llvm-project

1
//===---- ParseStmtAsm.cpp - Assembly Statement Parser --------------------===//
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 parsing for GCC and Microsoft inline assembly.
10
//
11
//===----------------------------------------------------------------------===//
12

13
#include "clang/AST/ASTContext.h"
14
#include "clang/Basic/Diagnostic.h"
15
#include "clang/Basic/TargetInfo.h"
16
#include "clang/Parse/Parser.h"
17
#include "clang/Parse/RAIIObjectsForParser.h"
18
#include "llvm/ADT/SmallString.h"
19
#include "llvm/ADT/StringExtras.h"
20
#include "llvm/MC/MCAsmInfo.h"
21
#include "llvm/MC/MCContext.h"
22
#include "llvm/MC/MCInstPrinter.h"
23
#include "llvm/MC/MCInstrInfo.h"
24
#include "llvm/MC/MCObjectFileInfo.h"
25
#include "llvm/MC/MCParser/MCAsmParser.h"
26
#include "llvm/MC/MCParser/MCTargetAsmParser.h"
27
#include "llvm/MC/MCRegisterInfo.h"
28
#include "llvm/MC/MCStreamer.h"
29
#include "llvm/MC/MCSubtargetInfo.h"
30
#include "llvm/MC/MCTargetOptions.h"
31
#include "llvm/MC/TargetRegistry.h"
32
#include "llvm/Support/SourceMgr.h"
33
#include "llvm/Support/TargetSelect.h"
34
using namespace clang;
35

36
namespace {
37
class ClangAsmParserCallback : public llvm::MCAsmParserSemaCallback {
38
  Parser &TheParser;
39
  SourceLocation AsmLoc;
40
  StringRef AsmString;
41

42
  /// The tokens we streamed into AsmString and handed off to MC.
43
  ArrayRef<Token> AsmToks;
44

45
  /// The offset of each token in AsmToks within AsmString.
46
  ArrayRef<unsigned> AsmTokOffsets;
47

48
public:
49
  ClangAsmParserCallback(Parser &P, SourceLocation Loc, StringRef AsmString,
50
                         ArrayRef<Token> Toks, ArrayRef<unsigned> Offsets)
51
      : TheParser(P), AsmLoc(Loc), AsmString(AsmString), AsmToks(Toks),
52
        AsmTokOffsets(Offsets) {
53
    assert(AsmToks.size() == AsmTokOffsets.size());
54
  }
55

56
  void LookupInlineAsmIdentifier(StringRef &LineBuf,
57
                                 llvm::InlineAsmIdentifierInfo &Info,
58
                                 bool IsUnevaluatedContext) override;
59

60
  StringRef LookupInlineAsmLabel(StringRef Identifier, llvm::SourceMgr &LSM,
61
                                 llvm::SMLoc Location,
62
                                 bool Create) override;
63

64
  bool LookupInlineAsmField(StringRef Base, StringRef Member,
65
                            unsigned &Offset) override {
66
    return TheParser.getActions().LookupInlineAsmField(Base, Member, Offset,
67
                                                       AsmLoc);
68
  }
69

70
  static void DiagHandlerCallback(const llvm::SMDiagnostic &D, void *Context) {
71
    ((ClangAsmParserCallback *)Context)->handleDiagnostic(D);
72
  }
73

74
private:
75
  /// Collect the appropriate tokens for the given string.
76
  void findTokensForString(StringRef Str, SmallVectorImpl<Token> &TempToks,
77
                           const Token *&FirstOrigToken) const;
78

79
  SourceLocation translateLocation(const llvm::SourceMgr &LSM,
80
                                   llvm::SMLoc SMLoc);
81

82
  void handleDiagnostic(const llvm::SMDiagnostic &D);
83
};
84
}
85

86
void ClangAsmParserCallback::LookupInlineAsmIdentifier(
87
    StringRef &LineBuf, llvm::InlineAsmIdentifierInfo &Info,
88
    bool IsUnevaluatedContext) {
89
  // Collect the desired tokens.
90
  SmallVector<Token, 16> LineToks;
91
  const Token *FirstOrigToken = nullptr;
92
  findTokensForString(LineBuf, LineToks, FirstOrigToken);
93

94
  unsigned NumConsumedToks;
95
  ExprResult Result = TheParser.ParseMSAsmIdentifier(LineToks, NumConsumedToks,
96
                                                     IsUnevaluatedContext);
97

98
  // If we consumed the entire line, tell MC that.
99
  // Also do this if we consumed nothing as a way of reporting failure.
100
  if (NumConsumedToks == 0 || NumConsumedToks == LineToks.size()) {
101
    // By not modifying LineBuf, we're implicitly consuming it all.
102

103
    // Otherwise, consume up to the original tokens.
104
  } else {
105
    assert(FirstOrigToken && "not using original tokens?");
106

107
    // Since we're using original tokens, apply that offset.
108
    assert(FirstOrigToken[NumConsumedToks].getLocation() ==
109
           LineToks[NumConsumedToks].getLocation());
110
    unsigned FirstIndex = FirstOrigToken - AsmToks.begin();
111
    unsigned LastIndex = FirstIndex + NumConsumedToks - 1;
112

113
    // The total length we've consumed is the relative offset
114
    // of the last token we consumed plus its length.
115
    unsigned TotalOffset =
116
        (AsmTokOffsets[LastIndex] + AsmToks[LastIndex].getLength() -
117
         AsmTokOffsets[FirstIndex]);
118
    LineBuf = LineBuf.substr(0, TotalOffset);
119
  }
120

121
  // Initialize Info with the lookup result.
122
  if (!Result.isUsable())
123
    return;
124
  TheParser.getActions().FillInlineAsmIdentifierInfo(Result.get(), Info);
125
}
126

127
StringRef ClangAsmParserCallback::LookupInlineAsmLabel(StringRef Identifier,
128
                                                       llvm::SourceMgr &LSM,
129
                                                       llvm::SMLoc Location,
130
                                                       bool Create) {
131
  SourceLocation Loc = translateLocation(LSM, Location);
132
  LabelDecl *Label =
133
      TheParser.getActions().GetOrCreateMSAsmLabel(Identifier, Loc, Create);
134
  return Label->getMSAsmLabel();
135
}
136

137
void ClangAsmParserCallback::findTokensForString(
138
    StringRef Str, SmallVectorImpl<Token> &TempToks,
139
    const Token *&FirstOrigToken) const {
140
  // For now, assert that the string we're working with is a substring
141
  // of what we gave to MC.  This lets us use the original tokens.
142
  assert(!std::less<const char *>()(Str.begin(), AsmString.begin()) &&
143
         !std::less<const char *>()(AsmString.end(), Str.end()));
144

145
  // Try to find a token whose offset matches the first token.
146
  unsigned FirstCharOffset = Str.begin() - AsmString.begin();
147
  const unsigned *FirstTokOffset =
148
      llvm::lower_bound(AsmTokOffsets, FirstCharOffset);
149

150
  // For now, assert that the start of the string exactly
151
  // corresponds to the start of a token.
152
  assert(*FirstTokOffset == FirstCharOffset);
153

154
  // Use all the original tokens for this line.  (We assume the
155
  // end of the line corresponds cleanly to a token break.)
156
  unsigned FirstTokIndex = FirstTokOffset - AsmTokOffsets.begin();
157
  FirstOrigToken = &AsmToks[FirstTokIndex];
158
  unsigned LastCharOffset = Str.end() - AsmString.begin();
159
  for (unsigned i = FirstTokIndex, e = AsmTokOffsets.size(); i != e; ++i) {
160
    if (AsmTokOffsets[i] >= LastCharOffset)
161
      break;
162
    TempToks.push_back(AsmToks[i]);
163
  }
164
}
165

166
SourceLocation
167
ClangAsmParserCallback::translateLocation(const llvm::SourceMgr &LSM,
168
                                          llvm::SMLoc SMLoc) {
169
  // Compute an offset into the inline asm buffer.
170
  // FIXME: This isn't right if .macro is involved (but hopefully, no
171
  // real-world code does that).
172
  const llvm::MemoryBuffer *LBuf =
173
      LSM.getMemoryBuffer(LSM.FindBufferContainingLoc(SMLoc));
174
  unsigned Offset = SMLoc.getPointer() - LBuf->getBufferStart();
175

176
  // Figure out which token that offset points into.
177
  const unsigned *TokOffsetPtr = llvm::lower_bound(AsmTokOffsets, Offset);
178
  unsigned TokIndex = TokOffsetPtr - AsmTokOffsets.begin();
179
  unsigned TokOffset = *TokOffsetPtr;
180

181
  // If we come up with an answer which seems sane, use it; otherwise,
182
  // just point at the __asm keyword.
183
  // FIXME: Assert the answer is sane once we handle .macro correctly.
184
  SourceLocation Loc = AsmLoc;
185
  if (TokIndex < AsmToks.size()) {
186
    const Token &Tok = AsmToks[TokIndex];
187
    Loc = Tok.getLocation();
188
    Loc = Loc.getLocWithOffset(Offset - TokOffset);
189
  }
190
  return Loc;
191
}
192

193
void ClangAsmParserCallback::handleDiagnostic(const llvm::SMDiagnostic &D) {
194
  const llvm::SourceMgr &LSM = *D.getSourceMgr();
195
  SourceLocation Loc = translateLocation(LSM, D.getLoc());
196
  TheParser.Diag(Loc, diag::err_inline_ms_asm_parsing) << D.getMessage();
197
}
198

199
/// Parse an identifier in an MS-style inline assembly block.
200
ExprResult Parser::ParseMSAsmIdentifier(llvm::SmallVectorImpl<Token> &LineToks,
201
                                        unsigned &NumLineToksConsumed,
202
                                        bool IsUnevaluatedContext) {
203
  // Push a fake token on the end so that we don't overrun the token
204
  // stream.  We use ';' because it expression-parsing should never
205
  // overrun it.
206
  const tok::TokenKind EndOfStream = tok::semi;
207
  Token EndOfStreamTok;
208
  EndOfStreamTok.startToken();
209
  EndOfStreamTok.setKind(EndOfStream);
210
  LineToks.push_back(EndOfStreamTok);
211

212
  // Also copy the current token over.
213
  LineToks.push_back(Tok);
214

215
  PP.EnterTokenStream(LineToks, /*DisableMacroExpansions*/ true,
216
                      /*IsReinject*/ true);
217

218
  // Clear the current token and advance to the first token in LineToks.
219
  ConsumeAnyToken();
220

221
  // Parse an optional scope-specifier if we're in C++.
222
  CXXScopeSpec SS;
223
  if (getLangOpts().CPlusPlus)
224
    ParseOptionalCXXScopeSpecifier(SS, /*ObjectType=*/nullptr,
225
                                   /*ObjectHasErrors=*/false,
226
                                   /*EnteringContext=*/false);
227

228
  // Require an identifier here.
229
  SourceLocation TemplateKWLoc;
230
  UnqualifiedId Id;
231
  bool Invalid = true;
232
  ExprResult Result;
233
  if (Tok.is(tok::kw_this)) {
234
    Result = ParseCXXThis();
235
    Invalid = false;
236
  } else {
237
    Invalid =
238
        ParseUnqualifiedId(SS, /*ObjectType=*/nullptr,
239
                           /*ObjectHadErrors=*/false,
240
                           /*EnteringContext=*/false,
241
                           /*AllowDestructorName=*/false,
242
                           /*AllowConstructorName=*/false,
243
                           /*AllowDeductionGuide=*/false, &TemplateKWLoc, Id);
244
    // Perform the lookup.
245
    Result = Actions.LookupInlineAsmIdentifier(SS, TemplateKWLoc, Id,
246
                                               IsUnevaluatedContext);
247
  }
248
  // While the next two tokens are 'period' 'identifier', repeatedly parse it as
249
  // a field access. We have to avoid consuming assembler directives that look
250
  // like '.' 'else'.
251
  while (Result.isUsable() && Tok.is(tok::period)) {
252
    Token IdTok = PP.LookAhead(0);
253
    if (IdTok.isNot(tok::identifier))
254
      break;
255
    ConsumeToken(); // Consume the period.
256
    IdentifierInfo *Id = Tok.getIdentifierInfo();
257
    ConsumeToken(); // Consume the identifier.
258
    Result = Actions.LookupInlineAsmVarDeclField(Result.get(), Id->getName(),
259
                                                 Tok.getLocation());
260
  }
261

262
  // Figure out how many tokens we are into LineToks.
263
  unsigned LineIndex = 0;
264
  if (Tok.is(EndOfStream)) {
265
    LineIndex = LineToks.size() - 2;
266
  } else {
267
    while (LineToks[LineIndex].getLocation() != Tok.getLocation()) {
268
      LineIndex++;
269
      assert(LineIndex < LineToks.size() - 2); // we added two extra tokens
270
    }
271
  }
272

273
  // If we've run into the poison token we inserted before, or there
274
  // was a parsing error, then claim the entire line.
275
  if (Invalid || Tok.is(EndOfStream)) {
276
    NumLineToksConsumed = LineToks.size() - 2;
277
  } else {
278
    // Otherwise, claim up to the start of the next token.
279
    NumLineToksConsumed = LineIndex;
280
  }
281

282
  // Finally, restore the old parsing state by consuming all the tokens we
283
  // staged before, implicitly killing off the token-lexer we pushed.
284
  for (unsigned i = 0, e = LineToks.size() - LineIndex - 2; i != e; ++i) {
285
    ConsumeAnyToken();
286
  }
287
  assert(Tok.is(EndOfStream));
288
  ConsumeToken();
289

290
  // Leave LineToks in its original state.
291
  LineToks.pop_back();
292
  LineToks.pop_back();
293

294
  return Result;
295
}
296

297
/// Turn a sequence of our tokens back into a string that we can hand
298
/// to the MC asm parser.
299
static bool buildMSAsmString(Preprocessor &PP, SourceLocation AsmLoc,
300
                             ArrayRef<Token> AsmToks,
301
                             SmallVectorImpl<unsigned> &TokOffsets,
302
                             SmallString<512> &Asm) {
303
  assert(!AsmToks.empty() && "Didn't expect an empty AsmToks!");
304

305
  // Is this the start of a new assembly statement?
306
  bool isNewStatement = true;
307

308
  for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
309
    const Token &Tok = AsmToks[i];
310

311
    // Start each new statement with a newline and a tab.
312
    if (!isNewStatement && (Tok.is(tok::kw_asm) || Tok.isAtStartOfLine())) {
313
      Asm += "\n\t";
314
      isNewStatement = true;
315
    }
316

317
    // Preserve the existence of leading whitespace except at the
318
    // start of a statement.
319
    if (!isNewStatement && Tok.hasLeadingSpace())
320
      Asm += ' ';
321

322
    // Remember the offset of this token.
323
    TokOffsets.push_back(Asm.size());
324

325
    // Don't actually write '__asm' into the assembly stream.
326
    if (Tok.is(tok::kw_asm)) {
327
      // Complain about __asm at the end of the stream.
328
      if (i + 1 == e) {
329
        PP.Diag(AsmLoc, diag::err_asm_empty);
330
        return true;
331
      }
332

333
      continue;
334
    }
335

336
    // Append the spelling of the token.
337
    SmallString<32> SpellingBuffer;
338
    bool SpellingInvalid = false;
339
    Asm += PP.getSpelling(Tok, SpellingBuffer, &SpellingInvalid);
340
    assert(!SpellingInvalid && "spelling was invalid after correct parse?");
341

342
    // We are no longer at the start of a statement.
343
    isNewStatement = false;
344
  }
345

346
  // Ensure that the buffer is null-terminated.
347
  Asm.push_back('\0');
348
  Asm.pop_back();
349

350
  assert(TokOffsets.size() == AsmToks.size());
351
  return false;
352
}
353

354
// Determine if this is a GCC-style asm statement.
355
bool Parser::isGCCAsmStatement(const Token &TokAfterAsm) const {
356
  return TokAfterAsm.is(tok::l_paren) || isGNUAsmQualifier(TokAfterAsm);
357
}
358

359
bool Parser::isGNUAsmQualifier(const Token &TokAfterAsm) const {
360
  return getGNUAsmQualifier(TokAfterAsm) != GNUAsmQualifiers::AQ_unspecified;
361
}
362

363
/// ParseMicrosoftAsmStatement. When -fms-extensions/-fasm-blocks is enabled,
364
/// this routine is called to collect the tokens for an MS asm statement.
365
///
366
/// [MS]  ms-asm-statement:
367
///         ms-asm-block
368
///         ms-asm-block ms-asm-statement
369
///
370
/// [MS]  ms-asm-block:
371
///         '__asm' ms-asm-line '\n'
372
///         '__asm' '{' ms-asm-instruction-block[opt] '}' ';'[opt]
373
///
374
/// [MS]  ms-asm-instruction-block
375
///         ms-asm-line
376
///         ms-asm-line '\n' ms-asm-instruction-block
377
///
378
StmtResult Parser::ParseMicrosoftAsmStatement(SourceLocation AsmLoc) {
379
  SourceManager &SrcMgr = PP.getSourceManager();
380
  SourceLocation EndLoc = AsmLoc;
381
  SmallVector<Token, 4> AsmToks;
382

383
  bool SingleLineMode = true;
384
  unsigned BraceNesting = 0;
385
  unsigned short savedBraceCount = BraceCount;
386
  bool InAsmComment = false;
387
  FileID FID;
388
  unsigned LineNo = 0;
389
  unsigned NumTokensRead = 0;
390
  SmallVector<SourceLocation, 4> LBraceLocs;
391
  bool SkippedStartOfLine = false;
392

393
  if (Tok.is(tok::l_brace)) {
394
    // Braced inline asm: consume the opening brace.
395
    SingleLineMode = false;
396
    BraceNesting = 1;
397
    EndLoc = ConsumeBrace();
398
    LBraceLocs.push_back(EndLoc);
399
    ++NumTokensRead;
400
  } else {
401
    // Single-line inline asm; compute which line it is on.
402
    std::pair<FileID, unsigned> ExpAsmLoc =
403
        SrcMgr.getDecomposedExpansionLoc(EndLoc);
404
    FID = ExpAsmLoc.first;
405
    LineNo = SrcMgr.getLineNumber(FID, ExpAsmLoc.second);
406
    LBraceLocs.push_back(SourceLocation());
407
  }
408

409
  SourceLocation TokLoc = Tok.getLocation();
410
  do {
411
    // If we hit EOF, we're done, period.
412
    if (isEofOrEom())
413
      break;
414

415
    if (!InAsmComment && Tok.is(tok::l_brace)) {
416
      // Consume the opening brace.
417
      SkippedStartOfLine = Tok.isAtStartOfLine();
418
      AsmToks.push_back(Tok);
419
      EndLoc = ConsumeBrace();
420
      BraceNesting++;
421
      LBraceLocs.push_back(EndLoc);
422
      TokLoc = Tok.getLocation();
423
      ++NumTokensRead;
424
      continue;
425
    } else if (!InAsmComment && Tok.is(tok::semi)) {
426
      // A semicolon in an asm is the start of a comment.
427
      InAsmComment = true;
428
      if (!SingleLineMode) {
429
        // Compute which line the comment is on.
430
        std::pair<FileID, unsigned> ExpSemiLoc =
431
            SrcMgr.getDecomposedExpansionLoc(TokLoc);
432
        FID = ExpSemiLoc.first;
433
        LineNo = SrcMgr.getLineNumber(FID, ExpSemiLoc.second);
434
      }
435
    } else if (SingleLineMode || InAsmComment) {
436
      // If end-of-line is significant, check whether this token is on a
437
      // new line.
438
      std::pair<FileID, unsigned> ExpLoc =
439
          SrcMgr.getDecomposedExpansionLoc(TokLoc);
440
      if (ExpLoc.first != FID ||
441
          SrcMgr.getLineNumber(ExpLoc.first, ExpLoc.second) != LineNo) {
442
        // If this is a single-line __asm, we're done, except if the next
443
        // line is MS-style asm too, in which case we finish a comment
444
        // if needed and then keep processing the next line as a single
445
        // line __asm.
446
        bool isAsm = Tok.is(tok::kw_asm);
447
        if (SingleLineMode && (!isAsm || isGCCAsmStatement(NextToken())))
448
          break;
449
        // We're no longer in a comment.
450
        InAsmComment = false;
451
        if (isAsm) {
452
          // If this is a new __asm {} block we want to process it separately
453
          // from the single-line __asm statements
454
          if (PP.LookAhead(0).is(tok::l_brace))
455
            break;
456
          LineNo = SrcMgr.getLineNumber(ExpLoc.first, ExpLoc.second);
457
          SkippedStartOfLine = Tok.isAtStartOfLine();
458
        } else if (Tok.is(tok::semi)) {
459
          // A multi-line asm-statement, where next line is a comment
460
          InAsmComment = true;
461
          FID = ExpLoc.first;
462
          LineNo = SrcMgr.getLineNumber(FID, ExpLoc.second);
463
        }
464
      } else if (!InAsmComment && Tok.is(tok::r_brace)) {
465
        // In MSVC mode, braces only participate in brace matching and
466
        // separating the asm statements.  This is an intentional
467
        // departure from the Apple gcc behavior.
468
        if (!BraceNesting)
469
          break;
470
      }
471
    }
472
    if (!InAsmComment && BraceNesting && Tok.is(tok::r_brace) &&
473
        BraceCount == (savedBraceCount + BraceNesting)) {
474
      // Consume the closing brace.
475
      SkippedStartOfLine = Tok.isAtStartOfLine();
476
      // Don't want to add the closing brace of the whole asm block
477
      if (SingleLineMode || BraceNesting > 1) {
478
        Tok.clearFlag(Token::LeadingSpace);
479
        AsmToks.push_back(Tok);
480
      }
481
      EndLoc = ConsumeBrace();
482
      BraceNesting--;
483
      // Finish if all of the opened braces in the inline asm section were
484
      // consumed.
485
      if (BraceNesting == 0 && !SingleLineMode)
486
        break;
487
      else {
488
        LBraceLocs.pop_back();
489
        TokLoc = Tok.getLocation();
490
        ++NumTokensRead;
491
        continue;
492
      }
493
    }
494

495
    // Consume the next token; make sure we don't modify the brace count etc.
496
    // if we are in a comment.
497
    EndLoc = TokLoc;
498
    if (InAsmComment)
499
      PP.Lex(Tok);
500
    else {
501
      // Set the token as the start of line if we skipped the original start
502
      // of line token in case it was a nested brace.
503
      if (SkippedStartOfLine)
504
        Tok.setFlag(Token::StartOfLine);
505
      AsmToks.push_back(Tok);
506
      ConsumeAnyToken();
507
    }
508
    TokLoc = Tok.getLocation();
509
    ++NumTokensRead;
510
    SkippedStartOfLine = false;
511
  } while (true);
512

513
  if (BraceNesting && BraceCount != savedBraceCount) {
514
    // __asm without closing brace (this can happen at EOF).
515
    for (unsigned i = 0; i < BraceNesting; ++i) {
516
      Diag(Tok, diag::err_expected) << tok::r_brace;
517
      Diag(LBraceLocs.back(), diag::note_matching) << tok::l_brace;
518
      LBraceLocs.pop_back();
519
    }
520
    return StmtError();
521
  } else if (NumTokensRead == 0) {
522
    // Empty __asm.
523
    Diag(Tok, diag::err_expected) << tok::l_brace;
524
    return StmtError();
525
  }
526

527
  // Okay, prepare to use MC to parse the assembly.
528
  SmallVector<StringRef, 4> ConstraintRefs;
529
  SmallVector<Expr *, 4> Exprs;
530
  SmallVector<StringRef, 4> ClobberRefs;
531

532
  // We need an actual supported target.
533
  const llvm::Triple &TheTriple = Actions.Context.getTargetInfo().getTriple();
534
  const std::string &TT = TheTriple.getTriple();
535
  const llvm::Target *TheTarget = nullptr;
536
  if (!TheTriple.isX86()) {
537
    Diag(AsmLoc, diag::err_msasm_unsupported_arch) << TheTriple.getArchName();
538
  } else {
539
    std::string Error;
540
    TheTarget = llvm::TargetRegistry::lookupTarget(TT, Error);
541
    if (!TheTarget)
542
      Diag(AsmLoc, diag::err_msasm_unable_to_create_target) << Error;
543
  }
544

545
  assert(!LBraceLocs.empty() && "Should have at least one location here");
546

547
  SmallString<512> AsmString;
548
  auto EmptyStmt = [&] {
549
    return Actions.ActOnMSAsmStmt(AsmLoc, LBraceLocs[0], AsmToks, AsmString,
550
                                  /*NumOutputs*/ 0, /*NumInputs*/ 0,
551
                                  ConstraintRefs, ClobberRefs, Exprs, EndLoc);
552
  };
553
  // If we don't support assembly, or the assembly is empty, we don't
554
  // need to instantiate the AsmParser, etc.
555
  if (!TheTarget || AsmToks.empty()) {
556
    return EmptyStmt();
557
  }
558

559
  // Expand the tokens into a string buffer.
560
  SmallVector<unsigned, 8> TokOffsets;
561
  if (buildMSAsmString(PP, AsmLoc, AsmToks, TokOffsets, AsmString))
562
    return StmtError();
563

564
  const TargetOptions &TO = Actions.Context.getTargetInfo().getTargetOpts();
565
  std::string FeaturesStr =
566
      llvm::join(TO.Features.begin(), TO.Features.end(), ",");
567

568
  std::unique_ptr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
569
  if (!MRI) {
570
    Diag(AsmLoc, diag::err_msasm_unable_to_create_target)
571
        << "target MC unavailable";
572
    return EmptyStmt();
573
  }
574
  // FIXME: init MCOptions from sanitizer flags here.
575
  llvm::MCTargetOptions MCOptions;
576
  std::unique_ptr<llvm::MCAsmInfo> MAI(
577
      TheTarget->createMCAsmInfo(*MRI, TT, MCOptions));
578
  // Get the instruction descriptor.
579
  std::unique_ptr<llvm::MCInstrInfo> MII(TheTarget->createMCInstrInfo());
580
  std::unique_ptr<llvm::MCSubtargetInfo> STI(
581
      TheTarget->createMCSubtargetInfo(TT, TO.CPU, FeaturesStr));
582
  // Target MCTargetDesc may not be linked in clang-based tools.
583

584
  if (!MAI || !MII || !STI) {
585
    Diag(AsmLoc, diag::err_msasm_unable_to_create_target)
586
        << "target MC unavailable";
587
    return EmptyStmt();
588
  }
589

590
  llvm::SourceMgr TempSrcMgr;
591
  llvm::MCContext Ctx(TheTriple, MAI.get(), MRI.get(), STI.get(), &TempSrcMgr);
592
  std::unique_ptr<llvm::MCObjectFileInfo> MOFI(
593
      TheTarget->createMCObjectFileInfo(Ctx, /*PIC=*/false));
594
  Ctx.setObjectFileInfo(MOFI.get());
595

596
  std::unique_ptr<llvm::MemoryBuffer> Buffer =
597
      llvm::MemoryBuffer::getMemBuffer(AsmString, "<MS inline asm>");
598

599
  // Tell SrcMgr about this buffer, which is what the parser will pick up.
600
  TempSrcMgr.AddNewSourceBuffer(std::move(Buffer), llvm::SMLoc());
601

602
  std::unique_ptr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
603
  std::unique_ptr<llvm::MCAsmParser> Parser(
604
      createMCAsmParser(TempSrcMgr, Ctx, *Str.get(), *MAI));
605

606
  std::unique_ptr<llvm::MCTargetAsmParser> TargetParser(
607
      TheTarget->createMCAsmParser(*STI, *Parser, *MII, MCOptions));
608
  // Target AsmParser may not be linked in clang-based tools.
609
  if (!TargetParser) {
610
    Diag(AsmLoc, diag::err_msasm_unable_to_create_target)
611
        << "target ASM parser unavailable";
612
    return EmptyStmt();
613
  }
614

615
  std::unique_ptr<llvm::MCInstPrinter> IP(
616
      TheTarget->createMCInstPrinter(llvm::Triple(TT), 1, *MAI, *MII, *MRI));
617

618
  // Change to the Intel dialect.
619
  Parser->setAssemblerDialect(1);
620
  Parser->setTargetParser(*TargetParser.get());
621
  Parser->setParsingMSInlineAsm(true);
622
  TargetParser->setParsingMSInlineAsm(true);
623

624
  ClangAsmParserCallback Callback(*this, AsmLoc, AsmString, AsmToks,
625
                                  TokOffsets);
626
  TargetParser->setSemaCallback(&Callback);
627
  TempSrcMgr.setDiagHandler(ClangAsmParserCallback::DiagHandlerCallback,
628
                            &Callback);
629

630
  unsigned NumOutputs;
631
  unsigned NumInputs;
632
  std::string AsmStringIR;
633
  SmallVector<std::pair<void *, bool>, 4> OpExprs;
634
  SmallVector<std::string, 4> Constraints;
635
  SmallVector<std::string, 4> Clobbers;
636
  if (Parser->parseMSInlineAsm(AsmStringIR, NumOutputs, NumInputs, OpExprs,
637
                               Constraints, Clobbers, MII.get(), IP.get(),
638
                               Callback))
639
    return StmtError();
640

641
  // Filter out "fpsw" and "mxcsr". They aren't valid GCC asm clobber
642
  // constraints. Clang always adds fpsr to the clobber list anyway.
643
  llvm::erase_if(Clobbers, [](const std::string &C) {
644
    return C == "fpsr" || C == "mxcsr";
645
  });
646

647
  // Build the vector of clobber StringRefs.
648
  ClobberRefs.insert(ClobberRefs.end(), Clobbers.begin(), Clobbers.end());
649

650
  // Recast the void pointers and build the vector of constraint StringRefs.
651
  unsigned NumExprs = NumOutputs + NumInputs;
652
  ConstraintRefs.resize(NumExprs);
653
  Exprs.resize(NumExprs);
654
  for (unsigned i = 0, e = NumExprs; i != e; ++i) {
655
    Expr *OpExpr = static_cast<Expr *>(OpExprs[i].first);
656
    if (!OpExpr)
657
      return StmtError();
658

659
    // Need address of variable.
660
    if (OpExprs[i].second)
661
      OpExpr =
662
          Actions.BuildUnaryOp(getCurScope(), AsmLoc, UO_AddrOf, OpExpr).get();
663

664
    ConstraintRefs[i] = StringRef(Constraints[i]);
665
    Exprs[i] = OpExpr;
666
  }
667

668
  // FIXME: We should be passing source locations for better diagnostics.
669
  return Actions.ActOnMSAsmStmt(AsmLoc, LBraceLocs[0], AsmToks, AsmStringIR,
670
                                NumOutputs, NumInputs, ConstraintRefs,
671
                                ClobberRefs, Exprs, EndLoc);
672
}
673

674
/// parseGNUAsmQualifierListOpt - Parse a GNU extended asm qualifier list.
675
///       asm-qualifier:
676
///         volatile
677
///         inline
678
///         goto
679
///
680
///       asm-qualifier-list:
681
///         asm-qualifier
682
///         asm-qualifier-list asm-qualifier
683
bool Parser::parseGNUAsmQualifierListOpt(GNUAsmQualifiers &AQ) {
684
  while (true) {
685
    const GNUAsmQualifiers::AQ A = getGNUAsmQualifier(Tok);
686
    if (A == GNUAsmQualifiers::AQ_unspecified) {
687
      if (Tok.isNot(tok::l_paren)) {
688
        Diag(Tok.getLocation(), diag::err_asm_qualifier_ignored);
689
        SkipUntil(tok::r_paren, StopAtSemi);
690
        return true;
691
      }
692
      return false;
693
    }
694
    if (AQ.setAsmQualifier(A))
695
      Diag(Tok.getLocation(), diag::err_asm_duplicate_qual)
696
          << GNUAsmQualifiers::getQualifierName(A);
697
    ConsumeToken();
698
  }
699
  return false;
700
}
701

702
/// ParseAsmStatement - Parse a GNU extended asm statement.
703
///       asm-statement:
704
///         gnu-asm-statement
705
///         ms-asm-statement
706
///
707
/// [GNU] gnu-asm-statement:
708
///         'asm' asm-qualifier-list[opt] '(' asm-argument ')' ';'
709
///
710
/// [GNU] asm-argument:
711
///         asm-string-literal
712
///         asm-string-literal ':' asm-operands[opt]
713
///         asm-string-literal ':' asm-operands[opt] ':' asm-operands[opt]
714
///         asm-string-literal ':' asm-operands[opt] ':' asm-operands[opt]
715
///                 ':' asm-clobbers
716
///
717
/// [GNU] asm-clobbers:
718
///         asm-string-literal
719
///         asm-clobbers ',' asm-string-literal
720
///
721
StmtResult Parser::ParseAsmStatement(bool &msAsm) {
722
  assert(Tok.is(tok::kw_asm) && "Not an asm stmt");
723
  SourceLocation AsmLoc = ConsumeToken();
724

725
  if (getLangOpts().AsmBlocks && !isGCCAsmStatement(Tok)) {
726
    msAsm = true;
727
    return ParseMicrosoftAsmStatement(AsmLoc);
728
  }
729

730
  SourceLocation Loc = Tok.getLocation();
731
  GNUAsmQualifiers GAQ;
732
  if (parseGNUAsmQualifierListOpt(GAQ))
733
    return StmtError();
734

735
  if (GAQ.isGoto() && getLangOpts().SpeculativeLoadHardening)
736
    Diag(Loc, diag::warn_slh_does_not_support_asm_goto);
737

738
  BalancedDelimiterTracker T(*this, tok::l_paren);
739
  T.consumeOpen();
740

741
  ExprResult AsmString(ParseAsmStringLiteral(/*ForAsmLabel*/ false));
742

743
  // Check if GNU-style InlineAsm is disabled.
744
  // Error on anything other than empty string.
745
  if (!(getLangOpts().GNUAsm || AsmString.isInvalid())) {
746
    const auto *SL = cast<StringLiteral>(AsmString.get());
747
    if (!SL->getString().trim().empty())
748
      Diag(Loc, diag::err_gnu_inline_asm_disabled);
749
  }
750

751
  if (AsmString.isInvalid()) {
752
    // Consume up to and including the closing paren.
753
    T.skipToEnd();
754
    return StmtError();
755
  }
756

757
  SmallVector<IdentifierInfo *, 4> Names;
758
  ExprVector Constraints;
759
  ExprVector Exprs;
760
  ExprVector Clobbers;
761

762
  if (Tok.is(tok::r_paren)) {
763
    // We have a simple asm expression like 'asm("foo")'.
764
    T.consumeClose();
765
    return Actions.ActOnGCCAsmStmt(
766
        AsmLoc, /*isSimple*/ true, GAQ.isVolatile(),
767
        /*NumOutputs*/ 0, /*NumInputs*/ 0, nullptr, Constraints, Exprs,
768
        AsmString.get(), Clobbers, /*NumLabels*/ 0, T.getCloseLocation());
769
  }
770

771
  // Parse Outputs, if present.
772
  bool AteExtraColon = false;
773
  if (Tok.is(tok::colon) || Tok.is(tok::coloncolon)) {
774
    // In C++ mode, parse "::" like ": :".
775
    AteExtraColon = Tok.is(tok::coloncolon);
776
    ConsumeToken();
777

778
    if (!AteExtraColon && ParseAsmOperandsOpt(Names, Constraints, Exprs))
779
      return StmtError();
780
  }
781

782
  unsigned NumOutputs = Names.size();
783

784
  // Parse Inputs, if present.
785
  if (AteExtraColon || Tok.is(tok::colon) || Tok.is(tok::coloncolon)) {
786
    // In C++ mode, parse "::" like ": :".
787
    if (AteExtraColon)
788
      AteExtraColon = false;
789
    else {
790
      AteExtraColon = Tok.is(tok::coloncolon);
791
      ConsumeToken();
792
    }
793

794
    if (!AteExtraColon && ParseAsmOperandsOpt(Names, Constraints, Exprs))
795
      return StmtError();
796
  }
797

798
  assert(Names.size() == Constraints.size() &&
799
         Constraints.size() == Exprs.size() && "Input operand size mismatch!");
800

801
  unsigned NumInputs = Names.size() - NumOutputs;
802

803
  // Parse the clobbers, if present.
804
  if (AteExtraColon || Tok.is(tok::colon) || Tok.is(tok::coloncolon)) {
805
    if (AteExtraColon)
806
      AteExtraColon = false;
807
    else {
808
      AteExtraColon = Tok.is(tok::coloncolon);
809
      ConsumeToken();
810
    }
811
    // Parse the asm-string list for clobbers if present.
812
    if (!AteExtraColon && isTokenStringLiteral()) {
813
      while (true) {
814
        ExprResult Clobber(ParseAsmStringLiteral(/*ForAsmLabel*/ false));
815

816
        if (Clobber.isInvalid())
817
          break;
818

819
        Clobbers.push_back(Clobber.get());
820

821
        if (!TryConsumeToken(tok::comma))
822
          break;
823
      }
824
    }
825
  }
826
  if (!GAQ.isGoto() && (Tok.isNot(tok::r_paren) || AteExtraColon)) {
827
    Diag(Tok, diag::err_expected) << tok::r_paren;
828
    SkipUntil(tok::r_paren, StopAtSemi);
829
    return StmtError();
830
  }
831

832
  // Parse the goto label, if present.
833
  unsigned NumLabels = 0;
834
  if (AteExtraColon || Tok.is(tok::colon)) {
835
    if (!AteExtraColon)
836
      ConsumeToken();
837

838
    while (true) {
839
      if (Tok.isNot(tok::identifier)) {
840
        Diag(Tok, diag::err_expected) << tok::identifier;
841
        SkipUntil(tok::r_paren, StopAtSemi);
842
        return StmtError();
843
      }
844
      LabelDecl *LD = Actions.LookupOrCreateLabel(Tok.getIdentifierInfo(),
845
                                                  Tok.getLocation());
846
      Names.push_back(Tok.getIdentifierInfo());
847
      if (!LD) {
848
        SkipUntil(tok::r_paren, StopAtSemi);
849
        return StmtError();
850
      }
851
      ExprResult Res =
852
          Actions.ActOnAddrLabel(Tok.getLocation(), Tok.getLocation(), LD);
853
      Exprs.push_back(Res.get());
854
      NumLabels++;
855
      ConsumeToken();
856
      if (!TryConsumeToken(tok::comma))
857
        break;
858
    }
859
  } else if (GAQ.isGoto()) {
860
    Diag(Tok, diag::err_expected) << tok::colon;
861
    SkipUntil(tok::r_paren, StopAtSemi);
862
    return StmtError();
863
  }
864
  T.consumeClose();
865
  return Actions.ActOnGCCAsmStmt(AsmLoc, false, GAQ.isVolatile(), NumOutputs,
866
                                 NumInputs, Names.data(), Constraints, Exprs,
867
                                 AsmString.get(), Clobbers, NumLabels,
868
                                 T.getCloseLocation());
869
}
870

871
/// ParseAsmOperands - Parse the asm-operands production as used by
872
/// asm-statement, assuming the leading ':' token was eaten.
873
///
874
/// [GNU] asm-operands:
875
///         asm-operand
876
///         asm-operands ',' asm-operand
877
///
878
/// [GNU] asm-operand:
879
///         asm-string-literal '(' expression ')'
880
///         '[' identifier ']' asm-string-literal '(' expression ')'
881
///
882
//
883
// FIXME: Avoid unnecessary std::string trashing.
884
bool Parser::ParseAsmOperandsOpt(SmallVectorImpl<IdentifierInfo *> &Names,
885
                                 SmallVectorImpl<Expr *> &Constraints,
886
                                 SmallVectorImpl<Expr *> &Exprs) {
887
  // 'asm-operands' isn't present?
888
  if (!isTokenStringLiteral() && Tok.isNot(tok::l_square))
889
    return false;
890

891
  while (true) {
892
    // Read the [id] if present.
893
    if (Tok.is(tok::l_square)) {
894
      BalancedDelimiterTracker T(*this, tok::l_square);
895
      T.consumeOpen();
896

897
      if (Tok.isNot(tok::identifier)) {
898
        Diag(Tok, diag::err_expected) << tok::identifier;
899
        SkipUntil(tok::r_paren, StopAtSemi);
900
        return true;
901
      }
902

903
      IdentifierInfo *II = Tok.getIdentifierInfo();
904
      ConsumeToken();
905

906
      Names.push_back(II);
907
      T.consumeClose();
908
    } else
909
      Names.push_back(nullptr);
910

911
    ExprResult Constraint(ParseAsmStringLiteral(/*ForAsmLabel*/ false));
912
    if (Constraint.isInvalid()) {
913
      SkipUntil(tok::r_paren, StopAtSemi);
914
      return true;
915
    }
916
    Constraints.push_back(Constraint.get());
917

918
    if (Tok.isNot(tok::l_paren)) {
919
      Diag(Tok, diag::err_expected_lparen_after) << "asm operand";
920
      SkipUntil(tok::r_paren, StopAtSemi);
921
      return true;
922
    }
923

924
    // Read the parenthesized expression.
925
    BalancedDelimiterTracker T(*this, tok::l_paren);
926
    T.consumeOpen();
927
    ExprResult Res = Actions.CorrectDelayedTyposInExpr(ParseExpression());
928
    T.consumeClose();
929
    if (Res.isInvalid()) {
930
      SkipUntil(tok::r_paren, StopAtSemi);
931
      return true;
932
    }
933
    Exprs.push_back(Res.get());
934
    // Eat the comma and continue parsing if it exists.
935
    if (!TryConsumeToken(tok::comma))
936
      return false;
937
  }
938
}
939

940
const char *Parser::GNUAsmQualifiers::getQualifierName(AQ Qualifier) {
941
  switch (Qualifier) {
942
    case AQ_volatile: return "volatile";
943
    case AQ_inline: return "inline";
944
    case AQ_goto: return "goto";
945
    case AQ_unspecified: return "unspecified";
946
  }
947
  llvm_unreachable("Unknown GNUAsmQualifier");
948
}
949

950
Parser::GNUAsmQualifiers::AQ
951
Parser::getGNUAsmQualifier(const Token &Tok) const {
952
  switch (Tok.getKind()) {
953
    case tok::kw_volatile: return GNUAsmQualifiers::AQ_volatile;
954
    case tok::kw_inline: return GNUAsmQualifiers::AQ_inline;
955
    case tok::kw_goto: return GNUAsmQualifiers::AQ_goto;
956
    default: return GNUAsmQualifiers::AQ_unspecified;
957
  }
958
}
959
bool Parser::GNUAsmQualifiers::setAsmQualifier(AQ Qualifier) {
960
  bool IsDuplicate = Qualifiers & Qualifier;
961
  Qualifiers |= Qualifier;
962
  return IsDuplicate;
963
}
964