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/SmallString.h"
  33 #include "llvm/ADT/StringExtras.h"
  34 #include "llvm/ADT/StringRef.h"
  35 #include "llvm/Support/ErrorHandling.h"
  36 #include "llvm/Support/SaveAndRestore.h"
  37 #include <cassert>
  38
  39 using namespace clang;
  40
  41 namespace {
  42
  43 /// PPValue - Represents the value of a subexpression of a preprocessor
  44 /// conditional and the source range covered by it.
  45 class PPValue {
  46   SourceRange Range;
  47   IdentifierInfo *II;
  48
  49 public:
  50   llvm::APSInt Val;
  51
  52   // Default ctor - Construct an 'invalid' PPValue.
  53   PPValue(unsigned BitWidth) : Val(BitWidth) {}
  54
  55   // If this value was produced by directly evaluating an identifier, produce
  56   // that identifier.
  57   IdentifierInfo *getIdentifier() const { return II; }
  58   void setIdentifier(IdentifierInfo *II) { this->II = II; }
  59
  60   unsigned getBitWidth() const { return Val.getBitWidth(); }
  61   bool isUnsigned() const { return Val.isUnsigned(); }
  62
  63   SourceRange getRange() const { return Range; }
  64
  65   void setRange(SourceLocation L) { Range.setBegin(L); Range.setEnd(L); }
  66   void setRange(SourceLocation B, SourceLocation E) {
  67     Range.setBegin(B); Range.setEnd(E);
  68   }
  69   void setBegin(SourceLocation L) { Range.setBegin(L); }
  70   void setEnd(SourceLocation L) { Range.setEnd(L); }
  71 };
  72
  73 } // end anonymous namespace
  74
  75 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
  76                                      Token &PeekTok, bool ValueLive,
  77                                      bool &IncludedUndefinedIds,
  78                                      Preprocessor &PP);
  79
  80 /// DefinedTracker - This struct is used while parsing expressions to keep track
  81 /// of whether !defined(X) has been seen.
  82 ///
  83 /// With this simple scheme, we handle the basic forms:
  84 ///    !defined(X)   and !defined X
  85 /// but we also trivially handle (silly) stuff like:
  86 ///    !!!defined(X) and +!defined(X) and !+!+!defined(X) and !(defined(X)).
  87 struct DefinedTracker {
  88   /// Each time a Value is evaluated, it returns information about whether the
  89   /// parsed value is of the form defined(X), !defined(X) or is something else.
  90   enum TrackerState {
  91     DefinedMacro,        // defined(X)
  92     NotDefinedMacro,     // !defined(X)
  93     Unknown              // Something else.
  94   } State;
  95   /// TheMacro - When the state is DefinedMacro or NotDefinedMacro, this
  96   /// indicates the macro that was checked.
  97   IdentifierInfo *TheMacro;
  98   bool IncludedUndefinedIds = false;
  99 };
 100
 101 /// EvaluateDefined - Process a 'defined(sym)' expression.
 102 static bool EvaluateDefined(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
 103                             bool ValueLive, Preprocessor &PP) {
 104   SourceLocation beginLoc(PeekTok.getLocation());
 105   Result.setBegin(beginLoc);
 106
 107   // Get the next token, don't expand it.
 108   PP.LexUnexpandedNonComment(PeekTok);
 109
 110   // Two options, it can either be a pp-identifier or a (.
 111   SourceLocation LParenLoc;
 112   if (PeekTok.is(tok::l_paren)) {
 113     // Found a paren, remember we saw it and skip it.
 114     LParenLoc = PeekTok.getLocation();
 115     PP.LexUnexpandedNonComment(PeekTok);
 116   }
 117
 118   if (PeekTok.is(tok::code_completion)) {
 119     if (PP.getCodeCompletionHandler())
 120       PP.getCodeCompletionHandler()->CodeCompleteMacroName(false);
 121     PP.setCodeCompletionReached();
 122     PP.LexUnexpandedNonComment(PeekTok);
 123   }
 124
 125   // If we don't have a pp-identifier now, this is an error.
 126   if (PP.CheckMacroName(PeekTok, MU_Other))
 127     return true;
 128
 129   // Otherwise, we got an identifier, is it defined to something?
 130   IdentifierInfo *II = PeekTok.getIdentifierInfo();
 131   MacroDefinition Macro = PP.getMacroDefinition(II);
 132   Result.Val = !!Macro;
 133   Result.Val.setIsUnsigned(false); // Result is signed intmax_t.
 134   DT.IncludedUndefinedIds = !Macro;
 135
 136   PP.emitMacroExpansionWarnings(PeekTok);
 137
 138   // If there is a macro, mark it used.
 139   if (Result.Val != 0 && ValueLive)
 140     PP.markMacroAsUsed(Macro.getMacroInfo());
 141
 142   // Save macro token for callback.
 143   Token macroToken(PeekTok);
 144
 145   // If we are in parens, ensure we have a trailing ).
 146   if (LParenLoc.isValid()) {
 147     // Consume identifier.
 148     Result.setEnd(PeekTok.getLocation());
 149     PP.LexUnexpandedNonComment(PeekTok);
 150
 151     if (PeekTok.isNot(tok::r_paren)) {
 152       PP.Diag(PeekTok.getLocation(), diag::err_pp_expected_after)
 153           << "'defined'" << tok::r_paren;
 154       PP.Diag(LParenLoc, diag::note_matching) << tok::l_paren;
 155       return true;
 156     }
 157     // Consume the ).
 158     PP.LexNonComment(PeekTok);
 159     Result.setEnd(PeekTok.getLocation());
 160   } else {
 161     // Consume identifier.
 162     Result.setEnd(PeekTok.getLocation());
 163     PP.LexNonComment(PeekTok);
 164   }
 165
 166   // [cpp.cond]p4:
 167   //   Prior to evaluation, macro invocations in the list of preprocessing
 168   //   tokens that will become the controlling constant expression are replaced
 169   //   (except for those macro names modified by the 'defined' unary operator),
 170   //   just as in normal text. If the token 'defined' is generated as a result
 171   //   of this replacement process or use of the 'defined' unary operator does
 172   //   not match one of the two specified forms prior to macro replacement, the
 173   //   behavior is undefined.
 174   // This isn't an idle threat, consider this program:
 175   //   #define FOO
 176   //   #define BAR defined(FOO)
 177   //   #if BAR
 178   //   ...
 179   //   #else
 180   //   ...
 181   //   #endif
 182   // clang and gcc will pick the #if branch while Visual Studio will take the
 183   // #else branch.  Emit a warning about this undefined behavior.
 184   if (beginLoc.isMacroID()) {
 185     bool IsFunctionTypeMacro =
 186         PP.getSourceManager()
 187             .getSLocEntry(PP.getSourceManager().getFileID(beginLoc))
 188             .getExpansion()
 189             .isFunctionMacroExpansion();
 190     // For object-type macros, it's easy to replace
 191     //   #define FOO defined(BAR)
 192     // with
 193     //   #if defined(BAR)
 194     //   #define FOO 1
 195     //   #else
 196     //   #define FOO 0
 197     //   #endif
 198     // and doing so makes sense since compilers handle this differently in
 199     // practice (see example further up).  But for function-type macros,
 200     // there is no good way to write
 201     //   # define FOO(x) (defined(M_ ## x) && M_ ## x)
 202     // in a different way, and compilers seem to agree on how to behave here.
 203     // So warn by default on object-type macros, but only warn in -pedantic
 204     // mode on function-type macros.
 205     if (IsFunctionTypeMacro)
 206       PP.Diag(beginLoc, diag::warn_defined_in_function_type_macro);
 207     else
 208       PP.Diag(beginLoc, diag::warn_defined_in_object_type_macro);
 209   }
 210
 211   // Invoke the 'defined' callback.
 212   if (PPCallbacks *Callbacks = PP.getPPCallbacks()) {
 213     Callbacks->Defined(macroToken, Macro,
 214                        SourceRange(beginLoc, PeekTok.getLocation()));
 215   }
 216
 217   // Success, remember that we saw defined(X).
 218   DT.State = DefinedTracker::DefinedMacro;
 219   DT.TheMacro = II;
 220   return false;
 221 }
 222
 223 /// EvaluateValue - Evaluate the token PeekTok (and any others needed) and
 224 /// return the computed value in Result.  Return true if there was an error
 225 /// parsing.  This function also returns information about the form of the
 226 /// expression in DT.  See above for information on what DT means.
 227 ///
 228 /// If ValueLive is false, then this value is being evaluated in a context where
 229 /// the result is not used.  As such, avoid diagnostics that relate to
 230 /// evaluation.
 231 static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
 232                           bool ValueLive, Preprocessor &PP) {
 233   DT.State = DefinedTracker::Unknown;
 234
 235   Result.setIdentifier(nullptr);
 236
 237   if (PeekTok.is(tok::code_completion)) {
 238     if (PP.getCodeCompletionHandler())
 239       PP.getCodeCompletionHandler()->CodeCompletePreprocessorExpression();
 240     PP.setCodeCompletionReached();
 241     PP.LexNonComment(PeekTok);
 242   }
 243
 244   switch (PeekTok.getKind()) {
 245   default:
 246     // If this token's spelling is a pp-identifier, check to see if it is
 247     // 'defined' or if it is a macro.  Note that we check here because many
 248     // keywords are pp-identifiers, so we can't check the kind.
 249     if (IdentifierInfo *II = PeekTok.getIdentifierInfo()) {
 250       // Handle "defined X" and "defined(X)".
 251       if (II->isStr("defined"))
 252         return EvaluateDefined(Result, PeekTok, DT, ValueLive, PP);
 253
 254       if (!II->isCPlusPlusOperatorKeyword()) {
 255         // If this identifier isn't 'defined' or one of the special
 256         // preprocessor keywords and it wasn't macro expanded, it turns
 257         // into a simple 0
 258         if (ValueLive) {
 259           PP.Diag(PeekTok, diag::warn_pp_undef_identifier) << II;
 260
 261           const DiagnosticsEngine &DiagEngine = PP.getDiagnostics();
 262           // If 'Wundef' is enabled, do not emit 'undef-prefix' diagnostics.
 263           if (DiagEngine.isIgnored(diag::warn_pp_undef_identifier,
 264                                    PeekTok.getLocation())) {
 265             const std::vector<std::string> UndefPrefixes =
 266                 DiagEngine.getDiagnosticOptions().UndefPrefixes;
 267             const StringRef IdentifierName = II->getName();
 268             if (llvm::any_of(UndefPrefixes,
 269                              [&IdentifierName](const std::string &Prefix) {
 270                                return IdentifierName.startswith(Prefix);
 271                              }))
 272               PP.Diag(PeekTok, diag::warn_pp_undef_prefix)
 273                   << AddFlagValue{llvm::join(UndefPrefixes, ",")} << II;
 274           }
 275         }
 276         Result.Val = 0;
 277         Result.Val.setIsUnsigned(false); // "0" is signed intmax_t 0.
 278         Result.setIdentifier(II);
 279         Result.setRange(PeekTok.getLocation());
 280         DT.IncludedUndefinedIds = true;
 281         PP.LexNonComment(PeekTok);
 282         return false;
 283       }
 284     }
 285     PP.Diag(PeekTok, diag::err_pp_expr_bad_token_start_expr);
 286     return true;
 287   case tok::eod:
 288   case tok::r_paren:
 289     // If there is no expression, report and exit.
 290     PP.Diag(PeekTok, diag::err_pp_expected_value_in_expr);
 291     return true;
 292   case tok::numeric_constant: {
 293     SmallString<64> IntegerBuffer;
 294     bool NumberInvalid = false;
 295     StringRef Spelling = PP.getSpelling(PeekTok, IntegerBuffer,
 296                                               &NumberInvalid);
 297     if (NumberInvalid)
 298       return true; // a diagnostic was already reported
 299
 300     NumericLiteralParser Literal(Spelling, PeekTok.getLocation(),
 301                                  PP.getSourceManager(), PP.getLangOpts(),
 302                                  PP.getTargetInfo(), PP.getDiagnostics());
 303     if (Literal.hadError)
 304       return true; // a diagnostic was already reported.
 305
 306     if (Literal.isFloatingLiteral() || Literal.isImaginary) {
 307       PP.Diag(PeekTok, diag::err_pp_illegal_floating_literal);
 308       return true;
 309     }
 310     assert(Literal.isIntegerLiteral() && "Unknown ppnumber");
 311
 312     // Complain about, and drop, any ud-suffix.
 313     if (Literal.hasUDSuffix())
 314       PP.Diag(PeekTok, diag::err_pp_invalid_udl) << /*integer*/1;
 315
 316     // 'long long' is a C99 or C++11 feature.
 317     if (!PP.getLangOpts().C99 && Literal.isLongLong) {
 318       if (PP.getLangOpts().CPlusPlus)
 319         PP.Diag(PeekTok,
 320              PP.getLangOpts().CPlusPlus11 ?
 321              diag::warn_cxx98_compat_longlong : diag::ext_cxx11_longlong);
 322       else
 323         PP.Diag(PeekTok, diag::ext_c99_longlong);
 324     }
 325
 326     // 'z/uz' literals are a C++2b feature.
 327     if (Literal.isSizeT)
 328       PP.Diag(PeekTok, PP.getLangOpts().CPlusPlus
 329                            ? PP.getLangOpts().CPlusPlus2b
 330                                  ? diag::warn_cxx20_compat_size_t_suffix
 331                                  : diag::ext_cxx2b_size_t_suffix
 332                            : diag::err_cxx2b_size_t_suffix);
 333
 334     // 'wb/uwb' literals are a C2x feature. We explicitly do not support the
 335     // suffix in C++ as an extension because a library-based UDL that resolves
 336     // to a library type may be more appropriate there.
 337     if (Literal.isBitInt)
 338       PP.Diag(PeekTok, PP.getLangOpts().C2x
 339                            ? diag::warn_c2x_compat_bitint_suffix
 340                            : diag::ext_c2x_bitint_suffix);
 341
 342     // Parse the integer literal into Result.
 343     if (Literal.GetIntegerValue(Result.Val)) {
 344       // Overflow parsing integer literal.
 345       if (ValueLive)
 346         PP.Diag(PeekTok, diag::err_integer_literal_too_large)
 347             << /* Unsigned */ 1;
 348       Result.Val.setIsUnsigned(true);
 349     } else {
 350       // Set the signedness of the result to match whether there was a U suffix
 351       // or not.
 352       Result.Val.setIsUnsigned(Literal.isUnsigned);
 353
 354       // Detect overflow based on whether the value is signed.  If signed
 355       // and if the value is too large, emit a warning "integer constant is so
 356       // large that it is unsigned" e.g. on 12345678901234567890 where intmax_t
 357       // is 64-bits.
 358       if (!Literal.isUnsigned && Result.Val.isNegative()) {
 359         // Octal, hexadecimal, and binary literals are implicitly unsigned if
 360         // the value does not fit into a signed integer type.
 361         if (ValueLive && Literal.getRadix() == 10)
 362           PP.Diag(PeekTok, diag::ext_integer_literal_too_large_for_signed);
 363         Result.Val.setIsUnsigned(true);
 364       }
 365     }
 366
 367     // Consume the token.
 368     Result.setRange(PeekTok.getLocation());
 369     PP.LexNonComment(PeekTok);
 370     return false;
 371   }
 372   case tok::char_constant:          // 'x'
 373   case tok::wide_char_constant:     // L'x'
 374   case tok::utf8_char_constant:     // u8'x'
 375   case tok::utf16_char_constant:    // u'x'
 376   case tok::utf32_char_constant: {  // U'x'
 377     // Complain about, and drop, any ud-suffix.
 378     if (PeekTok.hasUDSuffix())
 379       PP.Diag(PeekTok, diag::err_pp_invalid_udl) << /*character*/0;
 380
 381     SmallString<32> CharBuffer;
 382     bool CharInvalid = false;
 383     StringRef ThisTok = PP.getSpelling(PeekTok, CharBuffer, &CharInvalid);
 384     if (CharInvalid)
 385       return true;
 386
 387     CharLiteralParser Literal(ThisTok.begin(), ThisTok.end(),
 388                               PeekTok.getLocation(), PP, PeekTok.getKind());
 389     if (Literal.hadError())
 390       return true;  // A diagnostic was already emitted.
 391
 392     // Character literals are always int or wchar_t, expand to intmax_t.
 393     const TargetInfo &TI = PP.getTargetInfo();
 394     unsigned NumBits;
 395     if (Literal.isMultiChar())
 396       NumBits = TI.getIntWidth();
 397     else if (Literal.isWide())
 398       NumBits = TI.getWCharWidth();
 399     else if (Literal.isUTF16())
 400       NumBits = TI.getChar16Width();
 401     else if (Literal.isUTF32())
 402       NumBits = TI.getChar32Width();
 403     else // char or char8_t
 404       NumBits = TI.getCharWidth();
 405
 406     // Set the width.
 407     llvm::APSInt Val(NumBits);
 408     // Set the value.
 409     Val = Literal.getValue();
 410     // Set the signedness. UTF-16 and UTF-32 are always unsigned
 411     // UTF-8 is unsigned if -fchar8_t is specified.
 412     if (Literal.isWide())
 413       Val.setIsUnsigned(!TargetInfo::isTypeSigned(TI.getWCharType()));
 414     else if (Literal.isUTF16() || Literal.isUTF32())
 415       Val.setIsUnsigned(true);
 416     else if (Literal.isUTF8()) {
 417       if (PP.getLangOpts().CPlusPlus)
 418         Val.setIsUnsigned(
 419             PP.getLangOpts().Char8 ? true : !PP.getLangOpts().CharIsSigned);
 420       else
 421         Val.setIsUnsigned(true);
 422     } else
 423       Val.setIsUnsigned(!PP.getLangOpts().CharIsSigned);
 424
 425     if (Result.Val.getBitWidth() > Val.getBitWidth()) {
 426       Result.Val = Val.extend(Result.Val.getBitWidth());
 427     } else {
 428       assert(Result.Val.getBitWidth() == Val.getBitWidth() &&
 429              "intmax_t smaller than char/wchar_t?");
 430       Result.Val = Val;
 431     }
 432
 433     // Consume the token.
 434     Result.setRange(PeekTok.getLocation());
 435     PP.LexNonComment(PeekTok);
 436     return false;
 437   }
 438   case tok::l_paren: {
 439     SourceLocation Start = PeekTok.getLocation();
 440     PP.LexNonComment(PeekTok);  // Eat the (.
 441     // Parse the value and if there are any binary operators involved, parse
 442     // them.
 443     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 444
 445     // If this is a silly value like (X), which doesn't need parens, check for
 446     // !(defined X).
 447     if (PeekTok.is(tok::r_paren)) {
 448       // Just use DT unmodified as our result.
 449     } else {
 450       // Otherwise, we have something like (x+y), and we consumed '(x'.
 451       if (EvaluateDirectiveSubExpr(Result, 1, PeekTok, ValueLive,
 452                                    DT.IncludedUndefinedIds, PP))
 453         return true;
 454
 455       if (PeekTok.isNot(tok::r_paren)) {
 456         PP.Diag(PeekTok.getLocation(), diag::err_pp_expected_rparen)
 457           << Result.getRange();
 458         PP.Diag(Start, diag::note_matching) << tok::l_paren;
 459         return true;
 460       }
 461       DT.State = DefinedTracker::Unknown;
 462     }
 463     Result.setRange(Start, PeekTok.getLocation());
 464     Result.setIdentifier(nullptr);
 465     PP.LexNonComment(PeekTok);  // Eat the ).
 466     return false;
 467   }
 468   case tok::plus: {
 469     SourceLocation Start = PeekTok.getLocation();
 470     // Unary plus doesn't modify the value.
 471     PP.LexNonComment(PeekTok);
 472     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 473     Result.setBegin(Start);
 474     Result.setIdentifier(nullptr);
 475     return false;
 476   }
 477   case tok::minus: {
 478     SourceLocation Loc = PeekTok.getLocation();
 479     PP.LexNonComment(PeekTok);
 480     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 481     Result.setBegin(Loc);
 482     Result.setIdentifier(nullptr);
 483
 484     // C99 6.5.3.3p3: The sign of the result matches the sign of the operand.
 485     Result.Val = -Result.Val;
 486
 487     // -MININT is the only thing that overflows.  Unsigned never overflows.
 488     bool Overflow = !Result.isUnsigned() && Result.Val.isMinSignedValue();
 489
 490     // If this operator is live and overflowed, report the issue.
 491     if (Overflow && ValueLive)
 492       PP.Diag(Loc, diag::warn_pp_expr_overflow) << Result.getRange();
 493
 494     DT.State = DefinedTracker::Unknown;
 495     return false;
 496   }
 497
 498   case tok::tilde: {
 499     SourceLocation Start = PeekTok.getLocation();
 500     PP.LexNonComment(PeekTok);
 501     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 502     Result.setBegin(Start);
 503     Result.setIdentifier(nullptr);
 504
 505     // C99 6.5.3.3p4: The sign of the result matches the sign of the operand.
 506     Result.Val = ~Result.Val;
 507     DT.State = DefinedTracker::Unknown;
 508     return false;
 509   }
 510
 511   case tok::exclaim: {
 512     SourceLocation Start = PeekTok.getLocation();
 513     PP.LexNonComment(PeekTok);
 514     if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true;
 515     Result.setBegin(Start);
 516     Result.Val = !Result.Val;
 517     // C99 6.5.3.3p5: The sign of the result is 'int', aka it is signed.
 518     Result.Val.setIsUnsigned(false);
 519     Result.setIdentifier(nullptr);
 520
 521     if (DT.State == DefinedTracker::DefinedMacro)
 522       DT.State = DefinedTracker::NotDefinedMacro;
 523     else if (DT.State == DefinedTracker::NotDefinedMacro)
 524       DT.State = DefinedTracker::DefinedMacro;
 525     return false;
 526   }
 527   case tok::kw_true:
 528   case tok::kw_false:
 529     Result.Val = PeekTok.getKind() == tok::kw_true;
 530     Result.Val.setIsUnsigned(false); // "0" is signed intmax_t 0.
 531     Result.setIdentifier(PeekTok.getIdentifierInfo());
 532     Result.setRange(PeekTok.getLocation());
 533     PP.LexNonComment(PeekTok);
 534     return false;
 535
 536   // FIXME: Handle #assert
 537   }
 538 }
 539
 540 /// getPrecedence - Return the precedence of the specified binary operator
 541 /// token.  This returns:
 542 ///   ~0 - Invalid token.
 543 ///   14 -> 3 - various operators.
 544 ///    0 - 'eod' or ')'
 545 static unsigned getPrecedence(tok::TokenKind Kind) {
 546   switch (Kind) {
 547   default: return ~0U;
 548   case tok::percent:
 549   case tok::slash:
 550   case tok::star:                 return 14;
 551   case tok::plus:
 552   case tok::minus:                return 13;
 553   case tok::lessless:
 554   case tok::greatergreater:       return 12;
 555   case tok::lessequal:
 556   case tok::less:
 557   case tok::greaterequal:
 558   case tok::greater:              return 11;
 559   case tok::exclaimequal:
 560   case tok::equalequal:           return 10;
 561   case tok::amp:                  return 9;
 562   case tok::caret:                return 8;
 563   case tok::pipe:                 return 7;
 564   case tok::ampamp:               return 6;
 565   case tok::pipepipe:             return 5;
 566   case tok::question:             return 4;
 567   case tok::comma:                return 3;
 568   case tok::colon:                return 2;
 569   case tok::r_paren:              return 0;// Lowest priority, end of expr.
 570   case tok::eod:                  return 0;// Lowest priority, end of directive.
 571   }
 572 }
 573
 574 static void diagnoseUnexpectedOperator(Preprocessor &PP, PPValue &LHS,
 575                                        Token &Tok) {
 576   if (Tok.is(tok::l_paren) && LHS.getIdentifier())
 577     PP.Diag(LHS.getRange().getBegin(), diag::err_pp_expr_bad_token_lparen)
 578         << LHS.getIdentifier();
 579   else
 580     PP.Diag(Tok.getLocation(), diag::err_pp_expr_bad_token_binop)
 581         << LHS.getRange();
 582 }
 583
 584 /// EvaluateDirectiveSubExpr - Evaluate the subexpression whose first token is
 585 /// PeekTok, and whose precedence is PeekPrec.  This returns the result in LHS.
 586 ///
 587 /// If ValueLive is false, then this value is being evaluated in a context where
 588 /// the result is not used.  As such, avoid diagnostics that relate to
 589 /// evaluation, such as division by zero warnings.
 590 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
 591                                      Token &PeekTok, bool ValueLive,
 592                                      bool &IncludedUndefinedIds,
 593                                      Preprocessor &PP) {
 594   unsigned PeekPrec = getPrecedence(PeekTok.getKind());
 595   // If this token isn't valid, report the error.
 596   if (PeekPrec == ~0U) {
 597     diagnoseUnexpectedOperator(PP, LHS, PeekTok);
 598     return true;
 599   }
 600
 601   while (true) {
 602     // If this token has a lower precedence than we are allowed to parse, return
 603     // it so that higher levels of the recursion can parse it.
 604     if (PeekPrec < MinPrec)
 605       return false;
 606
 607     tok::TokenKind Operator = PeekTok.getKind();
 608
 609     // If this is a short-circuiting operator, see if the RHS of the operator is
 610     // dead.  Note that this cannot just clobber ValueLive.  Consider
 611     // "0 && 1 ? 4 : 1 / 0", which is parsed as "(0 && 1) ? 4 : (1 / 0)".  In
 612     // this example, the RHS of the && being dead does not make the rest of the
 613     // expr dead.
 614     bool RHSIsLive;
 615     if (Operator == tok::ampamp && LHS.Val == 0)
 616       RHSIsLive = false;   // RHS of "0 && x" is dead.
 617     else if (Operator == tok::pipepipe && LHS.Val != 0)
 618       RHSIsLive = false;   // RHS of "1 || x" is dead.
 619     else if (Operator == tok::question && LHS.Val == 0)
 620       RHSIsLive = false;   // RHS (x) of "0 ? x : y" is dead.
 621     else
 622       RHSIsLive = ValueLive;
 623
 624     // Consume the operator, remembering the operator's location for reporting.
 625     SourceLocation OpLoc = PeekTok.getLocation();
 626     PP.LexNonComment(PeekTok);
 627
 628     PPValue RHS(LHS.getBitWidth());
 629     // Parse the RHS of the operator.
 630     DefinedTracker DT;
 631     if (EvaluateValue(RHS, PeekTok, DT, RHSIsLive, PP)) return true;
 632     IncludedUndefinedIds = DT.IncludedUndefinedIds;
 633
 634     // Remember the precedence of this operator and get the precedence of the
 635     // operator immediately to the right of the RHS.
 636     unsigned ThisPrec = PeekPrec;
 637     PeekPrec = getPrecedence(PeekTok.getKind());
 638
 639     // If this token isn't valid, report the error.
 640     if (PeekPrec == ~0U) {
 641       diagnoseUnexpectedOperator(PP, RHS, PeekTok);
 642       return true;
 643     }
 644
 645     // Decide whether to include the next binop in this subexpression.  For
 646     // example, when parsing x+y*z and looking at '*', we want to recursively
 647     // handle y*z as a single subexpression.  We do this because the precedence
 648     // of * is higher than that of +.  The only strange case we have to handle
 649     // here is for the ?: operator, where the precedence is actually lower than
 650     // the LHS of the '?'.  The grammar rule is:
 651     //
 652     // conditional-expression ::=
 653     //    logical-OR-expression ? expression : conditional-expression
 654     // where 'expression' is actually comma-expression.
 655     unsigned RHSPrec;
 656     if (Operator == tok::question)
 657       // The RHS of "?" should be maximally consumed as an expression.
 658       RHSPrec = getPrecedence(tok::comma);
 659     else  // All others should munch while higher precedence.
 660       RHSPrec = ThisPrec+1;
 661
 662     if (PeekPrec >= RHSPrec) {
 663       if (EvaluateDirectiveSubExpr(RHS, RHSPrec, PeekTok, RHSIsLive,
 664                                    IncludedUndefinedIds, PP))
 665         return true;
 666       PeekPrec = getPrecedence(PeekTok.getKind());
 667     }
 668     assert(PeekPrec <= ThisPrec && "Recursion didn't work!");
 669
 670     // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if
 671     // either operand is unsigned.
 672     llvm::APSInt Res(LHS.getBitWidth());
 673     switch (Operator) {
 674     case tok::question:       // No UAC for x and y in "x ? y : z".
 675     case tok::lessless:       // Shift amount doesn't UAC with shift value.
 676     case tok::greatergreater: // Shift amount doesn't UAC with shift value.
 677     case tok::comma:          // Comma operands are not subject to UACs.
 678     case tok::pipepipe:       // Logical || does not do UACs.
 679     case tok::ampamp:         // Logical && does not do UACs.
 680       break;                  // No UAC
 681     default:
 682       Res.setIsUnsigned(LHS.isUnsigned() || RHS.isUnsigned());
 683       // If this just promoted something from signed to unsigned, and if the
 684       // value was negative, warn about it.
 685       if (ValueLive && Res.isUnsigned()) {
 686         if (!LHS.isUnsigned() && LHS.Val.isNegative())
 687           PP.Diag(OpLoc, diag::warn_pp_convert_to_positive) << 0
 688             << toString(LHS.Val, 10, true) + " to " +
 689                toString(LHS.Val, 10, false)
 690             << LHS.getRange() << RHS.getRange();
 691         if (!RHS.isUnsigned() && RHS.Val.isNegative())
 692           PP.Diag(OpLoc, diag::warn_pp_convert_to_positive) << 1
 693             << toString(RHS.Val, 10, true) + " to " +
 694                toString(RHS.Val, 10, false)
 695             << LHS.getRange() << RHS.getRange();
 696       }
 697       LHS.Val.setIsUnsigned(Res.isUnsigned());
 698       RHS.Val.setIsUnsigned(Res.isUnsigned());
 699     }
 700
 701     bool Overflow = false;
 702     switch (Operator) {
 703     default: llvm_unreachable("Unknown operator token!");
 704     case tok::percent:
 705       if (RHS.Val != 0)
 706         Res = LHS.Val % RHS.Val;
 707       else if (ValueLive) {
 708         PP.Diag(OpLoc, diag::err_pp_remainder_by_zero)
 709           << LHS.getRange() << RHS.getRange();
 710         return true;
 711       }
 712       break;
 713     case tok::slash:
 714       if (RHS.Val != 0) {
 715         if (LHS.Val.isSigned())
 716           Res = llvm::APSInt(LHS.Val.sdiv_ov(RHS.Val, Overflow), false);
 717         else
 718           Res = LHS.Val / RHS.Val;
 719       } else if (ValueLive) {
 720         PP.Diag(OpLoc, diag::err_pp_division_by_zero)
 721           << LHS.getRange() << RHS.getRange();
 722         return true;
 723       }
 724       break;
 725
 726     case tok::star:
 727       if (Res.isSigned())
 728         Res = llvm::APSInt(LHS.Val.smul_ov(RHS.Val, Overflow), false);
 729       else
 730         Res = LHS.Val * RHS.Val;
 731       break;
 732     case tok::lessless: {
 733       // Determine whether overflow is about to happen.
 734       if (LHS.isUnsigned())
 735         Res = LHS.Val.ushl_ov(RHS.Val, Overflow);
 736       else
 737         Res = llvm::APSInt(LHS.Val.sshl_ov(RHS.Val, Overflow), false);
 738       break;
 739     }
 740     case tok::greatergreater: {
 741       // Determine whether overflow is about to happen.
 742       unsigned ShAmt = static_cast<unsigned>(RHS.Val.getLimitedValue());
 743       if (ShAmt >= LHS.getBitWidth()) {
 744         Overflow = true;
 745         ShAmt = LHS.getBitWidth()-1;
 746       }
 747       Res = LHS.Val >> ShAmt;
 748       break;
 749     }
 750     case tok::plus:
 751       if (LHS.isUnsigned())
 752         Res = LHS.Val + RHS.Val;
 753       else
 754         Res = llvm::APSInt(LHS.Val.sadd_ov(RHS.Val, Overflow), false);
 755       break;
 756     case tok::minus:
 757       if (LHS.isUnsigned())
 758         Res = LHS.Val - RHS.Val;
 759       else
 760         Res = llvm::APSInt(LHS.Val.ssub_ov(RHS.Val, Overflow), false);
 761       break;
 762     case tok::lessequal:
 763       Res = LHS.Val <= RHS.Val;
 764       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 765       break;
 766     case tok::less:
 767       Res = LHS.Val < RHS.Val;
 768       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 769       break;
 770     case tok::greaterequal:
 771       Res = LHS.Val >= RHS.Val;
 772       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 773       break;
 774     case tok::greater:
 775       Res = LHS.Val > RHS.Val;
 776       Res.setIsUnsigned(false);  // C99 6.5.8p6, result is always int (signed)
 777       break;
 778     case tok::exclaimequal:
 779       Res = LHS.Val != RHS.Val;
 780       Res.setIsUnsigned(false);  // C99 6.5.9p3, result is always int (signed)
 781       break;
 782     case tok::equalequal:
 783       Res = LHS.Val == RHS.Val;
 784       Res.setIsUnsigned(false);  // C99 6.5.9p3, result is always int (signed)
 785       break;
 786     case tok::amp:
 787       Res = LHS.Val & RHS.Val;
 788       break;
 789     case tok::caret:
 790       Res = LHS.Val ^ RHS.Val;
 791       break;
 792     case tok::pipe:
 793       Res = LHS.Val | RHS.Val;
 794       break;
 795     case tok::ampamp:
 796       Res = (LHS.Val != 0 && RHS.Val != 0);
 797       Res.setIsUnsigned(false);  // C99 6.5.13p3, result is always int (signed)
 798       break;
 799     case tok::pipepipe:
 800       Res = (LHS.Val != 0 || RHS.Val != 0);
 801       Res.setIsUnsigned(false);  // C99 6.5.14p3, result is always int (signed)
 802       break;
 803     case tok::comma:
 804       // Comma is invalid in pp expressions in c89/c++ mode, but is valid in C99
 805       // if not being evaluated.
 806       if (!PP.getLangOpts().C99 || ValueLive)
 807         PP.Diag(OpLoc, diag::ext_pp_comma_expr)
 808           << LHS.getRange() << RHS.getRange();
 809       Res = RHS.Val; // LHS = LHS,RHS -> RHS.
 810       break;
 811     case tok::question: {
 812       // Parse the : part of the expression.
 813       if (PeekTok.isNot(tok::colon)) {
 814         PP.Diag(PeekTok.getLocation(), diag::err_expected)
 815             << tok::colon << LHS.getRange() << RHS.getRange();
 816         PP.Diag(OpLoc, diag::note_matching) << tok::question;
 817         return true;
 818       }
 819       // Consume the :.
 820       PP.LexNonComment(PeekTok);
 821
 822       // Evaluate the value after the :.
 823       bool AfterColonLive = ValueLive && LHS.Val == 0;
 824       PPValue AfterColonVal(LHS.getBitWidth());
 825       DefinedTracker DT;
 826       if (EvaluateValue(AfterColonVal, PeekTok, DT, AfterColonLive, PP))
 827         return true;
 828
 829       // Parse anything after the : with the same precedence as ?.  We allow
 830       // things of equal precedence because ?: is right associative.
 831       if (EvaluateDirectiveSubExpr(AfterColonVal, ThisPrec,
 832                                    PeekTok, AfterColonLive,
 833                                    IncludedUndefinedIds, PP))
 834         return true;
 835
 836       // Now that we have the condition, the LHS and the RHS of the :, evaluate.
 837       Res = LHS.Val != 0 ? RHS.Val : AfterColonVal.Val;
 838       RHS.setEnd(AfterColonVal.getRange().getEnd());
 839
 840       // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if
 841       // either operand is unsigned.
 842       Res.setIsUnsigned(RHS.isUnsigned() || AfterColonVal.isUnsigned());
 843
 844       // Figure out the precedence of the token after the : part.
 845       PeekPrec = getPrecedence(PeekTok.getKind());
 846       break;
 847     }
 848     case tok::colon:
 849       // Don't allow :'s to float around without being part of ?: exprs.
 850       PP.Diag(OpLoc, diag::err_pp_colon_without_question)
 851         << LHS.getRange() << RHS.getRange();
 852       return true;
 853     }
 854
 855     // If this operator is live and overflowed, report the issue.
 856     if (Overflow && ValueLive)
 857       PP.Diag(OpLoc, diag::warn_pp_expr_overflow)
 858         << LHS.getRange() << RHS.getRange();
 859
 860     // Put the result back into 'LHS' for our next iteration.
 861     LHS.Val = Res;
 862     LHS.setEnd(RHS.getRange().getEnd());
 863     RHS.setIdentifier(nullptr);
 864   }
 865 }
 866
 867 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that
 868 /// may occur after a #if or #elif directive.  If the expression is equivalent
 869 /// to "!defined(X)" return X in IfNDefMacro.
 870 Preprocessor::DirectiveEvalResult
 871 Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
 872   SaveAndRestore PPDir(ParsingIfOrElifDirective, true);
 873   // Save the current state of 'DisableMacroExpansion' and reset it to false. If
 874   // 'DisableMacroExpansion' is true, then we must be in a macro argument list
 875   // in which case a directive is undefined behavior.  We want macros to be able
 876   // to recursively expand in order to get more gcc-list behavior, so we force
 877   // DisableMacroExpansion to false and restore it when we're done parsing the
 878   // expression.
 879   bool DisableMacroExpansionAtStartOfDirective = DisableMacroExpansion;
 880   DisableMacroExpansion = false;
 881
 882   // Peek ahead one token.
 883   Token Tok;
 884   LexNonComment(Tok);
 885
 886   // C99 6.10.1p3 - All expressions are evaluated as intmax_t or uintmax_t.
 887   unsigned BitWidth = getTargetInfo().getIntMaxTWidth();
 888
 889   PPValue ResVal(BitWidth);
 890   DefinedTracker DT;
 891   SourceLocation ExprStartLoc = SourceMgr.getExpansionLoc(Tok.getLocation());
 892   if (EvaluateValue(ResVal, Tok, DT, true, *this)) {
 893     // Parse error, skip the rest of the macro line.
 894     SourceRange ConditionRange = ExprStartLoc;
 895     if (Tok.isNot(tok::eod))
 896       ConditionRange = DiscardUntilEndOfDirective();
 897
 898     // Restore 'DisableMacroExpansion'.
 899     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 900
 901     // We cannot trust the source range from the value because there was a
 902     // parse error. Track the range manually -- the end of the directive is the
 903     // end of the condition range.
 904     return {false,
 905             DT.IncludedUndefinedIds,
 906             {ExprStartLoc, ConditionRange.getEnd()}};
 907   }
 908
 909   // If we are at the end of the expression after just parsing a value, there
 910   // must be no (unparenthesized) binary operators involved, so we can exit
 911   // directly.
 912   if (Tok.is(tok::eod)) {
 913     // If the expression we parsed was of the form !defined(macro), return the
 914     // macro in IfNDefMacro.
 915     if (DT.State == DefinedTracker::NotDefinedMacro)
 916       IfNDefMacro = DT.TheMacro;
 917
 918     // Restore 'DisableMacroExpansion'.
 919     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 920     return {ResVal.Val != 0, DT.IncludedUndefinedIds, ResVal.getRange()};
 921   }
 922
 923   // Otherwise, we must have a binary operator (e.g. "#if 1 < 2"), so parse the
 924   // operator and the stuff after it.
 925   if (EvaluateDirectiveSubExpr(ResVal, getPrecedence(tok::question),
 926                                Tok, true, DT.IncludedUndefinedIds, *this)) {
 927     // Parse error, skip the rest of the macro line.
 928     if (Tok.isNot(tok::eod))
 929       DiscardUntilEndOfDirective();
 930
 931     // Restore 'DisableMacroExpansion'.
 932     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 933     return {false, DT.IncludedUndefinedIds, ResVal.getRange()};
 934   }
 935
 936   // If we aren't at the tok::eod token, something bad happened, like an extra
 937   // ')' token.
 938   if (Tok.isNot(tok::eod)) {
 939     Diag(Tok, diag::err_pp_expected_eol);
 940     DiscardUntilEndOfDirective();
 941   }
 942
 943   // Restore 'DisableMacroExpansion'.
 944   DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
 945   return {ResVal.Val != 0, DT.IncludedUndefinedIds, ResVal.getRange()};
 946 }