xref: /freebsd/contrib/llvm-project/clang/lib/AST/FormatString.cpp (revision aae38d10b4eebf81c0942947e8b83a9bb8651d88) 
1 // FormatString.cpp - Common stuff for handling printf/scanf formats -*- C++ -*-
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 // Shared details for processing format strings of printf and scanf
10 // (and friends).
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "FormatStringParsing.h"
15 #include "clang/Basic/LangOptions.h"
16 #include "clang/Basic/TargetInfo.h"
17 #include "llvm/Support/ConvertUTF.h"
18 
19 using clang::analyze_format_string::ArgType;
20 using clang::analyze_format_string::FormatStringHandler;
21 using clang::analyze_format_string::FormatSpecifier;
22 using clang::analyze_format_string::LengthModifier;
23 using clang::analyze_format_string::OptionalAmount;
24 using clang::analyze_format_string::PositionContext;
25 using clang::analyze_format_string::ConversionSpecifier;
26 using namespace clang;
27 
28 // Key function to FormatStringHandler.
29 FormatStringHandler::~FormatStringHandler() {}
30 
31 //===----------------------------------------------------------------------===//
32 // Functions for parsing format strings components in both printf and
33 // scanf format strings.
34 //===----------------------------------------------------------------------===//
35 
36 OptionalAmount
37 clang::analyze_format_string::ParseAmount(const char *&Beg, const char *E) {
38   const char *I = Beg;
39   UpdateOnReturn <const char*> UpdateBeg(Beg, I);
40 
41   unsigned accumulator = 0;
42   bool hasDigits = false;
43 
44   for ( ; I != E; ++I) {
45     char c = *I;
46     if (c >= '0' && c <= '9') {
47       hasDigits = true;
48       accumulator = (accumulator * 10) + (c - '0');
49       continue;
50     }
51 
52     if (hasDigits)
53       return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg,
54           false);
55 
56     break;
57   }
58 
59   return OptionalAmount();
60 }
61 
62 OptionalAmount
63 clang::analyze_format_string::ParseNonPositionAmount(const char *&Beg,
64                                                      const char *E,
65                                                      unsigned &argIndex) {
66   if (*Beg == '*') {
67     ++Beg;
68     return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
69   }
70 
71   return ParseAmount(Beg, E);
72 }
73 
74 OptionalAmount
75 clang::analyze_format_string::ParsePositionAmount(FormatStringHandler &H,
76                                                   const char *Start,
77                                                   const char *&Beg,
78                                                   const char *E,
79                                                   PositionContext p) {
80   if (*Beg == '*') {
81     const char *I = Beg + 1;
82     const OptionalAmount &Amt = ParseAmount(I, E);
83 
84     if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
85       H.HandleInvalidPosition(Beg, I - Beg, p);
86       return OptionalAmount(false);
87     }
88 
89     if (I == E) {
90       // No more characters left?
91       H.HandleIncompleteSpecifier(Start, E - Start);
92       return OptionalAmount(false);
93     }
94 
95     assert(Amt.getHowSpecified() == OptionalAmount::Constant);
96 
97     if (*I == '$') {
98       // Handle positional arguments
99 
100       // Special case: '*0$', since this is an easy mistake.
101       if (Amt.getConstantAmount() == 0) {
102         H.HandleZeroPosition(Beg, I - Beg + 1);
103         return OptionalAmount(false);
104       }
105 
106       const char *Tmp = Beg;
107       Beg = ++I;
108 
109       return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1,
110                             Tmp, 0, true);
111     }
112 
113     H.HandleInvalidPosition(Beg, I - Beg, p);
114     return OptionalAmount(false);
115   }
116 
117   return ParseAmount(Beg, E);
118 }
119 
120 
121 bool
122 clang::analyze_format_string::ParseFieldWidth(FormatStringHandler &H,
123                                               FormatSpecifier &CS,
124                                               const char *Start,
125                                               const char *&Beg, const char *E,
126                                               unsigned *argIndex) {
127   // FIXME: Support negative field widths.
128   if (argIndex) {
129     CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
130   }
131   else {
132     const OptionalAmount Amt =
133       ParsePositionAmount(H, Start, Beg, E,
134                           analyze_format_string::FieldWidthPos);
135 
136     if (Amt.isInvalid())
137       return true;
138     CS.setFieldWidth(Amt);
139   }
140   return false;
141 }
142 
143 bool
144 clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
145                                                FormatSpecifier &FS,
146                                                const char *Start,
147                                                const char *&Beg,
148                                                const char *E) {
149   const char *I = Beg;
150 
151   const OptionalAmount &Amt = ParseAmount(I, E);
152 
153   if (I == E) {
154     // No more characters left?
155     H.HandleIncompleteSpecifier(Start, E - Start);
156     return true;
157   }
158 
159   if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
160     // Warn that positional arguments are non-standard.
161     H.HandlePosition(Start, I - Start);
162 
163     // Special case: '%0$', since this is an easy mistake.
164     if (Amt.getConstantAmount() == 0) {
165       H.HandleZeroPosition(Start, I - Start);
166       return true;
167     }
168 
169     FS.setArgIndex(Amt.getConstantAmount() - 1);
170     FS.setUsesPositionalArg();
171     // Update the caller's pointer if we decided to consume
172     // these characters.
173     Beg = I;
174     return false;
175   }
176 
177   return false;
178 }
179 
180 bool
181 clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
182                                                   FormatSpecifier &FS,
183                                                   const char *&I,
184                                                   const char *E,
185                                                   const LangOptions &LO) {
186   if (!LO.OpenCL)
187     return false;
188 
189   const char *Start = I;
190   if (*I == 'v') {
191     ++I;
192 
193     if (I == E) {
194       H.HandleIncompleteSpecifier(Start, E - Start);
195       return true;
196     }
197 
198     OptionalAmount NumElts = ParseAmount(I, E);
199     if (NumElts.getHowSpecified() != OptionalAmount::Constant) {
200       H.HandleIncompleteSpecifier(Start, E - Start);
201       return true;
202     }
203 
204     FS.setVectorNumElts(NumElts);
205   }
206 
207   return false;
208 }
209 
210 bool
211 clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
212                                                   const char *&I,
213                                                   const char *E,
214                                                   const LangOptions &LO,
215                                                   bool IsScanf) {
216   LengthModifier::Kind lmKind = LengthModifier::None;
217   const char *lmPosition = I;
218   switch (*I) {
219     default:
220       return false;
221     case 'h':
222       ++I;
223       if (I != E && *I == 'h') {
224         ++I;
225         lmKind = LengthModifier::AsChar;
226       } else if (I != E && *I == 'l' && LO.OpenCL) {
227         ++I;
228         lmKind = LengthModifier::AsShortLong;
229       } else {
230         lmKind = LengthModifier::AsShort;
231       }
232       break;
233     case 'l':
234       ++I;
235       if (I != E && *I == 'l') {
236         ++I;
237         lmKind = LengthModifier::AsLongLong;
238       } else {
239         lmKind = LengthModifier::AsLong;
240       }
241       break;
242     case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
243     case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
244     case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
245     case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
246     case 'q': lmKind = LengthModifier::AsQuad;       ++I; break;
247     case 'a':
248       if (IsScanf && !LO.C99 && !LO.CPlusPlus11) {
249         // For scanf in C90, look at the next character to see if this should
250         // be parsed as the GNU extension 'a' length modifier. If not, this
251         // will be parsed as a conversion specifier.
252         ++I;
253         if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
254           lmKind = LengthModifier::AsAllocate;
255           break;
256         }
257         --I;
258       }
259       return false;
260     case 'm':
261       if (IsScanf) {
262         lmKind = LengthModifier::AsMAllocate;
263         ++I;
264         break;
265       }
266       return false;
267     // printf: AsInt64, AsInt32, AsInt3264
268     // scanf:  AsInt64
269     case 'I':
270       if (I + 1 != E && I + 2 != E) {
271         if (I[1] == '6' && I[2] == '4') {
272           I += 3;
273           lmKind = LengthModifier::AsInt64;
274           break;
275         }
276         if (IsScanf)
277           return false;
278 
279         if (I[1] == '3' && I[2] == '2') {
280           I += 3;
281           lmKind = LengthModifier::AsInt32;
282           break;
283         }
284       }
285       ++I;
286       lmKind = LengthModifier::AsInt3264;
287       break;
288     case 'w':
289       lmKind = LengthModifier::AsWide; ++I; break;
290   }
291   LengthModifier lm(lmPosition, lmKind);
292   FS.setLengthModifier(lm);
293   return true;
294 }
295 
296 bool clang::analyze_format_string::ParseUTF8InvalidSpecifier(
297     const char *SpecifierBegin, const char *FmtStrEnd, unsigned &Len) {
298   if (SpecifierBegin + 1 >= FmtStrEnd)
299     return false;
300 
301   const llvm::UTF8 *SB =
302       reinterpret_cast<const llvm::UTF8 *>(SpecifierBegin + 1);
303   const llvm::UTF8 *SE = reinterpret_cast<const llvm::UTF8 *>(FmtStrEnd);
304   const char FirstByte = *SB;
305 
306   // If the invalid specifier is a multibyte UTF-8 string, return the
307   // total length accordingly so that the conversion specifier can be
308   // properly updated to reflect a complete UTF-8 specifier.
309   unsigned NumBytes = llvm::getNumBytesForUTF8(FirstByte);
310   if (NumBytes == 1)
311     return false;
312   if (SB + NumBytes > SE)
313     return false;
314 
315   Len = NumBytes + 1;
316   return true;
317 }
318 
319 //===----------------------------------------------------------------------===//
320 // Methods on ArgType.
321 //===----------------------------------------------------------------------===//
322 
323 clang::analyze_format_string::ArgType::MatchKind
324 ArgType::matchesType(ASTContext &C, QualType argTy) const {
325   if (Ptr) {
326     // It has to be a pointer.
327     const PointerType *PT = argTy->getAs<PointerType>();
328     if (!PT)
329       return NoMatch;
330 
331     // We cannot write through a const qualified pointer.
332     if (PT->getPointeeType().isConstQualified())
333       return NoMatch;
334 
335     argTy = PT->getPointeeType();
336   }
337 
338   switch (K) {
339     case InvalidTy:
340       llvm_unreachable("ArgType must be valid");
341 
342     case UnknownTy:
343       return Match;
344 
345     case AnyCharTy: {
346       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
347         // If the enum is incomplete we know nothing about the underlying type.
348         // Assume that it's 'int'.
349         if (!ETy->getDecl()->isComplete())
350           return NoMatch;
351         argTy = ETy->getDecl()->getIntegerType();
352       }
353 
354       if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
355         switch (BT->getKind()) {
356           default:
357             break;
358           case BuiltinType::Char_S:
359           case BuiltinType::SChar:
360           case BuiltinType::UChar:
361           case BuiltinType::Char_U:
362             return Match;
363         }
364       return NoMatch;
365     }
366 
367     case SpecificTy: {
368       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
369         // If the enum is incomplete we know nothing about the underlying type.
370         // Assume that it's 'int'.
371         if (!ETy->getDecl()->isComplete())
372           argTy = C.IntTy;
373         else
374           argTy = ETy->getDecl()->getIntegerType();
375       }
376       argTy = C.getCanonicalType(argTy).getUnqualifiedType();
377 
378       if (T == argTy)
379         return Match;
380       // Check for "compatible types".
381       if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
382         switch (BT->getKind()) {
383           default:
384             break;
385           case BuiltinType::Char_S:
386           case BuiltinType::SChar:
387           case BuiltinType::Char_U:
388           case BuiltinType::UChar:
389             return T == C.UnsignedCharTy || T == C.SignedCharTy ? Match
390                                                                 : NoMatch;
391           case BuiltinType::Short:
392             return T == C.UnsignedShortTy ? Match : NoMatch;
393           case BuiltinType::UShort:
394             return T == C.ShortTy ? Match : NoMatch;
395           case BuiltinType::Int:
396             return T == C.UnsignedIntTy ? Match : NoMatch;
397           case BuiltinType::UInt:
398             return T == C.IntTy ? Match : NoMatch;
399           case BuiltinType::Long:
400             return T == C.UnsignedLongTy ? Match : NoMatch;
401           case BuiltinType::ULong:
402             return T == C.LongTy ? Match : NoMatch;
403           case BuiltinType::LongLong:
404             return T == C.UnsignedLongLongTy ? Match : NoMatch;
405           case BuiltinType::ULongLong:
406             return T == C.LongLongTy ? Match : NoMatch;
407         }
408       return NoMatch;
409     }
410 
411     case CStrTy: {
412       const PointerType *PT = argTy->getAs<PointerType>();
413       if (!PT)
414         return NoMatch;
415       QualType pointeeTy = PT->getPointeeType();
416       if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
417         switch (BT->getKind()) {
418           case BuiltinType::Void:
419           case BuiltinType::Char_U:
420           case BuiltinType::UChar:
421           case BuiltinType::Char_S:
422           case BuiltinType::SChar:
423             return Match;
424           default:
425             break;
426         }
427 
428       return NoMatch;
429     }
430 
431     case WCStrTy: {
432       const PointerType *PT = argTy->getAs<PointerType>();
433       if (!PT)
434         return NoMatch;
435       QualType pointeeTy =
436         C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
437       return pointeeTy == C.getWideCharType() ? Match : NoMatch;
438     }
439 
440     case WIntTy: {
441       QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
442 
443       if (C.getCanonicalType(argTy).getUnqualifiedType() == WInt)
444         return Match;
445 
446       QualType PromoArg = argTy->isPromotableIntegerType()
447                               ? C.getPromotedIntegerType(argTy)
448                               : argTy;
449       PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
450 
451       // If the promoted argument is the corresponding signed type of the
452       // wint_t type, then it should match.
453       if (PromoArg->hasSignedIntegerRepresentation() &&
454           C.getCorrespondingUnsignedType(PromoArg) == WInt)
455         return Match;
456 
457       return WInt == PromoArg ? Match : NoMatch;
458     }
459 
460     case CPointerTy:
461       if (argTy->isVoidPointerType()) {
462         return Match;
463       } if (argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
464             argTy->isBlockPointerType() || argTy->isNullPtrType()) {
465         return NoMatchPedantic;
466       } else {
467         return NoMatch;
468       }
469 
470     case ObjCPointerTy: {
471       if (argTy->getAs<ObjCObjectPointerType>() ||
472           argTy->getAs<BlockPointerType>())
473         return Match;
474 
475       // Handle implicit toll-free bridging.
476       if (const PointerType *PT = argTy->getAs<PointerType>()) {
477         // Things such as CFTypeRef are really just opaque pointers
478         // to C structs representing CF types that can often be bridged
479         // to Objective-C objects.  Since the compiler doesn't know which
480         // structs can be toll-free bridged, we just accept them all.
481         QualType pointee = PT->getPointeeType();
482         if (pointee->getAsStructureType() || pointee->isVoidType())
483           return Match;
484       }
485       return NoMatch;
486     }
487   }
488 
489   llvm_unreachable("Invalid ArgType Kind!");
490 }
491 
492 ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const {
493   // Check for valid vector element types.
494   if (T.isNull())
495     return ArgType::Invalid();
496 
497   QualType Vec = C.getExtVectorType(T, NumElts);
498   return ArgType(Vec, Name);
499 }
500 
501 QualType ArgType::getRepresentativeType(ASTContext &C) const {
502   QualType Res;
503   switch (K) {
504     case InvalidTy:
505       llvm_unreachable("No representative type for Invalid ArgType");
506     case UnknownTy:
507       llvm_unreachable("No representative type for Unknown ArgType");
508     case AnyCharTy:
509       Res = C.CharTy;
510       break;
511     case SpecificTy:
512       Res = T;
513       break;
514     case CStrTy:
515       Res = C.getPointerType(C.CharTy);
516       break;
517     case WCStrTy:
518       Res = C.getPointerType(C.getWideCharType());
519       break;
520     case ObjCPointerTy:
521       Res = C.ObjCBuiltinIdTy;
522       break;
523     case CPointerTy:
524       Res = C.VoidPtrTy;
525       break;
526     case WIntTy: {
527       Res = C.getWIntType();
528       break;
529     }
530   }
531 
532   if (Ptr)
533     Res = C.getPointerType(Res);
534   return Res;
535 }
536 
537 std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
538   std::string S = getRepresentativeType(C).getAsString();
539 
540   std::string Alias;
541   if (Name) {
542     // Use a specific name for this type, e.g. "size_t".
543     Alias = Name;
544     if (Ptr) {
545       // If ArgType is actually a pointer to T, append an asterisk.
546       Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
547     }
548     // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
549     if (S == Alias)
550       Alias.clear();
551   }
552 
553   if (!Alias.empty())
554     return std::string("'") + Alias + "' (aka '" + S + "')";
555   return std::string("'") + S + "'";
556 }
557 
558 
559 //===----------------------------------------------------------------------===//
560 // Methods on OptionalAmount.
561 //===----------------------------------------------------------------------===//
562 
563 ArgType
564 analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
565   return Ctx.IntTy;
566 }
567 
568 //===----------------------------------------------------------------------===//
569 // Methods on LengthModifier.
570 //===----------------------------------------------------------------------===//
571 
572 const char *
573 analyze_format_string::LengthModifier::toString() const {
574   switch (kind) {
575   case AsChar:
576     return "hh";
577   case AsShort:
578     return "h";
579   case AsShortLong:
580     return "hl";
581   case AsLong: // or AsWideChar
582     return "l";
583   case AsLongLong:
584     return "ll";
585   case AsQuad:
586     return "q";
587   case AsIntMax:
588     return "j";
589   case AsSizeT:
590     return "z";
591   case AsPtrDiff:
592     return "t";
593   case AsInt32:
594     return "I32";
595   case AsInt3264:
596     return "I";
597   case AsInt64:
598     return "I64";
599   case AsLongDouble:
600     return "L";
601   case AsAllocate:
602     return "a";
603   case AsMAllocate:
604     return "m";
605   case AsWide:
606     return "w";
607   case None:
608     return "";
609   }
610   return nullptr;
611 }
612 
613 //===----------------------------------------------------------------------===//
614 // Methods on ConversionSpecifier.
615 //===----------------------------------------------------------------------===//
616 
617 const char *ConversionSpecifier::toString() const {
618   switch (kind) {
619   case dArg: return "d";
620   case DArg: return "D";
621   case iArg: return "i";
622   case oArg: return "o";
623   case OArg: return "O";
624   case uArg: return "u";
625   case UArg: return "U";
626   case xArg: return "x";
627   case XArg: return "X";
628   case fArg: return "f";
629   case FArg: return "F";
630   case eArg: return "e";
631   case EArg: return "E";
632   case gArg: return "g";
633   case GArg: return "G";
634   case aArg: return "a";
635   case AArg: return "A";
636   case cArg: return "c";
637   case sArg: return "s";
638   case pArg: return "p";
639   case PArg:
640     return "P";
641   case nArg: return "n";
642   case PercentArg:  return "%";
643   case ScanListArg: return "[";
644   case InvalidSpecifier: return nullptr;
645 
646   // POSIX unicode extensions.
647   case CArg: return "C";
648   case SArg: return "S";
649 
650   // Objective-C specific specifiers.
651   case ObjCObjArg: return "@";
652 
653   // FreeBSD kernel specific specifiers.
654   case FreeBSDbArg: return "b";
655   case FreeBSDDArg: return "D";
656   case FreeBSDrArg: return "r";
657   case FreeBSDyArg: return "y";
658 
659   // GlibC specific specifiers.
660   case PrintErrno: return "m";
661 
662   // MS specific specifiers.
663   case ZArg: return "Z";
664   }
665   return nullptr;
666 }
667 
668 Optional<ConversionSpecifier>
669 ConversionSpecifier::getStandardSpecifier() const {
670   ConversionSpecifier::Kind NewKind;
671 
672   switch (getKind()) {
673   default:
674     return None;
675   case DArg:
676     NewKind = dArg;
677     break;
678   case UArg:
679     NewKind = uArg;
680     break;
681   case OArg:
682     NewKind = oArg;
683     break;
684   }
685 
686   ConversionSpecifier FixedCS(*this);
687   FixedCS.setKind(NewKind);
688   return FixedCS;
689 }
690 
691 //===----------------------------------------------------------------------===//
692 // Methods on OptionalAmount.
693 //===----------------------------------------------------------------------===//
694 
695 void OptionalAmount::toString(raw_ostream &os) const {
696   switch (hs) {
697   case Invalid:
698   case NotSpecified:
699     return;
700   case Arg:
701     if (UsesDotPrefix)
702         os << ".";
703     if (usesPositionalArg())
704       os << "*" << getPositionalArgIndex() << "$";
705     else
706       os << "*";
707     break;
708   case Constant:
709     if (UsesDotPrefix)
710         os << ".";
711     os << amt;
712     break;
713   }
714 }
715 
716 bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target,
717                                              const LangOptions &LO) const {
718   switch (LM.getKind()) {
719     case LengthModifier::None:
720       return true;
721 
722     // Handle most integer flags
723     case LengthModifier::AsShort:
724       // Length modifier only applies to FP vectors.
725       if (LO.OpenCL && CS.isDoubleArg())
726         return !VectorNumElts.isInvalid();
727 
728       if (Target.getTriple().isOSMSVCRT()) {
729         switch (CS.getKind()) {
730           case ConversionSpecifier::cArg:
731           case ConversionSpecifier::CArg:
732           case ConversionSpecifier::sArg:
733           case ConversionSpecifier::SArg:
734           case ConversionSpecifier::ZArg:
735             return true;
736           default:
737             break;
738         }
739       }
740       LLVM_FALLTHROUGH;
741     case LengthModifier::AsChar:
742     case LengthModifier::AsLongLong:
743     case LengthModifier::AsQuad:
744     case LengthModifier::AsIntMax:
745     case LengthModifier::AsSizeT:
746     case LengthModifier::AsPtrDiff:
747       switch (CS.getKind()) {
748         case ConversionSpecifier::dArg:
749         case ConversionSpecifier::DArg:
750         case ConversionSpecifier::iArg:
751         case ConversionSpecifier::oArg:
752         case ConversionSpecifier::OArg:
753         case ConversionSpecifier::uArg:
754         case ConversionSpecifier::UArg:
755         case ConversionSpecifier::xArg:
756         case ConversionSpecifier::XArg:
757         case ConversionSpecifier::nArg:
758           return true;
759         case ConversionSpecifier::FreeBSDrArg:
760         case ConversionSpecifier::FreeBSDyArg:
761           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS4();
762         default:
763           return false;
764       }
765 
766     case LengthModifier::AsShortLong:
767       return LO.OpenCL && !VectorNumElts.isInvalid();
768 
769     // Handle 'l' flag
770     case LengthModifier::AsLong: // or AsWideChar
771       if (CS.isDoubleArg()) {
772         // Invalid for OpenCL FP scalars.
773         if (LO.OpenCL && VectorNumElts.isInvalid())
774           return false;
775         return true;
776       }
777 
778       switch (CS.getKind()) {
779         case ConversionSpecifier::dArg:
780         case ConversionSpecifier::DArg:
781         case ConversionSpecifier::iArg:
782         case ConversionSpecifier::oArg:
783         case ConversionSpecifier::OArg:
784         case ConversionSpecifier::uArg:
785         case ConversionSpecifier::UArg:
786         case ConversionSpecifier::xArg:
787         case ConversionSpecifier::XArg:
788         case ConversionSpecifier::nArg:
789         case ConversionSpecifier::cArg:
790         case ConversionSpecifier::sArg:
791         case ConversionSpecifier::ScanListArg:
792         case ConversionSpecifier::ZArg:
793           return true;
794         case ConversionSpecifier::FreeBSDrArg:
795         case ConversionSpecifier::FreeBSDyArg:
796           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS4();
797         default:
798           return false;
799       }
800 
801     case LengthModifier::AsLongDouble:
802       switch (CS.getKind()) {
803         case ConversionSpecifier::aArg:
804         case ConversionSpecifier::AArg:
805         case ConversionSpecifier::fArg:
806         case ConversionSpecifier::FArg:
807         case ConversionSpecifier::eArg:
808         case ConversionSpecifier::EArg:
809         case ConversionSpecifier::gArg:
810         case ConversionSpecifier::GArg:
811           return true;
812         // GNU libc extension.
813         case ConversionSpecifier::dArg:
814         case ConversionSpecifier::iArg:
815         case ConversionSpecifier::oArg:
816         case ConversionSpecifier::uArg:
817         case ConversionSpecifier::xArg:
818         case ConversionSpecifier::XArg:
819           return !Target.getTriple().isOSDarwin() &&
820                  !Target.getTriple().isOSWindows();
821         default:
822           return false;
823       }
824 
825     case LengthModifier::AsAllocate:
826       switch (CS.getKind()) {
827         case ConversionSpecifier::sArg:
828         case ConversionSpecifier::SArg:
829         case ConversionSpecifier::ScanListArg:
830           return true;
831         default:
832           return false;
833       }
834 
835     case LengthModifier::AsMAllocate:
836       switch (CS.getKind()) {
837         case ConversionSpecifier::cArg:
838         case ConversionSpecifier::CArg:
839         case ConversionSpecifier::sArg:
840         case ConversionSpecifier::SArg:
841         case ConversionSpecifier::ScanListArg:
842           return true;
843         default:
844           return false;
845       }
846     case LengthModifier::AsInt32:
847     case LengthModifier::AsInt3264:
848     case LengthModifier::AsInt64:
849       switch (CS.getKind()) {
850         case ConversionSpecifier::dArg:
851         case ConversionSpecifier::iArg:
852         case ConversionSpecifier::oArg:
853         case ConversionSpecifier::uArg:
854         case ConversionSpecifier::xArg:
855         case ConversionSpecifier::XArg:
856           return Target.getTriple().isOSMSVCRT();
857         default:
858           return false;
859       }
860     case LengthModifier::AsWide:
861       switch (CS.getKind()) {
862         case ConversionSpecifier::cArg:
863         case ConversionSpecifier::CArg:
864         case ConversionSpecifier::sArg:
865         case ConversionSpecifier::SArg:
866         case ConversionSpecifier::ZArg:
867           return Target.getTriple().isOSMSVCRT();
868         default:
869           return false;
870       }
871   }
872   llvm_unreachable("Invalid LengthModifier Kind!");
873 }
874 
875 bool FormatSpecifier::hasStandardLengthModifier() const {
876   switch (LM.getKind()) {
877     case LengthModifier::None:
878     case LengthModifier::AsChar:
879     case LengthModifier::AsShort:
880     case LengthModifier::AsLong:
881     case LengthModifier::AsLongLong:
882     case LengthModifier::AsIntMax:
883     case LengthModifier::AsSizeT:
884     case LengthModifier::AsPtrDiff:
885     case LengthModifier::AsLongDouble:
886       return true;
887     case LengthModifier::AsAllocate:
888     case LengthModifier::AsMAllocate:
889     case LengthModifier::AsQuad:
890     case LengthModifier::AsInt32:
891     case LengthModifier::AsInt3264:
892     case LengthModifier::AsInt64:
893     case LengthModifier::AsWide:
894     case LengthModifier::AsShortLong: // ???
895       return false;
896   }
897   llvm_unreachable("Invalid LengthModifier Kind!");
898 }
899 
900 bool FormatSpecifier::hasStandardConversionSpecifier(
901     const LangOptions &LangOpt) const {
902   switch (CS.getKind()) {
903     case ConversionSpecifier::cArg:
904     case ConversionSpecifier::dArg:
905     case ConversionSpecifier::iArg:
906     case ConversionSpecifier::oArg:
907     case ConversionSpecifier::uArg:
908     case ConversionSpecifier::xArg:
909     case ConversionSpecifier::XArg:
910     case ConversionSpecifier::fArg:
911     case ConversionSpecifier::FArg:
912     case ConversionSpecifier::eArg:
913     case ConversionSpecifier::EArg:
914     case ConversionSpecifier::gArg:
915     case ConversionSpecifier::GArg:
916     case ConversionSpecifier::aArg:
917     case ConversionSpecifier::AArg:
918     case ConversionSpecifier::sArg:
919     case ConversionSpecifier::pArg:
920     case ConversionSpecifier::nArg:
921     case ConversionSpecifier::ObjCObjArg:
922     case ConversionSpecifier::ScanListArg:
923     case ConversionSpecifier::PercentArg:
924     case ConversionSpecifier::PArg:
925       return true;
926     case ConversionSpecifier::CArg:
927     case ConversionSpecifier::SArg:
928       return LangOpt.ObjC;
929     case ConversionSpecifier::InvalidSpecifier:
930     case ConversionSpecifier::FreeBSDbArg:
931     case ConversionSpecifier::FreeBSDDArg:
932     case ConversionSpecifier::FreeBSDrArg:
933     case ConversionSpecifier::FreeBSDyArg:
934     case ConversionSpecifier::PrintErrno:
935     case ConversionSpecifier::DArg:
936     case ConversionSpecifier::OArg:
937     case ConversionSpecifier::UArg:
938     case ConversionSpecifier::ZArg:
939       return false;
940   }
941   llvm_unreachable("Invalid ConversionSpecifier Kind!");
942 }
943 
944 bool FormatSpecifier::hasStandardLengthConversionCombination() const {
945   if (LM.getKind() == LengthModifier::AsLongDouble) {
946     switch(CS.getKind()) {
947         case ConversionSpecifier::dArg:
948         case ConversionSpecifier::iArg:
949         case ConversionSpecifier::oArg:
950         case ConversionSpecifier::uArg:
951         case ConversionSpecifier::xArg:
952         case ConversionSpecifier::XArg:
953           return false;
954         default:
955           return true;
956     }
957   }
958   return true;
959 }
960 
961 Optional<LengthModifier> FormatSpecifier::getCorrectedLengthModifier() const {
962   if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) {
963     if (LM.getKind() == LengthModifier::AsLongDouble ||
964         LM.getKind() == LengthModifier::AsQuad) {
965       LengthModifier FixedLM(LM);
966       FixedLM.setKind(LengthModifier::AsLongLong);
967       return FixedLM;
968     }
969   }
970 
971   return None;
972 }
973 
974 bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
975                                                 LengthModifier &LM) {
976   assert(isa<TypedefType>(QT) && "Expected a TypedefType");
977   const TypedefNameDecl *Typedef = cast<TypedefType>(QT)->getDecl();
978 
979   for (;;) {
980     const IdentifierInfo *Identifier = Typedef->getIdentifier();
981     if (Identifier->getName() == "size_t") {
982       LM.setKind(LengthModifier::AsSizeT);
983       return true;
984     } else if (Identifier->getName() == "ssize_t") {
985       // Not C99, but common in Unix.
986       LM.setKind(LengthModifier::AsSizeT);
987       return true;
988     } else if (Identifier->getName() == "intmax_t") {
989       LM.setKind(LengthModifier::AsIntMax);
990       return true;
991     } else if (Identifier->getName() == "uintmax_t") {
992       LM.setKind(LengthModifier::AsIntMax);
993       return true;
994     } else if (Identifier->getName() == "ptrdiff_t") {
995       LM.setKind(LengthModifier::AsPtrDiff);
996       return true;
997     }
998 
999     QualType T = Typedef->getUnderlyingType();
1000     if (!isa<TypedefType>(T))
1001       break;
1002 
1003     Typedef = cast<TypedefType>(T)->getDecl();
1004   }
1005   return false;
1006 }
1007