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::ConversionSpecifier;
25 using namespace clang;
26 
27 // Key function to FormatStringHandler.
~FormatStringHandler()28 FormatStringHandler::~FormatStringHandler() {}
29 
30 //===----------------------------------------------------------------------===//
31 // Functions for parsing format strings components in both printf and
32 // scanf format strings.
33 //===----------------------------------------------------------------------===//
34 
35 OptionalAmount
ParseAmount(const char * & Beg,const char * E)36 clang::analyze_format_string::ParseAmount(const char *&Beg, const char *E) {
37   const char *I = Beg;
38   UpdateOnReturn <const char*> UpdateBeg(Beg, I);
39 
40   unsigned accumulator = 0;
41   bool hasDigits = false;
42 
43   for ( ; I != E; ++I) {
44     char c = *I;
45     if (c >= '0' && c <= '9') {
46       hasDigits = true;
47       accumulator = (accumulator * 10) + (c - '0');
48       continue;
49     }
50 
51     if (hasDigits)
52       return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg,
53           false);
54 
55     break;
56   }
57 
58   return OptionalAmount();
59 }
60 
61 OptionalAmount
ParseNonPositionAmount(const char * & Beg,const char * E,unsigned & argIndex)62 clang::analyze_format_string::ParseNonPositionAmount(const char *&Beg,
63                                                      const char *E,
64                                                      unsigned &argIndex) {
65   if (*Beg == '*') {
66     ++Beg;
67     return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
68   }
69 
70   return ParseAmount(Beg, E);
71 }
72 
73 OptionalAmount
ParsePositionAmount(FormatStringHandler & H,const char * Start,const char * & Beg,const char * E,PositionContext p)74 clang::analyze_format_string::ParsePositionAmount(FormatStringHandler &H,
75                                                   const char *Start,
76                                                   const char *&Beg,
77                                                   const char *E,
78                                                   PositionContext p) {
79   if (*Beg == '*') {
80     const char *I = Beg + 1;
81     const OptionalAmount &Amt = ParseAmount(I, E);
82 
83     if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
84       H.HandleInvalidPosition(Beg, I - Beg, p);
85       return OptionalAmount(false);
86     }
87 
88     if (I == E) {
89       // No more characters left?
90       H.HandleIncompleteSpecifier(Start, E - Start);
91       return OptionalAmount(false);
92     }
93 
94     assert(Amt.getHowSpecified() == OptionalAmount::Constant);
95 
96     if (*I == '$') {
97       // Handle positional arguments
98 
99       // Special case: '*0$', since this is an easy mistake.
100       if (Amt.getConstantAmount() == 0) {
101         H.HandleZeroPosition(Beg, I - Beg + 1);
102         return OptionalAmount(false);
103       }
104 
105       const char *Tmp = Beg;
106       Beg = ++I;
107 
108       return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1,
109                             Tmp, 0, true);
110     }
111 
112     H.HandleInvalidPosition(Beg, I - Beg, p);
113     return OptionalAmount(false);
114   }
115 
116   return ParseAmount(Beg, E);
117 }
118 
119 
120 bool
ParseFieldWidth(FormatStringHandler & H,FormatSpecifier & CS,const char * Start,const char * & Beg,const char * E,unsigned * argIndex)121 clang::analyze_format_string::ParseFieldWidth(FormatStringHandler &H,
122                                               FormatSpecifier &CS,
123                                               const char *Start,
124                                               const char *&Beg, const char *E,
125                                               unsigned *argIndex) {
126   // FIXME: Support negative field widths.
127   if (argIndex) {
128     CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
129   }
130   else {
131     const OptionalAmount Amt =
132       ParsePositionAmount(H, Start, Beg, E,
133                           analyze_format_string::FieldWidthPos);
134 
135     if (Amt.isInvalid())
136       return true;
137     CS.setFieldWidth(Amt);
138   }
139   return false;
140 }
141 
142 bool
ParseArgPosition(FormatStringHandler & H,FormatSpecifier & FS,const char * Start,const char * & Beg,const char * E)143 clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
144                                                FormatSpecifier &FS,
145                                                const char *Start,
146                                                const char *&Beg,
147                                                const char *E) {
148   const char *I = Beg;
149 
150   const OptionalAmount &Amt = ParseAmount(I, E);
151 
152   if (I == E) {
153     // No more characters left?
154     H.HandleIncompleteSpecifier(Start, E - Start);
155     return true;
156   }
157 
158   if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
159     // Warn that positional arguments are non-standard.
160     H.HandlePosition(Start, I - Start);
161 
162     // Special case: '%0$', since this is an easy mistake.
163     if (Amt.getConstantAmount() == 0) {
164       H.HandleZeroPosition(Start, I - Start);
165       return true;
166     }
167 
168     FS.setArgIndex(Amt.getConstantAmount() - 1);
169     FS.setUsesPositionalArg();
170     // Update the caller's pointer if we decided to consume
171     // these characters.
172     Beg = I;
173     return false;
174   }
175 
176   return false;
177 }
178 
179 bool
ParseVectorModifier(FormatStringHandler & H,FormatSpecifier & FS,const char * & I,const char * E,const LangOptions & LO)180 clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
181                                                   FormatSpecifier &FS,
182                                                   const char *&I,
183                                                   const char *E,
184                                                   const LangOptions &LO) {
185   if (!LO.OpenCL)
186     return false;
187 
188   const char *Start = I;
189   if (*I == 'v') {
190     ++I;
191 
192     if (I == E) {
193       H.HandleIncompleteSpecifier(Start, E - Start);
194       return true;
195     }
196 
197     OptionalAmount NumElts = ParseAmount(I, E);
198     if (NumElts.getHowSpecified() != OptionalAmount::Constant) {
199       H.HandleIncompleteSpecifier(Start, E - Start);
200       return true;
201     }
202 
203     FS.setVectorNumElts(NumElts);
204   }
205 
206   return false;
207 }
208 
209 bool
ParseLengthModifier(FormatSpecifier & FS,const char * & I,const char * E,const LangOptions & LO,bool IsScanf)210 clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
211                                                   const char *&I,
212                                                   const char *E,
213                                                   const LangOptions &LO,
214                                                   bool IsScanf) {
215   LengthModifier::Kind lmKind = LengthModifier::None;
216   const char *lmPosition = I;
217   switch (*I) {
218     default:
219       return false;
220     case 'h':
221       ++I;
222       if (I != E && *I == 'h') {
223         ++I;
224         lmKind = LengthModifier::AsChar;
225       } else if (I != E && *I == 'l' && LO.OpenCL) {
226         ++I;
227         lmKind = LengthModifier::AsShortLong;
228       } else {
229         lmKind = LengthModifier::AsShort;
230       }
231       break;
232     case 'l':
233       ++I;
234       if (I != E && *I == 'l') {
235         ++I;
236         lmKind = LengthModifier::AsLongLong;
237       } else {
238         lmKind = LengthModifier::AsLong;
239       }
240       break;
241     case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
242     case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
243     case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
244     case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
245     case 'q': lmKind = LengthModifier::AsQuad;       ++I; break;
246     case 'a':
247       if (IsScanf && !LO.C99 && !LO.CPlusPlus11) {
248         // For scanf in C90, look at the next character to see if this should
249         // be parsed as the GNU extension 'a' length modifier. If not, this
250         // will be parsed as a conversion specifier.
251         ++I;
252         if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
253           lmKind = LengthModifier::AsAllocate;
254           break;
255         }
256         --I;
257       }
258       return false;
259     case 'm':
260       if (IsScanf) {
261         lmKind = LengthModifier::AsMAllocate;
262         ++I;
263         break;
264       }
265       return false;
266     // printf: AsInt64, AsInt32, AsInt3264
267     // scanf:  AsInt64
268     case 'I':
269       if (I + 1 != E && I + 2 != E) {
270         if (I[1] == '6' && I[2] == '4') {
271           I += 3;
272           lmKind = LengthModifier::AsInt64;
273           break;
274         }
275         if (IsScanf)
276           return false;
277 
278         if (I[1] == '3' && I[2] == '2') {
279           I += 3;
280           lmKind = LengthModifier::AsInt32;
281           break;
282         }
283       }
284       ++I;
285       lmKind = LengthModifier::AsInt3264;
286       break;
287     case 'w':
288       lmKind = LengthModifier::AsWide; ++I; break;
289   }
290   LengthModifier lm(lmPosition, lmKind);
291   FS.setLengthModifier(lm);
292   return true;
293 }
294 
ParseUTF8InvalidSpecifier(const char * SpecifierBegin,const char * FmtStrEnd,unsigned & Len)295 bool clang::analyze_format_string::ParseUTF8InvalidSpecifier(
296     const char *SpecifierBegin, const char *FmtStrEnd, unsigned &Len) {
297   if (SpecifierBegin + 1 >= FmtStrEnd)
298     return false;
299 
300   const llvm::UTF8 *SB =
301       reinterpret_cast<const llvm::UTF8 *>(SpecifierBegin + 1);
302   const llvm::UTF8 *SE = reinterpret_cast<const llvm::UTF8 *>(FmtStrEnd);
303   const char FirstByte = *SB;
304 
305   // If the invalid specifier is a multibyte UTF-8 string, return the
306   // total length accordingly so that the conversion specifier can be
307   // properly updated to reflect a complete UTF-8 specifier.
308   unsigned NumBytes = llvm::getNumBytesForUTF8(FirstByte);
309   if (NumBytes == 1)
310     return false;
311   if (SB + NumBytes > SE)
312     return false;
313 
314   Len = NumBytes + 1;
315   return true;
316 }
317 
318 //===----------------------------------------------------------------------===//
319 // Methods on ArgType.
320 //===----------------------------------------------------------------------===//
321 
322 clang::analyze_format_string::ArgType::MatchKind
matchesType(ASTContext & C,QualType argTy) const323 ArgType::matchesType(ASTContext &C, QualType argTy) const {
324   // When using the format attribute in C++, you can receive a function or an
325   // array that will necessarily decay to a pointer when passed to the final
326   // format consumer. Apply decay before type comparison.
327   if (argTy->canDecayToPointerType())
328     argTy = C.getDecayedType(argTy);
329 
330   if (Ptr) {
331     // It has to be a pointer.
332     const PointerType *PT = argTy->getAs<PointerType>();
333     if (!PT)
334       return NoMatch;
335 
336     // We cannot write through a const qualified pointer.
337     if (PT->getPointeeType().isConstQualified())
338       return NoMatch;
339 
340     argTy = PT->getPointeeType();
341   }
342 
343   switch (K) {
344     case InvalidTy:
345       llvm_unreachable("ArgType must be valid");
346 
347     case UnknownTy:
348       return Match;
349 
350     case AnyCharTy: {
351       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
352         // If the enum is incomplete we know nothing about the underlying type.
353         // Assume that it's 'int'.
354         if (!ETy->getDecl()->isComplete())
355           return NoMatch;
356         argTy = ETy->getDecl()->getIntegerType();
357       }
358 
359       if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
360         switch (BT->getKind()) {
361           default:
362             break;
363           case BuiltinType::Char_S:
364           case BuiltinType::SChar:
365           case BuiltinType::UChar:
366           case BuiltinType::Char_U:
367           case BuiltinType::Bool:
368             return Match;
369         }
370       return NoMatch;
371     }
372 
373     case SpecificTy: {
374       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
375         // If the enum is incomplete we know nothing about the underlying type.
376         // Assume that it's 'int'.
377         if (!ETy->getDecl()->isComplete())
378           argTy = C.IntTy;
379         else
380           argTy = ETy->getDecl()->getIntegerType();
381       }
382       argTy = C.getCanonicalType(argTy).getUnqualifiedType();
383 
384       if (T == argTy)
385         return Match;
386       // Check for "compatible types".
387       if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
388         switch (BT->getKind()) {
389           default:
390             break;
391           case BuiltinType::Char_S:
392           case BuiltinType::SChar:
393           case BuiltinType::Char_U:
394           case BuiltinType::UChar:
395           case BuiltinType::Bool:
396             if (T == C.UnsignedShortTy || T == C.ShortTy)
397               return NoMatchTypeConfusion;
398             return T == C.UnsignedCharTy || T == C.SignedCharTy ? Match
399                                                                 : NoMatch;
400           case BuiltinType::Short:
401             return T == C.UnsignedShortTy ? Match : NoMatch;
402           case BuiltinType::UShort:
403             return T == C.ShortTy ? Match : NoMatch;
404           case BuiltinType::Int:
405             return T == C.UnsignedIntTy ? Match : NoMatch;
406           case BuiltinType::UInt:
407             return T == C.IntTy ? Match : NoMatch;
408           case BuiltinType::Long:
409             return T == C.UnsignedLongTy ? Match : NoMatch;
410           case BuiltinType::ULong:
411             return T == C.LongTy ? Match : NoMatch;
412           case BuiltinType::LongLong:
413             return T == C.UnsignedLongLongTy ? Match : NoMatch;
414           case BuiltinType::ULongLong:
415             return T == C.LongLongTy ? Match : NoMatch;
416         }
417       return NoMatch;
418     }
419 
420     case CStrTy: {
421       const PointerType *PT = argTy->getAs<PointerType>();
422       if (!PT)
423         return NoMatch;
424       QualType pointeeTy = PT->getPointeeType();
425       if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
426         switch (BT->getKind()) {
427           case BuiltinType::Char_U:
428           case BuiltinType::UChar:
429           case BuiltinType::Char_S:
430           case BuiltinType::SChar:
431             return Match;
432           default:
433             break;
434         }
435 
436       return NoMatch;
437     }
438 
439     case WCStrTy: {
440       const PointerType *PT = argTy->getAs<PointerType>();
441       if (!PT)
442         return NoMatch;
443       QualType pointeeTy =
444         C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
445       return pointeeTy == C.getWideCharType() ? Match : NoMatch;
446     }
447 
448     case WIntTy: {
449       QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
450 
451       if (C.getCanonicalType(argTy).getUnqualifiedType() == WInt)
452         return Match;
453 
454       QualType PromoArg = argTy->isPromotableIntegerType()
455                               ? C.getPromotedIntegerType(argTy)
456                               : argTy;
457       PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
458 
459       // If the promoted argument is the corresponding signed type of the
460       // wint_t type, then it should match.
461       if (PromoArg->hasSignedIntegerRepresentation() &&
462           C.getCorrespondingUnsignedType(PromoArg) == WInt)
463         return Match;
464 
465       return WInt == PromoArg ? Match : NoMatch;
466     }
467 
468     case CPointerTy:
469       if (argTy->isVoidPointerType()) {
470         return Match;
471       } if (argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
472             argTy->isBlockPointerType() || argTy->isNullPtrType()) {
473         return NoMatchPedantic;
474       } else {
475         return NoMatch;
476       }
477 
478     case ObjCPointerTy: {
479       if (argTy->getAs<ObjCObjectPointerType>() ||
480           argTy->getAs<BlockPointerType>())
481         return Match;
482 
483       // Handle implicit toll-free bridging.
484       if (const PointerType *PT = argTy->getAs<PointerType>()) {
485         // Things such as CFTypeRef are really just opaque pointers
486         // to C structs representing CF types that can often be bridged
487         // to Objective-C objects.  Since the compiler doesn't know which
488         // structs can be toll-free bridged, we just accept them all.
489         QualType pointee = PT->getPointeeType();
490         if (pointee->getAsStructureType() || pointee->isVoidType())
491           return Match;
492       }
493       return NoMatch;
494     }
495   }
496 
497   llvm_unreachable("Invalid ArgType Kind!");
498 }
499 
makeVectorType(ASTContext & C,unsigned NumElts) const500 ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const {
501   // Check for valid vector element types.
502   if (T.isNull())
503     return ArgType::Invalid();
504 
505   QualType Vec = C.getExtVectorType(T, NumElts);
506   return ArgType(Vec, Name);
507 }
508 
getRepresentativeType(ASTContext & C) const509 QualType ArgType::getRepresentativeType(ASTContext &C) const {
510   QualType Res;
511   switch (K) {
512     case InvalidTy:
513       llvm_unreachable("No representative type for Invalid ArgType");
514     case UnknownTy:
515       llvm_unreachable("No representative type for Unknown ArgType");
516     case AnyCharTy:
517       Res = C.CharTy;
518       break;
519     case SpecificTy:
520       Res = T;
521       break;
522     case CStrTy:
523       Res = C.getPointerType(C.CharTy);
524       break;
525     case WCStrTy:
526       Res = C.getPointerType(C.getWideCharType());
527       break;
528     case ObjCPointerTy:
529       Res = C.ObjCBuiltinIdTy;
530       break;
531     case CPointerTy:
532       Res = C.VoidPtrTy;
533       break;
534     case WIntTy: {
535       Res = C.getWIntType();
536       break;
537     }
538   }
539 
540   if (Ptr)
541     Res = C.getPointerType(Res);
542   return Res;
543 }
544 
getRepresentativeTypeName(ASTContext & C) const545 std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
546   std::string S = getRepresentativeType(C).getAsString(C.getPrintingPolicy());
547 
548   std::string Alias;
549   if (Name) {
550     // Use a specific name for this type, e.g. "size_t".
551     Alias = Name;
552     if (Ptr) {
553       // If ArgType is actually a pointer to T, append an asterisk.
554       Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
555     }
556     // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
557     if (S == Alias)
558       Alias.clear();
559   }
560 
561   if (!Alias.empty())
562     return std::string("'") + Alias + "' (aka '" + S + "')";
563   return std::string("'") + S + "'";
564 }
565 
566 
567 //===----------------------------------------------------------------------===//
568 // Methods on OptionalAmount.
569 //===----------------------------------------------------------------------===//
570 
571 ArgType
getArgType(ASTContext & Ctx) const572 analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
573   return Ctx.IntTy;
574 }
575 
576 //===----------------------------------------------------------------------===//
577 // Methods on LengthModifier.
578 //===----------------------------------------------------------------------===//
579 
580 const char *
toString() const581 analyze_format_string::LengthModifier::toString() const {
582   switch (kind) {
583   case AsChar:
584     return "hh";
585   case AsShort:
586     return "h";
587   case AsShortLong:
588     return "hl";
589   case AsLong: // or AsWideChar
590     return "l";
591   case AsLongLong:
592     return "ll";
593   case AsQuad:
594     return "q";
595   case AsIntMax:
596     return "j";
597   case AsSizeT:
598     return "z";
599   case AsPtrDiff:
600     return "t";
601   case AsInt32:
602     return "I32";
603   case AsInt3264:
604     return "I";
605   case AsInt64:
606     return "I64";
607   case AsLongDouble:
608     return "L";
609   case AsAllocate:
610     return "a";
611   case AsMAllocate:
612     return "m";
613   case AsWide:
614     return "w";
615   case None:
616     return "";
617   }
618   return nullptr;
619 }
620 
621 //===----------------------------------------------------------------------===//
622 // Methods on ConversionSpecifier.
623 //===----------------------------------------------------------------------===//
624 
toString() const625 const char *ConversionSpecifier::toString() const {
626   switch (kind) {
627   case dArg: return "d";
628   case DArg: return "D";
629   case iArg: return "i";
630   case oArg: return "o";
631   case OArg: return "O";
632   case uArg: return "u";
633   case UArg: return "U";
634   case xArg: return "x";
635   case XArg: return "X";
636   case fArg: return "f";
637   case FArg: return "F";
638   case eArg: return "e";
639   case EArg: return "E";
640   case gArg: return "g";
641   case GArg: return "G";
642   case aArg: return "a";
643   case AArg: return "A";
644   case cArg: return "c";
645   case sArg: return "s";
646   case pArg: return "p";
647   case PArg:
648     return "P";
649   case nArg: return "n";
650   case PercentArg:  return "%";
651   case ScanListArg: return "[";
652   case InvalidSpecifier: return nullptr;
653 
654   // POSIX unicode extensions.
655   case CArg: return "C";
656   case SArg: return "S";
657 
658   // Objective-C specific specifiers.
659   case ObjCObjArg: return "@";
660 
661   // FreeBSD kernel specific specifiers.
662   case FreeBSDbArg: return "b";
663   case FreeBSDDArg: return "D";
664   case FreeBSDrArg: return "r";
665   case FreeBSDyArg: return "y";
666 
667   // GlibC specific specifiers.
668   case PrintErrno: return "m";
669 
670   // MS specific specifiers.
671   case ZArg: return "Z";
672   }
673   return nullptr;
674 }
675 
676 Optional<ConversionSpecifier>
getStandardSpecifier() const677 ConversionSpecifier::getStandardSpecifier() const {
678   ConversionSpecifier::Kind NewKind;
679 
680   switch (getKind()) {
681   default:
682     return None;
683   case DArg:
684     NewKind = dArg;
685     break;
686   case UArg:
687     NewKind = uArg;
688     break;
689   case OArg:
690     NewKind = oArg;
691     break;
692   }
693 
694   ConversionSpecifier FixedCS(*this);
695   FixedCS.setKind(NewKind);
696   return FixedCS;
697 }
698 
699 //===----------------------------------------------------------------------===//
700 // Methods on OptionalAmount.
701 //===----------------------------------------------------------------------===//
702 
toString(raw_ostream & os) const703 void OptionalAmount::toString(raw_ostream &os) const {
704   switch (hs) {
705   case Invalid:
706   case NotSpecified:
707     return;
708   case Arg:
709     if (UsesDotPrefix)
710         os << ".";
711     if (usesPositionalArg())
712       os << "*" << getPositionalArgIndex() << "$";
713     else
714       os << "*";
715     break;
716   case Constant:
717     if (UsesDotPrefix)
718         os << ".";
719     os << amt;
720     break;
721   }
722 }
723 
hasValidLengthModifier(const TargetInfo & Target,const LangOptions & LO) const724 bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target,
725                                              const LangOptions &LO) const {
726   switch (LM.getKind()) {
727     case LengthModifier::None:
728       return true;
729 
730     // Handle most integer flags
731     case LengthModifier::AsShort:
732       // Length modifier only applies to FP vectors.
733       if (LO.OpenCL && CS.isDoubleArg())
734         return !VectorNumElts.isInvalid();
735 
736       if (Target.getTriple().isOSMSVCRT()) {
737         switch (CS.getKind()) {
738           case ConversionSpecifier::cArg:
739           case ConversionSpecifier::CArg:
740           case ConversionSpecifier::sArg:
741           case ConversionSpecifier::SArg:
742           case ConversionSpecifier::ZArg:
743             return true;
744           default:
745             break;
746         }
747       }
748       LLVM_FALLTHROUGH;
749     case LengthModifier::AsChar:
750     case LengthModifier::AsLongLong:
751     case LengthModifier::AsQuad:
752     case LengthModifier::AsIntMax:
753     case LengthModifier::AsSizeT:
754     case LengthModifier::AsPtrDiff:
755       switch (CS.getKind()) {
756         case ConversionSpecifier::dArg:
757         case ConversionSpecifier::DArg:
758         case ConversionSpecifier::iArg:
759         case ConversionSpecifier::oArg:
760         case ConversionSpecifier::OArg:
761         case ConversionSpecifier::uArg:
762         case ConversionSpecifier::UArg:
763         case ConversionSpecifier::xArg:
764         case ConversionSpecifier::XArg:
765         case ConversionSpecifier::nArg:
766           return true;
767         case ConversionSpecifier::FreeBSDrArg:
768         case ConversionSpecifier::FreeBSDyArg:
769           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
770         default:
771           return false;
772       }
773 
774     case LengthModifier::AsShortLong:
775       return LO.OpenCL && !VectorNumElts.isInvalid();
776 
777     // Handle 'l' flag
778     case LengthModifier::AsLong: // or AsWideChar
779       if (CS.isDoubleArg()) {
780         // Invalid for OpenCL FP scalars.
781         if (LO.OpenCL && VectorNumElts.isInvalid())
782           return false;
783         return true;
784       }
785 
786       switch (CS.getKind()) {
787         case ConversionSpecifier::dArg:
788         case ConversionSpecifier::DArg:
789         case ConversionSpecifier::iArg:
790         case ConversionSpecifier::oArg:
791         case ConversionSpecifier::OArg:
792         case ConversionSpecifier::uArg:
793         case ConversionSpecifier::UArg:
794         case ConversionSpecifier::xArg:
795         case ConversionSpecifier::XArg:
796         case ConversionSpecifier::nArg:
797         case ConversionSpecifier::cArg:
798         case ConversionSpecifier::sArg:
799         case ConversionSpecifier::ScanListArg:
800         case ConversionSpecifier::ZArg:
801           return true;
802         case ConversionSpecifier::FreeBSDrArg:
803         case ConversionSpecifier::FreeBSDyArg:
804           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
805         default:
806           return false;
807       }
808 
809     case LengthModifier::AsLongDouble:
810       switch (CS.getKind()) {
811         case ConversionSpecifier::aArg:
812         case ConversionSpecifier::AArg:
813         case ConversionSpecifier::fArg:
814         case ConversionSpecifier::FArg:
815         case ConversionSpecifier::eArg:
816         case ConversionSpecifier::EArg:
817         case ConversionSpecifier::gArg:
818         case ConversionSpecifier::GArg:
819           return true;
820         // GNU libc extension.
821         case ConversionSpecifier::dArg:
822         case ConversionSpecifier::iArg:
823         case ConversionSpecifier::oArg:
824         case ConversionSpecifier::uArg:
825         case ConversionSpecifier::xArg:
826         case ConversionSpecifier::XArg:
827           return !Target.getTriple().isOSDarwin() &&
828                  !Target.getTriple().isOSWindows();
829         default:
830           return false;
831       }
832 
833     case LengthModifier::AsAllocate:
834       switch (CS.getKind()) {
835         case ConversionSpecifier::sArg:
836         case ConversionSpecifier::SArg:
837         case ConversionSpecifier::ScanListArg:
838           return true;
839         default:
840           return false;
841       }
842 
843     case LengthModifier::AsMAllocate:
844       switch (CS.getKind()) {
845         case ConversionSpecifier::cArg:
846         case ConversionSpecifier::CArg:
847         case ConversionSpecifier::sArg:
848         case ConversionSpecifier::SArg:
849         case ConversionSpecifier::ScanListArg:
850           return true;
851         default:
852           return false;
853       }
854     case LengthModifier::AsInt32:
855     case LengthModifier::AsInt3264:
856     case LengthModifier::AsInt64:
857       switch (CS.getKind()) {
858         case ConversionSpecifier::dArg:
859         case ConversionSpecifier::iArg:
860         case ConversionSpecifier::oArg:
861         case ConversionSpecifier::uArg:
862         case ConversionSpecifier::xArg:
863         case ConversionSpecifier::XArg:
864           return Target.getTriple().isOSMSVCRT();
865         default:
866           return false;
867       }
868     case LengthModifier::AsWide:
869       switch (CS.getKind()) {
870         case ConversionSpecifier::cArg:
871         case ConversionSpecifier::CArg:
872         case ConversionSpecifier::sArg:
873         case ConversionSpecifier::SArg:
874         case ConversionSpecifier::ZArg:
875           return Target.getTriple().isOSMSVCRT();
876         default:
877           return false;
878       }
879   }
880   llvm_unreachable("Invalid LengthModifier Kind!");
881 }
882 
hasStandardLengthModifier() const883 bool FormatSpecifier::hasStandardLengthModifier() const {
884   switch (LM.getKind()) {
885     case LengthModifier::None:
886     case LengthModifier::AsChar:
887     case LengthModifier::AsShort:
888     case LengthModifier::AsLong:
889     case LengthModifier::AsLongLong:
890     case LengthModifier::AsIntMax:
891     case LengthModifier::AsSizeT:
892     case LengthModifier::AsPtrDiff:
893     case LengthModifier::AsLongDouble:
894       return true;
895     case LengthModifier::AsAllocate:
896     case LengthModifier::AsMAllocate:
897     case LengthModifier::AsQuad:
898     case LengthModifier::AsInt32:
899     case LengthModifier::AsInt3264:
900     case LengthModifier::AsInt64:
901     case LengthModifier::AsWide:
902     case LengthModifier::AsShortLong: // ???
903       return false;
904   }
905   llvm_unreachable("Invalid LengthModifier Kind!");
906 }
907 
hasStandardConversionSpecifier(const LangOptions & LangOpt) const908 bool FormatSpecifier::hasStandardConversionSpecifier(
909     const LangOptions &LangOpt) const {
910   switch (CS.getKind()) {
911     case ConversionSpecifier::cArg:
912     case ConversionSpecifier::dArg:
913     case ConversionSpecifier::iArg:
914     case ConversionSpecifier::oArg:
915     case ConversionSpecifier::uArg:
916     case ConversionSpecifier::xArg:
917     case ConversionSpecifier::XArg:
918     case ConversionSpecifier::fArg:
919     case ConversionSpecifier::FArg:
920     case ConversionSpecifier::eArg:
921     case ConversionSpecifier::EArg:
922     case ConversionSpecifier::gArg:
923     case ConversionSpecifier::GArg:
924     case ConversionSpecifier::aArg:
925     case ConversionSpecifier::AArg:
926     case ConversionSpecifier::sArg:
927     case ConversionSpecifier::pArg:
928     case ConversionSpecifier::nArg:
929     case ConversionSpecifier::ObjCObjArg:
930     case ConversionSpecifier::ScanListArg:
931     case ConversionSpecifier::PercentArg:
932     case ConversionSpecifier::PArg:
933       return true;
934     case ConversionSpecifier::CArg:
935     case ConversionSpecifier::SArg:
936       return LangOpt.ObjC;
937     case ConversionSpecifier::InvalidSpecifier:
938     case ConversionSpecifier::FreeBSDbArg:
939     case ConversionSpecifier::FreeBSDDArg:
940     case ConversionSpecifier::FreeBSDrArg:
941     case ConversionSpecifier::FreeBSDyArg:
942     case ConversionSpecifier::PrintErrno:
943     case ConversionSpecifier::DArg:
944     case ConversionSpecifier::OArg:
945     case ConversionSpecifier::UArg:
946     case ConversionSpecifier::ZArg:
947       return false;
948   }
949   llvm_unreachable("Invalid ConversionSpecifier Kind!");
950 }
951 
hasStandardLengthConversionCombination() const952 bool FormatSpecifier::hasStandardLengthConversionCombination() const {
953   if (LM.getKind() == LengthModifier::AsLongDouble) {
954     switch(CS.getKind()) {
955         case ConversionSpecifier::dArg:
956         case ConversionSpecifier::iArg:
957         case ConversionSpecifier::oArg:
958         case ConversionSpecifier::uArg:
959         case ConversionSpecifier::xArg:
960         case ConversionSpecifier::XArg:
961           return false;
962         default:
963           return true;
964     }
965   }
966   return true;
967 }
968 
getCorrectedLengthModifier() const969 Optional<LengthModifier> FormatSpecifier::getCorrectedLengthModifier() const {
970   if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) {
971     if (LM.getKind() == LengthModifier::AsLongDouble ||
972         LM.getKind() == LengthModifier::AsQuad) {
973       LengthModifier FixedLM(LM);
974       FixedLM.setKind(LengthModifier::AsLongLong);
975       return FixedLM;
976     }
977   }
978 
979   return None;
980 }
981 
namedTypeToLengthModifier(QualType QT,LengthModifier & LM)982 bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
983                                                 LengthModifier &LM) {
984   assert(isa<TypedefType>(QT) && "Expected a TypedefType");
985   const TypedefNameDecl *Typedef = cast<TypedefType>(QT)->getDecl();
986 
987   for (;;) {
988     const IdentifierInfo *Identifier = Typedef->getIdentifier();
989     if (Identifier->getName() == "size_t") {
990       LM.setKind(LengthModifier::AsSizeT);
991       return true;
992     } else if (Identifier->getName() == "ssize_t") {
993       // Not C99, but common in Unix.
994       LM.setKind(LengthModifier::AsSizeT);
995       return true;
996     } else if (Identifier->getName() == "intmax_t") {
997       LM.setKind(LengthModifier::AsIntMax);
998       return true;
999     } else if (Identifier->getName() == "uintmax_t") {
1000       LM.setKind(LengthModifier::AsIntMax);
1001       return true;
1002     } else if (Identifier->getName() == "ptrdiff_t") {
1003       LM.setKind(LengthModifier::AsPtrDiff);
1004       return true;
1005     }
1006 
1007     QualType T = Typedef->getUnderlyingType();
1008     if (!isa<TypedefType>(T))
1009       break;
1010 
1011     Typedef = cast<TypedefType>(T)->getDecl();
1012   }
1013   return false;
1014 }
1015