1 //===--- SemaStmtAsm.cpp - Semantic Analysis for Asm Statements -----------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 //  This file implements semantic analysis for inline asm statements.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "clang/Sema/SemaInternal.h"
15 #include "clang/Sema/Scope.h"
16 #include "clang/Sema/ScopeInfo.h"
17 #include "clang/Sema/Initialization.h"
18 #include "clang/Sema/Lookup.h"
19 #include "clang/AST/TypeLoc.h"
20 #include "clang/Lex/Preprocessor.h"
21 #include "clang/Basic/TargetInfo.h"
22 #include "llvm/ADT/ArrayRef.h"
23 #include "llvm/ADT/BitVector.h"
24 #include "llvm/ADT/SmallString.h"
25 #include "llvm/MC/MCAsmInfo.h"
26 #include "llvm/MC/MCContext.h"
27 #include "llvm/MC/MCExpr.h"
28 #include "llvm/MC/MCInst.h"
29 #include "llvm/MC/MCInstPrinter.h"
30 #include "llvm/MC/MCInstrInfo.h"
31 #include "llvm/MC/MCObjectFileInfo.h"
32 #include "llvm/MC/MCRegisterInfo.h"
33 #include "llvm/MC/MCStreamer.h"
34 #include "llvm/MC/MCSubtargetInfo.h"
35 #include "llvm/MC/MCSymbol.h"
36 #include "llvm/MC/MCTargetAsmParser.h"
37 #include "llvm/MC/MCParser/MCAsmLexer.h"
38 #include "llvm/MC/MCParser/MCAsmParser.h"
39 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
40 #include "llvm/Support/SourceMgr.h"
41 #include "llvm/Support/TargetRegistry.h"
42 #include "llvm/Support/TargetSelect.h"
43 using namespace clang;
44 using namespace sema;
45 
46 /// CheckAsmLValue - GNU C has an extremely ugly extension whereby they silently
47 /// ignore "noop" casts in places where an lvalue is required by an inline asm.
48 /// We emulate this behavior when -fheinous-gnu-extensions is specified, but
49 /// provide a strong guidance to not use it.
50 ///
51 /// This method checks to see if the argument is an acceptable l-value and
52 /// returns false if it is a case we can handle.
53 static bool CheckAsmLValue(const Expr *E, Sema &S) {
54   // Type dependent expressions will be checked during instantiation.
55   if (E->isTypeDependent())
56     return false;
57 
58   if (E->isLValue())
59     return false;  // Cool, this is an lvalue.
60 
61   // Okay, this is not an lvalue, but perhaps it is the result of a cast that we
62   // are supposed to allow.
63   const Expr *E2 = E->IgnoreParenNoopCasts(S.Context);
64   if (E != E2 && E2->isLValue()) {
65     if (!S.getLangOpts().HeinousExtensions)
66       S.Diag(E2->getLocStart(), diag::err_invalid_asm_cast_lvalue)
67         << E->getSourceRange();
68     else
69       S.Diag(E2->getLocStart(), diag::warn_invalid_asm_cast_lvalue)
70         << E->getSourceRange();
71     // Accept, even if we emitted an error diagnostic.
72     return false;
73   }
74 
75   // None of the above, just randomly invalid non-lvalue.
76   return true;
77 }
78 
79 /// isOperandMentioned - Return true if the specified operand # is mentioned
80 /// anywhere in the decomposed asm string.
81 static bool isOperandMentioned(unsigned OpNo,
82                          ArrayRef<GCCAsmStmt::AsmStringPiece> AsmStrPieces) {
83   for (unsigned p = 0, e = AsmStrPieces.size(); p != e; ++p) {
84     const GCCAsmStmt::AsmStringPiece &Piece = AsmStrPieces[p];
85     if (!Piece.isOperand()) continue;
86 
87     // If this is a reference to the input and if the input was the smaller
88     // one, then we have to reject this asm.
89     if (Piece.getOperandNo() == OpNo)
90       return true;
91   }
92   return false;
93 }
94 
95 StmtResult Sema::ActOnGCCAsmStmt(SourceLocation AsmLoc, bool IsSimple,
96                                  bool IsVolatile, unsigned NumOutputs,
97                                  unsigned NumInputs, IdentifierInfo **Names,
98                                  MultiExprArg constraints, MultiExprArg exprs,
99                                  Expr *asmString, MultiExprArg clobbers,
100                                  SourceLocation RParenLoc) {
101   unsigned NumClobbers = clobbers.size();
102   StringLiteral **Constraints =
103     reinterpret_cast<StringLiteral**>(constraints.data());
104   Expr **Exprs = exprs.data();
105   StringLiteral *AsmString = cast<StringLiteral>(asmString);
106   StringLiteral **Clobbers = reinterpret_cast<StringLiteral**>(clobbers.data());
107 
108   SmallVector<TargetInfo::ConstraintInfo, 4> OutputConstraintInfos;
109 
110   // The parser verifies that there is a string literal here.
111   if (!AsmString->isAscii())
112     return StmtError(Diag(AsmString->getLocStart(),diag::err_asm_wide_character)
113       << AsmString->getSourceRange());
114 
115   for (unsigned i = 0; i != NumOutputs; i++) {
116     StringLiteral *Literal = Constraints[i];
117     if (!Literal->isAscii())
118       return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
119         << Literal->getSourceRange());
120 
121     StringRef OutputName;
122     if (Names[i])
123       OutputName = Names[i]->getName();
124 
125     TargetInfo::ConstraintInfo Info(Literal->getString(), OutputName);
126     if (!Context.getTargetInfo().validateOutputConstraint(Info))
127       return StmtError(Diag(Literal->getLocStart(),
128                             diag::err_asm_invalid_output_constraint)
129                        << Info.getConstraintStr());
130 
131     // Check that the output exprs are valid lvalues.
132     Expr *OutputExpr = Exprs[i];
133     if (CheckAsmLValue(OutputExpr, *this)) {
134       return StmtError(Diag(OutputExpr->getLocStart(),
135                   diag::err_asm_invalid_lvalue_in_output)
136         << OutputExpr->getSourceRange());
137     }
138 
139     OutputConstraintInfos.push_back(Info);
140   }
141 
142   SmallVector<TargetInfo::ConstraintInfo, 4> InputConstraintInfos;
143 
144   for (unsigned i = NumOutputs, e = NumOutputs + NumInputs; i != e; i++) {
145     StringLiteral *Literal = Constraints[i];
146     if (!Literal->isAscii())
147       return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
148         << Literal->getSourceRange());
149 
150     StringRef InputName;
151     if (Names[i])
152       InputName = Names[i]->getName();
153 
154     TargetInfo::ConstraintInfo Info(Literal->getString(), InputName);
155     if (!Context.getTargetInfo().validateInputConstraint(OutputConstraintInfos.data(),
156                                                 NumOutputs, Info)) {
157       return StmtError(Diag(Literal->getLocStart(),
158                             diag::err_asm_invalid_input_constraint)
159                        << Info.getConstraintStr());
160     }
161 
162     Expr *InputExpr = Exprs[i];
163 
164     // Only allow void types for memory constraints.
165     if (Info.allowsMemory() && !Info.allowsRegister()) {
166       if (CheckAsmLValue(InputExpr, *this))
167         return StmtError(Diag(InputExpr->getLocStart(),
168                               diag::err_asm_invalid_lvalue_in_input)
169                          << Info.getConstraintStr()
170                          << InputExpr->getSourceRange());
171     }
172 
173     if (Info.allowsRegister()) {
174       if (InputExpr->getType()->isVoidType()) {
175         return StmtError(Diag(InputExpr->getLocStart(),
176                               diag::err_asm_invalid_type_in_input)
177           << InputExpr->getType() << Info.getConstraintStr()
178           << InputExpr->getSourceRange());
179       }
180     }
181 
182     ExprResult Result = DefaultFunctionArrayLvalueConversion(Exprs[i]);
183     if (Result.isInvalid())
184       return StmtError();
185 
186     Exprs[i] = Result.take();
187     InputConstraintInfos.push_back(Info);
188   }
189 
190   // Check that the clobbers are valid.
191   for (unsigned i = 0; i != NumClobbers; i++) {
192     StringLiteral *Literal = Clobbers[i];
193     if (!Literal->isAscii())
194       return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
195         << Literal->getSourceRange());
196 
197     StringRef Clobber = Literal->getString();
198 
199     if (!Context.getTargetInfo().isValidClobber(Clobber))
200       return StmtError(Diag(Literal->getLocStart(),
201                   diag::err_asm_unknown_register_name) << Clobber);
202   }
203 
204   GCCAsmStmt *NS =
205     new (Context) GCCAsmStmt(Context, AsmLoc, IsSimple, IsVolatile, NumOutputs,
206                              NumInputs, Names, Constraints, Exprs, AsmString,
207                              NumClobbers, Clobbers, RParenLoc);
208   // Validate the asm string, ensuring it makes sense given the operands we
209   // have.
210   SmallVector<GCCAsmStmt::AsmStringPiece, 8> Pieces;
211   unsigned DiagOffs;
212   if (unsigned DiagID = NS->AnalyzeAsmString(Pieces, Context, DiagOffs)) {
213     Diag(getLocationOfStringLiteralByte(AsmString, DiagOffs), DiagID)
214            << AsmString->getSourceRange();
215     return StmtError();
216   }
217 
218   // Validate tied input operands for type mismatches.
219   for (unsigned i = 0, e = InputConstraintInfos.size(); i != e; ++i) {
220     TargetInfo::ConstraintInfo &Info = InputConstraintInfos[i];
221 
222     // If this is a tied constraint, verify that the output and input have
223     // either exactly the same type, or that they are int/ptr operands with the
224     // same size (int/long, int*/long, are ok etc).
225     if (!Info.hasTiedOperand()) continue;
226 
227     unsigned TiedTo = Info.getTiedOperand();
228     unsigned InputOpNo = i+NumOutputs;
229     Expr *OutputExpr = Exprs[TiedTo];
230     Expr *InputExpr = Exprs[InputOpNo];
231 
232     if (OutputExpr->isTypeDependent() || InputExpr->isTypeDependent())
233       continue;
234 
235     QualType InTy = InputExpr->getType();
236     QualType OutTy = OutputExpr->getType();
237     if (Context.hasSameType(InTy, OutTy))
238       continue;  // All types can be tied to themselves.
239 
240     // Decide if the input and output are in the same domain (integer/ptr or
241     // floating point.
242     enum AsmDomain {
243       AD_Int, AD_FP, AD_Other
244     } InputDomain, OutputDomain;
245 
246     if (InTy->isIntegerType() || InTy->isPointerType())
247       InputDomain = AD_Int;
248     else if (InTy->isRealFloatingType())
249       InputDomain = AD_FP;
250     else
251       InputDomain = AD_Other;
252 
253     if (OutTy->isIntegerType() || OutTy->isPointerType())
254       OutputDomain = AD_Int;
255     else if (OutTy->isRealFloatingType())
256       OutputDomain = AD_FP;
257     else
258       OutputDomain = AD_Other;
259 
260     // They are ok if they are the same size and in the same domain.  This
261     // allows tying things like:
262     //   void* to int*
263     //   void* to int            if they are the same size.
264     //   double to long double   if they are the same size.
265     //
266     uint64_t OutSize = Context.getTypeSize(OutTy);
267     uint64_t InSize = Context.getTypeSize(InTy);
268     if (OutSize == InSize && InputDomain == OutputDomain &&
269         InputDomain != AD_Other)
270       continue;
271 
272     // If the smaller input/output operand is not mentioned in the asm string,
273     // then we can promote the smaller one to a larger input and the asm string
274     // won't notice.
275     bool SmallerValueMentioned = false;
276 
277     // If this is a reference to the input and if the input was the smaller
278     // one, then we have to reject this asm.
279     if (isOperandMentioned(InputOpNo, Pieces)) {
280       // This is a use in the asm string of the smaller operand.  Since we
281       // codegen this by promoting to a wider value, the asm will get printed
282       // "wrong".
283       SmallerValueMentioned |= InSize < OutSize;
284     }
285     if (isOperandMentioned(TiedTo, Pieces)) {
286       // If this is a reference to the output, and if the output is the larger
287       // value, then it's ok because we'll promote the input to the larger type.
288       SmallerValueMentioned |= OutSize < InSize;
289     }
290 
291     // If the smaller value wasn't mentioned in the asm string, and if the
292     // output was a register, just extend the shorter one to the size of the
293     // larger one.
294     if (!SmallerValueMentioned && InputDomain != AD_Other &&
295         OutputConstraintInfos[TiedTo].allowsRegister())
296       continue;
297 
298     // Either both of the operands were mentioned or the smaller one was
299     // mentioned.  One more special case that we'll allow: if the tied input is
300     // integer, unmentioned, and is a constant, then we'll allow truncating it
301     // down to the size of the destination.
302     if (InputDomain == AD_Int && OutputDomain == AD_Int &&
303         !isOperandMentioned(InputOpNo, Pieces) &&
304         InputExpr->isEvaluatable(Context)) {
305       CastKind castKind =
306         (OutTy->isBooleanType() ? CK_IntegralToBoolean : CK_IntegralCast);
307       InputExpr = ImpCastExprToType(InputExpr, OutTy, castKind).take();
308       Exprs[InputOpNo] = InputExpr;
309       NS->setInputExpr(i, InputExpr);
310       continue;
311     }
312 
313     Diag(InputExpr->getLocStart(),
314          diag::err_asm_tying_incompatible_types)
315       << InTy << OutTy << OutputExpr->getSourceRange()
316       << InputExpr->getSourceRange();
317     return StmtError();
318   }
319 
320   return Owned(NS);
321 }
322 
323 // isMSAsmKeyword - Return true if this is an MS-style inline asm keyword. These
324 // require special handling.
325 static bool isMSAsmKeyword(StringRef Name) {
326   bool Ret = llvm::StringSwitch<bool>(Name)
327     .Cases("EVEN", "ALIGN", true) // Alignment directives.
328     .Cases("LENGTH", "SIZE", "TYPE", true) // Type and variable sizes.
329     .Case("_emit", true) // _emit Pseudoinstruction.
330     .Default(false);
331   return Ret;
332 }
333 
334 // getIdentifierInfo - Given a Name and a range of tokens, find the associated
335 // IdentifierInfo*.
336 static IdentifierInfo *getIdentifierInfo(StringRef Name,
337                                          ArrayRef<Token> AsmToks,
338                                          unsigned Begin, unsigned End) {
339   for (unsigned i = Begin; i <= End; ++i) {
340     IdentifierInfo *II = AsmToks[i].getIdentifierInfo();
341     if (II && II->getName() == Name)
342       return II;
343   }
344   return 0;
345 }
346 
347 // getSpelling - Get the spelling of the AsmTok token.
348 static StringRef getSpelling(Sema &SemaRef, Token AsmTok) {
349   StringRef Asm;
350   SmallString<512> TokenBuf;
351   TokenBuf.resize(512);
352   bool StringInvalid = false;
353   Asm = SemaRef.PP.getSpelling(AsmTok, TokenBuf, &StringInvalid);
354   assert (!StringInvalid && "Expected valid string!");
355   return Asm;
356 }
357 
358 // Determine if we should bail on this MSAsm instruction.
359 static bool bailOnMSAsm(std::vector<StringRef> Piece) {
360   for (unsigned i = 0, e = Piece.size(); i != e; ++i)
361     if (isMSAsmKeyword(Piece[i]))
362       return true;
363   return false;
364 }
365 
366 // Determine if we should bail on this MSAsm block.
367 static bool bailOnMSAsm(std::vector<std::vector<StringRef> > Pieces) {
368   for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
369     if (bailOnMSAsm(Pieces[i]))
370       return true;
371   return false;
372 }
373 
374 // Determine if this is a simple MSAsm instruction.
375 static bool isSimpleMSAsm(std::vector<StringRef> &Pieces,
376                           const TargetInfo &TI) {
377   if (isMSAsmKeyword(Pieces[0]))
378       return false;
379 
380   for (unsigned i = 1, e = Pieces.size(); i != e; ++i)
381     if (!TI.isValidGCCRegisterName(Pieces[i]))
382       return false;
383   return true;
384 }
385 
386 // Determine if this is a simple MSAsm block.
387 static bool isSimpleMSAsm(std::vector<std::vector<StringRef> > Pieces,
388                           const TargetInfo &TI) {
389   for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
390     if (!isSimpleMSAsm(Pieces[i], TI))
391       return false;
392   return true;
393 }
394 
395 // Break the AsmSting into pieces (i.e., mnemonic and operands).
396 static void buildMSAsmPieces(StringRef Asm, std::vector<StringRef> &Pieces) {
397   std::pair<StringRef,StringRef> Split = Asm.split(' ');
398 
399   // Mnemonic
400   Pieces.push_back(Split.first);
401   Asm = Split.second;
402 
403   // Operands
404   while (!Asm.empty()) {
405     Split = Asm.split(", ");
406     Pieces.push_back(Split.first);
407     Asm = Split.second;
408   }
409 }
410 
411 static void buildMSAsmPieces(std::vector<std::string> &AsmStrings,
412                              std::vector<std::vector<StringRef> > &Pieces) {
413   for (unsigned i = 0, e = AsmStrings.size(); i != e; ++i)
414     buildMSAsmPieces(AsmStrings[i], Pieces[i]);
415 }
416 
417 // Build the individual assembly instruction(s) and place them in the AsmStrings
418 // vector.  These strings are fed to the AsmParser.
419 static void buildMSAsmStrings(Sema &SemaRef, ArrayRef<Token> AsmToks,
420                               std::vector<std::string> &AsmStrings,
421                      std::vector<std::pair<unsigned,unsigned> > &AsmTokRanges) {
422   assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");
423 
424   SmallString<512> Asm;
425   unsigned startTok = 0;
426   for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
427     bool isNewAsm = i == 0 || AsmToks[i].isAtStartOfLine() ||
428       AsmToks[i].is(tok::kw_asm);
429 
430     if (isNewAsm) {
431       if (i) {
432         AsmStrings.push_back(Asm.str());
433         AsmTokRanges.push_back(std::make_pair(startTok, i-1));
434         startTok = i;
435         Asm.clear();
436       }
437       if (AsmToks[i].is(tok::kw_asm)) {
438         i++; // Skip __asm
439         assert (i != e && "Expected another token");
440       }
441     }
442 
443     if (i && AsmToks[i].hasLeadingSpace() && !isNewAsm)
444       Asm += ' ';
445 
446     StringRef Spelling = getSpelling(SemaRef, AsmToks[i]);
447     Asm += Spelling;
448   }
449   AsmStrings.push_back(Asm.str());
450   AsmTokRanges.push_back(std::make_pair(startTok, AsmToks.size()-1));
451 }
452 
453 #define DEF_SIMPLE_MSASM(STR)                                                \
454   MSAsmStmt *NS =                                                            \
455     new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true,   \
456                             /*IsVolatile*/ true, AsmToks, Inputs, Outputs,   \
457                             InputExprs, OutputExprs, STR, Constraints,       \
458                             Clobbers, EndLoc);
459 
460 StmtResult Sema::ActOnMSAsmStmt(SourceLocation AsmLoc, SourceLocation LBraceLoc,
461                                 ArrayRef<Token> AsmToks,SourceLocation EndLoc) {
462   SmallVector<StringRef, 4> Constraints;
463   std::vector<std::string> InputConstraints;
464   std::vector<std::string> OutputConstraints;
465   SmallVector<StringRef, 4> Clobbers;
466   std::set<std::string> ClobberRegs;
467 
468   // FIXME: Use a struct to hold the various expression information.
469   SmallVector<IdentifierInfo*, 4> Inputs;
470   SmallVector<IdentifierInfo*, 4> Outputs;
471   SmallVector<Expr*, 4> InputExprs;
472   SmallVector<Expr*, 4> OutputExprs;
473   SmallVector<std::string, 4> InputExprNames;
474   SmallVector<std::string, 4> OutputExprNames;
475   SmallVector<unsigned, 4> InputExprStrIdx;
476   SmallVector<unsigned, 4> OutputExprStrIdx;
477 
478   // Empty asm statements don't need to instantiate the AsmParser, etc.
479   StringRef EmptyAsmStr;
480   if (AsmToks.empty()) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
481 
482   std::vector<std::string> AsmStrings;
483   std::vector<std::pair<unsigned,unsigned> > AsmTokRanges;
484   buildMSAsmStrings(*this, AsmToks, AsmStrings, AsmTokRanges);
485 
486   std::vector<std::vector<StringRef> > Pieces(AsmStrings.size());
487   buildMSAsmPieces(AsmStrings, Pieces);
488 
489   bool IsSimple = isSimpleMSAsm(Pieces, Context.getTargetInfo());
490 
491   // AsmParser doesn't fully support these asm statements.
492   if (bailOnMSAsm(Pieces)) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
493 
494   // Initialize targets and assembly printers/parsers.
495   llvm::InitializeAllTargetInfos();
496   llvm::InitializeAllTargetMCs();
497   llvm::InitializeAllAsmParsers();
498 
499   // Get the target specific parser.
500   std::string Error;
501   const std::string &TT = Context.getTargetInfo().getTriple().getTriple();
502   const llvm::Target *TheTarget(llvm::TargetRegistry::lookupTarget(TT, Error));
503 
504   OwningPtr<llvm::MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TT));
505   OwningPtr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
506   OwningPtr<llvm::MCObjectFileInfo> MOFI(new llvm::MCObjectFileInfo());
507   OwningPtr<llvm::MCSubtargetInfo>
508     STI(TheTarget->createMCSubtargetInfo(TT, "", ""));
509 
510   for (unsigned StrIdx = 0, e = AsmStrings.size(); StrIdx != e; ++StrIdx) {
511     llvm::SourceMgr SrcMgr;
512     llvm::MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
513     llvm::MemoryBuffer *Buffer =
514       llvm::MemoryBuffer::getMemBuffer(AsmStrings[StrIdx], "<inline asm>");
515 
516     // Tell SrcMgr about this buffer, which is what the parser will pick up.
517     SrcMgr.AddNewSourceBuffer(Buffer, llvm::SMLoc());
518 
519     OwningPtr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
520     OwningPtr<llvm::MCAsmParser>
521       Parser(createMCAsmParser(SrcMgr, Ctx, *Str.get(), *MAI));
522     OwningPtr<llvm::MCTargetAsmParser>
523       TargetParser(TheTarget->createMCAsmParser(*STI, *Parser));
524     // Change to the Intel dialect.
525     Parser->setAssemblerDialect(1);
526     Parser->setTargetParser(*TargetParser.get());
527 
528     // Prime the lexer.
529     Parser->Lex();
530 
531     // Parse the opcode.
532     StringRef IDVal;
533     Parser->ParseIdentifier(IDVal);
534 
535     // Canonicalize the opcode to lower case.
536     SmallString<128> OpcodeStr;
537     for (unsigned i = 0, e = IDVal.size(); i != e; ++i)
538       OpcodeStr.push_back(tolower(IDVal[i]));
539 
540     // Parse the operands.
541     llvm::SMLoc IDLoc;
542     SmallVector<llvm::MCParsedAsmOperand*, 8> Operands;
543     bool HadError = TargetParser->ParseInstruction(OpcodeStr.str(), IDLoc,
544                                                    Operands);
545     // If we had an error parsing the operands, fail gracefully.
546     if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
547 
548     // Match the MCInstr.
549     unsigned Kind;
550     unsigned ErrorInfo;
551     SmallVector<llvm::MCInst, 2> Instrs;
552     HadError = TargetParser->MatchInstruction(IDLoc, Kind, Operands, Instrs,
553                                               ErrorInfo,
554                                               /*matchingInlineAsm*/ true);
555     // If we had an error parsing the operands, fail gracefully.
556     if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
557 
558     // Get the instruction descriptor.
559     llvm::MCInst Inst = Instrs[0];
560     const llvm::MCInstrInfo *MII = TheTarget->createMCInstrInfo();
561     const llvm::MCInstrDesc &Desc = MII->get(Inst.getOpcode());
562     llvm::MCInstPrinter *IP =
563       TheTarget->createMCInstPrinter(1, *MAI, *MII, *MRI, *STI);
564 
565     // Build the list of clobbers, outputs and inputs.
566     unsigned NumDefs = Desc.getNumDefs();
567     for (unsigned i = 1, e = Operands.size(); i != e; ++i) {
568       if (Operands[i]->isToken() || Operands[i]->isImm())
569         continue;
570 
571       // FIXME: The getMCInstOperandNum() function does not work with tied
572       // operands or custom converters.
573       unsigned NumMCOperands;
574       unsigned MCIdx = TargetParser->getMCInstOperandNum(Kind, Inst, Operands,
575                                                          i, NumMCOperands);
576       assert (NumMCOperands && "Expected at least 1 MCOperand!");
577 
578       for (unsigned j = MCIdx, e = MCIdx + NumMCOperands; j != e; ++j) {
579         const llvm::MCOperand &Op = Inst.getOperand(j);
580 
581         // Skip immediates.
582         if (Op.isImm() || Op.isFPImm())
583           continue;
584 
585         // Skip invalid register operands.
586         if (Op.isReg() && Op.getReg() == 0)
587           continue;
588 
589         // Register/Clobber.
590         if (Op.isReg() && NumDefs && (j < NumDefs)) {
591           std::string Reg;
592           llvm::raw_string_ostream OS(Reg);
593           IP->printRegName(OS, Op.getReg());
594 
595           StringRef Clobber(OS.str());
596           if (!Context.getTargetInfo().isValidClobber(Clobber))
597             return StmtError(
598               Diag(AsmLoc, diag::err_asm_unknown_register_name) << Clobber);
599           ClobberRegs.insert(Reg);
600           continue;
601         }
602         // Expr/Input or Output.
603         if (Op.isExpr()) {
604           const llvm::MCExpr *Expr = Op.getExpr();
605           const llvm::MCSymbolRefExpr *SymRef;
606           if ((SymRef = dyn_cast<llvm::MCSymbolRefExpr>(Expr))) {
607             StringRef Name = SymRef->getSymbol().getName();
608             IdentifierInfo *II = getIdentifierInfo(Name, AsmToks,
609                                                    AsmTokRanges[StrIdx].first,
610                                                    AsmTokRanges[StrIdx].second);
611             if (II) {
612               CXXScopeSpec SS;
613               UnqualifiedId Id;
614               SourceLocation Loc;
615               Id.setIdentifier(II, AsmLoc);
616               ExprResult Result = ActOnIdExpression(getCurScope(), SS, Loc, Id,
617                                                     false, false);
618               if (!Result.isInvalid()) {
619                 // FIXME: Determine the proper constraints.
620                 bool isMemDef = (i == 1) && Desc.mayStore();
621                 if (isMemDef) {
622                   Outputs.push_back(II);
623                   OutputExprs.push_back(Result.take());
624                   OutputExprNames.push_back(Name.str());
625                   OutputExprStrIdx.push_back(StrIdx);
626                   OutputConstraints.push_back("=r");
627                 } else {
628                   Inputs.push_back(II);
629                   InputExprs.push_back(Result.take());
630                   InputExprNames.push_back(Name.str());
631                   InputExprStrIdx.push_back(StrIdx);
632                   InputConstraints.push_back("r");
633                 }
634               }
635             }
636           }
637         }
638       }
639     }
640   }
641   for (std::set<std::string>::iterator I = ClobberRegs.begin(),
642          E = ClobberRegs.end(); I != E; ++I)
643     Clobbers.push_back(*I);
644 
645   // Merge the output and input constraints.  Output constraints are expected
646   // first.
647   for (std::vector<std::string>::iterator I = OutputConstraints.begin(),
648          E = OutputConstraints.end(); I != E; ++I)
649     Constraints.push_back(*I);
650 
651   for (std::vector<std::string>::iterator I = InputConstraints.begin(),
652          E = InputConstraints.end(); I != E; ++I)
653     Constraints.push_back(*I);
654 
655   // Enumerate the AsmString expressions.
656   unsigned OpNum = 0;
657   for (unsigned i = 0, e = OutputExprNames.size(); i != e; ++i, ++OpNum) {
658     unsigned StrIdx = OutputExprStrIdx[i];
659     // Iterate over the assembly instruction pieces, skipping the mnemonic.
660     for (unsigned j = 1, f = Pieces[StrIdx].size(); j != f; ++j) {
661       // If the operand and the expression name match, then rewrite the operand.
662       if (OutputExprNames[i] == Pieces[StrIdx][j]) {
663         SmallString<32> Res;
664         llvm::raw_svector_ostream OS(Res);
665         OS << '$' << OpNum;
666         OutputExprNames[i] = OS.str();
667         Pieces[StrIdx][j] = OutputExprNames[i];
668         break;
669       }
670       // Check to see if the expression is a substring of the asm piece.
671       std::pair< StringRef, StringRef > Split =	Pieces[StrIdx][j].split(' ');
672       bool isKeyword = llvm::StringSwitch<bool>(Split.first)
673         .Cases("BYTE", "byte", "WORD", "word", "DWORD", true)
674         .Cases("dword", "QWORD", "qword", "XWORD", "xword", true)
675         .Cases("XMMWORD", "xmmword", "YMMWORD", "ymmword", true)
676         .Default(false);
677       if (isKeyword &&
678           Split.second.find_first_of(OutputExprNames[i]) != StringRef::npos) {
679         // Is is a substring, do the replacement.
680         SmallString<32> Res;
681         llvm::raw_svector_ostream OS(Res);
682         OS << '$' << OpNum;
683         std::string piece = Pieces[StrIdx][j].str();
684         size_t found = piece.find(InputExprNames[i]);
685         piece.replace(found, InputExprNames[i].size(), OS.str());
686         OutputExprNames[i] = piece;
687         Pieces[StrIdx][j] = OutputExprNames[i];
688         break;
689       }
690     }
691   }
692   for (unsigned i = 0, e = InputExprNames.size(); i != e; ++i, ++OpNum) {
693     unsigned StrIdx = InputExprStrIdx[i];
694     // Iterate over the assembly instruction pieces, skipping the mnemonic.
695     for (unsigned j = 1, f = Pieces[StrIdx].size(); j != f; ++j) {
696       // If the operand and the expression name match, then rewrite the operand.
697       if (InputExprNames[i] == Pieces[StrIdx][j]) {
698         SmallString<32> Res;
699         llvm::raw_svector_ostream OS(Res);
700         OS << '$' << OpNum;
701         InputExprNames[i] = OS.str();
702         Pieces[StrIdx][j] = InputExprNames[i];
703         break;
704       }
705       // Check to see if the expression is a substring of the asm piece.
706       std::pair< StringRef, StringRef > Split =	Pieces[StrIdx][j].split(' ');
707       bool isKeyword = llvm::StringSwitch<bool>(Split.first)
708         .Cases("BYTE", "byte", "WORD", "word", "DWORD", true)
709         .Cases("dword", "QWORD", "qword", "XWORD", "xword", true)
710         .Cases("XMMWORD", "xmmword", "YMMWORD", "ymmword", true)
711         .Default(false);
712       if (isKeyword &&
713           Split.second.find_first_of(InputExprNames[i]) != StringRef::npos) {
714         // It is a substring, do the replacement.
715         SmallString<32> Res;
716         llvm::raw_svector_ostream OS(Res);
717         OS << '$' << OpNum;
718         std::string piece = Pieces[StrIdx][j].str();
719         size_t found = piece.find(InputExprNames[i]);
720         piece.replace(found, InputExprNames[i].size(), OS.str());
721         InputExprNames[i] = piece;
722         Pieces[StrIdx][j] = InputExprNames[i];
723         break;
724       }
725     }
726   }
727 
728   // Emit the IR assembly string.
729   std::string AsmString;
730   for (unsigned i = 0, e = Pieces.size(); i != e; ++i) {
731     // Skip empty asm stmts.
732     if (Pieces[i].empty()) continue;
733 
734     if (i > 0)
735       AsmString += "\n\t";
736 
737     // Emit the mnemonic.
738     AsmString += Pieces[i][0];
739     if (Pieces[i].size() > 1)
740       AsmString += ' ';
741 
742     // Emit the operands adding $$ to constants.
743     for (unsigned j = 1, f = Pieces[i].size(); j != f; ++j) {
744       if (j > 1) AsmString += ", ";
745       unsigned Val;
746       if (!Pieces[i][j].getAsInteger(0, Val))
747         AsmString += "$$";
748 
749       AsmString += Pieces[i][j];
750     }
751   }
752 
753   MSAsmStmt *NS =
754     new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, IsSimple,
755                             /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
756                             InputExprs, OutputExprs, AsmString, Constraints,
757                             Clobbers, EndLoc);
758   return Owned(NS);
759 }
760