LLVM 22.0.0git
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 <optional>
28#include <string_view>
29#include <tuple>
30
31using namespace llvm;
32using namespace ms_demangle;
33
34static bool startsWithDigit(std::string_view S) {
35 return !S.empty() && std::isdigit(S.front());
36}
37
38struct NodeList {
39 Node *N = nullptr;
40 NodeList *Next = nullptr;
41};
42
43static bool consumeFront(std::string_view &S, char C) {
44 if (!llvm::itanium_demangle::starts_with(S, C))
45 return false;
46 S.remove_prefix(1);
47 return true;
48}
49
50static bool consumeFront(std::string_view &S, std::string_view C) {
51 if (!llvm::itanium_demangle::starts_with(S, C))
52 return false;
53 S.remove_prefix(C.size());
54 return true;
55}
56
57static bool consumeFront(std::string_view &S, std::string_view PrefixA,
58 std::string_view PrefixB, bool A) {
59 const std::string_view &Prefix = A ? PrefixA : PrefixB;
60 return consumeFront(S, Prefix);
61}
62
63static bool startsWith(std::string_view S, std::string_view PrefixA,
64 std::string_view PrefixB, bool A) {
65 const std::string_view &Prefix = A ? PrefixA : PrefixB;
66 return llvm::itanium_demangle::starts_with(S, Prefix);
67}
68
69bool Demangler::isMemberPointer(std::string_view MangledName, bool &Error) {
70 Error = false;
71 const char F = MangledName.front();
72 MangledName.remove_prefix(1);
73 switch (F) {
74 case '$':
75 // This is probably an rvalue reference (e.g. $$Q), and you cannot have an
76 // rvalue reference to a member.
77 return false;
78 case 'A':
79 // 'A' indicates a reference, and you cannot have a reference to a member
80 // function or member.
81 return false;
82 case 'P':
83 case 'Q':
84 case 'R':
85 case 'S':
86 // These 4 values indicate some kind of pointer, but we still don't know
87 // what.
88 break;
89 default:
90 // isMemberPointer() is called only if isPointerType() returns true,
91 // and it rejects other prefixes.
93 }
94
95 // If it starts with a number, then 6 indicates a non-member function
96 // pointer, and 8 indicates a member function pointer.
97 if (startsWithDigit(MangledName)) {
98 if (MangledName[0] != '6' && MangledName[0] != '8') {
99 Error = true;
100 return false;
101 }
102 return (MangledName[0] == '8');
103 }
104
105 // Remove ext qualifiers since those can appear on either type and are
106 // therefore not indicative.
107 consumeFront(MangledName, 'E'); // 64-bit
108 consumeFront(MangledName, 'I'); // restrict
109 consumeFront(MangledName, 'F'); // unaligned
110 demanglePointerAuthQualifier(MangledName);
111
112 if (MangledName.empty()) {
113 Error = true;
114 return false;
115 }
116
117 // The next value should be either ABCD (non-member) or QRST (member).
118 switch (MangledName.front()) {
119 case 'A':
120 case 'B':
121 case 'C':
122 case 'D':
123 return false;
124 case 'Q':
125 case 'R':
126 case 'S':
127 case 'T':
128 return true;
129 default:
130 Error = true;
131 return false;
132 }
133}
134
136consumeSpecialIntrinsicKind(std::string_view &MangledName) {
137 if (consumeFront(MangledName, "?_7"))
139 if (consumeFront(MangledName, "?_8"))
141 if (consumeFront(MangledName, "?_9"))
143 if (consumeFront(MangledName, "?_A"))
145 if (consumeFront(MangledName, "?_B"))
147 if (consumeFront(MangledName, "?_C"))
149 if (consumeFront(MangledName, "?_P"))
151 if (consumeFront(MangledName, "?_R0"))
153 if (consumeFront(MangledName, "?_R1"))
155 if (consumeFront(MangledName, "?_R2"))
157 if (consumeFront(MangledName, "?_R3"))
159 if (consumeFront(MangledName, "?_R4"))
161 if (consumeFront(MangledName, "?_S"))
163 if (consumeFront(MangledName, "?__E"))
165 if (consumeFront(MangledName, "?__F"))
167 if (consumeFront(MangledName, "?__J"))
170}
171
172static bool startsWithLocalScopePattern(std::string_view S) {
173 if (!consumeFront(S, '?'))
174 return false;
175
176 size_t End = S.find('?');
177 if (End == std::string_view::npos)
178 return false;
179 std::string_view Candidate = S.substr(0, End);
180 if (Candidate.empty())
181 return false;
182
183 // \?[0-9]\?
184 // ?@? is the discriminator 0.
185 if (Candidate.size() == 1)
186 return Candidate[0] == '@' || (Candidate[0] >= '0' && Candidate[0] <= '9');
187
188 // If it's not 0-9, then it's an encoded number terminated with an @
189 if (Candidate.back() != '@')
190 return false;
191 Candidate.remove_suffix(1);
192
193 // An encoded number starts with B-P and all subsequent digits are in A-P.
194 // Note that the reason the first digit cannot be A is two fold. First, it
195 // would create an ambiguity with ?A which delimits the beginning of an
196 // anonymous namespace. Second, A represents 0, and you don't start a multi
197 // digit number with a leading 0. Presumably the anonymous namespace
198 // ambiguity is also why single digit encoded numbers use 0-9 rather than A-J.
199 if (Candidate[0] < 'B' || Candidate[0] > 'P')
200 return false;
201 Candidate.remove_prefix(1);
202 while (!Candidate.empty()) {
203 if (Candidate[0] < 'A' || Candidate[0] > 'P')
204 return false;
205 Candidate.remove_prefix(1);
206 }
207
208 return true;
209}
210
211static bool isTagType(std::string_view S) {
212 switch (S.front()) {
213 case 'T': // union
214 case 'U': // struct
215 case 'V': // class
216 case 'W': // enum
217 return true;
218 }
219 return false;
220}
221
222static bool isCustomType(std::string_view S) { return S[0] == '?'; }
223
224static bool isPointerType(std::string_view S) {
225 if (llvm::itanium_demangle::starts_with(S, "$$Q")) // foo &&
226 return true;
227
228 switch (S.front()) {
229 case 'A': // foo &
230 case 'P': // foo *
231 case 'Q': // foo *const
232 case 'R': // foo *volatile
233 case 'S': // foo *const volatile
234 return true;
235 }
236 return false;
237}
238
239static bool isArrayType(std::string_view S) { return S[0] == 'Y'; }
240
241static bool isFunctionType(std::string_view S) {
242 return llvm::itanium_demangle::starts_with(S, "$$A8@@") ||
243 llvm::itanium_demangle::starts_with(S, "$$A6");
244}
245
247demangleFunctionRefQualifier(std::string_view &MangledName) {
248 if (consumeFront(MangledName, 'G'))
250 else if (consumeFront(MangledName, 'H'))
253}
254
255static std::pair<Qualifiers, PointerAffinity>
256demanglePointerCVQualifiers(std::string_view &MangledName) {
257 if (consumeFront(MangledName, "$$Q"))
258 return std::make_pair(Q_None, PointerAffinity::RValueReference);
259
260 const char F = MangledName.front();
261 MangledName.remove_prefix(1);
262 switch (F) {
263 case 'A':
264 return std::make_pair(Q_None, PointerAffinity::Reference);
265 case 'P':
266 return std::make_pair(Q_None, PointerAffinity::Pointer);
267 case 'Q':
268 return std::make_pair(Q_Const, PointerAffinity::Pointer);
269 case 'R':
270 return std::make_pair(Q_Volatile, PointerAffinity::Pointer);
271 case 'S':
272 return std::make_pair(Qualifiers(Q_Const | Q_Volatile),
274 }
275 // This function is only called if isPointerType() returns true,
276 // and it only returns true for the six cases listed above.
278}
279
280std::string_view Demangler::copyString(std::string_view Borrowed) {
281 char *Stable = Arena.allocUnalignedBuffer(Borrowed.size());
282 // This is not a micro-optimization, it avoids UB, should Borrowed be an null
283 // buffer.
284 if (Borrowed.size())
285 std::memcpy(Stable, Borrowed.data(), Borrowed.size());
286
287 return {Stable, Borrowed.size()};
288}
289
291Demangler::demangleSpecialTableSymbolNode(std::string_view &MangledName,
294 switch (K) {
296 NI->Name = "`vftable'";
297 break;
299 NI->Name = "`vbtable'";
300 break;
302 NI->Name = "`local vftable'";
303 break;
305 NI->Name = "`RTTI Complete Object Locator'";
306 break;
307 default:
309 }
310 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
312 STSN->Name = QN;
313 bool IsMember = false;
314 if (MangledName.empty()) {
315 Error = true;
316 return nullptr;
317 }
318 char Front = MangledName.front();
319 MangledName.remove_prefix(1);
320 if (Front != '6' && Front != '7') {
321 Error = true;
322 return nullptr;
323 }
324
325 std::tie(STSN->Quals, IsMember) = demangleQualifiers(MangledName);
326 if (!consumeFront(MangledName, '@'))
327 STSN->TargetName = demangleFullyQualifiedTypeName(MangledName);
328 return STSN;
329}
330
332Demangler::demangleLocalStaticGuard(std::string_view &MangledName,
333 bool IsThread) {
336 LSGI->IsThread = IsThread;
337 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, LSGI);
340 LSGVN->Name = QN;
341
342 if (consumeFront(MangledName, "4IA"))
343 LSGVN->IsVisible = false;
344 else if (consumeFront(MangledName, "5"))
345 LSGVN->IsVisible = true;
346 else {
347 Error = true;
348 return nullptr;
349 }
350
351 if (!MangledName.empty())
352 LSGI->ScopeIndex = demangleUnsigned(MangledName);
353 return LSGVN;
354}
355
357 std::string_view Name) {
359 Id->Name = Name;
360 return Id;
361}
362
364 IdentifierNode *Identifier) {
366 QN->Components = Arena.alloc<NodeArrayNode>();
367 QN->Components->Count = 1;
368 QN->Components->Nodes = Arena.allocArray<Node *>(1);
369 QN->Components->Nodes[0] = Identifier;
370 return QN;
371}
372
374 std::string_view Name) {
376 return synthesizeQualifiedName(Arena, Id);
377}
378
380 TypeNode *Type,
381 std::string_view VariableName) {
383 VSN->Type = Type;
384 VSN->Name = synthesizeQualifiedName(Arena, VariableName);
385 return VSN;
386}
387
389Demangler::demangleUntypedVariable(ArenaAllocator &Arena,
390 std::string_view &MangledName,
391 std::string_view VariableName) {
392 NamedIdentifierNode *NI = synthesizeNamedIdentifier(Arena, VariableName);
393 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
395 VSN->Name = QN;
396 if (consumeFront(MangledName, "8"))
397 return VSN;
398
399 Error = true;
400 return nullptr;
401}
402
404Demangler::demangleRttiBaseClassDescriptorNode(ArenaAllocator &Arena,
405 std::string_view &MangledName) {
408 RBCDN->NVOffset = demangleUnsigned(MangledName);
409 RBCDN->VBPtrOffset = demangleSigned(MangledName);
410 RBCDN->VBTableOffset = demangleUnsigned(MangledName);
411 RBCDN->Flags = demangleUnsigned(MangledName);
412 if (Error)
413 return nullptr;
414
416 VSN->Name = demangleNameScopeChain(MangledName, RBCDN);
417 consumeFront(MangledName, '8');
418 return VSN;
419}
420
422Demangler::demangleInitFiniStub(std::string_view &MangledName,
423 bool IsDestructor) {
426 DSIN->IsDestructor = IsDestructor;
427
428 bool IsKnownStaticDataMember = false;
429 if (consumeFront(MangledName, '?'))
430 IsKnownStaticDataMember = true;
431
432 SymbolNode *Symbol = demangleDeclarator(MangledName);
433 if (Error)
434 return nullptr;
435
436 FunctionSymbolNode *FSN = nullptr;
437
438 if (Symbol->kind() == NodeKind::VariableSymbol) {
439 DSIN->Variable = static_cast<VariableSymbolNode *>(Symbol);
440
441 // Older versions of clang mangled this type of symbol incorrectly. They
442 // would omit the leading ? and they would only emit a single @ at the end.
443 // The correct mangling is a leading ? and 2 trailing @ signs. Handle
444 // both cases.
445 int AtCount = IsKnownStaticDataMember ? 2 : 1;
446 for (int I = 0; I < AtCount; ++I) {
447 if (consumeFront(MangledName, '@'))
448 continue;
449 Error = true;
450 return nullptr;
451 }
452
453 FSN = demangleFunctionEncoding(MangledName);
454 if (FSN)
455 FSN->Name = synthesizeQualifiedName(Arena, DSIN);
456 } else {
457 if (IsKnownStaticDataMember) {
458 // This was supposed to be a static data member, but we got a function.
459 Error = true;
460 return nullptr;
461 }
462
463 FSN = static_cast<FunctionSymbolNode *>(Symbol);
464 DSIN->Name = Symbol->Name;
465 FSN->Name = synthesizeQualifiedName(Arena, DSIN);
466 }
467
468 return FSN;
469}
470
471SymbolNode *Demangler::demangleSpecialIntrinsic(std::string_view &MangledName) {
473
474 switch (SIK) {
476 return nullptr;
478 return demangleStringLiteral(MangledName);
483 return demangleSpecialTableSymbolNode(MangledName, SIK);
485 return demangleVcallThunkNode(MangledName);
487 return demangleLocalStaticGuard(MangledName, /*IsThread=*/false);
489 return demangleLocalStaticGuard(MangledName, /*IsThread=*/true);
491 TypeNode *T = demangleType(MangledName, QualifierMangleMode::Result);
492 if (Error)
493 break;
494 if (!consumeFront(MangledName, "@8"))
495 break;
496 if (!MangledName.empty())
497 break;
498 return synthesizeVariable(Arena, T, "`RTTI Type Descriptor'");
499 }
501 return demangleUntypedVariable(Arena, MangledName,
502 "`RTTI Base Class Array'");
504 return demangleUntypedVariable(Arena, MangledName,
505 "`RTTI Class Hierarchy Descriptor'");
507 return demangleRttiBaseClassDescriptorNode(Arena, MangledName);
509 return demangleInitFiniStub(MangledName, /*IsDestructor=*/false);
511 return demangleInitFiniStub(MangledName, /*IsDestructor=*/true);
514 // It's unclear which tools produces these manglings, so demangling
515 // support is not (yet?) implemented.
516 break;
518 DEMANGLE_UNREACHABLE; // Never returned by consumeSpecialIntrinsicKind.
519 }
520 Error = true;
521 return nullptr;
522}
523
525Demangler::demangleFunctionIdentifierCode(std::string_view &MangledName) {
526 assert(llvm::itanium_demangle::starts_with(MangledName, '?'));
527 MangledName.remove_prefix(1);
528 if (MangledName.empty()) {
529 Error = true;
530 return nullptr;
531 }
532
533 if (consumeFront(MangledName, "__"))
534 return demangleFunctionIdentifierCode(
536 if (consumeFront(MangledName, "_"))
537 return demangleFunctionIdentifierCode(MangledName,
539 return demangleFunctionIdentifierCode(MangledName,
541}
542
544Demangler::demangleStructorIdentifier(std::string_view &MangledName,
545 bool IsDestructor) {
547 N->IsDestructor = IsDestructor;
548 return N;
549}
550
552Demangler::demangleConversionOperatorIdentifier(std::string_view &MangledName) {
555 return N;
556}
557
559Demangler::demangleLiteralOperatorIdentifier(std::string_view &MangledName) {
562 N->Name = demangleSimpleString(MangledName, /*Memorize=*/false);
563 return N;
564}
565
567Demangler::translateIntrinsicFunctionCode(char CH,
569 using IFK = IntrinsicFunctionKind;
570 if (!(CH >= '0' && CH <= '9') && !(CH >= 'A' && CH <= 'Z')) {
571 Error = true;
572 return IFK::None;
573 }
574
575 // Not all ? identifiers are intrinsics *functions*. This function only maps
576 // operator codes for the special functions, all others are handled elsewhere,
577 // hence the IFK::None entries in the table.
578 static IFK Basic[36] = {
579 IFK::None, // ?0 # Foo::Foo()
580 IFK::None, // ?1 # Foo::~Foo()
581 IFK::New, // ?2 # operator new
582 IFK::Delete, // ?3 # operator delete
583 IFK::Assign, // ?4 # operator=
584 IFK::RightShift, // ?5 # operator>>
585 IFK::LeftShift, // ?6 # operator<<
586 IFK::LogicalNot, // ?7 # operator!
587 IFK::Equals, // ?8 # operator==
588 IFK::NotEquals, // ?9 # operator!=
589 IFK::ArraySubscript, // ?A # operator[]
590 IFK::None, // ?B # Foo::operator <type>()
591 IFK::Pointer, // ?C # operator->
592 IFK::Dereference, // ?D # operator*
593 IFK::Increment, // ?E # operator++
594 IFK::Decrement, // ?F # operator--
595 IFK::Minus, // ?G # operator-
596 IFK::Plus, // ?H # operator+
597 IFK::BitwiseAnd, // ?I # operator&
598 IFK::MemberPointer, // ?J # operator->*
599 IFK::Divide, // ?K # operator/
600 IFK::Modulus, // ?L # operator%
601 IFK::LessThan, // ?M operator<
602 IFK::LessThanEqual, // ?N operator<=
603 IFK::GreaterThan, // ?O operator>
604 IFK::GreaterThanEqual, // ?P operator>=
605 IFK::Comma, // ?Q operator,
606 IFK::Parens, // ?R operator()
607 IFK::BitwiseNot, // ?S operator~
608 IFK::BitwiseXor, // ?T operator^
609 IFK::BitwiseOr, // ?U operator|
610 IFK::LogicalAnd, // ?V operator&&
611 IFK::LogicalOr, // ?W operator||
612 IFK::TimesEqual, // ?X operator*=
613 IFK::PlusEqual, // ?Y operator+=
614 IFK::MinusEqual, // ?Z operator-=
615 };
616 static IFK Under[36] = {
617 IFK::DivEqual, // ?_0 operator/=
618 IFK::ModEqual, // ?_1 operator%=
619 IFK::RshEqual, // ?_2 operator>>=
620 IFK::LshEqual, // ?_3 operator<<=
621 IFK::BitwiseAndEqual, // ?_4 operator&=
622 IFK::BitwiseOrEqual, // ?_5 operator|=
623 IFK::BitwiseXorEqual, // ?_6 operator^=
624 IFK::None, // ?_7 # vftable
625 IFK::None, // ?_8 # vbtable
626 IFK::None, // ?_9 # vcall
627 IFK::None, // ?_A # typeof
628 IFK::None, // ?_B # local static guard
629 IFK::None, // ?_C # string literal
630 IFK::VbaseDtor, // ?_D # vbase destructor
631 IFK::VecDelDtor, // ?_E # vector deleting destructor
632 IFK::DefaultCtorClosure, // ?_F # default constructor closure
633 IFK::ScalarDelDtor, // ?_G # scalar deleting destructor
634 IFK::VecCtorIter, // ?_H # vector constructor iterator
635 IFK::VecDtorIter, // ?_I # vector destructor iterator
636 IFK::VecVbaseCtorIter, // ?_J # vector vbase constructor iterator
637 IFK::VdispMap, // ?_K # virtual displacement map
638 IFK::EHVecCtorIter, // ?_L # eh vector constructor iterator
639 IFK::EHVecDtorIter, // ?_M # eh vector destructor iterator
640 IFK::EHVecVbaseCtorIter, // ?_N # eh vector vbase constructor iterator
641 IFK::CopyCtorClosure, // ?_O # copy constructor closure
642 IFK::None, // ?_P<name> # udt returning <name>
643 IFK::None, // ?_Q # <unknown>
644 IFK::None, // ?_R0 - ?_R4 # RTTI Codes
645 IFK::None, // ?_S # local vftable
646 IFK::LocalVftableCtorClosure, // ?_T # local vftable constructor closure
647 IFK::ArrayNew, // ?_U operator new[]
648 IFK::ArrayDelete, // ?_V operator delete[]
649 IFK::None, // ?_W <unused>
650 IFK::None, // ?_X <unused>
651 IFK::None, // ?_Y <unused>
652 IFK::None, // ?_Z <unused>
653 };
654 static IFK DoubleUnder[36] = {
655 IFK::None, // ?__0 <unused>
656 IFK::None, // ?__1 <unused>
657 IFK::None, // ?__2 <unused>
658 IFK::None, // ?__3 <unused>
659 IFK::None, // ?__4 <unused>
660 IFK::None, // ?__5 <unused>
661 IFK::None, // ?__6 <unused>
662 IFK::None, // ?__7 <unused>
663 IFK::None, // ?__8 <unused>
664 IFK::None, // ?__9 <unused>
665 IFK::ManVectorCtorIter, // ?__A managed vector ctor iterator
666 IFK::ManVectorDtorIter, // ?__B managed vector dtor iterator
667 IFK::EHVectorCopyCtorIter, // ?__C EH vector copy ctor iterator
668 IFK::EHVectorVbaseCopyCtorIter, // ?__D EH vector vbase copy ctor iter
669 IFK::None, // ?__E dynamic initializer for `T'
670 IFK::None, // ?__F dynamic atexit destructor for `T'
671 IFK::VectorCopyCtorIter, // ?__G vector copy constructor iter
672 IFK::VectorVbaseCopyCtorIter, // ?__H vector vbase copy ctor iter
673 IFK::ManVectorVbaseCopyCtorIter, // ?__I managed vector vbase copy ctor
674 // iter
675 IFK::None, // ?__J local static thread guard
676 IFK::None, // ?__K operator ""_name
677 IFK::CoAwait, // ?__L operator co_await
678 IFK::Spaceship, // ?__M operator<=>
679 IFK::None, // ?__N <unused>
680 IFK::None, // ?__O <unused>
681 IFK::None, // ?__P <unused>
682 IFK::None, // ?__Q <unused>
683 IFK::None, // ?__R <unused>
684 IFK::None, // ?__S <unused>
685 IFK::None, // ?__T <unused>
686 IFK::None, // ?__U <unused>
687 IFK::None, // ?__V <unused>
688 IFK::None, // ?__W <unused>
689 IFK::None, // ?__X <unused>
690 IFK::None, // ?__Y <unused>
691 IFK::None, // ?__Z <unused>
692 };
693
694 int Index = (CH >= '0' && CH <= '9') ? (CH - '0') : (CH - 'A' + 10);
695 switch (Group) {
697 return Basic[Index];
699 return Under[Index];
701 return DoubleUnder[Index];
702 }
704}
705
707Demangler::demangleFunctionIdentifierCode(std::string_view &MangledName,
709 if (MangledName.empty()) {
710 Error = true;
711 return nullptr;
712 }
713 const char CH = MangledName.front();
714 switch (Group) {
716 MangledName.remove_prefix(1);
717 switch (CH) {
718 case '0':
719 case '1':
720 return demangleStructorIdentifier(MangledName, CH == '1');
721 case 'B':
722 return demangleConversionOperatorIdentifier(MangledName);
723 default:
725 translateIntrinsicFunctionCode(CH, Group));
726 }
728 MangledName.remove_prefix(1);
730 translateIntrinsicFunctionCode(CH, Group));
732 MangledName.remove_prefix(1);
733 switch (CH) {
734 case 'K':
735 return demangleLiteralOperatorIdentifier(MangledName);
736 default:
738 translateIntrinsicFunctionCode(CH, Group));
739 }
740 }
741
743}
744
745SymbolNode *Demangler::demangleEncodedSymbol(std::string_view &MangledName,
747 if (MangledName.empty()) {
748 Error = true;
749 return nullptr;
750 }
751
752 // Read a variable.
753 switch (MangledName.front()) {
754 case '0':
755 case '1':
756 case '2':
757 case '3':
758 case '4': {
759 StorageClass SC = demangleVariableStorageClass(MangledName);
760 return demangleVariableEncoding(MangledName, SC);
761 }
762 }
763 FunctionSymbolNode *FSN = demangleFunctionEncoding(MangledName);
764
765 IdentifierNode *UQN = Name->getUnqualifiedIdentifier();
768 static_cast<ConversionOperatorIdentifierNode *>(UQN);
769 if (FSN)
770 COIN->TargetType = FSN->Signature->ReturnType;
771 }
772 return FSN;
773}
774
775SymbolNode *Demangler::demangleDeclarator(std::string_view &MangledName) {
776 // What follows is a main symbol name. This may include namespaces or class
777 // back references.
778 QualifiedNameNode *QN = demangleFullyQualifiedSymbolName(MangledName);
779 if (Error)
780 return nullptr;
781
782 SymbolNode *Symbol = demangleEncodedSymbol(MangledName, QN);
783 if (Error)
784 return nullptr;
785 Symbol->Name = QN;
786
790 static_cast<ConversionOperatorIdentifierNode *>(UQN);
791 if (!COIN->TargetType) {
792 Error = true;
793 return nullptr;
794 }
795 }
796 return Symbol;
797}
798
799SymbolNode *Demangler::demangleMD5Name(std::string_view &MangledName) {
800 assert(llvm::itanium_demangle::starts_with(MangledName, "??@"));
801 // This is an MD5 mangled name. We can't demangle it, just return the
802 // mangled name.
803 // An MD5 mangled name is ??@ followed by 32 characters and a terminating @.
804 size_t MD5Last = MangledName.find('@', strlen("??@"));
805 if (MD5Last == std::string_view::npos) {
806 Error = true;
807 return nullptr;
808 }
809 const char *Start = MangledName.data();
810 const size_t StartSize = MangledName.size();
811 MangledName.remove_prefix(MD5Last + 1);
812
813 // There are two additional special cases for MD5 names:
814 // 1. For complete object locators where the object name is long enough
815 // for the object to have an MD5 name, the complete object locator is
816 // called ??@...@??_R4@ (with a trailing "??_R4@" instead of the usual
817 // leading "??_R4". This is handled here.
818 // 2. For catchable types, in versions of MSVC before 2015 (<1900) or after
819 // 2017.2 (>= 1914), the catchable type mangling is _CT??@...@??@...@8
820 // instead of_CT??@...@8 with just one MD5 name. Since we don't yet
821 // demangle catchable types anywhere, this isn't handled for MD5 names
822 // either.
823 consumeFront(MangledName, "??_R4@");
824
825 assert(MangledName.size() < StartSize);
826 const size_t Count = StartSize - MangledName.size();
827 std::string_view MD5(Start, Count);
829 S->Name = synthesizeQualifiedName(Arena, MD5);
830
831 return S;
832}
833
834SymbolNode *Demangler::demangleTypeinfoName(std::string_view &MangledName) {
835 assert(llvm::itanium_demangle::starts_with(MangledName, '.'));
836 consumeFront(MangledName, '.');
837
838 TypeNode *T = demangleType(MangledName, QualifierMangleMode::Result);
839 if (Error || !MangledName.empty()) {
840 Error = true;
841 return nullptr;
842 }
843 return synthesizeVariable(Arena, T, "`RTTI Type Descriptor Name'");
844}
845
846// Parser entry point.
847SymbolNode *Demangler::parse(std::string_view &MangledName) {
848 // Typeinfo names are strings stored in RTTI data. They're not symbol names.
849 // It's still useful to demangle them. They're the only demangled entity
850 // that doesn't start with a "?" but a ".".
851 if (llvm::itanium_demangle::starts_with(MangledName, '.'))
852 return demangleTypeinfoName(MangledName);
853
854 if (llvm::itanium_demangle::starts_with(MangledName, "??@"))
855 return demangleMD5Name(MangledName);
856
857 // MSVC-style mangled symbols must start with '?'.
858 if (!llvm::itanium_demangle::starts_with(MangledName, '?')) {
859 Error = true;
860 return nullptr;
861 }
862
863 consumeFront(MangledName, '?');
864
865 // ?$ is a template instantiation, but all other names that start with ? are
866 // operators / special names.
867 if (SymbolNode *SI = demangleSpecialIntrinsic(MangledName))
868 return SI;
869
870 return demangleDeclarator(MangledName);
871}
872
873TagTypeNode *Demangler::parseTagUniqueName(std::string_view &MangledName) {
874 if (!consumeFront(MangledName, ".?A")) {
875 Error = true;
876 return nullptr;
877 }
878 consumeFront(MangledName, ".?A");
879 if (MangledName.empty()) {
880 Error = true;
881 return nullptr;
882 }
883
884 return demangleClassType(MangledName);
885}
886
887// <type-encoding> ::= <storage-class> <variable-type>
888// <storage-class> ::= 0 # private static member
889// ::= 1 # protected static member
890// ::= 2 # public static member
891// ::= 3 # global
892// ::= 4 # static local
893
895Demangler::demangleVariableEncoding(std::string_view &MangledName,
896 StorageClass SC) {
898
899 VSN->Type = demangleType(MangledName, QualifierMangleMode::Drop);
900 VSN->SC = SC;
901
902 if (Error)
903 return nullptr;
904
905 // <variable-type> ::= <type> <cvr-qualifiers>
906 // ::= <type> <pointee-cvr-qualifiers> # pointers, references
907 switch (VSN->Type->kind()) {
909 PointerTypeNode *PTN = static_cast<PointerTypeNode *>(VSN->Type);
910
911 Qualifiers ExtraChildQuals = Q_None;
912 PTN->Quals = Qualifiers(VSN->Type->Quals |
913 demanglePointerExtQualifiers(MangledName));
914
915 bool IsMember = false;
916 std::tie(ExtraChildQuals, IsMember) = demangleQualifiers(MangledName);
917
918 if (PTN->ClassParent) {
919 QualifiedNameNode *BackRefName =
920 demangleFullyQualifiedTypeName(MangledName);
921 (void)BackRefName;
922 }
923 PTN->Pointee->Quals = Qualifiers(PTN->Pointee->Quals | ExtraChildQuals);
924
925 break;
926 }
927 default:
928 VSN->Type->Quals = demangleQualifiers(MangledName).first;
929 break;
930 }
931
932 return VSN;
933}
934
935// Sometimes numbers are encoded in mangled symbols. For example,
936// "int (*x)[20]" is a valid C type (x is a pointer to an array of
937// length 20), so we need some way to embed numbers as part of symbols.
938// This function parses it.
939//
940// <number> ::= [?] <non-negative integer>
941//
942// <non-negative integer> ::= <decimal digit> # when 1 <= Number <= 10
943// ::= <hex digit>+ @ # when Number == 0 or >= 10
944//
945// <hex-digit> ::= [A-P] # A = 0, B = 1, ...
946std::pair<uint64_t, bool>
947Demangler::demangleNumber(std::string_view &MangledName) {
948 bool IsNegative = consumeFront(MangledName, '?');
949
950 if (startsWithDigit(MangledName)) {
951 uint64_t Ret = MangledName[0] - '0' + 1;
952 MangledName.remove_prefix(1);
953 return {Ret, IsNegative};
954 }
955
956 uint64_t Ret = 0;
957 for (size_t i = 0; i < MangledName.size(); ++i) {
958 char C = MangledName[i];
959 if (C == '@') {
960 MangledName.remove_prefix(i + 1);
961 return {Ret, IsNegative};
962 }
963 if ('A' <= C && C <= 'P') {
964 Ret = (Ret << 4) + (C - 'A');
965 continue;
966 }
967 break;
968 }
969
970 Error = true;
971 return {0ULL, false};
972}
973
974uint64_t Demangler::demangleUnsigned(std::string_view &MangledName) {
975 bool IsNegative = false;
976 uint64_t Number = 0;
977 std::tie(Number, IsNegative) = demangleNumber(MangledName);
978 if (IsNegative)
979 Error = true;
980 return Number;
981}
982
983int64_t Demangler::demangleSigned(std::string_view &MangledName) {
984 bool IsNegative = false;
985 uint64_t Number = 0;
986 std::tie(Number, IsNegative) = demangleNumber(MangledName);
987 if (Number > INT64_MAX)
988 Error = true;
989 int64_t I = static_cast<int64_t>(Number);
990 return IsNegative ? -I : I;
991}
992
993// First 10 strings can be referenced by special BackReferences ?0, ?1, ..., ?9.
994// Memorize it.
995void Demangler::memorizeString(std::string_view S) {
996 if (Backrefs.NamesCount >= BackrefContext::Max)
997 return;
998 for (size_t i = 0; i < Backrefs.NamesCount; ++i)
999 if (S == Backrefs.Names[i]->Name)
1000 return;
1002 N->Name = S;
1003 Backrefs.Names[Backrefs.NamesCount++] = N;
1004}
1005
1007Demangler::demangleBackRefName(std::string_view &MangledName) {
1008 assert(startsWithDigit(MangledName));
1009
1010 size_t I = MangledName[0] - '0';
1011 if (I >= Backrefs.NamesCount) {
1012 Error = true;
1013 return nullptr;
1014 }
1015
1016 MangledName.remove_prefix(1);
1017 return Backrefs.Names[I];
1018}
1019
1020void Demangler::memorizeIdentifier(IdentifierNode *Identifier) {
1021 // Render this class template name into a string buffer so that we can
1022 // memorize it for the purpose of back-referencing.
1024 Identifier->output(OB, OF_Default);
1025 std::string_view Owned = copyString(OB);
1026 memorizeString(Owned);
1027 std::free(OB.getBuffer());
1028}
1029
1031Demangler::demangleTemplateInstantiationName(std::string_view &MangledName,
1032 NameBackrefBehavior NBB) {
1033 assert(llvm::itanium_demangle::starts_with(MangledName, "?$"));
1034 consumeFront(MangledName, "?$");
1035
1036 BackrefContext OuterContext;
1037 std::swap(OuterContext, Backrefs);
1038
1040 demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
1041 if (!Error)
1042 Identifier->TemplateParams = demangleTemplateParameterList(MangledName);
1043
1044 std::swap(OuterContext, Backrefs);
1045 if (Error)
1046 return nullptr;
1047
1048 if (NBB & NBB_Template) {
1049 // NBB_Template is only set for types and non-leaf names ("a::" in "a::b").
1050 // Structors and conversion operators only makes sense in a leaf name, so
1051 // reject them in NBB_Template contexts.
1054 Error = true;
1055 return nullptr;
1056 }
1057
1058 memorizeIdentifier(Identifier);
1059 }
1060
1061 return Identifier;
1062}
1063
1065Demangler::demangleSimpleName(std::string_view &MangledName, bool Memorize) {
1066 std::string_view S = demangleSimpleString(MangledName, Memorize);
1067 if (Error)
1068 return nullptr;
1069
1071 Name->Name = S;
1072 return Name;
1073}
1074
1075static bool isRebasedHexDigit(char C) { return (C >= 'A' && C <= 'P'); }
1076
1079 return (C <= 'J') ? (C - 'A') : (10 + C - 'K');
1080}
1081
1082uint8_t Demangler::demangleCharLiteral(std::string_view &MangledName) {
1083 assert(!MangledName.empty());
1084 if (!llvm::itanium_demangle::starts_with(MangledName, '?')) {
1085 const uint8_t F = MangledName.front();
1086 MangledName.remove_prefix(1);
1087 return F;
1088 }
1089
1090 MangledName.remove_prefix(1);
1091 if (MangledName.empty())
1092 goto CharLiteralError;
1093
1094 if (consumeFront(MangledName, '$')) {
1095 // Two hex digits
1096 if (MangledName.size() < 2)
1097 goto CharLiteralError;
1098 std::string_view Nibbles = MangledName.substr(0, 2);
1099 if (!isRebasedHexDigit(Nibbles[0]) || !isRebasedHexDigit(Nibbles[1]))
1100 goto CharLiteralError;
1101 // Don't append the null terminator.
1102 uint8_t C1 = rebasedHexDigitToNumber(Nibbles[0]);
1103 uint8_t C2 = rebasedHexDigitToNumber(Nibbles[1]);
1104 MangledName.remove_prefix(2);
1105 return (C1 << 4) | C2;
1106 }
1107
1108 if (startsWithDigit(MangledName)) {
1109 const char *Lookup = ",/\\:. \n\t'-";
1110 char C = Lookup[MangledName[0] - '0'];
1111 MangledName.remove_prefix(1);
1112 return C;
1113 }
1114
1115 if (MangledName[0] >= 'a' && MangledName[0] <= 'z') {
1116 char Lookup[26] = {'\xE1', '\xE2', '\xE3', '\xE4', '\xE5', '\xE6', '\xE7',
1117 '\xE8', '\xE9', '\xEA', '\xEB', '\xEC', '\xED', '\xEE',
1118 '\xEF', '\xF0', '\xF1', '\xF2', '\xF3', '\xF4', '\xF5',
1119 '\xF6', '\xF7', '\xF8', '\xF9', '\xFA'};
1120 char C = Lookup[MangledName[0] - 'a'];
1121 MangledName.remove_prefix(1);
1122 return C;
1123 }
1124
1125 if (MangledName[0] >= 'A' && MangledName[0] <= 'Z') {
1126 char Lookup[26] = {'\xC1', '\xC2', '\xC3', '\xC4', '\xC5', '\xC6', '\xC7',
1127 '\xC8', '\xC9', '\xCA', '\xCB', '\xCC', '\xCD', '\xCE',
1128 '\xCF', '\xD0', '\xD1', '\xD2', '\xD3', '\xD4', '\xD5',
1129 '\xD6', '\xD7', '\xD8', '\xD9', '\xDA'};
1130 char C = Lookup[MangledName[0] - 'A'];
1131 MangledName.remove_prefix(1);
1132 return C;
1133 }
1134
1135CharLiteralError:
1136 Error = true;
1137 return '\0';
1138}
1139
1140wchar_t Demangler::demangleWcharLiteral(std::string_view &MangledName) {
1141 uint8_t C1, C2;
1142
1143 C1 = demangleCharLiteral(MangledName);
1144 if (Error || MangledName.empty())
1145 goto WCharLiteralError;
1146 C2 = demangleCharLiteral(MangledName);
1147 if (Error)
1148 goto WCharLiteralError;
1149
1150 return ((wchar_t)C1 << 8) | (wchar_t)C2;
1151
1152WCharLiteralError:
1153 Error = true;
1154 return L'\0';
1155}
1156
1157static void writeHexDigit(char *Buffer, uint8_t Digit) {
1158 assert(Digit <= 15);
1159 *Buffer = (Digit < 10) ? ('0' + Digit) : ('A' + Digit - 10);
1160}
1161
1162static void outputHex(OutputBuffer &OB, unsigned C) {
1163 assert (C != 0);
1164
1165 // It's easier to do the math if we can work from right to left, but we need
1166 // to print the numbers from left to right. So render this into a temporary
1167 // buffer first, then output the temporary buffer. Each byte is of the form
1168 // \xAB, which means that each byte needs 4 characters. Since there are at
1169 // most 4 bytes, we need a 4*4+1 = 17 character temporary buffer.
1170 char TempBuffer[17];
1171
1172 ::memset(TempBuffer, 0, sizeof(TempBuffer));
1173 constexpr int MaxPos = sizeof(TempBuffer) - 1;
1174
1175 int Pos = MaxPos - 1; // TempBuffer[MaxPos] is the terminating \0.
1176 while (C != 0) {
1177 for (int I = 0; I < 2; ++I) {
1178 writeHexDigit(&TempBuffer[Pos--], C % 16);
1179 C /= 16;
1180 }
1181 }
1182 TempBuffer[Pos--] = 'x';
1183 assert(Pos >= 0);
1184 TempBuffer[Pos--] = '\\';
1185 OB << std::string_view(&TempBuffer[Pos + 1]);
1186}
1187
1188static void outputEscapedChar(OutputBuffer &OB, unsigned C) {
1189 switch (C) {
1190 case '\0': // nul
1191 OB << "\\0";
1192 return;
1193 case '\'': // single quote
1194 OB << "\\\'";
1195 return;
1196 case '\"': // double quote
1197 OB << "\\\"";
1198 return;
1199 case '\\': // backslash
1200 OB << "\\\\";
1201 return;
1202 case '\a': // bell
1203 OB << "\\a";
1204 return;
1205 case '\b': // backspace
1206 OB << "\\b";
1207 return;
1208 case '\f': // form feed
1209 OB << "\\f";
1210 return;
1211 case '\n': // new line
1212 OB << "\\n";
1213 return;
1214 case '\r': // carriage return
1215 OB << "\\r";
1216 return;
1217 case '\t': // tab
1218 OB << "\\t";
1219 return;
1220 case '\v': // vertical tab
1221 OB << "\\v";
1222 return;
1223 default:
1224 break;
1225 }
1226
1227 if (C > 0x1F && C < 0x7F) {
1228 // Standard ascii char.
1229 OB << (char)C;
1230 return;
1231 }
1232
1233 outputHex(OB, C);
1234}
1235
1236static unsigned countTrailingNullBytes(const uint8_t *StringBytes, int Length) {
1237 const uint8_t *End = StringBytes + Length - 1;
1238 unsigned Count = 0;
1239 while (Length > 0 && *End == 0) {
1240 --Length;
1241 --End;
1242 ++Count;
1243 }
1244 return Count;
1245}
1246
1247static unsigned countEmbeddedNulls(const uint8_t *StringBytes,
1248 unsigned Length) {
1249 unsigned Result = 0;
1250 for (unsigned I = 0; I < Length; ++I) {
1251 if (*StringBytes++ == 0)
1252 ++Result;
1253 }
1254 return Result;
1255}
1256
1257// A mangled (non-wide) string literal stores the total length of the string it
1258// refers to (passed in NumBytes), and it contains up to 32 bytes of actual text
1259// (passed in StringBytes, NumChars).
1260static unsigned guessCharByteSize(const uint8_t *StringBytes, unsigned NumChars,
1261 uint64_t NumBytes) {
1262 assert(NumBytes > 0);
1263
1264 // If the number of bytes is odd, this is guaranteed to be a char string.
1265 if (NumBytes % 2 == 1)
1266 return 1;
1267
1268 // All strings can encode at most 32 bytes of data. If it's less than that,
1269 // then we encoded the entire string. In this case we check for a 1-byte,
1270 // 2-byte, or 4-byte null terminator.
1271 if (NumBytes < 32) {
1272 unsigned TrailingNulls = countTrailingNullBytes(StringBytes, NumChars);
1273 if (TrailingNulls >= 4 && NumBytes % 4 == 0)
1274 return 4;
1275 if (TrailingNulls >= 2)
1276 return 2;
1277 return 1;
1278 }
1279
1280 // The whole string was not able to be encoded. Try to look at embedded null
1281 // terminators to guess. The heuristic is that we count all embedded null
1282 // terminators. If more than 2/3 are null, it's a char32. If more than 1/3
1283 // are null, it's a char16. Otherwise it's a char8. This obviously isn't
1284 // perfect and is biased towards languages that have ascii alphabets, but this
1285 // was always going to be best effort since the encoding is lossy.
1286 unsigned Nulls = countEmbeddedNulls(StringBytes, NumChars);
1287 if (Nulls >= 2 * NumChars / 3 && NumBytes % 4 == 0)
1288 return 4;
1289 if (Nulls >= NumChars / 3)
1290 return 2;
1291 return 1;
1292}
1293
1294static unsigned decodeMultiByteChar(const uint8_t *StringBytes,
1295 unsigned CharIndex, unsigned CharBytes) {
1296 assert(CharBytes == 1 || CharBytes == 2 || CharBytes == 4);
1297 unsigned Offset = CharIndex * CharBytes;
1298 unsigned Result = 0;
1299 StringBytes = StringBytes + Offset;
1300 for (unsigned I = 0; I < CharBytes; ++I) {
1301 unsigned C = static_cast<unsigned>(StringBytes[I]);
1302 Result |= C << (8 * I);
1303 }
1304 return Result;
1305}
1306
1308Demangler::demangleVcallThunkNode(std::string_view &MangledName) {
1311 FSN->Signature = Arena.alloc<ThunkSignatureNode>();
1313
1314 FSN->Name = demangleNameScopeChain(MangledName, VTIN);
1315 if (!Error)
1316 Error = !consumeFront(MangledName, "$B");
1317 if (!Error)
1318 VTIN->OffsetInVTable = demangleUnsigned(MangledName);
1319 if (!Error)
1320 Error = !consumeFront(MangledName, 'A');
1321 if (!Error)
1322 FSN->Signature->CallConvention = demangleCallingConvention(MangledName);
1323 return (Error) ? nullptr : FSN;
1324}
1325
1327Demangler::demangleStringLiteral(std::string_view &MangledName) {
1328 // This function uses goto, so declare all variables up front.
1330 std::string_view CRC;
1331 uint64_t StringByteSize;
1332 bool IsWcharT = false;
1333 bool IsNegative = false;
1334 size_t CrcEndPos = 0;
1335 char F;
1336
1338
1339 // Prefix indicating the beginning of a string literal
1340 if (!consumeFront(MangledName, "@_"))
1341 goto StringLiteralError;
1342 if (MangledName.empty())
1343 goto StringLiteralError;
1344
1345 // Char Type (regular or wchar_t)
1346 F = MangledName.front();
1347 MangledName.remove_prefix(1);
1348 switch (F) {
1349 case '1':
1350 IsWcharT = true;
1352 case '0':
1353 break;
1354 default:
1355 goto StringLiteralError;
1356 }
1357
1358 // Encoded Length
1359 std::tie(StringByteSize, IsNegative) = demangleNumber(MangledName);
1360 if (Error || IsNegative || StringByteSize < (IsWcharT ? 2 : 1))
1361 goto StringLiteralError;
1362
1363 // CRC 32 (always 8 characters plus a terminator)
1364 CrcEndPos = MangledName.find('@');
1365 if (CrcEndPos == std::string_view::npos)
1366 goto StringLiteralError;
1367 CRC = MangledName.substr(0, CrcEndPos);
1368 MangledName.remove_prefix(CrcEndPos + 1);
1369 if (MangledName.empty())
1370 goto StringLiteralError;
1371
1372 if (IsWcharT) {
1373 Result->Char = CharKind::Wchar;
1374 if (StringByteSize > 64)
1375 Result->IsTruncated = true;
1376
1377 while (!consumeFront(MangledName, '@')) {
1378 // For a wide string StringByteSize has to have an even length.
1379 if (StringByteSize % 2 != 0)
1380 goto StringLiteralError;
1381 if (StringByteSize == 0)
1382 goto StringLiteralError;
1383 if (MangledName.size() < 2)
1384 goto StringLiteralError;
1385 wchar_t W = demangleWcharLiteral(MangledName);
1386 if (StringByteSize != 2 || Result->IsTruncated)
1387 outputEscapedChar(OB, W);
1388 StringByteSize -= 2;
1389 if (Error)
1390 goto StringLiteralError;
1391 }
1392 } else {
1393 // The max byte length is actually 32, but some compilers mangled strings
1394 // incorrectly, so we have to assume it can go higher.
1395 constexpr unsigned MaxStringByteLength = 32 * 4;
1396 uint8_t StringBytes[MaxStringByteLength];
1397
1398 unsigned BytesDecoded = 0;
1399 while (!consumeFront(MangledName, '@')) {
1400 if (MangledName.size() < 1 || BytesDecoded >= MaxStringByteLength)
1401 goto StringLiteralError;
1402 StringBytes[BytesDecoded++] = demangleCharLiteral(MangledName);
1403 }
1404
1405 if (StringByteSize > BytesDecoded)
1406 Result->IsTruncated = true;
1407
1408 unsigned CharBytes =
1409 guessCharByteSize(StringBytes, BytesDecoded, StringByteSize);
1410 assert(StringByteSize % CharBytes == 0);
1411 switch (CharBytes) {
1412 case 1:
1413 Result->Char = CharKind::Char;
1414 break;
1415 case 2:
1416 Result->Char = CharKind::Char16;
1417 break;
1418 case 4:
1419 Result->Char = CharKind::Char32;
1420 break;
1421 default:
1423 }
1424 const unsigned NumChars = BytesDecoded / CharBytes;
1425 for (unsigned CharIndex = 0; CharIndex < NumChars; ++CharIndex) {
1426 unsigned NextChar =
1427 decodeMultiByteChar(StringBytes, CharIndex, CharBytes);
1428 if (CharIndex + 1 < NumChars || Result->IsTruncated)
1429 outputEscapedChar(OB, NextChar);
1430 }
1431 }
1432
1433 Result->DecodedString = copyString(OB);
1434 std::free(OB.getBuffer());
1435 return Result;
1436
1437StringLiteralError:
1438 Error = true;
1439 std::free(OB.getBuffer());
1440 return nullptr;
1441}
1442
1443// Returns MangledName's prefix before the first '@', or an error if
1444// MangledName contains no '@' or the prefix has length 0.
1445std::string_view Demangler::demangleSimpleString(std::string_view &MangledName,
1446 bool Memorize) {
1447 std::string_view S;
1448 for (size_t i = 0; i < MangledName.size(); ++i) {
1449 if (MangledName[i] != '@')
1450 continue;
1451 if (i == 0)
1452 break;
1453 S = MangledName.substr(0, i);
1454 MangledName.remove_prefix(i + 1);
1455
1456 if (Memorize)
1457 memorizeString(S);
1458 return S;
1459 }
1460
1461 Error = true;
1462 return {};
1463}
1464
1466Demangler::demangleAnonymousNamespaceName(std::string_view &MangledName) {
1467 assert(llvm::itanium_demangle::starts_with(MangledName, "?A"));
1468 consumeFront(MangledName, "?A");
1469
1471 Node->Name = "`anonymous namespace'";
1472 size_t EndPos = MangledName.find('@');
1473 if (EndPos == std::string_view::npos) {
1474 Error = true;
1475 return nullptr;
1476 }
1477 std::string_view NamespaceKey = MangledName.substr(0, EndPos);
1478 memorizeString(NamespaceKey);
1479 MangledName = MangledName.substr(EndPos + 1);
1480 return Node;
1481}
1482
1484Demangler::demangleLocallyScopedNamePiece(std::string_view &MangledName) {
1485 assert(startsWithLocalScopePattern(MangledName));
1486
1488 consumeFront(MangledName, '?');
1489 uint64_t Number = 0;
1490 bool IsNegative = false;
1491 std::tie(Number, IsNegative) = demangleNumber(MangledName);
1492 assert(!IsNegative);
1493
1494 // One ? to terminate the number
1495 consumeFront(MangledName, '?');
1496
1497 assert(!Error);
1498 Node *Scope = parse(MangledName);
1499 if (Error)
1500 return nullptr;
1501
1502 // Render the parent symbol's name into a buffer.
1504 OB << '`';
1505 Scope->output(OB, OF_Default);
1506 OB << '\'';
1507 OB << "::`" << Number << "'";
1508
1509 Identifier->Name = copyString(OB);
1510 std::free(OB.getBuffer());
1511 return Identifier;
1512}
1513
1514// Parses a type name in the form of A@B@C@@ which represents C::B::A.
1516Demangler::demangleFullyQualifiedTypeName(std::string_view &MangledName) {
1518 demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1519 if (Error)
1520 return nullptr;
1521 assert(Identifier);
1522
1523 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1524 if (Error)
1525 return nullptr;
1526 assert(QN);
1527 return QN;
1528}
1529
1530// Parses a symbol name in the form of A@B@C@@ which represents C::B::A.
1531// Symbol names have slightly different rules regarding what can appear
1532// so we separate out the implementations for flexibility.
1534Demangler::demangleFullyQualifiedSymbolName(std::string_view &MangledName) {
1535 // This is the final component of a symbol name (i.e. the leftmost component
1536 // of a mangled name. Since the only possible template instantiation that
1537 // can appear in this context is a function template, and since those are
1538 // not saved for the purposes of name backreferences, only backref simple
1539 // names.
1541 demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
1542 if (Error)
1543 return nullptr;
1544
1545 QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1546 if (Error)
1547 return nullptr;
1548
1550 if (QN->Components->Count < 2) {
1551 Error = true;
1552 return nullptr;
1553 }
1555 static_cast<StructorIdentifierNode *>(Identifier);
1556 Node *ClassNode = QN->Components->Nodes[QN->Components->Count - 2];
1557 SIN->Class = static_cast<IdentifierNode *>(ClassNode);
1558 }
1559 assert(QN);
1560 return QN;
1561}
1562
1564Demangler::demangleUnqualifiedTypeName(std::string_view &MangledName,
1565 bool Memorize) {
1566 // An inner-most name can be a back-reference, because a fully-qualified name
1567 // (e.g. Scope + Inner) can contain other fully qualified names inside of
1568 // them (for example template parameters), and these nested parameters can
1569 // refer to previously mangled types.
1570 if (startsWithDigit(MangledName))
1571 return demangleBackRefName(MangledName);
1572
1573 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1574 return demangleTemplateInstantiationName(MangledName, NBB_Template);
1575
1576 return demangleSimpleName(MangledName, Memorize);
1577}
1578
1580Demangler::demangleUnqualifiedSymbolName(std::string_view &MangledName,
1581 NameBackrefBehavior NBB) {
1582 if (startsWithDigit(MangledName))
1583 return demangleBackRefName(MangledName);
1584 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1585 return demangleTemplateInstantiationName(MangledName, NBB);
1586 if (llvm::itanium_demangle::starts_with(MangledName, '?'))
1587 return demangleFunctionIdentifierCode(MangledName);
1588 return demangleSimpleName(MangledName, /*Memorize=*/(NBB & NBB_Simple) != 0);
1589}
1590
1592Demangler::demangleNameScopePiece(std::string_view &MangledName) {
1593 if (startsWithDigit(MangledName))
1594 return demangleBackRefName(MangledName);
1595
1596 if (llvm::itanium_demangle::starts_with(MangledName, "?$"))
1597 return demangleTemplateInstantiationName(MangledName, NBB_Template);
1598
1599 if (llvm::itanium_demangle::starts_with(MangledName, "?A"))
1600 return demangleAnonymousNamespaceName(MangledName);
1601
1602 if (startsWithLocalScopePattern(MangledName))
1603 return demangleLocallyScopedNamePiece(MangledName);
1604
1605 return demangleSimpleName(MangledName, /*Memorize=*/true);
1606}
1607
1609 size_t Count) {
1610 NodeArrayNode *N = Arena.alloc<NodeArrayNode>();
1611 N->Count = Count;
1612 N->Nodes = Arena.allocArray<Node *>(Count);
1613 for (size_t I = 0; I < Count; ++I) {
1614 N->Nodes[I] = Head->N;
1615 Head = Head->Next;
1616 }
1617 return N;
1618}
1619
1621Demangler::demangleNameScopeChain(std::string_view &MangledName,
1622 IdentifierNode *UnqualifiedName) {
1623 NodeList *Head = Arena.alloc<NodeList>();
1624
1625 Head->N = UnqualifiedName;
1626
1627 size_t Count = 1;
1628 while (!consumeFront(MangledName, "@")) {
1629 ++Count;
1630 NodeList *NewHead = Arena.alloc<NodeList>();
1631 NewHead->Next = Head;
1632 Head = NewHead;
1633
1634 if (MangledName.empty()) {
1635 Error = true;
1636 return nullptr;
1637 }
1638
1639 assert(!Error);
1640 IdentifierNode *Elem = demangleNameScopePiece(MangledName);
1641 if (Error)
1642 return nullptr;
1643
1644 Head->N = Elem;
1645 }
1646
1648 QN->Components = nodeListToNodeArray(Arena, Head, Count);
1649 return QN;
1650}
1651
1652FuncClass Demangler::demangleFunctionClass(std::string_view &MangledName) {
1653 const char F = MangledName.front();
1654 MangledName.remove_prefix(1);
1655 switch (F) {
1656 case '9':
1658 case 'A':
1659 return FC_Private;
1660 case 'B':
1661 return FuncClass(FC_Private | FC_Far);
1662 case 'C':
1663 return FuncClass(FC_Private | FC_Static);
1664 case 'D':
1666 case 'E':
1668 case 'F':
1670 case 'G':
1672 case 'H':
1674 case 'I':
1675 return FuncClass(FC_Protected);
1676 case 'J':
1677 return FuncClass(FC_Protected | FC_Far);
1678 case 'K':
1680 case 'L':
1682 case 'M':
1684 case 'N':
1686 case 'O':
1688 case 'P':
1690 case 'Q':
1691 return FuncClass(FC_Public);
1692 case 'R':
1693 return FuncClass(FC_Public | FC_Far);
1694 case 'S':
1695 return FuncClass(FC_Public | FC_Static);
1696 case 'T':
1697 return FuncClass(FC_Public | FC_Static | FC_Far);
1698 case 'U':
1699 return FuncClass(FC_Public | FC_Virtual);
1700 case 'V':
1702 case 'W':
1704 case 'X':
1706 case 'Y':
1707 return FuncClass(FC_Global);
1708 case 'Z':
1709 return FuncClass(FC_Global | FC_Far);
1710 case '$': {
1712 if (consumeFront(MangledName, 'R'))
1713 VFlag = FuncClass(VFlag | FC_VirtualThisAdjustEx);
1714 if (MangledName.empty())
1715 break;
1716 const char F = MangledName.front();
1717 MangledName.remove_prefix(1);
1718 switch (F) {
1719 case '0':
1720 return FuncClass(FC_Private | FC_Virtual | VFlag);
1721 case '1':
1722 return FuncClass(FC_Private | FC_Virtual | VFlag | FC_Far);
1723 case '2':
1724 return FuncClass(FC_Protected | FC_Virtual | VFlag);
1725 case '3':
1726 return FuncClass(FC_Protected | FC_Virtual | VFlag | FC_Far);
1727 case '4':
1728 return FuncClass(FC_Public | FC_Virtual | VFlag);
1729 case '5':
1730 return FuncClass(FC_Public | FC_Virtual | VFlag | FC_Far);
1731 }
1732 }
1733 }
1734
1735 Error = true;
1736 return FC_Public;
1737}
1738
1740Demangler::demangleCallingConvention(std::string_view &MangledName) {
1741 if (MangledName.empty()) {
1742 Error = true;
1743 return CallingConv::None;
1744 }
1745
1746 const char F = MangledName.front();
1747 MangledName.remove_prefix(1);
1748 switch (F) {
1749 case 'A':
1750 case 'B':
1751 return CallingConv::Cdecl;
1752 case 'C':
1753 case 'D':
1754 return CallingConv::Pascal;
1755 case 'E':
1756 case 'F':
1757 return CallingConv::Thiscall;
1758 case 'G':
1759 case 'H':
1760 return CallingConv::Stdcall;
1761 case 'I':
1762 case 'J':
1763 return CallingConv::Fastcall;
1764 case 'M':
1765 case 'N':
1766 return CallingConv::Clrcall;
1767 case 'O':
1768 case 'P':
1769 return CallingConv::Eabi;
1770 case 'Q':
1772 case 'S':
1773 return CallingConv::Swift;
1774 case 'W':
1776 }
1777
1778 return CallingConv::None;
1779}
1780
1782Demangler::demangleVariableStorageClass(std::string_view &MangledName) {
1783 assert(MangledName.front() >= '0' && MangledName.front() <= '4');
1784
1785 const char F = MangledName.front();
1786 MangledName.remove_prefix(1);
1787 switch (F) {
1788 case '0':
1790 case '1':
1792 case '2':
1794 case '3':
1795 return StorageClass::Global;
1796 case '4':
1798 }
1800}
1801
1802std::pair<Qualifiers, bool>
1803Demangler::demangleQualifiers(std::string_view &MangledName) {
1804 if (MangledName.empty()) {
1805 Error = true;
1806 return std::make_pair(Q_None, false);
1807 }
1808
1809 const char F = MangledName.front();
1810 MangledName.remove_prefix(1);
1811 switch (F) {
1812 // Member qualifiers
1813 case 'Q':
1814 return std::make_pair(Q_None, true);
1815 case 'R':
1816 return std::make_pair(Q_Const, true);
1817 case 'S':
1818 return std::make_pair(Q_Volatile, true);
1819 case 'T':
1820 return std::make_pair(Qualifiers(Q_Const | Q_Volatile), true);
1821 // Non-Member qualifiers
1822 case 'A':
1823 return std::make_pair(Q_None, false);
1824 case 'B':
1825 return std::make_pair(Q_Const, false);
1826 case 'C':
1827 return std::make_pair(Q_Volatile, false);
1828 case 'D':
1829 return std::make_pair(Qualifiers(Q_Const | Q_Volatile), false);
1830 }
1831 Error = true;
1832 return std::make_pair(Q_None, false);
1833}
1834
1835// <variable-type> ::= <type> <cvr-qualifiers>
1836// ::= <type> <pointee-cvr-qualifiers> # pointers, references
1837TypeNode *Demangler::demangleType(std::string_view &MangledName,
1838 QualifierMangleMode QMM) {
1839 Qualifiers Quals = Q_None;
1840 bool IsMember = false;
1841 if (QMM == QualifierMangleMode::Mangle) {
1842 std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1843 } else if (QMM == QualifierMangleMode::Result) {
1844 if (consumeFront(MangledName, '?'))
1845 std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1846 }
1847
1848 if (MangledName.empty()) {
1849 Error = true;
1850 return nullptr;
1851 }
1852
1853 TypeNode *Ty = nullptr;
1854 if (isTagType(MangledName))
1855 Ty = demangleClassType(MangledName);
1856 else if (isPointerType(MangledName)) {
1857 if (isMemberPointer(MangledName, Error))
1858 Ty = demangleMemberPointerType(MangledName);
1859 else if (!Error)
1860 Ty = demanglePointerType(MangledName);
1861 else
1862 return nullptr;
1863 } else if (isArrayType(MangledName))
1864 Ty = demangleArrayType(MangledName);
1865 else if (isFunctionType(MangledName)) {
1866 if (consumeFront(MangledName, "$$A8@@"))
1867 Ty = demangleFunctionType(MangledName, true);
1868 else {
1869 assert(llvm::itanium_demangle::starts_with(MangledName, "$$A6"));
1870 consumeFront(MangledName, "$$A6");
1871 Ty = demangleFunctionType(MangledName, false);
1872 }
1873 } else if (isCustomType(MangledName)) {
1874 Ty = demangleCustomType(MangledName);
1875 } else {
1876 Ty = demanglePrimitiveType(MangledName);
1877 }
1878
1879 if (!Ty || Error)
1880 return Ty;
1881 Ty->Quals = Qualifiers(Ty->Quals | Quals);
1882 return Ty;
1883}
1884
1885bool Demangler::demangleThrowSpecification(std::string_view &MangledName) {
1886 if (consumeFront(MangledName, "_E"))
1887 return true;
1888 if (consumeFront(MangledName, 'Z'))
1889 return false;
1890
1891 Error = true;
1892 return false;
1893}
1894
1896Demangler::demangleFunctionType(std::string_view &MangledName,
1897 bool HasThisQuals) {
1899
1900 if (HasThisQuals) {
1901 FTy->Quals = demanglePointerExtQualifiers(MangledName);
1902 FTy->RefQualifier = demangleFunctionRefQualifier(MangledName);
1903 FTy->Quals = Qualifiers(FTy->Quals | demangleQualifiers(MangledName).first);
1904 }
1905
1906 // Fields that appear on both member and non-member functions.
1907 FTy->CallConvention = demangleCallingConvention(MangledName);
1908
1909 // <return-type> ::= <type>
1910 // ::= @ # structors (they have no declared return type)
1911 bool IsStructor = consumeFront(MangledName, '@');
1912 if (!IsStructor)
1913 FTy->ReturnType = demangleType(MangledName, QualifierMangleMode::Result);
1914
1915 FTy->Params = demangleFunctionParameterList(MangledName, FTy->IsVariadic);
1916
1917 FTy->IsNoexcept = demangleThrowSpecification(MangledName);
1918
1919 return FTy;
1920}
1921
1923Demangler::demangleFunctionEncoding(std::string_view &MangledName) {
1924 FuncClass ExtraFlags = FC_None;
1925 if (consumeFront(MangledName, "$$J0"))
1926 ExtraFlags = FC_ExternC;
1927
1928 if (MangledName.empty()) {
1929 Error = true;
1930 return nullptr;
1931 }
1932
1933 FuncClass FC = demangleFunctionClass(MangledName);
1934 FC = FuncClass(ExtraFlags | FC);
1935
1936 FunctionSignatureNode *FSN = nullptr;
1937 ThunkSignatureNode *TTN = nullptr;
1938 if (FC & FC_StaticThisAdjust) {
1939 TTN = Arena.alloc<ThunkSignatureNode>();
1940 TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1941 } else if (FC & FC_VirtualThisAdjust) {
1942 TTN = Arena.alloc<ThunkSignatureNode>();
1943 if (FC & FC_VirtualThisAdjustEx) {
1944 TTN->ThisAdjust.VBPtrOffset = demangleSigned(MangledName);
1945 TTN->ThisAdjust.VBOffsetOffset = demangleSigned(MangledName);
1946 }
1947 TTN->ThisAdjust.VtordispOffset = demangleSigned(MangledName);
1948 TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1949 }
1950
1951 if (FC & FC_NoParameterList) {
1952 // This is an extern "C" function whose full signature hasn't been mangled.
1953 // This happens when we need to mangle a local symbol inside of an extern
1954 // "C" function.
1955 FSN = Arena.alloc<FunctionSignatureNode>();
1956 } else {
1957 bool HasThisQuals = !(FC & (FC_Global | FC_Static));
1958 FSN = demangleFunctionType(MangledName, HasThisQuals);
1959 }
1960
1961 if (Error)
1962 return nullptr;
1963
1964 if (TTN) {
1965 *static_cast<FunctionSignatureNode *>(TTN) = *FSN;
1966 FSN = TTN;
1967 }
1968 FSN->FunctionClass = FC;
1969
1971 Symbol->Signature = FSN;
1972 return Symbol;
1973}
1974
1975CustomTypeNode *Demangler::demangleCustomType(std::string_view &MangledName) {
1976 assert(llvm::itanium_demangle::starts_with(MangledName, '?'));
1977 MangledName.remove_prefix(1);
1978
1979 CustomTypeNode *CTN = Arena.alloc<CustomTypeNode>();
1980 CTN->Identifier = demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1981 if (!consumeFront(MangledName, '@'))
1982 Error = true;
1983 if (Error)
1984 return nullptr;
1985 return CTN;
1986}
1987
1988// Reads a primitive type.
1990Demangler::demanglePrimitiveType(std::string_view &MangledName) {
1991 if (consumeFront(MangledName, "$$T"))
1993
1994 const char F = MangledName.front();
1995 MangledName.remove_prefix(1);
1996 switch (F) {
1997 case 'X':
1999 case 'D':
2001 case 'C':
2003 case 'E':
2005 case 'F':
2007 case 'G':
2009 case 'H':
2011 case 'I':
2013 case 'J':
2015 case 'K':
2017 case 'M':
2019 case 'N':
2021 case 'O':
2023 case '_': {
2024 if (MangledName.empty()) {
2025 Error = true;
2026 return nullptr;
2027 }
2028 const char F = MangledName.front();
2029 MangledName.remove_prefix(1);
2030 switch (F) {
2031 case 'N':
2033 case 'J':
2035 case 'K':
2037 case 'W':
2039 case 'Q':
2041 case 'S':
2043 case 'U':
2045 case 'P':
2047 case 'T':
2049 }
2050 break;
2051 }
2052 }
2053 Error = true;
2054 return nullptr;
2055}
2056
2057TagTypeNode *Demangler::demangleClassType(std::string_view &MangledName) {
2058 TagTypeNode *TT = nullptr;
2059
2060 const char F = MangledName.front();
2061 MangledName.remove_prefix(1);
2062 switch (F) {
2063 case 'T':
2065 break;
2066 case 'U':
2068 break;
2069 case 'V':
2071 break;
2072 case 'W':
2073 if (!consumeFront(MangledName, '4')) {
2074 Error = true;
2075 return nullptr;
2076 }
2078 break;
2079 default:
2080 assert(false);
2081 }
2082
2083 TT->QualifiedName = demangleFullyQualifiedTypeName(MangledName);
2084 return TT;
2085}
2086
2087// <pointer-type> ::= E? <pointer-cvr-qualifiers> <ext-qualifiers> <type>
2088// # the E is required for 64-bit non-static pointers
2089PointerTypeNode *Demangler::demanglePointerType(std::string_view &MangledName) {
2091
2092 std::tie(Pointer->Quals, Pointer->Affinity) =
2093 demanglePointerCVQualifiers(MangledName);
2094
2095 if (consumeFront(MangledName, "6")) {
2096 Pointer->Pointee = demangleFunctionType(MangledName, false);
2097 return Pointer;
2098 }
2099
2100 Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
2101 Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
2102
2103 Pointer->PointerAuthQualifier = createPointerAuthQualifier(MangledName);
2104
2105 Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Mangle);
2106 return Pointer;
2107}
2108
2110Demangler::demangleMemberPointerType(std::string_view &MangledName) {
2112
2113 std::tie(Pointer->Quals, Pointer->Affinity) =
2114 demanglePointerCVQualifiers(MangledName);
2116
2117 Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
2118 Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
2119
2120 // isMemberPointer() only returns true if there is at least one character
2121 // after the qualifiers.
2122 if (consumeFront(MangledName, "8")) {
2123 Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
2124 Pointer->Pointee = demangleFunctionType(MangledName, true);
2125 } else {
2126 Qualifiers PointeeQuals = Q_None;
2127 bool IsMember = false;
2128 std::tie(PointeeQuals, IsMember) = demangleQualifiers(MangledName);
2129 assert(IsMember || Error);
2130 Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
2131
2132 Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Drop);
2133 if (Pointer->Pointee)
2134 Pointer->Pointee->Quals = PointeeQuals;
2135 }
2136
2137 return Pointer;
2138}
2139
2141Demangler::demanglePointerExtQualifiers(std::string_view &MangledName) {
2142 Qualifiers Quals = Q_None;
2143 if (consumeFront(MangledName, 'E'))
2144 Quals = Qualifiers(Quals | Q_Pointer64);
2145 if (consumeFront(MangledName, 'I'))
2146 Quals = Qualifiers(Quals | Q_Restrict);
2147 if (consumeFront(MangledName, 'F'))
2148 Quals = Qualifiers(Quals | Q_Unaligned);
2149
2150 return Quals;
2151}
2152
2153std::optional<PointerAuthQualifierNode::ArgArray>
2154Demangler::demanglePointerAuthQualifier(std::string_view &MangledName) {
2155 if (!consumeFront(MangledName, "__ptrauth"))
2156 return std::nullopt;
2157
2158 constexpr unsigned NumArgs = PointerAuthQualifierNode::NumArgs;
2160
2161 for (unsigned I = 0; I < NumArgs; ++I) {
2162 bool IsNegative = false;
2163 uint64_t Value = 0;
2164 std::tie(Value, IsNegative) = demangleNumber(MangledName);
2165 if (IsNegative)
2166 return std::nullopt;
2167
2168 Array[I] = Value;
2169 }
2170
2171 return Array;
2172}
2173
2175Demangler::createPointerAuthQualifier(std::string_view &MangledName) {
2176 constexpr unsigned NumArgs = PointerAuthQualifierNode::NumArgs;
2177 std::optional<PointerAuthQualifierNode::ArgArray> Vals =
2178 demanglePointerAuthQualifier(MangledName);
2179
2180 if (!Vals)
2181 return nullptr;
2182
2183 PointerAuthQualifierNode *PtrAuthQual =
2186 PtrAuthQual->Components = Array;
2187 Array->Count = NumArgs;
2188 Array->Nodes = Arena.allocArray<Node *>(NumArgs);
2189
2190 for (unsigned I = 0; I < NumArgs; ++I)
2191 Array->Nodes[I] = Arena.alloc<IntegerLiteralNode>((*Vals)[I], false);
2192
2193 return PtrAuthQual;
2194}
2195
2196ArrayTypeNode *Demangler::demangleArrayType(std::string_view &MangledName) {
2197 assert(MangledName.front() == 'Y');
2198 MangledName.remove_prefix(1);
2199
2200 uint64_t Rank = 0;
2201 bool IsNegative = false;
2202 std::tie(Rank, IsNegative) = demangleNumber(MangledName);
2203 if (IsNegative || Rank == 0) {
2204 Error = true;
2205 return nullptr;
2206 }
2207
2208 ArrayTypeNode *ATy = Arena.alloc<ArrayTypeNode>();
2209 NodeList *Head = Arena.alloc<NodeList>();
2210 NodeList *Tail = Head;
2211
2212 for (uint64_t I = 0; I < Rank; ++I) {
2213 uint64_t D = 0;
2214 std::tie(D, IsNegative) = demangleNumber(MangledName);
2215 if (Error || IsNegative) {
2216 Error = true;
2217 return nullptr;
2218 }
2219 Tail->N = Arena.alloc<IntegerLiteralNode>(D, IsNegative);
2220 if (I + 1 < Rank) {
2221 Tail->Next = Arena.alloc<NodeList>();
2222 Tail = Tail->Next;
2223 }
2224 }
2225 ATy->Dimensions = nodeListToNodeArray(Arena, Head, Rank);
2226
2227 if (consumeFront(MangledName, "$$C")) {
2228 bool IsMember = false;
2229 std::tie(ATy->Quals, IsMember) = demangleQualifiers(MangledName);
2230 if (IsMember) {
2231 Error = true;
2232 return nullptr;
2233 }
2234 }
2235
2236 ATy->ElementType = demangleType(MangledName, QualifierMangleMode::Drop);
2237 return ATy;
2238}
2239
2240// Reads a function's parameters.
2242Demangler::demangleFunctionParameterList(std::string_view &MangledName,
2243 bool &IsVariadic) {
2244 // Empty parameter list.
2245 if (consumeFront(MangledName, 'X'))
2246 return nullptr;
2247
2248 NodeList *Head = Arena.alloc<NodeList>();
2249 NodeList **Current = &Head;
2250 size_t Count = 0;
2251 while (!Error && !llvm::itanium_demangle::starts_with(MangledName, '@') &&
2252 !llvm::itanium_demangle::starts_with(MangledName, 'Z')) {
2253 ++Count;
2254
2255 if (startsWithDigit(MangledName)) {
2256 size_t N = MangledName[0] - '0';
2257 if (N >= Backrefs.FunctionParamCount) {
2258 Error = true;
2259 return nullptr;
2260 }
2261 MangledName.remove_prefix(1);
2262
2263 *Current = Arena.alloc<NodeList>();
2264 (*Current)->N = Backrefs.FunctionParams[N];
2265 Current = &(*Current)->Next;
2266 continue;
2267 }
2268
2269 size_t OldSize = MangledName.size();
2270
2271 *Current = Arena.alloc<NodeList>();
2272 TypeNode *TN = demangleType(MangledName, QualifierMangleMode::Drop);
2273 if (!TN || Error)
2274 return nullptr;
2275
2276 (*Current)->N = TN;
2277
2278 size_t CharsConsumed = OldSize - MangledName.size();
2279 assert(CharsConsumed != 0);
2280
2281 // Single-letter types are ignored for backreferences because memorizing
2282 // them doesn't save anything.
2283 if (Backrefs.FunctionParamCount <= 9 && CharsConsumed > 1)
2284 Backrefs.FunctionParams[Backrefs.FunctionParamCount++] = TN;
2285
2286 Current = &(*Current)->Next;
2287 }
2288
2289 if (Error)
2290 return nullptr;
2291
2292 NodeArrayNode *NA = nodeListToNodeArray(Arena, Head, Count);
2293 // A non-empty parameter list is terminated by either 'Z' (variadic) parameter
2294 // list or '@' (non variadic). Careful not to consume "@Z", as in that case
2295 // the following Z could be a throw specifier.
2296 if (consumeFront(MangledName, '@'))
2297 return NA;
2298
2299 if (consumeFront(MangledName, 'Z')) {
2300 IsVariadic = true;
2301 return NA;
2302 }
2303
2305}
2306
2308Demangler::demangleTemplateParameterList(std::string_view &MangledName) {
2309 NodeList *Head = nullptr;
2310 NodeList **Current = &Head;
2311 size_t Count = 0;
2312
2313 while (!llvm::itanium_demangle::starts_with(MangledName, '@')) {
2314 if (consumeFront(MangledName, "$S") || consumeFront(MangledName, "$$V") ||
2315 consumeFront(MangledName, "$$$V") || consumeFront(MangledName, "$$Z")) {
2316 // parameter pack separator
2317 continue;
2318 }
2319
2320 ++Count;
2321
2322 // Template parameter lists don't participate in back-referencing.
2323 *Current = Arena.alloc<NodeList>();
2324
2325 NodeList &TP = **Current;
2326
2327 // <auto-nttp> ::= $ M <type> <nttp>
2328 const bool IsAutoNTTP = consumeFront(MangledName, "$M");
2329 if (IsAutoNTTP) {
2330 // The deduced type of the auto NTTP parameter isn't printed so
2331 // we want to ignore the AST created from demangling the type.
2332 //
2333 // TODO: Avoid the extra allocations to the bump allocator in this case.
2334 (void)demangleType(MangledName, QualifierMangleMode::Drop);
2335 if (Error)
2336 return nullptr;
2337 }
2338
2339 TemplateParameterReferenceNode *TPRN = nullptr;
2340 if (consumeFront(MangledName, "$$Y")) {
2341 // Template alias
2342 TP.N = demangleFullyQualifiedTypeName(MangledName);
2343 } else if (consumeFront(MangledName, "$$B")) {
2344 // Array
2345 TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2346 } else if (consumeFront(MangledName, "$$C")) {
2347 // Type has qualifiers.
2348 TP.N = demangleType(MangledName, QualifierMangleMode::Mangle);
2349 } else if (startsWith(MangledName, "$1", "1", !IsAutoNTTP) ||
2350 startsWith(MangledName, "$H", "H", !IsAutoNTTP) ||
2351 startsWith(MangledName, "$I", "I", !IsAutoNTTP) ||
2352 startsWith(MangledName, "$J", "J", !IsAutoNTTP)) {
2353 // Pointer to member
2354 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2355 TPRN->IsMemberPointer = true;
2356
2357 if (!IsAutoNTTP)
2358 MangledName.remove_prefix(1); // Remove leading '$'
2359
2360 // 1 - single inheritance <name>
2361 // H - multiple inheritance <name> <number>
2362 // I - virtual inheritance <name> <number> <number>
2363 // J - unspecified inheritance <name> <number> <number> <number>
2364 char InheritanceSpecifier = MangledName.front();
2365 MangledName.remove_prefix(1);
2366 SymbolNode *S = nullptr;
2367 if (llvm::itanium_demangle::starts_with(MangledName, '?')) {
2368 S = parse(MangledName);
2369 if (Error || !S->Name) {
2370 Error = true;
2371 return nullptr;
2372 }
2373 memorizeIdentifier(S->Name->getUnqualifiedIdentifier());
2374 }
2375
2376 switch (InheritanceSpecifier) {
2377 case 'J':
2378 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2379 demangleSigned(MangledName);
2381 case 'I':
2382 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2383 demangleSigned(MangledName);
2385 case 'H':
2386 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2387 demangleSigned(MangledName);
2389 case '1':
2390 break;
2391 default:
2393 }
2395 TPRN->Symbol = S;
2396 } else if (llvm::itanium_demangle::starts_with(MangledName, "$E?")) {
2397 consumeFront(MangledName, "$E");
2398 // Reference to symbol
2399 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2400 TPRN->Symbol = parse(MangledName);
2402 } else if (startsWith(MangledName, "$F", "F", !IsAutoNTTP) ||
2403 startsWith(MangledName, "$G", "G", !IsAutoNTTP)) {
2404 TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2405
2406 // Data member pointer.
2407 if (!IsAutoNTTP)
2408 MangledName.remove_prefix(1); // Remove leading '$'
2409 char InheritanceSpecifier = MangledName.front();
2410 MangledName.remove_prefix(1);
2411
2412 switch (InheritanceSpecifier) {
2413 case 'G':
2414 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2415 demangleSigned(MangledName);
2417 case 'F':
2418 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2419 demangleSigned(MangledName);
2420 TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2421 demangleSigned(MangledName);
2422 break;
2423 default:
2425 }
2426 TPRN->IsMemberPointer = true;
2427
2428 } else if (consumeFront(MangledName, "$0", "0", !IsAutoNTTP)) {
2429 // Integral non-type template parameter
2430 bool IsNegative = false;
2431 uint64_t Value = 0;
2432 std::tie(Value, IsNegative) = demangleNumber(MangledName);
2433
2434 TP.N = Arena.alloc<IntegerLiteralNode>(Value, IsNegative);
2435 } else {
2436 TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2437 }
2438 if (Error)
2439 return nullptr;
2440
2441 Current = &TP.Next;
2442 }
2443
2444 // The loop above returns nullptr on Error.
2445 assert(!Error);
2446
2447 // Template parameter lists cannot be variadic, so it can only be terminated
2448 // by @ (as opposed to 'Z' in the function parameter case).
2449 assert(llvm::itanium_demangle::starts_with(
2450 MangledName, '@')); // The above loop exits only on '@'.
2451 consumeFront(MangledName, '@');
2452 return nodeListToNodeArray(Arena, Head, Count);
2453}
2454
2455void Demangler::dumpBackReferences() {
2456 std::printf("%d function parameter backreferences\n",
2457 (int)Backrefs.FunctionParamCount);
2458
2459 // Create an output stream so we can render each type.
2460 OutputBuffer OB;
2461 for (size_t I = 0; I < Backrefs.FunctionParamCount; ++I) {
2462 OB.setCurrentPosition(0);
2463
2464 TypeNode *T = Backrefs.FunctionParams[I];
2465 T->output(OB, OF_Default);
2466
2467 std::string_view B = OB;
2468 std::printf(" [%d] - %.*s\n", (int)I, (int)B.size(), B.data());
2469 }
2470 std::free(OB.getBuffer());
2471
2472 if (Backrefs.FunctionParamCount > 0)
2473 std::printf("\n");
2474 std::printf("%d name backreferences\n", (int)Backrefs.NamesCount);
2475 for (size_t I = 0; I < Backrefs.NamesCount; ++I) {
2476 std::printf(" [%d] - %.*s\n", (int)I, (int)Backrefs.Names[I]->Name.size(),
2477 Backrefs.Names[I]->Name.data());
2478 }
2479 if (Backrefs.NamesCount > 0)
2480 std::printf("\n");
2481}
2482
2483std::optional<size_t>
2484llvm::getArm64ECInsertionPointInMangledName(std::string_view MangledName) {
2485 std::string_view ProcessedName{MangledName};
2486
2487 // We only support this for MSVC-style C++ symbols.
2488 if (!consumeFront(ProcessedName, '?'))
2489 return std::nullopt;
2490
2491 // The insertion point is just after the name of the symbol, so parse that to
2492 // remove it from the processed name.
2493 Demangler D;
2494 D.demangleFullyQualifiedSymbolName(ProcessedName);
2495 if (D.Error)
2496 return std::nullopt;
2497
2498 return MangledName.length() - ProcessedName.length();
2499}
2500
2501char *llvm::microsoftDemangle(std::string_view MangledName, size_t *NMangled,
2502 int *Status, MSDemangleFlags Flags) {
2503 Demangler D;
2504
2505 std::string_view Name{MangledName};
2506 SymbolNode *AST = D.parse(Name);
2507 if (!D.Error && NMangled)
2508 *NMangled = MangledName.size() - Name.size();
2509
2510 if (Flags & MSDF_DumpBackrefs)
2511 D.dumpBackReferences();
2512
2514 if (Flags & MSDF_NoCallingConvention)
2516 if (Flags & MSDF_NoAccessSpecifier)
2518 if (Flags & MSDF_NoReturnType)
2519 OF = OutputFlags(OF | OF_NoReturnType);
2520 if (Flags & MSDF_NoMemberType)
2521 OF = OutputFlags(OF | OF_NoMemberType);
2522 if (Flags & MSDF_NoVariableType)
2523 OF = OutputFlags(OF | OF_NoVariableType);
2524
2525 int InternalStatus = demangle_success;
2526 char *Buf;
2527 if (D.Error)
2528 InternalStatus = demangle_invalid_mangled_name;
2529 else {
2530 OutputBuffer OB;
2531 AST->output(OB, OF);
2532 OB += '\0';
2533 Buf = OB.getBuffer();
2534 }
2535
2536 if (Status)
2537 *Status = InternalStatus;
2538 return InternalStatus == demangle_success ? Buf : nullptr;
2539}
assert(UImm &&(UImm !=~static_cast< T >(0)) &&"Invalid immediate!")
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 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(PassOpts->AAPipeline)
#define CH(x, y, z)
Definition: SHA256.cpp:34
static int Lookup(ArrayRef< TableEntry > Table, unsigned Opcode)
Lightweight error class with error context and mandatory checking.
Definition: Error.h:159
Definition: MD5.h:42
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:75
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
@ OB
OB - OneByte - Set if this instruction has a one byte opcode.
Definition: X86BaseInfo.h:732
StorageClass
Definition: XCOFF.h:171
This is an optimization pass for GlobalISel generic memory operations.
Definition: AddressRanges.h:18
@ Offset
Definition: DWP.cpp:477
@ Length
Definition: DWP.cpp:477
DEMANGLE_ABI std::optional< size_t > getArm64ECInsertionPointInMangledName(std::string_view MangledName)
@ demangle_success
Definition: Demangle.h:31
@ demangle_invalid_mangled_name
Definition: Demangle.h:29
DEMANGLE_ABI 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:40
@ MSDF_NoReturnType
Definition: Demangle.h:45
@ MSDF_DumpBackrefs
Definition: Demangle.h:42
@ MSDF_NoMemberType
Definition: Demangle.h:46
@ MSDF_NoVariableType
Definition: Demangle.h:47
@ MSDF_NoCallingConvention
Definition: Demangle.h:44
@ MSDF_NoAccessSpecifier
Definition: Demangle.h:43
void swap(llvm::BitVector &LHS, llvm::BitVector &RHS)
Implement std::swap in terms of BitVector swap.
Definition: BitVector.h:858
#define N
NodeList * Next
NamedIdentifierNode * Names[Max]
void output(OutputBuffer &OB, OutputFlags Flags) const override
Definition: regcomp.c:186