| 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/ADT/StringExtras.h" |
| 18 | #include "llvm/Support/ConvertUTF.h" |
| 19 | #include <optional> |
| 20 | |
| 21 | using clang::analyze_format_string::ArgType; |
| 22 | using clang::analyze_format_string::ConversionSpecifier; |
| 23 | using clang::analyze_format_string::FormatSpecifier; |
| 24 | using clang::analyze_format_string::FormatStringHandler; |
| 25 | using clang::analyze_format_string::LengthModifier; |
| 26 | using clang::analyze_format_string::OptionalAmount; |
| 27 | using namespace clang; |
| 28 | |
| 29 | // Key function to FormatStringHandler. |
| 30 | FormatStringHandler::~FormatStringHandler() {} |
| 31 | |
| 32 | //===----------------------------------------------------------------------===// |
| 33 | // Functions for parsing format strings components in both printf and |
| 34 | // scanf format strings. |
| 35 | //===----------------------------------------------------------------------===// |
| 36 | |
| 37 | OptionalAmount clang::analyze_format_string::ParseAmount(const char *&Beg, |
| 38 | const char *E) { |
| 39 | const char *I = Beg; |
| 40 | UpdateOnReturn<const char *> UpdateBeg(Beg, I); |
| 41 | |
| 42 | unsigned accumulator = 0; |
| 43 | bool hasDigits = false; |
| 44 | |
| 45 | for (; I != E; ++I) { |
| 46 | char c = *I; |
| 47 | if (c >= '0' && c <= '9') { |
| 48 | hasDigits = true; |
| 49 | accumulator = (accumulator * 10) + (c - '0'); |
| 50 | continue; |
| 51 | } |
| 52 | |
| 53 | if (hasDigits) |
| 54 | return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg, |
| 55 | false); |
| 56 | |
| 57 | break; |
| 58 | } |
| 59 | |
| 60 | return OptionalAmount(); |
| 61 | } |
| 62 | |
| 63 | static bool ParseWidthModifier(const char *&I, const char *E, |
| 64 | unsigned &BitWidth, unsigned &ModifierLength) { |
| 65 | StringRef W = StringRef(I, E - I).take_while(F: llvm::isDigit); |
| 66 | if (W.empty() || W.front() == '0') |
| 67 | return false; |
| 68 | |
| 69 | if (W.getAsInteger(Radix: 10, Result&: BitWidth)) |
| 70 | return false; |
| 71 | |
| 72 | I = W.end(); |
| 73 | ModifierLength += W.size(); |
| 74 | |
| 75 | return true; |
| 76 | } |
| 77 | |
| 78 | OptionalAmount clang::analyze_format_string::ParseNonPositionAmount( |
| 79 | const char *&Beg, const char *E, unsigned &argIndex) { |
| 80 | if (*Beg == '*') { |
| 81 | ++Beg; |
| 82 | return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false); |
| 83 | } |
| 84 | |
| 85 | return ParseAmount(Beg, E); |
| 86 | } |
| 87 | |
| 88 | OptionalAmount clang::analyze_format_string::ParsePositionAmount( |
| 89 | FormatStringHandler &H, const char *Start, const char *&Beg, const char *E, |
| 90 | PositionContext p) { |
| 91 | if (*Beg == '*') { |
| 92 | const char *I = Beg + 1; |
| 93 | const OptionalAmount &Amt = ParseAmount(Beg&: I, E); |
| 94 | |
| 95 | if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) { |
| 96 | H.HandleInvalidPosition(startPos: Beg, posLen: I - Beg, p); |
| 97 | return OptionalAmount(false); |
| 98 | } |
| 99 | |
| 100 | if (I == E) { |
| 101 | // No more characters left? |
| 102 | H.HandleIncompleteSpecifier(startSpecifier: Start, specifierLen: E - Start); |
| 103 | return OptionalAmount(false); |
| 104 | } |
| 105 | |
| 106 | assert(Amt.getHowSpecified() == OptionalAmount::Constant); |
| 107 | |
| 108 | if (*I == '$') { |
| 109 | // Handle positional arguments |
| 110 | |
| 111 | // Special case: '*0$', since this is an easy mistake. |
| 112 | if (Amt.getConstantAmount() == 0) { |
| 113 | H.HandleZeroPosition(startPos: Beg, posLen: I - Beg + 1); |
| 114 | return OptionalAmount(false); |
| 115 | } |
| 116 | |
| 117 | const char *Tmp = Beg; |
| 118 | Beg = ++I; |
| 119 | |
| 120 | return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1, |
| 121 | Tmp, 0, true); |
| 122 | } |
| 123 | |
| 124 | H.HandleInvalidPosition(startPos: Beg, posLen: I - Beg, p); |
| 125 | return OptionalAmount(false); |
| 126 | } |
| 127 | |
| 128 | return ParseAmount(Beg, E); |
| 129 | } |
| 130 | |
| 131 | bool clang::analyze_format_string::ParseFieldWidth( |
| 132 | FormatStringHandler &H, FormatSpecifier &CS, const char *Start, |
| 133 | const char *&Beg, const char *E, unsigned *argIndex) { |
| 134 | // FIXME: Support negative field widths. |
| 135 | if (argIndex) { |
| 136 | CS.setFieldWidth(ParseNonPositionAmount(Beg, E, argIndex&: *argIndex)); |
| 137 | } else { |
| 138 | const OptionalAmount Amt = ParsePositionAmount( |
| 139 | H, Start, Beg, E, p: analyze_format_string::FieldWidthPos); |
| 140 | |
| 141 | if (Amt.isInvalid()) |
| 142 | return true; |
| 143 | CS.setFieldWidth(Amt); |
| 144 | } |
| 145 | return false; |
| 146 | } |
| 147 | |
| 148 | bool clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H, |
| 149 | FormatSpecifier &FS, |
| 150 | const char *Start, |
| 151 | const char *&Beg, |
| 152 | const char *E) { |
| 153 | const char *I = Beg; |
| 154 | |
| 155 | const OptionalAmount &Amt = ParseAmount(Beg&: I, E); |
| 156 | |
| 157 | if (I == E) { |
| 158 | // No more characters left? |
| 159 | H.HandleIncompleteSpecifier(startSpecifier: Start, specifierLen: E - Start); |
| 160 | return true; |
| 161 | } |
| 162 | |
| 163 | if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') { |
| 164 | // Warn that positional arguments are non-standard. |
| 165 | H.HandlePosition(startPos: Start, posLen: I - Start); |
| 166 | |
| 167 | // Special case: '%0$', since this is an easy mistake. |
| 168 | if (Amt.getConstantAmount() == 0) { |
| 169 | H.HandleZeroPosition(startPos: Start, posLen: I - Start); |
| 170 | return true; |
| 171 | } |
| 172 | |
| 173 | FS.setArgIndex(Amt.getConstantAmount() - 1); |
| 174 | FS.setUsesPositionalArg(); |
| 175 | // Update the caller's pointer if we decided to consume |
| 176 | // these characters. |
| 177 | Beg = I; |
| 178 | return false; |
| 179 | } |
| 180 | |
| 181 | return false; |
| 182 | } |
| 183 | |
| 184 | bool clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H, |
| 185 | FormatSpecifier &FS, |
| 186 | const char *&I, |
| 187 | const char *E, |
| 188 | const LangOptions &LO) { |
| 189 | if (!LO.OpenCL) |
| 190 | return false; |
| 191 | |
| 192 | const char *Start = I; |
| 193 | if (*I == 'v') { |
| 194 | ++I; |
| 195 | |
| 196 | if (I == E) { |
| 197 | H.HandleIncompleteSpecifier(startSpecifier: Start, specifierLen: E - Start); |
| 198 | return true; |
| 199 | } |
| 200 | |
| 201 | OptionalAmount NumElts = ParseAmount(Beg&: I, E); |
| 202 | if (NumElts.getHowSpecified() != OptionalAmount::Constant) { |
| 203 | H.HandleIncompleteSpecifier(startSpecifier: Start, specifierLen: E - Start); |
| 204 | return true; |
| 205 | } |
| 206 | |
| 207 | FS.setVectorNumElts(NumElts); |
| 208 | } |
| 209 | |
| 210 | return false; |
| 211 | } |
| 212 | |
| 213 | bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS, |
| 214 | const char *&I, |
| 215 | const char *E, |
| 216 | const LangOptions &LO, |
| 217 | bool IsScanf) { |
| 218 | LengthModifier::Kind lmKind = LengthModifier::None; |
| 219 | const char *lmPosition = I; |
| 220 | switch (*I) { |
| 221 | default: |
| 222 | return false; |
| 223 | case 'h': |
| 224 | ++I; |
| 225 | if (I != E && *I == 'h') { |
| 226 | ++I; |
| 227 | lmKind = LengthModifier::AsChar; |
| 228 | } else if (I != E && *I == 'l' && LO.OpenCL) { |
| 229 | ++I; |
| 230 | lmKind = LengthModifier::AsShortLong; |
| 231 | } else { |
| 232 | lmKind = LengthModifier::AsShort; |
| 233 | } |
| 234 | break; |
| 235 | case 'l': |
| 236 | ++I; |
| 237 | if (I != E && *I == 'l') { |
| 238 | ++I; |
| 239 | lmKind = LengthModifier::AsLongLong; |
| 240 | } else { |
| 241 | lmKind = LengthModifier::AsLong; |
| 242 | } |
| 243 | break; |
| 244 | case 'j': |
| 245 | lmKind = LengthModifier::AsIntMax; |
| 246 | ++I; |
| 247 | break; |
| 248 | case 'z': |
| 249 | lmKind = LengthModifier::AsSizeT; |
| 250 | ++I; |
| 251 | break; |
| 252 | case 't': |
| 253 | lmKind = LengthModifier::AsPtrDiff; |
| 254 | ++I; |
| 255 | break; |
| 256 | case 'L': |
| 257 | lmKind = LengthModifier::AsLongDouble; |
| 258 | ++I; |
| 259 | break; |
| 260 | case 'q': |
| 261 | lmKind = LengthModifier::AsQuad; |
| 262 | ++I; |
| 263 | break; |
| 264 | case 'a': |
| 265 | if (IsScanf && !LO.C99 && !LO.CPlusPlus11) { |
| 266 | // For scanf in C90, look at the next character to see if this should |
| 267 | // be parsed as the GNU extension 'a' length modifier. If not, this |
| 268 | // will be parsed as a conversion specifier. |
| 269 | ++I; |
| 270 | if (I != E && (*I == 's' || *I == 'S' || *I == '[')) { |
| 271 | lmKind = LengthModifier::AsAllocate; |
| 272 | break; |
| 273 | } |
| 274 | --I; |
| 275 | } |
| 276 | return false; |
| 277 | case 'm': |
| 278 | if (IsScanf) { |
| 279 | lmKind = LengthModifier::AsMAllocate; |
| 280 | ++I; |
| 281 | break; |
| 282 | } |
| 283 | return false; |
| 284 | // printf: AsInt64, AsInt32, AsInt3264 |
| 285 | // scanf: AsInt64 |
| 286 | case 'I': |
| 287 | if (I + 1 != E && I + 2 != E) { |
| 288 | if (I[1] == '6' && I[2] == '4') { |
| 289 | I += 3; |
| 290 | lmKind = LengthModifier::AsInt64; |
| 291 | break; |
| 292 | } |
| 293 | if (IsScanf) |
| 294 | return false; |
| 295 | |
| 296 | if (I[1] == '3' && I[2] == '2') { |
| 297 | I += 3; |
| 298 | lmKind = LengthModifier::AsInt32; |
| 299 | break; |
| 300 | } |
| 301 | } |
| 302 | ++I; |
| 303 | lmKind = LengthModifier::AsInt3264; |
| 304 | break; |
| 305 | case 'H': |
| 306 | if (LO.C23) { |
| 307 | lmKind = LengthModifier::AsDecimal32; |
| 308 | ++I; |
| 309 | break; |
| 310 | } |
| 311 | return false; |
| 312 | case 'D': |
| 313 | if (LO.C23) { |
| 314 | ++I; |
| 315 | if (I != E && *I == 'D') { |
| 316 | ++I; |
| 317 | lmKind = LengthModifier::AsDecimal128; |
| 318 | } else { |
| 319 | lmKind = LengthModifier::AsDecimal64; |
| 320 | } |
| 321 | break; |
| 322 | } |
| 323 | return false; |
| 324 | case 'w': |
| 325 | if (LO.C23) { |
| 326 | const char *WidthModifier = I + 1; |
| 327 | unsigned BitWidth = 0; |
| 328 | unsigned ModifierLength = 1; |
| 329 | |
| 330 | LengthModifier::Kind WidthKind = LengthModifier::AsIntN; |
| 331 | if (WidthModifier != E && *WidthModifier == 'f') { |
| 332 | WidthModifier = I + 2; |
| 333 | ModifierLength = 2; |
| 334 | WidthKind = LengthModifier::AsFastIntN; |
| 335 | } |
| 336 | |
| 337 | if (ParseWidthModifier(I&: WidthModifier, E, BitWidth, ModifierLength)) { |
| 338 | I = WidthModifier; |
| 339 | FS.setLengthModifier( |
| 340 | LengthModifier(lmPosition, WidthKind, BitWidth, ModifierLength)); |
| 341 | return true; |
| 342 | } |
| 343 | } |
| 344 | lmKind = LengthModifier::AsWide; |
| 345 | ++I; |
| 346 | break; |
| 347 | } |
| 348 | LengthModifier lm(lmPosition, lmKind); |
| 349 | FS.setLengthModifier(lm); |
| 350 | return true; |
| 351 | } |
| 352 | |
| 353 | bool clang::analyze_format_string::ParseUTF8InvalidSpecifier( |
| 354 | const char *SpecifierBegin, const char *FmtStrEnd, unsigned &Len) { |
| 355 | if (SpecifierBegin + 1 >= FmtStrEnd) |
| 356 | return false; |
| 357 | |
| 358 | const llvm::UTF8 *SB = |
| 359 | reinterpret_cast<const llvm::UTF8 *>(SpecifierBegin + 1); |
| 360 | const llvm::UTF8 *SE = reinterpret_cast<const llvm::UTF8 *>(FmtStrEnd); |
| 361 | const char FirstByte = *SB; |
| 362 | |
| 363 | // If the invalid specifier is a multibyte UTF-8 string, return the |
| 364 | // total length accordingly so that the conversion specifier can be |
| 365 | // properly updated to reflect a complete UTF-8 specifier. |
| 366 | unsigned NumBytes = llvm::getNumBytesForUTF8(firstByte: FirstByte); |
| 367 | if (NumBytes == 1) |
| 368 | return false; |
| 369 | if (SB + NumBytes > SE) |
| 370 | return false; |
| 371 | |
| 372 | Len = NumBytes + 1; |
| 373 | return true; |
| 374 | } |
| 375 | |
| 376 | //===----------------------------------------------------------------------===// |
| 377 | // Methods on ArgType. |
| 378 | //===----------------------------------------------------------------------===// |
| 379 | |
| 380 | static bool namedTypeToLengthModifierKind(ASTContext &Ctx, QualType QT, |
| 381 | LengthModifier::Kind &K) { |
| 382 | if (!Ctx.getLangOpts().C99 && !Ctx.getLangOpts().CPlusPlus) |
| 383 | return false; |
| 384 | for (/**/; const auto *TT = QT->getAs<TypedefType>(); QT = TT->desugar()) { |
| 385 | const auto *TD = TT->getDecl(); |
| 386 | const auto *DC = TT->getDecl()->getDeclContext(); |
| 387 | if (DC->isTranslationUnit() || DC->isStdNamespace()) { |
| 388 | StringRef Name = TD->getIdentifier()->getName(); |
| 389 | if (Name == "size_t" ) { |
| 390 | K = LengthModifier::AsSizeT; |
| 391 | return true; |
| 392 | } else if (Name == "ssize_t" /*Not C99, but common in Unix.*/) { |
| 393 | K = LengthModifier::AsSizeT; |
| 394 | return true; |
| 395 | } else if (Name == "ptrdiff_t" ) { |
| 396 | K = LengthModifier::AsPtrDiff; |
| 397 | return true; |
| 398 | } else if (Name == "intmax_t" ) { |
| 399 | K = LengthModifier::AsIntMax; |
| 400 | return true; |
| 401 | } else if (Name == "uintmax_t" ) { |
| 402 | K = LengthModifier::AsIntMax; |
| 403 | return true; |
| 404 | } |
| 405 | } |
| 406 | } |
| 407 | if (const auto *PST = QT->getAs<PredefinedSugarType>()) { |
| 408 | using Kind = PredefinedSugarType::Kind; |
| 409 | switch (PST->getKind()) { |
| 410 | case Kind::SizeT: |
| 411 | case Kind::SignedSizeT: |
| 412 | K = LengthModifier::AsSizeT; |
| 413 | return true; |
| 414 | case Kind::PtrdiffT: |
| 415 | K = LengthModifier::AsPtrDiff; |
| 416 | return true; |
| 417 | } |
| 418 | llvm_unreachable("unexpected kind" ); |
| 419 | } |
| 420 | return false; |
| 421 | } |
| 422 | |
| 423 | // Check whether T and E are compatible size_t/ptrdiff_t types. E must be |
| 424 | // consistent with LE. |
| 425 | // T is the type of the actual expression in the code to be checked, and E is |
| 426 | // the expected type parsed from the format string. |
| 427 | static clang::analyze_format_string::ArgType::MatchKind |
| 428 | matchesSizeTPtrdiffT(ASTContext &C, QualType T, QualType E) { |
| 429 | using MatchKind = clang::analyze_format_string::ArgType::MatchKind; |
| 430 | |
| 431 | if (!T->isIntegerType() || T->isBooleanType()) |
| 432 | return MatchKind::NoMatch; |
| 433 | |
| 434 | if (C.hasSameType(T1: T, T2: E)) |
| 435 | return MatchKind::Match; |
| 436 | |
| 437 | if (C.getCorrespondingSignedType(T: T.getCanonicalType()) != |
| 438 | C.getCorrespondingSignedType(T: E.getCanonicalType())) |
| 439 | return MatchKind::NoMatch; |
| 440 | |
| 441 | return MatchKind::NoMatchSignedness; |
| 442 | } |
| 443 | |
| 444 | clang::analyze_format_string::ArgType::MatchKind |
| 445 | ArgType::matchesType(ASTContext &C, QualType argTy) const { |
| 446 | // When using the format attribute in C++, you can receive a function or an |
| 447 | // array that will necessarily decay to a pointer when passed to the final |
| 448 | // format consumer. Apply decay before type comparison. |
| 449 | if (argTy->canDecayToPointerType()) |
| 450 | argTy = C.getDecayedType(T: argTy); |
| 451 | |
| 452 | if (Ptr) { |
| 453 | // It has to be a pointer. |
| 454 | const PointerType *PT = argTy->getAs<PointerType>(); |
| 455 | if (!PT) |
| 456 | return NoMatch; |
| 457 | |
| 458 | // We cannot write through a const qualified pointer. |
| 459 | if (PT->getPointeeType().isConstQualified()) |
| 460 | return NoMatch; |
| 461 | |
| 462 | argTy = PT->getPointeeType(); |
| 463 | } |
| 464 | |
| 465 | if (const auto *OBT = argTy->getAs<OverflowBehaviorType>()) |
| 466 | argTy = OBT->getUnderlyingType(); |
| 467 | |
| 468 | switch (K) { |
| 469 | case InvalidTy: |
| 470 | llvm_unreachable("ArgType must be valid" ); |
| 471 | |
| 472 | case UnsupportedTy: |
| 473 | return NoMatch; |
| 474 | |
| 475 | case UnknownTy: |
| 476 | return Match; |
| 477 | |
| 478 | case AnyCharTy: { |
| 479 | if (const auto *ED = argTy->getAsEnumDecl()) { |
| 480 | // If the enum is incomplete we know nothing about the underlying type. |
| 481 | // Assume that it's 'int'. Do not use the underlying type for a scoped |
| 482 | // enumeration. |
| 483 | if (!ED->isComplete()) |
| 484 | return NoMatch; |
| 485 | if (!ED->isScoped()) |
| 486 | argTy = ED->getIntegerType(); |
| 487 | } |
| 488 | |
| 489 | if (const auto *BT = argTy->getAs<BuiltinType>()) { |
| 490 | // The types are perfectly matched? |
| 491 | switch (BT->getKind()) { |
| 492 | default: |
| 493 | break; |
| 494 | case BuiltinType::Char_S: |
| 495 | case BuiltinType::SChar: |
| 496 | case BuiltinType::UChar: |
| 497 | case BuiltinType::Char_U: |
| 498 | return Match; |
| 499 | case BuiltinType::Bool: |
| 500 | if (!Ptr) |
| 501 | return Match; |
| 502 | break; |
| 503 | } |
| 504 | // "Partially matched" because of promotions? |
| 505 | if (!Ptr) { |
| 506 | switch (BT->getKind()) { |
| 507 | default: |
| 508 | break; |
| 509 | case BuiltinType::Int: |
| 510 | case BuiltinType::UInt: |
| 511 | return MatchPromotion; |
| 512 | case BuiltinType::Short: |
| 513 | case BuiltinType::UShort: |
| 514 | case BuiltinType::WChar_S: |
| 515 | case BuiltinType::WChar_U: |
| 516 | return NoMatchPromotionTypeConfusion; |
| 517 | } |
| 518 | } |
| 519 | } |
| 520 | return NoMatch; |
| 521 | } |
| 522 | |
| 523 | case SpecificTy: { |
| 524 | if (TK == TypeKind::SizeT || TK == TypeKind::PtrdiffT) { |
| 525 | return matchesSizeTPtrdiffT(C, T: argTy, E: T); |
| 526 | } |
| 527 | |
| 528 | if (const auto *ED = argTy->getAsEnumDecl()) { |
| 529 | // If the enum is incomplete we know nothing about the underlying type. |
| 530 | // Assume that it's 'int'. Do not use the underlying type for a scoped |
| 531 | // enumeration as that needs an exact match. |
| 532 | if (!ED->isComplete()) |
| 533 | argTy = C.IntTy; |
| 534 | else if (!ED->isScoped()) |
| 535 | argTy = ED->getIntegerType(); |
| 536 | } |
| 537 | |
| 538 | if (argTy->isSaturatedFixedPointType()) |
| 539 | argTy = C.getCorrespondingUnsaturatedType(Ty: argTy); |
| 540 | |
| 541 | argTy = C.getCanonicalType(T: argTy).getUnqualifiedType(); |
| 542 | |
| 543 | if (T == argTy) |
| 544 | return Match; |
| 545 | if (const auto *BT = argTy->getAs<BuiltinType>()) { |
| 546 | // Check if the only difference between them is signed vs unsigned |
| 547 | // if true, return match signedness. |
| 548 | switch (BT->getKind()) { |
| 549 | default: |
| 550 | break; |
| 551 | case BuiltinType::Bool: |
| 552 | if (Ptr && (T == C.UnsignedCharTy || T == C.SignedCharTy)) |
| 553 | return NoMatch; |
| 554 | [[fallthrough]]; |
| 555 | case BuiltinType::Char_S: |
| 556 | case BuiltinType::SChar: |
| 557 | if (T == C.UnsignedShortTy || T == C.ShortTy) |
| 558 | return NoMatchTypeConfusion; |
| 559 | if (T == C.UnsignedCharTy) |
| 560 | return NoMatchSignedness; |
| 561 | if (T == C.SignedCharTy) |
| 562 | return Match; |
| 563 | break; |
| 564 | case BuiltinType::Char_U: |
| 565 | case BuiltinType::UChar: |
| 566 | if (T == C.UnsignedShortTy || T == C.ShortTy) |
| 567 | return NoMatchTypeConfusion; |
| 568 | if (T == C.UnsignedCharTy) |
| 569 | return Match; |
| 570 | if (T == C.SignedCharTy) |
| 571 | return NoMatchSignedness; |
| 572 | break; |
| 573 | case BuiltinType::Short: |
| 574 | if (T == C.UnsignedShortTy) |
| 575 | return NoMatchSignedness; |
| 576 | break; |
| 577 | case BuiltinType::UShort: |
| 578 | if (T == C.ShortTy) |
| 579 | return NoMatchSignedness; |
| 580 | break; |
| 581 | case BuiltinType::Int: |
| 582 | if (T == C.UnsignedIntTy) |
| 583 | return NoMatchSignedness; |
| 584 | break; |
| 585 | case BuiltinType::UInt: |
| 586 | if (T == C.IntTy) |
| 587 | return NoMatchSignedness; |
| 588 | break; |
| 589 | case BuiltinType::Long: |
| 590 | if (T == C.UnsignedLongTy) |
| 591 | return NoMatchSignedness; |
| 592 | break; |
| 593 | case BuiltinType::ULong: |
| 594 | if (T == C.LongTy) |
| 595 | return NoMatchSignedness; |
| 596 | break; |
| 597 | case BuiltinType::LongLong: |
| 598 | if (T == C.UnsignedLongLongTy) |
| 599 | return NoMatchSignedness; |
| 600 | break; |
| 601 | case BuiltinType::ULongLong: |
| 602 | if (T == C.LongLongTy) |
| 603 | return NoMatchSignedness; |
| 604 | break; |
| 605 | } |
| 606 | // "Partially matched" because of promotions? |
| 607 | if (!Ptr) { |
| 608 | switch (BT->getKind()) { |
| 609 | default: |
| 610 | break; |
| 611 | case BuiltinType::Bool: |
| 612 | if (T == C.IntTy || T == C.UnsignedIntTy) |
| 613 | return MatchPromotion; |
| 614 | break; |
| 615 | case BuiltinType::Int: |
| 616 | case BuiltinType::UInt: |
| 617 | if (T == C.SignedCharTy || T == C.UnsignedCharTy || T == C.ShortTy || |
| 618 | T == C.UnsignedShortTy || T == C.WCharTy || T == C.WideCharTy) |
| 619 | return MatchPromotion; |
| 620 | break; |
| 621 | case BuiltinType::Char_U: |
| 622 | if (T == C.UnsignedIntTy) |
| 623 | return MatchPromotion; |
| 624 | if (T == C.UnsignedShortTy) |
| 625 | return NoMatchPromotionTypeConfusion; |
| 626 | break; |
| 627 | case BuiltinType::Char_S: |
| 628 | if (T == C.IntTy) |
| 629 | return MatchPromotion; |
| 630 | if (T == C.ShortTy) |
| 631 | return NoMatchPromotionTypeConfusion; |
| 632 | break; |
| 633 | case BuiltinType::Half: |
| 634 | case BuiltinType::Float: |
| 635 | if (T == C.DoubleTy) |
| 636 | return MatchPromotion; |
| 637 | break; |
| 638 | case BuiltinType::Short: |
| 639 | case BuiltinType::UShort: |
| 640 | if (T == C.SignedCharTy || T == C.UnsignedCharTy) |
| 641 | return NoMatchPromotionTypeConfusion; |
| 642 | break; |
| 643 | case BuiltinType::WChar_U: |
| 644 | case BuiltinType::WChar_S: |
| 645 | if (T != C.WCharTy && T != C.WideCharTy) |
| 646 | return NoMatchPromotionTypeConfusion; |
| 647 | } |
| 648 | } |
| 649 | } |
| 650 | return NoMatch; |
| 651 | } |
| 652 | |
| 653 | case CStrTy: |
| 654 | if (const auto *PT = argTy->getAs<PointerType>(); |
| 655 | PT && PT->getPointeeType()->isCharType()) |
| 656 | return Match; |
| 657 | return NoMatch; |
| 658 | |
| 659 | case WCStrTy: |
| 660 | if (const auto *PT = argTy->getAs<PointerType>(); |
| 661 | PT && |
| 662 | C.hasSameUnqualifiedType(T1: PT->getPointeeType(), T2: C.getWideCharType())) |
| 663 | return Match; |
| 664 | return NoMatch; |
| 665 | |
| 666 | case WIntTy: { |
| 667 | QualType WInt = C.getCanonicalType(T: C.getWIntType()).getUnqualifiedType(); |
| 668 | |
| 669 | if (C.getCanonicalType(T: argTy).getUnqualifiedType() == WInt) |
| 670 | return Match; |
| 671 | |
| 672 | QualType PromoArg = C.isPromotableIntegerType(T: argTy) |
| 673 | ? C.getPromotedIntegerType(PromotableType: argTy) |
| 674 | : argTy; |
| 675 | PromoArg = C.getCanonicalType(T: PromoArg).getUnqualifiedType(); |
| 676 | |
| 677 | // If the promoted argument is the corresponding signed type of the |
| 678 | // wint_t type, then it should match. |
| 679 | if (PromoArg->hasSignedIntegerRepresentation() && |
| 680 | C.getCorrespondingUnsignedType(T: PromoArg) == WInt) |
| 681 | return Match; |
| 682 | |
| 683 | return WInt == PromoArg ? Match : NoMatch; |
| 684 | } |
| 685 | |
| 686 | case CPointerTy: |
| 687 | if (const auto *PT = argTy->getAs<PointerType>()) { |
| 688 | QualType PointeeTy = PT->getPointeeType(); |
| 689 | if (PointeeTy->isVoidType() || (!Ptr && PointeeTy->isCharType())) |
| 690 | return Match; |
| 691 | return NoMatchPedantic; |
| 692 | } |
| 693 | |
| 694 | // nullptr_t* is not a double pointer, so reject when something like |
| 695 | // void** is expected. |
| 696 | // In C++, nullptr is promoted to void*. In C23, va_arg(ap, void*) is not |
| 697 | // undefined when the next argument is of type nullptr_t. |
| 698 | if (!Ptr && argTy->isNullPtrType()) |
| 699 | return C.getLangOpts().CPlusPlus ? MatchPromotion : Match; |
| 700 | |
| 701 | if (argTy->isObjCObjectPointerType() || argTy->isBlockPointerType()) |
| 702 | return NoMatchPedantic; |
| 703 | |
| 704 | return NoMatch; |
| 705 | |
| 706 | case ObjCPointerTy: { |
| 707 | if (argTy->getAs<ObjCObjectPointerType>() || |
| 708 | argTy->getAs<BlockPointerType>()) |
| 709 | return Match; |
| 710 | |
| 711 | // Handle implicit toll-free bridging. |
| 712 | if (const PointerType *PT = argTy->getAs<PointerType>()) { |
| 713 | // Things such as CFTypeRef are really just opaque pointers |
| 714 | // to C structs representing CF types that can often be bridged |
| 715 | // to Objective-C objects. Since the compiler doesn't know which |
| 716 | // structs can be toll-free bridged, we just accept them all. |
| 717 | QualType pointee = PT->getPointeeType(); |
| 718 | if (pointee->isStructureType() || pointee->isVoidType()) |
| 719 | return Match; |
| 720 | } |
| 721 | return NoMatch; |
| 722 | } |
| 723 | } |
| 724 | |
| 725 | llvm_unreachable("Invalid ArgType Kind!" ); |
| 726 | } |
| 727 | |
| 728 | static analyze_format_string::ArgType::MatchKind |
| 729 | integerTypeMatch(ASTContext &C, QualType A, QualType B, bool CheckSign) { |
| 730 | using MK = analyze_format_string::ArgType::MatchKind; |
| 731 | |
| 732 | uint64_t IntSize = C.getTypeSize(T: C.IntTy); |
| 733 | uint64_t ASize = C.getTypeSize(T: A); |
| 734 | uint64_t BSize = C.getTypeSize(T: B); |
| 735 | if (std::max(a: ASize, b: IntSize) != std::max(a: BSize, b: IntSize)) |
| 736 | return MK::NoMatch; |
| 737 | if (CheckSign && A->isSignedIntegerType() != B->isSignedIntegerType()) |
| 738 | return MK::NoMatchSignedness; |
| 739 | if (ASize != BSize) |
| 740 | return MK::MatchPromotion; |
| 741 | return MK::Match; |
| 742 | } |
| 743 | |
| 744 | analyze_format_string::ArgType::MatchKind |
| 745 | ArgType::matchesArgType(ASTContext &C, const ArgType &Other) const { |
| 746 | using AK = analyze_format_string::ArgType::Kind; |
| 747 | |
| 748 | // Per matchesType. |
| 749 | if (K == AK::InvalidTy || Other.K == AK::InvalidTy) |
| 750 | return NoMatch; |
| 751 | if (K == AK::UnsupportedTy || Other.K == AK::UnsupportedTy) |
| 752 | return NoMatch; |
| 753 | if (K == AK::UnknownTy || Other.K == AK::UnknownTy) |
| 754 | return Match; |
| 755 | |
| 756 | // Handle whether either (or both, or neither) sides has Ptr set, |
| 757 | // in addition to whether either (or both, or neither) sides is a SpecificTy |
| 758 | // that is a pointer. |
| 759 | ArgType Left = *this; |
| 760 | bool LeftWasPointer = false; |
| 761 | ArgType Right = Other; |
| 762 | bool RightWasPointer = false; |
| 763 | if (Left.Ptr) { |
| 764 | Left.Ptr = false; |
| 765 | LeftWasPointer = true; |
| 766 | } else if (Left.K == AK::SpecificTy && Left.T->isPointerType()) { |
| 767 | Left.T = Left.T->getPointeeType(); |
| 768 | LeftWasPointer = true; |
| 769 | } |
| 770 | if (Right.Ptr) { |
| 771 | Right.Ptr = false; |
| 772 | RightWasPointer = true; |
| 773 | } else if (Right.K == AK::SpecificTy && Right.T->isPointerType()) { |
| 774 | Right.T = Right.T->getPointeeType(); |
| 775 | RightWasPointer = true; |
| 776 | } |
| 777 | |
| 778 | if (LeftWasPointer != RightWasPointer) |
| 779 | return NoMatch; |
| 780 | |
| 781 | // Ensure that if at least one side is a SpecificTy, then Left is a |
| 782 | // SpecificTy. |
| 783 | if (Right.K == AK::SpecificTy) |
| 784 | std::swap(a&: Left, b&: Right); |
| 785 | |
| 786 | if (Left.K == AK::SpecificTy) { |
| 787 | if (Right.K == AK::SpecificTy) { |
| 788 | if (Left.TK != TypeKind::DontCare) { |
| 789 | return matchesSizeTPtrdiffT(C, T: Right.T, E: Left.T); |
| 790 | } else if (Right.TK != TypeKind::DontCare) { |
| 791 | return matchesSizeTPtrdiffT(C, T: Left.T, E: Right.T); |
| 792 | } |
| 793 | |
| 794 | auto Canon1 = C.getCanonicalType(T: Left.T); |
| 795 | auto Canon2 = C.getCanonicalType(T: Right.T); |
| 796 | if (Canon1 == Canon2) |
| 797 | return Match; |
| 798 | |
| 799 | auto *BT1 = QualType(Canon1)->getAs<BuiltinType>(); |
| 800 | auto *BT2 = QualType(Canon2)->getAs<BuiltinType>(); |
| 801 | if (BT1 == nullptr || BT2 == nullptr) |
| 802 | return NoMatch; |
| 803 | if (BT1 == BT2) |
| 804 | return Match; |
| 805 | |
| 806 | if (!LeftWasPointer && BT1->isInteger() && BT2->isInteger()) |
| 807 | return integerTypeMatch(C, A: Canon1, B: Canon2, CheckSign: true); |
| 808 | return NoMatch; |
| 809 | } else if (Right.K == AK::AnyCharTy) { |
| 810 | if (!LeftWasPointer && Left.T->isIntegerType()) |
| 811 | return integerTypeMatch(C, A: Left.T, B: C.CharTy, CheckSign: false); |
| 812 | return NoMatch; |
| 813 | } else if (Right.K == AK::WIntTy) { |
| 814 | if (!LeftWasPointer && Left.T->isIntegerType()) |
| 815 | return integerTypeMatch(C, A: Left.T, B: C.WIntTy, CheckSign: true); |
| 816 | return NoMatch; |
| 817 | } |
| 818 | // It's hypothetically possible to create an AK::SpecificTy ArgType |
| 819 | // that matches another kind of ArgType, but in practice Clang doesn't |
| 820 | // do that, so ignore that case. |
| 821 | return NoMatch; |
| 822 | } |
| 823 | |
| 824 | return Left.K == Right.K ? Match : NoMatch; |
| 825 | } |
| 826 | |
| 827 | ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const { |
| 828 | // Check for valid vector element types. |
| 829 | if (T.isNull()) |
| 830 | return ArgType::Invalid(); |
| 831 | |
| 832 | QualType Vec = C.getExtVectorType(VectorType: T, NumElts); |
| 833 | return ArgType(Vec, Name); |
| 834 | } |
| 835 | |
| 836 | ArgType ArgType::makeIntNType(ASTContext &Ctx, const LengthModifier &LengthMod, |
| 837 | bool Signed) { |
| 838 | bool IsFast = LengthMod.getKind() == LengthModifier::AsFastIntN; |
| 839 | QualType Ty = |
| 840 | IsFast ? Ctx.getLeastIntTypeForBitwidth(DestWidth: LengthMod.getBitWidth(), Signed) |
| 841 | : Ctx.getIntTypeForBitwidth(DestWidth: LengthMod.getBitWidth(), Signed); |
| 842 | if (Ty.isNull()) |
| 843 | return ArgType::Invalid(); |
| 844 | |
| 845 | ArgType Res(Ty); |
| 846 | Res.TK = IsFast ? (Signed ? TypeKind::FastIntN : TypeKind::FastUIntN) |
| 847 | : (Signed ? TypeKind::IntN : TypeKind::UIntN); |
| 848 | Res.BitWidth = LengthMod.getBitWidth(); |
| 849 | return Res; |
| 850 | } |
| 851 | |
| 852 | QualType ArgType::getRepresentativeType(ASTContext &C) const { |
| 853 | QualType Res; |
| 854 | switch (K) { |
| 855 | case InvalidTy: |
| 856 | llvm_unreachable("No representative type for Invalid ArgType" ); |
| 857 | case UnknownTy: |
| 858 | llvm_unreachable("No representative type for Unknown ArgType" ); |
| 859 | case UnsupportedTy: |
| 860 | llvm_unreachable("No representative type for Unsupported ArgType" ); |
| 861 | case AnyCharTy: |
| 862 | Res = C.CharTy; |
| 863 | break; |
| 864 | case SpecificTy: |
| 865 | if (TK == TypeKind::PtrdiffT || TK == TypeKind::SizeT) |
| 866 | // Using Name as name, so no need to show the uglified name. |
| 867 | Res = T->getCanonicalTypeInternal(); |
| 868 | else |
| 869 | Res = T; |
| 870 | break; |
| 871 | case CStrTy: |
| 872 | Res = C.getPointerType(T: C.CharTy); |
| 873 | break; |
| 874 | case WCStrTy: |
| 875 | Res = C.getPointerType(T: C.getWideCharType()); |
| 876 | break; |
| 877 | case ObjCPointerTy: |
| 878 | Res = C.ObjCBuiltinIdTy; |
| 879 | break; |
| 880 | case CPointerTy: |
| 881 | Res = C.VoidPtrTy; |
| 882 | break; |
| 883 | case WIntTy: { |
| 884 | Res = C.getWIntType(); |
| 885 | break; |
| 886 | } |
| 887 | } |
| 888 | |
| 889 | if (Ptr) |
| 890 | Res = C.getPointerType(T: Res); |
| 891 | return Res; |
| 892 | } |
| 893 | |
| 894 | std::string ArgType::getRepresentativeTypeName(ASTContext &C) const { |
| 895 | std::string S; |
| 896 | if (K != UnsupportedTy) |
| 897 | S = getRepresentativeType(C).getAsString(Policy: C.getPrintingPolicy()); |
| 898 | |
| 899 | std::string Alias; |
| 900 | if (Name) { |
| 901 | // Use a specific name for this type, e.g. "size_t". |
| 902 | Alias = Name; |
| 903 | } else { |
| 904 | const char *Prefix = nullptr; |
| 905 | switch (TK) { |
| 906 | case TypeKind::IntN: |
| 907 | Prefix = "int" ; |
| 908 | break; |
| 909 | case TypeKind::UIntN: |
| 910 | Prefix = "uint" ; |
| 911 | break; |
| 912 | case TypeKind::FastIntN: |
| 913 | Prefix = "int_fast" ; |
| 914 | break; |
| 915 | case TypeKind::FastUIntN: |
| 916 | Prefix = "uint_fast" ; |
| 917 | break; |
| 918 | case TypeKind::DontCare: |
| 919 | case TypeKind::SizeT: |
| 920 | case TypeKind::PtrdiffT: |
| 921 | break; |
| 922 | } |
| 923 | if (Prefix) { |
| 924 | Alias = Prefix; |
| 925 | Alias += std::to_string(val: BitWidth); |
| 926 | Alias += "_t" ; |
| 927 | } |
| 928 | } |
| 929 | if (!Alias.empty()) { |
| 930 | if (Ptr) { |
| 931 | // If ArgType is actually a pointer to T, append an asterisk. |
| 932 | Alias += (Alias[Alias.size() - 1] == '*') ? "*" : " *" ; |
| 933 | } |
| 934 | // If Alias is the same as the underlying type, e.g. wchar_t, then drop it. |
| 935 | if (S == Alias) |
| 936 | Alias.clear(); |
| 937 | } |
| 938 | |
| 939 | if (Alias.empty()) |
| 940 | return std::string("'" ) + S + "'" ; |
| 941 | |
| 942 | return std::string("'" ) + Alias + "'" + |
| 943 | (K == UnsupportedTy ? "" : " (aka '" + S + "')" ); |
| 944 | } |
| 945 | |
| 946 | //===----------------------------------------------------------------------===// |
| 947 | // Methods on OptionalAmount. |
| 948 | //===----------------------------------------------------------------------===// |
| 949 | |
| 950 | ArgType |
| 951 | analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const { |
| 952 | return Ctx.IntTy; |
| 953 | } |
| 954 | |
| 955 | //===----------------------------------------------------------------------===// |
| 956 | // Methods on LengthModifier. |
| 957 | //===----------------------------------------------------------------------===// |
| 958 | |
| 959 | StringRef analyze_format_string::LengthModifier::toString() const { |
| 960 | switch (kind) { |
| 961 | case AsChar: |
| 962 | return "hh" ; |
| 963 | case AsShort: |
| 964 | return "h" ; |
| 965 | case AsShortLong: |
| 966 | return "hl" ; |
| 967 | case AsLong: // or AsWideChar |
| 968 | return "l" ; |
| 969 | case AsLongLong: |
| 970 | return "ll" ; |
| 971 | case AsQuad: |
| 972 | return "q" ; |
| 973 | case AsIntMax: |
| 974 | return "j" ; |
| 975 | case AsSizeT: |
| 976 | return "z" ; |
| 977 | case AsPtrDiff: |
| 978 | return "t" ; |
| 979 | case AsInt32: |
| 980 | return "I32" ; |
| 981 | case AsInt3264: |
| 982 | return "I" ; |
| 983 | case AsInt64: |
| 984 | return "I64" ; |
| 985 | case AsLongDouble: |
| 986 | return "L" ; |
| 987 | case AsDecimal32: |
| 988 | return "H" ; |
| 989 | case AsDecimal64: |
| 990 | return "D" ; |
| 991 | case AsDecimal128: |
| 992 | return "DD" ; |
| 993 | case AsIntN: |
| 994 | case AsFastIntN: |
| 995 | return StringRef(Position, getLength()); |
| 996 | case AsAllocate: |
| 997 | return "a" ; |
| 998 | case AsMAllocate: |
| 999 | return "m" ; |
| 1000 | case AsWide: |
| 1001 | return "w" ; |
| 1002 | case None: |
| 1003 | return "" ; |
| 1004 | } |
| 1005 | llvm_unreachable("Invalid LengthModifier Kind!" ); |
| 1006 | } |
| 1007 | |
| 1008 | //===----------------------------------------------------------------------===// |
| 1009 | // Methods on ConversionSpecifier. |
| 1010 | //===----------------------------------------------------------------------===// |
| 1011 | |
| 1012 | const char *ConversionSpecifier::toString() const { |
| 1013 | switch (kind) { |
| 1014 | case bArg: |
| 1015 | return "b" ; |
| 1016 | case BArg: |
| 1017 | return "B" ; |
| 1018 | case dArg: |
| 1019 | return "d" ; |
| 1020 | case DArg: |
| 1021 | return "D" ; |
| 1022 | case iArg: |
| 1023 | return "i" ; |
| 1024 | case oArg: |
| 1025 | return "o" ; |
| 1026 | case OArg: |
| 1027 | return "O" ; |
| 1028 | case uArg: |
| 1029 | return "u" ; |
| 1030 | case UArg: |
| 1031 | return "U" ; |
| 1032 | case xArg: |
| 1033 | return "x" ; |
| 1034 | case XArg: |
| 1035 | return "X" ; |
| 1036 | case fArg: |
| 1037 | return "f" ; |
| 1038 | case FArg: |
| 1039 | return "F" ; |
| 1040 | case eArg: |
| 1041 | return "e" ; |
| 1042 | case EArg: |
| 1043 | return "E" ; |
| 1044 | case gArg: |
| 1045 | return "g" ; |
| 1046 | case GArg: |
| 1047 | return "G" ; |
| 1048 | case aArg: |
| 1049 | return "a" ; |
| 1050 | case AArg: |
| 1051 | return "A" ; |
| 1052 | case cArg: |
| 1053 | return "c" ; |
| 1054 | case sArg: |
| 1055 | return "s" ; |
| 1056 | case pArg: |
| 1057 | return "p" ; |
| 1058 | case PArg: |
| 1059 | return "P" ; |
| 1060 | case nArg: |
| 1061 | return "n" ; |
| 1062 | case PercentArg: |
| 1063 | return "%" ; |
| 1064 | case ScanListArg: |
| 1065 | return "[" ; |
| 1066 | case InvalidSpecifier: |
| 1067 | return nullptr; |
| 1068 | |
| 1069 | // POSIX unicode extensions. |
| 1070 | case CArg: |
| 1071 | return "C" ; |
| 1072 | case SArg: |
| 1073 | return "S" ; |
| 1074 | |
| 1075 | // Objective-C specific specifiers. |
| 1076 | case ObjCObjArg: |
| 1077 | return "@" ; |
| 1078 | |
| 1079 | // FreeBSD kernel specific specifiers. |
| 1080 | case FreeBSDbArg: |
| 1081 | return "b" ; |
| 1082 | case FreeBSDDArg: |
| 1083 | return "D" ; |
| 1084 | case FreeBSDrArg: |
| 1085 | return "r" ; |
| 1086 | case FreeBSDyArg: |
| 1087 | return "y" ; |
| 1088 | |
| 1089 | // GlibC specific specifiers. |
| 1090 | case PrintErrno: |
| 1091 | return "m" ; |
| 1092 | |
| 1093 | // MS specific specifiers. |
| 1094 | case ZArg: |
| 1095 | return "Z" ; |
| 1096 | |
| 1097 | // ISO/IEC TR 18037 (fixed-point) specific specifiers. |
| 1098 | case rArg: |
| 1099 | return "r" ; |
| 1100 | case RArg: |
| 1101 | return "R" ; |
| 1102 | case kArg: |
| 1103 | return "k" ; |
| 1104 | case KArg: |
| 1105 | return "K" ; |
| 1106 | } |
| 1107 | return nullptr; |
| 1108 | } |
| 1109 | |
| 1110 | std::optional<ConversionSpecifier> |
| 1111 | ConversionSpecifier::getStandardSpecifier() const { |
| 1112 | ConversionSpecifier::Kind NewKind; |
| 1113 | |
| 1114 | switch (getKind()) { |
| 1115 | default: |
| 1116 | return std::nullopt; |
| 1117 | case DArg: |
| 1118 | NewKind = dArg; |
| 1119 | break; |
| 1120 | case UArg: |
| 1121 | NewKind = uArg; |
| 1122 | break; |
| 1123 | case OArg: |
| 1124 | NewKind = oArg; |
| 1125 | break; |
| 1126 | } |
| 1127 | |
| 1128 | ConversionSpecifier FixedCS(*this); |
| 1129 | FixedCS.setKind(NewKind); |
| 1130 | return FixedCS; |
| 1131 | } |
| 1132 | |
| 1133 | //===----------------------------------------------------------------------===// |
| 1134 | // Methods on OptionalAmount. |
| 1135 | //===----------------------------------------------------------------------===// |
| 1136 | |
| 1137 | void OptionalAmount::toString(raw_ostream &os) const { |
| 1138 | switch (hs) { |
| 1139 | case Invalid: |
| 1140 | case NotSpecified: |
| 1141 | return; |
| 1142 | case Arg: |
| 1143 | if (UsesDotPrefix) |
| 1144 | os << "." ; |
| 1145 | if (usesPositionalArg()) |
| 1146 | os << "*" << getPositionalArgIndex() << "$" ; |
| 1147 | else |
| 1148 | os << "*" ; |
| 1149 | break; |
| 1150 | case Constant: |
| 1151 | if (UsesDotPrefix) |
| 1152 | os << "." ; |
| 1153 | os << amt; |
| 1154 | break; |
| 1155 | } |
| 1156 | } |
| 1157 | |
| 1158 | bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target, |
| 1159 | const LangOptions &LO) const { |
| 1160 | switch (LM.getKind()) { |
| 1161 | case LengthModifier::None: |
| 1162 | return true; |
| 1163 | |
| 1164 | // Handle most integer flags |
| 1165 | case LengthModifier::AsShort: |
| 1166 | // Length modifier only applies to FP vectors. |
| 1167 | if (LO.OpenCL && CS.isDoubleArg()) |
| 1168 | return !VectorNumElts.isInvalid(); |
| 1169 | |
| 1170 | if (CS.isFixedPointArg()) |
| 1171 | return true; |
| 1172 | |
| 1173 | if (Target.getTriple().isOSMSVCRT()) { |
| 1174 | switch (CS.getKind()) { |
| 1175 | case ConversionSpecifier::cArg: |
| 1176 | case ConversionSpecifier::CArg: |
| 1177 | case ConversionSpecifier::sArg: |
| 1178 | case ConversionSpecifier::SArg: |
| 1179 | case ConversionSpecifier::ZArg: |
| 1180 | return true; |
| 1181 | default: |
| 1182 | break; |
| 1183 | } |
| 1184 | } |
| 1185 | [[fallthrough]]; |
| 1186 | case LengthModifier::AsChar: |
| 1187 | case LengthModifier::AsLongLong: |
| 1188 | case LengthModifier::AsQuad: |
| 1189 | case LengthModifier::AsIntMax: |
| 1190 | case LengthModifier::AsSizeT: |
| 1191 | case LengthModifier::AsPtrDiff: |
| 1192 | switch (CS.getKind()) { |
| 1193 | case ConversionSpecifier::bArg: |
| 1194 | case ConversionSpecifier::BArg: |
| 1195 | case ConversionSpecifier::dArg: |
| 1196 | case ConversionSpecifier::DArg: |
| 1197 | case ConversionSpecifier::iArg: |
| 1198 | case ConversionSpecifier::oArg: |
| 1199 | case ConversionSpecifier::OArg: |
| 1200 | case ConversionSpecifier::uArg: |
| 1201 | case ConversionSpecifier::UArg: |
| 1202 | case ConversionSpecifier::xArg: |
| 1203 | case ConversionSpecifier::XArg: |
| 1204 | case ConversionSpecifier::nArg: |
| 1205 | return true; |
| 1206 | case ConversionSpecifier::FreeBSDrArg: |
| 1207 | case ConversionSpecifier::FreeBSDyArg: |
| 1208 | return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS(); |
| 1209 | default: |
| 1210 | return false; |
| 1211 | } |
| 1212 | |
| 1213 | case LengthModifier::AsShortLong: |
| 1214 | return LO.OpenCL && !VectorNumElts.isInvalid(); |
| 1215 | |
| 1216 | // Handle 'l' flag |
| 1217 | case LengthModifier::AsLong: // or AsWideChar |
| 1218 | if (CS.isDoubleArg()) { |
| 1219 | // Invalid for OpenCL FP scalars. |
| 1220 | if (LO.OpenCL && VectorNumElts.isInvalid()) |
| 1221 | return false; |
| 1222 | return true; |
| 1223 | } |
| 1224 | |
| 1225 | if (CS.isFixedPointArg()) |
| 1226 | return true; |
| 1227 | |
| 1228 | switch (CS.getKind()) { |
| 1229 | case ConversionSpecifier::bArg: |
| 1230 | case ConversionSpecifier::BArg: |
| 1231 | case ConversionSpecifier::dArg: |
| 1232 | case ConversionSpecifier::DArg: |
| 1233 | case ConversionSpecifier::iArg: |
| 1234 | case ConversionSpecifier::oArg: |
| 1235 | case ConversionSpecifier::OArg: |
| 1236 | case ConversionSpecifier::uArg: |
| 1237 | case ConversionSpecifier::UArg: |
| 1238 | case ConversionSpecifier::xArg: |
| 1239 | case ConversionSpecifier::XArg: |
| 1240 | case ConversionSpecifier::nArg: |
| 1241 | case ConversionSpecifier::cArg: |
| 1242 | case ConversionSpecifier::sArg: |
| 1243 | case ConversionSpecifier::ScanListArg: |
| 1244 | case ConversionSpecifier::ZArg: |
| 1245 | return true; |
| 1246 | case ConversionSpecifier::FreeBSDrArg: |
| 1247 | case ConversionSpecifier::FreeBSDyArg: |
| 1248 | return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS(); |
| 1249 | default: |
| 1250 | return false; |
| 1251 | } |
| 1252 | |
| 1253 | case LengthModifier::AsLongDouble: |
| 1254 | switch (CS.getKind()) { |
| 1255 | case ConversionSpecifier::aArg: |
| 1256 | case ConversionSpecifier::AArg: |
| 1257 | case ConversionSpecifier::fArg: |
| 1258 | case ConversionSpecifier::FArg: |
| 1259 | case ConversionSpecifier::eArg: |
| 1260 | case ConversionSpecifier::EArg: |
| 1261 | case ConversionSpecifier::gArg: |
| 1262 | case ConversionSpecifier::GArg: |
| 1263 | return true; |
| 1264 | // GNU libc extension. |
| 1265 | case ConversionSpecifier::dArg: |
| 1266 | case ConversionSpecifier::iArg: |
| 1267 | case ConversionSpecifier::oArg: |
| 1268 | case ConversionSpecifier::uArg: |
| 1269 | case ConversionSpecifier::xArg: |
| 1270 | case ConversionSpecifier::XArg: |
| 1271 | return !Target.getTriple().isOSDarwin() && |
| 1272 | !Target.getTriple().isOSWindows(); |
| 1273 | default: |
| 1274 | return false; |
| 1275 | } |
| 1276 | |
| 1277 | case LengthModifier::AsIntN: |
| 1278 | case LengthModifier::AsFastIntN: { |
| 1279 | if (!LO.C23) |
| 1280 | return false; |
| 1281 | |
| 1282 | TargetInfo::IntType TargetType = |
| 1283 | LM.getKind() == LengthModifier::AsIntN |
| 1284 | ? Target.getIntTypeByWidth(BitWidth: LM.getBitWidth(), /*IsSigned=*/true) |
| 1285 | : Target.getLeastIntTypeByWidth(BitWidth: LM.getBitWidth(), |
| 1286 | /*IsSigned=*/true); |
| 1287 | if (TargetType == TargetInfo::NoInt) |
| 1288 | return false; |
| 1289 | |
| 1290 | switch (CS.getKind()) { |
| 1291 | case ConversionSpecifier::bArg: |
| 1292 | case ConversionSpecifier::BArg: |
| 1293 | case ConversionSpecifier::dArg: |
| 1294 | case ConversionSpecifier::iArg: |
| 1295 | case ConversionSpecifier::oArg: |
| 1296 | case ConversionSpecifier::uArg: |
| 1297 | case ConversionSpecifier::xArg: |
| 1298 | case ConversionSpecifier::XArg: |
| 1299 | case ConversionSpecifier::nArg: |
| 1300 | return true; |
| 1301 | default: |
| 1302 | return false; |
| 1303 | } |
| 1304 | } |
| 1305 | |
| 1306 | case LengthModifier::AsAllocate: |
| 1307 | switch (CS.getKind()) { |
| 1308 | case ConversionSpecifier::sArg: |
| 1309 | case ConversionSpecifier::SArg: |
| 1310 | case ConversionSpecifier::ScanListArg: |
| 1311 | return true; |
| 1312 | default: |
| 1313 | return false; |
| 1314 | } |
| 1315 | |
| 1316 | case LengthModifier::AsMAllocate: |
| 1317 | switch (CS.getKind()) { |
| 1318 | case ConversionSpecifier::cArg: |
| 1319 | case ConversionSpecifier::CArg: |
| 1320 | case ConversionSpecifier::sArg: |
| 1321 | case ConversionSpecifier::SArg: |
| 1322 | case ConversionSpecifier::ScanListArg: |
| 1323 | return true; |
| 1324 | default: |
| 1325 | return false; |
| 1326 | } |
| 1327 | case LengthModifier::AsInt32: |
| 1328 | case LengthModifier::AsInt3264: |
| 1329 | case LengthModifier::AsInt64: |
| 1330 | switch (CS.getKind()) { |
| 1331 | case ConversionSpecifier::dArg: |
| 1332 | case ConversionSpecifier::iArg: |
| 1333 | case ConversionSpecifier::oArg: |
| 1334 | case ConversionSpecifier::uArg: |
| 1335 | case ConversionSpecifier::xArg: |
| 1336 | case ConversionSpecifier::XArg: |
| 1337 | return Target.getTriple().isOSMSVCRT(); |
| 1338 | default: |
| 1339 | return false; |
| 1340 | } |
| 1341 | case LengthModifier::AsWide: |
| 1342 | switch (CS.getKind()) { |
| 1343 | case ConversionSpecifier::cArg: |
| 1344 | case ConversionSpecifier::CArg: |
| 1345 | case ConversionSpecifier::sArg: |
| 1346 | case ConversionSpecifier::SArg: |
| 1347 | case ConversionSpecifier::ZArg: |
| 1348 | return Target.getTriple().isOSMSVCRT(); |
| 1349 | default: |
| 1350 | return false; |
| 1351 | } |
| 1352 | |
| 1353 | case LengthModifier::AsDecimal32: |
| 1354 | case LengthModifier::AsDecimal64: |
| 1355 | case LengthModifier::AsDecimal128: |
| 1356 | switch (CS.getKind()) { |
| 1357 | case ConversionSpecifier::aArg: |
| 1358 | case ConversionSpecifier::AArg: |
| 1359 | case ConversionSpecifier::eArg: |
| 1360 | case ConversionSpecifier::EArg: |
| 1361 | case ConversionSpecifier::fArg: |
| 1362 | case ConversionSpecifier::FArg: |
| 1363 | case ConversionSpecifier::gArg: |
| 1364 | case ConversionSpecifier::GArg: |
| 1365 | return LO.C23; |
| 1366 | default: |
| 1367 | return false; |
| 1368 | } |
| 1369 | } |
| 1370 | llvm_unreachable("Invalid LengthModifier Kind!" ); |
| 1371 | } |
| 1372 | |
| 1373 | bool FormatSpecifier::hasStandardLengthModifier() const { |
| 1374 | switch (LM.getKind()) { |
| 1375 | case LengthModifier::None: |
| 1376 | case LengthModifier::AsChar: |
| 1377 | case LengthModifier::AsShort: |
| 1378 | case LengthModifier::AsLong: |
| 1379 | case LengthModifier::AsLongLong: |
| 1380 | case LengthModifier::AsIntMax: |
| 1381 | case LengthModifier::AsSizeT: |
| 1382 | case LengthModifier::AsPtrDiff: |
| 1383 | case LengthModifier::AsLongDouble: |
| 1384 | case LengthModifier::AsDecimal32: |
| 1385 | case LengthModifier::AsDecimal64: |
| 1386 | case LengthModifier::AsDecimal128: |
| 1387 | case LengthModifier::AsIntN: |
| 1388 | case LengthModifier::AsFastIntN: |
| 1389 | return true; |
| 1390 | case LengthModifier::AsAllocate: |
| 1391 | case LengthModifier::AsMAllocate: |
| 1392 | case LengthModifier::AsQuad: |
| 1393 | case LengthModifier::AsInt32: |
| 1394 | case LengthModifier::AsInt3264: |
| 1395 | case LengthModifier::AsInt64: |
| 1396 | case LengthModifier::AsWide: |
| 1397 | case LengthModifier::AsShortLong: // ??? |
| 1398 | return false; |
| 1399 | } |
| 1400 | llvm_unreachable("Invalid LengthModifier Kind!" ); |
| 1401 | } |
| 1402 | |
| 1403 | bool FormatSpecifier::hasStandardConversionSpecifier( |
| 1404 | const LangOptions &LangOpt) const { |
| 1405 | switch (CS.getKind()) { |
| 1406 | case ConversionSpecifier::bArg: |
| 1407 | case ConversionSpecifier::BArg: |
| 1408 | case ConversionSpecifier::cArg: |
| 1409 | case ConversionSpecifier::dArg: |
| 1410 | case ConversionSpecifier::iArg: |
| 1411 | case ConversionSpecifier::oArg: |
| 1412 | case ConversionSpecifier::uArg: |
| 1413 | case ConversionSpecifier::xArg: |
| 1414 | case ConversionSpecifier::XArg: |
| 1415 | case ConversionSpecifier::fArg: |
| 1416 | case ConversionSpecifier::FArg: |
| 1417 | case ConversionSpecifier::eArg: |
| 1418 | case ConversionSpecifier::EArg: |
| 1419 | case ConversionSpecifier::gArg: |
| 1420 | case ConversionSpecifier::GArg: |
| 1421 | case ConversionSpecifier::aArg: |
| 1422 | case ConversionSpecifier::AArg: |
| 1423 | case ConversionSpecifier::sArg: |
| 1424 | case ConversionSpecifier::pArg: |
| 1425 | case ConversionSpecifier::nArg: |
| 1426 | case ConversionSpecifier::ObjCObjArg: |
| 1427 | case ConversionSpecifier::ScanListArg: |
| 1428 | case ConversionSpecifier::PercentArg: |
| 1429 | case ConversionSpecifier::PArg: |
| 1430 | return true; |
| 1431 | case ConversionSpecifier::CArg: |
| 1432 | case ConversionSpecifier::SArg: |
| 1433 | return LangOpt.ObjC; |
| 1434 | case ConversionSpecifier::InvalidSpecifier: |
| 1435 | case ConversionSpecifier::FreeBSDbArg: |
| 1436 | case ConversionSpecifier::FreeBSDDArg: |
| 1437 | case ConversionSpecifier::FreeBSDrArg: |
| 1438 | case ConversionSpecifier::FreeBSDyArg: |
| 1439 | case ConversionSpecifier::PrintErrno: |
| 1440 | case ConversionSpecifier::DArg: |
| 1441 | case ConversionSpecifier::OArg: |
| 1442 | case ConversionSpecifier::UArg: |
| 1443 | case ConversionSpecifier::ZArg: |
| 1444 | return false; |
| 1445 | case ConversionSpecifier::rArg: |
| 1446 | case ConversionSpecifier::RArg: |
| 1447 | case ConversionSpecifier::kArg: |
| 1448 | case ConversionSpecifier::KArg: |
| 1449 | return LangOpt.FixedPoint; |
| 1450 | } |
| 1451 | llvm_unreachable("Invalid ConversionSpecifier Kind!" ); |
| 1452 | } |
| 1453 | |
| 1454 | bool FormatSpecifier::hasStandardLengthConversionCombination() const { |
| 1455 | if (LM.getKind() == LengthModifier::AsLongDouble) { |
| 1456 | switch (CS.getKind()) { |
| 1457 | case ConversionSpecifier::dArg: |
| 1458 | case ConversionSpecifier::iArg: |
| 1459 | case ConversionSpecifier::oArg: |
| 1460 | case ConversionSpecifier::uArg: |
| 1461 | case ConversionSpecifier::xArg: |
| 1462 | case ConversionSpecifier::XArg: |
| 1463 | return false; |
| 1464 | default: |
| 1465 | return true; |
| 1466 | } |
| 1467 | } |
| 1468 | return true; |
| 1469 | } |
| 1470 | |
| 1471 | std::optional<LengthModifier> |
| 1472 | FormatSpecifier::getCorrectedLengthModifier() const { |
| 1473 | if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) { |
| 1474 | if (LM.getKind() == LengthModifier::AsLongDouble || |
| 1475 | LM.getKind() == LengthModifier::AsQuad) { |
| 1476 | LengthModifier FixedLM(LM); |
| 1477 | FixedLM.setKind(LengthModifier::AsLongLong); |
| 1478 | return FixedLM; |
| 1479 | } |
| 1480 | } |
| 1481 | |
| 1482 | return std::nullopt; |
| 1483 | } |
| 1484 | |
| 1485 | bool FormatSpecifier::namedTypeToLengthModifier(ASTContext &Ctx, QualType QT, |
| 1486 | LengthModifier &LM) { |
| 1487 | if (LengthModifier::Kind Out = LengthModifier::Kind::None; |
| 1488 | namedTypeToLengthModifierKind(Ctx, QT, K&: Out)) { |
| 1489 | LM.setKind(Out); |
| 1490 | return true; |
| 1491 | } |
| 1492 | return false; |
| 1493 | } |
| 1494 | |