Home | History | Annotate | Download | only in Analysis
      1 //= ScanfFormatString.cpp - Analysis of printf format strings --*- C++ -*-===//
      2 //
      3 //                     The LLVM Compiler Infrastructure
      4 //
      5 // This file is distributed under the University of Illinois Open Source
      6 // License. See LICENSE.TXT for details.
      7 //
      8 //===----------------------------------------------------------------------===//
      9 //
     10 // Handling of format string in scanf and friends.  The structure of format
     11 // strings for fscanf() are described in C99 7.19.6.2.
     12 //
     13 //===----------------------------------------------------------------------===//
     14 
     15 #include "clang/Analysis/Analyses/FormatString.h"
     16 #include "FormatStringParsing.h"
     17 #include "clang/Basic/TargetInfo.h"
     18 
     19 using clang::analyze_format_string::ArgType;
     20 using clang::analyze_format_string::FormatStringHandler;
     21 using clang::analyze_format_string::LengthModifier;
     22 using clang::analyze_format_string::OptionalAmount;
     23 using clang::analyze_format_string::ConversionSpecifier;
     24 using clang::analyze_scanf::ScanfConversionSpecifier;
     25 using clang::analyze_scanf::ScanfSpecifier;
     26 using clang::UpdateOnReturn;
     27 using namespace clang;
     28 
     29 typedef clang::analyze_format_string::SpecifierResult<ScanfSpecifier>
     30         ScanfSpecifierResult;
     31 
     32 static bool ParseScanList(FormatStringHandler &H,
     33                           ScanfConversionSpecifier &CS,
     34                           const char *&Beg, const char *E) {
     35   const char *I = Beg;
     36   const char *start = I - 1;
     37   UpdateOnReturn <const char*> UpdateBeg(Beg, I);
     38 
     39   // No more characters?
     40   if (I == E) {
     41     H.HandleIncompleteScanList(start, I);
     42     return true;
     43   }
     44 
     45   // Special case: ']' is the first character.
     46   if (*I == ']') {
     47     if (++I == E) {
     48       H.HandleIncompleteScanList(start, I - 1);
     49       return true;
     50     }
     51   }
     52 
     53   // Special case: "^]" are the first characters.
     54   if (I + 1 != E && I[0] == '^' && I[1] == ']') {
     55     I += 2;
     56     if (I == E) {
     57       H.HandleIncompleteScanList(start, I - 1);
     58       return true;
     59     }
     60   }
     61 
     62   // Look for a ']' character which denotes the end of the scan list.
     63   while (*I != ']') {
     64     if (++I == E) {
     65       H.HandleIncompleteScanList(start, I - 1);
     66       return true;
     67     }
     68   }
     69 
     70   CS.setEndScanList(I);
     71   return false;
     72 }
     73 
     74 // FIXME: Much of this is copy-paste from ParsePrintfSpecifier.
     75 // We can possibly refactor.
     76 static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
     77                                                 const char *&Beg,
     78                                                 const char *E,
     79                                                 unsigned &argIndex,
     80                                                 const LangOptions &LO,
     81                                                 const TargetInfo &Target) {
     82 
     83   using namespace clang::analyze_scanf;
     84   const char *I = Beg;
     85   const char *Start = nullptr;
     86   UpdateOnReturn <const char*> UpdateBeg(Beg, I);
     87 
     88     // Look for a '%' character that indicates the start of a format specifier.
     89   for ( ; I != E ; ++I) {
     90     char c = *I;
     91     if (c == '\0') {
     92         // Detect spurious null characters, which are likely errors.
     93       H.HandleNullChar(I);
     94       return true;
     95     }
     96     if (c == '%') {
     97       Start = I++;  // Record the start of the format specifier.
     98       break;
     99     }
    100   }
    101 
    102     // No format specifier found?
    103   if (!Start)
    104     return false;
    105 
    106   if (I == E) {
    107       // No more characters left?
    108     H.HandleIncompleteSpecifier(Start, E - Start);
    109     return true;
    110   }
    111 
    112   ScanfSpecifier FS;
    113   if (ParseArgPosition(H, FS, Start, I, E))
    114     return true;
    115 
    116   if (I == E) {
    117       // No more characters left?
    118     H.HandleIncompleteSpecifier(Start, E - Start);
    119     return true;
    120   }
    121 
    122   // Look for '*' flag if it is present.
    123   if (*I == '*') {
    124     FS.setSuppressAssignment(I);
    125     if (++I == E) {
    126       H.HandleIncompleteSpecifier(Start, E - Start);
    127       return true;
    128     }
    129   }
    130 
    131   // Look for the field width (if any).  Unlike printf, this is either
    132   // a fixed integer or isn't present.
    133   const OptionalAmount &Amt = clang::analyze_format_string::ParseAmount(I, E);
    134   if (Amt.getHowSpecified() != OptionalAmount::NotSpecified) {
    135     assert(Amt.getHowSpecified() == OptionalAmount::Constant);
    136     FS.setFieldWidth(Amt);
    137 
    138     if (I == E) {
    139       // No more characters left?
    140       H.HandleIncompleteSpecifier(Start, E - Start);
    141       return true;
    142     }
    143   }
    144 
    145   // Look for the length modifier.
    146   if (ParseLengthModifier(FS, I, E, LO, /*scanf=*/true) && I == E) {
    147       // No more characters left?
    148     H.HandleIncompleteSpecifier(Start, E - Start);
    149     return true;
    150   }
    151 
    152   // Detect spurious null characters, which are likely errors.
    153   if (*I == '\0') {
    154     H.HandleNullChar(I);
    155     return true;
    156   }
    157 
    158   // Finally, look for the conversion specifier.
    159   const char *conversionPosition = I++;
    160   ScanfConversionSpecifier::Kind k = ScanfConversionSpecifier::InvalidSpecifier;
    161   switch (*conversionPosition) {
    162     default:
    163       break;
    164     case '%': k = ConversionSpecifier::PercentArg;   break;
    165     case 'A': k = ConversionSpecifier::AArg; break;
    166     case 'E': k = ConversionSpecifier::EArg; break;
    167     case 'F': k = ConversionSpecifier::FArg; break;
    168     case 'G': k = ConversionSpecifier::GArg; break;
    169     case 'X': k = ConversionSpecifier::XArg; break;
    170     case 'a': k = ConversionSpecifier::aArg; break;
    171     case 'd': k = ConversionSpecifier::dArg; break;
    172     case 'e': k = ConversionSpecifier::eArg; break;
    173     case 'f': k = ConversionSpecifier::fArg; break;
    174     case 'g': k = ConversionSpecifier::gArg; break;
    175     case 'i': k = ConversionSpecifier::iArg; break;
    176     case 'n': k = ConversionSpecifier::nArg; break;
    177     case 'c': k = ConversionSpecifier::cArg; break;
    178     case 'C': k = ConversionSpecifier::CArg; break;
    179     case 'S': k = ConversionSpecifier::SArg; break;
    180     case '[': k = ConversionSpecifier::ScanListArg; break;
    181     case 'u': k = ConversionSpecifier::uArg; break;
    182     case 'x': k = ConversionSpecifier::xArg; break;
    183     case 'o': k = ConversionSpecifier::oArg; break;
    184     case 's': k = ConversionSpecifier::sArg; break;
    185     case 'p': k = ConversionSpecifier::pArg; break;
    186     // Apple extensions
    187       // Apple-specific
    188     case 'D':
    189       if (Target.getTriple().isOSDarwin())
    190         k = ConversionSpecifier::DArg;
    191       break;
    192     case 'O':
    193       if (Target.getTriple().isOSDarwin())
    194         k = ConversionSpecifier::OArg;
    195       break;
    196     case 'U':
    197       if (Target.getTriple().isOSDarwin())
    198         k = ConversionSpecifier::UArg;
    199       break;
    200   }
    201   ScanfConversionSpecifier CS(conversionPosition, k);
    202   if (k == ScanfConversionSpecifier::ScanListArg) {
    203     if (ParseScanList(H, CS, I, E))
    204       return true;
    205   }
    206   FS.setConversionSpecifier(CS);
    207   if (CS.consumesDataArgument() && !FS.getSuppressAssignment()
    208       && !FS.usesPositionalArg())
    209     FS.setArgIndex(argIndex++);
    210 
    211   // FIXME: '%' and '*' doesn't make sense.  Issue a warning.
    212   // FIXME: 'ConsumedSoFar' and '*' doesn't make sense.
    213 
    214   if (k == ScanfConversionSpecifier::InvalidSpecifier) {
    215     // Assume the conversion takes one argument.
    216     return !H.HandleInvalidScanfConversionSpecifier(FS, Beg, I - Beg);
    217   }
    218   return ScanfSpecifierResult(Start, FS);
    219 }
    220 
    221 ArgType ScanfSpecifier::getArgType(ASTContext &Ctx) const {
    222   const ScanfConversionSpecifier &CS = getConversionSpecifier();
    223 
    224   if (!CS.consumesDataArgument())
    225     return ArgType::Invalid();
    226 
    227   switch(CS.getKind()) {
    228     // Signed int.
    229     case ConversionSpecifier::dArg:
    230     case ConversionSpecifier::DArg:
    231     case ConversionSpecifier::iArg:
    232       switch (LM.getKind()) {
    233         case LengthModifier::None:
    234           return ArgType::PtrTo(Ctx.IntTy);
    235         case LengthModifier::AsChar:
    236           return ArgType::PtrTo(ArgType::AnyCharTy);
    237         case LengthModifier::AsShort:
    238           return ArgType::PtrTo(Ctx.ShortTy);
    239         case LengthModifier::AsLong:
    240           return ArgType::PtrTo(Ctx.LongTy);
    241         case LengthModifier::AsLongLong:
    242         case LengthModifier::AsQuad:
    243           return ArgType::PtrTo(Ctx.LongLongTy);
    244         case LengthModifier::AsInt64:
    245           return ArgType::PtrTo(ArgType(Ctx.LongLongTy, "__int64"));
    246         case LengthModifier::AsIntMax:
    247           return ArgType::PtrTo(ArgType(Ctx.getIntMaxType(), "intmax_t"));
    248         case LengthModifier::AsSizeT:
    249           // FIXME: ssize_t.
    250           return ArgType();
    251         case LengthModifier::AsPtrDiff:
    252           return ArgType::PtrTo(ArgType(Ctx.getPointerDiffType(), "ptrdiff_t"));
    253         case LengthModifier::AsLongDouble:
    254           // GNU extension.
    255           return ArgType::PtrTo(Ctx.LongLongTy);
    256         case LengthModifier::AsAllocate:
    257         case LengthModifier::AsMAllocate:
    258         case LengthModifier::AsInt32:
    259         case LengthModifier::AsInt3264:
    260         case LengthModifier::AsWide:
    261           return ArgType::Invalid();
    262       }
    263 
    264     // Unsigned int.
    265     case ConversionSpecifier::oArg:
    266     case ConversionSpecifier::OArg:
    267     case ConversionSpecifier::uArg:
    268     case ConversionSpecifier::UArg:
    269     case ConversionSpecifier::xArg:
    270     case ConversionSpecifier::XArg:
    271       switch (LM.getKind()) {
    272         case LengthModifier::None:
    273           return ArgType::PtrTo(Ctx.UnsignedIntTy);
    274         case LengthModifier::AsChar:
    275           return ArgType::PtrTo(Ctx.UnsignedCharTy);
    276         case LengthModifier::AsShort:
    277           return ArgType::PtrTo(Ctx.UnsignedShortTy);
    278         case LengthModifier::AsLong:
    279           return ArgType::PtrTo(Ctx.UnsignedLongTy);
    280         case LengthModifier::AsLongLong:
    281         case LengthModifier::AsQuad:
    282           return ArgType::PtrTo(Ctx.UnsignedLongLongTy);
    283         case LengthModifier::AsInt64:
    284           return ArgType::PtrTo(ArgType(Ctx.UnsignedLongLongTy, "unsigned __int64"));
    285         case LengthModifier::AsIntMax:
    286           return ArgType::PtrTo(ArgType(Ctx.getUIntMaxType(), "uintmax_t"));
    287         case LengthModifier::AsSizeT:
    288           return ArgType::PtrTo(ArgType(Ctx.getSizeType(), "size_t"));
    289         case LengthModifier::AsPtrDiff:
    290           // FIXME: Unsigned version of ptrdiff_t?
    291           return ArgType();
    292         case LengthModifier::AsLongDouble:
    293           // GNU extension.
    294           return ArgType::PtrTo(Ctx.UnsignedLongLongTy);
    295         case LengthModifier::AsAllocate:
    296         case LengthModifier::AsMAllocate:
    297         case LengthModifier::AsInt32:
    298         case LengthModifier::AsInt3264:
    299         case LengthModifier::AsWide:
    300           return ArgType::Invalid();
    301       }
    302 
    303     // Float.
    304     case ConversionSpecifier::aArg:
    305     case ConversionSpecifier::AArg:
    306     case ConversionSpecifier::eArg:
    307     case ConversionSpecifier::EArg:
    308     case ConversionSpecifier::fArg:
    309     case ConversionSpecifier::FArg:
    310     case ConversionSpecifier::gArg:
    311     case ConversionSpecifier::GArg:
    312       switch (LM.getKind()) {
    313         case LengthModifier::None:
    314           return ArgType::PtrTo(Ctx.FloatTy);
    315         case LengthModifier::AsLong:
    316           return ArgType::PtrTo(Ctx.DoubleTy);
    317         case LengthModifier::AsLongDouble:
    318           return ArgType::PtrTo(Ctx.LongDoubleTy);
    319         default:
    320           return ArgType::Invalid();
    321       }
    322 
    323     // Char, string and scanlist.
    324     case ConversionSpecifier::cArg:
    325     case ConversionSpecifier::sArg:
    326     case ConversionSpecifier::ScanListArg:
    327       switch (LM.getKind()) {
    328         case LengthModifier::None:
    329           return ArgType::PtrTo(ArgType::AnyCharTy);
    330         case LengthModifier::AsLong:
    331         case LengthModifier::AsWide:
    332           return ArgType::PtrTo(ArgType(Ctx.getWideCharType(), "wchar_t"));
    333         case LengthModifier::AsAllocate:
    334         case LengthModifier::AsMAllocate:
    335           return ArgType::PtrTo(ArgType::CStrTy);
    336         case LengthModifier::AsShort:
    337           if (Ctx.getTargetInfo().getTriple().isOSMSVCRT())
    338             return ArgType::PtrTo(ArgType::AnyCharTy);
    339         default:
    340           return ArgType::Invalid();
    341       }
    342     case ConversionSpecifier::CArg:
    343     case ConversionSpecifier::SArg:
    344       // FIXME: Mac OS X specific?
    345       switch (LM.getKind()) {
    346         case LengthModifier::None:
    347         case LengthModifier::AsWide:
    348           return ArgType::PtrTo(ArgType(Ctx.getWideCharType(), "wchar_t"));
    349         case LengthModifier::AsAllocate:
    350         case LengthModifier::AsMAllocate:
    351           return ArgType::PtrTo(ArgType(ArgType::WCStrTy, "wchar_t *"));
    352         case LengthModifier::AsShort:
    353           if (Ctx.getTargetInfo().getTriple().isOSMSVCRT())
    354             return ArgType::PtrTo(ArgType::AnyCharTy);
    355         default:
    356           return ArgType::Invalid();
    357       }
    358 
    359     // Pointer.
    360     case ConversionSpecifier::pArg:
    361       return ArgType::PtrTo(ArgType::CPointerTy);
    362 
    363     // Write-back.
    364     case ConversionSpecifier::nArg:
    365       switch (LM.getKind()) {
    366         case LengthModifier::None:
    367           return ArgType::PtrTo(Ctx.IntTy);
    368         case LengthModifier::AsChar:
    369           return ArgType::PtrTo(Ctx.SignedCharTy);
    370         case LengthModifier::AsShort:
    371           return ArgType::PtrTo(Ctx.ShortTy);
    372         case LengthModifier::AsLong:
    373           return ArgType::PtrTo(Ctx.LongTy);
    374         case LengthModifier::AsLongLong:
    375         case LengthModifier::AsQuad:
    376           return ArgType::PtrTo(Ctx.LongLongTy);
    377         case LengthModifier::AsInt64:
    378           return ArgType::PtrTo(ArgType(Ctx.LongLongTy, "__int64"));
    379         case LengthModifier::AsIntMax:
    380           return ArgType::PtrTo(ArgType(Ctx.getIntMaxType(), "intmax_t"));
    381         case LengthModifier::AsSizeT:
    382           return ArgType(); // FIXME: ssize_t
    383         case LengthModifier::AsPtrDiff:
    384           return ArgType::PtrTo(ArgType(Ctx.getPointerDiffType(), "ptrdiff_t"));
    385         case LengthModifier::AsLongDouble:
    386           return ArgType(); // FIXME: Is this a known extension?
    387         case LengthModifier::AsAllocate:
    388         case LengthModifier::AsMAllocate:
    389         case LengthModifier::AsInt32:
    390         case LengthModifier::AsInt3264:
    391         case LengthModifier::AsWide:
    392           return ArgType::Invalid();
    393         }
    394 
    395     default:
    396       break;
    397   }
    398 
    399   return ArgType();
    400 }
    401 
    402 bool ScanfSpecifier::fixType(QualType QT, QualType RawQT,
    403                              const LangOptions &LangOpt,
    404                              ASTContext &Ctx) {
    405 
    406   // %n is different from other conversion specifiers; don't try to fix it.
    407   if (CS.getKind() == ConversionSpecifier::nArg)
    408     return false;
    409 
    410   if (!QT->isPointerType())
    411     return false;
    412 
    413   QualType PT = QT->getPointeeType();
    414 
    415   // If it's an enum, get its underlying type.
    416   if (const EnumType *ETy = PT->getAs<EnumType>())
    417     PT = ETy->getDecl()->getIntegerType();
    418 
    419   const BuiltinType *BT = PT->getAs<BuiltinType>();
    420   if (!BT)
    421     return false;
    422 
    423   // Pointer to a character.
    424   if (PT->isAnyCharacterType()) {
    425     CS.setKind(ConversionSpecifier::sArg);
    426     if (PT->isWideCharType())
    427       LM.setKind(LengthModifier::AsWideChar);
    428     else
    429       LM.setKind(LengthModifier::None);
    430 
    431     // If we know the target array length, we can use it as a field width.
    432     if (const ConstantArrayType *CAT = Ctx.getAsConstantArrayType(RawQT)) {
    433       if (CAT->getSizeModifier() == ArrayType::Normal)
    434         FieldWidth = OptionalAmount(OptionalAmount::Constant,
    435                                     CAT->getSize().getZExtValue() - 1,
    436                                     "", 0, false);
    437 
    438     }
    439     return true;
    440   }
    441 
    442   // Figure out the length modifier.
    443   switch (BT->getKind()) {
    444     // no modifier
    445     case BuiltinType::UInt:
    446     case BuiltinType::Int:
    447     case BuiltinType::Float:
    448       LM.setKind(LengthModifier::None);
    449       break;
    450 
    451     // hh
    452     case BuiltinType::Char_U:
    453     case BuiltinType::UChar:
    454     case BuiltinType::Char_S:
    455     case BuiltinType::SChar:
    456       LM.setKind(LengthModifier::AsChar);
    457       break;
    458 
    459     // h
    460     case BuiltinType::Short:
    461     case BuiltinType::UShort:
    462       LM.setKind(LengthModifier::AsShort);
    463       break;
    464 
    465     // l
    466     case BuiltinType::Long:
    467     case BuiltinType::ULong:
    468     case BuiltinType::Double:
    469       LM.setKind(LengthModifier::AsLong);
    470       break;
    471 
    472     // ll
    473     case BuiltinType::LongLong:
    474     case BuiltinType::ULongLong:
    475       LM.setKind(LengthModifier::AsLongLong);
    476       break;
    477 
    478     // L
    479     case BuiltinType::LongDouble:
    480       LM.setKind(LengthModifier::AsLongDouble);
    481       break;
    482 
    483     // Don't know.
    484     default:
    485       return false;
    486   }
    487 
    488   // Handle size_t, ptrdiff_t, etc. that have dedicated length modifiers in C99.
    489   if (isa<TypedefType>(PT) && (LangOpt.C99 || LangOpt.CPlusPlus11))
    490     namedTypeToLengthModifier(PT, LM);
    491 
    492   // If fixing the length modifier was enough, we are done.
    493   if (hasValidLengthModifier(Ctx.getTargetInfo())) {
    494     const analyze_scanf::ArgType &AT = getArgType(Ctx);
    495     if (AT.isValid() && AT.matchesType(Ctx, QT))
    496       return true;
    497   }
    498 
    499   // Figure out the conversion specifier.
    500   if (PT->isRealFloatingType())
    501     CS.setKind(ConversionSpecifier::fArg);
    502   else if (PT->isSignedIntegerType())
    503     CS.setKind(ConversionSpecifier::dArg);
    504   else if (PT->isUnsignedIntegerType())
    505     CS.setKind(ConversionSpecifier::uArg);
    506   else
    507     llvm_unreachable("Unexpected type");
    508 
    509   return true;
    510 }
    511 
    512 void ScanfSpecifier::toString(raw_ostream &os) const {
    513   os << "%";
    514 
    515   if (usesPositionalArg())
    516     os << getPositionalArgIndex() << "$";
    517   if (SuppressAssignment)
    518     os << "*";
    519 
    520   FieldWidth.toString(os);
    521   os << LM.toString();
    522   os << CS.toString();
    523 }
    524 
    525 bool clang::analyze_format_string::ParseScanfString(FormatStringHandler &H,
    526                                                     const char *I,
    527                                                     const char *E,
    528                                                     const LangOptions &LO,
    529                                                     const TargetInfo &Target) {
    530 
    531   unsigned argIndex = 0;
    532 
    533   // Keep looking for a format specifier until we have exhausted the string.
    534   while (I != E) {
    535     const ScanfSpecifierResult &FSR = ParseScanfSpecifier(H, I, E, argIndex,
    536                                                           LO, Target);
    537     // Did a fail-stop error of any kind occur when parsing the specifier?
    538     // If so, don't do any more processing.
    539     if (FSR.shouldStop())
    540       return true;
    541       // Did we exhaust the string or encounter an error that
    542       // we can recover from?
    543     if (!FSR.hasValue())
    544       continue;
    545       // We have a format specifier.  Pass it to the callback.
    546     if (!H.HandleScanfSpecifier(FSR.getValue(), FSR.getStart(),
    547                                 I - FSR.getStart())) {
    548       return true;
    549     }
    550   }
    551   assert(I == E && "Format string not exhausted");
    552   return false;
    553 }
    554