clang/lib/Lex/PPExpressions.cpp

   1 //===--- PPExpressions.cpp - Preprocessor Expression Evaluation -----------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file implements the Preprocessor::EvaluateDirectiveExpression method,
  10 // which parses and evaluates integer constant expressions for #if directives.
  11 //
  12 //===----------------------------------------------------------------------===//
  13 //
  14 // FIXME: implement testing for #assert's.
  15 //
  16 //===----------------------------------------------------------------------===//
  17
  18 #include "clang/Basic/IdentifierTable.h"
  19 #include "clang/Basic/SourceLocation.h"
  20 #include "clang/Basic/SourceManager.h"
  21 #include "clang/Basic/TargetInfo.h"
  22 #include "clang/Basic/TokenKinds.h"
  23 #include "clang/Lex/CodeCompletionHandler.h"
  24 #include "clang/Lex/LexDiagnostic.h"
  25 #include "clang/Lex/LiteralSupport.h"
  26 #include "clang/Lex/MacroInfo.h"
  27 #include "clang/Lex/PPCallbacks.h"
  28 #include "clang/Lex/Preprocessor.h"
  29 #include "clang/Lex/Token.h"
  30 #include "llvm/ADT/APSInt.h"
  31 #include "llvm/ADT/STLExtras.h"
  32 #include "llvm/ADT/StringExtras.h"
  33 #include "llvm/ADT/StringRef.h"
  34 #include "llvm/Support/ErrorHandling.h"
  35 #include "llvm/Support/SaveAndRestore.h"
  36 #include <cassert>
  37
  38 using namespace clang;
  39
  40 namespace {
  41
  42 /// PPValue - Represents the value of a subexpression of a preprocessor
  43 /// conditional and the source range covered by it.
  44 class PPValue {
  45   SourceRange Range;
  46   IdentifierInfo *II = nullptr;
  47
  48 public:
  49   llvm::APSInt Val;
  50
  51   // Default ctor - Construct an 'invalid' PPValue.
  52   PPValue(unsigned BitWidth) : Val(BitWidth) {}
  53
  54   // If this value was produced by directly evaluating an identifier, produce
  55   // that identifier.
  56   IdentifierInfo *getIdentifier() const { return II; }
  57   void setIdentifier(IdentifierInfo *II) { this->II = II; }
  58
  59   unsigned getBitWidth() const { return Val.getBitWidth(); }
  60   bool isUnsigned() const { return Val.isUnsigned(); }
  61
  62   SourceRange getRange() const { return Range; }
  63
  64   void setRange(SourceLocation L) { Range.setBegin(L); Range.setEnd(L); }
  65   void setRange(SourceLocation B, SourceLocation E) {
  66     Range.setBegin(B); Range.setEnd(E);
  67   }
  68   void setBegin(SourceLocation L) { Range.setBegin(L); }
  69   void setEnd(SourceLocation L) { Range.setEnd(L); }
  70 };
  71
  72 } // end anonymous namespace
  73
  74 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
  75                                      Token &PeekTok, bool ValueLive,
  76                                      bool &IncludedUndefinedIds,
  77                                      Preprocessor &PP);
  78
  79 /// DefinedTracker - This struct is used while parsing expressions to keep track
  80 /// of whether !defined(X) has been seen.
  81 ///
  82 /// With this simple scheme, we handle the basic forms:
  83 ///    !defined(X)   and !defined X
  84 /// but we also trivially handle (silly) stuff like:
  85 ///    !!!defined(X) and +!defined(X) and !+!+!defined(X) and !(defined(X)).
  86 struct DefinedTracker {
  87   /// Each time a Value is evaluated, it returns information about whether the
  88   /// parsed value is of the form defined(X), !defined(X) or is something else.
  89   enum TrackerState {
  90     DefinedMacro,        // defined(X)
  91     NotDefinedMacro,     // !defined(X)
  92     Unknown              // Something else.
  93   } State;
  94   /// TheMacro - When the state is DefinedMacro or NotDefinedMacro, this
  95   /// indicates the macro that was checked.
  96   IdentifierInfo *TheMacro;
  97   bool IncludedUndefinedIds = false;
  98 };
  99
 100 /// EvaluateDefined - Process a 'defined(sym)' expression.
 101 static bool EvaluateDefined(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
 102                             bool ValueLive, Preprocessor &PP) {
 103   SourceLocation beginLoc(PeekTok.getLocation());
 104   Result.setBegin(beginLoc);
 105
 106   // Get the next token, don't expand it.
 107   PP.LexUnexpandedNonComment(PeekTok);
 108
 109   // Two options, it can either be a pp-identifier or a (.
 110   SourceLocation LParenLoc;
 111   if (PeekTok.is(tok::l_paren)) {
 112     // Found a paren, remember we saw it and skip it.
 113     LParenLoc = PeekTok.getLocation();
 114     PP.LexUnexpandedNonComment(PeekTok);
 115   }
 116
 117   if (PeekTok.is(tok::code_completion)) {
 118     if (PP.getCodeCompletionHandler())
 119       PP.getCodeCompletionHandler()->CodeCompleteMacroName(false);
 120     PP.setCodeCompletionReached();
 121     PP.LexUnexpandedNonComment(PeekTok);
 122   }
 123
 124   // If we don't have a pp-identifier now, this is an error.
 125   if (PP.CheckMacroName(PeekTok, MU_Other))
 126     return true;
 127
 128   // Otherwise, we got an identifier, is it defined to something?
 129   IdentifierInfo *II = PeekTok.getIdentifierInfo();
 130   MacroDefinition Macro = PP.getMacroDefinition(II);
 131   Result.Val = !!Macro;
 132   Result.Val.setIsUnsigned(false); // Result is signed intmax_t.
 133   DT.IncludedUndefinedIds = !Macro;
 134
 135   PP.emitMacroExpansionWarnings(
 136       PeekTok,
 137       (II->getName() == "INFINITY" || II->getName() == "NAN") ? true : false);
 138
 139   // If there is a macro, mark it used.
 140   if (Result.Val != 0 && ValueLive)
 141     PP.markMacroAsUsed(Macro.getMacroInfo());
 142
 143   // Save macro token for callback.
 144   Token macroToken(PeekTok);
 145
 146   // If we are in parens, ensure we have a trailing ).
 147   if (LParenLoc.isValid()) {
 148     // Consume identifier.
 149     Result.setEnd(PeekTok.getLocation());
 150     PP.LexUnexpandedNonComment(PeekTok);
 151
 152     if (PeekTok.isNot(tok::r_paren)) {
 153       PP.Diag(PeekTok.getLocation(), diag::err_pp_expected_after)
 154           << "'defined'" << tok::r_paren;
 155       PP.Diag(LParenLoc, diag::note_matching) << tok::l_paren;
 156       return true;
 157     }
 158     // Consume the ).
 159     PP.LexNonComment(PeekTok);
 160     Result.setEnd(PeekTok.getLocation());
 161   } else {
 162     // Consume identifier.
 163     Result.setEnd(PeekTok.getLocation());
 164     PP.LexNonComment(PeekTok);
 165   }
 166
 167   // [cpp.cond]p4:
 168   //   Prior to evaluation, macro invocations in the list of preprocessing
 169   //   tokens that will become the controlling constant expression are replaced
 170   //   (except for those macro names modified by the 'defined' unary operator),
 171   //   just as in normal text. If the token 'defined' is generated as a result
 172   //   of this replacement process or use of the 'defined' unary operator does
 173   //   not match one of the two specified forms prior to macro replacement, the
 174   //   behavior is undefined.
 175   // This isn't an idle threat, consider this program:
 176   //   #define FOO
 177   //   #define BAR defined(FOO)
 178   //   #if BAR
 179   //   ...
 180   //   #else
 181   //   ...
 182   //   #endif
 183   // clang and gcc will pick the #if branch while Visual Studio will take the
 184   // #else branch.  Emit a warning about this undefined behavior.
 185   if (beginLoc.isMacroID()) {
 186     bool IsFunctionTypeMacro =
 187         PP.getSourceManager()
 188             .getSLocEntry(PP.getSourceManager().getFileID(beginLoc))
 189             .getExpansion()
 190             .isFunctionMacroExpansion();
 191     // For object-type macros, it's easy to replace
 192     //   #define FOO defined(BAR)
 193     // with
 194     //   #if defined(BAR)
 195     //   #define FOO 1
 196     //   #else
 197     //   #define FOO 0
 198     //   #endif
 199     // and doing so makes sense since compilers handle this differently in
 200     // practice (see example further up).  But for function-type macros,
 201     // there is no good way to write
 202     //   # define FOO(x) (defined(M_ ## x) && M_ ## x)
 203     // in a different way, and compilers seem to agree on how to behave here.
 204     // So warn by default on object-type macros, but only warn in -pedantic
 205     // mode on function-type macros.
 206     if (IsFunctionTypeMacro)
 207       PP.Diag(beginLoc, diag::warn_defined_in_function_type_macro);
 208     else
 209       PP.Diag(beginLoc, diag::warn_defined_in_object_type_macro);
 210   }
 211
 212   // Invoke the 'defined' callback.
 213   if (PPCallbacks *Callbacks = PP.getPPCallbacks()) {
 214     Callbacks->Defined(macroToken, Macro,
 215                        SourceRange(beginLoc, PeekTok.getLocation()));
 216   }
 217
 218   // Success, remember that we saw defined(X).
 219   DT.State = DefinedTracker::DefinedMacro;
 220   DT.TheMacro = II;
 221   return false;
 222 }
 223
 224 /// EvaluateValue - Evaluate the token PeekTok (and any others needed) and
 225 /// return the computed value in Result.  Return true if there was an error
 226 /// parsing.  This function also returns information about the form of the
 227 /// expression in DT.  See above for information on what DT means.
 228 ///
 229 /// If ValueLive is false, then this value is being evaluated in a context where
 230 /// the result is not used.  As such, avoid diagnostics that relate to
 231 /// evaluation.
 232 static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
 233                           bool ValueLive, Preprocessor &PP) {
 234   DT.State = DefinedTracker::Unknown;
 235
 236   Result.setIdentifier(nullptr);
 237
 238   if (PeekTok.is(tok::code_completion)) {
 239     if (PP.getCodeCompletionHandler())
 240       PP.getCodeCompletionHandler()->CodeCompletePreprocessorExpression();
 241     PP.setCodeCompletionReached();
 242     PP.LexNonComment(PeekTok);
 243   }
 244
 245   switch (PeekTok.getKind()) {
 246   default:
 247     // If this token's spelling is a pp-identifier, check to see if it is
 248     // 'defined' or if it is a macro.  Note that we check here because many
 249     // keywords are pp-identifiers, so we can't check the kind.
 250     if (IdentifierInfo *II = PeekTok.getIdentifierInfo()) {
 251       // Handle "defined X" and "defined(X)".
 252       if (II->isStr("defined"))
 253         return EvaluateDefined(Result, PeekTok, DT, ValueLive, PP);
 254
 255       if (!II->isCPlusPlusOperatorKeyword()) {
 256         // If this identifier isn't 'defined' or one of the special
 257         // preprocessor keywords and it wasn't macro expanded, it turns
 258         // into a simple 0
 259         if (ValueLive) {
 260           PP.Diag(PeekTok, diag::warn_pp_undef_identifier) << II;
 261
 262           const DiagnosticsEngine &DiagEngine = PP.getDiagnostics();
 263           // If 'Wundef' is enabled, do not emit 'undef-prefix' diagnostics.
 264           if (DiagEngine.isIgnored(diag::warn_pp_undef_identifier,
 265                                    PeekTok.getLocation())) {
 266             const std::vector<std::string> UndefPrefixes =
 267                 DiagEngine.getDiagnosticOptions().UndefPrefixes;
 268             const StringRef IdentifierName = II->getName();
 269             if (llvm::any_of(UndefPrefixes,
 270                              [&IdentifierName](const std::string &Prefix) {
 271                                return IdentifierName.starts_with(Prefix);
 272                              }))
 273               PP.Diag(PeekTok, diag::warn_pp_undef_prefix)
 274                   << AddFlagValue{llvm::join(UndefPrefixes, ",")} << II;
 275           }
 276         }
 277         Result.Val = 0;
 278         Result.Val.setIsUnsigned(false); // "0" is signed intmax_t 0.
 279         Result.setIdentifier(II);
 280         Result.setRange(PeekTok.getLocation());
 281         DT.IncludedUndefinedIds = true;
 282         PP.LexNonComment(PeekTok);
 283         return false;
 284       }
 285     }
 286     PP.Diag(PeekTok, diag::err_pp_expr_bad_token_start_expr);
 287     return true;
 288   case tok::eod:
 289   case tok::r_paren:
 290     // If there is no expression, report and exit.
 291     PP.Diag(PeekTok, diag::err_pp_expected_value_in_expr);
 292     return true;
 293   case tok::numeric_constant: {
 294     SmallString<64> IntegerBuffer;
 295     bool NumberInvalid = false;
 296     StringRef Spelling = PP.getSpelling(PeekTok, IntegerBuffer,
 297                                               &NumberInvalid);
 298     if (NumberInvalid)
 299       return true; // a diagnostic was already reported
 300
 301     NumericLiteralParser Literal(Spelling, PeekTok.getLocation(),
 302                                  PP.getSourceManager(), PP.getLangOpts(),
 303                                  PP.getTargetInfo(), PP.getDiagnostics());
 304     if (Literal.hadError)
 305       return true; // a diagnostic was already reported.
 306
 307     if (Literal.isFloatingLiteral() || Literal.isImaginary) {
 308       PP.Diag(PeekTok, diag::err_pp_illegal_floating_literal);
 309       return true;
 310     }
 311     assert(Literal.isIntegerLiteral() && "Unknown ppnumber");
 312
 313     // Complain about, and drop, any ud-suffix.
 314     if (Literal.hasUDSuffix())
 315       PP.Diag(PeekTok, diag::err_pp_invalid_udl) << /*integer*/1;
 316
 317     // 'long long' is a C99 or C++11 feature.
 318     if (!PP.getLangOpts().C99 && Literal.isLongLong) {
 319       if (PP.getLangOpts().CPlusPlus)
 320         PP.Diag(PeekTok,
 321              PP.getLangOpts().CPlusPlus11 ?
 322              diag::warn_cxx98_compat_longlong : diag::ext_cxx11_longlong);
 323       else
 324         PP.Diag(PeekTok, diag::ext_c99_longlong);
 325     }
 326
 327     // 'z/uz' literals are a C++23 feature.
 328     if (Literal.isSizeT)
 329       PP.Diag(PeekTok, PP.getLangOpts().CPlusPlus
 330                            ? PP.getLangOpts().CPlusPlus23
 331                                  ? diag::warn_cxx20_compat_size_t_suffix
 332                                  : diag::ext_cxx23_size_t_suffix
 333                            : diag::err_cxx23_size_t_suffix);
 334
 335     // 'wb/uwb' literals are a C23 feature.
 336     // '__wb/__uwb' are a C++ extension.
 337     if (Literal.isBitInt)
 338       PP.Diag(PeekTok, PP.getLangOpts().CPlusPlus ? diag::ext_cxx_bitint_suffix
 339                        : PP.getLangOpts().C23
 340                            ? diag::warn_c23_compat_bitint_suffix
 341                            : diag::ext_c23_bitint_suffix);
 342
 343     // Parse the integer literal into Result.
 344     if (Literal.GetIntegerValue(Result.Val)) {
 345       // Overflow parsing integer literal.
 346       if (ValueLive)
 347         PP.Diag(PeekTok, diag::err_integer_literal_too_large)
 348             << /* Unsigned */ 1;
 349       Result.Val.setIsUnsigned(true);
 350     } else {
 351       // Set the signedness of the result to match whether there was a U suffix
 352       // or not.
 353       Result.Val.setIsUnsigned(Literal.isUnsigned);
 354
 355       // Detect overflow based on whether the value is signed.  If signed
 356       // and if the value is too large, emit a warning "integer constant is so
 357       // large that it is unsigned" e.g. on 12345678901234567890 where intmax_t
 358       // is 64-bits.
 359       if (!Literal.isUnsigned && Result.Val.isNegative()) {
 360         // Octal, hexadecimal, and binary literals are implicitly unsigned if
 361         // the value does not fit into a signed integer type.
 362         if (ValueLive && Literal.getRadix() == 10)
 363           PP.Diag(PeekTok, diag::ext_integer_literal_too_large_for_signed);
 364         Result.Val.setIsUnsigned(true);
 365       }
 366     }
 367
 368     // Consume the token.
 369     Result.setRange(PeekTok.getLocation());
 370     PP.LexNonComment(PeekTok);
 371     return false;
 372   }
 373   case tok::char_constant:          // 'x'
 374   case tok::wide_char_constant:     // L'x'
 375   case tok::utf8_char_constant:     // u8'x'
 376   case tok::utf16_char_constant:    // u'x'
 377   case tok::utf32_char_constant: {  // U'x'
 378     // Complain about, and drop, any ud-suffix.
 379     if (PeekTok.hasUDSuffix())
 380       PP.Diag(PeekTok, diag::err_pp_invalid_udl) << /*character*/0;
 381
 382     SmallString<32> CharBuffer;
 383     bool CharInvalid = false;
 384     StringRef ThisTok = PP.getSpelling(PeekTok, CharBuffer, &CharInvalid);
 385     if (CharInvalid)
 386       return true;
 387
 388     CharLiteralParser Literal(ThisTok.begin(), ThisTok.end(),
 389                               PeekTok.getLocation(), PP, PeekTok.getKind());
 390     if (Literal.hadError())
 391       return true;  // A diagnostic was already emitted.
 392
 393     // Character literals are always int or wchar_t, expand to intmax_t.
 394     const TargetInfo &TI = PP.getTargetInfo();
 395     unsigned NumBits;
 396     if (Literal.isMultiChar())
 397       NumBits = TI.getIntWidth();
 398     else if (Literal.isWide())
 399       NumBits = TI.getWCharWidth();
 400     else if (Literal.isUTF16())
 401       NumBits = TI.getChar16Width();
 402     else if (Literal.isUTF32())
 403       NumBits = TI.getChar32Width();
 404     else // char or char8_t
 405       NumBits = TI.getCharWidth();
 406
 407     // Set the width.
 408     llvm::APSInt Val(NumBits);
 409     // Set the value.
 410     Val = Literal.getValue();
 411     // Set the signedness. UTF-16 and UTF-32 are always unsigned
 412     // UTF-8 is unsigned if -fchar8_t is specified.
 413     if (Literal.isWide())
 414       Val.setIsUnsigned(!TargetInfo::isTypeSigned(TI.getWCharType()));
 415     else if (Literal.isUTF16() || Literal.isUTF32())
 416       Val.setIsUnsigned(true);
 417     else if (Literal.isUTF8()) {
 418       if (PP.getLangOpts().CPlusPlus)
 419         Val.setIsUnsigned(
 420             PP.getLangOpts().Char8 ? true : !PP.getLangOpts().CharIsSigned);
 421       else
 422         Val.setIsUnsigned(true);
 423     } else
 424       Val.setIsUnsigned(!PP.getLangOpts().CharIsSigned);
 425
 426     if (Result.Val.getBitWidth() > Val.getBitWidth()) {
 427       Result.Val = Val.extend(Result.Val.getBitWidth());
 428     } else {
 429       assert(Result.Val.getBitWidth() == Val.getBitWidth() &&
 430              "intmax_t smaller than char/wchar_t?");
 431       Result.Val = Val;
 432     }
 433
 434     // Consume the token.
 435     Result.setRange(PeekTok.getLocation());
 436     PP.LexNonComment(PeekTok);
 437     return false;
 438   }
 439   case tok::l_paren: {
 440     SourceLocation Start = PeekTok.getLocation();
 441     PP.LexNonComment(PeekTok);  // Eat the (.
 442     // Parse the value and if there are any binary operators involved, parse
 443     // them.
 444     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 445
 446     // If this is a silly value like (X), which doesn't need parens, check for
 447     // !(defined X).
 448     if (PeekTok.is(tok::r_paren)) {
 449       // Just use DT unmodified as our result.
 450     } else {
 451       // Otherwise, we have something like (x+y), and we consumed '(x'.
 452       if (EvaluateDirectiveSubExpr(Result, 1, PeekTok, ValueLive,
 453                                    DT.IncludedUndefinedIds, PP))
 454         return true;
 455
 456       if (PeekTok.isNot(tok::r_paren)) {
 457         PP.Diag(PeekTok.getLocation(), diag::err_pp_expected_rparen)
 458           << Result.getRange();
 459         PP.Diag(Start, diag::note_matching) << tok::l_paren;
 460         return true;
 461       }
 462       DT.State = DefinedTracker::Unknown;
 463     }
 464     Result.setRange(Start, PeekTok.getLocation());
 465     Result.setIdentifier(nullptr);
 466     PP.LexNonComment(PeekTok);  // Eat the ).
 467     return false;
 468   }
 469   case tok::plus: {
 470     SourceLocation Start = PeekTok.getLocation();
 471     // Unary plus doesn't modify the value.
 472     PP.LexNonComment(PeekTok);
 473     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 474     Result.setBegin(Start);
 475     Result.setIdentifier(nullptr);
 476     return false;
 477   }
 478   case tok::minus: {
 479     SourceLocation Loc = PeekTok.getLocation();
 480     PP.LexNonComment(PeekTok);
 481     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 482     Result.setBegin(Loc);
 483     Result.setIdentifier(nullptr);
 484
 485     // C99 6.5.3.3p3: The sign of the result matches the sign of the operand.
 486     Result.Val = -Result.Val;
 487
 488     // -MININT is the only thing that overflows.  Unsigned never overflows.
 489     bool Overflow = !Result.isUnsigned() && Result.Val.isMinSignedValue();
 490
 491     // If this operator is live and overflowed, report the issue.
 492     if (Overflow && ValueLive)
 493       PP.Diag(Loc, diag::warn_pp_expr_overflow) << Result.getRange();
 494
 495     DT.State = DefinedTracker::Unknown;
 496     return false;
 497   }
 498
 499   case tok::tilde: {
 500     SourceLocation Start = PeekTok.getLocation();
 501     PP.LexNonComment(PeekTok);
 502     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 503     Result.setBegin(Start);
 504     Result.setIdentifier(nullptr);
 505
 506     // C99 6.5.3.3p4: The sign of the result matches the sign of the operand.
 507     Result.Val = ~Result.Val;
 508     DT.State = DefinedTracker::Unknown;
 509     return false;
 510   }
 511
 512   case tok::exclaim: {
 513     SourceLocation Start = PeekTok.getLocation();
 514     PP.LexNonComment(PeekTok);
 515     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 516     Result.setBegin(Start);
 517     Result.Val = !Result.Val;
 518     // C99 6.5.3.3p5: The sign of the result is 'int', aka it is signed.
 519     Result.Val.setIsUnsigned(false);
 520     Result.setIdentifier(nullptr);
 521
 522     if (DT.State == DefinedTracker::DefinedMacro)
 523       DT.State = DefinedTracker::NotDefinedMacro;
 524     else if (DT.State == DefinedTracker::NotDefinedMacro)
 525       DT.State = DefinedTracker::DefinedMacro;
 526     return false;
 527   }
 528   case tok::kw_true:
 529   case tok::kw_false:
 530     Result.Val = PeekTok.getKind() == tok::kw_true;
 531     Result.Val.setIsUnsigned(false); // "0" is signed intmax_t 0.
 532     Result.setIdentifier(PeekTok.getIdentifierInfo());
 533     Result.setRange(PeekTok.getLocation());
 534     PP.LexNonComment(PeekTok);
 535     return false;
 536
 537   // FIXME: Handle #assert
 538   }
 539 }
 540
 541 /// getPrecedence - Return the precedence of the specified binary operator
 542 /// token.  This returns:
 543 ///   ~0 - Invalid token.
 544 ///   14 -> 3 - various operators.
 545 ///    0 - 'eod' or ')'
 546 static unsigned getPrecedence(tok::TokenKind Kind) {
 547   switch (Kind) {
 548   default: return ~0U;
 549   case tok::percent:
 550   case tok::slash:
 551   case tok::star:                 return 14;
 552   case tok::plus:
 553   case tok::minus:                return 13;
 554   case tok::lessless:
 555   case tok::greatergreater:       return 12;
 556   case tok::lessequal:
 557   case tok::less:
 558   case tok::greaterequal:
 559   case tok::greater:              return 11;
 560   case tok::exclaimequal:
 561   case tok::equalequal:           return 10;
 562   case tok::amp:                  return 9;
 563   case tok::caret:                return 8;
 564   case tok::pipe:                 return 7;
 565   case tok::ampamp:               return 6;
 566   case tok::pipepipe:             return 5;
 567   case tok::question:             return 4;
 568   case tok::comma:                return 3;
 569   case tok::colon:                return 2;
 570   case tok::r_paren:              return 0;// Lowest priority, end of expr.
 571   case tok::eod:                  return 0;// Lowest priority, end of directive.
 572   }
 573 }
 574
 575 static void diagnoseUnexpectedOperator(Preprocessor &PP, PPValue &LHS,
 576                                        Token &Tok) {
 577   if (Tok.is(tok::l_paren) && LHS.getIdentifier())
 578     PP.Diag(LHS.getRange().getBegin(), diag::err_pp_expr_bad_token_lparen)
 579         << LHS.getIdentifier();
 580   else
 581     PP.Diag(Tok.getLocation(), diag::err_pp_expr_bad_token_binop)
 582         << LHS.getRange();
 583 }
 584
 585 /// EvaluateDirectiveSubExpr - Evaluate the subexpression whose first token is
 586 /// PeekTok, and whose precedence is PeekPrec.  This returns the result in LHS.
 587 ///
 588 /// If ValueLive is false, then this value is being evaluated in a context where
 589 /// the result is not used.  As such, avoid diagnostics that relate to
 590 /// evaluation, such as division by zero warnings.
 591 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
 592                                      Token &PeekTok, bool ValueLive,
 593                                      bool &IncludedUndefinedIds,
 594                                      Preprocessor &PP) {
 595   unsigned PeekPrec = getPrecedence(PeekTok.getKind());
 596   // If this token isn't valid, report the error.
 597   if (PeekPrec == ~0U) {
 598     diagnoseUnexpectedOperator(PP, LHS, PeekTok);
 599     return true;
 600   }
 601
 602   while (true) {
 603     // If this token has a lower precedence than we are allowed to parse, return
 604     // it so that higher levels of the recursion can parse it.
 605     if (PeekPrec < MinPrec)
 606       return false;
 607
 608     tok::TokenKind Operator = PeekTok.getKind();
 609
 610     // If this is a short-circuiting operator, see if the RHS of the operator is
 611     // dead.  Note that this cannot just clobber ValueLive.  Consider
 612     // "0 && 1 ? 4 : 1 / 0", which is parsed as "(0 && 1) ? 4 : (1 / 0)".  In
 613     // this example, the RHS of the && being dead does not make the rest of the
 614     // expr dead.
 615     bool RHSIsLive;
 616     if (Operator == tok::ampamp && LHS.Val == 0)
 617       RHSIsLive = false;   // RHS of "0 && x" is dead.
 618     else if (Operator == tok::pipepipe && LHS.Val != 0)
 619       RHSIsLive = false;   // RHS of "1 || x" is dead.
 620     else if (Operator == tok::question && LHS.Val == 0)
 621       RHSIsLive = false;   // RHS (x) of "0 ? x : y" is dead.
 622     else
 623       RHSIsLive = ValueLive;
 624
 625     // Consume the operator, remembering the operator's location for reporting.
 626     SourceLocation OpLoc = PeekTok.getLocation();
 627     PP.LexNonComment(PeekTok);
 628
 629     PPValue RHS(LHS.getBitWidth());
 630     // Parse the RHS of the operator.
 631     DefinedTracker DT;
 632     if (EvaluateValue(RHS, PeekTok, DT, RHSIsLive, PP)) return true;
 633     IncludedUndefinedIds = DT.IncludedUndefinedIds;
 634
 635     // Remember the precedence of this operator and get the precedence of the
 636     // operator immediately to the right of the RHS.
 637     unsigned ThisPrec = PeekPrec;
 638     PeekPrec = getPrecedence(PeekTok.getKind());
 639
 640     // If this token isn't valid, report the error.
 641     if (PeekPrec == ~0U) {
 642       diagnoseUnexpectedOperator(PP, RHS, PeekTok);
 643       return true;
 644     }
 645
 646     // Decide whether to include the next binop in this subexpression.  For
 647     // example, when parsing x+y*z and looking at '*', we want to recursively
 648     // handle y*z as a single subexpression.  We do this because the precedence
 649     // of * is higher than that of +.  The only strange case we have to handle
 650     // here is for the ?: operator, where the precedence is actually lower than
 651     // the LHS of the '?'.  The grammar rule is:
 652     //
 653     // conditional-expression ::=
 654     //    logical-OR-expression ? expression : conditional-expression
 655     // where 'expression' is actually comma-expression.
 656     unsigned RHSPrec;
 657     if (Operator == tok::question)
 658       // The RHS of "?" should be maximally consumed as an expression.
 659       RHSPrec = getPrecedence(tok::comma);
 660     else  // All others should munch while higher precedence.
 661       RHSPrec = ThisPrec+1;
 662
 663     if (PeekPrec >= RHSPrec) {
 664       if (EvaluateDirectiveSubExpr(RHS, RHSPrec, PeekTok, RHSIsLive,
 665                                    IncludedUndefinedIds, PP))
 666         return true;
 667       PeekPrec = getPrecedence(PeekTok.getKind());
 668     }
 669     assert(PeekPrec <= ThisPrec && "Recursion didn't work!");
 670
 671     // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if
 672     // either operand is unsigned.
 673     llvm::APSInt Res(LHS.getBitWidth());
 674     switch (Operator) {
 675     case tok::question:       // No UAC for x and y in "x ? y : z".
 676     case tok::lessless:       // Shift amount doesn't UAC with shift value.
 677     case tok::greatergreater: // Shift amount doesn't UAC with shift value.
 678     case tok::comma:          // Comma operands are not subject to UACs.
 679     case tok::pipepipe:       // Logical || does not do UACs.
 680     case tok::ampamp:         // Logical && does not do UACs.
 681       break;                  // No UAC
 682     default:
 683       Res.setIsUnsigned(LHS.isUnsigned() || RHS.isUnsigned());
 684       // If this just promoted something from signed to unsigned, and if the
 685       // value was negative, warn about it.
 686       if (ValueLive && Res.isUnsigned()) {
 687         if (!LHS.isUnsigned() && LHS.Val.isNegative())
 688           PP.Diag(OpLoc, diag::warn_pp_convert_to_positive) << 0
 689             << toString(LHS.Val, 10, true) + " to " +
 690                toString(LHS.Val, 10, false)
 691             << LHS.getRange() << RHS.getRange();
 692         if (!RHS.isUnsigned() && RHS.Val.isNegative())
 693           PP.Diag(OpLoc, diag::warn_pp_convert_to_positive) << 1
 694             << toString(RHS.Val, 10, true) + " to " +
 695                toString(RHS.Val, 10, false)
 696             << LHS.getRange() << RHS.getRange();
 697       }
 698       LHS.Val.setIsUnsigned(Res.isUnsigned());
 699       RHS.Val.setIsUnsigned(Res.isUnsigned());
 700     }
 701
 702     bool Overflow = false;
 703     switch (Operator) {
 704     default: llvm_unreachable("Unknown operator token!");
 705     case tok::percent:
 706       if (RHS.Val != 0)
 707         Res = LHS.Val % RHS.Val;
 708       else if (ValueLive) {
 709         PP.Diag(OpLoc, diag::err_pp_remainder_by_zero)
 710           << LHS.getRange() << RHS.getRange();
 711         return true;
 712       }
 713       break;
 714     case tok::slash:
 715       if (RHS.Val != 0) {
 716         if (LHS.Val.isSigned())
 717           Res = llvm::APSInt(LHS.Val.sdiv_ov(RHS.Val, Overflow), false);
 718         else
 719           Res = LHS.Val / RHS.Val;
 720       } else if (ValueLive) {
 721         PP.Diag(OpLoc, diag::err_pp_division_by_zero)
 722           << LHS.getRange() << RHS.getRange();
 723         return true;
 724       }
 725       break;
 726
 727     case tok::star:
 728       if (Res.isSigned())
 729         Res = llvm::APSInt(LHS.Val.smul_ov(RHS.Val, Overflow), false);
 730       else
 731         Res = LHS.Val * RHS.Val;
 732       break;
 733     case tok::lessless: {
 734       // Determine whether overflow is about to happen.
 735       if (LHS.isUnsigned())
 736         Res = LHS.Val.ushl_ov(RHS.Val, Overflow);
 737       else
 738         Res = llvm::APSInt(LHS.Val.sshl_ov(RHS.Val, Overflow), false);
 739       break;
 740     }
 741     case tok::greatergreater: {
 742       // Determine whether overflow is about to happen.
 743       unsigned ShAmt = static_cast<unsigned>(RHS.Val.getLimitedValue());
 744       if (ShAmt >= LHS.getBitWidth()) {
 745         Overflow = true;
 746         ShAmt = LHS.getBitWidth()-1;
 747       }
 748       Res = LHS.Val >> ShAmt;
 749       break;
 750     }
 751     case tok::plus:
 752       if (LHS.isUnsigned())
 753         Res = LHS.Val + RHS.Val;
 754       else
 755         Res = llvm::APSInt(LHS.Val.sadd_ov(RHS.Val, Overflow), false);
 756       break;
 757     case tok::minus:
 758       if (LHS.isUnsigned())
 759         Res = LHS.Val - RHS.Val;
 760       else
 761         Res = llvm::APSInt(LHS.Val.ssub_ov(RHS.Val, Overflow), false);
 762       break;
 763     case tok::lessequal:
 764       Res = LHS.Val <= RHS.Val;
 765       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 766       break;
 767     case tok::less:
 768       Res = LHS.Val < RHS.Val;
 769       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 770       break;
 771     case tok::greaterequal:
 772       Res = LHS.Val >= RHS.Val;
 773       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 774       break;
 775     case tok::greater:
 776       Res = LHS.Val > RHS.Val;
 777       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 778       break;
 779     case tok::exclaimequal:
 780       Res = LHS.Val != RHS.Val;
 781       Res.setIsUnsigned(false);  // C99 6.5.9p3, result is always int (signed)
 782       break;
 783     case tok::equalequal:
 784       Res = LHS.Val == RHS.Val;
 785       Res.setIsUnsigned(false);  // C99 6.5.9p3, result is always int (signed)
 786       break;
 787     case tok::amp:
 788       Res = LHS.Val & RHS.Val;
 789       break;
 790     case tok::caret:
 791       Res = LHS.Val ^ RHS.Val;
 792       break;
 793     case tok::pipe:
 794       Res = LHS.Val | RHS.Val;
 795       break;
 796     case tok::ampamp:
 797       Res = (LHS.Val != 0 && RHS.Val != 0);
 798       Res.setIsUnsigned(false);  // C99 6.5.13p3, result is always int (signed)
 799       break;
 800     case tok::pipepipe:
 801       Res = (LHS.Val != 0 || RHS.Val != 0);
 802       Res.setIsUnsigned(false);  // C99 6.5.14p3, result is always int (signed)
 803       break;
 804     case tok::comma:
 805       // Comma is invalid in pp expressions in c89/c++ mode, but is valid in C99
 806       // if not being evaluated.
 807       if (!PP.getLangOpts().C99 || ValueLive)
 808         PP.Diag(OpLoc, diag::ext_pp_comma_expr)
 809           << LHS.getRange() << RHS.getRange();
 810       Res = RHS.Val; // LHS = LHS,RHS -> RHS.
 811       break;
 812     case tok::question: {
 813       // Parse the : part of the expression.
 814       if (PeekTok.isNot(tok::colon)) {
 815         PP.Diag(PeekTok.getLocation(), diag::err_expected)
 816             << tok::colon << LHS.getRange() << RHS.getRange();
 817         PP.Diag(OpLoc, diag::note_matching) << tok::question;
 818         return true;
 819       }
 820       // Consume the :.
 821       PP.LexNonComment(PeekTok);
 822
 823       // Evaluate the value after the :.
 824       bool AfterColonLive = ValueLive && LHS.Val == 0;
 825       PPValue AfterColonVal(LHS.getBitWidth());
 826       DefinedTracker DT;
 827       if (EvaluateValue(AfterColonVal, PeekTok, DT, AfterColonLive, PP))
 828         return true;
 829
 830       // Parse anything after the : with the same precedence as ?.  We allow
 831       // things of equal precedence because ?: is right associative.
 832       if (EvaluateDirectiveSubExpr(AfterColonVal, ThisPrec,
 833                                    PeekTok, AfterColonLive,
 834                                    IncludedUndefinedIds, PP))
 835         return true;
 836
 837       // Now that we have the condition, the LHS and the RHS of the :, evaluate.
 838       Res = LHS.Val != 0 ? RHS.Val : AfterColonVal.Val;
 839       RHS.setEnd(AfterColonVal.getRange().getEnd());
 840
 841       // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if
 842       // either operand is unsigned.
 843       Res.setIsUnsigned(RHS.isUnsigned() || AfterColonVal.isUnsigned());
 844
 845       // Figure out the precedence of the token after the : part.
 846       PeekPrec = getPrecedence(PeekTok.getKind());
 847       break;
 848     }
 849     case tok::colon:
 850       // Don't allow :'s to float around without being part of ?: exprs.
 851       PP.Diag(OpLoc, diag::err_pp_colon_without_question)
 852         << LHS.getRange() << RHS.getRange();
 853       return true;
 854     }
 855
 856     // If this operator is live and overflowed, report the issue.
 857     if (Overflow && ValueLive)
 858       PP.Diag(OpLoc, diag::warn_pp_expr_overflow)
 859         << LHS.getRange() << RHS.getRange();
 860
 861     // Put the result back into 'LHS' for our next iteration.
 862     LHS.Val = Res;
 863     LHS.setEnd(RHS.getRange().getEnd());
 864     RHS.setIdentifier(nullptr);
 865   }
 866 }
 867
 868 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that
 869 /// may occur after a #if or #elif directive.  If the expression is equivalent
 870 /// to "!defined(X)" return X in IfNDefMacro.
 871 Preprocessor::DirectiveEvalResult
 872 Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro,
 873                                           Token &Tok, bool &EvaluatedDefined,
 874                                           bool CheckForEoD) {
 875   SaveAndRestore PPDir(ParsingIfOrElifDirective, true);
 876   // Save the current state of 'DisableMacroExpansion' and reset it to false. If
 877   // 'DisableMacroExpansion' is true, then we must be in a macro argument list
 878   // in which case a directive is undefined behavior.  We want macros to be able
 879   // to recursively expand in order to get more gcc-list behavior, so we force
 880   // DisableMacroExpansion to false and restore it when we're done parsing the
 881   // expression.
 882   bool DisableMacroExpansionAtStartOfDirective = DisableMacroExpansion;
 883   DisableMacroExpansion = false;
 884
 885   // Peek ahead one token.
 886   LexNonComment(Tok);
 887
 888   // C99 6.10.1p3 - All expressions are evaluated as intmax_t or uintmax_t.
 889   unsigned BitWidth = getTargetInfo().getIntMaxTWidth();
 890
 891   PPValue ResVal(BitWidth);
 892   DefinedTracker DT;
 893   SourceLocation ExprStartLoc = SourceMgr.getExpansionLoc(Tok.getLocation());
 894   if (EvaluateValue(ResVal, Tok, DT, true, *this)) {
 895     // Parse error, skip the rest of the macro line.
 896     SourceRange ConditionRange = ExprStartLoc;
 897     if (Tok.isNot(tok::eod))
 898       ConditionRange = DiscardUntilEndOfDirective(Tok);
 899
 900     // Restore 'DisableMacroExpansion'.
 901     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 902
 903     // We cannot trust the source range from the value because there was a
 904     // parse error. Track the range manually -- the end of the directive is the
 905     // end of the condition range.
 906     return {std::nullopt,
 907             false,
 908             DT.IncludedUndefinedIds,
 909             {ExprStartLoc, ConditionRange.getEnd()}};
 910   }
 911
 912   EvaluatedDefined = DT.State != DefinedTracker::Unknown;
 913
 914   // If we are at the end of the expression after just parsing a value, there
 915   // must be no (unparenthesized) binary operators involved, so we can exit
 916   // directly.
 917   if (Tok.is(tok::eod)) {
 918     // If the expression we parsed was of the form !defined(macro), return the
 919     // macro in IfNDefMacro.
 920     if (DT.State == DefinedTracker::NotDefinedMacro)
 921       IfNDefMacro = DT.TheMacro;
 922
 923     // Restore 'DisableMacroExpansion'.
 924     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 925     bool IsNonZero = ResVal.Val != 0;
 926     SourceRange ValRange = ResVal.getRange();
 927     return {std::move(ResVal.Val), IsNonZero, DT.IncludedUndefinedIds,
 928             ValRange};
 929   }
 930
 931   // Otherwise, we must have a binary operator (e.g. "#if 1 < 2"), so parse the
 932   // operator and the stuff after it.
 933   if (EvaluateDirectiveSubExpr(ResVal, getPrecedence(tok::question),
 934                                Tok, true, DT.IncludedUndefinedIds, *this)) {
 935     // Parse error, skip the rest of the macro line.
 936     if (Tok.isNot(tok::eod))
 937       DiscardUntilEndOfDirective(Tok);
 938
 939     // Restore 'DisableMacroExpansion'.
 940     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 941     SourceRange ValRange = ResVal.getRange();
 942     return {std::nullopt, false, DT.IncludedUndefinedIds, ValRange};
 943   }
 944
 945   if (CheckForEoD) {
 946     // If we aren't at the tok::eod token, something bad happened, like an extra
 947     // ')' token.
 948     if (Tok.isNot(tok::eod)) {
 949       Diag(Tok, diag::err_pp_expected_eol);
 950       DiscardUntilEndOfDirective(Tok);
 951     }
 952   }
 953
 954   EvaluatedDefined = EvaluatedDefined || DT.State != DefinedTracker::Unknown;
 955
 956   // Restore 'DisableMacroExpansion'.
 957   DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 958   bool IsNonZero = ResVal.Val != 0;
 959   SourceRange ValRange = ResVal.getRange();
 960   return {std::move(ResVal.Val), IsNonZero, DT.IncludedUndefinedIds, ValRange};
 961 }
 962
 963 Preprocessor::DirectiveEvalResult
 964 Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro,
 965                                           bool CheckForEoD) {
 966   Token Tok;
 967   bool EvaluatedDefined;
 968   return EvaluateDirectiveExpression(IfNDefMacro, Tok, EvaluatedDefined,
 969                                      CheckForEoD);
 970 }