SemaStmtAsm.cpp revision 153f8ecb5169347e41a0734ee1698498d88b96c5 
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/MCInst.h"
28#include "llvm/MC/MCInstPrinter.h"
29#include "llvm/MC/MCInstrInfo.h"
30#include "llvm/MC/MCObjectFileInfo.h"
31#include "llvm/MC/MCRegisterInfo.h"
32#include "llvm/MC/MCStreamer.h"
33#include "llvm/MC/MCSubtargetInfo.h"
34#include "llvm/MC/MCTargetAsmParser.h"
35#include "llvm/MC/MCParser/MCAsmLexer.h"
36#include "llvm/MC/MCParser/MCAsmParser.h"
37#include "llvm/Support/SourceMgr.h"
38#include "llvm/Support/TargetRegistry.h"
39#include "llvm/Support/TargetSelect.h"
40using namespace clang;
41using namespace sema;
42
43/// CheckAsmLValue - GNU C has an extremely ugly extension whereby they silently
44/// ignore "noop" casts in places where an lvalue is required by an inline asm.
45/// We emulate this behavior when -fheinous-gnu-extensions is specified, but
46/// provide a strong guidance to not use it.
47///
48/// This method checks to see if the argument is an acceptable l-value and
49/// returns false if it is a case we can handle.
50static bool CheckAsmLValue(const Expr *E, Sema &S) {
51  // Type dependent expressions will be checked during instantiation.
52  if (E->isTypeDependent())
53    return false;
54
55  if (E->isLValue())
56    return false;  // Cool, this is an lvalue.
57
58  // Okay, this is not an lvalue, but perhaps it is the result of a cast that we
59  // are supposed to allow.
60  const Expr *E2 = E->IgnoreParenNoopCasts(S.Context);
61  if (E != E2 && E2->isLValue()) {
62    if (!S.getLangOpts().HeinousExtensions)
63      S.Diag(E2->getLocStart(), diag::err_invalid_asm_cast_lvalue)
64        << E->getSourceRange();
65    else
66      S.Diag(E2->getLocStart(), diag::warn_invalid_asm_cast_lvalue)
67        << E->getSourceRange();
68    // Accept, even if we emitted an error diagnostic.
69    return false;
70  }
71
72  // None of the above, just randomly invalid non-lvalue.
73  return true;
74}
75
76/// isOperandMentioned - Return true if the specified operand # is mentioned
77/// anywhere in the decomposed asm string.
78static bool isOperandMentioned(unsigned OpNo,
79                         ArrayRef<AsmStmt::AsmStringPiece> AsmStrPieces) {
80  for (unsigned p = 0, e = AsmStrPieces.size(); p != e; ++p) {
81    const AsmStmt::AsmStringPiece &Piece = AsmStrPieces[p];
82    if (!Piece.isOperand()) continue;
83
84    // If this is a reference to the input and if the input was the smaller
85    // one, then we have to reject this asm.
86    if (Piece.getOperandNo() == OpNo)
87      return true;
88  }
89  return false;
90}
91
92StmtResult Sema::ActOnAsmStmt(SourceLocation AsmLoc, bool IsSimple,
93                              bool IsVolatile, unsigned NumOutputs,
94                              unsigned NumInputs, IdentifierInfo **Names,
95                              MultiExprArg constraints, MultiExprArg exprs,
96                              Expr *asmString, MultiExprArg clobbers,
97                              SourceLocation RParenLoc) {
98  unsigned NumClobbers = clobbers.size();
99  StringLiteral **Constraints =
100    reinterpret_cast<StringLiteral**>(constraints.get());
101  Expr **Exprs = exprs.get();
102  StringLiteral *AsmString = cast<StringLiteral>(asmString);
103  StringLiteral **Clobbers = reinterpret_cast<StringLiteral**>(clobbers.get());
104
105  SmallVector<TargetInfo::ConstraintInfo, 4> OutputConstraintInfos;
106
107  // The parser verifies that there is a string literal here.
108  if (!AsmString->isAscii())
109    return StmtError(Diag(AsmString->getLocStart(),diag::err_asm_wide_character)
110      << AsmString->getSourceRange());
111
112  for (unsigned i = 0; i != NumOutputs; i++) {
113    StringLiteral *Literal = Constraints[i];
114    if (!Literal->isAscii())
115      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
116        << Literal->getSourceRange());
117
118    StringRef OutputName;
119    if (Names[i])
120      OutputName = Names[i]->getName();
121
122    TargetInfo::ConstraintInfo Info(Literal->getString(), OutputName);
123    if (!Context.getTargetInfo().validateOutputConstraint(Info))
124      return StmtError(Diag(Literal->getLocStart(),
125                            diag::err_asm_invalid_output_constraint)
126                       << Info.getConstraintStr());
127
128    // Check that the output exprs are valid lvalues.
129    Expr *OutputExpr = Exprs[i];
130    if (CheckAsmLValue(OutputExpr, *this)) {
131      return StmtError(Diag(OutputExpr->getLocStart(),
132                  diag::err_asm_invalid_lvalue_in_output)
133        << OutputExpr->getSourceRange());
134    }
135
136    OutputConstraintInfos.push_back(Info);
137  }
138
139  SmallVector<TargetInfo::ConstraintInfo, 4> InputConstraintInfos;
140
141  for (unsigned i = NumOutputs, e = NumOutputs + NumInputs; i != e; i++) {
142    StringLiteral *Literal = Constraints[i];
143    if (!Literal->isAscii())
144      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
145        << Literal->getSourceRange());
146
147    StringRef InputName;
148    if (Names[i])
149      InputName = Names[i]->getName();
150
151    TargetInfo::ConstraintInfo Info(Literal->getString(), InputName);
152    if (!Context.getTargetInfo().validateInputConstraint(OutputConstraintInfos.data(),
153                                                NumOutputs, Info)) {
154      return StmtError(Diag(Literal->getLocStart(),
155                            diag::err_asm_invalid_input_constraint)
156                       << Info.getConstraintStr());
157    }
158
159    Expr *InputExpr = Exprs[i];
160
161    // Only allow void types for memory constraints.
162    if (Info.allowsMemory() && !Info.allowsRegister()) {
163      if (CheckAsmLValue(InputExpr, *this))
164        return StmtError(Diag(InputExpr->getLocStart(),
165                              diag::err_asm_invalid_lvalue_in_input)
166                         << Info.getConstraintStr()
167                         << InputExpr->getSourceRange());
168    }
169
170    if (Info.allowsRegister()) {
171      if (InputExpr->getType()->isVoidType()) {
172        return StmtError(Diag(InputExpr->getLocStart(),
173                              diag::err_asm_invalid_type_in_input)
174          << InputExpr->getType() << Info.getConstraintStr()
175          << InputExpr->getSourceRange());
176      }
177    }
178
179    ExprResult Result = DefaultFunctionArrayLvalueConversion(Exprs[i]);
180    if (Result.isInvalid())
181      return StmtError();
182
183    Exprs[i] = Result.take();
184    InputConstraintInfos.push_back(Info);
185  }
186
187  // Check that the clobbers are valid.
188  for (unsigned i = 0; i != NumClobbers; i++) {
189    StringLiteral *Literal = Clobbers[i];
190    if (!Literal->isAscii())
191      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
192        << Literal->getSourceRange());
193
194    StringRef Clobber = Literal->getString();
195
196    if (!Context.getTargetInfo().isValidClobber(Clobber))
197      return StmtError(Diag(Literal->getLocStart(),
198                  diag::err_asm_unknown_register_name) << Clobber);
199  }
200
201  AsmStmt *NS =
202    new (Context) AsmStmt(Context, AsmLoc, IsSimple, IsVolatile, NumOutputs,
203                          NumInputs, Names, Constraints, Exprs, AsmString,
204                          NumClobbers, Clobbers, RParenLoc);
205  // Validate the asm string, ensuring it makes sense given the operands we
206  // have.
207  SmallVector<AsmStmt::AsmStringPiece, 8> Pieces;
208  unsigned DiagOffs;
209  if (unsigned DiagID = NS->AnalyzeAsmString(Pieces, Context, DiagOffs)) {
210    Diag(getLocationOfStringLiteralByte(AsmString, DiagOffs), DiagID)
211           << AsmString->getSourceRange();
212    return StmtError();
213  }
214
215  // Validate tied input operands for type mismatches.
216  for (unsigned i = 0, e = InputConstraintInfos.size(); i != e; ++i) {
217    TargetInfo::ConstraintInfo &Info = InputConstraintInfos[i];
218
219    // If this is a tied constraint, verify that the output and input have
220    // either exactly the same type, or that they are int/ptr operands with the
221    // same size (int/long, int*/long, are ok etc).
222    if (!Info.hasTiedOperand()) continue;
223
224    unsigned TiedTo = Info.getTiedOperand();
225    unsigned InputOpNo = i+NumOutputs;
226    Expr *OutputExpr = Exprs[TiedTo];
227    Expr *InputExpr = Exprs[InputOpNo];
228
229    if (OutputExpr->isTypeDependent() || InputExpr->isTypeDependent())
230      continue;
231
232    QualType InTy = InputExpr->getType();
233    QualType OutTy = OutputExpr->getType();
234    if (Context.hasSameType(InTy, OutTy))
235      continue;  // All types can be tied to themselves.
236
237    // Decide if the input and output are in the same domain (integer/ptr or
238    // floating point.
239    enum AsmDomain {
240      AD_Int, AD_FP, AD_Other
241    } InputDomain, OutputDomain;
242
243    if (InTy->isIntegerType() || InTy->isPointerType())
244      InputDomain = AD_Int;
245    else if (InTy->isRealFloatingType())
246      InputDomain = AD_FP;
247    else
248      InputDomain = AD_Other;
249
250    if (OutTy->isIntegerType() || OutTy->isPointerType())
251      OutputDomain = AD_Int;
252    else if (OutTy->isRealFloatingType())
253      OutputDomain = AD_FP;
254    else
255      OutputDomain = AD_Other;
256
257    // They are ok if they are the same size and in the same domain.  This
258    // allows tying things like:
259    //   void* to int*
260    //   void* to int            if they are the same size.
261    //   double to long double   if they are the same size.
262    //
263    uint64_t OutSize = Context.getTypeSize(OutTy);
264    uint64_t InSize = Context.getTypeSize(InTy);
265    if (OutSize == InSize && InputDomain == OutputDomain &&
266        InputDomain != AD_Other)
267      continue;
268
269    // If the smaller input/output operand is not mentioned in the asm string,
270    // then we can promote the smaller one to a larger input and the asm string
271    // won't notice.
272    bool SmallerValueMentioned = false;
273
274    // If this is a reference to the input and if the input was the smaller
275    // one, then we have to reject this asm.
276    if (isOperandMentioned(InputOpNo, Pieces)) {
277      // This is a use in the asm string of the smaller operand.  Since we
278      // codegen this by promoting to a wider value, the asm will get printed
279      // "wrong".
280      SmallerValueMentioned |= InSize < OutSize;
281    }
282    if (isOperandMentioned(TiedTo, Pieces)) {
283      // If this is a reference to the output, and if the output is the larger
284      // value, then it's ok because we'll promote the input to the larger type.
285      SmallerValueMentioned |= OutSize < InSize;
286    }
287
288    // If the smaller value wasn't mentioned in the asm string, and if the
289    // output was a register, just extend the shorter one to the size of the
290    // larger one.
291    if (!SmallerValueMentioned && InputDomain != AD_Other &&
292        OutputConstraintInfos[TiedTo].allowsRegister())
293      continue;
294
295    // Either both of the operands were mentioned or the smaller one was
296    // mentioned.  One more special case that we'll allow: if the tied input is
297    // integer, unmentioned, and is a constant, then we'll allow truncating it
298    // down to the size of the destination.
299    if (InputDomain == AD_Int && OutputDomain == AD_Int &&
300        !isOperandMentioned(InputOpNo, Pieces) &&
301        InputExpr->isEvaluatable(Context)) {
302      CastKind castKind =
303        (OutTy->isBooleanType() ? CK_IntegralToBoolean : CK_IntegralCast);
304      InputExpr = ImpCastExprToType(InputExpr, OutTy, castKind).take();
305      Exprs[InputOpNo] = InputExpr;
306      NS->setInputExpr(i, InputExpr);
307      continue;
308    }
309
310    Diag(InputExpr->getLocStart(),
311         diag::err_asm_tying_incompatible_types)
312      << InTy << OutTy << OutputExpr->getSourceRange()
313      << InputExpr->getSourceRange();
314    return StmtError();
315  }
316
317  return Owned(NS);
318}
319
320// isMSAsmKeyword - Return true if this is an MS-style inline asm keyword. These
321// require special handling.
322static bool isMSAsmKeyword(StringRef Name) {
323  bool Ret = llvm::StringSwitch<bool>(Name)
324    .Cases("EVEN", "ALIGN", true) // Alignment directives.
325    .Cases("LENGTH", "SIZE", "TYPE", true) // Type and variable sizes.
326    .Case("_emit", true) // _emit Pseudoinstruction.
327    .Default(false);
328  return Ret;
329}
330
331static StringRef getSpelling(Sema &SemaRef, Token AsmTok) {
332  StringRef Asm;
333  SmallString<512> TokenBuf;
334  TokenBuf.resize(512);
335  bool StringInvalid = false;
336  Asm = SemaRef.PP.getSpelling(AsmTok, TokenBuf, &StringInvalid);
337  assert (!StringInvalid && "Expected valid string!");
338  return Asm;
339}
340
341static bool bailOnMSAsm(std::vector<StringRef> Piece) {
342  for (unsigned i = 0, e = Piece.size(); i != e; ++i)
343    if (isMSAsmKeyword(Piece[i]))
344      return true;
345  return false;
346}
347
348static bool bailOnMSAsm(std::vector<std::vector<StringRef> > Pieces) {
349  for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
350    if (bailOnMSAsm(Pieces[i]))
351      return true;
352  return false;
353}
354
355static bool isSimpleMSAsm(std::vector<StringRef> &Pieces,
356                          const TargetInfo &TI) {
357  if (isMSAsmKeyword(Pieces[0]))
358      return false;
359
360  for (unsigned i = 1, e = Pieces.size(); i != e; ++i)
361    if (!TI.isValidGCCRegisterName(Pieces[i]))
362      return false;
363  return true;
364}
365
366static bool isSimpleMSAsm(std::vector<std::vector<StringRef> > Pieces,
367                          const TargetInfo &TI) {
368  for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
369    if (!isSimpleMSAsm(Pieces[i], TI))
370      return false;
371  return true;
372}
373
374// Break the AsmSting into pieces.
375static void buildMSAsmPieces(StringRef Asm, std::vector<StringRef> &Pieces) {
376  std::pair<StringRef,StringRef> Split = Asm.split(' ');
377
378  // Mnemonic
379  Pieces.push_back(Split.first);
380  Asm = Split.second;
381
382  // Operands
383  while (!Asm.empty()) {
384    Split = Asm.split(", ");
385    Pieces.push_back(Split.first);
386    Asm = Split.second;
387  }
388}
389
390// Build the unmodified MSAsmString.
391static std::string buildMSAsmString(Sema &SemaRef,
392                                    ArrayRef<Token> AsmToks,
393                                    std::vector<std::string> &AsmStrings) {
394  assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");
395
396  SmallString<512> Res;
397  SmallString<512> Asm;
398  for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
399    bool isNewAsm = i == 0 || AsmToks[i].isAtStartOfLine() ||
400      AsmToks[i].is(tok::kw_asm);
401
402    if (isNewAsm) {
403      if (i) {
404        AsmStrings.push_back(Asm.c_str());
405        Res += Asm;
406        Asm.clear();
407        Res += '\n';
408      }
409      if (AsmToks[i].is(tok::kw_asm)) {
410        i++; // Skip __asm
411        assert (i != e && "Expected another token");
412      }
413    }
414
415    if (i && AsmToks[i].hasLeadingSpace() && !isNewAsm)
416      Asm += ' ';
417
418    Asm += getSpelling(SemaRef, AsmToks[i]);
419  }
420  AsmStrings.push_back(Asm.c_str());
421  Res += Asm;
422  return Res.c_str();
423}
424
425#define DEF_SIMPLE_MSASM                                                   \
426  MSAsmStmt *NS =                                                          \
427    new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true, \
428                            /*IsVolatile*/ true, AsmToks, Inputs, Outputs, \
429                            AsmString, Clobbers, EndLoc);
430
431StmtResult Sema::ActOnMSAsmStmt(SourceLocation AsmLoc,
432                                SourceLocation LBraceLoc,
433                                ArrayRef<Token> AsmToks,
434                                SourceLocation EndLoc) {
435  // MS-style inline assembly is not fully supported, so emit a warning.
436  Diag(AsmLoc, diag::warn_unsupported_msasm);
437  SmallVector<StringRef,4> Clobbers;
438  std::set<std::string> ClobberRegs;
439  SmallVector<IdentifierInfo*, 4> Inputs;
440  SmallVector<IdentifierInfo*, 4> Outputs;
441
442  // Empty asm statements don't need to instantiate the AsmParser, etc.
443  if (AsmToks.empty()) {
444    StringRef AsmString;
445    DEF_SIMPLE_MSASM;
446    return Owned(NS);
447  }
448
449  unsigned NumAsmStrings;
450  std::vector<std::string> AsmStrings;
451  std::string AsmString = buildMSAsmString(*this, AsmToks, AsmStrings);
452  NumAsmStrings = AsmStrings.size();
453
454  std::vector<std::vector<StringRef> > Pieces;
455  Pieces.resize(NumAsmStrings);
456
457
458  for (unsigned i = 0; i != NumAsmStrings; ++i)
459    buildMSAsmPieces(AsmStrings[i], Pieces[i]);
460
461  bool IsSimple = isSimpleMSAsm(Pieces, Context.getTargetInfo());
462
463  // AsmParser doesn't fully support these asm statements.
464  if (bailOnMSAsm(Pieces)) { DEF_SIMPLE_MSASM; return Owned(NS); }
465
466  // Initialize targets and assembly printers/parsers.
467  llvm::InitializeAllTargetInfos();
468  llvm::InitializeAllTargetMCs();
469  llvm::InitializeAllAsmParsers();
470
471  // Get the target specific parser.
472  std::string Error;
473  const std::string &TT = Context.getTargetInfo().getTriple().getTriple();
474  const llvm::Target *TheTarget(llvm::TargetRegistry::lookupTarget(TT, Error));
475
476  OwningPtr<llvm::MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TT));
477  OwningPtr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
478  OwningPtr<llvm::MCObjectFileInfo> MOFI(new llvm::MCObjectFileInfo());
479  OwningPtr<llvm::MCSubtargetInfo>
480    STI(TheTarget->createMCSubtargetInfo(TT, "", ""));
481
482  for (unsigned i = 0, e = AsmStrings.size(); i != e; ++i) {
483    llvm::SourceMgr SrcMgr;
484    llvm::MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
485    llvm::MemoryBuffer *Buffer =
486      llvm::MemoryBuffer::getMemBuffer(AsmStrings[i], "<inline asm>");
487
488    // Tell SrcMgr about this buffer, which is what the parser will pick up.
489    SrcMgr.AddNewSourceBuffer(Buffer, llvm::SMLoc());
490
491    OwningPtr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
492    OwningPtr<llvm::MCAsmParser>
493      Parser(createMCAsmParser(SrcMgr, Ctx, *Str.get(), *MAI));
494    OwningPtr<llvm::MCTargetAsmParser>
495      TargetParser(TheTarget->createMCAsmParser(*STI, *Parser));
496    // Change to the Intel dialect.
497    Parser->setAssemblerDialect(1);
498    Parser->setTargetParser(*TargetParser.get());
499
500    // Prime the lexer.
501    Parser->Lex();
502
503    // Parse the opcode.
504    StringRef IDVal;
505    Parser->ParseIdentifier(IDVal);
506
507    // Canonicalize the opcode to lower case.
508    SmallString<128> Opcode;
509    for (unsigned i = 0, e = IDVal.size(); i != e; ++i)
510      Opcode.push_back(tolower(IDVal[i]));
511
512    // Parse the operands.
513    llvm::SMLoc IDLoc;
514    SmallVector<llvm::MCParsedAsmOperand*, 8> Operands;
515    bool HadError = TargetParser->ParseInstruction(Opcode.str(), IDLoc,
516                                                   Operands);
517    // If we had an error parsing the operands, fail gracefully.
518    if (HadError) { DEF_SIMPLE_MSASM; return Owned(NS); }
519
520    // Match the MCInstr.
521    unsigned ErrorInfo;
522    SmallVector<llvm::MCInst, 2> Instrs;
523    HadError = TargetParser->MatchInstruction(IDLoc, Operands, Instrs,
524                                              ErrorInfo,
525                                              /*matchingInlineAsm*/ true);
526    // If we had an error parsing the operands, fail gracefully.
527    if (HadError) { DEF_SIMPLE_MSASM; return Owned(NS); }
528
529    // Get the instruction descriptor.
530    llvm::MCInst Inst = Instrs[0];
531    const llvm::MCInstrInfo *MII = TheTarget->createMCInstrInfo();
532    const llvm::MCInstrDesc &Desc = MII->get(Inst.getOpcode());
533    llvm::MCInstPrinter *IP =
534      TheTarget->createMCInstPrinter(1, *MAI, *MII, *MRI, *STI);
535
536    // Build the list of clobbers.
537    for (unsigned i = 0, e = Desc.getNumDefs(); i != e; ++i) {
538      const llvm::MCOperand &Op = Inst.getOperand(i);
539      if (!Op.isReg())
540        continue;
541
542      std::string Reg;
543      llvm::raw_string_ostream OS(Reg);
544      IP->printRegName(OS, Op.getReg());
545
546      StringRef Clobber(OS.str());
547      if (!Context.getTargetInfo().isValidClobber(Clobber))
548        return StmtError(Diag(AsmLoc, diag::err_asm_unknown_register_name) <<
549                         Clobber);
550      ClobberRegs.insert(Reg);
551    }
552  }
553  for (std::set<std::string>::iterator I = ClobberRegs.begin(),
554         E = ClobberRegs.end(); I != E; ++I)
555    Clobbers.push_back(*I);
556
557  MSAsmStmt *NS =
558    new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, IsSimple,
559                            /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
560                            AsmString, Clobbers, EndLoc);
561  return Owned(NS);
562}
563