X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FX86%2FAsmParser%2FX86AsmParser.cpp;h=b84bcf9dd7e68d15d3b58b5b1ded3cb260c02a7b;hb=23a4885f59511eb4781e534a3b2a5910733cb357;hp=da9b8b57dd58b9c0ad21480d598cd2bc2da6236a;hpb=c3a9574eed4420985533b7223fda094d3c191f68;p=oota-llvm.git diff --git a/lib/Target/X86/AsmParser/X86AsmParser.cpp b/lib/Target/X86/AsmParser/X86AsmParser.cpp index da9b8b57dd5..b84bcf9dd7e 100644 --- a/lib/Target/X86/AsmParser/X86AsmParser.cpp +++ b/lib/Target/X86/AsmParser/X86AsmParser.cpp @@ -8,7 +8,11 @@ //===----------------------------------------------------------------------===// #include "MCTargetDesc/X86BaseInfo.h" +#include "X86AsmInstrumentation.h" +#include "X86AsmParserCommon.h" +#include "X86Operand.h" #include "llvm/ADT/APFloat.h" +#include "llvm/ADT/STLExtras.h" #include "llvm/ADT/SmallString.h" #include "llvm/ADT/SmallVector.h" #include "llvm/ADT/StringSwitch.h" @@ -16,6 +20,7 @@ #include "llvm/MC/MCContext.h" #include "llvm/MC/MCExpr.h" #include "llvm/MC/MCInst.h" +#include "llvm/MC/MCInstrInfo.h" #include "llvm/MC/MCParser/MCAsmLexer.h" #include "llvm/MC/MCParser/MCAsmParser.h" #include "llvm/MC/MCParser/MCParsedAsmOperand.h" @@ -27,19 +32,23 @@ #include "llvm/Support/SourceMgr.h" #include "llvm/Support/TargetRegistry.h" #include "llvm/Support/raw_ostream.h" +#include using namespace llvm; namespace { -struct X86Operand; static const char OpPrecedence[] = { - 0, // IC_PLUS - 0, // IC_MINUS - 1, // IC_MULTIPLY - 1, // IC_DIVIDE - 2, // IC_RPAREN - 3, // IC_LPAREN + 0, // IC_OR + 1, // IC_AND + 2, // IC_LSHIFT + 2, // IC_RSHIFT + 3, // IC_PLUS + 3, // IC_MINUS + 4, // IC_MULTIPLY + 4, // IC_DIVIDE + 5, // IC_RPAREN + 6, // IC_LPAREN 0, // IC_IMM 0 // IC_REGISTER }; @@ -47,10 +56,22 @@ static const char OpPrecedence[] = { class X86AsmParser : public MCTargetAsmParser { MCSubtargetInfo &STI; MCAsmParser &Parser; + const MCInstrInfo &MII; ParseInstructionInfo *InstInfo; + std::unique_ptr Instrumentation; private: + SMLoc consumeToken() { + SMLoc Result = Parser.getTok().getLoc(); + Parser.Lex(); + return Result; + } + enum InfixCalculatorTok { - IC_PLUS = 0, + IC_OR = 0, + IC_AND, + IC_LSHIFT, + IC_RSHIFT, + IC_PLUS, IC_MINUS, IC_MULTIPLY, IC_DIVIDE, @@ -79,7 +100,7 @@ private: PostfixStack.push_back(std::make_pair(Op, Val)); } - void popOperator() { InfixOperatorStack.pop_back_val(); } + void popOperator() { InfixOperatorStack.pop_back(); } void pushOperator(InfixCalculatorTok Op) { // Push the new operator if the stack is empty. if (InfixOperatorStack.empty()) { @@ -117,12 +138,12 @@ private: if (StackOp == IC_RPAREN) { ++ParenCount; - InfixOperatorStack.pop_back_val(); + InfixOperatorStack.pop_back(); } else if (StackOp == IC_LPAREN) { --ParenCount; - InfixOperatorStack.pop_back_val(); + InfixOperatorStack.pop_back(); } else { - InfixOperatorStack.pop_back_val(); + InfixOperatorStack.pop_back(); PostfixStack.push_back(std::make_pair(StackOp, 0)); } } @@ -175,6 +196,30 @@ private: Val = Op1.second / Op2.second; OperandStack.push_back(std::make_pair(IC_IMM, Val)); break; + case IC_OR: + assert (Op1.first == IC_IMM && Op2.first == IC_IMM && + "Or operation with an immediate and a register!"); + Val = Op1.second | Op2.second; + OperandStack.push_back(std::make_pair(IC_IMM, Val)); + break; + case IC_AND: + assert (Op1.first == IC_IMM && Op2.first == IC_IMM && + "And operation with an immediate and a register!"); + Val = Op1.second & Op2.second; + OperandStack.push_back(std::make_pair(IC_IMM, Val)); + break; + case IC_LSHIFT: + assert (Op1.first == IC_IMM && Op2.first == IC_IMM && + "Left shift operation with an immediate and a register!"); + Val = Op1.second << Op2.second; + OperandStack.push_back(std::make_pair(IC_IMM, Val)); + break; + case IC_RSHIFT: + assert (Op1.first == IC_IMM && Op2.first == IC_IMM && + "Right shift operation with an immediate and a register!"); + Val = Op1.second >> Op2.second; + OperandStack.push_back(std::make_pair(IC_IMM, Val)); + break; } } } @@ -184,6 +229,10 @@ private: }; enum IntelExprState { + IES_OR, + IES_AND, + IES_LSHIFT, + IES_RSHIFT, IES_PLUS, IES_MINUS, IES_MULTIPLY, @@ -206,11 +255,12 @@ private: StringRef SymName; bool StopOnLBrac, AddImmPrefix; InfixCalculator IC; + InlineAsmIdentifierInfo Info; public: IntelExprStateMachine(int64_t imm, bool stoponlbrac, bool addimmprefix) : State(IES_PLUS), PrevState(IES_ERROR), BaseReg(0), IndexReg(0), TmpReg(0), Scale(1), Imm(imm), Sym(0), StopOnLBrac(stoponlbrac), - AddImmPrefix(addimmprefix) {} + AddImmPrefix(addimmprefix) { Info.clear(); } unsigned getBaseReg() { return BaseReg; } unsigned getIndexReg() { return IndexReg; } @@ -218,11 +268,77 @@ private: const MCExpr *getSym() { return Sym; } StringRef getSymName() { return SymName; } int64_t getImm() { return Imm + IC.execute(); } - bool isValidEndState() { return State == IES_RBRAC; } + bool isValidEndState() { + return State == IES_RBRAC || State == IES_INTEGER; + } bool getStopOnLBrac() { return StopOnLBrac; } bool getAddImmPrefix() { return AddImmPrefix; } bool hadError() { return State == IES_ERROR; } + InlineAsmIdentifierInfo &getIdentifierInfo() { + return Info; + } + + void onOr() { + IntelExprState CurrState = State; + switch (State) { + default: + State = IES_ERROR; + break; + case IES_INTEGER: + case IES_RPAREN: + case IES_REGISTER: + State = IES_OR; + IC.pushOperator(IC_OR); + break; + } + PrevState = CurrState; + } + void onAnd() { + IntelExprState CurrState = State; + switch (State) { + default: + State = IES_ERROR; + break; + case IES_INTEGER: + case IES_RPAREN: + case IES_REGISTER: + State = IES_AND; + IC.pushOperator(IC_AND); + break; + } + PrevState = CurrState; + } + void onLShift() { + IntelExprState CurrState = State; + switch (State) { + default: + State = IES_ERROR; + break; + case IES_INTEGER: + case IES_RPAREN: + case IES_REGISTER: + State = IES_LSHIFT; + IC.pushOperator(IC_LSHIFT); + break; + } + PrevState = CurrState; + } + void onRShift() { + IntelExprState CurrState = State; + switch (State) { + default: + State = IES_ERROR; + break; + case IES_INTEGER: + case IES_RPAREN: + case IES_REGISTER: + State = IES_RSHIFT; + IC.pushOperator(IC_RSHIFT); + break; + } + PrevState = CurrState; + } void onPlus() { IntelExprState CurrState = State; switch (State) { @@ -314,7 +430,7 @@ private: } PrevState = CurrState; } - void onDispExpr(const MCExpr *SymRef, StringRef SymRefName) { + void onIdentifierExpr(const MCExpr *SymRef, StringRef SymRefName) { PrevState = State; switch (State) { default: @@ -329,7 +445,7 @@ private: break; } } - void onInteger(int64_t TmpInt) { + bool onInteger(int64_t TmpInt, StringRef &ErrMsg) { IntelExprState CurrState = State; switch (State) { default: @@ -337,6 +453,10 @@ private: break; case IES_PLUS: case IES_MINUS: + case IES_OR: + case IES_AND: + case IES_LSHIFT: + case IES_RSHIFT: case IES_DIVIDE: case IES_MULTIPLY: case IES_LPAREN: @@ -346,9 +466,15 @@ private: assert (!IndexReg && "IndexReg already set!"); IndexReg = TmpReg; Scale = TmpInt; + if(Scale != 1 && Scale != 2 && Scale != 4 && Scale != 8) { + ErrMsg = "scale factor in address must be 1, 2, 4 or 8"; + return true; + } // Get the scale and replace the 'Register * Scale' with '0'. IC.popOperator(); } else if ((PrevState == IES_PLUS || PrevState == IES_MINUS || + PrevState == IES_OR || PrevState == IES_AND || + PrevState == IES_LSHIFT || PrevState == IES_RSHIFT || PrevState == IES_MULTIPLY || PrevState == IES_DIVIDE || PrevState == IES_LPAREN || PrevState == IES_LBRAC) && CurrState == IES_MINUS) { @@ -361,6 +487,7 @@ private: break; } PrevState = CurrState; + return false; } void onStar() { PrevState = State; @@ -434,11 +561,17 @@ private: break; case IES_PLUS: case IES_MINUS: + case IES_OR: + case IES_AND: + case IES_LSHIFT: + case IES_RSHIFT: case IES_MULTIPLY: case IES_DIVIDE: case IES_LPAREN: // FIXME: We don't handle this type of unary minus, yet. if ((PrevState == IES_PLUS || PrevState == IES_MINUS || + PrevState == IES_OR || PrevState == IES_AND || + PrevState == IES_LSHIFT || PrevState == IES_RSHIFT || PrevState == IES_MULTIPLY || PrevState == IES_DIVIDE || PrevState == IES_LPAREN || PrevState == IES_LBRAC) && CurrState == IES_MINUS) { @@ -472,38 +605,49 @@ private: MCAsmLexer &getLexer() const { return Parser.getLexer(); } bool Error(SMLoc L, const Twine &Msg, - ArrayRef Ranges = ArrayRef(), + ArrayRef Ranges = None, bool MatchingInlineAsm = false) { if (MatchingInlineAsm) return true; return Parser.Error(L, Msg, Ranges); } + bool ErrorAndEatStatement(SMLoc L, const Twine &Msg, + ArrayRef Ranges = None, + bool MatchingInlineAsm = false) { + Parser.eatToEndOfStatement(); + return Error(L, Msg, Ranges, MatchingInlineAsm); + } + X86Operand *ErrorOperand(SMLoc Loc, StringRef Msg) { Error(Loc, Msg); return 0; } + X86Operand *DefaultMemSIOperand(SMLoc Loc); + X86Operand *DefaultMemDIOperand(SMLoc Loc); X86Operand *ParseOperand(); X86Operand *ParseATTOperand(); X86Operand *ParseIntelOperand(); X86Operand *ParseIntelOffsetOfOperator(); + bool ParseIntelDotOperator(const MCExpr *Disp, const MCExpr *&NewDisp); X86Operand *ParseIntelOperator(unsigned OpKind); - X86Operand *ParseIntelMemOperand(unsigned SegReg, int64_t ImmDisp, - SMLoc StartLoc); - X86Operand *ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End); + X86Operand *ParseIntelSegmentOverride(unsigned SegReg, SMLoc Start, unsigned Size); + X86Operand *ParseIntelMemOperand(int64_t ImmDisp, SMLoc StartLoc, + unsigned Size); + bool ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End); X86Operand *ParseIntelBracExpression(unsigned SegReg, SMLoc Start, int64_t ImmDisp, unsigned Size); - X86Operand *ParseIntelVarWithQualifier(const MCExpr *&Disp, - StringRef &Identifier); + bool ParseIntelIdentifier(const MCExpr *&Val, StringRef &Identifier, + InlineAsmIdentifierInfo &Info, + bool IsUnevaluatedOperand, SMLoc &End); + X86Operand *ParseMemOperand(unsigned SegReg, SMLoc StartLoc); X86Operand *CreateMemForInlineAsm(unsigned SegReg, const MCExpr *Disp, unsigned BaseReg, unsigned IndexReg, unsigned Scale, SMLoc Start, SMLoc End, - unsigned Size, StringRef SymName); - - bool ParseIntelDotOperator(const MCExpr *Disp, const MCExpr **NewDisp, - SmallString<64> &Err); + unsigned Size, StringRef Identifier, + InlineAsmIdentifierInfo &Info); bool ParseDirectiveWord(unsigned Size, SMLoc L); bool ParseDirectiveCode(StringRef IDVal, SMLoc L); @@ -511,26 +655,47 @@ private: bool processInstruction(MCInst &Inst, const SmallVectorImpl &Ops); + /// Wrapper around MCStreamer::EmitInstruction(). Possibly adds + /// instrumentation around Inst. + void EmitInstruction(MCInst &Inst, + SmallVectorImpl &Operands, + MCStreamer &Out); + bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, SmallVectorImpl &Operands, MCStreamer &Out, unsigned &ErrorInfo, - bool MatchingInlineAsm); + bool MatchingInlineAsm) override; - /// isSrcOp - Returns true if operand is either (%rsi) or %ds:%(rsi) - /// in 64bit mode or (%esi) or %es:(%esi) in 32bit mode. - bool isSrcOp(X86Operand &Op); + /// doSrcDstMatch - Returns true if operands are matching in their + /// word size (%si and %di, %esi and %edi, etc.). Order depends on + /// the parsing mode (Intel vs. AT&T). + bool doSrcDstMatch(X86Operand &Op1, X86Operand &Op2); - /// isDstOp - Returns true if operand is either (%rdi) or %es:(%rdi) - /// in 64bit mode or (%edi) or %es:(%edi) in 32bit mode. - bool isDstOp(X86Operand &Op); + /// Parses AVX512 specific operand primitives: masked registers ({%k}, {z}) + /// and memory broadcasting ({1to}) primitives, updating Operands vector if required. + /// \return \c true if no parsing errors occurred, \c false otherwise. + bool HandleAVX512Operand(SmallVectorImpl &Operands, + const MCParsedAsmOperand &Op); bool is64BitMode() const { // FIXME: Can tablegen auto-generate this? return (STI.getFeatureBits() & X86::Mode64Bit) != 0; } - void SwitchMode() { - unsigned FB = ComputeAvailableFeatures(STI.ToggleFeature(X86::Mode64Bit)); + bool is32BitMode() const { + // FIXME: Can tablegen auto-generate this? + return (STI.getFeatureBits() & X86::Mode32Bit) != 0; + } + bool is16BitMode() const { + // FIXME: Can tablegen auto-generate this? + return (STI.getFeatureBits() & X86::Mode16Bit) != 0; + } + void SwitchMode(uint64_t mode) { + uint64_t oldMode = STI.getFeatureBits() & + (X86::Mode64Bit | X86::Mode32Bit | X86::Mode16Bit); + unsigned FB = ComputeAvailableFeatures(STI.ToggleFeature(oldMode | mode)); setAvailableFeatures(FB); + assert(mode == (STI.getFeatureBits() & + (X86::Mode64Bit | X86::Mode32Bit | X86::Mode16Bit))); } bool isParsingIntelSyntax() { @@ -546,19 +711,24 @@ private: /// } public: - X86AsmParser(MCSubtargetInfo &sti, MCAsmParser &parser) - : MCTargetAsmParser(), STI(sti), Parser(parser), InstInfo(0) { + X86AsmParser(MCSubtargetInfo &sti, MCAsmParser &parser, + const MCInstrInfo &mii, + const MCTargetOptions &Options) + : MCTargetAsmParser(), STI(sti), Parser(parser), MII(mii), InstInfo(0) { // Initialize the set of available features. setAvailableFeatures(ComputeAvailableFeatures(STI.getFeatureBits())); + Instrumentation.reset( + CreateX86AsmInstrumentation(Options, Parser.getContext(), STI)); } - virtual bool ParseRegister(unsigned &RegNo, SMLoc &StartLoc, SMLoc &EndLoc); - virtual bool ParseInstruction(ParseInstructionInfo &Info, StringRef Name, - SMLoc NameLoc, - SmallVectorImpl &Operands); + bool ParseRegister(unsigned &RegNo, SMLoc &StartLoc, SMLoc &EndLoc) override; - virtual bool ParseDirective(AsmToken DirectiveID); + bool + ParseInstruction(ParseInstructionInfo &Info, StringRef Name, SMLoc NameLoc, + SmallVectorImpl &Operands) override; + + bool ParseDirective(AsmToken DirectiveID) override; }; } // end anonymous namespace @@ -569,423 +739,63 @@ static unsigned MatchRegisterName(StringRef Name); /// } -static bool isImmSExti16i8Value(uint64_t Value) { - return (( Value <= 0x000000000000007FULL)|| - (0x000000000000FF80ULL <= Value && Value <= 0x000000000000FFFFULL)|| - (0xFFFFFFFFFFFFFF80ULL <= Value && Value <= 0xFFFFFFFFFFFFFFFFULL)); -} - -static bool isImmSExti32i8Value(uint64_t Value) { - return (( Value <= 0x000000000000007FULL)|| - (0x00000000FFFFFF80ULL <= Value && Value <= 0x00000000FFFFFFFFULL)|| - (0xFFFFFFFFFFFFFF80ULL <= Value && Value <= 0xFFFFFFFFFFFFFFFFULL)); -} - -static bool isImmZExtu32u8Value(uint64_t Value) { - return (Value <= 0x00000000000000FFULL); -} - -static bool isImmSExti64i8Value(uint64_t Value) { - return (( Value <= 0x000000000000007FULL)|| - (0xFFFFFFFFFFFFFF80ULL <= Value && Value <= 0xFFFFFFFFFFFFFFFFULL)); -} - -static bool isImmSExti64i32Value(uint64_t Value) { - return (( Value <= 0x000000007FFFFFFFULL)|| - (0xFFFFFFFF80000000ULL <= Value && Value <= 0xFFFFFFFFFFFFFFFFULL)); -} -namespace { - -/// X86Operand - Instances of this class represent a parsed X86 machine -/// instruction. -struct X86Operand : public MCParsedAsmOperand { - enum KindTy { - Token, - Register, - Immediate, - Memory - } Kind; - - SMLoc StartLoc, EndLoc; - SMLoc OffsetOfLoc; - StringRef SymName; - bool AddressOf; - - struct TokOp { - const char *Data; - unsigned Length; - }; - - struct RegOp { - unsigned RegNo; - }; - - struct ImmOp { - const MCExpr *Val; - }; - - struct MemOp { - unsigned SegReg; - const MCExpr *Disp; - unsigned BaseReg; - unsigned IndexReg; - unsigned Scale; - unsigned Size; - }; - - union { - struct TokOp Tok; - struct RegOp Reg; - struct ImmOp Imm; - struct MemOp Mem; - }; - - X86Operand(KindTy K, SMLoc Start, SMLoc End) - : Kind(K), StartLoc(Start), EndLoc(End) {} - - StringRef getSymName() { return SymName; } - - /// getStartLoc - Get the location of the first token of this operand. - SMLoc getStartLoc() const { return StartLoc; } - /// getEndLoc - Get the location of the last token of this operand. - SMLoc getEndLoc() const { return EndLoc; } - /// getLocRange - Get the range between the first and last token of this - /// operand. - SMRange getLocRange() const { return SMRange(StartLoc, EndLoc); } - /// getOffsetOfLoc - Get the location of the offset operator. - SMLoc getOffsetOfLoc() const { return OffsetOfLoc; } - - virtual void print(raw_ostream &OS) const {} - - StringRef getToken() const { - assert(Kind == Token && "Invalid access!"); - return StringRef(Tok.Data, Tok.Length); - } - void setTokenValue(StringRef Value) { - assert(Kind == Token && "Invalid access!"); - Tok.Data = Value.data(); - Tok.Length = Value.size(); - } - - unsigned getReg() const { - assert(Kind == Register && "Invalid access!"); - return Reg.RegNo; - } - - const MCExpr *getImm() const { - assert(Kind == Immediate && "Invalid access!"); - return Imm.Val; - } - - const MCExpr *getMemDisp() const { - assert(Kind == Memory && "Invalid access!"); - return Mem.Disp; - } - unsigned getMemSegReg() const { - assert(Kind == Memory && "Invalid access!"); - return Mem.SegReg; - } - unsigned getMemBaseReg() const { - assert(Kind == Memory && "Invalid access!"); - return Mem.BaseReg; - } - unsigned getMemIndexReg() const { - assert(Kind == Memory && "Invalid access!"); - return Mem.IndexReg; - } - unsigned getMemScale() const { - assert(Kind == Memory && "Invalid access!"); - return Mem.Scale; - } - - bool isToken() const {return Kind == Token; } - - bool isImm() const { return Kind == Immediate; } - - bool isImmSExti16i8() const { - if (!isImm()) - return false; - - // If this isn't a constant expr, just assume it fits and let relaxation - // handle it. - const MCConstantExpr *CE = dyn_cast(getImm()); - if (!CE) - return true; - - // Otherwise, check the value is in a range that makes sense for this - // extension. - return isImmSExti16i8Value(CE->getValue()); - } - bool isImmSExti32i8() const { - if (!isImm()) - return false; - - // If this isn't a constant expr, just assume it fits and let relaxation - // handle it. - const MCConstantExpr *CE = dyn_cast(getImm()); - if (!CE) - return true; - - // Otherwise, check the value is in a range that makes sense for this - // extension. - return isImmSExti32i8Value(CE->getValue()); - } - bool isImmZExtu32u8() const { - if (!isImm()) - return false; - - // If this isn't a constant expr, just assume it fits and let relaxation - // handle it. - const MCConstantExpr *CE = dyn_cast(getImm()); - if (!CE) - return true; - - // Otherwise, check the value is in a range that makes sense for this - // extension. - return isImmZExtu32u8Value(CE->getValue()); - } - bool isImmSExti64i8() const { - if (!isImm()) - return false; - - // If this isn't a constant expr, just assume it fits and let relaxation - // handle it. - const MCConstantExpr *CE = dyn_cast(getImm()); - if (!CE) +static bool CheckBaseRegAndIndexReg(unsigned BaseReg, unsigned IndexReg, + StringRef &ErrMsg) { + // If we have both a base register and an index register make sure they are + // both 64-bit or 32-bit registers. + // To support VSIB, IndexReg can be 128-bit or 256-bit registers. + if (BaseReg != 0 && IndexReg != 0) { + if (X86MCRegisterClasses[X86::GR64RegClassID].contains(BaseReg) && + (X86MCRegisterClasses[X86::GR16RegClassID].contains(IndexReg) || + X86MCRegisterClasses[X86::GR32RegClassID].contains(IndexReg)) && + IndexReg != X86::RIZ) { + ErrMsg = "base register is 64-bit, but index register is not"; return true; - - // Otherwise, check the value is in a range that makes sense for this - // extension. - return isImmSExti64i8Value(CE->getValue()); - } - bool isImmSExti64i32() const { - if (!isImm()) - return false; - - // If this isn't a constant expr, just assume it fits and let relaxation - // handle it. - const MCConstantExpr *CE = dyn_cast(getImm()); - if (!CE) + } + if (X86MCRegisterClasses[X86::GR32RegClassID].contains(BaseReg) && + (X86MCRegisterClasses[X86::GR16RegClassID].contains(IndexReg) || + X86MCRegisterClasses[X86::GR64RegClassID].contains(IndexReg)) && + IndexReg != X86::EIZ){ + ErrMsg = "base register is 32-bit, but index register is not"; return true; - - // Otherwise, check the value is in a range that makes sense for this - // extension. - return isImmSExti64i32Value(CE->getValue()); - } - - bool isOffsetOf() const { - return OffsetOfLoc.getPointer(); - } - - bool needAddressOf() const { - return AddressOf; - } - - bool isMem() const { return Kind == Memory; } - bool isMem8() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 8); - } - bool isMem16() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 16); - } - bool isMem32() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 32); - } - bool isMem64() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 64); - } - bool isMem80() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 80); - } - bool isMem128() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 128); - } - bool isMem256() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 256); - } - - bool isMemVX32() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 32) && - getMemIndexReg() >= X86::XMM0 && getMemIndexReg() <= X86::XMM15; - } - bool isMemVY32() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 32) && - getMemIndexReg() >= X86::YMM0 && getMemIndexReg() <= X86::YMM15; - } - bool isMemVX64() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 64) && - getMemIndexReg() >= X86::XMM0 && getMemIndexReg() <= X86::XMM15; - } - bool isMemVY64() const { - return Kind == Memory && (!Mem.Size || Mem.Size == 64) && - getMemIndexReg() >= X86::YMM0 && getMemIndexReg() <= X86::YMM15; - } - - bool isAbsMem() const { - return Kind == Memory && !getMemSegReg() && !getMemBaseReg() && - !getMemIndexReg() && getMemScale() == 1; - } - - bool isReg() const { return Kind == Register; } - - void addExpr(MCInst &Inst, const MCExpr *Expr) const { - // Add as immediates when possible. - if (const MCConstantExpr *CE = dyn_cast(Expr)) - Inst.addOperand(MCOperand::CreateImm(CE->getValue())); - else - Inst.addOperand(MCOperand::CreateExpr(Expr)); - } - - void addRegOperands(MCInst &Inst, unsigned N) const { - assert(N == 1 && "Invalid number of operands!"); - Inst.addOperand(MCOperand::CreateReg(getReg())); - } - - void addImmOperands(MCInst &Inst, unsigned N) const { - assert(N == 1 && "Invalid number of operands!"); - addExpr(Inst, getImm()); - } - - void addMem8Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem16Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem32Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem64Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem80Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem128Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMem256Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMemVX32Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMemVY32Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMemVX64Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - void addMemVY64Operands(MCInst &Inst, unsigned N) const { - addMemOperands(Inst, N); - } - - void addMemOperands(MCInst &Inst, unsigned N) const { - assert((N == 5) && "Invalid number of operands!"); - Inst.addOperand(MCOperand::CreateReg(getMemBaseReg())); - Inst.addOperand(MCOperand::CreateImm(getMemScale())); - Inst.addOperand(MCOperand::CreateReg(getMemIndexReg())); - addExpr(Inst, getMemDisp()); - Inst.addOperand(MCOperand::CreateReg(getMemSegReg())); - } - - void addAbsMemOperands(MCInst &Inst, unsigned N) const { - assert((N == 1) && "Invalid number of operands!"); - // Add as immediates when possible. - if (const MCConstantExpr *CE = dyn_cast(getMemDisp())) - Inst.addOperand(MCOperand::CreateImm(CE->getValue())); - else - Inst.addOperand(MCOperand::CreateExpr(getMemDisp())); - } - - static X86Operand *CreateToken(StringRef Str, SMLoc Loc) { - SMLoc EndLoc = SMLoc::getFromPointer(Loc.getPointer() + Str.size()); - X86Operand *Res = new X86Operand(Token, Loc, EndLoc); - Res->Tok.Data = Str.data(); - Res->Tok.Length = Str.size(); - return Res; - } - - static X86Operand *CreateReg(unsigned RegNo, SMLoc StartLoc, SMLoc EndLoc, - bool AddressOf = false, - SMLoc OffsetOfLoc = SMLoc(), - StringRef SymName = StringRef()) { - X86Operand *Res = new X86Operand(Register, StartLoc, EndLoc); - Res->Reg.RegNo = RegNo; - Res->AddressOf = AddressOf; - Res->OffsetOfLoc = OffsetOfLoc; - Res->SymName = SymName; - return Res; - } - - static X86Operand *CreateImm(const MCExpr *Val, SMLoc StartLoc, SMLoc EndLoc){ - X86Operand *Res = new X86Operand(Immediate, StartLoc, EndLoc); - Res->Imm.Val = Val; - return Res; - } - - /// Create an absolute memory operand. - static X86Operand *CreateMem(const MCExpr *Disp, SMLoc StartLoc, SMLoc EndLoc, - unsigned Size = 0, - StringRef SymName = StringRef()) { - X86Operand *Res = new X86Operand(Memory, StartLoc, EndLoc); - Res->Mem.SegReg = 0; - Res->Mem.Disp = Disp; - Res->Mem.BaseReg = 0; - Res->Mem.IndexReg = 0; - Res->Mem.Scale = 1; - Res->Mem.Size = Size; - Res->SymName = SymName; - Res->AddressOf = false; - return Res; - } - - /// Create a generalized memory operand. - static X86Operand *CreateMem(unsigned SegReg, const MCExpr *Disp, - unsigned BaseReg, unsigned IndexReg, - unsigned Scale, SMLoc StartLoc, SMLoc EndLoc, - unsigned Size = 0, - StringRef SymName = StringRef()) { - // We should never just have a displacement, that should be parsed as an - // absolute memory operand. - assert((SegReg || BaseReg || IndexReg) && "Invalid memory operand!"); - - // The scale should always be one of {1,2,4,8}. - assert(((Scale == 1 || Scale == 2 || Scale == 4 || Scale == 8)) && - "Invalid scale!"); - X86Operand *Res = new X86Operand(Memory, StartLoc, EndLoc); - Res->Mem.SegReg = SegReg; - Res->Mem.Disp = Disp; - Res->Mem.BaseReg = BaseReg; - Res->Mem.IndexReg = IndexReg; - Res->Mem.Scale = Scale; - Res->Mem.Size = Size; - Res->SymName = SymName; - Res->AddressOf = false; - return Res; + } + if (X86MCRegisterClasses[X86::GR16RegClassID].contains(BaseReg)) { + if (X86MCRegisterClasses[X86::GR32RegClassID].contains(IndexReg) || + X86MCRegisterClasses[X86::GR64RegClassID].contains(IndexReg)) { + ErrMsg = "base register is 16-bit, but index register is not"; + return true; + } + if (((BaseReg == X86::BX || BaseReg == X86::BP) && + IndexReg != X86::SI && IndexReg != X86::DI) || + ((BaseReg == X86::SI || BaseReg == X86::DI) && + IndexReg != X86::BX && IndexReg != X86::BP)) { + ErrMsg = "invalid 16-bit base/index register combination"; + return true; + } + } } -}; - -} // end anonymous namespace. - -bool X86AsmParser::isSrcOp(X86Operand &Op) { - unsigned basereg = is64BitMode() ? X86::RSI : X86::ESI; - - return (Op.isMem() && - (Op.Mem.SegReg == 0 || Op.Mem.SegReg == X86::DS) && - isa(Op.Mem.Disp) && - cast(Op.Mem.Disp)->getValue() == 0 && - Op.Mem.BaseReg == basereg && Op.Mem.IndexReg == 0); + return false; } -bool X86AsmParser::isDstOp(X86Operand &Op) { - unsigned basereg = is64BitMode() ? X86::RDI : X86::EDI; +bool X86AsmParser::doSrcDstMatch(X86Operand &Op1, X86Operand &Op2) +{ + // Return true and let a normal complaint about bogus operands happen. + if (!Op1.isMem() || !Op2.isMem()) + return true; - return Op.isMem() && - (Op.Mem.SegReg == 0 || Op.Mem.SegReg == X86::ES) && - isa(Op.Mem.Disp) && - cast(Op.Mem.Disp)->getValue() == 0 && - Op.Mem.BaseReg == basereg && Op.Mem.IndexReg == 0; + // Actually these might be the other way round if Intel syntax is + // being used. It doesn't matter. + unsigned diReg = Op1.Mem.BaseReg; + unsigned siReg = Op2.Mem.BaseReg; + + if (X86MCRegisterClasses[X86::GR16RegClassID].contains(siReg)) + return X86MCRegisterClasses[X86::GR16RegClassID].contains(diReg); + if (X86MCRegisterClasses[X86::GR32RegClassID].contains(siReg)) + return X86MCRegisterClasses[X86::GR32RegClassID].contains(diReg); + if (X86MCRegisterClasses[X86::GR64RegClassID].contains(siReg)) + return X86MCRegisterClasses[X86::GR64RegClassID].contains(diReg); + // Again, return true and let another error happen. + return true; } bool X86AsmParser::ParseRegister(unsigned &RegNo, @@ -1015,7 +825,7 @@ bool X86AsmParser::ParseRegister(unsigned &RegNo, RegNo = MatchRegisterName(Tok.getString().lower()); if (!is64BitMode()) { - // FIXME: This should be done using Requires and + // FIXME: This should be done using Requires and // Requires so "eiz" usage in 64-bit instructions can be also // checked. // FIXME: Check AH, CH, DH, BH cannot be used in an instruction requiring a @@ -1097,6 +907,22 @@ bool X86AsmParser::ParseRegister(unsigned &RegNo, return false; } +X86Operand *X86AsmParser::DefaultMemSIOperand(SMLoc Loc) { + unsigned basereg = + is64BitMode() ? X86::RSI : (is32BitMode() ? X86::ESI : X86::SI); + const MCExpr *Disp = MCConstantExpr::Create(0, getContext()); + return X86Operand::CreateMem(/*SegReg=*/0, Disp, /*BaseReg=*/basereg, + /*IndexReg=*/0, /*Scale=*/1, Loc, Loc, 0); +} + +X86Operand *X86AsmParser::DefaultMemDIOperand(SMLoc Loc) { + unsigned basereg = + is64BitMode() ? X86::RDI : (is32BitMode() ? X86::EDI : X86::DI); + const MCExpr *Disp = MCConstantExpr::Create(0, getContext()); + return X86Operand::CreateMem(/*SegReg=*/0, Disp, /*BaseReg=*/basereg, + /*IndexReg=*/0, /*Scale=*/1, Loc, Loc, 0); +} + X86Operand *X86AsmParser::ParseOperand() { if (isParsingIntelSyntax()) return ParseIntelOperand(); @@ -1113,6 +939,8 @@ static unsigned getIntelMemOperandSize(StringRef OpStr) { .Cases("XWORD", "xword", 80) .Cases("XMMWORD", "xmmword", 128) .Cases("YMMWORD", "ymmword", 256) + .Cases("ZMMWORD", "zmmword", 512) + .Cases("OPAQUE", "opaque", -1U) // needs to be non-zero, but doesn't matter .Default(0); return Size; } @@ -1121,42 +949,40 @@ X86Operand * X86AsmParser::CreateMemForInlineAsm(unsigned SegReg, const MCExpr *Disp, unsigned BaseReg, unsigned IndexReg, unsigned Scale, SMLoc Start, SMLoc End, - unsigned Size, StringRef SymName) { - bool NeedSizeDir = false; - if (const MCSymbolRefExpr *SymRef = dyn_cast(Disp)) { - const MCSymbol &Sym = SymRef->getSymbol(); - // FIXME: The SemaLookup will fail if the name is anything other then an - // identifier. - // FIXME: Pass a valid SMLoc. - bool IsVarDecl = false; - unsigned tLength, tSize, tType; - SemaCallback->LookupInlineAsmIdentifier(Sym.getName(), NULL, tLength, tSize, - tType, IsVarDecl); + unsigned Size, StringRef Identifier, + InlineAsmIdentifierInfo &Info){ + // If this is not a VarDecl then assume it is a FuncDecl or some other label + // reference. We need an 'r' constraint here, so we need to create register + // operand to ensure proper matching. Just pick a GPR based on the size of + // a pointer. + if (isa(Disp) && !Info.IsVarDecl) { + unsigned RegNo = + is64BitMode() ? X86::RBX : (is32BitMode() ? X86::EBX : X86::BX); + return X86Operand::CreateReg(RegNo, Start, End, /*AddressOf=*/true, + SMLoc(), Identifier, Info.OpDecl); + } + + // We either have a direct symbol reference, or an offset from a symbol. The + // parser always puts the symbol on the LHS, so look there for size + // calculation purposes. + const MCBinaryExpr *BinOp = dyn_cast(Disp); + bool IsSymRef = + isa(BinOp ? BinOp->getLHS() : Disp); + if (IsSymRef) { if (!Size) { - Size = tType * 8; // Size is in terms of bits in this context. - NeedSizeDir = Size > 0; - } - // If this is not a VarDecl then assume it is a FuncDecl or some other label - // reference. We need an 'r' constraint here, so we need to create register - // operand to ensure proper matching. Just pick a GPR based on the size of - // a pointer. - if (!IsVarDecl) { - unsigned RegNo = is64BitMode() ? X86::RBX : X86::EBX; - return X86Operand::CreateReg(RegNo, Start, End, /*AddressOf=*/true, - SMLoc(), SymName); + Size = Info.Type * 8; // Size is in terms of bits in this context. + if (Size) + InstInfo->AsmRewrites->push_back(AsmRewrite(AOK_SizeDirective, Start, + /*Len=*/0, Size)); } } - if (NeedSizeDir) - InstInfo->AsmRewrites->push_back(AsmRewrite(AOK_SizeDirective, Start, - /*Len=*/0, Size)); - // When parsing inline assembly we set the base register to a non-zero value // if we don't know the actual value at this time. This is necessary to // get the matching correct in some cases. BaseReg = BaseReg ? BaseReg : 1; return X86Operand::CreateMem(SegReg, Disp, BaseReg, IndexReg, Scale, Start, - End, Size, SymName); + End, Size, Identifier, Info.OpDecl); } static void @@ -1191,6 +1017,7 @@ RewriteIntelBracExpression(SmallVectorImpl *AsmRewrites, } } assert (Found && "Unable to rewrite ImmDisp."); + (void)Found; } else { // We have a symbolic and an immediate displacement, but no displacement // before the bracketed expression. Put the immediate displacement @@ -1220,8 +1047,7 @@ RewriteIntelBracExpression(SmallVectorImpl *AsmRewrites, } } -X86Operand * -X86AsmParser::ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End) { +bool X86AsmParser::ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End) { const AsmToken &Tok = Parser.getTok(); bool Done = false; @@ -1243,7 +1069,7 @@ X86AsmParser::ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End) { Done = true; break; } - return ErrorOperand(Tok.getLoc(), "Unexpected token!"); + return Error(Tok.getLoc(), "unknown token in expression"); } case AsmToken::EndOfStatement: { Done = true; @@ -1252,47 +1078,89 @@ X86AsmParser::ParseIntelExpression(IntelExprStateMachine &SM, SMLoc &End) { case AsmToken::Identifier: { // This could be a register or a symbolic displacement. unsigned TmpReg; - const MCExpr *Disp = 0; + const MCExpr *Val; SMLoc IdentLoc = Tok.getLoc(); StringRef Identifier = Tok.getString(); if(!ParseRegister(TmpReg, IdentLoc, End)) { SM.onRegister(TmpReg); UpdateLocLex = false; break; - } else if (!getParser().parsePrimaryExpr(Disp, End)) { - if (X86Operand *Err = ParseIntelVarWithQualifier(Disp, Identifier)) - return Err; - - SM.onDispExpr(Disp, Identifier); + } else { + if (!isParsingInlineAsm()) { + if (getParser().parsePrimaryExpr(Val, End)) + return Error(Tok.getLoc(), "Unexpected identifier!"); + } else { + // This is a dot operator, not an adjacent identifier. + if (Identifier.find('.') != StringRef::npos) { + return false; + } else { + InlineAsmIdentifierInfo &Info = SM.getIdentifierInfo(); + if (ParseIntelIdentifier(Val, Identifier, Info, + /*Unevaluated=*/false, End)) + return true; + } + } + SM.onIdentifierExpr(Val, Identifier); UpdateLocLex = false; break; } - return ErrorOperand(Tok.getLoc(), "Unexpected identifier!"); + return Error(Tok.getLoc(), "Unexpected identifier!"); } - case AsmToken::Integer: + case AsmToken::Integer: { + StringRef ErrMsg; if (isParsingInlineAsm() && SM.getAddImmPrefix()) InstInfo->AsmRewrites->push_back(AsmRewrite(AOK_ImmPrefix, Tok.getLoc())); - SM.onInteger(Tok.getIntVal()); + // Look for 'b' or 'f' following an Integer as a directional label + SMLoc Loc = getTok().getLoc(); + int64_t IntVal = getTok().getIntVal(); + End = consumeToken(); + UpdateLocLex = false; + if (getLexer().getKind() == AsmToken::Identifier) { + StringRef IDVal = getTok().getString(); + if (IDVal == "f" || IDVal == "b") { + MCSymbol *Sym = + getContext().GetDirectionalLocalSymbol(IntVal, IDVal == "b"); + MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; + const MCExpr *Val = + MCSymbolRefExpr::Create(Sym, Variant, getContext()); + if (IDVal == "b" && Sym->isUndefined()) + return Error(Loc, "invalid reference to undefined symbol"); + StringRef Identifier = Sym->getName(); + SM.onIdentifierExpr(Val, Identifier); + End = consumeToken(); + } else { + if (SM.onInteger(IntVal, ErrMsg)) + return Error(Loc, ErrMsg); + } + } else { + if (SM.onInteger(IntVal, ErrMsg)) + return Error(Loc, ErrMsg); + } break; + } case AsmToken::Plus: SM.onPlus(); break; case AsmToken::Minus: SM.onMinus(); break; case AsmToken::Star: SM.onStar(); break; case AsmToken::Slash: SM.onDivide(); break; + case AsmToken::Pipe: SM.onOr(); break; + case AsmToken::Amp: SM.onAnd(); break; + case AsmToken::LessLess: + SM.onLShift(); break; + case AsmToken::GreaterGreater: + SM.onRShift(); break; case AsmToken::LBrac: SM.onLBrac(); break; case AsmToken::RBrac: SM.onRBrac(); break; case AsmToken::LParen: SM.onLParen(); break; case AsmToken::RParen: SM.onRParen(); break; } if (SM.hadError()) - return ErrorOperand(Tok.getLoc(), "Unexpected token!"); + return Error(Tok.getLoc(), "unknown token in expression"); - if (!Done && UpdateLocLex) { - End = Tok.getLoc(); - Parser.Lex(); // Consume the token. - } + if (!Done && UpdateLocLex) + End = consumeToken(); } - return 0; + return false; } X86Operand *X86AsmParser::ParseIntelBracExpression(unsigned SegReg, SMLoc Start, @@ -1309,10 +1177,10 @@ X86Operand *X86AsmParser::ParseIntelBracExpression(unsigned SegReg, SMLoc Start, // may have already parsed an immediate displacement before the bracketed // expression. IntelExprStateMachine SM(ImmDisp, /*StopOnLBrac=*/false, /*AddImmPrefix=*/true); - if (X86Operand *Err = ParseIntelExpression(SM, End)) - return Err; + if (ParseIntelExpression(SM, End)) + return 0; - const MCExpr *Disp; + const MCExpr *Disp = 0; if (const MCExpr *Sym = SM.getSym()) { // A symbolic displacement. Disp = Sym; @@ -1320,17 +1188,23 @@ X86Operand *X86AsmParser::ParseIntelBracExpression(unsigned SegReg, SMLoc Start, RewriteIntelBracExpression(InstInfo->AsmRewrites, SM.getSymName(), ImmDisp, SM.getImm(), BracLoc, StartInBrac, End); - } else { - // An immediate displacement only. - Disp = MCConstantExpr::Create(SM.getImm(), getContext()); } - // Parse the dot operator (e.g., [ebx].foo.bar). - if (Tok.getString().startswith(".")) { - SmallString<64> Err; + if (SM.getImm() || !Disp) { + const MCExpr *Imm = MCConstantExpr::Create(SM.getImm(), getContext()); + if (Disp) + Disp = MCBinaryExpr::CreateAdd(Disp, Imm, getContext()); + else + Disp = Imm; // An immediate displacement only. + } + + // Parse struct field access. Intel requires a dot, but MSVC doesn't. MSVC + // will in fact do global lookup the field name inside all global typedefs, + // but we don't emulate that. + if (Tok.getString().find('.') != StringRef::npos) { const MCExpr *NewDisp; - if (ParseIntelDotOperator(Disp, &NewDisp, Err)) - return ErrorOperand(Tok.getLoc(), Err); + if (ParseIntelDotOperator(Disp, NewDisp)) + return 0; End = Tok.getEndLoc(); Parser.Lex(); // Eat the field. @@ -1340,159 +1214,201 @@ X86Operand *X86AsmParser::ParseIntelBracExpression(unsigned SegReg, SMLoc Start, int BaseReg = SM.getBaseReg(); int IndexReg = SM.getIndexReg(); int Scale = SM.getScale(); - - if (isParsingInlineAsm()) - return CreateMemForInlineAsm(SegReg, Disp, BaseReg, IndexReg, Scale, Start, - End, Size, SM.getSymName()); - - // handle [-42] - if (!BaseReg && !IndexReg) { - if (!SegReg) - return X86Operand::CreateMem(Disp, Start, End, Size); - else - return X86Operand::CreateMem(SegReg, Disp, 0, 0, 1, Start, End, Size); + if (!isParsingInlineAsm()) { + // handle [-42] + if (!BaseReg && !IndexReg) { + if (!SegReg) + return X86Operand::CreateMem(Disp, Start, End, Size); + else + return X86Operand::CreateMem(SegReg, Disp, 0, 0, 1, Start, End, Size); + } + StringRef ErrMsg; + if (CheckBaseRegAndIndexReg(BaseReg, IndexReg, ErrMsg)) { + Error(StartInBrac, ErrMsg); + return 0; + } + return X86Operand::CreateMem(SegReg, Disp, BaseReg, IndexReg, Scale, Start, + End, Size); } - return X86Operand::CreateMem(SegReg, Disp, BaseReg, IndexReg, Scale, Start, - End, Size); + + InlineAsmIdentifierInfo &Info = SM.getIdentifierInfo(); + return CreateMemForInlineAsm(SegReg, Disp, BaseReg, IndexReg, Scale, Start, + End, Size, SM.getSymName(), Info); } // Inline assembly may use variable names with namespace alias qualifiers. -X86Operand *X86AsmParser::ParseIntelVarWithQualifier(const MCExpr *&Disp, - StringRef &Identifier) { - // We should only see Foo::Bar if we're parsing inline assembly. - if (!isParsingInlineAsm()) - return 0; +bool X86AsmParser::ParseIntelIdentifier(const MCExpr *&Val, + StringRef &Identifier, + InlineAsmIdentifierInfo &Info, + bool IsUnevaluatedOperand, SMLoc &End) { + assert (isParsingInlineAsm() && "Expected to be parsing inline assembly."); + Val = 0; - // If we don't see a ':' then there can't be a qualifier. - if (getLexer().isNot(AsmToken::Colon)) - return 0; + StringRef LineBuf(Identifier.data()); + SemaCallback->LookupInlineAsmIdentifier(LineBuf, Info, IsUnevaluatedOperand); - bool Done = false; const AsmToken &Tok = Parser.getTok(); - AsmToken IdentEnd = Tok; - while (!Done) { - switch (getLexer().getKind()) { - default: - Done = true; - break; - case AsmToken::Colon: - getLexer().Lex(); // Consume ':'. - if (getLexer().isNot(AsmToken::Colon)) - return ErrorOperand(Tok.getLoc(), "Expected ':' token!"); - getLexer().Lex(); // Consume second ':'. - if (getLexer().isNot(AsmToken::Identifier)) - return ErrorOperand(Tok.getLoc(), "Expected an identifier token!"); - break; - case AsmToken::Identifier: - IdentEnd = Tok; - getLexer().Lex(); // Consume the identifier. - break; - } + + // Advance the token stream until the end of the current token is + // after the end of what the frontend claimed. + const char *EndPtr = Tok.getLoc().getPointer() + LineBuf.size(); + while (true) { + End = Tok.getEndLoc(); + getLexer().Lex(); + + assert(End.getPointer() <= EndPtr && "frontend claimed part of a token?"); + if (End.getPointer() == EndPtr) break; } - unsigned Len = IdentEnd.getLoc().getPointer() - Identifier.data(); - Identifier = StringRef(Identifier.data(), Len + IdentEnd.getString().size()); + // Create the symbol reference. + Identifier = LineBuf; MCSymbol *Sym = getContext().GetOrCreateSymbol(Identifier); MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None; - Disp = MCSymbolRefExpr::Create(Sym, Variant, getParser().getContext()); - return 0; + Val = MCSymbolRefExpr::Create(Sym, Variant, getParser().getContext()); + return false; } -/// ParseIntelMemOperand - Parse intel style memory operand. -X86Operand *X86AsmParser::ParseIntelMemOperand(unsigned SegReg, - int64_t ImmDisp, - SMLoc Start) { - const AsmToken &Tok = Parser.getTok(); - SMLoc End; +/// \brief Parse intel style segment override. +X86Operand *X86AsmParser::ParseIntelSegmentOverride(unsigned SegReg, + SMLoc Start, + unsigned Size) { + assert(SegReg != 0 && "Tried to parse a segment override without a segment!"); + const AsmToken &Tok = Parser.getTok(); // Eat colon. + if (Tok.isNot(AsmToken::Colon)) + return ErrorOperand(Tok.getLoc(), "Expected ':' token!"); + Parser.Lex(); // Eat ':' + + int64_t ImmDisp = 0; + if (getLexer().is(AsmToken::Integer)) { + ImmDisp = Tok.getIntVal(); + AsmToken ImmDispToken = Parser.Lex(); // Eat the integer. - unsigned Size = getIntelMemOperandSize(Tok.getString()); - if (Size) { - Parser.Lex(); - assert ((Tok.getString() == "PTR" || Tok.getString() == "ptr") && - "Unexpected token!"); - Parser.Lex(); + if (isParsingInlineAsm()) + InstInfo->AsmRewrites->push_back( + AsmRewrite(AOK_ImmPrefix, ImmDispToken.getLoc())); + + if (getLexer().isNot(AsmToken::LBrac)) { + // An immediate following a 'segment register', 'colon' token sequence can + // be followed by a bracketed expression. If it isn't we know we have our + // final segment override. + const MCExpr *Disp = MCConstantExpr::Create(ImmDisp, getContext()); + return X86Operand::CreateMem(SegReg, Disp, /*BaseReg=*/0, /*IndexReg=*/0, + /*Scale=*/1, Start, ImmDispToken.getEndLoc(), + Size); + } } - // Parse ImmDisp [ BaseReg + Scale*IndexReg + Disp ]. - if (getLexer().is(AsmToken::Integer)) { - if (isParsingInlineAsm()) - InstInfo->AsmRewrites->push_back(AsmRewrite(AOK_ImmPrefix, - Tok.getLoc())); - int64_t ImmDisp = Tok.getIntVal(); - Parser.Lex(); // Eat the integer. - if (getLexer().isNot(AsmToken::LBrac)) - return ErrorOperand(Start, "Expected '[' token!"); + if (getLexer().is(AsmToken::LBrac)) return ParseIntelBracExpression(SegReg, Start, ImmDisp, Size); + + const MCExpr *Val; + SMLoc End; + if (!isParsingInlineAsm()) { + if (getParser().parsePrimaryExpr(Val, End)) + return ErrorOperand(Tok.getLoc(), "unknown token in expression"); + + return X86Operand::CreateMem(Val, Start, End, Size); } + InlineAsmIdentifierInfo Info; + StringRef Identifier = Tok.getString(); + if (ParseIntelIdentifier(Val, Identifier, Info, + /*Unevaluated=*/false, End)) + return 0; + return CreateMemForInlineAsm(/*SegReg=*/0, Val, /*BaseReg=*/0,/*IndexReg=*/0, + /*Scale=*/1, Start, End, Size, Identifier, Info); +} + +/// ParseIntelMemOperand - Parse intel style memory operand. +X86Operand *X86AsmParser::ParseIntelMemOperand(int64_t ImmDisp, SMLoc Start, + unsigned Size) { + const AsmToken &Tok = Parser.getTok(); + SMLoc End; + + // Parse ImmDisp [ BaseReg + Scale*IndexReg + Disp ]. if (getLexer().is(AsmToken::LBrac)) - return ParseIntelBracExpression(SegReg, Start, ImmDisp, Size); + return ParseIntelBracExpression(/*SegReg=*/0, Start, ImmDisp, Size); + assert(ImmDisp == 0); - if (!ParseRegister(SegReg, Start, End)) { - // Handel SegReg : [ ... ] - if (getLexer().isNot(AsmToken::Colon)) - return ErrorOperand(Start, "Expected ':' token!"); - Parser.Lex(); // Eat : - if (getLexer().isNot(AsmToken::LBrac)) - return ErrorOperand(Start, "Expected '[' token!"); - return ParseIntelBracExpression(SegReg, Start, ImmDisp, Size); + const MCExpr *Val; + if (!isParsingInlineAsm()) { + if (getParser().parsePrimaryExpr(Val, End)) + return ErrorOperand(Tok.getLoc(), "unknown token in expression"); + + return X86Operand::CreateMem(Val, Start, End, Size); } - const MCExpr *Disp = 0; + InlineAsmIdentifierInfo Info; StringRef Identifier = Tok.getString(); - if (getParser().parsePrimaryExpr(Disp, End)) + if (ParseIntelIdentifier(Val, Identifier, Info, + /*Unevaluated=*/false, End)) return 0; - if (!isParsingInlineAsm()) - return X86Operand::CreateMem(Disp, Start, End, Size); + if (!getLexer().is(AsmToken::LBrac)) + return CreateMemForInlineAsm(/*SegReg=*/0, Val, /*BaseReg=*/0, /*IndexReg=*/0, + /*Scale=*/1, Start, End, Size, Identifier, Info); + + Parser.Lex(); // Eat '[' - if (X86Operand *Err = ParseIntelVarWithQualifier(Disp, Identifier)) - return Err; + // Parse Identifier [ ImmDisp ] + IntelExprStateMachine SM(/*ImmDisp=*/0, /*StopOnLBrac=*/true, + /*AddImmPrefix=*/false); + if (ParseIntelExpression(SM, End)) + return 0; + + if (SM.getSym()) { + Error(Start, "cannot use more than one symbol in memory operand"); + return 0; + } + if (SM.getBaseReg()) { + Error(Start, "cannot use base register with variable reference"); + return 0; + } + if (SM.getIndexReg()) { + Error(Start, "cannot use index register with variable reference"); + return 0; + } - return CreateMemForInlineAsm(/*SegReg=*/0, Disp, /*BaseReg=*/0,/*IndexReg=*/0, - /*Scale=*/1, Start, End, Size, Identifier); + const MCExpr *Disp = MCConstantExpr::Create(SM.getImm(), getContext()); + // BaseReg is non-zero to avoid assertions. In the context of inline asm, + // we're pointing to a local variable in memory, so the base register is + // really the frame or stack pointer. + return X86Operand::CreateMem(/*SegReg=*/0, Disp, /*BaseReg=*/1, /*IndexReg=*/0, + /*Scale=*/1, Start, End, Size, Identifier, + Info.OpDecl); } /// Parse the '.' operator. bool X86AsmParser::ParseIntelDotOperator(const MCExpr *Disp, - const MCExpr **NewDisp, - SmallString<64> &Err) { + const MCExpr *&NewDisp) { const AsmToken &Tok = Parser.getTok(); int64_t OrigDispVal, DotDispVal; // FIXME: Handle non-constant expressions. - if (const MCConstantExpr *OrigDisp = dyn_cast(Disp)) { + if (const MCConstantExpr *OrigDisp = dyn_cast(Disp)) OrigDispVal = OrigDisp->getValue(); - } else { - Err = "Non-constant offsets are not supported!"; - return true; - } + else + return Error(Tok.getLoc(), "Non-constant offsets are not supported!"); - // Drop the '.'. - StringRef DotDispStr = Tok.getString().drop_front(1); + // Drop the optional '.'. + StringRef DotDispStr = Tok.getString(); + if (DotDispStr.startswith(".")) + DotDispStr = DotDispStr.drop_front(1); // .Imm gets lexed as a real. if (Tok.is(AsmToken::Real)) { APInt DotDisp; DotDispStr.getAsInteger(10, DotDisp); DotDispVal = DotDisp.getZExtValue(); - } else if (Tok.is(AsmToken::Identifier)) { - // We should only see an identifier when parsing the original inline asm. - // The front-end should rewrite this in terms of immediates. - assert (isParsingInlineAsm() && "Unexpected field name!"); - + } else if (isParsingInlineAsm() && Tok.is(AsmToken::Identifier)) { unsigned DotDisp; std::pair BaseMember = DotDispStr.split('.'); if (SemaCallback->LookupInlineAsmField(BaseMember.first, BaseMember.second, - DotDisp)) { - Err = "Unable to lookup field reference!"; - return true; - } + DotDisp)) + return Error(Tok.getLoc(), "Unable to lookup field reference!"); DotDispVal = DotDisp; - } else { - Err = "Unexpected token type!"; - return true; - } + } else + return Error(Tok.getLoc(), "Unexpected token type!"); if (isParsingInlineAsm() && Tok.is(AsmToken::Identifier)) { SMLoc Loc = SMLoc::getFromPointer(DotDispStr.data()); @@ -1502,7 +1418,7 @@ bool X86AsmParser::ParseIntelDotOperator(const MCExpr *Disp, Val)); } - *NewDisp = MCConstantExpr::Create(OrigDispVal + DotDispVal, getContext()); + NewDisp = MCConstantExpr::Create(OrigDispVal + DotDispVal, getContext()); return false; } @@ -1512,17 +1428,14 @@ X86Operand *X86AsmParser::ParseIntelOffsetOfOperator() { const AsmToken &Tok = Parser.getTok(); SMLoc OffsetOfLoc = Tok.getLoc(); Parser.Lex(); // Eat offset. - assert (Tok.is(AsmToken::Identifier) && "Expected an identifier"); const MCExpr *Val; + InlineAsmIdentifierInfo Info; SMLoc Start = Tok.getLoc(), End; StringRef Identifier = Tok.getString(); - if (getParser().parsePrimaryExpr(Val, End)) - return ErrorOperand(Start, "Unable to parse expression!"); - - const MCExpr *Disp = 0; - if (X86Operand *Err = ParseIntelVarWithQualifier(Disp, Identifier)) - return Err; + if (ParseIntelIdentifier(Val, Identifier, Info, + /*Unevaluated=*/false, End)) + return 0; // Don't emit the offset operator. InstInfo->AsmRewrites->push_back(AsmRewrite(AOK_Skip, OffsetOfLoc, 7)); @@ -1530,9 +1443,10 @@ X86Operand *X86AsmParser::ParseIntelOffsetOfOperator() { // The offset operator will have an 'r' constraint, thus we need to create // register operand to ensure proper matching. Just pick a GPR based on // the size of a pointer. - unsigned RegNo = is64BitMode() ? X86::RBX : X86::EBX; + unsigned RegNo = + is64BitMode() ? X86::RBX : (is32BitMode() ? X86::EBX : X86::BX); return X86Operand::CreateReg(RegNo, Start, End, /*GetAddress=*/true, - OffsetOfLoc, Identifier); + OffsetOfLoc, Identifier, Info.OpDecl); } enum IntelOperatorKind { @@ -1551,39 +1465,24 @@ X86Operand *X86AsmParser::ParseIntelOperator(unsigned OpKind) { const AsmToken &Tok = Parser.getTok(); SMLoc TypeLoc = Tok.getLoc(); Parser.Lex(); // Eat operator. - assert (Tok.is(AsmToken::Identifier) && "Expected an identifier"); - const MCExpr *Val; - AsmToken StartTok = Tok; + const MCExpr *Val = 0; + InlineAsmIdentifierInfo Info; SMLoc Start = Tok.getLoc(), End; StringRef Identifier = Tok.getString(); - if (getParser().parsePrimaryExpr(Val, End)) - return ErrorOperand(Start, "Unable to parse expression!"); + if (ParseIntelIdentifier(Val, Identifier, Info, + /*Unevaluated=*/true, End)) + return 0; - const MCExpr *Disp = 0; - if (X86Operand *Err = ParseIntelVarWithQualifier(Disp, Identifier)) - return Err; - - unsigned Length = 0, Size = 0, Type = 0; - if (const MCSymbolRefExpr *SymRef = dyn_cast(Val)) { - const MCSymbol &Sym = SymRef->getSymbol(); - // FIXME: The SemaLookup will fail if the name is anything other then an - // identifier. - // FIXME: Pass a valid SMLoc. - bool IsVarDecl; - if (!SemaCallback->LookupInlineAsmIdentifier(Sym.getName(), NULL, Length, - Size, Type, IsVarDecl)) - // FIXME: We don't warn on variables with namespace alias qualifiers - // because support still needs to be added in the frontend. - if (Identifier.equals(StartTok.getString())) - return ErrorOperand(Start, "Unable to lookup expr!"); - } - unsigned CVal; + if (!Info.OpDecl) + return ErrorOperand(Start, "unable to lookup expression"); + + unsigned CVal = 0; switch(OpKind) { default: llvm_unreachable("Unexpected operand kind!"); - case IOK_LENGTH: CVal = Length; break; - case IOK_SIZE: CVal = Size; break; - case IOK_TYPE: CVal = Type; break; + case IOK_LENGTH: CVal = Info.Length; break; + case IOK_SIZE: CVal = Info.Size; break; + case IOK_TYPE: CVal = Info.Type; break; } // Rewrite the type operator and the C or C++ type or variable in terms of an @@ -1597,11 +1496,11 @@ X86Operand *X86AsmParser::ParseIntelOperator(unsigned OpKind) { X86Operand *X86AsmParser::ParseIntelOperand() { const AsmToken &Tok = Parser.getTok(); - SMLoc Start = Tok.getLoc(), End; - StringRef AsmTokStr = Tok.getString(); + SMLoc Start, End; // Offset, length, type and size operators. if (isParsingInlineAsm()) { + StringRef AsmTokStr = Tok.getString(); if (AsmTokStr == "offset" || AsmTokStr == "OFFSET") return ParseIntelOffsetOfOperator(); if (AsmTokStr == "length" || AsmTokStr == "LENGTH") @@ -1612,14 +1511,23 @@ X86Operand *X86AsmParser::ParseIntelOperand() { return ParseIntelOperator(IOK_TYPE); } + unsigned Size = getIntelMemOperandSize(Tok.getString()); + if (Size) { + Parser.Lex(); // Eat operand size (e.g., byte, word). + if (Tok.getString() != "PTR" && Tok.getString() != "ptr") + return ErrorOperand(Start, "Expected 'PTR' or 'ptr' token!"); + Parser.Lex(); // Eat ptr. + } + Start = Tok.getLoc(); + // Immediate. if (getLexer().is(AsmToken::Integer) || getLexer().is(AsmToken::Minus) || getLexer().is(AsmToken::LParen)) { AsmToken StartTok = Tok; IntelExprStateMachine SM(/*Imm=*/0, /*StopOnLBrac=*/true, /*AddImmPrefix=*/false); - if (X86Operand *Err = ParseIntelExpression(SM, End)) - return Err; + if (ParseIntelExpression(SM, End)) + return 0; int64_t Imm = SM.getImm(); if (isParsingInlineAsm()) { @@ -1633,6 +1541,13 @@ X86Operand *X86AsmParser::ParseIntelOperand() { } if (getLexer().isNot(AsmToken::LBrac)) { + // If a directional label (ie. 1f or 2b) was parsed above from + // ParseIntelExpression() then SM.getSym() was set to a pointer to + // to the MCExpr with the directional local symbol and this is a + // memory operand not an immediate operand. + if (SM.getSym()) + return X86Operand::CreateMem(SM.getSym(), Start, End, Size); + const MCExpr *ImmExpr = MCConstantExpr::Create(Imm, getContext()); return X86Operand::CreateImm(ImmExpr, Start, End); } @@ -1643,23 +1558,22 @@ X86Operand *X86AsmParser::ParseIntelOperand() { "before bracketed expr."); // Parse ImmDisp [ BaseReg + Scale*IndexReg + Disp ]. - return ParseIntelMemOperand(/*SegReg=*/0, Imm, Start); + return ParseIntelMemOperand(Imm, Start, Size); } // Register. unsigned RegNo = 0; if (!ParseRegister(RegNo, Start, End)) { // If this is a segment register followed by a ':', then this is the start - // of a memory reference, otherwise this is a normal register reference. + // of a segment override, otherwise this is a normal register reference. if (getLexer().isNot(AsmToken::Colon)) return X86Operand::CreateReg(RegNo, Start, End); - getParser().Lex(); // Eat the colon. - return ParseIntelMemOperand(/*SegReg=*/RegNo, /*Disp=*/0, Start); + return ParseIntelSegmentOverride(/*SegReg=*/RegNo, Start, Size); } // Memory operand. - return ParseIntelMemOperand(/*SegReg=*/0, /*Disp=*/0, Start); + return ParseIntelMemOperand(/*Disp=*/0, Start, Size); } X86Operand *X86AsmParser::ParseATTOperand() { @@ -1698,6 +1612,73 @@ X86Operand *X86AsmParser::ParseATTOperand() { } } +bool +X86AsmParser::HandleAVX512Operand(SmallVectorImpl &Operands, + const MCParsedAsmOperand &Op) { + if(STI.getFeatureBits() & X86::FeatureAVX512) { + if (getLexer().is(AsmToken::LCurly)) { + // Eat "{" and mark the current place. + const SMLoc consumedToken = consumeToken(); + // Distinguish {1to} from {%k}. + if(getLexer().is(AsmToken::Integer)) { + // Parse memory broadcasting ({1to}). + if (getLexer().getTok().getIntVal() != 1) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected 1to at this point"); + Parser.Lex(); // Eat "1" of 1to8 + if (!getLexer().is(AsmToken::Identifier) || + !getLexer().getTok().getIdentifier().startswith("to")) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected 1to at this point"); + // Recognize only reasonable suffixes. + const char *BroadcastPrimitive = + StringSwitch(getLexer().getTok().getIdentifier()) + .Case("to8", "{1to8}") + .Case("to16", "{1to16}") + .Default(0); + if (!BroadcastPrimitive) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Invalid memory broadcast primitive."); + Parser.Lex(); // Eat "toN" of 1toN + if (!getLexer().is(AsmToken::RCurly)) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected } at this point"); + Parser.Lex(); // Eat "}" + Operands.push_back(X86Operand::CreateToken(BroadcastPrimitive, + consumedToken)); + // No AVX512 specific primitives can pass + // after memory broadcasting, so return. + return true; + } else { + // Parse mask register {%k1} + Operands.push_back(X86Operand::CreateToken("{", consumedToken)); + if (X86Operand *Op = ParseOperand()) { + Operands.push_back(Op); + if (!getLexer().is(AsmToken::RCurly)) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected } at this point"); + Operands.push_back(X86Operand::CreateToken("}", consumeToken())); + + // Parse "zeroing non-masked" semantic {z} + if (getLexer().is(AsmToken::LCurly)) { + Operands.push_back(X86Operand::CreateToken("{z}", consumeToken())); + if (!getLexer().is(AsmToken::Identifier) || + getLexer().getTok().getIdentifier() != "z") + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected z at this point"); + Parser.Lex(); // Eat the z + if (!getLexer().is(AsmToken::RCurly)) + return !ErrorAndEatStatement(getLexer().getLoc(), + "Expected } at this point"); + Parser.Lex(); // Eat the } + } + } + } + } + } + return true; +} + /// ParseMemOperand: segment: disp(basereg, indexreg, scale). The '%ds:' prefix /// has already been parsed if present. X86Operand *X86AsmParser::ParseMemOperand(unsigned SegReg, SMLoc MemStart) { @@ -1755,10 +1736,11 @@ X86Operand *X86AsmParser::ParseMemOperand(unsigned SegReg, SMLoc MemStart) { // If we reached here, then we just ate the ( of the memory operand. Process // the rest of the memory operand. unsigned BaseReg = 0, IndexReg = 0, Scale = 1; - SMLoc IndexLoc; + SMLoc IndexLoc, BaseLoc; if (getLexer().is(AsmToken::Percent)) { SMLoc StartLoc, EndLoc; + BaseLoc = Parser.getTok().getLoc(); if (ParseRegister(BaseReg, StartLoc, EndLoc)) return 0; if (BaseReg == X86::EIZ || BaseReg == X86::RIZ) { Error(StartLoc, "eiz and riz can only be used as index registers", @@ -1801,6 +1783,11 @@ X86Operand *X86AsmParser::ParseMemOperand(unsigned SegReg, SMLoc MemStart) { } // Validate the scale amount. + if (X86MCRegisterClasses[X86::GR16RegClassID].contains(BaseReg) && + ScaleVal != 1) { + Error(Loc, "scale factor in 16-bit address must be 1"); + return 0; + } if (ScaleVal != 1 && ScaleVal != 2 && ScaleVal != 4 && ScaleVal != 8){ Error(Loc, "scale factor in address must be 1, 2, 4 or 8"); return 0; @@ -1831,24 +1818,26 @@ X86Operand *X86AsmParser::ParseMemOperand(unsigned SegReg, SMLoc MemStart) { SMLoc MemEnd = Parser.getTok().getEndLoc(); Parser.Lex(); // Eat the ')'. - // If we have both a base register and an index register make sure they are - // both 64-bit or 32-bit registers. - // To support VSIB, IndexReg can be 128-bit or 256-bit registers. - if (BaseReg != 0 && IndexReg != 0) { - if (X86MCRegisterClasses[X86::GR64RegClassID].contains(BaseReg) && - (X86MCRegisterClasses[X86::GR16RegClassID].contains(IndexReg) || - X86MCRegisterClasses[X86::GR32RegClassID].contains(IndexReg)) && - IndexReg != X86::RIZ) { - Error(IndexLoc, "index register is 32-bit, but base register is 64-bit"); - return 0; - } - if (X86MCRegisterClasses[X86::GR32RegClassID].contains(BaseReg) && - (X86MCRegisterClasses[X86::GR16RegClassID].contains(IndexReg) || - X86MCRegisterClasses[X86::GR64RegClassID].contains(IndexReg)) && - IndexReg != X86::EIZ){ - Error(IndexLoc, "index register is 64-bit, but base register is 32-bit"); - return 0; - } + // Check for use of invalid 16-bit registers. Only BX/BP/SI/DI are allowed, + // and then only in non-64-bit modes. Except for DX, which is a special case + // because an unofficial form of in/out instructions uses it. + if (X86MCRegisterClasses[X86::GR16RegClassID].contains(BaseReg) && + (is64BitMode() || (BaseReg != X86::BX && BaseReg != X86::BP && + BaseReg != X86::SI && BaseReg != X86::DI)) && + BaseReg != X86::DX) { + Error(BaseLoc, "invalid 16-bit base register"); + return 0; + } + if (BaseReg == 0 && + X86MCRegisterClasses[X86::GR16RegClassID].contains(IndexReg)) { + Error(IndexLoc, "16-bit memory operand may not include only index register"); + return 0; + } + + StringRef ErrMsg; + if (CheckBaseRegAndIndexReg(BaseReg, IndexReg, ErrMsg)) { + Error(BaseLoc, ErrMsg); + return 0; } return X86Operand::CreateMem(SegReg, Disp, BaseReg, IndexReg, Scale, @@ -1945,43 +1934,35 @@ ParseInstruction(ParseInstructionInfo &Info, StringRef Name, SMLoc NameLoc, if (getLexer().isNot(AsmToken::EndOfStatement) && !isPrefix) { // Parse '*' modifier. - if (getLexer().is(AsmToken::Star)) { - SMLoc Loc = Parser.getTok().getLoc(); - Operands.push_back(X86Operand::CreateToken("*", Loc)); - Parser.Lex(); // Eat the star. - } - - // Read the first operand. - if (X86Operand *Op = ParseOperand()) - Operands.push_back(Op); - else { - Parser.eatToEndOfStatement(); - return true; - } - - while (getLexer().is(AsmToken::Comma)) { - Parser.Lex(); // Eat the comma. - - // Parse and remember the operand. - if (X86Operand *Op = ParseOperand()) - Operands.push_back(Op); - else { - Parser.eatToEndOfStatement(); - return true; + if (getLexer().is(AsmToken::Star)) + Operands.push_back(X86Operand::CreateToken("*", consumeToken())); + + // Read the operands. + while(1) { + if (X86Operand *Op = ParseOperand()) { + Operands.push_back(Op); + if (!HandleAVX512Operand(Operands, *Op)) + return true; + } else { + Parser.eatToEndOfStatement(); + return true; } - } + // check for comma and eat it + if (getLexer().is(AsmToken::Comma)) + Parser.Lex(); + else + break; + } - if (getLexer().isNot(AsmToken::EndOfStatement)) { - SMLoc Loc = getLexer().getLoc(); - Parser.eatToEndOfStatement(); - return Error(Loc, "unexpected token in argument list"); - } - } + if (getLexer().isNot(AsmToken::EndOfStatement)) + return ErrorAndEatStatement(getLexer().getLoc(), + "unexpected token in argument list"); + } - if (getLexer().is(AsmToken::EndOfStatement)) - Parser.Lex(); // Consume the EndOfStatement - else if (isPrefix && getLexer().is(AsmToken::Slash)) - Parser.Lex(); // Consume the prefix separator Slash + // Consume the EndOfStatement or the prefix separator Slash + if (getLexer().is(AsmToken::EndOfStatement) || + (isPrefix && getLexer().is(AsmToken::Slash))) + Parser.Lex(); if (ExtraImmOp && isParsingIntelSyntax()) Operands.push_back(X86Operand::CreateImm(ExtraImmOp, NameLoc, NameLoc)); @@ -2014,103 +1995,103 @@ ParseInstruction(ParseInstructionInfo &Info, StringRef Name, SMLoc NameLoc, delete &Op; } } - // Transform "ins[bwl] %dx, %es:(%edi)" into "ins[bwl]" - if (Name.startswith("ins") && Operands.size() == 3 && - (Name == "insb" || Name == "insw" || Name == "insl")) { - X86Operand &Op = *(X86Operand*)Operands.begin()[1]; - X86Operand &Op2 = *(X86Operand*)Operands.begin()[2]; - if (Op.isReg() && Op.getReg() == X86::DX && isDstOp(Op2)) { - Operands.pop_back(); - Operands.pop_back(); - delete &Op; - delete &Op2; - } - } - // Transform "outs[bwl] %ds:(%esi), %dx" into "out[bwl]" - if (Name.startswith("outs") && Operands.size() == 3 && - (Name == "outsb" || Name == "outsw" || Name == "outsl")) { - X86Operand &Op = *(X86Operand*)Operands.begin()[1]; - X86Operand &Op2 = *(X86Operand*)Operands.begin()[2]; - if (isSrcOp(Op) && Op2.isReg() && Op2.getReg() == X86::DX) { - Operands.pop_back(); - Operands.pop_back(); - delete &Op; - delete &Op2; + // Append default arguments to "ins[bwld]" + if (Name.startswith("ins") && Operands.size() == 1 && + (Name == "insb" || Name == "insw" || Name == "insl" || + Name == "insd" )) { + if (isParsingIntelSyntax()) { + Operands.push_back(X86Operand::CreateReg(X86::DX, NameLoc, NameLoc)); + Operands.push_back(DefaultMemDIOperand(NameLoc)); + } else { + Operands.push_back(X86Operand::CreateReg(X86::DX, NameLoc, NameLoc)); + Operands.push_back(DefaultMemDIOperand(NameLoc)); } } - // Transform "movs[bwl] %ds:(%esi), %es:(%edi)" into "movs[bwl]" - if (Name.startswith("movs") && Operands.size() == 3 && - (Name == "movsb" || Name == "movsw" || Name == "movsl" || - (is64BitMode() && Name == "movsq"))) { - X86Operand &Op = *(X86Operand*)Operands.begin()[1]; - X86Operand &Op2 = *(X86Operand*)Operands.begin()[2]; - if (isSrcOp(Op) && isDstOp(Op2)) { - Operands.pop_back(); - Operands.pop_back(); - delete &Op; - delete &Op2; + // Append default arguments to "outs[bwld]" + if (Name.startswith("outs") && Operands.size() == 1 && + (Name == "outsb" || Name == "outsw" || Name == "outsl" || + Name == "outsd" )) { + if (isParsingIntelSyntax()) { + Operands.push_back(DefaultMemSIOperand(NameLoc)); + Operands.push_back(X86Operand::CreateReg(X86::DX, NameLoc, NameLoc)); + } else { + Operands.push_back(DefaultMemSIOperand(NameLoc)); + Operands.push_back(X86Operand::CreateReg(X86::DX, NameLoc, NameLoc)); } } - // Transform "lods[bwl] %ds:(%esi),{%al,%ax,%eax,%rax}" into "lods[bwl]" - if (Name.startswith("lods") && Operands.size() == 3 && + + // Transform "lods[bwlq]" into "lods[bwlq] ($SIREG)" for appropriate + // values of $SIREG according to the mode. It would be nice if this + // could be achieved with InstAlias in the tables. + if (Name.startswith("lods") && Operands.size() == 1 && (Name == "lods" || Name == "lodsb" || Name == "lodsw" || - Name == "lodsl" || (is64BitMode() && Name == "lodsq"))) { - X86Operand *Op1 = static_cast(Operands[1]); - X86Operand *Op2 = static_cast(Operands[2]); - if (isSrcOp(*Op1) && Op2->isReg()) { - const char *ins; - unsigned reg = Op2->getReg(); - bool isLods = Name == "lods"; - if (reg == X86::AL && (isLods || Name == "lodsb")) - ins = "lodsb"; - else if (reg == X86::AX && (isLods || Name == "lodsw")) - ins = "lodsw"; - else if (reg == X86::EAX && (isLods || Name == "lodsl")) - ins = "lodsl"; - else if (reg == X86::RAX && (isLods || Name == "lodsq")) - ins = "lodsq"; - else - ins = NULL; - if (ins != NULL) { - Operands.pop_back(); - Operands.pop_back(); - delete Op1; - delete Op2; - if (Name != ins) - static_cast(Operands[0])->setTokenValue(ins); + Name == "lodsl" || Name == "lodsd" || Name == "lodsq")) + Operands.push_back(DefaultMemSIOperand(NameLoc)); + + // Transform "stos[bwlq]" into "stos[bwlq] ($DIREG)" for appropriate + // values of $DIREG according to the mode. It would be nice if this + // could be achieved with InstAlias in the tables. + if (Name.startswith("stos") && Operands.size() == 1 && + (Name == "stos" || Name == "stosb" || Name == "stosw" || + Name == "stosl" || Name == "stosd" || Name == "stosq")) + Operands.push_back(DefaultMemDIOperand(NameLoc)); + + // Transform "scas[bwlq]" into "scas[bwlq] ($DIREG)" for appropriate + // values of $DIREG according to the mode. It would be nice if this + // could be achieved with InstAlias in the tables. + if (Name.startswith("scas") && Operands.size() == 1 && + (Name == "scas" || Name == "scasb" || Name == "scasw" || + Name == "scasl" || Name == "scasd" || Name == "scasq")) + Operands.push_back(DefaultMemDIOperand(NameLoc)); + + // Add default SI and DI operands to "cmps[bwlq]". + if (Name.startswith("cmps") && + (Name == "cmps" || Name == "cmpsb" || Name == "cmpsw" || + Name == "cmpsl" || Name == "cmpsd" || Name == "cmpsq")) { + if (Operands.size() == 1) { + if (isParsingIntelSyntax()) { + Operands.push_back(DefaultMemSIOperand(NameLoc)); + Operands.push_back(DefaultMemDIOperand(NameLoc)); + } else { + Operands.push_back(DefaultMemDIOperand(NameLoc)); + Operands.push_back(DefaultMemSIOperand(NameLoc)); } + } else if (Operands.size() == 3) { + X86Operand &Op = *(X86Operand*)Operands.begin()[1]; + X86Operand &Op2 = *(X86Operand*)Operands.begin()[2]; + if (!doSrcDstMatch(Op, Op2)) + return Error(Op.getStartLoc(), + "mismatching source and destination index registers"); } } - // Transform "stos[bwl] {%al,%ax,%eax,%rax},%es:(%edi)" into "stos[bwl]" - if (Name.startswith("stos") && Operands.size() == 3 && - (Name == "stos" || Name == "stosb" || Name == "stosw" || - Name == "stosl" || (is64BitMode() && Name == "stosq"))) { - X86Operand *Op1 = static_cast(Operands[1]); - X86Operand *Op2 = static_cast(Operands[2]); - if (isDstOp(*Op2) && Op1->isReg()) { - const char *ins; - unsigned reg = Op1->getReg(); - bool isStos = Name == "stos"; - if (reg == X86::AL && (isStos || Name == "stosb")) - ins = "stosb"; - else if (reg == X86::AX && (isStos || Name == "stosw")) - ins = "stosw"; - else if (reg == X86::EAX && (isStos || Name == "stosl")) - ins = "stosl"; - else if (reg == X86::RAX && (isStos || Name == "stosq")) - ins = "stosq"; - else - ins = NULL; - if (ins != NULL) { - Operands.pop_back(); - Operands.pop_back(); - delete Op1; - delete Op2; - if (Name != ins) - static_cast(Operands[0])->setTokenValue(ins); + + // Add default SI and DI operands to "movs[bwlq]". + if ((Name.startswith("movs") && + (Name == "movs" || Name == "movsb" || Name == "movsw" || + Name == "movsl" || Name == "movsd" || Name == "movsq")) || + (Name.startswith("smov") && + (Name == "smov" || Name == "smovb" || Name == "smovw" || + Name == "smovl" || Name == "smovd" || Name == "smovq"))) { + if (Operands.size() == 1) { + if (Name == "movsd") { + delete Operands.back(); + Operands.back() = X86Operand::CreateToken("movsl", NameLoc); } + if (isParsingIntelSyntax()) { + Operands.push_back(DefaultMemDIOperand(NameLoc)); + Operands.push_back(DefaultMemSIOperand(NameLoc)); + } else { + Operands.push_back(DefaultMemSIOperand(NameLoc)); + Operands.push_back(DefaultMemDIOperand(NameLoc)); + } + } else if (Operands.size() == 3) { + X86Operand &Op = *(X86Operand*)Operands.begin()[1]; + X86Operand &Op2 = *(X86Operand*)Operands.begin()[2]; + if (!doSrcDstMatch(Op, Op2)) + return Error(Op.getStartLoc(), + "mismatching source and destination index registers"); } } @@ -2222,10 +2203,68 @@ processInstruction(MCInst &Inst, case X86::SBB16i16: return convert16i16to16ri8(Inst, X86::SBB16ri8); case X86::SBB32i32: return convert32i32to32ri8(Inst, X86::SBB32ri8); case X86::SBB64i32: return convert64i32to64ri8(Inst, X86::SBB64ri8); + case X86::VMOVAPDrr: + case X86::VMOVAPDYrr: + case X86::VMOVAPSrr: + case X86::VMOVAPSYrr: + case X86::VMOVDQArr: + case X86::VMOVDQAYrr: + case X86::VMOVDQUrr: + case X86::VMOVDQUYrr: + case X86::VMOVUPDrr: + case X86::VMOVUPDYrr: + case X86::VMOVUPSrr: + case X86::VMOVUPSYrr: { + if (X86II::isX86_64ExtendedReg(Inst.getOperand(0).getReg()) || + !X86II::isX86_64ExtendedReg(Inst.getOperand(1).getReg())) + return false; + + unsigned NewOpc; + switch (Inst.getOpcode()) { + default: llvm_unreachable("Invalid opcode"); + case X86::VMOVAPDrr: NewOpc = X86::VMOVAPDrr_REV; break; + case X86::VMOVAPDYrr: NewOpc = X86::VMOVAPDYrr_REV; break; + case X86::VMOVAPSrr: NewOpc = X86::VMOVAPSrr_REV; break; + case X86::VMOVAPSYrr: NewOpc = X86::VMOVAPSYrr_REV; break; + case X86::VMOVDQArr: NewOpc = X86::VMOVDQArr_REV; break; + case X86::VMOVDQAYrr: NewOpc = X86::VMOVDQAYrr_REV; break; + case X86::VMOVDQUrr: NewOpc = X86::VMOVDQUrr_REV; break; + case X86::VMOVDQUYrr: NewOpc = X86::VMOVDQUYrr_REV; break; + case X86::VMOVUPDrr: NewOpc = X86::VMOVUPDrr_REV; break; + case X86::VMOVUPDYrr: NewOpc = X86::VMOVUPDYrr_REV; break; + case X86::VMOVUPSrr: NewOpc = X86::VMOVUPSrr_REV; break; + case X86::VMOVUPSYrr: NewOpc = X86::VMOVUPSYrr_REV; break; + } + Inst.setOpcode(NewOpc); + return true; + } + case X86::VMOVSDrr: + case X86::VMOVSSrr: { + if (X86II::isX86_64ExtendedReg(Inst.getOperand(0).getReg()) || + !X86II::isX86_64ExtendedReg(Inst.getOperand(2).getReg())) + return false; + unsigned NewOpc; + switch (Inst.getOpcode()) { + default: llvm_unreachable("Invalid opcode"); + case X86::VMOVSDrr: NewOpc = X86::VMOVSDrr_REV; break; + case X86::VMOVSSrr: NewOpc = X86::VMOVSSrr_REV; break; + } + Inst.setOpcode(NewOpc); + return true; + } } } static const char *getSubtargetFeatureName(unsigned Val); + +void X86AsmParser::EmitInstruction( + MCInst &Inst, SmallVectorImpl &Operands, + MCStreamer &Out) { + Instrumentation->InstrumentInstruction(Inst, Operands, getContext(), MII, + Out); + Out.EmitInstruction(Inst, STI); +} + bool X86AsmParser:: MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, SmallVectorImpl &Operands, @@ -2234,7 +2273,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, assert(!Operands.empty() && "Unexpect empty operand list!"); X86Operand *Op = static_cast(Operands[0]); assert(Op->isToken() && "Leading operand should always be a mnemonic!"); - ArrayRef EmptyRanges = ArrayRef(); + ArrayRef EmptyRanges = None; // First, handle aliases that expand to multiple instructions. // FIXME: This should be replaced with a real .td file alias mechanism. @@ -2248,7 +2287,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, Inst.setOpcode(X86::WAIT); Inst.setLoc(IDLoc); if (!MatchingInlineAsm) - Out.EmitInstruction(Inst); + EmitInstruction(Inst, Operands, Out); const char *Repl = StringSwitch(Op->getToken()) @@ -2284,7 +2323,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, Inst.setLoc(IDLoc); if (!MatchingInlineAsm) - Out.EmitInstruction(Inst); + EmitInstruction(Inst, Operands, Out); Opcode = Inst.getOpcode(); return false; case Match_MissingFeature: { @@ -2336,25 +2375,25 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, unsigned Match1, Match2, Match3, Match4; Match1 = MatchInstructionImpl(Operands, Inst, ErrorInfoIgnore, - isParsingIntelSyntax()); + MatchingInlineAsm, isParsingIntelSyntax()); // If this returned as a missing feature failure, remember that. if (Match1 == Match_MissingFeature) ErrorInfoMissingFeature = ErrorInfoIgnore; Tmp[Base.size()] = Suffixes[1]; Match2 = MatchInstructionImpl(Operands, Inst, ErrorInfoIgnore, - isParsingIntelSyntax()); + MatchingInlineAsm, isParsingIntelSyntax()); // If this returned as a missing feature failure, remember that. if (Match2 == Match_MissingFeature) ErrorInfoMissingFeature = ErrorInfoIgnore; Tmp[Base.size()] = Suffixes[2]; Match3 = MatchInstructionImpl(Operands, Inst, ErrorInfoIgnore, - isParsingIntelSyntax()); + MatchingInlineAsm, isParsingIntelSyntax()); // If this returned as a missing feature failure, remember that. if (Match3 == Match_MissingFeature) ErrorInfoMissingFeature = ErrorInfoIgnore; Tmp[Base.size()] = Suffixes[3]; Match4 = MatchInstructionImpl(Operands, Inst, ErrorInfoIgnore, - isParsingIntelSyntax()); + MatchingInlineAsm, isParsingIntelSyntax()); // If this returned as a missing feature failure, remember that. if (Match4 == Match_MissingFeature) ErrorInfoMissingFeature = ErrorInfoIgnore; @@ -2371,7 +2410,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, if (NumSuccessfulMatches == 1) { Inst.setLoc(IDLoc); if (!MatchingInlineAsm) - Out.EmitInstruction(Inst); + EmitInstruction(Inst, Operands, Out); Opcode = Inst.getOpcode(); return false; } @@ -2478,11 +2517,9 @@ bool X86AsmParser::ParseDirective(AsmToken DirectiveID) { } else if (IDVal.startswith(".intel_syntax")) { getParser().setAssemblerDialect(1); if (getLexer().isNot(AsmToken::EndOfStatement)) { - if(Parser.getTok().getString() == "noprefix") { - // FIXME : Handle noprefix + // FIXME: Handle noprefix + if (Parser.getTok().getString() == "noprefix") Parser.Lex(); - } else - return true; } return false; } @@ -2496,7 +2533,7 @@ bool X86AsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) { for (;;) { const MCExpr *Value; if (getParser().parseExpression(Value)) - return true; + return false; getParser().getStreamer().EmitValue(Value, Size); @@ -2504,8 +2541,10 @@ bool X86AsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) { break; // FIXME: Improve diagnostic. - if (getLexer().isNot(AsmToken::Comma)) - return Error(L, "unexpected token in directive"); + if (getLexer().isNot(AsmToken::Comma)) { + Error(L, "unexpected token in directive"); + return false; + } Parser.Lex(); } } @@ -2515,22 +2554,29 @@ bool X86AsmParser::ParseDirectiveWord(unsigned Size, SMLoc L) { } /// ParseDirectiveCode -/// ::= .code32 | .code64 +/// ::= .code16 | .code32 | .code64 bool X86AsmParser::ParseDirectiveCode(StringRef IDVal, SMLoc L) { - if (IDVal == ".code32") { + if (IDVal == ".code16") { Parser.Lex(); - if (is64BitMode()) { - SwitchMode(); + if (!is16BitMode()) { + SwitchMode(X86::Mode16Bit); + getParser().getStreamer().EmitAssemblerFlag(MCAF_Code16); + } + } else if (IDVal == ".code32") { + Parser.Lex(); + if (!is32BitMode()) { + SwitchMode(X86::Mode32Bit); getParser().getStreamer().EmitAssemblerFlag(MCAF_Code32); } } else if (IDVal == ".code64") { Parser.Lex(); if (!is64BitMode()) { - SwitchMode(); + SwitchMode(X86::Mode64Bit); getParser().getStreamer().EmitAssemblerFlag(MCAF_Code64); } } else { - return Error(L, "unexpected directive " + IDVal); + Error(L, "unknown directive " + IDVal); + return false; } return false;