
2169 ATy->Dimensions = nodeListToNodeArray(Arena, Head, Rank);
2170
2171 if (consumeFront(MangledName, "$$C")) {
2172 bool IsMember = false;
2173 std::tie(ATy->Quals, IsMember) = demangleQualifiers(MangledName);
2174 if (IsMember) {
2175 Error = true;
2176 return nullptr;
2177 }
2178 }
2179
2180 ATy->ElementType = demangleType(MangledName, QualifierMangleMode::Drop);
2181 return ATy;
2182}
2183
2184// Reads a function's parameters.
2186Demangler::demangleFunctionParameterList(std::string_view &MangledName,
2187 bool &IsVariadic) {
2188 // Empty parameter list.
2189 if (consumeFront(MangledName, 'X'))
2190 return nullptr;
2191
2192 NodeList *Head = Arena.alloc<NodeList>();
2193 NodeList **Current = &Head;
2194 size_t Count = 0;
2195 while (!Error && !llvm::itanium_demangle::starts_with(MangledName, '@') &&
2196 !llvm::itanium_demangle::starts_with(MangledName, 'Z')) {
2197 ++Count;
2198
2199 if (startsWithDigit(MangledName)) {
2200 size_t N = MangledName[0] - '0';
2201 if (N >= Backrefs.FunctionParamCount) {
2202 Error = true;
2203 return nullptr;
2204 }
2205 MangledName.remove_prefix(1);
2206
2207 *Current = Arena.alloc<NodeList>();
2208 (*Current)->N = Backrefs.FunctionParams[N];
2209 Current = &(*Current)->Next;
2210 continue;
2211 }
2212
2213 size_t OldSize = MangledName.size();
2214
2215 *Current = Arena.alloc<NodeList>();
2216 TypeNode *TN = demangleType(MangledName, QualifierMangleMode::Drop);
2217 if (!TN || Error)
2218 return nullptr;
2219
2220 (*Current)->N = TN;
2221
2222 size_t CharsConsumed = OldSize - MangledName.size();
2223 assert(CharsConsumed != 0);
2224
2225 // Single-letter types are ignored for backreferences because memorizing
2226 // them doesn't save anything.
2227 if (Backrefs.FunctionParamCount <= 9 && CharsConsumed > 1)
2228 Backrefs.FunctionParams[Backrefs.FunctionParamCount++] = TN;
2229
2230 Current = &(*Current)->Next;
2231 }
2232
2233 if (Error)
2234 return nullptr;
2235
2236 NodeArrayNode *NA = nodeListToNodeArray(Arena, Head, Count);
2237 // A non-empty parameter list is terminated by either 'Z' (variadic) parameter
2238 // list or '@' (non variadic). Careful not to consume "@Z", as in that case
2239 // the following Z could be a throw specifier.
2240 if (consumeFront(MangledName, '@'))
2241 return NA;
2242
2243 if (consumeFront(MangledName, 'Z')) {
2244 IsVariadic = true;
2245 return NA;
2246 }
2247
2249}
2250
2252Demangler::demangleTemplateParameterList(std::string_view &MangledName) {
2253 NodeList *Head = nullptr;
2254 NodeList **Current = &Head;
2255 size_t Count = 0;
2256
2257 while (!llvm::itanium_demangle::starts_with(MangledName, '@')) {
2258 if (consumeFront(MangledName, "$S") || consumeFront(MangledName, "$$V") ||
2259 consumeFront(MangledName, "$$$V") || consumeFront(MangledName, "$$Z")) {
2260 // parameter pack separator
2261 continue;
2262 }
2263
2264 ++Count;
2265
2266 // Template parameter lists don't participate in back-referencing.
2267 *Current = Arena.alloc<NodeList>();
2268
2269 NodeList &TP = **Current;
2270
2271 // <auto-nttp> ::= $ M <type> <nttp>
2272 const bool IsAutoNTTP = consumeFront(MangledName, "$M");
2273 if (IsAutoNTTP) {
2274 // The deduced type of the auto NTTP parameter isn't printed so
2275 // we want to ignore the AST created from demangling the type.
2276 //
2277 // TODO: Avoid the extra allocations to the bump allocator in this case.
2278 (void)demangleType(MangledName, QualifierMangleMode::Drop);
2279 if (Error)
2280 return nullptr;
2281 }
2282
2283 TemplateParameterReferenceNode *TPRN = nullptr;
2284 if (consumeFront(MangledName, "$$Y")) {
2285 // Template alias
2286 TP.N = demangleFullyQualifiedTypeName(MangledName);
2287 } else if (consumeFront(MangledName, "$$B")) {
2288 // Array
2289 TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2290 } else if (consumeFront(MangledName, "$$C")) {
2291 // Type has qualifiers.
2292 TP.N = demangleType(MangledName, QualifierMangleMode::Mangle);
2293 } else if (startsWith(MangledName, "$1", "1", !IsAutoNTTP) ||
2294 startsWith(MangledName, "$H", "H", !IsAutoNTTP) ||
2295 startsWith(MangledName, "$I", "I", !IsAutoNTTP) ||
2296 startsWith(MangledName, "$J", "J", !IsAutoNTTP)) {
2297 // Pointer to member
2298 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2299 TPRN->IsMemberPointer = true;
2300
2301 if (!IsAutoNTTP)
2302 MangledName.remove_prefix(1); // Remove leading '$'
2303
2304 // 1 - single inheritance <name>
2305 // H - multiple inheritance <name> <number>
2306 // I - virtual inheritance <name> <number> <number>
2307 // J - unspecified inheritance <name> <number> <number> <number>
2308 char InheritanceSpecifier = MangledName.front();
2309 MangledName.remove_prefix(1);
2310 SymbolNode *S = nullptr;
2311 if (llvm::itanium_demangle::starts_with(MangledName, '?')) {
2312 S = parse(MangledName);
2313 if (Error || !S->Name) {
2314 Error = true;
2315 return nullptr;
2316 }
2317 memorizeIdentifier(S->Name->getUnqualifiedIdentifier());
2318 }
2319
2320 switch (InheritanceSpecifier) {
2321 case 'J':
2322 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2323 demangleSigned(MangledName);
2325 case 'I':
2326 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2327 demangleSigned(MangledName);
2329 case 'H':
2330 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2331 demangleSigned(MangledName);
2333 case '1':
2334 break;
2335 default:
2337 }
2339 TPRN->Symbol = S;
2340 } else if (llvm::itanium_demangle::starts_with(MangledName, "$E?")) {
2341 consumeFront(MangledName, "$E");
2342 // Reference to symbol
2343 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2344 TPRN->Symbol = parse(MangledName);
2346 } else if (llvm::itanium_demangle::starts_with(MangledName, "$F") ||
2347 llvm::itanium_demangle::starts_with(MangledName, "$G")) {
2348 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2349
2350 // Data member pointer.
2351 MangledName.remove_prefix(1);
2352 char InheritanceSpecifier = MangledName.front();
2353 MangledName.remove_prefix(1);
2354
2355 switch (InheritanceSpecifier) {
2356 case 'G':
2357 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2358 demangleSigned(MangledName);
2360 case 'F':
2361 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2362 demangleSigned(MangledName);
2363 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2364 demangleSigned(MangledName);
2365 break;
2366 default:
2368 }
2369 TPRN->IsMemberPointer = true;
2370
2371 } else if (consumeFront(MangledName, "$0", "0", !IsAutoNTTP)) {
2372 // Integral non-type template parameter
2373 bool IsNegative = false;
2374 uint64_t Value = 0;
2375 std::tie(Value, IsNegative) = demangleNumber(MangledName);
2376
2377 TP.N = Arena.alloc<IntegerLiteralNode>(Value, IsNegative);
2378 } else {
2379 TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2380 }
2381 if (Error)
2382 return nullptr;
2383
2384 Current = &TP.Next;
2385 }
2386
2387 // The loop above returns nullptr on Error.
2388 assert(!Error);
2389
2390 // Template parameter lists cannot be variadic, so it can only be terminated
2391 // by @ (as opposed to 'Z' in the function parameter case).
2392 assert(llvm::itanium_demangle::starts_with(
2393 MangledName, '@')); // The above loop exits only on '@'.
2394 consumeFront(MangledName, '@');
2395 return nodeListToNodeArray(Arena, Head, Count);
2396}
2397
2398void Demangler::dumpBackReferences() {
2399 std::printf("%d function parameter backreferences\n",
2400 (int)Backrefs.FunctionParamCount);
2401
2402 // Create an output stream so we can render each type.
2403 OutputBuffer OB;
2404 for (size_t I = 0; I < Backrefs.FunctionParamCount; ++I) {
2405 OB.setCurrentPosition(0);
2406
2407 TypeNode *T = Backrefs.FunctionParams[I];
2408 T->output(OB, OF_Default);
2409
2410 std::string_view B = OB;
2411 std::printf(" [%d] - %.*s\n", (int)I, (int)B.size(), B.data());
2412 }
2413 std::free(OB.getBuffer());
2414
2415 if (Backrefs.FunctionParamCount > 0)
2416 std::printf("\n");
2417 std::printf("%d name backreferences\n", (int)Backrefs.NamesCount);
2418 for (size_t I = 0; I < Backrefs.NamesCount; ++I) {
2419 std::printf(" [%d] - %.*s\n", (int)I, (int)Backrefs.Names[I]->Name.size(),
2420 Backrefs.Names[I]->Name.data());
2421 }
2422 if (Backrefs.NamesCount > 0)
2423 std::printf("\n");
2424}
2425
2426char *llvm::microsoftDemangle(std::string_view MangledName, size_t *NMangled,
2427 int *Status, MSDemangleFlags Flags) {
2428 Demangler D;
2429
2430 std::string_view Name{MangledName};
2431 SymbolNode *AST = D.parse(Name);
2432 if (!D.Error && NMangled)
2433 *NMangled = MangledName.size() - Name.size();
2434
2435 if (Flags & MSDF_DumpBackrefs)
2436 D.dumpBackReferences();
2437
2439 if (Flags & MSDF_NoCallingConvention)
2441 if (Flags & MSDF_NoAccessSpecifier)
2443 if (Flags & MSDF_NoReturnType)
2444 OF = OutputFlags(OF | OF_NoReturnType);
2445 if (Flags & MSDF_NoMemberType)
2446 OF = OutputFlags(OF | OF_NoMemberType);
2447 if (Flags & MSDF_NoVariableType)
2448 OF = OutputFlags(OF | OF_NoVariableType);
2449
2450 int InternalStatus = demangle_success;
2451 char *Buf;
2452 if (D.Error)
2453 InternalStatus = demangle_invalid_mangled_name;
2454 else {
2455 OutputBuffer OB;
2456 AST->output(OB, OF);
2457 OB += '\0';
2458 Buf = OB.getBuffer();
2459 }
2460
2461 if (Status)
2462 *Status = InternalStatus;
2463 return InternalStatus == demangle_success ? Buf : nullptr;
2464}
static GCRegistry::Add< OcamlGC > B("ocaml", "ocaml 3.10-compatible GC")
static GCRegistry::Add< ErlangGC > A("erlang", "erlang-compatible garbage collector")
static GCRegistry::Add< StatepointGC > D("statepoint-example", "an example strategy for statepoint")
#define DEMANGLE_FALLTHROUGH
#define DEMANGLE_UNREACHABLE
std::string Name
bool End
Definition: ELF_riscv.cpp:480
Qualifiers
#define F(x, y, z)
Definition: MD5.cpp:55
#define I(x, y, z)
Definition: MD5.cpp:58
static bool startsWithLocalScopePattern(std::string_view S)
static bool isArrayType(std::string_view S)
static unsigned countEmbeddedNulls(const uint8_t *StringBytes, unsigned Length)
static bool startsWithDigit(std::string_view S)
static QualifiedNameNode * synthesizeQualifiedName(ArenaAllocator &Arena, IdentifierNode *Identifier)
static void outputEscapedChar(OutputBuffer &OB, unsigned C)
static bool isCustomType(std::string_view S)
static void outputHex(OutputBuffer &OB, unsigned C)
static std::pair< Qualifiers, PointerAffinity > demanglePointerCVQualifiers(std::string_view &MangledName)
static bool isMemberPointer(std::string_view MangledName, bool &Error)
static VariableSymbolNode * synthesizeVariable(ArenaAllocator &Arena, TypeNode *Type, std::string_view VariableName)
static unsigned decodeMultiByteChar(const uint8_t *StringBytes, unsigned CharIndex, unsigned CharBytes)
static void writeHexDigit(char *Buffer, uint8_t Digit)
static FunctionRefQualifier demangleFunctionRefQualifier(std::string_view &MangledName)
static bool isRebasedHexDigit(char C)
static NodeArrayNode * nodeListToNodeArray(ArenaAllocator &Arena, NodeList *Head, size_t Count)
static uint8_t rebasedHexDigitToNumber(char C)
static unsigned countTrailingNullBytes(const uint8_t *StringBytes, int Length)
static NamedIdentifierNode * synthesizeNamedIdentifier(ArenaAllocator &Arena, std::string_view Name)
static bool startsWith(std::string_view S, std::string_view PrefixA, std::string_view PrefixB, bool A)
static bool consumeFront(std::string_view &S, char C)
static bool isFunctionType(std::string_view S)
static bool isPointerType(std::string_view S)
static unsigned guessCharByteSize(const uint8_t *StringBytes, unsigned NumChars, uint64_t NumBytes)
static SpecialIntrinsicKind consumeSpecialIntrinsicKind(std::string_view &MangledName)
static bool isTagType(std::string_view S)
if(VerifyEach)
#define CH(x, y, z)
Definition: SHA256.cpp:34
assert(ImpDefSCC.getReg()==AMDGPU::SCC &&ImpDefSCC.isDef())
static bool isMemberPointer(uint32_t Attrs)
static int Lookup(ArrayRef< TableEntry > Table, unsigned Opcode)
Lightweight error class with error context and mandatory checking.
Definition: Error.h:160
Definition: MD5.h:41
The instances of the Type class are immutable: once they are created, they are never changed.
Definition: Type.h:45
LLVM Value Representation.
Definition: Value.h:74
T * alloc(Args &&... ConstructorArgs)
char * allocUnalignedBuffer(size_t Size)
#define INT64_MAX
Definition: DataTypes.h:71
@ Tail
Attemps to make calls as fast as possible while guaranteeing that tail call optimization can always b...
Definition: CallingConv.h:76
@ C
The default llvm calling convention, compatible with C.
Definition: CallingConv.h:34
@ SC
CHAIN = SC CHAIN, Imm128 - System call.
@ OB
OB - OneByte - Set if this instruction has a one byte opcode.
Definition: X86BaseInfo.h:748
StorageClass
Definition: XCOFF.h:170
This is an optimization pass for GlobalISel generic memory operations.
Definition: AddressRanges.h:18
@ Offset
Definition: DWP.cpp:480
@ Length
Definition: DWP.cpp:480
@ demangle_success
Definition: Demangle.h:29
@ demangle_invalid_mangled_name
Definition: Demangle.h:27
char * microsoftDemangle(std::string_view mangled_name, size_t *n_read, int *status, MSDemangleFlags Flags=MSDF_None)
Demangles the Microsoft symbol pointed at by mangled_name and returns it.
MSDemangleFlags
Definition: Demangle.h:37
@ MSDF_NoReturnType
Definition: Demangle.h:42
@ MSDF_DumpBackrefs
Definition: Demangle.h:39
@ MSDF_NoMemberType
Definition: Demangle.h:43
@ MSDF_NoVariableType
Definition: Demangle.h:44
@ MSDF_NoCallingConvention
Definition: Demangle.h:41
@ MSDF_NoAccessSpecifier
Definition: Demangle.h:40
void swap(llvm::BitVector &LHS, llvm::BitVector &RHS)
Implement std::swap in terms of BitVector swap.
Definition: BitVector.h:860
#define N
NodeList * Next
NamedIdentifierNode * Names[Max]
void output(OutputBuffer &OB, OutputFlags Flags) const override
Definition: regcomp.c:192