lib/Sema/SemaStmtAsm.cpp

//===--- SemaStmtAsm.cpp - Semantic Analysis for Statements ---------------===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//  This file implements semantic analysis for inline asm statements.
//
//===----------------------------------------------------------------------===//

#include "clang/Sema/SemaInternal.h"
#include "clang/Sema/Scope.h"
#include "clang/Sema/ScopeInfo.h"
#include "clang/Sema/Initialization.h"
#include "clang/Sema/Lookup.h"
#include "clang/AST/TypeLoc.h"
#include "clang/Lex/Preprocessor.h"
#include "clang/Basic/TargetInfo.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/BitVector.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/MC/MCAsmInfo.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCInstPrinter.h"
#include "llvm/MC/MCInstrInfo.h"
#include "llvm/MC/MCObjectFileInfo.h"
#include "llvm/MC/MCRegisterInfo.h"
#include "llvm/MC/MCStreamer.h"
#include "llvm/MC/MCSubtargetInfo.h"
#include "llvm/MC/MCTargetAsmParser.h"
#include "llvm/MC/MCParser/MCAsmLexer.h"
#include "llvm/MC/MCParser/MCAsmParser.h"
#include "llvm/Support/SourceMgr.h"
#include "llvm/Support/TargetRegistry.h"
#include "llvm/Support/TargetSelect.h"
using namespace clang;
using namespace sema;

/// CheckAsmLValue - GNU C has an extremely ugly extension whereby they silently
/// ignore "noop" casts in places where an lvalue is required by an inline asm.
/// We emulate this behavior when -fheinous-gnu-extensions is specified, but
/// provide a strong guidance to not use it.
///
/// This method checks to see if the argument is an acceptable l-value and
/// returns false if it is a case we can handle.
static bool CheckAsmLValue(const Expr *E, Sema &S) {
  // Type dependent expressions will be checked during instantiation.
  if (E->isTypeDependent())
    return false;

  if (E->isLValue())
    return false;  // Cool, this is an lvalue.

  // Okay, this is not an lvalue, but perhaps it is the result of a cast that we
  // are supposed to allow.
  const Expr *E2 = E->IgnoreParenNoopCasts(S.Context);
  if (E != E2 && E2->isLValue()) {
    if (!S.getLangOpts().HeinousExtensions)
      S.Diag(E2->getLocStart(), diag::err_invalid_asm_cast_lvalue)
        << E->getSourceRange();
    else
      S.Diag(E2->getLocStart(), diag::warn_invalid_asm_cast_lvalue)
        << E->getSourceRange();
    // Accept, even if we emitted an error diagnostic.
    return false;
  }

  // None of the above, just randomly invalid non-lvalue.
  return true;
}

/// isOperandMentioned - Return true if the specified operand # is mentioned
/// anywhere in the decomposed asm string.
static bool isOperandMentioned(unsigned OpNo,
                         ArrayRef<AsmStmt::AsmStringPiece> AsmStrPieces) {
  for (unsigned p = 0, e = AsmStrPieces.size(); p != e; ++p) {
    const AsmStmt::AsmStringPiece &Piece = AsmStrPieces[p];
    if (!Piece.isOperand()) continue;

    // If this is a reference to the input and if the input was the smaller
    // one, then we have to reject this asm.
    if (Piece.getOperandNo() == OpNo)
      return true;
  }
  return false;
}

StmtResult Sema::ActOnAsmStmt(SourceLocation AsmLoc, bool IsSimple,
                              bool IsVolatile, unsigned NumOutputs,
                              unsigned NumInputs, IdentifierInfo **Names,
                              MultiExprArg constraints, MultiExprArg exprs,
                              Expr *asmString, MultiExprArg clobbers,
                              SourceLocation RParenLoc, bool MSAsm) {
  unsigned NumClobbers = clobbers.size();
  StringLiteral **Constraints =
    reinterpret_cast<StringLiteral**>(constraints.get());
  Expr **Exprs = exprs.get();
  StringLiteral *AsmString = cast<StringLiteral>(asmString);
  StringLiteral **Clobbers = reinterpret_cast<StringLiteral**>(clobbers.get());

  SmallVector<TargetInfo::ConstraintInfo, 4> OutputConstraintInfos;

  // The parser verifies that there is a string literal here.
  if (!AsmString->isAscii())
    return StmtError(Diag(AsmString->getLocStart(),diag::err_asm_wide_character)
      << AsmString->getSourceRange());

  for (unsigned i = 0; i != NumOutputs; i++) {
    StringLiteral *Literal = Constraints[i];
    if (!Literal->isAscii())
      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
        << Literal->getSourceRange());

    StringRef OutputName;
    if (Names[i])
      OutputName = Names[i]->getName();

    TargetInfo::ConstraintInfo Info(Literal->getString(), OutputName);
    if (!Context.getTargetInfo().validateOutputConstraint(Info))
      return StmtError(Diag(Literal->getLocStart(),
                            diag::err_asm_invalid_output_constraint)
                       << Info.getConstraintStr());

    // Check that the output exprs are valid lvalues.
    Expr *OutputExpr = Exprs[i];
    if (CheckAsmLValue(OutputExpr, *this)) {
      return StmtError(Diag(OutputExpr->getLocStart(),
                  diag::err_asm_invalid_lvalue_in_output)
        << OutputExpr->getSourceRange());
    }

    OutputConstraintInfos.push_back(Info);
  }

  SmallVector<TargetInfo::ConstraintInfo, 4> InputConstraintInfos;

  for (unsigned i = NumOutputs, e = NumOutputs + NumInputs; i != e; i++) {
    StringLiteral *Literal = Constraints[i];
    if (!Literal->isAscii())
      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
        << Literal->getSourceRange());

    StringRef InputName;
    if (Names[i])
      InputName = Names[i]->getName();

    TargetInfo::ConstraintInfo Info(Literal->getString(), InputName);
    if (!Context.getTargetInfo().validateInputConstraint(OutputConstraintInfos.data(),
                                                NumOutputs, Info)) {
      return StmtError(Diag(Literal->getLocStart(),
                            diag::err_asm_invalid_input_constraint)
                       << Info.getConstraintStr());
    }

    Expr *InputExpr = Exprs[i];

    // Only allow void types for memory constraints.
    if (Info.allowsMemory() && !Info.allowsRegister()) {
      if (CheckAsmLValue(InputExpr, *this))
        return StmtError(Diag(InputExpr->getLocStart(),
                              diag::err_asm_invalid_lvalue_in_input)
                         << Info.getConstraintStr()
                         << InputExpr->getSourceRange());
    }

    if (Info.allowsRegister()) {
      if (InputExpr->getType()->isVoidType()) {
        return StmtError(Diag(InputExpr->getLocStart(),
                              diag::err_asm_invalid_type_in_input)
          << InputExpr->getType() << Info.getConstraintStr()
          << InputExpr->getSourceRange());
      }
    }

    ExprResult Result = DefaultFunctionArrayLvalueConversion(Exprs[i]);
    if (Result.isInvalid())
      return StmtError();

    Exprs[i] = Result.take();
    InputConstraintInfos.push_back(Info);
  }

  // Check that the clobbers are valid.
  for (unsigned i = 0; i != NumClobbers; i++) {
    StringLiteral *Literal = Clobbers[i];
    if (!Literal->isAscii())
      return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
        << Literal->getSourceRange());

    StringRef Clobber = Literal->getString();

    if (!Context.getTargetInfo().isValidClobber(Clobber))
      return StmtError(Diag(Literal->getLocStart(),
                  diag::err_asm_unknown_register_name) << Clobber);
  }

  AsmStmt *NS =
    new (Context) AsmStmt(Context, AsmLoc, IsSimple, IsVolatile, MSAsm,
                          NumOutputs, NumInputs, Names, Constraints, Exprs,
                          AsmString, NumClobbers, Clobbers, RParenLoc);
  // Validate the asm string, ensuring it makes sense given the operands we
  // have.
  SmallVector<AsmStmt::AsmStringPiece, 8> Pieces;
  unsigned DiagOffs;
  if (unsigned DiagID = NS->AnalyzeAsmString(Pieces, Context, DiagOffs)) {
    Diag(getLocationOfStringLiteralByte(AsmString, DiagOffs), DiagID)
           << AsmString->getSourceRange();
    return StmtError();
  }

  // Validate tied input operands for type mismatches.
  for (unsigned i = 0, e = InputConstraintInfos.size(); i != e; ++i) {
    TargetInfo::ConstraintInfo &Info = InputConstraintInfos[i];

    // If this is a tied constraint, verify that the output and input have
    // either exactly the same type, or that they are int/ptr operands with the
    // same size (int/long, int*/long, are ok etc).
    if (!Info.hasTiedOperand()) continue;

    unsigned TiedTo = Info.getTiedOperand();
    unsigned InputOpNo = i+NumOutputs;
    Expr *OutputExpr = Exprs[TiedTo];
    Expr *InputExpr = Exprs[InputOpNo];

    if (OutputExpr->isTypeDependent() || InputExpr->isTypeDependent())
      continue;

    QualType InTy = InputExpr->getType();
    QualType OutTy = OutputExpr->getType();
    if (Context.hasSameType(InTy, OutTy))
      continue;  // All types can be tied to themselves.

    // Decide if the input and output are in the same domain (integer/ptr or
    // floating point.
    enum AsmDomain {
      AD_Int, AD_FP, AD_Other
    } InputDomain, OutputDomain;

    if (InTy->isIntegerType() || InTy->isPointerType())
      InputDomain = AD_Int;
    else if (InTy->isRealFloatingType())
      InputDomain = AD_FP;
    else
      InputDomain = AD_Other;

    if (OutTy->isIntegerType() || OutTy->isPointerType())
      OutputDomain = AD_Int;
    else if (OutTy->isRealFloatingType())
      OutputDomain = AD_FP;
    else
      OutputDomain = AD_Other;

    // They are ok if they are the same size and in the same domain.  This
    // allows tying things like:
    //   void* to int*
    //   void* to int            if they are the same size.
    //   double to long double   if they are the same size.
    //
    uint64_t OutSize = Context.getTypeSize(OutTy);
    uint64_t InSize = Context.getTypeSize(InTy);
    if (OutSize == InSize && InputDomain == OutputDomain &&
        InputDomain != AD_Other)
      continue;

    // If the smaller input/output operand is not mentioned in the asm string,
    // then we can promote the smaller one to a larger input and the asm string
    // won't notice.
    bool SmallerValueMentioned = false;

    // If this is a reference to the input and if the input was the smaller
    // one, then we have to reject this asm.
    if (isOperandMentioned(InputOpNo, Pieces)) {
      // This is a use in the asm string of the smaller operand.  Since we
      // codegen this by promoting to a wider value, the asm will get printed
      // "wrong".
      SmallerValueMentioned |= InSize < OutSize;
    }
    if (isOperandMentioned(TiedTo, Pieces)) {
      // If this is a reference to the output, and if the output is the larger
      // value, then it's ok because we'll promote the input to the larger type.
      SmallerValueMentioned |= OutSize < InSize;
    }

    // If the smaller value wasn't mentioned in the asm string, and if the
    // output was a register, just extend the shorter one to the size of the
    // larger one.
    if (!SmallerValueMentioned && InputDomain != AD_Other &&
        OutputConstraintInfos[TiedTo].allowsRegister())
      continue;

    // Either both of the operands were mentioned or the smaller one was
    // mentioned.  One more special case that we'll allow: if the tied input is
    // integer, unmentioned, and is a constant, then we'll allow truncating it
    // down to the size of the destination.
    if (InputDomain == AD_Int && OutputDomain == AD_Int &&
        !isOperandMentioned(InputOpNo, Pieces) &&
        InputExpr->isEvaluatable(Context)) {
      CastKind castKind =
        (OutTy->isBooleanType() ? CK_IntegralToBoolean : CK_IntegralCast);
      InputExpr = ImpCastExprToType(InputExpr, OutTy, castKind).take();
      Exprs[InputOpNo] = InputExpr;
      NS->setInputExpr(i, InputExpr);
      continue;
    }

    Diag(InputExpr->getLocStart(),
         diag::err_asm_tying_incompatible_types)
      << InTy << OutTy << OutputExpr->getSourceRange()
      << InputExpr->getSourceRange();
    return StmtError();
  }

  return Owned(NS);
}

// isMSAsmKeyword - Return true if this is an MS-style inline asm keyword. These
// require special handling.
static bool isMSAsmKeyword(StringRef Name) {
  bool Ret = llvm::StringSwitch<bool>(Name)
    .Cases("EVEN", "ALIGN", true) // Alignment directives.
    .Cases("LENGTH", "SIZE", "TYPE", true) // Type and variable sizes.
    .Case("_emit", true) // _emit Pseudoinstruction.
    .Default(false);
  return Ret;
}

static StringRef getSpelling(Sema &SemaRef, Token AsmTok) {
  StringRef Asm;
  SmallString<512> TokenBuf;
  TokenBuf.resize(512);
  bool StringInvalid = false;
  Asm = SemaRef.PP.getSpelling(AsmTok, TokenBuf, &StringInvalid);
  assert (!StringInvalid && "Expected valid string!");
  return Asm;
}

static void patchMSAsmStrings(Sema &SemaRef, bool &IsSimple,
                              SourceLocation AsmLoc,
                              ArrayRef<Token> AsmToks,
                              const TargetInfo &TI,
                              std::vector<llvm::BitVector> &AsmRegs,
                              std::vector<llvm::BitVector> &AsmNames,
                              std::vector<std::string> &AsmStrings) {
  assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");

  // Assume simple asm stmt until we parse a non-register identifer (or we just
  // need to bail gracefully).
  IsSimple = true;

  SmallString<512> Asm;
  unsigned NumAsmStrings = 0;
  for (unsigned i = 0, e = AsmToks.size(); i != e; ++i) {

    // Determine if this should be considered a new asm.
    bool isNewAsm = i == 0 || AsmToks[i].isAtStartOfLine() ||
      AsmToks[i].is(tok::kw_asm);

    // Emit the previous asm string.
    if (i && isNewAsm) {
      AsmStrings[NumAsmStrings++] = Asm.c_str();
      if (AsmToks[i].is(tok::kw_asm)) {
        ++i; // Skip __asm
        assert (i != e && "Expected another token.");
      }
    }

    // Start a new asm string with the opcode.
    if (isNewAsm) {
      AsmRegs[NumAsmStrings].resize(AsmToks.size());
      AsmNames[NumAsmStrings].resize(AsmToks.size());

      StringRef Piece = AsmToks[i].getIdentifierInfo()->getName();
      // MS-style inline asm keywords require special handling.
      if (isMSAsmKeyword(Piece))
        IsSimple = false;

      // TODO: Verify this is a valid opcode.
      Asm = Piece;
      continue;
    }

    if (i && AsmToks[i].hasLeadingSpace())
      Asm += ' ';

    // Check the operand(s).
    switch (AsmToks[i].getKind()) {
    default:
      IsSimple = false;
      Asm += getSpelling(SemaRef, AsmToks[i]);
      break;
    case tok::comma: Asm += ","; break;
    case tok::colon: Asm += ":"; break;
    case tok::l_square: Asm += "["; break;
    case tok::r_square: Asm += "]"; break;
    case tok::l_brace: Asm += "{"; break;
    case tok::r_brace: Asm += "}"; break;
    case tok::numeric_constant:
      Asm += getSpelling(SemaRef, AsmToks[i]);
      break;
    case tok::identifier: {
      IdentifierInfo *II = AsmToks[i].getIdentifierInfo();
      StringRef Name = II->getName();

      // Valid register?
      if (TI.isValidGCCRegisterName(Name)) {
        AsmRegs[NumAsmStrings].set(i);
        Asm += Name;
        break;
      }

      IsSimple = false;

      // MS-style inline asm keywords require special handling.
      if (isMSAsmKeyword(Name)) {
        IsSimple = false;
        Asm += Name;
        break;
      }

      // FIXME: Why are we missing this segment register?
      if (Name == "fs") {
        Asm += Name;
        break;
      }

      // Lookup the identifier.
      // TODO: Someone with more experience with clang should verify this the
      // proper way of doing a symbol lookup.
      DeclarationName DeclName(II);
      Scope *CurScope = SemaRef.getCurScope();
      LookupResult R(SemaRef, DeclName, AsmLoc, Sema::LookupOrdinaryName);
      if (!SemaRef.LookupName(R, CurScope, false/*AllowBuiltinCreation*/))
        break;

      assert (R.isSingleResult() && "Expected a single result?!");
      NamedDecl *Decl = R.getFoundDecl();
      switch (Decl->getKind()) {
      default:
        assert(0 && "Unknown decl kind.");
        break;
      case Decl::Var: {
      case Decl::ParmVar:
        AsmNames[NumAsmStrings].set(i);

        VarDecl *Var = cast<VarDecl>(Decl);
        QualType Ty = Var->getType();
        (void)Ty; // Avoid warning.
        // TODO: Patch identifier with valid operand.  One potential idea is to
        // probe the backend with type information to guess the possible
        // operand.
        break;
      }
      }
      break;
    }
    }
  }

  // Emit the final (and possibly only) asm string.
  AsmStrings[NumAsmStrings] = Asm.c_str();
}

// Build the unmodified MSAsmString.
static std::string buildMSAsmString(Sema &SemaRef,
                                    ArrayRef<Token> AsmToks,
                                    unsigned &NumAsmStrings) {
  assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");
  NumAsmStrings = 0;

  SmallString<512> Asm;
  for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
    bool isNewAsm = i == 0 || AsmToks[i].isAtStartOfLine() ||
      AsmToks[i].is(tok::kw_asm);

    if (isNewAsm) {
      ++NumAsmStrings;
      if (i)
        Asm += '\n';
      if (AsmToks[i].is(tok::kw_asm)) {
        i++; // Skip __asm
        assert (i != e && "Expected another token");
      }
    }

    if (i && AsmToks[i].hasLeadingSpace() && !isNewAsm)
      Asm += ' ';

    Asm += getSpelling(SemaRef, AsmToks[i]);
  }
  return Asm.c_str();
}

StmtResult Sema::ActOnMSAsmStmt(SourceLocation AsmLoc,
                                SourceLocation LBraceLoc,
                                ArrayRef<Token> AsmToks,
                                SourceLocation EndLoc) {
  // MS-style inline assembly is not fully supported, so emit a warning.
  Diag(AsmLoc, diag::warn_unsupported_msasm);
  SmallVector<StringRef,4> Clobbers;
  std::set<std::string> ClobberRegs;
  SmallVector<IdentifierInfo*, 4> Inputs;
  SmallVector<IdentifierInfo*, 4> Outputs;

  // Empty asm statements don't need to instantiate the AsmParser, etc.
  if (AsmToks.empty()) {
    StringRef AsmString;
    MSAsmStmt *NS =
      new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true,
                              /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
                              AsmString, Clobbers, EndLoc);
    return Owned(NS);
  }

  unsigned NumAsmStrings;
  std::string AsmString = buildMSAsmString(*this, AsmToks, NumAsmStrings);

  bool IsSimple;
  std::vector<llvm::BitVector> Regs;
  std::vector<llvm::BitVector> Names;
  std::vector<std::string> PatchedAsmStrings;

  Regs.resize(NumAsmStrings);
  Names.resize(NumAsmStrings);
  PatchedAsmStrings.resize(NumAsmStrings);

  // Rewrite operands to appease the AsmParser.
  patchMSAsmStrings(*this, IsSimple, AsmLoc, AsmToks,
                    Context.getTargetInfo(), Regs, Names, PatchedAsmStrings);

  // patchMSAsmStrings doesn't correctly patch non-simple asm statements.
  if (!IsSimple) {
    MSAsmStmt *NS =
      new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true,
                              /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
                              AsmString, Clobbers, EndLoc);
    return Owned(NS);
  }

  // Initialize targets and assembly printers/parsers.
  llvm::InitializeAllTargetInfos();
  llvm::InitializeAllTargetMCs();
  llvm::InitializeAllAsmParsers();

  // Get the target specific parser.
  std::string Error;
  const std::string &TT = Context.getTargetInfo().getTriple().getTriple();
  const llvm::Target *TheTarget(llvm::TargetRegistry::lookupTarget(TT, Error));

  OwningPtr<llvm::MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TT));
  OwningPtr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
  OwningPtr<llvm::MCObjectFileInfo> MOFI(new llvm::MCObjectFileInfo());
  OwningPtr<llvm::MCSubtargetInfo>
    STI(TheTarget->createMCSubtargetInfo(TT, "", ""));

  for (unsigned i = 0, e = PatchedAsmStrings.size(); i != e; ++i) {
    llvm::SourceMgr SrcMgr;
    llvm::MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
    llvm::MemoryBuffer *Buffer =
      llvm::MemoryBuffer::getMemBuffer(PatchedAsmStrings[i], "<inline asm>");

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

    OwningPtr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
    OwningPtr<llvm::MCAsmParser>
      Parser(createMCAsmParser(SrcMgr, Ctx, *Str.get(), *MAI));
    OwningPtr<llvm::MCTargetAsmParser>
      TargetParser(TheTarget->createMCAsmParser(*STI, *Parser));
    // Change to the Intel dialect.
    Parser->setAssemblerDialect(1);
    Parser->setTargetParser(*TargetParser.get());

    // Prime the lexer.
    Parser->Lex();

    // Parse the opcode.
    StringRef IDVal;
    Parser->ParseIdentifier(IDVal);

    // Canonicalize the opcode to lower case.
    SmallString<128> Opcode;
    for (unsigned i = 0, e = IDVal.size(); i != e; ++i)
      Opcode.push_back(tolower(IDVal[i]));

    // Parse the operands.
    llvm::SMLoc IDLoc;
    SmallVector<llvm::MCParsedAsmOperand*, 8> Operands;
    bool HadError = TargetParser->ParseInstruction(Opcode.str(), IDLoc,
                                                   Operands);
    assert (!HadError && "Unexpected error parsing instruction");

    // Match the MCInstr.
    SmallVector<llvm::MCInst, 2> Instrs;
    HadError = TargetParser->MatchInstruction(IDLoc, Operands, Instrs);
    assert (!HadError && "Unexpected error matching instruction");
    assert ((Instrs.size() == 1) && "Expected only a single instruction.");

    // Get the instruction descriptor.
    llvm::MCInst Inst = Instrs[0];
    const llvm::MCInstrInfo *MII = TheTarget->createMCInstrInfo();
    const llvm::MCInstrDesc &Desc = MII->get(Inst.getOpcode());
    llvm::MCInstPrinter *IP =
      TheTarget->createMCInstPrinter(1, *MAI, *MII, *MRI, *STI);

    // Build the list of clobbers.
    for (unsigned i = 0, e = Desc.getNumDefs(); i != e; ++i) {
      const llvm::MCOperand &Op = Inst.getOperand(i);
      if (!Op.isReg())
        continue;

      std::string Reg;
      llvm::raw_string_ostream OS(Reg);
      IP->printRegName(OS, Op.getReg());

      StringRef Clobber(OS.str());
      if (!Context.getTargetInfo().isValidClobber(Clobber))
        return StmtError(Diag(AsmLoc, diag::err_asm_unknown_register_name) <<
                         Clobber);
      ClobberRegs.insert(Reg);
    }
  }
  for (std::set<std::string>::iterator I = ClobberRegs.begin(),
         E = ClobberRegs.end(); I != E; ++I)
    Clobbers.push_back(*I);

  MSAsmStmt *NS =
    new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, IsSimple,
                            /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
                            AsmString, Clobbers, EndLoc);
  return Owned(NS);
}